Principles of Development

This page intentionally left blank

Principles of Development
Fourth Edition

Lewis Wolpert | Cheryll Tickle

Thomas Jessell
Peter Lawrence
Elliot Meyerowitz
Elizabeth Robertson
Jim Smith
Great Clarendon Street, Oxford ox2 6dp
Oxford University Press is a department of the University of Oxford.
It furthers the University’s objective of excellence in research, scholarship,
and education by publishing worldwide in
Oxford New York
Auckland Cape Town Dar es Salaam Hong Kong Karachi
Kuala Lumpur Madrid Melbourne Mexico City Nairobi
New Delhi Shanghai Taipei Toronto
With offices in
Argentina Austria Brazil Chile Czech Republic France Greece
Guatemala Hungary Italy Japan Poland Portugal Singapore
South Korea Switzerland Thailand Turkey Ukraine Vietnam
Oxford is a registered trade mark of Oxford University Press
in the UK and in certain other countries
Published in the United States
by Oxford University Press Inc., New York

© Oxford University Press 2011

The moral rights of the authors have been asserted
Database right Oxford University Press (maker)
All rights reserved. No part of this publication may be reproduced,
stored in a retrieval system, or transmitted, in any form or by any means,
without the prior permission in writing of Oxford University Press,
or as expressly permitted by law, or under terms agreed with the appropriate
reprographics rights organization. Enquiries concerning reproduction
outside the scope of the above should be sent to the Rights Department,
Oxford University Press, at the address above
You must not circulate this book in any other binding or cover
and you must impose the same condition on any acquirer
British Library Cataloguing in Publication Data
Data available
Library of Congress Cataloging in Publication Data
Data available
Typeset by Techset Composition Ltd
Printed in the United States
on acid-free paper by
Courier, Kendallville, IN

ISBN 978–0–19–954907–8 (pbk)

ISBN 978–0–19–955428–7 (hbk)

1 3 5 7 9 10 8 6 4 2
Preface

Developmental biology is at the core of all biology of multicellular organisms. It deals

with the process by which the genes in the fertilized egg control cell behavior in the
embryo and so determine the character of the animal or plant. Evolution operates by
changing the development of the organism so that a better-adapted form develops.
With the application of advances in cell and molecular biology and, most recently,
genomics, the progress in developmental biology in recent years has been remarkable,
and an enormous amount of information is now available. In this fourth edition we
have included many recent advances, for example in the understanding of stem cells
(Chapter 10) and expanded our consideration of development in relation to evolution
(Chapter 15). We have also highlighted, wherever possible, the relevance to human
embryonic development.
Principles of Development is designed for undergraduates and the emphasis is on
principles and key concepts. Central to our approach is that development can be best
understood by understanding how genes control cell behavior. We have assumed that
students have some basic familiarity with cell biology and genetics, but all key con-
cepts, such as the control of gene activity, are explained in the text.
Conscious of the pressures on students, we have tried to make the principles as
clear as possible and to provide numerous summaries, in both words and pictures,
and have tried to avoid providing too much detail. The illustrations in this book are
a special feature and have been carefully designed and chosen to illuminate both
experiments and mechanisms.
We have resisted the temptation to cover every aspect of development and have,
instead, focused on those systems that best illuminate common principles. Indeed, a
theme that runs throughout the book is that universal principles govern the process
of development. At all stages, what we included has been guided by what we believe
undergraduates should know about development.
We have thus concentrated our attention on vertebrates and Drosophila, but not to
the exclusion of other organisms, such as the nematode and the sea urchin, where
they best illustrate a concept. Because Drosophila development is so well understood
and has been so influential, in in this edition, as in the previous edition, we have
started the book with it rather than with vertebrates. An important feature of our book
is the inclusion of plant development, which is usually neglected in general textbooks
of developmental biology. There have been striking advances in the understanding
of plant development in recent times, and some unique and important features have
emerged.
In this edition, we have made a new separate chapter to describe the embryology
and genetics of our vertebrate model organisms together with some of the important
methods used to study them (Chapter 3). The mechanisms involved in early verte-
brate development are then considered in the two subsequent chapters (Chapters 4
and 5). Other chapters have been extensively revised. Many new diagrams and pho-
tographs are included throughout the book.
The emphasis in the book is on early development and the laying down of body
plans and organ systems, such as limbs and the nervous system, but we also include
later aspects of development, including growth and regeneration. The book concludes
with a consideration of evolution and development.
vi Preface

In providing further reading, our prime concern has been to guide the student to
helpful papers rather than to give credit to all the scientists who have made major
contributions: to those whom we have neglected, we apologize.
For this new edition, Cheryll Tickle has joined Lewis Wolpert as a main co-author.
Each chapter has also been reviewed by a number of experts (see page xix), to whom
we give thanks. We made the initial revisions, which were then deciphered, edited,
and incorporated by our editor Eleanor Lawrence, whose expertise and influence
pervades the book. Eleanor’s input has been invaluable in ensuring that the informa-
tion in the book will be readily accessible to students. The new illustrations were
brilliantly drawn or adapted by Matthew McClements, who created the illustrations
for the first edition.
We are indebted to Bethan Lee and Jonathan Crowe at Oxford University Press, for
their help and patience throughout the preparation of this new edition.

L.W.

London
July 2010

C.T.

Bath
July 2010

www.oxfordtextbooks.co.uk/orc/wolpert4e/

This book is augmented by additional online materials for both adopters of the book
and their students.

For registered adopters:

Electronic artwork
Figures from the book are available to download, for use in lectures.

Journal Clubs
Discussion questions focused around primary literature articles that link to topics
featured in the book, to guide the process of assimilating knowledge from the research
literature.

For students:

Flashcard glossary
Flashcards, which can be downloaded to mobile devices, to help students test their
recall of key terminology.

Multiple choice questions

An extensive bank of questions to help students check their understanding of con-
cepts introduced in the book.

Web links and web activities

Links to web sites that augment the printed text, with commentary to explain the
relevance of the sites to the topics in the book, and activities to encourage active
engagement with the information held on these sites.
About the authors

Lewis Wolpert is Emeritus Professor of Biology as Applied to Medicine, in the Depart-

ment of Anatomy and Developmental Biology, University College London, London,
UK. He is the author of The Triumph of the Embryo, A Passion for Science, The
­Unnatural Nature of Science, and Six Impossible Things Before Breakfast.

Cheryll Tickle is Foulerton Research Professor of The Royal Society at the University
of Bath, Bath, UK.

Thomas Jessell is Professor of Biochemistry and Molecular Biophysics, a member of

the Center for Neurobiology and Behaviour, and a Howard Hughes Medical Institute
investigator at the Department of Biochemistry and Molecular Biophysics, University
of Columbia Medical Center, New York, USA. He is an author of Principles of Neural
Science and Essentials of Neural Science and Behaviour.

Peter Lawrence is in the Department of Zoology, University of Cambridge and

­Emeritus member of the Medical Research Council Laboratory of Molecular Biology,
Cambridge, UK. He is the author of The Making of a Fly.

Elliot Meyerowitz is the George W. Beadle Professor of Biology and Chair of the Divi-
sion of Biology at the California Institute of Technology, Pasadena, CA, USA.

Elizabeth Robertson is a Wellcome Trust Principal Fellow and Professor at the

Sir William Dunn School of Pathology at the University of Oxford, Oxford, UK.

Jim Smith is Director of the Medical Research Council National Institute for Medical
Research, London, UK.

Eleanor Lawrence is a freelance science writer and editor.

Matthew McClements is an illustrator who specializes in design for scientific, techni-

cal, and medical communication.
This page intentionally left blank
Summary of contents

Preface v
About the authors vii
Contents xi
Reviewer acknowledgements xix
Figure acknowledgements xx

Chapter 1 History and basic concepts 1

Chapter 2 Development of the Drosophila body plan 35

Chapter 3 Vertebrate development I: life cycles and experimental

techniques 93

Chapter 4 Vertebrate development II: axes and germ layers 128

Chapter 5 vertebrate development III: patterning the early nervous

system and the somites 173

Chapter 6 Development of nematodes, sea urchins, and ascidians 215

Chapter 7 Plant development 255

Chapter 8 Morphogenesis: change in form in the early embryo 289

Chapter 9 Germ cells, fertilization, and sex 329

Chapter 10 Cell differentiation and stem cells 365

Chapter 11 Organogenesis 411

Chapter 12 Development of the nervous system 468

Chapter 13 Growth and post-embryonic development 505

Chapter 14 Regeneration 535

Chapter 15 Evolution and development 556

Glossary 589
Index 605
This page intentionally left blank
Contents

Preface v 1.15 Patterning can involve the interpretation of positional

About the authors vii information 25

Summary of contents ix nn Box 1E Signal transduction and intracellular signaling 26

Reviewer acknowledgements xix 1.16 Lateral inhibition can generate spacing patterns 27

Figure acknowledgements xx 1.17 Localization of cytoplasmic determinants and asymmetric

cell division can make daughter cells different from each other 27
Chapter 1 History and basic concepts 1 nn Box 1F When development goes awry 28
nn Box 1A Basic stages of Xenopus laevis development 2 1.18 The embryo contains a generative rather than a
The origins of developmental biology 4 descriptive program 29
1.1 Aristotle first defined the problem of epigenesis and 1.19 The reliability of development is achieved by a
preformation 4 variety of means 30
1.2 Cell theory changed the conception of embryonic 1.20 The complexity of embryonic development is due
development and heredity 5 to the complexity of cells themselves 30
1.3 Two main types of development were originally proposed 6 1.21 Development is intimately involved in evolution 31
nn Box 1B The mitotic cell cycle 7 Summary 32
1.4 The discovery of induction showed that one group of
Summary to Chapter 1 32
cells could determine the development of neighboring cells 8
1.5 The study of development was stimulated by the coming
together of genetics and development 8 Chapter 2 Development of the Drosophila
body plan 35
1.6 Development is studied mainly through selected model
organisms 9 Drosophila life cycle and overall development 36
1.7 The first developmental genes were identified as 2.1 The early Drosophila embryo is a multinucleate
spontaneous mutations 11 syncytium 36
Summary 13 2.2 Cellularization is followed by gastrulation and
A conceptual tool kit 13 segmentation 38
1.8 Development involves the emergence of pattern, 2.3 After hatching, the Drosophila larva develops through
change in form, cell differentiation, and growth 14 several larval stages, pupates, and then undergoes
nn Box 1C Germ layers 15 metamorphosis to become an adult 38
1.9 Cell behavior provides the link between gene action 2.4 Many developmental genes have been identified in
and developmental processes 17 Drosophila through induced large-scale genetic screening 39
1.10 Genes control cell behavior by specifying which proteins Setting up the body axes 40
are made 17
2.5 The body axes are set up while the Drosophila embryo
1.11 The expression of developmental genes is under tight is still a syncytium 40
control 19
2.6 Maternal factors set up the body axes and direct the
nn Box 1D Tracking gene expression in embryos 20 early stage of Drosophila development 41
1.12 Development is progressive and the fate of cells
nn Box 2A Mutagenesis and genetic screening strategy
becomes determined at different times 21
for identifying developmental mutants in Drosophila 42
1.13 Inductive interactions can make cells different from
2.7 Three classes of maternal genes specify the antero-
each other 23
posterior axis 43
1.14 The response to inductive signals depends on the
2.8 Bicoid protein provides an antero-posterior gradient of a
state of the cell 25
morphogen 44
xii Contents

2.9 The posterior pattern is controlled by the gradients of 2.24 Segmentation genes stabilize parasegment boundaries
Nanos and Caudal proteins 46 and set up a focus of signaling at the boundary that
2.10 The anterior and posterior extremities of the embryo patterns the segment 74
are specified by cell-surface receptor activation 47 2.25 Insect epidermal cells become individually polarized
2.11 The dorso-ventral polarity of the embryo is specified in an antero-posterior direction in the plane of the epithelium 77
by localization of maternal proteins in the egg vitelline envelope 48 nn Box 2F Planar cell polarity in Drosophila 78
2.12 Positional information along the dorso-ventral axis 2.26 Some insects use different mechanisms for patterning
is provided by the Dorsal protein 49 the body plan 79
nn Box 2B The Toll signaling pathway: a multifunctional Summary 80
pathway 50 Specification of segment identity 81
Summary 50 2.27 Segment identity in Drosophila is specified by Hox genes 81
Localization of maternal determinants during oogenesis 51 2.28 Homeotic selector genes of the bithorax complex are
2.13 The antero-posterior axis of the Drosophila egg is responsible for diversification of the posterior segments 82
specified by signals from the preceding egg chamber and 2.29 The Antennapedia complex controls specification of
by interactions of the oocyte with follicle cells 52 anterior regions 84
2.14 Localization of maternal mRNAs to either end of the 2.30 The order of Hox gene expression corresponds to the
egg depends on the reorganization of the oocyte cytoskeleton 53 order of genes along the chromosome 84
2.15 The dorso-ventral axis of the egg is specified by 2.31 The Drosophila head region is specified by genes
movement of the oocyte nucleus followed by signaling other than the Hox genes 85
between oocyte and follicle cells 54
Summary 85
Summary 56
Summary to Chapter 2 86
Patterning the early embryo 56
2.16 The antero-posterior axis is divided up into broad Chapter 3 Vertebrate development I: life cycles
regions by gap-gene expression 57 and experimental techniques 93
2.17 Bicoid protein provides a positional signal for the Vertebrate life cycles and outlines of development 94
anterior expression of zygotic hunchback 57
3.1 The frog Xenopus laevis is the model amphibian for
2.18 The gradient in Hunchback protein activates and developmental studies 96
represses other gap genes 58
3.2 The zebrafish embryo develops around a large mass of yolk 101
2.19 The expression of zygotic genes along the dorso-
3.3 Birds and mammals resemble each other and differ
ventral axis is controlled by Dorsal protein 59
from Xenopus in some important features of early
nn Box 2C P-element-mediated transformation 60 development 103
nn Box 2D Targeted gene expression and misexpression 3.4 The early chicken embryo develops as a flat disc of
screening 61 cells overlying a massive yolk 103
2.20 The Decapentaplegic protein acts as a morphogen 3.5 Early development in the mouse involves the allocation
to pattern the dorsal region 63 of cells to form the placenta and extra-embryonic membranes 109
Summary 65 Experimental approaches to studying vertebrate
Activation of the pair-rule genes and the establishment development 113
of parasegments 66 3.6 Not all techniques are equally applicable to all vertebrates 114
2.21 Parasegments are delimited by expression of pair-rule nn Box 3A Gene-expression profiling by DNA
genes in a periodic pattern 66 microarray 115
2.22 Gap-gene activity positions stripes of pair-rule 3.7 Fate mapping and lineage tracing reveal which cells in
gene expression 68 the early embryo give rise to which adult structures 117
Summary 70 nn Box 3B Insertional mutagenesis and gene knock-outs
Segmentation genes and compartments 70 in mice: the Cre/loxP system 118
2.23 Expression of the engrailed gene delimits a 3.8 Developmental genes can be identified by spontaneous
cell-lineage boundary and defines a compartment 70 mutation and by large-scale mutagenesis screens 119
nn Box 2E Genetic mosaics and mitotic recombination 73 nn Box 3C Large-scale mutagenesis in zebrafish 120
 Contents xiii

3.9 Transgenic techniques enable animals to be produced 4.15 Members of the TGF-b family have been identified
with mutations in specific genes 120 as mesoderm inducers 157
3.10 Gene function can also be tested by transient 4.16 The zygotic expression of mesoderm-inducing and
transgenesis and gene silencing 123 patterning signals in Xenopus is activated by the combined
3.11 Gene regulatory networks in embryonic development actions of maternal VegT and Wnt signaling 158
can be revealed by chromatin immunoprecipitation techniques 124 4.17 Signals from the organizer pattern the mesoderm
Summary to Chapter 3 124 dorso-ventrally by antagonizing the effects of ventral signals 159
4.18 Threshold responses to gradients of signaling
Chapter 4 Vertebrate development II: axes and proteins are likely to pattern the mesoderm 161
germ layers 128 nn Box 4E A zebrafish gene regulatory network 162
Setting up the body axes 129 4.19 Mesoderm induction and patterning in the chick
4.1 The animal–vegetal axis is maternally determined in and mouse occurs during primitive-streak formation 163
Xenopus and zebrafish 129 Summary 165
4.2 Localized stabilization of the transcriptional regulator Summary to Chapter 4 166
β-catenin specifies the future dorsal side and the location
of the main embryonic organizer in Xenopus and zebrafish 130 Chapter 5 Vertebrate development III: patterning
nn Box 4A Intercellular protein signals in vertebrate the early nervous system and the somites 173
development 131 The role of the organizer and neural induction 175
4.3 Signaling centers develop on the dorsal side of Xenopus 5.1 The inductive capacity of the organizer changes
and zebrafish blastulas 133 during gastrulation 175
4.4 The antero-posterior and dorso-ventral axes of the chick 5.2 The neural plate is induced in the ectoderm 179
blastoderm are related to the primitive streak 136
nn Box 5A Chromatin-remodeling complexes 182
4.5 The definitive antero-posterior and dorso-ventral axes
nn Box 5B The FGF signaling pathway 184
of the mouse embryo are not recognizable early in development 138
5.3 The nervous system is initially patterned by signals
4.6 Movement of the distal visceral endoderm indicates the
from the mesoderm 185
definitive antero-posterior axis in the mouse embryo 140
5.4 Neural crest cells arise from the borders of the neural plate 186
4.7 The bilateral symmetry of the early embryo is broken
to produce left–right asymmetry of internal organs 141 Summary 186

nn Box 4B Fine-tuning Nodal signaling 143 Somite formation and antero-posterior patterning 187

Summary 145 5.5 Somites are formed in a well-defined order along the
antero-posterior axis 187
The origin and specification of the germ layers 145
nn Box 5C The Notch signaling pathway 190
4.8 A fate map of the amphibian blastula is constructed
by following the fate of labeled cells 146 5.6 Identity of somites along the antero-posterior axis is
specified by Hox gene expression 191
4.9 The fate maps of vertebrates are variations on a basic plan 147
nn Box 5D Retinoic acid; a small-molecule intercellular signal 192
4.10 Cells of early vertebrate embryos do not yet have
their fates determined and regulation is possible 149 nn Box 5E The Hox genes 194

4.11 In Xenopus the endoderm and ectoderm are specified 5.7 Deletion or overexpression of Hox genes causes
by maternal factors, but the mesoderm is induced from changes in axial patterning 197
ectoderm by signals from the vegetal region 150 5.8 Hox gene expression is activated in an anterior to
nn Box 4C Identical twins 152 posterior pattern 198

nn Box 4D Preimplantation genetic screening 153 5.9 The fate of somite cells is determined by signals
from the adjacent tissues 199
4.12 Mesoderm induction occurs during a limited period
in the blastula stage 154 Summary 201

4.13 Zygotic gene expression is turned on in Xenopus The initial regionalization of the vertebrate brain 202
at the mid-blastula transition 155 5.10 Local signaling centers pattern the brain along the
4.14 Mesoderm-inducing and patterning signals in Xenopus antero-posterior axis 203
are produced by the vegetal region, the organizer, and the 5.11 The hindbrain is segmented into rhombomeres by
ventral mesoderm 156 boundaries of cell-lineage restriction 203
xiv Contents

nn Box 5F Eph receptors and their ephrin ligands 205 Summary 243
5.12 Hox genes provide positional information in the Ascidians 244
developing hindbrain 206 6.15 Animal–vegetal and antero-posterior axes in the
5.13 Neural crest cells from the hindbrain migrate to ascidian embryo are defined before first cleavage 244
populate the branchial arches 207 6.16 In ascidians, muscle is specified by localized
5.14 The embryo is patterned by the neurula stage into cytoplasmic factors 246
organ-forming regions that can still regulate 208 6.17 Notochord, neural precursors, and mesenchyme in
Summary 209 ascidians require inducing signals from neighboring cells 247

Summary to Chapter 5 209 Summary 249

Summary to Chapter 6 249
Chapter 6 Development of nematodes, sea
urchins, and ascidians 215 Chapter 7 Plant development 255
Nematodes 216 7.1 The model plant Arabidopsis thaliana has a short
6.1 The antero-posterior axis in Caenorhabditis elegans life-cycle and a small diploid genome 256
is determined by asymmetric cell division 218 Embryonic development 258

nn Box 6A Gene silencing by antisense RNA and RNA 7.2 Plant embryos develop through several distinct stages 258
interference 220 nn Box 7A Angiosperm embryogenesis 259
6.2 The dorso-ventral axis in Caenorhabditis elegans is 7.3 Gradients of the signal molecule auxin establish the
determined by cell–cell interactions 221 embryonic apical–basal axis 260
6.3 Both asymmetric divisions and cell–cell interactions 7.4 Plant somatic cells can give rise to embryos and seedlings 263
specify cell fate in the early nematode embryo 223 nn Box 7B Transgenic plants 264
6.4 Hox genes specify positional identity along the Summary 264
antero-posterior axis in Caenorhabditis elegans 226
Meristems 265
6.5 The timing of events in nematode development is
7.5 A meristem contains a small, central zone of self-
under genetic control that involves microRNAs 226
renewing stem cells 265
nn Box 6B Gene silencing by microRNAs 228
7.6 The size of the stem-cell area in the meristem is kept
6.6 Vulval development is initiated by the induction of a constant by a feedback loop to the organizing center 266
small number of cells by short-range signals from a single
7.7 The fate of cells from different meristem layers can
inducing cell 229
be changed by changing their position 267
Summary 231
7.8 A fate map for the embryonic shoot meristem can be
Echinoderms 232 deduced using clonal analysis 268
6.7 The sea-urchin embryo develops into a free-swimming 7.9 Meristem development is dependent on signals
larva 232 from other parts of the plant 270
6.8 The sea-urchin egg is polarized along the 7.10 Gene activity patterns the proximo-distal and adaxial–
animal–vegetal axis 235 abaxial axes of leaves developing from the shoot meristem 270
6.9 The sea-urchin fate map is finely specified, 7.11 The regular arrangement of leaves on a stem is
yet considerable regulation is possible 236 generated by regulated auxin transport 272

6.10 The vegetal region of the sea-urchin embryo acts 7.12 Root tissues are produced from Arabidopsis root
as an organizer 236 apical meristems by a highly stereotyped pattern of cell
divisions 273
6.11 The sea-urchin vegetal region is demarcated by
the nuclear accumulation of β-catenin 238 7.13 Root hairs are specified by a combination of positional
information and lateral inhibition 275
6.12 The genetic control of the skeletogenic pathway
is known in considerable detail 238 Summary 276

6.13 The oral–aboral axis in sea urchins is related to the Flower development and control of flowering 276
plane of the first cleavage 241 7.14 Homeotic genes control organ identity in the flower 277
6.14 The oral ectoderm acts as an organizing region for 7.15 The Antirrhinum flower is patterned dorso-ventrally
the oral–aboral axis 242 as well as radially 279
 Contents xv

nn Box 7C The basic model for the patterning of the Cell migration 318
Arabidopsis flower 280 8.13 Neural crest migration is controlled by environmental
7.16 The internal meristem layer can specify floral meristem cues 318
patterning 281
Summary 320
7.17 The transition of a shoot meristem to a floral meristem
Directed dilation 320
is under environmental and genetic control 281
8.14 Later extension and stiffening of the notochord
Summary 283
occurs by directed dilation 321
Summary to Chapter 7 284
8.15 Circumferential contraction of hypodermal cells
elongates the nematode embryo 321
Chapter 8 Morphogenesis: change in form in
8.16 The direction of cell enlargement can determine the
the early embryo 289
form of a plant leaf 322
nn Box 8A Change in cell shape and cell movement 291
Summary 323
Cell adhesion 292
Summary to Chapter 8 323
8.1 Sorting out of dissociated cells demonstrates differences
in cell adhesiveness in different tissues 292
Chapter 9 Germ cells, fertilization, and sex 329
nn Box 8B Cell-adhesion molecules and cell junctions 293
The development of germ cells 330
8.2 Cadherins can provide adhesive specificity 294
9.1 Germ-cell fate is specified in some embryos by a distinct
Summary 295
germ plasm in the egg 331
Cleavage and formation of the blastula 295
9.2 In mammals germ cells are induced by cell–cell
8.3 The orientation of the mitotic spindle determines the interactions during development 333
plane of cleavage at cell division 297
9.3 Germ cells migrate from their site of origin to the gonad 334
8.4 Cells become polarized in the sea-urchin blastula and
9.4 Germ cells are guided to their final destination by
the mouse morula 298
chemical signals 334
8.5 Fluid accumulation as a result of tight-junction formation
9.5 Germ-cell differentiation involves a halving of
and ion transport forms the blastocoel of the mammalian
chromosome number by meiosis 335
blastocyst 300
nn Box 9A Polar bodies 338
8.6 Internal cavities can be created by cell death 301
9.6 Oocyte development can involve gene amplification and
Summary 302
contributions from other cells 338
Gastrulation movements 302
9.7 Factors in the cytoplasm maintain the totipotent potential of
8.7 Gastrulation in the sea urchin involves cell migration
the egg 339
and invagination 303
9.8 In mammals some genes controlling embryonic growth
8.8 Mesoderm invagination in Drosophila is due to changes in
are ‘imprinted’ 339
cell shape that are controlled by genes that pattern the
dorso-ventral axis 306 Summary 342

8.9 Germ-band extension in Drosophila involves myosin- Fertilization 342

dependent remodeling of cell junctions and cell intercalation 307 9.9 Fertilization involves cell-surface interactions between
8.10 Dorsal closure in Drosophila and ventral closure in egg and sperm 343
Caenorhabditis elegans are brought about by the action 9.10 Changes in the egg envelope at fertilization block
of filopodia 308 polyspermy 345
8.11 Vertebrate gastrulation involves several different 9.11 Sperm–egg fusion causes a calcium wave that results
types of tissue movement 309 in egg activation 346
nn Box 8C Convergent extension 310 Summary 348
Summary 315 Determination of the sexual phenotype 348
Neural tube formation 316 9.12 The primary sex-determining gene in mammals is on
8.12 Neural tube formation is driven by changes in cell the Y chromosome 349
shape and convergent extension 316 9.13 Mammalian sexual phenotype is regulated by gonadal
Summary 318 hormones 349
xvi Contents

9.14 The primary sex-determining signal in Drosophila is 10.15 The differentiated state of a cell can change by
the number of X chromosomes, and is cell autonomous 351 transdifferentiation 397
9.15 Somatic sexual development in Caenorhabditis is 10.16 Embryonic stem cells can proliferate and differentiate
determined by the number of X chromosomes 353 into many cell types in culture 399
9.16 Most flowering plants are hermaphrodites, but some nn Box 10B Testing ES cell potential in tetraploid blastocysts 399
produce unisexual flowers 354 10.17 Stem cells could be a key to regenerative medicine 400
9.17 Determination of germ-cell sex depends on both genetic nn Box 10C Induced pluripotent stem cells 401
constitution and intercellular signals 355
10.18 Various approaches can be used to generate
9.18 Various strategies are used for dosage compensation differentiated cells for cell-replacement therapies 403
of X-linked genes 356
Summary 405
Summary 359
Summary to Chapter 10 406
Summary to Chapter 9 360

Chapter 11 Organogenesis 411

Chapter 10 Cell differentiation and stem cells 365
The vertebrate limb 412
The control of gene expression 368
11.1 The vertebrate limb develops from a limb bud 412
10.1 Control of transcription involves both general and
11.2 Genes expressed in the lateral plate mesoderm are
tissue-specific transcriptional regulators 368
involved in specifying the position and type of limb 413
10.2 External signals can activate gene expression 370
11.3 The apical ectodermal ridge is required for limb
10.3 The maintenance and inheritance of patterns of gene outgrowth 415
activity depend on chemical and structural modifications to
11.4 Patterning of the limb bud involves positional
chromatin, as well as on gene-regulatory proteins 371
information 416
nn Box 10A Histones and Hox genes 374
11.5 How position along the proximo-distal axis of the
Summary 374 limb bud is specified is still a matter of debate 416
Models of cell differentiation 375 11.6 The polarizing region specifies position along the
10.4 All blood cells are derived from multipotent stem cells 375 limb’s antero-posterior axis 419
10.5 Colony-stimulating factors and intrinsic changes control 11.7 Sonic hedgehog produced by the polarizing region
differentiation of the hematopoietic lineages 378 is likely to be the primary morphogen patterning the
10.6 Developmentally regulated globin gene expression is antero-posterior axis of the limb 420
controlled by regulatory sequences far distant from the nn Box 11A Positional information and morphogen gradients 421
coding regions 379 11.8 Transcription factors might specify digit identity 422
10.7 The epithelia of adult mammalian skin and gut are nn Box 11B Too many fingers: mutations that affect antero-
continually replaced by derivatives of stem cells 382 posterior patterning can cause polydactyly 423
10.8 The MyoD family of genes determines differentiation nn Box 11C Sonic hedgehog signaling and the primary cilium 424
into muscle 385
11.9 The dorso-ventral axis of the limb is controlled by
10.9 The differentiation of muscle cells involves withdrawal the ectoderm 426
from the cell cycle, but is reversible 387
11.10 Development of the limb is integrated by interactions
10.10 Skeletal muscle and neural cells can be renewed between signaling centers 427
from stem cells in adults 388
11.11 Different interpretations of the same positional signals
10.11 Embryonic neural crest cells differentiate into a wide give different limbs 427
range of different cell types 389
11.12 Hox genes establish the polarizing region and also
10.12 Programmed cell death is under genetic control 392 provide a code for limb patterning 428
Summary 393 11.13 Self-organization may be involved in the development
The plasticity of gene expression 394 of the limb bud 430

10.13 Nuclei of differentiated cells can support 11.14 Limb muscle is patterned by the connective tissue 431
development 394 nn Box 11D Reaction–diffusion mechanisms 432
10.14 Patterns of gene activity in differentiated cells can 11.15 The initial development of cartilage, muscles, and
be changed by cell fusion 396 tendons is autonomous 433
 Contents xvii

11.16 Joint formation involves secreted signals and 12.2 The development of neurons in Drosophila involves
mechanical stimuli 433 asymmetric cell divisions and timed changes in gene
11.17 Separation of the digits is the result of programmed expression 472
cell death 434 12.3 Specification of vertebrate neuronal precursors also
Summary 434 involves lateral inhibition 473

Insect wings and legs 435 nn Box 12A Specification of the sensory organs of adult
Drosophila 474
11.18 Positional signals from compartment boundaries
pattern the wing imaginal disc 436 12.4 Neurons are formed in the proliferative zone of the
vertebrate neural tube and migrate outwards 475
11.19 A signaling center at the boundary between dorsal
and ventral compartments patterns the Drosophila wing nn Box 12B Timing the birth of cortical neurons 478
along the dorso-ventral axis 438 12.5 The pattern of differentiation of cells along the dorso-
11.20 The leg disc is patterned in a similar manner to the ventral axis of the spinal cord depends on ventral and dorsal
wing disc, except for the proximo-distal axis 439 signals 478

11.21 Butterfly wing markings are organized by additional 12.6 Neuronal subtypes in the ventral spinal cord are
positional fields 441 specified by the ventral to dorsal gradient of Shh 480

11.22 Different imaginal discs can have the same positional 12.7 Spinal cord motor neurons at different dorso-ventral
values 442 positions project to different trunk and limb muscles 481

Summary 443 12.8 Antero-posterior pattern in the spinal cord is determined

in response to secreted signals from the node and adjacent
Vertebrate and insect eyes 444
mesoderm 482
11.23 The vertebrate eye develops from the neural tube
Summary 483
and the ectoderm of the head 445
Axon navigation 484
11.24 Patterning of the Drosophila eye involves cell–cell
interactions 448 12.9 The growth cone controls the path taken by a
growing axon 485
11.25 Eye development in Drosophila is initiated by the
actions of the same transcription factors that specify 12.10 Motor neuron axons in the chick limb are guided by
eye-precursor cells in vertebrates 450 ephrin–Eph interactions 486

Summary 451 12.11 Axons crossing the midline are both attracted and
repelled 488
Internal organs: insect tracheal system, vertebrate
lungs, kidneys, blood vessels, heart, and teeth 451 12.12 Neurons from the retina make ordered connections
with visual centers in the brain 489
11.26 The Drosophila tracheal system is a model for
branching morphogenesis 452 Summary 492

11.27 The vertebrate lung also develops by branching of Synapse formation and refinement 493
epithelial tubes 453 12.13 Synapse formation involves reciprocal interactions 493
11.28 The development of kidney tubules involves reciprocal 12.14 Many motor neurons die during normal development 496
induction by the ureteric bud and surrounding mesenchyme 454
12.15 Neuronal cell death and survival involve both intrinsic
11.29 The vascular system develops by vasculogenesis and extrinsic factors 497
followed by angiogenesis 456
12.16 The map from eye to brain is refined by neural activity 497
11.30 The development of the vertebrate heart involves
Summary 499
specification of a mesodermal tube that is patterned along
its long axis 457 Summary to Chapter 12 500

11.31 A homeobox gene code specifies tooth identity 459

Chapter 13 Growth and post-embryonic
Summary 461 development 505
Summary to Chapter 11 461 Growth 505
13.1 Tissues can grow by cell proliferation, cell enlargement,
Chapter 12 Development of the nervous system 468 or accretion 506
Specification of cell identity in the nervous system 469 13.2 Cell proliferation is controlled by regulating entry into
12.1 Neurons in Drosophila arise from proneural clusters 470 the cell cycle 506
xviii Contents

13.3 Cell division in early development can be controlled by Regeneration in Hydra 548
an intrinsic developmental program 508 14.7 Hydra grows continuously but regeneration does
13.4 Organ size can be controlled by both intrinsic growth not require growth 548
programs and extracellular signals 509 14.8 The head region of Hydra acts both as an organizing
13.5 The amount of nourishment an embryo receives can region and as an inhibitor of inappropriate head formation 549
have profound effects in later life 511 14.9 Genes controlling regeneration in Hydra are similar
13.6 Determination of organ size involves coordination to those expressed in vertebrate embryos 551
of cell growth, cell division, and cell death 512 Summary 552
13.7 Body size is also controlled by the neuroendocrine Summary to Chapter 14 552
system in both insects and mammals 513
nn Box 13A Gradients of signaling molecules could Chapter 15 Evolution and development 556
determine organ size 514 nn Box 15A ‘Darwin’s finches’ 558
13.8 Growth of the long bones occurs in the growth plates 517 The evolution of development 559
13.9 Growth of vertebrate striated muscle is dependent 15.1 Genomic evidence is throwing light on the origin
on tension 519 of metazoans 559
13.10 Cancer can result from mutations in genes that 15.2 Multicellular organisms evolved from single-celled
control cell multiplication and differentiation 520 ancestors 560
13.11 Hormones control many features of plant growth 522 Summary 562
Summary 523 The evolutionary modification of embryonic development 562
Molting and metamorphosis 523 15.3 Hox gene complexes have evolved through gene
13.12 Arthropods have to molt in order to grow 524 duplication 563
13.13 Metamorphosis is under environmental and 15.4 Changes in Hox genes generated the elaboration of
hormonal control 524 vertebrate and arthropod body plans 565
Summary 527 15.5 The position and number of paired appendages in
Aging and senescence 527 insects is dependent on Hox gene expression 567

13.14 Genes can alter the timing of senescence 529 15.6 The basic body plan of arthropods and vertebrates is
similar, but the dorso-ventral axis is inverted 568
13.15 Cell senescence blocks cell multiplication 530
15.7 Limbs evolved from fins 569
Summary 531
15.8 Vertebrate and insect wings make use of evolutionarily
Summary to Chapter 13 531
conserved developmental mechanisms 574
15.9 The evolution of developmental differences can be based
Chapter 14 Regeneration 535
on changes in just a few genes 575
Limb and organ regeneration 536
15.10 Embryonic structures have acquired new functions
14.1 Amphibian limb regeneration involves cell during evolution 576
dedifferentiation and new growth 537
Summary 578
14.2 The limb blastema gives rise to structures with
Changes in the timing of developmental processes 579
positional values distal to the site of amputation 540
15.11 Evolution can be due to changes in the timing of
14.3 Retinoic acid can change proximo-distal positional
developmental events 579
values in regenerating limbs 542
15.12 The evolution of life histories has implications for
14.4 Insect limbs intercalate positional values by both
development 581
proximo-distal and circumferential growth 544
Summary 581
14.5 Heart regeneration in the zebrafish involves the
resumption of cell division by cardiomyocytes 546 Summary to Chapter 15 582

14.6 The mammalian peripheral nervous system can

regenerate 547 Glossary 587
Summary 547 Index 603
Reviewer acknowledgements

Many thanks to the following who kindly reviewed various Spencer M. Mass, SUNY New Paltz
parts of the book: Elliot Meyerowitz, California Institute of Technology
Michelle Arbeitman, University of Southern California Andre Pires da Silva, University of Texas Arlington
Haini Cai, University of Georgia Gary Radice, University of Richmond, Virginia
Scott Dougan, University of Georgia Ann Rougvie, University of Minnesota
Robert Drewell, Harvey Mudd College Margaret Saha, College of William and Mary
Deborah Garrity, Colorado State University Craig M. Scott, Clarion University
Ivan Gepner, Monmouth University Diane C. Slusarski, University of Iowa
Tanja Godenschwege, Florida Atlantic University David Stein, University of Texas Austin
Peter Holland, University of Oxford Leslie Stevens, University of Texas Austin
Douglas Houston, University of Iowa Daniel P. Szeto, Purdue University
Margaret Johnson, University of Alabama Keiko Torri, University of Washington
Douglas R. Kankel, Yale University Lance Urven, Marian University of Wisconsin
David Leaf, Western Washington University Matthew Wawersik, College of William and Mary
Barbara Lom, Davidson College Athula H. Wikramanayake, University of Miami
Morris F. Maduro, University of California, Riverside Nina Zanetti, Siena College
Vicki J. Martin, Appalachian State University Ted Zerucha, Appalachian State University
 Figure acknowledgements

Chapter 1 Fig. 2.26 from Yu, D., and Small, S.: Precise registration of
Fig. 1.1 a) courtesy of Alpesh Doshi, CRGH, London; b) reproduced gene expression boundaries by a repressive morphogen in
courtesy of the MRC/Wellcome-funded Human Developmental Drosophila. Curr. Biol, 2008, 18: 868–876.
Biology Resource Box 2D from Blagburn, J.M.: Engrailed expression in subsets of
adult Drosophila sensory neurons: an enhancer-trap study.
Chapter 2 Invert Neurosa. 2008, 8(3): 133–146.

Fig. 2.1 top photograph reproduced with permission from Turner, Fig. 2.30 illustration after Lawrence, P.: The Making of a Fly. Oxford:
F.R., Mahowald, A.P.: Scanning electron microscopy of Blackwell Scientific Publications, 1992.
Drosophila embryogenesis. I. The structure of the egg envelopes Fig. 2.33 from Yu, D., and Small, S.: Precise registration of
and the formation of the cellular blastoderm. Dev. Biol. 1976, gene expression boundaries by a repressive morphogen in
50: 95–108. Middle photograph reproduced with permission from Drosophila. Curr. Biol, 2008, 18: 868–876.
Turner, F.R., Mahowald, A.P.: Scanning electron microscopy of Fig. 2.38 from Goodman, R.M., Thombrel, S., Firtinal, Z., Grayl,
Drosophila melanogaster embryogenesis. III. Formation of the D., Betts, D., and Roebuck, J.: Sprinter: a novel transmembrane
head and caudal segments. Dev. Biol. 1979, 68: 96–109. protein required for Wg secretion and signalling. Dev, 2006,
Fig. 2.4 left photograph reproduced with permission from Turner, 133: 4901–4911.
F.R., Mahowald, A.P.: Scanning electron microscopy of Fig. 2.41 illustration after Lawrence, P.: The Making of a Fly. Oxford:
Drosophila melanogaster embryogenesis. II. Gastrulation and Blackwell Scientific Publications, 1992.
segmentation. Dev. Biol. 1977, 57: 403–416. Center photograph
Fig. 2.47 illustration after Lawrence, P.: The Making of a Fly. Oxford:
reproduced with permission from Alberts, B., Bray, D., Lewis, J.,
Blackwell Scientific Publications, 1992. Photograph reproduced
Raft, M., Roberts, K., Watson, J.D.: Molecular Biology of the Cell,
with permission from Bender, W., Akam, A., Karch, F., Beachy,
3rd edition. New York, Garland Publishing, 1994.
P.A., Peifer, M., Spierer, P., Lewis, E.B., Hogness, D.S.: Molecular
Fig. 2.10 reproduced with permission from Spriov, A., Fahmy, K., genetics of the bithorax complex in Drosophila melanogaster.
Schneider, M., Frei, E., Noll, Baumgartner, S. Formation of the Science 1983, 221: 23–29 (image on front cover). © 1983 American
bicoid morphogen gradient: an mRNA gradient dictates the Association for the Advancement of Science.
protein gradient. Dev, 2009, 136: 605–614.
Chapter 3
Fig. 2.12 photograph reproduced with permission from Griffiths,
A.J.H., Miller, J.H., Suzuki, D.T., Lewontin, R.C., Gelbart, W.M.: Fig. 3.3 top photograph reproduced with permission from Alberts, B.,
An Introduction to Genetic Analysis, 6th edition. New York: W.H. Bray, D., Lewis, J., Raff, M., Roberts, K., Watson, J.D.: Molecular
Freeman & Co., 1996. Biology of the Cell, 3rd edition. New York: Garland Publishing,
1994.
Fig. 2.13 reproduced with permission from Surkova, S., Kosman, D.,
Kozlov, M., Myasnikova, E., Samsonova, A.A., Spriov, A., Vanario- Fig. 3.5 photographs reproduced with permission from Kessel, R.G.,
Alonso, C.E., Samsonova, M., Reinitz, J. Charaterization of the Shih, C.Y.: Scanning Electron Microscopy in Biology: A Student’s
Drosophila segment determination morpheme. Dev. Biol. 2008, Atlas of Biological Organization. London, Springer-Verlag, 1974.
313: 844–862. © 1974 Springer-Verlag GmbH & Co. KG.
Fig. 2.17 illustration after González-Reyes, A., Elliott, H., St Johnston, Fig. 3.6 illustration after Balinsky, B.I.: An Introduction to
D.: Polarization of both major body axes in Drosophila by Embryology. Fourth edition. Philadelphia, W.B. Saunders, 1975.
gurken-torpedo signalling. Nature 1995, 375: 654–658. Fig. 3.14 top photograph reproduced with permission from Kispert,
Fig. 2.21 illustration after St Johnston, D.: Moving messages: the A., Ortner, H., Cooke, J., Herrmann, B.G.: The chick Brachyury
intracellular localization of mRNAs. Nat. Rev. Mol. Cell Biol. 2005, gene: developmental expression pattern and response to axial
6: 363–375. induction by localized activin. Dev. Biol. 1995, 168: 406–415.

Fig. 2.22 from Martin, S.G., Leclerc, V., Smith-Litière, K., St Johnston, Fig. 3.17 illustration after Patten, B.M.: Early Embryology of the Chick.
D.: The identification of novel genes required for Drosophila New York, Mc Graw-Hill, 1971.
anteroposterior axis formation in a germline clone screen using Fig. 3.21 top photograph reproduced with permission from Bloom,
GFP-Staufen. Dev. 2003, 130, 4201–4215. T.L.: The effects of phorbol ester on mouse blastomeres: a role
 Figure acknowledgements xxi

for protein kinase C in compaction? Development 1989, 106: 159– Fig. 5.15 illustration after Mangold, O.: Über die induktionsfahigkeit
171 Published by permission of The Company of Biologists Ltd. der verschiedenen bezirke der neurula von urodelen.
Fig. 3.23 illustration after Hogan, B., Beddington, R., Costantini, Naturwissenschaften 1933, 21: 761–766.
F., Lacy, E.: Manipulating the Mouse Embryo: A Laboratory Fig. 5.16 illustration after Kelly, O.G., Melton, D.A.: Induction and
Manual, 2nd edition. New York: Cold Spring Harbor Laboratory patterning of the vertebrate nervous system. Trends Genet. 1995,
Press, 1994. 11: 273–278.
Fig. 3.24 illustration adapted, with permission, from McMahon, A.P.: Fig. 5.21 illustration from Deschamps, J., van Nes, J.: Developmental
Mouse development. Winged-helix in axial patterning. Curr. regulation of the Hox genes during axial morphogenesis in the
Biol. 1994, 4: 903–906. mouse. Development 2005, 132: 2931–2942.
Fig. 3.26 illustration after Kaufman, M.H.: The Atlas of Mouse Fig. 5.26 illustration after Burke, A.C., Nelson, C.E., Morgan, B.A.,
Development. London: Academic Press, 1992. Tabin, C.: Hox genes and the evolution of vertebrate axial
morphology. Development 1995, 121: 333–346.
Chapter 4
Fig. 5.31 illustration after Johnson, R.L., Laufer, E., Riddle, R.D.,
Fig. 4.14 illustration after Rodriguez, T.A., Srinivas, S., Clements, Tabin, C.: Ectopic expression of Sonic hedgehog alters dorsal-
M.P., Smith, J.C., Beddington, R.S.: Induction and migration ventral patterning of somites. Cell 1994, 79: 1165–1173.
of the anterior visceral endoderm is regulated by the extra-
Fig. 5.33 illustration adapted, with permission, from Lumsden, A.:
embryonic ectoderm. Development 2005, 132: 2513–2520.
Cell lineage restrictions in the chick embryo hindbrain. Phil.
Fig. 4.37 photograph reproduced with permission from Smith W.C., Trans. R. Soc. Lond. B 1991, 331: 281–286.
Harland, R.M.: Expression cloning of noggin, a new dorsalizing
Fig. 5.34 illustration adapted, with permission, from Lumsden, A.:
factor localized to the Spemann organizer in Xenopus embryos.
Cell lineage restrictions in the chick embryo hindbrain. Phil.
Cell 1992, 70: 829–840. © 1992 Cell Press.
Trans. R. Soc. Lond. B 1991, 331: 281–286.
Fig. 4.38 adapted from Heasman, J.: Patterning the early zenopus
Fig. 5.36 photograph reproduced with permission from Lumsden,
embryo. Dev 2006, 133: 1205–1217.
A., Krumlauf, R.: Patterning the vertebrate neuraxis. Science
Fig. 4.44 illustration after Beddington, S.P., Robertson, E.J.: Axis 1996, 274: 1109–1115 (image on front cover). © 1996 American
development and early asymmetry in mammals. Cell 1999, 96: Association for the Advancement of Science.
195–209.
Fig. 5.37 illustration after Krumlauf, R.: Hox genes and pattern
Box 4C adapted from Larsen, W.J.: Human Embryology, 2nd edn, formation in the branchial region of the vertebrate head. Trends
Churchill Livingstone, 1997, p. 481. Genet. 1993, 9: 106–112.
Box 4E from Morley, R.H., Lachani, K., Keefe, D., Gilchrist, M.J.,
Flicek, P., Smith, J.C., Wardle, F.C.: A gene regulatory network
directed by zebrafish No tail accounts for its roles in mesoderm
Chapter 6
formation. Proc Natl Acad Sci USA, 2009, 106: 3829–3834.
Fig. 6.5 photograph reproduced with permission from Strome, S.,
Chapter 5 Wood, W.B.: Generation of asymmetry and segregation of germ-
Fig. 5.2 photograph reproduced with permission from Hausen, P., line granules in early C. elegans embryos. Cell 1983 35: 15–25.
Riebesell, M.: The Early Development of Xenopus laevis. Berlin: © 1983 Cell Press.
Springer-Verlag, 1991. Fig. 6.6 illustration after Sulston, J.E., Schierenberg, E., White, J.G.,
Fig. 5.7 from Stern, C.D.: Neural induction: old problem, new Thompson, J.N.: The embryonic cell lineage of the nematode C.
findings, yet more questions. Dev, 2005, 132, 2007–21. elegans. Dev. Biol. 1983, 100: 69–119.
Fig. 5.9 from Kiecker, C., and Niehrs, C.: A morphogen gradient of Fig. 6.7 photographs reproduced with permission from Wood W.B:
Wnt/beta- catenin signalling regulates anteroposterior neural Evidence from reversal of handedness in C. elegans embryos for
patterning in Xenopus. Dev, 2001, 128: 4189–4201. early cell interactions determining cell fates. Nature 1991, 349:
Fig. 5.12 illustration after Kintner, C.R., Dodd, J.: Hensen’s node 536–538. © 1991 Macmillan Magazines Ltd.
induces neural tissue in Xenopus ectoderm. Implications for the Fig. 6.8 illustration after Mello, C.C., Draper, B.W., Priess, J.R.: The
action of the organizer in neural induction. Development 1991, maternal genes apx-1 and glp-1 and establishment of dorsal-
113: 1495–1505. ventral polarity in the early C. elegans embryo. Cell 1994, 77:
Fig. 5.13 adapted from Delfino-Machin, M. et. al.: Specification and 95–106.
maintenance of the spinal cord stem zone. Development 2005, Fig. 6.10 from Mizumoto, K., and Sawa, H.: Two bs or not two bs:
132: 4273–4283. regulation of asymmetric division by b-catenin. TRENDS in Cell
Fig. 5.14 illustration from Kudoh, T. et. al.: Combinatorial Fgf and Biology, 2007, Vol. 17 No. 10.
Bmp signalling patterns the gastrula ectoderm into prospective Fig. 6.12 illustration after Bürglin, T.R., Ruvkun, G.: The
neural and epidermal domains. Development 2004, 131: 3581– Caenorhabditis elegans homeobox gene cluster. Curr. Opin. Gen.
3592. Dev. 1993, 3: 615–620.
xxii Figure acknowledgments

Fig. 6.18 top and middle photographs reproduced with permission after Sachs, T.: Pattern Formation in Plant Tissues. Cambridge:
from Jim Coffman. Lower photograph reproduced with permission Cambridge University Press, 1994.
from Oxford Scientific Films. Fig. 7.20 adapted from Heisler, M.G. et al.: Patterns of auxin
Fig. 6.23 illustration from Oliveri, P., Davidson, E.H.: Gene transport and gene expression during primordium development
regulatory network controlling embryonic specification in the revealed by live imaging of the Arabadopsis inflorescence
sea urchin. Curr. Opin. Genet. Dev. 2004, 14: 351–360. meristem. Curr. Biol. 2005, 15: 1899–1911.
Fig. 6.24 adapted from Oliveri, P., Tu, Q., Davidson, E.H.: Global Fig. 7.22 illustration after Scheres, B., Wolkenfelt, H., Willemsen, V.,
regulatory logic for specification of an embryonic cell lineage. Terlouw, M., Lawson, E., Dean, C., Weisbeek, P.: Embryonic origin
Proc. Natl Acad. Sci. USA 2008, 105: 5955–5962. of the Arabidopsis primary root and root meristem initials.
Fig. 6.25 illustration from Oliveri, P., Davidson, E.H.: Gene Development 1994, 120: 2475–2487.
regulatory network controlling embryonic specification in the Fig. 7.24 photograph reproduced with permission from Meyerowitz,
sea urchin. Curr. Opin. Genet. Dev. 2004, 14: 351–360. E.M., Bowman, J.L., Brockman, L.L., Drews, G.N., Jack, T.,
Fig. 6.28 middle photograph reproduced with permission from Corbo, Sieburth, L.E., Weigel, D.: A genetic and molecular model for
J.C., Levine, M., Zeller, R.W.: Characterization of a notochord- flower development in Arabidopsis thaliana. Development Suppl.
specific enhancer from the Brachyury promoter region of the 1991, 157–167. Published by permission of The Company of
ascidian, Ciona intestinalis. Development 1997, 124: 589–602. Biologists Ltd.
Published by permission of The Company of Biologists Ltd. Top Fig. 7.25 illustration after Coen, E.S., Meyerowitz, E.M.: The war of
photograph reproduced with permission from Shigeki Fujiwara and the whorls: genetic interactions controlling flower development.
Naoki Shimozono. Lower photograph reproduced with permission Nature 1991, 353: 31–37.
from Andrew Martinez. Fig. 7.26 photographs reproduced with permission from Meyerowitz,
Fig. 6.30 illustration from Nishida, H.: Specification of embryonic E.M., Bowman, J.L., Brockman, L.L., Drews, G.N., Jack, T.,
axis and mosaic development in ascidians. Dev. Dyn. 2005, 233: Sieburth, L.E., Weigel, D.: A genetic and molecular model for
1177–1193. flower development in Arabidopsis thaliana. Development Suppl.
Fig. 6.31 illustration after Conklin, E.G.: The organization and cell 1991, 157–167. Published by permission of The Company of
lineage of the ascidian egg. J. Acad. Nat. Sci. Philadelphia 1905, Biologists Ltd. (left panel); center panel from Bowman, J.L., Smyth,
13: 1–119. D.R., Meyerowitz, E.M.: Genes directing flower development in
Arabidopsis. Plant Cell 1989, 1: 37–52. Published by permission of
Fig. 6.32 Picco, V., Hudson, C., Yasuo, H.: Ephrin-Eph signalling
The American Society of Plant Physiologists.
drives the asymmetric division of notochord/neural precursors
in Ciona embryos. Development 2007, 134: 1491–1497. Fig. 7.29 adapted from Lohmann, J.U., Weigel, D.: Building beauty:
the genetic control of floral patterning. Dev. Cell 2002, 2: 135–142.
Chapter 7 Fig. 7.30 photograph reproduced with permission from Coen, E.S.,
Fig. 7.4 illustration after Scheres, B., Wolkenfelt, H., Willemsen, V., Meyerowitz, E.M.: The war of the whorls: genetic interactions
Terlouw, M., Lawson, E., Dean, C., Weisbeek, P.: Embryonic origin controlling flower development. Nature 1991, 353: 31–37. © 1991
of the Arabidopsis primary root and root meristem initials. Macmillan Magazines Ltd.
Development 1994, 120: 2475–2487.
Fig. 7.31 illustration after Drews, G.N., Goldberg, R.B.: Genetic
Fig. 7.5 illustration after Friml, J., et al.: Efflux-dependent auxin control of flower development. Trends Genet. 1989 5: 256–261.
gradients establish the apical-basal axis of Arabidopsis. Nature
Fig. 7.35 illustration from Bl·zquez, M.A.: The right time and place
2003, 426: 147–153.
for making flowers. Science 2005, 309: 1024–1025.
Fig. 7.6 adapted from Chapman, E.J. and Estelle, M.: Cytokinin and
auxin intersection in root meristems. Genome Biol. 2009, 10: 210. Chapter 8
doi:10.1186/gb-2009-10-2-210 Fig. 8.3 from Foty, R.A., and Steinberg, M.S.: The differential
Fig. 7.7 from Friml, J., et al: Efflux-dependent auxin gradients adhesion hypothesis: a direct evaluation. Developmental Biology
establish the apical-basal axis of Arabidopsis. Nature 2003, 426: 278 (2005) 255– 263.
147–153. Fig. 8.6 illustration after Strome, S.: Determination of cleavage
Fig. 7.10 illustration after Alberts, B., Bray, D., Lewis, J., Raff, M., planes. Cell 1993, 72: 3–6.
Roberts, K., Watson, J.D.: Molecular Biology of the Cell, 2nd edition. Fig. 8.8 from Raff, E.C., Villinski, J.T., Turner, F.R., Danoghue, P.C.J.,
New York: Garland Publishing, 1989. and Raff, R.A.: Experimental taphonomy shows the feasibility of
Fig. 7.14 illustration after Steeves, T.A., Sussex, I.M.: Patterning in fossil embryos. PNAS 2006, 103: 5846–5851.
Plant Development. Cambridge: Cambridge University Press, 1989. Fig. 8.9 photograph reproduced with permission from Bloom T.L.:
Fig. 7.16 illustration after McDaniel, C.N., Poethig, R.S.: Cell lineage The effects of phorbol ester on mouse blastomeres: a role
patterns in the shoot apical meristem of the germinating maize for protein kinase C in compaction? Development 1989, 106:
embryo. Planta 1988, 175: 13–22. 159–171.
Fig. 7.18 top panel, illustration after Poethig, R.S., Sussex, I.M.: The Fig. 8.11 from Eckert, J.J., and Fleming, T.P..: Tight junction
cellular parameters of leaf development in tobacco: a clonal biogenesis during early development. Biochim. Biophys. Acta.
analysis. Planta 1985, 165: 170–184. Bottom panel, illustration 2008, 1778: 717–728.
 Figure acknowledgements xxiii

Fig. 8.12 from Alberts, B. et al.: Molecular Biology of the Cell, Fifth Fig. 9.6 illustration after Wylie, C.C., Heasman, J.: Migration,
Edition, Garland Science, New York, 2008. proliferation, and potency of primordial germ cells. Semin. Dev.
Fig. 8.13 illustration after Coucouvanis, E., Martin, G.R.: Signals for Biol. 1993, 4: 161–170.
death and survival: a two-step mechanism for cavitation in the Fig. 9.7 adapted from Weidinger, G., Köprunner, M., Thisse, C.,
vertebrate embryo. Cell 1995, 83: 279–287. Thisse, B., and Razl, E.: Regulation of zebrafish primordial germ
Fig. 8.17 photograph reproduced with permission from Merrill, J.B., cell migration by attraction towards an intermediate target.
Santos, L.L.: A scanning electron micrographical overview of Development , 2002, 129: 25–36.
cellular and extracellular patterns during blastulation and Fig. 9.14 illustration after Alberts, B., Bray, D., Lewis, J., Raft, M.,
gastrulation in the sea urchin, Lytechinus variegatus. In The Roberts, K., Watson, J.D.: Molecular Biology of the Cell, 2nd edition.
Cellular and Molecular Biology of Invertebrate Development. Edited New York: Garland Publishing, 1989.
by Sawyer, R.H. and Showman, R.M. University of South Carolina Fig. 9.19 illustration after Goodfellow, P.N., Lovell-Badge, R.: SRY
Press, 1985; pp. 3–33. and sex determination in mammals. Ann. Rev. Genet. 1993, 27:
Fig. 8.19 illustration after Odell, G.M., Oster, G., Alberch, P., 71–92.
Burnside, B.: The mechanical basis of morphogenesis. I. Fig. 9.21 illustration after Higgins, S.J., Young, P., Cunha, G.R.:
Epithelial folding and invagination. Dev. Biol. 1981, 85: 446–462. Induction of functional cytodifferentiation in the epithelium of
Fig. 8.21 photographs reproduced with permission from Leptin, tissue recombinants II. Instructive induction of Wolffian duct
M., Casal, J., Grunewald, B., Reuter, R.: Mechanisms of early epithelia by neonatal seminal vesicle mesenchyme. Development
Drosophila mesoderm formation. Development Suppl. 1992, 1989, 106: 235–250.
23–31. Published by permission of The Company of Biologists Ltd. Fig. 9.25 illustration after Cline, T.W.: The Drosophila sex
Fig. 8.22 illustration after Bertet, C., Sulak, L., Lecuit, T. determination signal: how do flies count to two? Trends Genet.
Myosindependent junction remodelling controls planar cell 1993 9: 385–390.
intercalation and axis elongation. Nature 2004, 429: 667–671. Fig. 9.29 illustration after Clifford, R., Francis, R., Schedl, T.: Somatic
Fig. 8.24 illustration after Balinsky, B.I.: An Introduction to control of germ cell development. Semin. Dev. Biol 1994, 5: 21–30.
Embryology, 4th edition. Philadelphia, W.B. Saunders, 1975. Fig. 9.31 illustration after Alberts, B., Bray, D., Lewis, J., Raft, M.,
Fig. 8.28 photograph reproduced with permission from Smith, J.C., Roberts, K., Watson, J.D.: Molecular Biology of the Cell, 2nd edition.
Cunliffe, V., O’Reilly, M-A.J., Schulte-Merker, S., Umbhauer, M.: New York: Garland Publishing, 1989.
Xenopus Brachyury. Semin. Dev. Biol. 1995, 6: 405–410. © 1995 by
permission of the publisher, Academic Press Ltd., London.
Chapter 10
Fig. 8.29 illustration after Montero, J.A., Heisenberg, C.P. Fig. 10.3 illustration after Tijian, R.: Molecular machines that
Gastrulation dynamics: cells move into focus. Trends Cell Biol. control genes. Sci. Am. 1995, 272: 54–61.
2004,14: 620–627.
Fig. 10.4 illustration after Alberts B., Bray, D., Lewis, J., Raff, M.,
Fig. 8.30 illustration after Wallingford, J.B., Fraser, S., Harland, R.M. Roberts, K., Watson, J.D.: Molecular Biology of the Cell, 2nd edition.
Convergent extension: the molecular control of polarized cell New York: Garland Publishing, 1989.
movement during embryonic development. Dev. Cell 2002, 2:
Fig. 10.6 illustration after Alberts, B., Bray, D., Lewis, J., Raft, M.,
695–706.
Roberts, K., Watson, J.D.: Molecular Biology of the Cell, 2nd edition.
Fig. 8.34 illustration after Schoenwolf, G.C., Smith, J.L.: Mechanisms New York: Garland Publishing, 1989.
of neurulation: traditional viewpoint and recent advances.
Fig. 10.8 illustration after Kluger, Y., Lian, Z., Zhang, X., Newburger,
Development 1990 109: 243–270.
P.E., Weissman, S.M.: A panorama of lineage-specific
Fig. 8.39 photographs reproduced with permission from Priess, J.R., transcription in hematopoiesis. BioEssays 2004, 26: 1276–1287.
Hirsh, D.I.: Caenorhabditis elegans morphogenesis: the role of the
Fig. 10.10 illustration after Metcalf, D.: Control of granulocytes and
cytoskeleton in elongation of the embryo. Dev. Biol 1986, 117:
macrophages: molecular, cellular, and clinical aspects. Science
156–173. © 1986 Academic Press.
1991, 254: 529–533.
Fig. 8.41 photographs reproduced with permission from Tsuge,
Fig. 10.13 illustration after Crossley, M., Orkin, S.H.: Regulation of
T., Tsukaya, H., Uchimaya, H.: Two independent and polarized
the b-globin locus. Curr. Opin. Genet. Dev. 1993, 3: 232–237.
processes of cell elongation regulate leaf blade expansion in
Arabidopsis thaliana (L.) Heynh. Development 1996, 122: Fig. 10.17 from Barker, N., van Esl, J.H., Kuipers, J., Kujala, P., van
1589–1600. Published by permission of The Company of den Born, M., Cozijnsen, M., Haegebarth, A., Korving, J., Begthell,
Biologists Ltd. H., Peters, P.J., and Clevers, H: Identification of stem cells in small
intestine and colon by marker gene Lgr5. Nature
Fig. 10.22 illustration after Taupin, P.: Adult neurogenesis in the
Chapter 9 mammalian central nervous system: functionality and potential
Fig. 9.5 adapted from Hogan, B.: Decisions, decisions. Nature 2002, clinical interest. Med. Sci. Monit. 2005, 11: 247–252.
418: 282, and Saitou, M., Barton, S. Surani, M.: A molecular Fig. 10.27 illustration after Riedl, S.J., Shi, Y.: Molecular
programme for the specification of germ cell fate in mice. mechanisms of caspase regulation during apoptosis. Nat. Rev.
Nature, 2002, 418: 293–300. Mol. Cell Biol. 2004, 5: 897–907.
xxiv Figure acknowledgments

Chapter 11 Fig. 12.21 photograph reproduced with permission from Serafini, T.,
Fig. 11.6 photograph reproduced with permission from Cohn, M.J., Colamarino, S.A., Leonardo, E.D., Wang, H., Beddington, R.,
Izpisúa-Belmonte, J.C., Abud, H., Heath, J.K., Tickle, C.: Fibroblast Skarnes, W.C., Tessier-Lavigne, M.: Netrin-1 is required for
growth factors induce additional limb development from the commissural axon guidance in the developing vertebrate
flank of chick embryos. Cell 1995, 80: 739–746. © 1995 Cell Press. nervous system. Cell 1996, 87: 1001–1014. © 1996 Cell Press.

Fig. 11.18 adapted from Wellik, D.M., Capecchi, M.R.: Hox10 and Fig. 12.28 illustration after Kandell, E.R., Schwartz, J.H., Jessell,
Hox11 genes are required to globally pattern the mammalian T.M.: Principles of Neural Science, 3rd edition. New York: Elsevier
skeleton. Science 2003, 301: 363–367. Science Publishing Co., Inc., 1991.

Fig. 11.21 photograph reproduced with permission from Garcia- Fig. 12.30 illustration after Li, Z., Sheng, M.: Some assembly
Martinez, V., Macias, D., Gañan, Y., Garcia-Lobo, J.M., Francia, required: the development of neuronal synapses. Nat. Rev. 2003,
M.V., Fernandez- Teran, M.A., Hurle, J.M.: Internucleosomal 4: 833–841.
DNA fragmentation and programmed cell death (apoptosis) in Fig. 12.32 illustration after Goodman, C.S., Shatz, C.J.:
the interdigital tissue of embryonic chick leg bud. J. Cell Sci. Developmental mechanisms that generate precise patterns of
1993, 106: 201–208. Published by permission of The Company of neuronal connectivity. Cell Suppl. 1993, 72: 77–98.
Biologists Ltd. Fig. 12.33 illustration after Kandell, E.R., Schwartz, J.H., Jessell,
Fig. 11.23 illustration after French, V., Daniels, G.: Pattern formation: T.M.: Essentials of Neural Science and Behavior. Norwalk,
the beginning and the end of insect limbs. Curr. Biol. 1994, 4: Connecticut: Appleton & Lange, 1991.
35–37. Fig. 12.34 illustration after Goodman, C.S., Shatz, C.J.:
Fig. 11.26 (a) photograph reproduced with permission from Nellen, Developmental mechanisms that generate precise patterns of
D., Burke, R., Struhl, G., Basler, K.: Direct and long-range action neuronal connectivity. Cell Suppl. 1993, 72: 77–98.
of a dpp morphogen gradient. Cell 1996, 85: 357–368. © 1996,
Chapter 13
Cell Press.
Fig. 13.2 adapted from D.O. Morgan, The Cell Cycle. Oxford
Fig. 11.26 (b–e) photograph reproduced with permission from Moser,
University Press, 2006.
M. and Campbell, G.: Generating and interpreting the Brinker
gradient in the Drosophila wing. Dev. Biol. 2005, 286: 647–658. Fig. 13.3 illustration after Edgar, B.A., Lehman, D.A., O’Farrell, P.H.:
Transcriptional regulation of string (cdc25): a link between
Fig. 11.29 photograph reproduced with permission from Zecca, M.,
developmental programming and the cell cycle. Development
Basler, K., Struhl, G.: Direct and long-range action of a wingless
1994, 120: 3131–3143.
morphogen gradient. Cell 1996, 87: 833–844. © 1996 Cell Press.
Fig. 13.5 photograph reproduced with permission from Harrison,
Fig. 11.31 illustration after Bryant, P.J. The polar coordinate model
R.G.: Organization and Development of the Embryo. New Haven:
goes molecular. Science 1993, 259: 471–472.
Yale University Press, 1969. © 1969 Yale University Press.
Fig. 11.40 adapted from Adler, R., and Valeria Canto-Soler, M.:
Fig. 13.6 illustration after Gray, H.: Gray’s Anatomy. Edinburgh:
Molecular mechanisms of optic vesicle development:
Churchill-Livingstone, 1995.
Complexities, ambiuities, and controversies. Developmental
Biology, 305: 1–13. Fig. 13.8 from Nijhout, H.F.: Size matters (but so does time), and
it’s OK to be different. Dev. Cell , 2008, 15: 491–492.
Fig. 11.41 photographs reproduced with permission from Gehring,
W.J.: New perspectives on eye development and the evolution of Fig. 13.11 illustration after Walls, G.A.: Here today, bone tomorrow.
eyes and photoreceptors. J. Hered. 2005, 96(3): 171–184. Curr. Biol. 1993, 3: 687–689.

Fig. 11.49 adapted from Kelly, R.G., Buckingham, M.E.: The anterior Fig. 13.12 illustration after Kronenberg, H.M.: Developmental
heart-forming field: voyage to the arterial pole of the heart. regulation of the growth plate. Nature (Insight) 2003, 423:
Trends Genet. 2002, 18: 210–216. 332–336.
Fig. 13.18 illustration after Tata, J.R.: Gene expression during
Chapter 12 metamorphosis: an ideal model for post-embryonic
development. BioEssays 1993, 15: 239–248.
Fig. 12.7 illustration after Rakic, P.: Mode of cell migration to the
superficial layers of fetal monkey neocortex. J. Comp. Neurol. Fig. 13.19 illustration after Tata, J.R.: Gene expression during
1972, 145: 61–83. metamorphosis: an ideal model for post-embryonic
development. BioEssays 1993, 15: 239–248.
Fig. 12.8 adapted from Honda, T., Tabata, H., Nakajima, K.: Cellular
and molecular mechanisms of neuronal migration in neocortical Chapter 14
development. Semin. Cell Dev. Biol. 2003, 14: 169–174. Fig. 14.5 from Kragl, M., Knapp, D., Nacul, E., Khattak, S., Maden,
Fig. 12.15 adapted from Dasen, J., Tice, B., Brenner-Morton, S., M., Epperlein, H., and Tanaka, E.: Cells keep a memory of their
Jessell, T.: A Hox regulatory network establishes motor neuron tissue origin during axolotl limb regeneration. Nature, 2009, 460:
pool identity and target-muscle connectivity. Cell 2005, 123: 60–65.
477–491. Fig. 14.11 photographs reproduced with permission from Pecorino,
Fig. 12.16 illustration after Alberts, B., Bray, D., Lewis, J., Raff, M., L.T. Entwistle A., Brockes, J.P.: Activation of a single retinoic acid
Roberts, K., Watson, J.D.: Molecular Biology of the Cell, 2nd edition. receptor isoform mediates proximodistal respecification. Curr.
New York: Garland Publishing, 1989. Biol.1996, 6: 563–569.
 Figure acknowledgements xxv

Fig. 14.13 illustration after French, V., Bryant, P.J., Bryant, S.V.: Fig. 15.11 adapted from Boisvert, C.A., Mark-Kurik, E., and Ahlberg,
Pattern regulation in epimorphic fields. Science 1976, 193: P.E.: The pectoral fin of Panderichthys and the origin of digits.
969–981. Nature 2008, 456: 636–638.
Fig. 14.14 illustration after French, V., Bryant, P.J., Bryant, S.V.: Fig. 15.12 illustration after Gerhart, J., Lowe, C., Kirschner, M.:
Pattern regulation in epimorphic fields. Science 1976, 193: Hemichordates and the origins of chordates. Curr. Opin. Genet.
969–981. Dev. 2005, 15: 461–467.
Fig. 14.15 photograph reproduced with permission from Müller, Fig. 15.16 illustration after Gregory, W.K.: Evolution Emerging. New
W.A.: Diacylglycerol-induced multihead formation in Hydra. York: Macmillan, 1957.
Development 1989, 105: 309–316. Published by permission of The Fig. 15.20 illustration after Gompel, N. et al.: Chance caught on
Company of Biologists Ltd. the wing: cis-regulatory evolution and the origin of pigment
Chapter 15 patterns in Drosophila. Nature, 2005, 433: 481–487.

Fig. 15.9 illustration after Akam, M.: Hox genes and the evolution Fig. 15.22 illustration after Larsen, W.J.: Human Embryology. New
of diverse body plans. Phil, Trans. R. Soc. Lond. B 1995, 349: York: Churchill Livingstone, 1993.
313–319. Fig. 15.23 illustration after Romer, A.S.: The Vertebrate Body.
Fig. 15.10 illustration after Ferguson, E.L.: Conservation of dorsa- Philadelphia: W.B. Saunders, 1949.
ventral patterning in arthropods and chordates. Curr. Opin.
Genet. Dev. 1996, 6: 424–431.
This page intentionally left blank
 1
History and basic concepts
nn The origins of nn A conceptual tool kit
developmental biology

The aim of this chapter is to provide a conceptual framework for the study of devel-
a
opment. We start with a brief history of the study of embryonic development, which
illustrates how some of the key questions in developmental biology were first for-
mulated, and continue with some of the essential principles of development. The
big question is how does a single cell—the fertilized egg—give rise to a multicellular
organism, in which a multiplicity of different cell types are organized into tissues and
organs to make up a three-dimensional body. This question can be studied from many
different viewpoints, all of which have to be fitted together to obtain a complete
picture of development: which genes are expressed, and when and where; how cells
communicate with each other; how a cell’s developmental fate is determined; how
cells proliferate and differentiate into specialized cell types; and how major changes in
body shape are produced. We shall see that an organism’s development is ultimately
driven by the regulated expression of its genes, determining which proteins are pres-
b
ent in which cells and when. In turn, proteins largely determine how a cell behaves.
The genes provide a generative program for development, not a blueprint, as their
actions are translated into developmental outcomes through cellular behavior such as
intercellular signaling, cell proliferation, cell differentiation, and cell movement.

The development of a multicellular organism from a single cell—the fertilized egg—is

a brilliant triumph of evolution. The fertilized egg divides to give rise to many mil-
lions of cells, which form structures as complex and varied as eyes, arms, heart, and
brain. This amazing achievement raises a multitude of questions. How do the cells
arising from division of the fertilized egg become different from each other? How do
they become organized into structures such as limbs and brains? What controls the
behavior of individual cells so that such highly organized patterns emerge? How are Fig. 1.1 Human fertilized egg and
the organizing principles of development embedded within the egg, and in particular embryo. (a) Human fertilized egg. The
sperm and egg nuclei have not yet fused.
within the genetic material, DNA? Much of the excitement in developmental biology
(b) Human embryo at around 51 days'
today comes from our growing understanding of how genes direct these developmen-
gestation (Carnegie stage 20), which is
tal processes, and genetic control is one of the main themes of this book. Thousands equivalent to a mouse embryo at 13.5 days
of genes are involved in controlling development, but we will focus only on those that post-fertilization. A human embryo at this
have key roles and illustrate general principles. stage is around 21–23 mm long.
Understanding how embryos develop is a huge intellectual challenge, and one of (a) Courtesy of Alpesh Doshi, CRGH, London.
the ultimate aims of the science of developmental biology is to understand how we (b) Reproduced courtesy of the MRC/
humans develop (Fig. 1.1). We need to understand human development for several Wellcome-funded Human Developmental
reasons. We need to properly understand why it sometimes goes wrong and why a Biology Resource.
2 Chapter 1 History and basic concepts

Box 1A Basic stages of Xenopus laevis development

Blastula

Egg
Sperm Animal pole

Gastrula (section)
blastocoel
Cleavage
Vegetal pole

Fertilization blastopore

Adult Gastrulation
mesoderm

Metamorphosis
Free-swimming
tadpole endoderm
Dorsal
Neurulation
ectoderm future gut
(section)
Ventral Organogenesis
Neurula
Anterior
Anterior neural
folds
brain
spinal cord
notochord
Ventral Dorsal
somites
Posterior
(dorsal view)
yolk mass

Posterior
Tailbud-stage embryo Tailbud-stage
(dorsal view with embryo
surface removed) (lateral view)

fetus may fail to be born or a baby be born with congenital abnormalities. The link
here with genetic control of development is very close, as mutations in genes can
lead to abnormal development; environmental factors, such as drugs and infections,
can affect it too. Another area of medical research related to developmental biology
is regenerative medicine—finding out how to use cells to repair damaged tissues and
organs. The focus of regenerative medicine is currently on stem cells. These have
many of the properties of embryonic cells, such as the ability to proliferate and to
develop into a range of different tissues, and are discussed in Chapter 10. Cancer cells
also display some properties of embryonic cells, such as the ability to divide indefi-
nitely, and so the study of embryonic cells and their development could lead to new
and better treatments for cancer, as many of the same genes are involved.
The development of an embryo from the fertilized egg is known as embryogenesis
(Box 1A). One of the first tasks is to lay down the overall body plan of the organism,
and we shall see that different organisms solve this fundamental problem in several
 The origins of developmental biology 3

Although vertebrate development is very varied, there are a num- internal organs, are still on the surface of the embryo. During the
ber of basic stages that can be illustrated by following the devel- next stage—­gastrulation—there is a dramatic rearrangement of
opment of the frog Xenopus laevis. The unfertilized egg is a large cells; the endoderm and mesoderm move inside, and the basic
cell. It has a pigmented upper surface (the animal pole) and a body plan of the tadpole is established. Internally, the mesoderm
lower region (the vegetal pole) characterized by an accumulation gives rise to a rod-like structure (the ­notochord), which runs from
of yolk granules. the head to the tail, and lies centrally beneath the future nervous
After fertilization of the egg by a sperm, and the fusion of system. On either side of the notochord are segmented blocks of
male and female nuclei, cleavage begins. Cleavages are mitotic mesoderm called somites, which will give rise to the muscles and
divisions in which cells do not grow between each division, and vertebral column, as well as the dermis of the skin (notochord and
so with successive cleavages the cells become smaller. After somites can be seen in the cutaway view of the later tailbud-stage
about 12 division cycles, the embryo, now known as a blastula, embryo).
consists of many small cells surrounding a fluid-filled cavity (the Shortly after gastrulation, the ectoderm above the notochord
blastocoel) above the larger yolky cells. Already, changes have folds to form a tube (the neural tube), which gives rise to the brain
occurred within the cells and they have interacted with each other and spinal cord—a process known as neurulation. By this time, other
so that some future tissue types have become partly specified. organs, such as limbs, eyes, and gills, are specified at their future
These are the three germ layers: mesoderm, endoderm, and locations, but only develop a little later, during organogenesis.
ectoderm. The animal region will give rise to ectoderm, which ­During organogenesis, specialized cells such as muscle, cartilage,
forms both the epidermis of the skin and the nervous system. and neurons differentiate. Within 48 hours the embryo has become
The future ­endoderm and mesoderm, which are destined to form a feeding tadpole with typical vertebrate features.

ways. The focus of this book is mainly on animal development, in particular that of
vertebrates—frogs, birds, fish, and mammals—whose early development is discussed
in Chapters 3 to 5. We also look at selected invertebrates, such as the sea urchin,
ascidians, and particularly the fruit fly and the nematode worm. Our understanding of
the genetic control of development is most advanced in these last two animals and the
main features of their early development are considered in Chapters 2 and 6, respec-
tively. They are also used throughout the book to illustrate particular aspects of devel-
opment. In Chapter 7 we look briefly at some aspects of plant development, which
differs in many respects from that of animals but involves similar basic principles.
Morphogenesis, or the development of form, is discussed in Chapter 8. In Chapter 9
we look at how sex is determined and how germ cells develop. The differentiation
of unspecialized cells into cells that carry out particular functions, such as muscle
cells and blood cells, is considered in Chapter 10. Structures such as the vertebrate
limb, and organs such as the heart, insect and vertebrate eyes, and the nervous sys-
tem, illustrate the problems of multicellular organization and tissue differentiation
in embryogenesis, and we consider some of these systems in detail in Chapters 11
and 12. The study of developmental biology, however, goes well beyond the devel-
opment of the embryo. Post-embryonic growth and aging, and how some animals
undergo metamorphosis, are covered in Chapter 13, and how animals can regener-
ate lost organs is discussed in Chapter 14. Taking a wider view, we shall consider in
Chapter 15 how developmental mechanisms have evolved and how they constrain the
very process of evolution itself.
One might ask whether it is necessary to cover so many different organisms in order
to understand the basic features of development. The answer is yes. Developmental
biologists do indeed believe that there are general principles of development that
apply to all animals, but life is too wonderfully diverse to find all the answers in a sin-
gle organism. As it is, developmental biologists have tended to focus their efforts on
a relatively small number of animals, chosen because they were convenient to study
and amenable to experimental manipulation or genetic analysis. This is why some
creatures, such as the frog Xenopus laevis (Fig. 1.2), the nematode worm Caenorhab-
ditis elegans, and the fruit fly Drosophila melanogaster, have such a dominant place Fig. 1.2 The South African claw-toed frog,
in developmental biology. Similarly, work with the thale-cress, Arabidopsis thaliana, Xenopus laevis. Scale bar = 1 cm.
has uncovered many features of plant development. Photograph courtesy of J. Smith.
4 Chapter 1 History and basic concepts

One of the most exciting and satisfying aspects of developmental biology is that
understanding a developmental process in one organism can help to illuminate sim-
ilar processes elsewhere—for example, giving insights into how humans develop.
Nothing illustrates this more dramatically than the influence that our understanding
of Drosophila development, and especially of its genetic basis, has had throughout
developmental biology. The identification of genes controlling early embryogenesis
in Drosophila has led to the discovery of related genes being used in similar ways in
the development of mammals and other vertebrates. Such discoveries encourage us
to believe in the existence of general developmental principles.
Frogs have long been a favorite organism for studying early development because their
eggs are large and their embryos are robust—easy to grow in a simple culture medium
and relatively easy to experiment on. Embryogenesis in the South African frog Xenopus
(Box 1A, pp. 2–3) illustrates some of the basic stages of development in all animals.
In the rest of this chapter we first look briefly at the history of embryology—as the
study of developmental biology has been called for most of its existence. The term
developmental biology itself is of much more recent origin and reflects the apprecia-
tion that development is not restricted to the embryo alone. Traditionally, embryology
described experimental results in terms of morphology and cell fate, but we now
understand development in terms of molecular genetics and cell biology as well. In
the second part of the chapter we will introduce some key concepts that are used
repeatedly in studying and understanding development.

The origins of developmental biology

Many questions in embryology were first posed hundreds, and in some cases thou-
sands, of years ago. Appreciating the history of these ideas helps us to understand
why we approach developmental problems in the way that we do today.

1.1 Aristotle first defined the problem of epigenesis and preformation

A scientific approach to explaining development started with Hippocrates in Greece

in the fifth-century bc. Using the ideas current at the time, he tried to explain develop-
ment in terms of the principles of heat, wetness, and solidification. About a century
later the study of embryology advanced when the Greek philosopher Aristotle formu-
lated a question that was to dominate much thinking about development until the end
of the nineteenth century. Aristotle addressed the problem of how the different parts
of the embryo were formed. He considered two possibilities: one was that everything
in the embryo was preformed from the very beginning and simply got bigger dur-
ing development; the other was that new structures arose progressively, a process he
termed epigenesis (which means ‘upon formation’) and that he likened metaphorically
to the ‘knitting of a net’. Aristotle favored epigenesis and his conjecture was correct.
Aristotle’s influence on European thought was enormous and his ideas remained
dominant well into the seventeenth century. The contrary view to epigenesis, namely
that the embryo was preformed from the beginning, was championed anew in the late
seventeenth century. Many could not believe that physical or chemical forces could
mold a living entity like the embryo. Along with the contemporaneous background of
belief in the divine creation of the world and all living things, was the belief that all
Fig. 1.3 Malpighi’s description of the
chick embryo. The figure shows Malpighi’s
embryos had existed from the beginning of the world, and that the first embryo of a
drawings, made in 1673, depicting the early species must contain all future embryos.
embryo (top), and at 2 days’ incubation Even the brilliant seventeenth-century Italian embryologist, Marcello Malpighi,
(bottom). His drawings accurately illustrate could not free himself from preformationist ideas. While he provided a remarkably
the shape and blood supply of the embryo. accurate description of the development of the chick embryo, he remained convinced,
Reprinted by permission of the President and against the evidence of his own observations, that the fully formed embryo was pres-
Council of the Royal Society. ent from the beginning (Fig. 1.3). He argued that at very early stages the parts were
 The origins of developmental biology 5

so small that they could not be seen, even with his best microscope. Other preforma-
tionists believed that the sperm contained the embryo, and some even claimed to see
a tiny human—an homunculus—in the head of each human sperm (Fig. 1.4).
The preformation/epigenesis issue was vigorously debated throughout the eigh-
teenth century. But the problem could not be resolved until one of the great advances
in biology had taken place—the recognition that living things, including embryos,
were composed of cells.

1.2 Cell theory changed the conception of embryonic development and heredity

The invention of the microscope around 1600 was essential for the discovery of cells,
but the ‘cell theory’ of life was only developed between 1820 and 1880 by, among oth-
ers, the German botanist, Matthias Schleiden, and the physiologist, Theodor Schwann.
It recognized that all living organisms consist of cells, that these are the basic units
of life, and that new cells can only be formed by the division of pre-existing cells.
The cell theory was one of the most illuminating advances in biology, and had an
enormous impact. Multicellular organisms, such as animals and plants, could now
be viewed as communities of cells. Development could not therefore be based on
preformation but must be epigenetic, because during development many new cells are
generated by division from the egg, and new types of cells are formed. A crucial step
forward in understanding development was the recognition, in the 1840s, that the egg
itself is but a single, albeit specialized, cell.
An important advance in embryology was the proposal by the nineteenth-century
German biologist, August Weismann, that the offspring does not inherit its charac-
Fig. 1.4 Some preformationists believed
teristics from the body (the soma) of the parent but only from the germ cells—egg
that an homunculus was curled up in the
and sperm. Weismann drew a fundamental distinction between germ cells and the
head of each sperm.
body cells or somatic cells (Fig. 1.5). Characteristics acquired by the body during an
An imaginative drawing, after Nicholas
animal’s life cannot be transmitted to the germline. As far as heredity is concerned,
Harspeler (1694).
the body is merely a carrier of germ cells. As the English novelist and essayist Samuel
Butler put it: ‘A hen is only an egg’s way of making another egg.’
Work on sea-urchin eggs showed that after fertilization the egg contains two nuclei,
which eventually fuse; one of these nuclei belongs to the egg, while the other comes
from the sperm. Fertilization therefore results in a single cell—the zygote—carrying
a nucleus with contributions from both parents, and it was concluded that the cell
nucleus must contain the physical basis of heredity. The climax of this line of research
was the demonstration, towards the end of the nineteenth century, that the chromo-
somes within the nucleus of the zygote are derived in equal numbers from the two
parental nuclei, and the recognition that this provided a physical basis for the trans-
mission of genetic characters according to the laws developed by the Austrian ­botanist

First generation Second generation Third generation Fig. 1.5 The distinction between germ
mutations in mutation in cells and somatic cells. In each generation
somatic cells do germ cell germ cells give rise to both somatic cells
not affect germline affects somatic ­and germ cells, but inheritance is through the
cells and
somatic cells

somatic cells

somatic cells

germline germ cells only (first panel). Changes that

occur due to a mutation (red) in a somatic
cell can be passed on to its daughter cells
but do not affect the germline, as shown in
the second panel. In contrast, a mutation in
the germline (green) will be present in every
cell in the body of the new organism to
which that cell contributes, and will also be
zygote zygote zygote passed on to future generations through the
germ cells germ cells germ cells
germline, as shown in the third panel.
6 Chapter 1 History and basic concepts

First cleavage Second cleavage

Fig. 1.6 Weismann’s theory of nuclear

determination. Weismann assumed that
there were factors in the nucleus that were
distributed asymmetrically to daughter cells
during cleavage and directed their future
Weismann’s nuclear determinants
development.

and monk, Gregor Mendel. The number of chromosomes is kept constant from gen-
eration to generation by a specialized type of cell division that produces the germ
cells, called meiosis, which halves the chromosome number; the full complement of
chromosomes is then restored at fertilization. The zygote and the somatic cells that
arise from it divide by the process of mitosis, which maintains chromosome number
(Box 1B, p. 7). Germ cells contain a single copy of each chromosome and are called
haploid, whereas germ-cell precursor cells and the other somatic cells of the body
contain two copies and are called diploid.

1.3 Two main types of development were originally proposed

The next big question was how cells became different from one another during
embryonic development. With the increasing emphasis on the role of the nucleus, in
the 1880s Weismann put forward a model of development in which the nucleus of
the zygote contained a number of special factors or determinants (Fig. 1.6). He pro-
posed that while the fertilized egg underwent the rapid cycles of cell division known
Fig. 1.7 Roux’s experiment to as cleavage, these determinants would be distributed unequally to the daughter cells
investigate Weismann’s theory of mosaic and so would control the cells’ future development. The fate of each cell was therefore
development. After the first cleavage of a predetermined in the egg by the factors it would receive during cleavage. This type of
frog embryo, one of the two cells is killed model was termed ‘mosaic,’ as the egg could be considered to be a mosaic of discrete
by pricking it with a hot needle; the other localized determinants. Central to Weismann’s theory was the assumption that early
remains undamaged. At the blastula stage cell divisions must make the daughter cells quite different from each other as a result
the undamaged cell can be seen to have of unequal distribution of nuclear components.
divided as normal into many cells that fill In the late 1880s, initial support for Weismann’s ideas came from experiments
half of the embryo. The development of the
carried out independently by the German embryologist, Wilhelm Roux, who experi-
blastocoel, a small fluid-filled space in the
mented with frog embryos. Having allowed the first cleavage of a fertilized frog egg,
center of the blastula, is also restricted to the
undamaged half. In the damaged half of the
Roux destroyed one of the two cells with a hot needle and found that the remaining
embryo, no cells appear to have formed. At cell developed into a well-formed half-larva (Fig. 1.7). He concluded that the ‘devel-
the neurula stage, the undamaged cell has opment of the frog is based on a mosaic mechanism, the cells having their character
developed into something resembling half a and fate determined at each cleavage.’
normal embryo.

Fertilized frog egg Two-cell stage Blastula stage (section) Neurula stage

hot needle blastocoel remains half

of embryo
killed
cell

neural
tube
 The origins of developmental biology 7

Box 1B The mitotic cell cycle

When a eukaryotic cell duplicates itself it goes through a fixed

sequence of events called the cell cycle. The cell grows in size,
the DNA is replicated, and the replicated chromosomes then
undergo mitosis and become segregated into two daughter Resting
nuclei. Only then can the cell divide to form two daughter cells, M
G0
which can go through the whole sequence again.
G2 Division
The standard eukaryotic mitotic cell cycle is divided into well-
marked phases. At the M phase, mitosis and cell cleavage give rise
to two new cells. The rest of the cell cycle, between one M phase Growth
factors
and the next, is called interphase. Replication of DNA occurs G1
during a defined period in interphase, the S phase. Preceding S I nte r p h a s e
S phase is a period known as G1 (the G stands for gap), and after it
another interval known as G2, after which the cells enter mitosis
(see figure). G1, S phase, and G2 collectively make up interphase,
the part of the cell cycle during which cells synthesize proteins
and grow, as well as replicating their DNA. When somatic cells
are not proliferating they are usually in a state known as G0, into Particular phases of the cell cycle are absent in some cells:
which they withdraw after mitosis. The decision to enter G0 or to ­ uring cleavage of the fertilized Xenopus egg, G1 and G2 are vir-
d
proceed through G1, may be controlled by both intracellular status tually absent and cells get smaller at each division. In Drosophila
and extracellular signals such as growth factors. Growth factors salivary glands there is no M phase, as the DNA replicates repeat-
enable the cell to proceed out of G0 and progress through the cell edly without mitosis or cell division, leading to the formation of
cycle. Cells such as neurons and skeletal muscle cells, which do giant polytene chromosomes.
not divide after differentiation, are permanently in G0.

But when Roux’s fellow countryman, Hans Driesch, repeated the experiment on sea-
urchin eggs, he obtained quite a different result (Fig. 1.8). He wrote later: ‘But things
turned out as they were bound to do and not as I expected; there was, typically, a
whole gastrula on my dish the next morning, differing only by its small size from a nor-
mal one; and this small but whole gastrula developed into a whole and typical larva.’

Normal development of sea urchin larva from two-cell stage

Driesch’s separation of cells at two-cell stage resulted in the death of one cell.
The surviving cell developed into a small but otherwise normal larva Fig. 1.8 The outcome of Driesch’s
experiment on sea urchin embryos, which
first demonstrated the phenomenon of
regulation. After separation of cells at the
two-cell stage, the remaining cell developed
into a small, but whole, normal larva. This
contradicted Roux’s earlier finding that if
one of the
separated cells one of the cells of a two-cell frog embryo is
usually died damaged, the remaining cell develops into a
half-embryo only (see Fig. 1.7).
8 Chapter 1 History and basic concepts

Dorsal lip of blastopore grafted from an

Driesch had completely separated the cells at the two-cell stage and obtained a
unpigmented species of newt to the normal but small larva. That was just the opposite of Roux’s result, and was the first
blastocoel roof of a pigmented species clear demonstration of the developmental process known as regulation. This refers
Triton cristatus to the ability of an embryo to restore normal development even if some portions are
gastrula removed or rearranged very early in development, and its basis is explained later in
the chapter. We shall see many examples of regulation throughout the book (frog
dorsal lip embryos do, in fact, regulate as well; an explanation of Roux’s experiment, and why
of blastopore
he got the result he did, is given later, in Section 4.3).
blastocoel But although Weismann was wrong about nuclear determinants, there are many
examples of developmentally important proteins and RNAs located in the cell cyto-
plasm that are distributed unequally at cell division and which make two daughter
cells different from each other. These are known as cytoplasmic determinants.

Triton taeniatus 1.4 The discovery of induction showed that one group of cells could
gastrula determine the development of neighboring cells

The fact that embryos can regulate implies that cells must interact with each other,
A secondary embryo is induced but the central importance of cell–cell interactions in embryonic development was not
really established until the discovery of the phenomenon of induction. This is where
one cell, or tissue, directs the development of another, neighboring, cell or tissue.
The importance of induction and other cell–cell interactions in development was
proved dramatically in 1924 when Hans Spemann and his assistant, Hilde Mangold,
carried out a famous transplantation experiment in amphibian embryos. They showed
that a partial second embryo could be induced by grafting one small region of an early
newt embryo onto another at the same stage (Fig. 1.9). The grafted tissue was taken
secondary (induced) embryo
from the dorsal lip of the blastopore—the slit-like invagination that forms where gas-
X Y trulation begins on the dorsal surface of the amphibian embryo (see Box 1A, pp. 2–3).
Primary Secondary This small region they called the organizer, as it seemed to be ultimately responsible
structures (induced)
structures for controlling the organization of a complete embryonic body; it is now known as the
neural tube
notochord Spemann–Mangold organizer, or just the Spemann organizer. For their discovery,
notochord
neural tube
Spemann received the Nobel Prize for Physiology or Medicine in 1935, the first Nobel
Prize ever given for embryological research. Sadly, Hilde Mangold had died earlier, in
an accident, and so could not be honored.

Fig. 1.9 The dramatic demonstration by 1.5 The study of development was stimulated by the coming together of
Spemann and Mangold of induction of genetics and development
a new main body axis by the organizer
When Mendel’s laws were rediscovered in 1900 there was a great surge of interest in
region in the early amphibian gastrula.
mechanisms of inheritance, particularly in relation to evolution, but less so in relation to
A piece of tissue (yellow) from the dorsal lip
of the blastopore of a newt (Triton cristatus) development. Genetics was seen as the study of the transmission of hereditary elements
gastrula is grafted to the opposite side of a from generation to generation, whereas embryology was the study of how an individual
gastrula of another, pigmented, newt species organism develops and, in particular, how cells in the early embryo became different
(Triton taeniatus, pink). The grafted tissue from each other. Genetics seemed, in this respect, to be irrelevant to development.
induces a new body axis containing neural The fledgling science of genetics was put on a firm conceptual and experimen-
tube and somites. The unpigmented graft tal footing in the first-quarter of the twentieth century by the American, Thomas
tissue forms a notochord at its new site Hunt Morgan. Morgan chose the fruit fly Drosophila melanogaster as his experimental
(see section in lower panel) but the neural organism. He noticed a fly with white eyes rather than the usual red eyes and by care-
tube and the other structures of the new
ful cross-breeding he showed that inheritance of this mutant trait was linked to the
axis have been induced from the pigmented
sex of the fly. He found three other sex-linked traits and worked out that they were
host tissue. The organizer region discovered
each determined by three distinct ‘genetic loci,’ which occupied different positions on
by Spemann and Mangold is known as the
Spemann organizer. the same chromosome, the fly’s X chromosome. The rather abstract hereditary ‘fac-
tors’ of Mendel had been given reality. But even though Morgan had originally trained
as an embryologist, he made little headway in explaining development in terms of
genetics. That had to wait until the nature of the gene was better understood.
 The origins of developmental biology 9

An important concept in understanding how genes influence physical and physi-

ological traits is the distinction between genotype and phenotype. This was first put
forward by the Danish botanist, Wilhelm Johannsen, in 1909. The genetic endowment
of an organism—the genetic information it acquires from its parents—is the genotype.
Its visible appearance, internal structure, and biochemistry comprise the phenotype.
While the genotype certainly controls development, environmental factors interact-
ing with the genotype influence the phenotype. Despite having identical genotypes,
identical twins can develop considerable differences in their phenotypes as they grow
up (Fig. 1.10), and these tend to become more evident with age.
Following Morgan’s discoveries in genetics, the problem of development could now
be posed in terms of the relationship between genotype and phenotype: how the genetic
endowment becomes ‘translated’ or ‘expressed’ during development to give rise to a
Fig. 1.10 The difference between
functioning organism. But the coming together of genetics and embryology was slow
genotype and phenotype. These identical
and tortuous. The discovery in the 1940s that genes are made of DNA and encode
twins have the same genotype because
proteins was a major turning point. It was already clear that the properties of a cell are one fertilized egg split into two during
determined by the proteins it contains, and so the fundamental role of genes in devel- development. Their slight difference in
opment could at last be appreciated. By controlling which proteins were made in a cell, appearance is due to nongenetic factors,
genes could control the changes in cell properties and behavior that occurred during such as environmental influences.
development. A further major advance in the 1960s was the discovery that some genes Photograph courtesy of Josè and Jaime
encode proteins that control the activity of other genes. A more recent discovery is that Pascual.
small RNAs called microRNAs are also important in the control of gene expression.

1.6 Development is studied mainly through selected model organisms

Although the development of various species has been studied at one time or another,
a relatively small number of organisms provide most of our knowledge about devel-
opmental mechanisms. We can thus regard them as ‘models’ for understanding the
processes involved, and they are often called model organisms. Sea urchins and
amphibians were the main animals used for the first experimental investigations
because their developing embryos are both easy to obtain and, in the case of amphib-
ians, sufficiently large and robust for relatively easy experimental manipulation, even
at quite late stages. Among vertebrates, the frog Xenopus laevis, the mouse (Mus
musculus), the chicken (Gallus gallus), and the zebrafish (Danio rerio), are the main
model organisms now studied. Among invertebrates, the fruit fly Drosophila melano-
gaster and the nematode Caenorhabditis elegans have been the focus of most atten-
tion, because a great deal is known about their developmental genetics and they
can be easily genetically modified. Two Nobel prizes have been awarded for discov-
eries about development in Drosophila and Caenorhabditis, respectively. With the
advent of modern methods of genetic analysis, there has also been a resurgence of
interest in the sea urchin Strongylocentrotus purpuratus and ascidians such as Ciona
­intestinalis. For plant developmental biology, Arabidopsis thaliana serves as the main
model ­organism. The life cycles and background details for these model organisms
are given in the relevant chapters later in the book. The evolutionary relationships of
these organisms are shown in Fig. 1.11.
The reasons for these choices are partly historical—once a certain amount of
research has been done on one animal it is more efficient to continue to study it
rather than start at the beginning again with another species—and partly a ques-
tion of ease of study and biological interest. Each species has its advantages and
disadvantages as a developmental model. The chick embryo, for example, has long
been studied as an example of vertebrate development because fertile eggs are eas-
ily available, the embryo withstands experimental microsurgical manipulation very
well, and it can be cultured outside the egg. A disadvantage, however, was that until
very recently little was known about the chick’s developmental genetics. In contrast,
we know a great deal about the genetics of the mouse, although the mouse is more
10 Chapter 1 History and basic concepts

Plants (Arabidopsis)
Nematodes (Caenorhabditis)
Insects (Drosophila)
Arthropods
Crustaceans
Annelids
Molluscs

Single-celled Echinoderms (sea urchin)

ancestors Mammals (mouse, human)
Amphibians (Xenopus)
Birds (chick)
Chordates
Reptiles
Fish (zebrafish)
Fig. 1.11 Phylogenetic tree showing the Ascidians (Ciona)
placement of the main developmental
model organisms. The organisms discussed Cnidarians (Hydra)
in this book are highlighted in blue.

­ ifficult to study in some ways, as development normally takes place entirely within
d
the mother. It is, however, possible to culture very early mouse embryos for a little
time outside the uterus and then re-implant them. Many developmental mutations
have been identified in the mouse, and it is also amenable to genetic modification
by transgenic techniques, by which genes can be introduced, deleted, or modified in
living organisms. It is also the best experimental model we have for studying mamma-
lian development, including that of humans. The zebrafish is a more recent addition
to the select list of vertebrate model systems; it is easy to breed in large numbers, the
embryos are transparent and so cell divisions and tissue movements can be followed
visually, and it has great potential for genetic investigations.
A major goal of developmental biology is to understand how genes control embry-
onic development, and to do this one must first identify which genes, out of the
many thousands in the organism, are critically and specifically involved in control-
ling development. This task can be approached in various ways, depending on the
organism involved, but the general starting point is to identify mutations that alter
development in some specific and informative way, as described in the following sec-
tion. Techniques for identifying such genes and for detecting and manipulating their
expression in the organism are described throughout the book, along with techniques
for genetic manipulation.
Some of our model organisms are more amenable to conventional genetic analy-
sis than others. Despite its importance in developmental biology, little conventional
genetics has been done on X. laevis, which has the disadvantage of being tetraploid
­(it has four sets of chromosomes in its somatic cells, as opposed to the two sets car-
ried by the cells of diploid organisms, such as humans and mice) and a relatively long
breeding period, taking 1–2 years to reach sexual maturity. Using modern genetic and
bioinformatics techniques, however, many developmental genes have been identified in­
X. laevis by direct DNA sequence comparison with known genes in Drosophila and
mice. The closely related frog, Xenopus tropicalis, is a more attractive organism for
genetic analysis; it is diploid and can also be genetically manipulated to produce trans-
genic organisms.
The complete set of genes characteristic of a particular organism is known as its
­genome. In sexually reproducing diploid species, two complete copies of the genome
 The origins of developmental biology 11

are carried by the chromosomes in each somatic cell—one copy inherited from the father
and one from the mother. The complete DNA sequences of the genomes of many model
species are now known, which helps enormously in identifying developmental genes.
Genome sequences have been completed for sea urchin, Drosophila, Caenorhabditis,
chicken, mouse, and human, and there is a draft genome sequence for X. tropicalis. The
zebrafish genome is being sequenced; a complication here is that a genome-duplication
event in the zebrafish’s evolutionary history means that there are duplicates of at least
2900 out of the estimated 20,000 protein-coding genes in its genome.
In general, when an important developmental gene has been identified in one ani-
mal, it has proved very rewarding to see whether a corresponding gene is present and
is acting in a developmental capacity in other animals. Such genes are often identi-
fied by a sufficient degree of nucleotide sequence similarity to indicate descent from
a common ancestral gene. Genes that meet this criterion are known as homologous
genes. As we shall see in Chapter 5, this approach identified a hitherto unsuspected
class of vertebrate genes that control the regular segmented pattern from head to tail
that is represented by the different types of vertebrae at different positions. These
genes were identified by their homology with genes that determine the identities of
the different body segments in Drosophila (described in Chapter 2).

1.7 The first developmental genes were identified as spontaneous mutations

Most of the organisms dealt with in this book are sexually reproducing diploids: their
somatic cells contain two copies of each gene, with the exception of those on the sex
chromosomes. In a diploid species, one copy, or allele, of each gene is contributed by
the male parent and the other by the female. For many genes there are several differ-
ent ‘normal’ alleles present in the population, which leads to the variation in normal
phenotype one sees within any sexually reproducing species. Occasionally, however,
a mutation will occur spontaneously in a gene and there will be marked change, usu-
ally deleterious, in the phenotype of the organism.
Many of the genes that affect development have been identified by spontaneous
mutations that disrupt their function and produce an abnormal phenotype. Mutations
are classified broadly according to whether they are dominant or recessive (Fig. 1.12).
Dominant and semi-dominant mutations are those that produce a distinctive pheno-
type when the mutation is present in just one of the alleles of a pair; that is, it exerts an

Recessive mutation (e.g. vestigial–) Semi-dominant mutation (e.g. Brachyury)

Genotype Phenotype Genotype Phenotype

wild type normal wild type normal

vg+ +

vg+ +

Fig. 1.12 Types of mutations. Left: a

heterozygous mutation normal heterozygous mutation deformed tail mutation is recessive when it only has an
vg+ +
effect in the homozygous state; that is, when
both copies of the gene carry the mutation.
A plus sign (+) denotes the ‘normal’ allele
vg– T
(wild type), and a minus sign (−) the recessive
allele. Right: by contrast, a dominant or semi-
homozygous mutation vestigial wings homozygous mutation embryonic lethal
dominant mutation produces an effect on the
vg– T
phenotype in the heterozygous state; that
is, when just one copy of the mutant gene is
vg– present. T denotes the mutant form of the
T
gene Brachyury.
12 Chapter 1 History and basic concepts

effect in the heterozygous state. Dominant developmental mutations can be missed as

they are often lethal in the embryo. By contrast, recessive mutations, such as vestigial,
which leads to the non-development of wings in Drosophila, alter the phenotype only
when both alleles are of the mutant type; this is known as the homozygous state.
In general, dominant mutations are more easily recognized, particularly if they
affect gross anatomy or coloration, provided that they do not cause the early death of
the embryo in the heterozygous state. Truly dominant mutations are rare, however.
A mutation in the mouse gene Brachyury is a classic example of a semi-dominant
mutation and was originally identified because mice heterozygous for this mutation
(symbolized by T) have short tails. When the mutation is homozygous it has a much
greater effect, and embryos die at an early stage, indicating that the gene is required
for normal embryonic development (Fig. 1.13). Once breeding studies had confirmed
that the Brachyury trait was due to a single gene, the gene could be mapped to a loca-
tion on a particular chromosome by classical genetic-mapping techniques. Identifying
recessive mutations is more laborious, as the heterozygote has a phenotype identical
to a normal wild-type animal, and a carefully worked-out breeding program is needed
to obtain homozygotes. In mammals, identifying potentially lethal recessive develop-
mental mutations requires careful observation and analysis, as the homozygotes may
die unnoticed inside the mother.
Many mutations in invertebrates were identified as conditional mutations. The
effects of these mutations only show up when the animal is kept in a particular
condition—most commonly at a higher temperature, in which case the mutation is
known as a temperature-sensitive mutation. At the usual ambient temperature, the
animal appears normal. Temperature sensitivity is usually due to the encoded mutant
protein being able to fold into a functional structure at the normal temperature, but
being less stable at the higher temperature.
Very rigorous criteria must be applied to identify mutations that are affecting a genu-
ine developmental process and not just affecting some vital but routine ‘housekeeping’
function without which the animal cannot survive. One simple criterion for a devel-
opmental mutation is that it leads to the death of the embryo, but mutations in genes
involved in vital housekeeping functions can also have this effect. Mutations that pro-
duce observable abnormal embryonic development are more promising candidates for
true developmental mutations. In later chapters we shall see how large-scale screening
for mutations after mutagenesis by chemicals or X-rays has identified many more devel-
opmental genes than would ever have been picked up by rare spontaneous mutations.

T +
+ +

Short tail

Fig. 1.13 Genetics of the semi-dominant

mutation Brachyury (T) in the mouse.
T T + +
A male heterozygote carrying the T mutation + + + +
merely has a short tail. When mated with
a female homozygous for the normal, or
wild-type, Brachyury gene (++), some of the
offspring will also be heterozygotes and
have short tails. Mating two heterozygotes
together will result in some of the homozygous
T T + T embryonic lethal
offspring being homozygous (T/T    ) for the + + +
T (abnormal posterior
mutation, resulting in a severe and lethal mesoderm)
developmental abnormality in which the
posterior mesoderm does not develop.
 A conceptual tool kit 13

In recent years, new techniques for identifying developmental genes have appeared.
These depend on knowing the existence and sequence of a gene (for example, from
genomic sequences) and working backwards to determine its function in the animal,
usually by removing the gene or blocking its function. Such methods are known as
reverse genetics, as opposed to the traditional methods of forward genetics outlined
above. Examples of reverse genetics include gene knock-out, in which the gene is
effectively deleted from the animal’s genome by transgenic techniques (discussed in
Section 3.9), and gene knockdown or gene silencing, in which gene expression is
prevented by techniques such as RNA interference or antisense RNA (discussed in
Box 6A, p. 220).

Summary

The study of embryonic development started with the Greeks more than 2000 years
ago. Aristotle put forward the idea that embryos were not contained completely pre-
formed in miniature within the egg, but that form and structure emerged gradually, as
the embryo developed. This idea was challenged in the seventeenth and eighteenth
centuries by those who believed in preformation, the idea that all embryos that had
been, or ever would be, had existed from the beginning of the world. The emergence
of the cell theory in the nineteenth century finally settled the issue in favor of epigen-
esis, and it was realized that the sperm and egg are single, albeit highly specialized,
cells. Some of the earliest experiments showed that very early sea-urchin embryos are
able to regulate, that is to develop normally, even if cells are removed or killed. This
established the important principle that development must depend, at least in part, on
communication between the cells of the embryo. Direct evidence for the importance of
cell–cell interactions came from the organizer graft experiment carried out by Spemann
and Mangold in 1924, showing that the cells of the amphibian organizer region could
induce a new partial embryo from host tissue when transplanted into another embryo.
The role of genes in controlling development, by determining which proteins are made,
has only been fully appreciated in the past 50 years and the study of the genetic
basis of development has been made much easier in recent times by the techniques of
molecular biology and the availability of the DNA sequences of whole genomes.

A conceptual tool kit

Development into a multicellular organism is the most complicated fate a single liv-
ing cell can undergo; in this lies both the fascination and the challenge of develop-
mental biology. Yet only a few basic principles are needed to start to make sense of
developmental processes. The rest of this chapter is devoted to introducing these key
concepts. These principles are encountered repeatedly throughout the book, as we
look at different organisms and developmental systems, and should be regarded as a
conceptual tool kit, essential for embarking on a study of development.
Genes control development by determining where and when proteins are synthe-
sized, and many thousands of genes are involved. Gene activity sets up intracellular
networks of interactions between proteins and genes, and between proteins and pro-
teins, that give cells their particular properties. One of these properties is the abil-
ity to communicate with, and respond to, other cells. It is these cell–cell interactions
that determine how the embryo develops; no developmental process can therefore be
attributed to the function of a single gene or single protein. The amount of genetic and
molecular information on developmental processes is now enormous. In this book,
we will be highly selective and describe only those molecular details that give an
­insight into the mechanisms of development and illustrate general principles.
14 Chapter 1 History and basic concepts

1.8 Development involves the emergence of pattern, change in form, cell

differentiation, and growth

In most embryos, fertilization is immediately followed by cell division. This stage

is known as cleavage, and divides the fertilized egg into a number of smaller cells
(Fig. 1.14). Unlike the cell divisions that take place during growth of a tissue, there is
no increase in cell size between each cleavage division; the cleavage cell cycles con-
sist simply of phases of DNA replication, mitosis, and cell division (see Box 1B, p. 7).
Early embryos are therefore no bigger than the zygote, and there is little increase in
size during embryogenesis unless the embryo has access to a substantial source of
material it can use to grow—for example, the large yolk available to the chick embryo
and the nutrients provided to the mammalian embryo via the placenta.
Development is essentially the emergence of organized structures from an initially
very simple group of cells. It is convenient to distinguish four main developmental
processes, which occur in roughly sequential order in development, although in reality
they overlap with, and influence, each other considerably. They are pattern formation,
morphogenesis, cell differentiation, and growth. Pattern formation is the process by
which cellular activity is organized in space and time so that a well-ordered structure
develops within the embryo, and is of considerable importance in the early embryo.
In the developing arm, for example, pattern formation enables cells to ‘know’ whether
to make an upper arm or fingers, and where the muscles should form. There is no
universal strategy of patterning; rather, it is achieved by various cellular and molecular
mechanisms in different organisms and at different stages of development.
Pattern formation initially involves laying down the overall body plan—defining
the main body axes of the embryo that run from head (anterior) to tail (posterior),
and from back (dorsal) to underside (ventral). Most of the animals in this book have a
head at one end and a tail at the other, with the left and right sides of the body being
outwardly bilaterally symmetrical—that is, a mirror image of each other. In such ani-
mals, the main body axis is the antero-posterior axis, which runs from the head to
the tail. Bilaterally symmetrical animals also have a dorso-ventral axis, running from
the back to the belly. A striking feature of these axes is that they are almost always at
right angles to one another and so can be thought of as making up a system of coor-
dinates on which any position in the body could be specified (Fig. 1.15). Internally,
animals have distinct differences between the left and right sides in the disposition of
their internal organs—the human heart, for example, is on the left side. In plants, the
main body axis runs from the growing tip (the apex) to the roots and is known as the
apical–basal axis. Plants also have radial symmetry, with a radial axis running from
the center of the stem outwards.
Even before the body axes become clear, eggs and embryos often show a distinct
polarity, which in this context means that they have an intrinsic orientation, with
one end different from the other. Polarity can be specified by a gradient in a pro-
tein or other molecule, as the slope of the gradient provides direction. Many cells in

Fig. 1.14 Light micrographs of cleaving

Xenopus eggs.
 A conceptual tool kit 15

Xenopus laevis tailbud-stage embryo

Dorsal (back)

Right

Anterior Posterior
(head) (tail)

Fig. 1.15 The main axes of a developing

embryo. The antero-posterior axis and the
Left
Ventral (front) dorso-ventral axis are at right angles to one
another, as in a coordinate system.

­ eveloping embryos have their own individual polarity, and this can result in polarity
d
along a sheet of cells. It is as if a direction arrow is present in all the cells, making
one end slightly different from the other. This type of polarity is known as planar
cell polarity and is discussed further in Chapters 2 and 8. Cell polarity is sometimes
visible in the structures they produce—as with the epidermal hairs on the edge of the
Drosophila wing, which all point in the same direction.
At the same time that the axes are being formed, cells are being allocated to the
different germ layers—ectoderm, mesoderm, and endoderm (Box 1C). During fur-
ther pattern formation, cells of these germ layers acquire different identities so that

Box 1C Germ layers

The concept of germ layers is useful to

Germ layers
distinguish between regions of the early
embryo that give rise to quite distinct types Dorsal Dorsal
Vertebrates Insects
of tissues. It applies to both vertebrates and
invertebrates. All the animals considered in
this book, except for the cnidarian Hydra,
are triploblasts, with three germ layers: the
endoderm, which gives rise to the gut and
its derivatives, such as the liver and lungs
in vertebrates; the mesoderm, which gives
rise to the skeleto-muscular system, connec-
coelom
tive tissues, and other internal organs, such Ventral Ventral
as the kidney and heart; and the ectoderm,
which gives rise to the epidermis of the Germ layers Organs
skin and nervous system. These are speci-
fied early in development. The distinction Endoderm gut, liver, lungs gut
between the different germ layers can be Mesoderm skeleton, muscle, kidney, heart, blood muscle, heart, blood
fuzzy and there are notable exceptions. The
Ectoderm epidermis of skin, nervous system cuticle, nervous system
neural crest in vertebrates, for example, is
ectodermal in origin but gives rise not only
to neural tissue but also to some skeletal
elements, which would normally be consid-
ered mesodermal in origin.
16 Chapter 1 History and basic concepts

mouth

Fig. 1.16 Gastrulation in the sea

urchin. Gastrulation transforms the spherical gut
blastula into a structure with a tube—the
gut—through the middle. The left-hand side
anus
of the embryo has been removed.

­ rganized spatial patterns of differentiated cells emerge, such as the arrangement of

o
skin, muscle, and cartilage in developing limbs, and the arrangement of neurons in
the nervous system. In the earliest stages of pattern formation, differences between
cells are not easily detected and probably consist of subtle differences caused by a
change in activity of a very few genes.
The second important developmental process is change in form, or morphogenesis
(Chapter 8). Embryos undergo remarkable changes in three-dimensional form—you
need only look at your hands and feet. At certain stages in development, there are
characteristic and dramatic changes in form, of which gastrulation is the most strik-
ing. Almost all animal embryos undergo gastrulation, during which endoderm and
Weeks after fertilization
8 12 16 Birth mesoderm move inside, the gut is formed, and the main body plan emerges. During
gastrulation, cells on the outside of the embryo move inwards and, in animals such
Change in size as the sea urchin, gastrulation even transforms a hollow spherical blastula into a gas-
trula with a tube through the middle—the gut (Fig. 1.16). Morphogenesis in animal
embryos can also involve extensive cell migration. Most of the cells of the human
face, for example, are derived from cells that migrated from a tissue called the neural
crest, which originates on the dorsal side of the embryo. Morphogenesis can also
involve developmentally programmed cell death, called apoptosis, which is respon-
sible, for example, for the separation of our fingers and toes.
The third developmental process we must consider here is cell differentiation, in
which cells become structurally and functionally different from each other, ending up
as distinct cell types, such as blood, muscle, or skin cells. Differentiation is a gradual
process, cells often going through several divisions between the time at which they
start differentiating and the time they are fully differentiated (when some cell types
Change in proportions
stop dividing altogether), and is discussed in Chapter 10. In humans, the fertilized egg
gives rise to at least 250 clearly distinguishable types of cell.
Pattern formation and cell differentiation are very closely interrelated, as we can see
by considering the difference between human arms and legs. Both contain exactly the
same types of cell—muscle, cartilage, bone, skin, and so on—yet the pattern in which
they are arranged is clearly different. It is essentially pattern formation that makes us
different from elephants and chimpanzees.
The fourth process is growth—the increase in size. In general there is little growth
during early embryonic development and the basic pattern and form of the embryo
is laid down on a small scale, always less than a millimeter in extent. Subsequent
growth can be brought about in various ways: cell multiplication, increase in cell size,
and deposition of extracellular materials, such as bone and shell. Growth can also be
Fig. 1.17 The human embryo changes
morphogenetic in that differences in growth rates between organs, or between parts
shape as it grows. From the time the body
plan is well-established at 8 weeks until birth
of the body, can generate changes in the overall shape of the embryo (Fig. 1.17), as
the embryo increases in length some tenfold we shall see in more detail in Chapter 13.
(upper panel), while the relative proportion These four developmental processes are neither independent of each other nor
of the head to the rest of the body decreases strictly sequential. In very general terms, however, one can think of pattern forma-
(lower panel). As a result, the shape of the tion in early development specifying differences between cells that lead to changes in
embryo changes. Scale bar = 10 cm. form, cell differentiation, and growth. But in any real developing system there will be
After Moore, K.L.: 1983. many twists and turns in this sequence of events.
 A conceptual tool kit 17

1.9 Cell behavior provides the link between gene action and developmental
processes
Gene expression within cells leads to the synthesis of proteins that are responsible for
particular cellular properties and behavior, which in turn determine the course of embry-
onic development. The past and current patterns of gene activity confer a certain state, or
identity, on a cell at any given time, which is reflected in its molecular organization—in
particular which proteins are present. As we shall see, embryonic cells and their prog-
eny undergo many changes in state as development progresses. Other categories of cell
behavior that will concern us are intercellular communication, also known as cell–cell
signaling, changes in cell shape and cell movement, cell proliferation, and cell death.
Changing patterns of gene activity during early development are essential for pat-
tern formation. They give cells identities that determine their future behavior and
lead eventually to their final differentiation. And, as we saw in the example of induc-
tion by the Spemann organizer, the capacity of cells to influence each other’s fate by
producing and responding to signals is crucial for development. By their response to
signals for cell movement or a change in shape, for example, cells generate the physi-
cal forces that bring about morphogenesis (Fig. 1.18). The curvature of a sheet of localized
contraction
cells into a tube, as happens in Xenopus and other vertebrates during formation of the
neural tube (see Box 1A, pp. 2–3), is the result of contractile forces generated by cells
changing their shape at certain positions within the cell layer. An important feature of Fig. 1.18 Localized contraction of
cell surfaces is the presence of adhesive proteins known as cell-adhesion molecules particular cells can cause a whole sheet
of cells to fold. Contraction of a line of cells
that serve a variety of functions: they hold cells together in tissues, they enable cells
at their apices due to the contraction of
to sense the nature of the surrounding extracellular matrix, and they serve to guide
cytoskeletal elements causes a furrow to
migratory cells such as the neural crest cells of vertebrates, which leave the neural
form in a sheet of epidermis.
tube to form structures elsewhere in the body.
We can therefore describe and explain developmental processes in terms of how
individual cells and groups of cells behave. Because the final structures generated by
development are themselves composed of cells, explanations and descriptions at the
cellular level can provide an account of how these adult structures are formed.
Because development can be understood at the cellular level, we can pose the
question of how genes control development in a more precise form. We can now ask
how genes are controlling cell behavior. The many possible ways in which a cell can
behave therefore provide the link between gene activity and the morphology of the
adult animal—the final outcome of development. Cell biology provides the means by
which the genotype becomes translated into the phenotype.

1.10 Genes control cell behavior by specifying which proteins are made
What a cell can do is determined very largely by the proteins present within it. The
hemoglobin in red blood cells enables them to transport oxygen; the cells lining the
gut secrete specialized digestive enzymes; and skeletal muscle cells are able to con-
tract because they contain contractile structures composed of the proteins myosin,
actin and tropomyosin, as well as other muscle-specific proteins needed to make
muscle function correctly. All these are specialized proteins that are not involved in
the ‘housekeeping’ activities that are common to all cells and keep them alive and
functioning. Housekeeping activities include the production of energy and the meta-
bolic pathways involved in the breakdown and synthesis of molecules necessary for
the life of the cell. Although there are qualitative and quantitative variations in house-
keeping proteins in different cells, they are not important players in development. In
development we are concerned primarily with those proteins that make cells different
from one another and which are often known as tissue-specific proteins.
Genes control development mainly by specifying which proteins are made in which
cells and when. In this sense the genes are passive participants in development,
compared with the proteins they encode, which are the agents that directly determine
18 Chapter 1 History and basic concepts

cell ­behavior, which includes determining which genes are expressed. To produce
a particular protein, its gene must be switched on and transcribed into messenger
RNA (mRNA); the mRNA must then be translated into protein. The initiation of
transcription, or switching on of a gene, involves combinations of specialized gene-
regulatory proteins binding to control regions in the DNA.
Both transcription and translation are subject to several layers of control, and trans-
lation does not automatically follow transcription. Figure 1.19 shows the main stages
in gene expression at which the production of a protein can be controlled. Even after
a gene has been transcribed, for example, the mRNA may be degraded before it can
be ­exported from the nucleus. Even if an mRNA reaches the cytoplasm, its ­translation
may be prevented or delayed. In the eggs of many animals, preformed mRNA is pre-
vented from being translated until after fertilization. Another process that determines
which proteins are produced is RNA processing. In eukaryotic organisms—all organ-
isms other than bacteria and archaea—the initial RNA transcripts of many genes can
Control regions be cut and spliced in different ways to give rise to two or more different mRNAs; in
this way, a number of different proteins with different properties can be produced
DNA from a single gene.
cis-regulatory
control region Even if a gene has been transcribed and the mRNA translated, the protein may still
not be able to function immediately. Many newly synthesized proteins require fur-
transcription RNA polymerase
factor ther post-translational modification before they acquire biological activity. One very
common permanent modification that occurs to proteins destined for the cell mem-
brane or for secretion is the addition of carbohydrate side chains, or glycosylation.
promoter region
Reversible post-translational modifications, such as phosphorylation, can also signifi-
Coding region cantly alter protein function. Together, alternative RNA splicing and post-­translational
modification mean that the number of functionally different proteins that can be
DNA produced is considerably greater than the number of protein-coding genes would indi-
cate. For many proteins, their localization in a particular part of the cell, for example
1 transcription the nucleus, is essential for them to carry out their function.
Some genes, such as those for ribosomal RNAs (rRNAs) and transfer RNAs (tRNAs),
RNA do not code for proteins; in this case the RNAs themselves are the end-products.
Another category of genes encodes microRNAs (miRNAs), small RNA molecules that
2 processing
Nucleus prevent the translation of specific mRNAs (see Box 6B, p. 228). Some microRNAs are
known to be involved in gene regulation in development.
mRNA An intriguing question is how many genes out of the total genome are develop-
mental genes—that is, genes specifically required for embryonic development. This
is not easy to estimate. In the early development of Drosophila, at least 60 genes are
3 transport
Cytoplasm directly involved in pattern formation up to the time that the embryo becomes divided
nuclear
envelope into segments. In Caenorhabditis, at least 50 genes are needed to specify a small

Fig. 1.19 Gene expression and protein synthesis. A protein-coding gene comprises a
coding region—a stretch of DNA that contains the instructions for making the protein—and
adjacent control regions. These include the promoter region, to which general transcription
factors and the enzyme RNA polymerase bind to start transcription, and other cis-regulatory
ribosome 4 translation control regions at which other transcription factors bind to switch the gene on or off. These
latter control regions may be thousands of base pairs away from the promoter. When the
gene is switched on, the DNA sequence is transcribed to produce RNA (1). The RNA formed
protein by transcription is spliced to remove introns and the non-coding region at the start of the
gene (yellow) and processed within the nucleus (2) to produce mRNA. This is exported from
the nucleus to the cytoplasm (3) for translation into protein at the ribosomes (4). Control of
5 modification gene expression and protein synthesis occurs mainly at the level of transcription but can also
occur at later stages. For example, mRNA may be degraded before it can be translated, or if it
is not translated immediately it may be stored in inactive form in the cytoplasm for translation
carbohydrate at a later stage. Some proteins require a further step of post-translational modification (5) to
become biologically active. A very common post-translational modification is the addition of
carbohydrate side-chains (glycosylation) as shown here.
 A conceptual tool kit 19

r­ eproductive structure known as the vulva. These are quite small numbers compared
to the ­thousands of genes that are active at the same time; some of these are essential
to development in that they are necessary for maintaining life, but they provide no
or little information that influences the course of development. Many genes change
their activity in a systematic way during development, suggesting that they could be
developmental genes. The total number of genes in the nematode and Drosophila is
around 19,000 and 15,000, respectively, and the number of genes involved in develop-
ment is likely to be in the thousands. A systematic analysis of most of the Caenorhab-
ditis genome suggests that approximately 9% of the genes (1722 out of 20,000) are
involved in development.
Developmental genes typically code for proteins involved in the regulation of
cell ­behavior—receptors, growth factors, intracellular signaling proteins, and gene-
­regulatory proteins. Many of these genes, especially those for receptors and signaling
molecules, are used throughout an organism’s life, but others are active only during
embryonic development.
Some of the techniques used to determine where and when a gene is active
are described in Box 1D (p. 20) and Box 3A (p.115). It is also possible to prevent
the expression of a given gene and so prevent its protein from being made. This can
be done using antisense RNAs, which are small RNA molecules of about 25 nucle-
otides ­complementary in sequence to part of a gene or mRNA. Such antisense RNAs
can block gene function by binding to the gene or its mRNA. Artificial RNA molecules
called morpholinos are used, which are more stable than normal RNA. Another tech-
nique for blocking gene expression is RNA interference, which operates in a similar
way, but uses a different type of small RNA called small interfering RNAs (siRNAs)
to exploit cells’ own metabolic pathways for degrading mRNAs. These techniques are
discussed in more detail in Box 6A, p. 220.

1.11 The expression of developmental genes is under tight control

All the somatic cells in an embryo are derived from the fertilized egg by successive
rounds of mitotic cell division. Thus, with rare exceptions, they all contain identical
genetic information, the same as that in the zygote. This is the principle of genetic
equivalence. The differences between cells must therefore be generated by differ-
ences in gene activity that lead to the synthesis of different proteins. This is the
principle of differential gene expression. Turning the correct genes on or off in the
right cells at the right time becomes the central issue in development. As we shall
see throughout the book, the genes do not provide a blueprint for development but
a set of instructions. Key elements in regulating the readout of these instructions are
the control ­regions that are located adjacent to developmental genes and genes for
specialized proteins. These are acted on by gene-regulatory proteins, or transcrip-
tion factors, which switch genes on or off by, respectively, activating or repressing
transcription. Some gene-regulatory proteins act by binding directly to the DNA of
the control regions (see Fig. 1.19) whereas others interact with transcription factors
already bound to the DNA.
Developmental genes are highly regulated to ensure they are switched on only at
the right time and place in development. This is a fundamental feature of develop-
ment. To achieve this, developmental genes usually have extensive and complex con-
trol regions composed of one or more regulatory modules, known as cis-regulatory
modules (the ‘cis’ refers to the fact that the regulatory region is on the same DNA
molecule as the gene it controls). Each module contains multiple binding sites for
different transcription factors, and it is the precise combination of factors that binds
that determines whether the gene is on or off. On average, a module will have binding
sites for four to eight different transcription factors, and these factors will in turn often
bind other co-activator or co-repressor proteins.
20 Chapter 1 History and basic concepts

Box 1D Tracking gene expression in embryos

In order to understand how gene expression is guiding development, it is essential

Single-stranded Mixture of single-
DNA probes stranded mRNA to know exactly where and when particular genes are active. Genes are switched on
specific for sequence A molecules in the cell and off during development and patterns of gene expression are continually chang-
ing. Several powerful techniques can show where a gene is being expressed in the
A* A* B
embryo.
A*
A One set of techniques uses in situ hybridization to detect the mRNA that is being
E transcribed from a gene. If a DNA or RNA probe is complementary in sequence to an
mRNA being transcribed in the cell, it will base-pair precisely (hybridize), with the
D mRNA (see figure, left). A DNA probe can therefore be used to locate its complemen-
tary mRNA in a tissue slice or a whole embryo. The probe may be labeled in various
C
ways—with a radioactive isotope, a fluorescence tag (fluorescence in situ hybridiza-
Mix tion, FISH; see Fig. 2.10), or an enzyme—to enable it to be detected. Radioactively
labeled probes are detected by autoradiography, whereas probes labeled with fluo-
rescent dyes are observed by fluorescence microscopy. Enzyme-labeled probes are
B detected by the enzyme’s conversion of a colorless substrate into colored product.
E
A* Labeling with fluorescent dyes or enzymes has the advantage over radioactive label-
A ing in that the expression of several different genes can be detected simultaneously,
by labeling the different probes with different colored dyes, and it also means that
researchers do not have to work with radioactive chemicals. To detect protein expres-
A* sion and localization, the detection agent is an antibody specific for the protein. Again,
A antibodies are tagged either with fluorescent dyes (see Fig. 2.33) or with enzymes
A
C (see Fig. 2.31). These techniques, and those for detecting mRNA, can only be applied
A*
to fixed tissue sections and embryos.
D
The pattern and timing of gene expression can also be followed by inserting a
‘reporter’ gene into an animal using transgenic ­techniques. The reporter gene codes
A* can only bind to its complementary
sequence, A for some easily detectable protein and is placed under the control of an appropriate
­promoter that will switch on the expression of the protein at a particular time and
place. One commonly used reporter gene is the lacZ ­sequence coding for the bacterial enzyme β-glactosidase. The presence of the
enzyme can be detected by treating the specimen after fixation with a modified substrate, which the enzyme reacts with to give a
blue-colored product (see Fig. 5.24).
Other very commonly used reporter genes are those for small
fluorescent proteins, such as green fluorescent protein (GFP), which a
fluoresce in various colors and were isolated from ­jellyfish and other
marine organisms. These proteins are extremely useful as they are
harmless to living cells and are readily be made ­visible by illuminat-
ing cultured cells, embryos or even whole ­animals with light of a
suitable wavelength. Gene expression can therefore be tracked in
living embryos and cells (see Fig. 2.22). The three scientists who
b
discovered and developed GFP and other fluorescent proteins for
use in imaging were awarded the 2008 Nobel Prize for Chemistry.
The zebrafish embryos illustrated show two different methods
of detecting expression of the same gene. In the embryo in the
upper photo, the expression of Sox10 mRNA (blue) in neural crest
cells has been detected by nucleic acid hybridization in a fixed
embryo. The lower photo is of a live transgenic zebrafish embryo
at the same stage of development, which carries a reporter trans-
gene in which a GFP coding sequence is attached to the promoter
region of Sox10. GFP is therefore expressed in the neural crest
cells that express Sox10.
Photographs courtesy of Robert Kelsh.
 A conceptual tool kit 21

These regulatory regions are called ‘modules’ as each can function somewhat
Positive feedback
independently of each other. This means that a gene with more than one regulatory
module is usually able to respond to different combinations of inputs, and so can be activator
expressed at different times and in different places within the embryo in response to
developmental signals. Different genes can have the same control module, which usu-
ally means that they will be expressed together, or different genes may have modules
that contain some, but not all, of the binding sites in common, introducing subtle
differences in the timing or location of expression. Thus, an organism’s genes are gene 1 gene 2
linked in complex interdependent networks of expression through their regulatory
modules and the proteins that bind to them. Common examples of such regulation
are positive-feedback and negative-feedback loops in which a transcription factor
respectively promotes or represses the expression of a gene whose product maintains
that gene expression (Fig. 1.20). Indeed, extensive networks of gene interactions for
various stages of development have now been worked out for animals such as the sea
urchin (see Chapter 6) and zebrafish (see Box 4E, p. 162).
As all the key steps in development reflect changes in gene activity, one might be
tempted to think of development simply in terms of mechanisms for controlling gene Negative feedback
expression. But this would be a big mistake. For gene expression is only the first
activator
step in a cascade of cellular processes that lead via protein synthesis to changes in
cell ­behavior and so direct the course of embryonic development. Proteins are the
machines that drive development. To think only in terms of genes is to ignore crucial
aspects of cell biology, such as change in cell shape, which may be initiated at several
steps removed from gene activity. In fact, there are very few cases where the complete
sequence of events from gene expression to altered cell behavior has been worked gene 3 gene 4
out. The route leading from gene activity to a structure such as the five-fingered hand
is tortuous.

1.12 Development is progressive and the fate of cells becomes determined at

different times

As embryonic development proceeds, the organizational complexity of the embryo

becomes vastly increased over that of the fertilized egg, as first recognized by Aris- Fig. 1.20 Simple genetic feedback
totle’s ‘epigenesis’ principle outlined earlier in the chapter. Many different cell types loops. Top: gene 1 is turned on by an
are formed, spatial patterns emerge, and there are major changes in shape. All this activating transcription factor (green); the
occurs more or less gradually, depending on the particular organism. But, in general, protein it produces (red) activates gene 2.
the embryo is first divided up into a few broad regions, such as the future germ layers The protein product of gene 2 (blue) not only
(mesoderm, ectoderm, and endoderm). Subsequently, the cells within these regions acts on further targets, but also activates
have their fates more and more finely determined. Determination implies a stable gene 1, forming a positive feedback loop
change in the internal state of a cell, and an alteration in the pattern of gene activity that will keep genes 1 and 2 switched on
even in the absence of the original activator.
is assumed to be the initial step, leading to a change in the proteins produced in the
Bottom: when the product of the gene at the
cell. Different mesoderm cells, for example, eventually become determined as muscle,
end of the pathway inhibits the first gene,
cartilage, bone, the fibroblasts of connective tissue, and the cells of the dermis of the
a negative feedback loop is formed. Arrows
skin. indicate activation; barred line indicates
It is important to understand clearly the distinction between the normal fate of a inhibition.
cell at any particular stage, and its state of determination. The fate of a group of cells
merely describes what they will normally develop into. By marking cells of the early
embryo one can find out, for example, which ectodermal cells will normally give rise
to the nervous system, and of those, which, in particular, will give rise to the neural
retina of the eye. However, that in no way implies that those cells can only develop
into a retina, or are already committed, or determined, to do so in the early embryo.
A group of cells is called specified if, when isolated and cultured in the neutral
environment of a simple culture medium away from the embryo, they develop more
or less according to their normal fate (Fig. 1.21). For example, cells at the animal pole
of the amphibian blastula (see Box 1A, pp. 2–3) are specified to form ectoderm, and
22 Chapter 1 History and basic concepts

Normal fate Region B not determined Region B determined Region B specified

transplanted region labeled transplanted region labeled

Region A

Fate map

Region B

Differentiated
tissue

Fig. 1.21 The distinction between cell

fate, determination, and specification. will form epidermis when isolated. Cells that are ‘specified’ in this technical sense
In this idealized system, regions A and B need not yet be ‘determined’, for influences from other cells can change their normal
differentiate into two different sorts of cells, fate; if tissue from the animal pole is put in contact with cells from the vegetal pole,
depicted as hexagons and squares. The fate some animal pole tissue will form mesoderm instead of epidermis. At a later stage
map (first panel) shows how they would
of development, however, the cells in the animal region have become determined as
normally develop. If cells from region B are
ectoderm and their fate cannot then be altered. Tests for specification rely on cultur-
grafted into region A and now develop as
ing the tissue in a neutral environment lacking any inducing signals, and this is often
A-type cells, the fate of region B has not yet
been determined (second panel). By contrast, difficult to achieve.
if region B cells are already determined The state of determination of cells at any developmental stage can be demonstrated
when they are grafted to region A, they will by transplantation experiments. At the blastula stage of the amphibian ­embryo, one
develop as B cells (third panel). Even if B cells can graft the ectodermal cells that give rise to the eye into the side of the body and
are not determined, they may be specified, in show that the cells develop according to their new position; that is, into mesodermal
that they will form B cells when cultured in cells. At this early stage, their potential for development is much greater than their
isolation from the rest of the embryo (fourth normal fate. However, if the same operation is done at a later stage, then the future
panel). eye region will form structures typical of an eye (Fig. 1.22). At the earlier stage
the cells were not yet determined as prospective eye cells, whereas later they had
become so.
It is a general feature of development that cells in the early embryo are less nar-
rowly determined than those at later stages; with time, cells become more and more
restricted in their developmental potential. We assume that determination involves
a change in the genes that are expressed by the cell and that this change fixes or
restricts the cell’s fate, thus reducing its developmental options.
We have already seen how, even at the two-cell stage, the cells of the sea-urchin
embryo do not seem to be determined. Each has the potential to generate a whole
new larva (see Section 1.3). Embryos like these, where the potential of cells is much
greater than that indicated by their normal fate, are termed regulative. Vertebrate
embryos are among those capable of considerable regulation. In contrast, those
embryos where, from a very early stage, the cells can develop only according to their
early fate are termed mosaic. The term mosaic has a long history (see Section 1.3). It
describes eggs and embryos that develop as if their pattern of future development is
 A conceptual tool kit 23

Fate mapping of normal development locates the presumptive eye region

presumptive Tailbud-stage embryo

presumptive eye region
epidermis
presumptive
neural plate
presumptive presumptive
endoderm mesoderm
blastopore normal eye

The presumptive eye region of a gastrula, transplanted into the trunk

of a neurula embryo, forms structures typical of the trunk

transplantation host tissue somitic tissue from graft

Gastrula Host neurula Fig. 1.22 Determination of the eye region

with time in amphibian development. If the
region of the gastrula that will normally give
A transplant of the eye region from a later-stage embryo develops as an eye rise to an eye is grafted into the trunk region
of a neurula (middle panel), the graft forms
transplantation eye-like structure
structures typical of its new location, such
as notochord and somites. If, however, the
eye region from a neurula is grafted into the
same site (bottom panel), it develops as an
Neurula Host neurula eye-like structure, since at this later stage it
has become determined.

laid down very early, even in the egg, as a spatial mosaic of different molecules in the
egg cytoplasm. These cytoplasmic determinants get distributed in a regular fashion
into different cells during cleavage, thus determining the fate of that cell lineage at
an early stage. The different parts of the embryo then develop quite independently
of each other. Caenorhabditis and ascidians are examples of embryos with signifi-
cant amounts of mosaic development and they are discussed in Chapter 6. In such
embryos, cell interactions may be quite limited. But the demarcation between regula-
tive and mosaic strategies of development is not always sharp, and in part reflects the
time when determination occurs; it occurs much earlier in mosaic systems.
The difference between regulative and mosaic embryos also reflects the relative
importance of cell–cell interactions in each system. The occurrence of regulation abso-
lutely requires interactions between cells, for how else could normal development
take place and how could deficiencies be recognized and restored? A truly mosaic
embryo, however, would in principle not require such interactions. No purely mosaic
embryos are known to exist.

1.13 Inductive interactions can make cells different from each other

Making cells different from one another is central to development. There are many
times in development where a signal from one group of cells influences the develop-
ment of an adjacent group of cells. This is known as induction, and the classic exam-
ple is the action of the Spemann organizer in amphibians (see Section 1.4). Inducing
signals may be propagated over several or even many cells, or be highly localized.
The inducing signals from the amphibian organizer affect many cells, whereas other
inducing signals may be transmitted from a single cell to its immediate neighbor. Two
24 Chapter 1 History and basic concepts

different types of inductions should be distinguished: permissive and instructive.

Cell signaling
Permissive inductions occur when a cell makes only one kind of response to a signal,
and makes it when a given level of signal is reached. By contrast, in an instruc-
tive induction, the cells respond differently to different concentrations of the signal.
‘Antagonistic’ signal molecules that block induction, for example, by preventing the
­inducing signal reaching the cell or binding to a cell-surface receptor, are also impor-
tant in controlling development.
Inducing signals are transmitted between cells in three main ways (Fig. 1.23).
First, the signal can be transmitted through the extracellular space in the form of a
Diffusion secreted diffusible molecule. Second, cells may interact directly with each other by
means of molecules located on their surfaces. In both these cases, the signal is gener-
ally received by receptor proteins in the cell membrane and is subsequently relayed
through intracellular signaling systems to produce the cellular response. Third, the
signal may pass from cell to cell directly. In most animal cells, this is through gap
signal junctions, which are specialized protein pores in the apposed plasma membranes
providing direct channels of communication between the cytoplasms of adjacent cells
through which small molecules can pass. Plant cells are connected by strands of
cytoplasm called plasmodesmata, through which even quite large molecules such as
Direct contact
proteins can pass directly from cell to cell.
In the case of signaling by a diffusible protein or by direct contact, the signal is
received at the cell membrane. If it is to alter gene expression in the nucleus, or
change cell behavior in other ways, that signal has to be transmitted from the mem-
signal
brane to the cell’s interior. This process is known generally as signal transduction
or intra­cellular signaling, and is carried out by relays of intracellular signaling mol-
ecules that are activated when the extracellular signal molecule binds to its recep-
tor. An example of one intracellular signaling pathway important in development is
Gap junction illustrated in Box 1E (p. 26)—the pathway activated by members of the Wnt family of
secreted signal proteins. The basic Wnt pathway, like many others we shall encoun-
ter, is used again and again in different developmental contexts in all the animals
included in this book.
Within the cell, signaling proteins and small-molecule second messengers, such as
cyclic AMP, interact with one another to transmit the signal onward in the cell. Activa-
tion and deactivation of the pathway components by phosphorylation is an important
feature of most signaling pathways. In the case of development, most of the signaling
leads to the activation or repression of genes by switching transcription on or off.
Fig. 1.23 An inducing signal can be Another important target of extracellular signals in development is the cytoskeleton,
transmitted from one cell to another the internal network of protein fibers whose rearrangement enables cells to change
in three main ways. The signal can be shape, to move, and to divide. Signaling pathways are also used to temporarily alter
a secreted diffusible molecule, which enzyme activity and metabolic activity within cells and to initiate nerve impulses in
interacts with a receptor on the target cell neurons. The various signals a cell may be receiving at any one time are integrated by
surface (second panel), or the signal can be cross-talk between the different intracellular signaling pathways to produce an appro-
produced by direct contact between two priate response. An example of what can happen when an important developmental
complementary proteins at the cell surfaces
signaling pathway is defective is illustrated in Box 1F (p. 28).
(third panel). If the signal involves a small
An important feature of induction is whether or not the responding cell is able
molecule it may pass directly from cell to
to respond to the inducing signal. This competence may depend, for example, on
cell through gap junctions in the plasma
membrane (fourth panel).
the presence of the appropriate receptor and transducing mechanism, or on the pres-
ence of the particular transcription factors needed for gene activation. A cell’s com-
petence for a particular response can change with time; for example, the Spemann
organizer can induce changes in the cells it affects only during a restricted window
of time.
In embryos, it seems that small is generally beautiful where signaling and pattern
formation are concerned. Most patterns are specified on a scale involving just tens of
cells and distances of only 100 to 500 µm. The final organism may be very big, but this
is almost entirely due to growth after the basic pattern has been formed.
 A conceptual tool kit 25

1.14 The response to inductive signals depends on the state of the cell

Although an inductive signal can be viewed as an instruction to cells about how to

behave, it is important to realize that responses to such a signal are entirely depen-
dent on the current state of the cells. Not only does the individual cell have to be
competent to respond, but the number of possible responses is usually very limited.
An inductive signal can only select one response from a small number of possible
cellular responses. Thus instructive inducing signals would be more accurately called
selective signals. A truly instructive signal would be one that provided the cell with
entirely new information and capabilities, by providing it with, for example, new
genes, which is not thought to occur during development.
The fact that, at any given time, an inductive signal selects one out of several pos-
sible responses has several important implications for biological economy. On the
one hand, it means that different signals can activate a particular gene at different
stages of development: the same gene is often turned on and off repeatedly during
development. On the other, the same signal can be used to elicit different responses in
different cells. A particular signaling molecule, for example, can act on several types
of cell, evoking a characteristic and different response from each, depending on their
developmental history. As we will see in future chapters, evolution has been lazy with
respect to this aspect of development, and a small number of intercellular signaling
molecules are used again and again for different purposes.

1.15 Patterning can involve the interpretation of positional information

One general mode of pattern formation can be illustrated by considering the pattern-
ing of a simple non-biological model—the French flag (Fig. 1.24). The French flag
has a simple pattern: one-third blue, one-third white, and one-third red, along just
one axis. Moreover, the flag comes in many sizes but always the same pattern, and
thus can be thought of as mimicking the capacity of an embryo to regulate. Given a
line of cells, any one of which can be blue, white, or red, and given also that the line
of cells can be of variable length, what sort of mechanism is required for the line to
develop the pattern of a French flag?
One solution is for the group of cells to acquire positional information. That is,
each cell acquires an identity, or positional value, that is related to its position along
the line with respect to the boundaries at either end. After they have acquired their
positional values, the cells interpret this information by differentiating according to
their genetic program. Those in the left-hand third of the line will become blue, those
in the middle-third white, and so on. Very good evidence that cells use positional sig-
nals comes from the regeneration of amphibian and insect limbs, in which the missing
regions are precisely replaced (discussed in Chapter 14).
Pattern formation using positional information implies at least two distinct stages:
first the positional value has to be related to some boundary, and then it has to be
interpreted. The separation of these two processes has an important implication: it
means that there does not need to be a set relation between the positional values and
how they are interpreted. In different circumstances, the same set of positional values
could be used to generate the Italian flag or another pattern. How positional values
will be interpreted will depend on the particular genetic instructions active in the
group of cells and thus the cells’ developmental history.
Cells could have their position specified by various mechanisms. The simplest
is based on a gradient of some substance. If the concentration of some chemical
decreases from one end of a line of cells to the other, then the concentration of that
chemical in any cell along the line effectively specifies the position of the cell with
respect to the boundary (Fig. 1.25). A chemical whose concentration varies and which
is involved in pattern formation is called a morphogen. In the case of the French flag, Fig. 1.24 The French flag.
26 Chapter 1 History and basic concepts

Box 1E Signal transduction and intracellular signaling

Secreted signaling proteins acting

Wnt absent Wnt present
on receptors in cell membranes
activate numerous different intra-
Wnt
cellular signaling pathways. The LRP6 Frizzled
example illustrated here is the
important ­developmental signal-
P
ing pathway initiated by the Wnt P P
P
family of intercellular signaling P
Axin P
proteins (pronounced ‘Wints’). Dishevelled
Named after the proteins encoded CK1� GSK-3 CK1�
by the Wingless gene in Droso-
P
phila and the Int gene (now Wnt1) Axin P
in mammals, Wnt proteins are P �-catenin
APC P degradation inhibited
involved in specifying cell fate in
GSK-3
a multiplicity of developmental
contexts in all multicellular ani- free �-catenin
�-catenin
mals studied so far. We shall meet degradation accumulates
many of the functions of the Wnt
Cytoplasm
family throughout the book. In
adult humans, Wnt is required to HDAC
maintain stem cells in their self- CTBP Gro
TCF TCF
renewing state, and incorrect con-
trol of the Wnt signaling pathway Nucleus
is implicated in common cancers,
such as colon cancer.
Wnt proteins can activate a number of different intracellular in the nucleus, transcriptional co-repressors bind to the TCF
signaling pathways: the one we will illustrate here is involved ­transcription factors and prevent gene expression.
in specifying cell fates in the early development of all animals. Like most extracellular signals, Wnt acts on its target cells by
It is known as the canonical Wnt/β-catenin pathway as it was binding to a specific transmembrane receptor protein (Frizzled)
the first to be discovered and leads to the accumulation of the at the cell surface (right-hand side of the figure). The signal that
protein β-catenin and its action as a transcriptional co-activator. Wnt has bound is transmitted across the membrane by Frizzled
Other Wnt pathways will be described when they are encoun- and an associated co-receptor, LRP, even though Wnt itself
tered in the course of the book. A simplified version of the does not enter the cell. Activation of these receptors causes
­canonical Wnt/β-catenin pathway is shown in the figure. As with the same protein kinases that phosphorylated β-catenin in the
many developmental signaling pathways, the initial Wnt signal is destruction complex to now become associated with the mem-
converted, or ‘transduced’, into changes in gene expression in the brane and phosphorylate the cytoplasmic tail of the activated
nucleus. In this case, the Wnt signal prevents the ­degradation of LRP. Protein phosphorylation, which changes the activity of
the protein β-catenin in the cytoplasm, and by doing so enables proteins or their interactions with other proteins, is ubiquitous
β-catenin to accumulate and to enter the nucleus, where it binds in signaling pathways as a means of transmitting the signal
to and activates transcription factors of the TCF (T-cell-specific onwards. The intracellular signaling protein Dishevelled (Dsh)
factor) family and so activates the expression of selected target and the protein Axin (which is a component of the destruction
genes. complex) are recruited to the tails of LRP and Frizzled at the
In the absence of a Wnt signal, β-catenin is bound by a ‘destruc- membrane. This prevents the destruction complex forming and
tion complex’ of proteins in the cytoplasm (left-hand side of the leaves β-catenin free to accumulate and to enter the nucleus.
figure). This includes protein kinases (CK1γ and GSK-3β) that Inside the nucleus, β-catenin binding to TCF displaces the co-
phosphorylate β-catenin, thus targeting it for ubiquitination repressors, lifts the repression and enables the target genes to
and degradation in the proteasome. In the absence of β-catenin be expressed.
 A conceptual tool kit 27

we assume a source of morphogen at one end and a sink at the other, and that the Each cell has the potential to develop
concentrations of morphogen at both ends are kept constant but are different from as blue, white, or red
each other. Then, as the morphogen diffuses down the line, its concentration at any
point effectively provides positional information. If the cells can respond to threshold
concentrations of the morphogen—for example, above a particular concentration the
cells develop as blue, while below this concentration they become white, and at yet
another, lower, concentration they become red—the line of cells will develop as a
French flag (see Fig. 1.25). Thresholds can represent the amount of morphogen that Position of each cell is defined by the
concentration of morphogen
must bind to receptors to activate an intracellular signaling system, or concentrations
of transcription factors required to activate particular genes. The use of threshold con- Concentration
centrations of transcription factors to specify position is most beautifully illustrated of morphogen

in the early development of Drosophila, as we shall see in Chapter 2. Other ways of

specifying positional information are by direct intercellular interactions and by timing
mechanisms.
The French flag model illustrates two important features of development in the real
1 2 3 4 5 6
world. The first is that, even if the length of the line varies, the system regulates and
the pattern will still form correctly, given that the boundaries of the system are prop-
erly defined by keeping the different concentrations of morphogen constant at either Positional value is interpreted by the cells
end. The second is that the system could also regenerate the complete original pattern which differentiate to form a pattern
if it were cut in half, provided that the boundary concentrations were re-established.
Concentration thresholds
It is therefore truly regulative. We have discussed it here as a one-dimensional pat- of morphogen
terning problem, but the model can easily be extended to provide patterning in two
dimensions (Fig. 1.26).
It is still unclear how positional information is specified: morphogen diffusion is
one of the mechanisms that have been proposed, but this may not be reliable enough.
We do not know how fine-grained the differences in position are. For example, does
every cell in the early embryo have a distinct positional value of its own and is it able
to make use of this difference?
Fig. 1.25 The French flag model of
pattern formation. Each cell in a line of
1.16 Lateral inhibition can generate spacing patterns cells has the potential to develop as blue,
white, or red. The line of cells is exposed to
Many structures, such as the feathers on a bird’s skin, are found to be more or less a concentration gradient of some substance
regularly spaced with respect to one another. One mechanism that gives rise to such and each cell acquires a positional value
spacing is called lateral inhibition (Fig. 1.27). Given a group of cells that all have the defined by the concentration at that point.
potential to differentiate in a particular way, for example as feathers, it is possible to Each cell then interprets the positional
regularly space the cells that will form feathers by a mechanism in which the first cells value it has acquired and differentiates
to begin to form feathers (which occurs essentially at random) inhibit adjacent cells into blue, white, or red, according to a
from doing so. This is reminiscent of the regular spacing of trees in a forest by compe- predetermined genetic program, thus forming
the French flag pattern. Substances that
tition for sunlight and nutrients. In embryos, lateral inhibition is often the result of the
can direct the development of cells in this
differentiating cell producing an inhibitory molecule that acts locally on the nearest
way are known as morphogens. The basic
neighbors to prevent them developing in a similar way. requirements of such a system are that
the concentration of substance at either
1.17 Localization of cytoplasmic determinants and asymmetric cell division end of the gradient must remain different
from each other but constant, thus fixing
can make daughter cells different from each other
boundaries to the system. Each cell must
Positional specification is just one way in which cells can be given a particular iden- also contain the necessary information to
tity. A separate mechanism is based on cytoplasmic localization and asymmetric interpret the positional values. Interpretation
cell division (Fig. 1.28). Asymmetric divisions are so called because they result in of the positional value is based upon
daughter cells having properties different from each other, independently of any envi- different threshold responses to different
concentrations of morphogen.
ronmental influence. The properties of such cells therefore depend on their lineage
or line of descent, and not on environmental cues. Although some asymmetric cell
divisions are also unequal divisions in that they produce cells of different sizes, this
is not usually the most important feature in animals; it is the unequal distribution
of cytoplasmic factors that makes the division asymmetric. An alternative way of
28 Chapter 1 History and basic concepts

Box 1F When development goes awry

The signaling pathway stimulated by the small secreted pro- As clinical geneticists continued to identify genes involved
tein Sonic hedgehog (Shh) (named after the Hedgehog protein in holoprosencephaly, more genes encoding components in
in Drosophila, which was the first member of this family to be the Shh pathway were implicated. In all cases, the cause of the
discovered) regulates many developmental processes in humans abnormality could be ascribed to a deficiency in Shh signaling,
and other vertebrates from very early in embryogenesis onwards. but ­different genes were affected in different cases. In some
In adults, Shh controls the growth of stem cells, and misregula- groups of patients, the abnormality was caused by mutations
tion of this signaling pathway leads to tumor formation. Like in the Shh gene itself, but in others the affected genes encoded
many other signaling pathways in development (for example, the either the receptor for Shh (Patched) or a transcription factor
Wnt pathway discussed in Box 1E, p. 26), the end results of Shh at the end of the pathway (Gli2). Mutations in a gene encod-
signaling are changes in gene expression. ing an ­enzyme required for the synthesis of cholesterol, which
Unraveling these important sig- promotes the activity of a protein called
naling pathways can help us under- Smoothened, which acts together with
stand the reasons for some develop- Patched to control the Shh pathway,
mental abnormalities, and Shh is a were found in some patients. (The full
good example. A relatively common Shh pathway is illustrated in Box 11C,
birth defect is a condition known as pp. 424–425). Shh signalling takes place
holoprosencephaly, which is char- on cellular structures called primary cilia,
acterized by a spectrum of midline which are immotile cilia present on most
brain and facial abnormalities. The vertebrate cell types, and patients with
severity of the condition varies genetic conditions in which formation of
greatly: at one end of the spectrum these structures is defective were also
the forebrain (the prosencephalon) found to have a characteristic facial phe-
fails to divide in the midline to form notype, although the alterations are very
right and left brain hemispheres, subtle and require sophisticated three-
there is complete absence of the dimensional modeling to detect. Thus,
midline of the face and the fetuses the discovery of genes that produce
do not survive; at the other end, similar symptoms in human patients can
the division of the forebrain is only highlight genes in a common develop-
mildly affected and the only observ- mental pathway.
able defect may be a cleft lip or the Facial defects similar to those pro-
presence of one incisor tooth rather duced by mutations in genes for Shh
than two. pathway proteins can also be caused
The first indications that Shh was involved in brain develop- by environmental factors. Pregnant ewes grazed in pastures
ment came in the early 1990s, when Shh was discovered to be in which the California False Hellebore (Veratrum californicum,
expressed in the ventral midline of the nervous system in mice. pictured) is growing may give birth to lambs with severe facial
In 1996, clinical geneticists identified one of the genes affected abnormalities, such as cyclopia (the presence of a single eye in
in patients with inherited holoprosencephaly—it turned out to be the midline of the face). The chemical in the plant that causes
Shh. At around the same time, it was found that when the Shh these ­defects is the alkaloid cyclopamine, and it too disrupts Shh
gene is completely knocked out experimentally in mice (see Sec- signaling. Cyclopamine interferes with the activity of Smooth-
tion 3.9 and Box 3B, p. 118 for the technique of gene knock- ened and reduces Shh signaling. This property of cyclopamine is
outs), forebrain division does not occur and the embryos have now being explored in a therapeutic context to ­develop drugs to
a condition known as cyclopia, with loss of midline facial struc- treat cancers that display abnormal Shh signaling.
tures and development of a single eye in the middle of the
face—a condition representing the most severe manifestation of Photograph courtesy of Jerry Friedman, Copyright Creative
­holoprosencephaly. Commons Attribution ShareAlike 3.00
 A conceptual tool kit 29

making the French flag pattern from the egg would be to have chemical differences
(representing blue, white, and red) distributed in the egg in the form of determinants
that foreshadowed the French flag. When the egg underwent cleavage, these cytoplas-
mic determinants would become distributed among the cells in a particular way and
a French flag would develop. This would require no interactions between the cells,
which would have their fates determined from the beginning.
Although such extreme examples of mosaic development (see Section 1.12) are not
known in nature, there are well-known cases where eggs or cells divide so that some
cytoplasmic determinant becomes unequally distributed between the two daughter
cells and they develop differently. This happens at the first cleavage of the nematode
egg, for example, and defines the antero-posterior axis of the embryo. The germ cells
of Drosophila are also specified by cytoplasmic determinants, in this case contained in
the cytoplasm located at the posterior end of the egg. And in Xenopus, a key determi-
nant in the first stages of embryonic development is a protein known as VegT, which
is localized in the vegetal region of the fertilized egg. In general, however, as develop-
ment proceeds, daughter cells become different from each other because of signals
from other cells or from their extracellular environment, rather than because of the
Fig. 1.26 Positional information can be
unequal distribution of cytoplasmic determinants. used to generate patterns. A good example,
A very particular type of asymmetric division is seen in stem cells. When they as shown here, is where the people seated
divide, these self-renewing cells can produce one new stem cell and another daugh- in a stadium each have a position defined by
ter cell that will differentiate into one or more cell types (Fig. 1.29). Stem cells that their row and seat numbers. Each position
are capable of giving rise to all the cell types in the body are present in very early has an instruction about which colored card
embryos, and are known as embryonic stem cells, whereas in adults, stem cells of to hold up, and this makes the pattern. If
apparently more limited potential are responsible for the continual renewal of tissues the instructions were changed, a different
such as blood, epidermis, and gut epithelium, and for the replacement of tissues such pattern would be formed, and so positional
as muscle when required. The difference in the behavior of the daughter cells pro- information can be used to produce an
enormous variety of patterns.
duced by division of a stem cell can be due either to asymmetric distribution of cyto-
plasmic determinants or to the effects of external signals. The generation of neurons
in Drosophila, for example, depends on the asymmetric distribution of cytoplasmic
determinants in the neuronal stem cells, whereas the production of the different types
of blood cells seems to be regulated to a great extent by extracellular signals.
Embryonic stem cells, which can give rise to all the types of cells in the body,
are called pluripotent, while stem cells that give rise to a more limited number of
cell types are known as multipotent. The hematopoietic stem cells in bone marrow,
which give rise to all the different types of blood cells, are an example of multipotent
stem cells. Stem cells are of great interest in relation to regenerative medicine, as they
could provide a means of repairing or replacing damaged organs and are discussed
in Chapter 10. For most animals, the only cell that can give rise to a complete new
organism is the fertilized egg, and this is described as totipotent.

1.18 The embryo contains a generative rather than a descriptive program

All the information for embryonic development is contained within the fertilized egg. Fig. 1.27 Lateral inhibition can give a
So how is this information interpreted to give rise to an embryo? One possibility is spacing pattern. Lateral inhibition occurs
that the structure of the organism is somehow encoded as a descriptive program in the when developing structures produce an
genome. Does the DNA contain a full description of the organism to which it will give inhibitor that prevents the formation of
rise: is it a blueprint for the organism? The answer is no. Instead, the genome con- any similar structures in the area adjacent
tains a program of instructions for making the organism—a generative program— to them, and the structures become evenly
spaced as a result.
that determines where and when different proteins are synthesized and thus controls
how cells behave.
A descriptive program, such as a blueprint or a plan, describes an object in some
detail, whereas a generative program describes how to make an object. For the same
object the programs are very different. Consider origami, the art of paper folding. By
30 Chapter 1 History and basic concepts

Asymmetric distribution of folding a piece of paper in various directions it is quite easy to make a paper hat or
cytoplasmic determinants a bird from a single sheet. To describe in any detail the final form of the paper with
the complex relationships between its parts is really very difficult, and not of much
help in explaining how to achieve it. Much more useful and easier to formulate are
instructions on how to fold the paper. The reason for this is that simple instructions
about folding have complex spatial consequences. In development, gene action simi-
larly sets in motion a sequence of events that can bring about profound changes in
the embryo. One can thus think of the genetic information in the fertilized egg as
equivalent to the folding instructions in origami; both contain a generative program
for making a particular structure.

1.19 The reliability of development is achieved by a variety of means

Cell division
Development is remarkably consistent and reliable; you need only look at the similar
lengths of your legs, each of which develops quite independently of the other for
some 15 years (see Chapter 13). Embryonic development thus needs to be reliable so
that the adult organism can function properly. For example, both wings of a bird must
be very similar in size and shape if it is to be able to fly satisfactorily. How such reli-
ability is achieved is an important problem in development.
Development needs to be reliable in the face of fluctuations that occur either
within the embryo or in the external environment. Internal fluctuations include small
changes in the concentrations of molecules, and also changes due, for example, to
mutations in genes not directly linked to the development of the organ in question.
Different daughter cells External factors that could perturb development include temperature and environ-
mental chemicals.
One of the central problems of reliability relates to pattern formation. How are cells
instructed to behave in a particular way in a particular position? Two ways in which
reliability is assured during development are apparent redundancy of mechanism and
negative feedback. Redundancy is when there are two or more ways of carrying out
a particular process; if one fails for any reason another will still function. It is like
having two batteries in your car instead of just one. True redundancy—such as hav-
ing two identical genes in a haploid genome with identical functions—is rare except
for cases such as rRNA, where there are often hundreds of copies of similar genes.
Fig. 1.28 Cell division with asymmetric
Apparent redundancy, on the other hand, where a given process can be specified by
distribution of cytoplasmic determinants. several different mechanisms, is probably one of the ways that embryos can achieve
If a particular molecule is distributed such precise and reliable results. It is like being able to draw a straight line with either
unevenly in the parent cell, cell division will a ruler or a piece of taut string. This type of situation is not true redundancy, but may
result in its being shared unequally between give the impression of being so if one mechanism is removed and the outcome is still
the cytoplasm of the two daughter cells. The apparently normal. Multiple ways of carrying out a process make it more robust and
more localized the cytoplasmic determinant resistant to environmental or genetic perturbation.
is in the parental cell, the more likely it will Negative feedback also has a role in ensuring consistency; here, the end product
be that one daughter cell will receive all of it of a process inhibits an earlier stage and thus keeps the level of product constant
and the other none, thus producing a distinct
(see, for example, Fig. 1.20). The classic example is in metabolism, where the end
difference between them.
product of a biochemical pathway inhibits an enzyme that acts early in the pathway.
Yet another reliability mechanism relates to the complexity of the networks of gene
activity that operate in development. There is evidence that these networks, which
involve multiple pathways, are robust and relatively insensitive to small changes in,
for example, the rates of the individual processes involved.

1.20 The complexity of embryonic development is due to the complexity of

cells themselves

Cells are, in a way, more complex than the embryo itself. In other words, the network
of interactions between proteins and DNA within any individual cell contains many
 A conceptual tool kit 31

more components and is very much more complex than the interactions between the
Stem cells
cells of the developing embryo. However clever you think cells are, they are almost
always far cleverer. Each of the basic cell activities involved in development, such as
S
cell division, response to cell signals, and cell movements, are the result of interac-
tions within a population of many different intracellular proteins whose composition S
varies over time and between different locations in the cell. Cell division, for example,
is a complex cell-biological program that takes place over a period of time, in a set S x
order of stages, and requires the construction and precise organization of specialized
S x
intracellular structures at mitosis.
Any given cell is expressing thousands of different genes at any given time. Much of S x
this gene expression may reflect an intrinsic program of activity, independent of exter-
nal signals. It is this complexity that determines how cells respond to the signals they x
receive; how a cell responds to a particular signal depends on its internal state. This
state can reflect the cell’s developmental history—cells have good memories—and
Fig. 1.29 Stem cells. Stem cells (S) are cells
so different cells can respond to the same signal in very different ways. We shall see
that both renew themselves and give rise to
many examples of the same signals being used over and over again by different cells
differentiated cell types. Thus, a daughter
at different stages of embryonic development, with different biological outcomes.
cell arising from stem-cell division may either
We have at present only a fragmentary picture of how all the genes and proteins in develop into another stem cell or give rise to
a cell, let alone a developing embryo, interact with one another. But new technologies a quite different type of cell (X).
are now making it possible to detect the simultaneous activity of hundreds of genes
in a given tissue. The discipline of systems biology is also beginning to develop tech-
niques to reconstruct the highly complex signaling networks used by cells. How to
interpret this information, and make biological sense of the patterns of gene activity
revealed, is a massive task for the future.

1.21 Development is intimately involved in evolution

The similarity in the developmental mechanisms and genes in such very different
animals as flies and vertebrates is the result of the evolutionary process. All animals
have evolved from a single multicellular ancestor and it is therefore inevitable that
some developmental mechanisms will be held in common by many different animals,
whereas new ones arose in different animal groups as evolution proceeded. The evo-
lution of development is discussed in more detail in Chapter 15.
The basic Darwinian theory of evolution by natural selection is that changes in
genes alter how an organism develops, and that changes in development in turn
determine how the adult will interact with their environment. If a developmental
change gives rise to adults better adapted to survive and reproduce in the prevailing
environment, it will be selected to be retained in the population. Thus changes in
development due to changes in the genes are fundamental to evolution.
Very clear examples of the workings of evolution in development can be seen in the
limbs of vertebrates. In Chapter 15 we shall see how analysis of fossil evidence shows
that the limbs of land vertebrates originally evolved from fins, and how the genetic
and developmental basis of this evolution is being reconstructed (see Section 15.7).
We shall also see how the basic pattern of the five-digit limb then evolved to give
limbs as different as those of bats, horses, and humans. In bats, the digits of the fore-
limb grow extremely long and support a leathery wing membrane, whereas in horses,
one digit of the ‘hand’ in the forelimb and the 'foot' in the hindlimb has become a
long bone that forms the lower part of the leg and the hoof, while some other digits
have been lost (see Fig. 15.14). The fossil record gives many other examples of evo-
lutionary changes.
While some major steps in evolution can be identified, the genes involved are not
easy to identify. There are also questions as to how big the genetic steps to new forms
were, and there are questions about how completely novel structures evolve. But
there is no doubt that it is all down to changes in embryonic development.
32 Chapter 1 History and basic concepts

Summary

Development results from the coordinated behavior of cells. The major processes
involved in development are pattern formation, morphogenesis or change in form, cell
differentiation, growth, cell migration, and cell death. These involve cell activities such
as intercellular signaling and turning genes on and off. Genes control cell behavior by
controlling where and when proteins are synthesized, and thus cell biology provides
the link between gene action and developmental processes. During development, cells
undergo changes in the genes they express, in their shape, in the signals they pro-
duce and respond to, in their rate of proliferation, and in their migratory behavior. All
these aspects of cell behavior are controlled largely by the presence of specific proteins;
gene activity controls which proteins are made. Since the somatic cells in the embryo all
generally contain the same genetic information, the changes that occur in development
are controlled by the differential activity of selected sets of genes in different groups
of cells. The genes’ control regions are fundamental to this process. Development is
progressive and the fate of cells becomes determined at different times. The potential
for development of cells in the early embryo is usually much greater than their normal
fate, but this potential becomes more restricted as development proceeds. Inductive
interactions, involving signals from one tissue or cell to another, are one of the main
ways of changing cell fate and directing development. Asymmetric cell divisions, in
which cytoplasmic components are unequally distributed to daughter cells, can also
make cells different. One widespread means of pattern generation is through positional
information; cells first acquire a positional value with respect to boundaries and then
interpret their positional values by behaving in different ways. Developmental signals
are more selective than instructive, choosing one or other of the developmental path-
ways open to the cell at that time. The embryo contains a generative, not a descriptive,
program—it is more like the instructions for making a structure by paper folding than a
blueprint. Various mechanisms, including apparent redundancy and negative feedback,
are involved in making development remarkably reliable. The complexity of develop-
ment lies within the cells. Evolution is intimately linked to development as heritable
changes in the adult organism must inevitably be the result of changes in the genetic
program of embryonic development.

Summary to Chapter 1

All the information for embryonic development is contained within the fertilized egg—the
diploid zygote. The genome of the zygote contains a program of instructions for making the
organism. In the working out of this developmental program, the cytoplasmic constituents of
the egg, and of the cells it gives rise to, are essential players along with the genes. Strictly
regulated gene activity, by controlling which proteins are synthesized, and where and when,
directs a sequence of cellular activities that brings about the profound changes that occur in
the embryo during development. The major processes involved in development are pattern
formation, morphogenesis, cell differentiation, growth, cell migration, and cell death, which
are controlled by communication between the cells of the embryo and changes in gene
expression. Development is progressive, with the fate of cells becoming more precisely
specified as it proceeds. Cells in the early embryo usually have a much greater potential for
development than is evident from their normal fate, and this enables embryos, particularly
vertebrate embryos, to develop normally even if cells are removed, added, or transplanted to
different positions. Cells’ developmental potential becomes much more restricted as devel-
opment proceeds. Many genes are involved in controlling the complex interactions that occur
during development, and reliability is achieved in various ways. Thousands of genes control
the development of animals and plants, and a full understanding of the processes involved
 A conceptual tool kit 33

is far from being achieved. While basic principles and certain developmental systems are
quite well understood, there are still many gaps. Changes in embryonic development are the
basis for the evolution of multicellular organisms.

n End of chapter questions c) Homo sapiens

d) Xenopus laevis
Long answer (concept questions)
2. Which of the following is a germ cell?
1. The ultimate goal of the science of developmental biology is to
understand human development. In your own words, discuss three a) cells of the gonads
areas of medicine that are impacted by the study of developmental b) intestinal cells
biology. c) sperm and eggs
2. The Greek philosopher Aristotle considered two opposing theo- d) zygotes
ries for development: preformation and epigenesis. Describe what is
meant by these two ideas; refer to the concept of the ‘homunculus’ in 3. Which statement best describes the relationship between geno-
describing preformation, and to the art of origami in describing epi- type and phenotype?
genesis. Which better describes our current view of development? a) ‘Genotype’ and ‘phenotype’ both describe the physical properties
of cells and organisms, just using different words or symbols.
3. Weismann’s concept of ‘determinants’ was supported by Roux’s
experiments demonstrating ‘mosaic development’ in the frog. b) Phenotype is entirely the consequence of a cell’s genotype,
How would determinants lead to mosaic development? Why does which is why we now recognize the principle of ‘genetic deter-
­Driesch’s concept of ‘regulation’ during development better explain minism.’
human twinning? c) The genes found in a cell are that cell’s genotype, and genes
directly control a cell’s phenotype.
4. Pattern formation is a central process in the study of develop-
ment. Describe three examples of pattern formation and explain d) The genotype determines what proteins a cell contains, and the
what ‘patterns’ are generated. proteins in turn determine the cell’s properties, or phenotype.

5. Briefly define the following cell behaviors: intercellular signaling, 4. Which would qualify as a dominant allele of a gene?
cell proliferation, cell differentiation, cell movement, changes in cell a) an allele that confers a phenotype on the organism when
shape, gene expression, and cell death. heterozygous; that is, when present in only one copy
6. Contrast the role of housekeeping genes and proteins during b) an allele that confers a phenotype only when both copies of the
development with the role of tissue-specific genes and proteins. gene are mutant; that is, when homozygous
Which class of genes and proteins is more important during devel- c) an allele that is commonly found in wild populations, as opposed
opment, housekeeping or tissue-specific? Give examples. to alleles found in laboratory strains or when a pathological con-
7. All cells in an organism contain the same genes; however, they dition is present
contain different types of cells. How is this paradox of ‘genetic equiv- d) an allele that confers a normal phenotype under some condi-
alence’ resolved by the concept of ‘differential gene expression’? tions, but confers a different, mutant phenotype when conditions
8. How do gene regulatory proteins, or transcription factors, work? such as elevated temperature are present
What is their relationship to control regions in DNA? 5. Which of the following basic developmental processes is most
9. Contrast cell fate with cell determination. What kinds of experi- dependent on cellular movements?
ments allow the distinction between fate and determination to be a) cell differentiation
made? b) growth
10. What are three ways in which inducing signals can be transmit- c) morphogenesis
ted in developing embryos? In what circumstance would an induc- d) pattern formation
ing signal be a ‘morphogen’?
6. Which aspect of cellular regulation operates after a protein has
11. What is the relationship between positional information and
been synthesized?
pattern formation? Give an example.
a) post-translational modification

Multiple choice (factual recall questions) b) RNA processing

NB There is only one right answer to each question. c) transcriptional control

d) translational control
1. Which of the following is a vertebrate ‘model organism’ used in
the study of development? 7. Which term could be used to refer to a segment of DNA?
a) Caenorhabditis elegans a) cis-regulatory module
b) Drosophila melanogaster b) differential gene expression
34 Chapter 1 History and basic concepts

c) gene regulatory protein Milestones in Development [http://www.nature.com/milestones/

d) transcription factor development/index.html] (date accessed 9 May 2010).
Needham, J.: A History of Embryology. Cambridge: Cambridge Uni-
8. A morphogen is versity Press, 1959.
a) a cell or group of cells that signal other cells to become deter- Sander, K.: ‘Mosaic work’ and ‘assimilating effects’ in embryo-
mined in a specific way genesis: Wilhelm Roux’s conclusions after disabling frog blas-
b) a gene that causes a cell or group of cells to adopt a particular tomeres. Roux’s Arch. Dev. Biol. 1991, 200: ­237–239.
shape Sander, K.: Shaking a concept: Hans Driesch and the varied fates
of sea urchin blastomeres. Roux’s Arch. Dev. Biol. 1992, 201:
c) a signaling molecule that causes the differentiation of neighbor-
265–267.
ing cells
d) a signaling molecule that confers concentration-dependent posi- A conceptual tool kit
tional information Alberts, B., et al.: Essential Cell Biology: An Introduction to the
Molecular Biology of the Cell. 5th edn. New York: Garland Science,
9. What distinguishes stem cells from other cell types?
2009.
a) Although all cell types can divide, only stem cells differentiate Howard, M.L., Davidson, E.H.: cis-Regulatory control circuits in
after division. development. Dev. Biol. 2004, 271: 109–118.
b) Stem cells are able to divide to produce populations of cells that Istrail, S., De-Leon, S.B., Davidson, E.H.: The regulatory genome
differentiate into a related set of cells. and the computer. Dev. Biol. 2007, 310: 187–195.
c) Stem cells divide asymmetrically to give rise to one daughter cell Jordan, J.D., Landau, E.M., Lyengar, R.: Signaling networks: the
that remains a stem cell, and a second daughter cell that differ- origins of cellular multitasking. Cell 2000, 103: 193–200.
entiates into one or more cell types. Levine, M., Tjian, R.: Transcriptional regulation and animal diver-
sity. Nature 2003, 424: 147–151.
d) Stem cells have a special type of cell division in which they form
Nelson, W.J.: Adaptation of core mechanisms to generate polarity.
small stems that then pinch off to give rise to the daughter cell.
Nature 2003, 422: 766–774.
10. How does organismal development evolve to give rise to new Papin, J.A., Hunter, T., Palsson, B.O., Subramanian, S.: Reconstruc-
forms and species? tion of cellular signalling networks and analysis of their prop-
a) A structure like the fin of a fish evolves into a structure like a erties. Nat. Rev. Mol. Biol. 2005, 6: 99–111.
human hand through the needs of the organism for a hand rather Volff, J.N. (Ed.): Vertebrate genomes. Genome Dyn 2006, 2: special
than a fin. issue.
Wolpert, L.: Do we understand development? Science 1994, 266:
b) Changes in genes that will be favored by natural selection will be
571–572.
induced, and thereby produce favorable changes in development.
Wolpert, L.: One hundred years of positional information. Trends
c) Changes in genes will alter an organism’s development, and if Genet. 1996, 12: ­359–364.
the new developmental patterns are favored by the environment, Xing, Y., Lee, C.: Relating alternative splicing to proteome com-
they will be retained through natural selection. plexity and genome evolution. Adv. Exp. Med. Biol. 2007, 623:
d) The characteristics acquired by the organism during its lifetime 36–49.
are passed on to the offspring; if the changes are favorable to the
Box 1E Signal transduction and intracellular signaling
organism, they will eventually come to be encoded in the genes
that govern development. Logan, C.Y., Nusse. R.: The Wnt signaling pathway in develop-
ment and disease. Annu. Rev. Cell Dev. Biol. 2004, 20: 781–810.
Multiple choice answer key Box 1F When development goes awry
1: d, 2: c, 3: d, 4: a, 5: c, 6: a, 7: a, 8: d, 9: c, 10: c. Cohen, M.M. Jr: Holoprosencephaly: clinical, anatomic, and
molecular dimensions. Birth Defects Res. A Clin. Mol. Teratol.
2006, 76: 658–673.
n Section further reading Geng, X., Oliver, G.: Pathogenesis of holoprosencephaly. J. Clin.
Invest. 2009, 119: ­1403–1413.
The origins of developmental biology
Roessler, E., Muenke, M.: How a Hedgehog might see holoprosen-
Hamburger, V.: The Heritage of Experimental Embryology: Hans Spe- cephaly. Hum. Mol. Genet. 2003, 12: R15–R25.
mann and the Organizer. New York: Oxford University Press, 1988.
Harris, H.: The Birth of the Cell. New Haven: Yale University Press,
1999.
 2
Development of the
Drosophila body plan
nn Drosophila life cycle and nn Patterning the early nn Segmentation genes and
overall development embryo compartments
nn Setting up the body axes nn Activation of the nn Specification of segment
nn Localization of maternal pair-rule genes and identity
determinants during the establishment of
oogenesis parasegments

The early development of the fruit fly Drosophila melanogaster is better understood
than that of any other animal of similar or greater morphological complexity. Its early
development exemplifies clearly many of the principles introduced in Chapter 1. The
genetic basis of development is particularly well understood in Drosophila, and many
developmental genes in other organisms, particularly vertebrates, have been identified
by their homology with Drosophila genes. In this chapter we look at the crucial role of
maternal gene products in the earliest development of the zygote and the early syn-
cytial blastoderm stage, and how they specify the main axes and regions of the body.
We shall see how gradients of maternal morphogens specify the antero-posterior and
dorso-ventral pattern of the early embryo, and how cis-acting regulatory elements
then control the precise spatial expression of the zygote’s own genes to divide the
embryo into segments. Another set of genes acting along the antero-posterior axis
then confers on each segment its unique character, as displayed by the appendages
the segment eventually bears in the adult, such as wings, legs, or antennae.

We are much more like flies in our development than you might think. Astonishing
discoveries in developmental biology over the past 25 years have revealed that many of
the genes that control the development of the fruit fly Drosophila are similar to those
controlling development in vertebrates, and indeed in many other animals. It seems that
once evolution finds a satisfactory way of patterning animal bodies, it tends to use the
same mechanisms and molecules over and over again with, of course, some important
modifications. While insect and vertebrate development might seem to be very different,
much has been learned from Drosophila that has illuminated vertebrate development.
From the genome sequence there are an estimated 14,000 protein-coding genes
in Drosophila, which is only twice the number in unicellular yeast and fewer than
the 19,000 genes estimated of the morphologically simpler nematode Caenorhab-
ditis. However, a recent large-scale analysis of the RNAs transcribed both during
36 Chapter 2 Development of the Drosophila body plan

­ evelopment and in adult Drosophila revealed a high incidence of alternative RNA

d
splicing, which would increase the number of different proteins that could be made.
In addition, around 1100 genes encoding functional RNAs, such as tRNAs and
­microRNAs, are present in the fly genome.
The pre-eminent place of Drosophila in modern developmental biology was rec-
ognized by the award of the 1995 Nobel Prize for Physiology or Medicine for work
that led to a fundamental understanding of how genes control development in the fly
embryo. This was only the second time that the Nobel Prize had been awarded for
work in developmental biology. While insect and vertebrate development may seem
to be very different, much has been learned that can be applied to vertebrate develop-
ment; for example, many intercellular signaling pathways are well conserved.
The first part of this chapter describes the life cycle of Drosophila and its overall
development. The remaining parts look at how the basic body plan of the Droso-
phila larva is established up to the stage at which the embryo becomes segmented,
and how the segments are patterned and acquire their unique identities. More about
Drosophila gastrulation, germ-cell development and sex-determination mechanisms,
adult organ development, neural development, and growth and metamorphosis will
be found in Chapters 8, 9, 11, 12, and 13, respectively.

Drosophila life cycle and overall development

The fruit fly Drosophila melanogaster is a small dipteran insect, about 3 mm long as
an adult, which undergoes embryonic development inside an egg and hatches from
the egg as a larva. This then goes through two more larval stages, growing bigger each
time, and eventually becomes a pupa, in which metamorphosis into the adult occurs.
The life cycle of Drosophila is shown in Fig. 2.1.

2.1 The early Drosophila embryo is a multinucleate syncytium

The Drosophila egg is rather oblong and the future anterior end is easily recognizable
by the micropyle, a nipple-shaped structure in the tough external coat surrounding
the egg. Sperm enter the anterior end of the egg through the micropyle. After fertiliza-
tion and fusion of the sperm and egg nuclei, the zygote nucleus undergoes a series
of rapid mitotic divisions, one about every 9 minutes, but, unlike in most embryos,
there is initially no cleavage of the cytoplasm and no formation of cell membranes
to separate the nuclei. The result after 12 nuclear divisions is a syncytium in which
around 6000 nuclei are present in a common cytoplasm (Fig. 2.2); the embryo essen-
tially remains a single cell during its early development. After nine divisions the nuclei
move to the periphery to form the syncytial blastoderm. This comprises a superficial
layer of nuclei and cytoplasm, and surrounds a central mass of yolky cytoplasm. The
syncytial blastoderm is equivalent to the blastula stage in embryos such as Xenopus
(see, for example, Box 1A, pp. 2–3). Shortly afterwards, membranes grow in from the
surface to enclose the nuclei and form cells, and the blastoderm becomes truly cel-
lular after 14 mitoses. Because of the formation of a syncytium, even large molecules
such as proteins can diffuse between nuclei during the first 3 hours of development
and, as we shall see, this is of great importance to early Drosophila development.
At the syncytial stage, a small number of nuclei move to the posterior end of the
embryo and become surrounded by cell membranes to form the pole cells, which end
up on the outer surface of the blastoderm (see Fig. 2.2). The pole cells eventually give
rise to the germ cells from which the gametes—sperm and eggs—develop in the adult
insect, while the blastoderm gives rise to the somatic cells of the embryo. This setting
aside of the germline from the rest of the embryo at a very early stage is a common
strategy in animal development.
 Drosophila life cycle and overall development 37

micropyle

Adult fly
Fertilized egg

Cleavage
Metamorphosis
Syncytial
blastoderm
0
9
60
Pupa Minutes stage 3–4
after 120
Days fertilization*
after
fertilization* 180 Gastrulation
Hours
after
fertilization*
10
Embryo
Larva
24
stage 13
third instar
Hatching

second instar

first instar

*At 25oC incubation

Fig 2.1 Life cycle of Drosophila melanogaster. After cleavage and The dorsal filaments are extraembryonic structures; a Drosophila,
gastrulation the embryo becomes segmented and hatches out as a second-instar larva (middle); and a Drosophila pupa (bottom). Scale
feeding larva. The larva grows and goes through two molts (instars), bars = 0.1 mm.
eventually forming a pupa that will metamorphose into the adult fly. Photographs courtesy of F. R. Turner (top, from Turner, F.R., et al.:
The photographs show scanning electron micrographs of: a Drosophila 1976; middle, from Turner, F.R., et al.: 1979).
egg before fertilization (top). The sperm enters through the micropyle.

30 minutes 70 minutes 90 minutes 2 hours 3 hours

Fusion of sperm Nuclear division Nuclei migrate to Syncytial Cellular
and egg nuclei creating syncytium periphery of cytoplasm blastoderm blastoderm

nucleus micropyle pole cells

Yolky cytoplasm

Fig 2.2 Cleavage of the Drosophila embryo. After fusion of the are delayed. After about 3 hours, cell membranes develop, giving rise
sperm and egg nuclei there is rapid nuclear division but no cell to the cellular blastoderm. About 15 pole cells, which will give rise to
membranes surround the nucleus, the result being a syncytium of germ cells, form a separate group at the posterior end of the embryo.
many nuclei in a common cytoplasm. After the ninth division the nuclei Times given are for incubation at 25°C.
move to the periphery to form the syncytial blastoderm, but some

In this chapter we will be concerned with the development of the embryo to the
stage at which it hatches out of the egg as a larva. Embryonic development naturally
takes place inside the opaque egg case, and to study it embryos are freed from the
egg case by treating with bleach and then placed in a special oil for observation.
38 Chapter 2 Development of the Drosophila body plan

D
This can be done at any stage and they will continue to develop and even hatch. The
A P embryo is about 0.5 mm long.

2.2 Cellularization is followed by gastrulation and segmentation

All the future tissues, except for the germline cells, are derived from the single epithe-
V
lial layer of the cellular blastoderm. For example, prospective mesoderm (see Box 1C,
p. 15) is located in the most ventral region, whereas the future midgut derives from
two regions of prospective endoderm, one at the anterior and the other at the pos-
terior end of the embryo. Endodermal and mesodermal tissues move to their future
positions inside the embryo during gastrulation, leaving ectoderm as the outer layer
(Fig. 2.3). Gastrulation starts at about 3 hours after fertilization, when the future
mesoderm in the ventral region invaginates to form a furrow along the ventral mid-
line. The mesodermal cells are initially internalized by formation of a tube of meso-
derm, which will be described in more detail in Chapter 8. The mesoderm cells then
separate from the surface layer of the tube and migrate under the ectoderm to internal
locations, where they later give rise to muscle and other connective tissues.
In insects, as in all arthropods, the main nerve cord lies ventrally, rather than
dorsally as in vertebrates. Shortly after the mesoderm has invaginated, ectodermal
cells of the ventral region, which will give rise to the nervous system, leave the sur-
face individually and form a layer of neuroblasts between the mesoderm and the
outer ectoderm. At the same time, two tube-like invaginations develop at the sites of
the future anterior and posterior midgut. These grow inward and eventually fuse to
form the endoderm of the midgut, while ectoderm is dragged inward behind them
at each end to form the foregut and the hindgut. The outer ectoderm layer develops
into the epidermis. There are no cell divisions during gastrulation, but once it is
completed, cells start to divide again. The cells of the epidermis only divide twice
more before they secrete a thin cuticle composed largely of protein and the polysac-
charide chitin.
Also during gastrulation, the ventral blastoderm or germ band, which comprises
the main trunk region, undergoes germ-band extension, which drives the posterior
trunk regions round the posterior end and onto what was the dorsal side (Fig. 2.4),
as described in Chapter 8. The germ band later retracts as embryonic development is
mesoderm nervous system completed. At the time of germ-band extension, the first external signs of segmentation
amnioserosa epidermis
can be seen. A series of evenly spaced grooves form more or less simultaneously and
gut germline
these demarcate parasegments, which later give rise to the segments of the larva and
adult. Parasegments and segments are out of register, so that a segment is formed by
Fig 2.3 Gastrulation in Drosophila.
cooperative development of the posterior region of one parasegment and the anterior
Gastrulation begins when the future
mesoderm invaginates in the ventral
region of the next. There are 14 parasegments: three contribute to mouthparts of the
region, first forming a furrow and then head, three to the thoracic region, and eight to the abdomen.
an internalized tube. The cells then leave
the tube and migrate internally under the
2.3 After hatching, the Drosophila larva develops through several larval
ectoderm. The nervous system comes from
stages, pupates, and then undergoes metamorphosis to become an adult
cells that leave the surface of the ventral
blastoderm and form a layer between the The larva (Fig. 2.5) hatches about 24 hours after fertilization, but the different regions
ventral ectoderm and mesoderm. The gut of the larval body are well-defined several hours before that. The head is a complex
forms from two invaginations at the anterior structure, largely hidden from view before the larva hatches. A specialized structure
and posterior end that fuse in the middle.
associated with the most anterior region of the head is the acron. Another specialized
The midgut region is derived from endoderm,
structure, called the telson, marks the posterior end of the larva. Between the head
whereas the foregut and hindgut are of
ectodermal origin. Regions striped blue and
and telson, three thoracic segments and eight abdominal segments can be distin-
gray represent tissue that will give rise to guished by specializations in the cuticle. On the ventral side of each segment are belts
both nervous system and epidermis. The of small tooth-like outgrowths called denticles, and other cuticular structures charac-
amnioserosa, an extra-embryonic membrane, teristic of each segment. As the larva feeds and grows, it molts, shedding its cuticle.
is discussed in Section 2.5. This process occurs twice, each stage being called an instar. After the third instar,
 Drosophila life cycle and overall development 39

Ventral view, early gastrula Lateral view, germ-band extension Lateral view, segmentation

Anterior Posterior Dorsal

germ band
ventral furrow

Ventral
mouthparts thorax abdomen

Fig 2.4 Gastrulation, germ-band extension,

the larva becomes a pupa, inside which metamorphosis into the adult fly occurs and and segmentation in the Drosophila
there is a major transformation of overall form. embryo. Gastrulation involves the future
The Drosophila larva has neither wings nor legs; these and other organs are formed mesoderm moving inside through the ventral
when the larva undergoes hormone-induced metamorphosis in the pupal stage. These furrow. During gastrulation, the ventral
structures are, however, already present in the larva as imaginal discs, small sheets of blastoderm (the germ band) extends, driving
prospective epidermal cells derived from the cellular blastoderm and usually contain- the posterior trunk region onto the dorsal
ing about 40 cells each at the time they are formed. These discs grow by cell prolifera- side and segmentation now takes place. Later
tion throughout larval life and form sacs of epithelia that fold to accommodate their the germ band shortens. Scale bar = 0.1 mm.

increase in size. There are imaginal discs for each of the six legs, two wings, and the Photographs courtesy of F.R. Turner (left
two halteres (balancing organs), and for the genital apparatus, eyes, antennae, and from Turner, F.R., et al.: 1977; middle from
Alberts, B., et al.: 1994).
other adult head structures (Fig. 2.6). At metamorphosis, these develop into the adult
organs. We will discuss the development of the imaginal discs in Chapter 11; they
provide continuity between the pattern of the larval body and that of the adult, even
though metamorphosis intervenes. A3 A4 A5
A1 A2 A6 A7
head A8

2.4 Many developmental genes have been identified in Drosophila through

induced large-scale genetic screening T1 T2 T3

Valuable though spontaneous mutations have been in the study of development, suit-
able developmentally informative mutations are rare. Many more developmental genes
have been identified by inducing random mutations in a large number of organisms Fig 2.5 Ventral view of a Drosophila
larva. T1 to T3 are the thoracic segments
by chemical treatments or irradiation with X-rays, and then screening for mutants of
and A1 to A8, the abdominal segments. The
developmental interest. The overall aim is to treat a large enough population so that, in
characteristic pattern of denticles can be
seen in the anterior region of each abdominal
segment. Scale bar = 0.1 mm.
Larva Metamorphosis Adult fly
Photograph courtesy of F.R. Turner.
salivary Imaginal discs for:
glands
mouthparts

antenna
eye

leg

Fig 2.6 Imaginal discs give rise to adult

haltere
structures at metamorphosis. The imaginal
discs in the Drosophila larva are small sheets
wing of epithelial cells; at metamorphosis they
give rise to a variety of adult structures.
genitalia The abdominal cuticle comes from groups of
tissue-forming cells (histoblasts) located in
each larval abdominal segment.
40 Chapter 2 Development of the Drosophila body plan

total, a mutation is induced in every gene in the genome. This sort of approach is best
used in organisms that breed rapidly and can be obtained and treated conveniently in
very large numbers.
Many of the developmental mutations that have led to our present understanding
of early Drosophila development came from a brilliantly successful screening program
that searched the Drosophila genome systematically for mutations affecting the pat-
terning of the early embryo. Its success was recognized with the award of a Nobel
Prize for Physiology or Medicine to Edward Lewis, Christiane Nüsslein-Volhard, and
Eric Wieschaus in 1995.
In this screening program thousands of flies were treated with a chemical mutagen,
and were bred and screened according to the strategy described in Box 2A (p. 42).
Given the number of progeny involved, it was important to devise a strategy that
would reduce the number of flies that had to be examined to find a mutation. So the
search was restricted to mutations on just one chromosome at a time. As described in
Box 2A, the program also incorporated a means of identifying flies homozygous for
the male-derived chromosomes carrying the induced mutations and, most important,
also included a way of automatically removing from the population flies that could
not be carrying a mutated chromosome.
A phenotypic character of particular value in screening for patterning mutants in Droso-
phila is the stereotyped pattern of small outgrowths, or denticles, on the larval segments,
as irregularities in these patterns enable mutants to be quickly spotted. In this way, the
key genes involved in patterning the early Drosophila embryo were first identified.
One type of mutation that affects Drosophila embryonic development needs a some-
what different type of screen. Maternal-effect mutations are mutations that when
expressed in the mother affect the embryo’s development. They are identified as
mutations that, when present in the mother, do not affect her appearance or normal
physiology but have effects on the development of her progeny. As we shall see later
in the chapter, some of these maternal genes exert their effects by being expressed
in the follicle cells of the ovary, which form a bag that contains the germline-derived
oocyte (the immature egg) and nurse cells. Others are expressed in the nurse cells
or in the oocyte, producing developmentally important mRNAs and proteins that are
deposited in the oocyte in a particular spatial pattern. This initial patterning process
is crucial for correct development.

Setting up the body axes

2.5 The body axes are set up while the Drosophila embryo is still a syncytium

The insect body, like that of vertebrates and most of the other organisms described
in this book, is bilaterally symmetrical. Like all animals with bilateral symmetry, the
Drosophila larva has two distinct and largely independent axes: the antero-posterior
and dorso-ventral axes, which are at right angles to each other. These axes are already
partly set up in the Drosophila egg, and become fully established and patterned in
the very early embryo, while it is still in the syncytial blastoderm stage. Along the
antero-posterior axis the embryo becomes divided into several broad regions, which
will become the head, thorax, and abdomen of the larva (Fig. 2.7). The thorax and
abdomen become divided into segments as the embryo develops, while two regions
of endoderm at each end of the embryo invaginate at gastrulation to form the gut (see
Fig. 2.3). Each segment, and the head, has its own unique character in the larva, as
revealed by both its external cuticular structures and its internal organization.
The dorso-ventral axis of the embryo becomes divided up into four regions early in
embryogenesis: from ventral to dorsal these are the mesoderm, which will form mus-
cles and other internal connective tissues; the neuroectoderm, which gives rise to the
larval nervous system; the dorsal ectoderm, which gives rise to the larval epidermis;
 Setting up the body axes 41

Fig 2.7 Patterning of the Drosophila

Egg Dorsal
embryo. The body plan is patterned along
two distinct axes. The antero-posterior and
Anterior Posterior dorso-ventral axes are at right angles to
each other and are laid down in the egg. In
the early embryo, the dorso-ventral axis is
divided into four regions: mesoderm (red),
Ventral ventral ectoderm (yellow), dorsal ectoderm
(orange), and amnioserosa (an extra-
Embryo amnioserosa parasegments embryonic membrane; green). The ventral
dorsal ectoderm visualized by regions
of specific gene activity ectoderm gives rise to both ventral epidermis
and neural tissue, the dorsal ectoderm to
epidermis. The antero-posterior axis becomes
Head Abdomen divided into different regions that later give
rise to the head, thorax, and abdomen. After
prospective posterior the initial division into broad body regions,
gut endoderm segmentation begins. The future segments
prospective
anterior gut Thorax ventral ectoderm can be visualized as transverse stripes by
endoderm staining for specific gene activity; these
mesoderm
stripes demarcate 14 parasegments, 10 of
First instar larva which are marked. The embryo develops
into a segmented larva. By the time the
larva hatches, the 14 parasegments have
acron been converted into thoracic (T1–T3) and
telson abdominal (A1–A8) segments, with each
segment being made up of the posterior
A8 half of one parasegment and the anterior
T1 T2 T3 A7 half of the next. Different segments are
A1 A5 A6
A2 A3 A4 distinguished by the patterns of bristles
Head Tail and denticles on the cuticle. Specialized
Thoracic segments Abdominal segments structures, the acron and telson, develop
at the head and tail ends, respectively.

and the amnioserosa, which gives rise to an extra-embryonic membrane on the dorsal
side of the embryo (see Fig. 2.7). Organization along the antero-posterior and dorso-
ventral axes of the early embryo develops more or less simultaneously, but is specified
by independent mechanisms and by different sets of genes in each axis.
The early development of Drosophila is peculiar to certain types of insects, as pat-
terning occurs within a multinucleate syncytial blastoderm (see Fig. 2.2). Only after the
beginning of segmentation does the embryo become truly multicellular. At the syncytial
stage, many proteins, including those that are not normally secreted from cells such as
transcription factors, can diffuse throughout the blastoderm and enter other nuclei. Con-
centration gradients of transcription factors that provide positional information for the
nuclei to interpret (see Section 1.15) can thus be set up in the syncytial blastoderm.
Early development is essentially two-dimensional as patterning occurs mainly in the
blastoderm, which is essentially a single layer of nuclei, and later cells. But the larva
itself is a three-dimensional object, with internal structures organized along the axis.
This third dimension develops later, at gastrulation (discussed in Chapter 8), when
parts of the surface layer move into the interior to form the gut, the mesodermal struc-
tures that will give rise to muscle, and the ectodermally derived nervous system.

2.6 Maternal factors set up the body axes and direct the early stage of
Drosophila development

The earliest stage of Drosophila embryonic development is guided by maternal ­factors,

mRNAs, and proteins that are synthesized and laid down in the egg by the mother (see
Section 2.4). In all, about 50 maternal genes are involved in setting up the two axes and
42 Chapter 2 Development of the Drosophila body plan

Box 2A Mutagenesis and genetic screening strategy for identifying developmental mutants in Drosophila

The mutagen ethyl methane sulfonate

(EMS) was applied to large numbers of male Male treated
with mutagen
flies homozygous for a recessive mutation (EMS) a DTS Viable below 29oC
on the selected chromosome. The chromo- a b
some marked in this way is designated
‘a’ in the figure. A recessive mutation is
chosen that gives adult flies an easily dis-
a* a* a*
tinguishable but viable phenotype when
homozygous (see Fig. 1.12). Here, we
Approximately 5000
shall use the example of a mutation that individual males outcrossed
a* DTS
gives white eyes in the homozygous state b b (number required for saturation
of chromosome with mutations)
(white -), instead of the normal red eyes of
the wild type.
The treated males, who now produce
sperm with a variety of induced muta-
tions on the ‘a’ chromosomes (a*), were
b a* a* a*
crossed with untreated females that car- F1 DTS
DTS
b b b b
ried different mutations (DTS and b) on
their two ‘a’ chromosomes, but were other-
Embryonic
wise wild type. These mutations track the lethal Dead at 29oC
untreated female-derived chromosomes
and automatically eliminate all embryos
a* a* b
carrying two female-derived chromosomes F2 a* b b
in subsequent generations. DTS is a domi-
nant temperature-sensitive mutation that Embryonic lethal Heterozygous Embryonic
with a pattern defect? viable stock lethal
causes death of the fly when the incuba-
tion temperature is raised to 29°C. b is a
non-developmental lethal recessive mutation, so that any flies chromosome a in males); heterozygous a* flies; and homozygous
homozygous for this female-derived chromosome will die as b flies (which die as embryos).
normal-looking embryos and be automatically eliminated. The If a* is indeed a patterning mutation that is lethal in the larva,
female flies also carried a balancer chromosome (not shown) then the culture tube in which the cross is made will contain no
that prevented recombination at meiosis; this is to prevent adult flies with the original male phenotype—white eyes, say.
recombination between male-derived and female-derived chro- Therefore tubes containing white-eyed flies can be discarded
mosomes in females. There is no recombination at meiosis in immediately, as the induced mutation a* they carry must have
male Drosophila. allowed them to develop to adulthood even when homozygous.
To identify the new recessive mutations (a*) caused by the If no white-eyed flies are present in a tube, then the homozygous
chemical treatment, a large number of the heterozygous males a*/a* embryos may have died or been arrested in their develop-
arising from this first cross were again outcrossed to DTS/b ment as a result of abnormal development, and the mutation is
females. Of the offspring of each individual cross, only a*/b flies of potential interest. The embryos of larvae from this cross can
survive when placed at 29°C; all other combinations die. The sur- then be examined for pattern defects, as in the example illus-
viving siblings were then intercrossed and the offspring of each trated here. The phenotypically wild-type adults in this tube are
cross were screened for patterning mutants. There are three pos- heterozygous for a* and are used as breeding stock to study the
sible outcomes: flies homozygous for the induced mutation a* mutant further. This whole program has to be repeated for each
(which will also be homozygous for the original mutation marking of the four chromosome pairs of Drosophila.

a basic framework of positional information, which is then interpreted by the embryo’s

own genetic program. In contrast to the maternal-effect genes, those genes expressed
in the embryo’s own nuclei during development are known as zygotic genes. All later
patterning, which involves expression of the zygotic genes, is built on the framework
of maternal gene products (Fig. 2.8). After development begins, the maternal RNAs
 Setting up the body axes 43

Maternal genes Gene Name of

Examples of regions of gene activity
group example

bicoid (bcd)

Zygotic genes
Gap
genes
hunchback (hb)

Fig 2.8 The sequential expression of

different sets of genes establishes the
Pair-rule body plan along the antero-posterior
genes axis. After fertilization, maternal gene
even-skipped (eve)
& fushi tarazu (ftz) products laid down in the egg, such as
bicoid mRNA, are translated. They provide
positional information which activates
the zygotic genes. The four main classes
of zygotic genes acting along the antero-
Segmentation
genes posterior axis are the gap genes, the pair-rule
genes, the segmentation genes, and the
wingless (wg)
selector, or homeotic, genes. The gap genes
define regional differences that result in
the expression of a periodic pattern of
gene activity by the pair-rule genes, which
Selector define the parasegments and foreshadow
genes segmentation. The segmentation genes
abdominal-A elaborate the pattern in the segments,
(abd-A)
and segment identity is determined by the
selector genes. The functions of each of
these classes of genes are discussed in this
chapter.

are translated, and the resulting proteins act on the embryo’s nuclei to activate zygotic
genes in a specific spatial pattern along each axis, thus setting the scene for the next
round of patterning.
Drosophila illustrates particularly well a general principle of development. The
embryo is patterned in a series of steps. Broad regional differences are established
first, and these are then refined to produce a larger number of smaller developmental
domains, each characterized by a unique profile of gene activity. Developmental genes
act in a strict temporal sequence. They form a hierarchy of gene activity in which the
action of one set of genes is essential for another set of genes to be activated, and thus
for the next stage of development to occur.

2.7 Three classes of maternal genes specify the antero-posterior axis

We will first look at how maternal gene products specify the antero-posterior axis of
the embryo. The expression of maternal genes during egg formation in the mother
creates differences in the egg along the antero-posterior axis even before it is fer-
tilized. These differences already distinguish the future anterior and posterior ends
of the adult. The roles of the maternal genes can be deduced from the effects of
44 Chapter 2 Development of the Drosophila body plan

Wild-type egg fate map Wild-type larva

Anterior Posterior

acron telson
head abdomen telson
thorax

Mutant eggs Mutant embryos Regions affected

bicoid mutant

Fig 2.9 The effects of mutations in

Anterior
the maternal gene system. Mutations
in maternal genes lead to deletions and
abnormalities in anterior, posterior, or
terminal structures. The wild-type fate map nanos mutant
shows which regions of the egg give rise to
particular regions and structures in the larva. Posterior
Regions that are affected in mutant eggs,
and which lead to lost or altered structures
in the larva, are shaded in red. In bicoid
mutants there is a partial loss of anterior torso mutant
structures and the appearance of a posterior
structure—the telson—at the anterior end. Terminal
nanos mutants lack a large part of the
posterior region. torso mutants lack both
acron and telson.

­ aternal-effect mutations (see Section 2.4) on the embryo. They fall into three classes:
m
those that affect anterior regions; those that affect posterior regions; and those that
affect both of the terminal regions (Fig. 2.9). Mutations of genes in the anterior class,
such as bicoid, lead to a reduction or loss of head and thoracic structures, and in
some cases their replacement with posterior structures. Posterior-group mutations,
such as nanos, cause the loss of abdominal regions, leading to a smaller than normal
larva, while those of the terminal class, such as torso, affect the acron and telson.
The idiosyncratic naming of genes in Drosophila usually reflects the attempts by the
discoverers to describe the mutant phenotype—nanos is Greek for dwarf, while torso
reflects the fact that both ends of the embryo are missing. In this chapter we meet
quite a number of gene names; all these are listed, together with their functions where
known, in the table at the end of this chapter (p. 87).

2.8 Bicoid protein provides an antero-posterior gradient of a morphogen

Maternal bicoid mRNA is localized at the anterior end of the unfertilized egg dur-
ing oogenesis. After fertilization it is translated, and the conventional view was that
Bicoid protein diffuses from the anterior end and forms a concentration gradient along
the antero-posterior axis. Very recent evidence, together with some older studies,
suggests, on the contrary, that the Bicoid-protein gradient is preceded by a gradient
in bicoid mRNA in the egg cortex, with the bicoid mRNA being proposed to be trans-
ported along cortical microtubules (Fig. 2.10). Translation of the mRNA gives rise to
the gradient in Bicoid protein, which provides the positional information required for
further patterning along this axis. Historically, the Bicoid gradient provided the first
concrete evidence for the existence of the morphogen gradients that had been postu-
lated to control pattern formation (see Section 1.15).
 Setting up the body axes 45

The role of the bicoid gene was first elucidated by a combination of genetic and
physical experiments on the Drosophila embryo. Female flies that do not express bicoid
produce embryos that have no proper head and thorax (see Fig. 2.9). In a separate line
of investigation into the role of localized cytoplasmic factors in anterior development,
normal eggs were pricked at the anterior ends and some cytoplasm allowed to leak out.
The embryos that developed bore a striking resemblance to bicoid mutant embryos.
This suggested that normal eggs have some factor(s) in the cytoplasm at their anterior
end that is absent in bicoid mutant eggs. This was confirmed by showing that anterior
cytoplasm from wild-type embryos could partially ‘rescue’ the bicoid mutant embryos—
in the sense they developed more normally—if it was injected into their anterior regions
(Fig. 2.11). Moreover, if normal anterior cytoplasm is injected into the middle of a
fertilized bicoid mutant egg, head structures developed at the site of injection and the
adjacent segments became thoracic segments, setting up a mirror-image body pattern at
the site of injection. The simplest interpretation of these experiments is that the bicoid
gene is necessary for the establishment of the anterior structures because it establishes
a gradient of Bicoid protein, whose source and highest level is at the anterior end.

A P

Wild-type egg Wild-type larva

bicoid mutant egg bicoid mutant larva

Fig 2.10 bicoid RNA forms a concentration
gradient in the early embryo. Panels a
to c show the formation of the gradient of
bicoid mRNA in Drosophila embryos (dorsal
side up and anterior to the left). bcd mRNA
transcripts (yellow) in the cortex are detected
by fluorescence in situ hybridization with a
Wild-type egg bicoid mutant egg Some anterior structures develop
labeled bcd cDNA in (a) an unfertilized egg,
and in embryos during interphase of (b)
nuclear cycle 9, and (c) after onset of nuclear
cycle 14. The corresponding formation of the
Bicoid protein gradient can be seen in the
nuclear cycle 14 embryo in (d), in which bcd
Wild-type egg bicoid mutant egg
mRNA is stained green and the Bcd protein
Bicoid Bicoid Thoracic Thoracic (inside nuclei) in red.
protein protein Head
segments segments Photographs reproduced from Spirov, A. et al.:
2009.
A P A P

Fig 2.11 The bicoid gene is necessary for the development of anterior structures. Embryos
whose mothers lack the bicoid gene lack anterior regions (second row). Transfer of anterior
cytoplasm from wild-type embryos to bicoid mutant embryos causes some anterior structure to
develop at the site of injection (third row). If wild-type anterior cytoplasm is transplanted to the
middle of a bicoid mutant egg or early embryo, head structures develop at the site of injection,
flanked on both sides by thoracic-type segments (fourth row). These results can be interpreted
in terms of the anterior cytoplasm setting up a gradient of Bicoid protein with the high point at
the site of injection (see graphs, bottom left panel). A, anterior; P, posterior.
46 Chapter 2 Development of the Drosophila body plan

bicoid mRNA was shown to be tightly localized to the anterior-most region of the
unfertilized egg by the technique of in situ hybridization (see Box 1D, p. 20). Stain-
ing with an antibody against Bicoid protein then showed that this is absent from the
unfertilized egg, but that after fertilization, bicoid mRNA is translated into protein.
After fertilization, bicoid mRNA diffuses from the anterior to form an anterior to
posterior gradient of RNA which is translated into an anterior to posterior gradient of
Bicoid protein. Bicoid is a transcription factor and so must enter the embryo’s nuclei
to carry out its function of switching on zygotic genes. By the time the syncytial blas-
toderm is formed, there is a clear gradient of intranuclear Bicoid protein concentration
along the antero-posterior axis, with its high point at the anterior end (see Fig. 2.10).
In an unprecedented technical feat, the dynamics of Bicoid-gradient ­formation have
recently been observed in living embryos by measuring the intranuclear concentra-
tions of Bicoid protein genetically tagged with green fluorescent protein (Bicoid–GFP).
The measurements indicate that the concentration of Bicoid protein inside a nucleus at
a given position along the gradient is somehow maintained at a constant level, despite
the increase in the number of nuclei at each mitotic cycle, and the outflow of Bicoid
into the cytoplasm from nuclei when nuclear membranes break down at mitosis.
Bicoid protein acts as a morphogen, as described in more detail later in the chapter.
It switches on particular zygotic genes at different threshold concentrations, thereby
initiating a new pattern of gene expression along the axis (the concept of thresholds
is discussed in Section 1.15). Thus, bicoid is a key maternal gene in early Drosophila
development. The other maternal genes of the anterior group are mainly involved in
the localization of bicoid mRNA to the anterior end of the egg during oogenesis and
in the control of setting up its mRNA gradient and translation after fertilization. Given
the crucial importance of bicoid in Drosophila development, it is worth noting that the
bicoid gene is only present in a small group of the most recently evolved dipterans—
such as fruit flies and blowflies. Later in this chapter we shall briefly discuss different
mechanisms of early development that occur in some other groups of insects. It is
not surprising that many different developmental mechanisms have evolved in such a
large and diverse group of animals as the insects.

2.9 The posterior pattern is controlled by the gradients

of Nanos and Caudal proteins

For proper patterning along an axis, both ends need to be specified, and Bicoid defines
only the anterior end of the antero-posterior axis. The posterior end is specified by the
actions of at least nine maternal genes—the posterior-group genes. Mutations in the
posterior-group genes result in larvae that are shorter than normal because there is no
abdomen (see Fig. 2.9). One function of the products of maternal posterior-group genes
(for example, oskar) is to is to localize nanos mRNA at the extreme posterior pole of
the unfertilized egg. Another action is to assemble the posterior germplasm in the egg.
This is cytoplasm that contains the so-called germline factors; in the embryo it is incor-
porated into the pole cells (see Section 2.1) and will give rise to eggs or sperm.
Like bicoid mRNA, nanos mRNA is translated only after fertilization and produces
a concentration gradient of Nanos protein, in this case with the highest level at the
posterior end of the embryo. But Nanos does not act directly as a morphogen to
specify the abdominal pattern; it has a quite different role. Its function is to suppress
translation of the maternal mRNA of another gene, hunchback, in the posterior region
of the embryo. Maternal hunchback mRNA is distributed throughout the embryo and
starts to be translated after fertilization. However, a little later, the Bicoid protein
activates expression of the embryo’s own hunchback genes in the anterior half of the
embryo. Restriction of Hunchback protein to the anterior region is crucial for correct
patterning along the antero-posterior axis. To avoid interference with this patterning
by maternal Hunchback protein in the posterior region, its translation is prevented
 Setting up the body axes 47

Maternal mRNA expression Protein expression

Fig 2.12 Establishment of a maternal
gradient in Hunchback protein. Left panel:
in the unfertilized egg, maternal hunchback
mRNA (turquoise) is present at a relatively
low level throughout the egg, whereas
nanos mRNA (yellow) is located posteriorly.
The photograph is an in situ hybridization
showing the location of nanos mRNA (black).
Right panel: after fertilization, nanos mRNA
is translated and Nanos protein blocks
translation of hunchback mRNA in the
posterior regions, giving rise to a shallow
hunchback nanos Hunchback Nanos
antero-posterior gradient in maternal
Hunchback protein. The photograph shows
the graded distribution of Nanos, detected
with a labeled antibody.
A P A P Photographs courtesy of R. Lehmann, from
Suzuki, D.T., et al.: 1996.

there, ­producing an antero-posterior gradient of maternal Hunchback protein. This

is the sole task of the Nanos protein (Fig. 2.12); it prevents hunchback mRNA transla-
tion by binding to a complex of the mRNA and Pumilio protein, which is encoded by
another of the posterior-group genes. If maternal Hunchback is completely removed
from embryos, then Nanos becomes completely dispensable for antero-posterior
­patterning.
Evolution can only work on what is already there; it cannot ‘look’ at the whole sys-
tem and redesign it economically. So, if a gene causes a difficulty by being expressed in
the ‘wrong’ place, the problem may be solved not by reorganizing the gene-­expression
pattern, but by bringing in a new function, such as nanos, to remove the unwanted
protein.
The fourth maternal product crucial to establishing the posterior end of the axis is
caudal mRNA. Again, this is only translated after fertilization. caudal mRNA itself is
uniformly distributed throughout the egg, but after fertilization, a posterior-to-anterior
gradient of Caudal is established by the specific inhibition of Caudal protein synthesis
by Bicoid, which binds to a site in the 3′ untranslated region of the caudal mRNA.
Because the concentration of Bicoid is low at the posterior end of the embryo, Caudal
protein concentration is highest there (Fig. 2.13). Mutations in the caudal gene result
in the abnormal development of abdominal segments.
Soon after fertilization, therefore, several gradients of maternal proteins have been
established along the antero-posterior axis. Two gradients—Bicoid and Hunchback
proteins—run in an anterior to posterior direction, while Caudal protein is graded
posterior to anterior. We next look at the quite different mechanism that specifies the
two termini of the embryo.

2.10 The anterior and posterior extremities of the embryo are specified by
cell-surface receptor activation
Fig 2.13 Restriction of Caudal protein to
A third group of maternal genes specifies the structures at the extreme ends of the
the posterior end of the embryo. Bicoid
antero-posterior axis—the acron and the head region at the anterior end, and the protein (green) prevents the synthesis of
telson and the most posterior abdominal segments at the posterior end. A key gene in Caudal protein (red) at the anterior end of
this group is torso; mutations in torso can result in embryos developing neither acron the embryo. The proteins were detected
nor telson (see Fig. 2.9). This indicates that the two terminal regions, despite their by fluorescent antibody staining on fixed
topographical separation, are not specified independently but use the same pathway. embryos.
The terminal regions are specified by an interesting mechanism that involves Photograph reproduced from Surkova, S.,
the localized activation of a receptor protein that is itself present throughout the et al.: 2008.
48 Chapter 2 Development of the Drosophila body plan

­ embrane of the fertilized egg. The activated receptor sends a signal to the adjacent
m
cytoplasm that specifies it as terminal. The receptor is known as Torso, as muta-
localized release of
ligand that activates tions in the maternal torso gene produce embryos lacking terminal regions. After
Torso receptor fertilization the maternal torso mRNA is translated and the Torso protein is uniformly
distributed throughout the fertilized egg plasma membrane. Torso is, however, only
activated at the ends of the fertilized egg because the protein ligand that stimulates it
is only present there.
vitelline This ligand for Torso is thought to be a fragment of a secreted protein known as Trunk.
membrane
trunk mRNA is deposited in the egg by the nurse cells and Trunk protein is secreted
ligand into the perivitelline space during oocyte development. The perivitelline space is the
space between the oocyte plasma membrane and a protective membrane of extracel-
perivitelline lular matrix called the vitelline membrane or vitelline envelope, which surrounds the
space oocyte. The Trunk protein is thought to be present throughout the perivitelline space,
Torso but the Trunk fragment that acts as a ligand for Torso is generated only at the poles of
receptor
the egg because the processing activity that produces it from Trunk is present only in
these two regions. Crucial to this processing is the activity of a protein called Torso-like,
which is produced exclusively by follicle cells adjacent to the two poles of the egg and
is present in the vitelline envelope at the poles of the fertilized egg. By the time develop-
ment begins after fertilization, small quantities of the Trunk ligand have been produced
at the poles and are present in the perivitelline space, where they can bind Torso.
Torso receptor activated at poles Because Trunk ligand is only present in small quantities, most of it becomes bound to
Torso at the poles, with little left to diffuse further away. In this way, a localized area of
receptor activation is set up at each pole (Fig. 2.14).
Stimulation of Torso by its ligand produces a signal that is transmitted across the
Fig 2.14 The receptor protein Torso is
involved in specifying the terminal regions plasma membrane to the interior of the developing embryo. This signal directs the
of the embryo. The receptor protein encoded activation of zygotic genes in nuclei at both poles, thus defining the two extremities of
by the gene torso is present throughout the embryo. The Torso protein is one of a large superfamily of transmembrane recep-
the egg plasma membrane. Its ligand is laid tors, known as receptor tyrosine kinases, whose cytoplasmic portions have tyrosine
down in the vitelline membrane at each protein kinase activity. The kinase is activated when the extracellular part of the
end of the egg during oogenesis. After receptor binds its ligand, and the cytoplasmic part of the receptor transmits the signal
fertilization, the ligand is released and onwards by phosphorylating cytoplasmic proteins.
diffuses across the perivitelline space to The ingenious mechanism for setting up a localized area of receptor activation is
activate the Torso protein at the ends of
not confined to determination of the terminal regions of the embryo, but is also used
the embryo only.
in setting up the dorso-ventral axis, which we consider next.

2.11 The dorso-ventral polarity of the embryo is specified by localization of

maternal proteins in the egg vitelline envelope

The dorso-ventral axis is specified by a different set of maternal genes from those
that specify the anterior-posterior axis. The basic mechanism is very similar to that
described in the previous section. The receptor involved in dorso-ventral axis orga-
nization is a maternal protein called Toll, which is present throughout the cell mem-
brane of the fertilized egg. The ventral end of the axis is determined by the localized
production in the ventral vitelline envelope of the ligand for this receptor. The ligand
is a protein fragment produced by processing of a maternal protein called Spätzle.
After fertilization, Spätzle itself is uniformly distributed throughout the extra-embry-
onic perivitelline space. The localized processing of Spätzle is controlled by a small
set of maternal genes that are expressed only in the follicle cells that surround the
future ventral region—about a third of the total surface of the developing egg. The key
gene here is pipe, which codes for an enzyme, a heparan sulfate sulfotransferase, that
is secreted into the oocyte vitelline envelope by these cells. The subsequent activity
of the Pipe enzyme leads, in some way not yet fully understood, to the localization
of a protease activity in the vitelline envelope on the ventral side of the embryo. The
processed Spätzle fragment is thus only present in the ventral perivitelline space.
 Setting up the body axes 49

Toll mRNA is laid down in the oocyte and is probably not translated until after fer-
tilization. Although Toll is present throughout the membrane of the fertilized egg it is
only activated in the future ventral region of the embryo, owing to the localization of its
ligand. Toll activation is greatest where the concentration of its ligand is highest, and falls
off rapidly, probably due to the limited amount of ligand being mopped up by the recep-
tors. Activation of Toll sends a signal to the adjacent cytoplasm of the embryo. At this
stage, the embryo is still a syncytial blastoderm and this signal causes a maternal gene
product in the cytoplasm—the Dorsal protein—to enter nearby nuclei (Fig. 2.15). This
protein is a transcription factor with a vital role in organizing the dorso-ventral axis.

2.12 Positional information along the dorso-ventral axis is provided by the

Dorsal protein

The initial dorso-ventral organization of the embryo is established at right angles

to the antero-posterior axis at about the same time that this axis is being divided into
terminal, anterior, and posterior regions. The embryo initially becomes divided into
four regions along the dorso-ventral axis and this patterning is controlled by the dis-
tribution of the maternal protein Dorsal (see Fig. 2.27).
Unlike Bicoid, Dorsal protein is uniformly distributed in the egg. Initially it is
restricted to the cytoplasm, but under the influence of signals from the ventrally acti-
vated Toll proteins it enters nuclei in a graded fashion, with the highest concentration
in ventral nuclei and the concentration progressively decreasing in a dorsal direction,
as the Toll signal becomes weaker (see Fig. 2.15). Thus there is little or no Dorsal in
nuclei in the dorsal regions of the embryo. The role of Toll was first established by
the observation that mutant embryos lacking it are strongly dorsalized—that is, no
ventral structures develop. In these embryos, Dorsal protein does not enter nuclei
but remains uniformly distributed in the cytoplasm. Transfer of wild-type cytoplasm
into Toll-mutant embryos results in specification of a new dorso-ventral axis, the ven-
tral region always corresponding to the site of injection. This occurs because in the
absence of Toll, the Spätzle fragments produced on the original ventral side diffuse
throughout the perivitelline space because there is no Toll protein to bind them. When
the wild-type cytoplasm is injected, the Toll proteins it contains enter the membrane

Toll activated ventrally

Concentration
of Dorsal
Cross-section of syncytial protein
blastoderm in nuclei
Dorsal vitelline Fig 2.15 Toll protein activation results in
protein in membrane
cytoplasm a gradient of intranuclear Dorsal protein
D D V
along the dorso-ventral axis. Before Toll
protein is activated, the Dorsal protein (red)
Spätzle is distributed throughout the peripheral
fragment
Dorsal (ligand) band of cytoplasm. The Toll protein is a
nuclei protein in receptor that is only activated in the ventral
nucleus region, by a maternally derived ligand (the
Spätzle fragment), which is processed in the
perivitelline space after fertilization. The
yolky
cytoplasm V V localized activation of Toll results in the entry
of Dorsal protein into nearby nuclei. The
activation of Toll intranuclear concentration of Dorsal protein
in ventral region by
Spätzle fragment is greatest in ventral nuclei, resulting in a
Toll receptor protein
ventral to dorsal gradient. D, dorsal; V, ventral.
50 Chapter 2 Development of the Drosophila body plan

Box 2B The Toll signaling pathway: a multifunctional pathway

The interaction between the Dorsal and Cactus proteins in the

Toll signaling pathway
Toll signaling pathway of Drosophila is of more than local inter-
est: Dorsal protein is a transcription factor with homology to the
Rel/NFκB family of vertebrate transcription factors, which are
involved in regulation of gene expression in immune responses,
Nucleus
and the Toll signaling pathway is also used in the adult fly in
defense against infection. So what might seem at first sight a
rather specialized mechanism for confining transcription factors P
to the cytoplasm until it is time for them to enter the nucleus Cactus
Dorsal degradation
during embryonic development is likely to be widely used for
Cactus
controlling gene expression and cell differentiation.
The Toll signaling pathway is thus a good example of a con- Cactus kinase Pelle
served intracellular signaling pathway that is used by multicel-
lular organisms in different contexts, in this case ranging from Cytoplasm
embryonic development to defense against disease. All mem- dMyD88
bers of the Rel/NFκB family are typically held inactive in the
cytoplasm until the cell is stimulated through an appropriate Spätzle Toll Perivitelline
space
receptor. This leads to degradation of the inhibitory protein,
which releases the transcription factor. This then enters the
nucleus and activates gene transcription (see figure). In the
Drosophila embryonic Toll pathway, Dorsal is held inactive in the
syncytium cytoplasm by the protein Cactus. When Toll is acti- In adult Drosophila, the Toll receptor is stimulated by fungal and
vated by binding the Spätzle fragment, its cytoplasmic domain bacterial infection and the signaling pathway results in the pro-
binds an adaptor protein, dMyD88 (or Tube), which in turn inter- duction of antimicrobial peptides. In humans, Toll-like ­receptors
acts with and activates the protein kinase Pelle. Activation of acting by essentially the same pathway are also involved in innate
Pelle leads, through several more intermediate steps, which immunity to microbial infection. IRAK, and IκB are the mammalian
have not yet been fully established, to the phosphorylation and homologs of Pelle, and Cactus, respectively, and play the same
degradation of Cactus. This releases Dorsal, which is then free roles in this pathway. In vertebrates, NFκB is also activated in
to enter a nucleus. response to signaling through receptors other than Toll.

at the site of the injection. Spätzle fragments then bind to these receptors, setting in
motion the chain of events that defines the ventral region at the site of injection.
In the absence of a signal from Toll protein, Dorsal protein is prevented from enter-
ing nuclei by being bound in the cytoplasm to another maternal gene product, the
Cactus protein. As a result of Toll activation, Cactus is degraded and no longer binds
Dorsal, which is then free to enter the nuclei. The pathway leading from Toll to the
activation of Dorsal is shown in Box 2B. In embryos lacking Cactus protein, almost all
of the Dorsal protein is found in the nuclei; there is a very poor concentration gradient
and the embryos are ventralized—that is, no dorsal structures develop.

Summary

Maternal genes act in the ovary of the mother fly to set up differences in the egg in the
form of localized deposits of mRNAs and proteins. After fertilization, maternal mRNAs
are translated and provide the embryonic nuclei with positional information in the form
of protein gradients or localized protein. Along the antero-posterior axis there is an
anterior to posterior gradient of maternal Bicoid protein, which controls patterning of
the anterior region. For normal development it is essential that maternal Hunchback
protein is absent from the posterior region and its suppression is the function of the
posterior to anterior gradient of Nanos. The extremities of the embryo are specified by
 Localization of maternal determinants during oogenesis 51

localized activation of the receptor protein Torso at the poles. The dorso-ventral axis is
established by intranuclear localization of the Dorsal protein in a graded manner (ven-
tral to dorsal), as a result of ventrally localized activation of the receptor protein Toll by
a fragment of the protein Spätzle.

Summary: maternal gene action in the fertilized egg of Drosophila

Antero-posterior Dorso-ventral

mRNAs: bicoid forms anterior to posterior gradient; Spätzle protein activates Toll
hunchback uniform; nanos and caudal uniform receptor on ventral side

Anterior to posterior gradient of Bicoid

protein formed. hunchback mRNA translation Dorsal protein enters ventral nuclei,
suppressed in posterior region by Nanos. caudal giving ventral to dorsal gradient
mRNA translation repressed by Bicoid

Termini: Torso receptor activated by Trunk at ends of egg

Localization of maternal determinants during oogenesis

Having considered the importance of localized maternal gene products in the egg
in setting the basic framework for development, we now look at how they come
to be localized so precisely. When the Drosophila egg is released from the ovary it
already has a well-defined organization. bicoid mRNA is located at the anterior end
and nanos and oskar mRNAs at the opposite end. Torso-like protein is present in the
vitelline envelope at both poles, and other maternal proteins are localized in the ven- Fig 2.16 Egg development in
tral vitelline envelope. Numerous other maternal mRNAs, such as those for caudal, Drosophila. Oocyte development begins
hunchback, Toll, torso, dorsal, and cactus, are distributed uniformly. How do these in a germarium, with stem cells at one
maternal ­mRNAs and proteins get laid down in the egg during oogenesis—its period end. One stem cell will divide four times to
of development in the ovary—and how are they localized in the correct places? give 16 cells with cytoplasmic connections
The development of an egg in the Drosophila ovary is shown in Fig. 2.16. A diploid between each other. One of the cells that
germline stem cell in the germarium divides asymmetrically to produce another stem is connected to four others will become
cell and a cell called the cystoblast, which undergoes a further four mitotic divisions the oocyte, the others will become nurse
to give 16 cells with cytoplasmic bridges between them; this group of cells is known cells. The nurse cells and oocyte become
surrounded by follicle cells and the resulting
as the germline cyst. One of these 16 cells will become the oocyte; the other 15 will
structure buds off from the germarium as
develop into nurse cells, which produce large quantities of proteins and RNAs that are
an egg chamber. Successively produced
exported into the oocyte through the cytoplasmic bridges. Somatic ovarian cells make egg chambers are still attached to each
up a sheath of follicle cells around the nurse cells and oocyte to form the egg ­chamber. other at the poles. The oocyte grows as the
The follicle cells have a key role in patterning the egg’s axes. During ­oogenesis, they nurse cells provide material through the
become subdivided into functionally different populations at various locations around cytoplasmic bridges. The follicle cells have a
key role in patterning the oocyte.

stalk follicle cell anterior follicle cell posterior

follicle cell
Germarium Egg chamber
nucleus

stem cells cyst oocyte

nurse cell oocyte A oocyte P
52 Chapter 2 Development of the Drosophila body plan

the oocyte; these subpopulations express different genes and thus have differing effects
on the parts of the oocyte adjacent to them (Fig. 2.17). Follicle cells also secrete the
materials of the vitelline envelope and egg case that surround the mature egg. During
most of the stages of development discussed here the oocyte is arrested in the first
­prophase stage of meiosis (see Fig. 9.8). Meiosis is completed after fertilization.

2.13 The antero-posterior axis of the Drosophila egg is specified by signals

Fig 2.17 Drosophila oocyte from the preceding egg chamber and by interactions of the oocyte with
development. A developing Drosophila follicle cells
oocyte (right) is shown attached to
its 15 nurse cells (left) and surrounded by a The antero-posterior axis is the first axis to be established in the oocyte. The first visi-
monolayer of 700 follicle cells. The oocyte ble sign of antero-posterior polarity is the movement of the oocyte from a central posi-
and follicle layer are cooperating at this tion surrounded by nurse cells to the posterior end of the developing egg chamber,
time to define the future dorso-ventral axis in direct contact with follicle cells. This rearrangement occurs while the egg chamber
of the egg and embryo, as indicated by the is becoming separated from the germarium, and is caused by preferential adhesion
expression of a gene only in the follicle cells
between the future posterior end of the oocyte and the adjacent posterior follicle cells.
overlying the dorsal anterior region of the
Oocyte antero-posterior polarity is the result of signaling from the anterior of the older
oocyte (blue staining).
egg chamber to the posterior of the younger chamber (Fig. 2.18). Two signaling path-
Photograph courtesy of A. Spradling.
ways are involved and act in relay. The germline cyst in the older chamber first signals
to its anterior follicle cells via a widely used signaling pathway, the Delta–Notch path-
way, which will be described in detail later in the book (see Box 5C, p. 190). This sig-
naling specifies several of the follicle cells to become specialized anterior polar follicle
cells. These specialized cells in turn signal to adjacent follicle cells via receptors that
stimulate another intracellular signaling pathway called the JAK–STAT pathway and
induce them to form a stalk between the two egg chambers. Signals, as yet unknown,
from the stalk cells then cause the younger egg chamber to round up and cause the
oocyte and the posterior follicle cells in this chamber to express the adhesion mol-
ecule E-cadherin, which anchors the oocyte in a posterior position. (The way in which
cadherins and other adhesion molecules work is explained in Box 8B, p. 293.) Thus,
an antero-posterior polarity is propagated from one egg chamber to the next. How the
very first egg chamber produced gets polarized is, however, not yet understood.
Once the oocyte has become located posteriorly, the next steps in its antero-
­posterior polarization are mediated by a protein called Gurken, which is a member
of the transforming growth factor-α (TGF-α) family (examples of the most commonly
used growth factor families in Drosophila are given in Fig. 2.19; we shall meet

germline cyst follicle cells

oocyte

egg chamber
anterior stalk posterior
polar cells polar cells posterior
follicle cells

Fig 2.18 Signals from older to younger egg chambers initially polarize the Drosophila
oocyte. As a germline cyst buds from the germarium, it signals through the Delta–Notch
pathway (small red arrows) to induce the formation of anterior polar follicle cells (red). These in
turn signal to the adjacent cells anterior to them and induce them to become stalk (green). The
stalk induces the adjacent younger cyst to round up and the oocyte and the posterior follicle
cells to produce cadherin, which positions the oocyte at the posterior of the egg chamber. The
yellow arrows indicate the overall direction of signaling from older to younger egg chambers.
 Localization of maternal determinants during oogenesis 53

Common intercellular signals used in Drosophila

Family and examples Receptors Examples of roles in development
Hedgehog family
Hedgehog Patched (together with Smoothened) Patterning of insect segments
Positional signaling in insect leg and wing discs
(see Chapter 11)
Wingless (Wnt) family
Wingless and six Frizzled (together with LRP6) Insect segment and imaginal disc specification
other Wnt proteins (see Chapter 11)
Other Wnts have roles in development
Delta and Serrate
Transmembrane Notch Roles at many stages in development
signaling proteins Specification of oocyte polarity
Transforming growth factor-a (TGF-a) family
Gurken, Spitz, Vein EGF receptor (receptor tyrosine kinase) Polarization of the oocyte
One only in Drosophila, known as Eye development, wing vein differentiation
DER or Torpedo. (see Chapter 11)
Transforming growth factor-b (TGF-b) family
Decapentaplegic Receptors are heterodimers of type I Patterning of the dorso-ventral axis
(e.g. Thick veins) and type II Patterning of imaginal discs (see Chapter 11)
(e.g. Punt) subunits
Serine/threonine kinases
Fibroblast growth factor (FGF) family
A small number of FGF FGF receptors (receptor tyrosine kinases) Migration of tracheal cells Fig 2.19 Common intercellular signals in
homologs (e.g. Branchless) Two in Drosophila, e.g. Breathless (see Chapter 11)
Drosophila.

many of these again in vertebrates and other animals). Early in oocyte develop-
ment, gurken mRNA is translated at the posterior end of the oocyte close to the
nucleus, ­producing a local posterior concentration of the protein, which is secreted
across the oocyte membrane. Gurken induces the terminal follicle cells to adopt a
posterior fate by locally stimulating the receptor protein Torpedo, which is present
on the surface of follicle cells (Fig. 2.20). Torpedo is a receptor tyrosine kinase and
is the Drosophila equivalent of the epidermal growth factor (EGF) receptor of mam-
mals. In response to Gurken signaling through Torpedo, the posterior follicle cells
produce an as-yet-­unidentified signal that induces a reorientation of the oocyte’s
microtubule cytoskeleton, so that about midway through oogenesis, the minus ends
of most microtubules are directed toward the anterior and the plus ends toward the
posterior.

2.14 Localization of maternal mRNAs to either end of the egg depends on the
reorganization of the oocyte cytoskeleton

The reorientation of the oocyte microtubule cytoskeleton and the correct localiza-
tion of mRNAs depends on a group of proteins known as the PAR proteins, which
are themselves localized at the posterior end of the oocyte. Among other functions,
these proteins are involved in determining antero-posterior polarity of cells in many
different situations in animal development. They were discovered in Caenorhabditis
elegans and are discussed in more detail in Chapter 6, in connection with their role in
controlling the first asymmetric division of the fertilized nematode egg.
Microtubule reorientation is essential for the final localization of maternal mRNAs,­
such as bicoid and oskar at either end of the egg. bicoid mRNA is originally made
by nurse cells located next to the anterior end of the developing oocyte, and is trans-
ferred from them to the oocyte. After the developing egg responds to the Gurken
signal, bicoid mRNA is transported along the reoriented microtubules, probably by
the motor protein dynein, to its final location at the anterior-most end of the egg
54 Chapter 2 Development of the Drosophila body plan

Specification of A–P and D–V axes

nurse cell

uncommitted Nucleus
terminal follicle cells Gurken
protein

microtubules
Terminal
Oocyte follicle cell
Fig 2.20 Specification of the antero-
posterior and dorso-ventral axes during
Drosophila oogenesis. The oocyte moves Gurken Torpedo
(receptor)
to the posterior end of the egg chamber and
anterior posterior
comes into contact with the polar follicle follicle cells follicle cells
cells. It is separated from follicle cells at
the anterior end (blue) by the nurse cells.
gurken mRNA in the oocyte is localized to
the posterior. It is translated and Gurken
protein is secreted locally. The binding of
this protein to the receptor protein Torpedo
on the adjacent follicle cells initiates their
specification as posterior terminal follicle
cells (yellow). These send a signal back to dorsal follicle cells
the oocyte that reorganizes the oocyte D Gurken
microtubule cytoskeleton. The oocyte nucleus protein
now moves to an anterior dorsal position and
gurken mRNA coming into the oocyte from
the nurse cells is transported to the region A P
surrounding the nucleus. Translation of the
mRNA and local release of Gurken protein
specifies the adjacent follicle cells as dorsal
follicle cells and that side of the oocyte as V ventral follicle cells
the future dorsal side.

(Fig. 2.21, first panel). Similarly, oskar mRNA is delivered into the oocyte by nurse
cells and transported towards the posterior end of the oocyte by kinesin via another
set of microtubules (see Fig. 2.20, second panel). Both these localizations require the
RNA-binding protein Staufen, which can itself be observed to localize to the posterior
and anterior ends of the oocyte (Fig. 2.22). Observations of oskar mRNA particles
showed that they were moved in all directions, but with a sufficient bias towards the
posterior end. One role of oskar mRNA and protein is to nucleate the assembly of the
germplasm at the posterior end of the egg. In the embryo this cytoplasm is incorpo-
rated into the pole cells (see Section 2.1) and directs them to form primordial germ
cells. The localization of oskar mRNA and assembly of the germplasm is also required
for the subsequent localization of nanos mRNA to the posterior end (see Section 2.9).
Unlike the transport of bicoid and oskar mRNA, nanos localization depends on the
actin microfilaments of the cytoskeleton.

2.15 The dorso-ventral axis of the egg is specified by movement of the

oocyte nucleus followed by signaling between oocyte and follicle cells

The setting-up of the egg’s dorso-ventral axis involves a further set of oocyte–­follicle
cell interactions, which occur after the posterior end of the oocyte has been ­specified
 Localization of maternal determinants during oogenesis 55

bicoid mRNA localization oskar mRNA localization gurken mRNA localization

– – –

+ + +
+ + +

+ + +

Fig 2.21 bicoid and oskar mRNAs

are localized to the anterior and
and which depend on the previous reorganization of the microtubule array. The
posterior ends of the oocyte
oocyte nucleus moves along the microtubules from the posterior of the oocyte to a respectively. Localization of maternal
site on the anterior margin (see Fig. 2.20). Gurken protein is expressed at this new mRNAs delivered to the oocyte by the nurse
site, possibly from mRNA that has been relocated to one side of the nucleus from cells is by transport along microtubules. The
elsewhere in the oocyte (Fig. 2.21, third panel). The locally produced Gurken acts motor protein dynein transports bicoid and
as a signal to the adjacent follicle cells, specifying them as dorsal follicle cells; the gurken mRNAs towards the minus ends of
side away from the nucleus thus becomes the ventral region by default. The ventral the microtubules. oskar mRNA is transported
follicle cells produce proteins, such as Pipe (see Section 2.11), that are deposited in towards the plus ends of the microtubules,
the ventral vitelline envelope of the oocyte and are instrumental in establishing the possibly by the motor protein kinesin. Three
ventral side of the axis. different populations of microtubules are
thought to be involved in the transport of
The Torso-like protein, which distinguishes the termini of the Drosophila embryo, is
bicoid, oskar and gurken mRNAs.
synthesized and secreted by follicle cells at both the posterior and anterior poles, but
not by the other follicle cells. Torso-like protein is thus deposited exclusively in the
vitelline envelope at both ends of the egg during oogenesis. After fertilization it acts
together with other proteins to cause the processing of Trunk and production of the
ligand for Torso (see Section 2.10).

(a) (b) (c)

Fig 2.22 The RNA-binding protein Staufen localizes to each egg chambers were counterstained with rhodamine-phalloidin,
pole of the Drosophila oocyte. The location of the Staufen protein which labels actin filaments (red). b, In a stage 9 egg chamber,
at different stages of oocyte development is made visible by linking Staufen localizes to the posterior of the oocyte. c, In a stage 10b
its gene to the coding sequence of green fluorescent protein (GFP) egg chamber, Staufen is localized to both the anterior and posterior
and making flies transgenic for this construct. a, In a stage 6 egg poles of the oocyte.
chamber, Staufen (green) accumulates in the oocyte. The fixed Photographs reproduced from Martin, S.G., et al .: 2003.
56 Chapter 2 Development of the Drosophila body plan

Summary

Drosophila oocytes develop inside individual egg chambers that are successively pro-
duced from a germarium, which contains germline stem cells that give rise to the oocyte
and nurse cells, and stem cells that give rise to the somatic follicle cells that surround the
oocyte. The nurse cells provide the oocyte with large amounts of mRNAs and proteins,
some of which become localized in particular sites. As a result of signals from the adja-
cent older egg chamber, an oocyte becomes localized posteriorly in its egg chamber as
a result of differential cell adhesion to posterior follicle cells. Subsequently the oocyte
sends a signal to these follicle cells, which respond with a signal that causes a reorganiza-
tion of the oocyte cytoskeleton that localizes bicoid mRNA to the anterior end and other
mRNAs to the posterior end of the oocyte, thus setting up the beginnings of the embry-
onic antero-posterior axis. The dorso-ventral axis of the oocyte is also initiated by a local
signal from the oocyte to follicle cells on the future dorsal side of the egg, thus specifying
them as dorsal follicle cells. Thus, directly or indirectly, follicle cells on the opposite side
of the oocyte specify the ventral side of the oocyte by deposition of maternal proteins
in the ventral vitelline envelope. Follicle cells at either end of the oocyte similarly specify
the termini by localized deposition of maternal protein in the vitelline envelope.

Summary: polarization of axes in the Drosophila oocyte

Antero-posterior Dorso-ventral

oocyte localized at posterior end of follicle

by cadherin

oocyte Gurken protein induces posterior

follicle cells via Torpedo

posterior signal from follicle cells reorganizes nucleus moves dorsally

oocyte cytoskeleton

oocyte Gurken induces dorsal follicle cells

bicoid mRNA localized in anterior,
oskar and other mRNAs in posterior
ventral follicle cells deposit ventral
proteins in oocyte viteline envelope

Termini: follicle cells at both ends of the egg deposit ligand for the Torso protein
in the vitelline envelope

Patterning the early embryo

Understanding in such detail how the main body axes of Drosophila are specified
was a major experimental achievement, and those who work on other animals like
the frog and chick, whose developmental genetics are less well-known, are justifiably
somewhat envious. We have seen how gradients of Bicoid, Hunchback, and Caudal
proteins are established along the antero-posterior axis, and how intranuclear Dorsal
protein is graded along the ventral to dorsal axis. This maternally derived framework
of positional information is interpreted and elaborated on by zygotic genes to give
each region of the embryo an identity. Most of the zygotic genes that are the first to
 Patterning the early embryo 57

be activated along the antero-posterior and dorso-ventral axes encode transcription

Gap gene expression
factors, which are thus localized along the axes and activate yet more zygotic genes.
We first consider patterning along the antero-posterior axis. hunchback

2.16 The antero-posterior axis is divided up into broad regions

by gap-gene expression

Patterning along the antero-posterior axis starts when the embryo is still acellular. The
gap genes are the first zygotic genes to be expressed along the antero-posterior axis,
giant
and all code for transcription factors. Gap genes were initially recognized by their
mutant phenotypes, in which quite large sections of the body pattern along the ante-
ro-posterior axis are missing. Although the mutant phenotype of a gap gene usually
shows a gap in the antero-posterior pattern in more-or-less the region in which the
gene is normally expressed, there are also more wide-ranging effects. This is because
gap-gene expression is also essential for later development along the axis. Krüppel
Gap-gene expression is initiated by the antero-posterior gradient of Bicoid protein
while the embryo is still essentially a single multinucleate cell. Bicoid primarily acti-
vates anterior expression of the gap gene hunchback, which in turn is instrumental in
switching on the expression of the other gap genes, among which are giant, Krüppel,­
and knirps, which are expressed in this sequence along the antero-posterior axis
knirps
(Fig. 2.23) (giant is in fact expressed in two bands, one anterior and one posterior,
but its posterior expression does not concern us here).
As the blastoderm is still acellular at this stage, the gap–gene proteins can diffuse
away from their site of synthesis. They are short-lived proteins with half-lives of min-
utes. Their distribution, therefore, extends only slightly beyond the region in which
the gene is expressed, and this typically gives a bell-shaped protein concentration tailless
profile. Zygotic hunchback protein is an exception, as it is expressed fairly uniformly
over a broad anterior region with a steep drop in concentration at the posterior bound-
ary of its expression. The control of zygotic hunchback expression by Bicoid is best
understood and will be considered first.

2.17 Bicoid protein provides a positional signal for the anterior expression of Fig 2.23 The expression of the gap
zygotic hunchback genes hunchback, Krüppel, giant, knirps,
The Bicoid protein induces expression of the zygotic hunchback genes over most of and tailless in the early Drosophila
embryo. Gap-gene expression at different
the anterior half of the embryo. This zygotic expression is superimposed on low levels
points along the antero-posterior axis is
of Hunchback protein expressed from the ubiquitous maternal hunchback mRNA,
controlled by the concentration of Bicoid
whose translation is suppressed posteriorly by Nanos (see Section 2.9). and Hunchback proteins, together with
The localized anterior expression of Hunchback is an interpretation of the posi- interactions between the gap genes
tional information provided by the Bicoid protein gradient. The hunchback gene themselves. The expression pattern of the
is switched on only when Bicoid, a transcription factor, is present above a certain gap genes provides an aperiodic pattern
threshold concentration. This level is attained only in the anterior half of the embryo, of transcription factors along the antero-
close to the site of Bicoid synthesis, which restricts hunchback expression to this posterior axis, which delimits broad body
region. regions.
The relationship between Bicoid concentration and hunchback gene expression
can be illustrated by looking at how hunchback expression changes when the Bicoid
concentration gradient is changed by increasing the maternal dosage of the bicoid
gene (Fig. 2.24). Expression of hunchback then extends more posteriorly because
the region in which the concentration of Bicoid is above the threshold for hunchback
activation is also extended in this direction.
A smooth gradient in Bicoid protein is translated into a sharp boundary of hunch-
back expression about halfway along the embryo. Neighboring nuclei along the antero-
­posterior axis experience Bicoid concentrations that are very similar, differing by only
58 Chapter 2 Development of the Drosophila body plan

about 10% from one cell to the next. How do nuclei distinguish such small ­differences
One dose of bicoid
and form a sharp boundary of gene expression against a background of biological
‘noise’, such as that resulting from small random variations in the dynamics of Bicoid
behavior. This question has been addressed by using expression of a Bicoid–GFP
hunchback expression construct (see Section 2.8) to directly observe and measure the distribution of Bicoid
protein and its concentrations in nuclei at different points along the antero-posterior
axis. Bicoid distribution, intranuclear concentrations at different positions, and the
sharp boundary of hunchback expression all turn out to be highly reproducible over
Concentration of Bicoid protein
a large number of embryos, which suggests that background noise is in fact kept low
by precise controls in the embryo. These might, for example, include communication
threshold
between nuclei.
Bicoid is a member of the homeodomain family of transcriptional activators and
A P activates the hunchback gene by binding to regulatory sites within the promoter
region. Direct evidence of hunchback activation by Bicoid was originally obtained by
experiments in which a fusion gene constructed from the hunchback promoter regions
Two doses of bicoid
and a bacterial reporter gene, lacZ, was introduced into the fly genome using trans-
genic techniques such as P-element-mediated transformation (Box 2C, p. 60; some
other related techniques for tracking gene expression and forcing new patterns of
hunchback expression gene expression in Drosophila are shown in Box 2D, p. 61). The large promoter region
required for completely normal gene expression can be whittled down to an essential
sequence of 263 base pairs that will still give almost normal activation of hunchback.
This sequence has several sites at which Bicoid can bind, and it seems likely that the
Concentration of Bicoid protein
threshold response involves cooperativity between the different binding sites: that is,
binding of Bicoid at one site makes binding of Bicoid at a nearby site easier and so
threshold facilitates further binding.
The regulatory region of a gene such as hunchback is yet another example of a
developmental switch that directs nuclei along a new developmental pathway. We
A P
will encounter many more examples of these transcriptional switches in Drosophila
early development.
Fig 2.24 Maternal Bicoid protein controls
zygotic hunchback expression. If the dose
of maternal bicoid is increased twofold, the 2.18 The gradient in Hunchback protein activates and represses other gap genes
extent of the Bicoid gradient also increases.
The Hunchback protein is a transcription factor and acts as a morphogen to which
The activity of the hunchback gene is
the other gap genes respond. The other gap genes are expressed in transverse stripes
determined by the threshold concentration
of Bicoid, so at the higher dose, its region of
across the antero-posterior axis (see Fig. 2.23). The stripes are delimited by mecha-
expression is extended toward the posterior nisms that depend on the control regions of these genes being sensitive to different
end because the region in which Bicoid concentrations of Hunchback protein, and also to other proteins, including Bicoid.
concentration exceeds the threshold level Expression of the Krüppel gene, for example, is activated by low levels of Hunchback,
also extends more posteriorly (see graph, but is repressed at high concentrations. Within this concentration window Krüppel
bottom panel). remains activated (Fig. 2.25, top panel). But below a lower threshold concentration of
Hunchback, Krüppel is not activated. In this way, the gradient in Hunchback protein
locates a band of Krüppel gene activity near the center of the embryo (Fig. 2.26).
Refinement of this spatial localization is brought about by repression of Krüppel by
other gap-gene proteins.
Such relationships were worked out by altering the concentration profile of Hunch-
back protein systematically, while all other known influences were eliminated or held
constant. Increasing the dose of Hunchback protein, for example, results in a posterior
shift in its concentration profile, and this results in a posterior shift in the posterior
boundary of Krüppel expression. In another set of experiments on embryos lacking
Bicoid protein (so that only the maternal Hunchback protein gradient is present), the
level of Hunchback is so low that Krüppel is even activated at the anterior end of the
embryo (Fig. 2.25, bottom panel).
 Patterning the early embryo 59

Fig 2.25 Krüppel gene activity is specified by Hunchback protein. Top panel: above a
Krüppel gene expression in wild type
threshold concentration of Hunchback protein, the Krüppel gene is repressed; at a lower
concentration, above another threshold value, it is activated. Bottom panel: in mutants lacking Krüppel
the bicoid gene, and thus also lacking zygotic hunchback gene expression, only maternal
Hunchback protein is present, which is located at the anterior end of the embryo at a relatively
low level. In these mutants, Krüppel is activated at the anterior end of the embryo, giving an
abnormal pattern.

Hunchback protein is also involved in specifying the anterior borders of the bands Concentration of
Hunchback protein
of expression of the gap genes knirps and giant, again by a mechanism involving
Repression
thresholds for repression and activation of these genes. At high concentrations of
Activation
Hunchback, knirps is repressed, and this specifies its anterior margin of expression.
The posterior margin of the knirps band is specified by a similar type of interaction
A P
with the product of another gap gene, tailless. Where the regions of expression of the
gap genes overlap, there is extensive cross-inhibition between them, their proteins
all being transcription factors. These interactions are essential to sharpen and stabi- Krüppel gene expression in bicoid mutant
lize the pattern of gap gene expression. For example, the anterior border of Krüppel
Krüppel
expression lies four to five nuclei posterior to nuclei that express giant, and is set by
low levels of Giant protein.
The antero-posterior axis thus becomes divided into a number of unique regions
on the basis of the overlapping and graded distributions of different transcrip-
tion factors. This beautifully elegant method of delimiting regions can, however,
only work in an embryo such as the acellular syncytial blastoderm of Drosophila, Concentration of maternal
where the transcription factors are able to diffuse freely throughout the embryo. Hunchback protein
This regional distribution of the gap-gene products provides the starting point for
the next stage in development along the antero-posterior axis—the activation of Activation
­the pair-rule genes, cellularization, and the beginning of segmentation. But before
A P
we discuss this, we return to the dorso-ventral axis at around this time point in
development and look at how it is patterned further as the syncytial blastoderm
becomes cellular.

2.19 The expression of zygotic genes along the dorso-ventral axis is

controlled by Dorsal protein

After Dorsal protein has entered the nuclei of the syncytial blastoderm (see Sec-
tion 2.12), its effects on gene expression divide the dorso-ventral axis into well
defined regions. This is also the stage at which the germ layers become distinguished,
as Dorsal also acts to specify the ventral-most cells as prospective mesoderm. It is
estimated that 30 genes are direct targets of the Dorsal gradient. Going from ventral
to dorsal, the main re­gions are mesoderm, ventral ectoderm, dorsal ectoderm, and
prospective amnioserosa. The mesoderm gives rise to internal soft tissues such as
muscle and connective tissue; the ventral ectoderm becomes the neuroectoderm,
which gives rise to all the nervous tissue as well as ventral epidermis; the dorsal
ectoderm gives rise only to epidermis. The third germ layer, the endoderm, which is
Fig. 2.26 Localization of Krüppel
located at either end of the embryo and which we do not consider here, gives rise to
expression by the Hunchback protein
the midgut (see Fig. 2.3). gradient. Wild-type embryos were
Patterning along the axes poses a problem like that of patterning the French flag stained with fluorescent antibodies to
(see Section 1.15). Expression of zygotic genes in localized regions along the dorso- simultaneously detect the Hb protein
ventral axis is initially controlled by the graded concentration of the intranuclear gradient (red) and the expression of Krüppel
Dorsal protein, which falls off rapidly in the dorsal half of the embryo; little Dorsal mRNA (green) early in nuclear cycle 14.
protein is found in nuclei above the equator. In the ventral region, Dorsal protein Photograph reproduced from Yu, D., and
has two main functions—it activates certain genes at specific positions in the region Small, S.: 2008.
60 Chapter 2 Development of the Drosophila body plan

Box 2C P-element-mediated transformation

Transgenic fruit flies have contributed greatly to Drosophila

developmental genetics. They are made by inserting a known white– white–
sequence of DNA into the Drosophila chromosomal DNA, using
as a carrier a transposon that occurs naturally in some strains
of Drosophila. This transposon is known as a P element, and the
technique as P-element-mediated transformation (see ­figure).
Two DNA plasmids injected
P elements can insert at almost any site on a chromosome, into fertilized egg
and can also hop from one site to another within the germ cells,
white+
an action that requires an enzyme called a transposase. As hop-
ping can cause genomic instability, carrier P elements have had
their own transposase gene removed. The transposase required
Defective P element, Helper P element
to insert the P element initially is instead provided by a helper no transposase makes transposase
but cannot insert
P element, which cannot itself insert into the host chromosomes
and is thus quickly lost from cells. The carrier and helper ele-
ments are injected together into the posterior end of the egg
where the germ cells are made.
As well as the gene to be inserted, an additional marker gene,
such as the wild-type white + gene, is added to the P element.
When white + is the marker, the P element is inserted into flies
homozygous for the mutant white - gene (which have white
eyes rather than the red eyes of the wild-type Drosophila). Red
eyes are dominant over white, and so flies in which the P ele- G0 generation
ment has become integrated into the chromosome, and is being
expressed, can be detected by their red eyes.
In the first generation, all flies have white eyes, as any P ele-
ment that has integrated is still restricted to the germ cells. But
in the second generation, a few flies will have wild-type red
eyes, showing that they carry the inserted P element in their G1 generation
somatic cells.
This technique can be used to increase the number of copies
of a particular gene, or to introduce a mutated gene that has
its control or coding regions altered in a known way, or to intro-
duce new genes. It is also possible to introduce genes that carry
a marker coding sequence such as lacZ (encoding the bacterial
enzyme β-galactosidase), whose expression is detectable by his-
tochemical staining (see Box 1D, p. 20). The P element itself can
white+
also be used as a mutagen, as its insertion into a gene usually (transformed)
destroys that gene’s function.
This approach has been adapted to large-scale screens that
look systematically for genes whose overexpression or mis- complement to the more conventional genetic screens described
expression in a particular tissue causes a mutant phenotype in Box 2A (p. 42), which generally detect loss-of-function muta-
(see Box 2D, p. 61). This is known as misexpression screening. tions. If the target flies also carry a mutation in a known gene,
In this case, flies expressing Gal4 in the tissue of interest are misexpression screening can be used to identify genes whose
crossed with large numbers of different lines of target flies car- overexpression enhances or suppresses the mutation. This
rying random insertions of the Gal4-binding site, and the prog- approach can identify genes whose products interact directly or
eny screened for a mutant phenotype. This approach is a useful are part of the same pathway.
 Patterning the early embryo 61

Box 2D Targeted gene expression and misexpression screening

The ability to turn on the expres-

sion of a gene in a particular place
and time during development is
very useful for analyzing its role
in development. This is called tar-
geted gene expression and can
be achieved in several ways. One
approach is to give the selected
gene a heat-shock promoter using
Pattern line Target lines
standard genetic techniques such
as P-element-­mediated transfor-
mation (see Box 2C, p. 60). This
endogenous
enables the gene to be switched gene
on by a sudden rise in the tem-
perature at which the embryos are Gal4 Gal4 sites
being kept. By adjusting the tem-
Pattern P element Target P element
perature, the timing of expression
of genes attached to this promoter
can be controlled; the effects of expressing a gene at different the target gene has Gal4-binding sites in its regulatory region. The
stages of development can be studied in this way. photo shows the use of this technique to investigate ­expression
Another approach to targeted gene expression uses the of engrailed in sensory neurons in the head of the adult fly.­
transcription factor Gal4 from yeast. This protein can activate A Drosophila strain carrying an engrailed-GAL4 coding fusion trans-
transcription of any gene whose promoter has a Gal4-binding gene was crossed with a strain carrying a Gal4-responsive GFP
site. In Drosophila, genes with Gal4-responsive promoters can transgene. In the progeny of these flies, GFP is expressed in those
be created by inserting the Gal4-binding site. To turn on the cells in which engrailed is expressed, making them glow green.
target gene, Gal4 itself has to be produced in the embryo. In A novel pattern of gene expression can also be obtained using
one approach, Gal4 can be produced in a designated region or the Gal4 system. Using the Gal4 system, for example, the pair-
at a particular time in development by introducing a P-element rule gene even-skipped was expressed in even-numbered rather
transgene in which the yeast Gal4-coding region is attached to a than odd-numbered parasegments, and this led to changes in
Drosophila regulatory region known to be activated in that situ- the pattern of denticles on the cuticle.
ation (see top figure). Photograph from Blagburn, J.M.: 2008.
A second, more versatile, approach
is based on the so-called enhancer-
trap technique. The Gal4-coding
sequence is attached to a vector that
integrates randomly into the Droso-
phila genome. The Gal4 gene will come
under the control of the promoter and
enhancer region adjacent to its site of
integration, and so Gal4 protein will be
produced where or when that gene is
normally expressed. A large number
of Drosophila lines with differing pat-
terns of Gal4 expression for different
purposes have been produced by this
technique. The selected target gene
will be silent in the absence of Gal4.
To activate it in a particular tissue, for
example, flies that express Gal4 in that
tissue are crossed with flies in which
62 Chapter 2 Development of the Drosophila body plan

Dorsal
and represses the activity of other genes, which are therefore only expressed in the
dorsal region (Fig. 2.27).
In the ventral-most region, where concentrations of intranuclear Dorsal protein are
highest, the zygotic genes twist and snail are activated by Dorsal protein in a strip
of nuclei along the ventral middle line of the embryo; soon after this the syncytial
blastoderm becomes cellular. This ventral strip of cells will form the mesoderm. The
­expression of twist and snail is required both for development of the cells as meso-
derm and for gastrulation, during which the ventral band of prospective mesoderm
cells moves into the interior of the embryo. In the future neuroectoderm, the gene
rhomboid is activated at low levels of Dorsal protein, but is not expressed in more
ventral regions because it is repressed there by the Snail protein.
The genes decapentaplegic, tolloid, and zerknüllt are repressed by Dorsal protein
and so their activity is confined to the more dorsal regions of the embryo, where there
is virtually no Dorsal protein in the nuclei. zerknüllt is expressed most dorsally and
appears to specify the amnioserosa. decapentaplegic is a key gene in the specification
Ventral of pattern in the dorsal part of the dorso-ventral axis and its role will be considered
in detail in Section 2.20.
amnioserosa (zerknüllt) Mutations in the maternal dorso-ventral genes can cause dorsalization or ventral-
dorsal ectoderm (decapentaplegic, tolloid) ization of the embryo (see Section 2.12). In dorsalized embryos, Dorsal protein is
excluded uniformly from the nuclei. This has a number of effects, one of which is that
neurectoderm (rhomboid, sog)
the decapentaplegic gene is no longer repressed and is expressed everywhere. In con-
mesoderm (twist, snail)
trast, twist and snail are not expressed at all in dorsalized embryos, as they need high
Dorsal gradient intranuclear levels of Dorsal protein to be activated. The opposite result is obtained in
mutant embryos where the Dorsal protein is present at high concentration in all the
Fig 2.27 The subdivision of the dorso- nuclei, and the embryos are ventralized; twist and snail are expressed throughout and
ventral axis into different regions by the decapentaplegic is not expressed at all (Fig. 2.28).
gradient in intranuclear Dorsal protein. Genes whose expression is regulated by the Dorsal protein, such as twist, snail,
A gradient of intranuclear Dorsal protein is and decapentaplegic, contain binding sites for the protein in their control regions
formed with its high point at the ventral
that activate or repress gene expression at particular concentrations of the Dorsal
midline and little or no Dorsal protein in
protein. This threshold effect on gene expression is the result of the integrating func-
the dorsal half of the blastoderm. In the
dorsal region, the genes tolloid, zerknüllt,
tion of these cooperative binding sites, as discussed earlier in regard to the activation
and decapentaplegic, which are repressed of hunchback by Bicoid. The ability of genes to respond in a threshold-like manner
elsewhere by Dorsal protein, are expressed. to varying concentrations of Dorsal protein is due to the presence of both high- and
In the ventral half of the blastoderm, the low-affinity binding sites for the protein in their regulatory DNA. In the most ventral
Dorsal protein activates the genes twist, part of the embryo (a strip 12–14 cells wide), where the concentration of Dorsal
snail, rhomboid, and short gastrulation protein is high, the spatial extent of gene expression is delimited by the low-affinity
(sog). twist is autoregulatory, maintaining sites, whereas high-affinity sites control expression in slightly more dorsal regions
its own expression, and also activates snail; (up to 20 cells from the ventral midline) where there is less Dorsal protein. It is very
Snail protein inhibits rhomboid and sog likely that the threshold response involves cooperativity between the different binding
expression, repressing these genes in the
sites; that is, binding at one site makes binding at a nearby site easier and so facili-
future mesoderm. twist and snail require high
tates further binding. Inhibitory interactions are also involved in delimiting areas of
levels of Dorsal for their activation whereas
rhomboid and sog can still be activated when
gene expression. For example, the regulatory regions of the rhomboid gene contain
lower levels of Dorsal are present. modules with binding sites for both Dorsal and Snail proteins; Dorsal activates while
Snail represses. Snail protein thus represses the expression of rhomboid in the ventral
region and this helps confine rhomboid expression to the neuroectoderm.
An approximately twofold difference in the level of Dorsal determines whether an
unspecified embryonic cell forms mesoderm or neuroectoderm. Five different thresh-
olds in Dorsal concentration pattern the future ventral midline and the neuroectoderm.
The neuroectoderm, for example, is subsequently subdivided into three layers along
the dorso-ventral axis, which will form three distinct columns in the future nerve cord
(we shall return to this topic in Chapter 12). This subdivision is primarily the result
of activation of genes for three different transcription factors at distinct thresholds of
the Dorsal gradient, and the pattern is maintained by regulatory ­interactions between
 Patterning the early embryo 63

Wild type Ventralized

D D

Dorsal Dorsal
protein protein

V V

threshold above which threshold above which Fig 2.28 The nuclear gradient in Dorsal
twist gene is activated twist gene is activated
threshold threshold protein is interpreted by the activation
above which Dorsal above which Dorsal of other genes, such as twist and
decapentaplegic protein decapentaplegic protein
gene is in gene is in decapentaplegic. Left panels: in normal
repressed nucleus repressed nucleus embryos, the twist gene is activated
above a certain threshold concentration
(green line) of Dorsal protein, whereas
D V D V
above a lower threshold (yellow line), the
decapentaplegic (dpp) gene is repressed.
dpp Right panels: in ventralized embryos, the
expression
twist Dorsal protein is present in all nuclei; twist
twist expression
expression
is now also expressed everywhere, whereas
decapentaplegic is not expressed at all,
because Dorsal protein is above the threshold
V V
level required to repress it everywhere.

these genes and with other genes, in which genes expressed in the more ventral
regions tend to repress those expressed more dorsally.
The gradient of Dorsal protein is therefore effectively acting as a morphogen gradi-
ent along the dorso-ventral axis, activating specific genes at different threshold con-
centrations, and so defining the dorso-ventral pattern. The regulatory sequences in
these genes can be thought of as developmental switches, which when thrown by the
binding of transcription factors activate genes and set cells off along new developmen-
tal pathways. The Dorsal protein gradient is one solution to the French flag problem.
But it is not the whole story—yet another gradient is also involved.

2.20 The Decapentaplegic protein acts as a morphogen to pattern the

dorsal region

As with the antero-posterior axis, each end of the dorso-ventral axis is specified by dif-
ferent proteins. The gradient of Dorsal protein, with its high point in the ventral-most
region, specifies the initial pattern of zygotic gene activity, and patterns the ventral
mesoderm and neuroectoderm. But the dorsal region is not similarly specified by a
low-level gradient in Dorsal. Indeed, there is little or no Dorsal protein in the nuclei of
the dorsal half of the embryo. The more dorsal part of the dorso-ventral pattern is deter-
mined by a gradient in the activity of another morphogen, the Decapentaplegic protein.
This specifies the dorsal ectoderm and the most dorsal region, the amnioserosa.
Soon after the gradient of intranuclear Dorsal protein has become established the
embryo becomes cellular, and transcription factors can no longer diffuse between
nuclei. Secreted or transmembrane proteins and their corresponding receptors must
now be used to transmit developmental signals between cells. Decapentaplegic is one
such secreted signaling protein (see Fig. 2.19). As we shall see in Chapter 4, Decapen-
taplegic is a homolog of bone morphogenetic protein-4 (BMP-4), a TGF-β cytokine in
64 Chapter 2 Development of the Drosophila body plan

vertebrates that is also involved in patterning the dorso-ventral axis. Decapentaplegic

is also involved in many other developmental processes throughout Drosophila devel-
opment, including the patterning of the wing imaginal discs, which is discussed in
Chapter 11.
The decapentaplegic (dpp) gene is expressed throughout the dorsal region where
Dorsal protein is not present in the nuclei. Decapentaplegic protein (Dpp) then dif-
fuses away from this source, forming a gradient of Dpp activity with its high point
in the dorsal region. The cells in the dorsal region have receptors that allow them to
accurately assess how much Dpp is present and to respond by activating transcription
of the appropriate genes, thus dividing up the dorso-ventral axis into different regions
characterized by different patterns of gene expression.
Evidence that a gradient in Dpp specifies dorsal pattern comes from experiments
in which dpp mRNA is introduced into an early wild-type embryo. As more mRNA
is introduced and the concentration of Decapentaplegic protein increases above the
normal level, the cells along the dorso-ventral axis adopt a more dorsal fate than they
would normally. Ventral ectoderm becomes dorsal ectoderm and, at very high concen-
trations of dpp mRNA all the ectoderm develops as the amnioserosa.
Dpp is initially produced uniformly throughout the dorsal region just as cellulariza-
tion is beginning, but within less than an hour its activity appears to be restricted to a
dorsal strip of some five to seven cells and is much lower in the adjacent prospective
dorsal ectoderm (Fig. 2.29). This sharp peak of Dpp concentration is not a matter

Decapentaplegic is expressed in the dorsal region Dpp activity is prevented from spreading Dpp, Sog and Tld interact, resulting in a sharp peak
where Dorsal is not present in the nuclei ventrally by Sog of Dpp activity in the dorsal-most cells

Dpp activity
Dpp protein
Tld protein
Sog binds
amnioserosa Dpp and Tld
at gradient
overlap
dorsal ectoderm

neurectoderm

Sog protein mesoderm

Dorsal protein decapentaplegic, tolloid sog

Fig 2.29 Decapentaplegic protein activity is restricted to the the neurectoderm. Right panel: Dpp bound to Sog is also thought to be
dorsal-most region of the embryo by the antagonistic activity of carried towards the dorsal region by the developing Sog gradient, thus
the Short gastrulation protein. Left panel: the gradient of intranuclear helping to concentrate its activity in a sharp peak in the dorsal-most
Dorsal protein leads to the expression of the gene short gastrulation region. Tld is also thought to be involved in creating the final sharp
(sog) in the non-neural ventral ectoderm of the early blastoderm. dorsal peak of Dpp activity. It binds to the Sog–Dorsal complex and
decapentaplegic and tolloid are expressed throughout the dorsal cleaves Sog. This releases Dorsal, which can then bind to its receptors.
ectoderm. Middle panel: sog encodes a secreted protein, Sog, that forms In the mid-blastoderm embryo the dorsal region has been subdivided
a ventral to dorsal gradient, while the secreted Decapentaplegic protein by Sog activity into a region of high Dpp signaling, which will become
(Dpp) is initially present throughout the dorsal region. The protease the amnioserosa, and a zone of lower Dpp signaling, which is the dorsal
Tolloid (Tld) is also expressed in the same region as Dpp. Where Sog ectoderm.
meets Dpp, it binds it and thus prevents Dpp interacting with its After Ashe and Levine: 1999.
receptors. This prevents Dpp signaling from spreading ventrally into
 Patterning the early embryo 65

of simple diffusion, however; instead, it illustrates how a gradient in activity can be

derived from a fairly uniform initial distribution of the protein by interactions with
other proteins. In addition, Decapentaplegic activity is also thought to be modified by
interactions with different forms of its receptors, further refining the activity gradient.
It also interacts with collagen in the extracellular matrix, which restricts its movement
and thus restricts its signaling range. Dpp is a good example of how the formation of
a signal gradient is usually more complicated than simple diffusion.
Producing the sharp transition between the most dorsal ectoderm cells and those
just lateral to it involves the proteins Short gastrulation (Sog), Twisted gastrulation
(Tsg), and Tolloid. Sog and Tsg are BMP-related proteins that can bind Dpp and thus
prevent it binding to its receptors, and in this sense are inhibitory. Tolloid is a metal-
loproteinase that degrades Sog when it is bound to Dpp, thus releasing Dpp.
Sog is expressed throughout the neurectoderm and binds Dpp, preventing its activ-
ity spreading into this region. Sog protein in turn is degraded by Tolloid, which is
expressed throughout the dorsal region; this sets up a gradient in Sog with a high point
in the neuroectoderm and a low point at the dorsal midline and ensures that Sog, along
with its bound Dpp, will be shuttled towards the dorsal region. Degradation of Sog or
Tsg by Tolloid releases free Dpp, which helps to generate a gradient of Dpp activity
with its high point in the dorsal-most region (see Fig. 2.29). Another factor proposed to
sharpen the gradient is the rapid internalization and degradation of Dpp when it binds
to its receptors. Dpp activity is also affected by its synergistic action with another mem-
ber of the TGF-β family, Screw, which forms heterodimers with Dpp that signal more
strongly than homodimers of Dpp or Screw alone; these are likely to be responsible
for most of the Dpp activity. Experimental observations and mathematical modeling
suggest that heterodimers are formed preferentially in the dorsal-most region, which
also helps explain the high Dpp activity here. Homodimers of Dpp and of Screw are
responsible for lower levels of signaling elsewhere in the prospective dorsal ectoderm.
Despite much work on this system however, the contributions of all these components
to generating the gradient in Dpp activity are not completely understood.
The Dpp/Sog pairing has a counterpart in vertebrates, the BMP-4/Chordin partner-
ship (Chordin is the vertebrate homolog of Sog), which is also involved in patterning the
dorso-ventral axis. The nerve cord runs dorsally in vertebrates, however, not ventrally
as in insects, and thus the dorso-ventral axis is reversed in vertebrates compared with
insects; we shall see in Chapter 4 how this reversal in anatomical position is reflected in
a reversal of the pattern of activity of BMP-4 and Chordin during early development.

Summary

Gradients of maternally derived transcription factors along the dorso-ventral and

antero-posterior axes provide positional information that activates zygotic genes at
specific locations along these axes. Along the antero-posterior axis the maternal gra-
dient in Bicoid protein initiates the activation of zygotic gap genes to specify general
body regions. Interactions between the gap genes, all of which code for transcription
factors, help to define their borders of expression. The dorso-ventral axis becomes
divided into four regions: ventral mesoderm, ventral ectoderm (neurectoderm), dorsal
ectoderm (dorsal epidermis), and amnioserosa. A ventral to dorsal gradient of mater-
nal Dorsal protein both specifies the ventral mesoderm and defines the dorsal region;
a second gradient, of the Decapentaplegic protein, specifies the dorsal ectoderm.
Patterning along the dorso-ventral and antero-posterior axes divides the embryo
into a number of discrete regions, each characterized by a unique pattern of zygotic
gene activity.
66 Chapter 2 Development of the Drosophila body plan

Summary: early expression of zygotic genes

Antero-posterior Dorso-ventral

Bicoid protein gradient switches ventro-dorsal gradient of intranuclear

hunchback on at high concentration Dorsal protein forms

Hunchback activates and represses ventral activation of twist, snail, and

gap genes like Krüppel, knirps, giant repression of decapentaplegic

gap-gene products and gap genes interact Decapentaplegic expressed dorsally

to sharpen expression boundaries

axis is divided into unique domains gradient of Decapentaplegic activity

containing different combinations patterns dorsal region
of transcription factors

dorso-ventral axis divided into prospective

mesoderm, neurectoderm, epidermis,
amnioserosa

Activation of the pair-rule genes and the establishment

of parasegments

The most obvious feature of a Drosophila larva is the regular segmentation of the lar-
val cuticle along the antero-posterior axis, each segment carrying cuticular structures
that define it as, for example, thorax or abdomen. This pattern of segmentation is
carried over into the adult, in which each segment has its own identity. Adult append-
ages such as wings, halteres, and legs are attached to particular segments (Fig. 2.30),
but the discernible segments of the mature larva are not in fact the first units of seg-
mentation along this axis. The basic developmental modules, whose definition we
will follow in some detail, are the parasegments, which are specified first and from
which the segments derive.

2.21 Parasegments are delimited by expression of pair-rule genes in a

periodic pattern

The first visible signs of segmentation in the embryo are transient grooves that appear
on the surface of the embryo after gastrulation. These grooves define the paraseg-
ments. There are 14 parasegments, which are the fundamental units in the segmenta-
tion of the Drosophila embryo. Once each parasegment is delimited, it behaves as an
independent developmental unit, under the control of a particular set of genes, and in
this sense at least the embryo can be thought of as being built up piecemeal of mod-
ules. The parasegments are initially similar but each will soon acquire its own unique
identity. The parasegments are out of register with the final segments by about half a
segment; each segment is made up of the posterior region of one parasegment and the
anterior region of the next (see Fig. 2.30). In the anterior head region the segmental
arrangement is lost when some of the anterior parasegments fuse.
The parasegments in the thorax and abdomen are delimited by the action of the
pair-rule genes, each of which is expressed in a series of seven transverse stripes
along the embryo, each stripe corresponding to every second parasegment. When
pair-rule gene expression is visualized by staining for the pair-rule proteins, a striking
zebra-striped embryo is revealed (Fig. 2.31).
 Activation of the pair-rule genes and the establishment of parasegments 67

Parasegments
Early embryo

Fig 2.30 The relationship between

even-skipped
parasegment expression 1 2 3 4 5 6 7 8 9 10 11 12 13 14 parasegments and segments in the early
engrailed a p a p a p a p a p a p a p a p a p a p a p a p a p a p embryo, late embryo, and adult fly. Initially,
parasegment expression
pair-rule genes are expressed in the embryo
engrailed as stripes in every second parasegment.
segment expression p a p a p a p a p a p a p a p a p a p a p a p a p a p a
larval and C1 C2 evenskipped (yellow), for example, is
C3 T1 T2 T3 A1 A2 A3 A4 A5 A6 A7 A8
adult segments expressed in odd-numbered parasegments.
The segmentation selector gene engrailed
(blue) is expressed in the anterior region
of every parasegment, and delimits the
anterior margin of each parasegment. Each
larval segment is composed of the posterior
Late embryo
region of one parasegment and the anterior
C2 region of the next. The anterior region of a
A8
A7 parasegment becomes the posterior portion
C3 A5 A6
C1 T1 T2 T3 A1 A2 A3 A4 of a segment. Segments are thus offset
from the original parasegments by about
half a segment, and engrailed is expressed
T1 T2 A2 A3 A4 A5 in the posterior region of each segment.
In this figure, a and p refer to the anterior
and posterior compartments of segments or
A6 parasegments. The segment specification
Adult A7 is carried over into the adult and results
A8 in particular appendages, such as legs and
C2
wings, developing on specific segments
C3 only. C1, C2, and C3 represent segments
which become fused to form the head region.
T3
T1 T2 T, thoracic segments; A, abdominal segments.
Segments
After Lawrence, P.: 1992.

Anterior Posterior
Fig 2.31 The striped patterns of activity
of pair-rule genes in the Drosophila embryo
just before cellularization. Parasegments
are delimited by pair-rule gene expression,
each pair-rule gene being expressed in
alternate parasegments. Expression of the
pair-rule genes even-skipped (blue) and
fushi tarazu (brown) is visualized by staining
with antibody for their protein products.
even-skipped is expressed in odd-numbered
parasegments, fushi tarazu in even-
numbered parasegments. Scale bar = 0.1 mm.
Photograph from Lawrence, P.: 1992.

The positions of the stripes of pair-rule gene expression are determined by the
pattern of gap-gene expression—a non-repeating pattern of gap-gene activity is con-
verted into repeating stripes of pair-rule gene expression. We now consider how this
is achieved.
68 Chapter 2 Development of the Drosophila body plan

2.22 Gap-gene activity positions stripes of pair-rule gene expression

Pair-rule genes are expressed in stripes, with a periodicity corresponding to alternate

parasegments. Mutations in these genes thus affect alternate segments. Some pair-
rule genes (e.g. even-skipped, eve) define odd-numbered parasegments, whereas others
(e.g. fushi tarazu) define even-numbered parasegments. The striped pattern of expres-
sion of pair-rule genes is present even before cells are formed, while the embryo is
still a syncytium, although cellularization occurs soon after expression begins. Each
pair-rule gene is expressed in seven stripes, each of which is only a few cells wide. For
some genes, such as eve, the anterior margin of the stripe corresponds to the anterior
boundary of a parasegment; the domains of expression of other pair-rule genes, how-
ever, cross parasegment boundaries.
The striped expression pattern appears gradually; the stripes of the eve gene are ini-
tially fuzzy, but eventually acquire a sharp anterior margin. At first sight this type of
patterning would seem to require some underlying periodic process, such as the set-
ting up of a wave-like concentration of a morphogen, with each stripe forming at the
crest of a wave. It was surprising, therefore, to discover that each stripe is specified
independently.
As an example of how the pair-rule stripes are generated we will look in detail
at the expression of the second eve stripe (Fig. 2.32). The appearance of this stripe
depends on the normal expression of bicoid and of the three gap genes hunchback,
Krüppel, and giant (only the anterior band of giant expression is involved in specify-
ing the second eve stripe; see Fig. 2.23). Bicoid and Hunchback proteins are required
to activate the eve gene, but they do not define the boundaries of the stripe. These are
defined by Krüppel and Giant proteins, by a mechanism based on repression of eve.
When concentrations of Krüppel and Giant are above certain threshold levels, eve is
repressed, even if Bicoid and Hunchback are present. The anterior edge of the stripe
is localized at the point of threshold concentration of Giant protein, whereas the pos-
terior border is similarly specified by the Krüppel protein.

bicoid giant hunchback Krüppel

eve
stripe 2

Fig 2.32 The specification of the second

1 2 3 4 5 6
even-skipped (eve) stripe by gap gene
proteins. The different concentrations of Parasegments
transcription factors encoded by the gap
genes hunchback, giant, and Krüppel localize
even-skipped—expressed in a narrow stripe
at a particular point along their gradients—in Anterior Posterior
parasegment 3. Bicoid and Hunchback
proteins activate the gene in a broad domain,
and the anterior and posterior borders are 13 14
formed through repression by Giant and 1 2 3 4 5 6 7 8 9 10 11 12
Krüppel proteins, respectively.
 Activation of the pair-rule genes and the establishment of parasegments 69

In contrast, the positioning of the third and fourth eve stripes depends on regu-
latory regions that are repressed by the high concentration of Hunchback in the
anterior region. The third stripe is expressed about halfway along the embryo,
where Hunchback concentration starts to drop sharply, while the fourth eve stripe is
expressed more posteriorly, where Hunchback is at an even lower level (Fig. 2.33).
The posterior boundary of the third eve stripe is delimited by repression by the gap
protein Knirps.
The independent localization of each of the stripes by the gap-gene transcription
Fig 2.33 Positioning of the third and
factors requires that, in each stripe, the pair-rule gene responds to different con-
fourth eve stripes by the Hunchback
centrations and combinations of the gap-gene transcription factors. The pair-rule
gradient. Embryo stained with fluorescent
genes thus require complex cis-regulatory control regions with multiple binding sites antibodies to simultaneously detect the
for each of the different factors. Examination of the regulatory regions of the eve Hb protein (red) and the mRNA of its target
gene reveals seven separate modules, each controlling the localization of a differ- gene eve (green). The position of the third
ent stripe. Regulatory regions of around 500 base pairs have been isolated that each and fourth eve stripes is due directly to lack
determines the expression of a single stripe (Fig. 2.34). The eve gene is an excellent of repression of eve by Hunchback as its
example of modular cis-regulatory regions controlling the expression of a gene at gradient drops off sharply.
different sites. Photograph from Danyang, Y., and Small, S.:
The presence of control regions, which, when activated, can lead to gene expres- 2008.
sion in a specific position in the embryo, is a fundamental principle controlling gene
action in development. Other examples of this are the localized expression of the gap
genes, and of the genes along the dorso-ventral axis.
Each regulatory region of such a gene contains binding sites for different transcrip-
tion factors, some of which activate the gene, while others repress it. In this way, the
combinatorial activity of the gap genes regulates pair-rule gene expression in each
parasegment. Some pair-rule genes, such as fushi tarazu, may not be regulated by the
gap genes directly, but may depend on the prior expression of primary pair-rule genes,
such as even-skipped and hairy. With the initiation of pair-rule gene expression, the
embryo becomes segmented; it is now divided into a number of unique regions,
which are characterized mainly by the combinations of transcription factors being
expressed in each. These include proteins encoded by the gap genes, the pair-rule
genes, and the genes expressed along the dorso-ventral axis.
The transcription factors encoded by the pair-rule genes set up the spatial frame-
work for the next round of patterning by transcriptional activation. This involves the
further patterning of the parasegments, the development of the final segmentation and
the acquisition of segment identity, which is considered in the following sections.

Binding of gap-gene proteins to one of the regulatory control regions in the promoter of even-skipped

–1550 activators –1070

base base
pairs repressors pairs

Bicoid Hunchback Giant Krüppel

Fig 2.34 Sites of action of activating and repressing transcription factors in the control
region of even-skipped involved in expression of the second even-skipped stripe. A control
region of around 500 base pairs, located between 1070 and 1550 base pairs upstream of the
transcription start site, directs formation of the second even-skipped stripe. Gene expression
occurs when the Bicoid and Hunchback transcription factors are present above a threshold
concentration, with the Giant and Krüppel proteins acting as repressors where they are above
threshold levels. The repressors may act by preventing binding of activators.
70 Chapter 2 Development of the Drosophila body plan

Summary

The activation of the pair-rule genes by the gap genes results in the transformation of
the embryonic pattern along the antero-posterior axis from a variable regionalization to a
periodic one. The pair-rule genes define 14 parasegments. Each parasegment is defined
by narrow stripes of pair-rule gene activity. These stripes are uniquely defined by the local
concentration of gap-gene transcription factors acting on the regulatory regions of the
pair-rule genes. Each pair-rule gene is expressed in alternate parasegments—some in odd-
numbered, others in even-numbered. Most pair-rule genes code for transcription factors.

Summary: pair-rule genes and segmentation

production of local combinations of gap-gene transcription factors

activation of each pair-rule gene in seven transverse stripes along the antero-posterior axis

pair-rule gene expression defines 14 parasegments, each pair-rule gene

being expressed in alternate parasegments

Segmentation genes and compartments

The expression of the pair-rule genes defines the anterior boundaries of all 14 paraseg-
ments but, like the gap genes, their activity is only temporary. Moreover, at this time
the blastoderm becomes cellularized. How, therefore, are the positions of paraseg-
ment boundaries fixed, and how do the final segment boundaries of the larval epi-
dermis become established? This is the role of the segmentation genes. Unlike the
gap genes and pair-rule genes, which encode transcription factors, the segmentation
genes are a diverse group of genes that bear no obvious relation to each other in their
protein products or mechanisms of operation.
Segmentation genes are activated in response to pair-rule gene expression. They
are each expressed in 14 transverse stripes, one stripe corresponding to each paraseg-
ment. During pair-rule gene expression the blastoderm becomes cellularized, so the
segmentation genes are acting in a cellular rather than a syncytial environment. One
of the segmentation genes to be activated by the pair-rule genes is engrailed, which is
expressed in the anterior region of every parasegment. engrailed is of particular inter-
est as its expression delimits a boundary of cell-lineage restriction and so defines a
developmental unit called a compartment, as discussed in the next section. Moreover,
engrailed is also a selector gene, a gene that confers a particular identity on a region
or regions by controlling the activity of other genes, and which continues to act for
an extended period.

2.23 Expression of the engrailed gene delimits a cell-lineage boundary and

defines a compartment

The engrailed gene has a key role in segmentation and, unlike the pair-rule and gap
genes, whose activity is transitory, it is expressed throughout the life of the fly. engrailed
activity first appears at the time of cellularization as a series of 14 transverse stripes.
Figure 2.35 shows expression at the later extended germ-band stage, when part of the
ventral blastoderm (the germ band) has extended over the dorsal side of the embryo.
engrailed is initially expressed in a single line of cells at the anterior margin of each
 Segmentation genes and compartments 71

Fig 2.35 The expression of the engrailed

gene in a late (stage 11) Drosophila
embryo. The gene is expressed in the
anterior region of each parasegment and the
transitory grooves between parasegments
can be seen. At this stage of development,
the germ band has temporarily extended and
is curved over the back of the embryo. Scale
transitory grooves
bar = 0.1 mm.
Photograph from Lawrence, P.: 1992.

parasegment, which is itself only about four cells wide (Fig. 2.36). It is likely that this
periodic pattern of engrailed activity is the result of the combinatorial action of transcrip-
tion factors encoded by pair-rule genes, including fushi tarazu and even-skipped. Evi-
dence that the pair-rule genes do control engrailed expression is provided, for example,
by embryos carrying mutations in fushi tarazu, in which engrailed expression is absent
only in even-numbered parasegments (in which fushi tarazu is normally expressed).
The anterior and posterior margins of the parasegment have a very important
property—­they are boundaries of cell-lineage restriction. Cells and their descen-
dants from one parasegment never move into adjacent ones. Such domains of lineage
restriction are known as compartments. A compartment can be defined as a region in

Fig 2.36 The expression of the pair-rule

genes fushi tarazu (blue), even-skipped
(pink), and engrailed (purple dots) in
parasegments. engrailed is expressed at the
Anterior Posterior anterior margin of each stripe and delimits
Parasegment the anterior border of each parasegment.
boundaries
The boundaries of the parasegments become
sharper and straighter later on.
fushi tarazu even skipped engrailed
After Lawrence, P.: 1992.
72 Chapter 2 Development of the Drosophila body plan

the embryo, larva, or adult that contains all the descendants of the cells present when
Wild-type wing with normal clone
the compartment is set up, and no others.
Anterior The existence of compartments can be detected by cell-lineage studies. Cell lineage
can be followed by marking single cells in the early embryo so that all their descen-
dants (clones) can be identified at later stages of development. One technique is to
inject the egg with a harmless fluorescent compound that is incorporated into all the
cells of the embryo. In the early embryo, a fine beam of ultraviolet light directed onto
a single cell activates the fluorescent compound. All the descendants of that cell will
Posterior be fluorescent and can therefore be identified. Examination of these clones, together
with an analysis of engrailed expression, shows that cells at the anterior margin of
Wild-type wing with Minute clone the parasegment have no descendants that lie on the other side of that margin. The
anterior margin is therefore a boundary of lineage restriction; that is, cells and their
descendants that are on one or other side of the boundary when it is formed cannot
cross it during subsequent development.
compart
ment bo The lineage restriction of the anterior margin of the parasegment is carried over into
undary
the segments of the larva and the adult. But because the anterior part of a paraseg-
ment becomes the posterior part of a segment, the lineage restriction within a seg-
ment is between anterior and posterior regions. The segment is thus divided into
anterior and posterior compartments, with engrailed expression defining the posterior
compartment (see Fig. 2.30).
engrailed mutant wing
In a compartment, the cells may all initially be under common genetic control. Imag-
ine a group of cells becoming divided into two separate regional populations by means
of signals that turn on different genes in each region. If there is never any mixing of the
descendants of the original cells across the boundary the regions are compartments.
Lineage restriction within compartments is clearly illustrated by the behavior of
cells in the wing of the adult fly. It is easier to distinguish lineage restriction in adult
structures than in embryonic and early larval structures, because in adults there has
been considerable cell division since the initial event that determined the compart-
ment in the embryo. Compartments are made visible by making genetically mosaic
Fig 2.37 The boundary between anterior
flies composed of two distinguishable kinds of cells (Box 2E, p. 73). Single cells in
and posterior compartments in the wing the embryonic blastoderm or in larval epidermis are given a distinctive phenotype by
can be demonstrated by marked cell X-ray-or laser-induced mitotic recombination. The fates of all the descendants of this
clones. Top panel: in the wild-type wing, marked cell are then followed. Their behavior depends on the stage of development at
clones marked by mitotic recombination which the founder nucleus was marked. Descendants of nuclei marked at early stages
in the embryo, which gives marked cells during cleavage become part of many tissues and organs, but those of nuclei marked
a phenotype different from other cells of at the blastoderm stage or later have a more restricted fate. They are found only in the
the wing, are too small to demonstrate the anterior or the posterior part of each segment (or of an appendage like a wing), never
compartment boundary. Middle panel: the throughout the whole segment.
use of the Minute technique (see Box 2E,
Cells of imaginal discs divide only about 10 times after the cellular blastoderm
opposite) produces an increased rate of cell
stage; thus clones of experimentally marked cells are small, even in the adult, and
division in the marked cell and gives clones
large enough to determine that cells from
it is not easy to detect a boundary of lineage restriction (Fig. 2.37, top panel). Clone
one compartment do not cross the boundary size can be increased by the Minute technique, which results in the marked cell divid-
into an adjoining compartment. Bottom panel: ing many more times than the other cells (see Box 2E, p. 73). A single clone of such
in the wing of an engrailed mutant in which cells can almost fill either the anterior or posterior part of the wing, and this makes
the engrailed protein is not produced, there the boundary that the cells never cross more evident: this boundary separates the
is no posterior compartment or boundary. anterior and posterior compartments (see Fig. 2.37, middle panel). In a normal wing,
Clones in the anterior part of the wing each compartment is constructed by all the descendants of a set of founding cells. The
cross over into the posterior region and the compartment boundary is remarkably sharp and straight and does not correspond to
posterior region is transformed into a more any structural features in the wing. These experiments also show that the pattern of
anterior-like structure, bearing anterior-type
the wing is in no way dependent on cell lineage. A single marked embryonic cell can
hairs on its margin. The engrailed gene is
give rise to about a twentieth of the cells of the adult wing or, using the Minute tech-
required for the maintenance of the character
of the posterior compartment, and for the
nique to increase clone size, about half the wing. The lineage of the wing cells in each
formation of the boundary. case is quite different, yet the wing’s pattern is quite normal. This is very different
from the nematode C. elegans, where the cell lineages in development are constant.
 Segmentation genes and compartments 73

Box 2E Genetic mosaics and mitotic recombination

Genetic mosaics are embryos derived

Proliferating epithelium in larva
from a single genome but in which there
is a mixture of cells with rearranged X-ray
or inactivated genes. In flies, genetic mutant allele
mosaics can be generated by inducing
rare mitotic recombination events in
the somatic cells of the embryo or larva.
The original method was to induce
chromosome breaks using ­X-rays just wild-type allele
after the chromosomes have replicated
to form two chromatids; this results
Mitotic recombination
in the exchange of material between
homologous chromosomes as the
break is repaired. Today, mitotic recom-
bination is induced in strains of trans-
genic flies that carry the gene for the
yeast recombinase FLP and its target
Mitosis
sequence, FRT, on their chromosomes.
Activation of the recombinase will
result in recombination involving the
FRT sequence. A mitotic recombination
event can generate a single cell with a
unique genetic constitution that will be
inherited by all the cell’s descendants,
which in flies usually form a coherent Proliferation
patch of tissue.
Easily distinguishable mutations like
the recessive multiple wing hairs can
Adult wing epithelium
be used to identify the marked clone; if
a cell homozygous for this mutation is
generated by mitotic recombination in
a heterozygous larva, all of its descen-
dant cells will have multiple hairs (see
Expression in wing
figure). Marked epidermal clones made
by this method are usually small, as
there is little cell proliferation after the
recombination event. Larger clones can
be made using the Minute technique.
The cells in flies carrying a mutation in
the Minute gene grow more slowly than
those in the wild type. By using flies
heterozygous for the Minute mutation,
clones can be made in which the mitotic
recombination event has generated a marked cell that is normal cells and thus see what structures they are able to contribute to.
because it has lost the Minute mutation, and so is wild type. This This can provide information on their state of determination or
normal cell proliferates faster than the slower-growing back- specification at different developmental stages. The technique
ground and thus large clones of the marked cells are produced can also be used to study the localized effects of homozygous
(see Fig. 2.37). The mitotic recombination technique has many mutations that are lethal if present in the homozygous state
applications. If clones of marked cells are generated at differ- throughout the whole animal.
ent stages of development, one can trace the fate of the altered Illustration after Lawrence, P.: 1992.
74 Chapter 2 Development of the Drosophila body plan

The compartment pattern in the adult Drosophila wing is carried over from its initial
specification in the imaginal discs. When epidermal cells are set aside in the ­embryo
to form the imaginal discs, each disc carries over the compartment pattern of the
parasegments from which it arises. A wing disc, for example, develops at the bound-
ary between the two parasegments that contribute to the second thoracic segment.
Thus, the wing is divided into an anterior compartment and a posterior ­compartment,
the compartment boundary (the old parasegment boundary) running in a straight line
more or less down the middle of the wing.
The specification of cells as the posterior compartment of a segment (the anterior
of a parasegment) initially occurs when the parasegments are set up, and is due to
the engrailed gene. Expression of engrailed is required both to confer a ‘posterior seg-
ment’ identity on the cells and to change their surface properties so that they cannot
mix with the cells adjacent to them, hence setting up the parasegment boundary.
Direct evidence for the role of engrailed comes from the behavior of clones of wing
cells in an engrailed mutant (see Fig. 2.37, bottom panel). In the absence of normal
engrailed expression, clones are not confined to anterior or posterior parts of the seg-
ment and there is no compartment boundary. Moreover, in engrailed mutants, the
posterior compartment is partly transformed so that it comes to resemble the pattern
of the anterior part of the wing. For example, bristles normally found only at the ante-
rior margin of the wing are also found at the posterior margin.
The engrailed gene needs to be expressed continuously throughout larval and pupal
stages and into the adult to maintain the character of the posterior compartment of
the segment. Thus, engrailed is also an example of a selector gene—a gene whose
activity is sufficient to cause cells to adopt a particular fate. Selector genes can control
the development of a region such as a compartment and, by controlling the activity of
other genes, give the region a particular identity.

2.24 Segmentation genes stabilize parasegment boundaries and set up a

focus of signaling at the boundary that patterns the segment

Each larval segment has a well-defined antero-posterior pattern, which is easily seen
on the ventral epidermis of the abdomen: the anterior region of each segment bears
denticles, while the posterior region is naked (Fig. 2.38). The rows of denticles, of
which there are six types, make a distinct pattern. Mutations in segmentation genes
often alter the denticle pattern, and this is how such genes were first discovered. For
example, a mutation in a segmentation gene known as wingless (named after its phe-
notype in the adult fly) results in the whole of the ventral abdomen being covered in
denticles, but in the posterior half of each segment the denticle pattern is reversed. In
this mutant, the anterior region of each segment has been duplicated in mirror-image
polarity, and the posterior region is lost.
The larval denticle patterns depend on the correct establishment and maintenance of
the embryonic parasegment boundaries. Establishment of the parasegment ­boundary

Fig 2.38 Each segment of the Drosophila

larva has a characteristic pattern of
denticles on its ventral surface. Denticles
are confined to the anterior regions of
segments, with each segment having its own
pattern.
Photograph from Goodman, R.M., et al.: 2006.
 Segmentation genes and compartments 75

Anterior Parasegment boundaries Posterior

Fig 2.39 The domains of expression of

engrailed, hedgehog
the segmentation genes. At the cellular
blastoderm stage, the engrailed gene is
expressed at the anterior margin of each
wingless
parasegment together with hedgehog,
while wingless is expressed at the posterior
margin. After gastrulation, patched is
patched expressed in all of the cells that express
neither engrailed nor hedgehog. About two
cell divisions occur between the time that
Segment boundaries the parasegments are delimited and the time
the embryo hatches.

depends on an intercellular signaling circuit being set up between adjacent cells, which
delimits the boundary between them. This circuit primarily involves the segmentation
genes wingless, hedgehog, and engrailed, which are expressed in restricted domains
within the parasegment (Fig. 2.39). engrailed, as we saw earlier, encodes a transcrip-
tion factor, while Wingless and Hedgehog are secreted signal proteins that act via
receptor proteins on cell surfaces to alter patterns of gene expression. Wingless is one
of the prototype members of the so-called Wnt family of signal proteins, which are
key players in development in vertebrates, as well as invertebrates, and are involved in
determining cell fate and cell differentiation in many different aspects of development.
Hedgehog similarly has homologs in other animals, including vertebrates.
We have seen how the pair-rule genes first delimit the parasegments by inducing
the expression of engrailed at the anterior margin of each parasegment (see Fig. 2.35).
Cells expressing engrailed now also express the secreted signal Hedgehog, which acti-
vates and maintains expression of the signal protein Wingless in the row of cells
immediately anterior to the Engrailed-expressing cells. The general signaling pathway Fig 2.40 The Hedgehog signaling
by which Hedgehog acts is shown in Fig. 2.40. The secreted Wingless protein in pathway. Left panel: in the absence of
turn provides a signal that feeds back over the boundary to maintain hedgehog and Hedgehog, the membrane protein Patched,
engrailed expression, thus stabilizing and maintaining the compartment boundary which is the receptor for Hedgehog, inhibits
(Fig. 2.41). At parasegment boundaries, and in many of their other developmental the membrane protein Smoothened. In the
absence of Smoothened activity, the
transcription factor Cubitus interruptus
No Hedgehog present: Ci is converted to a repressor Hedgehog present: Ci acts as a gene activator (Ci) is held in the cytoplasm in two protein
Hh complexes—one associated with Smoothened
Smoothened
and another with the protein Suppressor of
Patched
fused (Su(fu)). In the absence of Hedgehog,
Ci in the Smoothened complex is also
phosphorylated by several protein kinases
Cos2 P (glycogen synthase kinase (GSK-3), protein
Ci Fu Cos2 P
kinase A (PKA) and casein kinase 1 (CK1)),
Ci Fu
PKA which results in the proteolytic cleavage of
Ci Su(fu) Ci and formation of the truncated protein
Su(fu) CiRep. This enters the nucleus and acts as a
GSK-3
PKA repressor of Hedgehog target genes. Right
CK1
P Ci panel: Hedgehog (Hh) binding to Patched
P Cytoplasm
P Ci CK1 lifts the inhibition on Smoothened and blocks
production of CiRep. Ci is released from both
Nucleus
proteolytic its complexes in the cytoplasm, enters the cell
cleavage target
CiRep genes nucleus, and acts as a gene activator. Genes
dpp, wg, en activated in response to Hedgehog signaling
include wingless (wg), decapentaplegic (dpp),
and engrailed (en).
76 Chapter 2 Development of the Drosophila body plan

Fig 2.41 Interactions between hedgehog, Wingless (Wg) and Hedgehog (Hh) feed back on In the absence of Wg activity, expression
each other; Wg also promotes expression of Hh and En is lost—leading to a loss of
wingless, and engrailed genes and of Engrailed (En) segment polarity and pattern
proteins at the compartment boundary
control denticle pattern. Top left panel: Hh
the engrailed gene, which encodes a
epidermal cells
homeodomain transcription factor, is
expressed in cells along the anterior margin Wg Hh Hh
En En
of the parasegment, which delimits the future
boundary of the posterior compartment in
the segment; these cells also express the compartment boundary Wg Wg
segmentation gene hedgehog and secrete and anterior boundary
of parasegment
the Hedgehog protein. Hedgehog protein anterior posterior
activates and maintains expression of the compartment compartment
of segment of segment
segmentation gene wingless in adjacent cells
across the compartment boundary. Wingless
protein feeds back on engrailed-expressing
cells to maintain expression of the engrailed
gene and hedgehog. These interactions
stabilize and maintain the compartment
boundary. Top right panel: in mutants where
the wingless gene is inactivated and the
Wingless protein is absent, neither hedgehog
nor engrailed genes are expressed. This leads
to loss of the compartment boundary and the
normally well defined pattern of denticles
within each abdominal segment. Bottom left
panel: in the wild-type larva the denticle
bands in the ventral cuticle are confined to
the anterior part of the segment and are
dependent on the activity of hedgehog and
wingless genes. Bottom right panel: in the
wingless mutant, denticles are present across
segment
the whole ventral surface of the segment in
what looks like a mirror-image repeat of the
anterior segment pattern.
A P A P
Photographs from Lawrence, P.: 1992.

functions, Wingless and other Wnt proteins act through the signaling pathway illus-
trated in Box 1E (p. 26), which is often called the canonical Wnt/β-catenin ­pathway
(β-catenin is called Armadillo in Drosophila). A computer model of the Wingless–
Hedgehog–Engrailed circuitry shows that it is remarkably robust and resistant to
­variation, such as variations in the levels of expression of the genes governing the
circuit’s behavior. Some time after the parasegment boundary has been established,
a deep groove develops at the posterior edge of each engrailed stripe and marks the
edge of each segment. This means that engrailed expression marks the posterior com-
partment of the larval segment.
In the embryo, therefore, the continuous epithelium of the abdomen becomes divided­
up into a succession of anterior and posterior compartments, with the earlier paraseg-
ment boundary forming the boundary between the anterior and posterior compartments
of a segment (see Fig. 2.30). Once compartment boundaries are consolidated, signaling
from the parasegment boundary sets up the pattern of each segment, leading eventually
to the differentiation of the epidermal cells to produce the cuticle pattern evident on the
larva. The pattern of denticle belts is specified in response to the signals provided by
Wingless and Hedgehog. At this later stage of embryonic development, Hedgehog and
Wingless expression are no longer dependent on each other and can be analyzed sepa-
rately. Wingless protein moves posteriorly over the compartment ­boundary to pattern
 Segmentation genes and compartments 77

the posterior compartment of the segment, and also moves anteriorly to pattern the
a1
anterior part of the segment immediately anterior to the cells in which it is expressed.
a2
Wingless moves a shorter distance posteriorly than anteriorly, as it is more rapidly
degraded in the posterior compartment, and so its effects on patterning extend over
A
a longer range in the anterior compartment. In the anterior compartment, ­Wingless a3
represses genes required for denticle formation, and the cells that lie immediately ante-
rior to the Wingless-expressing line of cells are specified as epidermal cells that will a4
produce smooth cuticle, while the anterior extent of Wingless signaling delimits the a5
a6
posterior edge of the denticle belts in that segment. In mutants that lack wingless func-
P
tion, denticles are produced in the normally smooth regions (see Fig. 2.41).
Hedgehog signaling does not pattern the posterior segment compartment in which
Hedgehog is actually expressed, as posterior compartment cells are ‘blind’ to its
Fig 2.42 The epidermis of each segment
action. The Hedgehog signals move anteriorly to maintain Wingless function and also
in the abdomen of the adult Drosophila
posteriorly to help pattern the anterior compartment of the next segment. The signals
is divided into an anterior and a posterior
provided by Hedgehog and Wingless lead, by a complex set of interactions, to the
compartment. A number of different regions,
expression of a number of genes in non-overlapping narrow stripes that specify the a1 to a6, can be distinguished in the anterior
larval segmental pattern of smooth cuticle or specific rows of denticles. compartment, characterized by different
Patterning within compartment boundaries is also visible in the pattern of the epi- bristles, pigmentation, and gene expression.
dermis on the abdominal segments of the adult fly (Fig. 2.42). The epidermis of each A, anterior compartment; P, posterior
segment is subdivided into a number of types—at least five in the anterior compart- compartment.
ment and three in the posterior compartment—which are distinguished by the pres-
ence or absence of pigmentation and the positioning of bristles and hairs.

2.25 Insect epidermal cells become individually polarized in an antero-

posterior direction in the plane of the epithelium

The insect epidermis in each segment not only becomes patterned into bands of epider-
mal cells of different types, but each cell acquires an individual antero-posterior polar-
ity, reflected in the fact that the hairs and bristles on the adult Drosophila abdomen all
point backwards (see Fig. 2.42). Many different types of cells have polarity, which sim-
ply means that one end of the cell is structurally or functionally different from the other.
For example, polarization is clearly seen in cells undergoing chemotaxis in response
to a gradient of a chemoattractant, where the advancing end of the cell is quite differ-
ent from the posterior end. This type of cell polarity is called planar cell polarity, to
distinguish it from the apical–basal polarity often present in an epithelial sheet. Planar
cell polarity is also clearly visible in the Drosophila wing, which is composed of a thin
sheet of epithelium. The wing epidermal cells all bear hairs and these hairs form at the
ends of the cells away from the body—the distal ends of the cells—and all point away
from the body (see Fig. 2.37). Another example is the insect compound eye, where the
individual units (ommatidia) in each half of the eye point away from the midline (see
Chapter 11), which is the result of the establishment of planar polarity in the developing
photoreceptor cells in the eye imaginal disc. In vertebrates, examples of planar polarity
occur in the stereocilia in the inner ear and in the scales of fish. Planar cell polarity
is also important in morphogenetic cell movements such as the tissue extension that
occurs at gastrulation in Xenopus, when the spherical frog blastula is being converted
into the more elongated gastrula; this will be discussed in Chapter 8.
The mechanism by which planar cell polarity is established is not yet completely
understood. Current hypotheses propose the existence of an informational gradi-
ent across the tissue that sets the overall direction of polarity, while local signaling
between one end of a cell and the other, and between one cell and its neighbor are
thought to set the polarity of the individual cell (see Box 2F, p. 78). In the case of the
adult insect abdominal epidermis, for example, the signaling gradients set up at com-
partment boundaries have been proposed to have a role in establishing planar cell
polarity, as well as in patterning the segment.
78 Chapter 2 Development of the Drosophila body plan

Box 2F Planar cell polarity in Drosophila

Cells in an epithelium have a dis- a c

tinct apical–basal polarity, but more A
relevant to patterning in develop-
ment is polarity in the plane of the
Fz–
epithelium—­planar cell polarity.
This is visible in the epidermis of
the adult fly, where the abdominal b
epidermis has hairs all pointing in
a posterior direction (see figure,
panel a), and the epidermis of the P
wing has hairs all pointing in a
distal direction—­away from the proximal base of the wing and known as Strabismus). An asymmetric­ interaction across the
towards its tip. Another example of planar cell polarity in Droso- junction between the Fz:Fmi and Fmi:Vang complexes seems
phila is the orientation of the ommatidia in the insect compound to be central to establishing cell polarity—by some mechanism
eye (see Chapter 11). that is still to be determined. The cytoplasmic signaling pro-
The mechanism for establishing planar polarity is not fully teins Dishevelled (Dsh) and Diego (Dgo) become associated with
understood, especially how the axis of polarity across the tissue Fz:Fmi and the cytoplasmic protein Prickled (Pk) associates with
is determined. At the cellular level, however, some key proteins Fmi:Vang (see lower figure). These cytoplasmic proteins seem
involved have been identified. The cell-surface receptor Frizzled not to be essential for establishing cell polarity per se but may
(Fz) is one. In the fly abdominal epidermis, if Fz is absent then tis- act to convert the polarity signal into some of the manifesta-
sue polarity is disrupted and the hairs point in random directions tions of polarity within the cell.
(see top figure, panel b). But if the gene for Fz is turned off in Yet another set of proteins, which include the membrane pro-
just a small clone of epidermal cells, the unaffected cells immedi- tein Four-jointed, and the atypical cadherins Dachsous and Fat,
ately posterior to them reverse polarity, showing that polarity is have also been implicated in determining planar cell polarity.
influenced by intercellular—cell to cell—communication (see figure, They form gradients along the axis of polarity in tissues, and one
panel c; red arrows indicate the reversed direction; the black arrow proposed role for these proteins has been as long-range graded
shows the normal direction of polarity from anterior (A) to poste- extracellular signals that set the direction of polarity by interact-
rior (P)). We met Fz earlier as the receptor for Wingless, but in the ing in some way with the proteins described above.
case of planar cell polarity it does not act through the canonical However, there is evidence that this system can operate inde-
Wnt pathway shown in Box 1E (p. 26) and is unlikely to be stimu- pendently in parallel with the Fz-Fmi-Vang system rather than
lated by Wingless or other Drosophila Wnts. in a strict linear relationship. A clone of cells overexpressing
Fz is thought to have at least three functions in planar cell polar- Dachsous or other proteins of that group can repolarize adjacent
ity. The first is to receive long-range patterning information from cells even when those cells lack both Fz and Fmi and so cannot
upstream cues, which might involve the generation of a gradient of form the junctional complexes. Unraveling planar cell polarity is
Fz activity across the whole axis of the tissue. Clonal repolarization still a fascinating work in progress.
experiments like those shown in the figure above suggest that the Determining cell polarity is an important aspect of development
direction arrow of polarity points down the Fz gradient; cells point in all animals. We shall see in Chapter 8, for example, how some of
their hairs in the direction of a lower level of Fz and away from a the same proteins regulate the changes in cell shape and polarity
higher level. Second, Fz is involved in the cell–cell communication that help to reshape vertebrate embryos during gastrulation.
that locally coordinates cell polarity. The third function of Fz is to
provide an intracellular cue for the growth of the hairs.
It is not yet known how the direction of polarity is set across
a tissue, but it is now clear that key interactions that determine
and maintain the polarity of individual cells take place at the
epidermal cell junctions. An asymmetry between the two sides
of a junction is established by interactions between proteins in
the adjacent cell membranes. In proximal to distal polarity in the Fmi Fmi
wing, for example, Fz becomes associated with an atypical cad- Dsh Fz Vang
herin called Flamingo (Fmi, also known as Starry Night) at the Dgo Pk
distal end of each epidermal cell, while at the proximal end, Fmi
Distal Proximal
is associated with the membrane protein Van Gogh (Vang, also
 Segmentation genes and compartments 79

Abnormal Oncopeltus with a

Normal Oncopeltus Abnormal hair pattern
gap in the segment boundary

Hair pattern Hair pattern

Segment
boundary
A P A P

Gradient interpretation Gradient interpretation

A P A P

Fig 2.43 Gradients could specify polarity

A natural case where loss of planar cell polarity can be observed is in adults of
in the segments of Oncopeltus. The hairs
the plant-feeding bug Oncopeltus, which, like Drosophila, have hairs covering each on the cuticle point in a posterior direction
­segment, all pointing in a posterior direction (Fig. 2.43, left panel). In Oncopeltus, some and this may reflect an underlying gradient
individuals naturally have a gap in the segment boundary; this results in a rather pre- in a morphogen that is partly maintained
cise pattern of hairs with altered orientation, many of the hairs near the gap pointing by the segment boundary (left panels).
in the reverse direction (Fig. 2.43, center and right panels). If the slope of a gradient When there is a gap in the boundary (center
within each segment gives the hairs their polarity, a gap in a segment boundary would panels) the sharp discontinuity in morphogen
result in a local change in the gradient, the sharp change in concentration at the normal concentration smoothes out, with a resulting
boundary being smoothed out and forming a local gradient running in the opposite local reversal of the gradient and of the
direction in which the hairs point. This is
direction. This would result in the hairs in this region pointing in the opposite direction.
illustrated in the photograph (right panel).
Mutations that cause similar reversals in polarity can be induced in clones of cells in
the Drosophila abdominal epidermis, and can be similarly explained by the disruption
of an informational gradient set up within a compartment. Genes essential for planar
cell polarity have been identified in Drosophila and other organisms, together with
some of the proteins involved in setting up the long-range gradient and in the signaling
pathways that interpret it to set individual cell polarity (see Box 2F, p. 78).

2.26 Some insects use different mechanisms for patterning the body plan

Drosophila belongs to an evolutionarily advanced group of insects, in which all the Long-germ insect fate map
segments are specified more or less at the same time in development, as shown both
by the striped patterns of pair-rule gene expression in the acellular blastoderm and
the appearance of all the segments shortly after gastrulation. This type of develop-
ment is known as long-germ development, as the blastoderm corresponds to the H Ab
Th
whole of the future embryo. Many other insects, such as the flour beetle Tribolium,
have a short-germ development. In short-germ development the blastoderm is short
and forms only the anterior segments. The posterior segments are added by growth
after completion of the blastoderm stage and gastrulation (Fig. 2.44). In spite of early
differences between them, the mature germ band stages of both long- and ­short-germ Short-germ insect fate map

Fig 2.44 Differences in the development of long-germ and short-germ insects. Top panel:
the general fate map of long-germ insects such as Drosophila shows that the whole of the
body plan—head (H), thorax (Th), and abdomen (Ab)—is present at the time of initial germ- Gz
band formation. Bottom panel: in short-germ insects, only the anterior regions of the body H Th
plan are present at this embryonic stage. Most of the abdominal segments develop later, after
gastrulation, from a posterior growth zone (Gz).
80 Chapter 2 Development of the Drosophila body plan

Krüppel hairy
Fig 2.45 Gap and pair-rule gene
expression in long-germ and short-
Drosophila
germ insects at the time of germ-band
formation. Krüppel (red) is a gap gene and
hairy (green) is a pair-rule gene. The position 1 2 3 4 5 6 7
of the Krüppel stripe in the short-germ
embryo of Tribolium indicates that posterior
regions of the body are not yet present
Tribolium A P
at this stage. Similarly, only three hairy
stripes, corresponding to the first three hairy 3
stripes in a long-germ embryo, are present in 2
1
Tribolium.

insect embryos look similar. This, therefore, is a common stage through which
all insect embryos develop. In the long-germ embryos of the wasp Nasonia, gradients
in the protein Orthodenticle pattern the anterior and posterior regions, in a manner
similar to the action of Bicoid in Drosophila.
The question clearly arises as to what processes are common to the specification
of the body plan in long- and short-germ insects. One clear difference is that whereas
the patterning of the body plan in the long-germ band of Drosophila takes place before
cell boundaries form, much of the body plan in short-germ insects is laid down at a
later stage, when the posterior segments are generated during growth. At this stage
the embryo is multicellular. So are the same genes involved?
There is very good evidence that the same genes and development processes are
involved in the patterning of Tribolium and Drosophila. For example, the gap gene
Krüppel is expressed at the posterior end of the Tribolium embryo at the blastoderm
stage and not in the middle region, as in Drosophila (Fig. 2.45). It therefore seems to
be specifying the same part of the body in the two insects. Similarly, only two repeats
of the pair-rule stripes are present at the blastoderm stage, together with a posterior
cap of pair-rule gene expression, in contrast to the seven in Drosophila. The genes
wingless and engrailed are also expressed in a similar relation to that in Drosophila.
Although not many genes have been studied in detail in other insects, at least one,
the gene engrailed, is known to be expressed in the posterior region of segments in a
variety of insects. The pair-rule gene even-skipped (see Section 2.21), while present in
the grasshopper (a short-germ insect), may not have a similar role in segmentation. It
is, however, involved later in the development of the nervous system in the grasshop-
per, and is also expressed at the posterior end of the growing germ band.
Differences in early development are much more dramatic in some other insects.
In certain parasitic wasps, the egg is small and undergoes cleavage to form a ball of
cells, which then falls apart. Each of the resulting small clusters of cells—there can
be as many as 400—can develop into a separate embryo. This wasp’s development
apparently does not depend on maternal information to specify the body axes, and in
this respect resembles that of the early mammalian embryo.

Summary

The segmentation genes are involved in patterning the parasegments. One of the first
genes to be activated is engrailed, which is expressed at the anterior margin of each
parasegment, delineating a group of cells which becomes the posterior compartment
of a segment. engrailed is also a selector gene in that it provides a group of cells with
a long-term regional identity. engrailed expression shows lineage restriction: cells
expressing engrailed define the posterior compartment of a segment, and cells never
cross from the posterior into the anterior compartment. engrailed is turned on by the
 Specification of segment identity 81

pair-rule genes, and its expression is maintained by the segmentation genes wingless
and hedgehog, which stabilize the compartment boundary. The compartment boundary
is involved in specifying pattern and cell polarity within the segment. Studies in other
insects also suggest that a discrete gradient of positional information is set up in each
segment, delimited by the boundaries, and this gradient patterns the segment. Adult
epidermal cells are polarized. Hairs in the adult fly body and wing point along an axis,
and this is due to a mechanism that gives rise to planar cell polarity. In contrast to
Drosophila, some insects have a short-germ pattern of development in which posterior
segments are added by growth after the cellular blastoderm stage.

Summary: gene expression defines segment compartments

pair-rule gene expression

segmentation and selector gene engrailed activated in anterior of each parasegment,

defining anterior compartment of parasegment and posterior compartment of segment

engrailed-expressing cells also express segmentation gene hedgehog

cells on other side of compartment boundary express segmentation gene wingless

engrailed expression maintained and compartment boundary stabilized by

Wingless and Hedgehog proteins

compartment boundary provides signaling center from which segment is patterned

Specification of segment identity

Each segment has a unique identity, most easily seen in the larva in the characteristic
pattern of denticles on the ventral surface. Since the same segmentation genes are
turned on in each segment, what makes the segments different from each other? Their
identity is specified by a class of master regulatory genes, known as homeotic ­selector
genes, which set the future developmental pathway of each segment. A selector gene
controls the activity of other genes and is required throughout development to main-
tain this pattern of gene expression. Homologs of the Drosophila selector genes that
control segment identity have subsequently been discovered in virtually all other
animals, where they broadly control identity along the antero-posterior axis, as we
shall see in Chapter 5.

2.27 Segment identity in Drosophila is specified by Hox genes

The first evidence for the existence of genes that specify segment identity came from
unusual and striking mutations in Drosophila that produced homeotic transforma-
tions—the conversion of one segment into another. Eventually the genes that produced
these mutations were identified and the intricate way in which they specified segment
identity was worked out. The discovery of these genes, now called the Hox genes, had
an enormous impact on developmental biology. Homologous genes were found in other
animals where they acted in a similar way, essentially specifying identity along the
antero-posterior axis. Hox genes are now considered as one of the fundamental defining
82 Chapter 2 Development of the Drosophila body plan

Fig 2.46 The Antennapedia and bithorax

homeotic selector gene complexes. The Anterior Posterior
order of the genes from 3′to 5′in each
complex reflects both the order of their 3’ lab pb Dfd Scr Antp Ubx abd-A Abd-B 5’
spatial expression (anterior to posterior) and
the timing of expression (3′earliest).
Antennapedia complex bithorax complex

lab labial Ubx Ultrabithorax

pb proboscipedia abd-A abdominal-A
Dfd Deformed Abd-B Abdominal-B
Scr Sex combs reduced
Antp Antennapedia

Wild type

T2 features of animals. In 1995, the American geneticist Edward Lewis shared the Nobel
wing
A Prize in Physiology or Medicine for his pioneering work on the Drosophila homeotic
gene complexes and how they function. In Drosophila, the Hox genes are organized
into two gene clusters, which together make up the HOM-C complex (Fig. 2.46; see also
Box 5E, p. 194). The Hox genes all encode transcription factors and take their name
from the distinctive homeobox DNA motif that encodes part of the DNA-binding site
P
of these proteins.
The two homeotic gene clusters in Drosophila are called the bithorax complex and
T3 haltere the Antennapedia complex and are named after the mutations that first revealed
A their existence. Flies with the bithorax mutation have part of the haltere (the balanc-
ing organ on the third thoracic segment) transformed into part of a wing (Fig. 2.47),
P whereas flies with the dominant Antennapedia mutation have their antennae trans-
formed into legs. Genes identified by such mutations are called ­homeotic genes
because when mutated they result in homeosis—the transformation of a whole seg-
bithorax mutant ment or structure into another related one, as in the transformation of antenna to
leg. These bizarre transformations arise out of the homeotic selector genes’ key role
T2 wing as identity specifiers. They control the activity of other genes in the segments, thus
determining, for example, that a particular imaginal disc will develop as wing or hal-
tere. The bithorax complex controls the development of parasegments 5–14, while the
Antennapedia complex controls the identity of the more anterior parasegments. The
action of the bithorax complex is the best understood and will be discussed first.

2.28 Homeotic selector genes of the bithorax complex are responsible for
T3 diversification of the posterior segments
transformed anterior compartment
The bithorax complex of Drosophila comprises three homeobox genes: Ultrabithorax
(Ubx), abdominal-A (abd-A), and Abdominal-B (Abd-B). These genes are expressed in
the parasegments in a combinatorial manner (Fig. 2.48, top panel). Ubx is expressed
posterior of haltere in all parasegments from 5 to 12, abd-A is expressed more posteriorly in paraseg-
ments 7–13, and Abd-B more posteriorly still from parasegment 10 onwards. Because

Fig 2.47 Homeotic transformation of the wing and haltere by mutations in the bithorax
complex. Top panel: in the normal adult, both wing and haltere are divided into an anterior
(A) and a posterior (P) compartment. Middle panel: the bithorax mutation transforms the
anterior compartment of the haltere into an anterior wing region. The mutation postbithorax
acts similarly on the posterior compartment, converting it into posterior wing (not illustrated).
Bottom panel: if both mutations are present together, the effect is additive and the haltere is
transformed into a complete wing, producing a four-winged fly.
Photograph courtesy of E. Lewis, from Bender, W., et al.: 1983. Illustration after Lawrence,
P.: 1992.
 Specification of segment identity 83

the genes are active to varying extents in different parasegments, their combined
activities define the character of each parasegment. Abd-B also suppresses Ubx, such
that Ubx expression is very low by parasegment 14, as Abd-B expression increases. Wild type
The pattern of activity of the bithorax complex genes is determined by gap and pair- Parasegments
7 8 9 10 11 12 13
6 14
rule genes. 5
4
The role of the bithorax complex was first indicated by classical genetic experi- 3

ments. In larvae lacking the whole of the bithorax complex (see Fig. 2.48, second
A8
panel), every parasegment from 5 to 13 develops in the same way and resembles T1 T2 T3 A1 A2 A3 A4 A5 A6
A7
Segments
parasegment 4. The bithorax complex is therefore essential for the diversification of
Abdominal-B
these segments, whose basic pattern is represented by parasegment 4. This paraseg-
abdominal-A
ment can be considered as being a type of ‘default’ state, which is modified in all the Ultrabithorax
parasegments posterior to it by the proteins encoded by the bithorax complex. It is
because the genes of the bithorax complex are able to superimpose a new identity on Ultrabithorax, abdominal-A,
the default state that they are called selector genes. and Abdominal-B absent
An indication of the role of each gene of the bithorax complex can be deduced by 4 4 4 4 4
4 4 4 14
looking at embryos constructed so that the genes are put one at a time into embryos 4
4
lacking the whole complex (see Fig. 2.48, bottom three panels). If only the Ubx gene 3

is present, the resulting larva has one parasegment 4, one parasegment 5, and eight A9
T1 T2 T2 T2 T2
parasegments 6. Clearly, Ubx has some effect on all parasegments from 5 backward T2 T2 T2 T2 T2 T2
and can specify parasegments 5 and 6. If abd-A and Ubx are put into the embryo,
then the larva has parasegments 4, 5, 6, 7, and 8, followed by five parasegments 9.
So abd-A affects parasegments from 7 backward, and in combination, Ubx and abd-A
can specify the character of parasegments 7, 8, and 9. Similar principles apply to
Abd-B, whose domain of influence extends from parasegment 10 backward, and Ultrabithorax added back
which is expressed most strongly in parasegment 14. Differences between individual
6 6 6 6 6
6 6 14
segments may reflect differences in the spatial and temporal pattern of Hox gene 5
6
4
expression. 3

These results illustrate an important principle, namely that the character of the
parasegments is specified by the genes of the bithorax complex acting in a combi- T1 T2 T3 A1
A1 A9
A1 A1 A1 A1 A1
natorial manner. Their combinatorial effect can also been seen by taking the genes
away one at a time, from the wild type. In the absence of Ubx, for example, paraseg-
ments 5 and 6 become converted to parasegment 4 (see Fig. 2.48, bottom panel).
There is a further effect on the cuticle pattern in parasegments 7–14 in that structures
characteristic of the thorax are now present in the abdomen, showing that Ubx exerts Ultrabithorax and abdominal-A
an effect in all these segments. Such abnormalities may result from expression of added back
‘nonsense’ combinations of the bithorax genes. For example, in such a mutant, abd-A 8 9 9 9 9 9 14
6 7
protein is present in parasegments 7–9 without Ubx protein, and this is a combina- 4
5

tion never found normally. The site of appendages, such as legs, is determined by the 3

Hox genes. For example, expression of the genes Antp and Ubx specifies the segments
T1 T2 T3 A9
from which the second and third pair of legs, respectively, will emerge. The down- A1
A2 A3 A4 A4 A4
A4 A4

stream targets of Hox proteins are beginning to be identified, and there may be a very
large number of genes whose expression is affected.

Fig 2.48 The spatial pattern of expression of genes of the bithorax complex
characterizes each parasegment. In the wild-type embryo (top panel) the expression of Ultrabithorax only absent
the genes Ultrabithorax, abdominal-A, and Abdominal-B are required to confer an identity
on each parasegment. Mutations in the bithorax complex result in homeotic transformations 4 7 8 9 10 11 12 13
14
4
in the character of the parasegments and the segments derived from them. When the 4
3
bithorax complex is completely absent (second panel), parasegments 5–13 are converted into
nine parasegments 4 (corresponding to segment T2 in the larva), as shown by the denticle
T1 T2 T2 A8
and bristle patterns on the cuticle. The three lower panels show the transformations caused T2 A7
A2 A3 A4 A5 A6
by the absence of different combinations of genes. When the Ultrabithorax. gene alone is
absent (bottom panel), parasegments 5 and 6 are converted into 4. In each case the spatial
extent of gene expression is detected by in situ hybridization (see Box 1D, p. 20). Note that the
specification of parasegment 14 is relatively unaffected by the bithorax complex.
84 Chapter 2 Development of the Drosophila body plan

While both the gap and pair-rule proteins initially control the pattern of Hox gene
expression, these proteins disappear within about 4 hours. The continued correct
expression of the Hox genes involves two groups of genes—the Polycomb and Tritho-
rax groups. The proteins of the Polycomb group maintain transcriptional repression of
Hox genes where they are initially off, while the Trithorax group maintain expression
in those cells where Hox genes have been turned on.

2.29 The Antennapedia complex controls specification of anterior regions

The Antennapedia complex comprises five homeobox genes (see Fig. 2.46), which
control the behavior of the parasegments anterior to parasegment 5 in a man-
ner similar to the bithorax complex. Since there are no novel principles involved,
we just mention its role briefly. Several genes within the complex are critically
involved in specifying particular parasegments. Mutations in the gene Deformed
affect the ectodermally derived structures of parasegments 0 and 1, those in Sex
combs reduced affect parasegments 2 and 3, and those in the Antp gene affect
parasegments 4 and 5, the parasegments that produce the leg imaginal discs. The
Antennapedia mutation that transforms antennae into legs in the adult fly results
from the misexpression of the Antp gene in anterior segments in which it is not
normally expressed.

2.30 The order of Hox gene expression corresponds to the order of genes
along the chromosome

The bithorax and Antennapedia complexes possess some striking features of gene
organization. Within both, the order of the genes in the complex is the same as the
spatial and temporal order in which they are expressed along the antero-posterior
axis during development. Ubx, for example, is on the 3′side of abd-A on the chromo-
some, is more anterior in its pattern of expression, and is activated earlier. As we will
see, the related Hox gene complexes of vertebrates, whose ancestors diverged from
those of arthropods hundreds of millions of years ago, show the same correspon-
dence between gene order and order of expression. This highly conserved temporal
and spatial co-linearity must be related to the mechanisms that control the expres-
sion of these genes.
The complex yet subtle control to which the genes of the bithorax complex are
subject is seen in experiments in which production of Ultrabithorax protein is forced
in all segments. This targeted gene misexpression is achieved by linking the Ultra-
bithorax protein-coding sequence to a heat-shock promoter—a promoter that is acti-
vated at 29°C—and introducing the novel DNA construct into the fly genome using
a P element (see Box 2D, p. 61). When this transgenic embryo is given a heat shock
for a few minutes, the extra Ubx gene is transcribed and the protein is made at high
levels in all cells. This has no effect on posterior parasegments (in which Ubx pro-
tein is normally present), with the exception of parasegment 5 which, for unknown
reasons, is transformed into parasegment 6 (this may reflect a quantitative effect of
the Ubx protein). However, all parasegments anterior to 5 are also transformed into
parasegment 6. While this seems a reasonably simple and expected result, consider
what happens to parasegment 13.
Transcription of Ubx is normally suppressed in parasegment 13 in wild-type embryos,
yet when the protein itself is produced in the parasegment as a result of heat shock,
it has no effect. By some mechanism, the Ultrabithorax protein is rendered inactive
in this parasegment. This phenomenon is quite common in the specification of the
parasegments and is known as ‘phenotypic suppression’ or ‘posterior dominance’: it
means that Hox gene products normally expressed in anterior regions are suppressed
by more posterior products.
 Specification of segment identity 85

While the roles of the bithorax and Antennapedia complexes in controlling seg-
ment identity are well established, we do not yet know very much about how they
interact with the genes acting next in the developmental pathway. These are the genes
that actually specify the structures that give the segments their unique identities.
What, for example, is the pathway that leads to a thoracic rather than an abdominal
segment structure? Does each Hox gene activate just a few or many target genes? Do
target genes with different functions act in concert to form a particular structure?
Answers to these questions will help us understand how changes in a single gene can
cause a homeotic transformation like antenna into leg. The role of the Hox genes in
the evolution of different body patterns and in the evolution of limbs from fins will be
considered in Chapter 15.

2.31 The Drosophila head region is specified by genes other than the
Hox genes

As mentioned earlier in the chapter, the Drosophila larval head anterior to the man-
dibles is formed by the three most anterior parasegments. The segmental structure is
also evident in the embryonic brain, which is composed of three segments (neurom-
eres). The specification of the anterior head and embryonic nervous system is not,
however, under the control of either the pair-rule genes or the Hox genes. Instead,
it appears to be determined at the blastoderm stage by the overlapping expression
of the genes orthodenticle, empty spiracles, and buttonhead, which all encode gene-
regulatory proteins and resemble the gap genes in their effect on phenotype; they
are sometimes referred to as the ‘head gap genes.’ Unlike the gap genes regionalizing
the trunk and abdomen, however, they do not regulate each other’s activity. Nor
is it known whether they work in a combinatorial way like the Hox genes to also
confer specific identities on the head segments. Orthodenticle and empty spiracles
have homologs that carry out similar functions in vertebrates, and it is clear that
this head-specification system, like the Hox genes, is of ancient evolutionary origin
in animals.
The genes orthodenticle and empty spiracles are examples of homeodomain tran-
scription factors that lie outside the Hox gene clusters. The DNA-binding homeodo-
main is not confined to the proteins encoded by the Hox gene clusters, and similar
domains are found in many other transcription factors involved in development in
both Drosophila and vertebrates.

Summary

Segment identity is conferred by action of selector or homeotic genes, which direct the
development of different parasegments so that each acquires a unique identity. Two
clusters of selector genes, known generally as the Hox genes, are involved in specifi-
cation of segment identity in Drosophila: the Antennapedia complex, which controls
parasegment identity in the head and first thoracic segment, and the bithorax complex,
which acts on the remaining parasegments. Segment identity seems to be determined
by the combination of Hox genes active in the segment. The spatial pattern of expres-
sion of selector genes along the embryo body is initially determined by the preceding
gap gene activity, but selector genes need to be switched on continuously throughout
development to maintain the required phenotype. Mutations in genes of the Antenna-
pedia and bithorax complexes can bring about homeotic transformations, in which one
segment or structure is changed into a related one, for example an antenna into a leg.
The Antennapedia and bithorax complexes are remarkable in that gene order on the
chromosome corresponds to the spatial and temporal order of gene expression along
the body.
86 Chapter 2 Development of the Drosophila body plan

Summary: homeotic genes and segment identity

gap and pair-rule gene expression delimits parasegments

selector genes of the HOM complex are expressed along the antero-posterior axis
in an order co-linear with the order of genes on the chromosome

Antennapedia complex bithorax complex

lab, pb, Dfd, Scr, Antp Ubx, abd-A, Abd-B

specify segment identity specify segment identity in second

in head and and third thoracic segments and
first thoracic segment abdominal segments

Summary to Chapter 2

The earliest stages of development in Drosophila take place when the embryo is a multinu-
cleate syncytium. During oogenesis, the maternal gene products are deposited in the egg
in a particular spatial pattern during oogenesis. This pattern will determine the main body
axes and lay down a framework of positional information which, after fertilization, acti-
vates a cascade of zygotic gene activity that further patterns the body. The first zygotic
genes to be activated along the antero-posterior axis are the gap genes, all encoding tran-
scription factors, whose pattern of expression divides the embryo into a number of regions.
The transition to a segmented body organization then starts with the activity of the pair-
rule genes, whose sites of expression are specified by the gap gene proteins and which
divide the axis into 14 parasegments. Along the dorso-ventral axis, zygotic gene expres-
sion also defines several regions, including the future mesoderm and future neural tissue.
At the time of pair-rule gene expression, the embryo becomes cellularized and is no longer
a syncytium. Segmentation genes pattern the segment and compartment boundaries are
involved in patterning and polarizing segments. The identity of the segments is deter-
mined by two gene complexes which contain homeotic selector genes, known generally as
the Hox genes. The spatial pattern of expression of these genes is largely determined by
gap gene activity.

n End of chapter questions 4. Developmental genes act hierarchically during pattern forma-
tion, first defining broad regions, which are secondarily refined to
Long answer (concept questions) form a larger number of smaller regions (see Section 2.6). Sum-
marize how this general principle is illustrated by early Drosophila
1. As we will see in future chapters, cleavage divides the eggs of
embryogenesis.
vertebrates and many other animals into a cluster of separate cells.
In contrast, the equivalent stage in the Drosophila embryo generates 5. The anterior end of the embryo is established by a locally high
a syncytium. What is a syncytium, and how does this influence the level of activity of the transcription factor Bicoid; the ventral side of
early development of Drosophila? the embryo is established by a locally high level of activity of the
transcription factor Dorsal. Contrast the mechanisms by which these
2. Much attention has been paid to the process of segmentation
activities come to be localized in the required places.
in the Drosophila embryo. What are segments, and what is their
significance to the strategy used by Drosophila to carry out pattern 6. Bicoid was the first morphogen whose mechanism of action was
formation? Do humans have segments? Give examples. molecularly characterized; an important outcome of this work was
the understanding of how a smoothly declining gradient of Bicoid
3. Briefly summarize the mutation screen devised by Nüsslein–­
can lead to a sharply delimited boundary of Hunchback expression.
Volhard and Wieschaus to find developmental mutants. What role did
Use the Bicoid–Hunchback system to illustrate how an activation
denticle patterns play in their search for developmental mutants?
 Specification of segment identity 87

Summary: main genes involved in specifying pattern in the early Drosophila embryo

Gene Maternal/ Nature Transcription Function (where known)

zygotic of protein factor (T), receptor (R),
or signal protein (S)
Antero-posterior bicoid M Homeodomain T Morphogen, provides positional information along AP axis
system hunchback M/Z Zinc fingers T Morphogen, provides positional information along AP axis
nanos M RNA-binding protein Helps to establish AP gradient of hunchback protein
caudal M Homeodomain T Involved in specifying posterior region

gurken M Secreted protein S Posterior

of TGF-α family oocyte–follicle cell signaling
oskar M Pole-cell determination

Terminal system torso M Receptor tyrosine R Terminal specification

kinase
trunk M S Ligand for torso

Gap genes hunchback Z Zinc fingers T

Krüppel Z Zinc fingers T
knirps Z Zinc fingers T Localize pair-rule gene expression
giant Z Leucine fingers T
tailless Z Zinc fingers T

Pair-rule genes even-skipped Z Homeodomain T Delimits odd-numbered parasegments

fushi tarazu Z Homeodomain T Delimits even-numbered parasegments
hairy Z Helix-loop-helix T

engrailed Z Homeodomain T Defines anterior region of parasegment and posterior

Segmentation
region of segment
genes
hedgehog Z Membrane or secreted S
wingless Z Secreted S
frizzled Z Membrane R Components of signaling pathways that pattern
gooseberry Z Homeodomain T segments and stabilize compartment boundaries
patched Z Membrane R
smoothened Z G-protein-coupled R
Selector genes
Ultrabithorax Z Homeodomain T
bithorax complex abdominal-A Z Homeodomain T Combinatorial activity confers identity on parasegments 5–13
Abdominal-B Z Homeodomain T

Deformed Z Homeodomain T
Antennapedia complex Sex combs Z Homeodomain T
Combinatorial activity confers identity on parasegments
reduced
anterior to 5
Antennapedia Z Homeodomain T
labial Z Homeodomain T

Polycomb group Z T
Maintenance genes Trithorax Z T
Maintain state of homeotic genes

Dorso-ventral system

Maternal genes Toll M Membrane R Activation results in dorsal protein entering nucleus
spätzle M Extracellular S Ligand for Toll protein
dorsal M T Morphogen, sets dorso-ventral polarity
tube M Adaptor protein
pelle M Protein kinase Components of the Toll signaling pathway leading to
cactus M Cytoplasmic inhibitor Dorsal protein entering nuclei
gurken M Secreted protein S Specifies oocyte axis
of TGF-α family
pipe M Sulfotransferase Enzyme Part of pathway leading to
Spätzle processing
Zygotic genes twist Z Helix-loop-helix T Define mesoderm
snail Z Zinc finger T
rhomboid Z Membrane protein S
zerknüllt Z Homeodomain T
decapenta- Secreted protein
plegic Z of TGF-β family S Confer regional identity on dorso-ventral axis
tolloid Z BMP-2 family S
short gastru- Z S
lation

AP, antero-posterior; TGF, transforming growth factor.

88 Chapter 2 Development of the Drosophila body plan

threshold can translate a smooth gradient into a sharp developmen- 2. The process of segmentation will form how many parasegments?
tal boundary. Be sure to incorporate the role of cooperative binding a) 8 in the abdomen
of Bicoid to the hunchback promoter in your answer.
b) 11 to form the thorax and abdomen
7. The dorsal–ventral axis is patterned by a gradient of nuclear c) 14: 3 for the mouth, 3 in the thorax, and 8 in the abdomen
localization of the transcription factor Dorsal. Outline the steps
d) 3: head, thorax, and abdomen
that lead to this gradient of nuclear localization: include the Pipe
enzyme, the Spätzle ligand, the Toll receptor, the Cactus protein, 3. When do the legs and wings of a fly form?
and finally Dorsal. a) during embryogenesis, after imaginal disc cells are segregated
8. The specification of the 14 parasegments along the anterior– away at the cellular blastoderm stage
posterior axis depends on precise spatial expression of the pair-rule b) during pupation, from the imaginal discs
genes. Pair-rule gene expression depends in turn on the gap genes.
c) during the larval instars
Using the second stripe of even-skipped expression as a model, sum-
marize the interactions that lead to proper expression of Bicoid, d) legs and wings are present during larval stages, but take their
Hunchback, Krüppel, and Giant. In particular, describe how the final form only after metamorphosis
anterior border of Krüppel is set, and how this boundary sets the 4. Which protein (gene) functions to determine an anterior fate in
posterior border of eve stripe 2 expression. the Drosophila embryo?
9. What is the special role of engrailed in the patterning of the Dros- a) Bicoid
ophila body plan? b) Caudal
10. Drosophila has played a central role in our understanding of c) Nanos
insect development, but is not necessarily representative of all d) Torso
insects. How does the long-germ development of Drosophila dif-
fer from the short-germ band type of development seen in beetles, 5. Maternal determinants are
grasshoppers, and wasps? a) genes that control differentiation as a female or male fruit fly
11. What is meant by a homeotic transformation? Give an example. b) proteins and mRNAs packaged into the egg by the mother
c) proteins and mRNAs that determine the organizing axes of the
12. Experiments demonstrate that the role of the Bithorax Complex
ovary
is to drive the identity of the parasegments in the posterior por-
tion of the animal away from the default identity of parasegment 4. d) substances present in the female that determine whether or not
Describe these experiments. How do these experiments illustrate the she will become a mother
importance of the combinations of gene products present in specify- 6. Which is the correct hierarchy of gene activity in early Droso-
ing segment identity? phila segmentation?
13. The generation of transgenic Drosophila using the P element a) gap, segment polarity, pair-rule, maternal
transposon has played a complementary role to traditional muta-
b) maternal, gap, pair-rule, segment polarity
tion analysis in deciphering the genetic regulation of Drosophila
development. Answer the following questions: Why is the P ele- c) maternal, pair-rule, gap, segment-polarity
ment important to this process? Why is the P element injected into d) segment polarity, pair-rule, gap, maternal
the posterior of the egg, before any cellularization has occurred?
7. A gap gene mutation could cause which of the following defects
Why are flies mutant for the white gene typically used as the
in the embryonic body plan?
hosts for these experiments? Last, what genetic crosses are neces-
sary before the consequences of the P element injection can be a) every other segment is missing, resulting in T1, T3, A2, A4, etc.,
observed? but no T2, A1, A3, and so on
b) patterning within each segment is abnormal, causing denticle
Multiple choice (factual recall questions) belts to form across the entire segment
NB There is only one right answer to each question. c) segments A2 through A6 are missing, but the rest of the pattern
is essentially normal
1. The ‘spinal cord’ of Drosophila—its ‘nerve cord’
d) the identity of one or more segments is transformed to that of a
a) consists of two parallel structures that run laterally, along the different segment
sides of the animal.
8. Segments are first positioned at the cell-by-cell level by the
b) is in no way analogous to the spinal cord of vertebrates, since
Drosophila has no spine! a) gap genes
c) runs along the dorsal side, opposite the mouth as is the case in b) maternal genes
vertebrates such as humans c) pair-rule genes
d) runs along the ventral surface, on the same side as the mouth d) segment polarity genes
 Specification of segment identity 89

9. Engrailed expression defines the 2.8 Bicoid protein provides an antero-posterior gradient of a
a) anterior compartment of the segment morphogen
Driever, W., Nüsslein-Volhard, C.: The bicoid protein determines
b) anterior margin of each segment
position in the Drosophila embryo in a concentration depen-
c) posterior compartment of the segment dent manner. Cell 1988, 54: 95–104.
d) posterior margin of each parasegment Ephrussi, A., St Johnston, D.: Seeing is believing: the bicoid mor-
phogen gradient matures. Cell 2004, 116: 143–152.
10. Anterior patterning in the long-germ band wasp Nasonia is Gibson, M.C.: Bicoid by the numbers: quantifying a morphogen
under the control of gradient. Cell 2007, 130: 14–16.
a) Bicoid Spirov, A., Fahmy, K., Schneider, M., Frei, E., Noll, M., Baumgartner, S.:
Formation of the bicoid morphogen gradient: an mRNA gradient
b) Engrailed
dictates the protein gradient. Development 2009, 136: 605–614.
c) Krüppel Teleman, A.A., Strigini, M., Cohen, S.M.: Shaping morphogen gra-
d) Orthodenticle dients. Cell 2001, 105: 559–562.

2.9 The posterior pattern is controlled by the gradients of

Multiple choice answer key Nanos and Caudal proteins
1: d, 2: c, 3: b, 4: a, 5: b, 6: b, 7: c, 8: c, 9: c, 10: d. Irish, V., Lehmann, R., Akam, M.: The Drosophila posterior-group
gene nanos functions by repressing hunchback activity. Na-
ture 1989, 338: 646–648.
n General further reading Murafta, Y., Wharton, R.P.: Binding of pumilio to maternal hunch-
back mRNA is required for posterior patterning in Drosophila
Adams, M.D., et al.: The genome sequence of Drosophila melano- embryos. Cell 1995, 80: 747–756.
gaster. Science 2000, 267: 2185–2195. Rivera-Pomar, R., Lu, X., Perrimon, N., Taubert, H., Jackle, H.: Acti-
Ashburner, M.: Drosophila. A Laboratory Handbook. New York: vation of posterior gap gene expression in the Drosophila blas-
Cold Spring Harbor Laboratory Press, 1989. toderm. Nature 1995, 376: 253–256.
Lawrence, P.A.: The Making of a Fly. Oxford: Blackwell Scientific Struhl, G.: Differing strategies for organizing anterior and pos-
Publications, 1992. terior body pattern in Drosophila embryos. Nature 1989, 338:
Matthews, K.A., Kaufman, T.C., Gelbart, W.M.: Research resources 741–744.
for Drosophila: the expanding universe. Nat Rev. Genet. 2005, 6:
2.10 The anterior and posterior extremities of the embryo
179–193.
are specified by cell-surface receptor activation
Stolc, V., et al.: A gene expression map for the euchromatic ge-
nome of Drosophila melanogaster. Science 2004, 306: 655–660. Casali, A., Casanova, J.: The spatial control of Torso RTK acti-
vation: a C-terminal fragment of the Trunk protein acts as a
signal for Torso receptor in the Drosophila embryo. Develop-
ment 2001, 128: 1709–1715.
n Section further reading Casanova, J., Struhl, G.: Localized surface activity of torso, a re-
ceptor tyrosine kinase, specifies terminal body pattern in Dros-
2.2 Cellularization is followed by gastrulation and
ophila. Genes Dev. 1989, 3: 2025–2038.
segmentation
Coppey, M., Boettiger, A.N., Berezhkovskii, A.M., Shvartsman, S.Y.:
Stathopoulos, A., Levine, M.: Whole-genome analysis of Drosophila­ Nuclear trapping shapes the terminal gradient in the Droso-
gastrulation. Curr. Opin. Genet. Dev. 2004, 14: 477–484. phila embryo. Curr. Biol. 2008, 18: 915–919.
Furriols, M., Casanova, J.: In and out of Torso RTK signalling.
2.4 Many developmental mutations have been identified in EMBO J. 2003, 22: 1947–1952.
Drosophila through induced large-scale genetic screening Furriols, M., Ventura, G., Casanova, J.: Two distinct but convergent
groups of cells trigger Torso receptor tyrosine kinase activa-
Nusslein-Volhard. C., Wieschaus, E.: Mutations affecting segment
tion by independently expressing torso-like. Proc. Natl Acad.
number and polarity in Drosophila. Nature 1980, 287: 795–801.
Sci. USA 2007, 104: 11660–11665.
Rørth, P.: A modular misexpression screen in Drosophila detecting Li, W.X.: Functions and mechanisms of receptor tyrosine kinase
tissue-specific phenotypes. Proc. Natl Acad. Sci. USA 1996, 93: Torso signaling: lessons from Drosophila embryonic terminal
12418–12422. development. Dev. Dyn. 2005, 232: 656–672.

2.7 Three classes of maternal genes specify the antero- 2.11 The dorso-ventral polarity of the embryo is specified by
posterior axis localization of maternal proteins in the egg vitelline envelope
St Johnston, D., Nüsslein-Volhard, C.: The origin of pattern and Anderson, K.V.: Pinning down positional information: dorsal-­ventral
polarity in the Drosophila embryo. Cell 1992, 68: 201–219. polarity in the Drosophila embryo. Cell 1998, 95: 439–442.
90 Chapter 2 Development of the Drosophila body plan

Sen, J., Goltz, J.S., Stevens, L., Stein, D.: Spatially restricted ex- Roth, S., Neuman-Silberberg, F.S., Barcelo, G., Schupbach, T.: Cor-
pression of pipe in the Drosophila egg chamber defines embry- nichon and the EGF receptor signaling process are necessary
onic dorsal-ventral polarity. Cell 1998, 95: 471–481. for both anterior-posterior and dorsal-ventral pattern forma-
tion in Drosophila. Cell 1995, 81: 967–978.
2.12 Positional information along the dorso-ventral axis is van Eeden, F., St Johnston, D.: The polarisation of the anterior-
provided by the Dorsal protein posterior and dorsal-ventral axes during Drosophila oogenesis.
Belvin, M.P., Anderson, K.V.: A conserved signaling pathway: Curr. Opin. Genet. Dev. 1999, 9: 396–404.
the Drosophila Toll-dorsal pathway. Annu. Rev. Cell Dev.
Biol. 1996, 12: 393–416. 2.16 The antero-posterior axis is divided up into broad
Imler, J.L., Hoffmann, J.A.: Toll receptors in Drosophila: a family regions by gap-gene expression
of molecules regulating development and immunity. Curr. Top. Hülskamp, M., Tautz, D.: Gap genes and gradients—the logic be-
Microbiol. Immunol. 2002, 270: 63–79. hind the gaps. BioEssays 1991, 13: 261–268.
Roth, S., Stein, D., Nüsslein-Volhard, C.: A gradient of nuclear lo-
calization of the dorsal protein determines dorso-ventral pat-
2.17 Bicoid protein provides a positional signal for the
tern in the Drosophila embryo. Cell 1989, 59: 1189–1202.
anterior expression of zygotic hunchback
Steward, R., Govind, R.: Dorsal-ventral polarity in the Drosophila
Brand, A.H., Perrimon, N.: Targeted gene expression as a means of
embryo. Curr. Opin. Genet. Dev. 1993, 3: 556–561.
altering cell fates and generating dominant phenotypes. Devel-
2.13 The antero-posterior axis of the Drosophila egg is opment 1993, 118: 401–415.
specified by signals from the preceding egg chamber and by Gibson, M.C.: Bicoid by the numbers: quantifying a morphogen
interactions of the oocyte with follicle cells gradient. Cell 2007, 130: 14–16.
González-Reyes, A., St Johnston, D.: The Drosophila AP axis is Gregor, T., Tank, D.W., Wieschaus, E.F., Bialek W.: Probing the lim-
polarised by the cadherin-mediated positioning of the oocyte. its to positional information. Cell 2007, 130: 153–164.
Development 1998, 125: 3635–3644. Ochoa-Espinosa, A., Yucel, G., Kaplan, L., Pare, A., Pura, N.,
Huynh, J.R., St Johnston, D.: The origin of asymmetry: early po- ­Oberstein, A., Papatsenko, D., Small, S.: The role of binding site
larization of the Drosophila germline cyst and oocyte. Curr. cluster strength in Bicoid-dependent patterning in Drosophila.
Biol. 2004, 14: R438–R449. Proc. Natl Acad. Sci. USA 2005, 102: 4960–4965.
Riechmann, V., Ephrussi, A.: Axis formation during Drosophila Simpson-Brose, M., Treisman, J., Desplan, C.: Synergy between the
­oogenesis. Curr. Opin. Genet. Dev. 2001, 11: 374–383. Hunchback and Bicoid morphogens is required for anterior
Torres, I.L., Lopez-Schier, H., St Johnston, D.: A Notch/Delta- patterning in Drosophila. Cell 1994, 78: 855–865.
­dependent relay mechanism establishes anterior-posterior po- Struhl, G., Struhl, K., Macdonald, P.M.: The gradient morphogen
larity in Drosophila. Dev. Cell 2005, 5: 547–558. Bicoid is a concentration-dependent transcriptional activator.
Zimyanin, V.L., Belaya, K., Pecreaux, J., Gilchrist, M.J., Clark, Cell 1989, 57: 1259–1273.
A., Davis, I., St Johnston, D.: In vivo imaging of oskar mRNA
transport reveals the mechanism of posterior localization. 2.18 The gradient in Hunchback protein activates and
Cell 2008, 134: 843–853. represses other gap genes
Rivera-Pomar, R., Jäckle, H.: From gradients to stripes in Droso-
2.14 Localization of maternal mRNAs to either end of the egg
phila embryogenesis: filling in the gaps. Trends Genet. 1996, 12:
depends on the reorganization of the oocyte cytoskeleton
478–483.
Becalska, A.N., Gavis, E.R.: Lighting up mRNA localization in Struhl, G., Johnston, P., Lawrence, P.A.: Control of Drosophila body
Drosophila oogenesis. Development 2009, 136: 2493–2503. pattern by the hunchback morphogen gradient. Cell 1992, 69:
Martin, S.G., Leclerc, V., Smith-Litière, K., St Johnston, D.: The 237–249.
identification of novel genes required for Drosophila antero- Wu, X., Vakani, R., Small, S.: Two distinct mechanisms for differen-
posterior axis formation in a germline clone screen using GFP- tial positioning of gene expression borders involving the Droso-
Staufen. Development 2003, 130: 4201–4215. phila gap protein giant. Development 1998, 125: 3765–3774.
St Johnston, D.: Moving messages: the intracellular localization
of mRNAs. Nat Rev. Mol. Cell Biol. 2005, 6: 363–375.
2.19 The expression of zygotic genes along the dorso-ventral
2.15 The dorso-ventral axis of the egg is specified by axis is controlled by Dorsal protein
movement of the oocyte nucleus followed by signaling Cowden, J., Levine, M.: Ventral dominance governs segmen-
between oocyte and follicle cells tal patterns of gene expression across the dorsal-ventral axis
González-Reyes, A., Elliott, H., St Johnston, D.: Polarization of both of the neurectoderm in the Drosophila embryo. Dev. Biol.
major body axes in Drosophila by gurken-torpedo signaling. 2003, 262: 335–349.
Nature 1995, 375: 654–658. Harland, R.M.: A twist on embryonic signalling. Nature 2001,
Jordan, K.C., Clegg, N.J., Blasi, J.A., Morimoto, A.M., Sen, J., 410: 423–424.
Stein, D., McNeill, H., Deng, W.M., Tworoger, M., Ruohola- Markstein, M., Zinzen, R., Markstein, P., Yee, K.P., Erives, A.,
Baker, H.: The homeobox gene mirror links EGF signalling to Stathopoulos, A., Levine, M.: A regulatory code for neurogenic
­embryonic dorso-ventral axis formation through Notch activa- gene expression in the Drosophila embryo. Development 2004,
tion. Nat. Genet. 2000, 24: 429–433. 131: 2387–2394.
 Specification of segment identity 91

Rusch, J., Levine, M.: Threshold responses to the dorsal regulatory Harrison, D.A., Perrimon, N.: Simple and efficient generation of
gradient and the subdivision of primary tissue territories in the marked clones in Drosophila. Curr. Biol. 1993, 3: 424–433.
Drosophila embryo. Curr. Opin. Genet. Dev. 1996, 6: 416–423. Vincent, J.P., O’Farrell, P.H.: The state of engrailed expression is
Stathopoulos, A., Levine, M.: Dorsal gradient networks in the not clonally transmitted during early Drosophila development.
Drosophila embryo. Dev. Biol. 2002, 246: 57–67. Cell 1992, 68: 923–931.

2.20 The Decapentaplegic protein acts as a morphogen to 2.24 Segmentation genes stabilize parasegment boundaries
pattern the dorsal region and set up a focus of signaling at the boundary that patterns
Affolter, M., Basler, K.: The Decapentaplegic morphogen gradient: the segment
from pattern formation to growth regulation. Nat. Rev. Genet. Alexandre, C., Lecourtois, M., Vincent, J.-P.: Wingless and hedge-
2007, 8: 663–674. hog pattern Drosophila denticle belts by regulating the produc-
Ashe, H.L., Levine, M.: Local inhibition and long-range enhance- tion of short-range signals. Development 1999, 126: 5689–5698.
ment of Dpp signal transduction by Sog. Nature 1999, 398: Beckett, K., Franch-Marro, X., Vincent, J.P.: Glypican-mediated en-
­427–431. docytosis of hedgehog has opposite effects in flies and mice.
Mizutani, C.M., Nie, Q., Wan, F.Y., Zhang, Y.T., Vilmos, P., Sousa- Trends Cell Biol. 2008, 18: 360–363.
Neves, R., Bier, E., Marsh, J.L., Lander, A.D.: Formation of the Bejsovec, A., Wieschaus, E.: Segment polarity gene interactions
BMP activity gradient in the Drosophila embryo. Dev. Cell 2005, modulate epidermal patterning in Drosophila embryos. Devel-
8: 915–924. opment 1993, 119: 501–517.
Shimmi, O., Umulis, D., Othmer, H., O’Connor, M.B.: Facilitated Briscoe, J., Therond, P.: Hedgehog signaling: from the Drosophila
transport of a Dpp/Scw heterodimer by Sog/Tsg leads to robust cuticle to anti-cancer drugs. Dev. Cell 2005, 8: 143–151.
patterning of the Drosophila blastoderm embryo. Cell 2005, Gordon, M., Nusse, R.: Wnt signaling: multiple pathways: multi-
120: 873–886. ple receptors, and multiple transcription factors. J. Biol. Chem.
Srinivasan, S., Rashka, K.E., Bier, E.: Creation of a Sog morphogen 2006, 281: 22429–22433.
gradient in the Drosophila embryo. Dev. Cell 2002, 2: 91–101. Hooper, J.E., Scott, M.P.: Communicating with Hedgehogs. Nat.
Wang, Y.-C., Ferguson, E.L.: Spatial bistability of Dpp-receptor in- Rev. Mol. Cell Biol. 2005, 6: 306–317.
teractions driving Drosophila dorsal-ventral patterning. Nature Johnson, R.L., Scott, M.P.: New players and puzzles in the
2005, 434: 229–234. Hedgehog signaling pathway. Curr. Opin. Genet. Dev. 1998, 8:
­450–456.
Wharton, K.A., Ray, R.P., Gelbart, W.M.: An activity gradient
Larsen, C.W., Hirst, E., Alexandre, C., Vincent, J.-P.: Segment
of decapentaplegic is necessary for the specification of dor-
boundary formation in Drosophila embryos. Development 2003,
sal pattern elements in the Drosophila embryo. Development
130: 5625–5635.
1993, 117: 807–822.
Lawrence, P.A., Casal, J., Struhl, G.: Hedgehog and engrailed: pat-
tern formation and polarity in the Drosophila abdomen. Devel-
2.22 Gap-gene activity positions stripes of pair-rule gene
opment 1999, 126: 2431–2439.
expression
Tolwinski, N.S., Wieschaus, E.: Rethinking Wnt signaling. Trends
Clyde, D.E., Corado, M.S., Wu, X., Paré, A., Papatsenko, D., Small, Genet. 2004, 20: 177–181
S.: A self-organizing system of repressor gradients establishes von Dassow, G., Meir, E., Munro, E.M., Odell, G.M.: The segment
segmental complexity in Drosophila. Nature 2003, 426: 849– polarity network is a robust developmental module. Nature
853. 2000, 406: 188–192.
Luengo Hendriks, C.L., Keränen, S.V., Fowlkes, C.C., Simirenko, L., Wehrli, M., Dougan, S.T., Caldwell, K., O’Keefe, L., Schwartz, S.,
Weber, G.H., DePace, A.H., Henriquez, C., Kaszuba, D.W., Vaizel-Ohayon, D., Schejter, E., Tomlinson, A., DiNardo, S.: ­arrow
Hamann, B., Eisen, M.B., Malik, J., Sudar D., Biggin, M.D., encodes an LDL-receptor-related protein essential for Wingless
Knowles, D.W.: Three-dimensional morphology and gene ex- signalling. Nature 2000, 407: 527–530.
pression in the Drosophila blastoderm at cellular resolution I: Zhu, A., Scott, M.: Incredible journey: how do developmental sig-
data acquisition pipeline. Genome Biol. 2006, 7: R123. nals travel through tissue? Genes Dev. 2004, 18: 2983–2997.
Small, S., Levine, M.: The initiation of pair-rule stripes in the Dros-
ophila blastoderm. Curr. Opin. Genet. Dev. 1991, 1: 255–260. 2.25 Insect epidermal cells become individually polarized in
Yu, D., Small, D.: Precise registration of gene expression bound- an antero-posterior direction in the plane of the epithelium &
aries by a repressive morphogen in Drosophila. Curr. Biol. Box 2F Planar cell polarity in Drosophila
2008, 18: 888–876.
Lawrence, P.A., Casal, J., Struhl, G.: Cell interactions and planar
polarity in the abdominal epidermis of Drosophila. Develop-
2.23 Expression of the engrailed gene delimits a cell-lineage ment 2004, 131: 4651–4664.
boundary and defines a compartment Lawrence, P.A., Struhl, G., Casal, J.: Planar polarity: one or two
Dahmann, C., Basler, K.: Compartment boundaries: at the edge of pathways. Nat. Rev. Genet. 2008, 8: 555–363.
development. Trends Genet. 1999, 15: 320–326. Ma, D., Yang, C., McNeill, H., Simon, M.A., Axelrod, J.D.: Fidelity
Gray, S., Cai, H., Barolo, S., Levine, M.: Transcriptional repression in planar cell polarity signalling. Nature 2003, 421: 543–547.
in the Drosophila embryo. Phil. Trans. R. Soc. Lond. 1995, 349: Strutt, D.: The planar polarity pathway. Curr. Biol. 2008, 16:
257–262. ­R898–R902.
92 Chapter 2 Development of the Drosophila body plan

Strutt, D., Strutt, H.: Differential activities of the core planar 2.28 Homeotic selector genes of the bithorax complex are
polarity proteins during Drosophila wing patterning. Dev. responsible for diversification of the posterior segments
Biol. 2007, 302: 181–194. Castelli-Gair, J., Akam, M.: How the Hox gene Ultrabithorax speci-
Wu, J., Mlodzik, M.: The Frizzled extracellular domain is a ligand fies two different segments: the significance of spatial and
for Van Gogh/Stm during nonautonomous planar cell polarity temporal regulation within metameres. Development 1995, 121:
signaling. Dev. Cell 2008, 15: 462–469. 2973–2982.
Duncan, I.: How do single homeotic genes control multiple seg-
2.26 Some insects use different mechanisms for patterning
ment identities? BioEssays 1996, 18: 91–94.
the body plan
Lawrence, P.A, Morata, G.: Homeobox genes: their function in
Akam, M., Dawes, R.: More than one way to slice an egg. Curr. Drosophila segmentation and pattern formation. Cell 1994, 78:
Biol. 1992, 8: 395–398. 181–189.
Angelini, D.R., Liu, P.Z., Hughes, C.L., Kaufman, T.C.: Hox gene Liang, Z., Biggin, M.D.: Eve and ftz regulate a wide array of
function and interaction in the milkweed bug Oncopeltus fas- genes in blastoderm embryos: the selector homeoproteins di-
ciatus (Hemiptera). Dev. Biol. 2005, 287: 440–455. rectly or indirectly regulate most genes in Drosophila. Develop-
Brown, S.J., Parrish, J.K., Beeman, R.W., Denell, R.E.: Molecular ment 1998, 125: 4471–4482.
characterization and embryonic expression of the even-skipped Mann, R.S., Morata, G.: The developmental and molecular biology
ortholog of Tribolium castaneum. Mech. Dev. 1997, 61: 165–173. of genes that subdivide the body of Drosophila. Annu. Rev. Cell
French, V.: Segmentation (and eve) in very odd insect embryos. Dev. Biol. 2000, 16: 143–271.
BioEssays 1996, 18: 435–438. Mannervik, M.: Target genes of homeodomain proteins. BioEssays
French, V.: Insect segmentation: genes, stripes and segments in 1999, 4: 267–270.
‘Hoppers’. Curr. Biol. 2001, 11: R910–R913. Simon, J.: Locking in stable states of gene expression: transcrip-
Lynch, J.A., Brent, A.E., Leaf, D.S., Pultz, M.A., Desplan, C.: Local- tional control during Drosophila development. Curr. Opin. Cell
ized maternal orthodenticle patterns anterior and posterior in Biol. 1995, 7: 376–385.
the long germ wasp Nasonia. Nature 2006, 439: 728–732.
Sander, K.: Pattern formation in the insect embryo. In Cell Pat- 2.30 The order of Hox gene expression corresponds to the
terning, Ciba Found. Symp. 29. London: Ciba Foundation, 1975: order of genes along the chromosome
241–263. Morata, G.: Homeotic genes of Drosophila. Curr. Opin. Genet. Dev.
Tautz, D., Sommer, R.J.: Evolution of segmentation genes in insects. 1993, 3: 606–614.
Trends Genet. 1995, 11: 23–27.
2.31 The Drosophila head region is specified by genes other
2.27 Segment identity in Drosophila is specified by Hox than the Hox genes
genes
Rogers B.T., Kaufman, T.C.: Structure of the insect head in on-
Lewis E.B.: A gene complex controlling segmentation in Droso- togeny and phylogeny: a view from Drosophila. Int. Rev. Cytol.
phila. Nature 1978, 276: 565–570. 1997, 174: 1–84.
Stark, A.: A single Hox locus in Drosophila produces functional
microRNAs from opposite DNA strands. Genes Dev. 2008, 22:
8–13.
 3
Vertebrate development I:
life cycles and experimental
techniques
nn Vertebrate life cycles and nn Experimental approaches
outlines of development to studying vertebrate
development

In this chapter we examine the similarities and differences in the outlines of develop-
ment of four vertebrate model animals: amphibians represented by the frog Xenopus,
fish by the zebrafish, birds by the chick, and mammals by the mouse. The second part
of the chapter introduces some of the experimental approaches used to investigate
vertebrate development.

We saw in Chapter 2 how early development in insects is largely under the control of
maternal factors that interact with each other to specify broadly different regions of
the body. This initial blueprint is then elaborated on by the embryo’s own genes. We
shall now look at how the same task of establishing the outline of the body plan is
achieved in early vertebrate development. All vertebrates, despite their many outward
differences, have a similar basic body plan. The defining structures are the segmented
backbone or vertebral column surrounding the spinal cord, with the brain at the
head end enclosed in a bony or cartilaginous skull (Fig. 3.1). These prominent struc-
tures mark the antero-posterior axis, the main body axis of vertebrates. The head is
at the anterior end of this axis, followed by the trunk with its paired appendages—limbs
in terrestrial vertebrates (with the exception of snakes) and fins in fish—and in many
vertebrates this axis terminates at the posterior end in a post-anal tail. The vertebrate
body also has a distinct dorso-ventral axis running from the back to the belly, with
the spinal cord running along the dorsal side and the mouth defining the ventral side.
The antero-posterior and dorso-ventral axes together define the left and right sides
of the animal. Vertebrates have a general bilateral symmetry around the dorsal mid-
line so that outwardly the right and left sides are mirror images of each other. Some
internal organs, such as lungs, kidneys, and gonads, are also present as symmetrically
paired structures, but single organs such as the heart and liver are arranged asym-
metrically with respect to the dorsal midline with the heart on the left and the liver
on the right.
94 Chapter 3 Vertebrate development I: life cycles and experimental techniques

In this and the following two chapters we will look at four vertebrates whose early
development has been particularly well studied—Xenopus, zebrafish, chicken, and
mouse. We also touch on the early development of a human embryo. In this chapter,
we will first outline the life cycles of these organisms and briefly describe the key
features of their early development, providing the context for the later discussions.
We will not at this stage consider the mechanisms of development, but focus on the
changes in form that occur in early embryonic development to give the well-defined
overall body plan of the vertebrate embryo, with its characteristic structures such
as the notochord and neural tube (see Fig. 3.8). We will take the embryo from the
earliest cell divisions of the fertilized egg, through the cell movements of gastrula-
tion, the dynamic process in which the embryo is rearranged to form the three main
germ layers in their correct positions in the body, to the end of neurulation, which
forms the neural tube—the earliest appearance of the nervous system. The germ
layers are the ectoderm, which gives rise to the nervous system and epidermis, the
mesoderm, which gives rise to the skeleton, muscles, heart, blood, and some other
internal organs and tissues, and the endoderm, which gives rise to the gut and asso-
ciated glands and organs (see Box 1C, p. 15). In the second half of the chapter we
will introduce some of the experimental techniques that are used to study vertebrate
development.
Chapter 4 then looks in detail at the first two key events in development: the estab-
lishment of the antero-posterior and dorso-ventral axes, and the specification and
early patterning of the germ layers. In particular, we will look at the setting up of the
Fig. 3.1 The skeleton of a 17.5-day (E17.5) organizer regions (see Section 1.4) and the induction of the mesodermal germ layer
mouse embryo illustrates the vertebrate and its patterning along the dorso-ventral axis. In Chapter 5 we continue the discus-
body plan. The skeletal elements in this
sion of early development in vertebrates by looking at the role of the organizer region
embryo have been stained with Alcian blue
in directing development along the antero-posterior axis during and after the large-
(which stains cartilage) and Alizarin red
scale reorganization of tissues that occurs during gastrulation. This reorganization
(which stains bone). The vertebral column,
which develops from blocks of somites, leads to the all-important induction of the prospective neural tissue from ectoderm—
is divided up into cervical (neck), thoracic from which all the nervous system develops—and the formation of the notochord
(chest), lumbar (lower back), and sacral (hip and somites from mesoderm, the somites subsequently developing into muscle and
and lower) regions. The paired limbs can also skeletal tissues (see Box 1B, p. 7). By this stage the embryo has become recogniz-
be seen. Scale bar = 1 mm. ably vertebrate. The development of individual structures and organs, such as limbs,
Photograph courtesy of M. Maden. eyes, heart, and the nervous system, is covered in later chapters. We shall take a
more detailed look at the mechanics of vertebrate gastrulation and the underlying
cell biology that allows such extensive remodeling of tissues in Chapter 8, along with
gastrulation in simpler non-vertebrate animals such as the sea urchin.

Vertebrate life cycles and outlines of development

All vertebrate embryos pass through a broadly similar set of developmental stages,
which are outlined for each organism in the life-cycle diagrams in this chapter (for
example, the one for Xenopus in Fig. 3.3). After fertilization, the zygote undergoes
cleavage. These are rapid cell divisions by which the embryo becomes divided into
a number of smaller cells. This is followed by gastrulation, a set of cell movements
that generates the three distinct cell layers: ectoderm, mesoderm, and endoderm (see
Box 1C, p. 15). By the end of gastrulation, the ectoderm covers the embryo, and the
mesoderm and endoderm have moved inside. The endoderm gives rise to the gut, and
to its derivatives such as liver and lungs; the mesoderm forms the skeleton, muscles,
connective tissues, kidneys, heart, and blood, as well as some other tissues; and the
ectoderm gives rise to the epidermis and the nervous system.
Figure 3.2 shows the differences in form of the early embryos from the four model
species and of a human embryo at the same stage. After gastrulation, all vertebrate
embryos pass through a stage at which they all more or less resemble each other
 Vertebrate life cycles and outlines of development 95

and show the specific embryonic features of the chordates, the phylum of animals to
which vertebrates belong. This stage is called the phylotypic stage. At the phylotypic
stage the head is distinct and the neural tube, the forerunner of the nervous system,
runs along the dorsal midline of the antero-posterior (head to tail) axis. Immediately
under the neural tube runs the notochord, a signature structure of chordates, flanked
on either side by the somites, blocks of tissue from which the muscles and skeleton
will derive. Features special to the different vertebrate groups, such as beaks, wings,
and fins, appear later.

Xenopus Zebrafish Chick Mouse Human

Fig. 3.2 Vertebrate embryos show considerable differences in thus already developed some extra-embryonic tissues required for
form before gastrulation but subsequently all go through a stage implantation. The mouse embryo proper is the small cup-shaped
at which they look similar. The eggs of the frog (Xenopus), zebrafish, structure at the center, seen here in cross-section as a U-shaped layer
chicken, and mouse are very different in size, while human eggs are of epithelium. The human embryo proper is a two-layered disc of cells.
about the same size as mouse eggs (top row). Scale bars in this row all Scale bars for mouse and human embryos = 100 µm. After gastrulation
represent 1 mm, except for the chicken egg, which represents 10 mm. and formation of the neural tube, vertebrate embryos pass through an
Their early development (second row) is rather different. In this row, embryonic stage at which they all look rather similar (third row), and
the embryos are shown in cross-section at the stage corresponding which is known as the phylotypic stage. The body has developed, and
roughly to the Xenopus blastula (left panel) just before gastrulation neural tube, somites, notochord, and head structures are present. Scale
commences. The main determinant of tissue arrangement is the bars = 1 mm. After this stage their development diverges again. Paired
amount of yolk (yellow) in the egg. The human and mouse embryos appendages, for example, develop into fins in fish, wings and legs in
(on the right) at this stage have implanted into the uterine wall and the chick and arms and legs in the human (bottom row).
96 Chapter 3 Vertebrate development I: life cycles and experimental techniques

The rod-shaped notochord is one of the earliest mesodermal structures that can
be recognized in vertebrate embryos. This structure does not persist and its cells
eventually become incorporated into the column of vertebrae that form the spine (see
Fig. 3.1). The rest of the vertebral column, the skeleton of the trunk, and the muscles
of the trunk and limbs develop from the somites, blocks of cells that form in an antero-
posterior sequence from the mesoderm on either side of the notochord. Towards the
end of gastrulation, the ectoderm overlying the notochord becomes specified as neu-
ral tissue and then rolls up to form the neural tube, from which the brain, spinal cord,
and the rest of the nervous system derive (see Fig. 3.7). The overall similarity of the
body plan in all vertebrates suggests that the developmental processes that establish it
are broadly similar in the different animals. This is largely the case, although there are
also considerable differences in development, especially at the earliest stages.
The differences in development in the model organisms particularly relate to how
and when the axes are set up, and how the germ layers are established, as we shall see
in Chapter 4, and are mainly due to the different modes of reproduction and the con-
sequent form of the earliest embryo. Yolk provides all the nutrients for fish, amphib-
ian, reptile, and bird embryonic development, and for the few egg-laying mammals
such as the platypus. The eggs of most mammals, by contrast, are small and non-
yolky, and the embryo is nourished for the first few days by fluids in the oviduct and
uterus. Once implanted in the uterus wall, the mammalian embryo develops special-
ized extra-embryonic membranes that surround and protect the embryo and through
which it receives nourishment from the mother via the placenta. Avian embryos also
develop extra-embryonic membranes for obtaining nutrients from the yolk, for oxygen
and carbon dioxide exchange through the permeable egg shell, and for waste dis-
posal. Birds and mammals, both of which form an extra-embryonic membrane called
an amnion, are known as amniotes, whereas amphibians and fish, which do not form
extra-embryonic membranes, are known as anamniotes.
To study development, it is necessary to have a reliable way of identifying and refer-
ring to a particular stage of development. Simply measuring the time from fertilization
is not satisfactory for most species, as there is considerable variation. Amphibians,
for example, will develop quite normally over a range of temperatures, but the rate
of development is quite different at different temperatures. Developmental biologists,
therefore, divide the normal embryonic development of each species into a series
of numbered stages, which are identified by their main features rather than by time
after fertilization. A stage-10 Xenopus embryo, for example, refers to an embryo at a
very early stage of gastrulation, whereas stage 54 is the fully developed tadpole with
legs. Numbered stages have similarly been characterized for the chick embryo, and
these provide a much finer staging system than measuring the time since the egg
was laid. Even for mouse embryos, which develop in a much more constant environ-
ment inside the mother, there is considerable variation in developmental timing, even
among embryos in the same litter. Therefore, although early mouse embryos are often
staged according to days post-conception, with the morning on which the vaginal
plug appears called 0.5 dpc, or E0.5 (embryonic day 0.5), this is only a guide, and
once somites have formed, somite number can be used as a more accurate indication
of developmental stage. Links to websites that describe all the stages for our four
model vertebrates are given in General further reading. The techniques that can be
used to study these four vertebrates are discussed in the later part of the chapter.

3.1 The frog Xenopus laevis is the model amphibian

for developmental studies

The amphibian species most commonly used for developmental work is the African
claw-toed frog, Xenopus laevis, which is completely aquatic and able to develop nor-
mally in tap water. Xenopus laevis is a tetraploid species, however, and the ­diploid
 Vertebrate life cycles and outlines of development 97

species X. tropicalis is becoming increasingly used for genetic studies (see Sec-
tion 1.6). Throughout this book, Xenopus will refer to X. laevis, unless otherwise
specified. A great advantage of Xenopus is that its fertilized eggs are easy to obtain;
females and males injected with the human hormone chorionic gonadotropin and
put together overnight will mate, and the female will lay hundreds of eggs in the
water, which are fertilized by sperm released by the male. Eggs can also be fertilized
in a dish by adding sperm to eggs released after hormonal stimulation of the female.
One advantage of artificial fertilization is that because the resulting embryos develop
synchronously, many embryos, all at the same stage, can be obtained. The eggs are
large (1.2–1.4 mm in diameter) and so are quite easily manipulated. The embryos of
Xenopus are extremely hardy and are highly resistant to infection after microsurgery.
It is also easy to culture fragments of early Xenopus embryos in a simple, chemically
defined solution.
The Xenopus life cycle and main developmental stages are summarized in Fig. 3.3.
Fig. 3.3 Life cycle of the African claw-toed
The mature Xenopus egg has a distinct polarity, with a dark, pigmented animal ­region frog Xenopus laevis. The numbered stages
and a pale, yolky, and heavier vegetal region (Fig. 3.4). The axis running from the refer to standardized stages of Xenopus
animal pole to the vegetal pole is known as the animal–vegetal axis. Before fertiliza- development. An illustrated list of all stages
tion, the egg is enclosed in a protective vitelline membrane, which is embedded in can be found on the Xenbase website listed
a gelatinous coat. Meiosis is not yet complete: the first meiotic division has resulted in General further reading. The photographs
in a small cell—a polar body—forming at the animal pole, but the second meiotic show: an embryo at the blastula stage (top,
division is completed only after fertilization, when the second polar body also forms scale bar = 0.5 mm); a tadpole at stage 41
at the animal pole (see Box 9A, p. 338). At fertilization, one sperm enters the egg in (middle, scale bar = 1 mm); and an adult frog
(bottom, scale bar = 1 cm).
the animal region. The egg completes meiosis and the egg and sperm nuclei fuse to
form the diploid zygote nucleus. Photographs courtesy of J. Slack (top, from
Alberts, B., et al.: 1994) and J. Smith (middle
and bottom).

Adult (stage 66)

Egg
(stage 1)

Metamorphosis Cleavage
Blastula
(stage 8)
60 0
Free-swimming tadpole (stage 45)
5
4 (at 25oC)
Days Hours
after after
fertilization fertilization
10
Tailbud embryo (stage 26)
1 stage 10
(section)
20 15

Gastrulation
Organogenesis Gastrula
Neurula (section)

(stage 16) (stage 12)

(dorsal view)
98 Chapter 3 Vertebrate development I: life cycles and experimental techniques

The first cleavage division of the fertilized egg occurs about 90 minutes after fer-
tilization and is along the plane of the animal–vegetal axis, dividing the embryo into
equal left and right halves (Fig. 3.5). Further cleavages follow rapidly at intervals of
about 20 minutes. The second cleavage is also along the animal–vegetal axis but at
right angles to the first. The third cleavage is equatorial, at right angles to the first two,
and divides the embryo into four animal cells and four larger vegetal cells. In Xenopus
there is no cell growth between cell divisions at this early stage and so continued
cleavage results in the formation of smaller and smaller cells; the cells deriving from
cleavage divisions in animal embryos are often called blastomeres. Cleavage occurs
synchronously and divisions occur in such a way that cells in the yolky vegetal half of
the embryo are larger than those in the animal half. Inside this spherical mass of cells
a fluid-filled cavity—the blastocoel—develops in the animal region, and the embryo
is now called a blastula.
At the end of blastula formation, the Xenopus embryo has gone through about
Fig. 3.4 A late-stage Xenopus oocyte. 12 cell divisions and is made up of several thousand cells. The mesoderm and endo-
The surface of the animal half (top) is derm, which will develop into internal structures, are located around the equator in
pigmented and the paler, vegetal half of the the marginal zone and in the vegetal region, respectively, while the ectoderm, which
egg is heavy with yolk. Scale bar = 1 mm.
will eventually cover the whole of the embryo, is still confined to the animal region
Photograph courtesy of J. Smith. (Fig. 3.6, first panel).
The next stage is gastrulation, which involves extensive movement and rear-
rangement of the germ layers specified in the blastula so that they become located
in their proper positions in the body. Gastrulation involves changes in three dimen-
sions, and so it can be difficult to visualize. The first external sign of gastrulation is
a small slit-like infolding—the blastopore—that forms on the surface of the blastula
on the future dorsal side (see Fig. 3.6, second panel). This region is of particular
Fig. 3.5 Cleavage of the Xenopus importance in development, as it is the site of the embryonic organizer, known as
embryo. The upper panels are diagrams the Spemann organizer in amphibians, without which dorsal and axial develop-
showing the fertilized egg and the first three ment will not occur. The famous experiment by Hans Spemann and Hilde Mangold
cleavage divisions seen from the side. Polar that demonstrated this is described in Fig. 1.9. Once gastrulation has started, the
bodies shown in the fertilized egg and the embryo is known as a gastrula. In Xenopus, the future endoderm and mesoderm
two-cell stage are attached to the animal in the marginal zone move inside the gastrula through the blastopore by rolling
pole. The photographs below show the under the lip as coherent sheets of cells. This type of inward movement is known
cleaving Xenopus egg, but are taken from as involution. Once inside, the tissues converge towards the midline and extend
various angles (see also Fig. 1.14).
along the antero-posterior axis beneath the dorsal ectoderm, elongating the embryo
Photographs courtesy of R. Kessel, from along the antero-posterior axis. At the same time, the ectoderm spreads downward
Kessel, R.G., et al.: 1974.

Cleavage in Xenopus

sperm entry polar bodies

point

1st three cleavages

are perpendicular
 Vertebrate life cycles and outlines of development 99

Blastula, stage 8 Early gastrula, stage 10 Gastrula, stage 11 Late gastrula, stage 12

marginal Animal pole blastocoel archenteron archenteron yolk plug

zone

dorsal
lip
yolky
Vegetal pole vegetal cells blastopore blastocoel

ectoderm mesoderm endoderm Fig. 3.6 Gastrulation in amphibians.

The blastula (first panel) contains several
thousand cells and there is a fluid-filled
to cover the whole embryo by a process known as epiboly. The involuting layer of
cavity, the blastocoel, beneath the cells at
dorsal endoderm is closely applied to the mesoderm, and the space between it and
the animal pole. Gastrulation begins (second
the yolky vegetal cells is known as the archenteron (see Fig. 3.6, third panel). This panel) at the blastopore, which forms on the
is the precursor of the gut cavity. dorsal side of the embryo. Future mesoderm
The inward movement of endoderm and mesoderm begins dorsally and then spreads and endoderm of the marginal zone move
to form a complete circle of involuting cells around the blastopore (see Fig. 3.6, fourth inside at this site through the dorsal lip of
panel). By the end of gastrulation, the blastopore has almost closed—the remaining the blastopore, the mesoderm ending up
slit will form the anus. The dorsal mesoderm now lies beneath the dorsal ectoderm, sandwiched between the endoderm and
the lateral and ventral mesoderm is in its definitive position in the body plan, and the ectoderm in the animal region (third panel).
ectoderm has spread to cover the entire embryo. There is still a large amount of yolk The tissue movements create a new internal
cavity—the archenteron—that will become
present, which provides nutrients until the larva—the tadpole—starts feeding. During
the gut. Endoderm in the ventral region also
gastrulation, the mesoderm in the dorsal region starts to develop into the notochord
moves inside through the ventral lip of the
and the somites, while the more lateral mesoderm—the lateral plate mesoderm— blastopore (fourth panel) and will eventually
will later form mesoderm-derived internal organs such as the kidneys, with anterior completely line the archenteron. At the end
lateral plate mesoderm giving rise to the heart. The cell and tissue movements that of gastrulation the blastocoel is considerably
occur during gastrulation in Xenopus and other animals are discussed in more detail reduced in size.
in Chapter 8. After Balinsky, B.I.: 1975.
Gastrulation is succeeded by neurulation—the formation of the neural tube, which
is the early embryonic precursor of the central nervous system. The embryo is then
called a neurula. The earliest visible sign of neurulation is the formation of the neu-
ral folds, which form on the edges of the neural plate, an area of columnar ecto-
derm cells overlying the notochord. The folds rise up, fold toward the midline and
fuse together to form the neural tube, which sinks beneath the epidermis (Fig. 3.7).
­Neural crest cells detach from the top of the neural tube on either side of the site of
fusion and migrate throughout the body to form various structures, as we describe in
Chapter 8. The anterior neural tube gives rise to the brain; further back, the neural
tube overlying the notochord will develop into the spinal cord. The formation of noto­
chord, somites, and neural tube is described in Chapter 5. The embryo now begins
to look something like a tadpole and we can recognize the main vertebrate features
(Fig. 3.8). At the anterior end, the brain is already divided up into a number of
­regions, and the eyes and ears have begun to develop. There are also three branchial
arches on each side, the most anterior of which will form the jaws. More posteriorly,
the somites and notochord are well developed. The mouth breaks through at stage 40,
about 2.5 days after fertilization. The post-anal tail of the tadpole is formed last. It
develops from the tailbud, which gives rise to a continuation of notochord, somites,
and neural tube in the tail. Further development gives rise to various organs and
tissues such as blood and heart, kidneys, lungs, and liver. After organ formation, or
100 Chapter 3 Vertebrate development I: life cycles and experimental techniques

organogenesis, is completed, the tadpole hatches out of its jelly covering and begins
to swim and feed. Later, the tadpole larva will undergo metamorphosis to produce the
adult frog; the tail regresses and the limbs grow (metamorphosis is described in more
detail in Chapter 13).

Sagittal section Dorsal view of embryo Transverse section

mesoderm Anterior
archenteron Dorsal
mesoderm
epidermis neural plate

A P

blastocoel yolk plug blastocoel

endoderm Ventral
Posterior
gut cavity notochord gut cavity notochord
Fig. 3.7 Neurulation in amphibians. Top
row: the neural plate develops neural folds
just as the notochord begins to form in the
midline (see middle row). Middle and bottom
rows: the neural folds come together in the
midline to form the neural tube, from which
the brain and spinal cord will develop. During
neurulation, the embryo elongates along the mesoderm blastopore epidermis endoderm
antero-posterior axis. The left panel shows neural tube neural tube
sections through the embryo in the planes notochord lateral plate
mesoderm somite
indicated by the yellow dashed lines in the
center panel. The center panel shows dorsal
surface views of the amphibian embryo.
The right panel shows sections through
the embryo in the planes indicated by the
green dashed lines in the center panel. This
diagram shows neurulation in a urodele epidermis endoderm epidermis endoderm
amphibian embryo rather than Xenopus, as
the neural folds in urodele embryos are more ectoderm mesoderm endoderm notochord
clearly defined.

Fig. 3.8 The early tailbud stage (stage 26)

Stage 26 Xenopus embryo (tailbud stage)
of a Xenopus embryo. At the anterior end, in
the head region, the future eye is prominent
and the ear vesicle (otic vesicle), which
will develop into the ear, has formed. The
brain is divided into forebrain, midbrain, and
hindbrain. Just posterior to the site at which
the mouth will form are the branchial arches, Schematic of Xenopus embryo with epidermis on the left-hand side removed
the first of which will form the lower jaw.
hindbrain ear vesicle notochord somite spinal cord
More posteriorly, a succession of somites Dorsal
midbrain tailbud
lies on either side of the notochord (which
is stained brown in the photograph). The forebrain
embryonic kidney (pronephros) is beginning
to form from lateral mesoderm. Ventral to
Anterior Posterior
these structures is the gut (not visible in
this picture). The tailbud will give rise to the eye vesicle
tail of the tadpole, forming a continuation of
somites, neural tube, and notochord. Scale adhesive organ
bar = 1 mm. branchial arches pronephros Ventral endoderm
(future kidney) epidermis
Photograph courtesy of B. Herrmann.
 Vertebrate life cycles and outlines of development 101

3.2 The zebrafish embryo develops around a large mass of yolk

The zebrafish has two great advantages as a vertebrate model for development: its
short life-cycle of approximately 12 weeks makes genetic analysis, including large-
scale genetic screening relatively easy (see Box 3C, p. 120), while the transparency
of the embryo allows the fate and movements of individual cells during development
Fig. 3.9 Life cycle of the zebrafish.
to be observed. Because of the feasibility of genetic analysis in the zebrafish, it has
The zebrafish embryo develops as a
turned out to be a useful model organism for studying some human diseases caused
mound-shaped blastoderm sitting on top of
by genetic defects, in particular certain blood and cardiovascular disorders. The life a large yolk cell. It develops rapidly and by
cycle is shown in Fig. 3.9. The zebrafish egg is about 0.7 mm in diameter, with a clear 2 days after fertilization the tiny fish, still
animal–vegetal axis: the cytoplasm and nucleus at the animal pole sit upon a large attached to the remains of its yolk, hatches
mass of yolk. After fertilization, the zygote undergoes cleavage, but cleavage does not out of the egg. The top photograph shows
extend into the yolk and results in a mound of blastomeres perched above a large yolk a zebrafish embryo at the sphere stage of
cell. The first five cleavages are all vertical, and the first horizontal cleavage gives rise development, with the embryo sitting on top
to the 64-cell stage about 2 hours after fertilization (Fig. 3.10). of the large yolk cell (scale bar = 0.5 mm).
Further cleavage leads to the sphere stage, in which the embryo is now in the form The middle photograph shows an embryo at
the 14-somite stage, showing developing
of a blastoderm of around 1000 cells lying on top of the yolk cell. The hemispherical
organ systems. Its transparency is useful for
blastoderm has an outer layer of flattened cells, one cell thick, known as the outer
observing cell behavior (scale bar = 0.5 mm).
enveloping layer, and a deep layer of more rounded cells from which the embryo
The bottom photograph shows an adult
develops (Fig. 3.11). During the early blastoderm stage, blastomeres at the margin of zebrafish (scale bar = 1 cm). An illustrated
the blastoderm merge and collapse into the yolk cell, forming a continuous layer of list of the numbered stages in zebrafish
multinucleated non-yolky cytoplasm underlying the blastoderm that is called the yolk development can be found on the website
syncytial layer. The blastoderm, together with the yolk syncytial layer, spreads in a listed in General further reading.
vegetal direction by epiboly and eventually covers the yolk cell. Photographs courtesy of C. Kimmel (top, from
Although the fish blastoderm and the amphibian blastula are different in shape, they Kimmel, C.B., et al.: 1995), N. Holder (middle),
are corresponding stages in development. In the fish, the endoderm is derived from the and M. Westerfield (bottom).

30 minutes
after fertilization

Adult

Cleavage

Sphere
stage
0 2 (1000 cells)
Free 90
swimming
4
Hours after
fertilization Shield
pigment cell 2 stage
Days after
6 (section)
fertilization

Hatching 8 Gastrulation
1
and epiboly
16

75%
epiboly
Body plan (section)
established
Major organs
visible
Organogenesis

14-somite stage
102 Chapter 3 Vertebrate development I: life cycles and experimental techniques

Fig. 3.10 Cleavage of the zebrafish

embryo is initially confined to the animal deep cells that lie right at the blastoderm margin. This narrow marginal layer is often
(top) half of the embryo. known as mesendoderm because individual cells in this region can give rise to both
Photographs courtesy of R. Kessel, from endoderm and mesoderm. Deep cells located between four and six cell diameters away
Kessel, R.G., et al.: 1974. from the margin give rise exclusively to mesoderm. This overlap between prospective
endoderm and mesoderm in the zebrafish differs from the arrangement in Xenopus in
which the endoderm and mesoderm occupy more distinct locations. The ectoderm in the
fish embryo, as in Xenopus, comes from cells in the animal region of the blastoderm.
By about 5.5 hours after fertilization the blastoderm has spread halfway to the vegetal
pole, and the deep-layer cells accumulate to form a thickening around the blastoderm
edge known as the embryonic germ ring (Fig. 3.11, second panel). At the same time,
deep-layer cells within the germ ring converge towards the dorsal side of the embryo,
eventually forming a compact shield-shaped region in the germ ring on the dorsal side
that becomes visible at about 6 hours after fertilization, marking the shield stage. The
shield region is analogous to the Spemann organizer of Xenopus. Gastrulation ensues and
Fig. 3.11 Epiboly and gastrulation in the
the mesendodermal cells and mesoderm cells roll under the blastoderm margin in the pro-
zebrafish. At the end of the first stages of
cess of involution and move into the interior under the ectoderm. The earliest cells that
cleavage, around 4.3 hours after fertilization,
the zebrafish embryo is composed of a involute become endoderm. As the mesendoderm is becoming internalized, the ectoderm
mound of blastomeres sitting on top of continues to undergo epiboly, spreading in the vegetal direction until it covers the whole
the yolk, separated from the yolk by a embryo, including the yolk (discussed in more detail in Chapter 8). Differences between
multinucleate layer of cytoplasm called the gastrulation in the zebrafish and ­Xenopus are that in the fish inward movement of cells
yolk syncytial layer (first panel). With further occurs all around the periphery of the blastoderm at about the same time, and that endo-
cleavage and spreading out of the layers derm is derived predominantly from the dorsal and lateral margins of the blastoderm.
of cells (epiboly), the upper half of the yolk Once internalized into the gastrulating embryo, the mesendodermal cells migrate
becomes covered by a blastoderm with a under the ectoderm towards the animal pole, the tissue converging to form the main
thickened edge, the germ ring, and a shield-
axis of the embryo, and extending and elongating the embryo in an antero-posterior
shaped region is visible in the blastoderm on
direction, as in Xenopus. The future mesodermal and endodermal cells are now beneath
the dorsal side (second panel). Gastrulation
the ectoderm, and by the time that the blastoderm has spread about three-quarters of
occurs by involution of cells in a ring around
the edge of the blastoderm (third panel). the way to the vegetal pole, a single layer of endoderm cells has formed in the embryo
The involuting cells converge on the dorsal next to the yolk with the more superficial cells becoming mesoderm. By 9 hours the
midline to form the body of the embryo notochord has become distinct in the dorsal midline of the embryo, and involution
encircling the yolk (fourth panel).

Blastoderm sits on yolk Spreading of blastoderm by epiboly Gastrulation begins with involution Convergence and extension

Animal pole
deep enveloping
layer layer

yolk syncytial layer shield

region

yolk
germ ring
Ventral Dorsal

Vegetal pole
Epiboly Involution Convergence and extension
 Vertebrate life cycles and outlines of development 103

of cells around the blastoderm margin is complete by 10 hours. Somite formation,

neurulation, and migration of neural crest cells then follow.
Over the next 12 hours the embryo elongates further, and rudiments of the primary
organ systems become recognizable. Somites first appear at the anterior end at about
10.5 hours, and new ones are formed at intervals of initially 20 minutes and then 30 min-
utes; by 18 hours, 18 somites are present. Formation of the gut in zebrafish differs in
several respects from that of Xenopus, chick, and mouse. In zebrafish, the gut starts to
develop at a relatively late stage in gastrulation—at the 18-somite stage—and is formed
by reorganization of cells within the internalized solid mass of endoderm to form a tubu-
lar structure. Neurulation in zebrafish begins, as in Xenopus, towards the end of gastru-
lation with the formation of the neural plate, an area of columnar ectoderm overlying the
notochord. But unlike Xenopus, the whole of the neural plate in the zebrafish first forms
a solid rod of cells that later becomes hollowed out internally to form the neural tube.
The nervous system develops rapidly. Optic vesicles, which give rise to the eyes, can be
distinguished at 12 hours as bulges from the brain, and by 18 hours the body starts to
twitch. At 48 hours the embryo hatches, and the young fish begins to swim and feed.

3.3 Birds and mammals resemble each other and differ from Xenopus in some
important features of early development

Before we describe the course of development in chick and mouse embryos separately,
it is worth emphasizing some features of their development that differ from Xenopus.
The first is the shape of the early embryo. The avian or mammalian structure corre-
sponding to the spherical amphibian blastula just before gastrulation is not a hollow
blastula but a layer of epithelium called the epiblast. The second difference is that
at this stage there are no distinct contiguous regions of the epiblast that correspond
to ectoderm, endoderm, and mesoderm. As we shall discuss in more detail in later
chapters, the timing of specification of the germ layers is somewhat different from
­Xenopus. There is also considerable cell proliferation in the chick and mouse epiblasts
before and during gastrulation, which causes cell mixing at this stage. The third differ-
ence, which is related to both the previous ones, is that the gastrulation process that
leads to internalization of endoderm and mesoderm and the organization of the germ
layers is rather different in appearance from that in Xenopus, with internalization of
cells occurring in a straight furrow, rather than through a circular blastopore.
In mouse and chick, the equivalent to the amphibian blastopore is the primitive
streak. The primitive streak is most easily visualized in the flat epiblast of the chick
embryo and can be seen in the top photograph in Fig. 3.14. At gastrulation, epiblast
cells converge on the primitive streak and pass through it as individual cells, spread-
ing out underneath the surface and forming a bottom layer of endoderm and a middle
layer of mesoderm (Fig. 3.12). Cells become specified as endoderm and mesoderm
during their passage through the streak, with the cells remaining on the surface
becoming ectoderm. Because of the essentially sheet-like form of the avian or mam-
malian epiblast at this stage, this initial phase of gastrulation does not directly form
a gut cavity in the same way as in the spherical Xenopus gastrula. The gut is formed
later, by a folding together of the lateral edges of the embryo, which results in a gut
cavity entirely surrounded by layers of endoderm, mesoderm, and ectoderm. We shall
now return to the chick to look at the course of its development in more detail.

3.4 The early chicken embryo develops as a flat disc of cells overlying a
massive yolk

Avian embryos are very similar to those of mammals in the morphological complexity of
the embryo and the general course of embryonic development, but are easier to obtain
and observe. Many observations can be carried out simply by making a window in the
104 Chapter 3 Vertebrate development I: life cycles and experimental techniques

Fig. 3.12 Ingression of mesoderm and

endoderm during gastrulation in the Hensen’s node primitive streak
chick embryo. Gastrulation begins with the
formation of the primitive streak, a region
epiblast (ectoderm)
of proliferating and migrating cells, which
elongates from the posterior marginal
zone. Future mesodermal and endodermal
cells migrate through the primitive streak
into the interior of the blastoderm. During
gastrulation, the primitive streak extends
about halfway across the area pellucida (see
Fig. 3.14). At its anterior end an aggregation
of cells, known as Hensen’s node, forms.
As the streak extends, cells of the epiblast endoblast
subgerminal
move toward the primitive streak (arrows), space endoderm
move through it, and then outward again migrating cells displacing
mesoderm endoblast
underneath the surface to give rise internally
to the mesoderm and endoderm internally,
the latter displacing the endoblast. ectoderm mesoderm prospective endoderm endoblast
Adapted from Balinsky, B.I., et al.: 1975.

egg shell, and the embryo can also be cultured outside the egg. This is particularly con-
venient for experimental microsurgical manipulations, tracing cell lineages by injecting
marker dyes, and investigations of the effects of introduced genes and other treatments
(see Section 3.6). Despite considerable differences in shape between chick and mouse
embryos in the very early stages of development (see Fig. 3.2), gastrulation and later
development are very similar in both, and studies of chick embryos complement those of
mouse. The chick’s flat early embryo is similar in shape to that of humans (see Fig. 3.2)
and the topology of early human embryonic development is in some ways more easily
compared with that of the chick than with the mouse, which has a cup-shaped embryo.
The large yolky egg cell is fertilized and begins to undergo cleavage while still in
the hen’s oviduct. Because of the massive yolk, cleavage is confined to a small patch
of cytoplasm several millimeters in diameter, which contains the nucleus and lies on
top of the yolk. Cleavage in the oviduct results in the formation of a disc of cells called
the blastoderm or blastodisc. During the 20-hour passage down the oviduct, the egg
becomes surrounded by extracellular albumen (egg white), the shell membranes, and
Structure of the fertilized hen’s egg when laid the shell (Fig. 3.13). At the time of laying, the blastoderm, which is analogous to the
blastoderm (embryo) early amphibian blastula, is composed of some 20,000–60,000 cells. The chick devel-
egg shell
egg white yolk opmental cycle is shown in Fig. 3.14.
The early cleavage furrows extend downward from the surface of the cytoplasm but
do not completely separate the cells, whose ventral faces initially remain open to the
yolk. The central region of the blastoderm, under which a cavity called the subgermi-
nal space develops, is translucent and is known as the area pellucida, in contrast to
the outer region, which is the darker area opaca (Fig. 3.15). A layer of cells called the
yolk hypoblast develops over the yolk to form the floor of the cavity. The hypoblast even-
air sac balancer
tually gives rise to extra-embryonic structures such as those that connect the embryo
shell membranes vitelline membrane
to its source of nutrients in the yolk. The embryo proper is formed from the remaining
upper layer of the blastoderm, known as the epiblast.
Fig. 3.13 The structure of a hen’s egg at
The first morphological structure that presages the antero-posterior polarity of the
the time of laying. Cleavage begins after
embryo is a crescent-shaped ridge of small cells called Koller’s sickle, located at
fertilization while the egg is still in the
the boundary between the area opaca and area pellucida at the posterior end of
oviduct. The albumen (egg white) and shell
are added during the egg’s passage down the
the embryo. Koller’s sickle defines the position in which the streak will form and
oviduct. At the time of laying, the embryo is a the region of the epiblast immediately adjacent to the sickle is known as the poste-
disc-shaped cellular blastoderm lying on top rior marginal zone. The streak is first visible as a denser region that then gradually
of a massive yolk, which is surrounded by the extends as a ­narrow stripe to just over half way across the area pellucida, eventually
egg white and shell. forming a furrow in the dorsal face of the epiblast (see Fig. 3.12).
 Vertebrate life cycles and outlines of development 105

Adult Egg Development

blastoderm within the oviduct
(embryo)

Cleavage

Chick
area
0 opaca
60
area
pellucida
Hours after
fertilization 20 posterior
marginal
21 Days after zone
laying Laying
Hours after Gastrulation
6 laying 16
Hatching
Hensen’s
50 node

Stage 30
primitive
streak

Organogenesis
Stage 4

Stage 14 (22 somites)

Fig. 3.14 Life cycle of the chicken.

Unlike Xenopus, cell proliferation and growth of the chick embryo continues through- The egg is fertilized in the hen and by the
out gastrulation. Epiblast cells converge on the primitive streak, and as the streak time it is laid cleavage is complete and a
moves forward from the posterior marginal zone, cells in the furrow move inward and cellular blastoderm lies on the yolk. After
spread out anteriorly and laterally beneath the upper layer, forming a layer of loosely gastrulation, the primitive streak forms.
connected cells, or mesenchyme, in the subgerminal space (see Figs 3.12 and 3.15). Regression of Hensen’s node is associated
The primitive streak is thus similar to the blastopore region of amphibians, but cells with somite formation. The photographs
move inwards individually, rather than as a coherent sheet, a type of inward move- show: the primitive streak (stained brown
ment known as ingression. The ingressing cells give rise to mesoderm and endoderm, by staining with antibody against Brachyury
protein) surrounded by the area pellucida
whereas the cells that remain on the surface of the epiblast give rise to the ectoderm.
(top, scale bar = 1 mm); a stage 14 embryo
The primitive streak in the chick embryo is fully extended by 16 hours after laying. At
(50–53 hours after laying) with 22 somites
the anterior end of the streak there is a condensation of cells known as Hensen’s node,
(the head region is well defined and the
where cells are also moving inwards. Hensen’s node is the major organizing center for transparent organ adjacent to it is the
the early chick embryo, equivalent to the Spemann organizer in amphibians, and is ventricular loop of the heart; middle,
formed of cells derived from Koller’s sickle as well as cells recruited from the epiblast. scale bar = 1 mm); a stage-35 embryo,
After the primitive streak has elongated to its full length, some cells from Hensen’s about 8.5–9 days after laying, with a
node start to migrate forwards along the midline under the epiblast, to give rise to the well-developed eye and beak (bottom,
prechordal plate mesoderm and the head process. The prechordal plate mesoderm is scale bar = 10 mm). A descriptive list of the
a looser mass of cells anterior to the head process, the anterior part of the notochord. numbered stages (Hamilton and Hamburger
After the head process forms, the primitive streak begins to regress, with ­Hensen’s stages) in chick development can be found on
the website listed in General further reading.
node moving back towards the posterior end of the embryo (Fig. 3.16). As the node
regresses, the notochord is laid down in its wake, lengthening the head process along Top photograph courtesy of B. Herrmann,
from Kispert, A., et al: 1995.
the dorsal midline. As the notochord is laid down, the mesoderm immediately on
each side of it begins to form the somites. These processes are discussed in detail in
Chapters 4 and 5. The first pair of somites is formed at about 24 hours after laying
and new ones are formed at intervals of 90 minutes. The rest of the mesoderm lateral
106 Chapter 3 Vertebrate development I: life cycles and experimental techniques

Fig. 3.15 Cleavage and epiblast formation

cleavage cleavage
in the chick embryo. By the time the egg furrow furrow
is laid, cleavage has divided the small area
of egg cytoplasm free from yolk into a disc-
shaped cellular blastoderm. The panels on
the left show the view of the embryo from
above, while the panels on the right show
Yolk Yolk
cross-sections through the embryo. The first
cleavage furrows extend downward from the
blastoderm
surface of the egg cytoplasm and initially
do not separate the blastoderm completely
from the yolk. In the cellular blastoderm the
central area overlying the subgerminal space
is called the area pellucida and the marginal
subgerminal space
region is called the area opaca. A layer of
cells develops immediately overlying the
yolk and is known as the hypoblast. This
area
will give rise to extra-embryonic structures, opaca
while the upper layers of the blastoderm—the
epiblast—give rise to the embryo. The chick area
pellucida
primitive streak starts to form when the
hypoblast becomes displaced forward from posterior
marginal
the posterior marginal zone by a new layer zone
of cells called the secondary hypoblast or Laying Laying
endoblast, which grows out from the zone.
The cells of the epiblast move toward the epiblast
primitive streak (arrows), move through it,
and then outward again underneath the Koller’s
surface to give rise to the mesoderm and sickle
endoderm internally, the latter displacing the
endoblast.
hypoblast

primitive
Fig. 3.16 Regression of Hensen’s streak cell movement
primitive
node. After extending about halfway across streak
the blastoderm, the primitive streak begins
to regress, with Hensen’s node moving in
a posterior direction as the head fold and
neural plate begin to form anterior to it. As mesenchyme (migrating cells) endoblast
the node moves backward, the notochord
develops in the area anterior to it and ectoderm mesoderm prospective endoderm endoblast
somites begin to form on either side of the
notochord.

Anterior
Hensen’s node

head fold
head
process

somites

regressing
primitive streak
primitive streak
Posterior
 Vertebrate life cycles and outlines of development 107

Fig. 3.17 Development of the neural tube

neural fold and mesoderm in the chick embryo. Once
the notochord has formed, neurulation
mesenchyme begins in an anterior to posterior direction.
neural groove The figure shows a series of sections along
the antero-posterior axis of a chick embryo.
foregut
Neural-tube formation is well advanced at the
notochord anterior end (top two sections), where the
extra-embryonic head fold has already separated the future
mesoderm
head from the rest of the blastoderm and
head fold the ventral body fold has brought endoderm
from both sides of the body together to form
neural plate the gut. During neurulation, the neural plate
changes shape: neural folds rise up on either
endoderm side and form a tube when they meet in
the midline. The mesenchyme in this region
will give rise to head structures. Further
back (middle sections), in the future trunk
region of the embryo, notochord and somites
neural fold
have formed and neurulation is starting. At
neural groove the posterior end, behind Hensen’s node
(bottom section), notochord formation,
somite
somite formation, and neurulation have
intermediate not yet begun. The mesoderm internalized
mesoderm
through the primitive streak starts to form
coelom structures appropriate to its position along
somatic the antero-posterior and dorso-ventral axes.
mesoderm
For example, in the future trunk region,
splanchnic
mesoderm the intermediate mesoderm will form the
mesodermal parts of the kidney, and the
somite
anterior splanchnic mesoderm will give rise to
the heart. The body fold will continue down
neural folds the length of the embryo, forming the gut
and also bringing paired organ rudiments that
Hensen’s node initially form on each side of the midline (e.g.
those of the heart and dorsal aorta) together
to form the final organs lying ventral to the
mesoderm gut. Blood islands, from which the first blood
blood island cells are produced, form from the ventral-
most part of the lateral mesoderm.
primitive groove
After Patten, B.M.: 1971.
primitive ridge

primitive
streak

to the somites is the lateral plate mesoderm, and this will develop into organs such
as heart, kidneys, and the vascular system and blood (Fig. 3.17). The node eventually
forms a center of stem cells in the tailbud, which gives rise to the post-anal tail. Fig. 3.18 Scanning electron micrograph
As the notochord forms, neural tissue begins to develop, first as the neural plate, which of chick showing early somites and neural
is first evident as a region of columnar ectoderm above the notochord. The neural plate tube. Blocks of somites can be seen adjacent
then folds upwards and inwards so that its sides come together and eventually fuse in the to the neural tube, with the notochord lying
dorsal midline to form the neural tube, initially leaving the anterior and posterior ends beneath it. The lateral plate mesoderm flanks
open. The fused neural tube becomes covered over by epidermis; Fig. 3.18 shows a section the somites. Scale bar = 0.1 mm.
through the chick embryo showing the fused neural tube and the notochord beneath it. Photograph courtesy of J. Wilting.
108 Chapter 3 Vertebrate development I: life cycles and experimental techniques

Just after the head process (head notochord) appears and Hensen’s node starts to
hindbrain
ear midbrain regress, all three germ layers in the head region start to fold ventrally, to begin to gen-
branchial erate the head fold (see Fig. 3.16, third image from left). This delimits a pouch lined
arches by endoderm, from which the pharynx and foregut will arise (see top of Fig. 3.17).
eye Eventually, a similar fold appears at the tail end to define the hindgut. In addition,
the sides of the embryo fold together to form the rest of the gut, which only remains
forebrain
open in the middle (umbilical region) and subsequently the mesoderm and ectoderm
grow over to form the ventral body wall. This key morphogenetic event is known as
heart ventral closure. As this folding takes place, the two heart rudiments that start out on
either side come together at the midline to form one organ lying ventral to the gut. By
wing bud just over 2 days after laying, the embryo has reached the 22-somite stage, the head is
well-developed, optic vesicles and auditory pits are present, and the heart and blood
vessels have formed. Blood vessels and blood islands, where the first blood cells are
being formed, have developed in the extra-embryonic tissues; the vessels connect up
somites
with those of the embryo to provide a circulation with a beating heart. At this stage
leg bud the embryo starts to turn on its side; the head develops a flexure so that the right eye
comes to lie uppermost towards the shell.
tailbud By 3–3.5 days after laying, 40 somites have formed, the head is now much more
developed with prominent eyes, and the limbs are beginning to develop (Fig. 3.19).
At 4 days after laying, extra-embryonic membranes have developed through which
Fig. 3.19 The 40-somite stage of the the embryo gets its nourishment from the yolk and which also provide protection
chick embryo. Development of the head (Fig. 3.20). The amnion surrounds a fluid-filled amniotic sac in which the embryo lies
region and the heart are quite well advanced and which provides mechanical protection; the chorion lies outside the amnion just
and the wing and leg buds are present as beneath the shell; the allantois receives excretory products and provides the site of
small protrusions. oxygen and carbon dioxide exchange; and the yolk sac surrounds the yolk. By about
10 days after laying, the embryo is very well-developed; the wings, legs, and beak are
now formed (see Fig 3.14, bottom panel. In the remaining time before hatching, the
embryo grows in size, the internal organs become fully developed, and down feathers
grow on the wings and body. The chick hatches 21 days after the egg is laid. A picture
Fig. 3.20 The extra-embryonic structures of a live quail embryo inside the egg obtained by magnetic resonance imaging can be
and circulation of the chick embryo. seen in Fig. 3.27.
A chick embryo at 4 days of incubation is
depicted in situ. The embryo lies within
the fluid-filled amniotic cavity enclosed by
the amnion, which provides a protective
chamber. The yolk is surrounded by the shell
yolk sac membrane. The vitelline vein takes
nutrients from the yolk sac to the embryo shell membrane amnion
and the blood is returned to the yolk sac
via the vitelline artery. The umbilical artery
O2 amniotic cavity
takes waste products to the allantois and CO2
chorion
the umbilical vein brings oxygen to the embryo
embryo. The arteries are shown in red allantois
and the veins in blue but this does not heart
denote the oxygenation status of the blood.
As the embryo grows, the amniotic cavity
enlarges; the allantois also increases in size
and its outer layer fuses with the chorion,
the membrane under the shell, while at the vitelline artery
umbilical artery
same time the yolk sac shrinks. Note that in umbilical vein vitelline vein
this diagram the allantois has been enlarged
so that the umbilical vessels can be shown
clearly.
yolk sac
After Patten, B.M.: 1951
 Vertebrate life cycles and outlines of development 109

3.5 Early development in the mouse involves the allocation of cells to form
the placenta and extra-embryonic membranes

The mouse has a life cycle of 9 weeks from fertilization to mature adult (Fig. 3.21), Fig. 3.21 The life cycle of the mouse. The
egg is fertilized in the oviduct, where cleavage
which is relatively short for a mammal. This makes genetic analysis relatively easy
also takes place before implantation of the
and is one reason the mouse has become the main model organism for mammalian
blastocyst in the uterine wall at 5 days after
development and was the first mammal after humans to have its complete genome fertilization. Gastrulation and organogenesis
sequenced. As in all our model organisms, knowledge of the genome sequence then take place over a period of about
enables profiles of all the genes expressed at a particular stage in development to 7 days and the remaining 6 days before birth
be determined (Box 3A, p. 115). Particular advantages of the mouse in this respect are largely a time of overall growth. After
are the relative ease with which mice with a particular genetic constitution can be gastrulation the mouse embryo undergoes a
routinely produced using transgenic techniques (see Section 3.9), and the abil- complicated movement known as turning, in
ity to produce mice in which particular genes are rendered completely inactive or which it becomes surrounded by its extra-
‘knocked out’. embryonic membranes (not shown here).
Mammalian eggs are much smaller than either chick or Xenopus, about 80–100 µm The photographs show (from top): a fertilized
mouse egg just before the first cleavage
in diameter for both mouse and human, and they contain no yolk. The unfertilized egg
(scale bar = 10 µm); the anterior view of a
is shed from the ovary into the oviduct and is surrounded by a protective external coat,
mouse embryo at E8 (scale bar = 0.1 mm); and
the zona pellucida, which is composed of mucopolysaccharides and glycoproteins. a mouse embryo at E14 (scale bar = 1 mm).
Fertilization takes place internally in the oviduct; meiosis is then completed and the An illustrated list of the stages in mouse
second polar body forms (see Box 9A, p. 338). Cleavage starts while the fertilized egg development can be found on the website
is still in the oviduct. Early cleavages are very slow compared with Xenopus and chick, listed in General further reading.
the first occurring about 24 hours after fertilization, the second about 20 hours later, Photographs courtesy of: T. Bloom (top, from
and subsequent cleavages at about 12-hour intervals. Cleavage produces a solid ball of Bloom, T.L.: 1989); N. Brown (middle); and
J. Wilting (bottom).

Adult Fertilized
polar bodies egg zona pellucida

Cleavage
Blastocyst

0
Birth 50 2
Implantation
4
uterine
Days after wall
18 fertilization 6

16 8
Fetal
growth and Gastrulation
development 14 10
12
Turning

Organogenesis
110 Chapter 3 Vertebrate development I: life cycles and experimental techniques

Two-cell Four-cell Eight-cell Compacted morula Blastocyst

polar
body

Fig. 3.22 Cleavage in the mouse

embryo. The photographs show the cells, or blastomeres, called a morula (Fig. 3.22). At the eight-cell stage the ­blastomeres
cleavage of a fertilized mouse egg from the increase the area of cell surface in contact with each other in a process called compac-
two-cell stage through to the formation of tion. After compaction, the cells are polarized; their exterior surfaces carry microvilli,
the blastocyst. After the eight-cell stage, whereas their inner surfaces are smooth. Further cleavages are somewhat variable and
compaction occurs, forming a solid ball of
are both radial and tangential, so that by the equivalent of the 32-cell stage the morula
cells called the morula, in which individual
contains about 10 internal cells and more than 20 outer cells.
cell outlines can no longer be discerned. The
internal cells of the morula give rise to the
A special feature of mammalian development is that the early cleavages give rise
inner cell mass, which can be seen as the to two distinct groups of cells—the trophectoderm and the inner cell mass. The
compact clump at the top of the blastocyst. internal cells of the morula give rise to the inner cell mass and the outer cells to
It is from this that the embryo proper forms. the trophectoderm. The trophectoderm will give rise to extra-embryonic structures,
The outer layer of the hollow blastocyst—the such as the placenta, through which the embryo gains nourishment from the mother.
trophectoderm—gives rise to extra-embryonic The embryo proper develops from a subset of cells in the inner cell mass. At this
structures. stage (E3.5) the mammalian embryo is known as a blastocyst (see Fig. 3.21). Fluid
Photographs courtesy of T. Fleming. is pumped by the trophectoderm into the interior of the blastocyst, which causes the
trophectoderm to expand and form a fluid-filled cavity (the blastocoel) containing the
inner cell mass at one end.
From E3.5 to E4.5 the inner cell mass becomes divided into two regions. The sur-
face layer in contact with the blastocoel becomes the primitive endoderm, and will
contribute to extra-embryonic membranes, whereas the remainder of the inner cell
mass—the primitive ectoderm or epiblast—will develop into the embryo proper as
well as giving rise to some extra-embryonic components. At this stage, about E4.5, the
embryo is released from the zona pellucida and implants into the uterine wall.
The course of early post-implantation development of the mouse embryo from
around E4.5 to E8.5 appears more complicated than that of the chick, partly because
of the need to produce a larger variety of extra-embryonic membranes, and partly
because the epiblast is distinctly cup-shaped in the early stages. This is a peculiarity
of mouse and other rodent embryos. Human and rabbit embryos, for example, are flat
blastodiscs, much more resembling that of the chick (see Fig. 3.2). Despite the differ-
ent topology, however, gastrulation and the later development of the mouse embryo
are in essence very similar to that of the chick.
The first 2 days of mouse post-implantation development are shown in Fig. 3.23.
After the initial adhesion of the blastocyst to the uterine epithelium, the cells of the
mural trophectoderm—the region not in contact with the inner cell mass—replicate
their DNA without cell division, giving rise to primary trophoblast giant cells that
invade the uterus wall and surround most of the conceptus, forming an interface
with the maternal tissue. The uterus wall envelops the blastocyst, and the polar
trophectoderm cells in contact with the inner cell mass continue to divide, forming
the ectoplacental cone and the extra-embryonic ectoderm, which both contribute
to the placenta. The outer cells of the ectoplacental cone differentiate into secondary
trophoblast giant cells. Some cells from the primitive endoderm migrate to cover the
whole inner surface of the mural trophectoderm. They become the parietal endo-
derm, which eventually becomes Reichert’s membrane, a sticky layer of cells and
 Vertebrate life cycles and outlines of development 111

Blastocyst at time of implantation Implanted embryo 5 days Egg cylinder 5.5 days Onset of gastrulation 6 days

Extra-embryonic

polar ectoplacental trophectoderm

trophectoderm cone Anterior Posterior
extra- extra-
embryonic embryonic
ectoderm ectoderm extra-
epiblast inner embryonic
cell visceral visceral ectoderm
primitive endoderm endoderm
endoderm mass
epiblast visceral primitive
blastocoel endoderm P streak
parietal epiblast
endoderm epiblast
(embryonic
trophoblast parietal Dorsal ectoderm)
mural giant cells endoderm side of
trophectoderm embryo proamniotic
A cavity
Embryonic

Fig. 3.23 Early post-implantation development of the mouse in contact with the epiblast proliferates and forms extra-embryonic
embryo. First panel: before implantation, the fertilized egg has tissues—the ectoplacental cone and extra-embryonic ectoderm—which
undergone cleavage to form a hollow blastocyst, in which a small contribute to the placenta. The epiblast elongates and develops an
group of cells, the inner cell mass, will give rise to the embryo, while internal cavity (proamniotic cavity), which gives it a cup-shaped form.
the rest of the blastocyst forms the trophectoderm, which will develop Third panel: the cylindrical structure containing both the epiblast and
into extra-embryonic structures. At the time of implantation the inner the extra-embryonic tissue derived from the polar trophectoderm is
cell mass divides into two regions: the primitive ectoderm or epiblast, known as the egg cylinder. Fourth panel: the beginning of gastrulation
which will develop into the embryo proper, and the primitive endoderm, is marked by the appearance of the primitive streak (brown) at the
which will contribute to extra-embryonic structures. Second panel: the posterior of the epiblast (P). It starts to extend anteriorly (arrow)
mural trophectoderm gives rise to trophoblast giant cells, which invade towards the bottom of the cylinder. For simplicity, the parietal
the uterine wall, helping to anchor the blastocyst to it. The blastocyst endoderm and trophoblast giant cells are not shown in this panel or in
becomes surrounded by the uterine wall. The polar trophectoderm subsequent figures.

e­ xtracellular ­matrix that has a barrier function. The remaining primitive endoderm
cells form the visceral endoderm, which covers the elongating egg cylinder contain-
ing the epiblast.
By E5, an internal cavity has formed inside the epiblast, which then becomes
­cup-shaped—U-shaped when seen in cross-section (see Fig. 3.23, second and third
panels). The epiblast is now a curved single layer of epithelium, which at this stage
contains about 1000 cells. The embryo develops from this layer. The first easily vis-
ible sign that marks the future antero-posterior axis is the appearance of the primi-
tive streak at about E6. The streak starts as a localized thickening at the edge of the
cup on one side; this side will correspond to the future posterior end of the embryo.
The initial development of the primitive streak in the mouse is similar to that in
the chick. Over the next 12–24 hours it elongates until it reaches the bottom of the
cup. A condensation of cells—the node—becomes distinguishable at the anterior end
of the extended streak and corresponds to Hensen’s node in the chick embryo. To
make primitive streak formation easier to compare with that of the chick, imagine
the epiblast cup spread out flat. In mouse gastrulation, as in chick, epiblast cells con-
verge on the primitive streak, and proliferating cells migrate through it to spread out
laterally and anteriorly between the ectoderm and the visceral endoderm to form a
mesodermal layer (Fig. 3.24). Development from E7 is shown in Fig. 3.25.
Some epiblast-derived cells pass through the streak and enter the visceral endo-
derm, gradually displacing it to form the definitive embryonic endoderm on the
outside of the cup, which is the future ventral side of the embryo. Cells migrating
anteriorly from the node form the head process and notochord, while prospective
mesodermal cells from the region surrounding the node migrate anteriorly to form the
somites. Cell proliferation continues during gastrulation, and the embryo anterior to
the node grows rapidly in size; as in the chick, the node eventually becomes a center
of stem cells in the tailbud, giving rise to the post-anal tail.
112 Chapter 3 Vertebrate development I: life cycles and experimental techniques

Early primitive streak (6 days after fertilization) Early head process (7.5 days after fertilization)

visceral
endoderm mesoderm mesoderm
primitive
streak

Posterior
Anterior
primitive
streak

mesoderm

epiblast notochord
(head process) node

Fig. 3.24 Gastrulation in the mouse embryo. Left panel: as in the chick, gastrulation in the
mouse embryo begins when epiblast cells converge on the posterior of the epiblast and move
under the surface, forming the denser primitive streak (brown) where the cells are becoming
internalized. Once inside, the proliferating cells spread out laterally between the epiblast
surface and the visceral endoderm to give a layer of prospective mesoderm (light brown).
Some of the internalized cells will eventually replace the visceral endoderm to give definitive
endoderm (not shown on these diagrams for simplicity), which will form the gut. Right panel: as
gastrulation proceeds, the primitive streak lengthens and reaches the bottom of the cup, with
the node at the anterior end. The node gives rise to the notochord, which forms the structure
known as the head process. Part of the visceral endoderm and the mesoderm has been cut
away in this diagram to show the node and notochord. Note that, given the topology of the
mouse embryo at this stage, the germ layers appear inverted (ectoderm on the inner surface of
the cup, endoderm on the outer) if compared with the frog gastrula.

7.2 days after fertilization (section) 7.7 days (lateral view) 8 days (embryo before turning) 10.5 days (lateral view, after turning)

cranial neural folds posterior A

stem cell fore mid
A P A P A zone P
brain brain
allantois optic
allantois
amniotic remains of vesicle otic
cavity primitive vesicle
heart streak
neurectoderm bulge branchial
mesoderm amnion arches
head mesoderm gut tailbud
proamniotic primitive fold endoderm notochord
cavity streak definitive
endoderm hindlimb forelimb
visceral somites
node endoderm notochord node bud P bud

Fig. 3.25 Schematic views of the early development of the mouse there has been further growth of the embryo anterior to the node, the
embryo to the completion of gastrulation and neurulation. First head is distinct, the neural folds have formed, the foregut and hindgut
panel: by 7.5 days, the primitive streak (brown) has extended to have closed, and somites are beginning to form on either side of the
the bottom of the epiblast and the node has formed. The anterior notochord. The embryo is covered by a layer of ectoderm that will form
ectoderm (blue) becomes prospective neurectoderm, which will give the epidermis, which is not shown on this diagram. Fourth panel: by
rise to the brain and spinal cord. Mesoderm is shown as light brown. 10.5 days, gastrulation and neurulation are complete. The embryo has
Second panel: the anterior part of the embryo grows in size and the undergone a complicated turning process around day 9 that brings the
head fold appears. Definitive endoderm (green) replaces visceral dorsal and ventral sides into their final positions. Scale bars for the
endoderm (yellow) to form an outer layer on the ventral surface of the first three panels = 100 µm; fourth panel, scale bar = 75 µm.
embryo. The notochord (red) begins to form. Third panel: by 8 days
 Experimental approaches to studying vertebrate development 113

8.5 days after fertilization 3–4 hours later 9 days after fertilization 9.5 days after fertilization

placenta
allantois
parietal yolk sac
visceral yolk sac

amnion
gut

Fig. 3.26 Turning in the mouse

Somite formation and organogenesis start at the anterior end of the mouse embryo embryo. Between 8.5 and 9.5 days, the
and proceed posteriorly. The first somite is formed at around E7.5 and new ones are mouse embryo becomes entirely enclosed in
formed at intervals of 120 minutes. At around E8, neural folds have also started to the protective amnion and amniotic fluid. The
form at the anterior end on the dorsal side of the embryo and a head is apparent (see visceral yolk sac, a major source of nutrition,
Fig. 3.25). The embryonic endoderm—initially exposed on the ventral surface of the surrounds the amnion and the allantois
embryo—becomes internalized to form the foregut and hindgut, the lateral surfaces connects the embryo to the placenta.
eventually folding together to internalize the gut completely with subsequent over- After Kaufman, M.H.: 1992.
growth of mesoderm and ectoderm to form the ventral body wall in the process of
ventral closure. The heart and liver move into their final positions relative to the gut,
and the head becomes distinct. By about midway through embryonic development,
gastrulation and neurulation are complete: the mouse embryo has a distinct head
and the forelimb buds are starting to develop. A complicated ‘turning’ process has
occurred around E9 that produces a more recognizable mouse embryo, surrounded
by its extra-embryonic membranes (Fig. 3.26). As a result of turning, the initial cup-
shaped epiblast has turned inside out so that the dorsal surface is now on the out-
side; the ventral surface, with the umbilical cord that connects it to the placenta, is
facing inwards. (Turning is another developmental quirk peculiar to rodents; human
embryos are surrounded by their extra-embryonic membranes from the beginning.)
Organogenesis in the mouse proceeds very much as in the chick embryo. From fertil-
ization to birth takes around 18 to 21 days, depending on the strain of mouse.

Experimental approaches to studying vertebrate development

This part of the chapter introduces some of the main techniques used to study the
developmental biology of vertebrates. It is not intended to be comprehensive, but to
indicate general experimental approaches and to describe in more detail a few tech-
niques that are very commonly used. Rather than reading it straight through, you may
wish to use it as a reference to come back to, to put into context some of the experi-
ments described in the rest of the book. Many of the techniques described here are
also applicable to non-vertebrates (see Chapters 2 and 6).
The earliest studies of vertebrate embryonic development involved careful and ­detailed
study of whole and dissected embryos under the microscope, and simply watching,
describing, and drawing the development of easily accessible embryos such as those of
newts and frogs. From the seventeenth century onwards, observations of animal and
human embryos by zoologists and physicians gradually yielded the comprehensive ana-
tomical descriptions of embryonic development that are the foundation of modern devel-
opmental biology. Hensen’s node and Koller’s sickle (see Section 3.4), for example, are
named after embryologists working in the early twentieth century. Careful observation is
just as important in developmental biology today, even if the need to draw by hand the
images you see in the microscope has been superseded by digital-imaging technology.
Once ways of detecting the expression of genes in situ were developed in the lat-
ter half of the twentieth century—both riboprobes to detect the gene transcripts and
114 Chapter 3 Vertebrate development I: life cycles and experimental techniques

i­mmunohistochemistry using antibodies to detect the expression of the proteins the

transcripts encode (see Box 1D, p. 20), anatomical descriptions of normal develop-
ment could be supplemented by mapping the expression patterns of developmentally
important genes. Projects are currently under way to produce comprehensive online
‘atlases’ of the normal expression patterns of hundreds of genes at different devel-
opmental stages for the mouse and the chick (see General further reading). Tech-
niques such as DNA microarray analysis can also detect and identify the expression
of large numbers of gene at the same time and so can provide information about all
the genes being expressed in a particular tissue or at a particular stage in development
(Box 3A, p. 115). Observation on its own cannot unravel the mechanisms underlying
developmental processes, however, and the only way to find out more about these
is to disturb the developmental process in some specific way and see what happens.
Techniques for interfering with development can be very broadly divided into two
types, and many experiments in developmental biology will use a combination of
these two approaches. Classical experimental embryological techniques manipulate
the embryo by physical intervention—by removing or adding cells to cleavage-stage
embryos or transplanting blocks of cells from one embryo to another, for example. The
other class of techniques is based on genetics and genomics, and is used to disturb
the expression of developmentally important genes by mutation, gene silencing, over-
expression or misexpression—­expressing the gene at a time or place where it is not
normally expressed.
We shall start by taking a brief look at experimental manipulation in Xenopus and
chick embryos, and then describe some of the techniques from molecular biology,
genetics, and genomics that have revolutionized the study of developmental biology
over the past few decades. Other technical advances that have benefited develop-
mental biology over the past 20 years are the great improvement in computer-aided
microscopic imaging techniques, the development of fluorescent labels in a vast range
of colors that allow the imaging of living embryos, and the introduction of new forms
of imaging into the field, such as magnetic resonance imaging (MRI) and optical pro-
jection tomography (OPT). Images of live avian embryos inside the egg can now be
obtained by magnetic resonance imaging (Fig. 3.27).

3.6 Not all techniques are equally applicable to all vertebrates

Fig. 3.27 A live 9-day quail embryo inside

You will soon notice as you read further in this book that experiments involving the
its egg. The image on the left was obtained microsurgical manipulation of vertebrate embryos largely feature Xenopus and the
by magnetic resonance imaging (MRI). chick. Few spontaneous developmental mutations are known in these animals, and
Photograph courtesy of Suzanne Duce. chick and amphibian embryos were being studied long before developmental genes had

lens beak
vitreous chamber of eye
wing
brain humerus
heart ventricle
ophthalmic vessel liver
spinal cord
allantois
gizzard
heart ventricle
liver
yolk
 Experimental approaches to studying vertebrate development 115

Box 3A Gene-expression profiling by DNA microarray

In this book we deal with the general prin-

ciples underlying development so we do
not describe every gene known so far to
be involved in laying down the body plans
of the animals we cover. However, now that
the genomes of the model vertebrates have Unfertilized Fertilized 2-cell 4-cell 8-cell Morula Blastocyst
egg egg
been sequenced, systematic genome-wide
approaches are being used to identify all
Extract mRNA. Copy into cDNA and amplify
the genes involved in a particular develop-
mental process. Genomic sequence informa-
tion can also be used to identify the targets
of the transcription factors that coordinate Hybridize to genomic microarray
gene expression in space and time.
The identification of all the genes expres­
sed in a particular tissue or at a particular Compare with a universal genomic reference sample to determine which genes are expressed at each stage
stage of development can be accomplished
by carrying out genome-wide screens for
gene expression using DNA microarrays,
often known as DNA chips. This technology enables the levels ­amplified as RNA or converted into cDNA and amplified by the
of RNA transcripts of thousands of genes to be measured simul- polymerase chain reaction (PCR). The nucleic acid is then tagged
taneously, and the main use of microarrays in developmental with a fluorescent dye and hybridized to the microarray. A refer-
biology is to monitor the changes in gene expression that ­occur ence sample of mRNA, similarly treated and tagged with a differ-
within a tissue or embryo at different stages of development ent fluorescent dye, is also hybridized to the same microarray.
or after experimental manipulation. Microarrays for studying The microarray is then scanned and the ratios of the signals from
gene expression come in various formats, but the most widely both dyes at each spot are recorded and converted into a relative
used are flat surfaces studded with a regular array of clusters or expression level for the test sample.
‘spots’ of DNA fragments of known sequence, each cluster repre- One such experiment, for example, aimed to find all the genes
senting a DNA sequence found in a specific protein-coding gene. expressed at particular stages in early mouse development.
These DNA fragments are known as the ‘probes,’ and a microar- Genes expressed in the unfertilized egg, fertilized egg, two-cell
ray can contain thousands to hundreds of thousands of differ- stage, four-cell stage, eight-cell stage, morula, and blastocyst
ent probes. Microarrays composed of hundreds of thousands of have been identified by extracting mRNA from 500 eggs or
oligonucleotide probes specially synthesized to cover the whole embryos at the appropriate stage, then labeling it and hybrid-
genome sequence of mouse and human are now commercially izing it to miniaturized microarrays of unique oligonucleotides of
available. Alternatively, for more limited applications, probes can 60 bases each, representing all of the mouse genes (see figure).
be made from cDNAs that have been made by reverse transcrip- The binding patterns of the labeled RNAs indicate which genes
tion of mRNA and so represent just the genes being expressed are expressed at the different stages. Figure 3.35 illustrates
by a given tissue. another technique that uses DNA microarrays to help recon-
To determine which genes are being expressed, the total struct the gene regulatory networks involved in particular devel-
mRNA from the tissue of interest is extracted, then either opmental processes.

been identified. X. laevis is also unsuitable for conventional genetic analysis because
of its long generation time (1–2 years) and its tetraploid genome. Amphibian and avian
embryos are, however, robust to surgical manipulation, and easily accessible to the
experimenter at all stages in their development, unlike those of mammals (Fig. 3.28).
Microsurgical manipulations in Xenopus, such as the removal of particular blastomeres,
or the grafting of cells from one embryo into another, provided much of the information
about the location and function of organizer regions in the blastula and their role in
development; several experiments of this type are illustrated in Chapter 4. Blastomere
removal identifies regions of the embryo that are essential for further development,
while transplantation experiments are used to test the developmental potential of a
116 Chapter 3 Vertebrate development I: life cycles and experimental techniques

The suitability of different vertebrate embryos for classical experimental embryological manipulation

Frog (Xenopus laevis) Zebrafish Chick Mouse

Are living embryos easily Yes. Eggs can be obtained and Yes. As for Xenopus For most of it. Eggs are fertilized Only very early stages. In vitro
available and observable at all externally fertilized and embryos internally and so the very early fertilized eggs can be grown in
stages of their development? develop to swimming tadpole stage stages of development—to culture up to the blastocyst stage,
in the aquarium. Large batches of blastoderm stage—occur within the after which they have to be replaced
eggs can be fertilized at the same mother hen. Laid eggs are easily in a surrogate mother mouse to
time and develop synchronously to obtainable, can be incubated in the continue development. Early isolated
give large numbers of embryos at the laboratory, and the development of embryos can be cultured for a limited
same developmental stage. the embryo observed at any time up time (see Section 3.6).
to hatching. Early embryos can also
be cultured up to about stage 10 out
of the egg.

Is the embryo amenable to Embryo can be manipulated Amenable to surgical manipulation Can be manipulated surgically up to Only very early embryos up to
experimental manipulation, e.g. surgically up to the neurula stage up to the neurula stage and to some quite late stages in embryonic blastocyst stage can be manipulated
blastomere removal or addition, (see Chapter 4). Embryos are extent in later stages. The embryo is development, including limb if they are to be replaced in a
or tissue grafting? comparatively large, and are highly transparent, which helps in observing development (see Chapters 4, 5, 11, surrogate mother for further
resistant to infection after developmental changes and the and 12). development (see Figs 3.34 and
microsurgery movement of labeled cells 4.12). Isolated postimplantation
embryos can be cultured for a limited
time (see Section 3.6).

Is fate mapping and lineage Yes, in the early embryo. Single cells Yes, as in Xenopus Relatively easy and cells can be In the past it was difficult. Now
tracing relatively easy? in the early blastula can be injected labeled at quite late embryonic becoming easier owing to transgenic
with a fluorescent non-toxic dye or stages by dye injection; the use of techniques for restricting reporter
RNA encoding fluorescent protein chick–quail chimeric embryos; grafts gene expression to particular cells
(GFP) (see Chapter 4) from GFP chickens (see below) or (see Section 3.9 and Box 3B) and also
electroporation of DNA encoding a inducible transgenes.
reporter protein or fluorescent
protein (see Section 3.7).

Fig. 3.28 The suitability of the

different model vertebrates for classical particular region of the embryo or to determine the time at which the fate of cells in
experimental embryological manipulation. a particular location becomes irreversibly determined (see Section 1.12). The cells of
early amphibian embryos carry their own store of nutrients in the form of yolk platelets
and so explants of embryonic tissue can be cultured in a simple salt medium for several
days. This allows experiments that investigate the inducing effects of one tissue on
another when placed together in culture, and experiments of this type in Xenopus are
described in Chapters 4 and 5. Similar explant experiments have been carried out with
chick embryonic tissue embedded in collagen gels and using cell culture medium.
The early chick embryo (blastoderm stage) can be removed from the egg and cultured
on its vitelline membrane with thin albumen in a watch glass. The embryo will then
develop for up to about 36 hours (Hamilton and Hamburger stage 10). Experiments in
which pieces of one chick embryo are grafted into another at different stages have also
provided much of the information on the organizer properties of different regions of
the embryo and the generation of the primitive streak (discussed in Chapters 4 and 5).
Older embryos can be accessed through a window cut in the egg shell and manipulated
while remaining in the egg (Fig. 3.29). The window is then re-sealed with adhesive tape
and the embryo allowed to continue development so that the effects of the procedures
can be assessed. The accessibility of the chick embryo allows many types of experi-
mental procedures: part of a developing structure such as a limb bud can be removed;
tissue can be grafted to new sites; and the effects of chemicals such as growth factors
on development can be investigated by implanting small inert beads soaked in these
agents into a specific location in the embryo (examples of all these types of experiment
Fig. 3.29 A chick embryo can be observed
in the study of chick limb development are described in Chapter 11).
through a window in the eggshell. The
After implantation, mouse development in vivo is hidden from view and can only
almost transparent 3.5-day embryo
(H&H stage 21) can be just seen lying on top
be followed by isolating embryos at different stages (see Fig. 4.12) and cells can be
of the yolk. After an experimental procedure injected into the inner cell mass of the blastocyst (see Section 3.9). Tissues from
has been carried out, the window can be very early embryos can be cultured for a limited time. For example, inner cell masses
sealed with adhesive tape and the embryo isolated from 3.5-day embryos and cultured in vitro for several days will develop cer-
left to develop further. tain embryonic and extraembryonic tissues and some structures typical of a normal
 Experimental approaches to studying vertebrate development 117

e­ mbryo at the corresponding stage. Mouse embryos between stages E6.5 and E12.5
have been cultured for 24–48 hours in roller tubes, which allows manipulations such
as node transplantation. Experimental investigation of living embryonic tissues at
later stages requires dissection of embryos and the growth of tissue or organ explants
(such as limb buds, embryonic kidney, lung) in vitro.

3.7 Fate mapping and lineage tracing reveal which cells in the early embryo
give rise to which adult structures

Fate maps like those that will be discussed in Chapter 4 show what a particular region
of an embryo will develop into during normal development. For both biological and
technical reasons, fate mapping in the early embryo is easiest in Xenopus. At the blas-
tula stage, cells that will eventually form the three germ layers are arranged in distinct
regions and are accessible from the surface of the embryo (see Fig. 3.6). Single blas-
tomeres can be easily labeled with a non-toxic marker such as DiI (which fluoresces
red) or fluorescein-linked dextran (see Fig. 4.18) or by expression of a fluorescent
protein such as GFP by injecting GFP RNA.
Fate mapping by dye injection into individual cells is more difficult in the chick
embryo as the cells are much smaller, but it has been used to trace the lineage of
cells derived from Hensen’s node. Chick–quail chimeras have also been used for cell-
­lineage tracing. Chick and quail embryos are very similar in their development and
early embryos composed of a mixture of chick and quail cells will develop normally
in culture. The cells of the two species can be distinguished by the different appear-
ance of their nuclei; quail cells have a prominent nucleolus that stains red with an
appropriate stain (see Fig. 5.6). Quail cells can also be distinguished from chick cells
by immunohistochemical staining with labeled antibodies for quail proteins. For lin-
eage tracing, cells are taken from a particular site in an early quail embryo and trans-
planted into the same site in a chick embryo of the same age. Later-stage embryos
are then stained and sectioned to determine where the quail cells have ended up.
Recently, transgenic chicken embryos expressing GFP in all cells have become avail-
able (see Section 3.9) and can be used for long-term fate mapping.
Defined small populations of cells, or even single cells, can also be conveniently marked
in chick embryos in the egg (in ovo) or in culture by introducing a DNA encoding a fluo-
rescent protein. Despite the small cell size in chick embryos compared with Xenopus
early blastomeres, DNA constructs can be targeted to a relatively small number of cells by
the technique of electroporation. The DNA is injected into the embryo at the desired site
with a fine pipette and pulses of electric current applied using very fine electrodes. The
current makes the membranes of nearby cells permeable to the injected DNA.
Cell-lineage tracing and fate mapping by dye injection or electroporation of nucleic
acids is technically much more difficult in mammalian embryos than in Xenopus or
the chick because the embryos are less accessible inside the mother. Nevertheless
these same techniques have been used to follow cell fate in mouse embryos between
E6.5 and E8.5, stages that can be maintained in culture (see Section 3.6). Researchers
have found other ingenious ways around the problem, although some of the tech-
niques used required a great deal of skill. For example, in experiments carried out
in the early 2000s to study the migration of neurons in the developing brain, a DNA
vector containing a GFP gene was injected into the lateral ventricles of the brains of
mouse embryos in utero (at around stage E14), accompanied by electroporation, with-
out harming the embryos. Sections of the brains of later-stage embryos and of mouse
pups after birth were examined by fluorescence microscopy to reveal the locations of
the GFP-labeled neurons, which originate in the layer of cells immediately under the
ventricle surface (see Chapter 12).
Lineage tracing in the mouse is now done much more conveniently using transgenic
embryos expressing a reporter gene in a particular set of cells under the control of the
118 Chapter 3 Vertebrate development I: life cycles and experimental techniques

Cre/loxP system (Box 3B) or a reporter gene linked to a suitable promoter (see Sec-
tion 3.9). Techniques such as these are continually being refined to give better resolu-
tion, and have proved particularly useful for tracing later events in embryogenesis,
such as organ development, the development of the nervous system, and cell behavior
in epithelia in which cells are continually being replaced, such as skin and gut (see
Section 10.7). Clonal cell-lineage analysis is now becoming possible using a transgenic
mouse in which a drug-inducible Cre recombinase (linked to the reporter gene LacZ
(see Box 1D, p. 20) or to GFP) is expressed in every cell; treatment of the pregnant
mouse with an appropriate dose of the drug will lead to a low frequency of recombina-
tion and generate clones of marked cells in the embryos. This strategy led to the discov-
ery of dorso-ventral compartments in the developing mouse limb (see Chapter 11).

Box 3B Insertional mutagenesis and gene knock-outs in mice: the Cre/loxP system

One very powerful and widely used way of

targeting a gene knock-out to a specific tis-
target gene
sue and/or a particular time in development
is provided by the Cre/loxP system. The tar-
get gene is first ‘floxed’ by inserting a loxP loxP loxP
sequence of 34 base pairs on either side of
the gene (hence ‘floxed’, for ‘flanked by lox        ’).
+ Cre
This targeting is carried out using embry-
onic stem cells in culture (see Section 3.9)
and transgenic mice are then produced by
introducing the genetically modified cells
into blastocysts (see Fig. 3.34) and ­deriving Heart-specific Cre Ubiquitous inducible Cre
lines of transgenic mice. The line of mice
with the floxed gene is then crossed with
another line of transgenic mice carrying
the gene for the recombinase Cre. loxP
sequences are recognized by Cre, and the + inducer
resulting recombination reaction excises all
the DNA between the two loxP sites. In the
offspring, if Cre is expressed in all cells, then
all cells will excise the floxed target, causing
a knock-out of the target gene in all cells. target gene expressed target gene deleted
However, if the gene for Cre is under the
control of a tissue-specific promoter, so that, for example, it is only However, it is unlikely that any gene is without any value at
expressed in heart tissue, the target gene will only be excised in all to an animal. It is much more likely that there is an altered
heart tissue (see figure). If the Cre gene is linked to an inducible ­phenotype in these apparently normal animals, which is too
control region, it is possible to induce excision of the target gene subtle to be detected under the artificial conditions of life in a
at will by exposing the mice to the inducing stimulus. laboratory. Redundancy is thus probably apparent rather than
A significant number of knock-outs of a single gene result in real. A further complication is the possibility that, under such cir-
mice developing without any obvious abnormality, or with fewer cumstances, related genes with similar functions may increase
and less severe abnormalities than might be expected from the their activity to compensate for the mutated gene.
normal pattern of gene activity. A striking example is that of As well as knocking out a coding sequence, the Cre/loxP
myoD, a key gene in muscle differentiation, which is described in method can be used to target promoters in such a way that the
Chapter 10. In myoD knock-outs, the mice are anatomically nor- associated gene is permanently turned on. This technique is
mal, although they do have a reduced survival rate. The most likely used in cell-lineage studies in mice to turn on a reporter gene
explanation for this is that the process of muscle differentiation like GFP or lacZ in particular cells at a particular stage in develop-
contains a certain amount of redundancy and that other genes can ment. The progeny of the labeled cells can then be distinguished
substitute for some of the functions of myoD (see Section 10.8). in later-stage embryos or after birth.
 Experimental approaches to studying vertebrate development 119

3.8 Developmental genes can be identified by spontaneous mutation and by

large-scale mutagenesis screens

Rare spontaneous mutations were for a long time the only way that genetic distur-
bances of development could be studied, and such mutations have been identified
in mice (see Fig. 1.12) and in chickens. Informative spontaneous mutations are rare,
however, even in mice, and as they are only noticed if the animal survives birth, muta-
tions in important developmental genes that lead to the death of a mouse embryo
in utero (embryonic lethal mutations) are likely to be missed. A lethal develop-
mental mutation in chicken was only discovered because of the reduced hatchability
of the eggs. Many more developmental mutations have been produced by inducing
random mutation in large numbers of organisms by chemical treatments or irradiation
by X-rays, and then screening for mutants of developmental interest. The aim, where
possible, is to treat a large enough population so that, in total, a mutation is induced
in every gene in the genome. This sort of approach is best used in organisms that
breed rapidly and that can conveniently be obtained and treated in very large num-
bers. Among vertebrates, this approach has been applied to both mice and zebrafish.
Zebrafish in particular, are a potentially valuable vertebrate system for large-scale
mutagenesis because large numbers can be handled, and the transparency and large
size of the embryos makes it easier to identify developmental abnormalities (Box 3C,
p. 120). However, unlike the genetic screening in Drosophila described earlier in the
book (see Box 2A, p. 42), there are no genetic means so far of automatically eliminat-
ing unaffected individuals in zebrafish. This means that all the progeny of a cross
have to be examined visually for any developmental abnormality.
Mutagenesis screens like those for zebrafish are also carried out in mice, but
because of the mouse’s larger size, longer generation time, and the fact that embryos
cannot be screened so easily, they are more difficult and costly to set up. Neverthe-
less, many such screens have been carried out. Chemical mutagenesis generates both
dominant mutations with varying degrees of gene function, which are particularly
relevant to human conditions, and recessive lack-of-function mutations. Dominant
mutations can be recognized in offspring in the F1 generation (Fig. 3.30), but more
complicated breeding programs like that described for zebrafish have to be used to
reveal recessive mutations.

male mice are

injected with the
mutagen ENU

a b c a b c a b c a* b c a b* c a b c*

mutagenized males
wild-type oocytes are mated to sperm carrying mutations*
wild-type females

Fig. 3.30 A mutagenesis screen for

a* b c a b* c a b c* dominant mutations in the mouse. Progeny
a b c a b c a b c G1 screening for
dominant mutations with a dominant mutation, such as the
dark brown coat color shown here, can be
detected in the first generation (G1). ENU,
ethylnitrosurea.
120 Chapter 3 Vertebrate development I: life cycles and experimental techniques

Box 3C Large-scale mutagenesis in zebrafish

A screening program using zebrafish

­involves breeding for three generations
(see figure). Male fish treated with a Mutagen
chemical mutagen are crossed with wild- treatment
of spermatogonia
type females; their F1 male offspring are
crossed again with wild-type females, and
the female and male siblings from each Gametes +*
of these crosses are themselves crossed.
The offspring from each of these pairs
are ­examined separately for homozygous F1 + +
mutant phenotypes. If the F1 fish carry a
+* +
mutation, then in 25% of the F2 matings
two heterozygotes will mate and 25%
of their offspring will be homozygous for
the mutation. Zebrafish can also be made
to develop as haploids by fertilizing the
F2 +or+ +or+
egg with sperm heavily irradiated with + +* + +*
UV light. This allows one to detect early-
acting recessive mutations without having Possible sibling matings
to breed the fish to obtain homozygous
+ + + + + + + +
embryos. +* + + +* + + +* +*
Two new techniques can be used to iden-
tify or create mutations in zebrafish genes
about which nothing may be known apart F3
from their sequences. TILLING (targeting-
induced local lesions in genomes) is used
to identify mutations generated by chemi-
cal mutagenesis in particular genes. First
developed for plants, it has also been
applied to Drosophila, Caenorhabditis, and
the rat. All that is necessary is knowledge of the DNA sequence mutation are used to generate families of fish with that muta-
of the gene. Male fish treated with a chemical mutagen are tion, which reveal any phenotypic change associated with it.
crossed with wild-type females. A sample of tissue from each F1 A relatively novel method for generating mutations relies on
male, which will be heterozygous for any point mutations gener- engineering targeting specificity into zinc-finger nucleases
ated, is collected and, at the same time, its sperm are frozen and (ZFNs), which cleave DNA, and using them to cleave and disrupt
stored. Mutations in genes of interest are detected by first mak- chosen genes in vivo. This approach has been used in zebrafish to
ing large amounts of the mutated DNA by the polymerase chain target the No-tail (Ntl ) gene, the zebrafish version of Brachyury.
reaction (PCR) and then hybridizing the mutated DNA with unmu- Like Brachyury in the mouse (see Fig. 1.13), No-tail is required
tated DNA from the same gene. Mismatched bases will reveal the for tail formation. About 30% of zebrafish embryos injected at
site of a mutation and the DNA is then sequenced to identify the the one-cell stage with mRNA for Ntl-specific ZFN lacked tails,
precise change. Males and/or frozen sperm corresponding to each showing that Ntl had been rendered non-functional.

3.9 Transgenic techniques enable animals to be produced with mutations in

specific genes

To study the function of a particular gene, it is more efficient to be able to specifi-

cally alter that gene in the animal rather than having to trust to random mutagenesis.
Strains of mice with a particular mutant genetic constitution can now be produced
 Experimental approaches to studying vertebrate development 121

Fig. 3.31 Chicks transgenic for green

fluorescent protein (GFP). Transgenic
chickens can be obtained by injecting DNA
carried by a self-inactivating lentiviral vector
into the subgerminal space beneath the
blastodisc in newly laid eggs. The chicks
illustrated are second-generation offspring
from an original transgenic bird, and carry the
GFP transgene in the germline. The bird in
the center is not transgenic.
Photograph from McGrew, J.M. et al.: 2004.

relatively routinely, and as long as the mutation is recessive, animals heterozygous

for the ­mutation can be maintained as breeding stock, even for mutations that may be
lethal to the embryo when homozygous. Animals into which an additional or altered
gene has been introduced are known as transgenic animals. Transgenic techniques
are most highly developed in mice, and we shall describe transgenesis here in rela-
tion to mice. Transgenic zebrafish can be obtained by injecting DNA into the one-
cell-stage embryo and the first transgene to be successfully delivered into zebrafish
was the ­delta-crystallin gene of the chicken. Transgenesis has quite recently been
demonstrated in principle in chickens, by the injection of lentiviral vectors contain-
ing the gene encoding GFP under the blastoderm of a newly laid egg (Fig. 3.31). And
although X. laevis is less amenable to germline transgenesis, the transgenic techniques
that have been developed for its close relative X. tropicalis, which is diploid and has
a much shorter generation time of 5–9 months, can also be used. For Xenopus, the
strategy is to introduce the DNA transgene into sperm nuclei and then transplant
these nuclei into unfertilized eggs. The relative suitability of our model vertebrates for
genetics-based techniques is summarized in Fig. 3.32.

The suitability of the different model vertebrates for genetics-based techniques for studying development

Frog (Xenopus laevis) Zebrafish Chick Mouse

Genome sequenced The diploid X. tropicalis genome has Yes Yes Yes
been sequenced

Spontaneous mutations No Yes Yes Yes

Induced mutations Yes Yes (see Box 3C) No Yes

(mutational screens)

Gene silencing or gene knock-out Gene silencing by morpholino antisense RNAs (MOs) (Box 6A). Gene silencing by MOs and RNA Gene knock-out by Cre/loxP system
in somatic cells interference (RNAi) (Box 6A) (requires making transgenic
animals first, Box 3B)

Germline gene knock-out No Yes Not yet Yes

(transgenic animals)

Gain-of-function By injection of mRNAs into the fertilized egg or very early embryo Not routinely possible so far but Transgenic mice with gain-of-
(e.g. misexpression and ectopic (see e.g. Fig. 4.4) transgenic chicks have been function mutations can be made
expression) of specific genes in produced (Fig. 3.31)
the whole embryo

Targeting of gene overexpres- Expression can be targeted to some No Yes in somatic tissue Transgenic mice with inducible or
sion or misexpression in time extent by injection of RNAs into cell-type specific expression of a
and space specific blastomeres transgene can be made

Embryonic stem cells can be No No Yes Yes. Mutant transgenic mice can also
cultured and differentiated in be made using ES cells mutated by
vivo homologous recombination (see
Section 3.9 and Chapter 10)

Fig. 3.32 The suitability of the different model vertebrates for genetics-based study of development.
122 Chapter 3 Vertebrate development I: life cycles and experimental techniques

Two main techniques for producing transgenic mice are currently in use. One is to
inject a transgene composed of DNA encoding the required gene and any neces-
sary regulatory regions directly into the male pronucleus of a fertilized egg, and then
­replace the egg in a surrogate mother. If the transgene becomes incorporated into the
genome it will be present in all the cells of the embryo, including the germline, and
will be expressed according to the promoter and other regulatory regions it contains.
In this way, mice that overexpress a given gene in particular cells or at particular times
during development can be produced, and the effect of misexpressing a gene in a tis-
sue or at a time where it is not usually active can be investigated.
A newer technique for producing transgenic mice uses embryonic stem cells (ES
cells) that have been mutated in vitro, and this is now the most commonly used
transgenic technique for producing specific gene knock-outs—the permanent dele-
tion of a gene function. ES cells are pluripotent cells derived from the inner cell mass
of a mouse blastocyst; they can be maintained in culture indefinitely and grown in
large numbers and are being investigated for their possible use in regenerative medi-
cine (discussed in Chapter 10). ES cells injected into the cavity of an early blastocyst
become incorporated into the inner cell mass and so will become part of all the
embryo’s tissues, even giving rise to germ cells and gametes. For example, if ES cells
from a brown-pigmented mouse are introduced into the inner cell mass of a blastocyst
from a white mouse, the mouse that develops from this embryo will be a chimera of
‘brown’ and ‘white’ cells. In the skin, this mosaicism is visible as patches of brown
and white hairs.
To generate transgenic mice with a particular mutation, the ES cells are mutated
while growing in culture and the mutated cells are introduced into a blastocyst, which
is then placed in a surrogate mother to continue its development. Mutations are tar-
geted to a particular gene by the technique of homologous recombination, in which
specially tailored DNA constructs are introduced into the cell by transfection. A trans-
fected DNA molecule will usually insert randomly in the genome, but by including
some sequence that is homologous with the desired target gene, it is possible to make
it insert at a predetermined site (Fig. 3.33). Homologous recombination between the
transfected DNA and the target gene in the ES cell results in an insertion that renders

Construct targeting DNA Targeting DNA introduced Select for drug-resistant cells
into ES cells

cloned gene exon

cells with
targeted insertion
drug-resistance
gene

targeting DNA
Fig. 3.33 Production of mutant mouse ES
cells by homologous recombination. The
gene to be targeted for gene knock-out is targeting
DNA
first cloned and a drug-resistance or other homologous
selectable marker is inserted into it, which recombination
also renders the gene non-functional. When exon target gene in
the modified gene is introduced into cultured chromosome
mouse ES cells, it will undergo homologous
recombination with the corresponding gene mutated
chromosome
in a small number of the ES cells. These can inactivated gene
then be selected by their drug resistance.
 Experimental approaches to studying vertebrate development 123

the gene non-functional, and this is now the main way of producing gene knock- ES cells (carrying a mutation in
outs. Homologous recombination works particularly well in mice. The DNA to be a single gene) in culture
introduced must contain enough sequence homology with its target to insert within
it in at least a few cells in the ES cell culture. The introduced DNA usually carries a
drug-­resistance gene so that cells containing the insertion can be selected by ­adding
the drug, which kills the other, unmodified, cells. The mutated ES cells are then
introduced into the inner cell mass of a blastocyst, producing a transgenic mouse car-
rying a mutation in a known gene (Fig. 3.34). When one gene is replaced by another
functional gene using these techniques, it is called a gene knock-in.
Because the animals produced by ES-cell transfer are chimeras composed of both
ES cells injected into inner cell
mutant and normal cells, they may show few, if any, effects of the mutation. If the mass of normal blastocyst
mutant gene has entered the germline, however, strains of mice heterozygous for the
altered gene can be intercrossed to produce homozygotes that are viable or fail to
develop, depending on the gene involved. We shall see examples of gene knock-outs
in mice being used to study the functions of Hox genes in regionalizing the somitic
mesoderm and the neural tube in Chapters 5 and 12.

3.10 Gene function can also be tested by transient transgenesis Chimeric animal produces gametes
and gene silencing carrying the mutation

Germline transgenesis is not the only way of altering gene expression, and produc-
ing transgenic mice is still a relatively expensive process. Techniques of transient
transgenesis or somatic transgenesis are easier and cheaper to implement, and are
particularly valuable in amphibians and the chick, in which germline transgenesis
has not been a practical option. The early Xenopus embryo has large blastomeres
and is very suitable for experiments in which the mRNA of a gene suspected of hav-
ing a developmental role is injected into a specific blastomere to see the effect of its
overexpression. mRNA injection is also used in Xenopus and other embryos to see if a
specific gene can ‘rescue’ normal development after a region of the early embryo has
Fig.3.34 Production of transgenic mice
been removed or the embryo’s own gene has been silenced. Experiments of this type
by the introduction of mutated ES cells
are described in Chapter 4.
into the blastocyst. ES cells injected into a
Transient transgenesis for both short-term and long-term overexpression or mis-
blastocyst will become part of the inner cell
expression of genes is also widely used to study gene function in the chick embryo. mass, and can give rise to all the cell types
For short-term expression, genes are delivered into cells by electroporation (see Sec- in the mouse, including the germline. The
tion 3.6). For longer-term overexpression, the gene is inserted into a replication- first generation of transgenic mice will be
­competent retroviral vector, which is then injected into the embryo. The infected cells chimeras of the blastocyst genotype and the
express the gene as part of the viral replication cycle and, as the virus replicates, it ES cell genotype, but if they carry the mutant
spreads the gene to neighboring cells. gene in the germline, they can be interbred
The transient counterpart to the gene knock-out is the technique of gene silencing over several generations to produce strains of
or gene knockdown, in which a gene is not mutated or removed from the genome, heterozygous or homozygous transgenic mice.
but its expression is blocked by targeting the mRNA and preventing its translation.
Techniques for doing this are discussed in Box 6A, p. 220. Because it is not possible to
generate loss-of-function mutations in Xenopus and chick embryos, gene silencing by
the injection of morpholino antisense RNAs is particularly useful and this technique
is also widely used in zebrafish. Morpholinos are stable RNAs designed to be comple-
mentary to a specific mRNA. When introduced into the cells of an embryo, they
hybridize with the target mRNA, thus preventing its translation into protein. Injection
of a morpholino into a fertilized Xenopus egg will knock down gene expression in all
the cells of the early embryo; electroporation of morpholinos into cells of early chick
embryos will produce local knockdown. The technique of RNA interference, which
is also described in Box 6A (p. 220) is also used in embryology. RNAi constructs
encoding the interfering RNAs have been electroporated into the neural tube of chick
embryos to knock down gene expression.
124 Chapter 3 Vertebrate development I: life cycles and experimental techniques

Fig. 3.35 ChIP-chip and ChIP-seq. These two variants of chromatin immunoprecipitation
Cross-link proteins to genomic DNA in vivo
are used to discover the binding sites in DNA for transcription factors and other DNA-binding
proteins. After cross-linking any bound proteins to the DNA by chemical treatment, the DNA
is fragmented and treated with an antibody specific for the protein of interest. The antibody
proteins complex is purified and the DNA extracted. It is then amplified by the polymerase chain reaction
(PCR) and either analyzed by DNA microarray (ChIP-chip) or by DNA sequencing (ChIP-seq).

3.11 Gene regulatory networks in embryonic development can be revealed by

chromatin immunoprecipitation techniques
Fragment genomic DNA and incubate with antibody
specific for a given transcription factor One powerful new approach to investigating development relies on a combination
of bioinformatics and experimental molecular and cell biology to describe the net-
antibody work of interactions between genes and their transcription factors, and so identify
gene regulatory networks. Once the genome sequence of an animal is known, the
technique of chromatin immunoprecipitation followed by DNA microarray analysis
(ChIP-chip), can be used to identify the target genes that a given transcription fac-
tor binds to in vivo (Fig. 3.35). Embryos at a given embryonic stage are treated so
that proteins bound to the cis-regulatory control elements in the DNA are chemically
linked to the DNA. The DNA is then isolated and sheared into small segments and
Immunopurify treated with an antibody specific for a given transcription factor. This will precipitate,
or ‘pull down,’ all the DNA segments to which that transcription factor was bound
in the embryo’s cells. The identity of the DNA can then be determined by digesting
away the protein, labeling the DNA segment and hybridizing it to DNA microarrays
containing genomic sequences lying upstream of known genes (see Box 3A, p. 115).
Alternatively, the protein-free DNA can be sequenced, when the technique is known
as ChIP-seq.
The associated gene can then be identified by searching the databases of genome
sequences. ChIP-seq has been used, for example, to identify the gene targets of the
transcription factor Ntl (No-tail), the zebrafish equivalent of Brachyury (see Fig. 1.13),
in mid-gastrula zebrafish embryos, and thus begin to uncover the gene regulatory net-
work involved in mesoderm specification and patterning in this model animal (see
Amplify DNA by PCR Box 4E, p. 162).

Summary to Chapter 3

All vertebrates have a similar basic body plan. The defining vertebrate structures are the
vertebral column, which surrounds the spinal cord, and the bony or cartilaginous skull
that forms the head and encloses the brain. The vertebrate body has three main axes
of symmetry—the antero-posterior axis running from head to tail, the dorso-ventral axis
running from back to belly, and bilateral symmetry around the dorsal midline, so that
Sequence
Fluorescently label DNA outwardly the right and left sides are mirror images of each other. Some internal organs,
and hybridize to DNA such as lungs, kidneys, and gonads, are also present as symmetrically paired structures,
Identify by searching microarray to identify
on a database but single organs, such as the heart and liver, are arranged asymmetrically with respect
to the dorsal midline. Early vertebrate embryos pass through a set of developmental
ChIP-seq ChIP-chip
stages—cleavage, gastrulation, and neurulation—to form embryos that broadly resemble
each other—a stage known as the phylotypic stage. There are considerable differences
in the development of the different model organisms before the phylotypic stage, which
relate to how and when the axes are set up, and how the germ layers are established.
These differences are mainly due to the different modes of reproduction of the animals
and the consequent form of the earliest embryo. Structures typical of the ­different
­vertebrate groups, such as fins, beaks, wings, and tails, develop after the phylotypic
 Experimental approaches to studying vertebrate development 125

stage. The body pattern and the initial organ rudiments are laid down when the embryo
is still very small. Growth in size takes place at later stages. The mechanisms ­underlying
­development can only be studied by disturbing the normal developmental process in
­specific ways and observing what happens. Techniques for interfering with develop-
ment can be broadly divided into two types: experimental embryological techniques that
manipulate the embryo by removing or adding cells to cleavage-stage embryos or trans-
planting blocks of cells from one embryo to another; and genetics-based techniques that
disturb the expression of developmentally important genes by mutation, gene silencing,
­overexpression or misexpression.

n End of chapter questions 10. Briefly summarize the production of transgenic mice and of
knock-out mice. How do the two processes differ, both in technique
Long answer (concept questions) and in outcome?
1. All vertebrates share certain features; what are those features?
Multiple choice (factual recall questions)
Referring to Fig. 3.2, identify those features in the row of illustra-
tions that shows the phylotypic stage. NB There is only one right answer to each question.

2. After fertilization, all embryos undergo cleavage. In the Xenopus

­ 1. In which portion of the frog’s life cycle would a frog appear most
embryo, what is the relationship of the animal-vegetal axis to the similar to a mammal?
plane of the first three cleavages? Describe how cleavage in the a) blastula
chick embryo differs from cleavage in frogs and mice; why do these
b) phylotypic stage
differences occur?
c) gastrulation
3. Compare the terms blastula, blastoderm, and blastocyst. To what
d) fertilized egg
organisms do they refer? What features do they share, and how do
they differ? 2. The darkly pigmented end of a Xenopus egg is called
a) dorsal
4. Gastrulation in Xenopus leads to the establishment of the three
germ layers through the process of involution. What are the three b) the animal pole
germ layers, and what is their relationship to each other in the gas- c) the vegetal pole
trula (stage 11)? What is the archenteron, and what will it become? d) yolk
5. Distinguish between the following regions in the chick embryo: 3. Which of the following choices puts the early developmental
epiblast and hypoblast; posterior marginal zone, Koller’s sickle, and stages of the frog in their correct order?
Hensen’s node; endoblast and endoderm; mesenchyme and meso-
a) blastula, cleavage, gastrula, neurula
derm.
b) cleavage, gastrulation, neurulation, organogenesis
6. Hensen’s node of the gastrulating chick embryo is conceptually
c) neurulation, gastrulation, cleavage, organogenesis
analogous to the blastopore of the Xenopus embryo, but despite this
analogy, several differences between the two structures exist. Elabo- d) tadpole, embryo, neurula, gastrula
rate on the following differences: (a) the ingression of cells in the 4. Cleavage in the zebrafish gives rise to the
chick versus the involution of cells in the frog; (b) the movement
a) 14-somite stage
of Hensen’s node in the chick versus the lack of movement of the
blastopore; (c) the formation of the notochord in two species. b) gastrula
c) sphere stage
7. During neurulation in the chick embryo, the precursors to several
adult structures first form. What structures will form from: ­neural d) shield stage
plate; somites; intermediate mesoderm; splanchnic mesoderm; 5. The primitive streak of mouse and chick embryos is equivalent to
notochord? (See Fig. 3.17.) which amphibian structure?
8. Distinguish between the inner cell mass and the trophectoderm in a) archenteron
the mouse embryo. Outline the events that will lead to the ­formation b) blastocoel
of the epiblast and a primitive streak. c) blastopore
9. What are some of the features that would make an animal species d) neural folds
a good ‘model organism’ for the study of development? For example,
6. At the blastocyst stage, the mammalian embryo is composed of
what features make a chicken an attractive model for development
compared to frogs? In contrast, what features does the mouse offer a) a blastocoel, yolk, and animal cap cells
over chickens? b) an epiblast overlying a hypoblast
126 Chapter 3 Vertebrate development I: life cycles and experimental techniques

c) ectodermal, mesodermal, and endodermal cells ­flipbook of zebrafish embryogenesis. Development 1996, 123:
d) trophectoderm and inner cell mass 1–461 [http://www.bio.umass.edu/biology/karlstrom/Karlstrom-
Lab.html] (accessed 13 May 2010).
7. The identification of all the genes expressed in a particular tissue Chick: list and description of Hamilton & Hamburger developmen-
or at a particular stage of development can be done using tal stages: UNSW Embryology – Chicken Developmental Stages
a) in situ hybridization [http://embryology.med.unsw.edu.au/Otheremb/chick1.htm]
b) insertional mutagenesis (accessed 13 May 2010).
Mouse: Edinburgh Mouse Atlas Project [http://genex.hgu.mrc.
c) microarrays
ac.uk] (accessed 13 May 2010).
d) zinc-finger nucleases Human: The Multidimensional Human Embryo [http://embryo.
8. The advantage of using the Cre/loxP system for targeted muta- soad.umich.edu] (accessed 13 May 2010); UNSW Embryology:
genesis is Carnegie staging of human embryos [http://embryology.med.
unsw.edu.au] (accessed 13 May 2010).
a) a gene can be deleted in a particular tissue or at a desired time
in development
b) the Cre/loxP system allows germline transgenesis in amphibians n Section further reading
and chickens
c) the Cre/loxP system is more specific than homologous recom- 3.1 The frog Xenopus laevis is the model amphibian for
bination developmental studies
d) the deletion of a gene by the Cre/loxP system is reversible Hausen, P., Riebesell, H.: The Early Development of Xenopus laevis.
Berlin: Springer-Verlag, 1991.
9. The least favorable organism for traditional genetic analysis is Nieuwkoop, P.D., Faber, J.: Normal Tables of Xenopus laevis.
a) Drosophila ­Amsterdam: North Holland, 1967.
b) mouse
3.2 The zebrafish embryo develops around a large mass of
c) Xenopus
yolk
d) zebrafish Kimmel, C.B., Ballard, W.W., Kimmel, S.R., Ullmann, B., Schilling,
10. Electroporation is T.F.: Stages of embryonic development of the zebrafish. Dev.
Dyn. 1995, 203: 253–310.
a) fusion of adjacent cells using an electrical pulse
Warga, R.M., Nusslein-Volhard, C.: Origin and development of the
b) injection of a gene into the nucleus of a cell zebrafish endoderm. Development 1999, 26: 827–838.
c) the introduction of a fluorescent lineage maker into cells Westerfield, M. (Ed.): The Zebrafish Book; A Guide for the Labora-
d) use of brief pulses of electricity to cause cells to take up DNA tory Use of Zebrafish (Brachydanio rerio). Eugene, Oregon: Uni-
versity of Oregon Press, 1989.

Multiple choice answer key 3.4 The early chicken embryo develops as a flat disc of cells
1: b, 2: b, 3: b, 4: c, 5: c, 6: d, 7: c, 8: a, 9: c, 10: d. overlying a massive yolk
Bellairs, R., Osmond, M.: An Atlas of Chick Development (2nd edn).
London: Academic Press, 2005.
n General further reading Chuai, M., Weijer, C.J.: The mechanisms underlying primitive
streak formation in the chick embryo. Curr. Top. Dev. Biol. 2008,
Bard, J.B.L.: Embryos. Color Atlas of Development. London: Wolfe, 81: 135–156.
1994. Hamburger, V., Hamilton, H.L.: A series of normal stages in the
Carlson, B.M.: Patten’s Foundations of Embryology. New York: development of a chick. J. Morph. 1951, 88: 49–92.
McGraw-Hill, 1996. Lillie, F.R.: Development of the Chick: An Introduction to Embryol-
ogy. New York: Holt, 1952.
Descriptions of developmental stages of frog, zebrafish, chick Patten, B.M.: The Early Embryology of the Chick. New York: McGraw-
and mouse embryos Hill, 1971.
Xenopus: developmental stages: Xenbase – Nieuwkoop and Faber Stern, C.D.: Cleavage and gastrulation in avian embryos (ver-
stage series [http://www.xenbase.org/xenbase/original/atlas/ sion 3.0). Encyclopedia of Life Sciences 2009, http://www.els.net/
NF/NF-all.html] (accessed 13 May 2010); movies of cleavage and (13 May 2010).
gastrulation [http://www.xenbase.org] (accessed 13 May 2010).
Zebrafish: developmental stages Karlstrom Lab [http://www.bio. 3.5 Early development in the mouse involves the allocation
umass.edu/biology/kunkel/fish/zebra/devstages.html] (accessed of cells to form the placenta and extra-embryonic membranes
13 May 2010); ZFIN Embryonic Developmental Stages [http:// Cross, J.C., Werb, Z., Fisher, S.J.: Implantation and the placenta:
zfin.org/zf_info/zfbook/stages/stages.html] (accessed 13 May key pieces of the developmental puzzle. Science 1994, 266:
2010); developmental movie: Karlstrom, R.O., Kane, D.A.: A 1508–1518.
 Experimental approaches to studying vertebrate development 127

Kaufman, M.H.: The Atlas of Mouse Development (2nd printing). Butte, A.: The use and analysis of microarray data. Nat. Rev.
London: Academic Press, 1994. Microarray Collection 2004 [http://www.nature.com/reviews/
Kaufman, M.H., Bard, J.B.L.: The Anatomical Basis of Mouse Devel- focus/microarrays/index.html] (accessed 13 May 2010).
opment. London: Academic Press, 1999.
Box 3B Insertional mutagenesis and gene knock-outs in mice:
Experimental approaches to studying vertebrate the Cre/loxP system
development Yu, Y., Bradley, A.: Engineering chromosomal rearrangements in
Hogan, H., Beddington, R., Costantini, F., Lacy, E.: Manipulating the mice. Nat. Rev. Genet. 2001, 2: 780–790.
Mouse Embryo. A Laboratory Manual (2nd edn). New York: Cold
Box 3C Large-scale mutagenesis in zebrafish
Spring Harbor Laboratory Press, 1994.
McGrew, M.J., Sherman, A., Ellard, F.M., Lillico, S.G., Gilhooley, Doyon, Y., McCammon, J.M., Miller, J.C., Faraji, F., Ngo, C., Katibah,
H.J., Kingsman, A.J., Mitrophanous, K.A., Sang, H.: Efficient pro- G.E., Amora, R., Hocking, T.D., Zhang, L., Rebar, E.J., Gregory,
duction of germline transgenic chickens using lentiviral vec- P.D., Urnov, F.D., Amacher, S.L.: Heritable targeted gene dis-
tors. EMBO Rep. 2004, 5: 728–733. ruption in zebrafish using designed zinc-finger nucleases.
Nat. Biotechnol. 2008, 26: 702–708.
Sharpe, P., Mason, I.: Molecular Embryology Methods and Protocols
Stemple, D.: Tilling—a high throughput harvest for functional
(2nd edn). New York: Humana Press; 2008.
genomics. Nat. Rev. Genet. 2004, 5: 1–6.
Southall, T.D., Brand, A.H.: Chromatin profiling in model organ-
isms. Brief. Funct. Genomics Proteomics 2007, 6: 133–140.
Stern, C.D.: The chick: a great model system just became even
greater. Dev. Cell 2004, 8: 9–17.

Box 3A Gene-expression profiling by DNA microarray

Hamatani, T., Carter, M.G., Sharov, A.A., Ko, M.S.: Dynamics of
global gene expression changes during mouse preimplantation
development. Dev. Cell 2004, 6: 117–131.
4
Vertebrate development II:
axes and germ layers
nn Setting up the body axes nn The origin and specification
of the germ layers

In this chapter we examine the mechanisms that specify the body axes and germ
­layers of the early embryo in Xenopus, zebrafish, chick, and mouse. We will first
look at axis formation and explore the relative roles of maternal factors in the egg,
external factors, and the embryo’s own developmental program. One of the key pro-
cesses of early vertebrate development is the development of the three germ layers—­
ectoderm, endoderm, and mesoderm—from the fertilized egg, and we shall see how
the different groups of animals achieve this in somewhat different ways. Despite its
outward bilateral symmetry, the vertebrate body has a left–right internal asymmetry,
as indicated by the positions of the heart and liver, and we shall see that the origins
of this asymmetry are established very early in embryonic life.

Vertebrate development proceeds with respect to well-defined antero-posterior (head

to tail) and dorso-ventral (back to belly) axes. In this chapter we will look in detail
at the first two key events in vertebrate development. The first is the establishment of
these axes, and the second is the specification and early patterning of the germ layers,
in particular the induction of the mesodermal germ layer and its patterning along the
dorso-ventral axis. In the next chapter (Chapter 5) we shall look at patterning of the
vertebrate embryo along the antero-posterior axis and the all-important induction of
the prospective nervous system from ectoderm.
For some vertebrates, the regions of the egg that will give rise to the endoderm and
ectoderm seem already at least partly specified, but the mesodermal germ layer is
not pre-specified in any vertebrate embryo, and its induction as part of the embryo’s
developmental program is one of the key events in early development. In Xenopus,
for example, factors in the animal region of the egg specify that region as prospective
ectoderm, while factors in the vegetal half of the egg specify that region as prospective
endoderm. The mesoderm, in contrast, is produced after development has begun by
signals from the vegetal-region cells acting on adjacent animal-region cells to specify
a band of mesoderm around the equator of the blastula. As the blastula develops,
the mesoderm becomes patterned along the dorso-ventral axis so that it becomes
subdivided into regions that give rise to different structures (Fig. 4.1). The timing
of mesoderm induction differs considerably between the different vertebrate groups,
however, although many of the signaling molecules used are the same.
 Setting up the body axes 129

Schematic longitudinal section of embryo

Xenopus fertilized egg Blastula just before gastrulation after gastrulation and neurulation

Animal Animal somitic mesoderm notochord

epidermis nervous prechordal plate Dorsal neural tube
system
forebrain

Ventral Dorsal Anterior Posterior

blood, notochord gut

kidney endoderm
somites endoderm Ventral ectoderm
Vegetal Vegetal

Fig. 4.1 Comparison of the axes and fate

Setting up the body axes maps of the Xenopus egg and blastula with
the structures in the tailbud embryo. The
We first consider how the antero-posterior and dorso-ventral axes are established in ver- Xenopus egg has an initial axis of symmetry
tebrate embryos and whether they are already present in the egg or are specified later. In running from the animal to the vegetal
pole—the animal–vegetal axis (first panel).
other words, to what extent is the embryo already patterned as a result of maternal fac-
This symmetry is broken by the specification
tors laid down in the egg during its development in the ovary? The maternal contribution
of one side of the blastula as the dorsal
to early development differs considerably among our four model organisms. How the
side (second panel). After gastrulation and
body axes will be established in amphibians and zebrafish is under the control of factors neurulation, the embryo has elongated and
already present in the egg, whereas birds and mammals follow a different developmental now has a distinct anterior (head) to posterior
strategy. Finally, we will briefly consider the intriguing question of how the left–right (tail) axis and a distinct dorso-ventral axis at
asymmetry, or ‘handedness’, of a number of internal organs might be determined. right angles to it (third panel). In Xenopus,
the endoderm and ectoderm are specified by
4.1 The animal–vegetal axis is maternally determined in Xenopus and maternal factors already present in the egg,
but the mesoderm is induced from ectoderm
zebrafish
by signals from the vegetal region during the
The very earliest stages in development of both Xenopus and the zebrafish are exclu- blastula stage.
sively under the control of maternal factors present in the egg. As we saw in Drosophila
(see Chapter 2), maternal factors are the mRNA and protein products of genes that are
expressed by the mother during oogenesis and laid down in the egg while it is being
formed. The Xenopus egg possesses a distinct polarity even before it is fertilized, and
this polarity influences the pattern of cleavage. One end of the egg, the animal pole,
which rotates to be uppermost when the egg is fertilized, has a heavily pigmented sur-
face, whereas most of the yolk is located toward the opposite, unpigmented end, the
vegetal pole (see Fig. 3.4). The animal half contains the egg nucleus, which is located
close to the animal pole. These differences define the animal–vegetal axis. The pig-
ment itself has no role in development but is a useful marker for real developmental
differences that exist between the animal and vegetal halves of the egg. The planes
of early cleavages that occur after fertilization are related to the animal–vegetal axis.
The first cleavage is parallel with the axis and often defines a plane corresponding
to the midline, dividing the embryo into left and right sides. The second cleavage is
in the same plane at right angles to the first and divides the egg into four cells, while
the third cleavage is at right angles to the axis and divides the embryo into animal and
vegetal halves, each composed of four blastomeres (see Fig. 3.5).
Maternal factors are of key importance in the very early development of Xenopus
and zebrafish because most of the embryo’s own genes do not begin to be transcribed
until after a stage called the mid-blastula transition, which will be discussed later in
this chapter. The Xenopus egg contains large amounts of maternal mRNAs and there
are also large amounts of stored proteins; there is, for example, sufficient histone
protein for the assembly of more than 10,000 nuclei, quite enough to see the embryo
through the first 12 cleavages until its own genes begin to be expressed. The egg also
contains a number of maternal mRNAs that encode proteins with specifically devel-
opmental roles. These mRNAs are localized along the animal–vegetal axis of the egg
during its development in the mother (oogenesis), with most of the developmentally
130 Chapter 4 Vertebrate development II: axes and germ layers

Animal pole
important maternal products ending up in the vegetal hemisphere of the mature egg.
If the vegetal-most cytoplasm is removed early in the first cell cycle leading up to the
first cleavage division of the fertilized egg, the embryo never develops beyond a radi-
ally symmetric mass of tissue that lacks any indication of structures characteristic of
nucleus
the antero-posterior axis, such as notochord or somites.
Localization of maternal mRNAs occurs by one of two routes. Early in oogenesis,
some mRNAs are concentrated in the cortex at the future vegetal pole via trans-
port involving a cytoplasmic region known as the message transport organizer region
(METRO), which is associated with the mitochondrial cloud—a region of the oocyte
Vegetal pole cytoplasm where mitochondria are being generated. The mechanism underlying the
selective trapping and transport of mRNAs by the METRO is not yet clear, but it may
involve association of the mRNAs with the cytoskeleton and small vesicles of endo-
Fig. 4.2 Distribution of mRNA for the
plasmic reticulum. Later in oogenesis, other mRNAs are transported to the vegetal
growth factor Vg-1 in the unfertilized
cortex along microtubules by kinesin motor proteins, a similar mechanism to that of
amphibian egg. In situ hybridization with
mRNA transport in the Drosophila oocyte (see Fig. 2.21).
a radioactive probe for maternal Vg-1
mRNA in a Xenopus late oocyte shows its The expression of the developmentally important maternal mRNAs in the Xenopus
localization in the cortex of the vegetal egg is regulated such that in most cases they are only translated after fertilization. The
region, as indicated by the yellow color proteins encoded by several of these mRNAs are candidates for signals that specify
(which represents silver grains viewed by early polarity and set further development in train. One maternal mRNA encodes the
special lighting). In this image, made before signaling protein Vg-1 (secreted signaling proteins used in vertebrate early develop-
the use of fluorescent labels to detect nucleic ment are listed in Box 4A, p. 131). Vg-1 mRNA is synthesized during early oogenesis
acid hybridization was commonplace, the and becomes localized in the vegetal cortex of mature oocytes (Fig. 4.2), moving into
section through the egg was covered with the vegetal cytoplasm before fertilization. It is translated after fertilization and func-
a radiographic emulsion after hybridization
tions as an early signal for some aspects of mesoderm induction. Another vegetally
of the probe, and the presence and density
localized mRNA encodes the protein Xwnt-11, which provides one of the early signals
of the radioactive probe at any location was
required for dorso-ventral axis specification. Xwnt-11 is a member of the Wnt family
determined by the amount of silver grains
deposited. Scale bar = 1 mm. of signaling proteins related to the Wingless protein in Drosophila. Other components
of the Wnt signaling pathway are also encoded by maternal mRNAs (see Box 1E,
Photograph courtesy of D. Melton.
p. 26). The other main class of maternal factors with developmental roles comprises
the transcription factors that switch on and regulate zygotic gene expression. Mater-
nal mRNA encoding the T-box transcription factor VegT is localized to the vegetal
hemisphere and is translated after fertilization. VegT has a key role in specifying both
endoderm and mesoderm and its localization in the vegetal region is crucial to this
function.
The animal–vegetal axis of the egg is in some sense related to the antero-posterior
axis of the tadpole, as the head will form from the animal region, but the axes of
the tadpole and the fertilized egg are not directly comparable (see Fig. 4.1). Which
side of the animal region will form the head is not determined until the dorsal side of
the embryo is specified, which does not happen until after fertilization. The precise
position of the embryo’s antero-posterior axis thus depends on the specification of
the dorso-ventral axis.
In the zebrafish fertilized egg, maternal factors are also distributed along the
­animal–vegetal axis and key factors for axis formation are present in the vegetal
region. As in Xenopus, removal of the vegetal-most yolk early in the first cell cycle of
the fertilized egg results in radially symmetric embryos that lack structures character-
istic of the antero-posterior axis.

4.2 Localized stabilization of the transcriptional regulator β-catenin specifies

the future dorsal side and the location of the main embryonic organizer in
Xenopus and zebrafish

The spherical unfertilized egg of Xenopus is radially symmetric about the animal–
vegetal axis, and this symmetry is broken only when the egg is fertilized. Sperm entry
sets in motion a series of events that defines the dorso-ventral axis of the gastrula
 Setting up the body axes 131

Box 4A Intercellular protein signals in vertebrate development

Proteins that are known to act

Common intercellular signaling proteins in vertebrate embryonic development
as signals between cells during
­development belong to seven main Signal molecule family Receptors Examples of roles in development
families. Some of these families, Fibroblast growth factor (FGF)
such as the fibroblast growth fac- Twenty-five FGF genes have been FGF receptors Roles at all stages in development.
tors (FGFs), were originally identi- identified in vertebrates, not all of which (receptor tyrosine kinases) Maintenance of mesoderm formation
are present in all vertebrates (this chapter). Induction of spinal cord
fied because they were essential (see Chapter 5 and Box 5B for signaling
for the survival and proliferation of pathway). Signal from apical ectodermal
ridge in limb bud (Chapter 11)
mammalian cells in tissue culture.
Epidermal growth factor (EGF)
The members of the seven families
Epidermal growth factor EGF receptors Cell proliferation and differentiation.
are either secreted or are signal- (receptor tyrosine kinases) Limb patterning (Chapter 11)
ing proteins spanning the plasma Transforming growth factor-� (TGF-�)
membrane, and they provide inter- A large protein family that includes activin, Type I and Type II receptor subunits Roles at all stages in development.
cellular signals at many stages of Vg-1, bone morphogenetic proteins (receptor serine/threonine kinases) Signaling pathway in Fig. 4.33.
(BMPs), Nodal, and Nodal-related proteins Receptors act as heterodimers Mesoderm induction and patterning
development. Other ­signal mol- (this chapter)
ecules, such as insulin and the
Hedgehog
insulin-like growth factors, and the
Sonic hedgehog and other members Patched Positional signaling in limb and neural
neurotrophins, are used by embryos tube (Chapters 11 and 12). Signaling
for particular tasks, but the seven pathway in Box 11C. Involved in
determination of left–right asymmetry
families listed in the table are the in chick embryos (this chapter)
main families involved throughout Wingless (Wnt)
vertebrate development in verte- Nineteen Wnt genes have been identified in Frizzled family of receptors Roles at all stages in development. For
brates. Homologs of these proteins vertebrates, not all of which are present in (seven-span transmembrane canonical and non-canonical signaling
all vertebrates proteins) pathways see Box 1E and Box 8C.
are also important in embryonic Dorso-ventral axis specification in
development in ­non-vertebrates Xenopus (this chapter). Limb
development (Chapter 11)
(see Chapter 2, Chapter 6, and
examples throughout the rest of Delta and Serrate
the book). Transmembrane signaling proteins Notch Roles at all stages in development.
Left–right asymmetry (this chapter).
The signaling molecules listed Somite development (see Chapter 5,
in the table all act by binding to signaling pathway in Box 5C)
recep­tors on the surface of a tar- Ephrins
get cell. This produces a signal that Transmembrane signaling proteins Eph receptors Formation of rhombomeres in embryonic
(receptor tyrosine kinases) hindbrain (Chapter 5). Guidance of
is passed, or transduced, across developing blood vessels (Chapter 11).
the cell membrane by the receptor Guidance of axon navigation in the
nervous system (Chapter 12)
to the intracellular biochemical sig-
naling pathways. Each type of sig-
nal protein has a corresponding receptor or set of receptors, and a covalently linked complex that binds to and activates the recep-
only cells with the appropriate receptors on their surface can tor, bringing two receptor proteins together to form a dimer (see
respond. The intracellular pathways can be complex and involve Fig. 4.33). In some cases, the active form of the protein signal is
numerous different proteins. We have already introduced the a heterodimer, made up of two different members of the same
canonical Wnt intracellular signaling pathway in Box 1E (p. 26) family. Secreted signals can diffuse or be transported short dis-
and the Nodal signaling pathway is illustrated in Fig. 4.33 and tances through tissue and set up concentration gradients. Some
Box 4B, p. 143. Figures illustrating the other signaling pathways signal molecules, however, remain bound to the cell surface and
are cross-referenced in the table. In the developmental context, thus can only interact with receptors on a cell that is in direct
the important responses of cells are either a change in gene contact. One such is the protein Delta, which is membrane-bound
expression, with specific genes being switched on or off, or a and interacts with the receptor protein Notch on an adjacent cell
change in the cytoskeleton, leading to changes in cell shape and (see Box 5C, p. 190 for the Notch signaling pathway).
motility. For consistency in this book we have hyphenated the abbre-
Some signaling proteins, such as those of the transforming viations for all secreted signaling proteins where the name
growth factor-β (TGF-β) family, act as dimers; two molecules form allows—for example Vg-1, BMP-4, and TGF-β.
132 Chapter 4 Vertebrate development II: axes and germ layers

Fertilization of egg 30o cortical rotation Cleavage Late blastula

Animal
sperm pigmented
cortex
Spemann
V V V organizer
Cytoplasm blastocoel

D D D
blastopore
cortex dorsalizing heavy yolk Activation of canonical Wnt
factors pathway on future dorsal side
Direction of
Vegetal cortical rotation

Fig. 4.3 The future dorsal side of the

amphibian embryo develops opposite (see Section 3.1 for a general outline of Xenopus development), with the dorsal side
the site of sperm entry as a result of the forming more or less opposite the sperm’s entry point.
redistribution of maternal dorsalizing Sperm entry can occur anywhere in the animal hemisphere of the Xenopus egg
factors by cortical rotation. After and it causes the outer layer of cytoplasm, the cortex, to loosen from the inner dense
fertilization (first panel), the cortical layer cytoplasm so that it is able to move independently. The cortex is a gel-like layer rich in
just under the cell membrane rotates by actin filaments and associated material. During the first cell-cycle, the cortex rotates
30° towards the future dorsal region. This about 30° against the underlying cytoplasm in a direction away from the site of sperm
results in dorsalizing factors, such as XWnt-
entry (Fig. 4.3). As a result of this cortical rotation, an array of sub-cortical micro-
11 mRNA and Dishevelled (Dsh) protein (a
tubules becomes oriented with their plus ends pointing away from the site of sperm
component of the Wnt signaling pathway)—
entry and they act as tracks for the directed movement of proteins and mRNAs.
indicated by green dots—being relocated
from their initial position at the vegetal Cortical rotation and transport along cortical microtubules leads to the relocation
pole to a position approximately opposite to of maternal factors originally located at the vegetal pole, such as XWnt-11 mRNA and
the site of sperm entry (second panel). The Dishevelled protein (a Wnt signaling pathway component), to sites on the equator
factors are transported further away from opposite the sperm entry point, creating asymmetry in the fertilized egg, such that
the vegetal pole along microtubules that lie when the second cleavage division occurs only two of the four blastomeres inherit
at the interface of the cortex and the yolky these maternal factors. These factors specify their new location as the future dorsal
cytoplasm. The Wnt pathway is activated side of the embryo, conferring a second axis of symmetry, the dorso-ventral axis, on
where these factors end up (third panel, red the fertilized egg (see Fig. 4.3). For this reason, they are often called dorsalizing fac-
dots indicate activation of the Wnt signaling
tors. Establishment of the future dorsal region is a key event in Xenopus development,
pathway), which results in an accumulation
as it sets up the conditions necessary for the formation of the main embryonic orga-
of b-catenin in nuclei on the future dorsal
nizer, the Spemann organizer, in this region. As we saw in Chapter 1, the embryonic
side of the blastula (not shown here). In the
blastula (fourth panel), gastrulation starts in organizer was originally discovered as a small region of the embryo, active in the
the dorsal region, adjacent to the site of the early stages of development, which, when grafted to another embryo, could induce
Spemann organizer, which is formed in this the formation of a partial second embryo (see Fig. 1.9).
region. V = ventral; D = dorsal. Any failure of cortical rotation and translocation of the dorsalizing factors blocks all
further anterior and dorsal development, leading to a highly abnormal and non-viable
embryo. Cortical rotation can be prevented by irradiating the ventral side of the egg
with ultraviolet (UV) light, which disrupts the microtubules responsible for transport-
ing the dorsalizing factors. Embryos developing from treated eggs are ventralized;
that is, they are deficient in structures normally formed on the dorsal side and develop
excessive amounts of blood-forming mesoderm, a tissue normally only present at the
embryo’s ventral midline. With increasing doses of radiation, both dorsal and anterior
structures are lost, and the ventralized embryo appears as little more than a small
distorted cylinder. The effects of UV irradiation are, however, ‘rescued’ by tipping
the egg. Because of gravity, the dense yolk slides against the cortex, thus effectively
restoring rotation and the localization of maternal factors.
The maternal dorsalizing factors act by stimulating a signaling pathway known as
the canonical Wnt/b-catenin pathway. This pathway leads to the local stabilization
of the protein b-catenin in the cells of the future dorsal region by preventing its degra-
dation there. The canonical Wnt pathway (see Box 1E, p. 26) is one of several signaling
pathways that can be stimulated by Wnt proteins and we have already seen it in action
 Setting up the body axes 133

in Drosophila in response to Wingless. b-Catenin has a dual role as a gene-regulatory Injection of �-catenin mRNA into a ventral
protein and as a protein that links cell-adhesion molecules to the ­cytoskeleton (see vegetal cell (32-cell Xenopus embryo)
Box 8B, p. 293). In the specification of the dorsal region it is acting as a gene-regulatory
protein. The dorsalizing effects of b-catenin can be seen when its mRNA is injected Animal

into ventral vegetal cells, which then form a second main body axis (Fig. 4.4).
Maternal b-catenin is initially present throughout the vegetal region, but when
bound to a ‘destruction complex’ of proteins that includes the protein kinase GSK-3b,
which phosphorylates it, the phosphorylated b-catenin is targeted for proteasomal Ventral Dorsal
degradation (see Box 1E, p. 26). Activation of the canonical Wnt signaling pathway by
the dorsalizing factors blocks the activity of the destruction complex, thus preventing
the degradation of b-catenin on the prospective dorsal side of the embryo. By the two-
cell stage, b-catenin has started to accumulate in the cytoplasm of these cells, and by
the end of the next two cleavage divisions it has entered the nuclei of the dorsal cells, Vegetal

where it helps switch on zygotic genes required for the next stage in development.
In Xenopus, maternal XWnt-11, translated in the dorsal region from its mRNA is a
likely activator of the canonical Wnt signaling pathway at this stage. Further evidence Twinned embryo develops with a duplicated axis
that a Wnt protein is required for dorsalization comes from the finding that inhibiting
expression of one of the Xenopus Wnt receptors, Xfz7 (Xenopus ­Frizzled 7), using
antisense morpholino oligonucleotides (see Box 6A, p. 220) causes loss of dorsal
structures. XWnt-11 mRNA is present in the ventral region as well, and Wnt signal-
ing might be limited to the dorsal region by maternal ‘antagonists’ or inhibitors of
Wnt signaling present in the ventral region. As we shall see thoughout this chapter,
inhibition of the activity of secreted signaling proteins is an important developmental
strategy for restricting signaling to specific regions. The Wnt pathway might also be
activated directly via pathway components such as Dishevelled, which is moved into
the future dorsal region as a result of cortical rotation (see Fig. 4.3), or by inhibitory
proteins acting on the GSK-3b complex. The key role played by inhibition of GSK-3b
activity in establishing the future dorsal region explains the long-established find-
ing that treatment of Xenopus embryos with lithium chloride at this stage dorsalizes
Fig. 4.4 Induction of a new dorsal side
them, promoting the formation of dorsal and anterior structures at the expense of
by injection of b-catenin mRNA. Injection
ventral and posterior structures. Lithium is known to block GSK-3b activity.
of mRNA encoding b-catenin into ventral
The dorso-ventral axis is specified similarly in zebrafish, although in zebrafish the vegetal cells can specify a new Nieuwkoop
site of sperm entry is not involved in determining the future dorsal side and there is no center (discussed in Section 4.3) at the site
cortical rotation. Specification of the dorsal region in the zebrafish involves activation of injection, leading to a twinned embryo.
of b-catenin by a maternal dorsalizing activity that moves in a microtubule-dependent Injection of some other vegetally localized
manner from the vegetal region towards the animal pole and comes to lie under the maternal mRNAs, such as Vg-1, have similar
cellular blastoderm. The stimulus for this movement is so far unknown. The maternal effects.
dorsalizing factors activate the b-catenin pathway in the dorsal yolk syncytial layer and
dorsal margin blastomeres, and b-catenin accumulates in nuclei there (see Fig. 4.7).

4.3 Signaling centers develop on the dorsal side of Xenopus

and zebrafish blastulas

The combined actions of b-catenin and other maternal factors present in the dorsal
side of the vegetal region lead to the formation of a signaling center in this region.
A signaling center is a region of an embryo that produces signaling molecules that act
as morphogens to pattern the embryo and set cell fates. The first signaling center that
develops in the dorsal–vegetal region in the Xenopus blastula is known as the blastula
organizer or Nieuwkoop center. The center is named after the Dutch ­embryologist
Pieter Nieuwkoop, who discovered the organizing properties of the vegetal region
through experiments in which explants of animal cap tissue were combined with
vegetal explants. Before those experiments, in the late 1960s, it had been thought that
mesoderm, like endoderm and ectoderm, was intrinsically specified by a particular
cytoplasmic region in the amphibian egg.
134 Chapter 4 Vertebrate development II: axes and germ layers

Xenopus embryo at the four-cell stage divided into dorsal and ventral halves The ventral half develops into a ventralized embryo.
The dorsal half develops into a dorsalized embryo

Ventral half lacks Nieuwkoop center

sperm entry
point ventralized embryo
dorsal half

dorsalized embryo

Ven
tral Nieuwkoop
Dor center
sal ventral half

Fig. 4.5 The Nieuwkoop center is essential

for normal development. If a Xenopus The Nieuwkoop center sets the initial dorso-ventral polarity in the blastula, and
embryo is divided into a dorsal and a ventral was originally thought to send a distinct signal required for establishment of the
half at the four-cell stage, the dorsal half Spemann organizer. The Spemann organizer arises just above the Nieuwkoop center
containing the Nieuwkoop center develops as at the late blastula–early gastrula stage and is crucial for further development of both
a dorsalized embryo lacking a gut, while the the antero-posterior and dorso-ventral axes of the embryo and the induction of the
ventral half, which has no Nieuwkoop center, nervous system from the ectoderm. More recent work indicating molecular similari-
is ventralized and lacks both dorsal and
ties between the Nieuwkoop center generated in tissue explants and the Spemann
anterior structures.
organizer as defined in vivo suggests, however, that it might be more appropriate to
consider the Nieuwkoop center as a version of the Spemann organizer that manifests
itself in explant experiments, or as an early stage in Spemann organizer formation.
Whatever the true nature of the Nieuwkoop center, the importance of the dorsal
region as an organizer is shown by experiments that divide the embryo into two at the
four-cell stage in such a way that one half contains the future dorsal region while the
other does not (Fig. 4.5). The dorsal half will develop most structures of the embryo,
although it will lack a gut, which develops from ventral blastomeres. The ventral half
will develop much more abnormally, with little relevance to the cells’ normal fates,
producing a distorted, radially symmetric ventralized embryo that lacks all dorsal and
anterior structures; it has developed in the same way as an embryo in which cortical
rotation has been completely blocked by UV irradiation. This experiment shows that
dorsal and ventral regions are established by the four-cell stage.
The influence of the Nieuwkoop center can also be seen in experiments in which
cells from the dorsal–vegetal region of a 32-cell Xenopus embryo are grafted into the
ventral side of another embryo. This gives rise to a twinned embryo with two dorsal
sides (Fig. 4.6). Cells from the graft itself contribute to endodermal tissues but not to
the mesodermal and neural tissues of the new axis. This shows that the grafted cells
are able to induce these fates in the cells of the host embryo. In contrast, ­grafting
ventral cells to the dorsal side has no effect. As we saw in Chapter 1, grafts of the

Transplantation of Nieuwkoop center to ventral side Twinned embryo develops

with a duplicated axis

Animal Animal

Fig. 4.6 The Nieuwkoop center can

Ventral Ventral
specify a new dorsal side. Grafting vegetal
cells containing the Nieuwkoop center from
the dorsal side of one 32-cell Xenopus Dorsal Dorsal Dorsal Dorsal
blastula to the ventral side of another results
in the formation of a second axis and the
host
development of a twinned embryo. The Vegetal Nieuwkoop Vegetal Nieuwkoop
center center
grafted cells signal but do not contribute to Donor (32-cell embryo) Recipient (32-cell embryo)
the second axis.
 Setting up the body axes 135

Spemann organizer also produce twinned embryos but in this case the grafted tissue
forms the notochord of the secondary axis.
We can now understand the result of Roux’s classic experiment (see Fig. 1.8) in
which he destroyed one cell of a frog embryo at the two-cell stage and got a half-
embryo rather than a half-sized whole embryo. The crucial feature of his experiment,
it turns out, was that the killed cell remained attached, but the embryo did not ‘know’
it was dead. The remaining living cell could still develop a functional dorsal organizer
but it developed as if the other half of the embryo was still there. If the killed cell
had been separated from the living blastomere, the embryo could have regulated and
developed into a small-sized whole embryo.
Although b-catenin becomes localized in nuclei in both the animal and vegetal
­regions and specifies a large region as dorsal, the future organizing centers—the
­Nieuwkoop center and the Spemann organizer—arise in the vegetal/equatorial region.
These centers develop in this region because organizing function depends not only on
b-catenin but also on the maternal transcription factor VegT, which is present through-
out the vegetal region from fertilization onwards. By the 16- to 32-cell stage, ­b-catenin
protein can be detected in nuclei on the dorsal side (Fig. 4.7), where in ­vegetal cells
expressing VegT it activates zygotic genes such as siamois, which is involved in the
induction of Spemann organizer function. VegT is also required to induce expression
of the genes encoding Nodal-related proteins, signaling proteins that induce ­mesoderm

Xenopus Zebrafish

Animal Animal

Ventral Dorsal Ventral Dorsal

Vegetal Vegetal

blastoderm Fig. 4.7 A comparison of the

Spemann establishment of the dorsal organizer
organizer in Xenopus and zebrafish embryos. Left
panels: in Xenopus, cortical rotation moves
dorsal determinants (green) towards the
future dorsal side of the embryo, creating a
large region where, starting at the 32-cell
Nieuwkoop stage, b-catenin moves from the cytoplasm
center
into nuclei (red). Cells in the region in which
b-catenin is intranuclear and where VegT
future
shield is also present express the transcription
factor Siamois, which defines the Nieuwkoop
center in the dorsal vegetal region (blue
shading). Right panels: in zebrafish, dorsal
determinants are transported towards
the future dorsal side of the embryo and
blastula yolk enter the blastoderm. At the mid-blastula
organizer syncytial transition, b-catenin is translocated into
layer dorsal yolk syncytial layer nuclei where it
induces the expression of the transcription
factor Dharma in the future shield region, the
dorsal determinants nuclear �-catenin Siamois/Dharma
future organizing region (blue shading).
136 Chapter 4 Vertebrate development II: axes and germ layers

(see Box 4A, p. 131). Named after the mouse protein Nodal, members of this protein
family are key factors in mesoderm induction in all vertebrates, and in the frog they are
known as the Xenopus Nodal-related factors (Xnrs). They are members of the TGF-b
family of signaling proteins and we will discuss them further when we look at the
induction and patterning of the mesoderm in the last part of this chapter.
In the zebrafish embryo, b-catenin becomes localized in nuclei of the dorsal yolk
syncytial layer and the marginal blastomeres overlying it, which will become the
shield region (see Section 3.2). The shield contains an organizing center analogous
to the Spemann organizer. b-catenin induces the expression of a gene encoding a
transcription factor, called Dharma (also known as Bozozok), which is essential for
inducing the organizing function of the shield and the development of head and trunk
regions (see Fig. 4.7).
Rotation in oviduct
4.4 The antero-posterior and dorso-ventral axes of the chick blastoderm are
related to the primitive streak

The chick embryo develops from a circular blastoderm sitting on the top of the yolk
(see Section 3.3). Like the amphibian blastula, the chick blastoderm is initially radi-
X Y ally symmetric. This symmetry is broken when the posterior end of the embryo is
specified. The future posterior end becomes evident soon after the egg is laid by the
appearance of a ridge of cells known as Koller’s sickle at one side of the blastoderm.
As we saw in Section 3.3, this sickle lies adjacent to the posterior marginal zone,
and marks the site at which the primitive streak will develop at gastrulation. Unlike
Xenopus and the zebrafish, in the chick the embryo’s own genes are expressed from
the start of cleavage and it is unlikely that maternal determinants are involved in early
X–Y cross-section during rotation
blastoderm
development to any great extent.
The primitive streak indicates the direction of the antero-posterior axis of the
P
embryo: the start of the streak marks the posterior end of the embryo and the streak
A
X Y elongates in an anterior direction. The location of the posterior marginal zone is there-
Yolk
fore crucial to setting the body axes, and it turns out that this is determined by gravity.
During the passage of the fertilized egg through the hen’s uterus, which takes about
Albumen 20 hours, the egg moves pointed end first and rotates slowly around its long axis, each
revolution taking about 6 minutes. Cleavage has already started at this stage, and the
blastoderm contains many thousands of cells by the time the egg is laid. As the egg
Development after laying
rotates, the blastoderm remains tilted at an angle of about 45° to the vertical. The
Surface view of yolk downward-pointing edge indicates the future head end of the embryo and the poste-
rior marginal zone develops at the edge of the blastoderm that is uppermost (Fig. 4.8).
primitive streak It is still completely unknown how this effect of gravity breaks the blastoderm’s initial
radial symmetry. Whatever the answer, the axis is not irreversibly determined by
these events. The chick blastoderm is highly regulative. Even at the stage when the
X P A Y
blastoderm contains around 20,000 cells, it can be cut into many fragments, each of
which will develop a complete embryonic axis.
The posterior marginal zone can be thought of as an organizing center analogous
in some ways with the Nieuwkoop center in Xenopus, since it can induce the for-
mation of a new body axis. If the posterior marginal zone from one blastoderm is
grafted to another position on a second blastoderm of the same age it can induce a
new primitive streak at this position (Fig. 4.9). The primitive streak is initiated by
Fig. 4.8 Gravity defines the antero-posterior
the concerted actions of various signaling proteins, which we shall return to later in
axis of the chick. Rotation of the egg in the
the chapter. One of the earliest signals involved in streak initiation in the chick is the
oviduct of the mother results in the blastoderm
being tilted in the direction of rotation, though TGF-b family member Vg-1, which is expressed in the posterior marginal zone. When
it tends to remain uppermost. The posterior cells expressing Vg-1 are grafted to another part of the marginal zone in another blas-
marginal zone (P) develops at that side of the toderm ­(essentially the same experiment as that shown in Fig. 4.9), they can induce a
blastoderm that was uppermost, and initiates complete new streak. We have already encountered this signaling protein in Xenopus,
the primitive streak. A = anterior. where it is similarly able to induce a new body axis (see Fig. 4.4).
 Setting up the body axes 137

Transplantation of posterior marginal zone cells from one chick blastoderm

to another can result in an additional primitive streak
area additional induced
opaca Anterior
primitive streak Fig. 4.9 The posterior marginal zone of
the chick blastodisc induces the primitive
streak. The posterior marginal zone marks
the posterior end of the antero-posterior
area axis. Grafting posterior marginal zone cells
pellucida expressing the signaling protein Vg-1 to a
lateral site in the marginal zone of another
posterior blastoderm can result in the formation of an
marginal zone extra primitive streak at that position. Two
streaks often do not develop, however, as
normal host
Posterior
primitive streak whichever forms first tends to inhibit the
development of the other.

In the chick, however, often only one axis develops in the grafted embryos—either
the host’s normal axis or the one induced by the graft. This suggests that the more
advanced of the two organizing centers inhibits streak formation elsewhere. It has
been shown that Vg-1 induces an inhibitory signal that can travel across the embryo,
a distance of around 3 mm in about 6 hours, to prevent another streak forming. Vg-1
is thus required for initiating streak formation but at the same time is involved in
inhibiting the formation of a second streak.
Once the primitive streak has formed and starts to elongate, the antero-posterior
axis is defined. After the node starts to regress, the head can be seen forming at the
anterior end (Fig. 4.10). The streak also defines the future dorso-ventral axis of the
embryo, with the most ventral structures (such as extra-embryonic mesoderm and
lateral plate) being derived from the more posterior end of the streak and dorsal struc-
tures such as the notochord from the node (see Fig. 4.21). The relation of the streak
to the dorso-ventral axis is the result of patterning of mesoderm cells while they are
in the primitive streak, and will be explained in more detail later in the chapter.
Primitive-streak formation only starts when the hypoblast becomes displaced away
from the posterior marginal region by growth of a layer of cells known as the endo-
blast (see Fig. 3.15). The hypoblast actively inhibits streak formation, as shown by
experiments in which its complete removal leads to the formation of multiple streaks
in random positions. This inhibition is due to a protein produced by the hypoblast

Stage 5 chick embryo (head process stage) Stage 7 chick embryo (head fold stage)

Anterior head fold

Hensen’s node head process Fig. 4.10 The direction of the primitive
neural streak marks the antero-posterior axis
fold of the chick embryo. After extending to
somite its full length, the primitive streak begins
to regress, with Hensen’s node moving in
a posterior direction (first panel). As the
node moves backward, the notochord (head
process) develops anterior to it, from cells of
the node that are left behind as it regresses.
The head fold and neural plate form in the
regressing regressing
primitive streak primitive streak epiblast anterior to the furthest extent of the
primitive streak (second panel). As the node
Posterior regresses, somites begin to form on either
side of the notochord.
138 Chapter 4 Vertebrate development II: axes and germ layers

which antagonizes the function of one of the signaling proteins required for streak
Signal produced by hypoblast inhibits formation
of the primitive streak formation, and which will be discussed in more detail later in the chapter. When
hypoblast has been replaced by endoblast, this inhibitory signal ceases and primitive
Koller’s sickle
streak formation can begin (Fig. 4.11).

4.5 The definitive antero-posterior and dorso-ventral axes of the mouse

hypoblast prospective endoblast embryo are not recognizable early in development

The mammalian egg differs considerably from the eggs of either Xenopus or the chick,
When the hypoblast is replaced by endoblast, as it contains no yolk and the early stages in development are concerned with generat-
primitive streak formation can start
ing extra-embryonic yolk sac membranes and the placenta connecting the embryo to
start of primitive streak the mother, so that the embryo can be nourished. There is no clear sign of polarity in
the mouse egg, and any contribution of maternal factors is difficult to determine as
zygotic gene activation occurs in the one-cell embryo and is essential for development
beyond the two-cell stage. Several claims have been made that the first cleavage of the
zygote is related to an axis of the embryo and that the first two cells (blastomeres) are
endoblast biased towards different fates. This is an interesting possibility but is not yet clear, as
different workers get different results. Furthermore, the highly regulative nature of early
Fig. 4.11 The primitive streak begins to development in the mouse argues against the importance of maternal determinants. It
form once the hypoblast has been replaced
has been known for a long time that both blastomeres of the two-cell mouse embryo
by endoblast. Signaling in the posterior
can give rise to normal embryos if separated. In addition, individual blastomeres from
marginal zone that enables posterior streak
formation is initially inhibited by a protein
mouse embryos up to the eight-cell stage can contribute extensively to both embry-
secreted by the hypoblast, a layer of cells onic and placental tissues, and normal development can still occur even after cells are
that initially lies below the blastodisc over removed or added to a preimplantation embryo—the blastocyst before it implants into
the underlying yolk (see Fig. 3.15). When the the wall of the uterus. This extensive regulative ability would not, however, preclude
hypoblast has been displaced by endoblast, the possibility that, in normal development, individual blastomeres in early embryos
which grows in from the posterior marginal have different developmental properties and fates and maternal determinants might be
zone, this protein is no longer made and the distributed differently between the blastomeres—the debate over this continues.
primitive streak can begin to form. The early cleavages in mouse embryos do not follow a well-ordered pattern. Some
are parallel to the egg’s surface, so that the morula has distinct outer and inner cell
populations. By the 32-cell stage, the morula has developed into a blastocyst, a hol-
Fig. 4.12 The specification of the low sphere of epithelium containing the inner cell mass—some 10–15 cells—attached
blastomeres that will form the inner cell to the epithelium at one end (Fig. 4.12, left panels). The epithelium of the blastocyst
mass of a mouse embryo depends on
whether they are on the outside or the
Normal development of the mouse embryo Experimental manipulation shows that
inside of the embryo. Left panels: following leads to the formation of the blastocyst cell fate depends on relative position
cleavage the mouse embryo is a solid ball
of cells (a morula). This develops into the eight-cell embryo
hollow blastocyst, which consists of an labeled
cells
inner cell mass surrounded by a layer of
trophectoderm. Right panels: to determine
whether the position of a blastomere in the
morula determines whether it will develop
into trophectoderm or inner cell mass,
labeled blastomeres (blue) from a four-cell
mouse embryo are separated and combined
with unlabeled blastomeres (gray) from
inner cell mass Fate of labeled cells’ descendants
another embryo. By following the fate of
the descendants of the labeled cells, it can 97%
be seen that blastomeres on the outside
of the aggregate more often give rise to 60%
trophectoderm, 97% of the labeled cells 40%
ending up in that layer. The reverse is true
3%
for the origin of the inner cell mass, which
derives predominantly from blastomeres on trophectoderm Troph- Inner Troph- Inner
ectoderm cell mass ectoderm cell mass
the inside.
 Setting up the body axes 139

will form the trophectoderm, which gives rise to extra-embryonic structures associ-
ated with implantation and formation of the placenta, a structure unique to mam-
malian development. The cells of the inner cell mass are pluripotent, meaning that
they can give rise to all the cell types in the embryo (see Section 1.17, p. 29), and they
remain pluripotent up to stage E4.5.
Inner cell mass cells can be isolated and grown in culture to produce pluripotent
embryonic stem cells (ES cells). As described in Chapter 10, these cells have been
shown to give rise to all tissues of the embryo, including the germ cells, if introduced
into a blastocyst, and this property is exploited to produce mice carrying particular
mutations or particular transgenes. ES cells that have been mutated or genetically
engineered in culture are introduced into a blastocyst, which is then placed in a sur-
rogate mother (see Fig. 3.34). If the introduced ES cells contribute to the germ cells
of the embryo, a line of mice can be bred that carry the mutation in all their cells.
Human ES cells are also being investigated for their medical therapeutic potential
(discussed in Chapter 10).
The specification of cells as either inner cell mass or trophectoderm in early mouse
embryos depends on their relative positions in the cleaving embryo. Determination of
their fate occurs after the 32-cell stage, and during earlier stages all the cells seem to Embryonic
be equivalent in their ability to give rise to either tissue. The most direct evidence for polar
trophectoderm
the effect of position comes from taking individual blastomeres from disaggregated epiblast inner
primitive cell
­eight-cell embryos, labeling them, and then combining the labeled blastomeres in endoderm mass
different positions with respect to unlabeled blastomeres from another embryo. If the blastocoel
labeled cells are placed on the outside of a group of unlabeled cells they usually give mural
trophectoderm
rise to trophectoderm; if they are placed inside, so that they are surrounded by unla- Abembryonic
beled cells, they more often give rise to inner cell mass (see Fig. 4.12, right panels). If
a whole embryo is surrounded by blastomeres from another embryo, it can all become
part of the inner cell mass of a giant embryo. Aggregates composed entirely of either
‘outside’ or ‘inside’ cells of early embryos can also develop into normal blastocysts, Proximal
showing that there is no specification of these cells, other than by their position, at
trophectoderm
this stage.
extra-
The segregation of trophectoderm and the pluripotent inner cell mass lineages is embryonic
ectoderm
the first differentiation event in mammalian embryonic development, and is essential visceral
endoderm
for the formation of the placenta. The transcription factor Cdx2 is involved in the
Anterior Posterior
differentiation of trophectoderm, while the transcription factor Oct3/4 is involved in
keeping the inner cell mass pluripotent. Oct3/4 is expressed from the two-cell stage epiblast

and all blastomeres at the eight-cell stage co-express both transcription factors. As blastocoel
the morula forms, Cdx2 expression levels become slightly higher in the outer cells,
Distal
possibly as a result of asymmetric cell divisions. Then, because of reciprocal nega-
tive feedback between Cdx2 and Oct3/4, Cdx expression levels increase and Oct3/4
expression levels decrease in the outer cells, whereas Oct3/4 levels of expression Fig. 4.13 The axes of the early mouse
increase and Cdx expression levels decrease in the inner cells. By the blastocyst stage, embryo. At the blastocyst stage (top), at
Cdx2 expression is confined to the trophectoderm and Oct3/4 to the cells of the inner about E4, the inner cell mass is confined to
cell mass. In the absence of Cdx2, trophectoderm initially forms in the blastocyst but the embryonic region and this defines an
does not differentiate and is not maintained. An additional transcription factor, Tead4, embryonic–abembryonic axis (which relates
is required for formation of the trophectoderm. geometrically, although not in terms of cell
Before the blastocyst stage the embryo is a symmetric spheroidal ball of cells. fate, to the dorso-ventral axis of the future
epiblast). The inner cell mass is oval, and thus
The first asymmetry appears in the blastocyst, where the blastocoel develops asym-
also has an axis of bilateral symmetry. At
metrically, leaving the inner cell mass attached to one part of the trophectoderm
about E6 days (bottom) the antero-posterior
(Fig. 4.13). The blastocyst now has one distinct axis running from the site where
axis in the epiblast becomes visible, with
the inner cell mass is attached—the embryonic pole—to the opposite, abembry- the formation of the primitive streak at the
onic, pole, with the blastocoel occupying most of the abembryonic half. This axis posterior end. The interior of the epiblast cup
is known as the ­embryonic–abembryonic axis. It corresponds in a geometric corresponds to the dorsal side of the future
sense to the dorso-ventral axis of the epiblast, but is not related to specification embryo, and the outer side to the future
of cell fate. ventral side.
140 Chapter 4 Vertebrate development II: axes and germ layers

By about 4.5 days after fertilization, the inner cell mass has become differentiated
into two tissues: primary or primitive endoderm (which will form extra-­embryonic
structures) forms its outer, or blastocoelic, surface, and this surrounds the epiblast,
from which the embryo and some extra-embryonic structures will develop. The
­blastocyst implants in the uterine wall, and the trophectoderm at the embryonic
pole proliferates to form the ectoplacental cone, producing extra-embryonic ecto-
derm that pushes the inner cell mass across the blastocoel. A cavity—the proam-
niotic cavity—is then formed within the epiblast as its cells proliferate. The mouse
epiblast is now a monolayer and is formed into a cup shape, U-shaped in section,
with a layer of visceral endoderm on the outside. At this stage, about E5.0–E5.25,
the egg cylinder has a proximo-distal polarity in relation to the site of implantation
with the ectoplacental cone at the proximal end, and the cup-shaped epiblast at the
distal end (see Fig. 4.13). Extra-embryonic ectoderm forms a distinct layer of cells
between the ectoplacental cone and the epiblast. Proximo-distal polarity within the
epiblast is generated by Nodal signaling. A small region of visceral endoderm over
the distal-most end of the epiblast cup is induced as distal visceral endoderm (DVE)
by Nodal signaling from the epiblast. The DVE then starts to express genes encoding
the homeodomain transcription factors Hex and Lim-1 (Lhx1), which are markers of
anterior tissues, and also expresses genes encoding extracellular antagonists of Wnt
and Nodal signaling.

4.6 Movement of the distal visceral endoderm indicates the definitive antero-
posterior axis in the mouse embryo

The true antero-posterior axis of the mouse embryo is thought to be formed at E5.5–
E6.0, about 12–24 hours before gastrulation begins, by a symmetry-breaking event
in which the DVE moves rapidly to one side of the epiblast cup, displacing the more
proximal visceral endoderm. This movement specifies the prospective anterior region
of the embryo, and the endodermal cells derived from the DVE are now called the
­anterior visceral endoderm (AVE). The AVE specifies this side of the epiblast as
­anterior, in part by preventing the formation of the primitive streak here (Fig. 4.14).
Active cell migration is the probable underlying mechanism for this asymmetric move-
ment, although increased cell proliferation may also be involved. The AVE continues
to express Hex and Lim1, together with a range of Wnt and Nodal antagonists. As a
result, the epiblast lying next to it is specified as anterior ectoderm, from which the
Fig. 4.14 The symmetry-breaking brain and other anterior structures will eventually develop.
event in the early mouse embryo is
Despite the differences in topology, this process is developmentally similar to that
the specification of the distal visceral
described earlier for the chick (see Section 4.4): the DVE/AVE is equivalent to the chick
endoderm. At around E5.5, before primitive
streak formation, visceral endoderm (yellow)
at the distal-most end of the cup is specified
as distal visceral endoderm (DVE; green) by Specification of distal visceral Proliferation and movement Specification of anterior ectoderm and
signals from the epiblast (red arrows) and endoderm (DVE) (5.5 days) of anterior visceral endoderm beginning of primitive streak (6.5 days)
(AVE) (6 days)
begins to proliferate and extend to one side
of the epiblast, where it becomes the anterior Proximal
visceral endoderm (AVE). The anterior fate is
extra-
restricted to the distal region by inhibitory embryonic
ectoderm
signals from the extra-embryonic ectoderm
to the proximal visceral endoderm (barred
blue lines). The extended AVE signals to
the adjacent epiblast (green arrows) and epiblast primitive
streak
induces anterior ectoderm (pale green), visceral
which includes neural ectoderm. At E6.5 the endoderm
primitive streak (brown) has begun to form anterior
DVE AVE ectoderm
on the opposite side of the epiblast, thus
Distal
marking the posterior end of the axis.
 Setting up the body axes 141

hypoblast, which also prevents primitive streak formation. In both cases, this influence
is due to these cells producing proteins that counteract Nodal signaling: Cerberus in
chick; Lefty-1 and Cerberus-like 1 in mouse. The movement of the DVE to the ante-
rior side thus leads directly to the initiation of the primitive streak at around E6.25,
and its positioning on the opposite side of the cup, which becomes the posterior end
of the embryo. When a mouse node, which develops as part of the primitive streak,
is transplanted to another part of the epiblast, a secondary axis is induced but this
almost invariably lacks the most anterior regions, including the forebrain. This has
been ­interpreted as evidence that determination of the anterior part of the embryo, by
the AVE movement described earlier, is developmentally independent of determination
of the rest of the body.
The extra-embryonic ectoderm in the E5.5 embryo seems to play a role in ini-
tially restricting the DVE to the most distal region of the visceral endoderm. Nodal
is present throughout the epiblast and could, in principle, induce DVE and subse-
quently AVE fate throughout the visceral endoderm. If the extra-embryonic ectoderm
is removed, the region of the visceral endoderm expressing anterior markers expands.
Signaling by bone morphogenetic proteins (BMPs) (see Box 4A, p. 131) in proximal
extra-embryonic ectoderm may contribute to restricting AVE fate and preventing the
genes specifying anterior fate being expressed in proximal regions of the visceral
endoderm. We shall return to this later in the chapter, when we discuss the formation
of the primitive streak in more detail.

4.7 The bilateral symmetry of the early embryo is broken to produce left–right
asymmetry of internal organs

Vertebrates are bilaterally symmetric about the midline of the body for many struc-
tures, such as eyes, ears, and limbs. But, while the vertebrate body is outwardly sym-
metric, most internal organs are in fact asymmetric with respect to the left and right
sides. This is known as left–right asymmetry. In mice and humans, for example, the
heart is on the left side, the right lung has more lobes than the left, the stomach and
spleen lie towards the left, and the bulk of the liver towards the right. This handed-
ness of organs is remarkably consistent, but there are rare individuals, about one
in 10,000 in humans, who have the condition known as situs inversus, a complete
mirror-image reversal of handedness. Such people are generally asymptomatic, even
though all their organs are reversed. A similar condition can be produced in mice that
carry the iv mutation, in which organ asymmetry is randomized (Fig. 4.15). In this
context, ‘randomized’ means that some iv mutant mice will have the normal organ
asymmetry and some the reverse.

Fig. 4.15 Left–right asymmetry of the

mouse heart is under genetic control. Each
photograph shows a mouse heart viewed
frontally after the loops have formed. The
normal asymmetry of the heart results in
it looping to the right, as indicated by the
arrow (left panel). 50% of mice that are
homozygous for the mutation in the iv gene
have hearts that loop to the left (right panel).
Scale bar = 0.1 mm.
Photographs courtesy of N. Brown.
142 Chapter 4 Vertebrate development II: axes and germ layers

Specification of left and right is fundamentally different from specifying the other
axes of the embryo, as left and right have meaning only after the antero-posterior and
dorso-ventral axes have been established. If one of these axes were reversed, then so
too would be the left–right axis (this is the reason that handedness is reversed when
you look in a mirror: your dorso-ventral axis is reversed, and so left becomes right and
vice versa). One suggestion to explain the development of left–right asymmetry is that
an initial asymmetry at the molecular level is converted to an asymmetry at the cel-
lular and multicellular level. If that were so, the asymmetric molecules or ­molecular
structure would need to be oriented with respect to both the antero-posterior and
dorso-ventral axes.
Although the mechanisms by which left–right symmetry is initially broken are
still not known, the subsequent cascade of events that leads to organ asymmetry is
better understood. The stimulation of a ‘leftward’ flow of extracellular fluid across
the embryonic midline by transient populations of ciliated cells has been shown to
be critical in mouse, zebrafish and Xenopus embryos in inducing asymmetric expres-
sion of genes involved in establishing left versus right. (see Fig. 4.17). In the mouse,
ciliated cells are found on the ventral surface of the node and create fluid flow in a
pit formed below the node where there is no underlying endoderm. In the zebrafish,
the ciliated epithelium lines a fluid-filled vesicle known as Kupffer’s vesicle, which
develops at the midline in the zebrafish embryo’s tailbud at the end of gastrula-
tion, while in Xenopus, a triangular region of ciliated mesoderm, the amphibian
gastrocoel roof plate, straddles the midline in the roof of the archenteron. In chick
embryos, however, motile cilia do not appear to be present on the ventral surface
of the node and instead passive leftward movement of cells appears to be involved
in establishing asymmetrical gene expression. Disrupting the action of the cilia in
either the Kupffer’s vesicle of zebrafish or the gastrocoel roof plate of Xenopus,
for example by using antisense morpholinos (see Box 6A, p. 220) to knock down
expression of ciliary dynein genes, leads to a breakdown in right–left asymmetry.
In mouse embryos, mutations that affect the motor protein dynein and block the
movement of cilia, as in iv mouse mutant and Kartagener’s syndrome in humans,
result in randomized asymmetry (see Fig. 4.15). Reversal of the leftward flow in the
ventral node by an imposed fluid movement also reverses left–right asymmetry in
mouse embryos.
One of the key proteins in establishing ‘leftness,’ and which comes to be expressed
at high levels on the left side of the embryo, is Nodal. Nodal is initially expressed
symmetrically on both sides of the node in both mouse and chick embryos, and adja-
cent to Kupffer’s vesicle in zebrafish, and the gastrocoel roof plate in Xenopus. Initial
small differences in Nodal expression on right and left sides induced by leftward
flow of fluid flow across the midline in mouse, zebrafish and Xenopus then become
amplified as a result of a positive feedback loop—as one effect of Nodal signaling
is to activate the Nodal gene, and so produce more Nodal protein (Box 4B, p. 143).
Asymmetric Nodal expression in the left side of the region is followed by widespread
and stable activation of Nodal expression in the left lateral-plate mesoderm together
with expression of the transcription factor Pitx2, another key determinant of leftness.
This left-sided pattern of gene expression in the lateral plate mesoderm that gives rise
to internal organs such as the heart is highly conserved, being found in the mouse,
zebrafish, Xenopus, and chick. It has been shown in chick embryos that if the expres-
sion of either Nodal or Pitx2 in the lateral plate mesoderm is made symmetric, then
organ asymmetry is randomized.
An early indication of left–right asymmetry in both Xenopus and chick embryos is
the asymmetric activity of a proton–potassium pump (H+/K+-ATPase). This ­ATPase can
be detected as early as the third cleavage division—the eight-cell stage—in ­Xenopus,
and treatment of early embryos with drugs that inhibit the pump cause ­randomization
 Setting up the body axes 143

Box 4B Fine-tuning Nodal signaling

Mouse Nodal and the corresponding Nodal-related proteins

in other vertebrates are secreted signaling proteins that are Cerberus-1
essential for the induction and patterning of the mesoderm and
1
the establishment of left–right asymmetry in early vertebrate
embryos. Signaling molecules that have such powerful effects Nodal pro-protein
convertase 2
on development must be tightly regulated to ensure that the
right amounts are present in the right place at the right time, and
prodomain Lefty1/2
the strength of Nodal signaling can be modulated and regulated
at several different levels as indicated on the figure. Nodal
One is the production of a functional protein (1). Like many
other growth factors, Nodal is synthesized and secreted as an
co-receptor
inactive precursor protein that has to be converted to the active
form by proteolytic cleavage; this is carried out by enzymes
called proprotein convertases. If these are not present, then cell membrane P P

even if the nodal gene is expressed, no functional protein will

be made.
A second level of control is due to extracellular inhibitors that
P
bind to Nodal and/or its receptors (2) and prevent the signal from
reaching the target cell, as discussed in this chapter. Like other Smad2,3 Smad2,3
TGF-b family members, Nodal binds to heterodimeric receptors
known as type I/II receptors, and initiates an intracellular signa­ Smad4
ling cascade in which proteins called Smads are phosphorylated
P
and enter the nucleus to regulate the expression of target genes nuclear
(see Fig. 4.33). Extracellular antagonists of Nodal signaling membrane
include the proteins Cerberus and Lefty.
A third level of control of Nodal signaling is through regulation 4
Nucleus
of its gene (3). The expression of lefty, Cerberus and the nodal
gene itself can all be increased in response to Nodal signaling,
and this can lead to either negative or positive feedback on the
amount of Nodal present. A negative-feedback loop, in which
Nodal mRNA
Nodal increases the expression of lefty, fine tunes the level of
P
Nodal signaling in the proximo-distal patterning of the mouse
epiblast. Here, Nodal induces the distal visceral endoderm (DVE),
which then secretes Cerberus-like 1 and Lefty-1, which in turn Target genes:
Nodal (positive feedback)
downregulate Nodal in an adjacent area of the epiblast, thus Gsc
3 Pitx
specifying this region as the anterior end of the antero-posterior
Lefty1/2 (negative feedback)
axis. A positive feedback loop, in which Nodal signaling upregu- Cerberus-1 (negative feedback)
lates the expression of Nodal, operates in the establishment of
left–right asymmetry and leads to high levels of Nodal on the left
side of the node in mouse embryos (see Section 4.7). Yet another ­ xpression or activity at some level (see summary table, p. 167),
e
level of fine tuning of Nodal signaling discovered more recently and also make positive and negative feedback loops with these
involves the specific inhibition of translation of nodal and lefty antagonists in a similar way to Nodal. In Xenopus, for example,
mRNAs by microRNAs (4 on the figure above) (see Box 6B, p. 228 the dorsalizing signals produced by Spemann’s organizer are
for how microRNAs act). proteins, such as Noggin and Chordin, that block the action of
Surprisingly many of the molecules that play key roles in shap- the ventralizing signal, BMP-4 (see Section 4.17). In proximo-
ing the vertebrate body plan act as extracellular antagonists, distal patterning of the epiblast in mouse embryos, the proxi-
or inhibitors, of intercellular signaling molecules and block or malizing signals Nodal and Wnt are blocked by the distally pro-
modulate signaling. Other widely used developmental signals duced ­extracellular antagonists Cerberus-like 1 and Dickkopf-1,
such as Wnts and BMPs have antagonists that counteract their ­respectively.
144 Chapter 4 Vertebrate development II: axes and germ layers

Fig. 4.16 Determination of left–right

asymmetry in the chick. The activity of a
proton–potassium pump (H+/K+-ATPase) in
node
Hensen’s node is reduced on the left side of
the node, leading to a membrane-potential streak
difference across the node and an increased
release of calcium (Ca2+) into the extracellular notochord Lefty
space on the left side of the node (left 2+
Ca
panel). This leads to a greater activation of H+/K+-ATPase
Notch signaling on the left side of the node,
which in turn activates expression of the Notch
H+/K+-ATPase
gene nodal in cells on the left side of the Ca2+ electrochemical activin
node. Nodal signaling together with Sonic potential
difference Nodal Shh Shh
hedgehog (Shh) then switches on nodal
expression in the lateral plate mesoderm on
the left side, which leads to expression of Pitx2
the transcription factor Pitx2, an important
determinant of leftness (right panel).
Shh activity is inhibited by activin on the Left Right
right-hand side. The Nodal antagonist Lefty
expressed in the left half of the notochord inhibition activation
and the floorplate of the neural tube provides
a midline barrier that prevents Nodal crossing
to the right-hand side. This is a simplified of asymmetry. How the asymmetry of the H+/K+-ATPase itself ­originates is not known,
version of a much more complex set of but one possibility is that the mRNA encoding the ion transporter and/or the ion
interactions. transporter protein itself might be moved by motor ­proteins along oriented cytoskel-
etal elements in the fertilized egg. In Xenopus, disruption of cortical actin at the first
cleavage division can affect left–right asymmetry.
In the chick, a detailed mechanism for left–right asymmetry based on the asym-
metric activity of an H+/K+-ATPase in Hensen’s node has been proposed (Fig. 4.16).
It has been found that the activity of the pump is reduced on the left side of the node,
leading to membrane depolarization and release of calcium from cells to the extracel-
lular space on the left-hand side. The release of calcium in turn leads to the expres-
sion of the Notch ligands, Delta-like and Serrate, in a spatial pattern that could
activate the transmembrane signaling protein Notch on the left side of the node
only (the Notch signaling pathway is illustrated in Box 5C, p. 190). Notch activity,
in conjunction with signaling by the secreted protein Sonic hedgehog (Shh) (see
Box 11C, pp. 424–425, for the Shh signaling pathway), which is expressed at higher
levels on the left side of the node, then leads to the left-sided expression of Nodal
in the lateral-plate mesoderm. Left-sided expression of Shh may be due to passive
leftward movement of Shh-expressing cells; Shh expression on the right side is also
repressed by signaling by the TGF-b family member activin. The left-sided expression
of Nodal induced by combined Notch and Shh signaling then in turn activates the
expression of the transcription factor Pitx2 on the left side. If a pellet of cells secret-
ing Shh is placed on the right side, then organ asymmetry is randomized. The right
half of the embryo is protected from Nodal signaling by the protein Lefty secreted
by the notochord and the floor of the neural tube. Lefty antagonizes Nodal signaling
by binding to the Nodal protein and so prevents Nodal binding to and activating its
receptor. Lefty might also interact with the receptor itself and so compete for Nodal
binding (see Box 4B, p. 143).
Studies in mouse and zebrafish also implicate the release of calcium from cells on
the left side of the node or Kupffer’s vesicle, respectively, as part of the symmetry-
breaking process. In the mouse, one hypothesis suggests that the leftward flow of
fluid activates sensory cilia on the left side of the node, leading to calcium release and
the upregulation of Nodal (Fig. 4.17). Other suggestions are that the leftward flow
leads to the accumulation of a morphogen on the left of the node.
 The origin and specification of the germ layers 145

Summary
Nodal Nodal
Setting up the body axes in vertebrates involves maternal factors, external influences,
and cell–cell interactions. In the amphibian embryo, maternal factors determine the
animal–vegetal axis, which corresponds approximately to the antero-posterior axis; the
dorso-ventral axis is specified by the site of sperm entry and the resulting cortical rota- L R
tion relocates maternal factors to the dorsal side, which leads to the establishment of the
Nieuwkoop center. In the zebrafish, maternal factors also specify the dorsal side and the
embryonic shield, the site of the embryonic organizer, although the symmetry-breaking
event has not yet been identified in this case. In chick embryos, gravity determines the Nodal Nodal
side of the blastoderm at which the posterior marginal zone, and thus the primitive streak, Ca2+
forms, marking the orientation of the antero-posterior axis. It is not clear yet whether Ca2+
specification of the axes in the mouse embryo involves any maternal component. The axes Ca2+
are definitively established later in the epiblast under the direction of signals from embry-
onic and extra-embryonic tissues. The symmetry-breaking event in the mouse appears
to be the specification of the anterior end of the embryo, which is coupled to the deter-
Fig. 4.17 In the mouse embryo, cilia-
mination of the position of the primitive streak at the posterior end of the epiblast. The directed leftward flow of extracellular fluid
generation of the consistent left–right organ asymmetry found in vertebrates is marked generates left–right asymmetry across the
by the expression of the secreted protein Nodal and the transcription factor Pitx2 on the midline. In the mouse, the initial event that
left side only. This asymmetry appears to be generated by various mechanisms, involv- breaks left–right symmetry is a short period
ing a proton–potassium pump asymmetry (in Xenopus and chick), the leftward movement of directed leftward flow of extracellular fluid
of cilia (in mouse, humans, zebrafish, and Xenopus), and asymmetrically located calcium caused by the movement of cilia on a pit on
signaling, Notch pathway activation, Sonic hedgehog signaling and Nodal signaling. the ventral side of the node at around the
late head-fold stage. Before this flow, there
is a low level of expression of nodal mRNA on
both sides of the node (top panel, indicated by
short arrows). There are two types of cilia on
Summary: vertebrate axis determination
the cells lining a pit on the ventral side of the
node: central motile cilia (blue) that generate
Dorso-ventral axis Antero-posterior axis the flow and peripheral immotile cilia (red).
Coordinated rotational beating of the central
Xenopus sperm entry point and cortical rotation specified in organizer cilia generates a leftward flow of extracellular
lead to specification of dorsal side opposite the
fluid. One hypothesis is that the immotile cilia
point of sperm entry
act as mechanosensors for the direction of
Zebrafish initial symmetry-breaking event unknown. specified in organizer flow and their stimulation results in release
Localization of maternal Ndr1 and nuclear of intracellular Ca2+ in the cells on the left
β-catenin specifies dorsal side and side of the node, as illustrated in this figure.
position of organizer Propagation of the Ca2+ signal to nearby
cells then leads to the upregulation of Nodal
Chick posterior marginal zone specifies ventral end gravity determines position of
expression on the left side (bottom panel,
of dorso-ventral axis posterior marginal zone and thus
the posterior end of the A–P axis larger arrow). Other hypotheses propose that
the leftward flow leads to the accumulation of
Mouse interaction between inner cell mass specification and movement a chemical morphogen on the left-hand side of
and trophectoderm of dorsal visceral endoderm the node, and that this morphogen directly or
indirectly causes the upregulation of Nodal.

The origin and specification of the germ layers

We have seen in the preceding sections how the main axes are laid down in various
vertebrate embryos. We now focus on the earliest patterning of the embryo with
respect to these axes: the specification of the three germ layers—endoderm, meso-
derm, and ectoderm—and their further diversification (see Box 1C, p. 15).
All the tissues of the body are derived from these three germ layers. The mesoderm
becomes subdivided into cells that give rise to notochord, muscle, heart and kidney,
and blood-forming tissues, among others. The ectoderm becomes subdivided into cells
146 Chapter 4 Vertebrate development II: axes and germ layers

that give rise to the epidermis of the skin and cells that develop into the nervous sys-
tem. The endoderm gives rise to the gut and its associated organs such as the lungs,
liver, and pancreas. We first look at the fate maps of early embryos of the different
vertebrates, which tell us which tissues are generated from particular regions of the
embryo (see Section 1.12). We then consider how the germ layers are specified and
subdivided, with the main focus on Xenopus, in which these processes are best under-
stood and where many of the genes and proteins involved have been identified. The
fate maps and the genes involved in germ-layer specification are more similar in all
our vertebrate models than are the mechanisms involved in early axis specification.

4.8 A fate map of the amphibian blastula is constructed by following the fate
of labeled cells

The external appearance of the Xenopus blastula at the 32-cell stage gives no indica-
tion of how the different regions will develop, but individual cells can be identified
with respect to the animal–vegetal axis (defined by pigmentation) and dorso-ventral
axes (defined by the sperm entry point). By following the fate of individual cells, or
groups of cells, we can make a map on the blastula surface showing the regions that
will give rise to, for example, notochord, somites, nervous system, or gut. The fate
map shows where the tissues of each germ layer normally come from, but it indicates
neither the full potential of each region for development nor to what extent its fate
is already specified or determined in the blastula. In other words, we know what
each of the early cells will give rise to—but the embryo does not. Early vertebrate
embryos have considerable capacity for regulation when pieces are removed or are
transplanted to a different part of the same embryo (see Section 1.12). This implies
considerable developmental plasticity at this early stage and also that the actual fate
of cells is heavily dependent on the signals they receive from neighboring cells.
One way of making a fate map is to stain various parts of the surface of the early
embryo with a lipophilic dye such as DiI, and observe where the labeled region ends
up. Individual cells can also be labeled by injection of stable high-molecular-weight
molecules, such as fluorescein-labeled dextran, which cannot pass through cell mem-
branes and so are restricted to the injected cell and its progeny; as flourescein fluo-
resces green when excited by UV light, the fluorescein-labeled dextran can be easily
detected under a fluorescence microscope. The fluorescent protein green fluorescent
protein (GFP), which produces a green light when excited by UV, is now widely used
for such purposes (see Box 1D, p. 20). Figure 4.18 shows a Xenopus embryo labeled
with fluorescein-dextran-amine for fate mapping.

Dorsal

A4 A3 A2 A1

B4 B1
B3 B2
Fig. 4.18 Fate mapping of the early
Xenopus embryo. Left panel: a single cell C4 C1
C3 C2
in the embryo, C3, is labeled by injection of
fluorescein-dextran-amine, which fluoresces
green under UV light. Right panel: a cross- D4 D1
D3 D2
section of the embryo, made at the tailbud
stage, shows that the labeled cell has given
rise to mesoderm cells on one side of the
embryo. Scale bar = 0.5 mm. Ventral
Photograph courtesy of L. Dale.
 The origin and specification of the germ layers 147

The fate map of the late Xenopus blastula (Fig. 4.19) shows that the yolky vegetal
Fate map: lateral view
region, which occupies the lower third of the spherical blastula, gives rise to most of
the endoderm. The yolk, which is present in all cells, provides the nutrition for the Animal
developing embryo, and is gradually used up as development proceeds. At the other
pole, the animal hemisphere becomes ectoderm, which becomes further diversified ectoderm
into epidermis and the future nervous tissue. The future mesoderm forms from a (epidermis)
neural ectoderm
belt-like region, the marginal zone, around the equator of the blastula. In Xenopus, (nervous system)
Ventral Dorsal
but not in all amphibians, a thin outer layer of presumptive endoderm overlies the mesoderm
mesoderm
presumptive mesoderm in the marginal zone. (blood, kidney) (somites)

The fate map of the blastula makes clear the function of gastrulation. At the blas-
endoderm
tula stage, the endoderm, which gives rise to the gut, is on the outside and so must
move inside. Similarly, the mesoderm, which will form internal tissues and organs, marginal notochord
zone Vegetal
must move inwards. During gastrulation, the marginal zone moves into the interior
through the dorsal lip of the blastopore, which lies above the Nieuwkoop center. The
fate map of the mesoderm (see Fig. 4.19) shows that it is subdivided into different Fate map: dorsal view
fates along the dorso-ventral axis of the blastula. The most dorsal mesoderm gives
Animal
rise to the notochord, followed, going ventrally, by somites (which give rise to muscle
tissue), lateral plate (which contains heart and kidney mesoderm), and blood islands
(tissue where hematopoiesis first occurs in the embryo). There are also important
differences between the future dorsal and ventral sides of the animal hemisphere: the
epidermis comes mainly from the ventral side of the animal hemisphere, whereas the
nervous system comes from the dorsal side. After neural tube formation the epidermis
spreads to cover the whole of the embryo.
The terms dorsal and ventral in relation to the fate map can be somewhat confus-
ing, because the fate map does not correspond exactly to a neat set of axes at right blastopore
angles to each other. As a result of cell movements during gastrulation, cells from Vegetal
the dorsal side of the blastula give rise to some ventral parts of the anterior end of
the embryo, such as the head, as well as to dorsal structures, and will also form some Fig. 4.19 Fate map of a late Xenopus
other ventral structures, such as the heart. The ventral region gives rise to ventral blastula. The ectoderm gives rise to the
structures in the anterior part of the embryo, but will also form some dorsal structures epidermis and nervous system. Along the
posteriorly. This is why the dorsalized embryos described in Section 4.2 also have dorso-ventral axis the mesoderm gives rise
overdeveloped anterior structures and lack posterior regions. Indeed, a somewhat dif- to notochord, somites, heart, kidneys, and
ferent interpretation of the standard fate map for Xenopus has been proposed in which blood. Note that blood can also form in more
the ‘dorso-ventral’ axis of the blastula is considered to be the future antero-posterior dorsal regions. In Xenopus, although not in
all amphibians, there is also endoderm (not
axis of the embryo.
shown here) overlying the mesoderm in the
marginal zone.
4.9 The fate maps of vertebrates are variations on a basic plan

Fate maps of the early embryos of zebrafish, chick, and mouse have been prepared
using techniques essentially similar to those used for Xenopus: cells in the early
embryo are labeled and their fate followed. In the zebrafish embryo there is extensive
cell-mixing during the transition from blastula to gastrula, and so it is not possible
to construct a reproducible fate map at cleavage stages but only from gastrula stage.
The zebrafish late blastula comprises a blastoderm of deep cells and a thin ­overlying
­enveloping layer of cells (Fig. 4.20). The overlying enveloping layer is largely ­protective
and is eventually lost. At the beginning of gastrulation, the fate of deep-layer cells,
from which all the cells of the embryo will come, is correlated with their position in
respect to the animal pole. Cells at the dorsal and lateral margin of the blastoderm
give rise to endoderm and mesoderm (mesendoderm), cells slightly further toward
the animal pole give rise to mesoderm, while ectodermal cells come from the blas-
toderm nearest the animal pole (see Fig. 4.20). In general terms, the fate map of the
zebrafish is rather similar to that of an amphibian, if one imagines the vegetal region
of the amphibian blastula being replaced by one large yolk cell.
148 Chapter 4 Vertebrate development II: axes and germ layers

It is difficult to produce a fate map of the chick embryo at the early blastoderm
Fate map of zebrafish at the early gastrula stage
stage, the stage that roughly corresponds to the Xenopus blastula, because cells that
Animal
will contribute to the different germ layers are mixed up in the epiblast. Roughly, cells
adjacent to the posterior marginal zone at the early blastula stage are equivalent to
ectoderm neural the dorsal cells of the Xenopus blastula (future notochord). There are extensive cell
(epidermis) ectoderm
movements both before and during the emergence of the primitive streak, but the
Ventral mesoderm Dorsal picture becomes clearer once the primitive streak has formed and cells have started to
(somites)
move inside and be determined as mesoderm and endoderm.
At the stage shown in Fig. 4.21, the chick blastoderm has become a three-layered
uncleaved endoderm and
yolk cell mesoderm structure, with the endoderm forming the bottom layer (not visible in this dorsal
(mesendoderm)
mesoderm view). Cells from the surface of the blastoderm have ingressed through the primitive
(blood forming) notochord
Vegetal streak into the interior and there formed the mesodermal and endodermal layers.
Most of the cells in the outer surface of the blastoderm are now prospective ecto-
Fig. 4.20 Fate map of zebrafish at the derm and will form neural tube and epidermis, but there are still regions of the outer
early gastrula stage. The three germ blastoderm that will move through the streak and give rise to mesoderm. Hensen’s
layers come from the blastoderm, which sits node at the anterior end of the streak is prospective mesoderm; as the node regresses
on the lower hemisphere composed of an
it leaves cells behind that will form the notochord and will also contribute to the
uncleaved yolk cell. The endoderm comes
somites. The mesoderm lying along the antero-posterior midline gives rise to somites
from the dorsal and lateral margins of the
and is flanked by cells that will form the lateral plate mesoderm (shown as a single
blastoderm, which also give rise to mesoderm
(mesendoderm), and some has already moved region in Fig. 4.21) and structures such as the heart and kidney. In the lowest layer of
inside. the embryo, closest to the yolk, the presumptive endoderm (not visible in Fig. 4.21) is
flanked by cells that will form extra-embryonic structures.
In the case of the mouse, most cells from the inner cell mass of an embryo less
than 3.5 days old can give rise to many different embryonic tissues, as well as to
some ­extra-embryonic structures, such as the visceral and parietal endoderm, and
so at this stage a fate map cannot be constructed. At 6–7 days’ gestation the mouse
epiblast becomes transformed into the three germ layers by gastrulation. Gastrulation
in the mouse is essentially very similar to gastrulation in the chick, but at this stage
the mouse epiblast is folded into a cup, which makes the process of primitive streak
and node formation more difficult to follow. Like the chick, there is extensive cell
movement and cell proliferation in the epiblast. A detailed fate map of this stage has
been established by tracing the descendants of single cells that have been labeled by
injection with a dye. But descendants of a single cell can spread widely and give rise
Fate map of a chick embryo when the to cells of different germ layers, so that only about 50% of the labeled clones have
primitive streak has fully formed
progeny in only one germ layer.
Anterior The fate map of the mouse at the primitive-streak stage is basically similar to that
ectoderm of the chick (Fig. 4.22). As the mouse epiblast at this stage is cup-shaped, it may be
(epidermis) neural
ectoderm helpful to look back at Figs 3.24 and 3.25, which show the cup-shaped embryo, to
Hensen’s
node help interpret its fate map. Figure 4.22 shows the embryo as if the cup were opened
(notochord out and flattened and viewed from the dorsal side, which is the interior of the cup.
and somites)
primitive Apart from the topology, there are other differences from the chick fate map that
streak are worth noting; one is that in the mouse the node gives rise exclusively to the
mesoderm notochord. Cells from the anterior and middle regions of the streak will move for-
(somites and
lateral plate) ward around the node to produce the other mesodermal tissues (see Fig. 3.26). The
extra-embryonic definitive gut endoderm and the midline axial mesoderm (at this stage called mes-
mesoderm endoderm) derive from the extreme anterior end of the primitive streak, whereas
Posterior the middle part of the streak gives rise mainly to somites and lateral-plate meso-
derm. The posterior part of the streak will become the tailbud, and also provides
the extra-embryonic mesoderm that forms the extra-embryonic membranes—the
Fig. 4.21 Fate map of a chick embryo
when the primitive streak has fully
amnion, visceral yolk sac, and allantois.
formed. The diagram shows a view of the We can see, therefore, that the fate maps of the different vertebrates are similar
dorsal surface of the embryo. Almost all the when we look at the relationship between the germ layers and the site of the ­inward
endoderm has already moved through the ­movement of cells at gastrulation (Fig. 4.23). The differences are due mainly to the
streak to form a lower layer, so is not shown. amount of yolk in the different eggs, which determines the patterns of cleavage
 The origin and specification of the germ layers 149

and influences the shape of the early embryo. The similarity in relationship between
Fate map of a mouse at the late gastrula stage
the germ layers suggests that similar mechanisms will be involved in their specifi-
cation. This is true to some extent, but as we shall see, there are still considerable Anterior
­differences. ectoderm neural
(epidermis) ectoderm

4.10 Cells of early vertebrate embryos do not yet have their fates determined
and regulation is possible

Early vertebrate embryos have considerable powers of regulation (see Section 1.3)
when parts of the embryo are removed or rearranged. Experiments show that at the
blastula stage, and even later, many cells are not yet specified or determined (see
­Section 1.12 and Fig. 1.22); their potential for development is greater than their posi-
tion on the fate map suggests.
node
The state of determination of cells, or of small regions of an embryo, can be studied (notochord)
by transplanting them to a different region of a host embryo and seeing how they endoderm mesoderm
develop. If they are already determined, they will develop according to their original (somites and
extra-embryonic primitive lateral plate)
position. If they are not yet determined, they will develop in line with their new posi- mesoderm streak
tion. This can be shown experimentally by introducing a single labeled cell from a Posterior
Xenopus blastula into the blastocoel of a later-stage embryo and following its fate.
The transferred cell divides, and during gastrulation its progeny become distributed
Fig. 4.22 Fate map of a mouse at the late
to different parts of the embryo. gastrula stage. The embryo is depicted as
In general, cells in transplants made from early blastulas are not yet determined; if the ‘cup’ has been flattened and is viewed
their progeny will differentiate according to the signals they receive at their new from the dorsal side. At this stage the
location. So cells from the vegetal pole, which would normally form endoderm, can primitive streak is at its full length.
contribute to a wide range of other tissues such as muscle or nervous system, when
grafted at an early stage. Similarly, early animal pole cells, whose normal fate is epi-
dermis or nervous tissue, can form endoderm or mesoderm. With time, cells gradually Fig. 4.23 The fate maps of vertebrate
become determined, so that similar cells taken from later-stage blastulas and early embryos at comparable developmental
gastrulas develop according to their fate at the time of transplantation. stages. In spite of all the differences
Fragments of a fertilized Xenopus egg that are only one-fourth of the normal in early development, the fate maps of
volume can develop into more-or-less normally proportioned but small embryos. vertebrate embryos at stages equivalent
There must therefore be a patterning mechanism involving cell interactions that to a late blastula or early gastrula show
can cope with such differences in size. There are, however, limits to the capacity strong similarities. All maps are shown in a
for regulation. Isolated animal and vegetal halves of an eight-cell Xenopus embryo dorsal view. The future notochord mesoderm
occupies a central dorsal position. The neural
do not develop normally; and while dorsal halves of eight-cell embryos regulate to
ectoderm lies adjacent to the notochord, with
produce a reasonably normal embryo, the ventral halves do not. As we saw in Sec-
the rest of the ectoderm anterior to it. The
tion 4.3, this is due to the development of the essential embryonic signaling centers
mouse fate map depicts the late gastrula
in the dorsal half. stage. The future epidermal ectoderm of the
zebrafish is on its ventral side.

Frog Zebrafish Chick Mouse

Anterior Anterior Anterior Anterior

endoderm ectoderm neural ectoderm notochord mesoderm extra-embryonic mesoderm inward movement of cells
150 Chapter 4 Vertebrate development II: axes and germ layers

The early chick embryo also has remarkable powers of regulation, and fragments of
the blastoderm can give rise to whole embryos. The capacity to regulate persists right
up to the time at which the primitive streak starts to form. In chick there is no mid-
blastula transition, as zygotic genes are expressed from the start, and although Vg-1
is normally expressed in the posterior marginal zone (see Section 4.4), a new site of
expression of Vg-1 appears when the embryo is cut into pieces.
Early mouse embryos can also regulate to achieve the correct size. Giant embryos
formed by aggregation of several embryos in early cleavage stages can achieve normal
size within about 6 days by reducing cell proliferation. The mouse embryo retains
considerable capacity to regulate until late in gastrulation. Even at the primitive-
streak stage, up to 80% of the cells of the epiblast can be destroyed by treatment with
the cytotoxic drug mitomycin C, and the embryo can still recover and develop with
relatively minor abnormalities.
Further evidence for regulation in mammalian embryos, including human, and in
birds comes from twinning. As explained in Box 4C (p. 152), human twins usually
result from division of the inner cell mass, but twinning also very rarely occurs as
late as 9–15 days gestation, when the primitive streak is about to form. These powers
of regulation of human embryos have enabled the development of reproductive tech-
nologies such as preimplantation genetic diagnosis (Box 4D, p. 153).
Because early vertebrate embryos show considerable capacity for regulation and
many of the cells are not determined, this implies that cell–cell communication must
determine cell fate. To study this question, one can create chimeric mice—that is,
Fig. 4.24 Fusion of mouse embryos gives mice that are mosaics of cells with two different genetic constitutions—by fusing
rise to a chimera. If a four-cell stage embryo
two embryos (Fig. 4.24). Chimeras made from a normal embryo and one that is
of an unpigmented strain of mouse is fused
­genetically similar, but homozygous for a mutated version of a single gene, can
with a similar embryo of a pigmented strain,
the resulting embryo will give rise to a
be used to find out whether the effects of the gene are cell-autonomous or non-
chimeric animal, with a mixture of pigmented cell-autonomous. If only the mutant cells exhibit the mutant phenotype and are
and unpigmented cells. The distribution not ‘rescued’ by the normal cells, the gene is acting cell-autonomously. This means
of the different cells in the skin gives this that the product of the gene is acting solely within the cell in which it is made, and
chimera a patchy coat. is not influencing other cells. The cells of a black mouse, for example, remain black
when put into a white mouse, and they do not change the white cells to black;
they are thus autonomous with respect to pigmentation (see Fig. 4.24). In contrast,
a gene is acting non-­autonomously when either the mutant cells in the chimera
appear to act normally or the normal cells start to show the mutant phenotype. Non-
­autonomous action is typically due to a gene product such as an intercellular sig-
naling protein that is secreted by one cell and acts on others, whereas autonomous
effects are due to proteins such as transcription factors that act only within the cell
that makes them.

4.11 In Xenopus the endoderm and ectoderm are specified by maternal

factors, but the mesoderm is induced from ectoderm by signals from the
vegetal region

When explants from different regions of the early Xenopus blastula are cultured in a
simple medium containing the necessary salts for ion balance, tissue from the region
nearest the animal pole will form a ball of epidermal cells, while explants from the
vegetal region are endodermal in their development (Fig. 4.25, top panels, blue and
yellow, respectively). These results are in line with the normal fates of these regions.
It is thus generally accepted that in Xenopus all the ectoderm and most of the endo-
derm is specified by maternal factors in the egg. There is no evidence that any sig-
nals from other regions of the embryo are necessary for their initial specification.
The mesoderm, however, is different. It is induced from prospective ectoderm in a
band round the equator of the blastula in response to signals from the prospective
 The origin and specification of the germ layers 151

Tissues formed from explants from a Xenopus late blastula

Animal ectoderm
1
animal cap cells
1

2 3 mesenchyme notochord
Ventral equatorial cells Dorsal 2 3

muscle neural
epidermis blood tube
4
4
vegetal cells
Vegetal undifferentiated vegetal tissue

Vegetal tissue induces mesoderm in animal cap

animal cap cells induced mesodermal tissue

muscle
mesenchyme notochord

vegetal cells

Fig. 4.25 Induction of mesoderm by the vegetal region in the Xenopus blastula. Top
panels: explants of animal cap cells (blue, 1) or vegetal cells (yellow, 2) on their own from a late
blastula form only ectoderm or endoderm, respectively. Explants from the equatorial region
(3 and 4), where animal and vegetal regions are adjacent, form both ectodermal (epidermis and
neural tube, blue) and mesodermal tissues (mesenchyme, blood cells (such as erythrocytes),
notochord, and muscle, red), showing that mesoderm induction has taken place. The reason for
the differences in the mesodermal tissues formed by ventral and dorsal explants is explained
later in the chapter. Bottom panels: the standard experiment for studying mesoderm induction
is to take explants from animal and vegetal regions that would normally make only ectoderm
and endoderm, respectively, and culture them together for 3 days. When such explants from
an early blastula are combined and cultured, mesodermal tissues including notochord, muscle,
blood, and loose mesenchyme (red) are induced from the animal cap tissue.

endoderm. A small amount of anterior endoderm, which gives rise to the pharyngeal
endoderm, is also induced from the ectoderm along with the mesoderm.
The formation of mesoderm can be observed by culturing explants from the blas-
tula. When explants from the equatorial region of the late blastula, where animal and
vegetal cells are adjacent, are cultured in isolation, they form ectodermal derivatives
and mesoderm (see Fig. 4.25, top panels). When explants of animal and vegetal
­regions distant from the equator are cultured in contact with each other, mesodermal
tissues are also formed (see Fig. 4.25, bottom panels). Confirmation that it is the
­animal cap cells, and not the vegetal cells, that are forming mesoderm was obtained
by pre-labeling the animal region of the blastula with a cell-lineage marker and show-
ing that the labeled cells form the mesoderm. Clearly, the vegetal region is producing
a signal or signals that can induce mesoderm. Mesoderm can be distinguished by its
152 Chapter 4 Vertebrate development II: axes and germ layers

Box 4C Identical twins

The existence of identical, or mon­

ozygotic, twins shows that early Zygote
human embryos, like early-stage
mouse embryos, can regulate.
Identical twins are produced when
an early embryo splits in half— Splitting occurs
at 2-cell stage
unlike non-identical, or dizygotic,
twins, which are the result of the
fertilization of two separate eggs.
Monozygotic twinning can occur Splitting in early blastocyst
yields two inner cell masses
at various stages in early devel-
opment and the stage at which
it has occurred can be deduced
by the arrangement of the extra-
­embryonic membranes and the pla-
centa. Identical twins with separate
amniotic and chorionic cavities and
separate placentas are produced by Later splitting yields two embryos
the splitting of the morula during from one inner cell mass
cleavage and before implantation,
as shown here for twinning at the
two-cell stage (figure, left). More
commonly, twins have their own
amniotic cavities but share a cho-
rionic cavity and a placenta, which
represents splitting of the very
early inner cell mass inside the
blastocyst (figure, center). A very
small percentage of identical twins
(about 4%) share both amniotic
and chorionic cavities and a pla-
centa, which means that twinning
must have occurred quite late, with
splitting of the inner cell mass at
epiblast stages, between 9–15 days
after fertilization ­(figure, right).
Separate amnions, chorions, Separate amnions Common amnion,
Figure adapted from Larsen, W.J.: and placentas Common chorion and placenta chorion, and placenta
1997.

histology as, after a few days’ culture, it can contain muscle, notochord, blood, and
loose mesenchyme (connective tissue). It can also be identified by the typical proteins
that cells of mesodermal origin produce, such as muscle-specific actin.
In this explant system, if the animal and vegetal explants are separated by a filter
with pores too small to allow cell contacts to develop, mesoderm induction still takes
place. This suggests that the mesoderm-inducing signals are in the form of secreted
molecules that diffuse across the extracellular space, and do not pass directly from
cell to cell via cell junctions (see Fig. 1.23). Differentiation of a mesodermal tissue
such as muscle in the explant system appears to depend on a community effect in the
responding cells. A few animal cap cells placed on vegetal tissue will not be induced
to express muscle-specific genes. Even when a small number of individual cells are
 The origin and specification of the germ layers 153

Box 4D Preimplantation genetic diagnosis

We now take in vitro fertilization (IVF) of human eggs as a treat-

ment for infertility almost for granted, even though it is only
30 years since the first IVF baby, Louise Brown, was born in the
United Kingdom. More recently, it has become possible to deter-
mine the genotype of embryos produced by IVF before implan-
tation without harming the embryo. This procedure is called
­preimplantation genetic diagnosis (PGD) and was developed
in the late 1980s for use with fertile couples at risk of passing
on a serious genetic disease to their children. The aim was to
provide an alternative to prenatal diagnosis and the potential
termination of an affected pregnancy. linked to an increased susceptibility to prostate cancer. By deter-
Because of the capacity of human embryos to regulate, one mining whether an embryo from a high-risk family carries a BRCA1
blastomere can be removed from an IVF-produced embryo during disease allele the genetic susceptibility to these cancers could, in
its early cleavage in vitro without affecting subsequent devel- principle, be eliminated from a family. At least one baby has been
opment. The DNA from this blastomere can be amplified in vitro born to such a high-risk couple after IVF and selection by PGD for
and tested for the presence or absence of mutations known to the absence of the BRCA1 disease allele. In the United Kingdom,
cause disease. Where there is a known high risk that the parents PGD has been licensed by the Human Fertilisation and Embryol-
will pass on a particular genetic disease—for example, when the ogy Authority (HFEA) for more than 60 genetic diseases.
parents are both carriers for the cystic fibrosis gene—PGD can be There are practical and ethical questions in relation to PGD,
used to ensure that an embryo that would develop the disease such as which genetic conditions it should be applied to. An
is not implanted into the mother. ­ethical question that arose recently concerned parents who
The demand for PGD is likely to increase because it can be wished to select an IVF embryo with the best HLA match to a
used not only to identify embryos with mutations that will inevi- sibling suffering from a rare blood disease, so that the IVF child
tably cause a potentially fatal disease in infancy and childhood, could eventually donate stem cells to their sibling. The HFEA
but also to identify embryos with mutations in genes that predis- made a general ruling allowing this on a case-by-case basis in
pose an individual to disease in later life. An example is the gene 2004. One UK family, who have a child with the rare condition
BRCA1, certain mutations in which predispose women to develop Diamond–Blackfan anemia, have since had an HLA-matched
breast and ovarian cancer, and account for 80% of these tumors daughter selected by PGD after IVF.
in women with a genetically inherited predisposition (5–10% of Photograph courtesy of Dr Malpani, MD, Malpani Infertility Clinic,
all breast and ovarian cancer). In males, mutations in BRCA1 are Mumbai, India. www.drmalpani.com

placed between two groups of vegetal cells, induction does not occur. By contrast,
larger aggregates of animal cap cells respond by strongly expressing muscle-specific
genes (Fig. 4.26).
In contrast to mesoderm, the ectoderm and most of the endoderm are maternally
specified in Xenopus. The transcription factor VegT has a key role in specifying the
vegetal region as endoderm. VegT is translated from maternal mRNA present in the
vegetal region of the egg and is inherited by the cells that develop from this region.
Injection of VegT mRNA into animal cap cells induces expression of endoderm-­specific
markers, whereas blocking the translation of VegT mRNA in the vegetal region, by
injecting oligonucleotides that are antisense to the mRNA (see Box 6A, p. 220), results
in a loss of endoderm. VegT is also essential for mesoderm induction, as we shall see
later. In VegT-depleted embryos, ectoderm forms throughout the embryo.
A strong candidate for a maternal determinant of ectoderm is the E3 ubiquitin-­ligase
Ectodermin, which is translated from maternal mRNA located at the animal pole of the
fertilized egg and becomes localized in the nuclei of cells throughout the animal cap
(Fig. 4.27). Ectodermin functions to counteract potential mesoderm-inducing signals,
and in this way limits mesoderm formation to the equatorial zone. Ectodermin adds
ubiquitin to Smad4, a gene regulatory protein activated as a result of BMP ­signaling
(see Fig. 4.33), and this affects the interactions of Smad4 with other proteins.
154 Chapter 4 Vertebrate development II: axes and germ layers

Another determinant of ectoderm is the zygotic transcription factor Foxl1e, which

animal cap cells
is expressed in the cells of the animal half of the late blastula and is required to
maintain their regional identity and their development as ectoderm (see Fig. 4.27).
In the absence of Foxl1e, the animal cells mix with the cells of other germ layers and
differentiate according to their new positions.

4.12 Mesoderm induction occurs during a limited period in the blastula stage

vegetal cells Mesoderm induction is one of the key events in vertebrate development, as the meso-
derm itself is a source of essential signals for further development of the embryo.
In Xenopus, mesoderm induction is largely completed before gastrulation begins.
Explant experiments like those described above have shown that in Xenopus there is
a period of about 7 hours at the blastula stage during which animal cells are compe-
dissociated animal cap cells tent to respond to a mesoderm-inducing signal; cells lose their competence to respond
about 11 hours after fertilization. An exposure of about 2 hours to inducing signal is
sufficient for the induction of at least some mesoderm.
One might expect that the timing of expression of mesoderm-specific genes, such
as those of muscle, would be closely coupled to the time at which the mesoderm is
induced—but it is not. Explant experiments have shown that, irrespective of when
during the competent period the cells are exposed to the 2-hour induction, muscle-
specific gene expression always starts at about 5 hours after the end of the competent
period, a time that corresponds to mid-gastrula stage in the embryo (Fig. 4.28). Muscle
Cap cells develop Cap cells develop gene expression can occur as early as 5 hours after induction, if induction occurs late
as non-muscle cells as muscle cells
in the period of competence, or as late as 9 hours after induction, if induction ­occurs
early in the competent period. These results suggest that there is an ­independent

Signals that specify ectoderm, endoderm, and mesoderm in Xenopus

Animal

Foxl1e
Fig. 4.26 The community effect. A single
animal cap cell or a small number of animal
rm

ect

cap cells in contact with vegetal tissue are

ode

ode
ect

not induced to become mesodermal cells

rm

Ectodermin Ectodermin
and do not begin to express mesodermal
markers such as muscle-specific proteins.
A minimum of about 200 animal cap cells Ventral mesoderm mesoderm Dorsal
must be present for induction of muscle
differentiation to occur.
Nodals
endoderm endoderm

VegT

Vegetal

Fig. 4.27 Signals for germ-layer specification in Xenopus. In the vegetal half of the blastula,
the maternal factor VegT is essential for both endoderm specification and, indirectly, for the
production of Nodal proteins, which are one of the main signals that induce mesoderm from the
prospective ectodermal cells of the animal cap. In the animal half of the blastula, the maternal
factor Ectodermin counteracts potential mesoderm-inducing signals, and limits mesoderm
formation to the equatorial zone. The zygotic transcription factor FoxI1e is expressed in the
animal region and is necessary for the development of the animal-half cells as ectoderm and for
preventing their mixing with more vegetal cells and developing as mesoderm or endoderm. In
turn, Nodal signaling inhibits the expression of FoxI1e in the mesoderm.
 The origin and specification of the germ layers 155

Fig. 4.28 Timing of muscle gene

4 5 6 7 8 9 10 11 12 13 14 15 16
expression is not linked to the time of
Hours after fertilization mesoderm induction. Animal cap cells
Animal cap cells isolated from an early Xenopus blastula are
from a Xenopus
blastula competent to respond to mesoderm-inducing
signals only for a period of about 7 hours,
For 7 hours cap cells are ‘competent’ Muscle gene between 4 and 11 hours after fertilization.
to be induced to form mesoderm expression in induced
cells always occurs at For expression of target genes to occur,
mid-gastrula stage
Two-hour exposure to inducing signal exposure to inducer must be for at least
anywhere within this period is 2 hours within this period. Irrespective
sufficient for mesoderm formation
of when the induction occurs within this
competence period, muscle-gene activation
occurs at the same time: 16 hours after
timing mechanism by which the cells monitor the time elapsed since fertilization and fertilization.
then, provided that they have received the mesoderm-inducing signal, they express
muscle-specific genes.
The extended period of competence gives the embryo latitude as to when meso-
derm induction actually takes place, and means that the timing of the inductive ­signal
does not have to be rigorously linked to a short time period. Given this latitude,
an independent timing mechanism for mesodermal gene expression is one way to
keep subsequent development coordinated, whatever the actual time of mesoderm
­induction.

4.13 Zygotic gene expression is turned on in Xenopus at the mid-blastula

transition

Another event that appears to be under intrinsic temporal control, although not linked
to mesoderm induction, is the beginning of transcription from the embryo’s own
genes. The Xenopus egg contains quite large amounts of maternal mRNAs, which are
laid down during oogenesis. After fertilization, the rate of protein synthesis increases
1.5-fold, and a large number of new proteins are synthesized by translation of the
maternal mRNA. There is, in fact, very little new mRNA synthesis until 12 cleavages
Concentration
have taken place and the embryo contains 4096 cells. At this stage, transcription from
zygotic genes begins, cell cycles become asynchronous, and various other changes
Threshold
occur. This is known as the mid-blastula transition, although it does in fact occur
in the late blastula, just before gastrulation starts. A few zygotic genes are expressed
earlier, most notably two Nodal-related genes, Xnr-5 and Xnr-6, which are expressed Event Time
as early as the 256-cell stage.
How is the mid-blastula transition triggered? Suppression of cleavage but not DNA
Fig. 4.29 A timing mechanism that could
synthesis does not alter the timing of transcriptional activation, and so the timing operate in development. A mechanism
is not linked directly to cell division. Nor are cell–cell interactions involved, as dis- based on an analogy to an egg-timer
sociated blastomeres undergo the transition at the same time as intact embryos. The could measure time to the mid-blastula
key factor in triggering the mid-blastula transition seems to be the ratio of DNA to transition. The decrease in the active
cytoplasm—the quantity of DNA present per unit mass of cytoplasm. concentration of some molecule, such as a
Direct evidence for this comes from increasing the amount of DNA artificially by transcriptional repressor, could occur with
allowing more than one sperm to enter the egg or by injecting extra DNA into the egg. time, and the transition could occur when the
In both cases, transcriptional activation occurs prematurely, suggesting that there may repressor reaches a critically low threshold
concentration. This would be equivalent
be some fixed amount of a general repressor of transcription present initially in the
to all the sand running into the bottom of
egg cytoplasm. As the egg cleaves, the amount of cytoplasm does not increase, but
the egg-timer. In the embryo a reduction in
the amount of DNA does. The amount of repressor in relation to DNA gets smaller
repressor activity per nucleus would in fact
and smaller until there is insufficient to bind to all the available sites on the DNA and occur, because repressor concentration in the
the repression is lifted. Timing of the mid-blastula transition thus seems to fit with an embryo as a whole remains constant but the
hourglass egg-timer model (Fig. 4.29). In such a model something has to accumulate, number of nuclei increases as a result of cell
in this case DNA, until a threshold is reached. The threshold is determined by the division. The amount of repressor per nucleus
initial concentration of the cytoplasmic factor, which does not increase. thus gets progressively smaller over time.
156 Chapter 4 Vertebrate development II: axes and germ layers

Zebrafish embryos also go through a mid-blastula transition at which zygotic tran-

scription begins. It occurs at the 512-cell stage, coincidentally with the formation of
the syncytial layer at the margin between the blastoderm and yolk (see Section 3.2).
Shortly after this, b-catenin activates zygotic genes such as dharma, chordino, and
nodal-related in the future dorsal region (see Fig. 4.7).

4.14 Mesoderm-inducing and patterning signals in Xenopus are produced by

the vegetal region, the organizer, and the ventral mesoderm

Once it had been established that mesoderm is induced by signals from the vegetal
region, the next question was the nature of those signals and how the mesoderm
becomes patterned along the dorso-ventral axis. In the Xenopus late blastula fate map,
the mesoderm is divided into a number of regions along the dorso-ventral axis (see
Fig. 4.19): the most dorsal region gives rise to the notochord; the next most dorsal to
the somites (embryonic tissue that produces the skeletal muscles and the skeleton of
the trunk); then comes mesoderm giving rise to organs such as the kidney; with the
most ventral regions giving rise to blood only (although significant amounts of blood
will also form from more dorsal mesoderm).
This patterning is not intrinsic to the mesoderm and has to be acquired. Experiments
that combined different parts of the vegetal region with animal cap explants suggested
differences in mesoderm-patterning activity along the dorso-ventral axis (Fig. 4.30).
For example, dorsal vegetal tissue containing the Nieuwkoop center induces noto-
chord and muscle from animal cap cells, whereas ventral vegetal tissue induces
mainly blood-forming tissue and little muscle. Evidence for mesoderm-­patterning
signals from the organizer came from combining a fragment of dorsal marginal zone
from a late blastula with a fragment of the ventral presumptive mesoderm. The ven-
tral fragment will now form substantial amounts of muscle, whereas ventral meso-
derm isolated from an early blastula and cultured on its own will produce mainly
blood-forming tissue and mesenchyme.
These types of experiments, together with other evidence, indicated that the induc-
tion and initial patterning of the mesoderm in Xenopus is due to the concerted actions

Dorsal vegetal cells induce muscle and notochord from animal cap cells

animal cap cells

Ventral Dorsal

dorsal vegetal cells

Fig. 4.30 Differences in mesoderm Ventral vegetal cells induce blood and associated tissue from animal cap cells
induction by dorsal and ventral vegetal
regions. The dorsal vegetal region of the animal cap cells
Xenopus blastula, which contains the
Nieuwkoop center, induces notochord and
Ventral Dorsal
muscle from animal cap tissues, while ventral
vegetal cells induce blood and associated
tissues. This is good evidence for different
ventral vegetal cells
inducing signals coming from the dorsal and
ventral vegetal regions.
 The origin and specification of the germ layers 157

Fig. 4.31 Signals involved in mesoderm

organizer induction and patterning originate from
Animal
tissue (O)
different regions of the blastula. Signals
from the vegetal region initially induce the
mesoderm from the prospective ectoderm
Ventral Dorsal
mesoderm O of the animal cap, with the organizer region
being specified on the dorsal-most side where
the inducing signal is strongest and of longest
Vegetal
duration. Signals from the ventral side of
the embryo pattern the mesoderm ventrally,
and the extent of their influence is limited
of signals emanating from different parts of the blastula (Fig. 4.31). The vegetal region
by signals emanating from the Spemann
initially produces signals that induce the mesoderm and establish the Spemann orga-
organizer region (0), which counteract them.
nizer on the dorsal side. Another set of signals gives the ventral mesoderm its char-
acter. At the same time the organizer emits yet another set of signals that counteract
the ventral ­signals, limiting their range of action and so, for example, enabling the
mesoderm adjacent to the organizer to give rise to somites. The outcome of all this
activity is that by late blastula–early gastrula stage there is a band of prospective
mesoderm around the equator of the embryo and this is becoming subdivided into
regions with different fates.
The importance of the signals from the organizer in organizing both the dorso-
­ventral and the antero-posterior axis of the embryo is shown by a graft of the orga-
nizer into the ventral marginal zone of an early gastrula (Fig. 4.32). As discussed
earlier, the organizer graft induces a complete new dorso-ventral axis. This, together
with the induction of anterior and neural structures by the organizer, which we shall
discuss in Chapter 5, results in a twinned embryo.

4.15 Members of the TGF-b family have been identified

as mesoderm inducers

The explant experiments discussed in the previous sections suggested that secreted
extracellular signaling proteins were involved in mesoderm induction. Two main
approaches­ were then used to try and identify these factors in Xenopus. One was to
apply candidate proteins directly to isolated animal caps in culture. The other was to
inject mRNA encoding the suspected inducers into animal cells of the early blastula
and then culture the cells. By itself, however, the ability to induce mesoderm in culture
does not prove that a particular protein is a natural inducer in the embryo. Rigorous
criteria must be met before such a conclusion can be reached. These include the pres-
ence of the protein signal and its receptor in the right concentration, place, and time
in the embryo; the demonstration that the appropriate cells can respond to it; and
the demonstration that blocking the response prevents induction. On all these criteria,
growth factors of the TGF-b family (see Box 4A, p. 131) have been identified as the most
likely candidates for mesoderm inducers. Members of this family known to be involved
in early Xenopus development include the Xenopus Nodal-related proteins (Xnr-1, -2,
-4, -5, -6), the protein Derrière, the bone morphogenetic proteins (BMPs), Vg-1, and
activin.

Fig. 4.32 Transplantation of the Spemann organizer can induce a new axis in
Xenopus. The Spemann organizer region produces signals required for mesoderm induction
and patterning, and their effect can be seen by transplanting the Spemann organizer into the
ventral region of another gastrula. The resulting embryo has two distinct heads, one of which
was induced by the Spemann organizer. The organizer therefore produces signals that not only
result in patterning the mesoderm dorso-ventrally, but also induce neural tissue and anterior
structures. Scale bar = 1 mm.
Photograph courtesy of J. Smith.
158 Chapter 4 Vertebrate development II: axes and germ layers

Fig. 4.33 Signaling by members of the TGF-b family of growth factors uses different
Nodal combinations of receptor subunits and transcriptional regulatory proteins to activate
BMP TGF-�
ligand Activin different sets of target genes. Members of this family involved in early vertebrate
dimer development, such as Nodal and the Nodal-related proteins, bone morphogenetic proteins
(BMPs), and activin, are dimeric ligands that act at cell-surface receptors. These receptors are
heterodimers of two different subunits—type I and type II—with intracellular serine/threonine
kinase domains. There are several different forms of type I and type II subunits and these
combine to form distinctive receptors for different TGF-b family members. Binding of ligand
causes subunit II to phosphorylate receptor subunit I, which in turn phosphorylates intracellular
P P P P signaling proteins called Smads. These are named for the Sma and Mad proteins in Drosophila,
type I type II type I type II in which they were discovered. The phosphorylated Smads themselves bind to another type of
serine/ Smad, sometimes called co-Smads, forming a transcriptional regulatory complex that enters the
threonine
kinase nucleus to either activate or repress target genes. Different receptors use different Smads, thus
domain leading to the activation of different sets of target genes. The identity of the type I subunit
determines which Smads get activated. The biological response depends on the combination of
Smad1,5,8 Smad2,3 the activated target genes and the particular cellular environment.
P P

The signaling pathways stimulated by TGF-b growth factors are illustrated in

Smad4 Smad4
Fig. 4.33. Confirmation of the role of TGF-b growth factors in mesoderm induction
P P came initially from experiments that blocked activation of the receptors for these
factors (Fig. 4.34). The activin type II receptor is a receptor for several TGF-b family
growth factors and is uniformly distributed throughout the early Xenopus blastula.
When mRNA for a mutant receptor subunit is injected into the early embryo and
P P blocks TGF- b family signaling, mesoderm formation is prevented. Because different
TGF-b growth factors can act through the same receptor, these experiments were not
able to pinpoint the individual proteins responsible.
target genes target genes
More recently, it has become possible to study mesoderm induction using more
Ventralization of Dorsal mesoderm specific inhibitors of Nodal family proteins, and to detect the presence of signal-
mesoderm induction
Bone and cartilage Cell-cycle ing by using labeled antibodies to detect phosphorylated Smads and to distinguish
formation arrest their type. This latter technique enables the spatial patterns and dynamics of normal
Apoptosis Extracellular
matrix formation growth factor signaling to be visualized directly, and it is now widely used to detect
signaling by TGF-b family members. For example, in Xenopus, signaling by BMPs can
be distinguished from that of Nodal-related proteins or activin by looking for phos-
phorylated Smad1 or Smad2, respectively (see Fig. 4.33).

4.16 The zygotic expression of mesoderm-inducing and patterning

signals in Xenopus is activated by the combined actions of maternal VegT
and Wnt signaling

One maternal vegetal factor of great importance in the earliest stage of mesoderm
induction is the transcription factor VegT (see Section 4.1). Directly and indirectly,
VegT activates the zygotic expression of the Xnr proteins and Derrière. If VegT is
severely ­depleted, expression of Xnr genes and Derrière is downregulated and almost
no mesoderm develops, showing that VegT is crucial for its induction. Injection of
mRNAs for the Xnrs or Derrière can rescue mesoderm formation in the VegT-depleted
embryos, suggesting that these are most likely to be the direct mesoderm inducers.
The rescue of head, trunk, and tail mesoderm by the mRNAs for the genes Xnr-1,
Xnr-4, and Xnr-5 indicates that they are involved in the general induction of all meso-
derm, whereas Derrière mRNA rescues only trunk and tail, indicating that on its own
it cannot induce the anterior-most mesoderm. However, Nodal-related proteins act as
dimers (see Fig. 4.33), and assays of their activity in blastula explants have shown
that they can heterodimerize and cooperate to induce mesoderm. FGF signaling (see
Box 5B, p. 184) also appears to be necessary, but not sufficient by itself, for general
mesoderm induction.
 The origin and specification of the germ layers 159

Fig. 4.34 A mutant activin receptor blocks mesoderm induction. Receptors for proteins of
Normal and defective
the TGF-b family function as dimers (see Fig. 4.33). Receptor function in a cell can be blocked type I receptor proteins
by introducing mRNA encoding a mutant receptor subunit that lacks most of the cytoplasmic
domain, and so cannot function. The mutant receptor subunit can bind ligand and forms normal mutant
type I activin truncated
heterodimers with normal receptor subunits, but the complex cannot signal. The mutant receptor receptor
subunit thus acts in a dominant-negative fashion preventing receptor function. When mRNA
encoding the mutant activin receptor subunit is injected into cells of the two-cell Xenopus
embryo, subsequent mesoderm formation is blocked. No mesoderm or axial structures are
formed except for the cement gland, the most anterior structure of the embryo.

Numerous experiments have now shown that Xnrs are expressed in the Xenopus kinase
domain
vegetal region in a graded fashion at the appropriate time for mesoderm induction, and
that the difference between a dorsal and a ventral mesoderm fate is likely to be due to
expression of a cocktail of different Xnrs or differences in Xnr levels along the dorso-
ventral axis. Nuclear b-catenin stimulates Xnr gene transcription, and thus Xnr levels No signal from mutant receptor even when
will be highest on the dorsal side, where the VegT signal, Xnrs, and the b-catenin sig- dimerized with a normal type II receptor subunit
nal overlap (Fig. 4.35). Signaling via Xnrs stimulates phosphorylation and activation
of Smad2, and a wave of Smad2 activation, detected by anti-­phospho-Smad2 antibod-
ies, appears to sweep from the dorsal to the ventral side of the embryo from early bound ligand
to mid-gastrulation. All this evidence is consistent with the idea that the Spemann
organizer is specified where Nodal-related signals emanating from the vegetal region
are most intense and of longest duration (see Fig. 4.35).

4.17 Signals from the organizer pattern the mesoderm dorso-ventrally by

antagonizing the effects of ventral signals

We can now consider the other signals that pattern the mesoderm along the dorso-
ventral axis once it has been induced. Ventralization of the mesoderm is promoted by
gradients of activity of two signaling molecules, BMP-4 and Xwnt-8, with their high
Signal No signal

Mid-blastula Mesoderm Gastrula mRNA encoding mutant receptor injected

induction into both cells of a two-cell embryo
�-catenin ectoderm CNS

mesoderm

Ventral Dorsal Spemann

organizer
VegT Xnr proteins endoderm

�-catenin Nodal-related HIGH Organizer

No mesoderm or axial structures

VegT Nodal-related LOW Ventral mesoderm

Fig. 4.35 A gradient in Nodal-related proteins may provide the initial signals in mesoderm
induction. Maternal VegT in the vegetal region activates the transcription of Nodal-related
genes (Xnrs). The presence of b-catenin on the dorsal side results in a dorsal-to-ventral
gradient in the Xnr proteins. These induce mesoderm and, at high doses, specify the Spemann
organizer on the dorsal side. This is a simplified version because other signals, such as Vg-1
and activin, also have roles. CNS, central nervous system.
160 Chapter 4 Vertebrate development II: axes and germ layers

points on the ventral side of the blastula. Initially, BMP-4 protein is made uniformly
Expression of genes encoding secreted proteins
throughout the late Xenopus blastula, and zygotic Xwnt-8 protein is made in the
Animal prospective mesoderm (Fig. 4.36). As noted in Section 4.11, the potential mesoderm-
inducing and patterning activity of BMP-4 is restricted to the equatorial zone by the
action of Ectodermin. When the action of BMP-4 is blocked throughout the blastula by
introducing a dominant-negative mutant receptor (see Fig. 4.34), the embryo is dor-
salized, with ventral cells differentiating as both muscle and notochord. Conversely,
Ventral Dorsal
overexpression of BMP-4 ventralizes the embryo.
The restriction of BMP and Wnt-8 signaling to the more ventral regions is due to
dorsalizing signals that are secreted by the organizer in response to Nodal signaling
organizer and inhibit, or antagonize, BMP and Wnt-8 activity. The first of these antagonists to
region
be discovered was the protein Noggin, which inhibits BMP-4 activity. The gene for
Vegetal
Noggin was first discovered in a screen for factors that could rescue UV-irradiated
ventralized Xenopus embryos, in which a dorsal region had not been specified by the
BMP-4 Xwnt-8
usual activity of the b-catenin pathway. The Noggin gene is strongly expressed in the
Chordin, Noggin, Frzb, Follistatin, Cerberus Spemann organizer (Fig. 4.37). Noggin protein cannot induce mesoderm in animal-
cap explants but it can dorsalize explants of ventral marginal zone tissue, thus making
Fig. 4.36 Distribution of mRNAs encoding
it a good candidate for a signal that patterns the mesoderm along the dorso-ventral
secreted proteins in the Xenopus axis. The organizer was subsequently found to secrete a mixture of antagonists—the
blastula. The secreted proteins that are BMP antagonists Noggin, Chordin, and Follistatin, and the Wnt antagonist Frizzled-
made in the organizer block the action of ­related protein (Frzb). The use of antagonists to restrict signaling to a particular
BMP-4 and Xwnt-8, proteins that are made region, or to modulate the level of signaling, is widespread in embryonic develop-
throughout the blastula and prospective ment, as noted earlier in relation to anterior specification and left–right asymmetry in
mesoderm, respectively, where their genes chick and mouse (see Sections 4.6 and 4.7).
are expressed. Noggin, Chordin, and Follistatin proteins interact with BMP-4 protein and prevent
it from binding to its receptor. In this way, a functional gradient of BMP-4 activity
is set up across the dorso-ventral axis with its highest point ventrally and little or
no activity in the presumptive dorsal mesoderm. Frzb generates a similar ventral-­
to-dorsal gradient of Wnt activity by binding to Wnt proteins and preventing them
acting dorsally. Yet another signaling antagonist is Cerberus, which inhibits the activ-
ity of BMPs, Nodal-related proteins, and Wnt proteins in prospective anterior tissues.
In Xenopus, Cerberus is expressed in the organizer and the anterior endoderm, and it
is involved in the suppression of mesoderm fate and induction of anterior structures,
in particular the head, as discussed in more detail in Chapter 5. The summary table
on p. 167 lists some of the main proteins so far identified as mesoderm inducers and
mesoderm-patterning factors in Xenopus, while Fig. 4.38 summarizes which signal-
ing systems are active and which are antagonized in the different germ layers at the
blastula stage.
The antagonistic action of Chordin on BMP-4 mirrors that of the Drosophila Chordin
homolog Sog on the BMP-4 homolog Decapentaplegic (Dpp) in the patterning of the
dorso-ventral axis in the fly (see Section 2.20). In the fly, however, the dorso-ventral
axis is inverted compared with that of vertebrates, and so Dpp specifies a dorsal fate
whereas BMP-4 specifies a ventral fate (the inversion of the dorso-ventral axis dur-
ing evolution is discussed in Section 15.6). Other components of the mechanism that
forms the gradient of Dpp activity in Drosophila are also conserved in vertebrates. Xol-
loid, for example, is the Xenopus equivalent of the fly metalloproteinase Tolloid, and
degrades Chordin. Xolloid is thought to act as a clearing agent for Chordin, reducing
the extent of its long-range diffusion and helping to maintain a gradient of Chordin
Fig. 4.37 Expression of noggin in the dorsalizing activity.
Xenopus blastula. Staining for expression of Mutagenesis screens have identified genes in zebrafish similar to those involved in
noggin mRNA locates it in the region of the mesoderm induction and dorso-ventral patterning in Xenopus, and they confirm the
Spemann organizer. Scale bar = 1 mm. general outline of germ-layer specification. The zebrafish signaling protein Nodal-
Photograph courtesy of R. Harland, from related 1 (Ndr1, also called Squint) is expressed in the dorsal yolk syncytial layer and
Smith, W.C., et al.: 1992. overlying cells as a result of b-catenin signaling. Together with another Nodal-related
 The origin and specification of the germ layers 161

protein, Ndr2 (Cyclops), Ndr1 is required for the specification of the marginal blasto-
Animal cells
meres as mesendoderm: prospective endoderm and mesoderm. High levels of Nodal
signaling are thought to specify endoderm and lower levels the mesoderm. Double
BMP Pro-neural fate
mutants of Ndr1 and Ndr2 lack both head and trunk mesoderm, but some mesoderm FGF
still develops in the tail region. The Ndr proteins signal via Smad2. Nuclear accumu-
lation of Smad2 and the formation of the complex between Smad2 and its co-Smad,
Smad4, have been visualized in living transgenic zebrafish embryos carrying a Smad2 BMP Pro-epidermal fate
Xnr
gene fused with a sequence encoding a fluorescent tag. This experiment revealed a
graded distribution of Nodal-type signaling, most probably due to Ndr1, in the blas-
toderm of these embryos, with highest activity at the margin of the blastoderm and Equatorial cells
grading off towards the animal pole, and confirmed that a gradient of Nodal signaling
patterns the zebrafish blastoderm into endoderm, mesoderm, and ectoderm. Nodal BMP Head mesoderm
signaling also appears to be highest and of longest duration in the dorsal marginal Wnt
Xnr
region of the blastoderm, suggesting that it may also help to pattern the mesoderm
dorso-ventrally. Many of the other proteins involved in Xenopus mesoderm induc-
BMP Notochord
tion and specification have been found to be involved, such as zebrafish versions of Wnt
Chordin and BMPs. Xnr

BMP Pro-myogenic fate

4.18 Threshold responses to gradients of signaling proteins are likely to FGF
pattern the mesoderm Xnr

The mesoderm-inducing and patterning signals described in previous sections exert

their developmental effects by inducing the expression of specific genes that are Vegetal cells
required for the further development of the mesoderm into, for example, notochord,
muscle, or blood. Expression of the gene Brachyury, which encodes a T-box family Xnr Pro-endodermal fate
transcription factor, is one of the earliest markers of mesoderm in all vertebrates, FGF
BMP
and Brachyury is thought to act as a key transcription factor in mesoderm specifica-
tion and patterning. The Xenopus version of the Brachyury gene, Xbra, is expressed
throughout the presumptive mesoderm in the late blastula and early gastrula from
Fig. 4.38 Summary of active and
soon after the mid-blastula transition (Fig. 4.39), later becoming confined to the
antagonized signals in the different germ
­notochord (the dorsal-most derivative of the mesoderm), and to the tailbud (posterior layers in a Xenopus blastula. Three main
mesoderm). Brachyury is essential for the development of posterior mesoderm. The families of secreted signaling proteins
gene was originally discovered in the mouse as a result of semi-dominant mutations are involved in germ-layer specification in
that produced short tails; when homozygous, this mutation is lethal in the embryo as Xenopus: FGFs, BMPs, and Nodals (Xnrs).
a result of a lack of development of posterior mesoderm (see Fig. 1.13). Loss of the Signals in green lettering are active. Those in
zebrafish ortholog of Brachyury, called No tail (Ntl), similarly results in fish lacking a black with a down arrow are inhibited.
tail. No tail and other zebrafish genes involved in specifying the mesoderm have been Adapted from Heasman, J.: 2006.
linked up into a gene regulatory network using the techniques of chromatin immuno-
precipitation followed by DNA sequencing (ChIP-seq) (Box 4E, p. 162).
One of the first zygotic genes to be expressed specifically in the Xenopus organizer,
and in other vertebrate embryonic organizer regions, is goosecoid, which encodes a
transcription factor somewhat similar to both the Gooseberry and Bicoid proteins of
Drosophila—hence the name. In line with its presence in the organizer, microinjec-
tion of goosecoid mRNA into the ventral region of a blastula mimics to some extent
transplantation of the Spemann organizer (see Fig. 4.32), resulting in the formation of
a secondary axis. Genes for other transcription factors are also specifically expressed
in the organizer region (Fig. 4.40 and summary table, p. 167).
It is still not clear how the secreted signaling proteins present in the mesoderm turn
on genes like goosecoid and Brachyury in the right place. For example, Brachyury can

Fig. 4.39 Expression of Brachyury in the Xenopus blastula. A cross-section through the
embryo along the animal–vegetal axis shows that Brachyury (red) is expressed in the prospective
mesoderm. Scale bar = 0.5 mm.
Photograph courtesy of M. Sargent and L. Essex.
162 Chapter 4 Vertebrate development II: axes and germ layers

Fig. 4.40 Expression of transcription factor genes in a late Xenopus blastula. The
Expression of genes encoding transcription factors
expression domains of a number of zygotic genes that code for transcription factors correspond
Animal quite well to demarcations on the specification map, as judged by the distribution of their
mRNAs. Brachyury is expressed in a ring around the embryo corresponding quite closely to
the future mesoderm (see also Fig. 8.28). A number of transcription factors are specifically
expressed in the region of the dorsal mesoderm that corresponds to the Spemann organizer
and are required for its function. Xnot protein appears to have a role in the specification of
Ventral Dorsal notochord; Goosecoid and Xlim1 function are required for a transplanted Spemann organizer to
be able to induce the formation of a new head.

organizer be experimentally switched on in Xenopus animal caps by activin, but what induces
region
it normally in the presumptive mesoderm in vivo is still unclear; a combination of
Vegetal
Vg-1 and Xnrs is thought to be most likely, and its expression is probably maintained
by FGF. As we have seen in the previous sections, there are gradients of activity of
Brachyury goosecoid, Xnot, Xlim1

Box 4E A zebrafish gene regulatory network

The zebrafish transcription factor

No-tail (Ntl) is a homolog of the
mouse mesodermal transcription
factor Brachyury and the technique
of chromatin immunoprecipitation
followed by DNA sequencing (ChIP-
seq, see Fig. 3.35) has been used to
uncover the targets of Ntl in meso-
derm specification, patterning and
differentiation. Ntl activates a net-
work of genes encoding transcrip-
tion factors (for example, Snail
and the T-box (Tbx) transcription
factors), intercellular signals (for
example, FGFs, Notch, and Wnt-11
in its later role as a ­signal direct-
ing convergent extension during
gastrulation, see Box 8C, p. 310),
and genes involved in the differ-
entiation of specific cell types (for
example, genes encoding the tran-
scription factors Pax3, FoxD3, and
Myf5 acting in muscle differentia-
tion, see Chapter 10).
In the figure, genes are grouped
to represent different aspects of
Ntl activity in the mesoderm: control of morphogenetic move- has been verified either genetically (in the case of signaling
ments, notochord specification, muscle specification, posterior molecules) or by transcription factor binding to the target gene
identity, and left–right patterning. Expression of all the genes promoter. Shaded boxes indicate that additional assays have
shown here is activated by Ntl. Double arrowheads leading from also shown direct regulation of the target gene by Ntl. Genes
a gene indicate that the gene product is an intercellular signaling whose encoded proteins regulate other genes have been colored
molecule. Solid arrowed lines joining two genes indicate that the for clarity. Note the number of gene products that feed back to
interaction has been verified both genetically and by demonstra- regulate expression of the ntl gene itself.
tion of direct binding of the encoded transcription factor to the
target gene promoter. Dashed lines indicate that the ­interaction Figure from Morley, R.H.: 2009.
 The origin and specification of the germ layers 163

Fig. 4.41 Graded responses of early Xenopus tissue to increasing concentrations of

Genes expressed by ectodermal cells in response
activin. When animal cap cells are treated with increasing concentrations of activin, particular to increasing concentration of activin
genes are activated at specific concentrations, as shown in the top panel. At intermediate
concentrations of activin, Brachyury is induced, whereas goosecoid, which is typical of the Activin
organizer region, is only induced at high concentrations. If beads releasing a low concentration concentration
of activin are placed in the center of a mass of animal cap cells (lower left panel), expression of
low-response genes such as Brachyury is induced immediately around the beads. With a high
concentration of activin in the beads (lower right panel), goosecoid and other high-response
genes are now expressed around the beads and the low-response genes farther away.

secreted signaling molecules throughout the mesoderm that could provide positional keratin Brachyury goosecoid
Epidermis Muscle Organizer
information for switching on developmental genes. (�-actin)
Experiments with the TGF-family member activin provide a beautiful example of
Response of ectodermal cells
how a diffusible protein could pattern a tissue by turning on particular genes at ­specific
threshold concentrations. Although activin itself is not likely to be responsible for
Low-concentration High-concentration
primary mesoderm patterning in vivo, it acts through the same receptor as the Xnrs, activin beads activin beads
and animal cap cells from a Xenopus blastula respond to increasing doses of activin animal cap cells
by activation of different mesodermal genes at different threshold ­concentrations. In
this explant system, increasing concentrations of activin specify several distinct cell
states that correspond to the different regions along the dorso-ventral axis. At the
lowest concentrations of activin, only epidermal genes are expressed; no mesoderm
is induced. Then, as the concentration increases, Brachyury is expressed, together
with muscle-specific genes such as that for actin. With a further increase in activin
low-response genes high-response genes
concentration, goosecoid is expressed, corresponding to the dorsal-most region of the e.g. Brachyury e.g. goosecoid
mesoderm—the Spemann organizer—which produces the notochord (Fig. 4.41). One
can see, therefore how graded growth-factor signaling along the dorso-ventral axis
could, in principle, activate transcription factors in particular regions and thus pattern
the tissues. A 1.5-fold increase in activin concentration is sufficient to cause a change
from formation of muscle to formation of notochord by explants of animal caps.
Just how gradients are set up in vivo with the necessary precision is not clear, how-
ever. Simple diffusion of morphogen molecules may play a part, but more complex
processes are likely to be involved, as we saw in the formation of the Dpp gradient
in Drosophila (see Section 2.20). Some common cellular mechanisms that could be
involved in gradient formation are indicated in Box 11A, p. 421.
How do cells distinguish between different concentrations of growth factor? Occu-
pation of just 100 activin receptors per cell is required to activate expression of
Brachyury, whereas 300 receptors must be occupied for goosecoid to be expressed.
The link between the strength of the signal and gene expression may not be so sim-
ple, however. There are additional layers of intracellular regulation; cells expressing
goosecoid at high activin concentrations also repress Brachyury, for example, and this
involves the action of the Goosecoid protein itself, together with other proteins.
Excellent evidence for a secreted morphogen turning on mesodermal genes at
­specific threshold concentrations also comes from zebrafish. The putative morpho-
gen here is the protein Ndr1, which is involved in patterning the mesoderm (see
­Section 4.17). Injection of Ndr1 mRNA into a single cell of an early zebrafish embryo
results in high-threshold Ndr1 target genes being activated in adjacent cells, whereas
low-threshold genes are activated in more distant cells.

4.19 Mesoderm induction and patterning in the chick and mouse occurs
during primitive-streak formation

In the chick, most mesoderm induction and patterning occurs at the primitive streak.
Chick epiblast isolated before streak formation will form some mesoderm containing­
blood vessels, blood cells, and some muscle, but no dorsal mesodermal structures,
such as notochord. Treatment of the isolated epiblast with activin, however, results in
164 Chapter 4 Vertebrate development II: axes and germ layers

Fig. 4.42 Signals at the posterior marginal zone of the chick epiblast that initiate
Signaling by epiblast cells at the
posterior marginal zone primitive streak formation. Epiblast cells in the posterior marginal zone overlying Koller’s
sickle secrete Vg-1 and Wnts. These signals induce expression of nodal in neighboring epiblast
cells, but Nodal protein function is blocked by Cerberus, which is produced by the hypoblast.
When the hypoblast has been displaced by endoblast, Nodal signaling in the epiblast and FGF
signals from Koller’s sickle induce the internalization of epiblast cells and formation of the
primitive streak.
Koller’s sickle

the additional appearance of notochord and more muscle, suggesting that TGF-b ­family
Nodal function antagonized by Cerberus
produced by hypoblast members such as Nodal act as mesoderm-inducing and/or mesoderm-­patterning sig-
nals in chick embryos as they do in Xenopus.
As we saw in Section 4.4, the posterior marginal zone in the chick embryo is crucial
for the formation of the primitive streak, and the zone also supplies signals for meso-
derm induction and patterning. The primitive streak is initiated in the posterior mar-
ginal zone by the actions of Wnt-8c, which is present throughout the marginal zone in
hypoblast prospective endoblast
a gradient with its high point at the posterior, and Vg-1, which is localized to the most
posterior region of the marginal zone. Once the hypoblast and its inhibitory signal
Hypoblast replaced by endoblast, Nodal and FGF
induce formation of primitive streak have been replaced by endoblast, Vg-1 and Wnt-8c induce the expression of Nodal
in the posterior marginal zone and the streak (Fig. 4.42). Fibroblast growth factor
start of primitive streak
(FGF) is expressed in Koller’s sickle and together with Nodal induces the internaliza-
tion of epiblast cells, by formation of the primitive streak and mesoderm. The protein
­Chordin is made in the cells at the tip of the streak. It antagonizes the signaling activ-
ity of bone morphogenetic proteins (BMPs), which are secreted by cells outside the
streak and tend to inhibit the formation of axial mesoderm, as they do in Xenopus. All
endoblast
these signals are required for streak formation and a fully functional Hensen’s node.
Vg-1 + Wnt Cerberus Nodal FGF Mesodermal patterning also occurs in the streak, with different regions of the streak
giving rise to different mesodermal tissues along the dorso-ventral axis: the tip of the
streak, including Hensen’s node, gives rise to the notochord and somites, and the
­posterior streak generates lateral mesoderm, including blood and blood vessels, and
extra-embryonic mesoderm (Fig. 4.43). Thus, although the antero-posterior axis of
the primitive streak marks the orientation of the future antero-posterior axis of the
embryo, dorso-ventral patterning of the mesoderm also initially occurs along this axis.
Fate map of chick primitive streak This occurs by the more posterior cells moving progressively forward and outwards
notochord, prechordal head mesoderm to take up their positions along the dorso-ventral axis, which is orthogonal to the
antero-posterior axis.
medial part In the mouse, mesoderm induction occurs only in the primitive streak, which is
of somite initiated at around E6, starting from one small region at one side of the epiblast (see
Fig. 3.24). This region is specified as a result of the determination of the anterior end
lateral part
of somite of the embryo by the movement of the distal visceral endoderm, which then becomes
the anterior visceral endoderm (AVE) (see Section 4.6). As the anterior ectoderm
kidney becomes determined, proximal epiblast cells that express genes characteristic of pro-
spective mesoderm start to converge on the posterior proximal region and internalize
lateral at the point of convergence to initiate the primitive streak (Fig. 4.44, first and second
plate
panels). BMP-4 is expressed in the extra-embryonic ectoderm above the proximal
rim of the epiblast and activates the expression of mesodermal markers in adjacent
proximal epiblast cells. Nodal is required both to establish the posterior end of the
extra-
embryonic primitive streak and for mesoderm formation, and is expressed in the primitive streak
mesoderm
along with Wnts. In mutants lacking Nodal function, mesoderm does not form. Genes
encoding secreted anatagonists of Nodal and Wnt signaling, such as Cerberus-like 1
Fig. 4.43 Mesoderm is patterned within
and Dickkopf, respectively, are expressed on the anterior side of the embryo and
the chick primitive streak. Different parts ­inhibit BMP-4, Nodal, and Wnt signaling in the AVE and its overlying ectoderm.
of the primitive streak give rise to mesoderm This allows the ectoderm to develop as anterior tissues. Altogether, this indicates
with different fates. The arrows indicate the that, as in other vertebrates, graded Nodal signals pattern the mouse embryo, with
direction of movement of the mesodermal cells. the highest levels of Nodal becoming restricted to the proximal posterior region of
 The origin and specification of the germ layers 165

E6 days E6.5 days E7 days Signaling within the E6.5-day streak

extra-embryonic
BMP-4
extra-
embryonic
Anterior Posterior mesoderm BMPR1

streak
proximal primitive primitive
epiblast streak streak
Brachyury (T)
Wnt-3 Nodal
organizer
anterior (node)
visceral endoderm anterior

node
ectoderm Gsc

axial mesendoderm ectoderm

Fig. 4.44 Specification of the primitive

streak in the mouse embryo. First panel: by
the epiblast and being required to induce primitive-streak formation. Dose-dependent
E6 the anterior visceral endoderm (AVE) has
Nodal ­signaling also patterns the mesendoderm and mesoderm, with definitive gut induced anterior character in the underlying
endoderm and prechordal plate being specified by high levels of Nodal signaling, the epiblast (small arrows) and early markers
node by intermediate levels of signaling, and paraxial and lateral-plate mesoderm by of the future primitive streak are restricted
low levels of Nodal signaling. to the proximal rim of the epiblast. Second
In the mouse, as in the chick, mesoderm patterning occurs in the primitive streak and panel: BMP-4 (red dots) is transiently
the position along the primitive streak at which cells ingress determines their eventual expressed in the adjacent extra-embryonic
fate. The earliest cells to ingress are those in the posterior streak which give rise to extra- ectoderm, and the posterior movement
embryonic tissues, while cells that ingress in more intermediate and anterior regions of (white arrow) of cells expressing primitive-
streak markers (purple) results in the
the streak form lateral mesoderm, and cells ingressing at the anterior end of the streak
primitive streak forming opposite the AVE
give rise to anterior mesendoderm. Once inside the embryo, these cells from the ante-
at 6.5 days. Third panel: by 7 days, extra-
rior end of the streak move anteriorly along the midline, pattern the overlying neurecto-
embryonic mesoderm (brown) is produced
derm, and give rise to dorsal mesoderm such as the notochord. At the anterior-most tip from the posterior end of the streak while
of the streak, cells that will give rise to the definitive endoderm also ingress. How all the the node (red) forms at the anterior end. The
different cell lineages are segregated at the tip of the streak is not yet understood. extreme anterior end of the streak gives
Having looked at the induction and initial patterning of the mesoderm, we are rise to axial mesendoderm, which will form
now in a position to consider the final emergence of the typical vertebrate body plan. prechordal plate and the gut endoderm.
In the next chapter, we will discuss patterning of the germ layers along the antero- Fourth panel: a possible scheme of the
posterior axis during gastrulation, the initial formation of the nervous system, and the molecular interactions within the streak at
­development of notochord and somites from mesoderm. 6.5 days. Gsc, goosecoid.

Summary

There are strong similarities in the fate maps of the frog, zebrafish, chick, and mouse.
Even though there is good evidence for the maternal specification of some regions
such as the future endoderm in amphibians, the embryo can still undergo considerable
regulation at the blastula stage. This implies that interactions between cells, rather
than intrinsic factors, have a central role even in early amphibian development. This
strategy is even more pronounced in the mouse and chick, where it appears to be posi-
tion that determines cell fate.
In Xenopus, the mesoderm and a small amount of endoderm are induced from pro-
spective ectoderm at the equator of the blastula by signals from the vegetal region.
Patterning of the mesoderm along the dorso-ventral axis produces regions that, from
ventral to dorsal, give rise to blood and blood vessels, internal organs such as kidneys,
the somites, and the organizer region in the most dorsal mesoderm, which produces
the notochord. Signals originating from the ventral side of the blastula, give the ventral
mesoderm its character. These signals are counteracted by signals emanating from the
organizer, which limit their influence.
166 Chapter 4 Vertebrate development II: axes and germ layers

Protein growth factors of the TGF-b family are excellent candidates for mesoderm-
inducing factors as well as for patterning the mesoderm. Other secreted proteins, such as
Noggin and Chordin, inhibit the action of BMP-4, and so are involved in specifying the dor-
sal mesoderm. Brachyury and goosecoid are early mesodermally expressed genes encod-
ing transcription factors, and their pattern of expression may be specified by gradients of
signaling proteins, the genes being turned on at particular threshold concentrations.

Summary: mesoderm induction in Xenopus

VegT in vegetal region

general inducing signal
e.g. Nodal-related
induction

ventral mesoderm dorsal mesoderm with

Spemann organizer

ventralizing signals dorsalizing signals

e.g. BMP-4, Xwnt-8 e.g. Noggin, Chordin
patterning of mesoderm

Summary to Chapter 4

All vertebrates have the same basic body plan. During early development, the antero-­posterior
and dorso-ventral axes of this body plan are set up. The mechanism appears to be ­different
in frog, chick, zebrafish, and mouse but can involve localized maternal ­determinants, external
signals, and cell-cell interactions. This early patterning also establishes bilateral asymmetry. It
is possible to construct a fate map in the early embryo for the three germ layers—mesoderm,
endoderm, and ectoderm. The fate maps of the different vertebrates have strong similarities.
At this early stage the embryos are still capable of considerable regulation and this empha-
sizes the essential role of cell–cell interactions in development. In Xenopus, signals emanating
from different regions of the blastula are involved in mesoderm induction and early pattern-
ing. Good candidates for these signals have been identified and include members of the TGF-b
family and Wnt proteins. At particular concentrations, these signals activate mesoderm-
­specific genes such as Brachyury and so gradients could pattern the mesoderm. The summary
table on p. 166 lists all genes considered in this chapter in relation to Xenopus.

n End of chapter questions a) The two dorsal cells of a four-cell embryo are isolated from the
two ventral cells. What does each half embryo develop into?
Long answer (concept questions) b) Cells from the animal region of a late blastula are cultured
1. The organization of the Xenopus embryo begins with the estab- together with cells from the dorsal vegetal region.
lishment of the dorso-ventral axis. A critical component of this event c) Cells from the dorsal vegetal region of a 32-cell embryo are
is the accumulation of b-catenin in cells of the dorsal marginal zone. grafted into the ventral marginal zone of an early gastrula.
Describe the series of events that lead to this accumulation. Include
4. What is the result of the following experiment?
the maternal deposition of factors into the oocyte, events accompany-
The dorsal lip of the blastopore from one early Xenopus gastrula is
ing sperm entry, cortical rotation, and the Wnt signaling pathway.
grafted into the ventral marginal zone of another early gastrula (see
2. Give a molecular explanation for the ventralizing effects of: UV also Section 1.4). What is the explanation?
irradiation of the egg, and treatment of the embryo with lithium.
5. Describe the developmental roles of the VegT protein found in
3. The blastula organizer of Xenopus, also known as the Nieuw­koop fertilized eggs and early embryos in Xenopus embryonic develop-
center, was defined by various experiments. Describe the results of ment. Include in your answer what kind of protein it is; whether
the following experiments: it is maternally supplied, or is transcribed and translated from the
 The origin and specification of the germ layers 167

Summary: main genes involved in patterning of axes and germ layers in Xenopus

Gene Maternal/ Type of protein Where Function of protein

zygotic expressed

Specification of germ layers and dorso-ventral axis

VegT M Transcription factor Vegetal region Endoderm specification; activates

expression of mesoderm inducers
Ectodermin M Ubiquitin ligase Animal hemisphere Ectoderm specification; inhibits
mesoderm formation
Vg-1 M TGF-β family Vegetal region; RNA enriched Mesoderm induction
dorsally after fertilization
Xwnt-11 M Wnt family Vegetal region; RNA Specification of dorsal structures and
translocated to dorsal side organizer formation
after fertilization
Dishevelled M Wnt pathway signaling Protein associated with Specification of dorsal structures and
protein vesicles that move to dorsal organizer formation
side of embryo after
fertilization
β-catenin M Acts in Wnt pathway; Protein enriched in dorsal Specification of dorsal structures and
regulates gene nuclei in response to organizer formation
expression Wnt signaling
GSK-3 M Protein kinase Protein depleted on Suppression of dorsalizing signals
dorsal side
axin M Binds β-catenin RNA throughout zygote; Suppression of dorsalizing signals
more protein dorsally
than ventrally
Derrière Z TGF-β family Vegetal hemisphere and Posterior-mesoderm induction
marginal region
Xnr-1,2,4,5,6 Z TGF-β family Vegetal hemisphere and Mesoderm induction
marginal region; higher
dorsally
FGF (several types) Z Secreted signals Vegetal hemisphere and Posterior-mesoderm development
marginal region

Mesoderm patterning
Brachyury Z Transcription factor Throughout prospective Posterior-mesoderm formation
mesoderm
Xwnt-8 Z Wnt family Ventral and lateral regions of Mesoderm ventralization
prospective mesoderm
BMP-4 Z TGF-β family Throughout late blastula; then Mesoderm ventralization
excluded from organizer
Activin Z TGF-β family Throughout late blastula/ Mesoderm induction and patterning
early gastrula

Organizer function
siamois Z Transcription factor Nieuwkoop center Induction of organizer
noggin Z Secreted signal Spemann organizer Mesoderm dorsalization by
antagonizing BMP-4
chordin Z Secreted signal Spemann organizer Mesoderm dorsalization by
antagonizing BMP-4
frizbee Z Secreted signal Spemann organizer Mesoderm dorsalization by
antagonizing Xwnt-8
cerberus Z Secreted signal Spemann organizer Promotes head development by inhibiting
Wnt, Nodal-related, and BMP signaling
goosecoid Z Transcription factor Spemann organizer Organizer function
Xlim-1 Z Transcription factor Spemann organizer Gastrulation and head formation
Xnot Z Transcription factor Spemann organizer Notochord specification
168 Chapter 4 Vertebrate development II: axes and germ layers

embryo’s own genes; where it is located in the fertilized egg; and If you had to choose between transcription factors, signals of the
what germ layers it is involved in specifying, whether alone or in TGF-b family, or signals of the Wnt family for further study into the
combination with other proteins. mechanisms that specify and pattern the mesoderm in vertebrates,
which would you choose? Justify your answer.
6. What is the posterior marginal zone of the chick embryo? What
is its significance to chick development? Describe the external forces
Multiple choice (factual recall questions)
that lead to its formation and the signaling molecules associated
with it. NB There is only one right answer to each question.

7. The establishment of left–right asymmetry is associated with 1. The nervous system, heart, and liver are derived from _____,
higher levels of the TGF-b family signaling molecule Nodal on the _______, and ______, respectively.
left side of the embryo, relative to the right. What are the influences a) all are derived from mesoderm
that led to this asymmetry of Nodal concentration? Include positive b) ectoderm, mesoderm, and endoderm
feedback, Notch signaling, Sonic hedgehog signaling, and the Lefty
protein in your answer. c) endoderm, mesoderm, and ectoderm
d) mesoderm, ectoderm, and endoderm
8. A simple view of gastrulation in Xenopus would be that the blas-
tula is like a soft ball, and the involution of cells during gastrulation 2. The positioning of the dorso-ventral axis in Xenopus is deter-
is analogous to poking your finger into the ball at the blastopore, mined by _____, while the positioning of the antero-posterior axis in
until you can touch the ventral surface. Examine the fate map of the chicks is determined by _____.
Xenopus embryo (Fig. 4.19); what is wrong with this simple view? a) gravity; point of sperm entry
Provide a more accurate description of the process in Xenopus (refer
to Fig. 3.6 for help). Where would the mesoderm have to be located b) maternal factors in both cases
in the blastula for the simple invagination method to work? c) position in the egg chamber; gravity

9. Specification, determination, and differentiation describe dif- d) the point of sperm entry; gravity
ferent states of development of a cell. Define each of these terms, 3. The ‘organizer’ in Xenopus is responsible for
and describe what experiments are used to distinguish these three
a) inducing a mesodermal fate in nearby cells
­different states of development (refer to Sections 1.8, 1.12, and 4.11
in constructing your answer)? Speculate on what sort of changes b) initiating involution and gastrulation
might have occurred at the molecular level in a cell between specifi- c) specifying the dorsal region of the embryo
cation and determination.
d) all of these events
10. Early amphibian development is driven largely by maternally
4. A key factor that specifies the organizer in Xenopus embryos is:
supplied factors until the mid-blastula transition. What is the mid-
blastula transition? What is the model that explains how it occurs? a) GSK-3b
What experimental evidence supports this model? b) Xfz7
11. ‘Dorsal vegetal tissue containing the Nieuwkoop center induces c) XWnt-11
notochord and muscle from animal cap cells, whereas ventral veg- d) b-catenin
etal tissue induces mainly blood-forming tissue and little muscle.’
(Section 4.14). What is the significance of these observations? What 5. How would you produce a chimeric mouse?
signaling events are required to explain these observations. a) Divide a morula into two separate clusters of blastomeres and let
each cluster form an embryo.
12. An early event in axis formation is the accumulation of the
transcription factor b-catenin in the nuclei of cells on one side of b) Introduce cells from the inner cell mass of one blastocyst into
the Xenopus embryo. Which side of the embryo does this specify? the inner cell mass of another blastocyst of the same genetic
Describe briefly how the actions of b-catenin lead to specification of constitution.
the Spemann organizer. c) Divide a fertilized mouse egg into two halves and let each
13. What is the developmental function of the TGF-b family mem- develop separately.
ber BMP-4 in the blastula? Since BMP-4 is initially present through- d) Introduce cells from the inner cell mass of one blastocyst into
out the blastula, how are its effects restricted to particular regions the inner cell mass of another blastocyst of a different genetic
(a) of the ectoderm, and (b) of the dorsal mesoderm? constitution.

14. The TGF-b signaling pathway is an example of the remarkable 6. If cells from the animal pole of a frog blastula (animal cap cells)
conservation of signaling pathways during evolution. Compare the are placed in direct contact with cells from the vegetal region, what
activities of BMP-4 in Xenopus and Decapentaplegic in Drosophila. is the result?
Are the antagonists of these signaling molecules in the two systems a) An embryo forms with only ectodermal and endodermal
also homologs? ­derivatives.
15. In this chapter, we have seen several examples of the importance b) Animal cap cells are induced by the vegetal cells to form
of transcriptional control and cell–cell signaling in ­development. ­mesodermal derivatives.
 The origin and specification of the germ layers 169

c) The embryo regulates and forms a normal embryo. Kodjabachian, L., Dawid, I.B., Toyama, R.: Gastrulation in zebrafish:
d) Vegetal hemisphere cells are induced by the animal cap cells to what mutants teach us. Dev. Biol. 1999, 126: 5309–5317.
form mesodermal derivatives. Logan, C.Y., Nusse, R.: The Wnt signalling pathway in develop-
ment and disease. Annu. Rev. Cell Dev. Biol. 2004, 20: 781–801.
7. Maternal factors are Pelegri, F.: Maternal factors in zebrafish development. Dev. Dyn.
a) the energy-rich yolk present in the eggs of many species 2003, 228: 535–554.
b) the genes contributed to the embryo by the maternal haploid Sokol, S.Y.: Wnt signaling and dorso-ventral axis specification in
genome vertebrates. Curr. Opin. Genet. Dev. 1999, 9: 405–410.
Tao, Q., Yokota, C., Puck, H., Kofron, M., Birsoy, B., Yan, D.,
c) hormonal influences encountered by the mammalian embryo in
Asashima,­ M., Wylie, C.C., Lin, X., Heasman, J.: Maternal wnt
the uterus
11 activates the canonical wnt signaling pathway required for
d) developmentally active mRNAs and proteins packaged into the axis formation in Xenopus embryos. Cell 2005, 120: 857–871.
egg by the mother

8. The activity of the protein Noggin is 4.3 Signaling centers develop on the dorsal side of Xenopus
a) to antagonize BMP-4, and block ventralization of the mesoderm and zebrafish blastulas
b) to signal through the Noggin pathway and establish the orga- Smith, J.: T-box genes: what they do and how they do it. Trends
nizer Genet. 1999, 15: 154–158.
Vonica, A., Gumbiner, B.M.: The Xenopus Nieuwkoop center and
c) to specify a mesodermal fate
Spemann-Mangold organizer share molecular components and
d) to turn on genes that determine the anterior-most cells in the a requirement for maternal Wnt activity. Dev. Biol. 2007, 312:
embryo, which will form the head 90–102.
9. Vg-1, Xnrs, BMPs and activin are all members of which family of
signaling molecules? 4.4 The antero-posterior and dorso-ventral axes of the chick
a) FGF blastoderm are related to the primitive streak
b) Hedgehog Bertocchini, F., Skromne, I., Wolpert, L., Stern, C.D.: Determina-
tion of embryonic polarity in a regulative system: evidence for
c) TGF-b
endogenous inhibitors acting sequentially during primitive
d) Wnt streak formation in the chick embryo. Development 2004, 131:
3381–3390.
Multiple choice answer key Khaner, O., Eyal-Giladi, H.: The chick’s marginal zone and primi-
1: b, 2: d, 3: b, 4: d, 5: d, 6: b, 7: d, 8: a, 9: c, 10: c. tive streak formation. I. Coordinative effect of induction and
inhibition. Dev. Biol. 1989, 134: 206–214.
Kochav, S., Eyal-Giladi, H.: Bilateral symmetry in chick embryo
n Section further reading determination by gravity. Science 1971, 171: 1027–1029.
Seleiro, E.A.P., Connolly, D.J., Cooke, J.: Early developmental
4.1 The animal–vegetal axis is maternally determined in expression and experimental axis determination by the chicken
Xenopus and zebrafish Vg-1 gene. Curr. Biol. 1996, 11: 1476–1486.
Stern, C.D.: Cleavage and gastrulation in avian embryos (ver-
Heasman, J.: Patterning the early Xenopus embryo. Development
sion 3.0). Encyclopedia of Life Sciences 2009 http://www.els.net/
2006, 133: 1205–1217.
(13 May 2010).
Schier, A.F., Talbot, W.S.: Molecular genetics of axis formation in
zebrafish. Annu. Rev. Genet. 2005, 39: 561–613.
Weaver, C., Kimelman, D.: Move it or lose it: axis specification in 4.5 The definitive antero-posterior and dorso-ventral
Xenopus. Development 2004, 131: 3491–3499. axes of the mouse embryo are not recognizable early in
development
4.2 Localized stabilization of the transcriptional regulator Arnold, S.J., Robertson, E.J.: Making a commitment: cell lineage
b-catenin specifies the future dorsal side and the location of allocation and axis patterning in the early mouse embryo.
the main embryonic organizer in Xenopus and zebrafish ­Nature Mol. Cell Biol. Rev. 2009, 10: 91–103.
Dosch, R., Wagner, D.S., Mintzer. K.A., Runke, G., Wiemelt, A.P., Beddington, R.S.P., Robertson, E.J.: Axis development and early
Mullins, M.C.: Maternal control of vertebrate development asymmetry in mammals. Cell 1999, 96: 195–209.
before the midblastula transition: mutants from the zebrafish Bischoff, M., Parfitt, D.E., Zernicka-Goetz, M.: Formation of the
I. Dev Cell 2004, 6: 771–780. embryonic-abembryonic axis of the mouse blastocyst: rela-
Gerhart, J., Danilchik, M., Doniach, T., Roberts, S., Browning, B., tionships between orientation of early cleavage divisions and
Stewart, R.: Cortical rotation of the Xenopus egg: consequences pattern of symmetric/asymmetric divisions. Development 2008,
for the antero-posterior pattern of embryonic dorsal develop- 135: 953–962.
ment. Development (Suppl.) 1989, 37–51. Deb, K., Sivaguru, M., Yul Yong, H., Roberts, M.: Cdx2 gene
Heasman, J.: Maternal determinants of embryonic cell fate. Semin. expression and trophectoderm lineage specification in mouse
Cell Dev. Biol. 2006, 17: 93–98. ­embryos. Science 2006, 311: 992–996.
170 Chapter 4 Vertebrate development II: axes and germ layers

Hillman, N., Sherman, M.I., Graham, C.: The effect of spatial Raya, A., Izpisua Belmonte, J.C.: Unveiling the establishment of
arrangement on cell determination during mouse develop- left-right asymmetry in the chick embryo. Mech. Dev. 2004, 121:
ment. J. Embryol. Exp. Morph. 1972, 28: 263–278. 1043–1054.
Perea-Gomez, A., Camus, A., Moreau, A., Grieve, K., Moneron, G., Raya, A., Izpisua Belmonte, J.C.: Insights into the establishment of
Dubois, A., Cibert, C., Collignon, J.: Initiation of gastrulation left-right asymmetries in vertebrates. Birth Defects Res C Embryo
in the mouse embryo is preceded by an apparent shift in the Today 2008, 84: 81–94.
orientation of the antero-posterior axis. Curr. Biol. 2004, 14: Schlueter, J., Brand, T.: Left-right axis development: examples of
197–207. similar and divergent strategies to generate asymmetric mor-
Rivera-Perez, J.A.: Axial specification in mice: ten years of phogenesis in chick and mouse embryos. Cytogenet. Genome
advances and controversies. J. Cell Physiol. 2007, 213: 654–660. Res. 2007, 117: 256–267.
Robertson, E.J., Norris, D.P., Brennan, J., Bikoff, E.K.: Control of Shen, M.M.: Nodal signaling: developmental roles and regulation.
early anterior-posterior patterning in the mouse embryo by Development 2007, 134: 1023–1034.
TGF-beta signalling. Phil. Trans. R Soc. Lond. B Biol. Sci. 2003,
358: 1351–1357. 4.8 A fate map of the amphibian blastula is constructed by
Rodriguez, T.A., Srinivas, S., Clements, M.P., Smith, J.C., Bedding- following the fate of labeled cells
ton, R.S.: Induction and migration of the anterior visceral endo- Dale, L., Slack, J.M.W.: Fate map for the 32 cell stage of Xenopus
derm is regulated by the extra-embryonic ectoderm. Develop- laevis. Development 1987, 99: 527–551.
ment 2005, 132: 2513–2520. Gerhart J.: Changing the axis changes the perspective. Dev. Dyn.
Rossant, J., Tam, P.P.L.: Blastocyst lineage formation, early embry- 2002, 225: 380–383.
onic asymmetries and axis patterning in the mouse. Develop- Lane, M.C., Smith, W.C.: The origins of primitive blood in
ment 2009, 136: 701–713. ­Xenopus: implications for axial patterning. Development 1999,
Srinivas, S., Rodriguez, T., Clements, M., Smith, J.C., Beddington, 126: 423–434.
R.S.P.: Active cell migration drives the unilateral movements of
the anterior visceral endoderm. Development 2004, 131: 1157– 4.9 The fate maps of vertebrates are variations on
1164. a basic plan
Takaoka, K., Yamamoto, M., Hamada, H.: Origin of body axes in Beddington, R.S.P., Morgenstern, J., Land, H., Hogan, A.: An in situ
the mouse embryo. Curr. Opin. Genet. Dev. 2007, 17: 344–350. transgenic enzyme marker for the midgestation mouse embryo
Zernicka-Goetz, M.: Developmental cell biology: cleavage pattern and the visualization of inner cell mass clones during early
and emerging asymmetry of the mouse embryo. Nat. Rev. Mol. organogenesis. Development 1989, 106: 37–46.
Cell Biol. 2005, 6: 919–928. Gardner, R.L., Rossant, J.: Investigation of the fate of 4–5 day post-
coitum mouse inner cell mass cells by blastocyst injection.
4.7 The bilateral symmetry of the early embryo is broken to
J. Embryol. Exp. Morph. 1979, 52: 141–152.
produce left–right asymmetry of internal organs
Kimmel, C.B., Warga, R.M., Schilling, T.F.: Origin and organization
Blum, M., Beyer, T., Weber, T., Vivk, P., Andre, P., Bitzer, E., of the zebrafish fate map. Development 1990, 108: 581–594.
­Schweickert, A.: Xenopus, an ideal model system to study verte- Lawson, K.A., Meneses, J.J., Pedersen, R.A.: Clonal analysis of
brate left-right asymmetry. Dev. Dyn. 2009, 238: 1215–1225. epiblast fate during germ layer formation in the mouse embryo.
Blum, M., Weber, T., Beyer, T., Vick, P.: Evolution of leftward flow. Development 1991, 113: 891–911.
Semin. Cell Dev. Biol. 2008, 20: 464–471. Smith, J.L., Gesteland, G.M., Schoenwolf, G.C.: Prospective fate
Brennan, J., Norris, D.P., Robertson, E.J.: Nodal activity in the node map of the mouse primitive streak at 7.5 days of gestation. Dev.
governs left-right asymmetry. Genes Dev. 2002, 16: 2339–2344. Dyn. 1994, 201: 279–289.
Brown, N.A., Wolpert, L.: The development of handedness in left/ Stern, C.D.: The marginal zone and its contribution to the hypo-
right asymmetry. Development 1990, 109: 1–9. blast and primitive streak of the chick embryo. Development
Gros, J., Feistel. K., Viebahn, C., Blum, M., Tabin, C.J.: Cell move- 1990, 109: 667–682.
ments at Hensen’s node establish left/right asymmetric gene Stern, C.D., Canning, D.R.: Origin of cells giving rise to mesoderm
expression in the chick. Science 2009, 324: 941–944. and endoderm in chick embryo. Nature 1990, 343: 273–275.
Levin, M., Palmer, A.P.: Left-right patterning from the inside out:
widespread evidence for intracellular control. BioEssays 2007, 4.10 Cells of early vertebrate embryos do not yet have their
29: 271–287. fates determined and regulation is possible
McGrath, J., Somlo, S., Makova, S., Tian, X., Brueckner, M.: Two Lewis, N.E., Rossant, J.: Mechanism of size regulation in mouse
populations of node monocilia initiate left-right asymmetry in embryo aggregates. J. Embryol. Exp. Morph. 1982, 72: 169–181.
the mouse. Cell 2003, 114: 61–73. Snape, A., Wylie, C.C., Smith, J.C., Heasman, J.: Changes in states
Rana, A.A., Barbera, J.P., Rodriguez, T.A., Lynch, D., Hirst, E., of commitment of single animal pole blastomeres of Xenopus
Smith, J.C., Beddington, R.S.P.: Targeted deletion of the novel laevis. Dev. Biol. 1987, 119: 503–510.
cytoplasmic dynein mD2LIC disrupts the embryonic organiser, Tam, P.P., Rossant, J.: Mouse embryonic chimeras: tools for study-
formation of body axes and specification of ventral cell fates. ing mammalian development. Development 2003, 130: 6155–
Development 2004, 131: 4999–5007. 6163.
 The origin and specification of the germ layers 171

Wylie, C.C., Snape, A., Heasman, J., Smith, J.C.: Vegetal pole cells Birsoy, B., Kofron, M., Schaible, K., Wylie, C., Heasman, J.: Vg1
and commitment to form endoderm in Xenopus laevis. Dev. is an essential signaling molecule in Xenopus development.
Biol. 1987, 119: 496–502. Development 2006, 133: 15–20.
Massagué, J.: How cells read TGF-b signals. Nat Rev. Mol. Cell Biol.
4.11 In Xenopus the endoderm and ectoderm are specified 2000, 1: 169–178.
by maternal factors, but the mesoderm is induced from Schier, A.F.: Nodal signaling in vertebrate development. Annu.
ectoderm by signals from the vegetal region Rev. Cell. Dev. Biol. 2003, 19: 589–621.
Dale, L.: Vertebrate development: multiple phases to endoderm
formation. Curr. Biol. 1999, 9: R812–R815. 4.16 The zygotic expression of mesoderm-inducing and
Dupont, S., Zacchigna, L., Cordenonsi, M., Soligo, S., Adorno, M., patterning signals in Xenopus is activated by the combined
Rugge, M., Piccolo, S.: Germ-layer specification and control of actions of maternal VegT and Wnt signaling
cell growth by Ectodermin, a Smad4 ubiquitin ligase. Cell 2005, De Robertis, E.M., Larrain, J., Oelgeschläger, M., Wessely, O.: The
121: 87–99. establishment of Spemann’s organizer and patterning of the
Mir, A., Kofron, M., Zorn, A.M., Bajzer, M., Haque, M., Heasman, J., vertebrate embryo. Nat Rev. Genet. 2000, 1: 171–181.
Wylie, C.C.: FoxI1e activates ectoderm formation and controls Fletcher, R.B., Harland, R.M.: The role of FGF signaling in the
cell position in the Xenopus blastula. Development 2007, 134: establishment and maintenance of mesodermal gene expres-
779–788. sion in Xenopus. Dev. Dyn. 2008, 237: 1243–1254.
White, J.A., Heasman, J.: Maternal control of pattern formation in Harvey, S.A., Smith, J.C.: Visualisation and quantification of mor-
Xenopus laevis. J. Exp. Zool. B 2008, 310: 73–84. phogen gradient formation in the zebrafish. PLoS Biol. 2009,
Xanthos, J.B., Kofron, M., Wylie, C., Heasman, J.: Maternal VegT 7: e101.
is the initiator of a molecular network specifying endoderm in Kofron, M., Demel, T., Xanthos, J., Lohr, J., Sun, B., Sive, H., Osada,
Xenopus laevis. Development 2001, 128: 167–180. S-I., Wright, C., Wylie, C., Heasman, J.: Mesoderm induction
in Xenopus is a zygotic event regulated by maternal VegT via
4.12 Mesoderm induction occurs during a limited period in TGFb growth factors. Development 1999, 126: 5759–5770.
the blastula stage Niehrs, C.: Regionally specific induction by the Spemann–­
Gurdon, J.B., Lemaire, P., Kato, K.: Community effects and related Mangold organizer. Nat. Rev. Genet. 2004, 5: 425–434.
phenomena in development. Cell 1993, 75: 831–834.
4.17 Signals from the organizer pattern the mesoderm
4.13 Zygotic gene expression is turned on in Xenopus at the dorso-ventrally by antagonizing the effects of ventral signals
mid-blastula transition De Robertis, E.M.: Spemann’s organizer and self-regulation in
Davidson, E.: Gene Activity In Early Development. New York: Aca- amphibian embryos. Nature Mol. Cell Biol. Rev. 2006, 7: 296–
demic Press, 1986. 302.
O’Boyle, S., Bree, R.T., McLoughlin, S., Grealy, M., Byrnes, L.: Iden- Gonzalez, E.M., Fekany-Lee, K., Carmany-Rampey, A., Erter, C.,
tification of zygotic genes expressed at the midblastula transi- Topczewski, J., Wright, C.V., Solnica-Krezel, L.: Head and trunk
tion in zebrafish. Biochem. Biophys. Res. Commun. 2007, 358: in zebrafish arise via coinhibition of BMP signaling by bozo-
462–468. zok and chordino. Genes Dev. 2000, 14: 3087–3092.
Yasuda, G.K., Schübiger, G.: Temporal regulation in the early Oelgeschläger, M., Larrain, J., Geissert, D., De Robertis, E.M.: The
embryo: is MBT too good to be true? Trends Genet. 1992, 8: evolutionarily conserved BMP-binding protein Twisted gastru-
124–127. lation promotes BMP signalling. Nature 2000, 405: 757–763.
Piccolo, S., Sasai, Y., Lu, B., De Robertis, E.M.: Dorsoventral pat-
4.14 Mesoderm-inducing and patterning signals in Xenopus terning in Xenopus: inhibition of ventral signals by direct bind-
are produced by the vegetal region, the organizer, and the ing of chordin to BMP-4. Cell 1996, 86: 589–598.
ventral mesoderm Piepenburg, O., Grimmer, D., Williams, P.H., Smith, J.C.: Activin
Agius, E., Oelgeschläger, M., Wessely, O., Kemp, C., De Robertis, redux: specification of mesodermal pattern in Xenopus by
E.M.: Endodermal Nodal-related signals and mesoderm induc- graded concentrations of endogenous activin B. Development
tion in Xenopus. Development 2000, 127: 1173–1183. 2004, 131: 4977–4986.
Heasman, J.: Patterning the early Xenopus embryo. Development Schier, A.F.: Axis formation and patterning in zebrafish. Curr.
2006, 133: 1205–1217. Opin. Genet. Dev. 2001, 11: 393–404.
Kimelman, D.: Mesoderm induction: from caps to chips. Nat. Rev. Zimmerman, L.B., De Jesús-Escobar, J.M., Harland, R.M.: The Spe-
Genet. 2006, 7: 360–372. mann organizer signal noggin binds and inactivates bone mor-
phogenetic protein 4. Cell 1996, 86: 599–606.
4.15 Members of the TGF-b family have been identified as
mesoderm inducers 4.18 Threshold responses to gradients of signaling proteins
Amaya, E., Musci, T.J., Kirschner, M.W.: Expression of a domi- are likely to pattern the mesoderm
nant negative mutant of the FGF receptor disrupts mesoderm Chen, Y., Schier, A.F.: The zebrafish Nodal signal Squint functions
­formation in Xenopus embryos. Cell 1991, 66: 257–270. as a morphogen. Nature 2001, 411: 607–609.
172 Chapter 4 Vertebrate development II: axes and germ layers

Green, J.B.A., New, H.V., Smith, J.C.: Responses of embryonic Chu, G.C., Dunn, N.R., Anderson, D.C., Oxburgh, L., Robertson,
­Xenopus cells to activin and FGF are separated by multiple E.J.: Differential requirements for Smad4 in TGFbeta-­dependent
dose thresholds and correspond to distinct axes of the meso- patterning of the early mouse embryo. Development 2004, 131:
derm. Cell 1992, 71: 731–739. 3501–3512.
Gurdon, J.B., Standley, H., Dyson, S., Butler, K., Langon, T., Ryan, K., Dunn, N.R., Vincent, S.D., Oxburgh, L., Robertson, E.J., Bikoff,
Stennard, F., Shimizu, K., Zorn, A.: Single cells can sense their E.K.: Combinatorial activities of Smad2 and Smad3 regulate
position in a morphogen gradient. Development 1999, 126: mesoderm formation and patterning in the mouse embryo.
5309–5317. Development 2004, 131: 1717–1728.
Jones, C.M., Armes, N., Smith, J.C.: Signaling by TGF-b family Lu, C.C., Robertson, E.J.: Multiple roles for Nodal in the epiblast
members: short-range effects of Xnr-2 and BMP-4 contrast with of the mouse embryo in the establishment of anterior-posterior
the long-range effects of activin. Curr. Biol. 1996, 6: 1468–1475. patterning. Dev. Biol. 2004, 273: 149–159.
Papin, C., Smith, J.C.: Gradual refinement of activin-induced Vincent, S.D., Dunn, N.R., Hayashi, S., Norris, D.P., Robertson, E.J.:
thresh­olds requires protein synthesis. Dev. Biol. 2000, 217: Cell fate decisions within the mouse organizer are governed by
166–172. graded Nodal signals. Genes Dev. 2003, 17: 1646–1662.
Schulte-Merker, S., Smith, J.C.: Mesoderm formation in response to Zakin, L., Reversade, B., Kuroda, H., Lyons, K.M., De Robertis, E.M.:
Brachyury requires FGF signalling. Curr. Biol. 1995, 5: 62–67. Sirenomelia in Bmp7 and Tsg compound mutant mice: require-
ment for Bmp signaling in the development of ventral posterior
Box 4E A zebrafish gene regulatory network mesoderm. Development 2005, 132: 2489–2499.
Morley, R.H., Lachanib, K., Keefe, D., Gilchrist, M.J., Flicek, P.,
Smith, J.C., Wardle, F.C.: A gene regulatory network directed
by zebrafish No tail accounts for its roles in mesoderm forma-
tion. Proc. Natl Acad. Sci. USA 2009, 106: 3829–3834.

4.19 Mesoderm induction and patterning in the chick and

mouse occurs during primitive-streak formation
Chapman, S.C., Matsumoto, K., Cai, Q., Schoenwolf, G.C.: Specifi-
cation of germ layer identity in the chick gastrula. BMC Dev.
Biol. 2007, 7: 91.
 5
Vertebrate development III:
patterning the early nervous
system and the somites
nn The role of the organizer nn Somite formation and nn Initial antero-posterior
and neural induction ­patterning patterning of the brain

During and after gastrulation, the vertebrate embryo becomes further patterned
along the antero-posterior and dorso-ventral axes. This patterning is carried out by
a combination of signals from various regions of the embryo and the interpretation
of positional identity by cells. A key developmental event that accompanies gastrula-
tion is the induction of the nervous system from the dorsal ectoderm by a complex
cascade of signals from adjacent tissues, including the organizer. At around the same
time, the main antero-posterior organization of the rest of the body starts to become
evident, with the segmentation of mesoderm along the antero-posterior axis into
blocks of tissue called somites—which give rise to the spinal column, ribs, and the
skeletal muscles of the trunk. The expression of genes involved in encoding position
along the antero-posterior axis is central to this patterning, as is the mechanism by
which their spatial patterns of expression are initially specified.

In Chapter 4, we saw how the body axes are set up and how the three germ layers are
initially specified in vertebrate embryos. Although amphibian, fish, chick, and mouse
embryos share some features at these stages, there are some significant differences.
As we approach the phylotypic stage—the embryonic stage common to all vertebrates
(see Fig. 3.2)—the similarities between vertebrate embryos become greater.
During gastrulation, the germ layers—mesoderm, endoderm, and ectoderm—move
to the positions in which they will develop into the structures of the larval or adult
body. The antero-posterior body axis of the vertebrate embryo emerges clearly, with
the head at one end and the future tail at the other (Fig. 5.1). In this chapter, we focus
mainly on the patterning of ectoderm that will develop into the nervous system, and on
the formation and patterning of the mesoderm-derived somites, blocks of tissue that
give rise to the skeleton and skeletal muscles of the trunk, and to some of the dermis of
the skin. The cell movements of gastrulation and the action of the organizer region are
crucial to establishing the vertebrate body plan and will be discussed in this chapter in
relation to their role in patterning. A detailed discussion of the mechanisms underlying
the movements of cells and tissues during gastrulation is deferred to Chapter 8.
174 Chapter 5 Vertebrate development III: patterning the early nervous system and the somites

Fig. 5.1 The final body plan of the

Fate map of amphibian blastula Schematic sagittal section of embryo
embryo emerges during gastrulation and just before gastrulation after gastrulation and neurulation
neurulation. Left, fate map of a late Xenopus
Animal somitic mesoderm notochord
blastula. Right, a schematic of a sagittal Dorsal
prechordal plate neural tube
section through a tailbud stage Xenopus
forebrain
embryo after gastrulation and neurulation.
The mesoderm and endoderm have moved Ventral Dorsal Anterior Posterior
inside. The mesoderm gives rise to the notochord gut
prechordal plate (brown), notochord (red), Ventral
somites endoderm ectoderm
somites (orange), and lateral mesoderm (not Vegetal
shown in this view). The endoderm (yellow)
moves inside to line the gut. The neural tube
(dark blue) forms from dorsal ectoderm and In Xenopus, the dorsal mesoderm (the organizer region) in the late blastula is
the ectoderm that will form the epidermis i­nternalized at gastrulation to form the prechordal plate mesoderm, which lies anterior
(light blue) covers the whole embryo. The
to the notochord and gives rise to the ventral head mesoderm, and the rigid rod-like
antero-posterior axis has emerged, with the
­notochord that runs along the dorsal midline. The notochord is flanked immediately
head at the anterior end.
on each side by more ventral mesoderm, which by the end of gastrulation is beginning
to segment to form blocks of somites, starting at the anterior end (see Fig. 5.1). The
vertebrate notochord is a transient structure, and its cells eventually become incorpo-
rated into the vertebrae and intervening discs of the spinal column. As gastrulation
proceeds, neurulation, the first stage of nervous system formation, begins. The ecto-
derm overlying the notochord rolls upwards on both sides of the dorsal midline to form
the neural folds, which meet over the midline to form a tubular structure, the neural
tube (see Figs. 3.7). The internal structure of the Xenopus embryo just after the end
of neurulation is illustrated in Fig. 5.2. The main structures that can be recognized at
this stage are the neural tube, the notochord, the somites, the lateral plate mesoderm,
which lies ventral to the somites on either side, and the endoderm lining the gut.
Both the mesodermal structures along the antero-posterior axis of the trunk and the
ectodermally derived nervous system have a distinct antero-posterior organization. An
initial antero-posterior patterning of the germ layers occurs during mesoderm induc-
tion, as mesoderm becomes differentiated from endoderm and ectoderm, while some
important aspects of later antero-posterior patterning in both the mesoderm and the
nervous system are controlled by the Hox genes, which are the vertebrate equivalent of
the Hox genes responsible for antero-posterior patterning in Drosophila (see Chapter 2).

Transverse section of a stage 22 Schematic interpretation of

Xenopus embryo photograph opposite

Dorsal central
canal neural tube
dermomyotome
notochord
sclerotome somite

Fig. 5.2 A cross-section through a stage

22 Xenopus embryo just after gastrulation
and neurulation are completed. The
germ layers are now all in place for future
archenteron
development and organogenesis. The (future gut) lumen
most dorsal parts of the somites have
already begun to differentiate into the
dermomyotome, which will give rise to the
yolky endoderm
trunk and limb muscles and the dermis,
as described later in the chapter. Scale
bar = 0.2 mm.
lateral plate
Photograph from Hausen, P. and Riebesell, ectoderm mesoderm
Ventral
M.: 1991.
 The role of the organizer and neural induction 175

In the first part of this chapter we consider the function of the embryonic organizer
in establishing the basic antero-posterior organization of vertebrate embryos, focusing
on its role in the induction of the nervous system. In the second part of the chapter
we describe the development of the somites and discuss the role of the Hox genes in
the antero-posterior patterning of the somitic mesoderm, as shown by their effects on
the somite-derived vertebrae of the spinal column. Finally, we will look at the antero-
posterior patterning of the hindbrain, in which Hox genes are again involved.

The role of the organizer and neural induction

We shall first consider the role of the organizer region, which is crucially important
both in neural induction and in organizing the antero-posterior axis. The Spemann
organizer of amphibians, the shield in zebrafish, Hensen’s node in the chick, and the
equivalent node region in the mouse all have a similar global organizing function
in vertebrate development. They can induce a complete body axis if transplanted to
another embryo at an appropriate stage (see Chapter 4), and so are able to organize
and coordinate both dorso-ventral and antero-posterior aspects of the body plan, as
well as induce neural tissue from ectoderm. In the mouse, the anterior visceral endo-
derm is required in addition to the node for the induction of structures anterior to the
end of the notochord, such as the head and forebrain (see Section 4.6).
During gastrulation, the ectoderm in the dorsal region of the embryo becomes spec-
ified as neuroectoderm, the neural plate (see Fig. 3.7). During the subsequent stage
of neurulation, the neural plate forms the neural tube, which eventually differentiates
into the central nervous system—the brain and the spinal cord—and the peripheral
nervous system. The neural tube throws off neural crest cells, which migrate through-
out the body to give rise to the sensory and autonomic parts of the peripheral nervous
system. It also gives rise to some non-neural structures, most notably some of the
tissues of the face and jaws, such as bone and cartilage, tissues that in other parts
of the body are derived from mesoderm. The contribution of the neural crest cells to
these tissues in the head is a rare example of cells from one germ layer giving rise to
tissues characteristic of another. That particular example will be discussed later in this
chapter, while the enormous differentiation potential of neural crest cells generally is
discussed in Chapter 10, and details of their migration routes in the rest of the body
are considered in Chapter 8.
The nervous system must develop in the correct relationship with other body struc-
tures, particularly the mesodermally derived structures that give rise to the skeleto-
muscular system. Thus, patterning of the nervous system must be linked to that of the
mesoderm, and this is coordinated through the organizer, which is involved in both.
In this part of the chapter, we consider the induction of the neuroectoderm and for-
mation of the neural tube, along with the specification of the prospective neural crest
cells. The antero-posterior patterning of the brain and neural crest will be discussed
in the last part of the chapter.
The function of the organizer has been best studied in amphibians, and we have
­already described its role in the dorso-ventral patterning of the mesoderm in Xenopus
in Chapter 4. We now discuss its profound effects on the antero-posterior axis. The
function of the equivalent region in the chick, Hensen’s node, has also been exten-
sively studied because chick embryos are particularly amenable to surgical proce-
dures, such as transplantation (see Section 3.6).

5.1 The inductive capacity of the organizer changes during gastrulation

In amphibians, the action of the Spemann organizer is dramatically demonstrated in

what is classically known as primary embryonic induction. The organizer is located
176 Chapter 5 Vertebrate development III: patterning the early nervous system and the somites

Organizer transplant from early gastrula Organizer transplant from late gastrula

Fig. 5.3 The inductive properties of the

organizer change during gastrulation.
A graft of the organizer region, from the D D
dorsal lip of the blastopore of an early frog
gastrula to the ventral side of another early
gastrula, results in the development of an A P A P
additional anterior axis at the site of the
graft (left panels). A graft from the dorsal lip
D
region of a late gastrula to an early gastrula
only induces formation of tail structures additional tail
(right panels). A = anterior; P = posterior;
D = dorsal; V = ventral.

in the dorsal lip of the blastopore and if the blastopore dorsal lip from a very early gas-
trula is grafted to the ventral side of the marginal zone of another early gastrula, it can
induce a complete second embryo with a well defined head, a central nervous system,
a trunk region, and a tail (Fig. 5.3, left panels, and see Fig. 4.32). This organizer
function is often called the ‘head organizer.’ Various other treatments, such as graft-
ing dorsal vegetal blastomeres containing the Nieuwkoop center to the ventral side,
produce a similar result (see Fig. 4.6), but what all these treatments have in common
is that, directly or indirectly, they result in the formation of a new Spemann-organizer
region. Classic experiments also showed that a graft of the blastopore dorsal lip from
a mid-gastrula to an early gastrula induces a trunk and tail but no head; this organizer
function is often called the ‘trunk organizer’. A grafted dorsal lip from a late gastrula
induces only a tail (Fig. 5.3, right panels). These results were interpreted to mean that
as gastrulation proceeds, the antero-posterior axis of the embryo becomes specified,
and so the cells that make up the organizer at later stages of gastrulation induce only
posterior structures. We now also know that both the quantity and the nature of the
inducing signals produced by the organizer change as gastrulation proceeds.
The cells present at the dorsal lip of the blastopore in a Xenopus early gastrula them-
selves give rise during gastrulation to anterior endoderm, prechordal plate ­mesoderm,
and the notochord (Fig. 5.4). In addition to providing cells for these axial structures,
the organizer region has patterning and inductive properties: it helps to pattern the
adjacent more ventral mesoderm, as we saw in Chapter 4, and it induces the neural
plate in the adjacent dorsal ectoderm and gives it an initial antero-­posterior pattern.
The organizer of the early gastrula is therefore a complex signaling center with differ-
ent parts expressing different genes, having different inductive capacities, and ­giving
rise to different structures. Because of these complex properties, ­understanding how
the organizer region controls the overall pattern of the antero-­posterior axis is not
straightforward. As gastrulation proceeds and cells move inwards, the cellular com-
position of the dorsal lip changes, and at one level this explains the different ­inductive
properties displayed by the organizer over time in the experiments ­described above (see
Fig. 5.3). For example, in the early Xenopus gastrula, the vegetal portion of the orga-
nizer that is fated to produce prechordal plate mesoderm (shown in brown in Fig. 5.4)
expresses proteins, such as the transcription factor XOtx2, that are ­characteristic of
 The role of the organizer and neural induction 177

Xenopus embryo before gastrulation Xenopus embryo after gastrulation

Fig. 5.4 Different parts of the Xenopus
anterior organizer region give rise to different
Animal neural plate neural plate
ectoderm blastocoel D
notochord
tissues. In the early gastrula (left panel) the
prechordal organizer is located in the dorsal lip of the
plate
mesoderm blastopore. The cells of the leading edge
Ventral Dorsal Future gut cavity (orange) are the first to internalize and give
A P rise to anterior endoderm of the neurula
organizer stage (right panel). The deep cells (brown)
leading edge cells
are next to internalize and give rise to the
prechordal plate, the mesoderm anterior to
ectoderm the notochord, which forms the ventral head
blastopore anterior mesoderm. The remainder of the organizer
Vegetal endoderm endoderm gives rise to notochord (red).
V
Adapted from Kiecker, C. and Niehrs, C: 2001.

Fig. 5.5 Notochord and head-fold

formation in the chick embryo. This
schematic diagram shows a sagittal section
through the chick embryo (inset, dorsal
view) at the stage of head-fold formation
ectoderm as Hensen’s node starts to regress. As the
head fold prechordal plate mesoderm (neural plate) Hensen’s node node regresses, the notochord (sometimes
called the head process at this stage) starts
to form anterior to it with the prechordal
plate mesoderm immediately anterior to the
Yolk notochord mesenchyme
Anterior (head process) (prospective somites) Posterior
notochord. The undifferentiated mesenchyme
on either side of the notochord will form
somites. In the main picture the anterior
anterior structures. Experiments investigating the inductive capacity of different parts somitic mesoderm (in which one somite has
of the organizer show that the ability to induce heads—‘head organizer’ function—is formed, see inset) has been cut away to
also restricted to this vegetal region. The more dorsal part of the organizer (shown in show the notochord.
red in Fig. 5.4) is characterized by expression of the transcription factor Xnot, and can
induce trunk and tail structures but not heads.
The avian equivalent of the Spemann organizer is Hensen’s node, the region at the
anterior end of the primitive streak in the chick blastoderm. The node contributes cells
to prechordal plate mesoderm, notochord, somites, and gut endoderm, as well as pro-
ducing inducing signals. In the chick embryo, the notochord forms in the dorsal mid-
line anterior to Hensen’s node, mostly from the axial mesoderm left behind as the node
and primitive streak regress (Fig. 5.5). The neural plate develops above it and somites
are formed from mesoderm on either side of the notochord. Notochord and somite
formation and neurulation proceed similarly in the mouse (see Figs 3.25 and 3.26).
The properties of the avian node have been investigated by transplanting quail
nodes to chick embryos. Quail cells can be distinguished from chick cells by their
distinctive nuclei, which can be detected in histological sections (Fig. 5.6) or by using
a species-specific antibody. When, for example, a node from a gastrula-stage quail
embryo is grafted beneath the lateral epiblast of a chick embryo at the same stage
of development, the transplanted node can induce the formation of a complete addi-
tional axis with a head. When the node is taken from an older embryo at the head-
process-stage, for example, then it only induces a trunk and not a head (Fig. 5.7).
Thus the outcome of grafting nodes at different stages is similar to that seen with Fig. 5.6 Photograph of quail-chick
organizer grafts of different stages in Xenopus. In the mouse, the node precursors can chimeric tissue. The quail cells are on the
induce a similar axis duplication on transplantation to the lateral epiblast of an early left and the chick cells are on the right. Note
embryo, with the exception of the forebrain, which requires additional signals from the heavily staining nuclei of the quail cells.
the anterior visceral endoderm (see Fig. 4.14). Photograph courtesy of Nicole Le Douarin.
178 Chapter 5 Vertebrate development III: patterning the early nervous system and the somites

Donor quail stage 4 Host chick stage 4 Donor quail stage 6

head fold

Hensen’s
node

notochord

primitive streak

Xenopus Mouse
gastrula gastrula
mesoderm
Fig. 5.7 Hensen’s node can induce a new axis in avian embryos. When Hensen’s node
from a stage-4 quail embryo (left) is grafted to the extra-embryonic area opaca (dark gray),
just outside the embryonic area pellucida (light gray), of a host chick embryo at the same
organizer stage of development (center), a complete new axis, including brain tissue, forms at the site
of transplantation (photo, embryo on the left). The embryo in the center of the photograph
Genes in organizer region
has formed from the original primitive streak of the host chick embryo. The two stripes of dark
purple staining in the left and center embryos indicate expression of Krox20 mRNA, a marker for
Brachyury Brachyury
Genes the hindbrain, as detected by in situ hybridization. Quail cells are distinguished from chick cells
encoding by staining reddish brown with anti-quail antibody. Although some of the tissue of the new
transcription goosecoid goosecoid
axis (embryo on left) is formed from the quail graft, most has been induced from chick tissue
factors
that does not normally form an embryo. A node from a quail embryo at the head-process stage
Xlim1 Lim1
of development (stage 6, right) grafted to a stage-4 chick embryo produces a new axis with
only a trunk (photo, right embryo) and most of the new tissue is derived from the quail graft,
Xnr3 Nodal as indicated by the reddish-brown staining throughout. Photograph from Stern, C.D.: 2005.
Genes
encoding chordin, chordin, A number of proteins are specifically expressed in the vertebrate organizer and are
secreted noggin noggin
proteins known to be required for its function. Many are common to all vertebrates (Fig. 5.8).
Cerberus Cerberus-related Goosecoid, for example, is an early and reliable marker of the organizer that is
­expressed, in Xenopus, in the cells that will give rise to foregut, prechordal plate, and
notochord, and which have internalized by the mid-gastrula stage. Although goosecoid
Fig. 5.8 Genes expressed in the
expression is required for head formation in the normal course of development, goose-
Spemann organizer region of the Xenopus
coid mRNA injected into ventral blastomeres of Xenopus embryos induces a secondary
gastrula, and in the node in the mouse
gastrula. There is a similar pattern of gene axis lacking a head. It is important to bear in mind, however, that although many of the
activity in the two animals, with homologous same proteins are produced in and around the organizer in different vertebrates, they
genes being expressed. The expression of do not all appear to have precisely the same functions in all our model animals.
some of these genes, such as Brachyury, is We saw in Chapter 4 that gradients of signaling by TGF-β family proteins (Nodal
not confined to the organizer. and BMPs) pattern the dorso-ventral axis of the Xenopus gastrula (see Sections 4.17
 The role of the organizer and neural induction 179

and 4.18). Now we see that a gradient of Wnt/β-catenin signaling patterns the antero-
D
posterior axis. So any account of early vertebrate development needs to incorporate
these orthogonal TGF-β gradients and Wnt gradients into a three-dimensional coor-
dinate system of positional information that functions during axial patterning in the
gastrula (Fig. 5.9). The essential ingredients for head formation in Xenopus are the
inhibition in anterior tissues of Nodal, BMP and Wnt signals, which are present in
the gastrula at this time (see Section 4.17): an extra head can be induced by the simul- A P
taneous inhibition of BMP and Wnt signaling, or of Nodal and BMP signaling in the
ventral region of an early gastrula. As we saw in Chapter 4, antagonists of BMPs and
Wnts are secreted by the Xenopus organizer: the proteins Chordin, Noggin, and Fol-
listatin can antagonize BMP, and Dickkopf1antagonizes Wnt. The protein Cerberus,
which is produced by the anterior endoderm, antagonizes Wnt, Nodal, and BMP sig- V

naling (see Chapter 4 summary figure, p. 167). The head will form where both BMP
Wnt BMP
and Wnt signaling are low, whereas the tail forms where both are high. The important
function of the Spemann organizer is to regulate these gradients by secreting growth-
Fig. 5.9 Double-gradient model of
factor antagonists. Morphogen gradients at right angles to each other also operate
embryonic axis formation. The model shows
to pattern the Drosophila wing (see Chapter 11), and represent an important way of
how gradients of Wnts (blue) and BMPs
patterning an embryonic tissue. (green) perpendicular to each other regulate
antero-posterior and dorso-ventral patterning
5.2 The neural plate is induced in the ectoderm in the amphibian embryo. The color scales of
the arrows indicate the signaling gradients;
The induction of neural tissue from ectoderm was first indicated by the organizer- arrows indicate the direction of spreading of
transplant experiment in frogs illustrated in Fig. 5.3; in the secondary embryo that the signals. For example, the formation of the
forms at the site of transplantation, a nervous system develops from the host ecto- tail requires a higher level of Wnt signaling
derm that would normally have formed ventral epidermis. This suggested that neural than does the head. Patterning begins at
tissue could be induced from as-yet-unspecified ectoderm by signals emanating from gastrula stages, but for clarity is depicted
the organizer mesoderm. The requirement for induction was confirmed by experi- here in an early neurula.
ments that replaced prospective neural plate ectoderm with prospective epidermis Adapted from Niehrs, C.: 2004.
before gastrulation; the transplanted prospective epidermis developed into neural tis-
sue (Fig. 5.10). This showed that the formation of the nervous system is dependent
on an inductive signal.
In the chick embryo, neural tissue can similarly be induced in the epiblast—in both
the area pellucida and the area opaca—by grafts from the primitive-streak mesoderm
(see Fig. 5.7). Inducing activity is initially located in the anterior primitive streak
and Hensen’s node. Later, during regression of the node, inducing activity becomes

Fragment of presumptive epidermis transplanted

Early gastrulas from one early gastrula to another

presumptive Fig. 5.10 The nervous system of Xenopus

neural ectoderm is induced during gastrulation. The left
panels show the normal developmental fate
V D of ectoderm at two different positions in
the early gastrula. The right panels show
presumptive the transplantation of a piece of ventral
epidermis
ectoderm, whose normal fate is to form
epidermis, from the ventral side of an early
Induction of neural tissue in gastrula to the dorsal side of another, where
Normal fate of ectoderm at neurula stage transplanted fragment it replaces a piece of dorsal ectoderm whose
A
normal fate is to form neural tissue. In its
neural tissue neural tissue new location, the transplanted prospective
epidermis develops not as epidermis but as
neural tissue, and forms part of a normal
nervous system. This shows that the ventral
tissue has not yet been determined at the
epidermis
time of transplantation, and that neural
P
tissue is induced during gastrulation.
180 Chapter 5 Vertebrate development III: patterning the early nervous system and the somites

much reduced and by the four-somite stage has completely disappeared. The compe-
tence of the ectoderm to respond to neural-inducing signals disappears at the time of
head-process formation, suggesting that in normal development neural induction has
ended by this time.
An enormous amount of effort was devoted in the 1930s and 1940s to trying to
identify the signals involved in neural induction in amphibians. Researchers were
encouraged by the finding that a dead organizer region could still induce neural tis-
sue. It seemed to be merely a matter of hard work to isolate the chemicals responsible.
The search proved fruitless, however, for it appeared that an enormous range of dif-
ferent substances were capable of varying degrees of neural induction. As it turned
out, this was because salamander ectoderm, the main experimental material used,
has a high propensity to develop into neural tissue on its own. This is not the case
with Xenopus ectoderm, although simply dissociating Xenopus ectodermal cells for a
few hours can result in their differentiation as neural cells after reaggregation. The
molecules responsible for neural induction have still not been definitively identified,
although there are now some strong candidates.
A key discovery in the study of neural induction was the finding that the BMP
­inhibitor Noggin, the first secreted protein isolated from the Spemann’s organizer,
could induce neural differentiation in ectoderm explants from Xenopus embryos. In the
late Xenopus blastula, BMPs are expressed throughout the ectoderm (see Fig. 4.36),
but their expression is subsequently lost in the neural plate. These results suggested
that neural plate could only develop if BMP signaling is absent. BMP inhibitors such
as Noggin and Chordin produced by the organizer therefore became attractive candi-
dates for inhibiting BMP signaling in the prospective neural plate and allowing neu-
ral induction to proceed. As BMP signaling maintains expression of the BMP genes,
­antagonizing BMP signaling would also shut down BMP expression.
These observations led to the so-called ‘default model’ for neural induction in Xeno-
pus. This proposed that the default state of the dorsal ectoderm is to develop as neural
tissue, but that this pathway is blocked by the presence of BMPs, which promote the
epidermal fate. The role of the organizer is to lift this block by producing proteins
that inhibit BMP activity; the region of ectoderm that comes under the influence of
the organizer will then develop as neural ectoderm. The organizer produces several
antagonists including the BMP antagonists Chordin and Noggin, the protein Cerberus,
which can antagonize BMP, Nodal, and Wnt signals, and the Wnt antagonist, Dick-
kopf1 (see Fig. 5.8, and Chapter 4 summary figure, p. 167). According to this model,
the antagonists secreted from the organizer act on ectoderm that lies ­adjacent to
the organizer at the beginning of gastrulation. As gastrulation proceeds, internalized
cells derived from the organizer continue to secrete these proteins, which continue
to act on the now-overlying ectoderm (see Fig. 5.4). Elimination of these organizer
antagonists individually in amphibians has rather modest effects on neural induction.
But when combinations of the BMP antagonists Chordin, Noggin, and Follistatin, or
of Cerberus, Chordin, and Noggin were simultaneously depleted in the organizer of
the frog Xenopus tropicalis by antisense morpholino oligonucleotides (see Box 6A,
p. 220), there was a dramatic failure of neural and other dorsal development and an
expansion of ventral and posterior fates. These experiments show that BMP inhibi-
tors, which are produced at the right place and at the right time, are required for
neural induction.
The default model is not the complete answer, however, as neural development in
both Xenopus and the chick also requires the growth factor FGF, even when BMP inhi-
bition is lifted by the presence of Noggin and Chordin. Precisely how FGF fits into the
picture in Xenopus is still much debated, with the most recent work suggesting that
BMP inhibition is most important for inducing anterior neural tissue, whereas FGF is
required in addition to the inhibition of BMPs for giving a posterior character to the
prospective neuroectoderm (Fig. 5.11).
 The role of the organizer and neural induction 181

Neural plate forms from prospective ectoderm where

BMP is expressed throughout the Xenopus blastula BMP signaling is inhibited by antagonists from
the organizer
Animal Animal prospective
neural
plate

Ventral Dorsal Ventral Dorsal

Fig. 5.11 Inhibition of BMP signaling is

required for induction of neural plate. BMP
organizer organizer (green) is expressed throughout the early
region region blastula. In the late blastula/early gastrula,
forming
Vegetal Vegetal BMP antagonists secreted from the organizer
inhibit BMP signaling in adjacent regions.
This negative action is not on its own enough
BMP-4 BMP-4 BMP antagonists inhibition of BMP signaling
to induce formation of neural tissue, which
BMP antagonists ectoderm neural plate mesoderm endoderm also requires a positive input from FGF
signaling (not shown).

In the chick, the earliest steps in neural induction occur before gastrulation in the
blastoderm, where FGF signaling is also involved in mesoderm induction (see Sec-
tion 4.19). Blastoderm cells are thought to be prevented from taking on a neural-plate
fate at that time because the chromosomal regions encoding neural-inducing genes
are at that time in a ‘silenced’ state in which genes cannot be switched on and tran-
scribed. Experiments in chick embryos suggest that the lifting of this inhibition during
gastrulation is associated with the removal of repressor proteins from the regulatory
regions of such genes and the remodeling of the chromatin into an active state by
chromatin-remodeling complexes (Box 5A, p. 182). Blastoderm cells that receive FGF
signals but have not formed mesoderm by this stage can now form neural plate.
One of the key genes activated by FGF signaling in neural induction in chick ­embryos
is Churchill, which encodes a zinc finger transcription factor. When Churchill expres-
sion is downregulated using an antisense morpholino oligonucleotide (see Box 6A,
p. 220), the neural plate does not form. The activation of Churchill as a result of FGF
signaling leads indirectly to repression of genes characteristic of mesoderm and acti-
vation of the gene for the neural-specific transcription factor Sox2, the earliest defini-
tive marker of neural plate in the chick.
Not all the effects of FGF in neural induction in chick or Xenopus embryos may
­result from a direct inducing effect. FGF activates the mitogen-activated protein kinase
(MAPK) intracellular signaling pathway (Box 5B, p. 184) and studies in both Xenopus
and mouse provide evidence that MAPK can interfere with BMP signaling by causing
an inhibitory phosphorylation of Smad1, which is part of the intracellular signaling
pathway stimulated by BMPs (see Fig. 4.33). Therefore, FGF signaling could also con-
tribute to the inhibition of BMP signaling, and could help to allow neural induction
this way (see Fig. 5.11). The MAPK signaling pathway is also activated when Xenopus
ectodermal cells are dissociated, which might explain, as noted earlier, why this dis-
sociation so easily induces a neural fate.
Neural induction is therefore a complex multistep process and, as discussed earlier,
the very first stages are likely to occur in the blastula or blastoderm, even ­before
a distinct organizer region becomes detectable. The individual signaling proteins have
multiple roles in development, and their roles change over time, which makes it dif-
ficult to disentangle their real contributions to any particular process. Their develop-
mental roles also differ between different vertebrates. An essential similarity in the
mechanism of neural induction among vertebrates is likely, however, as Hensen’s
node from a chick embryo can induce neural gene expression in Xenopus ectoderm
182 Chapter 5 Vertebrate development III: patterning the early nervous system and the somites

Box 5A Chromatin-remodeling complexes

Whether a gene is expressed or

not depends not only on whether remodeled nucleosomes
appropriate gene-specific DNA-
­binding regulatory proteins are chromatin-
present, but also on the state of remodeling
complex
the chromatin—the complex of DNA
and proteins of which the chromo-
gene activator protein
somes are composed. Chromatin
proteins include the histones,
which package DNA into structures
called nucleosomes, and other
proteins that bind directly to DNA promoter DNA histone removal
histone chaperone
or to proteins bound to DNA, and
affect chromatin structure and the
availability of the DNA for tran-
scription. In highly simplified terms, euchromatin is chromatin active or inactive forms of the proteins, in different ­combinations,
in which genes are available for transcription whereas in hetero- were transfected into the area opaca of early chick epiblasts,
chromatin they are unavailable—or ‘silenced’. an area outside the embryo that would never normally express
In the case of the chick embryo, the gene Sox2 is an early Sox2, to see whether they could induce Sox2 expression. From
marker for neural plate but is not expressed before gastrula- the results of these experiments, a series of steps leading from
tion even though FGF, which is involved in neural-plate induc- inactive Sox2 in the early embryo to active Sox2 in the neural
tion, is active early on. Experiments investigating the regulation plate could be proposed. The early exposure to FGF is proposed
of Sox2 in cultured chick embryos have shown that, like many to prime Sox2 for eventual expression by ­inducing an activa-
other genes, its expression requires the action of a chromatin- tor protein that binds to the Sox2 control region along with
remodeling protein called Brm. Brm has ATPase activity and is an Brm, but the gene is kept inactive by the binding of ­repressive
essential component of numerous multiprotein ATP-dependent ­heterochromatin-inducing proteins, which are ubiquitously
chromatin-­remodeling complexes that are active in cells. ­expressed throughout the embryo. The inhibition is lifted by the
Such complexes can be recruited to promoter sites in the DNA specifically timed expression of a protein that disrupts the com-
(see Fig. 1.19) and they are thought to use the energy of ATP plex, removing the repressors and enabling the activator pro-
­hydrolysis to loosen the attachment of DNA to histones and tein, in conjunction with the now active chromatin-remodeling
thus enable nucleosomes to be slid along DNA (see figure above, complex, to switch on gene expression (see figure below). This
upper arrow) or the histone core to be removed from the DNA general strategy for preventing the premature ­expression of a
(see figure, lower arrow). This frees the promoter DNA to bind crucial gene in the presence of potential activating signals is
other gene regulatory proteins, RNA polymerase, and the rest likely to be widely deployed in development.
of the transcriptional machinery, thus enabling the gene to be The recruitment of chromatin-modeling complexes and
transcribed. ­heterochromatin-inducing proteins to specific sites on the chro-
Chromatin-remodeling complexes work in conjunction with mosome often depends on pre-existing chemical modifications
other gene regulatory proteins. To determine the effects of can- to chromatin such as DNA methylation and histone methylation
didate regulatory proteins on Sox2 expression, DNAs ­encoding and acetylation, which are discussed in Box 10A, p. 374.

Gene inactive Gene active

additional
activator heterochromatin- activator
protein inducing protein

Brm HP1

control region gene transcription

 The role of the organizer and neural induction 183

Pieces of chick epiblast are sandwiched between Fig. 5.12 Hensen’s node from a chick
mRNAs expressed in Xenopus
pieces of Xenopus animal cap tissue ectoderm in response to chick grafts embryo can induce gene expression
characteristic of neural tissue in Xenopus
ectoderm. Tissues from different parts of

(a) Non-streak epiblast

the primitive-streak stage of a chick epiblast

Stage 30 Xenopus
control (no graft)
Hensen’s node
are placed between two fragments of animal
(a) (c) cap tissue (prospective ectoderm) from

(c) Stage 4
(b) Streak
a Xenopus blastula. The induction in the
Xenopus ectoderm of genes that characterize
NF-3 the nervous system is detected by looking for
(b)
expression of mRNAs for neural cell adhesion
molecule (N-CAM) and neurogenic factor-3
(NF-3), which are expressed specifically in
neural tissue in stage-30 Xenopus embryos.
N-CAM Only transplants from Hensen’s node induce
the expression of these neural markers in the
Xenopus ectoderm. (EF-1α is a ubiquitous
transcription factor expressed in all cells and is
EF-1α
used as a control.)
After Kintner, C.R., et al.: 1991.

(Fig. 5.12), which suggests that there has been an evolutionary conservation of induc-
ing signals. Moreover, early nodes induce gene expression characteristic of anterior
amphibian neural structures, whereas older nodes induce expression typical of poste-
rior structures. These results are in line with the theory that the vertebrate node speci-
fies different antero-posterior positional values at different times, and they confirm
the essential similarity of Hensen’s node and the Spemann organizer.
In Xenopus, the elongation of the posterior neural plate is part of the general elongation
of the embryo along the antero-posterior axis and contributes to the generation of the spi-
nal cord (see Fig. 3.7). This elongation process is discussed in more detail in ­Chapter 8.
In chick and mouse, on the other hand, the whole spinal cord is generated from a small
region of proliferating stem cells that develops in the ectoderm at the posterior end of the
neural plate on both sides of the node and primitive streak (Fig. 5.13). This stem zone
becomes distinct just before somite formation begins and moves posteriorly along with
the node, leaving behind neural progenitor cells that produce the spinal cord.
In zebrafish, there appears to be a clear difference in the mechanism of induction of
anterior neural tissue close to the shield in the dorsal region, and the induction of the
posterior neural ectoderm that will give rise to the spinal cord. Whereas the ­organizer

Anterior Fig. 5.13 The spinal cord is formed from

a zone of stem cells that arises in the
Hensen’s neural fold head fold
node posterior neural plate. In the chick embryo,
neural tube a restricted region of stem cells forms in the
closed neural tube posterior neural plate alongside the node
prospective (spinal cord, and anterior primitive streak just before
stem zone somites undifferentiated
neural cells) the onset of somitogenesis (yellow). This
stem-cell region comes to express Sax1, a
transition zone
(progenitor gene encoding a homeodomain-containing
neural cells)
protein, and moves posteriorly with the node
primitive
Hensen’s as it regresses, leaving behind progenitor
node stem zone
streak neural cells that will form the spinal cord.
Posterior The complete spinal cord is formed from this
stem-cell zone.
Adapted from Delfino-Machin, M., et al.:
Sax1
2005.
184 Chapter 5 Vertebrate development III: patterning the early nervous system and the somites

Box 5B The FGF signaling pathway

The developmentally important growth factors FGF and EGF (epi-

cell
dermal growth factor), along with numerous other growth and signal
membrane
differentiation factors, signal through transmembrane receptors receptor Raf
with intracellular tyrosine kinase domains. Several different sig-
naling pathways lead from these receptors, which control many
Ras
aspects of cell behavior throughout an animal’s life. The signaling
P P Sos
pathway shown here is a simplified version of a pathway triggered tyrosine MAPKK
kinase
by FGF that leads to changes in gene expression that promote domain Grb P

cell survival, cell growth, cell division, or differentiation, depend-

ing on the cells involved and the developmental context. There Cytoplasm MAPK
are multiple different forms of FGF and FGF receptors, which are P
P
also used in different contexts. For example, in the chick, FGF-8 is
involved in somite formation and in signaling from the organizing
center in the brain (see Sections 5.5 and 5.10) whereas FGF-10 is
crucial for limb development (discussed in Chapter 11).
P
The intracellular signaling pathway from these receptors illus- Nucleus P
trated here is often known as the Ras–MAPK pathway because
of the key involvement of the small G protein Ras and the acti- Phosphorylation
vation of a cascade of serine/threonine kinases which ends in P P
P Transcription
transcription P
the activation of a mitogen-activated protein kinase (MAPK). factor
The name comes from the fact FGF and other growth factors
can act as mitogens, agents that stimulate cell proliferation. In
one variation or another, the central Ras–MAPK module occurs
in the signaling pathways from many different tyrosine-kinase of the first serine/theronine kinase in the cascade, which in the
receptors. mammalian pathway is called Raf. This phosphorylates and acti-
Binding of extracellular FGF to its receptor results in dimeriza- vates the next protein kinase, the MAPK kinase (MAPKK), which
tion of two ligand-bound receptor molecules, activating the phosphorylates and activates MAPK. MAPK may then phosphory-
­intracellular tyrosine kinase domains which then phosphorylate late other kinases and can also enter the nucleus and phosphory-
each other. The phosphorylated receptor tails recruit adaptor late transcription factors, thus activating gene expression. There
proteins (Grb and Sos), which in turn recruit and activate Ras at are several different MAPKs in mammalian cells that are used in
the plasma membrane. This results in the binding and activation different pathways and target different transcription factors.

in zebrafish contributes to the induction of anterior neural tissue by inhibiting BMP

signaling, the ectoderm that develops into the spinal cord is some distance away
from the organizer on the ventral-vegetal side of the embryo, where organizer signals
do not reach (Fig. 5.14). The initiator of neural development in this ventral-vegetal

Fig. 5.14 Prospective spinal cord in the

zebrafish embryo is distant from the Zebrafish late gastrula 1-day embryo
organizer. In the zebrafish, the ectoderm that
will form the spinal cord is situated on the neural tissue head
other side of the gastrula from the organizer, brain
tail
in the ventral–vegetal ectoderm, and is too epidermis
far away to be influenced by signals from the
organizer. FGF signals in the ventral–vegetal posterior V D
neural
region induce this ectoderm as neuroectoderm tissue
and BMPs promote its development as mesoderm
posterior neural tissue (spinal cord). The yolk
epidermis is not shown here for simplicity. By FGF trunk
activity spinal cord
this stage it covers the whole embryo.
BMP activity
Adapted from Kudoh, T., et al.: 2004.
 The role of the organizer and neural induction 185

Neurula anterior mesoderm grafted into early gastrula induces an extra head with eyes and forebrain
A

Fig. 5.15 Induction of the nervous

induced extra head system by the mesoderm is region
P
specific. Mesoderm from different positions
along the dorsal antero-posterior axis of early
Neurula posterior mesoderm grafted into early gastrula induces a trunk and tail newt neurulas induces structures specific
to its region of origin when transplanted to
A
induced trunk and tail ventral regions of early gastrulas. Anterior
mesoderm induces a head with a brain (top
panels), whereas posterior mesoderm induces
a posterior trunk with a spinal cord ending in
a tail (bottom panels).
P
After Mangold, O.: 1933.

ectoderm is FGF, and BMP signals, which are high in the ventral region, in this case
appear to push the neuroectoderm towards a posterior fate.

5.3 The nervous system is initially patterned by signals from the mesoderm

Whatever the mechanisms of neural induction eventually turn out to be, it is clear that
the neural plate can be patterned by signals from the mesoderm. Pieces of mesoderm
taken from different positions along the antero-posterior axis of a newt neurula and
placed in the blastocoel of an early newt embryo induce neural structures at the site of
transplantation. The structures that are formed correspond more or less to the original
position of the transplanted mesoderm: pieces of anterior mesoderm induce a head
with a brain, whereas posterior pieces induce a trunk with a spinal cord (Fig. 5.15).
Another indication of positional specificity comes from the observation that pieces
of neural plate induce similar regional neural structures in adjacent ectoderm when
transplanted beneath the ectoderm of another gastrula.
Both qualitative and quantitative differences in signaling by the mesoderm can
account for antero-posterior neural patterning. Qualitatively different inducers are
secreted by the mesoderm at different positions along the antero-posterior axis. In
Xenopus, Wnt antagonists such as Cerberus, Dickkopf1, and Frizzled-related pro-
tein are predominantly secreted from anterior endoderm and mesoderm beneath the
Graded signals from mesoderm give neural tissue
­future head, whereas BMP antagonists are released from mesoderm beneath both a posterior identity
head and trunk. The combinatorial action of BMP and Wnt antagonists, therefore,
could promote development of a head with a brain, whereas BMP antagonists alone
neural tissue neural plate
induce a trunk with a spinal cord. Mouse embryos mutant for Dickkopf1 are headless,
indicating that Wnt inhibition is necessary for head formation. However, when over-
expressed in Xenopus, Dickkopf1 is only able to induce extra heads when in combina-
tion with BMP inhibitors. This shows that it is the combined inhibition of BMP and
mesoderm
Wnt signaling that mimics the action of the anterior mesoderm.
Quantitative differences in Wnt/β-catenin signaling are present along the axis. Mov-
ing more posteriorly along the axis, the levels of Wnt/β-catenin signaling progressively
Fig. 5.16 Mesoderm patterns the neural
increase, resulting in a gradient with the highest level of Wnt/β-catenin signaling at
plate. Quantitative differences in a signal
the posterior end of the embryo (see Fig. 5.9). There is evidence that high levels of
along the axis could help pattern the
Wnt signaling give prospective neural tissue a more posterior identity (Fig. 5.16). In neural plate. For example, the levels of
Xenopus animal cap explants taken from embryos injected with mRNA for the BMP Wnt/β-catenin signaling increase from
antagonist Noggin, and used to represent anterior neuralized tissue, increasing levels anterior to posterior and the high levels
of Wnt signaling induce the expression of different posterior marker genes. Wnts have at the posterior end could confer a more
also been shown to posteriorize the neural plate in chick and zebrafish. Evidence posterior identity on the neural plate.
186 Chapter 5 Vertebrate development III: patterning the early nervous system and the somites

from Xenopus, chick and zebrafish indicates that FGFs also act as posteriorizing fac-
neural plate tors. Thus, both quantitative and qualitative differences in signaling can account for
antero-posterior neural patterning and several different signals are involved.

5.4 Neural crest cells arise from the borders of the neural plate

Neural crest cells are induced at the borders of the neural plate, which at neurulation
rise up to form the neural folds and come together to form the dorsal portion of the
neural tube (Fig. 5.17). From there, they migrate to give rise to an enormous range
of different tissues and cell types. Neural crest cells are, for example, the origin of
the neurons and glia of the autonomic nervous system, the glial Schwann cells of the
­peripheral nervous system, the pigment cells (melanocytes) of the skin, the adrenaline
notochord (epinephrine)-producing cells of the adrenal glands, and, most surprisingly, much of the
bone, cartilage, muscle, and connective tissue of the face—cell types that are more com-
Dorsal monly of mesodermal origin. The migration of neural crest cells is described in more
epidermis
detail in Chapter 8 and their differentiation in Chapter 10. The migration of neural crest
from the hindbrain to the branchial arches, from which some of the structures and tis-
sues of the vertebrate head are derived, is touched on briefly later in this chapter in rela-
tion to the antero-posterior patterning of the hindbrain and the cells derived from it.
neural The induction of the neural crest is a multi-step process that in Xenopus starts at the
crest early gastrula stage and continues until neural tube closure. A current view of neural
cells
crest induction is that crest cells form in a band in the ectoderm along each lateral
border of the neural plate, at a position where BMP signaling is just above the level
neural tube that blocks neural plate formation (see Fig. 5.11). As might be expected from their role
Ventral in neural plate induction, Wnt and FGF signals also appear to be involved in neural
crest specification. Induction of neural crest is marked by expression of the transcrip-
Fig. 5.17 Neural crest cells are specified tion factors Sox9 and Sox10, which in turn activate the gene snail, an early marker of
at the lateral borders of the neural neural crest. This gene is the vertebrate version of the Drosophila gene snail (see Sec-
plate. First panel: prospective neural crest tion 2.19). As we shall see in Chapter 8, the Snail protein is involved in enabling the
cells (green) are specified at the sides of morphological transition that an epithelial cell undergoes to become a migratory cell.
the neural plate (blue) where levels of BMP Such epithelial-to-mesenchymal transitions occur in many situations during animal
are just high enough to prevent neural plate
development, and neural crest migration is just one example.
specification. Second panel: formation of the
neural folds carries these cells to the dorsal
crests of the neural tube, from which they Summary
soon migrate away (third panel).
Patterning of the early vertebrate embryo, along both the antero-posterior and dorso-
ventral axes, is closely related to the action of the Spemann organizer and its morpho-
genesis during gastrulation. When grafted to the ventral side of an early gastrula, the
Spemann organizer induces both a new dorso-ventral axis and a new antero-posterior
axis, with the development of a second twinned embryo. The organizer secretes growth
factor antagonists, which set up gradients of TGF-β and Wnt signaling to pattern the
embryo. In the zebrafish, the embryonic shield is the organizer, while in chick develop-
ment, Hensen’s node serves a function similar to that of the Spemann organizer, and it
too can specify a new antero-posterior axis. In the mouse, the node can specify a new
axis apart from the most anterior forebrain, for which induction by the anterior visceral
endoderm is also necessary.
The vertebrate nervous system, which forms from the neural plate, is induced both
by early signals within the ectoderm and by signals from the mesoderm that comes to
lie beneath prospective neural plate ectoderm during gastrulation. In Xenopus, inhibi-
tion of BMP signaling by proteins such as Noggin produced by the organizer is required
for neural induction. Patterning of the neural plate involves both quantitative and quali-
tative differences in signaling by the mesoderm with high levels of Wnt and FGF signal-
ing specifying the more posterior structures.
 Somite formation and antero-posterior patterning 187

Somite formation and antero-posterior patterning

By the end of gastrulation, the vertebrate embryo has elongated along the antero-
­posterior axis, the neural tube is evident, the notochord has formed, and the rest of
the mesoderm is located on either side of it, eventually completely surrounding the
endodermally derived gut. The fate maps of vertebrates (see Figs 4.19–4.23) show that
the notochord develops from the most dorsal region of the mesoderm, and somites
from more ventro-lateral mesoderm on either side of the midline, which is known as
the paraxial mesoderm. Lateral to the paraxial mesoderm lies the intermediate meso-
derm that gives rise to the kidney, and more lateral still is the lateral plate mesoderm,
which becomes divided into somatic mesoderm and splanchnic mesoderm by forma-
tion of the coelom (see Fig. 3.17, which shows these regions of mesoderm in sections
of the chick embryo). The anterior lateral plate mesoderm gives rise to the heart
while limb buds arise from lateral plate mesoderm at the appropriate levels along the
antero-posterior axis of the body.
The somites give rise to the bone and cartilage of the trunk, including the spinal col-
umn, to the skeletal muscles, and to the dermis of the skin on the dorsal side of the body,
and their patterning provides much of the body’s antero-posterior organization. The ver-
tebrae, for example, have characteristic shapes at different positions along the spine. We
will first examine the development of the somites and how their different developmental
identities along the antero-posterior axis are specified by Hox genes. We conclude our
look at somite formation by discussing how individual somites are patterned and which
structures they give rise to. We end this part of the chapter by revisiting the embryonic fused
neural
brain and looking at aspects of its early patterning along the antero-posterior axis. tube
S1
S2 direction of
5.5 Somites are formed in a well-defined order along the antero-posterior axis S3 somite
S4 formation
Somite number can be highly variable between different vertebrates; birds and ­humans S5
S6
have around 50 whereas snakes have up to several hundred. Much of the work on
somite formation has been done on the chick because of the ease with which the pro- pre-somitic
mesoderm
cess can be observed, and so we will use the chick as our main model organism here.
In the chick embryo, somite formation occurs on either side of the notochord in the
paraxial mesoderm anterior to the regressing Hensen’s node (Fig. 5.18). The first five
somites in the chick contribute to the most posterior part of the skull and, anterior
to this, the paraxial mesoderm does not segment. It will form some facial muscles
and bones—the rest of the skull is of neural crest origin. Somite formation proceeds
in an anterior to posterior direction in the unsegmented mesoderm—the pre-somitic Hensen’s
­mesoderm—lying between the most recently formed somite and the node. As the node
node regresses, the length of this region remains fairly constant. Changes in cell shape
and intercellular contacts in the pre-somitic mesoderm result in the formation of dis-
tinct blocks of cells—the somites. Somites are formed in pairs, one on either side
of the notochord, with the two somites in a pair forming simultaneously. A pair of
somites forms every 90 minutes in the chick, every 120 minutes in the mouse, every Fig. 5.18 Somites form in pairs from the
45 minutes in Xenopus, and every 30 minutes in zebrafish. paraxial mesoderm. Somites are blocks of
The sculpting of the somites involves tissue separation and cell movements, as well tissue that form from the mesoderm lying
as integration of cells at the anterior and posterior somite borders. Somite formation immedately on either side of the notochord,
in ­Xenopus, in particular, displays several unique features. A group of cells of the the paraxial mesoderm. They are made
in an anterior to posterior direction from
unsegmented mesoderm, which are oriented perpendicular to the notochord, sepa-
unsegmented pre-somitic mesoderm which
rates from the rest of the mesoderm by formation of an intersomitic furrow and forms
is produced from a pool of stem cells around
a distinct block—the prospective somite. The block then undergoes a 90° rotation. the node. The first five somites give rise to
This rotation involves a series of cellular movements and reorientations that have not the posterior part of the skull; the remaining
been well investigated, but which seem to depend on cells sensing their locations somites produce the muscles and bones of
within the block and undergoing morphological changes and cellular rearrangements the trunk and tail and the muscles of the
according to their positions. limbs.
188 Chapter 5 Vertebrate development III: patterning the early nervous system and the somites

Pre-somitic mesoderm is inverted Somites form in reverse order

developing optic
neural tube vesicle

developing
brain

notochord

1
X Y
2
pre-somitic neural
mesoderm 3
tube
Y X 4

Order of somite formation

5
Hensen’s 10
node
9
8
7
6
11
12
13
14

Fig. 5.19 The temporal order of somite

formation is specified early in embryonic
development. Somite formation in the chick
In the chick, the cells that give rise to the somites originate in the epiblast on either
proceeds in an antero-posterior direction. side of the anterior primitive streak and move into it at gastrulation to form a popula-
Somites form sequentially in the pre-somitic tion of somitogenic stem cells located around Hensen’s node. The stem cells divide,
region between the last-formed somite and and those that remain in the stem-cell region continue to be self-renewing stem cells,
Hensen’s node, which moves posteriorly. If but those left behind as the node regresses form the pre-somitic mesoderm. As new
the antero-posterior axis of the pre-somitic cells are being added to the pre-somitic mesoderm at the posterior end of the chick
mesoderm is rotated through 180°, as shown embryo, somites are forming at the anterior end. In this way, the length of the pre-
by the arrow, the temporal order of somite somitic mesoderm remains more or less constant, enough to make about 12 somites.
formation is not altered—somite 6 still
The sequence of somite formation in the unsegmented region is unaffected by
develops before somite 10.
transverse cuts in the plate of pre-somitic mesoderm, suggesting that somite forma-
tion is an autonomous process and that, at this time, no extracellular signal specifying
antero-posterior position or timing is involved. Even if a piece of the unsegmented
mesoderm is rotated through 180°, each somite still forms at the normal time, but
with the sequence of formation running in the opposite direction to normal in the
inverted tissue (Fig. 5.19). So, before somite formation begins, a molecular pattern
that specifies the time of formation of each somite has already been laid down in
the pre-somitic mesoderm; we shall return to this patterning process later. Given the
­existence of this pattern, the prospective identity of each somite must be related to the
temporal order in which their cells entered the pre-somitic mesoderm.
Somite formation is largely determined by an internal ‘clock’ in the pre-somitic
­mesoderm. This clock is represented by periodic cycles of gene expression, such as
that of the genes c-hairy1 and c-hairy2 in the chick embryo, whose expression sweeps
from the posterior to the anterior end of the pre-somitic mesoderm with a period
of 90 minutes, the time it takes for a pair of somites to form. In a newly formed
somite, c-hairy 1 expression becomes restricted to the posterior end of the somite,
where it ­persists, while a new wave of c-hairy 1 expression starts at the tail end of the
pre-somitic mesoderm (Fig. 5.20). Oscillating gene expression has also been detected
 Somite formation and antero-posterior patterning 189

SI SI SII SII
S0 S0 SI SI
S0
S0
pre-somitic
mesoderm
notochord

Differentiation of
somite SII A somitic derivatives
formation SI
every 90 S0 P Border formation Anterior
minutes S-I
S-II Segment specification

pre-somitic mesoderm

somitic stem cells

FGF Hensen's node
gradient
0 hour
1.5 hours 3 hours
6 hours
9 hours

18 hours
Posterior

Fig. 5.20 Somite formation in the chick. As shown in the left- anterior (gray) and posterior (yellow) halves. The top right-hand panel
hand panel, the somites are generated successively from pre-somitic shows stages in one of the cycles of c-hairy 1 expression (yellow) that
mesoderm, which is derived from somitic stem cells in the primitive sweep from posterior to anterior of the pre-somitic mesoderm every
streak. As pre-somitic cells are released into the posterior pre-somitic 90 minutes. During each cycle, a given pre-somitic cell experiences
mesoderm, a new pair of somites buds from the anterior end every distinct phases when c-hairy 1 is expressed and when it is not
90 minutes. SI, the most recently formed somite; SII, the last but expressed. The lower right-hand panel shows the history of a pre-
one somite formed; S0, somite in the process of formation, whose somitic cell (red dot) from the time it enters the pre-somitic mesoderm
boundaries are not yet set; S-I, S–II, blocks of pre-somitic cells that will until it is incorporated into a somite. Somitic cells in the anterior
form somites. The position in which the border is formed to generate somites will have experienced fewer cycles of c-hairy 1 expression
each somite is specified by a threshold level of FGF-8 gradient (green before they leave the node region for the pre-somitic mesoderm than
dotted line). At formation, each somite acquires antero-posterior will cells in posterior somites, and this could define a clock that is both
polarity, after which it can respond to the signals that pattern it along linked to somite segmentation and ‘tells’ the somite its position along
the antero-posterior and dorso-ventral axes. The somite has distinct the antero-posterior axis.

in the mouse and zebrafish pre-somitic mesoderm. How these gene oscillations are
related to regular somite segmentation is not yet clear, and the underlying mechanism
for determining the segment boundary may differ in different vertebrates.
In the mouse and chick, another of the genes with cyclic expression encodes Luna-
tic fringe, an enzyme that modulates activity of the Notch–Delta signaling pathway
(Box 5C, p. 190). This pathway is widely involved in determining cell fate and delim-
iting boundaries during embryonic development. As we shall see later, somites are
functionally divided into anterior and posterior halves with different fates, and one
function of Notch signaling in mice seems to be to establish this anterior–posterior
boundary. Both Notch and Delta are transmembrane proteins, and this pathway is an
example of the transmission of signals by direct cell–cell contact. In zebrafish, one
proposed function of Notch activity during somitogenesis is the synchronization of
oscillating gene expression between neighboring somitic stem cells. Similarly, mouse
embryos in which Delta–Notch signaling is either made constitutively active or abol-
ished entirely often do not form somites, and if they do, the somites vary in size and
shape and are different on each side of the body.
The timing and position of somite formation seem to be determined by the interac-
tion of the segmentation clock with the intercellular signaling protein FGF-8. In the
chick and the mouse at this time there is an FGF-8 gradient in both mesoderm and
190 Chapter 5 Vertebrate development III: patterning the early nervous system and the somites

Box 5C The Notch signaling pathway

The family of Notch transmembrane

Notch inactive Notch bound to ligand
signaling proteins are involved in
numerous developmental decisions
that set cell fate, ranging from neu-
ronal precursors in Drosophila and
glycosylation
vertebrates to the differentiation of
epidermal cells in mammalian skin. Ligand
(e.g. Delta)
The ligands for Notch are also mem-
Notch
brane-bound proteins and so Notch
signaling requires direct cell–cell con-
tact. This means that Notch delivers
Notch
a very spatially precise signal that intracellular
domain
can set the fate of a single cell out Cleavage and
release of Notch
of many, as we shall see when we intracellular domain

look at the development of the Dros-

Cytoplasm Nucleus co-activator
ophila eye in Chapter 11 and neuro- complex
mastermind
blast specification in Chapter 12. repressor co-activator
CSL
The Notch ligands are the mem-
bers of the Delta and Serrate/Jag-
expression of target genes
ged1 families. Binding of ligand on
one cell to Notch on an adjacent
cell activates intracellular signaling from the receptor. The Notch the Notch-activated transcription factors CBF1/RBPkJ in mam-
receptor has a large extracellular domain with several glycosyla- mals, Suppressor of Hairless (Su(H)) in Drosophila and LAG-1 in
tion sites that are specifically modified by glycosyltransferases, C. ­elegans). In the absence of Notch signaling, CSL is complexed
including those of the Fringe family. Glycosylation modulates with repressor proteins. Activation of the CSL complex by binding
Notch signaling by affecting the ability of the extracellular of the Notch intracellular domain involves release of the repres-
­domain to bind to different ligands, facilitating its binding to sors and recruitment of the co-activator protein Mastermind and
Delta ligands while reducing the ability to bind to Serrate ligands. other co-activators. The outcome of the Notch signaling pathway
Activation of the Notch receptor involves the enzymatic cleav- is the transcription of specific genes. Notch signaling is very
age of the cytoplasmic tail of the receptor (the Notch intracellu- versatile. In different organisms and in different developmental
lar domain) and its translocation to the nucleus to bind and acti- ­circumstances, activation of Notch is used to switch a wide range
vate a transcription factor of the CSL family (CSL is named after of different genes on or off.

ectoderm, with its high point at the node and decreasing anteriorly. The gradient
regresses in a posterior direction along with the node, probably because FGF-8 mRNA
is only made in cells in and around the node and is progressively degraded in the
cells that are left behind when the node moves posteriorly. The result is a gradient
in intracellular FGF-8 mRNA, which after translation and secretion of FGF-8 gives an
extracellular head-to-tail FGF-8 gradient with its high point eventually residing in the
tailbud of the embryo (Fig. 5.21). Somite formation occurs where FGF-8 reaches a
sufficiently low threshold level. As the gradient moves along with the node, therefore,
it successively defines regions in the pre-somitic mesoderm where somites start to
form. The small secreted signaling molecule retinoic acid (Box 5D, pp. 192–193) has
an antagonistic effect on FGF action. It forms a gradient in the opposite direction to
FGF, and by antagonizing FGF, prevents the pre-somitic region from continually get-
ting longer. Retinoic acid is synthesized in the somites and diffuses both posteriorly
and anteriorly. It is also thought to be involved in maintaining the bilateral symmetry
of somite development by buffering the somites, in some way not yet understood,
against the signals that are setting up left–right asymmetry in the lateral mesoderm at
this stage in development (see Chapter 4).
 Somite formation and antero-posterior patterning 191

Fig. 5.21 FGF and retinoic acid gradients

8.5-day mouse embryo
help to pattern the antero-posterior axis
Anterior in the mouse embryo. An antero-posterior
S6
S7 retinoic acid gradient of FGF develops in the embryo, with
S8 production in its high point at the node. This schematic
S9
mesoderm diagram shows the dorsal view of the
S10
posterior half of an 8.5-day mouse embryo
retinoic acid
neural gradient with 10 pairs of somites formed and the
tube retinoic neural tube partly closed but still open at
acid the posterior end of the future spinal cord.
degradation
The FGF gradient is formed by cells in and
FGF
gradient around the node synthesizing FGF mRNA,
node
which is then gradually degraded when cells
leave the region anteriorly to form the pre-
somitic mesoderm. This results in a gradient
primitive of translated and secreted FGF protein
streak that is continuously moving posteriorly as
Posterior the embryo elongates. The gradient in the
mesoderm fades out at around the position
of the last-formed somite, suggesting that
high FGF intermediate FGF low FGF
somites are formed when the FGF level
drops to some low threshold. Retinoic acid
Somites differentiate into particular axial structures depending on their position synthesized and secreted by the somites
along the antero-posterior axis. The easiest somite-derived structures to detect are the forms an opposing gradient that antagonizes
vertebrae. The anterior-most somites contribute to the most posterior region of the the actions of FGF.
skull, those posterior to them form cervical vertebrae, and those more posterior still Adapted from Deschamps, J. and Van Nes, J.:
develop as thoracic vertebrae with ribs. Specification by position has occurred dur- 2005.
ing gastrulation before somite formation itself begins; for example, if unsegmented
somitic mesoderm from the presumptive thoracic region of the chick embryo is grafted
to replace the presumptive mesoderm of the neck region, it will still form thoracic ver-
tebrae with ribs (Fig. 5.22). How then is the pre-somitic mesoderm patterned so that
somites acquire their identity and form particular vertebrae?

5.6 Identity of somites along the antero-posterior axis is specified by Hox

gene expression
The antero-posterior patterning of the mesoderm is most clearly seen in the differences
in the vertebrae, each individual vertebra having well defined anatomical characteristics

Pre-somitic mesoderm from a thoracic vertebrae-forming Skeleton of recipient embryo at 9 days with
region is transplanted from a stage 10 chick embryo thoracic vertebrae developing from donor
into the cervical region of a stage 7 embryo tissue transplanted to cervical region

cervical vertebrae
vertebrae developing
from donor
tissue

thoracic
vertebrae
Fig. 5.22 The pre-somitic mesoderm
has acquired a positional identity before
somite formation. Pre-somitic mesoderm
that will give rise to thoracic vertebrae is
grafted to an anterior region of a younger
embryo that would normally develop into
cervical vertebrae. The grafted mesoderm
develops according to its original position and
forms ribs in the cervical region.
192 Chapter 5 Vertebrate development III: patterning the early nervous system and the somites

Box 5D Retinoic acid: a small-molecule intercellular signal

Retinoic acid is a small diffusible

Retinoic acid synthetic and signaling pathways
molecule derived from vitamin A
(retinol). Animals obtain vitamin A Extracellular space RAR RXR Nucleus
through their diet, and it has been retinol
known for a long time through retinol-binding protein
work with pigs that vitamin A defi- expression of
retinol target genes
ciency can lead to developmental transport
receptor
defects. It subsequently emerged
that an excess of vitamin A also Cytoplasm
leads to developmental defects.
Sadly, babies with craniofacial Cytoplasm
intercellular
malformations have been born to signaling
women who had been treated for
retinaldehyde retinoic
severe skin diseases with synthetic acid
vitamin A derivatives, not knowing
they were pregnant. CRBP
ADH or RALDH CYP26
Retinol is present at high levels­ RDHIO RDH retinoic acid
metabolites
in chicken eggs and in mammalian
embryos is supplied by the mater-
nal circulation. Retinol binds to
retinol-binding proteins in the extracellular space and is trans- is then metabolized to retinoic acid by retinaldehyde dehydro-
ferred into cells by the transport receptor Simulated by retinoic genases (RALDH). There are three vertebrate retinaldehyde
acid 6 (STRA6). Once inside the cell, retinol is bound by an intra- dehydrogenases, encoded by the genes Raldh1, 2, and 3, and
cellular retinol-binding protein (CRBP). Retinoic acid is produced expression of these genes is tightly regulated both temporally
in a two-step reaction (see figure, above), in which retinol is and spatially in vertebrate embryos to ensure that retinoic acid
first metabolized to retinaldehyde by retinol dehydrogenases is produced at very precise locations and specific times during
(mainly RDH10) or alcohol dehydrogenase (ADH); retinaldehyde development.

depending on its location along the axis. The most anterior vertebrae are specialized for
attachment and articulation of the skull, while the cervical vertebrae of the neck are fol-
lowed by the rib-bearing thoracic vertebrae and then those of the lumbar region, which
do not bear ribs, and finally, those of the sacral and caudal regions. ­Patterning of the
skeleton along the body axis is based on the mesodermal cells acquiring a positional
value that reflects their position along the axis and so determines their subsequent
development. Mesodermal cells that will form thoracic vertebrae, for example, have
different positional values from those that will form cervical vertebrae.
Patterning along the antero-posterior axis in all animals involves the expression of
genes that specify positional identity along the axis. The concept of positional identity,
or positional value, has important implications for developmental strategy; it implies
that a cell or a group of cells in the embryo acquires a unique state related to its posi-
tion at a given time, and that this determines its later development (see Section 1.15).
Homeotic genes that specify positional identity along the antero-posterior axis were
originally identified in Drosophila (see Chapter 2) and it was later discovered that the
related Hox genes are involved in patterning the antero-posterior axis in vertebrates.
The Hox genes are members of the large family of homeobox genes that are involved
in many aspects of development (Box 5E, p. 194). As we shall see later, patterning of
vertebrate embryos along the antero-posterior axis by Hox genes and other homeobox
genes is not confined to mesodermal structures; the hindbrain, for example, is also
divided into distinct regions by characteristic patterns of Hox gene expression.
 Somite formation and antero-posterior patterning 193

Retinoic acid is broken down by enzymes of the P450 cyto- Among the genes with a retinoic-
chrome system, mainly by the subfamily of CYP26 enzymes acid-response element is the gene
which, like the RALDHs, are expressed in very precise patterns for RARβ itself, and so one direct
in the embryo. Intricate homeostatic buffering mechanisms ­response to retinoic acid is a switch-
involving regulation of expression of the genes encoding retinoic-­ ing on of the RARβ gene. Expres-
acid-­metabolizing enzymes and end-product feedback serve to sion of RARβ can therefore be used
maintain appropriate levels of retinoic acid. If retinoic acid lev- as a reporter of retinoic acid signal-
els increase, Raldh gene expression is reduced and expression ing and transgenic mice have been
of Cyp26 genes is increased; when retinoic acid levels fall the made in which the reporter gene
reverse happens. lacZ (encoding β-galactosidase, see
Retinoic acid acting as an intercellular signal is shown in the Box 1D, p. 20) is hooked up to the
figure, although it probably also acts as a signal within the RARβ retinoic-acid-response element. β-Galactosidase can then
cells that produce it. Retinoic acid is lipid-soluble and so can be revealed in embryos and tissues by a histochemical reaction
pass through cell membranes. Inside a cell it is bound by pro- which gives a blue color, showing which cells were responding
teins that transport it to the nucleus, where it binds receptors to retinoic acid signaling. The photograph above shows an E9.5
that act directly as transcriptional regulators. Mammals and mouse embryo.
birds have three retinoic acid receptor (RAR) proteins—α , β and Other important developmental signals that act through intra-
γ—which act as dimers in combination with the retinoid X recep- cellular receptors of the same superfamily as the RAR and RXR
tor proteins (RXR; also α , β and γ). In the absence of retinoic proteins are the steroid hormones (such as estrogen and tes-
acid, the heterodimeric RAR:RXR receptors are bound to regula- tosterone, discussed in Chapters 9 and 13) and the thyroid hor-
tory DNA sequences known as retinoic-acid-response elements mones (thyroxine and tri-iodothyroxine; discussed in ­Chapter 13).
(RAREs) and repress the associated genes. Binding of retinoic Like retinoic acid these molecules are lipid soluble; they diffuse
acid to its nuclear receptor leads to activation of the gene asso- through the plasma membrane unaided and bind to intracellu-
ciated with the response element. Gene activation by retinoic lar receptor proteins. The ligand–receptor ­complex is then able
acid can involve not only the direct initiation of transcription but to act as a transcriptional regulator, binding directly to specific
also the recruitment of co-activator proteins that induce chro- response elements in the DNA to activate (or in some cases
matin remodeling, which enables the gene to be expressed (see repress) transcription of a number of genes ­simultaneously.
Box 5A, p. 182). Photograph courtesy of Pascal Dolle.

The Hox genes are the most striking example of a widespread conservation of devel-
opmental genes in animals. Although it was widely believed that there are common
mechanisms underlying the development of all animals, it seemed at first ­unlikely How patterns of gene expression can specify the
that genes involved in specifying the positional identity of the segments of Drosophila identity of four distinct regions: W, X, Y, and Z
would specify positional identity in vertebrates. However, the strategy of comparing
gene I gene II gene III gene IV
genes by sequence homology proved extremely successful in identifying vertebrate
Hox genes related to the homeotic genes of Drosophila. Numerous other genes first
identified in Drosophila, in which the genetic basis of development is far better under-
gene I
stood than in any other animal, have also proved to have counterparts involved in
gene II
development in vertebrates.
gene III
Most vertebrates have four separate clusters of Hox genes that are thought to have
arisen by duplications of the clusters themselves during vertebrate evolution, and gene IV

some of the Hox genes within each cluster may also have arisen by duplication (see
Box 5E, p. 194). The zebrafish is unusual in having seven Hox gene clusters, as a W X Y Z Tissue
result of further duplication. A particular feature of Hox gene expression in both Head Tail
­insects and vertebrates is that the genes in the clusters are expressed in a temporal
and spatial order that reflects their order on the chromosome. This is a unique feature
Fig. 5.23 Gene activity can provide
in development, as it is the only known case where a spatial arrangement of genes on positional values. The model shows how the
a chromosome corresponds to a spatial pattern in the embryo. pattern of gene expression along a tissue can
A simple idealized model illustrates the key features by which a Hox gene cluster specify the distinct regions W, X, Y, and Z. For
could record positional identity by a combinatorial mechanism. Consider four genes—I, example, only gene I is expressed in region W
II, III, and IV—arranged along a chromosome in that order (Fig. 5.23). The genes are but all four genes are expressed in region Z.
194 Chapter 5 Vertebrate development III: patterning the early nervous system and the somites

Box 5E The Hox genes

The proteins encoded by the Hox

genes of vertebrates belong to
Drosophila
a large group of gene-regulatory
A P
proteins that all contain a simi-
lar DNA-binding region of around
60 amino acids. This region, known
as the ­homeodomain, contains a
helix-turn-helix DNA-binding motif bcd
that is characteristic of many 3’ lab pb zen Dfd Scr ftz Antp Ubx abd-A Abd-B 5’
DNA-­binding proteins. The home­ Antennapedia complex bithorax complex
odomain is encoded by a DNA
motif of around 180 base pairs
termed the homeobox, a name Mouse
that came originally from the fact Hoxa, chromosome 6
that this gene family was dis- 3’ a1 a2 a3 a4 a5 a6 a7 a9 a10 a11 a13 5’
covered through mutations that
produce a homeotic transformation— Hoxb, chromosome 11
a mutation in which one structure b1 b2 b3 b4 b5 b6 b7 b8 b9 b13
replaces another. Hoxc, chromosome 15
Clusters of homeotic genes invol­
c4 c5 c6 c8 c9 c10 c11 c12 c13
ved in specifying segment identity
were first discovered in Drosophila. Hoxd, chromosome 2
For example, in one homeotic d1 d3 d4 d8 d9 d10 d11 d12 d13
mutant in Drosophila, a segment
in the fly’s body that does not nor-
mally bear wings is transformed
to resemble the adjacent segment
that does bear wings, resulting in a A P
fly with four wings instead of two.
The Hox genes in Drosophila
are organized into two distinct
gene complexes, the Antenna-
pedia complex and the bithorax complex (known collectively in the ­different ­clusters (e.g. Hoxa4, Hoxb4, Hoxc4, Hoxd4) are
as the HOM-C cluster). In vertebrates, the Hox genes are known as a paralogous subgroup. In the mouse and human
­organized into clusters, known as the Hox complexes, and there are 13 paralogous subgroups.
the homeoboxes of the genes are related in DNA sequence The Hox genes and their role in development are of ancient
to the ­homeoboxes of genes of the two gene complexes in origin. The mouse and frog genes are similar to each other and to
Drosophila. In each Hox cluster the order of the genes from those of Drosophila, both in their coding sequences and in their
3′to 5′in the DNA is the order in which they are expressed order on the chromosome. In both Drosophila and vertebrates,
along the antero-posterior axis and specify positional identity. these homeotic genes are involved in specifying regional iden-
In the mouse, there are four unlinked Hox complexes, des- tity along the antero-posterior axis. Many other genes contain a
ignated Hoxa, Hoxb, Hoxc, and Hoxd (originally called Hox1, homeobox but they do not, however, belong to a homeotic com-
Hox2, Hox3, and Hox4), located on chromosomes 6, 11, 15, plex, nor are they involved in homeotic transformations. Other
and 2, respectively (see figure; on chromosomes 7, 17, 12, subfamilies of homeobox genes in vertebrates include the Pax
and 2 in the human genome). All Hox genes thus resemble genes, which contain a homeobox typical of the Drosophila gene
each other to some extent; the homology is most marked paired. All these genes encode transcription factors with various
within the homeobox and less marked in sequences outside it. functions in development and cell ­differentiation.
Genes that have arisen by duplication and divergence within a Illustration after Veraksa et al.: 2000.
­species are known as ­paralogs, and the corresponding genes
 Somite formation and antero-posterior patterning 195

expressed in a corresponding order along the antero-posterior axis of a tissue. Thus,

gene I is expressed throughout the tissue with its anterior boundary at the anterior end.
Gene II has its anterior boundary in a more posterior position and expression continues
posteriorly. The same principles apply to the two other genes. This pattern of expres-
sion defines four distinct regions, coded for by the expression of different combinations
of genes. If the amount of gene product is varied within each expression domain, for
example by interactions between the genes, many more regions can be specified.
The role of the Hox genes in vertebrate axial patterning has been best studied in the
mouse, because it is possible to either knock out particular Hox genes or alter their
expression in transgenic mice (see Section 3.9 and Box 3B, p. 118). As in all verte-
brates, the Hox genes start to be expressed in mesoderm cells of mouse embryos at an
early stage of gastrulation when these cells begin to move internally away from the
primitive streak and towards the anterior. The ‘anterior’ Hox genes are expressed first.
More posterior Hox genes are expressed as gastrulation proceeds and clearly defined
patterns of Hox gene expression are most easily seen in the mesoderm after somite
formation and the neural tube after neurulation.
Figure 5.24 shows the expression patterns of three Hox genes in mouse embryos at
day E9.5. Typically, the expression of each gene is characterized by a sharp anterior
border and extends posteriorly. Although there is considerable overlap in the spatial
pattern of expression of Hox genes, particularly of neighboring genes in a cluster,
almost every region in the mesoderm along the antero-posterior axis is characterized
by a specific combination of expressed Hox genes, as shown by a summary picture
of the anterior borders of expression of Hox genes in mouse embryonic mesoderm
(Fig. 5.25). For example, the most anterior somites are characterized by expression of
Hoxa1 and Hoxb1 and no other Hox genes are expressed in this region. By contrast, all
Hox genes are expressed in the more posterior regions. The Hox genes can therefore
provide a code for regional identity. The spatial and temporal order of expression is
similar in the mesoderm and the ectodermally derived nervous system, but the posi-
tions of the boundaries between regions of gene expression in the two germ layers
do not always correspond. The most anterior region of the body in which Hox genes
are expressed is the hindbrain; other homeobox genes such as Emx and Otx, but not
Hox genes, are expressed in more anterior regions of the vertebrate body—midbrain,
forebrain, and the anterior part of the head.
If we focus on just the Hoxa complex of genes, we see that the most anterior bor-
der of expression in the mesoderm is that of Hoxa1 in the posterior head mesoderm,
while Hoxa11, a very posterior gene in the Hoxa complex, has its anterior border of
expression in the sacral region (see Fig. 5.25). This exceptional correspondence, or
co-linearity, between the order of the genes on the chromosome and their order of
spatial and temporal expression along the antero-posterior axis is typical of all the Hox

Hoxb1 Hoxb4 Hoxb9

Fig. 5.24 Hox gene expression in the
b1 b2 b3 b4 b5 b6 b7 b8 b9 b13
mouse embryo after neurulation. The
three panels show lateral views of 9.5 dpc
mouse embryos stained for expression of a
LacZ reporter (blue) linked to promoters for
the Hoxb1, Hoxb4, and Hoxb9 genes. These
genes are expressed in the neural tube and
the mesoderm. The arrowheads indicate the
anterior boundary of expression of each gene
within the neural tube. The position of the
three genes within the Hoxb gene complex is
indicated (inset). Scale bar = 0.5 mm.
Photographs courtesy of A. Gould.
196 Chapter 5 Vertebrate development III: patterning the early nervous system and the somites

Anterior Vertebral regions Posterior

cervical thoracic lumbar sacral caudal

Hox genes

b1
a1
a3
d3
d4
b4
a4
b5
a5
c5
c6
Fig. 5.25 A summary of Hox gene a6
expression along the antero-posterior a7
b9
axis of the mouse mesoderm. The anterior
b7
border of expression of each gene is shown Anterior c8
by the red blocks. Expression usually margins
of expression c9
extends backwards some distance but the d8
posterior margin of expression may be poorly a10
defined. The pattern of Hox gene expression d9
specifies the identity of the tissues at d10
d11
different positions. This figure represents an
a11
overall picture of Hox gene expression and d12
is not a snapshot of gene expression at any d13
particular time.

clusters. The genes of each Hox complex are expressed in an orderly sequence, with
the gene lying most 3′in the cluster being expressed the earliest and in the most ante-
rior position. The correct expression of the Hox genes is dependent on their position in
Fig. 5.26 Patterns of Hox gene
the cluster, and anterior genes (3′) are expressed before more posterior (5′) genes.
expression in the mesoderm of chick and
mouse embryos, and their relation to
Support for the idea that the Hox genes are involved in controlling regional iden-
regionalization. The pattern of Hox gene tity comes from comparing their patterns of expression in mouse and chick with the
expression varies along the antero-posterior well defined anatomical regions—cervical, thoracic, sacral, and lumbar (Fig. 5.26).
axis. The vertebrae (squares) of the spinal Hox gene expression corresponds well with the different regions. For example, even
column are derived from the embryonic though the number of cervical vertebrae in birds (14) is twice that of mammals, the
somites (circles), 40 of which are shown anterior boundaries of Hoxc5 and Hoxc6 gene expression in both chick and mouse
here. The vertebrae have characteristic lie on either side of the cervical/thoracic boundary. A correspondence between Hox
shapes in each of the five regions along the gene expression and region is also similarly conserved among vertebrates at other
antero-posterior axis: cervical (C), thoracic (T),
anatomical boundaries.
lumbar (L), sacral (S), and caudal (Ca). Which
It must be emphasized that the summary picture of Hox gene expression given in
somites form which vertebrae differs in chick
Fig. 5.25 does not represent a ‘snapshot’ of expression at a particular time but rather
and mouse. For example, thoracic vertebrae
start at somite 19 in the chick, but at somite
12 in the mouse. Despite this difference, Hox genes
a4 a9 a10 a11
the transition from one region to another b4 b9
corresponds to a similar pattern of Hox gene c4 c5 c6 c8 c9 c10 c11
d4 d9 d10 d11d12 mesoderm
expression in chick and mouse, such that the C1 T1 L1 S1 Ca1
anterior boundaries of expression of Hoxc5 Chick vertebrae
5 19 26 30 39
and Hoxc6 lie on either side of the transition somites
5 12 25 31 35
from cervical to thoracic vertebrae (light blue
Mouse vertebrae
to mid blue) in both species. Similarly, the C1 T1 L1 S1 Ca1
anterior boundaries of Hoxd9 and Hoxd10 a4 a9
b4 b7
are at the transition between the lumbar and c4 c5 c6 c8 c9 mesoderm
sacral regions (green to dark blue) in both d8 d9 d10 d11d12
chick and mouse. Hox genes
C=cervical T=thoracic L=lumbar S=sacral Ca=caudal
After Burke, A.C.: 1995.
 Somite formation and antero-posterior patterning 197

an integrated picture of the overall pattern of expression. Some genes are switched on
early and expression is then reduced, whereas others are switched on considerably
later; the most posterior Hox genes, such as Hoxd12 and Hoxd13, are expressed in
the post-anal tail. Moreover, this summary picture reflects only the general ­expression
of the genes in any particular region; not all Hox genes expressed in a region are
expressed in all the cells. Nevertheless, the overall pattern suggests that the combina-
tion of Hox genes expressed provides positional identity. In the cervical region, for
example, each somite, and thus each vertebra, could be specified by a unique pattern
of Hox gene expression or Hox code.
As we saw in Fig. 5.22, grafting experiments show that the character of the somites
is already determined in the pre-somitic mesoderm and that somitic tissue transplanted
to other levels along the axis retains its original identity. As would be expected if the
Hox code specifies vertebral identity, the transplanted pre-somitic mesoderm also
retains its original pattern of Hox gene expression.

5.7 Deletion or overexpression of Hox genes causes changes in axial patterning

If the Hox genes do provide positional values that determine a region’s subsequent
development, then one would expect morphological changes if their pattern of
­expression is altered. This is indeed the case. To investigate how Hox genes control
patterning, their expression can be abolished by mutation, or they can be expressed
in abnormal positions. Hox gene expression can be eliminated from the developing
mouse embryo by gene knock-out (see Section 3.9 and Box 3B, p. 118). Experiments
along these lines indicate that the absence of given Hox genes affects patterning in a
way that agrees with the idea that they are normally providing cells with positional
identity. In some cases, however, knocking out an individual Hox gene produces little
effect, or affects only one tissue, and more substantial changes in patterning are only
seen when two or more genes, particularly paralogous genes from different Hox
complexes (see Box 5E, p. 194), are knocked out together.
Mice in which Hoxd3 is deleted show structural defects in the first and second cervi-
cal vertebra, where this gene is normally strongly expressed (see Fig. 5.25), and in the
sternum (breastbone). More posterior structures, where the inactivated gene is also
normally expressed, show no evident defects, however. This observation illustrates a
general principle of Hox gene expression, which is that more posteriorly expressed
Hox genes tend to over-ride the function of more anterior Hox genes whenever they
are coexpressed. This phenomenon is known as posterior dominance or posterior
prevalence. Other Hox genes are expressed together with Hoxd3 in the lumbar and
sacral regions of the body and this explains why there are no posterior defects when
Hoxd3 is deleted. Another general point illustrated by the Hoxd3 knock-out is that
the effects of a Hox gene knock-out can be tissue specific, in that only certain tis-
sues in which the Hox gene is normally expressed are affected, whereas other tissues
at the same position along the antero-posterior axis appear normal. For example,
­although Hoxd3 is expressed in neural tube, branchial arches, and paraxial mesoderm
at the same antero-posterior level in the body, only the vertebrae are defective in a
Hoxd3 knock-out. The apparent absence of an effect in such cases may be due to
redundancy, with paralogous genes from another complex being able to compensate.
Hoxa3 is expressed strongly in the same tissues as Hoxd3 in the cervical region, but
when Hoxa3 alone is knocked out there are no defects in cervical vertebrae; instead
other defects are observed, including reductions in cartilage elements derived from
the second branchial arch. When both Hoxd3 and Hoxa3 are knocked out together,
much more severe defects are seen in both vertebrae and branchial arch derivatives,
suggesting that the two genes normally function together in these tissues.
Loss of Hox gene function often results in homeotic transformation—the conver-
sion of one body part into another. This is the case in the Hoxd3 knock-out mouse we
198 Chapter 5 Vertebrate development III: patterning the early nervous system and the somites

have just considered. Detailed analysis of the cervical region in these mice ­revealed
Normal
that the atlas vertebra (the first cervical vertebra) appears to be transformed into the
basal occipital bone at the base of the skull, and that the axis vertebra (the second
cervical vertebra) has acquired morphological features resembling the atlas. Thus, in
the absence of Hoxd3 expression, cells acquire a more anterior positional value and
develop into more anterior structures. The double knock-out of Hoxa3 and Hoxd3
13th 1st results in a deletion of the atlas. The complete absence of this bone in these double
thoracic lumbar
vertebra vertebra knock-outs suggests that one target of Hox gene action is the cell proliferation ­required
to build such a structure from the somite cells. Unfortunately, very few direct gene
targets transcriptionally regulated by Hox proteins have yet been identified.
Another example of homeotic transformation is seen in Hoxc8-knock-out mice. In nor-
mal embryos, Hoxc8 is expressed in the thoracic and more posterior regions of the embryo
from late gastrulation onward (see Fig. 5.25). Mice homozygous for deletion of Hoxc8
die within a few days of birth, and have abnormalities in patterning between the seventh
Hoxc8 deleted thoracic vertebra and the first lumbar vertebra. The most obvious homeotic transforma-
tions are the attachment of an eighth pair of ribs to the sternum and the development of
a 14th pair of ribs on the first lumbar vertebra (Fig. 5.27). Thus, as with the knock-outs
discussed above, the absence of Hoxc8 has led to some of the cells that would normally
express it acquiring a more anterior positional value, and developing accordingly.
Some homeotic transformations only occur if all the paralogous genes are knocked
out together. In mice, in the absence of all the Hox10 paralogous subgroup genes, there
are no lumbar vertebrae and there are ribs on all posterior vertebrae; in the absence
of all the Hox11 paralogous subgroup genes several sacral vertebrae become lumbar.
These homeotic transformations do not occur if only some members of the paralogous
subgroup are mutated, again suggesting redundancy of Hox gene function.
In contrast to the effects of knocking out Hox genes, the ectopic expression of Hox
genes in anterior regions that normally do not express them can result in transfor-
mations of anterior structures into structures that are normally more posterior. For
Fig. 5.27 Homeotic transformation of
example, when Hoxa7, whose normal anterior border of expression is in the thoracic
vertebrae due to deletion of Hoxc8 in the
region (see Fig. 5.25), is expressed throughout the whole antero-posterior axis, the
mouse. In loss-of-function homozygous
basal occipital bone of the skull is transformed into a pro-atlas structure, normally the
mutants of Hoxc8, the first lumbar vertebra
(yellow) is transformed into a rib-bearing next more posterior skeletal structure.
‘thoracic’ vertebra. The mutation has resulted
in the transformation of the lumbar vertebra
5.8 Hox gene expression is activated in an anterior to posterior pattern
into a more anterior structure.
In all vertebrates, the Hox genes begin to be expressed at an early stage of gas-
trulation, when mesodermal cells begin their gastrulation movements. The genes
­expressed most anteriorly, which correspond to those at the 3′end of a cluster, are
expressed first. If an ‘early’ Hoxd gene is relocated to the 5′end of the Hoxd complex,
for example, its spatial and temporal expression pattern then resembles that of the
neighboring Hoxd13 (see Box 5E, p. 194). This shows that the structure of the Hox
complex is crucial in determining the pattern of Hox gene expression.
Unlike the situation in Drosophila, where the activation of Hox genes clearly ­depends
on factors unequally distributed along the antero-posterior axis (see Chapter 2), the
mechanism of Hox gene activation in vertebrates is more complex and less well
­understood. One way the antero-posterior pattern of Hox gene expression might be
established in the somitic mesoderm is through linking gene activation to the time spent
by cells in the somite stem-cell region (see Fig. 5.20). In the chick and the mouse, the
whole of the mesoderm alongside the dorsal midline derives from a small population
of stem cells located in the anterior primitive streak and later in the tailbud. It has been
shown that genes of the Hoxb cluster follow a strict temporal order of co-linear activa-
tion in this stem-cell region, the 3′genes being activated before the 5′ones. Expression
of genes that are activated in the stem-cell region is maintained when the cells leave
to form the pre-somitic mesoderm, and in this way the temporal pattern of Hox gene
 Somite formation and antero-posterior patterning 199

activation can be converted into positional information; the cells expressing different
Hox genes are generated progressively along the antero-posterior axis as structures are
laid down in an antero-posterior sequence. This hypothesis is similar to one proposed
for specifying position along the proximo-distal axis of the vertebrate limb bud (see
Chapter 11).
A gradient in retinoic acid from anterior to posterior along the trunk region of the
mouse embryo (see Fig. 5.21) could also be involved in activating Hox genes in an
antero-posterior pattern. Retinoic acid has been shown experimentally to alter the
­expression of Hox genes. More recently, genes encoding retinoic acid receptors and
other proteins of the retinoid signaling pathway have been knocked out in mouse
embryos, and this results in homeotic transformations of the axial skeleton. Mouse
­embryos lacking the gene Cyp26a1, which encodes an enzyme that degrades retinoic
acid (see Box 5D, pp. 192–193), have many severe defects, including posterior ­homeotic
transformations of cervical vertebrae into vertebrae carrying ribs. These transforma-
tions occur in regions where retinoic acid signaling is higher than normal and where
anterior expansion of Hoxb4 expression has also been seen. A role for retinoic acid
in Hox gene activation is supported by studies in zebrafish. The zebrafish mutation
giraffe, which affects the Cyp26 gene, has anteriorly extended Hox gene ­expression
and vertebral transformations. From vertebrate genome sequences, predicted genes
for microRNAs (miRNAs) have been detected within Hox clusters and it has been
proposed that the encoded miRNAs may be involved in post-transcriptional regulation
of Hox protein expression (see Box 6B, p. 228, for how miRNAs work). More of the
predicted target genes of a given miRNA tend to lie on the 3′side of the miRNA gene
rather than the 5′side. Given this arrangement, the Hox miRNAs may help to repress
the function of more anterior Hox genes, and this could explain the general principle
of posterior prevalence—that more posteriorly expressed Hox genes tend to over-ride
the function of more anterior Hox genes wherever they are coexpressed.
We have concentrated here on the expression of Hox genes in the mesoderm, but
they are also expressed in a patterned way in the neural tube after its induction, and
we shall look at this aspect of antero-posterior regionalization later in the chapter.

5.9 The fate of somite cells is determined by signals from the adjacent tissues

We shall now return to the individual somite and consider how it is patterned to give
rise to different tissues. This patterning process is quite independent of the antero- Delta1 expression
posterior patterning of the whole pre-somitic mesoderm by Hox gene expression.
A P
Individual somites are subdivided into anterior and posterior halves with different Notch-Delta
signaling
properties, and this subdivision is linked to the segmentation clock. In the mouse, pre-somitic mesoderm
the transcription factor Mesp2 is expressed in a somite-wide stripe in the pre-somitic Mesp2 expression
mesoderm that will give rise to the next somite (S0), and its expression then shrinks
Mesp2
to just the anterior part of the somite (Fig. 5.28). Mesp2 activates a cascade of gene expression
expression that leads to its expression being maintained in the anterior part of the somite
somite. The restricted expression of Mesp2 is thought to be a consequence of Notch–
Delta signaling. The differences between the two halves of the somite are later used by Fig. 5.28 Division of somites into anterior
neural crest cells and motor nerves as guidance cues to generate the periodic arrange- and posterior halves. Positive and negative
ment of spinal nerves and ganglia (see Chapter 8). The subdivision is also important feedback loops between Delta1 and Mesp2
(right) are essential for dividing the somite
for vertebra formation, as each vertebra derives from the anterior half of one somite
into an anterior half expressing Mesp2 and
and the posterior half of the preceding one, a process known as ‘resegmentation’.
a posterior half expressing Delta1 (left),
Individual somites are also patterned along their medio-lateral axis. But before we
shown in dorsal views of the tail regions
consider the individual somite, we will first look at how the mesoderm as a whole is of E9.5 mouse embryos. Mesp2 is initially
patterned medio-laterally, such that paraxial mesoderm is formed on either side of the expressed in a somite-wide stripe in
notochord (medially) and lateral plate mesoderm further away (laterally). In chick pre-somitic mesoderm and then as the
embryos, the mesoderm cells along the medio-lateral axis of the somite originate from somite forms Mesp expression is maintained
different sites along the primitive streak and are brought together during gastrulation. in the anterior half.
200 Chapter 5 Vertebrate development III: patterning the early nervous system and the somites

2-day chick embryo 3-day embryo 4-day embryo

Dorsal
epidermis dermomyotome myotome dermomyotome
neural
tube

Medial Lateral
migrating
muscle
somite cells
sclerotome
notochord sclerotome
Ventral
sclerotome dermomyotome

Fig. 5.29 The fate map of a somite in The cells of the medial part of the somite come from the primitive streak close to
the chick embryo. The ventral medial Hensen’s node, whereas the cells in the lateral part come from slightly more posterior
quadrant (blue) gives rise to the sclerotome in the primitive streak, and cells that form the lateral plate mesoderm originate even
cells, which migrate to form the cartilage of more posteriorly (see Fig. 4.43). The level of BMP-4 signaling mediates medio-lateral
the vertebrae. The rest of the somite—the patterning of the mesoderm; it is high in the lateral plate mesoderm and lower in pre-
dermomyotome—forms the myotome, which somitic mesoderm, the decrease being due to the expression of the BMP antagonist
gives rise to all the trunk muscles, while the
Noggin in the pre-somitic mesoderm. The importance of low-level BMP signaling for
epithelial dermomyotome region gives rise to
somite formation is shown by the fact that inhibition of BMP by Noggin is sufficient
the dermis of the skin. The dermomyotome
also gives rise to muscle cells that migrate
to convert lateral plate mesoderm into somites.
into the limb buds. The somites of the vertebrate embryo give rise to major axial structures: the car-
tilage cells of the embryonic axial skeleton—the vertebrae and ribs; all the skeletal
muscles, including those of the limbs; and much of the dermis. The fate maps for par-
ticular somites have been made by grafting somites from a quail into a corresponding
position in a chick embryo at a similar stage of development and following the fate of
the quail cells. More recently, transgenic chickens expressing GFP have been created
and grafting GFP-expressing chick tissues into normal chick embryos will allow long-
term fate mapping experiments in the same avian species.
Cells located in the dorsal and lateral regions of a newly formed somite make up the
dermomyotome, which expresses the gene Pax3, a homeobox-containing gene of the
paired family (see Box 5E, p. 194). The dermomyotome gives rise to the myotome,
from which muscle cells originate, and it later forms the epithelial sheet over the
Dorsal myotome from which the dermis originates. The dermomyotome also contains cells
grafted notochord dermomyotome
that contribute to the vasculature. Cells from the medial region of the myotome form
mainly axial and back (epaxial) muscles, whereas lateral myotome cells migrate to
give rise to abdominal and limb (hypaxial) muscles. All committed muscle precursors
express muscle-specific transcription factors of the MyoD family. The ventral part of
neural tube
the medial somite contains sclerotome cells that express the Pax1 gene and migrate
ventrally to surround the notochord and develop into vertebrae and ribs (Fig. 5.29).
sclerotome sclerotome Which cells will form cartilage, muscle, or dermis is not yet determined at the time
of somite formation. This is clearly shown by experiments in which the dorso-ventral
notochord floor plate
Ventral orientation of newly formed somites is inverted but normal development nevertheless
­ensues. Specification of the fate of somite tissue requires signals from adjacent tissues. In
Fig. 5.30 A signal from the notochord the chick, determination of myotome occurs within hours of somite formation, whereas
induces sclerotome formation. A graft the future sclerotome is only determined later. Both the neural tube and the notochord
of an additional notochord to the dorsal
produce signals that pattern the somite and are required for its future development. If
region of a somite in a 10-somite chick
the notochord and neural tube are removed, the cells in the somites undergo apoptosis;
embryo suppresses the formation of the
dermomyotome from the dorsal portion
neither vertebrae nor axial muscles develop, although limb musculature still does.
of the somite, and instead induces the The role of the notochord in specifying the fate of somitic cells has been shown by
formation of sclerotome, which develops into experiments in the chick, in which an extra notochord is implanted to one side of the
cartilage. The graft also affects the shape of neural tube, adjacent to the somite. This has a dramatic effect on somite differentia-
the neural tube and its dorso-ventral pattern. tion, provided the operation is carried out on unsegmented pre-somitic ­mesoderm.
 Somite formation and antero-posterior patterning 201

When the somite develops, there is an almost complete conversion to cartilage

Dorsal
­precursors (Fig. 5.30), suggesting that the notochord is an inducer of cartilage. In nor- ectoderm
mal development, cartilage-inducing signals come from the notochord and the ventral
region of the neural tube, the floor plate. Other signals come from the dorsal neural
tube and the overlying ectoderm to specify the medial part of the dermomyotome,
while signals from the lateral plate mesoderm are involved in specifying the lateral
dermomyotome
demomyotome (Fig. 5.31).
Some of these signals have been identified. In the chick, both the notochord and the
sclerotome
floor plate of the neural tube express Sonic hedgehog (Shh), a secreted protein that is
a key molecule for positional signaling in a number of developmental situations. We
Ventral
encountered Shh in Chapter 1 as a signal involved in craniofacial development (see
Box 1F, p. 28) and also met it in Chapter 4 with respect to the asymmetry of structures
Wnts BMP-4 Shh
about the midline. We shall meet it again in Chapter 11 (see Box 11C, pp. 424–425 for
the Shh signaling pathway) in connection with limb development, and in Chapter 12
in the dorso-ventral patterning of the neural tube. In somite patterning, high levels Fig. 5.31 The somite is patterned by
of Shh signaling specify the ventral region of the somite and are required for sclero- signals secreted by adjacent tissues. The
tome development. Low levels of Shh signaling, together with signals from the dorsal sclerotome is thought to be specified by a
diffusible signal, Sonic hedgehog protein (Shh)
neural tube and from the overlying non-neural ectoderm, specify the dorsal region
(red and orange arrows), produced by the
of the somite as dermomyotome (see Fig. 5.31). Wnt signals are good candidates for
notochord (red) and the floor plate (orange) of
the dorsal signals. BMP-4 is the signal from the lateral plate mesoderm. Tendons arise
the neural tube. Signals from the dorsal neural
from cells that come from the dorso-lateral domain of the sclerotome and specifically tube and ectoderm such as Wnts (blue arrows)
express the transcription factor Scleraxis. This tendon-progenitor region is induced at would specify the dermomyotome, together
the boundary of the sclerotome and myotome. with lateral signals such as BMP-4 (green
Regulation of expression of the Pax homeobox genes in the somite by signals from arrow) from the lateral plate mesoderm.
the notochord and neural tube seems to be important in determining cell fate. Pax3 After Johnson, R.L.: 1994.
is initially expressed in all cells that will form somites. Its expression is then modu-
lated by signals from BMP-4 and Wnt proteins so that it becomes confined to muscle
precursors. It is then further downregulated in cells that differentiate as the muscles
of the back, but remains switched on in the migrating presumptive muscle cells that
populate the limbs. Mice that lack a functional Pax3 gene—Splotch mutants—lack
limb muscles. In the chick, Pax1 has been implicated in the formation of the scapula,
a key element in the shoulder girdle, part of which is contributed by somites. All
the scapula-forming cells express Pax1. But unlike the Pax1-­expressing cells of the
vertebrae, which are of sclerotomal origin, the blade of the scapula is formed from
dermomyotome cells of chick somites 17–24, while the head of the scapula is derived
from lateral plate mesoderm.

Summary

Somites are blocks of mesodermal tissue that are formed after gastrulation. They form
sequentially in pairs on either side of the notochord, starting adjacent to the posterior
region of the hindbrain at the anterior end of the embryo. The somites give rise to the
vertebrae, to the muscles of the trunk and limbs, and to most of the dermis of the skin.
The pre-somitic mesoderm is patterned with respect to the antero-posterior axis while
cells are in the node before the somites form. The first manifestation of this pattern is
the expression of the Hox genes in the pre-somitic mesoderm. The positional identity
of the somites is specified by the combinatorial expression of genes of the Hox com-
plexes along the antero-posterior axis, from next to the hindbrain to the posterior end
of the embryo, with the order of expression of these genes along the axis correspond-
ing to their order along the chromosome. Mutation or overexpression of a Hox gene
results, in general, in localized defects in the anterior parts of the regions in which the
gene is expressed, and can cause homeotic transformations. We can think of Hox genes
as ­providing positional information that specifies the identity of a region and its later
202 Chapter 5 Vertebrate development III: patterning the early nervous system and the somites

development. They act on downstream gene targets about which we know relatively
little.
Patterning of the mesoderm with respect to the medio-lateral axis and the formation
of somites in the paraxial mesoderm depends on the inhibition of signals that would
otherwise specify lateral plate mesoderm. Individual somites are also patterned by sig-
nals from the notochord, neural tube, and ectoderm, which induce particular regions of
each somite to give rise to muscle, cartilage, or dermis.

The initial regionalization of the vertebrate brain

The developing vertebrate brain becomes divided into three main regions—forebrain
(prosencephalon), midbrain (mesencephalon), and hindbrain (rhombencephalon)—
illustrated here for 2-day and 4.5-day chick embryos (Fig. 5.32). Considerable lateral
expansions develop from the embryonic forebrain and these are the rudiments of the
optic vesicles, from which the eyes will form. The embryonic forebrain gives rise to
the cerebral hemispheres (telencephalon), and the thalamus and the hypothalamus.
The cortex of the cerebral hemispheres is where the highest-level processing centers
for sensory information, motor control, learning, and memory are located, while the
hypothalamus is the brain’s link with the endocrine system, and controls the release
of hormones from the pituitary. The embryonic midbrain gives rise to the tectum,
which is the site of integration and relay centers for signals coming to and from the
hindbrain and also for inputs from the sensory organs, such as the eye (see Chap-
ter 12). The embryonic hindbrain is characterized by a series of seven or eight tran-
sient swellings, known as rhombomeres, and gives rise to the cerebellum, the pons
and the medulla oblongata, which are concerned with the regulation of basic body
activities that are independent of conscious control, such as regulation of muscle tone
and posture (by the cerebellum), and heartbeat, breathing, and blood pressure.

Regionalization of the embryonic brain

Stage HH13 (2 days) Stage HH24 (4.5 days)

prosencephalon Midbrain
Fig. 5.32 Local signaling centers in the
developing vertebrate brain. Lateral view of
PrZLI
the brain of a 2-day chick embryo (HH stage tectum
13; left panel) and of a 4.5-day chick embryo midbrain-
hindbrain
(HH stage 24: right panel). The two signaling mesencephalon prechordal boundary
plate
centers—the isthmus at the midbrain– thalamus
hindbrain boundary (MHB) and the zona ZLI
Forebrain

midbrain- prethalamus
limitans intrathalamica (ZLI) in the forebrain— hindbrain cerebellum
are indicated by dark blue lines. Bi-directional boundary
r1 telencephalon
signaling from these centers (indicated by
black arrows) is responsible for patterning r2
Hindbrain

the adjacent brain regions and involves FGF-8 r3 medulla ventral midline
in the MHB and Sonic hedgehog (Shh) in the rhombencephalon r4 motor nuclei
ZLI. Wnt-1 is also expressed on the anterior r5
side of the MHB. The boundary between notochord notochord
r6
Otx2 expression in the midbrain (violet) and
r7
Gbx2 expression in the anterior hindbrain floor plate
(yellow) is linked to formation of the MHB
signaling center.
Adapted from Kiecker, C., and Lumsden, A.: Fgf8 Wnt1 Shh Gbx2 Otx2
2005.
 The initial regionalization of the vertebrate brain 203

5.10 Local signaling centers pattern the brain along the antero-posterior axis

There are two local organizing regions in the developing midbrain and forebrain; one,
known as the isthmus, is located at the midbrain–hindbrain boundary and regulates
patterning of the posterior midbrain and the anterior hindbrain; the other, known as
the zona limitans intrathalamica, is located in the forebrain and regulates the pat-
terning of forebrain on both the anterior and posterior sides (see Fig. 5.32). The isth-
mus is already a major signaling center in the 2-day chick brain and is still active in
brain of 4.5-day chick embryos, whereas the zona limitans intrathalamica is formed
later and acts as a signaling center in the 4.5-day chick brain.
The signaling properties of the isthmus were first shown by grafting experiments
in 2-day chick embryos. If the isthmus is grafted into the anterior midbrain or pos-
terior forebrain, it converts that tissue into posterior midbrain, whereas if grafted
to the posterior hindbrain, the isthmus converts that tissue into anterior hindbrain
(which develops into the cerebellum). FGF-8 is expressed at the isthmus and is
a good candidate for a bidirectional signal controlling the patterning of posterior
midbrain and anterior hindbrain. Induced expression of FGF-8 in the chick ante-
rior midbrain, for example, results in the expression of genes characteristic of the
posterior midbrain. The zebrafish mutant acerebellar has a mutation that abolishes
expression of FGF-8 almost completely, and as the name suggests, mutant fish lack
the anterior hindbrain (r1, from which the cerebellum derives) and its associated
organizer.
The midbrain–hindbrain organizer region develops at the border between the
expression domains of transcription factors Otx2 (on the midbrain side) and Gbx2
(on the hindbrain side). Initial division into an Otx2-expressing anterior territory and
a Gbx2-expressing posterior territory occurs in the neural plate, and is one of the
earliest patterning events in brain development. Later, in the neural tube, FGF-8 is
expressed on the hindbrain side of the border and Wnt-1 on the midbrain side (see
Fig. 5.32). The other local signaling center, the zona limitans intrathalamica, signals
bi-directionally via Shh, and is essential for formation of the pre-thalamus anteriorly
and the thalamus posteriorly (see Fig. 5.32).

Cranial region of chick embryo (stage HH18, 3 days)

5.11 The hindbrain is segmented into rhombomeres by boundaries
of cell-lineage restriction
rhombomeres midbrain
r1
of hindbrain
Patterning of the posterior region of the vertebrate head and the hindbrain involves
optic
segmentation along the antero-posterior axis. The type of patterning of the nervous r2 vesicle
r3
system does not occur along the spinal cord, where the pattern of dorsal root ganglia r4
r5
r6
and ventral motor nerves at regular intervals—one pair per somite—is imposed by the r7 forebrain
somites. In the chick embryo, three segmented systems can be seen in the posterior r8
head region by 3 days of development (Fig. 5.33). The mesoderm on either side of s2
b1
the notochord is subdivided into somites, the hindbrain (the rhombencephalon) is b2 branchial
somites

divided into eight rhombomeres (r1 to r8), and a series of branchial arches (b1 to b4) b3 arches
b4
that are populated by neural crest cells migrating from the hindbrain have formed.
The branchial arches give rise to the tissues of the head and throat other than the
brain. A similar division of the hindbrain into rhombomeres occurs in the other model
Fig. 5.33 The nervous system in a 3-day
vertebrates.
chick embryo (HH stage 18). At this stage
Development of the hindbrain region of the head involves several interacting com- in development the hindbrain is divided into
ponents. The neural tube produces both the segmentally arranged cranial nerves eight rhombomeres (r1 to r8). The positions
that innervate the face and neck, and also neural crest cells, which give rise both to of three cranial nerves that arise from the
­peripheral nerves and most of the facial skeleton. In addition, the otic vesicle which hindbrain (V, VII, IX) are shown in green; b1 to
develops opposite rhombomeres 5 and 6 gives rise to the ear. The main skeletal ele- b4 are the four branchial arches; b1 gives rise
ments of the head in this region develop from the first three branchial arches. For to the jaws; s = somite.
example, the first arch gives rise to the jaws, while the second arch develops into Adapted from Lumsden, A.: 1991.
204 Chapter 5 Vertebrate development III: patterning the early nervous system and the somites

the bony parts of the ear. This region of the head is a particularly valuable model for
studying patterning along the antero-posterior axis because of the presence of the
numerous different structures ordered along it.
Immediately after the neural tube of the chick embryo closes in this region, the
future hindbrain becomes transversely constricted to define eight rhombomeres (see
Fig. 5.33). The cellular basis for these constrictions is not entirely understood but is
likely to involve differential cell division or changes in cell shape. The boundaries
between rhombomeres subsequently become clearly differentiated from the adjacent
neuroepithelium, with a marked increase in the extracellular space between cells.
Cell-lineage tracing experiments show that once formed, individual rhombomeres are
compartments of cell-lineage restriction; that is, cells and their descendants are con-
fined within the rhombomere in which they originate and do not cross from one side
of the boundary to the other. Tracking the descendants of individual labeled cells
shows that before constrictions become visible, the descendants of a given cell can
form a clone that spans a boundary. After the constrictions appear, however, clones
of labeled cells are confined to a single rhombomere (Fig. 5.34). This lineage restric-
tion suggests that each rhombomere is acting as an independent developmental unit.
In this respect a rhombomere is behaving rather like a compartment, which is an
important feature of insect development, as we saw in Chapter 2, but seems to be rare
in vertebrates. The rhombomeres are no longer visible morphologically after about
4.5 days in the chick embryo (see Fig. 5.32).
Clues to how cells from adjacent rhombomeres might be prevented from mixing
come from the discovery that the cells of odd-numbered rhombomeres seem to share
some adhesive property that is not shared by even-numbered rhombomeres. This was
originally revealed by experiments in which an odd-numbered and an even-numbered
rhombomere from non-adjacent positions were placed together—and a new boundary
formed between them. No new boundary formed when two odd-numbered rhombom-
eres were placed together, however, suggesting that their cells are compatible with
each other. The boundary region itself is unlikely to be acting purely as a mechanical
barrier, as cells from adjacent rhombomeres do not mix even when the boundary

Single neuroepithelial cells labeled at early and late stages and the labeled clones mapped at 2 days

Early-stage labeling Pattern of clones Late-stage labeling

r1

Fig. 5.34 Lineage restriction in

rhombomeres of the embryonic chick
hindbrain. Single cells are injected with a
label (rhodamine-labeled dextran) at an early
stage (left panel) or a later stage (right panel)
of neurulation, and their descendants are
mapped 2 days later. Cells injected before
rhombomere boundaries form give rise to
some clones that span two rhombomeres
(dark red) as well as those that do not
cross boundaries (red). Clones marked after
rhombomere formation never cross the
r7
boundary of the rhombomere that they
originate in (blue).
Adapted from Lumsden, A.: 1991.
 The initial regionalization of the vertebrate brain 205

­ etween them is surgically removed. The finding that particular cell-surface proteins of
b
the ephrin family and their Eph receptors are expressed separately in alternate rhom-
bomeres provides one possible mechanism for preventing cell mixing. Ephrins and
their receptors are transmembrane signaling proteins that interact with each other on
adjacent cells and can both signal to their respective cells. They direct either repulsive
or attractive cell–cell interactions (Box 5F, p. 205). In particular, the EphA4 receptor,
which is strongly expressed in rhombomeres r3 and r5, and its ligand ephrin B2, which
is expressed in r2, r4, and r6, could provide repulsive interactions at each rhombomere
boundary and so prevent cell mixing (see Fig. 5.34). The expression of EphA4 in r3
and r5 is under the control of the transcription factor Krox20, which is itself expressed

Box 5F Eph receptors and their ephrin ligands

Eph receptors and their ephrin ligands are transmembrane pro-

Bidirectional signaling by ephrins
teins that can generate both repulsive and adhesive interactions and their receptors
between cells. Because they are both attached to a cell surface,
they can only act through direct cell–cell contact, like Notch and
Delta. They are best known as contact-dependent guidance mole­
cules for growing axons in the nervous system, especially in the
guidance of axons of retinal neurons to their correct locations in ephrin B2 EphA4
the brain’s visual-processing centers; their role in axon guidance
is discussed in Chapter 12. Among other roles, they act as guid-
ance molecules for migrating neural crest cells, and appear to be
tyrosine
involved in the final stages of somite formation—the separation kinase
of the new somite from the pre-somitic mesoderm. And according domain
to work in mice, Ephs and ephrins are required to correctly set up
r2 r3
the local neuronal circuits in the spinal cord that enable you to
walk with alternate legs, rather than have to hop like a rabbit.
The 14 different Eph receptors in mammals fall into two struc-
tural classes: EphAs (EphA1–EphA8 and EphA10) and EphBs The repulsion commonly mediated by Eph–ephrin pairs is some-
(EphB1–Eph B4 and EphB6). Ephrins similarly can be classified thing of a paradox, given their inherent affinity for each other.
into ephrin As (ephrins A1–A5) and ephrin Bs (ephrins B1–B3). One solution to the paradox was the discovery that repulsion by
The A-type receptors typically bind to ephrin As, and the B-type ephrin As involves cleavage of their ectodomain, which enables
receptors to ephrin Bs, but EphA4 is an exception, being able to cell disengagement. Internalization of Eph–ephrin complexes by
bind both ephrin A and B ligands. Ephrins binding to their Eph endocytosis is another possible solution. Repulsive interactions
receptors can generate a bidirectional signal, in which both the seem generally to require a relatively high level of signaling from
Eph and the ephrin send a signal to their own cell (see figure). the Eph tyrosine kinase. If this is lacking or at a low level, then
Unlike most of the signaling receptors we have encountered so the binding of Eph and ephrin will tend to promote cell adhe-
far, the main target of Eph–ephrin signaling is the actin cytoskel- sion. An example of this has been found in the closure of the
eton, whose rearrangement causes the changes in cell morphol- neural tube in mice, in which a form of EphA7 lacking its kinase
ogy and behavior that underlie repulsion. domain is produced by the cells of the neural folds as a result
Ephs are receptor tyrosine kinases, with an intracellular of alternative RNA splicing. The neural fold cells also express
­kinase domain, whereas ephrins signal by associating with ephrin A5, and the adhesive interaction between EphA7 and
and activating a cytoplasmic kinase. Ephrin–Eph contact is best ephrin A5 on opposing folds is essential for neural tube closure.
known for producing repulsion between the interacting cells, as Mouse mutants that lack ephrin A5 have neural tube defects
occurs at the interface between different rhombomeres (see resembling the human condition of anencephaly, in which the
Section 5.11), but Eph–ephrin interactions can also result in cranial neural tube fails to close, the forebrain does not develop,
­attraction and ­adhesion. Whether a particular Eph–ephrin pairing and the ­fetus is usually stillborn. In tissues other than the neural
repels or attracts is not straightforward, and seems to depend on tube, in which EphA7 is produced with an intact tyrosine kinase
a range of different factors, including the cell types involved, the ­domain, its interaction with ephrin A5 causes repulsion between
numbers of receptors present, and their degree of clustering. the ­interacting cells.
206 Chapter 5 Vertebrate development III: patterning the early nervous system and the somites

in two stripes in the neural plate that become r3 and r5. In turn, separate sets of tran-
scription factors have been shown to activate Krox20 in r3 and r5, respectively.
The division of the hindbrain into rhombomeres has functional significance. A sim-
ilar pattern of neuronal cell types develops along the antero-posterior axis of each
rhombomere, and at the same time, each rhombomere acquires a unique identity in
respect of the particular cranial nerves and other tissues it gives rise to. The delimita-
tion and future development of individual rhombomeres is under the control of Hox
and other transcription factors, and we shall next consider how the Hox genes provide
positional information and identity both to the developing hindbrain itself and to the
neural crest cells that migrate away from it.

5.12 Hox genes provide positional information in the developing hindbrain

Visible constriction of the hindbrain into rhombomeres is preceded by the patterned

expression of transcription factor genes, including Krox20 and Hox genes, that ­initially
divides the neural tube into the prospective rhombomere segments. Hox gene expres-
sion provides at least part of molecular basis for the identities of both the rhombom-
eres and the neural crest at different positions in the hindbrain. No Hox genes are
expressed in the most anterior part of the head or in r1, but Hox genes from four
paralogous groups (1, 2, 3, and 4) are expressed in the rest of the mouse embryonic
hindbrain in a well-defined pattern, which closely correlates with the segmental pat-
tern (Fig. 5.35). The most anterior Hox gene expressed at this stage is Hoxa2, which
is expressed in r2. In general, the different paralogous groups have different anterior
margins of expression. For example, the anterior border of Hoxb2 expression is at the
boundary between r2 and r3, whereas the anterior margin of expression of Hoxb3 is
at the border between r4 and r5 (Fig. 5.36). Transplantation of rhombomeres from an
anterior to a more posterior position alters the pattern of Hox gene expression so that
it becomes the same as that normally expressed at the new location.
Studies at the molecular level have provided some indication to how Hox gene
expression is controlled in the individual rhombomeres. For example, although the
Hoxb2 gene is expressed in the three contiguous rhombomeres r3, r4, and r5, its
­expression in r3 and r5 is controlled quite independently from that in r4. The reg-
ulatory region of the Hoxb2 gene carries two separate cis-regulatory modules, one
of which regulates its expression in r3 and r5, while expression in r4 is controlled
through the other. In r3 and r5, Hoxb2 is activated in part by the transcription factor
Krox20, which is expressed in these rhombomeres but not in r4 (see Fig. 5.35 and
Section 5.11), and there are binding sites for Krox20 in the corresponding regulatory

Expression of Hox gene subgroups 1 to 4 in the hindbrain (r1 to r8)

Fig. 5.35 Expression of Hox genes in the
hindbrain. The expression of genes of three Paralogous subgroups
paralogous Hox complexes in the hindbrain 1 2 3 4
(rhombomeres r1 to r8) is shown. Hoxa1
Hoxa
and Hoxd1 are not expressed at this stage.
Note that there are no Hox genes expressed Hoxb
in r1. The Hox genes whose anterior
Hoxd
expression starts in each rhombomere are
listed underneath. With the exception of
Hoxb1, which is only expressed in r4, all Krox20 Krox20
the Hox genes listed are also expressed
in all rhombomeres posterior to their most r1 r2 r3 r4 r5 r6 r7 r8 (spinal cord) rhombomeres
anterior expression. The gene encoding the
transcription factor Krox20 is expressed in Hoxa2 Hoxb2 Hoxb1 Hoxb3 Hoxb4
Hoxa3 Hoxa4
rhombomeres 3 and 5. Hoxd3 Hoxd4
After Krumlauf, R.: 1993.
 The initial regionalization of the vertebrate brain 207

control region of Hoxb2. In the case of Hoxb4, whose anterior boundary of expression r2 r3 r5 r6
r4
is at the boundary between r6 and r7, there is evidence that the gene is switched on
by signals both from within the neural tube and from adjacent somites. One of these
signals could be retinoic acid, which is produced by the somites and has been estab-
lished to have a role in patterning the hindbrain. An absence of retinoic acid leads
to loss of hindbrain rhombomeres, whereas an excess causes a transformation of cell
fate from anterior to posterior.
Evidence that Hox genes determine the fate and future development of cells in Fig. 5.36 Gene expression in the
the rhombomeres comes from misexpression experiments in which Hoxb1 (normally hindbrain. The photograph shows a coronal
section through the hindbrain of a E9.5
expressed in r4) was expressed in r2. Each pair of rhombomeres produces motor
mouse embryo, which is transgenic for two
axons that project to a single branchial arch: axons from rhombomeres r2/r3 project
reporter constructs. The first construct
to the first branchial arch, whereas those from r4/r5 project to the second arch (see contains the lacZ gene under the control
Fig. 5.33). Hoxb1 is normally expressed in r4 but not in r2. If Hoxb1 is misexpressed of an enhancer from Hoxb2, which directs
in an r2 rhombomere, however, this then sends axons to the second arch. This is yet expression in rhombomeres 3 and 5 (revealed
another example of a homeotic transformation produced by misexpression of Hox as blue staining). The second construct
genes. contains an alkaline phosphatase gene under
Hox genes are not expressed in the anterior-most neural tissue—r1 of the hind- the control of an enhancer from Hoxb1, which
brain, the midbrain and forebrain. Instead, homeodomain transcription factors such directs expression in rhombomere 4 (revealed
as Otx and Emx are expressed anterior to the hindbrain and specify antero-posterior as greenish-brown staining). A similar
enhancer directing expression in rhombomere
pattern in the anterior brain. As we saw earlier, the posterior limit of Otx2 expression
4 exists for Hoxb2. Anterior is to the left, and
delineates the midbrain–hindbrain boundary in the chick (see Fig. 5.32). The Droso-
the positions of five of the rhombomeres are
phila orthodenticle gene and the vertebrate Otx genes are homologous and provide
indicated (r2 to r6). Scale bar = 0.1 mm.
a good example of the conservation of gene function during evolution. orthodenticle
Photograph courtesy of J. Sharpe, from
is expressed in the posterior region of the future Drosophila brain, and mutations
Lumsden, A. and Krumlauf, R.: 1996.
in the gene result in a greatly reduced brain. In mice, Otx1 and Otx2 are expressed
in overlapping domains in the developing forebrain and midbrain, and mutation in
Otx1 leads to brain abnormalities and epilepsy. Mice with a defective Otx gene can be
partly rescued by replacing it with orthodenticle, even though the sequence similar-
ity of the two proteins is confined to the homeodomain region. Human Otx can even
rescue orthodenticle mutants in Drosophila.

5.13 Neural crest cells from the hindbrain migrate to populate the
branchial arches

The pattern of Hox gene expression in the epidermal ectoderm of the branchial
arches at any particular position along the antero-posterior axis is similar to that
in the neural tube and neural crest. Neurons and neural crest cells obtain their
positional values from the Hox genes they express as part of the neural tube and
retain it when they migrate. Neural crest from the hindbrain migrates to populate
the branchial arches, where it will contribute to mesodermal tissues of the face and
skull. This migration is accomplished in about 10 hours between 1 and 2 days of
incubation in the chick embryo. The ectoderm may acquire its positional value from
the crest cells that come to lie beneath it, and thus come to express the same pattern
of Hox genes.
The cranial neural crest that migrates from the rhombomeres of the dorsal region
of the hindbrain has a segmental arrangement, correlating closely with the rhombom-
ere from which the crest cells come. This has been revealed by labeling chick neural
crest cells in vivo and following their migration pathways. Crest cells from rhombom-
eres 2, 4, and 6 populate the first, second, and third branchial arches, respectively
(Fig. 5.37).
The crest cells have already acquired a positional value before they begin to
­migrate. When crest cells of rhombomere 4 are replaced by cells from rhombomere 2
taken from another embryo, these cells enter the second branchial arch but develop
into structures characteristic of the first arch, to which they would normally have
208 Chapter 5 Vertebrate development III: patterning the early nervous system and the somites

Expression of Hox genes of paralogous subgroups 1 to 4 in the hindbrain (r1 to r8), migrating neural crest cells,
and the surface ectoderm of the branchial arches (b1 to b4)

r1 r2 r3 r4 r5 r6 r7 r8 (spinal cord) rhombomeres

neural crest cells

Fig. 5.37 Expression of Hox genes in
the branchial region of the head. The b4
b1 b2 b3 branchial arches
expression of Hox genes in the hindbrain
(rhombomeres r1 to r8), neural crest, surface ectoderm
branchial arches (b1 to b4), and surface
ectoderm is shown. The arrows indicate
the migration of neural crest cells into the
Anterior Posterior
branchial arches. Note the absence of neural
crest migration from r3 and r5.
Hox1 Hox2 Hox3 Hox4
After Krumlauf, R.: 1993.

­ igrated. This can result in the development of an additional lower jaw in the chick
m
embryo. However, neural crest cells have some developmental plasticity and their
ultimate differentiation depends on signals from the tissues into which they migrate.
Gene knock-outs in mice have also shown that the Hox genes are involved in pat-
terning the hindbrain region, though the results are not always easy to interpret; knock-
out of a particular Hox gene can affect different populations of neural crest cells in the
same animal, such as those that will form neurons and those that will form skeletal
structures. Knock-out of the Hoxa2 gene, for example, results in skeletal defects in
that region of the head corresponding to the normal domain of expression of the gene,
which extends from rhombomere 2 backwards. Segmentation itself is not affected,
but the skeletal elements in the second branchial arch, all of which come from neural
crest cells derived from rhombomere 4, are abnormal. The usual elements, such as the
Dorsal view, neurula stage Xenopus embryo
stapes of the inner ear are absent, but instead, some of the skeletal elements normally
Anterior formed by the first branchial arch develop, such as Meckel’s cartilage, a precursor of an
eye eye element in the lower jaw. Thus, suppression of Hoxa2 causes a partial homeotic trans-
formation of one segment into another. The converse effect is seen when Hoxa2 is mis-
expressed in a more anterior position. Expression of Hoxa2 in the first branchial arch,
in which it is not normally expressed leads to transformation of the first-arch cartilages,
forelimb forelimb such as the quadrate and Meckel’s cartilage, which is a precursor element of the lower
jaw, into second-arch cartilages such as those of the tongue skeleton. In ­addition, if
Hox2a is transfected into prospective anterior neural crest, the cells do not develop in
their normal way and the skeletal structures they give rise to are abolished.
hindlimb hindlimb These observations, together with those described earlier in this chapter, show
that during gastrulation the cells of vertebrates acquire positional values along the
Posterior antero-posterior axis, and that this positional identity is encoded by the genes of the
Hox complexes. Many of the anatomical differences between vertebrates are probably
Fig. 5.38 The Xenopus embryo has simply due to differences in the subsequent gene targets of Hox gene actions, which
become regionalized by the neurula result in the emergence of different but homologous skeletal structures—the mam-
stage. Various organs such as limbs, heart,
malian jaw or the bird’s beak, for example.
and eyes will develop from specific regions
(red) of the neurula after gastrulation is
complete. Some of these regions, like the 5.14 The embryo is patterned by the neurula stage into organ-forming
limb buds, are already determined at this regions that can still regulate
stage and will not form any other structure.
The boundaries of the regions are not sharply At the neurula stage in Xenopus, the body plan has been established and the ­regions
defined, however, and within each region or of the embryo that will form limbs, eyes, heart, and other organs have become
‘field’ considerable regulation is still possible. ­determined (Fig. 5.38). This contrasts sharply with the blastula stage, at which time
 The initial regionalization of the vertebrate brain 209

no such determination has occurred. The basic vertebrate phylotypic body plan is
thus established during gastrulation. Although the positions of various organs are
fixed, there is not yet any overt sign of differentiation. Nevertheless, numerous graft-
ing experiments have shown that the potential to form a given organ is now confined
to specific regions. Each of these regions has, however, considerable capacity for regu-
lation, so that if part of the region is removed a normal structure can still form. For
example, the region of the neurula that will form a forelimb will, when transplanted to
a different region, still develop into a limb. If part of a future limb region is removed,
the remaining part can still regulate to develop a normal limb. The development of
limbs and other organs is discussed in Chapter 11.

Summary

The developing brain is divided into three regions—forebrain, midbrain, and hindbrain—
along the antero-posterior axis. Signaling centers located at the midbrain–hindbrain
boundary and within the forebrain produce signals that pattern adjacent regions of the
hindbrain, midbrain, and forebrain, respectively. The hindbrain is segmented into rhom-
bomeres, with the cells of each rhombomere respecting their boundaries. A Hox gene
code provides positional values for the rhombomeres of the hindbrain, and the neu-
ral crest cells that derive from them, while other genes specify more anterior ­regions.
Neural crest cells that migrate from the hindbrain populate the branchial arches in a
position-dependent fashion. By the neurula stage, after gastrulation, the body plan
has been established.

Summary: patterning of the vertebrate axial body plan

gastrulation and organizer activity

the Hox gene complexes are expressed along the antero-posterior axis

Hox gene expression establishes positional identity for mesoderm, endoderm, and ectoderm

mesoderm develops into notochord, early signals and mesoderm induce

somites, and lateral plate mesoderm neural plate from ectoderm

somites receive signals from notochord, mesoderm signals give regional

neural tube, and ectoderm identity to neural tube

somite develops into sclerotome rhombomeres and neural crest in the

and dermomyotome hindbrain are characterized by regional
patterns of Hox gene expression

Summary to Chapter 5

The germ layers specified during blastula formation become further patterned along the
antero-posterior and dorso-ventral axes during gastrulation. The primary embryonic orga-
nizer is involved in the initial patterning that underlies the regionalization of the antero-
posterior body axis. Positional identity of cells along the antero-posterior axis is encoded
by the combinatorial expression of genes of the four Hox gene complexes. There is both
spatial and temporal co-linearity between the order of Hox genes on the chromosomes and
210 Chapter 5 Vertebrate development III: patterning the early nervous system and the somites

the order in which they are expressed along the axis from the hindbrain backwards. Inac-
tivation or overexpression of Hox genes can lead both to localized abnormalities and to
homeotic transformations of one ‘segment’ of the axis into another, indicating that these
genes are crucial in specifying regional identity. At the end of gastrulation, the basic body
plan has been laid down and the nervous system induced. Specific regions of each somite
give rise to cartilage, muscle, and dermis, and these regions are specified by signals from
the notochord, neural tube, lateral plate mesoderm, and ectoderm. Induction and patterning
of the nervous system involves both signals in the early embryo and from the underlying
mesoderm. Signaling centers pattern regions of the hindbrain, midbrain, and forebrain and
in the hindbrain, Hox gene expression provides positional values for both neural tissue and
neural crest cells.

n End of chapter questions region. How do the gene expression patterns illustrate the tempo-
ral gradient of expression along the Hox clusters? How do the gene
Long answer (concept questions) expression patterns illustrate how different combinations of gene
expression can produce spatial differences in cell identity along an
1. Distinguish between the notochord, the spinal cord, and the ver-
axis?
tebral column.
10. Compare the consequences of Hoxc8 gene deletion in the mouse
2. Describe the relationships between the following structures: neu-
(Fig. 5.27) with deletion of the Ubx gene in Drosophila (Fig. 2.48).
roectoderm, neural plate, neural folds, neural tube, and neural crest.
Are the transformations seen toward anterior or posterior identities?
3. What experiment defines the phenomenon of primary embryonic What do these transformations say about the roles of Hox genes in
induction? How do the results of this experiment change depending identity specification? Do these transformations in mice qualify as
on the stage of the gastrula from which the dorsal lip of the blasto- homeotic transformations?
pore is derived?
11. How does the zebrafish giraffe (gir) mutation support the
4. Review what we have learned about the specification of anterior- ­hypothesis that a gradient of retinoic acid may be involved in the
most mesoderm in the Xenopus gastrula including goosecoid activa- ­co-linearity of the vertebrate axis and Hox gene expression? Recall-
tion in response to TGF-β family signaling, and the requirement for ing that genes are typically named for the mutant phenotype pro-
antagonists of ventral signals for head formation. duced, speculate on the phenotype of the gir mutation. If a deletion
5. The ‘default model’ for neural induction in Xenopus proposes (a loss-of-function) of the Cyp26a1 gene in mice causes cervical ver-
that the development of dorsal ectoderm as neural tissue is the tebrae to become more posterior rib-bearing vertebrae, would the
­default pathway. However, there are signals from both the vegetal gir mutation similarly be a loss-of-function, or, conversely, might it
cells and from ventral mesoderm, which could prevent dorsal ecto- instead be a mutation that causes increased expression of the gene
derm from forming. In light of this, outline the influences required (gain of function)? Explain.
for dorsal ectoderm to form. Include Ectodermin, Chordin (or nog- 12. Describe the effects of the experiment shown in Fig. 5.30, in
gin), and Cerberus. which a notochord from a donor embryo is grafted adjacent to the
6. The text states (Section 5.1): ‘The important function of the Spe- dorsal region of the neural tube in a host embryo. What signaling
mann organizer is to regulate these [BMP and Wnt] gradients by molecule is proposed to be responsible for the observed induction?
secreting growth-factor antagonists.’ Discuss this statement. 13. What kind of proteins are the Pax proteins? What is their rela-
7. Somite formation in chicks proceeds from anterior to posterior in tionship to Hox proteins? What is the role of Pax3 in the somite?
response to a gradient of FGF-8. Describe this gradient: Where does 14. What are two signaling centers that form in the developing
the FGF-8 originate? Why does it form a gradient? How does the vertebrate brain? What signals do they emit? In what way do they
gradient trigger somite formation? qualify as true signal centers (or ‘organizers’)?
8. The Hox gene clusters of mice (Box 5E) illustrate several mecha- 15. What are rhombomeres? What similarities can you list between
nisms that operate in genome evolution: genes can become duplicat- segmentation of the vertebrate hindbrain into rhombomeres and
ed to give rise to paralogs, the entire cluster can become duplicated segmentation in the embryo Drosophila?
to give rise to paralogous groups of genes, and within any given
cluster, genes can be lost. Give an example of each of these three 16. What are ephrins? How do they differ from most of the signal-
mechanisms. ing molecules that have been discussed? (Hint: are they secreted?)
What is their role in the formation of rhombomeres?
9. The Hox genes are expressed in the vertebrate mesoderm both
temporally and spatially in order, from one end of the Hox gene com- 17. Describe an experiment that demonstrates a homeotic transfor-
plex to the other. Refer to Fig. 5.26 and compare Hox gene ­expression mation of rhombomeres. Outline the strategy used and compare this
in the mouse posterior cervical region to that in the ­anterior thoracic with other strategies that result in homeotic transformations.
 The initial regionalization of the vertebrate brain 211

Multiple choice (factual recall questions) b) with higher numbers, such as Hoxa13
NB There is only one right answer to each question. c) with letters, so that anterior genes are Hoxa1–13, and posterior
genes are Hoxp1–13
1. Somites will give rise to
d) with lower numbers, such as Hoxa1
a) heart and blood
b) internal organs of the abdomen 9. Neural crest cells will contribute to which tissues or structures?
c) the vertebral column and skeletal muscles a) melanocytes in the skin
d) the spinal cord b) mesoderm of the face and skull
c) peripheral nervous system
2. The equivalents of the Xenopus organizer in chicks and zebrafish
are d) all of these
a) Hensen’s node and the embryonic shield
b) the hypoblast and the yolk syncytial layer Multiple choice answer key

c) the notochord in both cases 1: c, 2: a, 3: d, 4: b, 5: a, 6: b, 7: d, 8: a, 9: d.

d) the primitive streak and the mesendoderm

3. Chordin acts n Section further reading

a) as a TGF-β signaling molecule
5.1 The inductive capacity of the organizer changes during
b) to antagonize Wnt signaling gastrulation
c) to bind DNA and activate transcription Agathon, A., Thisse, C., Thisse, B.: The molecular nature of the
d) to block the activity of BMP-4 zebrafish tail organizer. Nature 2003, 424: 448–452.
Beddington, R.S.P., Robertson, E.H.: Axis development and early
4. Cerberus acts asymmetry in mammals. Cell 1999, 96: 195–209.
a) as a TGF-β signaling molecule Brickman, J.M., Jones, C.M., Clements, M., Smith, J.C., Bedding-
b) to antagonize Wnt signaling ton, R.S.P.: Hex is a transcriptional repressor that contributes
to anterior identity and suppresses Spemann organiser func-
c) to bind DNA and activate transcription
tion. Development 2000, 127: 2303–2315.
d) to enhance the activity of BMP-4
Glinka, A., Wu, W., Delius, H., Monaghan, P., Blumenstock, C.,
5. Somite formation in chicks proceeds Niehrs, C.: Dickkopf-1 is a member of a new family of secreted
proteins and functions in head induction. Nature 1998, 391:
a) from anterior to posterior, as Hensen’s node retreats
357–362.
b) from posterior to anterior, as the primitive streak elongates
Glinka, A., Wu, W., Onichtchouk, D., Blumenstock, C., Niehrs, C.:
c) from the dorsal midline outward, into lateral plate Head induction by simultaneous repression of Bmp- and Wnt-
d) simultaneously along the axis signalling in Xenopus. Nature 1997, 389: 517–519.
Griffin, K., Patient, R., Holder, N.: Analysis of FGF function in
6. In the early blastula, Wnt signaling and the subsequent nuclear
normal and no tail zebrafish embryos reveals separate mecha-
accumulation of β-catenin specifies a dorsal fate, whereas after gas-
nisms for formation of the trunk and tail. Development 1995,
trulation, Wnt/β-catenin signaling specifies
121: 2983–2994.
a) a dorsal fate Jansen, H.J., Wacker, S.A., Bardine, N., Durston, A.J.: The role of
b) a posterior fate the Spemann organizer in anterior-posterior patterning of the
trunk. Mech. Dev. 2007, 124: 668–681.
c) a ventral fate
Jones, C.M., Broadbent, J., Thomas, P.Q., Smith, J.C., Beddington,
d) an anterior fate R.S.P.: An anterior signalling centre in Xenopus revealed by the
7. The sclerotome of the somite gives rise to _____, whereas the homeobox gene XHex. Curr. Biol. 1999, 9: 946–954.
myotome gives rise to _____ Joubin, K., Stern, C.D.: Molecular interactions continuously ­define
the organizer during the cell movements of gastrulation. Cell
a) muscles, vertebrae
1999, 98: 559–571.
b) peripheral nervous system, muscles Kiecker, C., Niehrs, C.: The role of prechordal mesendoderm
c) spinal cord, limb muscles in neural patterning. Curr. Opin. Neurobiol. 2001, 11: 27–33.
d) vertebrae, muscles of the back Mukhopadhyay, M., Shtrom, S., Rodriguez-Esteban, C., Chen, L.,
Tsuku, T., Gomer, L., Dorward, D.W., Glinka, A., Grinberg, A.,
8. Hox genes that are expressed more anteriorly comprise Hox Huang, S.-P., Niehrs, C., Izpisúa Belmonte, J.-C., Westphal, H.:
genes Dickkopf1 is required for embryonic head induction and limb
a) with all possible numbers and letters, since there is no systematic morphogenesis in the mouse. Dev. Cell 2001, 1: 423–434.
relationship between the gene designations and the expression Niehrs, C.: Regionally specific induction by the Spemann-­Mangold
patterns of the Hox genes organizer. Nat. Rev. Genet. 2004, 5: 425–434.
212 Chapter 5 Vertebrate development III: patterning the early nervous system and the somites

Piccolo, S., Agius, E., Leyns, L., Bhattacharya, S., Grunz, H., Bou- Kudoh, T., Concha, M.L., Houart, C., Dawid, I.B., Wilson, S.W.:
wmeester, T., De Robertis, E.M.: The head inducer Cerberus is Combinatorial Fgf and Bmp signalling patterns the gastrula
a multifunctional antagonist of Nodal, BMP and Wnt signals. ectoderm into prospective neural and epidermal domains. De-
Nature 1999, 397: 707–710. velopment 2004, 131: 3581–3592.
Schneider, V.A., Mercola, M.: Spatially distinct head and heart in- Pera, E.M., Ikeda, A., Eivers, E., De Robertis, E.M.: Integration of
ducers within the Xenopus organizer region. Curr. Biol. 1999, IGF, FGF, and anti-BMP signals via Smad1 phosphorylation in
9: 800–809. neural induction. Genes Dev. 2003, 17: 3023–3028.
Sasai, Y., De Robertis, E.M.: Ectodermal patterning in vertebrate
5.2 The neural plate is induced in the ectoderm embryos. Dev. Biol. 1997, 182: 5–20.
Bachiller, D., Klingensmith, J., Kemp, C., Belo, J.A., Anderson, R.M., Sharman, A.C., Brand, M.: Evolution and homology of the nervous
May, S.R., MacMahon, J.A., McMahon, A.P., Harland, R.M., Ros- system: cross-phylum rescues of otd/Otx genes. Trends Genet.
sant, J., De Robertis, E.M.: The organizer factors Chordin and 1998, 14: 211–214.
Noggin are required for mouse forebrain development. Nature Sheng, G., dos Reis, M., Stern, C.D.: Churchill, a zinc finger tran-
2000, 403: 658–661. scriptional activator, regulates the transition between gastrula-
Delfino-Machin, M., Lunn, J.S., Breitkreuz, D.N., Akai, J., Storey, tion and neurulation. Cell 2003, 115: 603–613.
K.G.: Specification and maintenance of the spinal cord stem Stern, C.D.: Initial patterning of the central nervous system: how
zone. Development 2005, 132: 4273–4283. many organizers? Nat Rev. Neurosci. 2001, 2: 92–98.
De Robertis, E.M.: Spemann’s organizer and self-regulation in Storey, K., Crossley, J.M., De Robertis, E.M., Norris, W.E., Stern,
­amphibian embryos. Nat. Rev. Mol. Cell Biol. 2006, 4: 296–302. C.D.: Neural induction and regionalization in the chick em-
De Robertis, E.M., Kuroda, H.: Dorsal-ventral patterning and neu- bryo. Development 1992, 114: 729–741.
ral induction in Xenopus embryos. Annu. Rev. Dev. Biol. 2004,
20: 285–308. 5.4 Neural crest cells arise from the borders of
Londin, E.R., Niemiec, J., Sirotkin, H.I.: Chordin, FGF signaling, the neural plate
and mesodermal factors cooperate in zebrafish neural induc- Huang, X., Saint-Jeannet, J-P.: Induction of the neural crest and
tion. Dev. Biol. 2005, 279: 1–19. the opportunities of life on the edge. Dev. Biol 2004, 275: 1–11.
Marchal, L., Luxardi, G., Thomé, V., Kodjabachian L.: BMP inhibi-
tion initiates neural induction via FGF signaling and Zic genes.
5.5 Somites are formed in a well-defined order along the
Proc. Natl Acad. Sci. USA 2009; 106: 17437–17442.
antero-posterior axis
Stern, C.: Neural induction: old problems, new findings, yet more
questions. Development 2005, 132: 2007–2021. Bray, S.J.: Notch signalling: a simple pathway becomes complex.
Streit, A., Berliner, A.J., Papanayotou, C., Sirulnik, A., Stern, C.D.: Nat. Rev. Mol. Cell Biol. 2006, 7: 678-689.
Initiation of neural induction by FGF signalling before gastru- Dequeant, M.L., Pourquié, O.: Segmental patterning of the verte-
lation. Nature 2000, 406: 74–78. brate embryonic axis. Nat. Rev. Genet. 2008, 9: 370–382.
Wilson, S.L., Rydström, A., Trimborn, T., Willert, K., Nusse, R., Dubrulle, J., Pourquié, O.: fgf8 mRNA decay establishes a gradient
Jessell, T.M., Edlund, T.: The status of Wnt signaling regulates that couples axial elongation to patterning in the vertebrate
neural and epidermal fates in the chick embryo. Nature 2001, embryo. Nature 2004, 427: 419–422.
411: 325–329. Dubrulle, J., Pourquié, O.: Coupling segmentation to axis forma-
tion. Development 2004, 131: 5783–5793.
Box 5A Chromatin-remodeling complexes Kawakami, Y., Raya, A., Raya, R.M., Rodriguez-Esteban, C., Izpisua
Ho, L., Crabtree, G.R.: Chromatin remodelling during develop- Belmonte, J.C.: Retinoic acid signalling links left-right asym-
ment. Nature 2010, 463: 474–484. metric patterning and bilaterally symmetric somitogenesis in
the zebrafish embryo. Nature 2005, 435: 165–171.
5.3 The nervous system is initially patterned by signals from Kieny, M., Mauger, A., Sengel, P.: Early regionalization of somitic
the mesoderm mesoderm as studied by the development of the axial skeleton
Ang, S.L., Rossant, J.: HNF-3β is essential for node and of the chick embryo. Dev Biol. 1972, 28: 142–161.
notochord formation in mouse development. Cell 1994, 78: Lai, E.: Notch signaling: control of cell communication and cell
561–574. fate. Development 2004, 131: 965–973.
Aybar, M.J., Mayor, R.: Early induction of neural crest cells: les- Lewis, J.: Autoinhibition with transcriptional delay: a simple
sons learned from frog, fish and chick. Curr. Opin. Genet. Dev. mechanism for the zebrafish somitogenesis oscillator. Curr.
2002, 12: 452–458. Biol 2003, 13: 1398–408.
Doniach, T.: Basic FGF as an inducer of antero-posterior neural Stern, C.D., Charité, J., Deschamps, J., Duboule, D., Durston,
pattern. Cell 1995, 85: 1067–1070. A.J., Kmita, M., Nicolas, J.-F., Palmeirim, I., Smith, J.C.,
Foley, A.C., Skromne, I., Stern, C.D.: Reconciling different models Wolpert, L.: Head–tail patterning of the vertebrate embryo:
of forebrain induction and patterning: a dual role for the hypo- one, two or many unsolved problems? Int. J. Dev Biol. 2006,
blast. Development 2000, 127: 3839–3854. 50: 3–15.
Kiecker, C., Niehrs, C.: A morphogen gradient of Wnt/beta-catenin Vermot, J., Pourquié, O.: Retinoic acid coordinates somitogenesis
signaling regulates anteroposterior neural patterning in Xeno- and left-right patterning in vertebrate embryos. Nature 2005,
pus. Development 2001, 128: 4189–4201. 435: 215–220.
 The initial regionalization of the vertebrate brain 213

5.6 Identity of somites along the antero-posterior axis is Wacker, S.A., Janse, H.J., McNulty, C.L., Houtzager, E., Durston,
specified by Hox gene expression A.J.: Timed interactions between the Hox expressing non-
Burke, A.C., Nelson, C.E., Morgan, B.A., Tabin, C.: Hox genes and ­organiser mesoderm and the Spemann organiser generate
the evolution of vertebrate axial morphology. Development ­positional information during vertebrate gastrulation. Dev.
1995, 121: 333–346. Biol. 2004, 268: 207–219.
Godsave, S., Dekker, E.J., Holling, T., Pannese, M., Boncinelli, E., Yekta, S., Tabin, C., Bartel, D.P.: MicroRNAs in the Hox network:
Durston, A.: Expression patterns of Hoxb in the Xenopus em- an apparent link to posterior prevalence. Nat. Rev. Genet. 2008,
bryo suggest roles in antero-posterior specification of the hind- 9: 789–796.
brain and in dorso-ventral patterning of the mesoderm. Dev.
5.9 The fate of somite cells is determined by signals from
Biol. 1994, 166: 465–476.
the adjacent tissues
Kondo, T., Duboule, D.: Breaking colinearity in the mouse HoxD
complex. Cell 1999, 97: 407–417. Brand-Saberi, B., Christ, B.: Evolution and development of distinct
Krumlauf, R.: Hox genes in vertebrate development. Cell 1994, 78: cell lineages derived from somites. Curr. Topics Dev. Biol. 2000,
191–201. 48: 1–42.
Nowicki, J.L., Burke, A.C.: Hox genes and morphological identity: Brent, A.E., Braun, T., Tabin, C.J.: Genetic analysis of interac-
axial versus lateral patterning in the vertebrate mesoderm. De- tions between the somitic muscle, cartilage and tendon cell
velopment 2000, 127: 4265–4275. lineages during mouse development. Development 2005, 132:
515–528.
Box 5D Retinoic acid—a small-molecule intercellular signal Huang, R., Zhi, Q., Patel, K., Wilting, J., Christ, B.: Dual origin and
Niederreither, K., Dolle, P.: Retinoic acid in development: towards segmental organisation of the avian scapula. Development 2000,
an integrated view. Nat. Rev. Genet. 2008, 9: 541–553. 127: 3789–3794.
Rossant, J., Zirngibl, R., Cado, D., Shago, M., Giguère, V.: Expres- Olivera-Martinez, I., Coltey, M., Dhouailly, D., Pourquié, O.: Medio-
sion of a retinoic acid response element-hsplacZ transgene de- lateral somitic origin of ribs and dermis determined by quail-
fines specific domains of transcriptional activity during mouse chick chimeras. Development 2000, 127: 4611–4617.
embryogenesis. Genes Dev. 1991, 5: 1333–1344. Pourquié, O., Fan, C.-M., Coltey, M., Hirsinger, E., Watanabe, Y.,
Bréant, C., Francis-West, P., Brickell, P., Tessier-Lavigne, M., Le
5.7 Deletion or overexpression of Hox genes causes changes
Douarin, N.M.: Lateral and axial signals involved in avian
in axial patterning
somite patterning: a role for BMP-4. Cell 1996, 84: 461–471.
Condie, B.G., Capecchi, M.R.: Mice with targeted disruptions in Tonegawa, A., Takahashi, Y.: Somitogenesis controlled by Noggin.
the paralogous genes Hoxa3 and Hoxd3 reveal synergistic in- Dev. Biol. 1998, 202: 172–182.
teractions. Nature 1994, 370: 304–307.
Duboule, D.: Vertebrate Hox genes and proliferation: an alter- 5.10 Local signaling centers pattern the brain along the
native pathway to homeosis? Curr. Opin. Genet. Dev. 1995, 5: antero-posterior axis
525–528. Brocolli, V., Boncinelli, E., Wurst, W.: The caudal limit of Otx2
Favier, B., Le Meur, M., Chambon, P., Dollé, P.: Axial skeleton ho- expression positions the isthmaic organizer. Nature 1999, 401:
meosis and forelimb malformations in Hoxd11 mutant mice. 164–168.
Proc. Natl Acad. Sci. USA 1995, 92: 310–314. Kiecker, C., Lumsden, A.: Compartments and their boundaries
Kessel, M., Gruss, P.: Homeotic transformations of murine verte- in vertebrate brain development. Nat. Rev. Neurosci. 2005, 6:
brae and concomitant alteration of the codes induced by retin- 553–564.
oic acid. Cell 1991, 67: 89–104. Rhinn, M., Brand, M.: The midbrain-hindbrain boundary orga-
Ruiz-i-Altaba, A., Jessell, T.: Retinoic acid modifies mesodermal pat- nizer. Curr. Opin. Neurobiol. 2001, 11: 34–42.
terning in early Xenopus embryos. Genes Dev. 1991, 5: 175–187.
Wellik, D.M., Capecchi, M.R.: Hox10 and Hox11 genes are required 5.11 The hindbrain is segmented into rhombomeres by
to globally pattern the mammalian skeleton. Science 2003, 301: boundaries of cell-lineage restriction
363–367. Cooke, J.E., Moens, C.B.: Boundary formation in the hindbrain:
5.8 Hox gene expression is activated in an anterior to Eph only it were simple. Trends Neurobiol. 2002, 25: 264–267.
posterior pattern Klein, R.: Neural development: bidirectional signals establish
boundaries. Curr. Biol. 1999, 9: R691–R694.
Duboule, D.: Vertebrate Hox gene regulation: clustering and/or
Lumsden, A.: Segmentation and compartition in the early avian
colinearity? Curr. Opin. Genet. Dev. 1998, 8: 514–518.
hindbrain. Mech. Dev. 2004, 121: 1081–1088.
Sakai, Y., Meno, C., Fujii, H., Nishino, J., Shiratori, H., Saijoh, Y.,
Xu, Q., Mellitzer, G., Wilkinson, D.G.: Roles of Eph receptors and
Rossant, J., Hamada, H.: The retinoic acid-inactivating enzyme
ephrins in segmental patterning. Phil Trans. Roy. Soc. B 2000,
CYP26 is essential for establishing an uneven distribution of
355: 993–1002.
retinoic acid along the anteroposterior axis within the mouse
embryo. Genes Dev. 2006, 15: 213–225.
Vasiliauskas, D., Stern, C.D.: Patterning the embryonic axis: FGF Box 5F Eph receptors and their ephrin ligands
signaling and how vertebrate embryos measure time. Cell 2001, Klein, R.: Eph/ephrin signaling in morphogenesis, neural devel-
106: 133–136. opment and plasticity. Curr. Opin. Cell Biol. 2004, 16: 580–589.
214 Chapter 5 Vertebrate development III: patterning the early nervous system and the somites

Kullander, K., Butt, S.J.B., Lebret, J.M., Lundefeld, L., Restrepo, E., Rijli, F.M., Mark, M., Lakkaraju, S., Dierich, A., Dolle, P., Chambon,
Ryderström, A., Klein, Rudiger, Kiehn, O.: Role of Eph4A and P.: A homeotic transformation is generated in the rostral bran-
ephrin B3 in local neuronal circuits that control walking. Sci- chial region of the head by disruption of Hoxa2, which acts as
ence 2003, 299: 1889–1892. a selector gene. Cell 1993, 75: 1333–1349.
Wassef, M.A., Chomette, D., Pouilhe, M., Stedman, A., Havis, E.,
5.12 Hox genes provide positional information in the Desmarquet-Trin Dinh, C., Schneider-Maunoury, S., Gilardi-
developing hindbrain ­Hebenstreit, P., Charnay, P., Ghislain, J.: Rostral hindbrain pattern-
Bell, E., Wingate, R.J., Lumsden, A.: Homeotic transformation of ing involves the direct activation of a Krox20 transcriptional
rhombomere identity after localized Hoxb1 misexpression. Sci- enhancer by Hox/Pbx and Meis factors. Development 2008, 135:
ence 1999, 284: 2168–2171. 3369–3378.
Grammatopoulos, G.A., Bell, E., Toole, L., Lumsden, A., Tucker, White, R.J., Schilling, T.F.: How degrading: Cyp26s in hindbrain
A.S.: Homeotic transformation of branchial arch identity after development. Dev. Dyn. 2008, 237: 2775–2790.
Hoxa2 overexpression. Development 2000, 127: 5355–5365.
Grapin-Botton, A., Bonnin, M-A., McNaughton, L.A., Krumlauf, R., 5.13 Neural crest cells from the hindbrain migrate to
Le Douarin, N.M.: Plasticity of transposed rhombomeres: Hox populate the branchial arches
gene induction is correlated with phenotypic modifications. Keynes, R., Lumsden, A.: Segmentation and the origin of region-
Development 1995, 121: 2707–2721. al diversity in the vertebrate central nervous system. Neuron
Hunt, P., Krumlauf, R.: Hox codes and positional specification 1990, 4: 1–9.
in vertebrate embryonic axes. Annu. Rev. Cell Biol. 1992, 8: Le Douarin, N.M., Creuzet, S., Couly, G., Dupin, E.: Neural crest
­227–256. plasticity and its limits. Development 2004, 131: 4637–4650.
Krumlauf, R.: Hox genes and pattern formation in the branchial
region of the vertebrate head. Trends Genet. 1993, 9: 106–112. 5.14 The embryo is patterned by the neurula stage into
Nonchev, S., Maconochie, M., Vesque, C., Aparicio, S., Ariza- organ-forming regions that can still regulate
­McNaughton, L., Manzanares, M., Maruthainar, K., Kuroiwa, A., De Robertis, E.M., Morita, E.A., Cho, K.W.Y.: Gradient fields and
Brenner, S., Charnay, P., Krumlauf, R.: The conserved role of homeobox genes. Development 1991, 112: 669–678.
Krox-20 in directing Hox gene expression during vertebrate
hindbrain segmentation. Proc. Natl Acad. Sci. USA 1996, 93:
9339–9345.
 6
Development of nematodes,
sea urchins, and ascidians
nn Nematodes nn Echinoderms nn Ascidians

Having looked at early embryonic development in Drosophila and vertebrates, we will

now examine some aspects of the early development of three other model invertebrate
organisms that introduce some different developmental mechanisms. The nematode
Caenorhabditis elegans is a very important model for the specification of fate associ-
ated with asymmetric cell division, as much of its early development involves patterning
on a cell-by-cell basis rather than by morphogens affecting groups of cells. Sea urchins,
which represent the echinoderms, are models for highly regulative development, and so
their development illustrates some principles common to that of vertebrates, which is
also highly regulative, but in a simpler system. Complete gene circuits underlying early
sea-urchin development have been identified. The ascidians are of interest as they are
chordates, as are the vertebrates; comparison of ascidian and vertebrate development
is therefore of great interest in understanding chordate evolution.

This chapter considers aspects of body-plan development in three model invertebrate

organisms—nematodes, sea urchins (representing echinoderms), and the ascidians
(commonly known as sea squirts), which are of interest as they are chordates. The
evolutionary relationships between the organisms discussed in this chapter are shown
in Fig. 1.11. All conform to the general plan of animal development: cleavage leads
to a blastula (or blastula-equivalent stage in Caenorhabditis), which then undergoes
gastrulation with the emergence of a body plan.
There is an old, and now less fashionable, distinction sometimes made between so-
called regulative and mosaic development—the former involving mainly cell–cell inter-
actions whereas the latter is based on localized cytoplasmic factors and their distribution
through asymmetric cell divisions (see Section 1.17). Nematodes and ascidians show
many instances of mosaic-like development, while sea-urchin development is highly reg-
ulative. There are, however, elements of both types of development in most embryos.
A feature of nematodes and ascidians is that cell fate is often specified on a cell-
by-cell basis, a typical characteristic of mosaic development, and in general does not
rely on positional information established by gradients of morphogens. This contrasts
with the specification of cell fate in groups of cells in Drosophila, vertebrates, and the
sea urchin. The early embryos of many invertebrates contain far fewer cells than those
of flies and vertebrates, with each cell acquiring a unique identity at an early stage of
development. In the nematode, for example, there are only 28 cells when gastrulation
216 Chapter 6 Development of nematodes, sea urchins, and ascidians

starts, compared with thousands in Drosophila. Specification on a cell-by-cell basis

often makes use of asymmetric cell division and the unequal distribution of cytoplas-
mic factors. Daughter cells resulting from asymmetric cell division often adopt different
fates not as a result of extracellular signals, but autonomously, as a result of the unequal
distribution of some factor between them. However, asymmetric cell division in the
early stages of development does not mean that cell–cell interactions are absent or
unimportant in these organisms. Differences in the fates of two daughter cells produced
by asymmetric division can also be specified by extracellular factors and cell signaling.
We begin with the nematode Caenorhabditis elegans, which has been studied inten-
sively and in which many key developmental genes and signaling pathways have
been identified. In this animal, specification is largely on a cell-by-cell basis. We then
consider sea urchins, which develop much more like vertebrates, in that their embryos
rely heavily on intercellular interactions and are highly regulative, and patterning
involves groups of cells. Finally we consider ascidians, with particular emphasis on
Fig. 6.1 Life cycle of the nematode
Caenorhabditis elegans. After cleavage the role of the localization of cytoplasmic determinants in their early development.
and embryogenesis there are four larval
stages (L1–L4) before the sexually mature
adult develops. Adults of C. elegans are Nematodes
usually hermaphrodite, although males
can develop. The photographs show: the The free-living soil nematode C. elegans, whose life cycle is shown in Fig. 6.1, is an
two-cell stage (top, scale bar = 10 µm); an important model organism in developmental biology. Its advantages are its suitability
embryo after gastrulation with the future for genetic analysis, the small number of cells and their invariant lineage, and the
larva curled up (middle, scale bar = 10 µm);
transparency of the embryo, which allows the formation of each cell to be observed.
and the four larval stages and adult (bottom,
This led Sydney Brenner to choose C. elegans as a model organism for studying
scale bar = 0.5 mm).
the genetic basis of development. Brenner and his co-workers, Robert Horvitz and­
Photographs courtesy of J. Ahringer.

Adult (hermaphrodite)
Fertilized
egg

Cleavage

Embryogenesis
0

10
L4 Hours after
fertilization Hatching
at 25oC

40 20

L1
30
L1 L2 L3

L4

L2
L3 Adult
 Nematodes 217

Posterior One cell dividing Two cells AB and P1 dividing Four cells 28-cell stage

P2 Ep

P1 Ea

AB ABp EMS

pronuclei ABa

Anterior

Fig. 6.2 Cleavage of the Caenorhabditis

John Sulston, were awarded the Nobel Prize in Physiology or Medicine in 2002 for elegans embryo. After fertilization, the
discoveries concerning genetic control of organ development and apoptosis—cell pronuclei of the sperm and egg fuse. The
­suicide—in their studies on the worm. egg then divides into a large anterior AB cell
C. elegans has a simple anatomy and the adults are about 1 mm long and just 70 µm and a smaller, posterior P1 cell. At the next
in diameter. Nematodes can be grown on agar plates in large numbers and early lar- cell division, AB divides into ABa and ABp,
val stages can be stored frozen and later resuscitated. C. elegans adults are mainly while P1 divides into P2 and EMS. Each of the
hermaphrodite; in nematodes, these are females that make their own sperm for a cells will continue to divide in a well-defined
pattern within these groups to give rise to
short time and then switch to making oocytes. They are capable of self-fertilization.
particular cell types and tissues. Gastrulation
Small numbers of males exist and can be used in mating experiments. Embryonic
begins at the 28-cell stage. By this stage,
development is rapid, the larva hatching after 15 hours at 20°C, although maturation further division of the EMS cell has produced
through larval stages to adulthood takes about 50 hours. the E cells, Ea and Eb, which will give rise to
The nematode egg is small, ovoid and only 50 µm long. Polar bodies are formed the gut, and MS cells (not labeled here) which
after fertilization. Before the male and female nuclei fuse, there is what appears to be give rise to a variety of other cell types.
an abortive cleavage, but after fusion of the nuclei, true cleavage begins (Fig. 6.2). Photographs courtesy of J. Ahringer.
The first cleavage is asymmetric and generates an anterior AB cell and a smaller pos-
terior P1 cell. At the second cleavage, AB divides to give ABa anteriorly and ABp pos-
teriorly, while P1 divides to give P2 and EMS. At this stage the main axes can already
be identified, as P2 is posterior and ABp dorsal.
The AB cells and the P2 cell will also divide in a well-defined pattern to give rise to the
other tissues of the worm. Gastrulation starts at the 28-cell stage, when the descendants
of the E cell (itself produced by division of the EMS cell) that will form the gut move
inside. Not all cells that are formed during embryonic development survive; programmed
death, or apoptosis, of specific cells is an integral feature of nematode development. Fig. 6.3 Caenorhabditis elegans
The newly hatched larva (Fig. 6.3), while similar in overall organization to the larva at the L1 stage (20 hours after
mature adult, is sexually immature and lacks a gonad and its associated structures, fertilization). The vulva will form from the
gonad primordium.

Anterior Posterior
pharynx hypodermis intestine

body muscle

Dorsal anus

nerve
ring
mouth excretory
cell
ganglion Ventral gonad primordium
218 Chapter 6 Development of nematodes, sea urchins, and ascidians

such as the vulva of hermaphrodites, which are required for reproduction. Post-embry-
onic development takes place during a series of four larval stages separated by molts.
The additional cells in the adult are derived largely from precursor blast cells (P cells)
that are distributed along the body axis. Each of these blast cells founds an invariant
lineage involving between one and eight cell divisions. The vulva, for example, is
derived from blast cells P5, P6, and P7. One can think of post-embryonic development
in the nematode as the addition of adult structures to the basic larval plan.
It is a triumph of direct observation that, with the aid of Nomarski interference
microscopy, the complete lineage of every cell in the nematode C. elegans has been
worked out. The pattern of cell division is largely invariant—it is virtually the same
in every embryo. The larva, when it hatches, is made up of 558 cells, and after four
further molts this number has increased to 959, excluding the germ cells, which vary
in number. This is not the total number of cells derived from the egg, as 131 cells
die during development as a result of apoptosis, which is discussed in more detail in
Chapter 10. As the fate of every cell at each stage is known, a fate map can be accu-
rately drawn at any stage and thus has a precision not found in any vertebrate. As
with any fate map, however, even where there is an invariant cell lineage, this preci-
sion in no way implies that the lineage must determine the fate or that the fate of the
cells cannot be altered. As we shall see, interactions between cells have a major role
in determining cell fate in the nematode.
The complete genome of C. elegans has been sequenced, and contains
nearly 20,000 predicted genes. C. elegans does not have extensive alternative RNA
splicing, so that most genes code for just one protein, which may be part of the reason
the anatomically simple worm apparently needs more genes than the more complex
fruit fly. Around 1700 genes have been identified as affecting development, two-thirds
of which were found using the technique of RNA interference (RNAi; Box 6A, p. 220).
Many nematode developmental genes are related to genes that control development
in Drosophila and other animals; they include the Hox genes (see Box 5E, p. 194),
and genes for signaling proteins of the TGF-β, Wnt, and Notch families (see Box 4A,
p. 131) and their associated intracellular pathways. One exception is the Hedgehog
signaling pathway (see Fig. 2.40), which is not present in Caenorhabditis, although
it does have genes for proteins that can be considered to be related to Hedgehog.
Although zygotic gene expression begins at the four-cell stage, maternal components
control almost all of the development up to gastrulation at the 28-cell stage.

6.1 The antero-posterior axis in Caenorhabditis elegans is determined by

asymmetric cell division

The first cleavage of the nematode egg is unequal, dividing the egg into a large ante-
rior AB cell and a smaller posterior P1 cell. This asymmetry defines the antero-pos-
terior axis and is determined at fertilization. The P1 cell behaves rather like a stem
cell; at each further division it produces one P-type cell and one daughter cell that
will embark on another developmental pathway. For the first three divisions, the
­P-cell daughters give rise to body cells, but after the fourth cleavage they only give
rise to germ cells (Fig. 6.4). Division of the AB cell gives rise to anterior and poste-
rior AB daughter cells. The anterior daughter, ABa, gives rise to typically ectodermal
tissues, such as the epidermis (called the hypodermis in nematodes) and nervous
system, but also to a small portion of the mesoderm of the pharynx. The poste-
rior daughter, ABp, also makes neurons and hypodermis as well as some specialized
cells. At second cleavage, P1 divides asymmetrically to give P2 and EMS, which will
subsequently divide into MS and E. MS gives rise to many of the body’s muscles
and the posterior half of the pharynx, while E is the precursor of the 20 cells of the
endoderm of the mid-gut. The C cell, which is derived from the P2 cell at the third
cleavage, forms hypodermis and body-wall muscle; D from the fourth cleavage of the
 Nematodes 219

Anterior Zygote Posterior

AB unequal cleavage
Fig. 6.4 Cell lineage and cell fate in the
P1
early Caenorhabditis elegans embryo. The
ABa ABp cleavage pattern is invariant. The first
EMS cleavage divides the fertilized egg into a
large anterior AB cell and a smaller posterior
P2
hypodermis, P1 cell and the descendants of these cells
neurons MS E
have constant lineages and fates. AB
muscle, glands, gut divides into ABa, which produces neurons,
coelomocytes C hypodermis, and muscle (of the pharynx), and
P3
ABp, which produces neurons, hypodermis,
and some specialized cells. P1 divides to give
EMS and P2. The EMS cell then divides to
give cells MS and E. MS gives rise to muscle,
neurons
glands, and coelomocytes (free-floating
spherical cells), and E produces the gut.
muscle
D P4 Further divisions of the P lineage are rather
like stem-cell divisions, with one daughter
hypodermis, muscle muscle of each division (C and D) giving rise to a
neurons
germ variety of tissues while the other (P2 and P3)
line continues to act as a stem cell. Eventually,
P4 gives rise to the germ cells.

P cell produces only muscle. All the cells undergo further invariant divisions, and at
about 100 minutes after fertilization gastrulation begins.
Before fertilization, there is no evidence of any asymmetry in the nematode egg, but
sperm entry sets up an antero-posterior polarity in the fertilized egg that determines
the position of the first cleavage division and the future embryonic antero-posterior
axis. This cleavage is both unequal and asymmetric, forming the large AB cell at the
anterior end and the smaller P1 cell at the posterior. The existence of polarity in the Chromosomes P granules
fertilized egg becomes evident before the first cleavage. A cap of actin microfilaments
a
forms at the future anterior end, and a set of cytoplasmic granules—the so-called
P granules, which contain maternal mRNAs and proteins required for development of sperm nucleus
the germline cells—become localized at the future posterior end. (The P granules are egg nucleus
not the determinants of polarization; their redistribution simply reflects it.) P gran- A P
ules remain in the P daughters of cell division, eventually becoming localized to the
P4 cell, which gives rise to the germline (Fig. 6.5). b
The initial polarization of the fertilized egg seems to be due to the centrosome
that the sperm nucleus brings into the egg. The centrosome, or cell center, is a small
structure composed of microtubules that organizes the microtubule cytoskeleton of
the cell; it duplicates to form the two centrosomes that nucleate each pole of the
microtubule spindle that organizes chromosome distribution in mitosis and meiosis
(the mitotic cell cycle is discussed in Box 1B, p. 7). The egg cortex, the layer of cyto- c
plasm immediately under the plasma membrane, contains a dense network of actin

Fig. 6.5 Localization of P granules after fertilization. The movement of P granules is

shown during the development of a fertilized egg of Caenorhabditis elegans from sperm
entry up to the 26-cell stage. In the left panel, the embryo is stained for DNA, to visualize
the chromosomes, and in the right panel for P granules. (a) Fertilized egg with egg nucleus at
d
anterior end and sperm nucleus at posterior end. At this stage the P granules are distributed
throughout the egg. (b) Fusion of sperm and egg nucleus. The P granules have already moved
to the posterior end. (c) Two-cell stage. The P granules are localized in the posterior cell.
(d) 26-cell stage. All the P granules are in the P4 cell, which will give rise to the germline only.
Photographs courtesy of W. Wood, from Strome, S., et al.: 1983.
220 Chapter 6 Development of nematodes, sea urchins, and ascidians

Box 6A Gene silencing by antisense RNA and RNA interference

Processing of incoming RNA Assembly of RISC Capture and degradation of mRNA

processing Cytoplasm Nucleus gene

enzyme (Dicer)

mRNA
double-stranded RNA
guide RNA RISC

short
interfering
RNAs

Gene knock-outs remove the function of a gene permanently which ­appeared to be rapidly degraded. RNAi is now known to
by disrupting its DNA. Another set of techniques for suppress- be caused by the degradation of double-stranded RNAs by a cel-
ing gene function rely on destroying or inhibiting the mRNA. lular enzyme called Dicer, which chops them into short lengths of
Because the gene itself is left untouched, and thus suppression around 21–23 nucleotides. This so-called short interfering RNA
is, in principle, reversible, this type of loss of function is gen- (siRNA) is unwound into single-stranded RNA, which becomes
erally known as gene silencing. All methods of gene silencing incorporated into a nuclease-containing protein complex known
rely on the introduction into embryos or cultured cells of an RNA as RISC (for RNA-induced silencing complex). The RNA-binding
with a sequence complementary to that of the specific mRNA component in RISC is a ribonuclease called Argonaute, also called
being targeted. Depending on the particular technique, the intro- Slicer in mammalian cells. The siRNA acts a ‘guide RNA’ to tar-
duced RNA either binds to the mRNA and prevents it from being get RISC to any mRNA containing an exactly complementary
translated, and/or targets a nuclease to bind to the mRNA and sequence. The mRNA is then degraded by the nuclease in the
degrade it. RISC (see figure). RNAi is particularly effective and easy to carry
The first technique of this type to be developed uses short out in C. elegans. Adult worms injected with double-stranded RNA
synthetic antisense RNAs, which are usually chemically modi- will show suppression of the corresponding gene in the embryos
fied (for example, morpholino RNAs) to increase their stability they produce. Worms can also be soaked in the appropriate RNA
within the cell. Morpholino oligonucleotides have been used or fed on bacteria expressing the required double-stranded RNA.
extensively to block specific gene expression, especially in sea The natural role of the cellular RNAi machinery is thought to be
urchins, Xenopus, and zebrafish. defense against transposons and viruses, many of which pro-
A more recent addition to the gene-silencing armoury is the duce double-stranded RNAs in the course of replication.
technique of RNA interference (RNAi), which recruits a natu- In plants, fungi, nematodes, Drosophila, and most non-
ral RNA-degrading mechanism that is apparently ubiquitous in ­mammalian cells, RNAi is carried out by introducing an appropri-
multicellular eukaryotes from plants to mammals. The phenom- ate double-stranded RNA, which is then processed into siRNAs
enon was first uncovered in plants during transgenic experi- within the cells. In mammalian cells, however, double-stranded
ments, when it was found that introduction of a gene very simi- RNAs trigger another defensive response that interferes with
lar to one of the plant’s own genes blocked the expression of the gene-silencing effect, and RNAi is instead achieved by intro-
both the introduced gene and the endogenous gene. It turned ducing the siRNAs themselves, or by expressing artificial DNA
out that suppression was occurring at the level of the mRNAs, constructs from which siRNAs are transcribed.

­ laments associated with the motor protein myosin, forming contractile macromolec-
fi
ular assemblies of actomyosin. Interaction of the centrosome with the actin network
causes the network to contract asymmetrically away from the point of sperm entry
about 30 minutes after fertilization. This leads to a general flow of cortical components
towards the future anterior end of the zygote, and cytoplasmic flow towards the future
posterior end. Among the cortical proteins that are relocated by this movement are a
group of maternal proteins known as PAR (partitioning) ­proteins. These are initially
 Nematodes 221

uniformly distributed, but become concentrated in anterior or posterior cortex after

cortical movement and control the asymmetric divisions in the early embryo.
The PAR proteins were discovered in a very ingenious way. Vast numbers of embryos
would have had to be screened to detect mutants defective in asymmetric develop-
ment. The process was streamlined by using a mutant—egl—in which the fertilized
eggs are not released and the larvae hatch out inside the parent, devouring it from the
inside and killing it. Mutations that halt development early on spare the parent, and
this is how the maternal par genes were discovered.
Partitioning-defective embryos have defects in the asymmetric divisions that divide
the zygote into blastomeres. If the normal pattern of division is not precisely main-
tained, future development is affected. The par genes encode several different types of
proteins, including protein kinases and other proteins involved in intracellular signal-
ing pathways. PAR proteins similar to those in C. elegans are involved in establishing
and maintaining cell polarity in many different situations, and are present in animals
ranging from nematodes to mammals. In C. elegans, the correct localization of PAR
proteins is required for the correct positioning and orientation of mitotic spindles
in the early cleavages to ensure division at the required place in the cell and in the
required plane.
The cortical flow resulting from sperm entry and actomyosin network contraction
leads to the anterior localization of PAR-3 and PAR-6, and the posterior localization of
PAR-1 and PAR-2. This asymmetric distribution then results, by mechanisms that are
not yet completely understood, in the positioning of the first mitotic spindle posterior to
the center of the cell, and thus to a first cleavage that is both unequal (producing cells
of different sizes) and functionally asymmetric (producing cells containing different
developmental determinants). The PAR proteins are thought to be involved in regulat-
ing the forces that pull on each pole of the spindle to position it in the cell. The mecha-
nisms that control the first two cleavage divisions are extremely complex, as a large-
scale RNAi experiment that individually suppressed around 98% of the nematode’s
genes showed that more than 600 gene products were involved in these two stages.

6.2 The dorso-ventral axis in Caenorhabditis elegans is determined by

cell–cell interactions

Despite the highly determinate cell lineage in the nematode, cell–cell interactions
are involved in specifying the dorso-ventral axis. At the time of the second cleavage,
if the future anterior ABa cell is pushed and rotated with a glass needle, not only
is the antero-posterior order of the daughter AB cells reversed, but the cleavage of
P1 is also affected, so that the position of the P1 daughter cell EMS relative to the
AB cells is inverted (Fig. 6.6). The manipulated embryo develops completely nor-
mally but ‘upside down’, with its dorso-ventral axis inverted within the eggshell.
ABp develops as the anterior cell and ABa as the posterior cell of the pair, and the
polarity of P2 has been reversed compared to its polarity in the unmanipulated
embryo. This shows that the dorso-ventral polarity of the embryo cannot already
be fixed at this stage and that the usual developmental fate of these cells can be
changed: this implies that their fate is specified by cell–cell interactions. The normal
development after this manipulation also implies that the left–right axis of symme-
try, which is defined in relation to the antero-posterior and dorso-ventral axes, is
not yet determined. Indeed, the left–right axis can be reversed by manipulation at a
slightly later stage, as we see next.
Adult worms have a well defined left–right asymmetry in their internal structures,
and the pattern of development on the left and right sides of the embryo is strikingly
different, more so than in many other animals. Not only do cell lineages on the left
and right differ, but some cells move from one side to the other during embryonic
development. The concept of left and right only has meaning once the ­antero-posterior
222 Chapter 6 Development of nematodes, sea urchins, and ascidians

Normal development Experimental manipulation

AB P1 AB P1

p p P1

a a

P1

Manipulation
p a

ABp EMS
ABa
Dorsal Ventral P2

Fig. 6.6 Reversal of dorso-ventral Ventral Dorsal

P2 ABp
polarity at the four-cell stage of the EMS ABa
nematode. In normal embryos, the AB cell
rotates at cleavage so that ABa is anterior.
If the AB cell is mechanically rotated in the
opposite direction at second cleavage, the
ABp cell is now anterior. This manipulation normal worm normal worm
also displaces the P1 cell so that when it
divides, the position of the EMS daughter Dorsal Ventral
cell is reversed with respect to the AB cells. Right Left
Development is still normal, but the dorso-
ventral axis is inverted (the worm develops Anterior Posterior Anterior Posterior
‘upside down’ in the egg case) a and p show
the anterior/posterior orientation of the AB
cell before and after manipulation. Left Right
Ventral Dorsal
After Sulston, J.E., et al.: 1983.

and dorso-ventral axes are defined (see Section 4.7). In C. elegans, specification of
left and right normally occurs at the third cleavage, and handedness can be reversed
by experimental manipulation at this stage. At third cleavage, ABa and ABp each
divides to produce laterally disposed right and left daughter cells (for example, ABa
produces ABal and ABar). The plane of cleavage is, however, slightly asymmetric, so
that the left daughter cell lies just a little anterior to its right-hand sister. If the cells are
manipulated with a glass rod during this cleavage, their positioning can be reversed
so that the right-hand cell lies slightly anterior (Fig. 6.7). This small manipulation is
sufficient to reverse the handedness of the animal.
While the molecular mechanisms specifying handedness are not known, mutation
of the maternal gene gpa-16 results in near-randomization of spindle orientation.
Using temperature-sensitive mutant alleles of gpa-16, and raising the embryos at a
temperature at which the first two cleavages are often normal, it proved possible to
pinpoint the third cleavage as the one at which handedness is determined. In mutant
worms that survived to maturity, handedness had been reversed from right to left in
around 50%, and this could be traced back to a skewing of the mitotic spindle axis
 Nematodes 223

Normal six-cell embryo Normal adult (ventral view)

EMS ABpl
intestine
ABal ABal
P2

ABar ABar

ABpr

AB spindle skewed with micromanipulator Reversed six-cell embryo Reversed adult (ventral view)

ABal

ABar
intestine

Fig. 6.7 Reversal of handedness in

through 90° at third cleavage. Embryos that died had complete misorientation of Caenorhabditis elegans. At the six-cell
spindle axes at the third cleavage. The protein GPA-16 is a G-protein α subunit which, stage in a normal embryo, the AB cell on the
together with other proteins, is involved in the positioning of centrosomes, and thus left side (Abal) is slightly anterior to that on
of the mitotic spindle. the right (top left panels, scale bar = 10 µm).
There are ways of generating functional asymmetry other than unequal cell divi- Manipulation that makes the AB cell on the
sion: for example, at a much later stage in nematode development a functional differ- right side (ABar) more anterior results in an
ence is generated between two taste receptors, one on the left and one on the right, animal with reversed handedness (bottom
left panels, scale bars = 10 µm). The right
so that each expresses genes for different chemoreceptors. This asymmetry in gene
panels show the resulting normal and
expression is controlled by a microRNA, lsy-6.
reversed adults; scale bars = 50 µm.
Photographs courtesy of W. Wood, from
6.3 Both asymmetric divisions and cell–cell interactions specify cell fate in the Wood, W.B.: 1991.
early nematode embryo

Although cell lineage in the nematode is invariant, experimental evidence, such as that
described above, shows that cell–cell interactions are crucial in specifying cell fate at
very early stages. Otherwise, the reversal of ABa and ABp by micromanipulation just as
they are being formed (see Fig. 6.7) would not give rise to a normal worm. ABa and ABp
must initially be equivalent, and their fate must be specified by interactions with adja-
cent cells. Evidence for such an interaction comes from removing P1 at the first cleavage:
the result is that the pharyngeal cells normally produced by ABa are not made.
What are the interactions that specify the non-equivalence of the two AB descen-
dants? If ABp is prevented from contacting P2 it develops as an ABa cell. Thus, the
P2 blastomere is responsible for specifying ABp. The induction of an ABp fate by P2
involves the proteins GLP-1 in the ABp cell and APX-1 in P2. These proteins are the
nematode counterparts of the receptor protein Notch and its ligand Delta, respectively
and, like Notch and Delta, are associated with the cell membrane. As we have seen in
previous chapters, interactions between Notch and its various ligands set cell fate in
many developmental situations throughout the animal kingdom.
GLP-1 is one of the earliest proteins to be spatially localized during nematode
embryogenesis. The maternal glp-1 mRNA is uniformly present throughout the
embryo but its translation is repressed in the P-cell lineage; at the two-cell stage the
GLP-1 protein is thus only present in the AB cell. This strategy is highly reminiscent of
the localization of maternal Hunchback protein in Drosophila (see Section 2.9), and,
as in that case, in C. elegans other maternal proteins binding to the 3′ untranslated
region of glp-1 mRNA suppress its translation.
224 Chapter 6 Development of nematodes, sea urchins, and ascidians

After second cleavage the ABa and ABp cells both contain GLP-1. The two cells are
then directed to become different by a local inductive signal, sent to the ABp cell from
ABp
GLP-1 the P2 cell at the four-cell stage (Fig. 6.8). This signal depends on cell contact: it is
ABa P2
GLP-1
delivered by APX-1 (produced by the P2 cell), which acts as an activating ligand for
APX-1 GLP-1 in the ABp cell membrane. As a result of this induction, the descendants of ABa
MOM-2
EMS and ABp respond differently to later signals from descendants of the EMS lineage.
MOM-5 The EMS cell gives rise to daughter cells (MS and E) that give rise to mesoderm
and endoderm, respectively (see Fig. 6.4). Specification of the EMS cell as a mesendo-
dermal precursor also involves localized maternal determinants in the EMS cell and
Fig. 6.8 The P2 cell is the source of inductive signals from the adjacent P2 cell (see Fig. 6.8). One maternal determinant
inductive signals that determine antero- involved in giving EMS its mesendodermal fate is the transcription factor SKN-1, the
posterior polarity and cell fate. ABp product of the skin excess (skn-1) gene. skn-1 mRNA is uniformly distributed at the
becomes different from ABa as a result of two-cell stage, but there are much higher levels of SKN-1 protein in the nucleus of
signaling from the adjacent P2 cell by the P1 than in AB. Maternal-effect mutations that abolish the function of skn-1 result in
Delta-like protein APX-1, which interacts with the EMS blastomere adopting a mesectodermal fate; that is, it develops much like its
the Notch-like receptor GLP-1 on the ABp
sister cell P2 (see Fig. 6.4), giving rise to mesoderm (body-wall muscle) and ectoderm
cell. P2 also sends a Wnt-like inductive signal,
(hypodermis—hence the name ‘skin excess’ for the skn-1 mutant), but no pharyngeal
MOM-2, to the EMS cell that interacts with
a Frizzled-like receptor, MOM-5, on EMS to
cells and, in most mutants, no endoderm. SKN-1 acts by switching on the zygotic
set up a posterior difference in the cell that genes med-1 and med-2, which encode transcription factors that are necessary for MS
determines the different fates of anterior and fate and which contribute to the E fate in parallel with other maternal inputs.
posterior daughter cells. These interactions Formation of endoderm from the EMS cell also requires an inductive signal. If
depend on cell contact. removed from the influence of its neighbors towards the end of the four-cell stage,
an EMS cell can develop in isolation to produce gut structures, but if isolated at the
beginning of that stage, it cannot. Removal of P2 at the early four-cell stage results in
no gut being formed, indicating that P2 is the cell that delivers the inductive signal.
This was confirmed by recombining isolated EMS and P2 cells, which restores devel-
MS
opment of endoderm; recombining EMS with other cells from the four-cell embryo
p has no effect. The P2 cell signal to the EMS cell is a Wnt protein (MOM-2), which
a interacts with a Frizzled receptor (MOM-5) on the EMS cell surface (see Fig. 6.8), the
a p point of contact specifying the future posterior cell, which will be E. Signaling by P2
a p thus serves to break two symmetries in the embryo: P2 makes ABp different from ABa,
a p
and it also polarizes the EMS cell so that its division produces a posterior E cell and
an anterior MS cell.
a p
cell death
So we begin to see how a combination of cell–cell interactions and localized cytoplas-
paapp mic determinants specify cell fate in the early nematode embryo. The result of killing
individual cells by laser ablation at the 32-cell stage, when gastrulation begins, suggests
a p that many of the cell lineages are now determined, because when a cell is destroyed at
this stage it cannot be replaced and its normal descendants are absent. Cell–cell inter-
actions are, however, required again later to effect cells’ final differentiation.
Cell differentiation in the nematode is closely linked to the pattern of cell division.
pharyngeal cell Each blastomere undergoes a unique and nearly invariant series of cleavages that suc-
paapaaa cessively divide cells into anterior and posterior daughter cells. Cell fate appears to be
specified by whether the final differentiated cell is descended through the anterior (a) or
Fig. 6.9 Cell fate is linked to the pattern posterior (p) cell at each division. In the lineage generated from the MS blastomere, for
of cell divisions. The figure illustrates part example, the cell resulting from the sequence p-a-a-p-p undergoes apoptosis (Fig. 6.9),
of the cell lineages generated by the MS whereas that resulting from p-a-a-p-a-a-a gives rise to a particular pharyngeal cell.
blastomere, which give rise to body muscle What is the causal relation between the pattern of division and cell differentiation? One
and the mesodermal cells of the pharynx.
answer appears to lie in the anterior daughter cell having a higher intranuclear level
Each division produces an anterior (a) and
of the maternal protein POP-1, a transcriptional regulatory protein, compared with its
a posterior (p) cell. The lineage paapp, for
example, always gives rise to a cell that
posterior sister. In the case of the EMS blastomere, for example, the anterior daughter
undergoes programmed cell death and dies cell, the MS blastomere, has a higher level of intranuclear POP-1 than the posterior
by apoptosis. The lineage paapaaa, on the daughter, the E blastomere, as a result of an asymmetric division of the EMS cell. In
other hand, always results in a particular a maternal-effect pop-1 mutant that does not have any POP-1 protein, MS adopts an
pharyngeal cell. E-like fate because the antero-posterior distinction is lost.
 Nematodes 225

MS cell E cell

LIT-1
P
SYS-1
P
WRM-1

POP-1

Cytoplasm Nucleus

Fig. 6.10 The Caenorhabditis Wnt/β-catenin asymmetry pathway determines anterior and
posterior cell fates in the MS and E cells. The Wnt (MOM-2) signal originally received by Frizzled
receptors (MOM-5) at the posterior end of the EMS cell (not shown) sets up anterior–posterior
differences in the EMS cell that become evident in its daughter MS (anterior) and E (posterior)
cells. POP-1 is the C. elegans versions of the transcription factor TCF. SYS-1 and WRM-1 are
nematode-specific β-catenins. In the E cell, SYS-1 is stabilized and accumulates in the nucleus.
In addition, a complex of WRM-1 and the protein kinase LIT-1 also accumulates in the nucleus
and phosphorylates POP-1. This leads to the export of POP-1 from the E-cell nucleus, reducing
the amount to a level at which SYS-1 can act as co-activator for POP-1 and switch on endoderm-
specifying target genes. In the MS cell, in contrast, neither SYS-1 nor the WRM-1–LIT-1 complex
a ABarp
accumulate in the nucleus and so levels of POP-1 remain high and target genes remain ABala
suppressed. The interactions between the various pathway components are still being worked out. ABpla C
E
ABplp
The differential distribution of POP-1 in the EMS daughter cells is the ultimate D
result of the P2 Wnt signal (MOM-2) acting on the receptor Frizzled (MOM-5) on the MS
ABara
EMS cell (see Fig. 6.8). POP-1 is the C. elegans version of the vertebrate transcription
factor TCF. In the ‘canonical Wnt pathway’ β-catenin enters the nucleus as a result of
Wnt signaling, interacts with TCF and converts it from a transcriptional repressor to
b hypodermis
a transcriptional activator (see Box 1E, p. 26). In the context of the EMS asymmetric
cell division, the Wnt signal acts via a different intracellular signaling pathway, the neurons

‘Wnt/β-catenin asymmetry pathway’ that has not been found in Drosophila or verte-
brates. This pathway has two branches that independently control the actions of two gut
nematode-specific β-catenins, WRM-1 and SYS-1. The MOM-2 Wnt signal received
at the posterior end of the EMS cell polarizes the cell, resulting in the differential
pharynx muscle
localization of cytoplasmic factors controlling the pathway in the anterior (MS) and
posterior (E) daughter cells after cell division. As a result of the localization, a com-
plex of WRM-1 and a protein kinase (LIT-1) accumulates in the E-cell nucleus, which Fig. 6.11 Fate map of the 80-cell gastrula
leads to the phosphorylation of POP-1 and its export from the nucleus (Fig. 6.10). The of Caenorhabditis elegans. In panel (a),
other β-catenin, SYS-1, also accumulates in the anterior nucleus, and is able to act the regions of the embryo are color-coded
according to their blastomere origin. In
as co-activator to the reduced amounts of POP-1 and help to switch on endoderm-
panel (b) they are colored according to the
specifying genes. In the MS cell, on the other hand, WRM-1–LIT-1 is exported from
organs or tissues they will ultimately give
the nucleus, thus keeping POP-1 levels high there and suppressing the endodermal
rise to. At this stage in development, cells
fate. The Wnt/β-catenin asymmetry pathway is thought to be a general mechanism of from different lineages that contribute to
regulating anterior/posterior cell fate throughout nematode development, switching the same tissue or organ have been brought
on different target genes in different situations. together in the embryo. For clarity, the
At the 80-cell gastrula stage a fate map can be made for the nematode embryo domains occupied by ABpra and ABprp are
(Fig. 6.11). At this stage, cells from different lineages that will contribute to the not shown in panel a. Anterior is to the left,
same organ, such as the pharynx, cluster together. The gut is made clonally from ventral down.
E ­derivatives only. Genes that appear to act as organ-identity genes may now be After Labouesse, M. and Mango, S.E.: 1999.
226 Chapter 6 Development of nematodes, sea urchins, and ascidians

expressed by the cells in the cluster. The pha-4 gene, for example, seems to be an
organ-­identity gene for the pharynx. Mutations in pha-4 result in loss of the pharynx,
whereas ectopic expression of normal pha-4 in all the cells of the embryo leads to all
cells ­expressing pharyngeal cell markers.

6.4 Hox genes specify positional identity along the antero-posterior axis in
Caenorhabditis elegans

The nematode body plan is very different from that of vertebrates or Drosophila, and
there is no segmental pattern along the antero-posterior axis; nevertheless, as in other
organisms, Hox genes are involved in specifying positional identity along this axis.
The nematode contains a large number of homeobox genes, of which only six are
orthologous with those in the Hox gene clusters of Drosophila or vertebrates. Four of
these genes—lin-39, ceh-13, mab-5, and egl-5—are relatively closely linked, with the
remaining two, php-3 and nob-1, some distance away on the same chromosome. Com-
parisons with other nematode species indicate loss of Hox genes and disruption of the
ancestral Hox gene cluster in the lineage leading to C. elegans and its close relatives.
Of the C. elegans Hox genes, only ceh-13, which is required for anterior organiza-
tion, is essential for embryonic development, and it seems that the other Hox genes
carry out their function of regional specification at the larval stage. With the exception
of ceh-13, the order of the genes along the chromosome is reflected in their spatial pat-
tern of expression, as shown for three Hox genes in Fig. 6.12. lin-39 appears to control
the fate of mid-body cells and is known to be involved in regulating the development
of the vulva in hermaphrodites (discussed later in this chapter), whereas mab-5 con-
trols the development of a region slightly more posterior, and egl-5 provides positional
identity for posterior structures. php-3 and nob-1 are also posterior group genes. As in
other organisms, mutations in Hox genes can cause cells in one region of the body to
adopt fates characteristic of other body regions. For example, a mutation in lin-39 can
result in larval mid-body cells expressing fates characteristic of more anterior or more
posterior body regions.
Despite the fact that nematode Hox gene expression occurs in a regional pattern,
the pattern may not be lineage dependent. For example, in the larva, cells expressing
the Hox gene mab-5 all occur in the same region (see Fig. 6.12) but are quite unrelated
by lineage. The expression of mab-5 is due to extracellular positional signals.

6.5 The timing of events in nematode development is under genetic control

that involves microRNAs

Embryonic development gives rise to a larva of 558 cells and there are then four larval
stages, which produce the adult. Because each cell in the developing nematode can be
identified by its lineage and position, genes that control the fates of individual cells at

lab pb Dfd Scr ftz Antp Ubx abd-A Abd-B

Drosophila
HOM-C cluster

Caenorhabditis
Fig. 6.12 The Caenorhabditis elegans Hox Hox cluster
gene cluster and its relation to the HOM-C lin-39 ceh-13 mab-5 egl-5 php-3 nob-1
gene cluster of Drosophila. The nematode
contains a cluster of six Hox genes, four of
which show homologies with Hox genes of Larva
the fly. The pattern of expression of three of
the genes in the larva is shown.
Anterior Posterior
After Bürglin, T.R., et al.: 1993.
 Nematodes 227

Wild type Precocious Retarded

lin-14(lf) lin-14(gf)
loss-of-function mutation gain-of-function mutation
Larval T.ap
stage
Fig. 6.13 Cell-lineage patterns in
2 wild-type and heterochronic mutants of
1 1
Caenorhabditis elegans. The lineage of the
X X T-blast cell (T.ap) continues through four
2 larval stages (left panel). Mutants in the gene
1
3 lin-14 show disturbance in the timing of cell
X
4 division, resulting in changes in the patterns
3 1 of cell lineage. Loss-of-function mutations
4 X result in a precocious lineage pattern, with
the pattern of development of early stages
1 being lost (center panel). Gain-of-function
X mutations result in retarded lineage patterns,
Key = X = cell death
with the patterns of the early larval stage
X being repeated (right panel).

specific times in development can also be identified. This enables the genetic control
of timing during development to be studied particularly easily in C. elegans. The order
of developmental processes is of central importance, as well as the time at which they
occur. Genes must be expressed both in the right place and at the right time. One
well-studied example of timing in nematode development is the generation of differ-
ent patterns of cell division and differentiation in the four larval stages of C. elegans,
which are easily distinguished in the developing cuticle.
Mutations in a small set of genes in C. elegans change the timing of cell divisions in
many tissues and cell types. Mutations that alter the timing of developmental events are
called heterochronic. The first two heterochronic genes to be discovered in C. ­elegans The stage-specific pattern of larval development
were lin-4 and lin-14, and we shall use them to illustrate both the phenomenon of het- is determined by the level of LIN-14 protein
erochrony and the control of this process by microRNAs. Different mutations in either
high L1
of these genes can produce either ‘retarded’ or ‘precocious’ development.
Changes in developmental timing resulting from mutations in lin-14 can be illus- LIN-14 medium L2

trated by what happens to the T-cell lineage (Fig. 6.13). In wild-type individuals, the low L3
T.ap cell gives rise to hypodermal cells, neurons, and their support cells by a pattern
of stereotyped divisions in larval stages L1 and L2. During larval stages L3 and L4,
The wild-type temporal gradient of LIN-14
some of the T-cell descendants divide further to give rise to other adult structures. may result from post-transcriptional repression
Loss-of-function mutations in lin-14 result in a precocious phenotype—for example, of lin-14 by lin-4, beginning early in larval
development
the pattern of cell divisions seen in L1 is lost and post-embryonic development starts
with the cell divisions normally seen in L2. In contrast, gain-of-function mutations in Wild type
lin-14 result in retarded development. Post-embryonic development begins normally,
LIN-14

but the cell-division patterns of the first or second larval stages are repeated. lin-4 repression
It has been suggested that the genes that control timing of developmental events
L1 L2 L3 L4
may do so by controlling the concentration of some substance, causing it to decrease
with time (Fig. 6.14). This temporal gradient could control development in much the
lin-14 (lf) mutant
Fig. 6.14 A model for the control of the temporal pattern of Caenorhabditis elegans larval
LIN-14

development. Top panel: the stage-specific pattern of larval development is determined by

a temporal gradient of the protein LIN-14, which decreases during larval development. A high L2 L3 L4
concentration in the early stages specifies the pattern of normal early-stage development as
shown in Fig. 6.13 (left panel). The reduction in LIN-14 at later stages is due to the inhibition
lin-14 (gf) mutant or lin-4 (lf) mutant
of lin-14 mRNA translation by lin-4 RNA. Bottom panels: loss-of-function (lf) mutations in
lin-14 result in the absence of the first larval stage (L1) lineage, whereas gain-of-function (gf)
LIN-14

mutations, which maintain a high level of LIN-14 throughout development, keep the lineage
in an L1 phase. Loss-of-function mutations in lin-4 result in a lifting of repression of lin-14, L1 L1 L1 L1 L1
a continued high activity of LIN-14, and a repetition of the L1 lineage.
228 Chapter 6 Development of nematodes, sea urchins, and ascidians

same way that a spatial gradient can control patterning. This type of timing mecha-
nism seems to operate in C. elegans, because the concentration of LIN-14 protein
drops tenfold between the first and later larval stages. Differences in the concentra-
tion of LIN-14 at different stages of development may specify the fates of cells, with
high concentrations specifying early fates and low concentrations later fates. Thus,
the decrease of LIN-14 during development could provide the basis of a precisely
ordered temporal sequence of cell activities. Dominant gain-of-function mutations of
lin-14 keep ­LIN-14 protein levels high, and so in these mutants the cells continue to
behave as if they are at an earlier larval stage. In contrast, loss-of-function mutations
result in abnormally low concentrations of LIN-14, and the larvae therefore behave as
if they were at a later larval stage.
Developmental timing in C. elegans was one of the first processes discovered in
which gene expression was found to be controlled by microRNAs (Box 6B). The
expression of lin-14 is controlled post-transcriptionally by lin-4, which encodes a
microRNA that represses the translation of lin-14 mRNA and downregulates the

Box 6B Gene silencing by microRNAs

Production of pre-miRNA Processing of pre-miRNA to miRNA Capture of mRNA and suppression of translation

gene

Dicer mRNA

Nucleus miRNA
RISC
pre-miRNA

Cytoplasm silenced or degraded mRNA

A new category of genes involved in development has been s­ equence. After initial processing in the nucleus, the transcript
recognized quite recently, which codes not for proteins but for is folded back on itself to form a double-stranded RNA hairpin,
specialized short RNAs, called microRNAs (miRNAs). These the pre-miRNA, which is exported to the cytoplasm (see fig-
regulate gene expression by preventing specific mRNA tran- ure). Here it is cleaved by the enzyme Dicer to produce a short
scripts from being translated. This natural method of gene single-­stranded miRNA about 22 nucleotides long. As in RNA
­silencing has similarities with RNA interference by small inter- interference, miRNAs become incorporated as guide RNAs in
fering RNAs (siRNAs) described in Box 6A (p. 220), but appears an ­RNA-induced silencing protein complex (RISC; see Box 6A,
to be a distinct phenomenon, although there is some overlap ­p. 220) and the complex is specifically targeted to an mRNA.
in the processing machinery. MicroRNAs were first identified in ­Unlike ­siRNAs, animal miRNAs are typically not perfectly comple-
Caenorhabditis elegans as the products of the genes let-7 and mentary to their target mRNA and have a few mismatched bases.
lin-4, which control the timing of development (see text). Devel- Once bound, the protein complex renders the mRNA inactive and
opmental roles are known for a few other miRNAs in nematodes, suppresses translation, and the mRNA may be degraded. miRNA
and for the miRNA bantam in Drosophila, which prevents cell can also repress transcription of genes.
death. There is now evidence that many human genes code Other small noncoding RNAs are known that have a role in reg-
for miRNAs and regulatory roles for human miRNAs are being ulating gene expression but work in a different way by physically
revealed in cancer. binding to the chromosomes to silence expression. Xist RNA,
The primary miRNA transcript is several hundred nucleotides which is involved in X-chromosome inactivation in mammals, is
long and contains an inverted repeat of the mature miRNA one example and is discussed in Chapter 9.
 Nematodes 229

concentration of LIN-14 protein. LIN-14 itself is a transcription factor, but no target

genes have yet been identified that can explain its ­heterochronic ­loss-of-function
and gain-of-function phenotypes. Increasing synthesis of lin-4 RNA during the
later larval stages generates the temporal gradient in LIN-14, as indicated by
the fact that loss-of-function mutations in lin-4 have the same effect as gain-of-
­function mutations in lin-14. Expression of another gene involved in timing, lin-41,
is also regulated post-transcriptionally by a microRNA, let-7, which represses its
­translation.

6.6 Vulval development is initiated by the induction of a small number of cells

by short-range signals from a single inducing cell

The vulva forms the external genitalia of the adult hermaphrodite worm and connects
with the uterus. It is an important example of a structure that is initially specified as
a small number of cells—just four, one inducing and three responding. Unlike the
other aspects of Caenorhabditis development discussed so far, the vulva is an adult
structure that develops in the last larval stage. The mature vulva contains 22 cells,
with a number of different cell types, and more than 40 genes are involved in its
development. It is derived from ectodermal cells originating from the AB blastomere.
Six of these precursor blast cells (P cells, not to be confused with the P (posterior)
cells of the early embryo) persist in the larva in a row aligned antero-posteriorly on
the ventral side of the larva, and posterior daughters of three of these—P5p, P6p, and
P7p—give rise to the vulva, each having a well defined lineage. One of the functions
of the Hox gene lin-39 is to prevent the fusion of the six P cells to form hypodermis
in the early larval stages, which is the fate of the rest of the ventral ectodermal cells
of this lineage. In discussing the development of the vulva, three distinct cell fates are
conventionally distinguished—primary (1°), secondary (2°), and tertiary (3°). The
primary and secondary fates refer to different cell types in the vulva, whereas the ter-
tiary fate is non-vulval. P6p normally gives rise to the primary lineage, whereas P5p
and P7p give rise to secondary lineages. The other three P cells give rise to the tertiary
lineage, divide once and then fuse to form parts of the hypodermis (Fig. 6.15). Ini-
tially, however, all six P cells are equivalent in their ability to develop as vulval cells.
One of the questions we address here is how the final three cells become selected as
the vulval precursor cells.

Anterior Posterior
Fig. 6.15 Development of the nematode
vulva. The vulva develops from three cells,
anchor cell P5p, P6p, and P7p, in the post-embryonic
stages of nematode development. Under
P3p P4p P5p P6p P7p P8p the influence of a fourth cell—the anchor
cell—P6p undergoes a primary pathway
of differentiation that gives rise to eight
vulval cells. P6p is flanked by P5p and P7p,
which each undergo a secondary pathway
of differentiation that gives rise to seven
tertiary tertiary secondary primary secondary tertiary
cells of different vulval cell type. Three other
hypodermis vulva hypodermis P cells nearby adopt a tertiary fate and give
rise to hypodermis.
230 Chapter 6 Development of nematodes, sea urchins, and ascidians

hours P5p P6p P7p Right

a
30
L3
A B B A
C D E F F E D C

35
anchor b
L4 FL cell FL
FT FT
Anterior Posterior

syncytial d
hypodermis

Left

vulF vulE vulD vulC vulB2 vulB1 vulA

Fig. 6.16 Formation of the vulva

by migration and fusion of specified The fates of the three vulval precursor cells are specified by an inductive signal
precursors. The left-hand diagram shows a from a fourth cell, the gonadal anchor cell, which confers a primary fate on P6p, the
lateral representation of the seven rings of cell nearest to it, and a secondary fate on P5p and P7p, the cells lying just beyond
the vulva at around 39 hours and the cell it (see Fig. 6.15). In addition, once induced, the primary cell inhibits its immediate
lineages that give rise to them. The daughter neighbors from expressing a primary fate. P cells that do not receive the anchor cell
cells of the A cell fuse, as do the daughter signal adopt a tertiary fate.
cells of the C cell. The open circles represent After induction, the lineage of these P cells is fixed. If one of the daughter cells is
nuclei. At a later stage there will be further
destroyed, the other does not change its fate. There is no evidence for cell interactions
fusion of cells in the rings. The blue bar on
in the further development of these three P-cell lineages, and cell fate is thus probably
the left shows the time in hours and larval
stages. The right panel shows a ventral view
specified by asymmetric cell divisions. The vulva is formed from the 22 cells derived
of the developing vulva. The sequence (a) to from the three lineages by a precise pattern of cell division, movement and cell fusion
(d) shows morphogenesis of the vulva over a that forms a structure of seven concentric rings stacked on each other to form a coni-
period of 5 hours, from 34 hours. Changes in cal structure (Fig. 6.16).
the shape of the cells and cell fusion give rise How are just three Pp cells specified, and how is the primary fate of the central cell
to the conical structure seen in the left panel. made different from the secondary fate of its neighbors? The six Pp cells are initially
After Sharma-Kishore, R., et al.: 1999. equivalent, in that any of them can give rise to vulval tissue. The key determining
signal is that provided by the anchor cell. The vital role of the anchor cell is shown by
cell-ablation experiments; when the anchor cell is destroyed with a laser microbeam,
the vulva does not develop.
Familiar signaling pathways are involved in vulval induction. The anchor cell’s
signal is the secreted product of the lin-3 gene, which is the counterpart of the
growth factor EGF in other animals. Mutations in lin-3 result in the same abnormal
development that follows removal of the anchor cell: no vulva is formed. The recep-
tor for the inductive signal is a transmembrane tyrosine kinase of the EGF receptor
(EGFR) family, encoded by the let-23 gene (Fig. 6.17). The inductive signal from the
anchor cell activates the EGFR intracellular signaling pathway most strongly in P6p,
but also detectably in P5p and P7p. The strongest activation of this pathway leads
to the P6p cell adopting a primary cell fate. P6p then sends a lateral signal to its
two neighbors that both prevents them from adopting a primary fate and induces a
secondary fate. This signal is composed of three proteins of the Delta family, which
interact with the transmembrane receptor LIN-12, a member of the Notch family,
on P5p and P7p. The canonical Wnt signaling pathway (see Box 1E, p. 26), involv-
ing a third C. elegans β-catenin (BAR-1), is involved in the adoption of cell fate in
response to this signal.
 Nematodes 231

anchor cell

inductive inductive
signal signal (LIN-3)

LET-23
3o 3o 2o 1o 2o 3o
lateral
inhibitory signal LIN-12
signal

hypodermis

Fig. 6.17 Intercellular interactions in vulval development. The anchor cell produces a
diffusible signal LIN-3, which induces a primary fate in the precursor cell closest to it by binding
to the receptor LET-23. It also induces a secondary fate in the two P cells slightly further
away, where the concentration of the signal is lower. The cell adopting a primary fate inhibits
adjacent cells from adopting the same fate by a mechanism of lateral inhibition involving LIN-12,
and also induces a secondary fate in these cells. A constitutive signal from the hypodermis
inhibits the development of both primary and secondary fates, but is overruled by the initial
inductive signal from the anchor cell.

Vulval development varies in many ways, both minor and major, among nematode
species as a result of evolutionary divergence and convergence within this large group
of organisms. Although the vulva has the same form in all species examined and is
derived from much the same small set of P cells, some species have a vulva located
posteriorly, whereas others, like C. elegans, have a centrally located vulva. The dis-
tinctive C. elegans mechanism of vulval induction, in which general vulval fate and a
primary cell fate are induced at the same time via a signal from a single gondal anchor
cell, is very recently evolved and probably arose within the genus Caenorhabditis.
Other species of nematode have a variety of vulval induction mechanisms that differ
from that of C. elegans to a greater or lesser extent.

Summary

Specification of cell fate in the nematode embryo is intimately linked to the pattern
of cleavage, and provides an excellent example of the subtle relationships between
maternally specified cytoplasmic differences and very local and immediate cell–cell
interactions. The antero-posterior axis is specified at the first cleavage by the site
of sperm entry and specification of the dorso-ventral and left–right axes involves
local cell–cell interactions. Gene products are asymmetrically distributed during the
early cleavage stages, but specification of cell fates in the early embryo is crucially
dependent on local cell–cell interactions, which are mediated by Wnt and Delta–Notch
signaling. The development of the gut, which is derived from a single cell, requires
an inductive signal by an adjacent cell. A small cluster of homeobox genes provides
positional identity along the antero-posterior axis of the larva. The timing of devel-
opmental events in the larva is controlled by changes in the levels of various LIN
proteins that decrease over time as a result of post-transcriptional repression by
microRNAs. Formation of the adult nematode vulva involves both a well defined
cell lineage and inductive cell interactions involving the EGF and Notch signaling
pathways. The detailed mechanism of vulval development varies among nematode
species.
232 Chapter 6 Development of nematodes, sea urchins, and ascidians

Summary: early nematode development: axis specification

Localization of PAR proteins in the fertilized egg

unequal and asymmetric first cleavage

Anterior Posterior
AB cell P1 cell

ABa and ABp EMS and P2

relative positions of daughter cells and EMS specifies dorso-ventral axis

relative positions of daughters of ABa and ABp specifies left–right handedness

asymmetric cell divisions and cell–cell interactions

P2 cell

EMS

MS E

asymmetric cell divisions

due to anterior POP-1
localization

Echinoderms

Echinoderms include the sea urchins and starfish. Because of their transparency and
ease of handling, sea-urchin embryos have long been used as a model developmen-
tal system. Another useful feature is that echinoderms are more closely related to
vertebrates than are the other main model invertebrate organisms, ­Drosophila and
Caenorhabditis. Echinoderms are deuterostomes, like vertebrates and ascidians, and
so these three groups have certain basic developmental features in common. Deu-
terostomes are coelomate animals that have radial cleavage of the egg, and in which
the primary invagination of the gut at gastrulation forms the anus, with the mouth
developing independently. Arthropods (which include ­Drosophila) and nematodes,
on the other hand, belong to the protostomes, in which cleavage of the zygote is not
radial and in which gastrulation primarily forms the mouth.
For some years the echinoderms were somewhat neglected as a general develop-
mental model as they were not amenable to investigation by conventional genetics
or by the transgenic techniques that have proved so successful in other organisms.
With the advent of new genetic techniques and the ability to identify genes by homol-
ogy with genes from other organisms, the genetic and molecular basis of sea urchin
development can now be studied. The genome sequence of the purple sea urchin
Strongylocentrotus purpuratus has been completed, and the complete gene-regulatory
network that governs early development is being worked out.

6.7 The sea-urchin embryo develops into a free-swimming larva

The fertilized sea-urchin egg is surrounded by a double membrane, inside which

the embryo develops to blastula stage. The blastula hatches from the membrane,
 Echinoderms 233

Fertilized
egg

Cleavage

Adults
2-cell
stage
0
50
120
Minutes
Metamorphosis after
fertilization
Days
after
fertilization Hours
5 after
Pluteus 64-cell
fertilization blastula
larva 24

48
Gastrulation
Hatching

Late Gastrula
gastrula

Fig. 6.18 Life cycle of the sea urchin

Strongylocentrotus purpuratus. Eggs
undergoes gastrulation, and develops into a bilaterally symmetrical free-swimming released by the female are fertilized
larva called a pluteus. This eventually undergoes metamorphosis into the radially externally by male sperm and develop into
symmetrical adult (Fig. 6.18). Developmental studies are confined to the develop- a blastula. After cleavage and gastrulation,
ment of the embryo into the larva, as metamorphosis is a complex and poorly the embryo hatches as a pluteus larva,
understood process. The sea-urchin embryo is classically regarded as a model of which then undergoes metamorphosis into
regulative development. It formed the basis for Driesch’s ideas at the beginning of the mature radially symmetrical adult. The
the twentieth century, that the position of the cells in an embryo determined their photographs show: the blastula (top); the
pluteus larva (middle); and the adult (bottom).
fate (see Section 1.3).
The sea-urchin egg divides by radial cleavage. The first three cleavages are symmet- Top and middle photographs courtesy of
Jim Coffman. Bottom Photograph from
ric but the fourth is asymmetric, producing four small cells at one pole of the egg, the
Oxford Scientific Films.
vegetal pole, thus defining the animal–vegetal axis of the egg. The first two cleavages
are at right angles to each other and divide the egg in the plane of the animal–vegetal
axis (Fig. 6.19). The third cleavage, in contrast, is equatorial and divides the embryo
into animal and vegetal halves. At the next cleavage the animal cells again divide in a
plane parallel to the animal–vegetal axis, but the vegetal cells divide asymmetrically
to produce four macromeres, which contain about 95% of the cytoplasm, and four
much smaller micromeres. At fifth cleavage, the micromeres divide asymmetrically
again, so that there are four small micromeres at the vegetal pole with four larger
micromeres above them. Continued cleavage results in a hollow spherical blastula
composed of about 1000 ciliated cells that form an epithelial sheet enclosing the
blastocoel.
Gastrulation in sea-urchin embryos, the mechanism of which we consider in
detail in Chapter 8, starts about 10 hours after fertilization, with the mesoderm and
­endoderm moving inside from the vegetal region. In S. purpuratus, the first event is
234 Chapter 6 Development of nematodes, sea urchins, and ascidians

Animal pole 4 cell 8 cell 16 cell 32 cell 64-cell blastula

Vegetal pole micromeres macromeres vegetal plate

Blastula Gastrula secondary mesenchyme mouth ectoderm Pluteus

Animal pole skeletal rod
O Ab O Ab

Vegetal pole
primary mesenchyme primary mesenchyme gut anus oral hood anus

Fig. 6.19 Development of the sea-urchin embryo. Top panels: plane of the animal–vegetal axis are shown, starting from a hollow
external view of cleavage to the 64-cell stage. The first two cleavages blastula. Gastrulation begins at the vegetal pole, with the entry of
divide the egg along the animal–vegetal axis. Prospective ectoderm is about 40 primary mesenchyme cells into the interior of the blastula.
in blue; endoderm in yellow, and mesoderm in red. Blastomeres that The gut invaginates from this site and fuses with the mouth, which
are colored in more than one color give rise to cells of all the respective invaginates from the opposite side of the embryo. The mouth defines
germ layers. The third cleavage divides the embryo into animal and one end of an oral (O)–aboral (Ab) axis, which will be the main axis of
vegetal halves. At the fourth cleavage, which is unequal, four small symmetry of the pluteus larva. The secondary mesenchyme comes
micromeres (red) are formed at the vegetal pole, only two of which can from the tip of the invagination. During further development, growth
be seen here. At fifth cleavage, the micromeres divide asymmetrically of skeletal rods laid down by the primary mesenchyme results in the
again, giving rise to four small micromeres and four large micromeres extension of the four ‘arms’ of the pluteus larva. The view of the
above them, not all of which can be seen here. Further cleavage results pluteus is from the oral side. The oral hood narrows into the mouth and
in a hollow blastula. Bottom panels: gastrulation and development channels food into it.
of the pluteus larva. Sections of the developing embryo through the

the entry into the blastocoel of about 32 primary mesenchyme (mesodermal) cells
at the vegetal pole (see Fig. 6.19). They migrate along the inner face of the blastula
wall to form a ring in the vegetal region and lay down calcareous rods that will form
the internal skeleton of the pluteus larva. The endoderm, together with the mesoder-
mal secondary mesenchyme, then starts to invaginate at the vegetal pole. Invagina-
tion involves an inward movement of cells in a sheet, rather like pushing a finger into
a balloon. In the embryo, the invagination is driven by local cellular forces, which
are discussed in Chapter 8. The invagination eventually stretches right across the
blastocoel, where it fuses with a small invagination in the future mouth region on
the ventral side. Thus the mouth, gut, and anus are formed. Before the invaginating
gut fuses with the mouth, secondary mesenchyme cells at the tip of the invagination
migrate out as single cells and give rise to mesoderm, such as muscle and pigment
cells. The embryo is then a feeding pluteus larva.
We will focus here on the initial patterning of the early sea-urchin embryo along its
two main axes. We shall see that, despite the great differences in final form between
sea urchins and vertebrates, some of the same basic developmental mechanisms
and gene circuits are used in both. Because of its body form, the sea-urchin pluteus
larva does not have antero-posterior and dorso-ventral axes like insect and vertebrate
embryos. Instead, the two axes that are defined in the embryo are the animal–vegetal
axis and an oral–aboral axis. The oral–aboral axis is defined by the position of the
mouth and in turn defines the larva’s plane of bilateral symmetry. An animal–vegetal
asymmetry is already present in the unfertilized egg, but the oral–aboral axis is only
 Echinoderms 235

established after fertilization and early cleavage. The sea urchin is also used as a
simple model of gastrulation and to study fertilization, and it is discussed in relation
to these topics in Chapters 8 and 9, respectively.

6.8 The sea-urchin egg is polarized along the animal–vegetal axis

The sea-urchin egg has a well-defined animal–vegetal polarity that appears to be

related to the site of attachment of the egg in the ovary. Polarity is marked in some
species by a fine canal at the animal pole, whereas in other species there is a band of
pigment granules in the vegetal region. Early development is intimately linked to this
egg axis. The first two planes of cleavage are always parallel to the animal–vegetal
axis, and at the fourth cleavage, which is unequal (see Fig. 6.19), the micromeres are
formed at the vegetal pole. The micromeres give rise to the primary mesenchyme, and
both the micromeres and the primary mesenchyme are initially specified by cytoplas-
mic factors localized at the vegetal pole of the egg.
The animal–vegetal axis is stable and cannot be altered by centrifugation, which
redistributes larger organelles such as mitochondria and yolk platelets. Isolated frag-
ments of egg also retain their original polarity. Blastomeres isolated at the two-cell
and four-cell stages, each of which contains a complete animal–vegetal axis, give rise
to mostly normal but small pluteus larvae (Fig. 6.20, left panel). Similarly, eggs that
are fused together with their animal–vegetal axes parallel to each other form giant,
but otherwise normal, larvae.
There is, in contrast, a striking difference in the development of animal and vegetal
halves if they are isolated at the eight-cell stage. An isolated animal half merely forms
a hollow sphere of ciliated ectoderm, whereas a vegetal half develops into a larva that
is variable in form but is usually vegetalized; that is, it has a large gut and skeletal Fig. 6.20 Development of isolated sea
rods, and a reduced ectoderm lacking the mouth region (see Fig. 6.20, right panel). urchin blastomeres. Left panel: if isolated at
On occasion, however, vegetal halves from the eight-cell stage can form normal plu- the four-cell stage, each blastomere develops
teus larvae if the third cleavage is slightly displaced toward the animal pole. into a small but normal larva. Right panel: an
Together, these observations show that there are maternally determined differences isolated animal half from the eight-cell stage
along the animal–vegetal axis that are necessary to specify either animal or vegetal forms a hollow sphere of ciliated ectoderm,
fates. Despite this, it is clear that the sea-urchin embryo has considerable capacity for whereas an isolated vegetal half usually
develops into a highly abnormal embryo with
regulation, implying the occurrence of cell–cell interactions.
a large gut, and some skeletal structures, but
with a reduced ectoderm.

Isolation at four-cell stage gives four small larvae Animal and vegetal halves develop differently when isolated

vegetal half

animal half

Animalized Vegetalized
236 Chapter 6 Development of nematodes, sea urchins, and ascidians

6.9 The sea-urchin fate map is finely specified, yet considerable regulation
Oral Animal Aboral
is possible

mesomeres At the 60-cell stage, five regions of cells can be distinguished along the animal–­vegetal
(ectoderm) axis and a simplified fate map can be constructed (Fig. 6.21). This is composed
Veg1 of three bands of cells: the small and large micromeres at the vegetal pole, which
vegetal (gut and
plate ectoderm) will give rise to mesoderm (the primary mesenchyme that forms the skeleton and
Veg2 some adult structures); the vegetal plate, which comprises the next two tiers of cells
(gut and
large micromeres secondary (Veg1 and Veg2) and gives rise to endoderm, mesoderm (the secondary mesenchyme
(primary Vegetal mesenchyme)
mesenchyme) that forms muscle and connective tissues), and some ectoderm; and the remainder of
small micromeres the embryo, which gives rise to ectoderm and is divided into future oral and aboral
regions. The ectoderm gives rise to the outer epithelium of the embryo and to neuro-
genic cells.
Using vital dyes as lineage tracers, the pattern of cleavage and of cell fates have
both been mapped. The pattern of cleavage has been shown to be invariant. Unlike
the nematode, however, this stereotypical cleavage pattern seems irrelevant to normal
development. Compressing an early embryo by squashing it with a cover slip, and
thus altering the pattern of cleavage, still results in a normal embryo after further
development. Thus the stereotypical cleavage pattern does not, in most cases, pre-
dict an invariance of cell fate, and the embryo is highly regulative: lineage and cell
fate are to a large extent separable. The regulative capacity is not uniform, however.
Fig. 6.21 Fate map of the sea-urchin
The micromeres, for example, have a fixed fate once they are formed and this fate
embryo. Lineage analysis has shown four
main regions at the 60-cell stage. The cannot be changed. They give rise only to primary mesenchyme, and no other fate
embryo is divided along the animal-vegetal has been observed, even if they are grafted to another site on the embryo. The fate
axis into three bands: the micromeres, of the vegetal macromeres is not fixed when they are formed. Vegetal macromeres
which give rise to the primary mesenchyme; labeled at the 16-cell stage give rise to endoderm, mesoderm, and ectoderm. If cells
the vegetal plate, which gives rise to are labeled at the 60-cell stage, the Veg2 cells derived from these macromeres become
the endoderm, secondary mesenchyme, either mesoderm (secondary mesenchyme) or endoderm, but never ectoderm, while
and some ectoderm; and the mesomeres, the Veg1 cells give rise to endoderm and ectoderm but not mesoderm. This shows
which give rise to ectoderm. The ectoderm that endomesoderm specification has occurred by this stage, but that the boundary
is divided into oral and aboral regions.
between endoderm and ectoderm has not yet been set. This occurs slightly later. Sin-
Prospective mesoderm is colored in red;
gle Veg1 cell progeny labeled after the eighth cleavage give rise to either endodermal
prospective endoderm in yellow; prospective
or ectodermal tissues, but not both.
neural tissue in dark blue; and prospective
epidermis in light blue. When isolated and cultured, some regions of the sea-urchin embryo develop more
or less in line with their normal fate. Micromeres isolated at the 16-cell stage will form
After Logan, C.Y. and McClay, D.R.: 1997.
primary mesenchyme cells and may even form skeletal rods. An isolated animal half
of presumptive ectoderm forms an animalized ciliated epithelial ball, but there is no
indication of mouth formation. Although these experiments suggest that localized
cytoplasmic factors are involved in specifying cell fate, this does not tell the full story,
as they do not reveal the role of inductive cell-cell interactions in the normal process
of development. The ability of a vegetal half to form an almost complete embryo
has already been noted. Another dramatic example of regulation is seen when the
animal and vegetal blastomeres from half of an eight-cell embryo are combined with
the animal blastomeres from another; this unlikely combination can develop into a
completely normal embryo. How does this occur? The answer, as we see next, is the
ability of the vegetal region to act as an organizer.

6.10 The vegetal region of the sea-urchin embryo acts as an organizer

The regulative capacities of the early sea-urchin embryo are due in large part to an
organizing center that develops in the vegetal region and which, like the amphibian
Spemann organizer (see Fig. 4.32), has the potential ability to induce the forma-
tion of an almost complete body axis. This center is initially set up by the activities
of ­maternal factors and becomes located in the micromeres formed at the fourth
 Echinoderms 237

Animal pole Animal pole

Vegetal pole Vegetal pole

animal half

micromeres micromeres

Larva Secondary gut induced

Fig. 6.22 The inductive action of

micromeres. Left panels: combining the four
micromeres from a 16-cell sea-urchin embryo
with an animal half from a 32-cell embryo
results in a normal larva. An animal half
cultured on its own merely forms ectoderm
(not shown). Right panels: micromeres
implanted into the side of another 32-cell
embryo induce the formation of another gut
at the implantation site.

c­ leavage division. It produces signals that induce the adjacent vegetal macromeres to
adopt an endomesodermal fate. The endomesoderm gives rise to the sea-urchin spic-
ules (skeletal rods), muscle, and gut. The signals from the organizer set in train a relay
of events that subsequently specifies the endomesoderm more finely as mesoderm or
endoderm, and helps set the boundary between endoderm and ectoderm. Ablation of
the micromeres at the fourth cleavage results in abnormal development, but if they
are removed at the sixth cleavage, after 2–3 hours of contact, gastrulation is delayed
but normal larvae can develop.
Striking evidence for an organizer came from experiments that combined microm-
eres isolated from a 16-cell embryo with an isolated animal half from a 32-cell
embryo. This combination can give rise to an almost normal larva (Fig. 6.22). The
­micromeres clearly induce some of the presumptive ectodermal cells in the animal
half to form a gut, and the ectoderm becomes correctly patterned. Further evidence
for the organizer-like properties of the micromeres comes from grafting them to the
side of an intact embryo. Here, they induce endoderm in the presumptive ectoderm,
which then invaginates to form a second gut. There is some evidence that the closer
the graft is to the vegetal region, the greater the size of the invagination, implying
a graded ability along the animal–vegetal axis to respond to a signal from microm-
eres.
The regulative capacity of the very early embryo can even compensate for loss of
the micromeres. If micromeres are removed as soon as they are formed at the fourth
cleavage (see Fig. 6.19), regulation occurs and a normal larva will still develop. The
most vegetal region assumes the property of micromeres, giving rise to skeleton-
forming cells and acquiring organizing properties.
238 Chapter 6 Development of nematodes, sea urchins, and ascidians

6.11 The sea-urchin vegetal region is demarcated by the nuclear

accumulation of β-catenin

The sea-urchin vegetal region and its organizing properties are specified by a mech-
anism very similar to the one that marks out the future dorsal region in the early
Xenopus embryo (discussed in Chapter 4). In the sea urchin, the early blastomeres
accumulate maternal β-catenin in their nuclei as a result of activation of part of
the canonical Wnt pathway (see Box 1E, p. 26). This accumulation is not uniform,
however; β-catenin concentrations are high in the micromere nuclei whereas it
is almost absent from nuclei in the most animal region of the future ectoderm. This
differential stabilization of intranuclear β-catenin along the animal–vegetal axis is
at least partly controlled by the local activation in the vegetal region of maternal
Dishevelled protein, which is an intracellular activator of the pathway by which
β-catenin is stabilized. Within the nucleus, β-catenin acts as a co-activator of the
transcription factor TCF to activate the expression of zygotic genes required to
specify a vegetal fate.
Treatments that inhibit β-catenin degradation, such as lithium chloride (see
­Section 4.2) cause vegetalization, resulting in sea-urchin embryos with a reduced
ectoderm and an enlarged gut. The main effect of lithium on whole embryos is to
expand the area of intranuclear β-catenin accumulation beyond the normal vegetal
region (the region receiving a Wnt signal), shifting the border between endoderm
and ectoderm toward the animal pole. Similarly, overexpression of β-catenin also
causes vegetalization and can induce endoderm in animal caps. In contrast, pre-
venting the nuclear accumulation of β-catenin in vegetal blastomeres completely
inhibits the development of both endoderm and mesoderm. Micromeres depleted
of β-catenin, for example, are unable to induce endoderm when transplanted to
animal regions.
Maternal β-catenin and the transcription factor Otx activate the gene pmar1 in the
micromeres at the fourth cleavage. pmar1 encodes a transcription factor that is pro-
duced exclusively in micromeres during cleavage and early blastula stages and is
required for the micromeres to express their organizer function and also for their
development as primary skeletogenic mesenchyme. The outline of endomesoderm
specification is illustrated in Fig. 6.23. Soon after they are formed, the micromeres
produce a signal, called the early signal (ES), which induces the endomesoderm state
in the Veg2 cells. A second signal, produced after the seventh cleavage, is the ligand
Delta, which acts on the Notch receptor in adjacent Veg2 cells to specify them as
pigment and secondary mesenchyme cells. At later stages Notch–Delta signaling is
also involved in endoderm specification and in delimiting the endoderm–­ectoderm
boundary. The secreted signaling protein Wnt-8, first expressed by micromeres after
the fourth cleavage, acts back on the micromeres that produce it to maintain the sta-
bilization of β-catenin and thus maintain micromere function. Wnt-8 acting later in
the Veg2 endoderm is involved in directing the specification of endoderm in adjacent
Veg1 cells. In this way, the sea urchin mesoderm and endoderm is specified by a series
of short-range signals that result in changes in gene expression.

6.12 The genetic control of the skeletogenic pathway is known in

considerable detail

The pluteus larva has an internal skeleton of rods composed of calcium magnesium
salts and matrix proteins, many of which are acidic glycoproteins. The skeletal mate-
rial is secreted by skeleton-forming (skeletogenic) primary mesenchyme cells, which
are derived from the micromeres. The gene-regulatory network underlying the devel-
opment of the skeletogenic cells, from the events occurring in the early blastula to
activation of genes for matrix proteins, has been worked out in some detail. The genes
in this developmental pathway, and the interactions between them, were identified
 Echinoderms 239

Micromeres Endomesoderm
Veg2 endomesoderm Veg1 endoderm

Maternal Nuclear
�-catenin/ Wnt-8
factors
TCF system

ES Endomesoderm specification genes

Fig. 6.23 An outline of endomesoderm

specification in the sea urchin. The
Early micromere Mesoderm Endoderm Wnt-8
specification Delta specification specification white boxes represent networks of genes
genes genes genes that are reponsible for the developmental
Endoderm functions indicated. The background shading
Late and specification represents the embryonic territories involved.
transient genes
Peach, the large micromere precursors that
micromere Endoderm Mesoderm
specification specification specification will give rise to skeletal tissue; yellow, the
genes genes genes early endomesoderm, which segregates at
Veg2 mesoderm Veg2 endoderm late cleavage into mesoderm (violet) and
endoderm (blue). The red arrows and blue
barred lines refer to gene interactions; the
Mesodermal Endodermal thin black arrows represent intercellular
Skeletogenic
genes differentiation differentiation signals. ES, early signal.
genes genes
Adapted from Oliveri, P. and Davidson, E.H.:
2004.

both by measuring gene activity directly and by using antisense morpholino RNAs
(see Box 6A, p. 220) to silence specific gene expression. Every known regulatory gene
that is expressed in the skeletogenic lineage has been incorporated into a genetic
regulatory network; here we focus on the activation of genes that specify the primary
mesenchyme as skeleton-forming tissue.
As described in Section 6.11, the zygotic gene pmar1 is activated exclusively in the
micromeres, and this is the first step in specifying the micromeres as future skeleton-
forming tissue (Fig. 6.24). If pmar1 mRNA is made to be present in all the cells,
the whole embryo turns into skeleton-forming mesenchyme. pmar1 encodes a tran-
scriptional repressor for the gene hesC, which itself encodes a repressor. The HesC
protein is repressing genes essential for the micromeres to develop as primary mesen-
chyme, and so the repression of hesC transcription by Pmar1 allows the activation of
these genes (for example, alx1, ets1 and tbr) in the micromeres only (see Fig. 6.24).
This is another example of a common developmental strategy by which development
is allowed to proceed by using repression to lift a pre-existing repression. Alx1 and
ets1 encode transcription factors, which activate the next tier of genes, and so on, in
conjunction with intercellular signaling, for example via the Notch–Delta pathway.
The end result is the activation of genes that encode skeletal matrix proteins and
other proteins involved in the laying down and mineralization of the skeleton. Some
of these genes, such as those for matrix proteins Sm27 and Sm50, can be seen at the
bottom of Fig. 6.24.
The network of interactions leading to skeletogenic mesoderm is just a small part of
the whole network that specifies the endomesoderm and endoderm (Fig. 6.25). The
complexity of these networks is hardly surprising given how complex the regulatory
regions of developmental genes can be. As we saw in Chapter 2, the regulatory region
of a Drosophila pair-rule gene, such as even-skipped, contains many binding sites for
transcription factors that can both activate and repress gene activity (see Fig. 2.34).
These target sites are grouped into discrete subregions—regulatory ­modules—which
240 Chapter 6 Development of nematodes, sea urchins, and ascidians

Otx Maternal inputs �-cat

Micromere specification
Pmar1 HesC

Fig. 6.24 Specification of the sea urchin Alx1 Ets1 Tbr Delta
micromeres and the activation of their
organizing and skeletogenic function. The
diagram shows a simplified version of the
initial specification of the micromere lineage
as discussed in the text. Arrows indicate that
the gene product activates its target gene,
a barred line indicates that it suppresses Tel Erg Hex Tgif
the gene. The maternal transcription
factors β-catenin and Otx are localized in
micromere nuclei and activate transcription
of pmar1 (red). This gene encodes a
transcriptional repressor that suppresses
Dri FoxB
expression of hesC (brown), the repressor of
micromere fate, which is otherwise active
throughout the embryo. Genes repressed by
HesC protein are indicated in blue. When this
repression is lifted in the micromeres, the
products of these genes activate further sets
of genes (orange and black) that carry out
the organizing and skeletogenic properties Sm27 Sm50 Msp130 Msp-L
of the micromeres. Genes at the bottom of
the diagram colored in black encode skeletal
matrix proteins (e.g. Sm27 and Sm50) and
other proteins involved in producing the Skeleton formation
endoskeleton. Ficolin CyP
Adapted from Oliveri, P., et al.: 2008.

each control expression of the gene at a particular site in the embryo. Many sea urchin
genes similarly have cis-regulatory control regions composed of a number of modules
that govern gene expression at different times and in different parts of the embryo.
Although all the genes controlling sea-urchin development have yet to be identi-
fied, many whose expression varies with both space and time have been analyzed in
detail. Their regulatory regions prove to be complex and modular, as illustrated by
the regulatory region of the sea-urchin Endo-16 gene, a gene coding for a secreted
glycoprotein of unknown function. Endo-16 is initially expressed in the vegetal
region of the blastula, in the presumptive endoderm (Fig. 6.26). After gastrulation,
­Endo-16 expression increases and becomes confined to the middle region of the gut.
The regulatory region controlling Endo-16 expression is about 2200 base pairs long
and contains at least 30 target sites, to which 13 different transcription factors can
bind. These regulatory sites seem to fall into several subregions, or modules. The
 Echinoderms 241

Maternal and early interactions Outside endomesoderm Ubiq

LiCl
� frizzled
GSK-3 R of mic
Mat c�

ECNS n�-TCF
Wnt8

PMC

Endomesoderm Veg1 Endo

Endo

Mes

Fig. 6.25 The gene regulatory network for

function of each module has been determined by attaching it to a reporter gene and sea-urchin endomesoderm specification. The
injecting the recombinant DNA construct into a sea-urchin egg. In such experiments, part of the network shaded in gray at the
module A, for example, is found to promote expression of its attached reporter gene top corresponds to the nuclear localization of
in the presumptive endoderm, whereas modules D and C prevent expression in the β-catenin under the control of maternal factors.
primary mesenchyme. Midgut expression is under control of module B. Modules D, C, The rest of the network refers to zygotic
E, and F are activated by the action of added lithium and are involved in the vegetal- gene expression. PMC, primary mesenchyme
(micromeres); Mes, secondary mesenchyme;
izing action of Endo-16.
Veg1 Endo, Veg1-derived endoderm; Endo, rest
The modular nature of the Endo-16 control region is similar to that of the pair-rule
of endoderm.
genes in Drosophila, where different modules are responsible for expression in each
Adapted from Olivieri, P. and Davidson, E.H.:
of the seven stripes. The sea-urchin example illustrates yet again the importance of
2004.
gene-control regions in integrating and interpreting developmental signals.

6.13 The oral–aboral axis in sea urchins is related to the plane of the
first cleavage

The other main axis in the sea urchin-embryo is the oral–aboral axis, which defines the
main axis of the pluteus larva and also defines the plane of larval bilateral symmetry.
The oral ectoderm gives rise to the cells of the stomodeum (the mouth), the ­epithelium
242 Chapter 6 Development of nematodes, sea urchins, and ascidians

promotes promotes
boosts expression shuts off expression shuts off expression at skeletogenic late expression expression basal
of A and B at ectoderm boundary mesenchyme boundary, PMCs in midgut in vegetal half promoter

G F E D C B A

ectoderm gut
vegetal plate Endo-16 gene expression
primary mesenchyme cells (PMCs)
Early development Late development

Fig. 6.26 Modular organization of the

that forms the larval outer layer in the oral region, and neurogenic structures in the
endo-16 gene regulatory region. The
larva. The aboral ectoderm gives rise only to epithelium, which will cover most of the
modular subregions are indicated A to G.
Thirteen different transcription factors bind
larval body. There are clear differences in skeletal ­patterning in the oral and aboral
to the more than 50 individual binding sites regions, and the oral side can be recognized well before invagination forms the mouth,
in these modules. The spatial domains of because the skeleton-forming mesenchyme migrates to form two columns of cells on
the embryo in which each module regulates the future oral face. The migration of these cells is considered in Chapter 8.
gene expression are indicated, together Unlike the animal–vegetal axis, the oral–aboral axis cannot be identified in the
with other regulatory modules. Early in egg and appears to be potentially labile up to as late as the 16-cell stage. This is
development, regulation of gene expression why ­Driesch could get a normal complete larva from one isolated cell of the two-
restricts Endo-16 activity to the vegetal plate cell embryo (see Fig. 1.8). In normal development, however, the axis is predicted
and the endoderm of the future gut; later in by the plane of the first cleavage. In the sea urchin S. purpuratus, a good correlation
development it is restricted to the midgut.
was found between the oral–aboral axis and the plane of first cleavage: the future
axis lies 45° clockwise from the first cleavage plane as viewed from the animal pole
(Fig. 6.27). In other sea urchin species, however, the relationship between the oral–
aboral axis and plane of cleavage is different; the axis may coincide with the plane of
cleavage or be at right angles to it.
plane of Oral–aboral One clear indication of the establishment of the oral–aboral axis is the expression
first cleavage axis
of the TGF-β family member Nodal exclusively in the presumptive oral ectoderm in
the early blastula, at around the 60- to 120-cell stage. The initial event that breaks the
radial symmetry of the fertilized egg and sets the oral end of the oral–aboral axis by
initiating Nodal expression may be an underlying mitochondrial gradient. Oral–aboral
45o
polarity correlates with a mitochondrial gradient (highest in the oral region), which
produces an intracellular gradient in hydrogen peroxide (H2O2) and reactive oxygen
species. Experimental destruction of H2O2 in the two-cell embryo suppresses the ini-
tial expression of Nodal and pushes oral–aboral polarity towards aboral. It is likely
that the stress-activated kinase p38, which is upregulated by H2O2, is an intermediary
in this process.

animal pole view

6.14 The oral ectoderm acts as an organizing region for the oral–aboral axis
Fig. 6.27 Position of the oral–aboral axis
Nodal is one of the earliest zygotic genes to be expressed in the sea urchin, and Nodal
of the sea urchin, Strongylocentrotus
signaling seems to be instrumental in organizing and patterning the oral–aboral axis.
purpuratus, embryo in relation to first
If Nodal activity is blocked, the embryo remains radially symmetrical and differentia-
cleavage. When viewed from the animal pole,
the oral–aboral axis is usually 45° clockwise
tion of ectoderm into oral and aboral ectoderm is blocked. Overexpression of Nodal
from the plane of first cleavage. Note that also abolishes the oral–aboral axis, but in this case all the ectoderm is converted to an
at this stage the polarity of the axis is not oral fate. But if Nodal expression is completely blocked up to the eight-cell stage and
determined, only the positions of the two Nodal is then expressed in a single blastomere, even in one that would not normally
ends. express it, development can be ‘rescued.’ The treated blastulas develop oral–aboral
 Echinoderms 243

polarity, with the Nodal-expressing cells becoming oral ectoderm, and will frequently
even develop into normal pluteus larvae.
These experiments suggest that Nodal activity sets up a signaling center with orga-
nizing properties along the oral–aboral axis. High levels of Nodal signaling in the oral
ectoderm result in the expression of the transcription factor Goosecoid (a characteris-
tic protein of the amphibian Spemann’s organizer) and the secreted signaling protein
BMP-2/4. In the sea urchin, Goosecoid is thought to suppress aboral fate in the oral
ectoderm, whereas BMP-2/4 is thought to diffuse from its source in the oral ectoderm
and induce an aboral fate in the rest of the ectoderm. Another protein expressed in
the prospective oral ectoderm as a consequence of Nodal activity is the TGF-β family
member Lefty, which is an antagonist of Nodal signaling and restricts Nodal activity
to the oral ectoderm. If the function of Lefty is blocked, most of the ectoderm develops
into oral ectoderm. This situation is reminiscent of the interactions involved in meso-
derm induction and patterning in vertebrates, showing how the same developmental
gene circuits can be deployed for different purposes in different species. In the early
sea-urchin embryo, Nodal appears to have no role in specifying and patterning the
mesoderm, which is one of its primary roles in vertebrates (see Chapter 4).
For correct overall development, patterning along one axis must be spatially and tem-
porally linked to what is happening along the other. In the sea urchin, the oral–aboral
axis is linked to the animal–vegetal axis by an indirect effect of the intranuclear local-
ization of β-catenin in the vegetal region (described in Section 6.11). In the early blas-
tula, the transcription factor FoxQ2 is initially produced throughout the animal hemi-
sphere and suppresses expression of the nodal gene there. As the wave of ­intranuclear
β-catenin localization proceeds from the vegetal pole towards the animal hemisphere,
the signals it generates restrict FoxQ2 expression to a small region at the animal pole
known as the animal plate (which eventually gives rise to neurogenic ectoderm). This
removal of FoxQ2 allows expression of Nodal in the prospective oral ectoderm.
There is a second role for Nodal signaling in the sea urchin—the establishment of
internal left–right asymmetry (see Section 4.7). Inside the sea-urchin larva, a struc-
ture called the adult rudiment, from which the adult tissues will be produced during
morphogenesis, is formed only on the left side. As in vertebrates, a regulatory circuit
comprising Nodal, Lefty, and Pitx2 regulates the generation of this asymmetry, but
the expression of these genes is reversed in the sea urchin compared with vertebrates.
Nodal signals produced in the ectoderm on the right side of the larva inhibit rudiment
formation on the right side.

Summary

Sea-urchin embryos have a remarkable capacity for regulation even though they also
have an invariant early cell lineage. The sea-urchin egg has a stable animal–vegetal
axis before fertilization, which appears to be specified by the localization of maternal
cytoplasmic factors in the egg. After fertilization, maternal factors specify the forma-
tion of an organizer region in the most vegetal region of the egg, where the micromeres
are formed. This organizer, like the Spemann organizer in amphibians, is able to specify
an almost complete body axis. Localized activation of the Wnt pathway and nuclear
accumulation of β-catenin specifies the vegetal region and the organizing properties
of the micromeres. The oral–aboral axis is determined after fertilization. Signals from
the micromeres induce and pattern the endomesoderm along the animal–vegetal axis,
whereas the oral–aboral axis is specified and patterned by the activation of the Nodal
signaling pathway in the future oral region. The site and time of expression of devel-
opmental genes are controlled by complex gene-regulatory networks, which have been
worked out in some detail.
244 Chapter 6 Development of nematodes, sea urchins, and ascidians

Summary: sea urchin early development

Animal–vegetal axis

localized maternal factors in egg

animal–vegetal axis

β-catenin accumulation

organizer specified in vegetal region (micromeres)

induction of endomesoderm

Ascidians

Adult ascidians, also known as sea squirts, are sessile marine animals. They are uro-
chordates, which are included in the same phylum—the Chordata—as the vertebrates
because their free-living, tadpole-like larvae possess a notochord, neural tube, and mus-
cles, and are rather similar to the tailbud stage of a vertebrate embryo. The larva, which
has about 2600 cells, undergoes metamorphosis into a sac-like sessile adult (Fig. 6.28).
Although they are chordates, some aspects of ascidian development are very differ-
ent from those of vertebrates. Ascidian embryos have an invariant cleavage pattern
(Fig. 6.29), and localized cytoplasmic factors appear to have a much more important
role in specifying cell fate. The genomes of the sea squirts Ciona intestinalis and C. savi-
gnyi have been sequenced, and C. intestinalis has an estimated 15,800 genes, which
is comparable to the other model invertebrates. Knowing the genome sequences will
enable the full panoply of genomic techniques to be applied to the study of ascidian
development, and enable its similarities to, and differences from, the development of
vertebrates and other chordates to be investigated at the genetic and molecular level.
Knowledge of the C. intestinalis genome has already been exploited in a large-
scale screen for cis-regulatory control elements controlling Hox gene expression. Nine
Hox genes have been identified in Ciona, seven of which are dispersed on the same
­chromosome, with the rest on another chromosome. Hox genes are expressed along
the neural tube during development, and some are expressed in the endoderm that
gives rise to the intestine. But, unlike flies and vertebrates, the timing of expression
of the various Hox genes in Ciona seems not to be coordinated, and several are miss-
ing. Some 670 transcription factors have already been identified from the genome
sequence, along with 119 major signal transduction proteins. All the major vertebrate
developmental intercellular signaling pathways are present in ascidians, including the
Hedgehog, Wnt, TGF-β, and Notch pathways.
Ascidian embryogenesis has long been regarded as a typical example of mosaic
development, with cytoplasmic factors specifying cell fate during cleavage, and cell
interactions playing only a relatively minor part. It is now clear, however, that cell
interactions are more important than was previously thought.

6.15 Animal–vegetal and antero-posterior axes in the ascidian embryo are

defined before first cleavage

The developmental axes in an ascidian embryo are related to the early pattern of cleav-
age. Like amphibian and sea-urchin eggs, the unfertilized ascidian egg is ­polarized
 Ascidians 245

Adult
(hermaphrodite) Fertilized
egg
Eight-cell
stage

2 Cleavage
20
Juvenile

Days*
after 110-cell
fertilization 6 stage
*Hours
after
fertilization

Metamorphosis
18 10
16

Early
tailbud
Hatching

Early
tadpole

Larva
*At 18o – 20oC incubation

Fig. 6.28 Life-cycle diagram of the ascidian Ciona The photographs show: (top) a 110-cell embryo; (middle) a larva (with
intestinalis. Ciona is hermaphrodite and eggs are fertilized externally. some notochord cells labeled green; scale bar = 0.1 mm); (bottom)
The fertilized egg takes about 18 hours to hatch into a larva, adult C. intestinalis.
depending on the temperature of the water. The free-swimming Top photograph courtesy of Shigeki Fujiwara and Naoki Shimozono;
larva undergoes metamorphosis into the sessile juvenile (about 2 cm middle photograph courtesy of J. Corbo, from Corbo, J.C., et al.: 1997;
tall) in around 20 days. Development of the juvenile into a sexually bottom photograph courtesy of Andrew Martinez.
mature adult (which is about 5–8 cm tall) takes a further 2 months.

along an animal–vegetal axis, with the prospective territories of ectoderm, mesoderm,

and endoderm lying along this axis from animal to vegetal. Another axis, which is
conventionally called the antero-posterior axis, is established perpendicular to this.
Note that this naming convention differs from that of the axes in amphibian embryos. Fig. 6.29 Cleavages in the egg of the
This difference is largely semantic, and is due to the way in which ascidian embryos ascidian Halocynthia roretzi. The first
gastrulate and to a difference in the position of the developing embryo used as a cleavage divides the fertilized egg along
reference point by ascidian researchers. The animal–vegetal axis of the egg is conven- the dorso-ventral plane into left and right
tionally considered as the future ventro-dorsal axis of the tadpole, as the vegetal pole halves. From this stage onwards the
becomes covered with dorsal tissues at gastrulation. development of the embryo is bilaterally
symmetrical. Corresponding right- and
left-half blastomeres are given the same
Animal 90o designation, but left-half blastomeres
A3 B3
polar bodies B4.2 and their descendants are conventionally
A4.2 underlined. The second cleavage divides the
B4.2
A AB2 AB2 A3 B3 embryo along the antero-posterior axis. Note
B4.1
A4.1 that the view of the embryo has been rotated
B4.1 by 90° towards the viewer, between the first
Left Right Anterior Posterior Anterior Posterior and second cleavages, so that the antero-
Vegetal
posterior axis can be shown clearly.
246 Chapter 6 Development of nematodes, sea urchins, and ascidians

Important maternal factors are localized in the vegetal region of the fertilized egg,
among them components of the Dishevelled–β-catenin signaling pathway. As in the
sea urchin, localized activation of this pathway stabilizes maternal β-catenin and
induces its preferential entry into nuclei in vegetal cells, where it is required to specify
endoderm in the vegetal region. In contrast, the development of an epidermal fate in
animal blastomeres requires suppression of β-catenin function.
The animal–vegetal and antero-posterior axes of the ascidian embryo are speci-
fied in the fertilized egg before first cleavage by movement of the cortical cyto-
plasm. The unfertilized egg is radially symmetrical along its animal–vegetal axis.
As in Xenopus, sperm entry at fertilization triggers a rotation of the egg cortex
that breaks this symmetry (see Section 4.2). In the ascidian, part of the cortical
cytoplasm moves towards the vegetal pole, generating a characteristic bulge at the
pole. This movement is associated with the cytoskeleton, principally cortical actin
filaments and a deeper network of intermediate filaments. The cortical region that
relocates is rich in mitochondria and is called the myoplasm. It contains maternal
cytoplasmic determinants, such as the mRNA macho-1, that specify muscle, and it
will eventually give rise to the muscle of the ascidian tadpole’s tail. Its position after
cortical movement specifies the posterior end of the antero-posterior axis, which is
also where gastrulation will start.
The first cleavage is in the plane of the animal–vegetal axis and gives rise to two
cells with similar developmental potential, each of which can regulate to grow into a
half-size tadpole if separated. The second cleavage is in the same plane but at right
angles to the first and delimits the anterior and posterior halves of the embryo. The
third cleavage demarcates the animal and vegetal halves (see Fig. 6.29). The planes
of further cleavages are strictly controlled; this stereotyped pattern of cleavage deter-
mines what cytoplasmic maternal determinants the resulting cells contain and thus
how they will develop. An important influence on further unequal cleavages in the
A P posterior region is the so-called chromosome-attracting body (CAB) in the myoplasm.
The CAB contains PAR proteins, which, as in nematodes (see Section 6.1), help posi-
tion the mitotic spindle so that the cell divides ­asymmetrically.
By the time the embryo reaches the 110-cell stage, at which gastrulation begins,
the fate of each blastomere has become restricted by a combination of the maternal
cytoplasmic determinants it contains and intercellular signaling, such that most of the
cells give rise to a single cell type in the larva (Fig. 6.30). When individual blastom-
eres from this stage are isolated and cultured they develop into the cell types specified
by the fate map.
Despite the prominent role of cytoplasmic determinants in specifying cell fate in
A P ascidians, the development of some types of mesodermal tissue at least depends on
inductive signals, as in vertebrates and echinoderms. Three distinct types of mesoder-
mal tissue are formed in ascidian embryos—the muscles of the larval tail, the mesen-
chymal tissue that gives rise to internal tissues, and the notochord. Here we shall look
at the ways in which the fate of these three tissues is determined.

6.16 In ascidians, muscle is specified by localized cytoplasmic factors

epidermis nerve brain endoderm
muscle notochord mesenchyme The fate of the myoplasm in the ascidian Styela is particularly easy to follow as
germ cells trunk lateral cell trunk ventral cell it contains yellow pigment granules that are visible in the embryo, and this was
the first indication, discovered at the beginning of the twentieth century, that a par-
Fig. 6.30 Fate map of the 110-cell ticular region of the ascidian egg cytoplasm could give rise to a particular tissue. A
ascidian embryo. Top panel: view looking hundred years later we are at last beginning to find out how. The cells that acquire
down on the animal pole. Almost all the cells the yellow myoplasm during cleavage give rise to the muscle cells of the larval tail
of the animal half become ectoderm. Bottom (Fig. 6.31). ­Before fertilization, the yellow granules are more or less uniformly dis-
panel: view from the vegetal pole. tributed throughout the egg; following fertilization and cortical rotation, there is a
Adapted from Nishida, H.: 2005. dramatic rearrangement of the myoplasm to form the yellow crescent at the equator.
 Ascidians 247

Yellow crescent at time Yellow crescent Gastrula—about Early tadpole

of first cleavage at 8-cell stage 22-cell stage 180 cells Dorsal view Lateral view

myoplasm myoplasm

Fig. 6.31 Muscle development and

By the eight-cell stage, the myoplasm is largely confined to the two vegetal posterior cytoplasmic determinants in the ascidian
cells, with a small amount in adjacent cells (see Fig. 6.31). In the species Halocynthia Styela. Following fertilization, the myoplasm,
roretzi, the cell lineage has been worked out in detail by the injection of a tracer into the which is colored with yellow granules,
early cells. The two posterior B4.1 cells (see Fig. 6.29) that contain myoplasm contribute moves laterally and toward the equator.
to the primary muscle cells, which are 28 of the 42 muscle cells that lie on each side of This movement forms a yellow crescent
the tail. The secondary muscle cells are derived from blastomeres adjacent to B4.1 at the at the future posterior end of the embryo.
eight-cell stage and end up at the top of the tail. The lineage is complex; for example, at Gastrulation starts at this site. The muscle of
the ascidian tadpole’s tail comes both from
the 128-cell stage one of the descendants of B4.1 is still an endomesodermal cell that will
cells that contain the yellow myoplasm and
give rise to both muscle and endoderm. So, while there is a good correlation between
the cells adjacent to them.
muscle development and myoplasm, these observations on their own did not establish
After Conklin, E.G.: 1905.
that it is something in the myoplasm that is causing differentiation into muscle.
Experiments that altered the distribution of the myoplasm also provided suggestive,
but not conclusive, evidence that the myoplasm on its own can specify muscle cells.
But the most persuasive evidence that the myoplasm specifies muscle is provided by
more recent experiments that deplete it of a key maternal mRNA—macho-1 mRNA.
This is localized in the myoplasm in the egg and its depletion results in loss of the pri-
mary muscle cells in the tail. If the posterior-vegetal cytoplasm containing the ­macho-1
mRNA is removed from the fertilized egg, the blastomeres that would normally
develop as muscle develop as nerve cord. Conversely, injection of macho-1 mRNA
into non-muscle cell lineages causes ectopic muscle differentiation.

6.17 Notochord, neural precursors, and mesenchyme in ascidians require

inducing signals from neighboring cells

When first visible in the early tadpole, the notochord of the Ciona larva consists of
a single row of 40 cells aligned along the center of the tail, which undergo shape
changes to eventually form a hollow tube. The notochord derives mainly from A lin-
eage cells with a small contribution from the B lineage and requires induction by
adjacent vegetal cells for its formation. Blastomeres that normally give rise to noto-
chord do not do so if they are isolated at the 32-cell stage, unless they are combined
with vegetal blastomeres. However, prospective notochord cells isolated at the 110-cell
stage do develop into notochord.
The specification of the notochord precursor cells of the A lineage provides a good
example of how a pattern of asymmetric cell division, together with signals from
neighboring cells, can allocate daughter cells to different germ layers. Blastomeres
A7.3 and A7.7 in the 64-cell embryo give rise to notochord only, whereas their sister
cells, A7.4 and A7.8, respectively, give rise to spinal cord. (This applies also to the
corresponding cells on the right-hand side of the embryo. We will follow events on
just one side for simplicity.) The division that creates these cells with very different
fates occurs between the 32-cell and 44-cell stage (Fig. 6.32). When A6.2 and A6.4 in
the 32-cell embryo divide, the posterior daughter of each division, the one adjacent
to the vegetal blastomeres contributes to the notochord. The anterior daughter, adja-
cent to animal cap cells, gives rise to neural tissue (spinal cord).
248 Chapter 6 Development of nematodes, sea urchins, and ascidians

8-cell stage 24-cell stage 32-cell stage 44-cell stage 110-cell stage

A6.2 A7.3 A7.4

a4.2 A7.8
A6.4
A7.7
A4.1 A6.2
A6.4

A7.4 neural
A6.2
A7.3 notochord

A7.8 neural
A6.4
A7.7 notochord

Fig. 6.32 A binary choice between notchord and neural fates in A6.4 divide along their antero-posterior axis to give rise to one
ascidian embryos. The embryo is viewed from the vegetal pole, in the (posterior) notochord precursor each (red, A7.3 and A7.7) and one
same orientation as shown in the lower panel in Fig. 6.30. Anterior is (anterior) neural precursor each (blue, A7.4 and A7.8). Each precursor
up. All notochord cells are derived from blastomeres A4.1 and B4.1, cell then divides in the medio-lateral plane to generate four notochord
formed at the eight-cell stage. In the A lineage, the notochord derives precursors and four neural precursors in the 110-cell embryo.
from an asymmetric division of the the A6.2 and A6.4 mother cells After Picco, V., et al.: 2007.
at the late 32-cell stage. These cells were formed from A4.1 by two
successive rounds of cell division. At the 44-cell stage, A6.2 and

As we saw with the asymmetric division of the Caenorhabditis EMS cell into an MS
cell and an E cell (see Section 6.3), the assignment of notchord and neural fates to the
daughters of A6.2 and A6.4 is the result of a polarization of the mother cells by inter-
cellular signals followed by a functionally asymmetric division. The growth factor FGF
can induce notochord in vitro and is probably the inducer of a notochord fate in vivo,
as preventing FGF signaling in the early embryo prevents notochord formation. If the
FGF signaling pathway is blocked, both the daughter cells of A6.2 and A6.4 are neural
precursors. The acquisition of a neural fate requires contact with anterior animal cap
cells and Eph–ephrin signaling, which suppresses the notochord fate. The involve-
ment of Eph receptors and their ephrin ligands as signals that determine cell fate in
the early embryo came as something of a surprise. In other chordates they are usually
involved in adhesive cell–cell interactions, such as those that delimit rhombomere
boundaries in the chick brain (see Section 5.11), and in the selective adhesive interac-
tions that guide growing axons to their targets during nervous system development
(see Chapter 12).
Although the early development of ascidians is very different from that of verte-
brates, the presence of a notochord and its induction by vegetal cells is a striking
parallel between the two groups. Moreover, the same genes are involved in notochord
specification in both ascidians and vertebrates. The gene Brachyury is expressed in
early mesoderm in vertebrates and then becomes confined to the notochord (see
Section 4.18). Expression of the ascidian homolog of Brachyury is first detected at
the 64-cell stage in the A-lineage precursors of the notochord, and this stage appears
to correspond with the time at which induction is complete. Ectopic expression of the
ascidian Brachyury gene can transform endoderm to notochord. Thus, there appear to
be similar mechanisms involved in notochord formation in all chordates.
Specification of mesenchymal precursors (which will give rise to internal tissues)
from a mesodermal precursor also requires signals from adjacent cells. The mesen-
chyme precursors are located adjacent to the endodermal cells at the vegetal pole
 Ascidians 249

of the 110-cell embryo (see Fig. 6.30). When blastomeres are isolated at the 32-cell
stage, all those blastomeres whose normal fate is mesenchyme develop into muscle. It
appears that a signal from the endoderm, probably FGF, normally suppresses muscle
formation in these blastomeres. In the absence of such signals they develop into mus-
cle, as a result of the presence of maternal muscle determinants, such as Macho-1,
within them. The interplay between intercellular signals and intracellular determi-
nants in setting cell fate is complicated. Embryos lacking Macho-1 form notochord
in place of mesenchyme, whereas the overexpression of macho-1 causes the usual
notochord precursors to develop as mesenchyme.

Summary

In the chordate ascidians, there is evidence that localized cytoplasmic factors are
involved in specifying cell fate, particularly of muscle, but that cell interactions are also
involved. The notochord develops through a well-defined cell lineage but also requires
induction. The ascidian homolog of the vertebrate gene Brachyury, which is involved
in specifying the notochord in vertebrates, is expressed in the presumptive ascidian
notochord after induction. The third type of mesodermal tissue in ascidian embryos, the
mesenchyme, also requires inductive signals from neighboring cells.

Summary: ascidian early development

animal–vegetal axis

sperm entry point determines posterior side

localized maternal factors in egg

cytoplasmic factors determine blastomere fate by 110-cell stage

myoplasm determines muscle fate notochord and mesenchyme formed

by induction

Summary to Chapter 6

In many invertebrates, such as the nematode and the ascidian, cytoplasmic localization
and cell lineage, together with local intercellular interactions, seem to be very important in
specifying cell fate. However, there is very little evidence in any of these animals for key
signaling regions corresponding to the vertebrate organizer, or for morphogen gradients as
in Drosophila. This may reflect the predominance of mechanisms that specify cell fate on a
cell-by-cell basis, rather than by groups of cells as in vertebrates and insects.
Although genetic evidence of the kind available in Drosophila is lacking in other inverte-
brates, except in the case of the nematode Caenorhabditis elegans, it is clear that maternal
genes are crucial in specifying early patterning. There are numerous examples of the first
cleavage in the embryo being related to the establishment of the embryo’s axes and evi-
dence that this plane of cleavage is maternally determined. Asymmetric cleavage associated
with the unequal distribution of cytoplasmic determinants seems to be very important at
early stages. Cytoplasmic movements that are involved in this localization of determinants
often occur after fertilization, and in this respect they may resemble the cortical rotation of
the amphibian egg.
250 Chapter 6 Development of nematodes, sea urchins, and ascidians

There can be no doubt that there are cell interactions, even at early stages, in all of the
organisms in this chapter; an example is the specification of left–right asymmetry in nema-
todes, where a small alteration in cell relationships can reverse their handedness. But in
most of these embryos the interactions are highly local, only affecting an adjacent cell. This
is unlike vertebrates, in which interactions may occur over several cell diameters and involve
groups of cells.
By contrast, in sea-urchin development, which is highly regulative, cell interactions play a
major role, and thus echinoderms resemble the vertebrates in this respect. The micromeres
at the vegetal pole have properties similar to those of the vertebrate organizer, and cells
are specified in groups rather than on a cell-by-cell basis. Ascidians are members of the
chordates, the same phylum as vertebrates, and show a mixture of mosaic development and
cell–cell interactions.

n End of chapter questions 9. What are the similarities between gastrulation in Strongylocentrotus
­
and Xenopus that make them both deuterostomes?
Long answer (concept questions) 10. Compare the first two cleavages in the frog and sea urchin. In
1. What is meant by asymmetric cell division: does it necessarily what ways are they similar or different morphologically? Are the
mean that one daughter is bigger than the other? What other pos- embryos similar or different molecularly; that is, have the deter-
sibility qualifies a cell division event as being asymmetric? How are minative events that occur in Xenopus by the four-cell stage also
both types of asymmetric cell division illustrated by the first division occurred in sea urchins? Describe the differences in terms of ‘regula-
in Caenorhabditis elegans? tion’ during development.
2. Compare and contrast RNAi and miRNA gene silencing. Incor- 11. The micromeres of sea urchins act as an organizer. Describe
porate these hints into your answer: experimental versus natural; two experiments that support this statement (see Fig. 6.22).
~22 nucleotides; RISC; and degradation versus translation.
12. pmar1 is a transcription factor in the paired class of homeo­
3. How is the antero-posterior axis of C. elegans determined? What domain factors, which acts as a switch in the specification of
is the role of the PAR proteins in axis determination? ­endomesoderm (versus ectoderm) in sea urchins. Outline the role
of β-catenin and Dishevelled in activation of pmar1 in the microm-
4. During studies of cell lineage in C. elegans, a convention arose
eres (note that Wnt is not involved at this point). What signaling
that allowed the unambiguous identification of each of its 558 cells
pathway is activated later in the micromeres in response to pmar1 ­
(at hatching). Key to this system is the designation of the anterior
expression (see Fig. 6.24)?
daughter of each cell division with an ‘a’, and the posterior daughter
with a ‘p.’ Refer to Fig. 6.9 and describe why the indicated pharyn- 13. Sea-urchin larvae have two overt axes, animal–vegetal and
geal cell is called ‘MS paapaaa.’ Similarly, why is the apoptotic cell oral–aboral, and a more subtle internal left–right asymmetry. Com-
called ‘MS paapp’? pare the molecular bases of these three axes with the three axes
of frogs (this may be most easily done in the form of a table, with
5. To what class of signaling proteins do the C. elegans proteins
the six types of axes on the side and the signaling molecules and
GLP-1 and APX-1 belong? What is the role of the GLP-1/APX-1 sys-
transcription factors along the top, or in some other form of your
tem in dorso-ventral cell fate specification? ‘APX’ stands for ‘anterior
own choosing).
pharynx in excess;’ why was the mutant gene named ‘APX’ (refer to
Section 6.1 and Fig. 6.4)? 14. What is the myoplasm of ascidian embryos? How is the myo-
plasm related to cortical rotation, and what tissue will it form?
6. What are heterochronic genes? What feature of C. elegans devel-
opment makes this a favorable organism for the identification of 15. Why are ascidians chordates, but not vertebrates? What simi-
such genes? larities and differences exist between the formation of the notochord
of ascidians and the notochord of amphibian or chick ­embryos?
7. lin-4 encodes a miRNA that represses translation of lin-14 mRNA.
Since lin-14 maintains the Tap cell in a first larval (L1) state, this
Multiple choice (factual recall questions)
repression is required for the cell to transition to a L2 identity.
Based on this mechanism, what is the phenotype of a lin-14 loss-of- NB There is only one right answer to each question.
­function mutant? What is the phenotype of a lin-4 loss-of-function 1. C. elegans is best known for
mutant? What would be the phenotype of a worm with loss-of-func-
tion mutations in both genes? Explain your answer. a) its formation of an inductive signaling center analogous to the
organizer of Xenopus
8. The anchor cell of C. elegans is required for the induction of
b) its fully defined cell lineage during development
vulval development. To trace this induction, indicate which cells
express the EGF, EGF receptor, Delta, and Notch proteins, and the c) the large size and easy manipulation of its blastomeres
consequences of these proteins in those cells. d) its mechanisms of metamorphosis
 Ascidians 251

2. P granules of nematodes are b) Blastomeres at the four-cell stage each still possesses a portion
a) determinants of the antero-posterior axis of the original animal–vegetal axis, and if isolated will go on the
form a fully formed, yet smaller, larva.
b) determinants of the germ line
c) Blastomeres from the animal or vegetal poles of the embryo can
c) precursors of TGF-β family signaling proteins be isolated and will go on to form an embryo, as long as either
d) remnants of the sperm pronucleus an animal or vegetal pole is included.
d) Up until the hatching of the embryo as a pluteus larva, any blas-
3. In flies, frogs, and chicks, gradients of morphogens determine
tomere can be separated from the others and will regulate to go
the future antero-posterior and dorso-ventral axes of the deve­
on and develop into a fully formed and full-sized larva.
loping embryo. How is the antero-posterior axis determined in
C. elegans? 8. In which ways is the organizer region of the sea urchin similar
a) Bicoid protein is translated in the anterior of the fertilized egg, to that of frogs?
leading to a gradient that determines the antero-posterior axis. a) The organizer accumulates β-catenin, which in turn activates the
b) Opposing gradients of chordin and BMP-4 establish the antero- expression of cell–cell signaling molecules.
posterior axis. b) The organizer forms near the dorsal lip of the blastopore.
c) Sperm entry leads to a reorganization of the cytoskeleton and c) The organizer induces the dorso-ventral axis.
redistribution of maternally packaged PAR proteins, which in d) The organizer will express high levels of Nodal.
turn determine the antero-posterior axis.
9. In the ascidian Styela the yellow-pigmented myoplasm marks
d) β-catenin becomes localized to the nucleus in the future anterior cells fated to become muscle cells in the tail; however, at the molec-
cells after fertilization. ular level, the key determinant of tail muscle seems to be:
4. What are the lin-4 and lin-14 genes of C. elegans? a) expression of the Brachyury gene
a) lin-4 and lin-14 are C. elegans versions of the Hox genes. b) induction of muscle by FGF
b) lin-4 and lin-14 are genes that are named for their control of the c) localization of β-catenin to the nuclei in the vegetal portion of
first division, and hence the lineage of the AB and P1 cells. the embryo
c) lin-4 controls the timing of transcription of lin-14, and so is d) the gene product of the macho-1 gene
referred to a ‘heterochronic’ gene. 10. The role of the Brachyury gene in ascidians is similar to that in
d) lin-4 encodes a miRNA that represses lin-14 translation, which in Xenopus in that
turn regulates timing during larval development. a) Brachyury is the master-switch gene for muscle development on
both organisms
5. In C. elegans, the only Hox gene utilized during embryogenesis is
b) Brachyury sets up the animal–vegetal axis
a) ceh-13, the C. elegans ortholog of the Drosophila labial gene
c) signaling that leads to induction of the mesoderm results in
b) egl-5, the C. elegans ortholog of the Drosophila Abd-B and the
expression of the Brachyury gene
vertebrate Hox9-13 genes
d) the Brachyury gene is expressed specifically in those cells that
c) lin-39, a C. elegans gene related to the Drosophila Deformed and
will form somites, and hence muscle
Sex combs reduced family of homeodomain proteins
d) mab-5, a C. elegans gene related to the Drosophila Antennapedia,­ Multiple choice answer key
Ultrabithorax, and abdominal-A family of homeodomain 1: b, 2: b, 3: c, 4: d, 5: a, 6: c, 7: b, 8: a, 9: d, 10: c.
­proteins

6. Which of these model organisms is closest evolutionarily to n Further reading

h­umans?
a) Fruit flies, because of their sophisticated nervous system and the Nematodes
presence of legs. Sommer, R.J.: Evolution of development in nematodes related to
b) Plants, because their early divergence from animal lineages C. elegans (December 14, 2005), WormBook, ed. The C. elegans
means they are closely related to animals, especially humans. Research Community, WormBook, doi/10.1895/wormbook.1.46.1,
http://www.wormbook.org (date accessed 13 May 2010).
c) Sea urchins, despite their radial symmetry and lack of limbs. Sulston, J.E., Scherienberg, E., White, J.G., Thompson, J.N.: The
d) Worms, despite their few cells, small genomes, and abbreviated embryonic cell lineage of the nematode Caenorhabditis ­elegans.
complement of Hox genes. Dev. Biol. 1983, 100: 64–119.
Sulston, J.: Cell lineage. In The Nematode Caenorhabditis elegans.
7. Which of the following statements about regulation in sea-urchin Edited by Wood, WB. New York: Cold Spring Harbor Laboratory
embryos is true? Press, 1988: 123–156.
a) Any blastomere at the eight-cell stage can still regulate to form Wood, W.B.: Embryology. In The Nematode Caenorhabditis elegans.
a full larva, but by the 60-cell stage, blastomeres will develop Edited by Wood, WB. New York: Cold Spring Harbor Laboratory
according to their presumptive fate. Press, 1988: 215–242.
252 Chapter 6 Development of nematodes, sea urchins, and ascidians

6.1 The antero-posterior axis in Caenorhabditis elegans is Kidd, A.R. 3rd, Miskowski, J.A., Siegfried, K.R., Sawa, H., ­Kimble, J.:
determined by asymmetric cell division A beta-catenin identified by functional rather than sequence
Cheeks, R.J., Canman, J.C., Gabriel, W.N., Meyer, N., Strome, S., criteria and its role in Wnt/MAPK signaling. Cell 2005,
Goldstein, B.: C. elegans PAR proteins function by mobilizing 121: ­761–772.
and stabilizing asymmetrically localized protein complexes. Lyczak, R., Gomes, J.E., Bowerman, B.: Heads or tails: cell polarity
Curr. Biol 2004, 14: 851–862. and axis formation in the early Caenorhabditis elegans embryo.
Cowan, C.R., Hyman, A.A.: Acto-myosin reorganization and PAR Dev. Cell 2002, 3: 157–166.
polarity in C. elegans. Development 2007, 134: 1035–1043. Maduro, M.F.: Structure and evolution of the C. elegans embryonic­
Goldstein, B., Macara, I.G.: The PAR proteins: fundamental play- emdomesoderm network. Biochim. Biophys. Acta 2008, 1789:
ers in animal cell polarization. Dev. Cell 2007, 13: 609–622. 250–260.
Munro, E., Nance, J., Priess, J.R.: Cortical flows powered by asym- Mello, G.C., Draper, B.W., Priess, J.R.: The maternal genes
metrical contraction transport PAR proteins to establish and apx-1 and glp-1 and the establishment of dorsal-ventral polar-
maintain anterior-posterior polarity in the early C. elegans ity in the early C. elegans embryo. Cell 1994, 77: 95–106.
embryo. Dev. Cell 2004, 7: 413–424. Mizumoto, K., Sawa, H.: Two βs or not two βs: regulation of asym-
metric division by β-catenin. Trends Cell Biol. 2007, 17: 465–
Box 6A Gene silencing by antisense RNA and RNA 473.
interference Mizumoto, K., Sawa, H.: Cortical β-catenin and APC regulate
Brennecke, J., Malone, C.D., Aravin, A.A., Sachidanandam, R., asymmetric nuclear β-catenin localization during asymmetric
Stark, A., Hannon, G.J.: An epigenetic role for maternally cell division in C. elegans. Dev. Cell 2007, 12: 287–299.
inherited piRNAs in transposon silencing. Science 2008, 322: Park, F.D., Priess, J.R.: Establishment of POP-1 asymmetry in early­
1387–1392. C. elegans embryos. Development 2003 130: 3547–3556.
DasGupta, R., Nybakken, K., Booker, M., Mathey-Prevot, B., Park, F.D., Tenlen, J.R., Priess, J.R.: C. elegans MOM-5/frizzled
­Gonsalves, F., Changkakoty, B., Perrimon, N.: A case study of functions in MOM-2/Wnt-independent cell polarity and is
the reproducibility of transcriptional reporter cell-based RNAi localized asymmetrically prior to cell division. Curr. Biol. 2004,
screens in Drosophila. Genome Biol. 2007, 8: R203. 14: 2252–2258.
Dykxhoorn, D.M., Novind, C.D., Sharp, P.A.: Killing the messenger: Phillips, B.T., Kidd III, A.R., King, R., Hardin, J., Kimble, J.: Recip-
short RNAs that silence gene expression. Nat. Rev. Mol. Cell. Biol rocal asymmetry of SYS-1/β-catenin and POP-1/TCF controls
2003, 4: 457–467. asymmetric divisions in Caenorhabditis elegans. Proc. Natl
Eisen, J.S., Smith, J.C.: Controlling morpholino experiments: don’t Acad. Sci. USA 2007, 104: 3231–3236.
stop making antisense. Development 2008, 135: 1735–1743.
Kamath, R.S., Fraser, A.G., Dong, Y., Poulin, G., Durbin, R., Gotta, M.,­
6.4 Hox genes specify positional identity along the antero-
Kanapin, A., Le Bot, N., Moreno, S., Sohrmann, M., Welchman,
posterior axis in Caenorhabditis elegans
D.P., Zipperlen, P., Ahringer, J.: Systematic functional analysis­
Aboobaker, A.A., Blaxter, M.L.: Hox gene loss during dynamic
of the Caenorhabditis elegans genome using RNAi. Nature
evolution of the nematode cluster. Curr. Biol. 2003, 13: 37–40.
2003, 421: 231–237.
Nature Insight: RNA interference. Nature 2004, 431: 337–370. Clark, S.G, Chisholm, A.D., Horvitz, H.R.: Control of cell fates in
Sonnichsen, B., et al.: Full-genome RNAi profiling of early embryo- the central body region of C. elegans by the homeobox gene
genesis in Caenorhabditis elegans. Nature 2005, 434: 462–469. lin-39. Cell 1993, 74: 43–55.
Cowing, D., Kenyon, C.: Correct Hox gene expression established
6.2 The dorso-ventral axis in Caenorhabditis elegans is independently of position in Caenorhabditis elegans. Nature
determined by cell–cell interactions 1996, 382: 353–356.
Bergmann, D.C., Lee, M., Robertson, B., Tsou, M.F., Rose, L.S., Salser, S.J., Kenyon, C.: A C. elegans Hox gene switches on, off,
Wood, W.B.: Embryonic handedness choice in C. elegans involves on and off again to regulate proliferation, differentiation and
the Galpha protein GPA-16. Development 2003, 130: 5731–5740. morphogenesis. Development 1996, 122: 1651–1661.
Wood, W.B.: Evidence from reversal of handedness in C. elegans Van Auken, K., Weaver, D.C., Edgar, L.G, Wood, W.B.: Caenorhab-
embryos for early cell interactions determining cell fates. N
­ ature ditis elegans embryonic axial patterning requires two recently
1991, 349: 536–538. discovered posterior-group Hox genes. Proc. Natl Acad. Sci.
USA 2000, 97: 4499–4503.
6.3 Both asymmetric divisions and cell–cell interactions
specify cell fate in the early nematode embryo 6.5 The timing of events in nematode development is under
Eisenmann, D.M.: Wnt signaling (June 25, 2005), WormBook, ed. genetic control that involves microRNAs
The C. elegans Research Community, WormBook, doi/10.1895/ Ambros, V.: Control of developmental timing in Caenorhabditis
wormbook.1.7.1, http://www.wormbook.org (date accessed elegans. Curr. Opin. Genet. Dev. 2000, 10: 428–433.
13 May 2010). Austin, J., Kenyon, C.: Developmental timekeeping: marking time
Evans, T.C., Crittenden, S.L., Kodoyianni, V., Kimble, J.: Transla- with antisense. Curr. Biol. 1994, 4: 366–396.
tional control of maternal glp-1 mRNA establishes an asymme- Grishok, A., Pasquinelli, A.E., Conte, D., Li, N, Parrish, S., Ha,
try in the C. elegans embryo. Cell 1994, 77: 183–194. I., Baillie, D.L., Fire, A., Ruvkun, G., Mello, C.C.: Genes and
 Ascidians 253

­ echanisms related to RNA interference regulate expression of

m 6.9 The sea urchin fate map is finely specified, yet
the small temporal RNAs that control C. elegans developmental considerable regulation is possible
timing. Cell 2001, 106: 23–34. Angerer, L.M., Angerer, R.C.: Patterning the sea-urchin embryo:
gene regulatory networks, signaling pathways, and cellular
Box 6B Gene silencing by microRNAs
interactions. Curr. Top. Dev. Biol. 2003, 53: 159–198.
Bartel, D.P.: MicroRNAS: genomics, biogenesis, mechanism and Angerer, L.M., Oleksyn, D.W., Logan, C.Y., McClay, D.R., Dale, L.,
function. Cell 2004, 116: 281–297. Angerer, R.C.: A BMP pathway regulates cell fate allocation
He, L., Hannon, G.J.: MicroRNAs: small RNAs with a big role in along the sea urchin animal-vegetal embryonic axis. Develop-
gene regulation. Nat Rev. Genet. 2004, 5: 522–531. ment 2000, 127: 1105–1114.
Liu, J.: Control of protein synthesis and mRNA degradation by Sweet, H.C., Hodor, P.G., Ettensohn, C.A.: The role of mi­cromere
microRNAs. Curr. Opin. Cell Biol. 2008, 20: 214–221. signaling in Notch activation and mesoderm specification
Murchison, E.P., Hannon, G.J.: miRNAs on the move: miRNA bio- during sea-urchin embryogenesis. Development 1999, 126:
genesis and the RNAi machinery. Curr. Opin. Cell Biol. 2004, 5255–5265.
16: 223–229. Wessel, G.M., Wikramanayake, A.: How to grow a gut: ontogeny
of the endoderm in the sea-urchin embryo. BioEssays 1999, 21:
6.6 Vulval development is initiated by the induction of a
459–471.
small number of cells by short-range signals from a single
inducing cell 6.10 The vegetal region of the sea-urchin embryo acts as an
Kenyon, C.: A perfect vulva every time: gradients and signaling organizer
cascades in C. elegans. Cell 1995, 82: 171–174. Ransick, A., Davidson, E.H.: A complete second gut induced by
Sharma-Kishore, R., White, J.G., Southgate, E., Podbilewicz, B.: For- transplanted micromeres in the sea-urchin embryo. Science 1993,
mation of the vulva in Caenorhabditis elegans: a paradigm for 259: 1134–1138.
organogenesis. Development 1999, 126: 691–699.
Sundaram, M., Han, M.: Control and integration of cell signal- 6.11 The sea-urchin vegetal region is specified by the
ing pathways during C. elegans vulval development. BioEssays nuclear accumulation of β-catenin
1996, 18: 473–480. Ettensohn, C.A.: The emergence of pattern in embryogenesis:
Yoo, A.S., Bais, C., Greenwald, I.: Crosstalk between the EGFR and regulation of beta-catenin localization during early sea urchin
LIN-12/Notch pathways in C. elegans vulval development. Sci- development. Sci STKE 2006, 14: pe48.
ence 2004, 303: 663–636. Weitzel, H.E., Illies, M.R., Byrum, C.A., Xu, R., Wikramanayake, A.H.,
Ettensohn, C.A.: Differential stability of beta-catenin along the
Echinoderms animal-vegetal axis of the sea-urchin embryo mediated by
Horstadius, S.: Experimental Embryology of Echinoderms. Oxford: dishevelled. Development 2004, 131: 2947–2955.
Clarendon Press, 1973.
6.12 The genetic control of the skeletogenic pathway is
6.7 The sea-urchin embryo develops into a free-swimming known in considerable detail
larva Davidson, E.H.: A view from the genome: spatial control of tran-
Raff, R.A., Raff, E.C.: Tinkering: new embryos from old—rapidly scription in sea urchin development. Curr. Opin. Genet. Dev. 1999,
and cheaply. Novartis Found. Symp. 2007, 284: 35–45; discus- 9: 530–541.
sion 45–54, 110–115. Ettensohn, C.A.: Lessons from a gene regulatory network: echino-
Wilt, F.H.: Determination and morphogenesis in the sea-urchin derm skeletogenesis provides insights into evolution, plasticity
embryo. Development 1987, 100: 559–576. and morphogenesis. Development 2009, 136: 11–21.
Sea Urchin Genome Sequencing Consortium: The genome of the Oliveri, P., Davidson, E.H.: Gene regulatory network controlling
sea urchin Strongylocentrotus purpuratus. Science 2006, 314: embryonic specification in the sea urchin. Curr. Opin. Genet.
941–952. Erratum in: Science 2007, 315: 766. Dev. 2004, 14: 351–360.
Sea urchin interactive online poster: Science 2006 www.sciencemag. Oliveri, P., Tu, Q., Davidson, E.H.: Global regulatory logic for
org/sciext/seaurchin (date accessed 13 May 2010). specification of an embryonic cell lineage. Proc. Natl Acad. Sci.
USA 2008, 105: 5955–5962.
6.8 The sea-urchin egg is polarized along the animal–vegetal Kirchnamer, C.V., Yuh, C.V., Davidson, E.H.: Modular cis-regulatory­
axis organization of developmentally expressed genes: two genes
Davidson, E.H., Cameron, R.S., Ransick, A.: Specification of cell transcribed territorially in the sea-urchin embryo, and addi-
fate in the sea-urchin embryo: summary and some proposed tional examples. Proc. Natl Acad. Sci. USA 1996, 93: 9322–
mechanisms. Development 1998, 125: 3269–3290. 9328.
Henry, J.J.: The development of dorsoventral and bilateral axial
properties in sea-urchin embryos. Semin. Cell Dev. Biol. 1998, 9: 6.13 The oral–aboral axis in sea urchins is related to the
43–52. plane of the first cleavage
Logan, C.Y., McClay, D.R.: The allocation of early blastomeres Cameron, R.A., Fraser, S.E., Britten, R.J., Davidson, E.H.: The oral-
to the ectoderm and endoderm is variable in the sea-urchin aboral axis of a sea-urchin embryo is specified by first cleav-
embryo. Development 1997, 124: 2213–2223. age. Development 1989, 106: 641–647.
254 Chapter 6 Development of nematodes, sea urchins, and ascidians

6.14 The oral ectoderm acts as an organizing region for the 6.15 Animal–vegetal and antero-posterior axes in the
oral–aboral axis ascidian embryo are defined before first cleavage
Coffman, J.A., Coluccio, A., Planchart, A., Robertson, A.J.: Oral- Nishida, H.: Specification of embryonic axis and mosaic develop-
­aboral axis specification in the sea-urchin embryo III. Role of ment in ascidians. Dev. Dyn. 2005, 233: 1177–1193.
mitochondrial redox signaling via H2O2. Dev. Biol. 2009, 330: Sardet, C., Paix, A., Prodon, F., Dru, P., Chenevert, J.: From oocyte
123–130. to 16-cell stage: cytoplasmic and cortical reorganizations that
Duboc, V., Lepage, T.: A conserved role for the nodal signaling path- pattern the ascidian embryo. Dev. Dyn. 2007, 236: 1716–1731.
way in the establishment of dorso-ventral and left–right axes in
deuterostomes. J. Exp. Zool. (Mol. Dev. Evol.) 2008, 310B: 41–53. 6.16 In ascidians, muscle is specified by localized
Duboc, V., Lapraz, F., Besnardeau, L., Lepage, T.: Lefty acts as an cytoplasmic factors
essential modulator of Nodal activity during sea urchin oral- Bates, W.R.: Development of myoplasm-enriched ascidian
aboral axis formation. Dev. Biol. 2008, 320: 49–59. embryos. Dev. Biol. 1988, 129: 241–252.
Duboc, V., Rottinger, E., Besnardeau, L., Lepage, T.: Nodal and di Gregorio, A., Levine, M.: Ascidian embryogenesis and the ori-
BMP2/4 signaling organizes the oral-aboral axis of the sea- gins of the chordate body plan. Curr. Opin. Genet. Dev. 1998, 7:
urchin embryo. Dev. Cell 2004, 6: 397–410. 457–463.
Yaguchi, S., Yaguchi, J., Angerer, R.C., Angerer, L.M.: A Wnt-FoxQ2- Kim, G.J., Nishida, H.: Suppression of muscle fate by cellular
nodal pathway links primary and secondary axis specification interaction as required for mesenchyme formation during
in sea-urchin embryos. Dev. Cell 2008, 14: 97–107. ascidian embryogenesis. Dev Biol. 1999, 214: 9–22.
Nishida, H., Sawada, K.: macho-1 encodes a localized mRNA in
Ascidians ascidian eggs that specifies muscle fate during embryogenesis.
Dehal, P., Satou, Y., Campbell, R.K., Chapman, J., Degnan, B., ­Nature 2001, 409: 724–728.
De Tomaso, A., Davidson, B., Di Gregoriao, A., Gelpke, M.,
­Goodstein, D.M., et al.: The draft genome of Ciona intestinalis: 6.17 Notochord, neural precursors, and mesenchyme in
insights into chordate and vertebrate origins. Science 2002, 298: ascidians require inducing signals from neighboring cells
2157–2167. Corbo, J.C., Levine, M., Zeller, R.W.: Characterization of a
Keys, D.N., Lee, B.I., Di Gregorio, A., Harafuji, N., Detter, J.C., notochord-specific enhancer from the Brachyury promoter
Wang, M., Kahsai, O., Ahn, S., Zhang, C, Doyle, S.A., Satoh, N., region of the ascidian, Ciona intestinalis. Development 1997,
Satou, Y, Saiga, H., Christian, A.T., Rokhsar, D.S., Hawkins, T.L., 124: ­589–602.
Levine, M., Richardson, P.M.: A saturation screen for cis-acting Kim, G.J., Yamada, A., Nishida, H.: An FGF signal from endoderm
regulatory DNA in the Hox genes of Ciona intestinalis. Proc. and localized factors in the posterior-vegetal egg cytoplasm
Natl Acad. Sci. USA 2005, 102: 679–683. pattern the mesodermal tissues in the ascidian embryo. Devel-
Lemaire, P., Smith, W.C., Nishida, H.: Ascidians and the plastic- opment 2000, 127: 2853–2862.
ity of the chordate developmental program. Curr. Biol. 2008, 18: Nakatani, Y., Nishida, H.: Induction of notochord during ascidian
R620–R631. embryogenesis. Dev. Biol. 1994, 166: 289–299.
Passamaneck, Y.J., Di Gregorio, A.: Ciona intestinalis: chordate Nakatani, Y., Yasuo, H., Satoh, N., Nishida, H.: Basic fibroblast
development made simple. Dev. Dyn. 2005, 233: 1–19. growth factor induces notochord formation and the expression
Satoh, N.: The Developmental Biology of Ascidians. Cambridge: of As-T, a Brachyury homolog, during ascidian embryogenesis.
Cambridge University Press, 1994. Development 1996, 122: 2023–2031.
Shoguchi, E., Hamaguchi, M., Satoh, N.: Genome-wide network of Picco, V., Hudson, C., Yasuo, H.: Ephrin–Eph signalling drives the
regulatory genes for construction of a chordate embryo. Dev. asymmetric division of notochord/neural precursors in Ciona
Biol. 2008, 316: 498–509. embryos. Development 2007, 134: 1491–1497.
 7
Plant development
nn Embryonic development nn Meristems nn Flower development and
control of flowering

Because of plant cells’ rigid cell walls and lack of cell movement within tissues,
a plant’s architecture is very much the result of patterns of oriented cell divisions.
Despite this apparent invariance, however, cell fate in development is largely deter-
mined by similar means as in animals—by a combination of positional signals and
­intercellular communication. As well as communicating by extracellular signals and
cell–surface interactions, plant cells are interconnected by cytoplasmic channels, which
allow movement of proteins such as transcription factors directly from cell to cell.

The plant kingdom is very large, ranging from the algae, many of which are unicel-
lular, to the multicellular land plants, which exist in a prodigious variety of forms.
Plants and animals probably evolved the process of multicellular development inde-
pendently, their last common ancestor being a unicellular eukaryote some 1.6 billion
years ago. Plant development is, therefore, of interest not simply for its own sake and
for its agricultural importance. By looking at the similarities and differences between
plant and animal development, a study of plant development can shed light on the
way that developmental mechanisms in two groups of multicellular organisms have
evolved independently and under different sets of developmental constraints.
Do plants and animals use the same developmental mechanisms? As we shall see
in this chapter, the logic behind the spatial layouts of gene expression that pattern a
developing flower is similar to that of Hox gene action in patterning the body axis in
animals, but the genes involved are completely different. Such similarities between
plant and animal development are due to the fact that the basic means of regulat-
ing gene expression are the same in both, and thus similar general mechanisms for
patterning gene expression in a multicellular tissue are bound to arise. As we shall
see, many of the general control mechanisms we have encountered in animal devel-
opment, such as asymmetric cell division, the response to positional signals, lateral
inhibition, and changes in gene expression in response to extracellular signals, are all
present in plants. Differences between plants and animals in the way development is
controlled arise from some different ways in which plant cells can communicate with
each other compared with animal cells, from the existence of rigid cell walls and the
lack of large-scale cell movements, to the fact that environment has a much greater
impact on plant development than on that of animals.
One general difference between plant and animal development is that most of the
development occurs not in the embryo but in the growing plant. Unlike an animal
embryo, the mature plant embryo inside a seed is not simply a smaller version of
256 Chapter 7 Plant development

the organism it will become. All the ‘adult’ structures of the plant—shoots, roots,
stalks, leaves, and flowers—are produced after germination from localized groups
of undifferentiated cells known as meristems. Two meristems are established in the
embryo: one at the tip of the root and the other at the tip of the shoot. These persist
in the adult plant, and almost all the other meristems, such as those in developing
leaves and flower shoots, are derived from them. Cells within meristems can divide
repeatedly and can potentially give rise to all plant tissues and organs. This means
that developmental patterning within meristems to produce organs such as leaves and
flowers continues throughout a plant’s life.
Plant and animal cells share many common internal features and much basic bio-
chemistry, but there are some crucial differences that have a bearing on plant develop-
ment. One of the most important is that plant cells are surrounded by a framework of
relatively rigid cell walls. There is therefore virtually no cell migration in plants, and
major changes in the shape of the developing plant cannot be achieved by the move-
ment and folding of sheets of cells. In plant development, form is largely generated
by differences in rates of cell division and by division in different planes, followed by
directed enlargement of the cells.
As in animal development, one of the main questions in plant development is how
cell fate is determined. Many structures in plants normally develop by stereotyped
patterns of cell division, but despite this observation, which implies the importance
of lineage, cell fate is, in many cases, also known to be determined by factors such as
position in the meristem and cell–cell signaling. The cell wall would seem to impose
a barrier to the passage of large signaling molecules, such as proteins, although it
is very thin in some regions, such as the meristems; but all known plant extracellu-
lar signaling molecules—such as auxins, gibberellins, cytokinins, and ethylene—are
small molecules that easily penetrate cell walls. Plant cells also communicate with
each other through fine cytoplasmic channels known as plasmodesmata, which link
neighboring plant cells through the cell wall, and they may be the channels by which
some developmentally important gene-regulatory proteins, and even mRNAs, move
directly between cells. The size of the channel varies and those in shoot cells have
the largest diameter.
Another important difference between plant and animal cells is that a complete,
fertile plant can develop from a single differentiated somatic cell and not just from
a fertilized egg. This suggests that, unlike the differentiated cells of adult animals,
some differentiated cells of the adult plant may retain totipotency. Perhaps they do
not become fully determined in the sense that adult animal cells do, or perhaps they
are able to escape from the determined state, although how this could be achieved
is as yet unknown. In any case, this difference between plants and animals illus-
trates the dangers of the wholesale application to plant development of concepts
derived from animal development. Nevertheless, the genetic analysis of development
in plants is turning up instances of genetic strategies for developmental patterning
rather similar to those of animals.
The small cress-like weed Arabidopsis thaliana has become the model plant for
genetic and developmental studies, and will provide many of the examples in this
chapter. We will begin by describing the main features of its morphology, life cycle,
and reproduction.

7.1 The model plant Arabidopsis thaliana has a short life-cycle and a small
diploid genome

The equivalent of Drosophila in the study of plant development is the small crucifer
Arabidopsis thaliana, commonly called thale cress, which is well-suited to genetic
and developmental studies. It is a diploid (unlike many plants, which are polyploid)
 Embryonic development 257

ovule
Adult flower Gametogenesis micropyle
egg cell
Fertilization
stamen embryo
sac
carpel chalaza pollen grain
pollen tube
petal pollen ovule
sepal

1
embryo
6

2
*Weeks (at 25oC) Embryogenesis
Growth and
maturation
5
3
cotyledons 4

hypocotyl

Germination
root
Seed

Seedling
*In continuous illumination

Fig. 7.1 Life cycle of Arabidopsis. In

and has a relatively small, compact genome that has been sequenced and which flowering plants, egg cells are contained
contains about 27,000 protein-coding genes. It is an annual, flowering in the first year separately in ovules inside the carpels.
of growth, and develops as a small ground-hugging rosette of leaves, from which a Fertilization of an egg cell by a male nucleus
branched flowering stem is produced with a flowerhead, or inflorescence, at the end from a pollen grain takes place inside the
of each branch. It develops rapidly, with a total life cycle in laboratory conditions of ovary. The egg then develops into an embryo
some 6–8 weeks, and like all flowering plants, mutant strains and lines can easily be contained within the ovule coat, forming a
stored in large quantities in the form of seeds. The life cycle of Arabidopsis is shown seed. Arabidopsis is a dicotyledon, and the
in Fig. 7.1. mature embryo has two wing-like cotyledons
(storage organs) at the apical (shoot) end of
Each Arabidopsis flower (Fig. 7.2) consists of four sepals surrounding four white
the main axis—the hypocotyl—which contains
petals; inside the petals are six stamens, the male sex organs, which produce pollen
a shoot meristem at one end and a root
containing the male gametes. Petals, sepals, and the other floral organs are thought to meristem at the other. Following germination,
derive evolutionarily from modified leaves. At the center of the flower are the female the seedling develops into a plant with roots,
sex organs, which consist of an ovary of two carpels, which contain the ovules. Each a stem, leaves, and flowers. The photograph
ovule contains an egg cell. Fertilization of an egg cell occurs when a pollen grain shows five mature Arabidopsis plants.
deposited on the carpel surface grows a tube that penetrates the carpel and delivers
two haploid pollen nuclei to an ovule. One nucleus fertilizes the egg cell, while the
other fuses with two other nuclei in the ovule. This forms a triploid cell that will
develop into a specialized nutritive tissue—the endosperm—that surrounds the fertil-
ized egg cells and provides the food source for embryonic development.
Following fertilization, the embryo develops inside the ovule, taking about 2 weeks
to form a mature seed, which is shed from the plant. The seed will remain dormant
until suitable external conditions trigger germination. The early stages of germination
and seedling growth rely on food supplies stored in the cotyledons (seed leaves),
which are storage organs developed by the embryo. Arabidopsis embryos have two
258 Chapter 7 Plant development

cotyledons, and thus belong to the large group of plants known as dicotyledons
(dicots for short). The other large group of flowering plants is the monocotyledons
(monocots), which have embryos with one cotyledon; monocots typically have long,
narrow leaves and include many important staple crops, such as wheat, rice, and
maize. Agriculturally important dicots include potato, tomato, sugar beet, and most
other vegetables.
At germination, the shoot and root elongate and emerge from the seed. Once the
shoot emerges above ground it starts to photosynthesize and forms the first true leaves
at the shoot apex. About 4 days after germination the seedling is a self-supporting
plant. Flower buds are usually visible on the young plant 3–4 weeks after germination
and will open within a week. Under ideal conditions, the complete Arabidopsis life
cycle thus takes about 6–8 weeks (see Fig. 7.1).

Embryonic development

Formation of the embryo, or embryogenesis, occurs inside the ovule, and the end
Fig. 7.2 An individual flower of result is a mature, dormant embryo enclosed in a seed awaiting germination. Dur-
Arabidopsis. Scale bar =  1 mm. ing embryogenesis, the shoot–root polarity of the plant body, which is known as
the apical–basal axis, is established, and the shoot and root meristems are formed.
Plants also possess radial symmetry, as seen in the concentric arrangement of the
different tissues in a plant stem, and this radial axis is also set up in the embryo.
Development of the Arabidopsis embryo involves a rather invariant pattern of cell
division (which is not the case in all plant embryos) and this enables structures in
the Arabidopsis seedling to be traced back to groups of cells in the early embryo to
provide a fate map (see Section 7.2).

7.2 Plant embryos develop through several distinct stages

Arabidopsis belongs to the angiosperms, or flowering plants, one of the two major
groups of seed-bearing plants; the other is the gymnosperms, or conifers. The typical
course of embryonic development in angiosperms is outlined in Box 7A. Like an ani-
mal zygote, the fertilized plant egg cell undergoes repeated cell divisions, cell growth,
and differentiation to form a multicellular embryo. The first division of the zygote is
at right angles to the long axis, dividing it into an apical cell and a basal cell, and
establishing an initial polarity that is carried over into the apical–basal polarity of the
embryo and into the apical–basal axis of the plant. In many species, the first zygotic
division is unequal, with the basal cell larger than the apical cell. The basal cell
divides to give rise to the suspensor, which may be several cells long (see Box 7A).
This attaches the embryo to maternal tissue and is a source of nutrients. The apical
cell divides vertically to form a two-celled proembryo, which will give rise to the rest
of the embryo. In some species, the basal cell contributes little to the further develop-
ment of the embryo, but in others, such as Arabidopsis, the topmost suspensor cell is
recruited into the embryo, where it is known as the hypophysis, and contributes to
the embryonic root meristem and root cap.
The next two divisions produce an eight-cell octant-stage embryo, which develops
into a globular-stage embryo of around 32 cells (Fig. 7.3). The embryo elongates and
the cotyledons start to develop as wing-like structures at one end, while an embry-
onic root forms at the other. This stage is known as the heart stage. Two groups of
undifferentiated cells capable of continued division are located at each end of this
axis; these are known as the apical meristems. The meristem lying between the coty-
ledons gives rise to the shoot, while the one at the opposite end of the axis, towards
the end of the embryonic root, will drive root growth at germination. The region in
between the embryonic root and the future shoot will become the seedling stem or
 Embryonic development 259

Box 7A Angiosperm embryogenesis

In flowering plants (angiosperms), the egg cell is contained within an ovule inside the
ovary in the flower (right inset in the bottom panel). At fertilization, a pollen grain depos-
cotyledons
ited on the surface of the stigma puts out a pollen tube, down which two male gametes
migrate into the ovule. One male gamete fertilizes the egg cell while the other combines
with another cell inside the ovule to form a specialized nutritive tissue, the endosperm, apical meristem
which surrounds, and provides the food source for, the developing embryo.
The small annual weed Capsella bursa-pastoris (shepherd’s purse) is a typical dicoty-
ledon. The first, asymmetric, cleavage divides the zygote transversely into an apical and hypocotyl
epidermis
a basal cell (bottom panel). The basal cell then divides several times to form a single row
of cells—the suspensor—which, in many angiosperm embryos, takes no further part in
cortex
embryonic development, but may have an absorptive function; in Capsella, however, it vascular
contributes to the root meristem. Most of the ­embryo is derived from the apical cell. This tissue

undergoes a series of stereotyped divisions, in which a precise pattern of cleavages in

root
different planes gives rise to the heart-shaped embryonic stage typical of dicotyledons.
This develops into a mature embryo that consists of a cylindrical body with a meristem at
­either end, and two cotyledons.
root meristem
The early embryo becomes differentiated along the radial axis into three main tissues—
the outer epidermis, the prospective vascular tissue, which runs through the center of
the main axis and cotyledons, and the ground tissue (prospective cortex) that surrounds it.
The ovule containing the embryo matures into a seed (left inset in the bottom panel), which remains dormant until suitable
external conditions trigger germination and growth of the seedling. A typical dicotyledon seedling (top right panel) ­comprises the
shoot apical meristem, two cotyledons, the trunk of the seedling (the hypocotyl), and the root apical meristem. The seedling may
be thought of as the phylotypic stage of flowering plants. The seedling body plan is simple. One axis—the apical–basal ­axis—defines
the main polarity of the plant. The shoot forms at the apical pole and the root at the basal pole. The shoot meristem is, therefore,
referred to in the seedling and adult plant simply as the apical meristem. A plant stem also has a radial axis, which is evident in the
radial symmetry in the hypocotyl and is continued in the root and shoot. In the center is the vascular tissue, which is surrounded
by cortex, and an outer covering of epidermis. At later stages, structures such as leaves and other organs have a dorso-ventral axis
running from the upper surface to the lower surface.

Egg Two cell Octant Globular Triangular Heart Torpedo Mature embryo

apical cell cotyledons shoot apical meristem epidermis

fertilized basal cell

egg

suspensor

basal cell

Seed Flower
embryo petal
root vascular
apical tissue
meristem
stamen stigma
style ground
ovary tissue

ovule
seed coat cotyledons sepal cotyledons
260 Chapter 7 Plant development

Globular stage Early heart stage Late heart stage Mature embryo stage
shoot meristem

cotyledons

suspensor

Fig. 7.3 Arabidopsis embryonic

development. Light micrographs (Nomarski hypocotyl. Almost all above ground adult plant structures are derived from the apical
optics) of cleared wild-type seeds of meristems. The main exception is radial growth in the stem, which is most evident
A. thaliana. The cotyledons can already be in woody plants, and which is produced by the cambium, a ring of secondary mer-
seen at the heart stage. The embryo proper istematic tissue in the stem. After the embryo is mature, the apical meristems remain
is attached to the seed coat through a quiescent until germination.
filamentous suspensor. Scale bar = 20 μm.
Seedling structures can be traced back to groups of cells in the early embryo to
Photographs courtesy of D. Meinke, from provide a fate map (Fig. 7.4). In Arabidopsis, patterns of cleavage up to the 16-cell
Meinke, D.W.: 1994.
stage are highly reproducible, and even at the octant stage it is possible to make a fate
map for the major regions of the seedling along the apical–basal axis. The upper tier
of cells gives rise to the cotyledons and the shoot meristem, the next tier is the origin
of the hypocotyl, and the bottom tier together with the region of the suspensor where
it joins the embryo will give rise to the root (see Section 7.3). At the heart stage, the
fate map is clear.
The radial pattern in the embryo comprises three concentric rings of tissue: the
outer epidermis, the ground tissue (cortex and endodermis), and the vascular tissue
at the center. This radial axis appears first at the octant stage, when adaxial (central)
and abaxial (peripheral) regions become established. Subsequently, periclinal divi-
sions, in which the plane of division is parallel to the outer surface, give rise to the
different rings of tissue, and anticlinal divisions, in which the plane of cell division is
at right angles to the outer surface, increase the number of cells in each ring of tissue
(Fig. 7.5). At the 16-cell stage the epidermal layer, or dermatogen, is established.
It is not known to what extent the fate of cells is determined at this stage or
whether their fate is dependent on the early pattern of asymmetric cell divisions. It
seems that cell lineage is not crucial, as mutations have been discovered that show
that cell division can be uncoupled from pattern formation. The fass mutation in
Arabidopsis disrupts the regular pattern of cell division by causing cells to divide in
random orientation, and produces a much fatter and shorter seedling than normal.
But although the fass seedling is misshapen, it has roots, shoots, and even eventu-
ally flowers in the correct places and the correct pattern of tissues along the radial
axis is maintained.

7.3 Gradients of the signal molecule auxin establish the embryonic

apical–basal axis

The small, organic molecule auxin (indoleacetic acid or IAA) is one of the most
important and ubiquitous chemical signals in plant development and plant growth. It
causes changes in gene expression by promoting the ubiquitination and degradation
 Embryonic development 261

Octant Globular Heart Torpedo

Seedling

cotyledons

Fig. 7.4 Fate map of the Arabidopsis

shoot embryo. The stereotyped pattern of cell
meristem division in dicotyledon embryos means that
hypocotyl
at the globular stage it is already possible
to map the three regions that will give rise
root to the cotyledons (dark green) and shoot
root meristem
meristem (red), the hypocotyl (yellow), and
the root meristem (purple) in the seedling.
After Scheres, B., et al.: 1994.

Periclinal divisions Anticlinal divisions

Fig. 7.5 Periclinal and anticlinal

divisions. Periclinal cell divisions are parallel
to the organ surface, whereas anticlinal
divisions are at right angles to the surface.

of transcriptional repressors known as Aux/IAA proteins. In the absence of auxin,

these bind to proteins called auxin-response factors (ARFs) to block the transcrip-
tion of so-called auxin-responsive genes. Auxin-stimulated degradation of Aux/IAA
proteins frees the auxin-response factors, which can then activate, or in some cases
repress, these genes (Fig. 7.6). Auxin-responsive genes include genes involved in the
regulation of cell division and cell expansion, as well as genes involved in specifying
cell fate. In some cases auxin appears to be acting as a classical morphogen, forming
a concentration gradient and specifying different fates according to a cell’s position
along the gradient.
The earliest known function of auxin in Arabidopsis is in the very first stage of
embryogenesis, where it establishes the apical–basal axis. Immediately after the first
division of the zygote, auxin is actively transported from the basal cell into the apical
cell, where it accumulates. It is transported out of the basal cell by the auxin-efflux
protein PIN7, which is localized in the apical plasma membrane of the basal cell. The
auxin is required to specify the apical cell, which gives rise to all the apical embryonic
structures such as shoot apical meristem and cotyledons. Through the subsequent cell
divisions, transport of auxin continues up through the suspensor cells and into the
262 Chapter 7 Plant development

cell at the base of the developing embryo until the globular-stage embryo of about
32 cells. The apical cells of the embryo then start to produce auxin, and the direction
TCTCTC of auxin transport is suddenly reversed. PIN7 proteins in the suspensor cells move to
ARF Auxin-responsive gene
the basal faces of the cells. Other PIN genes are activated, and the concerted actions
of the PIN proteins cause auxin from the apical region to be transported into the basal
AUX/IAA
region of the globular embryo, from which will develop the hypocotyl, root meristem,
Auxin and embryonic root (Fig. 7.7).
Expression of the homeobox genes WOX2 and WOX8 in the asymmetrically dividing
Ub zygote seems to be essential for the formation of the embryonic auxin gradient. Both
Ub Ub
AUX/IAA Ub genes are expressed in the zygote, but after division, WOX2 expression is ­restricted to
degradation AUX/IAA the apical cell and its descendants whereas WOX8 is expressed only in the basal cell
Auxin lineage. Expression of WOX8 appears to be required for the continued expression of
SCF TIR1 WOX2 and for the establishment of an auxin gradient, most probably by effects of the
WOX proteins on expression of the PIN proteins that direct auxin transport.
The importance of auxin in specifying the basal region of the embryo is illustrated
by the effects of mutations in the cellular machinery for auxin transport or auxin
TCTCTC response. The gene MONOPTEROS (MP), for example, encodes one of the auxin-
ARF ARF transcription response factors, ARF5. Embryos mutant for MP lack the hypocotyl and the root
meristem, and abnormal cell divisions are observed in the regions of the octant-stage
embryo that would normally give rise to these structures.
Fig. 7.6 Auxin signaling pathway. In the An early step in axis formation in the embryo is the specification of cells as ­potential
absence of auxin, AUX/IAA proteins bind shoot or root. The mutation topless-1 causes a dramatic switching of cell fate, trans-
to and repress the activity of transcription
forming the whole of the apical region into a second root region and abolish­ing
factors in the AUXIN RESPONSE FACTOR
­development of cotyledons and shoot meristem. The normal function of the TOPLESS
(ARF) family, which bind TGTCTC-containing
protein is to repress expression of root-promoting genes in the top half of the early
DNA sequence elements in the promoters
of auxin-responsive genes. Auxin targets embryo. It does this by acting as a transcriptional co-repressor, forming a complex
the ubiquitin ligase complex SCF/TIR1 together with Aux/IAA proteins and the auxin-response factors they regulate.
to the AUX/IAA protein, causing it to be The topless-1 mutation is temperature-sensitive, and so can be made to act at dif-
ubiquitinated (Ub). This modification targets ferent times during development by simply changing the temperature at which the
AUX/IAA for degradation and thus lifts the embryos are grown. Such manipulations show that the apical region can be respeci-
repression of the auxin-responsive gene, fied as root between the globular and heart stage, even after it has begun to show
which can then be transcribed. molecular signs of developing cotyledons and shoot meristem. This suggests that
Adapted from Chapman, E.J. and Estelle, M.: even though apical cell fate is normally specified by this stage of embryonic develop-
2009. ment, the decision is not irreversible.
Mutations in the gene SHOOT MERISTEMLESS (STM) completely block the for-
Proembryo Young globular embryo mation of the shoot apical meristem, but have no effect on the root meristem or
Apical hydrophysis other parts of the embryo. STM encodes a transcription factor that is also required
to ­maintain cells in the pluripotent state in the adult shoot meristem. The pattern of
STM expression develops gradually, which is also typical of several other genes that
characterize the shoot apical meristem. Expression is first detected in the globular
suspensor
cell stage in one or two cells and only later in the central region between the two cotyle-
dons (Fig. 7.8).

Fig. 7.7 The role of auxin in patterning the early embryo. Left panel: auxin produced in the
original basal cell accumulates in the two-celled proembryo (green) through transport in the
basal to apical direction mediated by the protein PIN7, which is located in the apical membranes
Basal of the basal and suspensor cells (purple arrows). Another PIN protein, PIN1, transports auxin
between the two cells of the proembryo (red arrows). Cell division distributes auxin into the
auxin response production of auxin developing embryo. Right panel: at the globular stage, free auxin starts to be produced at the
and accumulation apical pole and the direction of auxin transport is reversed. PIN7 becomes localized in basal
auxin efflux mediated by: membranes of the suspensor cells and the proteins PIN1 and PIN4 (orange arrows) transport
PIN1 PIN4 PIN7 auxin from the apical region into the most basal cell of the embryo, known as the hypophysis,
where it accumulates.
 Embryonic development 263

7.4 Plant somatic cells can give rise to embryos and seedlings

As gardeners well know, plants have amazing powers of regeneration. A complete,

new plant can develop from a small piece of stem or root, or even from the cut edge
of a leaf. This reflects an important difference between the developmental potential
of plant and animal cells. In animals, with few exceptions, cell determination and
differentiation are irreversible. By contrast, many somatic plant cells remain toti-
potent. Cells from roots, leaves, stems, and even, for some species, a single, iso-
lated ­protoplast—a cell from which the cell wall has been removed by enzymatic
treatment—can be grown in culture and induced by treatment with the appropriate
growth hormones to give rise to a new plant (Fig. 7.9). Plant cells proliferating in
culture can give rise to cell clusters that pass through a stage resembling normal
embryonic development, although the pattern of cell divisions are not the same.
These ‘embryoids’ can then develop into seedlings. Plant cells can also form callus,
an apparently disorganized mass of cells, and the callus can form new shoot or root
meristems, and therefore new shoots and roots. The ability of plants to regenerate
from somatic cells means that transgenic plants can be easily generated using the
plant pathogenic bacterium Agrobacterium tumefaciens as the carrier of the gene to Fig. 7.8 Section through a late heart-stage
be transferred (Box 7B, p. 264). Arabidopsis embryo showing expression of
The ability of single somatic cells to give rise to whole plants has two important SHOOT MERISTEMLESS (STM). At this stage,
implications for plant development. The first is that maternal determinants may be the STM RNA (stained red) is expressed
of little or no importance in plant embryogenesis, as it is unlikely that every somatic in cells located between the cotyledons.
cell would still be carrying such determinants. Second, it suggests that many cells in Scale bar = 25 μm.
the adult plant body are not fully determined with respect to their fate, but remain Photograph courtesy of K. Barton. From
­totipotent. Of course, this totipotency seems only to be expressed under special condi- Long, J.A. and Barton, K.B.: 1998.
tions, but it is, nevertheless, quite unlike the behavior of animal cells. It is as if such
plant cells have no long-term developmental memory, or that such memory is easily
reset.

Callus separated and Further culturing

Tissue sample from any region of an adult plant is cultured Callus forms generates new plant
single cells cultured

bud tissue
sample

leaf callus

stem

root

Fig. 7.9 Cultured somatic cells from a mature plant can form an The callus cells are then separated and grown in liquid culture, again
embryo and regenerate a new plant. The illustration shows the containing the appropriate growth hormones. In suspension culture,
generation of a plant from single cells. If a small piece of tissue from some of the callus cells divide to form small cell clusters. These cell
a plant stem or leaf is placed on a solid agar medium containing the clusters can resemble the globular stage of a dicotyledon embryo, and
appropriate nutrients and growth hormones, the cells will start to with further culture on solid medium, develop through the heart-
divide to form a disorganized mass of cells known as a callus. shaped and later stages to regenerate a complete new plant.
264 Chapter 7 Plant development

Box 7B Transgenic plants

One of the most common ways of generating transgenic plants i­ntegrated. Genes experimentally inserted into the T-DNA will
containing new and modified genes is through infection of plant therefore also be transferred into the plant cell chromosomes.
tissue in culture with the bacterium Agrobacterium tumefaciens, Ti plasmids, modified so that they do not cause tumors but still
the causal agent of crown gall tumors. Agrobacterium is a natu- retain the ability to transfer T-DNA, are widely used as vectors
ral genetic engineer. It contains a plasmid—the Ti plasmid—that for gene transfer in dicotyledonous plants. The genetically modi-
contains the genes required for the transformation and prolifera- fied plant cells of the callus can then be grown into a complete
tion of infected cells to form a callus. During infection, a portion new transgenic plant that carries the introduced gene in all its
of this plasmid—the T-DNA (shown in red below)—is transferred cells and can transmit it to the next generation.
into the genome of the plant cell, where it becomes stably

Leaf discs infected with Agrobacterium

Leaf discs removed and placed in culture medium containing recombinant plasmid Transgenic plants grow

plasmid

T-DNA

Calluses form; cells contain T-DNA

Summary

Early embryonic development in many plants is characterized by asymmetric cell divi-

sion of the zygote, which specifies apical and basal regions. Embryonic development
in flowering plants establishes the shoot and root meristems from which the adult
plant develops. Auxin gradients are involved in specifying the apical–basal axis of the
­Arabidopsis embryo. One major difference between plants and animals is that, in culture,
a single somatic plant cell can develop through an embryo-like stage and regenerate a
complete new plant, indicating that some differentiated plant cells retain totipotency.

Summary: early development in flowering plants

first asymmetric cell division and auxin signaling in embryo establishes apical–basal axis

embryonic cell fate is determined by position

shoot and root meristems of seedling give rise to all adult plant structures
 Meristems 265

Meristems apical
meristem
In plants, most of the adult structures are derived from just two regions of the embryo, axillary bud
the embryonic shoot and root meristems, which are maintained after germination. The
embryonic shoot meristem, for example, becomes the shoot apical meristem of the node
growing plant, giving rise to all the stems, leaves, and flowers. As the shoot grows, leaf
lateral outgrowths from the meristem give rise to leaves and to side shoots. In flower-
ing shoots, the vegetative meristem becomes converted into one capable of producing
node
floral meristems that make flowers, not leaves. In Arabidopsis, for example, the shoot
apical meristem changes from a vegetative meristem that makes leaves around it in a
spiral pattern to an inflorescence meristem that then produces floral meristems, and
thus flowers, around it in a spiral pattern. The first stages of future organs are known
as primordia (singular primordium). Each primordium consists of a small number node
of founder cells that produce the new structure by cell division and cell enlargement,
accompanied by differentiation. internode
There is usually a time delay between the initiation of two successive leaves in a (stem)
shoot apical meristem, and this results in a plant shoot being composed of repeated
modules. Each module consists of an internode (the cells produced by the meristem node
between successive leaf initiations), a node and its associated leaf, and an axillary
bud (Fig. 7.10). The axillary bud itself contains a meristem, known as the lateral
Fig. 7.10 Plant shoots grow in a modular
shoot meristem, which forms at the base of the leaf, and can produce a side shoot
fashion. The shoot apical meristem produces
when the inhibitory influence of the main shoot tip is removed. Root growth is not so
a repeated basic structural module. The
obviously modular, but similar considerations apply, as new lateral meristems initi- vegetative shoot module typically consists of
ated behind the root apical meristem give rise to lateral roots. internode, node, leaf, and axillary bud (from
Shoot apical meristems and root apical meristems operate on the same principles, which a side branch may develop). Successive
but there are some significant differences between them. We will use the shoot apical modules are shown here in different shades
meristem to illustrate the basic principles of meristem structure and properties, and of green. As the plant grows, the internodes
then discuss roots. behind the meristem lengthen and the leaves
expand.
After Alberts, B., et al.: 2002.
7.5 A meristem contains a small, central zone of self-renewing stem cells

Shoot apical meristems are rarely more than 250 μm in diameter in angiosperms
and contain a few hundred relatively small, undifferentiated cells that are capable
of cell division. Most of the cell divisions in normal plant development occur within
meristems, or soon after a cell leaves a meristem, and much of a plant’s growth
in size is due to cell elongation and enlargement. Cells leave the periphery of the
meristem to form organs such as leaves or flowers, and are replaced from a small
central zone of slowly dividing, self-renewing stem cells or initials at the tip of the
meristem (Fig. 7.11). In Arabidopsis this zone comprises around 12 to 20 cells. Initials
behave in the same way as animal stem cells (see Chapter 10). They can divide to
give one daughter that remains a stem cell and one that loses the stem-cell property.
This daughter cell continues to divide and its descendants are displaced towards the leaf
primordium L1
peripheral zone of the meristem, where they become founder cells for a new organ
or internode, leave the meristem, and differentiate. A very small number of long-term L2
stem cells at the meristem center may persist for the whole life of the plant.
Meristem stem cells are maintained in the self-renewing state by cells underly-
ing the central zone that form the organizing center. As we shall see, it is the

Fig. 7.11 Organization of the Arabidopsis shoot meristem. A longitudinal section is shown.
The meristem has three main layers, L1, L2, and an inner layer, as indicated by the yellow
lines, and is divided functionally into a central zone (red), a rib zone (yellow), and a peripheral
zone (blue). The stem cells or initials lie in the central zone, while the peripheral zone contains
proliferating cells that will give rise to leaves and side shoots. The rib zone gives rise to the central zone rib zone peripheral zone
central tissues of the plant stem.
266 Chapter 7 Plant development

microenvironment maintained by the organizing center that gives stem cells their
identity.
The undetermined state of meristem stem cells is confirmed by the fact that mer-
istems are capable of regulation. If, for example, a seedling shoot meristem is divided
into two or four parts by vertical incision, each part becomes reorganized into a
complete meristem, which gives rise to a normal shoot. Provided some subpopula-
tion of organizing center cells plus overlying stem cells is present, a normal meristem
will regenerate. If a shoot apical meristem is completely removed, no new apical
meristem form, but the incipient meristem at the base of the leaf is now able to
develop and form a new side shoot. In the presence of the original meristem, this
prospective meristem remains inactive, as active meristems inhibit the development
of other meristems nearby—as a result of auxin transport from the shoot apex, among
other factors. This regulative behavior is in line with cell–cell interactions being a
major determinant of cell fate in the meristem.

7.6 The size of the stem-cell area in the meristem is kept constant by a
feedback loop to the organizing center

Numerous genes that control the behavior of the cells in the meristem are known. The
gene STM, which is involved in specifying the shoot meristem in the development of
the Arabidopsis embryo (see Section 7.3) is, for example, expressed throughout adult
shoot meristems but is suppressed as soon as cells become part of an organ primor-
dium. Its role seems to be to maintain meristematic cells in an undifferentiated state,
as loss of STM function results in all the meristem cells being incorporated into organ
primordia. In contrast, mutations in the Arabidopsis CLAVATA (CLV) genes increase
the size of the meristem as a result of an increase in the number of stem cells. Nor-
mally, despite the continual exit of cells from the stem-cell pool, the number of stem
cells in a meristem is kept roughly the same throughout a plant’s life, by division of
the remaining stem cells. The role of the CLV genes in regulating stem-cell numbers
stem cells (CLV3) CLV3 protein is understood in some detail, and involves feedback between the stem cells and the
peripheral zone organizing center (WUS) organizing center that underlies them.
In Arabidopsis, cells of the organizing center express a homeobox transcription
leaf primordium factor, WUSCHEL (WUS). This is required to produce a signal (as yet unknown) that
gives the overlying cells their stem-cell identity. Mutations in WUS result in termina-
tion of the shoot meristem and cessation of growth as a result of the loss of stem
cells, while its overexpression increases stem-cell numbers. The stem cells express
CLAVATA3 (CLV3), which encodes a secreted protein that acts indirectly to repress
WUS. This feedback loop could control WUS activity in the organizing center and sup-
press WUS activation in neighboring cells, thus limiting the extent of WUS expression
(Fig. 7.12, top panel). In turn, this would regulate the extent of the stem-cell zone

Fig. 7.12 Regulation of the stem-cell population in a shoot meristem. Intercellular signals
control the position and size of the stem-cell population in the Arabidopsis shoot meristem. Top
panel: the organizing center (purple) expresses the transcription factor WUS, which induces the
CLV3 production of an as yet unknown intercellular signal (red arrow) that maintains the overlying
expression stem cells (orange). The stem cells express and secrete the signal protein CLAVATA3 (CLV3;
orange dots), which moves laterally and downwards and indirectly represses transcription of
the WUS gene in the surrounding cells, acting through its cell-surface receptor proteins CLV1
and CLV2. CLV3 thus limits the extent of the area specified as stem cells. The descendants
+CLV1/2 ? of the stem cells are continuously displaced into the peripheral zone of the meristem (pale
yellow), where they are recruited into leaf primordia. Bottom panel: the negative-feedback loop
whereby WUS expression is restricted by CLV3. Expression of WUS in the organizing center
is responsible for the production of an unknown signal that induces the expression of CLV3.
WUS CLV3 in turn signals through CLV1/2 to suppress the expression of WUS.
Adapted from Brand, U., et al.: 2000.
 Meristems 267

above the organizing center. If stem-cell numbers temporarily fall, for example, less
CLV3 is produced and WUS activity increases, with a consequent increase in stem-cell
numbers. More CLV3 is then produced and limits the extent of WUS activity. Other
CLAVATA proteins are involved in the feedback loop (Fig. 7.12, bottom panel).
Meristems can be induced elsewhere in the plant by the misexpression of genes
involved in specifying stem-cell identity, yet another indication that stem-cell iden-
tity is conferred by cell–cell interactions and not by an embryonically specified cell
lineage.

7.7 The fate of cells from different meristem layers can be changed by
changing their position

Other evidence that the fate of a meristematic cell is determined by its position in the
meristem, and thus the intercellular signals it is exposed to, comes from observing the
fates of cells in the different meristem layers. As well as being organized into central
and peripheral zones, the apical meristem of a dicotyledon, such as Arabidopsis, is
composed of three distinct layers of cells (Fig. 7.13). The outermost layer, L1, is just
one cell thick. Layer L2, just beneath L1, is also one cell thick. In both L1 and L2,
cell divisions are anticlinal—that is, the new wall is in a plane perpendicular to the

Lateral view Apical view

L L

M
L L

L
M L
L

Fig. 7.13 Apical meristem of

leaf primordium Arabidopsis. Top panels: scanning electron
L1 micrographs showing the organization of the
tunica meristem at the young vegetative apex of
L2 Arabidopsis. The plant is a clavata1 mutant,
which has a broadened apex that allows for
L3 corpus anticlinal divisions
of tunica layers a clearer visualization of the leaf primordia
L1 and L2 (L) and the meristem (M). Scale bar = 10 μm.
Bottom panel: diagram of a vertical section
through the apex of a shoot. The three-
layered structure of the meristem is apparent
in the most apical region. In layer 1 (L1)
and layer 2 (L2), the plane of cell division
is anticlinal; that is, at right angles to the
surface of the shoot. Cells in layer 3 (L3,
the corpus) can divide in any plane. A leaf
primordium is shown forming at one side of
the meristem.
Photographs courtesy of M. Griffiths.
268 Chapter 7 Plant development

layer—thus maintaining the two-layer organization. The innermost layer is L3, in

2n nuclei 4n nuclei
which the cells can divide in any plane. L1 and L2 are often known as the tunica, and
L3 as the corpus.
To find which tissues each layer can give rise to, the fates of cells in the different
layers can be followed by marking one layer with a distinguishable mutation, such as
L1 a change in pigmentation or a polyploid nucleus. When a complete layer is genetically
L2 different from the others in this way, the organism is known as a periclinal chimera
(Fig. 7.14) and the fate of cells from this layer can be traced. As we saw in relation to
Fig. 7.14 A periclinal meristem chimera
animal embryos, chimeras are organisms composed of cells of two different genotypes
composed of cells of two different (see Section 3.9). Plant chimeras can be made by inducing mutation in the apical
genotypes. In L1 the cells are diploid meristem of a seed or shoot tip by X-irradiation, or by treatment with chemicals such
whereas the cells of L2 are tetraploid—that as colchicine that induce polyploidy.
is, they have double the normal chromosome Layer L1 gives rise to the epidermis that covers all structures produced by the shoot,
number—and are larger and easily recognized. while L2 and L3 both contribute to cortex and vascular structures. Leaves and flowers
After Steeves, T.A., et al.: 1989. are produced mostly from L2; L3 contributes mainly to the stem. Although the three
layers maintain their identity in the central region of the meristem over long periods
of growth, cells in either L1 or L2 occasionally divide periclinally; new cell walls are
formed parallel to the surface of the meristem, and thus one of the new cells invades
an adjacent layer. This migrant cell now develops according to its new position, show-
ing that cell fate is not necessarily determined by the meristem layer in which the
cell originated, and that intercellular signaling is involved in specifying or changing
its fate in its new position. The anticlinal pattern of cell division in L2 becomes dis-
rupted when leaf primordia start to form, when the cells divide periclinally, as well
as anticlinally.
The transcription factor Knotted-1 in maize is homologous to Arabidopsis STM and,
like STM, it is expressed throughout the shoot meristem to keep cells in an undifferen-
tiated state. It is one example of a transcription factor that moves directly from cell to
cell. The KNOTTED-1 gene is expressed in all layers except L1, but the protein is also
found in L1, suggesting that it can move between cells, perhaps via plasmodesmata.
In Knotted-1 gain-of-function mutants, in which the gene is misexpressed in leaves,
Knotted-1 protein fused to green fluorescent protein has also been observed to move
from the inner layers of the leaf to the epidermis, but not in the opposite direction.

7.8 A fate map for the embryonic shoot meristem can be deduced using
clonal analysis

Much of our knowledge about the general properties of meristems outlined in previous
sections comes from studies some decades ago that determined how the embryonic
shoot apical meristem is related to the development of a plant over its whole lifetime.
What was the fate of individual ‘embryonic initials’, as the stem cells were known
at the time? Did particular regions in the embryonic meristem give rise to particular
parts of the adult plant? Individual initials can be marked by mutagenesis of seeds by
­X-irradiation or transposon activation, to give cells with a different color from the rest
of the plant, for example. If the marked meristem cell and its immediate progeny pop-
ulate only part of one layer of the meristem (in contrast to a periclinal chimera), then
this area will give rise to visible sectors of marked cells in stems and organs as the
Fig. 7.15 Tobacco plant mericlinal
plant grows; this type of chimera is known as a mericlinal chimera (Fig. 7.15). The
chimera. This plant has grown from an
embryonic shoot meristem in which an
fate of individually marked initials in mericlinal chimeras can be determined using
albino mutation has occurred in a cell of the clonal analysis in a manner similar to its use in Drosophila (see Box 2E, p. 73).
L2 layer. The affected area occupies about a In maize, the marked sectors usually start at the base of an internode and extend
third of the total circumference of the shoot, apically, terminating within a leaf. Some sectors include just a single internode and
suggesting that there are three apical initial leaf, representing the progeny of an embryonic initial that is lost from the meristem
cells in the embryonic shoot meristem. before the generation of the next leaf primordium. Others, on the other hand, extend
Photograph courtesy of S. Poethig. through numerous internodes, showing that some embryonic initials remain in the
 Meristems 269

meristem for a long time, contributing to a succession of nodes and internodes. In

sunflowers, marked clones have been observed to extend through several internodes
up into the flower, showing that a single initial can contribute to both leaves and
flowers.
From the analysis of hundreds of mericlinal chimeras, fate maps of the embryonic
shoot meristems of several species were constructed, which shed light on the proper-
ties of the shoot meristem and how it behaves during normal development. These fate
maps are probabilistic because it was not possible to know the location of the marked
cell in the embryonic meristem, which is inaccessible inside the seed.
The probabilistic fate map for the maize embryonic shoot apical meristem indicates
that the three most apical cells in L1 give rise to the male inflorescence (the tassel and
spike; Fig. 7.16). The remainder of the maize meristem can be divided into five tiers
of cells that produce internodes and leaves, and which form overlapping concentric
domains on the fate map. The outermost domain contributes to the earliest internode–
leaf modules, while the inner domains give rise to internodes and leaves successively
higher up the stem. A fate map has been similarly constructed for the embryonic
shoot apical meristem of Arabidopsis (Fig. 7.17). Most of the Arabidopsis embryonic
meristem gives rise to the first six leaves, whereas the remainder of the shoot, includ-
ing all the flowerheads, is derived from a very small number of embryonic cells at
the center of the meristem. Unlike maize, the number of leaves in ­Arabidopsis is not
fixed—growth is said to be indeterminate. There is no relation between particular

central spike
tassel base
Domains
6 18
5 17
4
3 16
2 15
leaf 14
primordium
1 13
12
11
female
10 inflorescence

9
8
7
6
Longitudinal section through Internode
embryonic shoot meristem number

Fig. 7.16 Probabilistic fate map of the shoot apical meristem in in the mature plant (see text). The fate of each domain, as estimated
the mature maize embryo. In maize, the primordia of the first six from the collective results of clonal analysis of many different plants, is
leaves are already present in the mature embryo and are excluded from shown on the right. Domain 6, comprising the three most apical L1 cells
the analysis. At the time the cells were marked, the meristem contained of the meristem, will give rise to the terminal male inflorescence. The
about 335 cells, which will give rise to 12 more leaves, the female fate of cells in the other domains is less circumscribed. Domain 5, for
inflorescences, and the terminal male inflorescence (the tassel and example, which consists of around eight L1 cells surrounding domain 6
spike). A longitudinal section through the embryonic apical dome—the and the underlying four or so L2 cells, can contribute to nodes 16–18;
shoot meristem—is shown on the left. Clonal analysis shows that it can domain 4 to nodes 14–18; and domain 3 to nodes 12–15. The female
be divided into six vertically stacked domains, each comprising a set of inflorescences, which develop in the leaf axils, are derived from the
initials that can give rise to a particular part of the plant. The number of corresponding domains.
initials in layers L1 and L2 of each domain at the embryonic stage can After McDaniel, C.N., et al.: 1988.
be estimated from the final extent of the corresponding marked sectors
270 Chapter 7 Plant development

cell lineages and particular structures, which indicates that position in the meristem is
1 L2 layer of meristem crucial in determining cell fate. One exception is that germ cells always arise from L2.
1 3
1 3 3 The L2 layer is a clone, and so there is, in the case of germ cells, a relation between
1 6 8 5 5 3
6 fate and a particular cell lineage.
1 8 i 7 5 3
6
i 7 2 The conclusions from the clonal analysis experiments are that the initials that con-
4 5
6 7 tribute to a particular structure are simply those that happen to be in the appropriate
4 6 5 2
4 4 2 region of the meristem at the time; they have not been prespecified in the embryo as,
4 2
2 say, flower or leaf.

7.9 Meristem development is dependent on signals from other parts of

the plant
i To what extent does the behavior of a meristem depend on other parts of the plant? It
seems to have some autonomy, because if a meristem is isolated from adjacent tissues
by excision it will continue to develop, although often at a much slower rate. Excised
shoot apical meristems of a variety of plants can be grown in culture, where they will
develop into shoots complete with leaves if the growth hormones auxin and cytokinin
7 8 are added. The behavior of the meristem in situ, however, is influenced in more subtle
ways by interactions with the rest of the plant.
1 As we saw earlier, the apical meristem of maize gives rise to a succession of nodes,
6
and terminates in the male flower. The number of nodes before flowering is usually
4
3 between 16 and 22. This number is not controlled by the meristem alone, however.
2 Evidence comes from culturing shoot tips consisting of the apical meristem and one
5 or two leaf primordia. Meristems taken from plants that have already formed as many
as 10 nodes still develop into normal maize plants with the full number of nodes.
Fig. 7.17 Probabilistic fate map of The isolated maize meristem has no memory of the number of nodes it has already
the embryonic shoot meristem of formed, and repeats the process from the beginning. The meristem itself is therefore
Arabidopsis. The L2 layer of the meristem not determined in the embryo with respect to the number of nodes it will form. In
is depicted as if flattened out and viewed the plant, control over the number of nodes that are formed must therefore involve
from above. The numbers indicate the leaf, signals from the rest of the plant to the meristem, finally directing the meristem to
as shown on the plant below, to which each
terminate node formation and form a tassel.
group of meristem cells contributes, and
indicate the sequence in which the leaves are
formed. The inflorescence shoot (i) is derived 7.10 Gene activity patterns the proximo-distal and adaxial–abaxial axes of
from a small number of cells in the center of leaves developing from the shoot meristem
the layer.
After Irish, V.E.: 1991.
Leaves develop from groups of founder cells within the peripheral zone of the shoot
apical meristem. The first indication of leaf initiation in the meristem is usually a swell-
ing of a region to the side of the apex, which forms the leaf primordium (Fig. 7.18).
This small protrusion is the result of increased localized cell multiplication and altered
patterns of cell division. It also reflects changes in polarized cell expansion.
Two new axes that relate to the future leaf are established in a leaf primordium.
These are the proximo-distal axis (leaf base to leaf tip) and the adaxial–abaxial
axis (upper surface to lower surface, sometimes called dorsal to ventral). The latter
is termed adaxial–abaxial as it is related to the radial axis of the shoot. The upper
surface of the leaf derives from cells near the center of this axis (adaxial), while the
lower surface derives from more peripheral cells (abaxial). The two leaf surfaces
carry out different functions and have different structures, with the top surface being
specialized for light capture and photosynthesis. In Arabidopsis, flattening of the leaf
along the adaxial–abaxial axis occurs after leaf primordia begin developing, but in
monocots like maize the leaf is flattened as it emerges. The establishment of adaxial–
abaxial polarity is likely to make use of positional information along the radial axis
of the meristem.
Arabidopsis leaf primordia emerge from the shoot meristem with distinct programs
of development in the adaxial and abaxial halves. This asymmetry can be seen from
 Meristems 271

Early development (lateral view)

adaxial
surface
abaxial
leaf shoot surface 1
4
primordium promeristem
1
1 3 2
2
1 2 3

Fig. 7.18 Leaf phyllotaxis. In shoots where

single leaves are arranged spirally up the
stem, the leaf primordia arise sequentially
Continuing development (transverse sections) in a mathematically regular pattern in the
meristem. Leaf primordia arise around the
sides of the apical dome, just outside the
1 promeristem region. A new leaf primordium
1 is formed slightly above and at a fixed
1 radial angle from the previous leaf, often
4 4 generating a helical arrangement of primordia
4 visible at the apex. Top panel: lateral views of
5 the shoot apex. Bottom panel: view looking
3 2 6 down on cross-sections through the apex
2 3 3 5 near the tip, at successive stages from the
2 top panel.
After Poethig, R.S., et al.: 1985 (top panel);
and Sachs, T.: 1994 (bottom panel).

the beginning, as the leaf primordium has a crescent shape in cross-section, with a
convex outer (abaxial) side and a concave inner (adaxial) surface (see Fig. 7.18). Dif-
ferent genes are expressed in the future adaxial and abaxial sides. For example, the
Arabidopsis gene FILAMENTOUS FLOWER (FIL) is normally expressed in the abaxial
side of the leaf primordium, and specifies an abaxial cell fate. Its ectopic expression
throughout the leaf primordium can cause all cells to adopt an abaxial cell fate, and
the leaf develops as an arrested cylindrical structure.
The specification of adaxial cell fate in Arabidopsis involves the genes ­PHABULOSA
(PHAB), PHAVOLUTA (PHAV), and REVOLUTA (REV). These encode transcription
factors and are initially expressed in the shoot meristem and in the adaxial side of
the primordium. Loss-of-function mutations in these genes result in radially sym-
metrical leaves with only abaxial cell types characteristic of the underside of the leaf.
A microRNA (see Box 6B, p. 228) is involved in the restriction of PHAB, PHAV, and
REV expression to the adaxial side, targeting and destroying their mRNAs on the
abaxial side and thus limiting their activity to the adaxial side.
It has been suggested that, in normal plants, the interaction between adaxial and
abaxial initial cells at the boundary between them initiates lateral growth, result-
ing in the formation of the leaf blade and the flattening of the leaf. The impor-
tance of boundaries in controlling pattern and form has already been seen in the
parasegments of Drosophila (see Section 2.24) and further examples will be found
in ­Chapter 11.
Development along the leaf proximo-distal axis also appears to be under genetic
control. Like other grasses, a maize leaf primordium is composed of prospective leaf-
sheath tissue proximal to the stem and prospective leaf-blade tissue distally. Muta-
tions in certain genes result in distal cells taking on more proximal identities; for
example, making sheath in place of blade. Similar proximo-distal shifts in pattern
occur in Arabidopsis as a result of mutation. Positional identity along the proximo-
distal axis may reflect the developmental age of the cells, distal cells adopting a differ-
ent fate from proximal cells because they mature later.
272 Chapter 7 Plant development

7.11 The regular arrangement of leaves on a stem is generated by regulated

auxin transport

As the shoot grows, leaves are generated within the meristem at regular intervals and
with a particular spacing. Leaves are arranged along a shoot in a variety of ways in
different plants, and the particular arrangement, known as phyllotaxy or phyllotaxis,
is reflected in the arrangement of leaf primordia in the meristem. Leaves can occur
singly at each node, in pairs, or in whorls of three or more. A common arrangement
is the positioning of single leaves spirally up the stem, which can sometimes form a
striking helical pattern in the shoot apex.
In plants in which leaves are borne spirally, a new leaf primordium forms at the
center of the first available space outside the central region of the meristem and
above the previous primordium (see Fig. 7.18). This pattern suggests a mechanism
for leaf arrangement based on lateral inhibition (see Section 1.16), in which each
leaf primordium inhibits the formation of a new leaf within a given distance. In this
model, inhibitory signals emanating from recently initiated primordia prevent leaves
from forming close to each other. There is some experimental evidence for this. In
ferns, leaf primordia are widely spaced, allowing experimental microsurgical interfer-
ence. Destruction of the site of the next primordium to be formed results in a shift
toward that site by the future primordium, whose position is closest to it (Fig. 7.19).
Recent studies indicate, however, that it is competition rather than inhibition that is
responsible.
As we have seen in the determination of apical–basal polarity in the embryo, auxin
is transported out of cells with the help of proteins such as PIN1 (see Section 7.3). In
the shoot, auxin produced in the shoot tip below the meristem is transported upwards
into the meristem, through the epidermis and the outermost meristem layer. The
direction of auxin flow in the shoot apex is controlled by PIN1 and follows a simple
rule: the side of the cell on which PIN1 is found is the side nearest the neighbor cell
with the highest auxin concentration. Thus auxin transport is always towards a region
of higher concentration. A high concentration of auxin is a primordium activator, and
initially, auxin is pumped towards a new primordium that is developing at a site of
high auxin concentration. This, however, depletes a zone of cells around the primor-
dium of auxin, such that cells nearer the center of the meristem now have more auxin
than the cells adaxial to the new primordium. This activates a feedback mechanism
that causes PIN1 to move to the other side of these cells, and auxin now flows out
of the new primordium towards the meristem, creating a new spot of high auxin
­concentration in the meristem farthest away from any new primordium (Fig. 7.20).

excision of region of leaf primordium 1

Fig. 7.19 Leaf primordia may be 1

positioned by lateral inhibition or by 4
competition. Leaf primordia on a fern shoot 4
tip form in a regular order in positions 1 to 4. 3
3
Primordia appear to form as far as possible
from existing primordia, so 2 forms almost
opposite 1. Normally, 4 will develop between 2 2
1 and 2, but if 1 is excised, 4 forms much
further away from 2. This result can be
interpreted either by the removal of lateral
inhibition by primordium 1 or by the removal
presumptive leaf primordium
of the competitive effect of 1 for some
primordium-inducing factor such as auxin.
 Meristems 273

Fig. 7.20 Auxin-dependent mechanism

high auxin
of phyllotaxis in Arabidopsis. Auxin is
low auxin
transported through PIN proteins towards
areas of high auxin concentration (dark
green), at which an organ primordium will
form. Cells around the developing primordium
become depleted of auxin (light green), which
causes the polarity of the PIN proteins to
reverse so that auxin then flows away from
the primordium. The red arrows indicate the
direction of PIN polarity. It is proposed that
This leads to auxin peaks occurring sequentially, at the regular positions later occu-
this pattern of auxin circulation could set
pied by new leaves.
up the regular pattern of leaf and flower
formation from meristems.
7.12 Root tissues are produced from Arabidopsis root apical meristems by a
highly stereotyped pattern of cell divisions

The organization of tissues in the Arabidopsis root tip is shown in Fig. 7.21. The
radial pattern comprises single layers of epidermal, cortical, endodermal, and peri-
cycle cells, with vascular tissue in the center (protophloem and protoxylem). Root
apical meristems resemble shoot apical meristems in many ways and give rise to the
root in a similar manner to shoot generation. But there are some important differences
between the root and shoot meristems. The shoot meristem is at the extreme tip of
the shoot, whereas the root meristem is covered by a root cap (which is itself derived
from one of the layers of the meristem); also, there is no obvious segmental arrange-
ment at the root tip resembling the node–internode–leaf module.
The root is set up early (see Section 7.2) and a well-organized embryonic root can
be identified in the late heart-stage embryo (Fig. 7.22). An antagonistic interaction
between auxin and cytokinin controls the establishment of the root stem-cell niche.
Clonal analysis has shown that the seedling root meristem can be traced back to a set
of embryonic initials that arise from a single tier of cells in the heart-stage embryo.
As in the shoot meristem, a root meristem is composed of an organizing center,
called the quiescent center in roots, in which the cells divide only very rarely, and
which is surrounded by stem-cell-like initials that give rise to the root tissue (see
Fig. 7.22). The quiescent center is essential for meristem function. When parts of the
meristem are removed by microsurgery, it can regenerate, but regeneration is always
preceded by the formation of a new quiescent center. Laser destruction of individual
protophloem
quiescent-center cells shows that, as in the shoot meristem, a key function of the protoxylem
quiescent center is to maintain the immediately adjacent initials in the stem-cell state pericycle
and prevent them from differentiating. endodermis
cortex
Each initial undergoes a stereotyped pattern of cell divisions to give rise to a epidermis
number of columns, or files, of cells in the growing root (see Fig. 7.21); each file of root cap
cells in the root thus has its origin in a single initial. Some initials give rise to both
endodermis and cortex, whereas others give rise to both epidermis and the root cap.
Before it leaves the meristem, therefore, the undifferentiated progeny of an endo-
dermis/cortex initial, for example, will divide asymmetrically to give one daughter
that produces cortex and one that produces endodermis. The gene SCARECROW is
necessary to ­confer this asymmetry on the dividing cell, and mutations in this gene
give roots with no distinct endodermis or cortex but with a tissue layer with charac-
teristics of both. root
The normal pattern of cell divisions is not obligatory, however. As discussed ear- apical
meristem
lier, fass mutants, which have disrupted cell divisions, still have relatively normal
root cap
Fig. 7.21 The structure of the root tip in Arabidopsis. Roots have a radial organization. In
the center of the growing root tip is the future vascular tissue (protoxylem and protophloem).
This is surrounded by further tissue layers.
274 Chapter 7 Plant development

Heart-shaped embryo

quiescent center stele initials

Fig. 7.22 Fate map of root regions in

the heart-stage Arabidopsis embryo. The
root grows by the division of a set of initial
root hair
cells. The root meristem comes from a small
number of cells in the heart-shaped embryo.
Each tissue in the root is derived from the
division of a particular initial cell. At the root apical root cap epidermis
center of the root meristem is a quiescent meristem initial cortex and and lateral
center, which does not divide. root cap columella endodermis root cap
root cap initial initials
After Scheres, B., et al.: 1994.

­ atterning in the root. In addition, laser ablation of individual meristem cells does not
p
lead to an abnormal root. The remaining initials undergo new patterns of cell divi-
sion that replace the progeny of the cells that have been destroyed. Such observations
show that, as in the shoot meristem, the fate of cells in the developing root meristem
depends on their recognition of positional signals and not on their lineage.
As we saw in Section 7.3, auxin gradients play a major role in patterning the embryo
and specifying the root region, and mutations that affect auxin localization lead to
root defects. At the globular stage of embryonic development, the auxin-transport pro-
tein PIN1 is localized in cells in the future root region and the highest level of auxin
is found adjacent to where the quiescent center will develop.
The role of auxin in root development continues into the adult plant. Auxin in
the root is transported out of cells via the PIN proteins, and enters adjacent cells.
Cells with raised auxin levels transport auxin better, probably due to an increase
in the number of PIN proteins in the membrane as auxin prevents their endocy-
tosis and recycling, and so there is a positive feedback loop that raises the local
concentration.
Auxin plays a key role in patterning the growing root. Modeling of auxin movement,
using the known distribution of PIN proteins in the membranes of different cell types,
has shown that PIN-directed flow can explain the formation and maintenance of a
stable auxin maximum at the quiescent center. Auxin is transported down the central
vascular tissue of the root tip, forming a concentration maximum at the quiescent
center, and then outwards and upwards through the outer layers of the root tip, form-
ing gradients along both the basal-apical axis and laterally in the root. This modeling
successfully simulates the effects of such auxin gradients on pattern formation, cel-
lular differentiation and root growth in real time.
An idea of how the auxin gradients might be translated into effects on cell fate and
cell behavior is provided by the graded distribution of the four PLETHORA-family
transcription factors in the root. These transcription factors are required for proper
root development and are expressed in a graded manner along the apical–basal axis,
with expression maxima for all at the quiescent center—the region of maximum auxin
concentration. Here they are essential for stem-cell maintenance and function. Lower
concentrations of PLETHORA proteins correspond to the meristem region, where cells
are proliferating, while even lower concentrations appear to be necessary for exit from
the meristem and cell differentiation in the elongation zone. Although not yet proven
experimentally, PLETHORA gene expression could provide a graded read-out of the
auxin gradient that helps direct root patterning.
 Meristems 275

Other influences on root patterning are the interplay between auxin and the hor-
mone cytokinin. Cytokinin helps regulate root meristem size and set the position
of the transition zone—where cells stop dividing and start elongating and differen-
tiating—by repressing both auxin transport and cellular responses to auxin in this
region.
Auxin is also involved in the ability of plants to regenerate from a small piece of
stem. In general, roots form from the end of the stem that was originally closest to the
root, whereas shoots tend to develop from dormant buds at the end that was near-
est to the shoot. This polarized regeneration is related to vascular differentiation and
to the polarized transport of auxin. Transport of auxin from its source in the shoot tip
toward the root leads to an accumulation of auxin at the ‘root’ end of the stem cutting,
where it induces the formation of roots. One hypothesis suggests that polarity is both
induced and expressed by the oriented flow of auxin.
One of the best examples of developmentally important transcription factor move-
ment from one cell to another is found in roots. As noted earlier, expression of the
gene SCARECROW is required for root cells to adopt an endodermal fate. This expres-
sion requires the transcriptional activator SHORT-ROOT (SHR). SHR is, however, not
synthesized in the prospective endodermal cells, but in the adjacent cells on the
inner side. SHR protein is transported from these cells outwards into the prospec-
tive endodermis, and this movement appears to be regulated and not simply due to
diffusion.

7.13 Root hairs are specified by a combination of positional information and

lateral inhibition

Root hairs are formed from epidermal cells at regular intervals around the root, and
this regularity is thought to be achieved by a combination of responses to positional
information and lateral inhibition by the movement of transcription factors between
cells. Files of cells that will make root hairs alternate with files of non-hair-producing
cells on the surface of the developing root. The importance of position is shown by
the fact that if an epidermal cell overlies a junction between two cortical cells it forms
a root hair, whereas if it contacts just one cortical cell it does not (Fig. 7.23). And if
a cell changes its position in relation to the cortex, its fate will also change, from a
potential hair-forming cell to a non-hair-forming cell and vice versa. Most cell divi-
sions in the future epidermis are horizontal, increasing the number of cells per file, lateral root cap
but occasionally a vertical anticlinal division occurs, pushing one of the daughter cells A T
T
into an adjacent file. The daughter cell then assumes a fate corresponding to its new epidermis
position in relation to the adjacent cortical cells.
The positional cues, as yet unknown, are thought to be detected by the epider- cortex
mal cells through the SCRAMBLED protein, which is a receptor-like protein kinase.
endodermis
SCRAMBLED activity influences the activity of a network of transcription factors that
control cell fate. Two groups of transcription factors have been identified by mutation pericycle
experiments, one group promoting a root-hair fate and one suppressing it. A key tran-
scription factor that seems to be regulated by SCRAMBLED activity is WEREWOLF,
Fig. 7.23 Organization of cell types in the
whose expression is suppressed in presumptive hair-forming cells, presumably in
root epidermis. The epidermis is composed
response to the positional signal. As well as promoting an atrichoblast fate, however, of two types of cells: trichoblasts (T), which
WEREWOLF is also required for the expression of transcription factors (CAPRICE, will form root hairs, and atrichoblasts (A),
TRYPTYCHON, and ENHANCER OF TRYPTYCHON) that are needed to specify hair which will not. Trichoblasts overlie the
cells. After positional signaling, these proteins will only be produced in the presump- junction between two cortical cells and
tive atrichoblasts, but they are thought to move laterally into the adjacent epidermal atrichoblasts are located over the outer
cells and promote these cells’ differentiation as trichoblasts by inhibiting genes that tangential wall of cortical cells.
would otherwise give an atrichoblast fate. After Dolan, L., Scheres, B.: 1998.
276 Chapter 7 Plant development

Summary

Meristems are the growing points of a plant. The apical meristems, found at the tips
of shoots and roots, give rise to all the plant organs—roots, stem, leaves, and flowers.
They consist of small groups of a few hundred undifferentiated cells that are capable
of repeated division. The center of the meristem is occupied by self-renewing stem
cells, which replace the cells that are lost from the meristem when organs are formed.
The fate of a cell in the shoot meristem depends upon its position in the meristem and
interactions with its neighbors, as when a cell is displaced from one layer to another it
adopts the fate of its new layer. Meristems can also regulate when parts are removed,
in line with cell–cell interactions determining cell fate. Fate maps of embryonic shoot
meristems show that they can be divided into domains, each of which normally con-
tributes to the tissues of a particular region of the plant, but the fate of the embry-
onic initials is not fixed. The shoot meristem gives rise to leaves in species-specific
patterns—phyllotaxy—which seem best accounted for by regulated transport of auxin.
Lateral inhibition is involved in the regular spacing of hairs on root and leaf surfaces.
In the root meristem, the cells are organized rather differently from those in the shoot
meristem, and there is a much more stereotyped pattern of cell division. A set of initial
cells maintains root structure by dividing along different planes.

Summary: meristems give rise to all adult tissues

Shoot meristem Root meristem

cell fate is determined by position stereotyped division of initials

along different planes

shoot meristem gives rise to stem internodes,

leaves, and flowers root meristem gives rise to
all root structures

arrangement of leaves on stem may be due

FM5
FM3 to an auxin-based mechanism at shoot apex
SAM
FM2 FM1
FM4

Fig. 7.24 Scanning electron micrograph of Flower development and control of flowering
an Arabidopsis inflorescence meristem.
The central inflorescence meristem (shoot Flowers contain the reproductive cells of higher plants and develop from the shoot
apical meristem, SAM) is surrounded by a meristem. In most plants, the transition from a vegetative shoot meristem to a floral
series of floral meristems (FM) of varying meristem that produces a flower is largely, or absolutely, under environmental con-
developmental ages. The inflorescence trol, with daylength and temperature being important determining factors. In a plant
meristem grows indeterminately, with cell such as Arabidopsis, in which each flowering shoot produces multiple flowers, the
divisions providing new cells for the stem
vegetative shoot meristem first becomes converted into an inflorescence meristem,
below, and new floral meristems on its flanks.
which then forms floral meristems, each of which develops completely into a single
The floral meristems (or floral primordia) arise
flower (Fig. 7.24). Floral meristems are thus determinate, unlike the indeterminate
one at a time in a spiral pattern. The most
mature of the developing flowers is on the shoot apical meristem. Flowers, with their arrangement of floral organs (sepals, pet-
right (FM1), showing the initiation of sepal als, stamens, and carpels), are rather complex structures, and it is a major challenge
primordia surrounding a still-undifferentiated to understand how they arise from the floral meristem.
floral meristem. Eventually such a floral The conversion of a vegetative shoot meristem into one that makes flowers involves
meristem will also form petal, stamen, and the induction of so-called meristem identity genes. A key regulator of floral induc-
carpel primordia. tion in Arabidopsis is the meristem identity gene LEAFY (LFY); a related gene in
Photograph from Meyerowitz, E.M., et al.: 1991. ­Antirrhinum is FLORICAULA (FLO). How environmental signals, such as daylength,
 Flower development and control of flowering 277

Fig. 7.25 Structure of an Arabidopsis flower. Arabidopsis flowers are radially symmetrical
and have an outer ring of four identical green sepals, enclosing four identical white petals, carpel
within which is a ring of six stamens, with two carpels in the center. Bottom: floral diagram
of the Arabidopsis flower representing a cross-section taken in the plane indicated in the top
diagram. This is a conventional representation of the arrangement of the parts of the flower,
showing the number of flower parts in each whorl and their arrangement relative to each other.
After Coen, E.S., et al.: 1991.
stamen

influence floral induction is discussed later. We will first consider the mechanisms that
pattern the flower, in particular those that specify the identity of the floral organs.

7.14 Homeotic genes control organ identity in the flower

petal
The individual parts of a flower each develop from a floral organ primordium pro-
duced by the floral meristem. Unlike leaf primordia, which are all identical, the floral
organ primordia must each be given a correct identity and be patterned according to
it. An Arabidopsis flower has four concentric whorls of structures (Fig. 7.25), which
reflect the arrangement of the floral organ primordia in the meristem. The sepals
sepal
(whorl 1) arise from the outermost ring of meristem tissue, and the petals (whorl 2)
from a ring of tissue lying immediately inside it. An inner ring of tissue gives rise
to the male reproductive organs—the stamens (whorl 3). The female reproductive
organs—the carpels (whorl 4)—develop from the center of the meristem. In a floral
meristem of Arabidopsis, there are 16 separate primordia, giving rise to a flower
with four sepals, four petals, six stamens and a pistil made up of two carpels (see sepal (Se)
Fig. 7.25).
stamen (St)
The primordia arise at specific positions within the meristem, where they develop
into their characteristic structures. After the emergence of the primordia in Antir-
rhinum, cell lineages become restricted to particular whorls, rather like the lineage
restriction to compartments in Drosophila (see Section 2.23). Lineage restriction 1 2 3 4 carpel (Ca)
occurs at the time when the pentagonal symmetry of the flower becomes visible and
genes that give the different floral organs their identity are expressed. The lineage
compartments within the floral meristem appear to be delineated by narrow bands of
non-dividing cells. petal (Pe)

Like the homeotic selector genes that specify segment identity in Drosophila, muta-
tions in floral identity genes cause homeotic mutations in which one type of flower X Y
part is replaced by another. In the Arabidopsis mutant apetala2, for example, the midline
sepals are replaced by carpels and the petals by stamens; in the pistillata mutant, whorls
1 2 3 4 4 3 2 1
petals are replaced by sepals and stamens by carpels. These mutations identified the
Se Pe St Ca St Pe Se
floral organ identity genes, and have enabled their mode of action to be determined.
Homeotic floral mutations in Arabidopsis fall into three classes, each of which
­affects the organs of two adjacent whorls (Fig. 7.26). The first class of mutations, of

Fig. 7.26 Homeotic floral mutations in

apetala2 mutant apetala3 mutant agamous mutant
Arabidopsis. Left panel: an apetala2 mutant
has whorls of carpels and stamens in place of
sepals and petals. Center panel: an apetala3
mutant has two whorls of sepals and two of
carpels. Right panel: agamous mutants have a
whorl of petals and sepals in place of stamens
and carpels. Transformations of whorls are
shown inset, and can be compared to the wild-
type arrangement, as shown in Fig. 7.25.
Photographs from Meyerowitz, E.M., et al.:
1991 (left panel), and Bowman, J.L., et al.:
Ca St St Ca Ca St St Ca Se Se Ca Ca Ca Ca Se Se Se Pe Pe Se Se Pe Pe Se
1989 (center panel).
278 Chapter 7 Plant development

which apetala2 is an example, affect whorls 1 and 2, giving carpels instead of sepals
C in whorl 1, and stamens instead of petals in whorl 2. The phenotype of the flower,
B B
A Midline A going from the outside to the center, is therefore carpel, stamen, stamen, carpel.
The second class of homeotic floral mutations affects whorls 2 and 3. In this class,
Se Pe St Ca Ca St Pe Se
apetala3 and pistillata give sepals instead of petals in whorl 2 and carpels instead of
1 2 3 4 4 3 2 1
stamens in whorl 3, with a phenotype sepal, sepal, carpel, carpel. The third class of
mutations affects whorls 3 and 4, and gives petals instead of stamens in whorl 3 and
sepals or variable structures in whorl 4. The mutant agamous, which belongs to this
class, has an extra set of sepals and petals in the center instead of the reproductive
organs.
These mutant phenotypes can be accounted for by an elegant model in which over-
lapping patterns of gene activity specify floral organ identity (Fig. 7.27) in a manner
highly reminiscent of the way in which Drosophila homeotic genes specify segment
Fig. 7.27 The three overlapping regions identity along the insect’s body. In detail, however, there are many differences, and
of the Arabidopsis floral meristem that quite different genes are involved. In this instance, plants and animals have, perhaps
have been identified by the homeotic floral not surprisingly, independently evolved a similar approach to patterning a multicel-
identity mutations. Region A corresponds to lular structure, but have recruited different proteins to carry it out.
whorls 1 and 2, B to whorls 2 and 3, and C to In essence, the floral meristem is divided by the expression patterns of the ­homeotic
whorls 3 and 4.
genes into three concentric overlapping regions, A, B, and C, which partition the
meristem into four non-overlapping regions corresponding to the four whorls. Each
of the A, B, and C regions corresponds to the zone of action of one class of homeotic
genes and the particular combinations of A, B, and C functions give each whorl a
unique identity and so specify organ identity. Of the genes mentioned in Fig. 7.26,
APETALA1 (AP1) and APETALA2 (AP2) are A-function genes, APETALA3 (AP3) and
PISTILLATA (P1) are B-function genes, and AGAMOUS (AG) is a C-function gene.
The expression of AP3 and AG in the developing flower is shown in Fig. 7.28. All the
homeotic genes, also known as floral organ identity genes, encode transcription
factors, and the B- and C-function proteins such as AP3 and AG contain a conserved
DNA-binding sequence known as the MADS box. MADS-box genes are present in
animals and yeast, but a role in development is known mainly in plants—although
a MADS-box transcription factor, MEF2, is involved in muscle differentiation in ani-
mals. The original simple model for specifying floral organ identity is presented in
more detail in Box 7C. Since the model was first proposed, more has become known
about the activities and functions of the genes identified by the homeotic muta-
tions, more genes controlling flower development have been discovered, and more
­‘functions’ added.
One question that arose when the original floral organ identity model was ­investigated
experimentally was why the ABC genes only showed their ­homeotic properties in the
floral meristem and did not convert leaves into floral organs when artificially over-
expressed in vegetative meristems, as might be expected for homeotic genes of this
type. The answer came with the discovery of the SEPALLATA (SEP) genes, which also
encode MADS-box proteins. These genes are required for the B and C functions and
are only active in floral meristems. The SEP proteins are thought to combine with B

Fig. 7.28 Expression of APETALA3 and

AGAMOUS during flower development. In
situ hybridization shows that AGAMOUS is
expressed in the central whorls (left panel),
whereas APETALA3 is expressed in the outer
whorls that give rise to petals and stamens
(right panel).
 Flower development and control of flowering 279

Fig. 7.29 Current status of the ABC model of floral organ identity. The regulatory genes
Floral meristem
LEAFY, WUSCHEL (WUS), and UNUSUAL FLORAL ORGANS (UFO) are expressed in specific
domains in the floral meristem, which, together with repression of APETALA1 by AGAMOUS,
results in the pattern of ABC functions. ABC proteins and the co-factor SEP proteins assemble UFO WUS UFO
into complexes that specify the different organ identities. LEAFY
Adapted from Lohmann, J.U., Weigel, D.: 2002

and C gene products to form active gene-regulatory complexes. A current view of the
mechanism of specifying floral organ identity is shown in Fig. 7.29. AP3/PI AP3/PI
There is a better understanding of the functions and patterning of the floral ­homeotic AP1 AG AP1
genes than when the ABC model was first proposed. The MADS-box homeotic A-class SEP3
gene AP1 has been found to have a dual role: it acts early with other genes to specify
general floral meristem identity and only later contributes to A function. It is induced
by the meristem identity gene LFY, which is expressed throughout the meristem, and
? AP1 AG PI
AP1 is actively inhibited in the central regions of floral meristems by AGAMOUS. The
AP1 ? AP3 SEP3
expression of the A-function gene APETALA2 (AP2) is translationally repressed by
a microRNA, keeping the AP2 protein at a low level. APETALA3 and PISTILLATA1
are thought to be activated as a result of a meristem identity gene called UNUSUAL AP1 PI AG SEP3
FLORAL ORGANS (UFO), which is expressed in the meristem in a pattern similar AP3 SEP3 SEP3 AG
to that of B-function genes (see Fig. 7.29). UFO encodes a component of ubiquitin
ligase, and is thought to exert its effects on flower development by targeting specific
proteins for degradation. As we saw in animals, in relation to the control of b-catenin
degradation (see Chapters 4 and 6), regulated degradation of proteins can be a power- Se Pe St Ca St Pe Se
ful developmental mechanism. In the center of the floral meristem, the expression of
AGAMOUS is partly controlled by WUS, which as we have seen earlier, is expressed in
the organizing center of the vegetative shoot meristem and continues to be expressed
in floral meristems.
Another group of genes that help pattern the floral organ primordia are genes
that control cell division. The gene SUPERMAN is one example, controlling cell
proliferation in stamen and carpel primordia, and in ovules. Plants with a muta-
tion in this gene have stamens instead of carpels in the fourth whorl. SUPERMAN
is expressed in the third whorl, and maintains the boundary between the third and
fourth whorls.
Despite the enormous variation in the flowers of different species, the mechanisms
underlying flower development seem to be very similar. For example, there are strik-
ing similarities between the genes controlling flower development of Arabidopsis and
Antirrhinum, despite the quite different final morphology of the snapdragon flower.
In developing Arabidopsis flowers, the patterns of activity of the corresponding genes
fit well with the cell-lineage restriction to whorls seen in Antirrhinum.

7.15 The Antirrhinum flower is patterned dorso-ventrally as well as radially

Like Arabidopsis flowers, those of Antirrhinum consist of four whorls, but unlike
Arabidopsis, they have five sepals, five petals, four stamens, and two united carpels
(Fig. 7.30, left). Floral homeotic mutations similar to those in Arabidopsis occur in
Antirrhinum, and floral organ identity is specified in the same way. Several of the
Antirrhinum homeotic genes have extensive homology with those of Arabidopsis, the
MADS box in particular being well conserved.
An extra element of patterning is required in the Antirrhinum flower, which has
a bilateral symmetry imposed on the basic radial pattern common to all flowers. In
whorl 2, the upper two petal lobes have a shape quite distinct from the lower three,
giving the flower its characteristic snapdragon appearance. In whorl 3, the uppermost
stamen is absent, as its development is aborted early on. The Antirrhinum flower
280 Chapter 7 Plant development

Box 7C The basic model for the patterning of the Arabidopsis flower

Wild type apetala2 mutant agamous mutant

sepal (Se)

petal (Pe)

stamen (St)

carpel (Ca)

Expression of wild-type Expression of functions Expression of functions

gene functions in homeotic mutant in homeotic mutant
c c
Function b b b b b b
a a a
Midline

Whorl Se Pe St Ca Ca St Pe Se Ca St St Ca Ca St St Ca Se Pe Pe Se Se Pe Pe Se
1 2 3 4 4 3 2 1 1 2 3 4 4 3 2 1 1 2 3 4 4 3 2 1

The floral meristem is divided into three overlapping regions, A, B, and C, each region
corresponding to a class of homeotic mutations, as shown in Fig. 7.26 (see text). Wild type
Three regulatory functions—a, b, and c—operate in regions A, B, and C, respectively, as AG
AP3/PI AP3/PI
shown in the panels above. In the wild-type flower (top left panel), it is assumed that AP1/AP2 AP1/AP2
Midline
a is expressed in whorls 1 and 2, b in 2 and 3, and c in whorls 3 and 4. In addition,
Se Pe St Ca Ca St Pe Se
a function inhibits c function in whorls 1 and 2 and c function inhibits a function in
1 2 3 4 4 3 2 1
whorls 3 and 4—that is, a and c functions are mutually exclusive. a alone specifies
sepals, a and b together specify petals, b and c stamens, and c alone carpels.
apetala2 mutant
The homeotic mutations eliminate the functions of a, b, or c, and alter the regions
AG
within the meristem where the various functions are expressed. Mutations in a, such AP3/PI AP3/PI
as apetala2 (see center top panel), result in an absence of function a, and c spreads
throughout the meristem, resulting in the half-flower pattern of carpel, stamen, sta- Ca St St Ca Ca St St Ca
men, carpel. Mutations in b, such as apetala3 (see Fig. 7.26), result in only a func- 1 2 3 4 4 3 2 1
tioning in whorls 1 and 2, and c in whorls 3 and 4, giving sepal, sepal, carpel, carpel. agamous mutant
Mutations in c genes (such as agamous), result in a activity in all whorls, giving the AP3/PI AP3/PI
AP1/AP2
phenotype sepal, petal, petal, sepal (see top right panel).
All the floral homeotic mutants discovered so far in Arabidopsis­can be quite sat- Se Pe Pe Se Se Pe Pe Se
isfactorily accounted for by this model (although there are small variations in gene 1 2 3 4 4 3 2 1
numbers and expression patterns in other species that allow mutant phenotypes apetala3/pistillata mutant
not seen in Arabidopsis), and particular genes can be assigned to each controlling AG
function. Function a corresponds to the activity of genes such as APETALA2, b to AP1/AP2 AP1/AP2
APETALA3 and PISTILLATA, and c to AGAMOUS. The model also accounts for the phe-
Se Se Ca Ca Ca Ca Se Se
notype of double mutants, such as apetala2 with apetala3, and apetala3 with pistil-
1 2 3 4 4 3 2 1
lata, as shown in the panels on the right.
This system emphasizes the similarity in function between the homeotic genes apetala2/apetala3 mutant
in animals and those controlling organ identity in flowers, although the genes them- AG
selves are completely different. The functional similarity with the Hox complex of
Drosophila is further illustrated by the role of the CURLY LEAF gene of ­Arabidopsis, Ca Ca Ca Ca Ca Ca Ca Ca
which is necessary for the stable maintenance of homeotic gene activity. CURLY 1 2 3 4 4 3 2 1
LEAF is related to the Polycomb family of genes in Drosophila and is similarly required
for stable repression of homeotic genes.
 Flower development and control of flowering 281

Fig. 7.30 Mutations in CYCLOIDEA make

Dorso-ventral axis the Antirrhinum flower symmetrical. In the
wild-type flower (left) the petal pattern is
different along the dorso-ventral axis. In the
mutant (right) the flower is symmetrical. All
the petals are like the most ventral one in the
wild type and are folded back.
Photograph courtesy of E. Coen, from Coen,
E.S., et al.: 1991.

Wild type CYCLOIDEA mutant

therefore has a distinct dorso-ventral axis. Another group of homeotic genes, differ- L1 inner core
ent from those that govern floral organ identity, appear to act in this dorso-ventral
L2 mantle sepal primordium
patterning. For example, mutations in the gene CYCLOIDEA, which is expressed in
the dorsal region, abolish dorso-ventral polarity and produce flowers that are more L3
radially symmetrical (Fig. 7.30, right).

7.16 The internal meristem layer can specify floral meristem patterning

Although all three layers of a floral meristem (Fig. 7.31) are involved in organo-
genesis, the contribution of cells from each layer to a particular structure may be
variable. Cells from one layer can become part of another layer without disrupting Fig. 7.31 Floral meristem. The meristem is
normal morphology, suggesting that a cell’s position in the meristem is the main composed of layers L1, L2, and L3. The inner
determinant of its future behavior. Some insight into positional signaling and pat- core cells are derived from L3. The sepal
terning in the floral meristem can be obtained by making periclinal chimeras (see primordia are just beginning to develop.
Section 7.7) from cells that have different genotypes and that give rise to different After Drews, G.N., et al.: 1989.
types of flower. From such chimeras, one can find out whether the cells develop
autonomously according to their own genotype, or whether their behavior is con-
trolled by signals from other cells.
As well as being produced by mutation, chimeras can also be generated by graft-
ing between two plants of different genotypes. A new shoot meristem can form at
the junction of the graft, and sometimes contains cells from both genotypes. Such
stamen

chimeras can be made between wild-type tomato plants and tomato plants carrying
carpel
sepal
petal

the mutation fasciated, in which the flower has an increased number of floral organs
Organ number

per whorl. This phenotype is also found in chimeras in which only layer L3 contains
fasciated cells (Fig. 7.32). The increased number of floral organs is associated with
an overall increase in the size of the floral meristem, and in the chimeric plants this Wild type fasciated Chimera:
cannot be achieved unless the fasciated cells of layer L3 induce the wild-type L1 fasciated
only in L3
cells to divide more frequently than normal. The mechanism of intercellular signal-
ing between L3 and L1 is not yet known. In Antirrhinum, the abnormal expression
of ­FLORICAULA in only one meristem layer can result in flower development. These Fig. 7.32 Floral organ number in chimeras
of wild-type and fasciated tomato
results illustrate the importance of signaling between layers in flower development.
plants. In the fasciated mutant there are
more organs in the flower than in wild-type
7.17 The transition of a shoot meristem to a floral meristem is under plants. In chimeras in which only layer L3 of
environmental and genetic control the floral meristem contains fasciated mutant
cells, the number of organs per flower is still
Flowering plants first grow vegetatively, during which time the apical meristem gener- increased, showing that L3 can control cell
ates leaves. Then, triggered by environmental signals such as increasing daylength, behavior in the outer layers of the meristem.
282 Chapter 7 Plant development

Fig. 7.33 Flowering can be controlled by daylength and LEAFY expression. As shown in
Long day Short day
the top panels, when wild-type Arabidopsis is grown under long-day conditions (left), few
lateral shoots are formed before the apical shoot meristem begins to form floral meristems.
When grown under short-day conditions, flowering is delayed and there are in consequence
more lateral shoots. The gene LEAFY is normally expressed only in inflorescence and floral
meristems, but if it is expressed throughout the plant (bottom panels), all shoot meristems
produced are converted to floral meristems in both daylengths.
Wild type

the plant switches to a reproductive phase and from then on the apical meristem
gives rise only to flowers. There are two types of transition from vegetative growth
to flowering. In the determinate type, the inflorescence meristem becomes a terminal
flower, whereas in the indeterminate type the inflorescence meristem gives rise to
a number of floral meristems. Arabidopsis is of the indeterminate type (Fig. 7.33).
A primary response to floral inductive signals in Arabidopsis is the expression of floral
meristem identity genes such as LEAFY and the dual-function AP1 (see Section 7.13),
which are necessary and sufficient for this transition. LEAFY potentially activates AP1
leaf flower shoot meristem
throughout the meristem while also activating AGAMOUS in the center of the flower.
AGAMOUS then represses the expression of AP1 in the center, helping to restrict its
floral organ identity function to region A (see Fig. 7.27). Mutations in floral meristem
identity genes partly transform flowers into shoots. In a leafy mutant, which lacks
LEAFY expressed throughout

LEAFY function, the flowers are transformed into spirally arranged sepal-like organs
along the stem, whereas expression of LEAFY throughout a plant is sufficient to con-
fer a floral fate on lateral shoot meristems and they develop as flowers (Fig. 7.33,
bottom panels).
In Arabidopsis, flowering is promoted by increasing daylength, which predicts the
end of winter and the onset of spring and summer (see Fig. 7.33). This behavior is
called photoperiodism. In some strains, flowering is also accelerated after the plant
has been exposed to a long period of cold temperature, a cue that winter has passed.
This phenomenon is known as vernalization. Grafting experiments have shown that
daylength is sensed not by the shoot meristem itself, but by the leaves. When the
period of continuous light reaches a certain length, a diffusible flower-inducing sig-
nal is produced that is transmitted through the phloem to the shoot meristem. The
pathway that triggers flowering involves the plant’s circadian clock, the internal
24-hour timer that causes many metabolic and physiological processes, including the
expression of some genes, to vary throughout the day. One of the genes regulated by
the circadian clock is CONSTANS (CO), which is a key gene in controlling the onset
of flowering and provides the link between the plant’s daylength-sensing mechanism
Short days and production of the flowering signal. The expression of CO oscillates on a 24-hour
cycle under the control of the circadian clock, and its timing is such that the peak CO
expression occurs towards the end of the afternoon. This means that in longer days,
peak expression occurs in the light, whereas in short days, it will already be dark at
this time. In the dark, the CO protein is degraded and so the circadian control ensures
that CO only accumulates to high enough levels to trigger the flowering pathway
when light conditions are favorable (Fig. 7.34).
CO protein degraded
CO is a transcription factor that activates a gene known as FLOWERING LOCUS T
(FT), producing the FT protein, which appears to act as the flowering signal. The FT pro-
Long days
tein is thought to travel from the leaf through the phloem to the shoot apical meristem,
where it acts in a complex with the transcription factor FLOWERING LOCUS D (FD),

Fig. 7.34 The initiation of flowering is under the dual control of daylength and the
circadian clock. The transcription factor CONSTANS (CO) is required for production of the
flowering signal and is expressed in leaves under the control of the circadian clock. In short
days, expression of the CO gene peaks in the dark and the protein is rapidly degraded. In long
CO protein accumulates
days, peak expression occurs in the light, and the CO protein accumulates.
 Flower development and control of flowering 283

shoot apex New meristem primordium

FD
FD
FD FT

FT
FT AP1 Fig. 7.35 Signals that initiate flowering in
FD Arabidopsis. When the daylength becomes
AP1 longer after winter, the transcription
factor CO accumulates in the leaf, which
leaf in turn activates the transcription of FT in
Flower formation leaf phloem cells. The FT protein travels
through the phloem to the shoot apex.
FT RNA This interacts with the transcription factor
FD to form a complex, which acts together
CO with the transcription factor LEAFY (whose
FT own expression in upregulated in the shoot
apex by FT) to activate key floral meristem
FT identity genes such as AP1, which convert
the vegetative meristem into one that will
produce floral meristems.

which is expressed in the meristem, to turn on the expression of genes such as AP1
that promote flowering (Fig. 7.35). If FT is activated in a single leaf, this is sufficient
to induce flowering. Induction of flowering also requires the downregulation of a set
of floral repressor genes such as FLOWERING LOCUS C (FLC). These suppress the
transition from a vegetative to the flowering state until the positive signals to flower
are received. FLC encodes a protein that binds to FT and suppresses its activity.
After cold exposure, for example, FLC activity is low, and the repression of FT can
be released.

Summary

Before flowering, which is triggered by environmental conditions such as daylength,

the vegetative shoot apical meristem becomes converted into an inflorescence
meristem, which either then becomes a flower or produces a series of floral mer-
istems, each of which develops into a single flower. Genes involved in the initiation
of flowering and patterning of the flower have been identified in both Arabidopsis
and ­Antirrhinum. Flowering is induced by daylength acting together with the plant’s
natural circadian rhythms of gene expression to turn on a gene in the leaves that
produces a flowering signal that is transported to the shoot meristem. This signal
turns on the expression of meristem identity genes that are required for the trans-
formation of the vegetative shoot meristem to an inflorescence meristem and the
formation of floral meristems from the inflorescence meristem. Homeotic floral organ
identity genes, which specify the organ types found in the flowers, have been iden-
tified from mutations that transform one flower part into another. On the basis of
these mutations, a model has been proposed in which the floral meristem is divided
into three concentric overlapping regions, in each of which certain floral identity
genes act in a combinatorial manner to specify the organ type appropriate to each
whorl. Studies with chimeric plants have shown that different meristem layers com-
municate with each other during flower development and that transcription factors
can move between cells.
284 Chapter 7 Plant development

Summary: flower development in Arabidopsis

vegetative meristem meristem

environmental identity
signals genes
inflorescence meristem floral meristems
floral identity genes

three concentric overlapping regions of gene expression established

flower consisting of four concentric whorls: sepals, petals, stamens, and carpels

Summary to Chapter 7

A distinctive feature of plant development is the presence of relatively rigid walls and the
absence of any cell migration. Another is that a single, isolated somatic cell from a plant
can regenerate into a complete new plant. Early embryonic development is characterized
by asymmetric cell division of the fertilized egg, which specifies the future apical and basal
regions. During early development of flowering plants, both asymmetric cell division and cell–
cell interactions are involved in patterning the body plan. During this process, the shoot and
root meristems are specified and these meristems give rise to all the organs of the plant—
stems, leaves, flowers, and roots. The shoot meristem gives rise to leaves in well-defined
positions, a process involving regulated transport of a morphogen, auxin. The shoot mer-
istem eventually becomes converted to an inflorescence meristem, which either becomes a
floral meristem (in determinate inflorescences) or gives rise to a series of floral meristems,
retaining its shoot meristem identity indefinitely (in indeterminate inflorescences). In floral
meristems, each of which develops into a flower, homeotic floral organ identity genes act in
combination to specify the floral organ types. Increasing daylength induces the synthesis of
a flowering signal in the leaves that is transported to the shoot meristem where it induces
flower formation.

n End of chapter questions 7. What has the study of mericlinal chimeras revealed about cell
specification during plant embryogenesis? Are cells specified in the
Long answer (concept questions) embryo to form leaves and flowers, in a way analogous, for example,
to the specification of cells as dorsal mesoderm in animal embryos?
1. What features led to the adoption of Arabidopsis thaliana as the
predominant model for plant development? 8. What are the terms applied to the top and bottom surfaces of
2. Distinguish between the following parts of a plant: shoot, root, a leaf, in reference to the radial axis of the shoot? How are the
node, leaf, meristem, sepal, petal, stamen, carpel. ­transcription factors PHAB, PHAV, and REV restricted to the top
­surface?
3. What is the role of auxin before the 32-cell stage in Arabidopsis
embryogenesis? What is the mechanism by which a differential in 9. How does auxin control the positioning of leaf primordia? Include
auxin concentrations is generated? What is the mechanism by which the role of the PIN transporter and the concept of lateral inhibition
auxins influence gene expression? in your answer.

4. Describe the process by which transgenic plants are pro- 10. What is SHORT-ROOT? How does it come to be present in
duced. Include the role of the Ti plasmid, and of Agrobacterium endodermal cells? Might plasmodesmata be involved? Is there an
­tumefaciens. analogous process involving the SHOOT MERISTEMLESS ortholog
KNOTTED-1 of maize?
5. What is a meristem? Describe the structure of the shoot meristem
of Arabidopsis. 11. What is the nature of the homeotic mutations that can occur in
Arabidopsis? What genes are involved?
6. Contrast the roles of the homeobox genes WUSCHEL (WUS) and
SHOOT MERISTEMLESS (STM) in formation and maintenance of 12. What is the ABC model for flower development? How does it
shoot meristems. illustrate combinatorial control of cellular identity?
 Flower development and control of flowering 285

13. The MADS-box is named for the prototypic proteins in which 6. The agamous mutation causes the formation of
it was found: MCM1 (Saccharomyces), AGAMOUS (Arabidopsis), a) flowers with only petals and sepals
DEFICIENS (Antirrhinum), and SRF (Homo). What is the function
b) flowers with only sepals and carpels
of the MADS-box in a protein? (Note that the MADS-box proteins are
not related to SMADS; see Fig. 4.33.) c) flowers with only stamens and carpels
d) plants completely lacking flowers
14. What modification of the ABC model derived from studies of Ara­
bidopsis is required to explain flower development in Antirrhinum? 7. Maintenance of the shoot meristems in adult Arabidopsis plants
relies on which of the following mechanisms?
15. Through what mechanism is the photoperiod interpreted to
trigger flower development in Arabidopsis? a) A homeobox transcription factor encoded by the WUSCHEL gene
is expressed in the organizing center and initiates a signal to the
overlying cells to behave as stem cells.
Multiple choice (factual recall questions)
b) A transcription factor encoded by the SHOOT MERISTEMLESS
NB There is only one right answer to each question gene is expressed in shoot meristem cells and maintains them in
1. How many genes are present in humans, Drosophila mela- their undifferentiated state.
nogaster, Caenorhabditis elegans, and Arabidopsis, respectively. c) Shoot meristems cells secrete proteins encoded by the CLAVATA
(Although the number for humans has not yet been presented in the family that antagonize WUSCHEL expression, thereby restricting
text, the number for the other organisms has been described.) the size of the shoot meristems.
a) 19,000 - 27,000 - 14,000 - 19,000 d) All of these are involved in specification and maintenance of
b) 21,000 - 14,000 - 19,000 - 27,000 shoot meristems.

c) 27,000 - 21,000 - 19,000 - 14,000 8. What is meant by the word ‘whorl’ in discussing floral meristems?
d) 14,000 - 19,000 - 21,000 - 27,000 a) Flowers consist of four different types of organs, which occur in
concentric rings called ‘whorls.’
2. Embryogenesis in plants occurs
b) The floral meristem as to rotate during flower formation, giving
a) in the ovule, after the seed is fertilized and shed by the plant the process the name ‘whorl.’
b) in the seed after it germinates c) The flowers of Arabidopsis appear as the stem elongates in a pat-
c) in the seed, after the seed is fertilized tern called a ‘whorl.’
d) inside the ovule, before the seed is shed by the plant d) The six stamens in a dicot flower like that of Arabidopsis form a
ring that is called the flower’s ‘whorl.’
3. Which statement is true about the totipotency of cells?
a) All animal and plant cells are totipotent. 9. In what way are the homeotic genes of flowering plants similar
to those of Drosophila and other animals?
b) In plants, many cells are totipotent, whereas in animals, only the
fertilized egg is totipotent. a) All homeotic genes encode transcription factors of the homeobox
class.
c) Mammalian embryonic stem cells are totipotent.
b) Homeotic genes in both plants and animals encode transcription
d) Only the stem cells of animals, and the meristem cells of plants
factors of the MADS-box type.
are totipotent.
c) Mutations in the homeotic genes of flowers cause transformation
4. One of the earliest events in Arabidopsis development is forma- of one organ into another.
tion of the _____ axis, in response to a gradient of _____. d) The homeotic genes of flowers are derived during evolution from
a) adaxial–abaxial, cytokinins the same primordial genes used in animals.
b) apical–basal, auxin 10. How is the ABC model for floral identity in Arabidopsis reminis-
c) apical–basal, Pin proteins cent of the models for homeotic gene function derived from studies
d) dorsal–ventral, miRNA in Drosophila?
a) In both organisms, each homeobox gene specifies the identity of
5. The fate map of the Arabidopsis embryo at the heart stage indi-
a different region of the adult.
cates that
b) In both organisms, one homeotic gene is expressed at the two
a) although none of the adult structures have formed, the regions ends, a second expressed in domains more central to that of
that will give rise to the meristems, which will in turn give rise to the first, and a third expressed most centrally, thus contributing
adult structures, can be identified unambiguous identities to all regions of the organism.
b) development in plants is so indeterminate that a true fate map c) In both organisms, it is often the combination of genes pres-
cannot be drawn ent that is critical in unambiguously specifying structures in the
c) the primordia of the leaves, stems, and roots have already formed adult.
d) the three germ layers that will give rise to roots, stems, and d) The key responsibility of the homeotic genes in both organisms
leaves have formed is the patterning of antero-posterior identity.
286 Chapter 7 Plant development

Multiple choice answer key Großhardt, R., Laux, T.: Stem cell regulation in the shoot mer-
1: b, 2: d, 3: b, 4: b, 5: a, 6: a, 7: d, 8: a, 9: c, 10: c. istem. J. Cell Sci. 2003, 116: 1659–1666.
Ma, H.: Gene regulation: Better late than never? Curr. Biol. 2000,
10: R365–R368.
n General further reading
7.6 The size of the stem-cell area in the meristem is kept
constant by a feedback loop to the organizing center
Meyerowitz, E.M.: Arabidopsis—a useful weed. Cell 1989, 56:
263–264. Brand, U., Fletcher, J.C., Hobe, M., Meyerowitz, E.M., Simon, R.:
Meyerowitz, E.M.: Plants compared to animals: the broader com- Dependence of stem cell fate in Arabidopsis on a feedback loop
parative view of development. Science 2002, 295: 1482–1485. regulated by CLV3 activity. Science 2000, 289: 635–644.
Carles, C.C., Fletcher, J.C.: Shoot apical meristem mainte-
nance: the art of dynamical balance. Trends Plant Sci. 2003, 8:
­394–401.
n Section further reading
Clark, S.E.: Cell signalling at the shoot meristem. Nat Rev. Mol.
Cell Biol. 2001, 2: 277–284.
7.1 The model plant Arabidopsis thaliana has a short
Lenhard, M., Laux, T.: Stem cell homeostasis in the Arabidopsis­
life-cycle and a small diploid genome & 7.2 Plant embryos
shoot meristem is regulated by intercellular movement of
develop through several distinct stages
CLAVATA3 and its sequestration by CLAVATA1. Development
Lloyd, C.: Plant morphogenesis: life on a different plane. Curr.
2003, 130: 3163–3173.
Biol. 1995, 5: 1085–1087.
Reddy, G.V., Meyerowitz, E.M.: Stem-cell homeostasis and growth
Mayer, U., Jürgens, G.: Pattern formation in plant embryogenesis:
dynamics can be uncoupled in the Arabidopsis shoot apex. Sci-
a reassessment. Semin. Cell Dev. Biol. 1998, 9: 187–193.
ence 2005, 310: 663–667.
Meyerowitz, E.M.: Genetic control of cell division patterns in
Schoof, H., Lenhard, M., Haecker, A., Mayer, K.F.X., Jürgens, G.,
developing plants. Cell 1997, 88: 299–308.
Laux, T.: The stem cell population of Arabidopsis shoot mer-
Torres-Ruiz, R.A., Jürgens, G.: Mutations in the FASS gene uncou-
istems is maintained by a regulatory loop between the CLAV-
ple pattern formation and morphogenesis in Arabidopsis devel-
ATA and WUSCHEL genes. Cell 2000, 100: 635–644.
opment. Development 1994, 120: 2967–2978.
Vernoux, T., Benfey, P.N.: Signals that regulate stem cell activ-
ity during plant development. Curr. Opin. Genet. Dev. 2005, 15:
7.3 Gradients of the signal molecule auxin establish the
388–394.
embryonic apical–basal axis
Breuninger, H., Rikirsch, E., Hermann, M., Ueda, M., Laux, T.:
7.7 The fate of cells from different meristem layers can be
Differential expression of WOX genes mediates apical-basal
changed by changing their position
axis formation in the Arabidopsis embryo. Dev. Cell 2008, 14:
Castellano, M.M., Sablowski, R.: Intercellular signalling in the
867–876.
transition from stem cells to organogenesis in meristems. Curr.
Friml, J., Vieten, A., Sauer, M., Weijers, D., Schwarz, H., Hamann,
Opin. Plant Biol. 2005, 8: 26–31.
T., Offringa, R., Jürgens, G.: Efflux-dependent auxin gradients
Gallagher, K.L., Benfey, P.N.: Not just another hole in the wall:
establish the apical-basal axis of Arabidopsis. Nature 2003, 426:
understanding intercellular protein trafficking. Genes Dev.
147–153.
2005, 19: 189–195.
Jenik, P.D., Barton, M.K.: Surge and destroy: the role of auxin in
Laux, T., Mayer, K.F.X.: Cell fate regulation in the shoot meristem.
plant embryogenesis. Development 2005, 132: 3577–3585.
Semin. Cell Dev. Biol. 1998, 9: 195–200.
Jürgens, G.: Axis formation in plant embryogenesis: cues and
Sinha, N.R., Williams, R.E., Hake, S.: Overexpression of the maize
clues. Cell 1995, 81: 467–470.
homeobox gene knotted-1, causes a switch from determinate to
Long, J.A., Moan, E.I., Medford, J.I., Barton, M.K.: A member of
indeterminate cell fates. Genes Dev. 1993, 7: 787–795.
the knotted class of homeodomain proteins encoded by the
Turner, I.J., Pumfrey, J.E.: Cell fate in the shoot apical meristem of
STM gene of Arabidopsis. Nature 1995, 379: 66–69.
Arabidopsis thaliana. Development 1992, 115: 755–764.
Szemenyei, H., Hannon, M., Long, J.A.: TOPLESS mediates aux-
Waites, R., Simon, R.: Signaling cell fate in plant meristems: three
in-dependent transcriptional repression during Arabidopsis
clubs on one tousle. Cell 2000, 103: 835–838.
embryogenesis. Science 2008, 319: 1384–1386.

7.4 Plant somatic cells can give rise to embryos and seedlings 7.8 A fate map for the embryonic shoot meristem can be
deduced using clonal analysis
Zimmerman, J.L.: Somatic embryogenesis: a model for early deve­
lop­ment in higher plants. Plant Cell 1993, 5: 1411–1423. Irish, V.F.: Cell lineage in plant development. Curr. Opin. Genet.
Dev. 1991, 1: 169–173.
7.5 A meristem contains a small central zone of self-renewing
stem cells 7.9 Meristem development is dependent on signals from
Byrne, M.E., Kidner, C.A., Martienssen, R.A.: Plant stem cells: other parts of the plant
divergent pathways and common themes in shoots and roots. Doerner, P.: Shoot meristems: intercellular signals keep the bal-
Curr. Opin. Genet. Dev. 2003, 13: 551–557. ance. Curr. Biol. 1999, 9: R377–R380.
 Flower development and control of flowering 287

Irish, E.E., Nelson, T.M.: Development of maize plants from cul- Scheres, B.: An auxin-dependent distal organizer of pattern and
tured shoot apices. Planta 1988, 175: 9–12. polarity in the Arabidopsis root. Cell 1999, 99: 463–472.
Sachs, T.: Pattern Formation in Plant Tissues. Cambridge: Cambridge Scheres, B., McKhann, H.I., van den Berg, C.: Roots redefined: ana-
University Press, 1994. tomical and genetic analysis of root development. Plant Physiol.
1996, 111: 959–964.
7.10 Gene activity patterns the proximo-distal and adaxial– van den Berg, C., Willemsen, V., Hendriks, G., Weisbeek, P.,
abaxial axes of leaves developing from the shoot meristem Scheres, B.: Short-range control of cell differentiation in the
Bowman, J.L.: Axial patterning in leaves and other lateral organs. Arabidopsis root meristem. Nature 1997, 390: 287–289.
Curr. Opin. Genet. Dev. 2000, 10: 399–404. Veit, B.: Plumbing the pattern of roots. Nature 2007, 449: 991–
Kidner, C.A., Martienssen, R.A.: Spatially restricted microRNA directs 992.
leaf polarity through ARGONAUTE1. Nature 2004, 428: 81–84.
Waites, R., Selvadurai, H.R., Oliver, I.R., Hudson, A.: The PHAN- 7.13 Root hairs are specified by a combination of positional
TASTICA gene encodes a MYB transcription factor involved in information and lateral inhibition
growth and dorsoventrality of lateral organs in Antirrhinum. Kwak, S.-H., Schiefelbein, J.: The role of the SCRAMBLED recep-
Cell 1998, 93: 779–789. tor-like kinase in patterning the Arabidopsis root epidermis.
Dev. Biol. 2007, 302: 118–131.
7.11 The regular arrangement of leaves on a stem is Schiefelbein, J., Kwak, S.-H., Wieckowski, Y., Barron, C., Bruex, A.:
generated by regulated auxin transport The gene regulatory network for root epidermal cell-
Berleth, T., Scarpella, E., Prusinkiewicz, P.: Towards the systems ­type pattern formation in Arabidopsis. J. Exp. Bot. 2009, 60:
biology of auxin-transport-mediated patterning. Trends Plant ­1515–1521.
Sci. 2007, 12: 151–159.
Heisler, M.G., Ohno, C., Das, P., Sieber, P., Reddy, G.V., Long, J.A., 7.14 Homeotic genes control organ identity in the flower
Meyerowitz, E.M.: Patterns of auxin transport and gene expres- Bowman, J.L., Sakai, H., Jack, T., Weigel, D., Mayer, U., Meyerow-
sion during primordium development revealed by live imaging itz, E.M.: SUPERMAN, a regulator of floral homeotic genes in
of the Arabidopsis inflorescence meristem. Curr. Biol. 2005, 15: Arabidopsis. Development 1992, 114: 599–615.
1899–1911. Breuil-Broyer, S., Morel, P., de Almeida-Engler, J., Coustham, V.,
Jönsson, H., Heisler, M.G., Shapiro, B.E., Meyerowitz, E.M., Negrutiu, I., Trehin, C.: High-resolution boundary analysis dur-
­Mjolsness, E.: An auxin-driven polarized transport model for ing Arabidopsis thaliana flower development. Plant J. 2004, 38:
phyllotaxis. Proc. Natl. Acad. Sci. USA 2006, 103: 1633–1638. 182–192.
Mitchison, G.J.: Phyllotaxis and the Fibonacci series. Science 1977, Coen, E.S., Meyerowitz, E.M.: The war of the whorls: genetic
196: 270–275. interactions controlling flower development. Nature 1991, 353:
Reinhardt, D.: Regulation of phyllotaxis. Int. J. Dev. Biol. 2005, 49: 31–37.
539–546. Irish, V.F.: Patterning the flower. Dev. Biol. 1999, 209: 211–220.
Scheres, B.: Non-linear signaling for pattern formation. Curr. Krizek, B.A., Meyerowitz, E.M.: The Arabidopsis homeotic genes
Opin. Plant Biol. 2000, 3: 412–417. APETALA3 and PISTILLATA are sufficient to provide the B class
Schiefelbein, J.: Cell-fate specification in the epidermis: a com- organ identity function. Development 1996, 122: 11–22.
mon patterning mechanism in the root and shoot. Curr. Opin. Krizek, B.A., Fletcher, J.C.: Molecular mechanisms of flower
Plant Biol. 2003, 6: 74–78. ­development: an armchair guide. Nat Rev. Genet. 2005, 6:
Smith, L.G., Hake, S.: The initiation and determination of leaves. 688–698.
Plant Cell 1992, 4: 1017–1027. Lohmann, J.U., Weigel, D.: Building beauty: the genetic control of
floral patterning. Dev. Cell 2002, 2: 135–142.
7.12 Root tissues are produced from Arabidopsis root apical Ma, H., dePamphilis, C.: The ABCs of floral evolution. Cell 2000,
meristems by a highly stereotyped pattern of cell divisions 101: 5–8.
Costa, S., Dolan, L.: Development of the root pole and patterning Meyerowitz, E.M., Bowman, J.L., Brockman, L.L., Drews, G.M.,
in Arabidopsis roots. Curr. Opin. Genet. Dev. 2000, 10: 405–409. Jack, T., Sieburth, L.E., Weigel, D.: A genetic and molecular
Dello Ioio, R., Nakamura, K., Moubayidin, L., Perilli, S., Taniguchi, M., model for flower development in Arabidopsis thaliana. Devel-
Morita, M.T., Aoyama, T., Costantino, P., Sabatini, S.: A genetic opment Suppl. 1991, 1: 157–167.
framework for the control of cell division and differentiation in Meyerowitz, E.M.: The genetics of flower development. Sci. Am.
the root meristem. Science 2008, 322: 1380–1384. 1994, 271: 40–47.
Dolan, L.: Positional information and mobile transcriptional reg- Sakai, H., Medrano, L.J., Meyerowitz, E.M.: Role of SUPERMAN in
ulators determine cell pattern in Arabidopsis root epidermis. maintaining Arabidopsis floral whorl boundaries. Nature 1994,
J. Exp. Bot. 2006, 57: 51–54. 378: 199–203.
Grieneisen, V.A., Xu, J., Marée, A.F.M., Hogeweg P., Scheres, B.: Vincent, C.A., Carpenter, R., Coen, E.S.: Cell lineage patterns and
Auxin transport is sufficient to generate a maximum and gradi- homeotic gene activity during Antirrhinum flower develop-
ent guiding root growth. Nature 2007, 449: 1008–1013. ment. Curr. Biol. 1995, 5: 1449–1458.
Sabatini, S., Beis, D., Wolkenfeldt, H., Murfett, J., Guilfoyle, T., Wagner, D., Sablowski, R.W.M., Meyerowitz, E.M.: Transcriptional
Malamy, J., Benfey, P., Leyser, O., Bechtold, N., Weisbeek, P., activation of APETALA 1. Science 1999, 285: 582–584.
288 Chapter 7 Plant development

7.15 The Antirrhinum flower is patterned dorso-ventrally as Corbesier, L., Vincent, C., Jang, S., Fornara, F., Fan, Q., Searle, I.,
well as radially Giakountis, A., Farrona, S., Gissot, L., Turnbull, C., Coupland, G.:
Coen, E.S.: Floral symmetry. EMBO J. 1996, 15: 6777–6788. FT protein movement contributes to long-distance signaling in
Luo, D., Carpenter, R., Vincent, C., Copsey, L., Coen, E.: Origin of floral induction of Arabidopsis. Science 2007, 316: 1030–1033.
floral asymmetry in Antirrhinum. Nature 1996, 383: 794–799. Hake, S.: Transcription factors on the move. Trends Genet. 2001,
17: 2–3.
7.16 The internal meristem layer can specify floral meristem Jaeger, K.E., Wigge, P.A.: FT protein acts as a long-range signal in
patterning Arabidopsis. Curr. Biol. 2007, 17: 1050–1054.
Szymkowiak, E.J., Sussex, I.M.: The internal meristem layer (L3) Kobayashi, Y., Weigel, D.: Move on up, it’s time for change -
determines floral meristem size and carpel number in tomato ­mobile signals controlling photoperiod-dependent flowering.
periclinal chimeras. Plant Cell 1992, 4: 1089–1100. Genes Dev. 2007, 21: 2371–2384.
Lohmann, J.U., Hong, R.L., Hobe, M., Busch, M.A., Parcy, F., Simon,
7.17 The transition of a shoot meristem to a floral meristem R., Weigel, D.: A molecular link between stem cell regulation
is under environmental and genetic control and floral patterning in Arabidopsis. Cell 2001, 105: 793–803.
An, H., Roussot, C., Suarez-Lopez, P., Corbesier, L., Vincent, C., Putterill, J., Laurie, R., Macknight, R.: It’s time to flower: the
Pineiro, M., Hepworth, S., Mouradov, A., Justin, S., Turnbull, C., genetic control of flowering time. BioEssays 2004, 26: 363–373.
Coupland, G.: CONSTANS acts in the phloem to regulate a sys- Sheldon, C.C., Hills, M.J., Lister, C., Dean, C., Dennis, E.S., Peacock,
temic signal that induces photoperiodic flowering of Arabidop- W.J.: Resetting of FLOWERING LOCUS C expression after epi-
sis. Development 2004, 131: 3615–3626. genetic repression by vernalization. Proc. Natl. Acad. Sci. USA
Becroft, P.W.: Intercellular induction of homeotic gene expression 2008, 105: 2214–2219.
in flower development. Trends Genet. 1995, 11: 253–255. Valverde, F., Mouradov, A., Soppe, W., Ravenscroft, D., Samach, A.,
Blázquez, M.A.: The right time and place for making flowers. Coupland, G.: Photoreceptor regulation of CONSTANS protein
­Science 2005, 309: 1024–1025. in photoperiodic flowering. Science 2004, 303: 1003–1006.
 8
Morphogenesis: change in form
in the early embryo
nn Cell adhesion nn Gastrulation movements nn Cell migration
nn Cleavage and formation of nn Neural tube formation nn Directed dilation
the blastula

Changes in form in animal embryos are brought about by cellular forces that are
generated in a variety of ways, including cell division, change in cell shape, rear-
rangement of cells within tissues, and the migration of individual cells from one
part of the embryo to another. These cellular forces are most frequently generated
within ­ep­ithelial sheets by contraction of the cells’ internal cytoskeleton; resisting
these forces are the cells’ own structure and the adhesive interactions that hold cells
together in tissues. The generation of hydrostatic pressure within a structure, cell
division, and cell migration also play a part in morphogenetic processes. The major
morphogenetic event in early animal embryos is gastrulation, when a wholesale reor-
ganization of the body plan occurs. Other examples of morphogenesis discussed in
this chapter include notochord and neural tube formation, and the migration of neural
crest cells to specific locations. Plants develop form through cell division and cell
enlargement.

So far, we have discussed early development mainly from the viewpoint of devel-
opmental patterning and the assignment of cell fate. In this chapter, we look at
embryonic development from a different perspective—the generation of form, or
morphogenesis. All animal embryos undergo a dramatic change in shape during
their early development. This occurs principally during gastrulation, the process
that transforms a two-dimensional sheet of cells into the complex three-dimensional
animal body. Gastrulation involves extensive rearrangements of cell layers and the
directed movement of cells from one location to another.
If pattern formation can be likened to painting, morphogenesis is more akin to
modeling a formless lump of clay into a recognizable shape. Change in form is largely
a problem in cell mechanics: understanding it requires an appreciation of the forces
involved in bringing about changes in cell shape and cell migration, the cellular
machinery that carries them out, and how this is controlled during development.
Two key cellular properties involved in changes in animal embryonic form are cell
adhesiveness and cell motility. Animal cells stick to one another, and to the extra-
cellular matrix, through interactions involving cell-surface proteins. Changes in the
290 Chapter 8 Morphogenesis: change in form in the early embryo

adhesion proteins at the cell surface can therefore determine the strength of cell–
cell adhesion and its specificity. These adhesive interactions affect the surface tension
at the cell membrane, a property that contributes to the mechanics of the cell rear-
rangements that generate the three distinct germ layers during gastrulation. Different
surface tensions are what keep two immiscible liquids such as water and oil separate,
and embryologists had long noted the ‘liquid-like’ behavior of the tissues in gastrulat-
ing frog embryos, while observing how the separate germ layers formed and remained
distinct from each other, slid over each other, spread out, and underwent internal
rearrangement to shape the embryo. The modern differential adhesion hypothesis
explains these liquid-like tissue dynamics as the consequences of surface tensions
and interfacial tensions generated by cells that tend to stick to each other but are also
able to change shape and to move.
The second key property—cell motility—encompasses both the migration of indi-
vidual cells and the ability of cells to change shape while remaining part of a tissue.
For example, folding of a cell sheet—a very common feature in embryonic devel-
opment—­is caused by changes in cell shape. Cells move and change shape by rear-
ranging their internal cytoskeleton, particularly those elements of it that can cause
contraction or constriction of part of the cell. Cellular contraction is generated by
actin filaments complexed with the motor protein myosin, forming contractile struc-
tures similar to, but simpler than, those in muscle. In non-muscle cells, contrac-
tile actin–myosin complexes (actomyosin) are concentrated in the cell cortex—the
region immediately underneath the cell membrane. Changes in shape that lead to
cell migration involve the extension of actin-based structures, such as filopodia
and lamellipodia, in the direction of movement (Box 8A). An additional force that
operates during morphogenesis, particularly in plants but also in some aspects of
animal embryogenesis, is hydrostatic pressure, which is generated by osmosis and
fluid accumulation. In plants there is no cell movement during growth, and changes
in form are generated by oriented cell division and cell expansion, as we saw in
Chapter 7.
Changes in embryonic form are the final consequence of the precise spatio-
­temporal expression of proteins that control cell adhesion, cell motility, oriented
cell division, and the generation of hydrostatic pressure. It is likely that changes in
form are brought about by an earlier patterning process that determines which cells
will express those proteins that are required to generate and harness the appropriate
forces. An attractive hypothesis is that pattern-determining genes, such as the Hox
genes, activate genes that control the expression of such proteins, and there is some
evidence for this.
In this chapter we shall discuss mainly the morphological changes that occur
during the development of the animal body plan. First, we will look at how cleav-
age of the zygote gives rise to the simple shape of the early embryo, of which the
spherical blastocyst of the mouse and the blastulas of sea urchin and amphibian
are good examples. We then consider the movements that occur during gastrula-
tion and during neurulation—the formation of the neural tube in vertebrates—
which involve folding of cell sheets and rearrangement of cell layers. In vertebrates,
migration of cells from the neural crest after neurulation generates a variety of
structures in the trunk and head, and we consider how these cells migrate to their
correct sites. Finally, we look at directed dilation, in which hydrostatic pressure
is the force driving changes in shape. Other morphogenetic mechanisms such as
cell growth, cell proliferation, and cell death will be considered in relation to the
development of particular organs and the growth of an organism as a whole (see
Chapters 11, 12, and 13).
We begin by considering how cells adhere to each other, and how differences in
adhesiveness and specificity of adhesion are involved in maintaining boundaries
­between tissues.
 Cell adhesion   291

Box 8A Change in cell shape and cell movement

Cells actively undergo major changes in shape during develop- a useful agent for investigating the role of actin networks. Actin
ment. The two main changes are associated with cell migration filaments can also assemble with myosin into contractile struc-
and with the infolding of epithelial sheets. Changes in shape are tures, which act as miniature muscles. For example, contraction
generated by the cytoskeleton, an intracellular protein framework of the contractile ring, bundles of actomyosin arranged in a ring
that also controls changes in cell movement. There are three in the cell cortex, pinches animal cells in two at cell division (see
principal types of protein polymers in the ­cytoskeleton—actin figure, left panel). The contraction of a similar ring of actomyosin
filaments (microfilaments), microtubules, and intermedi- around the apical end of a cell leads to apical constriction and
ate filaments—as well as many other proteins that interact with elongation of the cell (see figure, center panel).
them. Actin filaments and ­microtubules are dynamic structures, Many embryonic cells can migrate over a solid substratum,
polymerizing and depolymerizing according to the cell’s require- such as the extracellular matrix. They move by extending a
ments. Intermediate filaments are more stable, forming rope-like thin sheet-like layer of cytoplasm known as a lamellipodium
structures that transmit mechanical forces, spread mechanical (see figure, right panel), or long, fine cytoplasmic processes
stress, and provide mechanical stability to the cell. Microtubules called filopodia. Both these temporary structures are pushed
play an important part in maintaining cell asymmetry and polar- outwards from the cell by the assembly of actin filaments. Con-
ity, and provide tracks along which motor proteins convey other traction of the actomyosin network at the front and the rear
molecules and even organelles. They also form the spindles that of the cell then draws the cell forward. To do this, the contrac-
segregate chromosomes at mitosis and meiosis. Polymerization tile system must be able to exert a force on the substratum
and depolymerization of actin ­filaments at the leading edges of and this occurs at focal contacts, points at which the advanc-
cells are involved in cell shape change and cell movements, and ing filopodia or lamellipodium are anchored to the surface
actin filaments also associate with the motor protein myosin over which the cell is moving. At focal contacts, integrins (see
to form contractile bundles of ­actomyosin, which are primarily Box 8B, p. 293) both adhere to extracellular matrix molecules
­responsible for force-generating contractions within cells. through their extracellular domains, and provide an anchor
Actin filaments are fine threads of protein about 7 nm in point for actin filaments through their cytoplasmic domains.
­diameter and are polymers of the globular protein actin. They are Integrins transduce signals from the extracellular matrix across
organized into bundles and three-dimensional networks, which, the plasma membrane at focal contacts, enabling the cell to
in most cells, lie mostly just beneath the plasma membrane, sense the environment over which it is traveling and to adjust
forming the gel-like cell cortex. Numerous actin-binding proteins its movement accordingly.
are associated with actin filaments, and are involved in bundling The small GTP-binding proteins known as the Rho-family
them together, forming networks, and aiding the ­polymerization ­GTPases, which include Rho, Rac, and Cdc42, have a key role
and depolymerization of the actin subunits. Actin filaments in regulating the actin cytoskeleton. Rac is required for exten-
can form rapidly by polymerization of actin subunits and can sion of lamellipodia, for example, and Cdc42 is necessary to
be equally rapidly depolymerized. This provides the cell with a maintain cell polarity. Rho is involved in the planar cell polarity
highly versatile system for assembling actin filaments in a vari- signaling pathway (see Box 8C, p. 310). Signals from the envi-
ety of different ways and in different locations, as required. The ronment relayed to the cytoskeleton via the Rho-family proteins
fungal drug cytochalasin D prevents actin polymerization and is are thus able to control cell movement.

Cleaving cell Apical constriction Migrating cell

focal contact
contractile network actin bundles

myosin
contractile ring
lamellipodium
292 Chapter 8 Morphogenesis: change in form in the early embryo

Cell adhesion

The late embryo and the adult are composed of a variety of differentiated cell types,
which are grouped together to form tissues such as skin, cartilage and muscle. The
integrity of tissues is maintained by adhesive interactions between cells and between
cells and the extracellular matrix; differences in cell adhesiveness also help maintain
the boundaries between different tissues and structures. Cells stick to each other
by means of cell-adhesion molecules, which are proteins on the cell surface that
can bind strongly to proteins on other cell surfaces or in the extracellular matrix
(Box 8B). The particular adhesion molecules expressed by a cell determine which
cells it can adhere to, and changes in the adhesion molecules expressed are involved
in many developmental phenomena. In epithelial tissues, with which we shall be
most concerned here, adjacent cells are joined together by specialized structures
called ­adhesive cell junctions that incorporate cell-adhesion molecules. The junc-
tions discussed in this chapter are the adherens junction, in which the adhesion
molecule is cadherin, which is linked intracellularly to the actin cytoskeleton, and the
tight ­junction, which joins epithelial cells together with a tight ‘seal’ that forms a
permeability barrier between the environments inside and outside the epithelium.

8.1 Sorting out of dissociated cells demonstrates differences in cell

adhesiveness in different tissues

Differences in cell adhesiveness can be illustrated by experiments in which different

tissues are confronted with one another in an artificial setting. Two pieces of early
endoderm from an amphibian blastula will fuse to form a smooth sphere if placed in
contact. In contrast, when a piece of early endoderm and a piece of early ectoderm
are combined they initially fuse, but in time the endodermal and ectodermal cells
separate, until only a narrow bridge connects the two types of tissue (Fig. 8.1).
In another type of experiment, cells from two different tissues are dissociated,
mixed together, and then allowed to reaggregate. When cells of amphibian pre-
sumptive epidermis and presumptive neural plate are dissociated, mixed, and left to
reaggregate, they sort out to reform the two different tissues (Fig. 8.2). The epidermal
cells are eventually found on the outer face of the aggregate, surrounding a mass
of neural cells—the same types of cell are now in contact with each other. Mixed
­ectodermal and mesodermal cells similarly sort themselves out to form a mass of cells
with ectoderm on the outside and mesoderm on the inside.
This sorting out is the combined result of cell movement and differences in adhe-
siveness. Initially, cells move about randomly in the mixed aggregate, exchanging
weaker for stronger adhesions. The adhesive interactions between cells produce

ectoderm

Fig. 8.1 Separation of embryonic tissues

with different adhesive properties. When
pieces of early ectoderm (blue) and early
endoderm (yellow) from an amphibian
blastula are placed together, they initially
endoderm
fuse, but then separate until only a narrow
strip of tissue joins the two tissues.
 Cell adhesion 293

Box 8B Cell-adhesion molecules and cell junctions

Three classes of adhesion molecules are particularly important­­ for normal strong cell–cell adhesion. In adherens junctions the
in development (see figure). The cadherins are transmem­brane connection is to cytoskeletal actin filaments, whereas in des-
proteins that, in the presence of calcium ions (Ca2+), adhere to mosomes and hemi-desmosomes it is to intermediate filaments
cadherins on the surface of another cell. Calcium-independent­ such as keratin. Nectin, a member of the immunoglobulin super-
cell-cell adhesion involves a different structural class of proteins— family of adhesion molecules, also clusters to adherens junctions
­members of the large immunoglobulin superfamily. The neu- in mammalian tissues and connects to the actin cytoskeleton.
ral cell-adhesion molecule (N-CAM), which was first isolated from Adhesion of a cell to the extracellular matrix, which contains
neural tissue, is a typical member of this family. Some immu- proteins such as collagen, fibronectin, laminin, and tenascin, as
noglobulin superfamily members, such as N-CAM, bind to simi- well as proteoglycans, is by the binding of integrins to these
lar molecules on other cells; others bind to a different class of matrix molecules. An integrin molecule is composed of two
adhesion molecule, the integrins. Integrins also act as receptors ­different subunits, an α subunit and a β subunit. Twenty-four dif-
for molecules of the extracellular matrix, which they can use to ferent integrins are known so far in vertebrates, made up from
mediate adhesion between a cell and its substratum. eight β subunits and 18 α subunits. Many extracellular matrix
About 30 different types of cadherins have been identified in molecules are recognized by more than one integrin.
vertebrates. Cadherins bind to each other through one or more Integrins not only bind to other molecules through their extra-
binding sites located within the extracellular amino-terminal cellular face, but they also associate with the actin filaments
100 amino acids. In general, a cadherin binds only to another or intermediate filaments of the cell’s cytoskeleton through
cadherin of the same type, but they can also bind to some ­other complexes of proteins that are in contact with the integrin’s
molecules. Cadherins are the adhesive components in both ­cytoplasmic region. This association enables integrins to transmit
­adherens junctions (adhesive cell–cell junctions that are present ­information about the extracellular environment, such as extra-
in many tissues) and desmosomes (cell–cell junctions present cellular matrix composition or the type of intercellular contact, to
mainly in epithelia). Hemidesmosomes are an integrin-based the cell. Integrins can, therefore, transmit signals from the matrix
type of junction that rivets epithelial cells firmly to the extracel- that affect cell shape, motility, metabolism, and cell differentia-
lular matrix of the basement membrane. tion. Integrins are also involved in adhesion between cells, binding
As cells approach and touch one another, the cadherins clus- either to adhesion molecules of the immunoglobulin superfamily
ter at the site of contact. They interact with the intracellular or via a shared ligand to integrins on another cell surface.
cytoskeleton through the connection of their cytoplasmic tails A third type of cell junction present in epithelia is the tight
with catenins and other proteins, and thus can be involved in junction, which forms a seal that prevents water and other
transmitting signals to the cytoskeleton. This adhesive and sig- molecules passing between the epithelial cells. Multiple lines of
naling role of α-, β-, and γ -catenins is separate from the role transmembrane proteins called claudins and occludins in adjoin-
of β-catenin as a gene-regulatory protein (see, for example, ing cell membranes form a continuous seal around the apical end
Section 4.2). The interaction with the cytoskeleton is required of the cell (see Fig. 8.12).

Cell–cell adhesion Cell–matrix adhesion

Calcium-independent
Adherens junction Desmosome adhesion
vinculin
immunoglobulin superfamily talin

catenins
� �
integrin
Ca2+ Ca2+ Ca2+ Ca2+

Ca2+ Ca2+ cadherin cadherin Ca2+ Ca2+

laminin
Extracellular matrix
actin plakoglobin
bundles
desmoplakin
immunoglobulin superfamily
intermediate filaments e.g. N-CAM
294 Chapter 8 Morphogenesis: change in form in the early embryo

­ ifferent degrees of surface tension that is sufficient to generate the sorting-out behav-
d
Tissues taken from amphibian neurulas
ior, just as two immiscible liquids, such as oil and water, separate out when mixed.
As the constituent cells form stable contacts with each other, the tissues become
reorganized and the strength of intercellular binding in the system as a whole is maxi-
mized. In general, if the adhesion between unlike cells is weaker than the average of
the adhesions between like cells, the cells will segregate according to type, with the
more ­cohesive tissue tending to be enveloped by the less cohesive one. In terms of
the surface tension, those cells with the higher surface tension become surrounded by
presumptive presumptive
epidermal cells neural plate those with lower surface tension.
It is important to note that the above experiments on isolated cells in vitro simply
indicate the different adhesiveness of different tissues; they do not define which tis-
sue will be on the inside or the outside in the embryo itself. This is because in a real
embryo, tissues are under pre-existing developmental constraints. In the embryo,
the most cohesive tissue is the ectoderm, which is derived from the outermost layer
Tissues disaggregated into single cells of cells, forms a continuous sheet, and thus remains on the outer surface of the
embryo, while the less cohesive mesoderm and endoderm become internalized.
If ectodermal and mesendodermal cells from zebrafish are isolated and mixed in
vitro, however, they will rearrange with the more cohesive ectodermal cells on the
inside.
The in vitro experiments also show how differential cell adhesion can, in principle,
stabilize boundaries between tissues. We have already seen an example of boundary
formation by a somewhat different mechanism in Chapter 5, in the case of the rhom-
bomere boundaries in the hindbrain. There, repulsive interactions between ephrins
and their Eph receptors across a rhombomere boundary prevent cells from differ-
Spontaneous reaggregation. Cells sort ent rhombomeres intermingling, whereas cohesive Eph–ephrin interactions within a
out with epidermal cells outermost
rhombomere promote cell adhesion (see Section 5.11).

8.2 Cadherins can provide adhesive specificity

Differential adhesiveness between cells is the result of differences in the types and
numbers of adhesion molecules on cell surfaces. Evidence that the cadherins can
provide adhesive specificity comes from studies in which cells with different cad-
herins on their surface are mixed together. Fibroblasts of the mouse L cell line do not
Fig. 8.2 Sorting out of different cell express cadherins on their surface and do not adhere strongly to each other. But if
types. Ectoderm from the presumptive the gene encoding E-cadherin is introduced (transfected) into L cells in a form that
epidermis (gray) and from the presumptive can be expressed, the cells produce that cadherin on their surface and stick together,
neural plate (blue) of early amphibian forming a structure resembling a compact epithelium. The adherence is both calcium
neurulas are disaggregated into single cells dependent, indicating that it is due to the cadherin, and specific, as the transfected
by treatment with an alkaline solution. The cells do not adhere to untreated L cells lacking surface cadherins.
cells, when mixed together, sort out with the When groups of L cells are transfected with different types of cadherin and mixed
epidermal cells on the outside. together in suspension, only those cells expressing the same cadherin adhere strongly
to each other: cells expressing E-cadherin adhere strongly to other cells expressing
E-cadherin, but only weakly to cells expressing P- or N-cadherin. The amount of
cadherin on the cell surface can also have an effect on adhesion. When cells express-
ing different amounts of the same cadherin are mixed together, those cells with more
cadherin on their surface form an inner ball, surrounded by the cells with less surface
cadherin (Fig. 8.3). Thus, quantitative differences in cell-adhesion molecules could
maintain differential cell adhesion.
Cadherin molecules bind to each other through their extracellular domains, but
adhesion is not exclusively controlled by these domains. The cytoplasmic domain
associates with the actin filaments of the cytoskeleton by means of a protein com-
plex containing catenins (see Box 8B, p. 293), and failure to make this association
results in weak adhesion. In early Xenopus blastulas, for example, E-cadherin is
 Cleavage and formation of the blastula 295

Fig. 8.3 Sorting out of cells carrying

different amounts of adhesion molecules
on their surfaces. When two cell lines with
different amounts of N-cadherin on their
surface are mixed, they reassort with the
cells containing the most N-cadherin (green
stained cells) ending up on the inside.
Photographs courtesy of M. Steinberg, from
Foty, R.A., Steinberg, M.S.: 2005.

e­ xpressed in the ectoderm just before gastrulation and N-cadherin appears in the
prospective neural plate. If E-cadherin lacking an extracellular domain is produced
in the blastulas from injected mutated mRNA, the mutant cadherin will compete
with the embryo’s own intact cadherin molecules for the cytoskeletal association
sites. The defective cadherin cannot influence cell–cell adhesion as it has no extra-
cellular domain, but it blocks the intact cadherin’s access to the cytoskeleton. The
result is disruption of the ectoderm during gastrulation, indicating that a cadherin
molecule must bind both to its partner on an adjacent cell and to the cytoskeleton
of its own cell to create a stable adhesion. The initial binding of the extracellular
cadherin domains transmits a signal to the cytoskeleton, which then stabilizes the
interaction.

Summary

The adhesion of cells to each other and to the extracellular matrix maintains the integ-
rity of tissues and the boundaries between them. The associations of cells with each
other are determined by the cell-adhesion molecules they express on their surface: cells
bearing different adhesion molecules, or different quantities of the same molecule, sort
out into separate tissues. This is due to the effects of intercellular adhesiveness on the
cortical cytosekeleton and thus on cell-surface tension and interfacial tension between
tissues. Cell–cell adhesion is due mainly to two classes of surface proteins: the cad-
herins, which bind in a calcium-dependent manner to identical cadherins on another cell
surface, and members of the immunoglobulin superfamily, some of which bind to simi-
lar molecules on other cells, while others bind to different molecules, such as integrins.
The binding of this second class of surface proteins is calcium independent. Adhesion
to the extracellular matrix is mediated by a third class of adhesion molecule—the inte-
grins. Stable cell–cell adhesion via cadherins involves both physical contact between
their external domains and the cytoplasmic domain of the cadherin molecule binding to
the cell cytoskeleton via a protein complex containing catenins.

Cleavage and formation of the blastula

The first step in animal embryonic development is the division of the fertilized egg
by cleavage into a number of smaller cells (blastomeres), leading in many animals
to the formation of a hollow sphere of cells—the blastula (see Chapter 3). In many
embryos, cleavage involves short mitotic cell cycles, in which cell division and mito-
sis succeed each other repeatedly without intervening periods of cell growth. During
296 Chapter 8 Morphogenesis: change in form in the early embryo

cleavage, therefore, the mass of the embryo does not increase. Early cleavage patterns
can vary widely between different groups of animals (Fig. 8.4). In radial cleavage,
cleavages occur at right angles to the egg surface and the first few cleavages pro-
duce tiers of blastomeres that sit directly over each other. This type of cleavage is
characteristic of the deuterostomes, such as sea urchins and vertebrates. The eggs of
molluscs and annelids, which are protostomes, illustrate another cleavage pattern,
called spiral cleavage, in which successive divisions are at planes at slight angles to
each other, producing a spiral arrangement of cells. In both radial and spiral cleav-
age, some divisions may be unequal. The first three cleavages in the sea urchin,
for example, give rise to equal-sized blastomeres, whereas the later cleavage that
forms the micromeres is unequal, one daughter cell being smaller than the other (see
Fig. 6.19). In nematodes, the very first cleavage of the egg is unequal, producing two
cells of unequal size (see Fig. 8.4). In the early Drosophila embryo, the nuclei undergo
repeated divisions without cell division, forming a syncytium, and separation into
individual cells only occurs later, with the growth of cell membranes between the
nuclei (see ­Section 2.1).
The amount of yolk in the egg can influence the pattern of cleavage. In yolky eggs
undergoing otherwise symmetric cleavage, a cleavage furrow develops in the least
yolky region and gradually spreads across the egg, but its progress is slowed or even
halted by the presence of the yolk. Cleavage may thus be incomplete for some time.
This effect is most pronounced in the heavily yolked eggs of birds and zebrafish,
where complete divisions are restricted to a region at one end of the egg, and the
embryo is formed as a cap of cells sitting on top of the yolk (see Figs. 3.11 and 3.13).
Even in moderately yolky eggs, such as those of amphibians, the presence of yolk can
influence cleavage patterns. In frog eggs, for example, the later cleavages are unequal
and asynchronous, resulting in an animal half composed of a mass of small cells and
a yolky vegetal region composed of fewer and larger cells.
Two key questions arise in relation to early cleavage: how are the positions of the
cleavage planes determined; and how does cleavage lead to a hollow blastula (or its
equivalent), which has a clear inside–outside polarity?

Radial cleavage—sea urchin Unequal cleavage—nematode Spiral cleavage—

annelids and molluscs

Fig. 8.4 Different patterns of early

cleavage are found in different animal
groups. Radial cleavage (for example, sea
urchin) gives tiers of blastomeres sitting
directly above each other. Unequal cleavage
(for example, nematode) results in one
daughter cell being larger than the other. In
spiral cleavage (for example, molluscs and
annelid worms), the mitotic apparatus is
oriented at a slight angle to the long axis of
the cell, resulting in a spiral arrangement of
blastomeres.
 Cleavage and formation of the blastula 297

8.3 The orientation of the mitotic spindle determines the plane of cleavage at
cell division

The orientation of the plane of cleavage at cell division is of great importance in

embryonic development. It not only determines the spatial arrangement of daughter
cells relative to each other, and whether they will be of equal or unequal size, but it
can also result in the apportionment of different cytoplasmic determinants to different
daughter cells, thus determining their fate. We have already seen such effects in the
unequal cleavages of early nematode development, which produce cells with different
fates (see Chapter 6). The plane of cleavage can also be of great importance in later
morphogenesis and growth. It determines, for example, whether an epithelial sheet
remains as a single cell layer or becomes multilayered.
Cell division always occurs halfway between the two poles of the mitotic spindle.
One possible explanation proposes that there is an overall increase in tension in the
cell cortex due to cytoskeletal contraction, which causes the cell to round up. The
asters at the poles of the spindle then interact with the cortex to relax the tension
equally in the regions immediately adjacent to themselves, leaving the region between
them to form a contractile ring and divide the cell into two. It has been shown experi-
mentally that the position of a cleavage furrow can be specified by the asters alone,
independently of the presence of a mitotic spindle. If a fertilized sea-urchin egg is
deformed by inserting a glass bead that interrupts the first cleavage furrow, this cre-
ates a horseshoe-shaped cell in which the two nuclei generated by the first mitotic
division are each segregated into an arm of the horseshoe (Fig. 8.5). At the next mito-
sis, spindles form normally in each arm of the horseshoe and each is bisected by a
cleavage furrow, but a third furrow also forms between the two adjacent asters at the
top of the horseshoe that are not connected by a spindle.
An aster is composed of microtubules that radiate out from a structure called a
centrosome, which is an organizing center for microtubule growth in the cell. Before
mitosis, the single centrosome in the cell becomes duplicated and the daughter cen-
trosomes move to opposite sides of the nucleus and form the asters, whose inter- Fig. 8.5 The positions of cleavage furrows
actions with cortical microtubules will anchor and help orient the mitotic spindle. are determined by the asters of the
The pattern of centrosome duplication and movement usually results in successive mitotic spindle. If the mitotic apparatus of a
planes of cell division being at right angles to each other. An example of this is divi- fertilized sea-urchin egg is displaced at the
sion of the nematode AB cell, one of the pair of cells formed by the first cleavage of first cleavage by a glass bead, the cleavage
the zygote, where the mitotic spindle in the dividing AB cell is oriented roughly at furrow forms only on the side of the egg to
which the mitotic apparatus has been moved.
right angles to that of the zygote (Fig. 8.6). However, in the other product of the first
At the next cleavage, furrows bisect each
cleavage, the P1 cell, the centrosomes initially migrate to the same positions as in the
mitotic spindle as expected, but an additional
AB cell but the nucleus and its associated centrosomes then rotates 90° towards the
furrow, indicated by the blue triangle, forms
anterior of the cell, so that the eventual plane of cleavage of P1 is the same as that of between the two adjacent asters, even
the zygote. The different planes of cleavage in the AB and P1 cells are likely to be though there is no spindle between them.
Chromosomes are not shown for simplicity.

Glass bead displaces the spindle Bead interrupts the cleavage furrow. At second cleavage, a furrow
prior to division A horseshoe-shaped cell with develops between asters whether Four daughter cells are formed
two nuclei is formed or not a spindle is present

glass bead additional cleavage furrow

spindle nucleus cleavage furrow

298 Chapter 8 Morphogenesis: change in form in the early embryo

Centrosomes divide and In the P1 cell, the The spindle of the P1 cell The P1 cell, but not the AB
migrate so that in the AB cell centrosome–nucleus complex is aligned along the same cell, divides in the same
Division of zygote
the new spindle is at 90o subsequently undergoes a axis as in the zygote plane as the first cell division
to the previous one rotation of the zygote

Anterior Posterior
AB P1 AB P1 AB P1 AB P1

nucleus

Fig. 8.6 Different planes of cleavage

in different cells are determined by the under the ultimate control of cytoplasmic factors, such as the PAR proteins, that are
behavior of the centrosomes. The first differentially distributed at the first cleavage and have effects on spindle positioning
cleavage of the nematode zygote divides it (see Section 6.1).
into an anterior AB cell and a posterior P1 In plants, the planes of successive cell divisions, along with the subsequent direc-
cell. Before the next division, the duplicated tion of cell elongation, are the main factors determining the forms of stems, roots,
centrosomes of the AB cell move apart in leaves, and flowers (see Chapter 7). When a plant cell divides, it is not pinched in half
such a way that the plane of cleavage will by a contractile actomyosin ring as in an animal cell. Instead, a new cell membrane
be at right angles to the first division. The
and cell wall is formed in situ midway between the two poles of the mitotic spindle.
duplicated centrosomes in the P1 cell initially
In plants, the plane of division seems to be determined before mitosis begins, as a cir-
move into a similar alignment, but before
mitosis, the nucleus and its associated
cumferential band of microtubules and actin filaments transiently forms in the cortex
centrosomes rotate 90° so that cleavage is in at the position of the future cell division.
the same plane as the first cleavage. A common form of early animal embryo is the hollow spherical blastula, composed
After Strome, S.: 1993. of an epithelial sheet enclosing a fluid-filled interior. The development of such a
structure from a fertilized egg depends both on particular patterns of cleavage and
on changes in the way cells pack together, as shown in schematic form in Fig. 8.7.
If the plane of cell division is always at right angles to the surface (radial), the cells
will remain in a single layer. As cleavage proceeds individual cells get smaller, but
the surface area of the cell sheet increases and a space—the blastocoel—forms in the
interior, which increases in volume at each round of cell division (see Fig. 8.7). This
is essentially how the sea urchin blastula is formed.

8.4 Cells become polarized in the sea-urchin blastula and the mouse morula

As a result of the pattern of cleavage, the surface of the sea-urchin egg, which is cov-
ered in microvilli, becomes the outer surface of the blastula. A layer of extracellular
Fig. 8.7 Cell packing can determine the
matrix called the hyaline layer, which contains the proteins hyalin and echinonectin, is
volume of a blastula. Left panel: when
secreted on to the external surface and the cells adhere to this sheet. Adhesive cell–cell
adjacent cells in a spherical sheet, such as
junctions develop between adjacent blastomeres, forming them into an epithelium, and
those of a blastula, make contact with each
other over large areas of cell surface, the
overall volume of the cell sheet is relatively Same number and total cell volume, Radial cleavage produces more cells,
small as there is little space at its center. Cleavage different cell shape, larger with the same total cell volume but
Middle panel: with a decrease in the area of blastocoel volume larger blastocoel volume
intercellular contact, the size of the internal
space (the blastocoel), and thus the overall
volume of the blastula, is greatly increased
without any corresponding increase in cell
number or total cell volume. Right panel: if blastocoel
the number of cells is increased by radial
cleavage, and the packing remains the same,
the blastocoel volume will increase further,
again without any increase in total cell
volume.
 Cleavage and formation of the blastula 299

the cell contents become polarized in an apical–basal direction, with the Golgi appara-
tus oriented toward the apical (outer) surface. The cells are held together laterally by
cadherin-based adherens junctions and by septate junctions, a type of cell junction
found in invertebrates, but not chordates, and which is thought to provide the same
type of permeability barrier as the vertebrate tight junction—preventing water, small
molecules, and ions from permeating between the epithelial cells. On the inner (basal)
surface of the epithelium, a basal lamina (an organized layer of extracellular matrix) is
laid down. As the blastula develops, cilia are produced on the outer surface (Fig. 8.8).
In the mouse embryo, the earliest sign of structural differentiation is at the eight-cell
stage, when the morula undergoes a process known as compaction (Fig. 8.9). Until
then, the blastomeres form a loosely packed ball, with each cell surface uniformly cov-
ered by microvilli. At compaction, blastomeres flatten against each other, maximizing
cell–cell contact, and cells become polarized, with microvilli becoming confined to
the apical surfaces (Fig. 8.10). Radial cleavages produce two polarized cells, but some Fig. 8.8 Blastula of a sea-urchin embryo,
cleavages now occur tangentially (parallel to the surface), each producing one outer Lytechinus pictus. A single layer of cells
polarized cell and one inner nonpolarized cell. As development proceeds to form the surrounds the hollow blastocoel. Scale
blastocyst, the nonpolarized cells become the inner cell mass, which will give rise to bar = 10 µm.
Photograph from Raff, E.C. et al.: 2006.

Eight-cell stage Compaction

Fig. 8.9 Compaction of the mouse

embryo. At the eight-cell stage the cells
have relatively smooth surfaces and microvilli
are distributed uniformly over the surface.
At compaction, microvilli are confined to the
outer surface and cells increase their area of
contact with one another. Scale bar = 10 µm.
Photographs courtesy of T. Bloom, from
Bloom, T.L: 1989.

Two cells Eight cells

(plus polar bodies) Four cells Eight cells (compacted)

Cell division Cell division Compaction Fig. 8.10 Polarization of cells during
and polarization polarized cell
cleavage of the mouse embryo. At the
eight-cell stage, compaction takes place, with
Radial cleavage Tangential cleavage the cells forming extensive contacts with
each other (top panel). The cells also become
polarized. For example, microvilli, which were
initially uniformly distributed over the cell
Polarized Polarized surface, become confined to the outward-
facing cell surface. Future cleavages can thus
plane of cleavage divide the cell either into two polarized cells
(by a radial cleavage, bottom left panel) or
into a polarized and a nonpolarized cell (by a
Polarized Nonpolarized
microvilli tangential cleavage, bottom right panel). The
tangential cleavages give rise to the inner
cell mass.
300 Chapter 8 Morphogenesis: change in form in the early embryo

the embryo proper, whereas the outer cells develop tight junctions and form the tro-
phectoderm, the outer epithelium of the blastocyst (see Fig. 3.22, first panel).
The formation of cadherin-based adherens junctions (see Box 8B, p. 293) between
blastomeres is a major feature of compaction. At the two-cell and four-cell stages,
­E-cadherin is uniformly distributed over blastomere surfaces and contact between
cells is not extensive. At the eight-cell stage, however, E-cadherin becomes restricted
to regions of intercellular contact, where it now acts for the first time as an ­adhesion
molecule, forming adherens junctions. The change in the adhesive properties of
­E-cadherin results from its becoming linked to cytoskeletal elements in the cortex
via β-catenin. Adheren-junction formation requires the action of protein kinase C-α .
If this kinase is activated before the eight-cell stage, compaction occurs prematurely,
whereas if it is inhibited, compaction fails to occur. It is likely that phosphorylation of
junctional components such as β-catenin by the kinase is required for assembly of the
adhesive form of the cadherin–catenin complex.
Compaction is associated with a radical remodeling of the cell cortex, which is
presumably involved in the relocalization of E-cadherin and the redistribution of
microvilli. The cytoskeletal proteins actin, spectrin, and myosin are cleared from
the region of intercellular contact and become concentrated in a band around the api-
cal region of the polarized cell. It has been proposed that both the cell flattening and
the redistribution of the cortical elements may be brought about by the contraction of
actin filaments, which draws the cortical elements to the apical pole.

8.5 Fluid accumulation as a result of tight-junction formation and ion

transport forms the blastocoel of the mammalian blastocyst

Accumulation of fluid in the interior space, or blastocoel, of a blastocyst or blastula

exerts an outward pressure on the blastula wall, and this hydrostatic pressure is one of
the forces that forms and maintains the spherical shape. The blastocoel is formed in the
preimplantation mammalian embryo by a cavitation process in which fluid accumulates
in the interior of the developing blastocyst. The inflow of water is linked to the active
transport of sodium ions and other electrolytes into the extracellular spaces (Fig. 8.11).
At the eight-cell stage, tight junctions start to be formed between the polarized
cells of the outer layer of the mammalian morula. These cell junctions are typical of
epithelia and form a seal that prevents water molecules, ions, and small molecules
permeating in between the cells (Fig. 8.12). By the 32-cell stage this barrier is fully
Fig. 8.11 The blastocoel of the
mammalian blastocyst is formed by the
inflow of water creating an internal fluid- Tight junction between mouse blastocyst cells Mouse blastocyst with tight junctions stained green
filled cavity. Left panel: active transport
+
of sodium ions (Na+) across the basolateral H2O Na
membranes of trophectoderm cells into the Isotonic
blastocoel cavity of the mouse blastocyst
results in an increase in ion concentration in
the extracellular fluid, which causes water tight junction
to flow in from the surrounding isotonic
solution. The blastocoel is sealed off from the
outside medium by tight junctions between
Na+ Cytoplasm
the trophectoderm cells at their outer edge,
and so the inflow of water creates creates
hydrostatic pressure that enlarges the
blastocoel
blastocoel. Right panel: a surface view of a cavity
mouse blastocyst shows the tight junctions
between epithelial cells.
Pumped Na+ movement Passive H2O movement
Photograph from Eckert, J.J., Fleming, T.P.:
2008
 Cleavage and formation of the blastula 301

formed. At the same time, the Na+/K+-ATPase sodium pump, and other membrane
interacting
transport proteins, become active in the basolateral membranes of the outer cells, plasma membranes
transporting sodium and other electrolytes into the extracellular space. The cellular
sodium is replaced by sodium from the fluid surrounding the blastocyst. As the ion
concentration in the blastocoel fluid increases, water is drawn into the blastocoel
through aquaporin water channels in the cell membranes, and the increase in hydro-
static pressure caused by the accumulating fluid stretches the surrounding epithelial intercellular
space
layer. A similar mechanism based on ion transport seems to operate in drawing fluid
into the blastocoel in the Xenopus blastula. In the case of Xenopus, a small blastocoel
can be observed by electron microscopy as early as the two-cell stage.
sealing strands
of tight-junction
proteins
8.6 Internal cavities can be created by cell death

Hollow structures can be created in several different ways during early embryonic
development. In the case of the mammalian neural tube, for example, the tubular
structure results from the folding of an epithelial sheet, as we shall see later in this cytoplasmic
chapter. We have just seen how the blastocoel is formed by a cavitation process in half of lipid bilayer

which water inflow enlarges an interior space. Another example of cavity formation,
this time produced by cell death, occurs during the development of the mouse epi- Fig. 8.12 The structure of the tight
thelial epiblast. junction. The apposing membranes of
The epiblast, which gives rise to the mouse embryo, is derived from the inner adjacent epithelial cells are held closely
cell mass (see Section 3.5). Initially, the epiblast is a solid cylindrical mass of cells, together by the tight junction, which is
but it develops into an epithelial layer enclosing a fluid-filled cavity. Formation of located towards the apex of the cell and
this cavity results from programmed cell death—apoptosis—of the cells in the center forms a tight seal between the extracellular
of the epiblast (Fig. 8.13). It is likely that the surrounding cell layer—the visceral and intracellular milieu. The membranes are
endoderm—sends a death signal to the epiblast, and that only the outer epiblast cells held together by horizontal rows (sealing
strands) of membrane proteins (occludins and
survive, as a result of their contact with the basement membrane. These survival sig-
claudins), which bind tightly to each other.
nals are delivered by integrins in the cell membranes that interact with components
Adapted from Alberts, B., et al.: 2008.
of the basement membrane. The phenomenon of programmed cell death and its role
in development is discussed further in later chapters.

5-day mouse embryo: 5.5-day embryo:

Visceral endoderm sends death Cells not in contact with 6-day embryo:
signal. Cells in contact with basement membrane Completed cavity formation
basement membrane survive undergo apoptosis

extra-embryonic
ectoderm

epiblast
basement
membrane

Fig. 8.13 Cavity formation in the epiblast

of the mouse embryo. Left panel: the cells
of the inner cell mass proliferate and form
a solid epiblast (blue), which is surrounded
by the visceral endoderm (yellow). Center
and right panels: programmed cell death—
apoptosis—in the epiblast results in the
blastocoel visceral formation of a cavity (white). As shown at
endoderm cell death
the bottom of the first panel, the signal for
cell death die may come from the visceral
endoderm (first panel). Only the epiblast cells
directly attached to the basement membrane
receive a rescue signal and survive (second
and third panels).
death signal rescue signal
After Coucouvanis, E., et al.: 1995.
302 Chapter 8 Morphogenesis: change in form in the early embryo

Summary

In many animals, the fertilized egg undergoes a cleavage stage that divides the egg
into a number of small cells (blastomeres) and eventually gives rise to a hollow blastula.
Various patterns of cleavage are found in different animal groups. In some animals, such
as nematodes and molluscs, the plane of cleavage is of importance in determining the
position of particular blastomeres in the embryo, and the distribution of cytoplasmic
determinants. The plane of cleavage in animal cells is determined by the orientation of
the mitotic spindle, which is in turn determined by the final positioning of the asters.
In plant cells, the plane of cell division seems to be determined before mitosis begins.
At the end of the cleavage period, the blastula essentially consists of a polarized epi-
thelium surrounding a fluid-filled blastocoel. Fluid accumulation in the blastocoel may
be partly due to active ion transport into the blastocoel drawing water in after it by
osmosis. Cavity formation in the early mouse epiblast is caused by cell death.

Summary: blastula formation

position of asters in dividing cells determines plane of cleavage

formation of epithelial sheet

Mouse Sea urchin

tangential cleavages in epithelium radial cleavages only

give rise to prospective inner cell
mass and outer trophectoderm

sodium ions pumped into blastocoel, inflation of internal volume by

inflation of blastocyst by osmosis radial cell division

blastocyst comprises a single layer blastula comprises a single-layer

of trophectoderm cells enclosing epithelium enclosing the fluid-filled
inner cell mass and fluid-filled blastocoel
blastocoel

Gastrulation movements

Movements of individual cells and cell sheets at gastrulation bring most of the tissues
of the blastula (or its equivalent stage) into their appropriate position in relation to the
body plan. The necessity for gastrulation is clear from fate maps of the blastula in, for
example, sea urchins and amphibians, where the presumptive endoderm, mesoderm,
and ectoderm can be identified as adjacent regions in an epithelial sheet, and the endo-
derm and mesoderm must therefore move from the surface to inside the early embryo
(see Chapters 4–6). After gastrulation, these tissues become completely rearranged in
relation to each other: for example, the endoderm develops into an internal gut and is
separated from the outer ectoderm by a layer of mesoderm. Gastrulation thus involves
dramatic changes in the overall structure of the embryo, converting it into a complex
three-dimensional structure. During gastrulation, a program of cell activity involving
changes in cell shape and adhesiveness remodels the embryo, so that the endoderm
and mesoderm move inside and only ectoderm remains on the outside. The primary
 Gastrulation movements 303

force for gastrulation is provided by changes in cell shape, as well as the ability of
Animal
individual cells to migrate (see Box 8A, p. 291). In some embryos, all this remodeling basal blastocoel
lamina
occurs with little or no accompanying increase in cell number or total cell mass.
In this section, we first consider gastrulation mechanisms in sea urchins and insects,
in which it is a relatively simple process. Xenopus serves as our model for the more
complex gastrulation process in vertebrates.
ectoderm
endoderm
8.7 Gastrulation in the sea urchin involves cell migration and invagination
mesoderm
Just before gastrulation begins, the late sea-urchin blastula consists of a single-layered
ciliated epithelium surrounding a central fluid-filled blastocoel. The future mesoderm
occupies the most vegetal region, with the future endoderm adjacent to it (Fig. 8.14).
The rest of the embryo gives rise to ectoderm. The epithelium is polarized in an hyaline layer Vegetal cilia
­apical–basal direction: on its apical surface is the hyaline layer, while the basal sur-
face facing the blastocoel is lined with a basal lamina. The epithelial cells are attached Fig. 8.14 The sea-urchin blastula before
laterally to each other by septate junctions and adherens junctions. gastrulation. The prospective endoderm
Gastrulation begins with an epithelial-to-mesenchymal transition, with the most (yellow) and mesoderm (orange) are at the
vegetal mesodermal cells becoming motile and mesenchymal in form. They become vegetal pole. The rest of the blastula is
prospective ectoderm (gray). There is an
detached from each other and from the hyaline layer, and migrate into the blastocoel
extracellular hyaline layer and a basal lamina
as single cells (Fig. 8.15) that have lost both their epithelial polarity and their cuboid
lines the blastocoel. The blastula surface is
shape. They are now known as primary mesenchyme (see Section 6.7). The transition
covered with cilia.
to primary mesenchyme cells and entry into the blastocoel is foreshadowed by intense
pulsatory activity on the inner face of these cells, while still part of the epithelium,
and on occasion a small transitory infolding or invagination is seen in the surface of Fig. 8.15 Sea-urchin gastrulation. Cells
the blastula before migration begins properly. Cell internalization requires loss of cell– of the vegetal mesoderm undergo the
cell adhesion and is associated with repression of cadherin expression, the removal of transition to primary mesenchyme cells
cadherin from the cell surface by endocytosis, and the loss of α- and β-catenins. and enter the blastocoel at the vegetal
The epithelial-to-mesenchyme transition is regulated in part by the sea-urchin ver- pole. This is followed by an invagination of
sion of the gene snail, which, like many other developmental genes, was first identi- the endoderm, which extends inside the
fied in Drosophila (see Section 2.19). Experiments using antisense morpholinos (see blastocoel toward the animal pole, forming a
Box 6A, p. 220) to knock down snail expression in early sea-urchin embryos show clear archenteron. Filopodia extending from
the secondary mesenchyme cells at the tip
that it is required for the repression of cadherin gene expression and for cadherin
of the invaginating endoderm contact the
endocytosis in the epithelial-to-mesenchymal transition. The role of the Snail family
blastocoel wall and draw the invagination to
of transcription factors in this type of cellular change is highly conserved across the the site of the future mouth, with which it
animal kingdom, as we shall see in this chapter. The actions of Snail and a related fuses, forming the gut. Scale bar = 50 µm.
Photographs courtesy of J. Morrill.

Blastula secondary mesenchyme secondary mesenchyme Gastrula

gut
filopodia

mouth

endoderm
mesoderm primary mesenchyme invagination of endoderm archenteron anus
304 Chapter 8 Morphogenesis: change in form in the early embryo

Vegetal view
Ventral
ventral extension

vegetal ring

filopodia Animal pole

Ventral

primary mesenchyme cells ring of primary mesenchyme ventral extension of primary mesenchyme Vegetal pole

Fig. 8.16 Migration of primary

mesenchyme in early sea-urchin t­ranscription factor, Slug, are also of medical interest. A similar epithelial-­
development. The primary mesenchyme to-mesenchymal transition takes place in cancer cells, enabling them to undergo
cells enter the blastocoel at the vegetal metastasis and migrate from the original site of the tumor to set up secondary cancers
pole and migrate over the blastocoel wall by elsewhere in the body. In these metastatic cells, the normal low expression of snail
filopodial extension and contraction. Within and slug has been increased.
a few hours they take up a well-defined After entry, the primary mesenchymal cells migrate within the blastocoel to form
ring-like pattern in the vegetal region with a characteristic pattern on the inner surface of the blastocoel wall. They first become
extensions along the ventral side.
arranged in a ring around the gut in the vegetal region at the ectoderm–endoderm
border. Some then migrate to form two extensions towards the animal pole on the
ventral (oral) side (Fig. 8.16). The migration path of individual cells varies consider-
ably from embryo to embryo, but their final pattern of distribution is fairly constant.
The primary mesenchyme cells later lay down the skeletal rods of the sea urchin
endoskeleton by secretion of matrix proteins (see Section 6.12).
Primary mesenchymal cells move over the inner surface of the blastocoel wall by
means of fine filopodia, which can be up to 40 µm long and can extend in several
directions. Filopodia contain cross-linked bundles of actin filaments and extend for-
ward by the rapid assembly of actin filaments that push out the leading edge of the
blastocoel wall filopodium. At any one time, each cell has on average six filopodia, most of which are
branched. When filopodia make contact with, and adhere to, the basal lamina lining
the blastocoel wall, they retract, drawing the cell body toward the point of contact.
As each cell extends several filopodia (Fig. 8.17), some or all of which may retract
on contact with the wall, there seems to be competition between the filopodia, the
cell being drawn towards that region of the wall where the filopodia make the most
stable contact. The movement of the primary mesenchymal cells therefore resembles
a random search for the most stable attachment. As the cells migrate, their filopodia
filopodium fuse with each other, forming cable-like extensions.
If primary mesenchymal cells are artificially introduced by injection at the animal
pole, they move in a directed manner to their normal positions in the vegetal region.
Fig. 8.17 Filopodia of sea-urchin
This suggests that guidance cues, possibly graded, are distributed in the basal lamina.
mesenchyme cells. The scanning electron
micrograph shows a group of primary Even cells that have already migrated will migrate again to form a similar pattern
mesenchyme cells, some of which have fused when introduced into a younger embryo. The stability of contacts between filopodia
together, moving over the blastocoel wall by and the blastocoel wall is one factor determining the pattern of cell migration. Analy-
means of their numerous filopodia, which can sis of movies of migrating cells suggests that the most stable contacts are made in
extend and contract. the regions where the cells finally accumulate, namely the vegetal ring and the two
Photograph courtesy of J. Morrill, from Morrill, ventro-lateral clusters (see Fig. 8.16). The signaling molecules fibroblast growth fac-
J.B., et al.: 1985. tor A (FGFA) and vascular epithelial growth factor (VEGF) have been implicated as
 Gastrulation movements 305

possible guidance cues. In the blastula, FGFA is expressed in ectodermal regions to

which the primary mesenchyme will migrate and the FGF receptor is expressed on
the primary mesenchyme cells. If signaling by FGFA is inhibited, then so is cell migra-
tion and skeleton formation. Studies in vertebrates have shown that FGFs can attract apical face
migrating mesenchymal cells, and it is likely that they can also do so in the sea urchin.
VEGF is expressed by the prospective ectoderm cells at the most ventral sites to which
the mesenchyme cells migrate, while the VEGF receptor is again expressed exclusively
in the prospective primary mesenchyme. As with FGF, if VEGF or its receptor are lack-
ing, the cells do not reach their correct positions and the skeleton does not form.
The entry of the primary mesenchyme is followed by the invagination and extension
localized contraction
of the endoderm to form the embryonic gut (the archenteron). The endoderm invagi-
nates as a continuous sheet of cells (see Fig. 8.15). Formation of the gut occurs in two
phases. During the initial phase, the endoderm invaginates to form a short, squat cylinder
extending up to halfway across the blastocoel. There is then a short pause, after which
extension continues. In this second phase, the cells at the tip of the invaginating gut,
which will later detach as the secondary mesenchyme, form long filopodia, which make
contact with the blastocoel wall. Filopodial extension and contraction pull the elongat-
ing gut across the blastocoel until it eventually comes in contact with and fuses with the
mouth region, which forms a small invagination on the ventral side of the embryo (see Fig. 8.18 Change in cell shape in a small
Fig. 8.15). During this process the number of invaginating cells doubles, and this is in number of cells can cause invagination of
endodermal cells. Bundles of contractile
part due to cells around the site of invagination contributing to the hindgut.
filaments composed of actin and myosin
How is invagination of the endoderm initiated? The simplest explanation is that a
contract at the outer edge of a small number
change in shape of the endodermal cells causes, and initially maintains, the change of adjaent cells, making them wedge-shaped.
in curvature of the cell sheet (Fig. 8.18). At the site of invagination, the cuboid cells As long as the cells remain mechanically
adopt a more elongated, wedge-shaped form, which is narrower at the outer (apical) linked to each other and to adjacent cells
face. This change in cell shape is the result of constriction of the cell at its apex by in the sheet, this local change in cell shape
contraction of actomyosin in the cortex (see Box 8A, p. 291). Such a shape change draws the sheet of cells inward at that point.
would be sufficient to pull the outer surface of the cell sheet inward and to maintain
invagination, as shown by a computer simulation in which apical constriction, mod-
eled to spread over the vegetal pole, results in an invagination (Fig. 8.19).
The primary invagination only takes the gut about a third of the way to its final
destination. The second phase of gastrulation involves two different mechanisms:
contraction of filopodia in contact with the future mouth region, as described above,
and a filopodia-independent process. Treatments that interfere with filopodial attach-
ment to the blastocoel wall result in failure of the gut to elongate completely, but it
still reaches about two-thirds of its complete length. Filopodia-independent extension
is due to active rearrangement of cells within the endodermal sheet. If a sector of
cells in the vegetal mesoderm is labeled with a fluorescent dye before gastrulation,
it is seen to become a long, narrow strip as the gut extends (Fig. 8.20). The sheet of
cells has elongated in one direction while becoming narrower in width. We will look
Fig. 8.19 Computer simulation of
at this type of cell rearrangement—known as convergent extension—in more detail
the role of apical constriction in
later in this chapter, in relation to similar phenomena in amphibian and zebrafish
invagination. Computer simulation of the
gastrulation (see Box 8C, p. 310).
spreading of apical constriction over a region
What guides the tip of the gut to the future mouth region? As the long filopodia at of a cell sheet shows how this can lead to an
the tip of the gut initially explore the blastocoel wall, they make more stable contacts invagination.
Illustration after Odell, G.M., et al: 1981.

cells undergoing apical constriction

306 Chapter 8 Morphogenesis: change in form in the early embryo

Blastula Gastrula

Fig. 8.20 Gut extension during sea-urchin

gastrulation. Labeling of cells in the vegetal
region of the blastula (green) shows that gut
extension involves cell rearrangement within
the endoderm, causing the labeled cells to
become distributed in a long narrow strip.

at the animal pole and then in the region where the future mouth will form. Filopodia
making contact there remain attached for 20–50 times longer than when attached
to other sites on the blastocoel wall. Gastrulation in the sea urchin clearly shows
how changes in cell shape, changes in cell adhesiveness, and cell migration all work
together to cause a major change in embryonic form.

8.8 Mesoderm invagination in Drosophila is due to changes in cell shape that

are controlled by genes that pattern the dorso-ventral axis

At the time gastrulation begins, the Drosophila embryo consists of a blastoderm

of about 6000 cells, which forms a superficial single-celled layer (see Section 2.1).
­Gastrulation begins with the invagination of a longitudinal strip of future mesodermal
cells (8–10 cells wide) on the ventral side of the embryo, to form a ventral furrow
and then a tube inside the body. The tube breaks up into individual cells, which
spread out to form a single layer of mesoderm on the interior face of the ectoderm
(Fig. 8.21). The gut develops slightly later, by invaginations of prospective endoderm
near the anterior and posterior ends of the embryo (see Fig. 2.3), which we shall not
consider here.
Invagination of the mesoderm occurs in two phases. First, the central strip of cells
develops flattened and smaller apical surfaces, and their nuclei move away from the
apical surface. These changes in cell shape result in formation of the furrow and its
folding into the interior of the embryo to form a tube, the central cells forming the
tube proper while the peripheral cells form a ‘stem’ (see Fig. 8.21, center). This first
phase takes about 30 minutes. During the second phase, which takes about an hour,
the tube dissociates into individual cells, which proliferate and spread out laterally.
The entry of cells into cell division is delayed during invagination and this delay is
essential for the cell-shape changes to occur.
Mutant Drosophila embryos that have been dorsalized or ventralized (see Sec-
tion 2.12) show that this behavior of the mesodermal cells is autonomous, as in the
Fig. 8.21 Drosophila gastrulation.
Transverse sections of a Drosophila embryo sea urchin (see Section 8.7), and is not affected by adjacent tissues. In dorsalized
undergoing gastrulation. The mesodermal embryos, which have no mesoderm, no cells undergo nuclear migration or apical
cells (stained brown) are present in a contraction. In ventralized embryos, in which most of the cells are mesoderm, these
longitudinal strip in the epithelium on the changes occur throughout the whole of the dorso-ventral axis.
ventral side of the embryo. A change in their
shape from cuboid to wedge-shaped, causes
an invagination, which develops into a ventral
furrow (left panel). Further apical contraction
of the cells results in the mesoderm forming
a tube on the inside of the embryo (center
panel). The mesodermal cells then start to
migrate individually to different sites (right
panel). Scale bars = 50 µm.
Photographs courtesy of M. Leptin, from
Leptin, M., et al.: 1992.
 Gastrulation movements 307

Gastrulation in Drosophila provided the first possibility of linking the action of

patterning genes directly to morphogenesis, in this case by their involvement in caus-
ing changes in cell shape and cell adhesiveness. Mesoderm invagination is affected
by mutations in the genes twist and snail, which are expressed in the prospective
mesoderm before gastrulation as part of the patterning of the dorso-ventral axis (see
Section 2.19). By visualizing the outlines of ventral cells by labeling cell membranes
with green fluorescent protein, apical constriction of the prospective mesoderm can
be seen to occur as a result of successive pulses of contraction. Labeling of myosin
and actin showed that the constriction was due to pulsed contractions of an actomyo-
sin network that formed under the apical cell surface. No contractile network formed
at all in loss-of-function snail mutants, whereas in twist mutants, pulsed contractions
still occurred but they did not produced apical constriction. Further investigations con-
firmed that while Snail was required to initiate the whole process, Twist was required
to stabilize the state of contraction between pulses. In the absence of Twist function,
the cell surface relaxes back to its original area after each contraction.
Twist and Snail also control the next stage of mesoderm internalization. When the
central part of the mesoderm is fully internalized, its cells make contact with the
ectoderm. Adherens junctions that had been established in the blastoderm are now
lost—one of the results of Snail activity (see Section 8.7)—and the mesoderm under-
goes an epithelial-to-mesenchymal transition, dissociating into individual cells that
spread out to cover the ectoderm as a single-cell layer. Remarkably like the situation
in the sea urchin, FGF signals from the ectoderm acting on an FGF receptor on the
mesodermal cells are needed for the invaginating cells to spread. Failure to activate
the FGF signaling pathway results in the mesodermal cells remaining near the site of
invagination. Expression of the FGF receptor in the mesoderm is induced by Twist and
Snail, while Snail also represses expression of FGF in the mesoderm, thus restricting
FGF to the ectoderm.
Mesodermal cell spreading involves the cells switching from expression of ectoder-
mal E-cadherin to N-cadherin, so that they no longer adhere to ectodermal cells. The
change from E- to N-cadherin is also under the control of Snail and Twist. The actions
of Snail suppress the production of E-cadherin in the mesoderm, while expression of
Twist induces the production of N-cadherin.

8.9 Germ-band extension in Drosophila involves myosin-dependent

remodeling of cell junctions and cell intercalation

Another dramatic change in shape that occurs in the Drosophila embryo is the exten-
sion of the germ band (see Fig. 2.4), which leads to an almost doubling in the length
of the epithelial layer that forms the thorax and abdomen of the embryo. This exten-
sion is driven not by cell division or changes in cell shape, but by convergent exten-
sion of the ventral part of the epithelium. Adjacent cells intercalate between each
other laterally—towards the midline—so that the tissue narrows, causing it to extend
in the antero-posterior direction. This remodeling occurs over the whole of the ven-
tral epithelium and is due to regulated changes in the cadherin-containing adherens
­junctions that normally hold the cells tightly together.
At the beginning of germ-band extension, the epidermal cells are packed in a regu-
lar hexagonal pattern, with their boundaries either parallel to the dorso-ventral axis or
at 60° to it, as shown in Fig. 8.22. When extension starts, the adherens junctions on
the faces parallel to the dorso-ventral axis shrink and disappear and the cells become
diamond-shaped. New junctions parallel to the antero-posterior axis appear, making
the boundaries hexagonal again and intercalating the cells along the dorso-ventral
axis—which makes the tissue extend along the antero-posterior axis. This ability to
undergo convergent extension along the antero-posterior axis reflects a prior antero-
posterior patterning of cells in the epidermal sheet.
308 Chapter 8 Morphogenesis: change in form in the early embryo

A P

Fig. 8.22 Junction remodeling enables cells to undergo intercalation in Drosophila germ-
band extension. The disassembly and reassembly of intercellular contacts in the germ-band
epithelium drives the tissue to elongate along the antero-posterior axis and narrow along
the dorso-ventral axis. The heavy black outline indicates how the four cells in contact with
each other in the left-hand diagram have changed their intercellular contacts after junction
remodeling (right-hand diagram).
a
Adapted from Bertet, C., et al.: 2004.

The mechanism of intercalation in the germ band involves the regulated local-
ization and activity of myosin, which co-localizes with β-catenin–E-cadherin–actin
complexes at the adherens junctions (see Box 8B, p. 293) and is enriched in shrink-
ing junctions. The proposed mechanism is that regulated myosin contraction at the
junctions prevents E-cadherin holding the cells together, enabling new contacts to be
b
made that result in cell intercalation.

8.10 Dorsal closure in Drosophila and ventral closure in Caenorhabditis

elegans are brought about by the action of filopodia

At about 11 hours after the beginning of Drosophila embryogenesis, when gastrulation

is complete and the germ band has retracted again, the amnioserosa on the dorsal sur-
c
face of the embryo is not yet covered by the epidermis. The epidermis now moves to
close this gap, zipping-up the opening from both ends (Fig. 8.23). About 2 hours after
this process starts, the two epithelial fronts have fused to form a seam along the dorsal
midline, with the segment pattern exactly matched. The movement of the epidermis
is due to the action of filopodia and lamellipodia, which extend and retract from the
cells along the edges of the epithelial sheet. Filopodia are active along the whole epi-
thelial front, extending up to 10 µm from the edge, and their behavior is similar to that
seen in sea urchin gastrulation (see Section 8.7). Laser ablation of filopodial cells does
not prevent closure, however, which is brought about by actomyosin-based contrac-
tion at the leading edges of the epidermal sheet and of the amnioserosa. The small
GTPase Rac plays a key role in organizing actin filaments at the leading faces of the
Fig. 8.23 Dorsal closure in Drosophila. The cells, and is also involved through the Jun N-terminal kinase (JNK) intracellular sig-
closure of the epidermis along the dorsal
naling pathway in controlling changes in cell shape and fusion. Dorsal closure occurs
midline over a period of about 2 hours is
with remarkable accuracy, almost down to single-cell resolution. This appears to be
shown in the micrographs in panels a–c.
achieved by cells at each edge that express the same genes identifying and fusing
Starting at both ends of the opening, the
edges of the epidermis come together over
with each other. For example, the gene engrailed is expressed in a series of transverse
the amnioserosa (seen as the central oval stripes along the length of the embryo, corresponding to the posterior compartment
area in a and b) and fuse to form the midline of each segment (see Section 2.23). Cells expressing engrailed on one edge interact
seam, matching the segment pattern on only with cells expressing engrailed on the other edge, and cells expressing the gene
either side. The bottom panel shows filopodia patched, which form a different set of stripes, fuse only with patch-expressing cells on
acting as a zipper. the other edge, thus neatly matching up the stripes across the seam.
Photographs courtesy of A. Jacinto, from The C. elegans embryo undergoes a process similar to dorsal closure, but on the
Jacinto, A., et al.: 2002. ventral surface. At the end of gastrulation, the epidermis only covers the dorsal region
 Gastrulation movements 309

and the ventral region is bare. The epidermis then spreads around the embryo until
its edges finally meet along the ventral midline. Time-lapse studies show that the
initial ventral migration is led by four cells, two on each side, which extend filopo-
dia towards the ventral midline. Blocking filopodial activity by laser ablation or by
cytochalasin D, which blocks the formation of actin filaments, prevents ventral clo-
sure, showing that the filopodia provide the driving force. Closure probably occurs by
a zipper mechanism similar to that in the fly. Animal pole
marginal blastocoel
zone
8.11 Vertebrate gastrulation involves several different types of tissue
movement

Gastrulation in vertebrates involves a much more dramatic and complex rearrange-

ment of tissues than in sea urchins, because of the need to produce a more complex
body plan. In amphibians, fish, and birds, there is also the added complication of
the presence of large amounts of yolk. But the outcome is the same: the transfor-
mation of a two-dimensional sheet of cells into a three-dimensional embryo, with
ectoderm, mesoderm, and endoderm in the correct positions for further development
Vegetal pole yolk cells
of body structure. The main movements of gastrulation that we will discuss here are
­involution, which is the rolling-in of coherent sheets of endoderm and mesoderm at
the blastopore, as occurs in Xenopus (Fig. 8.24); epiboly, which is the spreading and
thinning of the ectodermal layers as the endoderm and mesoderm move inside and
which can be observed in gastrulating frogs and zebrafish; and convergent extension
(Box 8C, p. 310), which elongates the body axis. Gastrulation in mammals and birds
occurs in the primitive streak (see Chapters 3 and 4) and involves the convergence of
epiblast cells on the midline, the separation or delamination of cells individually from
the epiblast, and their internalization, or ingression, followed by migration internally
and convergent extension. We will begin with gastrulation in Xenopus.
In the late Xenopus blastula, the presumptive endoderm extends from the most veg-
bottle cells blastopore
etal region to cover the presumptive mesoderm. During gastrulation, the presumptive
endoderm moves inside through the blastopore to line the gut, while all the cells in
Epiboly archenteron
the equatorial band of mesoderm move inside to form a layer of mesoderm underly-
ing the ectoderm, extending antero-posteriorly along the dorsal midline of the embryo
(Fig. 8.25).
Gastrulation starts at a site on the dorsal side of the blastula, toward the vegetal
pole. The first visible sign is the formation of bottle-shaped cells (bottle cells, see
Fig. 8.25, second panel) by some of the presumptive mesodermal cells. The formation
of the bottle shape is due to apical constriction of the cell (see Box 8A, p. 291). As in
the sea urchin and Drosophila, this change in cell shape forms a groove in the blastula
surface. In Xenopus, this starts as a small groove—the blastopore—whose dorsal lip
Convergent Involution
is the site of the Spemann organizer. The layer of mesoderm and endoderm starts extension
to involute around the blastopore, rolling into the interior of the blastula against its
own under surface (see Fig. 8.25). As gastrulation proceeds, the region of involution Convergent
extension yolk plug

Fig. 8.24 Tissue movements during gastrulation of Xenopus. In the late blastula the future
mesoderm (red) is in the marginal zone, overlaid by presumptive endoderm (yellow). Ectoderm is
shown in blue. Gastrulation is initiated by the formation of bottle cells in the blastopore region,
which is followed by the involution of mesoderm over the dorsal lip of the blastopore. Marginal
zone endoderm and mesoderm move inside over the dorsal lip of the blastopore (events in this
region are shown in more detail in Fig. 8.25). The marginal zone endoderm, which was on the Epiboly
surface of the blastula, now lies ventral to the mesoderm and forms the roof of the archenteron
or future gut. At the same time, the ectoderm of the animal cap spreads downward. The
mesoderm converges and extends along the antero-posterior axis. The region of involution
blastocoel
spreads ventrally to include more endoderm, and forms a circle around a plug of yolky vegetal
cells. The ectoderm spreads by epiboly.
mesoderm ectoderm endoderm
After Balinsky, B.I.: 1975.
310 Chapter 8 Morphogenesis: change in form in the early embryo

Box 8C Convergent extension

Convergent extension plays a key

role in gastrulation and other mor- active lamellipodia inactive boundary
phogenetic processes. It is a mecha-
nism for elongating a sheet of cells

Antero-posterior axis
in one direction while narrowing its
width, and occurs by rearrangement
of cells within the sheet, rather than
by cell migration or cell division (see
top figure). It occurs, for example, in
the extension of the mesoderm that Medio-lateral
axis
elongates the antero-posterior axis in
amphibian embryos (see Fig. 8.27). For convergent extension to take place, the axes along which the cells will intercalate and extend
must already have been defined. The cells first become elongated in a direction at right angles to the antero-posterior axis—the medio-
lateral direction. They also become aligned parallel to one another in a direction perpendicular to the direction of tissue extension (see
top figure). Active movement is largely confined to each end of these elongated bipolar cells, which form active protrusions or filopodia,
enabling them to exert traction on neighboring cells on each side and on the underly-
Radial Medio-lateral
intercalation intercalation ing substratum and to shuffle in between each other—or intercalate—always along the
medio-lateral axis, with some cells moving medially and some laterally. This rearrange-
ment is known as medio-lateral interacalation. There are some small protrusions on
the anterior and posterior faces of the cells but these seem to hold apposed surfaces
together rather than being involved in cell movement. At a tissue boundary, one end
of the cell is captured and movement ceases, but cells can still exert traction on each
other to draw the boundaries closer together.
The boundary of a tissue undergoing convergent extension is maintained by the
behavior of cells at the boundary. When an active cell tip is at the boundary of the
tissue, its movement ceases and the cell becomes monopolar (see top figure), with
only the tip facing into the tissue still active. It is as if the cell tip that has reached the
boundary has become fixed there. Precisely what defines a boundary is not clear, but
boundaries between mesoderm, ectoderm, and endoderm seem to be specified before
gastrulation commences.
A mechanically similar process called Wnt
radial intercalation (see figure on left) Frizzled Vang/Strabismus
causes the thinning of a multi-cell layer
into a thinner sheet and its consequent extension around the edges, as seen dur-
ing epiboly of the ectoderm during frog and zebrafish gastrulation (see, for example,
adaptor
Fig. 8.24). Radial intercalation occurs in the multilayered ectoderm of the animal cap,
in which cells intercalate in a direction perpendicular to the surface, moving from one Dishevelled Daam1
layer into that immediately above. This leads to an increase in the surface area of the
cell sheet, as well as its thinning, and is in part the cause of epiboly. JNK
The molecular mechanisms controlling convergent extension in vertebrates are RhoA
related to those controlling planar cell polarity in Drosophila (see Box 2F, p. 78), and
involve signaling by Wnts through Frizzled via the non-canonical Wnt planar cell polarity
pathway (see figure on right). The much-simplified pathway shown here is a general
compilation from different organisms. Signaling by Wnt through its receptor Frizzled (Fz) Rok2
via Dishevelled activates RhoA and the Rho kinase Rok2, leading to changes in the cell’s
actin cytoskeleton. The transmembrane protein Van Gogh (Vang, also called Strabismus)
cytoskeletal
and other proteins identified from Drosophila are also components of the vertebrate reorganization
pathway, although how they all interact with each other is not yet clear. Vang is thought API
to interact with Dishevelled to stimulate the activation of another protein kinase, the Jun
N-terminal kinase (JNK), which can activate the transcription factor AP1.
 Gastrulation movements 311

Fig. 8.25 Tissue movement in the dorsal

region during formation of the blastopore
and gastrulation in Xenopus. First panel: late
blastula before gastrulation. In the marginal
zone, presumptive endoderm (yellow) overlies
the presumptive mesoderm (red). Second
panel: bottle cells at the site of the blastopore
archenteron undergo apical constriction and elongate,
migratory
cells causing the involution of surrounding cells
and the formation of a groove which defines
the dorsal lip of the blastopore. Third panel:
as gastrulation proceeds, the presumptive
endoderm and mesoderm involute and move
anteriorly under the ectoderm (blue). Fourth
panel: the archenteron—the future gut
marginal
zone Involution lined by endoderm—starts to form and the
mesoderm converges and extends along the
bottle cells blastopore antero-posterior axis. At the leading edge
of the mesoderm are migratory cells, which
give rise to the mesoderm of the head. The
mesoderm ectoderm endoderm ectoderm continues to move down by epiboly
to cover the whole embryo.
After Hardin, J.D., et al.: 1988.
spreads laterally and ventrally to involve the vegetal endoderm, and the blastopore
eventually forms a circle around a plug of yolky cells (see Fig. 8.24). In time, the blas-
topore contracts, forcing the yolky cells into the interior where they form the floor of Convergence and extension of equatorial
the gut. presumptive mesoderm during gastrulation
The first mesodermal cells to involute migrate as individual cells over the blastocoel Ventral Dorsal
roof to give rise to very anterior mesodermal structures in the head. Behind them is notochord
lateral D somite
mesoderm that enters, together with the overlying endoderm, as a single multilayered plate
B
sheet. For this cell sheet, going through the narrow blastopore is rather like going
through a funnel, and the cells become rearranged by convergent extension, which A
is a major feature of gastrulation. The mesoderm is initially in the form of an equato- C
midline
rial ring, but during gastrulation it converges and extends along the antero-posterior
axis (Fig. 8.26). Thus, cells that are initially on opposite sides of the embryo come to
lie next to each other (Fig. 8.27), whereas others become separated from each other Anterior extension
along the long axis of the embryo. The migrating mesoderm cells are polarized in B
the direction of the animal pole and their migration depends on interaction with the
fibronectin fibrils in the extracellular matrix lining the blastocoel roof. The assembly
of fibronectin fibrils in the matrix is due to tension at the surface of the blastocoel roof
epithelium, caused by cell–cell adhesion, being transmitted via integrins in the cell
membrane that interact with fibronectin.
In Xenopus, convergent extension occurs in both the mesoderm and endoderm as
A
they involute, and in the future neural ectoderm that gives rise to the spinal cord.
C
Together with the elongation of the notochord, all these processes elongate the embryo
convergence
in an antero-posterior direction. One can appreciate the dramatic nature of convergent
Posterior
extension by viewing the early gastrula from the blastopore. The future mesoderm
can be identified by the expression of the gene Brachyury, and can be seen as a nar-
Fig. 8.26 Convergent extension of the
row ring around the blastopore (Fig. 8.28). As the Brachyury-expressing ring of tissue
mesoderm. Initially the mesoderm (red,
enters the gastrula it converges into a narrow band along the dorsal midline of the
orange and yellow) is in an equatorial ring,
embryo, extending in the antero-posterior direction, and eventually only the prospec-
but during gastrulation it converges and
tive notochordal mesoderm along the midline itself continues to express Brachyury. extends along the antero-posterior axis.
As gastrulation proceeds, the mesoderm comes to lie immediately beneath the ecto- A–D are reference points, used to illustrate
derm, while the endoderm immediately below it lines the roof of the archenteron, the extent of movement during convergent
the future gut (see Fig. 8.25). Even though they are in contact, the mesoderm and extension. In the bottom panel, D is hidden
ectoderm do not adhere to each other and so can continue to move independently of behind C.
312 Chapter 8 Morphogenesis: change in form in the early embryo

Fig. 8.27 The rearrangement of mesoderm

Fate map of late blastula of Xenopus Sections of tailbud-stage Xenopus
and endoderm during gastrulation in
Xenopus. The presumptive mesoderm (red Animal Anterior
and orange) is present in a ring around E
brain
the blastula underlying the endoderm E A
(yellow), which is shown peeled back. During
notochord
gastrulation, both these tissues move inside
spinal cord
the embryo through the blastopore and C
F
D
completely change their shape, converging B
somites
and extending along the antero-posterior A
axis, so that points A and B move apart. Thus, C D
yolk mass
the two points C and D, originally on opposite dorsal lip
sides of the blastula, come to lie next to each Vegetal
other. In the neural ectoderm (blue) there is B F
neural ectoderm mesoderm Posterior
also convergent extension, as shown by the
endoderm mesoderm (notochord) Lateral view Dorsal view
movement of points E and F. Note that little
cell division or cell growth has occurred in
these stages, and all these changes have
each other. As the mesoderm and endoderm move inside, the ectoderm of the ani-
occurred by rearrangement of cells within the
mal cap region spreads down over the vegetal region by epiboly and will eventually
tissues.
cover the whole embryo. Epiboly involves both stretching of the cells themselves and
cell rearrangements, in which cells lower down in the ectodermal layers intercalate
between the cells immediately above them—a process called radial intercalation (see
Box 8C, p. 310). Together, these processes lead to thinning of the ectoderm layer and
an increase in its surface area.
Gastrulation in the zebrafish has both similarities and differences to that of Xeno­
pus. It begins by the blastoderm at the animal pole starting to spread out over the
yolk cell and down towards the vegetal pole. As in Xenopus, this epibolic spreading
is due to cells of the deep layer undergoing radial intercalation, leading to the layer
becoming thinner and thus increasing its surface area (Fig. 8.29, left panel). By the
time the blastoderm margin reaches about half-way down the yolk cell, the deep-layer
Fig. 8.28 Expression of Brachyury
cells have accumulated around the blastoderm edge to form a thickening known as
during Xenopus gastrulation illustrates
the embryonic germ ring, and have formed two cell layers—prospective ectoderm
convergent extension. Left: before
gastrulation, the expression of Brachyury overlying mesendoderm—under the superficial enveloping layer. The mesendoder-
(dark stain) marks the future mesoderm, mal cells now move internally at the edge of the germ ring and then move upwards
which is present as an equatorial ring when under the ectoderm towards the future anterior end of the embryo, forming a layer
viewed from the vegetal pole. Right: as called the hypoblast (Fig. 8.29, center panel). The ectoderm, now sometimes called
gastrulation proceeds, the mesoderm that the epiblast, continues to undergo epiboly and move downwards. The next step is
will form the notochord (delimited by blue convergent extension of the mesendodermal and ectodermal layers towards the dor-
lines) converges and extends along the sal midline and along the future antero-posterior axis, which elongates the embryo
midline. Scale bar = 1 mm. (Fig. 8.29, right panel).
Photograph courtesy of J. Smith, from Smith, In amphibian and zebrafish gastrulation, cells undergoing convergent extension
J.C., et al.: 1995. along the antero-posterior axis are polarized in the medio-lateral direction (see
Box 8C, p. 310). The contractile forces generated at the tips of the elongated cells
both move the cells towards the midline and extend the tissue. One can visualize
the process of convergent extension in terms of lines of cells pulling on one another
in a direction at right angles to the direction of extension. Because the cells at the
boundary are trapped at one end, this tensile force causes the tissue to narrow and
hence extend anteriorly. In fish embryos, convergent extension also involves an ini-
tial directed migration of individual mesodermal cells from lateral regions towards
the midline, a movement known as dorsal convergence, followed by medio-lateral
intercalation as mesodermal cells become incorporated into the future notochord at
the midline (Fig. 8.30).
The presumptive notochord mesoderm is among the first of the tissues to internal-
ize during gastrulation, and in amphibians it is the first of the dorsal structures to
differentiate. Initially it can be distinguished from the adjacent somitic mesoderm by
 Gastrulation movements 313

Fig. 8.29 Zebrafish gastrulation

movements. Left panel: the first sign of
gastrulation in the zebrafish embryo is
spreading of the blastoderm over the surface
of the yolk cell. This epiboly is caused by
a thinning of the deep layer by a radial
intercalation process. Center panel: when
the blastoderm has reached about half the
way down the yolk cell, separate layers
of prospective mesendoderm (brown) and
Radial intercalation Epiboly Convergent extension ectoderm (light blue) have formed and the
mesendoderm starts to internalize. Ectoderm
yolk does not internalize and will continue to move
syncytial down and spread out over the yolk cell. It will
layer
eventually cover the whole embryo. Right
panel: convergent extension (blue arrow) of
mesendoderm
the mesendoderm towards the dorsal midline
ectoderm results in elongation of the embryo.
After Montero, J.A., et al.: 2004.

enveloping Xenopus Zebrafish

layer
medio-lateral medio-lateral
intercalation intercalation

boundary
epiblast
deep layer hypoblast
directed
migration

the packing of its cells, and by the slight gap that forms the boundary between the two
tissues, possibly reflecting differences in cell adhesiveness. The notochord elongates
and develops into a stiff rod composed of a stack of thin flat cells shaped like pizza
slices.
After an initial elongation by medio-lateral intercalation and convergent extension,
the Xenopus notochord undergoes a further dramatic narrowing in width accompa-
nied by an increase in height (Fig. 8.31). The cells become elongated at right angles
to the main antero-posterior axis, foreshadowing the later pizza-slice arrangement of
cells (see Fig. 8.31, bottom). The cells intercalate between their neighbors again, thus
causing further convergent extension. A later stage in notochord development, which
stiffens and further elongates the rod of mesoderm, involves a process called directed
dilation and is discussed later in this chapter (Section 8.14).
The medio-lateral polarization of individual cells requires the vertebrate version Fig. 8.30 Convergent extension in
of the Wnt-mediated planar cell polarity pathway (see Box 8C, p. 310). When this Xenopus and zebrafish. The left panels show
pathway is rendered non-functional, Xenopus mesodermal cells fail to elongate or convergent extension of mesoderm in Xenopus
during the formation of the notochord, which
to develop polarized lamellipodia. But another polarization system is also at work
occurs by medio-lateral intercalation of cells in
in the convergent extension of the vertebrate body axis, which maintains the correct
a coherent tissue sheet. The right panels show
direction of extension in this population of continually rearranging and moving cells.
formation of the notochord by convergent
Before gastrulation, the organizer mesoderm is patterned along the antero-posterior extension in the zebrafish, which occurs
axis as reflected, for example, in a gradient of Chordin expression in the prospective by directed movement of loosely packed
notochord mesoderm—the chordamesoderm—which helps to pattern the overlying mesenchymal mesodermal cells from the
neural ectoderm by antagonizing the activity of BMPs (see Section 5.2). Antero-poste- lateral regions towards the midline, insertion
rior patterning seems to be a prerequisite for cell intercalation and convergent exten- into the future notochord and medio-lateral
sion. This can be demonstrated by disaggregating explants of ‘anterior’ and ‘posterior’ intercalation within the notochord boundary.
chordamesoderm from a Xenopus early gastrula and then mixing the cells together. After Wallingford, J.B., et al.: 2002.
314 Chapter 8 Morphogenesis: change in form in the early embryo

Notochord rearrangement and cell shape changes

Fig. 8.31 Changes in cell arrangement and

cell shape during notochord development
in Xenopus. During convergent extension of
the notochord, its height increases. At the
same time, the cells become elongated. The
bottom of the figure shows the eventual
arrangement of cells, and their pizza-slice
shape.
After Keller, R., et al.: 1989.

The chordamesoderm is situated at and just anterior to the upper lip of the blastopore.
The cells recognize each other by their current antero-posterior positional identity and
sort out to end up in the correct position. It was found that only combined explants
of anterior and posterior tissue could undergo intercalation and convergent exten-
sion; combinations of ‘all anterior’ or ‘all posterior’ cells did not. These experiments
suggest that the patterning along the antero-posterior axis is required to set up the
medio-lateral polarity in individual cells that enables them to intercalate. How cells
integrate the cues they receive from the antero-posterior polarization system and from
the Wnt-mediated planar polarity system has still to be worked out.
In the chick and the mouse (and in humans), gastrulation is indicated by forma-
tion of the primitive streak, as described in Chapter 3. As the streak moves forward,
epiblast cells separate from each other in the streak region and move into the interior
as individual cells. The epithelial epiblast—the future ectoderm—gives rise to both
the mesoderm and endoderm, as well as the ectoderm. Epiblast cells become speci-
fied as mesodermal and endodermal cells in the streak and move into the interior.
The mechanisms of streak formation, movement of epiblast cells, and delamina-
tion are not as clearly understood as are the involution movements in Xenopus.
One view is that convergence at the posterior end of the streak provides much of
the force for streak extension, and that the contraction and delamination of cells
in the streak draws lateral epiblast cells towards and into the streak. In the chick
Fig. 8.32 Polonaise movements in the epiblast, primitive-streak formation is preceded by extensive bilaterally symmetric
chick epiblast during primitive-streak movements of anterior epiblast cells outwards from the midline and backwards
formation. Left panel: cell movements in towards the posterior edge where the streak will begin (Fig. 8.32). These move-
the epiblast before laying start to bring cells
towards the site at which the primitive streak
will form. Second panel: around 6–7 hours ‘Polonaise’ movements in chick epiblast
after laying (stage 2 embryo) cells in the
epiblast undergo characteristic extensive Chick epiblast Chick epiblast Chick epiblast
before laying 6–7 hours after laying 12–13 hours after laying
‘polonaise’ movements outwards from the
midline and back towards the start of the
primitive streak; the region forming the
streak starts to converge and extend by
medio-lateral intercalation (see Fig. 8.30
and Box 8C, p. 310). Third panel: at around
12–13 hours after laying (stage 3) the streak
has extended to about half its full length;
as it extends, the epiblast cells in front of it
move forward. primitive
streak-forming region streak
Adapted from Voiculescu, O., et al.: 2007.
 Gastrulation movements 315

ments are known as ‘polonaise’ movements, as they resemble a formal eighteenth-

century dance of that name. High-resolution time-lapse videos of the chick epiblast
show that convergent extension is the most likely mechanism generating the cell
movements. The intercalation of cells close to the streak generates forces in the
epithelium that cause the movement of cells elsewhere in the epiblast. As with axis
elongation in amphibians and zebrafish, this example of convergent extension in the
chick involves the Wnt planar cell polarity pathway, and inhibition of this pathway
blocks streak formation.

Summary

At the end of cleavage the animal embryo is essentially a closed sheet of cells, which
is often in the form of a sphere enclosing a fluid-filled interior. Gastrulation converts
this sheet into a solid three-dimensional embryonic animal body. During gastrulation,
cells move from the exterior into the interior of the embryo, and the endodermal and
mesodermal germ layers take up their appropriate positions inside the embryo. Gas-
trulation results from a well-defined spatio-temporal pattern of changes in cell shape,
cell movements, and changes in cell adhesiveness, and the main forces underlying the
morphological changes are generated by changes in cell motility and localized cellular
contractions. In the sea urchin, gastrulation occurs in two phases. In the first, changes
in shape and adhesion of a small region of prospective mesoderm cells in the blastula
wall result in an epithelial-to-mesenchyme transition and migration of these cells into
the interior and along the inner face of the blastula wall, followed by invagination of
the adjacent part of the blastula epithelium, the prospective endoderm, to form the
gut. Filopodia on the migrating primary mesenchyme cells explore the environment,
drawing the cells to those regions of the wall where the filopodia make the most sta-
ble attachment; in this way guidance cues in the extracellular matrix lead the migrat-
ing cells to their correct locations. In the second phase, extension of the gut to reach
the prospective oral (mouth) region on the opposite side of the embryo occurs by cell
­rearrangement within the endoderm, which causes the tissue to narrow and lengthen
(convergent extension), and by traction of filopodia extending from the gut tip against
the blastocoel wall. Invagination of the mesoderm in Drosophila occurs by a similar
mechanism to that in sea urchin, whereas germ-band extension is due to ­myosin-driven
remodeling of cell junctions and intercalation of cells, leading to convergent extension.
Dorsal closure in Drosophila is driven by filopodial extension and retraction. Gastru-
lation in vertebrates involves more complex movements of cells and cell sheets and
also results in elongation of the embryo along the antero-posterior axis. In Xenopus it
involves three processes: involution, in which a double-layered sheet of endoderm and
mesoderm rolls into the interior over the lip of the blastopore; convergent extension
of endoderm and mesoderm in the antero-posterior direction to form the roof of the
gut, and the notochord and somitic mesoderm, respectively; and epiboly—the spread-
ing of the ectoderm from the animal cap region to cover the whole outer surface of
the embryo. Both convergent extension and epiboly are due to cell intercalation, the
cells becoming rearranged with respect to their neighbors. Convergent extension of
the antero-posterior axis at gastrulation involves both the vertebrate version of the
Wnt planar cell polarity signaling pathway and signals delivered by the pre-existing
antero-posterior patterning in the mesoderm. Gastrulation in the chick and the mouse
involves delamination of individual cells from the epiblast and their movement through
the primitive streak into the interior of the embryo, where they form mesoderm and
endoderm.
316 Chapter 8 Morphogenesis: change in form in the early embryo

Summary: gastrulation in sea urchin and Xenopus

Sea urchin Xenopus and zebrafish

formation of blastopore: mesodermal cells formation of blastopore: mesoderm and

migrate individually into the interior; endoderm move inside by
endoderm moves inside by involution or internalization
invagination of the epithelial sheet

convergent extension of mesoderm

convergent extension of endoderm along antero-posterior axis
and traction via filopodia
completes gut extension
extension of ectoderm over
whole surface by epiboly
neural plate

Neural tube formation

neural groove
Neurulation in vertebrates results in the formation of the neural tube—a tube of epi-
thelium derived from the dorsal ectoderm—which develops into the brain and spinal
cord (see Section 5.10). Following induction by the mesoderm during gastrulation, the
neural fold ectoderm that will give rise to the neural tube initially appears as a thickened plate of
tissue—the neural plate—in which the cells have become more columnar than the
cells of the adjacent ectoderm.
The vertebrate neural tube is formed by two different mechanisms in different
regions of the body. The anterior neural tube forms the brain and the central nervous
system in the trunk, and is essentially formed by the folding of the neural plate into
a tube. At neurulation, the edges of the neural plate become raised above the surface,
forming two parallel neural folds with a depression—the neural groove—between
them (Fig. 8.33). The neural folds eventually come together along the dorsal midline
of the embryo and fuse at their edges to form the neural tube, which then separates
from the adjacent ectoderm. The surface layer of ectoderm becomes epidermis. The
posterior neural tube in the lumbar and tail region, in contrast, develops from a solid
rod of cells which develops an interior cavity or lumen. There are, however, variations
in different vertebrates; in zebrafish and other fishes, for example, the whole of the
neural plate first forms a solid rod, which later becomes hollow.
neural crest cells ectoderm

8.12 Neural tube formation is driven by changes in cell shape

and convergent extension

neural tube The curvature of the neural plate and formation of the neural folds are associated
with changes in cell shape. The cells at the edge of the plate are constricted at the
apical surface, making them wedge-shaped (Fig. 8.34). This change in cell shape is, in
principle, sufficient to draw the edges of the neural plate up into folds. Like the axial
mesoderm, the neural tube undergoes convergent extension during its formation,
Fig. 8.33 Neural tube formation results from which both extends it in the antero-posterior direction, and brings the neural folds
formation and fusion of neural folds. close together, enabling them to fuse. The cells of the neural plate seem to become
Bending of the neural plate inwards at both
oriented for convergent extension in response to signals from midline structures such
sides creates the neural groove, with the
as the underlying notochord. Like convergent extension in the mesoderm, convergent
neural folds at the edges. The neural folds
come together and fuse to form a tube of
extension and closure of the neural tube require a functional Wnt planar polarity
epithelium—the neural tube. This detaches from pathway (see Box 8C, p. 310). If this pathway is defective, closure is incomplete, as
the rest of the ectoderm, which becomes the observed in mouse embryos lacking Dishevelled, Flamingo, or the vertebrate Vang
epidermis. Neural crest cells detach from the proteins, Vang-like 1 and 2, in which the neural tube remains open from the midbrain
dorsal neural tube and migrate away from it. to the lower spine, a condition known as craniorachischisis.
 Neural tube formation 317

Full-streak-stage chick embryo Streak has started regressing Somites start to appear

Anterior
head fold

Hensen’s neural
node furrow
somite

neural fold

primitive streak
Posterior

neural furrow surface ectoderm

neural fold neural fold

neural plate hinge
point

hinge
points

notochord

Fig. 8.34 Change in cell shape in the

neural plate during chick neurulation.
Neurulation does not occur along the whole of the neural plate at the same time, Top: surface views of the chick epiblast.
Bottom: cross-sections taken through the
but starts in the region of the midbrain, and proceeds anteriorly and posteriorly.
epiblast at the sites indicated by the dashed
Changes in shape of neural furrow cells are observed in chick embryos, but whether
lines in the top diagrams. Cells anterior to
they are a cause or a result of folding is not yet clear. The detailed cellular mecha-
Hensen’s node become elongated to form the
nism underlying the shape changes is not yet established, and in the chick there is neural plate. Cells in the center of the neural
some evidence that tissues lateral to the neural plate exert forces that help to throw plate become wedge-shaped, defining the
up the neural folds. Cells in the midline of the chick neural furrow, the so-called hinge point at which the neural plate bends.
hinge point, become wedge-shaped. Later, wedge-shaped cells are seen at additional Additional hinge points with wedge-shaped
hinge points on the sides of the furrow, where it curves round further to form a tube cells also form on the sides of the furrow.
(see Fig. 8.34). When the folds meet they zip together so closely that the lumen After Schoenwolf, G.C., et al.: 1990.
almost disappears; it only opens up again after the folds fuse. As in many other situ-
ations, apical constriction of cells in the neural furrow and hinge points is thought
to be due to the purse-string mechanism of actomyosin contraction illustrated in
Box 8A, p. 291.
The neural tube, which is initially part of the ectoderm, separates from the pre-
sumptive epidermis after its formation as a result of changes in cell adhesiveness. In
the chick, neural plate cells, like the rest of the ectoderm, initially have the adhesion
molecule L-CAM on their surface. However, as the neural folds develop, the neural
plate ectoderm begins to express both N-cadherin and N-CAM, whereas the adja-
cent ectoderm expresses E-cadherin. These differences may enable the neural tube to
separate from the surrounding ectoderm and allow it to sink beneath the surface, the
rest of the ectoderm reforming over it in a continuous layer. Changes in adhesiveness
could even provide part of the mechanism for neural tube formation, in a manner
similar to that described in Section 8.1 for the sorting out of cells.
318 Chapter 8 Morphogenesis: change in form in the early embryo

Summary

In vertebrates, induction of neural tissue by the mesoderm during gastrulation is fol-

lowed by neurulation—the formation of the neural tube, which will eventually form the
brain and spinal cord. Neurulation in mammals, birds, and amphibians involves both the
folding of the neural plate into a tube (in the head and trunk), and the formation of a
central cavity in a solid rod of cells (in the tail). The formation of the neural folds and
their coming together in the midline to form the neural tube is driven by changes in cell
shape and convergent extension within the tube itself, as well as by forces generated
by adjacent tissues. Separation of the neural tube from the overlying ectoderm requires
changes in cell adhesiveness.

Cell migration

Cell migration is a major feature of animal morphogenesis, with cells moving over
relatively long distances from one site to another. The example we shall discuss here
Quail Chick
is the migration of neural crest cells in vertebrate embryos to produce an enormously
diverse range of end products, including cartilage in the face, melanocytes in the skin
and the peripheral and the autonomic nervous systems. We will leave until Chapter 12
the migration of immature neurons that is so fundamental to the morphogenesis of
the nervous system. Other important examples of cell migration dealt with elsewhere
in the book include germ-cell migration (see Chapter 9) and muscle-cell migration in
vertebrate limbs (see Chapter 11).

8.13 Neural crest migration is controlled by environmental cues

Neural crest cells of vertebrates have their origin at the edges of the neural plate and
first become recognizable during neurulation (see Section 5.4). At the time of closure
of the neural tube in vertebrate embryos, neural crest cells can be seen on each side
of the midline (Fig. 8.35). They undergo an epithelial-to-mesenchymal transition and
leave the midline, migrating away from it on either side. As we saw earlier, epithelial-
to-mesenchymal transitions involve the transcription factors Slug and Snail, which
control processes that transform nonmotile epithelial cells into migrating cells (see
Section 8.7).
Neural crest cells give rise to a wide variety of different cell types, including cartilage
in the head, pigment cells in the dermis, the medullary cells of the adrenal gland, glial
Schwann cells, and to neurons of both the somato-sensory and the autonomic periph-
eral nervous systems. The differentiation of neural crest cells is discussed in Chapter 10.
Here we focus on the migration pathways of the crest cells in the trunk region of the
chick embryo. We have already discussed the migration of neural crest cells from the
hindbrain region of the neural tube to form the branchial arches (see Section 5.13).
Various strategies have been used to follow the migration of neural crest cells. For
Fig. 8.35 Following cell-migration
example, because quail cells have a nuclear marker that distinguishes them from
pathways by grafting a piece of quail
chick cells, grafting a neural tube from a quail embryo into a chick embryo allows the
neural tube to a chick host. A piece of
subsequent migration pathways of the quail neural crest cells in the chick embryo to
neural tube from a quail embryo is grafted
to a similar position in a chick host. The
be followed (see Fig. 8.35). It is also possible to identify migrating chick neural crest
photograph shows migration of the quail cells by tagging them with labeled monoclonal antibodies, by labeling them with a
neural crest cells (red arrows). Their migration fluorescent dye such as DiI, or by expressing a photoactivatable green fluorescent pro-
can be tracked as quail cells have a nuclear tein (pa-GFP) in the neural tube. pa-GFP can then be made to fluoresce when required
marker that distinguishes them from chick by a precisely focused beam of light, and the migration patterns of a small group of
cells. neural crest cells can be followed. Migrating neural crest cells in the zebrafish embryo
Photograph courtesy of N. Le Douarin. can be seen in Box 1D (p. 20).
 Cell migration 319

dorsal root
ganglion
ectoderm
pigment
1 cell
Fig. 8.36 Neural crest migration in the
trunk of the chick embryo. One group of
cells (1) migrates under the ectoderm to
somite
give rise to pigment cells (shown in dotted
outline). The other group of cells (2) migrates
neural over the neural tube and then through the
tube anterior half of the somite; the cells do
not migrate through the posterior half of
2 the somite. Those that migrate along this
e rior pathway give rise to dorsal root ganglia,
noto- Post
chord sympathetic ganglia, and cells of the adrenal
rior
Ante cortex, and their future sites are also shown
ectoderm in dotted outline. Those cells opposite the
aorta
posterior regions of a somite migrate in both
scelerotome
dermomyotome sympathetic directions along the neural tube until they
ganglion come to the anterior region of a somite. This
adreno- results in a segmental pattern of migration
medullary
cells fate of neural crest cells and is responsible for the segmental
arrangement of ganglia.

There are two main migratory pathways for neural crest cells in the trunk of the
chick embryo (Fig. 8.36). One goes dorso-laterally under the ectoderm and over
the somites; cells that migrate this way mainly give rise to pigment cells, which popu- Anterior
dorsal root ganglion
late the skin and feathers. The other pathway is more ventral, primarily giving rise neural tube
to sympathetic and sensory ganglion cells. Some crest cells move into the somites
to form dorsal root ganglia; others migrate through the somites to form sympathetic
ganglia and adrenal medulla, but appear to avoid the region around the notochord. somite
Trunk neural crest selectively migrates through the anterior half of the somite and not
through the posterior half. Within each somite, neural crest cells are found only in the
Anterior
anterior half, even when they originate in neural crest adjacent to the posterior half of
the somite. This behavior is unlike that of the neural crest cells taking the dorsal path-
way, which migrate over the whole dorso-lateral surface of the somite. Neural crest
migration to the anterior of each somite results in the distinct segmental arrangement
of spinal ganglia in vertebrates, with one pair of ganglia for each pair of somites—one Posterior
segment—in the embryo (Fig. 8.37). The segmental pattern of migration is due to Posterior
different guidance cues on the two halves of each somite. If the somites are rotated
through 180° so that their antero-posterior axis is reversed, the crest cells still migrate Fig. 8.37 Segmental arrangement of
through the original anterior halves only. dorsal root ganglia is due to the migration
Interactions between ephrins and Ephs are likely to contribute to the expulsion of of crest cells through the anterior half
neural crest cells from the posterior halves of the somites. In the chick, ephrin B1 of the somite only. Neural crest cells
is localized to the posterior halves of somites and acts as a repulsive guidance cue, cannot migrate through the posterior (gray)
half of a somite but can migrate in either
interacting with EphB3 on the neural crest cells, whereas the anterior halves of the
direction along the neural tube and through
somites, through which the neural crest cells do travel, express EphB3. Membrane
the anterior half of somites (yellow). The
proteins called semaphorins are also involved in repelling neural crest cells from
dorsal root ganglion in a given segment is
the posterior half of the somite. Interactions such as these provide a molecular basis thus made up of crest cells from the posterior
for the segmental arrangement of the spinal ganglia. region of the segment anterior to it, neural
The neural tube and notochord also both influence neural crest migration. If the early crest cells immediately adjacent to it (white),
neural tube is inverted through 180°, before neural crest migration starts, so that the and crest cells from the posterior of its own
dorsal surface is now the ventral surface, one might think that the cells that normally segment.
320 Chapter 8 Morphogenesis: change in form in the early embryo

migrate ventrally, now being nearer to their destination, would move ventrally. But this
is not the case, and many of the crest cells move upward through the sclerotome in a
ventral-to-dorsal direction, staying confined to the anterior half of each somite. This
suggests that the neural tube somehow influences the direction of migration of neural
crest cells. The notochord also exerts an influence, inhibiting neural crest cell migration
over a distance of about 50 µm, and thus preventing the cells from approaching it.
The direction of neural crest migration does involve some chemotaxis towards spe-
cific chemical cues. The migration of streams of neural crest cells in particular direc-
tions mainly results from the cells becoming polarized in the direction of migration,
probably due to the actions of the Wnt-mediated planar cell polarity pathway (see
Box 8C, p. 310). Inhibition of components of this pathway, such as Wnts and Frizzled,
inhibit neural crest migration.
Neural crest cells are guided by interactions with the extracellular matrix over
which they are moving, as well as by cell–cell interactions. Neural crest cells can
interact with extracellular matrix molecules by means of their cell-surface integrins
(see Box 8B, p. 293). Avian neural crest cells cultured in vitro adhere to, and migrate
efficiently on, fibronectin, laminin, and various collagens. Blocking adhesion of neu-
ral crest cells to fibronectin or laminin by blocking the integrin β1 subunit in vivo
causes severe defects in the head region but not in the trunk, suggesting that the crest
cells in these two regions adhere by different mechanisms, probably involving other
integrins. It is striking how neural crest cells in culture will preferentially migrate
along a track of fibronectin, although the role of this extracellular matrix molecule in
guiding the cells in the embryo is still unclear.

Summary

Streams of neural crest cells migrate from both sides of the dorsal neural tube to form
a wide variety of tissues in different parts of the body. Their migration is guided by
signals from other cells and by interactions with the extracellular matrix. In neural crest
cells migrating in the trunk, repulsive interactions between ephrins in the posterior
halves of the somites and Eph receptors on the neural crest cells prevent their migra-
tion over or through the posterior halves. Thus, neural crest cells that will give rise to
the dorsal root ganglia of the peripheral nervous system accumulate adjacent to the
anterior half of each somite, resulting in a segmental arrangement of paired ganglia
along the antero-posterior axis.

Summary: directed cell migration of vertebrate neural crest

neural crest cells

migration through anterior migration over somites

of somites

dorsal ganglion cells pigment cells

Directed dilation

Hydrostatic pressure can provide the force for morphogenesis in a variety of situa-
tions. We have already seen how an increase in hydrostatic pressure inside the mam-
malian blastocyst causes an increase in volume and maintains the blastocyst in a
 Directed dilation 321

roughly spherical shape (see Section 8.5). Here, we consider examples of directed Internal pressure applied to structure which
dilation, where an increase in internal hydrostatic pressure causes a distinct asym- is strong around its circumference and
weak along its length
metric change in shape as a result of the particular properties of the structure. For
example, if the circumferential resistance of a stretchable tube to internal pressure is strong circumferentially
much greater than the resistance lengthways, then an increase in the internal pressure
will cause an increase in length (Fig. 8.38).

weak longitudinally
8.14 Later extension and stiffening of the notochord occurs
by directed dilation pressure
After the Xenopus notochord has formed, its volume increases threefold, and there is
considerable further lengthening as it straightens and becomes stiffer. At this stage,
the notochord has become surrounded by a sheath of extracellular material, which Expansion along length
restricts circumferential expansion but does allow expansion in the antero-posterior
direction. The cells within the notochord develop fluid-filled vacuoles and expand
in volume as a result. They thus exert hydrostatic pressure on the notochord sheath,
resulting in directed dilation: circumferential expansion of the notochord is prevented
by the resistance of the sheath, and this ensures that the increase in volume (dilation) Fig. 8.38 Directed dilation. Hydrostatic
is directed along the notochord’s long axis. pressure inside a constraining sheath or
The vacuoles in the notochord cells are filled with glycosaminoglycans which, membrane can lead to elongation of the
because of their high carbohydrate content, tend to attract water into the vacuoles structure. If the circumferential resistance
by osmosis. It is this that produces the hydrostatic pressure that causes the increase is much greater than the longitudinal
in cell volume, and the consequent stiffening and straightening of the notochord. resistance, as it is in the notochord sheath,
the rod of cells inside the sheath lengthens.
Changes in the structure of the sheath during the period of notochord elongation
fit well with the proposed hydrostatic mechanism. The sheath contains both gly-
cosaminoglycans, which have little tensile strength, and the fibrous protein collagen,
which has a high tensile strength; during notochord dilation the density of collagen
fibers increases, providing resistance to circumferential expansion. The crucial role
of the sheath in dilation and elongation is shown by the fact that if it is digested
away, the notochord buckles and folds and the notochord cells, instead of being flat,
become rounded.

8.15 Circumferential contraction of hypodermal cells elongates

the nematode embryo

During the early development of the nematode there is little change in body shape
from the ovoid form of the fertilized egg, even during gastrulation. After gastrula-
tion, about 5 hours after fertilization, the embryo begins to elongate rapidly along its
antero-posterior axis. Elongation takes about 2 hours, during which time the embryo
decreases in circumference about threefold and undergoes a fourfold increase in
length.
This elongation is brought about by a change in shape of the hypodermal (epider-
mal) cells that make up the outermost layer of the embryo; their destruction by laser
ablation prevents elongation. During embryo elongation, these cells change shape
so that, instead of being elongated in the circumferential direction, they become
elongated along the antero-posterior axis (Fig. 8.39). Throughout this elongation the
hypodermal cells remain attached to each other by cell junctions. The junctions are
also linked within the cells by actin-containing fibers that run circumferentially, and
these fibers appear to shorten as the cells elongate. The disruption of actin filaments
by cytochalasin D treatment blocks elongation, and so it is very likely that their con-
traction brings about the change in cell shape. Circumferential contraction of the
hypodermal cells causes an increase in hydrostatic pressure within the embryo, forc-
ing an extension in an antero-posterior direction. Circumferentially oriented microtu-
bules may also have a mechanical role in constraining the expansion, in the same way
322 Chapter 8 Morphogenesis: change in form in the early embryo

Fig. 8.39 Increase in nematode body

length by directed dilation. The change
in body shape over 2 hours is illustrated
in the top panel. The increase in length is
due to circumferential contraction of the
hypodermal cells, as shown in the bottom
panel. The change in shape of a single cell
can be seen in the cell marked with arrows.
Scale bar = 10 µm.
Photographs courtesy of J. Priess, from
Priess, J.R., et al.: 1986.

as the sheath of the Xenopus notochord described earlier. Increase in nematode body
length is thus another example of directed dilation.

8.16 The direction of cell enlargement can determine the form of a plant leaf

Cell enlargement is a major process in plant growth and morphogenesis, providing

up to a 50-fold increase in the volume of a tissue. The driving force for expansion is
the hydrostatic pressure—turgor pressure—exerted on the cell wall as the protoplast
swells as a result of the entry of water into cell vacuoles by osmosis (Fig. 8.40).
Plant-cell expansion involves synthesis and deposition of new cell-wall material, and
is an example of directed dilation. The direction of cell growth is determined by the
orientation of the cellulose fibrils in the cell wall. Enlargement occurs primarily in a
direction at right angles to the fibrils, where the wall is weakest. The orientation of
cellulose fibrils in the cell wall is thought to be determined by the microtubules of
the cell’s cytoskeleton, which are responsible for positioning the enzyme assemblies
that synthesize cellulose at the cell wall. Plant growth hormones, such as ethylene
and gibberellins alter the orientation in which the fibrils are laid down and so can
alter the direction of expansion. Auxin aids expansion by loosening the structure of
the cell wall.
The development of a leaf involves a complex pattern of cell division and cell
elongation, with cell elongation playing a central part in the expansion of the leaf
blade. Two mutations that affect the shape of the blade by affecting the direction of
cell elongation have been identified. Leaves of the Arabidopsis mutant angustifolia
are similar in length to the wild type but are much narrower (Fig. 8.41). In contrast,
rotundifolia mutations reduce the length of the leaf relative to its width. Neither of
these mutations affects the number of cells in the leaf. Examination of the cells in the
developing leaf shows that these mutations are affecting the direction of elongation
of the enlarging cells.

Fig. 8.40 Enlargement of a plant cell.

Plant cells expand as water enters the cell
water
vacuoles and thus causes an increase in
intracellular hydrostatic pressure. The cell
elongates in a direction perpendicular to
the orientation of the cellulose fibrils in its
cell wall.
 Directed dilation 323

Wild type

angustifolia mutation

Fig. 8.41 The shape of the leaves of

Arabidopsis is affected by mutations
rotundifolia mutation affecting cell elongation. The angustifolia
mutation results in cells with narrow leaves,
while rotundifolia mutations cause short, fat
leaves to develop.
Photograph courtesy of H. Tsukaya, from
Tsuge, T., et al.: 1996.

Summary

Directed dilation results from an increase in hydrostatic pressure inside a structure, and
unequal peripheral resistance to this pressure. Extension of the notochord is brought
about by directed dilation, in which the notochord interior increases in volume while
its circumferential expansion is constrained by the notochord sheath, forcing it to elon-
gate. Similarly, the nematode embryo elongates after gastrulation as a result of a cir-
cumferential contraction of the outer hypodermal cells that generates pressure on the
internal cells, forcing the embryo to extend in an antero-posterior direction. In plants,
the shape of a leaf is determined by the direction of cell enlargement, which is dictated
by the distribution of cellulose fibrils in the cell wall. Directed dilation resulting from
internal pressure elongates the cell where the cell wall is weakest.

Summary: directed dilation

circumferential restraint
circumferential restraint circumferential restraint
by sheath by plant cell wall
hydrostatic pressure

notochord elongation cell elongation

nematode elongation

Summary to Chapter 8
Changes in the shape of animal embryos as they develop are mainly due to forces generated
by changes in cell shape, cell motility, and differences in cell adhesiveness. Formation of a
hollow, spherical blastula or blastocyst is the result of particular patterns of cell division,
cell packing, and cell polarization, and the spherical shape is maintained by flow of fluid
324 Chapter 8 Morphogenesis: change in form in the early embryo

into the interior and the hydrostatic pressure generated. Gastrulation involves major move-
ments of cells, such that the future endoderm and mesoderm move inside the embryo to
their appropriate positions in relation to the main body plan. Convergent extension has a
key role in many developmental contexts, such as elongation of the antero-posterior body
axis during vertebrate gastrulation, lengthening and closure of the neural tube, and germ-
band extension in Drosophila. Formation of the vertebrate neural tube from the neural plate
involves cell shape changes that cause the edges of the plate to curve upwards and finally
fuse. The role of cell migration in morphogenesis is represented in this chapter by the neural
crest cells, which leave the neural tube to give rise to a wide variety of tissues throughout
the body. The phenomenon of directed dilation, involving hydrostatic pressure, is respon-
sible for the later elongation of the notochord, the elongation of the nematode embryo and
the direction of enlargement of plant cells.

n End of chapter questions how this mechanism is used to initiate gastrulation in sea urchins,
Drosophila, and Xenopus.
Long answer (concept questions) 9. How do changes in cell shape and changes in cell adhesiveness
1. Name the four types of adhesive interactions between cells and collaborate in the formation of the neural tube?
between cells and the extracellular matrix. What are the trans-
10. Describe the migration route of the neural crest cells that will
membrane proteins that characterize each of these adhesive inter-
form the dorsal root ganglia. What role do Ephs and ephrins play in
actions? What are the particular features of an adherens junction:
this migratory pathway?
what cell adhesion molecule is used, what cytoskeletal compo-
nent is involved, and what links the adhesion molecule to the 11. What is the influence of directed dilation on formation of the
­cytoskeleton? notochord? What role do sugars (glycosaminoglycans) play in this
process?
2. Describe the results of an experiment in which epidermis and
neural plate from a neurula stage amphibian embryo are first disag- 12. In your own words, describe the cell movements that
gregated into single cells, then mixed together and reaggregated. ­occur during gastrulation, from the first appearance of the blas-
How is this experiment interpreted in terms of the relative strength topore, to the first appearance of the neural folds. Do this first for
of ‘self’ adhesiveness in these two tissues? How does this result Xenopus, then describe the major differences seen in chicks and
reflect the natural organization of these tissues in the embryo? zebrafish.

3. Define ‘radial cleavage.’ In the radial cleavage of an embryo such

as the sea urchin, how are the centrosomes positioned at the first Multiple choice (factual recall questions)
cleavage, with respect to the animal–vegetal axis? The second cleav-
NB There is only one right answer to each question.
age? The third cleavage?
1. Adherens junctions play in important role in development; they
4. What is compaction of the mouse embryo? How is compaction are composed of
related to two important subsequent events in mouse embryogen-
esis: the formation of the blastocoel, and the formation of the inner a) cadherins linked to other cadherins extracellularly, and to actin
cell mass? filaments intracellularly
b) cadherins linked to other cadherins extracellularly, and to inter-
5. Contrast the properties of an epithelial tissue with those of a mediate filaments intracellularly
mesenchymal tissue. How are cadherins related to the transition of
c) immunoglobulin superfamily (IgSF) molecules linked to other
epithelium to mesenchyme; how is this transition regulated, or trig-
IgSF molecules extracellularly, and to the cytoskeleton intracel-
gered (think: ‘snail’)?
lularly
6. Refer to the fate maps of sea urchins (Fig. 8.14) and Xenopus d) integrins linked to extracellular matrix molecules extracellularly,
(Fig. 8.24): in sea urchins, mesoderm is most vegetal and endo- and to the cytoskeleton intracellularly
derm lies between mesoderm and ectoderm; in Xenopus, endo-
derm is most vegetal and mesoderm lies between the endoderm 2. If embryos are disaggregated with chemicals or proteases, and the
and the ectoderm. Can you briefly describe the differences in individual cells are mixed together in culture, what will ­happen?
gastrulation in these two organisms that explain these differing a) The cells will associate with one another randomly.
­arrangements? b) The cells will often sort themselves so that like cells are
7. Contrast epiboly and convergent extension. How are these two ­together.
processes used during Xenopus gastrulation? c) The cells will reaggregate to form a normal embryo capable of
8. If cells in a flat sheet constrict one surface (the apical surface), continuing development.
the sheet will develop a depression in that region. Give examples of d) The cells will regulate to form one or more normal embryos.
 Directed dilation 325

3. What type of cleavage is used in frog embryos? b) Changes in cell shape in the neural plate and changes in adhe-
a) radial sion molecule expression in the neural tube are required during
formation of the neural tube.
b) spiral
c) Changes in cell shape in the neural plate are the only process
c) superficial
required for neural tube formation.
d) unequal
d) Neural tube formation is a passive by-product of convergent
4. How do the endodermal cells of the sea urchin embryo carry out extension in the notochord.
gut formation during gastrulation? 10. Neural crest cells taking the dorsal-lateral route will become
a) Changes in cell shape initiate the invagination; convergent exten- a) adrenal medulla
sion extends the sheet of cells into the blastocoel, and finally,
filopodia make contact with the future mouth region and pull the b) dorsal root ganglia
tip of the gut to that point. c) melanocytes
b) They crawl inside the blastocoel and form a solid rod, which is d) sympathetic ganglia
then hollowed out to form the tube of the gut.
Multiple choice answer key
c) They move inside the blastocoel, and form the skeletal structures
that will support the adult animal. 1: a, 2: b, 3: a, 4: a, 5: b, 6: c, 7: d, 8: a, 9: b, 10: c.
d) They move into the blastocoel as mesenchyme and crawl ven-
trally, to mark the location of the future mouth.
n General further reading
5. During gastrulation in the frog, the very first cells to move into
the interior of the embryo through the blastopore come from the Alberts, B. et al.: Molecular Biology of the Cell (5th edn). New York:
surface layer of cells in the marginal zone; they will become: Garland Science, 2008.
a) ectoderm Aman, A., Piotrowski, T.: Cell migration during morphogenesis.
b) endoderm Dev. Biol. 2010, 341: 20–33.
Engler, A.J., Humbert, P.O., Wehrle-Haller, B., Weaver, V.M.: Mul-
c) mesoderm tiscale modeling of form and function. Science 2009, 324:
d) yolk 208–212.
Mattila, P.K., Lappalainen, P.: Filopodia: molecular architecture and
6. Gastrulation in sea urchins, Drosophila, and Xenopus all begin
cellular functions. Nat. Rev. Mol. Cell Biol. 2008, 9: 446–454.
with a change in cell shape, in which the apical surface of an epithe-
lial sheet contracts. This process is called Stern, C. et al.: Gastrulation: From Cells to Embryos. New York: Cold
Spring Harbor Laboratory Press; 2004.
a) convergent extension
b) epiboly
c) invagination
n Section further reading
d) involution 8.1 Sorting out of dissociated cells demonstrates differences
7. In Xenopus, the mesoderm moves in through the blastopore by in cell adhesiveness in different tissues
rolling around the dorsal lip in a process called Davis, G.S., Phillips, H.M., Steinberg, M.S.: Germ-layer surface ten-
a) convergent extension sions and ‘tissue affinities’ in Rana pipiens gastrulae: quantita-
tive measurements. Dev. Biol. 1997, 192: 630–644.
b) epiboly
Steinberg, M.S.: Differential adhesion in morphogenesis: a mod-
c) invagination ern view. Curr. Opin. Genet. Dev. 2007, 17: 281–286.
d) involution
Box 8B Cell-adhesion molecules and cell junctions
8. In Xenopus, the elongation of the mesoderm toward the anterior
results from the intercalation of cells during a process called Hynes, R.O: Integrins: bidirectional allosteric signaling mecha-
nisms. Cell 2002, 110: 673–689.
a) convergent extension
Tepass, U., Truong, K., Godt, D., Ikura, M., Peifer, M.: Cadherins in
b) epiboly embryonic and neural morphogenesis. Nat Rev. Mol. Cell Biol.
c) invagination 2000, 1: 91–100.
d) involution Thiery, J.P.: Cell adhesion in development: a complex signaling
network. Curr. Opin. Genet. Dev. 2003, 13: 365–371.
9. In chicks, the formation of the neural tube relies on what cell
biological processes? 8.2 Cadherins can provide adhesive specificity
a) Changes in adhesiveness between the prospective neural tube Duguay, D., Foty, R.A., Steinberg, M.S.: Cadherin-mediated cell
cells and the prospective epidermis in the ectoderm are the only adhesion and tissue segregation: qualitative and quantitative
changes required for neural tube formation. determinants. Dev. Biol. 2003, 253: 309–323.
326 Chapter 8 Morphogenesis: change in form in the early embryo

Levine, E., Lee, C.H., Kintner, C., Gumbiner, B.M.: Selective dis- physically plausible in the sea urchin Stronglyocentrotus pur-
ruption of E-cadherin function in early Xenopus embryos by a puratus. Dev. Biol. 1998, 204: 235–250.
dominant negative mutant. Development 1994, 120: 901–909. Duloquin, L., Lhomond, G., Gache, C.: Localized VEGF signaling
Takeichi, M., Nakagawa, S., Aono, S., Usui, T., Uemura, T.: Pattern- from ectoderm to mesenchyme cells controls morphogenesis
ing of cell assemblies regulated by adhesion receptors of the cad- of the sea urchin embryo skeleton. Development 2007, 134:
herin superfamily. Proc. Roy. Soc. Lond. B 2000, 355: 885–896. 2293–2302.
Ettensohn, C.A.: Cell movements in the sea urchin embryo. Curr.
8.3 The orientation of the mitotic spindle determines the Opin. Genet. Dev. 1999, 9: 461–465.
plane of cleavage at cell division Gustafson, T., Wolpert, L.: Studies on the cellular basis of mor-
Backues, S.K., Konopka, C.A., McMichael, C.M., Bednarek, S.Y.: phogenesis in the sea urchin embryo. Directed movements of
Bridging the divide between cytokinesis and cell expansion. primary mesenchyme cells in normal and vegetalized larvae.
Curr. Opin. Plant. Biol. 2007, 10: 607–615. Exp. Cell Res. 1999, 253: 288–295.
Galli, M., van den Heuvel, S.: Determination of the cleavage plane in Mattila, P.K., Lappalainen, P.: Filopodia: molecular architecture and
early C. elegans embryos. Annu. Rev. Genet. 2008, 42: 389–411. cellular functions. Nat. Rev. Mol. Cell Biol. 2008, 9: 446–454.
Glotzer, M.: Cleavage furrow positioning. J. Cell Biol. 2004, 164: Miller, J.R., McClay, D.R.: Changes in the pattern of adherens
347–351. junction-associated β-catenin accompany morphogenesis in
Gönczy, P., Rose, L.S.: Asymmetric cell division and axis forma- the sea urchin embryo. Dev. Biol. 1997, 192: 310–322.
tion in the embryo. In WormBook (ed. The C. elegans Research Odell, G.M., Oster, G., Alberch, P., Burnside, B.: The mechanical
Community) (October 15, 2005). doi/10.1895/wormbook.1.30.1, basis of morphogenesis. I. Epithelial folding and invagination.
http://www.wormbook.org (date accessed 21 May 2010). Dev. Biol. 1981, 85: 446–462.
Raftopoulou, M., Hall, A.: Cell migration: Rho GTPases lead the
8.4 Cells become polarized in the sea-urchin blastula and the way. Dev Biol. 2004, 265: 23–32.
mouse morula Röttinger, E., Saudemont, A., Duboc, V., Besnardeau, L., McClay,
Pauken, C.M., Capco, D.G.: Regulation of cell adhesion during D., Lepage, T.: FGF signals guide migration of mesenchymal
embryonic compaction of mammalian embryos: roles for PKC cells, control skeletal morphogenesis and regulate gastrula-
and beta-catenin. Mol. Reprod. Dev. 1999, 54: 135–144. tion during sea urchin development. Development 2008, 135:
Sutherland, A.E., Speed, T.P., Calarco, P.G.: Inner cell allocation in 353–365.
the mouse morula: the role of oriented division during fourth
8.8 Mesoderm invagination in Drosophila is due to changes
cleavage. Dev. Biol. 1990, 137: 13–25.
in cell shape that are controlled by genes that pattern the
8.5 Fluid accumulation as a result of tight-junction formation dorso-ventral axis
and ion transport forms the blastocoel of the mammalian Martin, A.C., Kaschube, M., Wieschaus, E.F.: Pulsed contractions
blastocyst of an actin–myosin network drives apical contraction. Nature
Barcroft, L.C., Offenberg, H., Thomsen, P., Watson, A.J.: Aquaporin 2009, 457: 495–499.
proteins in murine trophectoderm mediate transepithelial water
8.9 Germ-band extension in Drosophila involves
movements during cavitation. Dev. Biol. 2003, 256: 342–354.
myosin-dependent remodeling of cell junctions and cell
Eckert, J.J., Fleming, T.P.: Tight junction biogenesis during early
intercalation
development. Biochim. Biophys. Acta 2008, 1778: 717–728.
Fleming, T.P., Sheth, B., Fesenko, I.: Cell adhesion in the preim- Bertet, C., Sulak, L., Lecuit, T.: Myosin-dependent junction remod-
plantation mammalian embryo and its role in trophectoderm elling controls planar cell intercalation and axis elongation.
differentiation and blastocyst morphogenesis. Front. Biosci. Nature 2004, 429: 667–671.
2001, 6: 1000–1007.
8.10 Dorsal closure in Drosophila and ventral closure in
Watson, A.J., Natale, D.R., Barcroft, L.C.: Molecular regulation of
Caenorhabditis elegans are brought about by the action
blastocyst formation. Anim. Reprod. Sci. 2004, 82–83: 583–592.
of filopodia
8.6 Internal cavities can be created by cell death Chin-Sang, I.D., Chisholm, A.D.: Form of the worm: genetics of
Coucouvanis, E., Martin, G.R.: Signals for death and survival: a epidermal morphogenesis in C. elegans. Trends Genet. 2000, 16:
two-step mechanism for cavitation in the vertebrate embryo. 544–551.
Cell 1995, 83: 279–287. Jacinto, A., Woolner, S., Martin, P.: Dynamic analysis of dorsal
closure in Drosophila: from genetics to cell biology. Dev. Cell
8.7 Gastrulation in the sea urchin involves cell migration and 2002, 3: 9–19.
invagination Peralta, X.G., Toyama, Y., Hutson, M.S., Montague, R., Venakides,
Davidson, L.A., Oster, G.F., Keller, R.E., Koehl, M.A.R.: Measure- S., Kiehart, D.P., Edwards, G.S.: Upregulation of forces and mor-
ments of mechanical properties of the blastula wall reveal phogenic asymmetries in dorsal closure during Drosophila
which hypothesized mechanisms of primary invagination are ­development. Biophys. J. 2007, 92: 2583–2596.
 Directed dilation 327

Woolner, S., Jacinto, A., Martin, P.: The small GTPase Rac plays Ninomiya, H., Elinson, R.P., Winklbauer, R.: Antero-posterior tis-
multiple roles in epithelial sheet fusion—dynamic studies of sue polarity links mesoderm convergent extension to axial pat-
Drosophila dorsal closure. Dev. Biol. 2005, 282: 163–173. terning. Nature 2004, 430: 364–367.
Simons, M., Mlodzik, M.: Planar cell polarity signaling: from fly
development to human disease. Annu. Rev. Genet. 2008, 42:
8.11 Vertebrate gastrulation involves several different types 517–540.
of tissue movement Torban, E., Kor, C., Gros, P.: Van Gogh-like2 (Strabismus) and its
Adams, D.S., Keller, R., Koehl, M.A.: The mechanics of notochord role in planar cell polarity and convergent extension in verte-
elongation, straightening, and stiffening in the embryo of brates. Trends Genet. 20: 570–577.
Xenopus laevis. Development 1990, 100: 115–130. Zallen, J.A.: Planar polarity and tissue morphogenesis. Cell 2007,
Dzamba, B.J., Jakab, K.R., Marsden, M., Schwartz, M.A., DeSimone, 129: 1051–1063.
D.W.: Cadherin adhesion, tissue tension, and noncanonical
Wnt signaling regulate fibronectin matrix organization. Dev. 8.12 Neural tube formation is driven by changes in cell shape
Cell. 2009, 16: 421–432. and convergent extension
Goto, T., Davidson, L., Asashima, M., Keller, R.: Planar cell polarity Colas, J.F., Schoenwolf, G.C.: Towards a cellular and molecular
genes regulate polarized extracellular matrix deposition dur- understanding of neurulation. Dev. Dyn. 2001, 221: 117–145.
ing frog gastrulation. Curr. Biol. 2005, 15: 787–793. Davidson, L.A., Keller, R.E.: Neural tube closure in Xenopus lae-
Heisenberg, C.P., Solnica-Krezel, L.: Back and forth between cell vis involves medial migration, directed protrusive activity, cell
fate specification and movement during vertebrate gastrula- intercalation and convergent extension. Development 1999,
tion. Curr. Opin. Genet. Dev. 2008, 18: 311–316. 126: 4547–4556.
Keller, R.: Cell migration during gastrulation. Curr. Opin. Cell Biol. Greene, N.D.E., Stanier, P., Copp, A.J.: Genetics of human neural
2005, 17: 533–541. tube defects. Hum. Mol. Genet. 2009, 18: R113–R129.
Haigo, S.L., Hildebrand, J.D., Harland, R.M., Wallingford, J.B.:
Keller, R., Cooper, M.S., D’Anilchik, M., Tibbetts, P., Wilson, P.A.:
Shroom induces apical constriction and is required for hinge-
Cell intercalation during notochord development in Xenopus
point formation during neural tube closure. Curr. Biol. 2003,
laevis. J. Exp. Zool. 1989, 251: 134–154.
13: 2125–2137.
Keller, R., Davidson, L.A., Shook, D.R.: How we are shaped:
Kibar, Z., Capra, V., Gros, P.: Toward understanding the genetic
the ­biomechanics of gastrulation. Differentiation 2003, 71: 171–
basis of neural tube defects. Clin. Genet. 2007, 71: 295–310.
205.
Rolo, A., Skoglund, P., Keller, R.: Morphogenetic movements dur-
Montero, J.A., Carvalho, L., Wilsch-Brauninger, M., Kilian, B.,
ing neural tube closure in Xenopus require myosin IIB. Dev.
­Mustafa, C., Heisenberg, C.P.: Shield formation at the onset of
Biol. 2009, 327: 327–338.
zebrafish gastrulation. Development 2005, 132: 1187–1198.
Torban, E., Patenaude, A.M., Leclerc, S., Rakowiecki, S., Gauthier,
Montero, J.A., Heisenberg, C.P.: Gastrulation dynamics: cells move S., Andelfinger, G., Epstein, D.J., Gros, P.: Genetic interaction
into focus. Trends Cell Biol. 2004, 14: 620–627. between members of the Vang1 family causes neural tube
Ninomiya, H., Winklbauer, R.: Epithelial coating controls mesen- ­defects in mice. Proc. Natl. Acad. Sci. USA 2008, 105: 3449–
chymal shape change through tissue-positioning effects and 3454.
reduction of surface-minimizing tension. Nat. Cell Biol. 2008, Wallingford, J.B., Harland, R.M.: Neural tube closure requires
10: 61–69. Dishevelled-dependent convergent extension of the midline.
Shih, J., Keller, R.: Gastrulation in Xenopus laevis: involution—a Development 2002, 129: 5815–5825.
current view. Dev. Biol. 1994, 5: 85–90.
Voiculescu, O., Bertocchini, F., Wolpert, L., Keller, R.E., Stern, C.D.: 8.13 Neural crest migration is controlled by
The amniote primitive streak is defined by epithelial cell inter- environmental cues
calation before gastrulation. Nature 2007, 449: 1049–1052. Bronner-Fraser, M.: Mechanisms of neural crest migration. Bio­
Wacker, S., Grimm, K., Joos, T., Winklbauer, R.: Development and Essays 1993, 15: 221–230.
control of tissue separation at gastrulation in Xenopus. Dev. Holder, N., Klein, R.: Eph receptors and ephrins: effectors of mor-
Biol. 2000, 224: 428–439. phogenesis. Development 1999, 126: 2033–2044.
Wallingford, J.B., Fraser, S.E., Harland, R.M.: Convergent exten- Kuan, C.Y., Tannahill, D., Cook, G.M., Keynes, R.J.: Somite polarity
sion: the molecular control of polarized cell movement during and segmental patterning of the peripheral nervous system.
embryonic development. Dev. Cell 2002, 2: 695–706. Mech. Dev. 2004, 121: 1055–1068.
Kuriyama, S., Mayor, R.: Molecular analysis of neural crest migra-
tion. Philos. Trans. R. Soc. Lond. B Biol. Sci. 2008, 363: 1349–
Box 8C Convergent extension 1362.
Keller, R., Davidson, L., Edlund, A., Elul, T., Ezin, M., Shook, D., Nagawa, S., Takeichi, M.: Neural crest emigration from the neural
Skoglund, P.: Mechanisms of convergence and extension by cell tube depends on regulated cadherin expression. Development
intercalation. Proc R. Soc. Lond. B 2000, 355: 897–922. 1998, 125: 2963–2971.
328 Chapter 8 Morphogenesis: change in form in the early embryo

Poliakoff, A., Cotrina, M., Wilkinson, D.G.: Diverse roles of Eph 8.15 Circumferential contraction of hypodermal cells
receptors and ephrins in the regulation of cell migration and elongates the nematode embryo
tissue assembly. Dev. Cell 2004, 7: 465–480. Priess, J.R., Hirsh, D.I.: Caenorhabditis elegans morphogenesis:
Thevenean, E., Marchant, L., Kuriyama, S., Gull, M., Moeppo, B., the role of the cytoskeleton in elongation of the embryo. Dev.
Parsons, M., Mayor, R.: Collective chemotaxis requires contact- Biol. 1986, 117: 156–173.
dependent cell polarity. Dev. Cell 2010, 19: 39–53.
Tucker, R.P.: Neural crest cells: a model for invasive behavior. Int. 8.16 The direction of cell enlargement can determine the
J. Biochem. Cell Biol. 2004, 36: 173–177. form of a plant leaf
Xu, Q., Mellitzer, G., Wilkinson, D.G.: Roles of Eph receptors and Jackson, D.: Designing leaves. Plant morphogenesis. Curr. Biol.
ephrins in segmental patterning. Proc. Roy. Soc. Lond. B 2000, 1996, 6: 917–919.
353: 993–1002. Tsuge, T., Tsukaya, H., Uchimiya, H.: Two independent and polar-
ized processes of cell elongation regulate leaf blade expansion
8.14 Later extension and stiffening of the notochord occurs in Arabidopsis thaliana (L.) Heynh. Development 1996, 122:
by directed dilation 1589–1600.
Adams, D.S., Keller, R., Koehl, M.A.: The mechanics of notochord
elongation, straightening and stiffening in the embryo of Xeno-
pus laevis. Development 1990, 110: 115–130.
 9
Germ cells, fertilization, and sex
nn The development nn Fertilization nn Determination of the
of germ cells sexual phenotype

Animal embryos develop from a single cell, the fertilized egg or zygote, which is
the product of the fusion of an egg and a sperm. In previous chapters we have
considered mainly how the basic body plan of the animal embryo is generated by
the development of the somatic cells and their assignment to the three germ lay-
ers. The zygote also gives rise to the germline cells—or germ cells—that develop
into eggs and sperm, and through which an organism’s genes are passed on to the
next generation. As we shall discuss in this chapter, in animals, germline cells are
specified and set apart in the early embryo, although functional mature eggs and
sperm are only produced by the adult. An important property of germ cells is that
they remain totipotent—able to give rise to all the different types of cells in the
body. Nevertheless, eggs and sperm in mammals have certain genes differentially
silenced during germ-cell development by a process known as genomic imprinting,
and we will look at the implications of this for development. Embryonic develop-
ment is initiated by the process of fertilization, and we shall discuss mechanisms in
sea urchins and mammals that ensure that only one sperm enters the egg. In the
animals we cover in this chapter, sex is determined by the chromosomal constitu-
tion, and we shall see that mechanisms of sex determination and of compensating
for the unequal complement of sex chromosomes between males and females differ
considerably in different species.

In sexually reproducing organisms, there is a fundamental distinction between the

germline cells (germ cells) and the somatic or body cells (see Section 1.2): the
former give rise to the gametes—eggs and sperm in animals—whereas somatic cells
make no genetic contribution to the next generation. Germ cells have three key func-
tions: the preservation of the genetic integrity of the germline, such as the prevention
of aging; the generation of genetic diversity; and the transmission of genetic informa-
tion to the next generation.
So far in this book, we have looked at the somatic development of our model organ-
isms. Not surprisingly, a great deal of the biology of animals and plants is devoted to
reproduction and sex, and in this chapter, we look at germ-cell formation, fertilization
of the egg by the sperm, and sex determination, mainly in the mouse, Drosophila, and
Caenorhabditis. As germ cells are the cells that will give rise to the next generation,
their development is crucial, and in animals, germ cells are usually specified and set
aside early in embryonic development. Plants, although reproducing sexually, differ
330 Chapter 9 Germ cells, fertilization, and sex

from most animals in that their germ cells are not specified early in embryonic devel-
opment, but during the development of the flowers (see Chapter 7). It is worth noting
that some simple animals, such as the coelenterate Hydra, can reproduce asexually,
by budding, and that even in some vertebrates, such as turtles, the eggs can develop
without being fertilized.
We will begin this chapter by considering how the germ cells are specified in the
early animal embryo and how they differentiate into eggs and sperm. Then, we will
look at the fertilization and activation of the egg by the sperm—the vital step that ini-
tiates development. Finally, we turn to sex determination—why males and females
are different from each other. In all the animals we look at here, sex is determined
genetically, by the number and type of specialized chromosomes known as the sex
chromosomes. Male and female embryos initially look the same, with sexual differ-
ences only emerging as a result of the activity of sex-determining genes located on
the sex chromosomes.

The development of germ cells

In all but the simplest animals, the cells of the germline are the only cells that can give
rise to a new organism. So, unlike somatic cells, which eventually all die, germ cells
in a sense outlive the bodies that produced them. They are, therefore, very special
cells. The outcome of germ-cell development is either a male gamete (the sperm in
animals) or a female gamete (the egg). The egg is a particularly remarkable cell, as it
ultimately gives rise to all the cells in an organism. In species whose embryos receive
no nutrition from the mother after fertilization, the egg must also provide everything
necessary for development, as the sperm contributes virtually nothing to the organ-
ism other than its chromosomes and a centrosome.
Animal germ cells typically differ from somatic cells by dividing less often during
early embryogenesis. Later, they are the only cells to undergo the type of nuclear
division called meiosis, by which gametes with half the number of chromosomes of
the germ-cell precursor are produced. The halving of chromosome number at meiosis
means that when two gametes come together at fertilization to form the zygote, the
diploid number of chromosomes is restored. If there were no reduction division in
germ-cell formation, the number of chromosomes in the somatic cells would double
in each generation.
Germ cells undergo meiosis and differentiate into eggs and sperm within special-
ized reproductive organs called gonads—the ovary in females and the testis in males.
But in many animals, including Drosophila and vertebrates, the germ cells are first
specified in a region quite distant from where the gonads will eventually form, and
migrate from their site of origin into the forming gonads. From the time they are first
specified as germline cells until the time that they enter the gonads, precursor germ
cells are known as primordial germ cells. In many animals, primordial germ cells
are formed in locations that seem to protect them from the inductive signals that
determine the fate of somatic cells, and instead shut down the somatic developmen-
tal program. In Caenorhabditis elegans, for example, there is evidence of a general
repression of transcription in the primordial germ cells. In some germlines, other
mechanisms that help suppress mutation and maintain genetic integrity are present.
In Drosophila, for example, the movement of transposons in the genome, which can
cause insertion mutations, is prevented by an RNA silencing pathway involving the
Piwi-interacting small RNAs (piRNAs).
Germ cells are specified very early in some animals, although not in mammals,
by cytoplasmic determinants present in the egg. We therefore start our discussion
of germ-cell development by looking at the specification of primordial germ cells by
special cytoplasm—the germ plasm—which is already present in these eggs.
 The development of germ cells 331

9.1 Germ-cell fate is specified in some embryos by a distinct germ plasm in Donor posterior pole plasm injected into
the egg anterior region of a second embryo

In flies, nematodes, fish, and frogs, molecules localized in specialized cytoplasm in

the egg are involved in specifying the germ cells. The clearest example of this is in
Drosophila, where primordial germ cells known as pole cells become distinct at the P
posterior pole of the egg about 90 minutes after fertilization, more than an hour ­before
the cellularization of the rest of the embryo (see Fig. 2.2). The cytoplasm at the pos-
terior pole is called the pole plasm and is distinguished by large organelles, the polar
granules, which contain both proteins and RNAs. Two key experiments demonstrate
that there is something special about the posterior cytoplasm. First, if the posterior end Y
of the egg is irradiated with ultraviolet light, which destroys the pole plasm activity,
no germ cells develop, although the somatic cells in that region do develop. ­Second,
if pole plasm from an egg is transferred to the anterior pole of another embryo, the
nuclei that become surrounded by the pole plasm are specified as germ-cell nuclei Anterior cells containing pole plasm transplanted
to the posterior region of a third embryo
(Fig. 9.1). If the anterior cells containing pole plasm are transplanted into the poste-
rior pole region of a third embryo, they develop as functional germ cells.
In Chapter 2, we saw how the main axes of the Drosophila egg are specified by the
follicle cells in the ovary, and how the mRNAs for proteins such as Bicoid and Nanos Y
become localized in the egg (see Section 2.14). The pole plasm also becomes local-
ized at the posterior end of the egg under the influence of the follicle cells. Several
maternal genes are involved in pole-plasm formation in Drosophila. Mutations in any
of at least eight genes result in the affected homozygous individual being ‘grandchild-
less.’ Its offspring lack a proper pole plasm, and although they may develop normally
G
in other ways, they lack germ cells and are therefore sterile. One of these eight genes
is oskar, which plays a central role in the organization and assembly of the pole
plasm; of the genes involved in pole-plasm formation, oskar is the only one to have its
mRNA localized at the posterior pole. The signal for localization is contained in the Adult fly develops germ cells with same genotype
3′untranslated region of the mRNA. Staufen protein is required for the localization as embryo Y among its own germ cells
of oskar mRNA and may act by linking the mRNA to the microtubule system that is
polarized along the antero-posterior axis (see Section 2.14). If the region of the oskar Fig. 9.1 Transplanted pole plasm can
gene that codes for the 3′localization signal is replaced by the bicoid 3′localization induce germ-cell formation in Drosophila.
Pole plasm from a fertilized egg of genotype
signal, flies made transgenic with this DNA construct have oskar mRNA localized at
P (pink) is transferred to the anterior end
the anterior end of the egg (Fig. 9.2).
of an early cleavage-stage embryo of
In the nematode, a germ-cell lineage is set up at the end of the fourth cleavage divi- genotype Y (yellow). After cellularization,
sion, with all the germ cells being derived from the P4 blastomere (see Section 6.1). cells containing pole plasm induced at the
The P4 cell is derived from three stem-cell-like divisions of the P1 cell. At each of anterior end of embryo Y are transferred to
these divisions, one daughter produces somatic cells, whereas the other divides again the posterior end (a site from which germ
to produce a somatic cell progenitor and a P cell. The egg contains P granules in its cells can migrate into the gonad) of another
cytoplasm that become asymmetrically distributed before each cleavage division, and embryo, of genotype G (green). The adult fly
are subsequently confined to the P-cell lineage (Fig. 9.3). The association of germ-cell that develops from embryo G contains germ
formation with the P granules suggested that they might have a role in germ-cell spec- cells of genotype Y as well as those of G.

ification, and at least one P-granule component, the product of the pgl-1 gene, has
been shown to be necessary for germ-cell development. PGL-1 protein may specify
germ cells by regulating some aspect of mRNA metabolism.
In both the fly and the nematode, repression of transcription is necessary to specify
a germ-cell fate. In the nematode, the gene pie-1 is involved in maintaining the stem-
cell property of the P blastomeres. It encodes a nuclear protein that is expressed
maternally and is not a component of P granules; the PIE-1 protein is present only in
the germline blastomeres, and represses new transcription of zygotic genes in these
blastomeres until PIE-1 disappears at around the 100-cell stage. This general repres-
sion may protect the germ cells from the actions of transcription factors that promote
development into somatic cells. In the absence of zygotic transcription, maternal
RNAs encode the germ-cell components.
332 Chapter 9 Germ cells, fertilization, and sex

bicoid gene female pronucleus female and male pronuclei AB cell P1 cell
DNA A P

RNA 5’ 3’
3’ untranslated region
containing bicoid localization signal
male pronucleus P granules cleavage furrow

oskar gene

P0

Transgenic fly containing DNA construct

AB P1

somatic
cells EMS P2

A P
oskar with normal oskar
bicoid 3’ MS E C P3
localization signal localization

somatic somatic somatic

cells cells cells D P4
Germ cells develop at both ends

pole cells induced at pole cells

anterior end of embryo
P granules and PIE-1 protein somatic germ cells
cells

Fig. 9.3 P granules and PIE-1 protein become asymmetrically distributed to germline
cells during cleavage of the nematode egg. Before fertilization, P granules are distributed
throughout the egg. After fertilization, P granules become localized at the posterior end of the
egg. At the first cleavage, they are only included in the P1 cell (top panel), and thus become
confined to the P cell lineage. The PIE-1 protein is only present in P cells. All germ cells are
Fig. 9.2 The gene oskar is involved in
derived from P4, which is formed at the fourth cleavage.
specifying the germ plasm in Drosophila.
In normal eggs, oskar mRNA is localized at the Xenopus and zebrafish also have distinct germ plasm. In Xenopus, distinct yolk-free
posterior end of the embryo, whereas bicoid patches of cytoplasm aggregate at the yolky vegetal pole after fertilization. When the
mRNA is at the anterior end. The localization blastomeres at the vegetal pole cleave, this cytoplasm is distributed asymmetrically,
signals for both bicoid and oskar mRNA are in
so that it is retained only in the most vegetal daughter cells, from which the germ
their 3′untranslated regions. By manipulating
cells are derived. Ultraviolet irradiation of the vegetal cytoplasm abolishes the forma-
the Drosophila DNA, the localization signal of
tion of germ cells, and transplantation of fresh vegetal cytoplasm into an irradiated
oskar can be replaced by that of bicoid (top
panel). A transgenic fly is made containing egg restores germ-cell formation. At gastrulation, the germ plasm is located in cells
the modified DNA. In its egg, oskar becomes in the floor of the blastocoel cavity, among the cells that give rise to the endoderm.
localized at the anterior end (middle panel). Cells containing the germ plasm are, however, not yet determined as germ cells and
The egg therefore has oskar mRNA at both can contribute to all three germ layers if transplanted to other sites. At the end of gas-
ends, and germ cells develop at both ends trulation, the primordial germ cells are determined, and migrate out of the presump-
of the embryo, as shown in the photograph tive endoderm and into the genital ridge, which develops from mesoderm lining the
(bottom panel). Thus, oskar alone is sufficient abdominal cavity and will form the gonads. However, there are quite big differences
to initiate the specification of germ cells. in germ-cell origin in different groups of amphibians. In the Urodeles—the tailed
Photograph courtesy of R. Lehmann. amphibians that include the newts—there is no evidence for cytoplasmic localization
of germ plasm, and the germ cells arise from lateral plate mesoderm. The evolution-
ary significance of this difference is not known.
 The development of germ cells 333

Distribution of germ plasm to specify germ cells in the zebrafish embryo

Zygote Early 4-cell stage 8-cell stage 32-cell stage 512-cell stage Sphere stage

cytokinetic ring cleavage furrow

Fig. 9.4 Distribution of germ plasm in

In the zebrafish, germ plasm containing a number of maternal mRNAs is present in
zebrafish embryo during cleavage. In this
several locations in the fertilized egg, and becomes localized to the distal ends of the
bird’s-eye view of the animal pole of the egg,
cleavage furrows in the first few cleavages. Its fate can be followed during early devel- maternal vasa mRNA (blue) is a marker for
opment by tracking one of its components, such as maternal vasa mRNA (Fig. 9.4). the germ plasm. The vasa mRNA becomes
By the 32-cell stage, germ plasm has been segregated into four cells. These cells then localized to the cleavage furrows at first
continue to divide asymmetrically, with the germ plasm segregating to one daughter and second divisions. As a furrow completes
cell at each division, so that even by the 512-cell stage there are still only four prospec- division the vasa mRNA forms tight clumps
tive germline cells in the embryo. At the sphere stage of the embryo (about 3.8 hours at the distal ends, at or just outside the cell
post-fertilization), the germ plasm expands to fill the cells, and equal divisions begin, boundary but under the yolk cell membrane.
eventually producing a total of around 30 primordial germ cells, which will migrate to At the 32-cell stage, the aggregates ingress
the prospective ovaries or testes. into four adjacent cells and segregate
asymmetrically at subsequent divisions,
ending up in four widely separated cells
9.2 In mammals germ cells are induced by cell–cell interactions at the 512-cell stage. At the sphere stage
during development (cleavage cycle 13), the vasa mRNA becomes
evenly distributed throughout the cells, and
There is no evidence for germ plasm in the mouse or other mammals or in the chick. subsequent cell divisions produce a total of
Indeed, it seems that, although germ plasm is present in several of our developmental around 30 primordial germ cells in the four
model organisms, it is the less prevalent mode of germline specification in animals gen- locations.
erally. Germ-cell specification in the mouse involves cell–cell interactions, as cultured Adapted from: Pelegri, F.: 2003.
embryonic stem cells can, when injected into the inner cell mass, give rise to both
germ cells and somatic cells (see Fig. 3.34). Mammalian germ cells were previously
identified mainly by their high concentration of the enzyme alkaline phosphatase,
which provides a convenient means of detecting them histochemically. Now that genes
involved in germline specification have been identified, it is possible to identify pri-
mordial germ cells at a much earlier stage by detecting the expression of such genes.
The earliest detectable primordial germ cells can be identified in the mouse proximal
epiblast just before the beginning of gastrulation. They form a cluster of six to eight cells
expressing the transcriptional repressor protein Blimp1, which is thought to specifically
repress genes associated with the somatic developmental program. The Blimp1-express-
ing cells are first found in the proximal epiblast at E6.25 days, in a region of prospective
extra-embryonic mesoderm that will, after cell movement, be adjacent to the beginning
of the primitive streak (Fig. 9.5). The Blimp1-positive cells proliferate and express the
transmembrane protein Fragilis, which may help to demarcate germ cells from their
somatic neighbors, and the protein Stella, which is exclusive to the germ-cell lineage.
By E7.5 days there are around 40 Stella-positive cells in the primitive streak and
these represent the full complement of primordial germ cells that will eventually
migrate to the mouse gonads (Fig. 9.5, third panel). Unlike somatic cells at this stage,
the Hox genes are repressed in the primordial germ cells, which may explain their
escape from a somatic cell fate and their retention of totipotency. Primordial germ
cells also express Oct4, which as we shall see in Chapter 10 is a transcription factor
associated with the maintenance of pluripotency in stem cells. The earliest events in
establishing human germ cells remain unknown, but migratory germ cells have been
reported at the fourth week of human embryonic development.
334 Chapter 9 Germ cells, fertilization, and sex

E6.25 days E7.25 days E8.5 days

extra-embryonic allantois
high Blimp1 and Stella
ectoderm
amnion

allantoic
bud PGCs

visceral epiblast
endoderm primitive visceral
streak endoderm embryo

Fig. 9.5 Germ-cell formation in the

mouse. First panel: a small number of 9.3 Germ cells migrate from their site of origin to the gonad
primordial germ cells (PGCs) (white)
In many animals, the primordial germ cells develop at some distance from the gonads,
expressing Blimp1 are first detectable
in the proximal epiblast at 6.25 days and only later migrate to them, where they differentiate into eggs or sperm. The rea-
after fertilization (E6.25). Second panel: son for this separation of site of origin from final destination is not known, but it may
during gastrulation, these cells and their be a mechanism for excluding germ cells from the general developmental upheaval
surrounding prospective extra-embryonic involved in laying down the body plan, or a mechanism for selecting the healthiest
mesoderm move to the posterior end of the germ cells, namely those that survive migration. The migration path of primordial germ
embryo above the primitive streak, where cells is controlled by their environment; in Xenopus, for example, primordial germ cells
the PGCs start also to express the germ- transplanted to the wrong place in the blastula do not end up in the gonad.
cell lineage gene stella. By E7.25, around The vertebrate gonad develops from mesoderm lining the abdominal cavity, which is
40 PGCs (orange) are present in the primitive
known as the genital ridge. In Xenopus, primordial germ cells originating in the floor of
streak. Third panel: at around E8.5, PGCs
the blastocoel migrate to the future gonad along a cell sheet that joins the gut to the geni-
start to migrate to the gonads.
tal ridge. Only a small number of cells start this journey, dividing about three times before
arrival, so that about 30 germ cells colonize the gonad—about the same number as in the
zebrafish. In contrast, the number of primordial germ cells that arrive at the genital ridge
of the mouse is about 8000, starting from a population of around 40 at the posterior end
of the primitive streak (see Fig. 9.5). In the mouse, the primordial germ cells enter the
Dorsal
hindgut, and migrate along the dorsal wall until they reach the genital ridges (Fig. 9.6).
neural tube In chick embryos, the pattern of migration is different again: the germ cells originate
in the epiblast and then migrate to the head end of the embryo. Most arrive at their
final destination in the gonads by circulating in the blood, leaving the bloodstream at
the hindgut and then migrating along an epithelial sheet to the gonad.
somite In zebrafish, small clusters of primordial germ cells are specified in four separate
positions relative to the embryonic axis (see Fig. 9.4). These cells migrate to form
genital bilateral lines of cells in the trunk at the level of somites 1–3 at around 6 hours post-
ridge
fertilization, and then reach their final position at the level of somites 8–10 at around
dorsal 12 hours, where they and the surrounding somatic cells form the gonads (Fig. 9.7).
mesentery hindgut
gut Zebrafish primordial germ cells face a challenging journey, as they originate in four
separate locations, and it may be no coincidence that the journey seems to be broken
Ventral
up into a number of distinct stages. Germ cells taken from the margin of a mid-
­blastula embryo can even find their way to the gonad after transplantation to the head
Fig. 9.6 Pathway of primordial germ-cell region (the animal pole) of a similar-stage embryo.
migration in the mouse embryo. In the final
stage of migration, the cells move from the
9.4 Germ cells are guided to their final destination by chemical signals
gut tube into the genital ridge, via the dorsal
mesentery. All migrating germ cells are continuously receiving signals for guidance, survival, and
After Wylie, C.C., et al.: 1993. proliferation from the tissues through which they migrate. Some of these signals have
 The development of germ cells 335

Shield-stage embryo 2-somite stage 6-somite stage 10-somite stage

site of
prospective
SDF-1a gonad

Fig. 9.7 Primordial germ cells in zebrafish

been identified; if they are lacking, germ cells are absent from the gonads or their num- migrate to the gonads in two stages. The
bers are greatly reduced. The main guidance cue in both zebrafish and in mice seems to diagrams show a dorsal view of the embryo.
be the chemoattractant protein SDF-1. In zebrafish, SDF-1 is secreted into the extracel- The chemoattractant SDF-1a (yellow) is
lular matrix by tissue cells in the regions through which the germ cells migrate and is expressed initially in a broad domain that
thought to be the long-range guidance cue for the clusters of primordial germ cells to includes the sites at which the primordial
reach their final positions from their different starting points (see Fig. 9.7). Experiments germ cells (PGCs, blue) are formed (see
that change the spatial expression of SDF-1 change the germ-cell migration patterns cor- Fig. 9.4). PGCs migrate towards the region of
respondingly. Also, knockdown by antisense RNA of either SDF-1 or its receptor CXCR4, highest SDF-1a, which is initially at the level
which is expressed on the surface of primordial germ cells, disrupted migration. of the first somite. Alterations in the pattern
of SDF-1a expression then cause the PGCs
In the mouse, the primordial germ cells in the primitive streak migrate anteriorly and
to migrate posteriorly to the level of the
become incorporated into the hindgut, which they eventually leave to migrate into the
eighth and tenth somites, where the gonads
genital ridge (see Fig. 9.6). The mouse version of SDF-1 seems to act as a guidance cue develop.
for this final part of the journey at least, as in animals lacking either SDF-1 or its recep-
tor, most germ cells get no further than the hindgut. The cell-surface receptor Kit, which
is also produced in migrating germ cells, and its ligand the Steel protein, produced by
the tissue cells over which the germ cells migrate, seem to be required for germ-cell
survival and proliferation rather than for guidance. SDF-1 and CXCR4 are not only of
interest to embryos; they are also expressed in some invasive human tumors, where
they contribute to the ability of the tumor cells to spread to other sites in the body.
In Drosophila, primordial germ cells are formed next to the posterior midgut, and
during gastrulation they are carried along the dorsal side of the embryo with the
midgut primordium. They then migrate across the dorsal midgut epithelium to the
adjacent mesoderm to join with gonadal precursor cells in forming the gonad. Screen-
ing for mutations that affect the direction of germ-cell migration has identified genes
that keep the cells on track. The gene wunen is involved in repelling germ cells from
the rest of the gut, and so preventing them from dispersing before they reach the
gonadal mesoderm, whereas expression of the enzyme HMGCoA reductase in the pro-
spective gonad is needed for the germ cells to be attracted to the prospective gonad.
Whether this enzyme is involved in producing a chemoattractant, or has some other
role to play, is so far not known.

9.5 Germ-cell differentiation involves a halving of chromosome number

by meiosis

The sperm and egg are haploid—that is, they contain just one copy of each chromo-
some in contrast to the two copies present in the diploid somatic cells. At fertilization,
therefore, the diploid chromosome number is reinstated. The primordial germ cells
described in the previous sections are diploid, and reduction from the diploid to the
haploid state during germ-cell development occurs during the specialized nuclear
division known as meiosis, which does not occur until after germ cells have reached
the gonads (Fig. 9.8). Meiosis comprises two cell divisions, in which the chromo-
somes are replicated before the first division, but not before the second, so that their
336 Chapter 9 Germ cells, fertilization, and sex

Fig. 9.8 Meiosis produces haploid cells. Meiosis reduces the number of chromosomes
Meiosis
from the diploid to the haploid number. Only one pair of homologous chromosomes is
Undifferentiated shown here for simplicity. Before the first meiotic division the DNA replicates, so that each
diploid paternal
germ cell homolog chromosome entering meiosis is composed of two identical chromatids. The paired homologous
chromosomes (known as a bivalent) undergo crossing over and recombination, and align on the
maternal meiotic spindle at the metaphase of the first meiotic division. The homologous chromosomes
homolog
separate, and each is segregated into a different daughter cell at the first cell division. There
is no DNA replication before the second meiotic division. The daughter chromatids of each
DNA replication
chromosome separate and segregate at the second cell division. The chromosome number of
the resulting daughter cells is thus halved.

chromatid
number is reduced by half. The cell divisions during meiosis can be unequal and give
rise to a small structure called a polar body (Box 9A, p. 338).
During prophase of the first meiotic division, replicated homologous chromosomes
Pairing of pair up and undergo recombination, in which corresponding DNA sequences are
homologous
chromosomes exchanged between the homologs (see Fig. 9.8). Homologous chromosomes usually
each carry different versions, or alleles, of many of the genes, and so meiotic recombi-
Prophase
of first nation generates chromosomes with new combinations of alleles. Meiosis, therefore,
meiotic Crossing over
division and results in gametes whose chromosomes carry different combinations of alleles com-
recombination
pared with the parent. This means that when two gametes come together at fertiliza-
bivalent tion, the resulting animal will differ in genetic constitution from either of its parents.
So although we might resemble our parents, we never look exactly like them. Meiotic
recombination is the main source of the genetic diversity present within populations
First of sexually reproducing organisms, including humans and other vertebrates.
metaphase Development of eggs and sperm follow different courses, even though they both
involve meiosis, and the course is determined by the sex of the embryo. The develop-
ment of the egg is known as oogenesis, and the main stages of mammalian oogenesis
are shown in Fig. 9.9 (left panel). In mammals, germ cells undergo a small number of
proliferative mitotic cell divisions as they migrate to the gonad. In the case of developing
eggs, the diploid germline cells, now known as oogonia, continue to divide mitotically
for a short time in the ovary. After entry into meiosis, they are called primary oocytes.
Within the ovary, individual oocytes are each enclosed in a sheath of somatic ovarian
cells called the follicle. Mammalian primary oocytes enter meiosis in the embryo but
become arrested in the prophase of the first meiotic division, the stage at which homolo-
Cell division 1 gous chromosomes pair up and recombination occurs (see Fig. 9.8). Arrest depends on
high levels of cyclic AMP within the oocyte, which are generated as a result of signaling
from G-protein-coupled receptors. The first meiotic division will not be completed until
Second
metaphase after ovulation in the adult, and the second meiotic division not until after fertilization.
Oocytes never proliferate again after entry into meiosis; thus, the number of oocytes
at this embryonic stage is generally considered to be the maximum number of eggs a
female mammal can ever have. In humans, most oocytes degenerate before puberty,
leaving about 400,000 out of an original 6–7 million to last a lifetime. This number
declines with age, with the decline becoming steeper after the mid-30 s until meno-
pause, typically in the 50 s (Fig. 9.10). In mammals and many other vertebrates, oocyte
development is held in suspension after birth for months (mice) or years (humans).
When the female becomes sexually mature, the oocytes start to undergo further devel-
opment, or maturation, as the result of hormonal stimuli. During maturation they grow
up to 10-fold in size in mammals, and very much larger in some animals, such as frogs.
Cell division 2
In some vertebrates, such as frogs and fish, large numbers of oocytes become mature
and are released, or ovulated, simultaneously at the end point of each reproductive
cycle. In mammals, only a relatively small number of oocytes mature during each cycle
and are released under the influence of luteinizing hormone. In mammals, meiosis
resumes at the time of ovulation; the released oocyte completes the first meiotic divi-
Haploid cells sion, with the production of one polar body, and ­proceeds as far as the metaphase
of the second meiotic division. Here meiosis is arrested again, and is only completed
 The development of germ cells 337

Oogenesis Spermatogenesis

Embryo Embryo
Primordial germ cells Primordial germ cells
multiply during multiply during
migration migration

Enters ovary Enters testis

Some mitotic
divisions of oogonia Arrest in G1
within ovary of cell cycle

After birth

Diploid spermatogonia
multiply by mitosis
in testis

No further multiplication
in embryo Stem
cell
Primary oocyte

Meiosis becomes
arrested in prophase
of division 1

Adult
Spermatocyte
Fig. 9.9 Oogenesis and spermatogenesis
in mammals. Left panel: after the germ
cells that form oocytes enter the embryonic
ovary, they divide mitotically a few times and
Further growth Meiotic division 1 then enter the prophase of the first meiotic
and development
of primary oocyte division. No further cell multiplication occurs.
Further development occurs in the sexually
egg coat mature adult female. This includes a 10-fold
increase in mass, the formation of external
cell coats, and the development of a layer of
cortical granules
Meiotic division 2 cortical granules located under the oocyte
plasma membrane. In each cycle, a group
Completion of of follicles starts to grow, oocyte growth
Ovulation division 1 of Spermatids
meiosis and maturation follows; a few eggs are
ovulated but most degenerate. Eggs continue
Secondary oocyte
to mature in the ovary under hormonal
influences, but become blocked in the
first second metaphase of meiosis, which is only
polar body Differentiation completed after fertilization. Polar bodies are
formed at meiosis (see Box 9A, p. 338). Right
Completion of panel: germ cells that develop into sperm
Fertilization division 2 enter the embryonic testis and become
of meiosis
arrested at the G1 stage of the cell cycle.
After birth, they begin to divide mitotically
second again, forming a population of stem cells
polar body (spermatogonia). These give off cells that
Mature sperm
then undergo meiosis and differentiate into
sperm. Sperm can therefore be produced
indefinitely.
338 Chapter 9 Germ cells, fertilization, and sex

Box 9A Polar bodies

Polar bodies are small cells formed by meio- Primary diploid First metaphase Second metaphase Haploid egg
sis during the development of an oocyte oocyte
into an egg. In this highly schematic illus-
First Second Second polar body
tration, the segregation of only one pair of meiotic First polar body meiotic
division division
chromosomes is shown for simplicity. There
are two cell divisions associated with meio-
sis, and one daughter from each division is
almost always very small compared with the
other, which becomes the egg—hence the
term polar bodies for these smaller cells. • Flatworms • Molluscs • Amphibians • Coelenterates
• Roundworms • Insects • Mammals • Echinoderms
The timing of meiosis in relation to the
development of the oocyte varies in dif- Examples of organisms that are fertilized at the stage of meiosis indicated
ferent animals, and in some species meio-
sis is completed and the second polar body formed only after little importance for later ­development, but in some animals the
­fertilization ­(see figure). In general, polar-body formation is of site of formation is a useful marker for the embryonic axes.

after fertilization, with the production of the second polar body. The egg and sperm
pronuclei then fuse to form the zygote nucleus. The time of fertilization in relation to
the stage of oocyte development differs in different animal groups (see Box 9A).
A major error in human oocyte meiosis results in the oocyte having an extra chro-
mosome 21, known as a trisomy of chromosome 21, as the result of non-disjunction of
homologous chromosomes 21 at the first meiotic division. This trisomy is the cause of
Down syndrome, and is one of the most common genetic causes of congenital malfor-
mations and learning disability. Trisomies and other errors in chromosome segregation
are relatively common in human eggs, far more so than in human sperm, and their
incidence rises exponentially as oocytes age. The majority of human embryos with too
many or too few chromosomes are not viable, because of the imbalance in the expres-
sion of large numbers of genes. They undergo developmental arrest at the preimplan-
tation stage, or fail to implant, or are spontaneously aborted during gestation.
Decline of human oocyte numbers with age The strategy for spermatogenesis—the production of sperm—is quite different
from oogenesis. Diploid germ cells that give rise to sperm do not enter meiosis in the
Oocyte 106 embryo, but become arrested at an early stage of the mitotic cell cycle in the embry-
population onic testis. They resume mitotic proliferation after birth. Later, in the sexually mature
(log 10) 5
10 animal, spermatogonial stem cells give rise to differentiating spermatocytes, which
undergo meiosis, each forming four haploid spermatids that mature into sperm (see
104
Fig. 9.9, right panel). Thus, unlike the fixed number of oocytes in female mammals,
sperm continue to be produced throughout the life of the organism.
103
birth 50 In Drosophila, there is a continuous production of both eggs and sperm from a
Age (years) population of stem cells in females and males, respectively. Oogenesis begins with
the division of a stem cell (see Fig. 2.16), and so there is no intrinsic limitation to the
Fig. 9.10 Decline in human oocyte
number of eggs that a female fly can produce.
numbers with age. The graph shows the
changes in numbers of oocytes in the human 9.6 Oocyte development can involve gene amplification and contributions
ovary with age, peaking at about 6–7 million from other cells
(in early fetal development) and then
decreasing by apoptotic cell death. At puberty
Eggs vary enormously in size among different animals, but they are always larger than
there remain about 400,000 oocytes and the somatic cells. A typical mammalian egg cell is about 0.1 mm in diameter, a frog’s
only a very small number of these, between egg about 1 mm, and a hen’s egg about 3 cm (i.e. the yolk; the white of the egg is
400 and 500, are released throughout a extracellular material) (see Fig. 3.2). To achieve growth to such a large size, a variety
woman’s lifetime. of mechanisms have evolved, some organisms using several of them together. One
 The development of germ cells 339

strategy is to increase the overall number of gene copies in the developing oocyte, as
this proportionately increases the amount of mRNA that can be transcribed, and thus
the amount of protein that can be synthesized. Vertebrate oocytes, for example, are
arrested in prophase of the first meiotic division and so have double the normal diploid
number of genes. Transcription and oocyte growth continue while meiosis is halted.
In addition, insects and amphibians produce many extra copies of selected genes
whose products are needed in very large quantities in the egg. During amphibian
­oocyte development, the rRNA genes are amplified from hundreds of copies to millions;
their subsequent translation produces enough ribosomal proteins to ensure sufficient
ribosomes for protein synthesis during oocyte growth and early embryonic develop-
ment. In insects, the genes that encode the proteins of the egg membrane, the chorion,
become amplified in the surrounding follicle cells, which produce the egg membrane.
Yet another strategy is for the oocyte to rely on the synthetic activities of other cells.
In insects, the nurse cells adjacent to the oocyte make many mRNAs and proteins and
deliver them to the oocyte (see Fig. 2.16). Yolk proteins in birds and amphibians are
made by liver cells, and are carried by the blood to the ovary, where the proteins enter
the oocyte by endocytosis and become packaged into yolk platelets. In C. elegans,
yolk proteins are made in intestinal cells, and in Drosophila in the fat body, from
where they are transported to the oocytes. In amphibian eggs, the oocyte is polarized
from an early stage onwards, and the yolk platelets accumulate at the vegetal pole.
In animals that depend on maternal determinants for their early development, these
mRNAs are produced by the mother and delivered into the egg, where they become
localized to the appropriate position by microtubule-based transport.

9.7 Factors in the cytoplasm maintain the totipotent potential of the egg
Like all other cells in the body, mature gametes are specialized cells that have under-
gone a program of differentiation. As we shall consider in more detail in Chapter 10,
except in rare instances, cell differentiation does not involve any alteration in the
sequence or amount of DNA, but instead involves epigenetic changes. These are
chemical modifications to DNA and to the chromosomal proteins associated with it
that change the structure of the chromosome locally so that some genes are selec-
tively silenced while others can still be expressed. Some somatic cell types, such as
muscle cells and nerve cells, do not divide further once differentiation is complete. In
cell types that retain the ability to divide, such as liver cells (hepatocytes) or connec-
tive tissue fibroblasts, the epigenetic changes associated with that cell type are passed
on to the daughter cells, so that they are also hepatocytes or fibroblasts, respectively.
Once differentiation is under way, none of these somatic cell types can in normal cir-
cumstances give rise to a completely different type of differentiated cell, let alone to
the whole range of cell types needed to build a new organism.
Unlike differentiated somatic cells, mature gametes must retain totipotent potential,
so that after fertilization their combined genomes can direct the development of a
complete individual. In the egg, factors present in the cytoplasm are responsible for
maintaining the egg genome in a totipotent state. This is demonstrated most dramati-
cally by the ability of the cytoplasm of an unfertilized egg to ‘reprogram’ the nucleus
from a differentiated somatic cell so that it reverts to a totipotent state and is able to
direct the development of a new organism. This forms the basis of animal cloning by
somatic cell nuclear transfer, which has been achieved in amphibians and some spe-
cies of mammals, including mice, and is described in Chapter 10.

9.8 In mammals some genes controlling embryonic growth are ‘imprinted’

Cloning mammals turned out to be much more difficult to achieve than some other ani-
mals, and this is most probably due to the phenomenon of genomic imprinting—by
which certain genes are switched off in either the egg or the sperm during their
340 Chapter 9 Germ cells, fertilization, and sex

Fig. 9.11 Paternal and maternal genomes Androgenetic—

Biparental Gynogenetic—
are both required for normal mouse two maternal genomes two paternal genomes
development. A normal biparental embryo
has contributions from both the paternal fertilized
and maternal nuclei in the zygote after pronuclei mouse egg
fertilization (left panel). Using nuclear
transplantation, an egg can be constructed
extra-embryonic
with two paternal or two maternal nuclei tissues
from an inbred strain. Embryos that develop
from an egg with two maternal genomes—
gynogenetic embryos (center panel)—have
underdeveloped extra-embryonic structures.
This results in development being blocked,
although the embryo itself is relatively
normal and well developed. Embryos that small placenta results in block
in embryonic development embryonic growth retarded
develop from eggs with two paternal
genomes—androgenetic embryos (right
panel)—have normal extra-embryonic
structures, but the embryo itself only
develops to a stage where a few somites
­ evelopment and remain silenced in the genome of the early embryo. Evidence for
d
have formed. imprinting in mammals came initially from the demonstration that the maternal and
paternal genomes make different contributions to embryonic development.
Mouse eggs can be manipulated by nuclear transplantation to have either two
paternal genomes or two maternal genomes, and can be reimplanted into a mouse for
further development. The embryos that result are known as androgenetic and gyno-
genetic embryos, respectively. Although both kinds of embryo have a diploid number
of chromosomes, their development is abnormal. The embryos with two paternal
genomes have well-developed extra-embryonic tissues, but the embryo itself is abnor-
mal, and does not proceed beyond a stage at which several somites are present. By
contrast, the embryos with diploid maternal genomes have relatively well-developed
embryos, but the extra-embryonic tissues—placenta and yolk sac—are poorly devel-
oped (Fig. 9.11). These results clearly show that both the maternal and the paternal
genome are necessary for normal mammalian development: the two parental genomes
make different contributions, and both are required for the normal development of
the embryo and the placenta. This is the apparent reason that mammals cannot be
naturally produced parthenogenetically, by activation of an unfertilized egg.
Such observations suggested that the paternal and maternal genomes must be
epigenetically modified during germ-cell differentiation. The paternal and maternal
genomes contain the same set of genes, but the imprinting process turns on or off
certain genes in either the sperm or the egg, so that they are, or are not, expressed
during development. Imprinting implies that the affected genes carry a ‘memory’ of
being in a sperm or an egg.
Imprinting is a reversible process. The reversibility is important, because in the next
generation any of the chromosomes could end up in male or female germ cells. Inher-
ited imprinting is probably erased during early germ-cell development, and imprinting
is later established afresh during germ-cell differentiation. When mammals are cloned
using donor nuclei from differentiated somatic cells, these nuclei will not have gone
through the normal reprogramming and imprinting process that occurs during germ-
cell formation, and this could account for the large number of failures and abnormali-
ties in the cloned embryos.
The imprinted genes affect not only early development but also the later growth of
the embryo. Further evidence that imprinted genes are involved in embryonic growth
comes from studies on chimeras made between normal embryos and androgenetic
or gynogenetic embryos. When inner cell mass cells from gynogenetic embryos are
injected into normal embryos, growth is retarded by as much as 50%. But when
­androgenetic inner cell mass cells are introduced into normal embryos, the chimera’s
 The development of germ cells 341

growth is increased by up to 50%. The imprinted genes on the male genome thus
lgf2 H19
significantly increase the growth of the embryo.
At least 80 imprinted genes have been identified in mammals, some of which encode Male ON OFF
non-coding RNAs (ncRNAs). Some imprinted genes are involved in growth control.
The insulin-like growth factor IGF-2 is required for embryonic growth; its gene, Igf2,
Female OFF ON
is imprinted in the maternal genome—that is, it is turned off, so that only the paternal
gene is active (Fig. 9.12). Direct evidence for the Igf2 gene being imprinted comes
from the observation that when sperm carrying a mutated, and thus defective, Igf2 Fig. 9.12 Imprinting of genes controlling
gene fertilize a normal egg, small offspring result. This is because only very low levels embryonic growth. In mouse embryos, the
of IGF-2 are produced from the imprinted maternal gene and this is not enough to paternal gene for insulin-like growth factor 2
(Igf2) is on, but the gene on the maternal
make up for the loss of IGF-2 expression from the paternal genome. In contrast, if the
chromosome is off. In contrast, the closely
non-functional mutant gene is carried by the egg’s genome, development is normal,
linked gene H19 is switched on in the
with the required IGF-2 activity being provided from the normal paternal gene. maternal genome but silenced in the paternal
Closely linked to the Igf2 gene on mouse chromosome 7 is the gene H19 (of genome.
unknown function), which is imprinted in the opposite direction. It is expressed from
the maternal chromosome but not from the paternal one (see Fig. 9.12). By manipu-
lating the chromosomes of an egg it was possible to make a diploid egg with two
maternal genomes but with one genome having the normal female imprinting of Igf2
and H19 while in the other Igf2 was not imprinted, as in the male germ cell (H19 was
absent in this genome). This egg could develop normally, unlike an embryo with two
female genomes.
A possible evolutionary explanation for the reciprocal imprinting of genes that con-
trol growth invokes parental-conflict theory: the theory that the reproductive strate-
gies of the father and mother are different. Paternal imprinting promotes embryonic
growth, whereas maternal imprinting reduces it, for example. The father wants to have
maximal growth for his own offspring, so that his genes have a good chance of surviv-
ing and being carried on. This can be achieved by having a large placenta, as a result
of producing growth hormone, whose production is stimulated by IGF-2. The mother,
who may mate with different males, benefits more by spreading her resources over all
her offspring, and so wishes to prevent too much growth in any one embryo. Thus,
a gene that promotes embryonic growth is turned off in the mother. Paternal genes
expressed in the offspring could be selected to extract more resources from mothers,
because an offspring’s paternal genes are less likely to be present in the mother’s other
children. There are, however, many effects of imprinted genes other than on growth.
Imprinting occurs during germ-cell differentiation, and so a mechanism is required
both for maintaining the imprinted condition throughout development and for wiping
it out during the next cycle of germ-cell development. One mechanism for maintaining
imprinting is DNA methylation, an epigenetic modification in which methyl groups
are attached to cytosines in DNA. As we shall see in Chapter 10, DNA methylation is
associated with gene silencing. Evidence that DNA methylation is required for imprint-
ing comes from mice in which the methylation process is aberrant. In these mice, the
Igf2 gene is no longer imprinted and is expressed from the maternal chromosome as
well as the paternal one. In addition to DNA methylation, other factors that have been
implicated in imprinting are non-coding regulatory RNAs (ncRNAs), repressor proteins
of the Polycomb group, and the chemical modification of histone proteins, the proteins
that help package DNA in the chromosomes. Some of these ncRNAs are involved in
imprinting adjacent genes. The ncRNA Air, for example, is encoded in the Igfr cluster
of imprinted genes and is necessary for their paternal imprinting.
A number of developmental disorders in humans are associated with imprinted
genes. One is Prader–Willi syndrome, which is linked to a loss of expression of the
paternal copy of a gene on chromosome 15, usually as a result of a deletion of a small
region of the chromosome containing that gene. Infants fail to thrive and later can
become extremely obese; they also show mental retardation and mental disturbances
such as obsessional-compulsive behavior. Angelman syndrome results from the loss
342 Chapter 9 Germ cells, fertilization, and sex

of the same region of the maternal chromosome 15. The effect in this case is severe
motor and mental retardation. Beckwith–Wiedemann syndrome is due to a general-
ized disruption of imprinting on a region of chromosome 7. There is excessive fetal
overgrowth and an increased predisposition to cancer.

Summary

In many animals, the germline cells are specified by localized cytoplasmic determinants
in the egg, whose localization is controlled in the mother by cells surrounding the
­oocyte. In contrast, germ cells in mammals are not specified by maternal determinants,
but by intercellular interactions in the embryo. Once the germ cells are determined,
they migrate from their site of origin to the gonads, where further development and
differentiation takes place. In the gonads, the diploid germ-cell precursors undergo
meiosis, eventually producing the haploid germ cells—eggs and sperm. The number of
oocytes in a female mammal is fixed before birth, whereas sperm production in male
mammals is continuous throughout adult life. Eggs are always larger than somatic cells,
and in some animal groups are very large indeed. In order to achieve their increased
size, specialized somatic cells surrounding the developing oocyte may provide some of
the constituents, such as yolk; in addition, some genes producing materials required in
large amounts may be amplified in the oocyte.
Both maternal and paternal genomes are necessary for normal mammalian develop-
ment. Embryos with diploid maternal or paternal genomes develop abnormally. Certain
genes in eggs and sperm are imprinted, so that the activity of the same gene is differ-
ent depending on whether it is of maternal or paternal origin. Several imprinted genes
are involved in growth control of the embryo. Improper imprinting can lead to develop-
mental abnormalities in humans.

Summary: specification of germ cells

Drosophila
germ plasm specified by oskar at posterior end of egg

C. elegans
germ cells specified germ plasm is characterized by polar granules and
by germ plasm PIE-1 protein, and segregates into P4 during cleavage

Xenopus
germ plasm localized in vegetal region of egg

Zebrafish
Mammals have no germ plasm germ plasm in cleavage furrows

Fertilization

Fertilization—the fusion of egg and sperm—is the trigger that initiates development.
In many marine organisms, such as sea urchins, sperm released into the water by the
males are attracted to eggs by chemotaxis—the sperm swim up a gradient of ­chemical
released by the egg. The membranes of the egg and sperm fuse, and the sperm nucleus
enters the egg cytoplasm, becoming the sperm pronucleus. In mammals and many
other animals, fertilization triggers the completion of meiosis in the egg. The set of
maternal chromosomes retained in the egg becomes the egg ­pronucleus, while the
other set is segregated into the second polar body (see Box 9A, p. 338). The sperm
 Fertilization 343

and egg pronuclei (see Fig. 1.1) fuse to form the zygotic nucleus, and the egg starts
dividing and embarks on its developmental program. Fertilization can be either exter-
nal, as in frogs, or internal, as in Drosophila, mammals, and birds. Of all the sperm
released by the male, only one fertilizes each egg. In many animals, including mam-
mals, sperm penetration activates a blocking mechanism in the egg that prevents any
further sperm entering. This is necessary because if more than one sperm nucleus
enters the egg there will be additional sets of chromosomes and centrosomes, result-
ing in abnormal development. In humans, embryos with such abnormalities fail to
develop. As we shall see in this chapter, there are multiple overlapping mechanisms
for ensuring that only one sperm nucleus contributes to the zygote.
Both eggs and sperm are structurally specialized for fertilization. The specializa-
tions of the egg are directed to preventing fertilization by more than one sperm, and
then to start development, while those of the sperm are directed to penetrating the
egg. The unfertilized egg is usually surrounded by several protective layers outside the
plasma membrane, and the eggs of many organisms have a layer of cortical granules
just beneath the plasma membrane, whose contents are released on fertilization and
help to block the entry of more sperm. These barriers together serve to make the egg
impenetrable to more than a small number of sperm.

9.9 Fertilization involves cell-surface interactions between egg and sperm

Sperm are motile cells, typically designed for activating the egg and at the same time
delivering their nucleus into the egg cytoplasm. They essentially consist of a nucleus,
mitochondria to provide an energy source, and a flagellum for movement (Fig. 9.13).
The anterior end is highly specialized to aid penetration. The sperm of C. elegans and
some other invertebrates are unusual as they more closely resemble normal cells and
move by ameboid motion.
After sperm have been deposited in the mammalian female reproductive tract, they
undergo a process known as capacitation, which facilitates fertilization by removing
certain inhibitory factors. There are very few mature eggs—usually one or two in
humans and about 10 in mice—waiting to be fertilized, and fewer than a hundred of
the millions of sperms deposited actually reach these eggs.
The sperm has to penetrate several physical barriers to enter the egg (Fig. 9.14). In
mammalian eggs, the first is a sticky layer of hyaluronic acid and embedded somatic
follicle cells, which are shed with the egg when it is released. These cells are called
cumulus cells, as in low-power microsopic images of a released egg it appears to be
surrounded by a ‘cloud’ of material. The first cloned mouse was called Cumulina as
she was produced using the nucleus of a cumulus cell (see Chapter 10). ­Hyaluronidase
activity on the surface of the sperm head helps it to penetrate this layer. The sperm
next encounters the zona pellucida, a layer of fibrous glycoproteins secreted by the
oocyte. This also acts as a physical barrier, but sperm are helped to penetrate it by the
acrosomal reaction—the release of enzymes contained in the acrosomal vesicle or
acrosome located in the sperm head (see Fig. 9.13).
Mammalian sperm carry various cell-surface proteins that are involved in binding
to, and penetrating, the zona pellucida. One of these is the adhesive protein SED1.
Fig. 9.13 A human sperm. The acrosomal
vesicle at the anterior end of the sperm
acrosomal midpiece plasma membrane
vesicle contains enzymes that are used to digest
the protective coats around the egg. The
plasma membrane on the head of the sperm
contains various specialized proteins that
bind to the egg coats and facilitate entry. The
nucleus mitochondria flagellum
sperm moves by its single flagellum, which is
powered by mitochondria. The overall length
Head Tail
from head to tail is about 60 µm.
344 Chapter 9 Germ cells, fertilization, and sex

1 2
Binding of
sperm to Acrosomal
zona pellucida reaction

cumulus cell layer

acrosomal vesicle
zona pellucida
follicle cell

5 3
egg nucleus
Fig. 9.14 Fertilization of a mammalian
egg. After penetrating the follicle-derived
cumulus cell layer, the sperm binds to
cortical granule
the zona pellucida (1). This triggers the
acrosomal reaction (2), in which enzymes Sperm nucleus Penetration
enters through
are released from the acrosomal vesicle and egg cytoplasm zona pellucida
break down the zona pellucida. This enables
the sperm to penetrate the zona pellucida (3),
and bind to the egg plasma membrane.
The plasma membrane of the sperm head
fuses with the egg plasma membrane (4).
This activates the egg, causing a release Fusion of
plasma
of cortical granules and the sperm nucleus membranes
then enters the egg along with its
mitochondria (5).
4
After Alberts, B., et al.: 1989.

As they progress through the epididymis of the testis, sperm become coated with this
protein, which is secreted by the epididymal epithelium. Sperm lacking SED1 fail to
bind the zona pellucida. After initial adhesion due to SED1 and other proteins, the
acrosomal reaction is triggered when proteins in the plasma membrane of the sperm
head bind to the zona pellucida glycoprotein ZP3. Binding to ZP3 triggers a signal-
ing pathway in the sperm that results in release of the contents of the acrosome by
exocytosis. The acrosomal enzymes break down the oligosaccharide side chains on
the zona pellucida glycoproteins, making a hole in the zona pellucida that enables the
sperm to approach the egg plasma membrane. In the sperm of many invertebrates,
such as the sea urchin, the acrosomal reaction also results in the extension of a rod-
like acrosomal process. This forms by polymerization of actin in the sperm cytoplasm
and facilitates contact with the egg membrane.
The acrosomal reaction also exposes proteins on the sperm surface that can bind to
the egg membrane and are involved in the fusion of sperm and egg membranes. One
of these proteins, called Izumo, has been shown to be essential for sperm–egg fusion
in mice. The ligand for Izumo on the egg plasma membrane is not yet known. The
only egg membrane protein identified so far as essential for fusion is the protein CD9,
but the sperm ligand for this receptor is similarly still unidentified.
Human and other mammalian eggs can be fertilized in culture and the very early
pre-blastocyst embryo transferred to the mother’s womb, where it implants and devel-
ops normally. This procedure of in vitro fertilization (IVF) has been of great help to
couples who have, for a variety of reasons, difficulty in conceiving. A human egg can
even be fertilized by injecting a single intact sperm directly into the egg in culture, a
technique known as intracytoplasmic sperm injection (ICSI), which is useful when
infertility is due to the sperm being unable to penetrate the egg.
 Fertilization 345

9.10 Changes in the egg envelope at fertilization block polyspermy

Although many sperm attach to the coats around the egg, and even reach the plasma
membrane, it is important that only one sperm nucleus fuses with the egg nucleus
as the unbalanced chromosomal constitution resulting from the fusion of more than
one sperm nucleus will be ultimately lethal to the embryo. Different organisms have
different ways of ensuring fertilization by only one sperm. In birds, for example, many
sperm penetrate the egg but only one sperm nucleus fuses with the egg nucleus; the
other sperm nuclei are destroyed in the cytoplasm. It is likely in this case that the
position at which the sperm enters the egg is important, with the sperm that enters
immediately into the germinal vesicle—the region containing the egg chromosomes—
acting as the fertilizing sperm. The DNA of sperm entering the cytoplasm outside this
region is probably degraded by cytoplasmic DNases, whose presence has been detected
in quail eggs. In sea urchins and mammals, on the other hand, the entry of more than
one sperm is prevented. This general strategy is called the block to polyspermy. In
many species, although not in mammals, the block to polyspermy consists of two
stages—an initial rapid reaction followed by a slower one.
The eggs of sea urchins and Xenopus have a two-stage block to polyspermy. They
are both externally fertilized and so are more likely than mammalian eggs to be bom-
barded simultaneously by large numbers of sperm. An immediate means of signaling
that one sperm has fused with the egg membrane is therefore desirable. The rapid
block to polyspermy in sea urchin is triggered within seconds by a transient depo-
larization of the egg plasma membrane that occurs on sperm–egg fusion. The electri-
cal membrane potential across the plasma membrane goes from −70 mV to +20 mV
within a few seconds of sperm entry (Fig. 9.15). The membrane potential then slowly
returns to its original level. If depolarization is prevented, polyspermy occurs, but
how depolarization blocks polyspermy is not yet clear. A similar rapid electrical block
to polyspermy occurs in Xenopus.
As the membrane repolarizes, an impenetrable membrane called the fertiliza-
tion membrane is formed around the egg. This is the result of the slow block to
polyspermy and is triggered by a wave of calcium release in the egg that is due to
sperm entry and leads to the cortical granules releasing their contents to the outside of
the plasma membrane by exocytosis. This type of reaction occurs in mammalian eggs
on fertilization, as well as in sea urchin and Xenopus. The wave of calcium also acti-
vates development, as we shall see in the next section. The unfertilized sea-urchin egg
is surrounded by a vitelline membrane, which corresponds to the zona pellucida of
mammalian oocytes. Release of the granule contents into the space between the plasma
membrane and the vitelline layer causes this layer to lift off the plasma ­membrane.
The cortical granule contents cross-link molecules in the vitelline ­membrane to pro-
duce a ‘hardened’ fertilization membrane, and also provide an additional jelly-like
hyaline layer between it and the egg’s plasma membrane (Fig. 9.16). Some of the
released enzymes also cleave sperm-binding proteins. Together, these changes ­prevent

+
Potential (mV)
+20

0 Fig. 9.15 Depolarization of the sea-urchin

egg plasma membrane at fertilization. The
resting membrane potential of the
–70
unfertilized sea-urchin egg is −70 mV. At
– fertilization, it changes rapidly to +20 mV,
and then slowly returns to the original value.
10 seconds Sperm added Fertilization membrane formation This depolarization may provide a fast block
to polyspermy.
346 Chapter 9 Germ cells, fertilization, and sex

Surface of the egg before After sperm entry, cortical granules Cortical granule material and The remaining cortical granule
sperm entry discharge and vitelline membrane the vitelline membrane material forms the hyaline layer
is raised form fertilization membrane
plasma
vitelline membrane microvillus membrane
fertilization
membrane

hyaline
layer

cortical granule

Fig. 9.16 The cortical reaction at

fertilization in the sea urchin. The egg additional sperm from gaining access to the egg. The sea-urchin fertilization mem-
is surrounded by a vitelline membrane, brane dissolves at the blastula stage, and the blastula ‘hatches’ out.
which lies outside the plasma membrane. Polyspermy is prevented in mammals both by mechanisms that regulate the num-
Membrane-bound cortical granules lie bers of sperm reaching the egg and by changes that occur to the zona pellucida and
just beneath the egg plasma membrane. egg membrane once the first sperm has entered the egg. There is no rapid electrical
At fertilization, the cortical granules fuse block to polyspermy in mammalian eggs, but there is a slow block very similar to that
with the plasma membrane, and some of in the sea urchin, which develops within 30 minutes to an hour after fertilization.
the contents are extruded by exocytosis. After the first sperm fuses with the plasma membrane, the granules in the egg cortex
These join with the vitelline membrane
are released by exocytosis and their contents form a layer immediately outside the egg
to form a tough fertilization membrane,
plasma membrane. This material modifies the chemical structure of the zona pellucida
which then lifts off the egg surface and
so that sperm can no longer bind to it. Mammalian eggs also have a membrane block
prevents further sperm entry. Other cortical
granule constituents give rise to a hyaline to polyspermy that develops over a similar timescale to the zona pellucida reaction. The
layer, which surrounds the egg under the membrane block has been demonstrated in mouse eggs whose zona pellucida has been
fertilization membrane. removed before fertilization in vitro. Challenge with a second dose of sperm after the
initial fertilization results in very few, if any, of these sperm entering the egg. The basis
for the membrane block to polyspermy in mammalian eggs is currently not known.

9.11 Sperm–egg fusion causes a calcium wave that results in egg activation

The activation of the egg at fertilization initiates a series of events that result in the
start of development. In the sea-urchin egg, for example, there is a several-fold increase
in protein synthesis, and there are often changes in egg structure, such as the cortical
rotation that occurs in amphibian eggs (see Section 4.2). Amphibian and mammalian
eggs, which are arrested in second meiotic metaphase (see Box 9A, p. 338), now
complete meiosis, after which the egg and sperm pronuclei fuse to form the diploid
zygotic genome, and the fertilized egg enters mitosis. In mice and humans, the pro-
nuclear membranes disappear before the pronuclei come together. In mammals, the
sperm mitochondria are destroyed, and so all mitochondria in the fertilized egg are of
maternal origin. The mitochondrial DNA genome is, therefore, inherited unchanged
down the female line apart from the occasional change in DNA sequence as a result
of mutation. This makes mitochondrial DNA an excellent material for studies that
use the rare DNA sequence changes accumulated in the mitochondrial genomes of
present-day humans to trace the movements of our earliest human ancestors out of
Africa and their spread to different parts of the world.
Fertilization and egg activation are associated with an explosive release of free calcium
ions (Ca2+) within the egg, producing a wave of Ca2+ that travels across it (Fig. 9.17). The
calcium release is triggered by sperm entry, and is both necessary and sufficient to initi-
ate the onset of development. In sea-urchin eggs, the wave starts at the point of sperm
entry and crosses the egg at a speed of 5–10 µm per second. In all mammals, oscillations
in calcium concentration occur for several hours after fertilization. Ca2+ release at fertil-
ization is probably triggered by the activity of a sperm-specific enzyme, phospholipase
C-ζ. This initiates a signaling pathway that leads to production of the second ­messenger
 Fertilization 347

15 seconds 19 seconds 25 seconds 31 seconds 36 seconds

Fig. 9.17 Calcium wave at fertilization.

inositol 1,4,5-trisphosphate, which acts on receptors in intracellular membranes to A series of images showing an intracellular
release calcium from stores in organelles such as the endoplasmic reticulum. calcium wave at fertilization in a sea-urchin
The sharp increase in free Ca2+ is crucial for egg activation. The eggs of many egg. The fertilizing sperm has fused just to
animals can be activated if the Ca2+ concentration in the egg cytosol is artificially the left of the top of the egg, and triggered
increased, for example, by direct injection of Ca2+. Conversely, preventing calcium the wave. Calcium ion concentration
is monitored with a calcium-sensitive
increase by injecting agents that bind it, such as the calcium chelator EGTA, blocks
fluorescent dye, using confocal fluorescence
activation. It has been known for many years that Xenopus eggs can be activated
microscopy. Calcium concentration is
simply by prodding them with a glass needle; this is due to a local influx in calcium
shown in false color: red is the highest
at the site of insertion that triggers a calcium wave. concentration, then yellow, green, and blue.
Calcium initiates the completion of meiosis in the fertilized egg by acting on proteins Times shown are seconds after sperm entry.
that control the cell cycle. The unfertilized Xenopus egg is maintained in the metaphase Photographs courtesy of M. Whitaker.
of the second meiotic division by the presence of high levels of a protein complex
called maturation-promoting factor (MPF), which is a complex of a cyclin-dependent
kinase (Cdk) and its partner cyclin. The effects of MPF are due to phosphorylation of
a variety of protein targets by the kinase. For the egg to complete meiosis, the level of
MPF activity must be reduced (Fig. 9.18). Similar Cdk–cyclin complexes control the
mitotic cell cycle (see Section 13.2). The calcium wave results in the activation of the
enzyme calmodulin-dependent protein kinase II. The activity of this kinase indirectly
results in the degradation of the cyclin component of MPF, which allows the egg to

MPF activity Progesterone stimulus Ca2+-induced cyclin degradation

Interphase arrest Metaphase arrest

male and female second

first polar body pronuclei polar body

Growth

Completion Entry into Meiosis II Fertilization Completion Mitosis Mitosis

of Meiosis I of Meiosis II
Immature
oocyte Mature oocyte Diploid zygote

Fig. 9.18 Profile of maturation-promoting factor (MPF) activity At fertilization, the calcium wave leads to completion of meiosis, and
in early Xenopus development. The immature Xenopus oocyte cell the second polar body is formed. The zygote starts to cleave rapidly by
cycle is arrested. On receipt of a hormonal progesterone stimulus it mitotic divisions. Maturation-promoting factor (MPF) rises sharply just
enters, and completes, the first meiotic division, with the formation before each division of the meiotic and mitotic cell cycles, remains high
of the first polar body. It enters the second meiotic division, but during mitosis, and then decreases abruptly and remains low between
becomes arrested again in metaphase. The egg is laid at this point. successive mitoses.
348 Chapter 9 Germ cells, fertilization, and sex

complete meiosis. The pronuclei then fuse, and the zygote moves on to the next stage
of its development, which is entry into the mitotic cell cycles of cleavage.

Summary

The fusion of sperm and egg at fertilization stimulates the egg to start dividing and
developing. Both sperm and egg have specialized structures relating to fertilization.
The initial binding of sperm to the mammalian egg is mediated by molecules in the
zona pellucida (in mammals) or the corresponding vitelline membrane (in sea urchins),
and leads to the release of the contents of the sperm acrosome, which facilitates the
penetration of the sperm through the layers surrounding the egg and allows it to reach
the egg plasma membrane. A block to polyspermy allows only one sperm to fuse with
the egg and deliver its nucleus into the egg cytoplasm. In sea urchins, the first block is
rapid and partial, and the second results from the release of egg cortical granule con-
tents to the exterior to form an impenetrable fertilization membrane. A key role in egg
activation after fertilization is played by the release of free calcium ions into the cyto-
sol, which spread in a wave from the site of sperm fusion. In mammals, and most other
vertebrates, fertilization triggers the completion of the second meiotic division; the
sperm and egg haploid pronuclei give rise to the zygote nucleus, and the egg divides.

Summary: fertilization in mammals

hyaluronidase activity on surface of mammalian sperm

sperm penetrates cumulus layer binding to zona pellucida elicits

acrosomal reaction in sperm

enzymes released from acrosome sperm head

sperm plasma membrane fuses with

egg plasma membrane

calcium waves
cortical reaction provides block to polyspermy

sperm nucleus delivered into cytoplasm

egg completes meiosis and development is initiated

Determination of the sexual phenotype

In organisms that produce two phenotypically different sexes, sexual development

involves the modification of a basic developmental program. Early development is
similar in both male and female embryos, with sexual differences only appearing at
later stages. In the organisms considered here, somatic sexual phenotype—that is,
the development of the individual as either male or female—is genetically fixed at
fertilization by the chromosomal content of the gametes that fuse to form the fertil-
ized egg. In mammals, for example, sex is determined by the Y chromosome: males
are XY, females are XX.
 Determination of the sexual phenotype 349

Even among vertebrates, however, sex is not always determined by which chro-
X X chromosome Y chromosome
mosomes are present; in alligators, it is determined by the environmental tempera-
ture during incubation of the embryo, and some fish can switch sex as adults in
response to environmental conditions. In insects, there is a wide range of different Y
sex-­determining mechanisms. Intriguing though these are, we focus here on those
organisms in which the genetic and molecular basis of sex determination is best
understood—mammals, Drosophila, and the nematode—in all of which sex is deter-
mined by chromosomal content, although by quite different mechanisms.
We first consider the determination of the somatic sexual phenotype. We then deal
with the determination of the sex of the germ cells—whether they become eggs or
sperm—and finally consider how the embryo compensates for the difference in chro-
mosomal composition between males and females. Fig. 9.19 The sex chromosomes in
humans. If two X chromosomes (XX) are
present, a female develops, whereas the
9.12 The primary sex-determining gene in mammals is on the Y chromosome presence of a Y chromosome (XY) leads to
development of a male. The inset shows a
The genetic sex of a mammal is established at the moment of conception, when the diagrammatic representation of the banding
sperm introduces either an X or a Y chromosome into the egg (Fig. 9.19). Eggs contain in the chromosomes, which represent regions
one X chromosome; if the sperm introduces another X the embryo will be female, if a of increased chromatin condensation.
Y it will be male. The presence of a Y chromosome causes testes to develop, and the
hormones they produce switch the development of all somatic tissues to a character-
istic male pathway and suppress female development. In the absence of a Y chromo-
some, the development of somatic tissues is along the female pathway. Specification
of a gonad as a testis is controlled by a single gene on the Y chromosome, the sex-
determining region of the Y chromosome (SRY in humans and Sry in mice), which
was formerly known as testis-determining factor.
Evidence that a region on the Y chromosome actively determines maleness first
came from two unusual human syndromes: Klinefelter syndrome, in which individu-
als have two X chromosomes and one Y (XXY), but are still males; and Turner syn-
drome, in which individuals have just one X chromosome (XO) and are female. Both Frequent Rare
these types of individual have some abnormalities; those with Klinefelter syndrome exchange exchange
are infertile males with small testes, whereas females with Turner syndrome do not
produce eggs. There are also rare cases of XY individuals who are female, and XX X Y X Y X Y
individuals who are phenotypically male. This is due to part of the Y chromosome
being lost (in XY females) or to part of the Y chromosome being transferred to the SRY SRY SRY
X chromosome (in XX males). This can happen during meiosis in the male germ cells
as the X and Y chromosomes are able to pair up, and crossing over can occur between
them. Very rarely, this crossing over transfers the SRY gene from the Y chromosome
onto the X (Fig. 9.20), thus leading to sex reversal.
The sex-determining region alone is sufficient to specify maleness, as shown by an
experiment in which the mouse equivalent of the SRY gene (Sry) was introduced into
the eggs of XX mice. These transgenic embryos developed as males, even though they
lacked all the other genes on the Y chromosome. The presence of the Sry gene, which
encodes a transcription factor, resulted in these XX embryos developing testes instead
of ovaries. In these embryos, as in normal males, Sry was expressed in the developing Fig. 9.20 Sex reversal in humans due to
gonad just before it started to differentiate and drove testis development by triggering chromosomal exchange. At meiosis in male
the differentiation of precursor cells into testis-specific Sertoli cells rather than ovarian germ cells, the X and Y chromosomes pair up
follicle cells. However, the Sry transgenic mice were not completely normal males, as (center panel) and there is crossing over of
other genes on the Y chromosome are necessary for the development of the sperm. the distal region (blue cross), which does not
These XX + Sry males were therefore infertile. affect sexual development (left panel). On
rare occasions, crossing over involves a larger
segment that includes the SRY gene (red
9.13 Mammalian sexual phenotype is regulated by gonadal hormones
cross), so that the X chromosome now carries
All mammals, whatever their genetic sex, start off as embryos along a sexually neu- this male-determining gene (right panel).
tral developmental pathway. The sex of the gonads is genetically determined, as the After Goodfellow, P.N., et al.: 1993.
350 Chapter 9 Germ cells, fertilization, and sex

presence of a Y chromosome results in the somatic cells of the embryo’s gonads

developing into testes rather than into ovaries. The testes secrete Müllerian-inhibiting
­substance, which suppresses female development by causing the embryonic precursor
of the female reproductive organs to regress. It also induces cells to become Leydig
cells, which secrete the male hormone testosterone, which stimulates the development
of male reproductive organs. In XX individuals, the absence of a Y means that ovaries
develop and all subsequent development is female.
The role of hormones in mammalian sexual development means that although the
sex of the gonads is genetically determined, all the other cells in the mammalian
body are neutral, irrespective of their chromosomal sex. It does not matter if they are
XX or XY, as any future sex-specific development they undergo is controlled by hor-
mones. The primary role of the testis in directing male development was originally
demonstrated by removing the prospective gonadal tissue from early rabbit embryos.
All the embryos developed as females, irrespective of their chromosomal constitu-
tion. To develop as a male, therefore, a testis has to be present. The testis exerts its
effect on sexual differentiation of somatic tissues primarily by secreting the hormone
testosterone.
The gonads in mammals develop in close association with the mesonephros; this
is an embryonic kidney that contributes to both the male and female reproductive
organs. Associated with the mesonephros on each side of the body are the Wolffian
ducts, which run down the body to the cloaca, an undifferentiated opening. Another
pair of ducts, the Müllerian ducts, run parallel to the Wolffian ducts and also open
into the cloaca. In early mammalian development, before gonadal differentiation,
both sets of ducts are present (Fig. 9.21). In females, in the absence of the testes,
the ­Müllerian ducts develop into the oviducts (Fallopian tubes), which transport
eggs from the ovaries to the uterus, while the Wolffian ducts degenerate.
In males, the expression of Sry results in the differentiation of Sertoli cells, which
are the somatic cells of the testis and are essential for testis formation and for sper-
matogenesis, as they retain the germ cells that migrate into the gonad. Sertoli cell
development involves upregulation of the transcription factor Sox9, which in turn
induces production and secretion of Müllerian-inhibiting substance. This protein
hormone in turn induces regression of the Müllerian duct, largely by apoptosis. The
Male/female interstitial cell lineage in the testis then differentiates into Leydig cells, which produce
indifferent stage
testosterone. Under the influence of testosterone, the Wolffian duct develops into
metanephric mesonephros
kidney the vas deferens, the duct that carries sperm to the penis. There is evidence that the
gonad extracellular signaling molecule Wnt-4 represses testosterone production in the undif-
ferentiated gonad, and that one of the functions of the Sry gene is to cause the down-
ureter
regulation of Wnt-4 expression, which occurs when Sertoli cells differentiate. FGF-9
is also required for Sertoli cell differentiation and male mice lacking the gene Fgf-9
Wolffian
duct
develop as females.
Müllerian
duct cloaca

Fig. 9.21 Development of the gonads and related structures in mammals. Top panel: early
in development, there is no difference between males and females in the structures that give
Male Female rise to the gonads and related organs. The future gonads lie adjacent to the mesonephros,
oviduct which are embryonic kidneys that are not functional in adult mammals. (The true kidney
testes ovaries develops from the metanephros, from which the ureter carries urine to the bladder.) Two sets
of ducts are present; the Wolffian ducts, which are associated with the mesonephros, and the
Müllerian ducts. Both ducts enter the cloaca. Bottom left panel: after testes develop in the
bladder bladder
male, their secretion of Müllerian-inhibiting substance results in degeneration of the Müllerian
duct by programmed cell death, whereas the Wolffian duct becomes the vas deferens, carrying
ureter ureter sperm from the testis. Bottom right panel: in females, the Wolffian duct disappears, also by
Wolffian
uterus
Müllerian programmed cell death, and the Müllerian duct becomes the oviduct. The uterus forms at the
urethra duct
duct
vagina (oviduct) end of the Müllerian ducts.
(vas deferens)
After Higgins, S.J., et al.: 1989.
 Determination of the sexual phenotype 351

The main secondary sexual characters that distinguish males and females are the Male/female indifferent
reduced size of mammary glands in males, and the development of a penis and a scro- stage
tum in males instead of the clitoris and labia of females (Fig. 9.22). This is due to the phallus
urethral
groove
action of the hormone testosterone. At early stages of embryonic development, the geni-
genital genital
tal regions of males and females are indistinguishable. Differences only arise after gonad swelling fold
development, as a result of the action of testosterone in males. For example, in humans, anus
the phallus gives rise to the clitoris in females and the end of the penis in males.
The role of hormones in sexual development is illustrated by rare cases of abnor-
mal sexual development. Certain XY males develop as phenotypic females in external Male Female
­appearance, even though they have testes and secrete testosterone. They have a muta- glans
tion that renders them insensitive to testosterone because they lack the testosterone body of scrotal labial
penis swelling swelling
receptor, which is present throughout the body. Conversely, genetic females with
genital fold
a completely normal XX constitution can develop as phenotypic males in external urethral groove
appearance, if they are exposed to male hormones during their embryonic develop- anus
ment.
Sex-specific behavior is also affected by the hormonal environment as a result of
prepuce
the effects of hormones on the brain. For example, male rats castrated after birth glans clitoris
develop the sexual behavioral characteristics of genetic females. labium
majus
labium
minus
9.14 The primary sex-determining signal in Drosophila is the number of
vagina
X chromosomes, and is cell autonomous
scrotum anus
The external sexual differences between Drosophila males and females are mainly in
the genital structures, although there are also some differences in bristle patterns and
Fig. 9.22 Development of the genitalia in
pigmentation, and male flies have a sex comb on the first pair of legs. In flies, sex deter- humans. At an early embryonic stage, the
mination of the somatic cells is cell autonomous—that is, it is specified on a cell-by-cell genitalia are the same in males and females
basis—and there is no process resembling the control of somatic sexual differentiation (top panel). After testis formation in males,
by hormones. Somatic sexual development is the result of a series of gene interactions the phallus and the genital fold give rise
that are initiated by the primary sex signal, and which act on a binary male/female to the penis, whereas in females they give
genetic switch. The end result is the expression of just a few effector genes, whose rise to the clitoris and the labia minus. The
activity controls the subsequent male or female differentiation of the somatic cells. genital swelling forms the scrotum in males
Like mammals, fruit flies have two unequally sized sex chromosomes, X and Y, and and the labia majus in females.

males are XY and females XX. But these similarities are misleading. In flies, sex is
not determined by the presence of a Y chromosome, but by the number of X chromo-
somes. Thus, XXY flies are female and X flies are male. The chromosomal composition
of each somatic cell determines its sexual development. This is beautifully illustrated
by the creation of genetic mosaics in which the left side of the animal is XX and the
right side X: the two halves develop as female and male, respectively (Fig. 9.23).
In flies, the presence of two X chromosomes results in the production of the protein
Sex-lethal (Sxl), whose gene is located on the X chromosome. This leads to female
development through a cascade of gene activation that first determines the sexual state
and then produces the sexual phenotype. At the end of the sex-­determination pathway sex comb
is the transformer (tra) gene, which determines how the mRNA of the ­doublesex (dsx)
gene is spliced; dsx encodes a transcription factor whose activity ultimately produces
most aspects of somatic sex. The dsx gene is active in both males and females, but
different protein products are produced in the two sexes as a result of the sex-specific
RNA splicing (Fig. 9.24). Males and females thus express similar but distinct Double-
sex (Dsx) proteins, which act in somatic cells to induce ­expression of sex-specific
genes, as well as to repress characteristics of the opposite sex. ­Production of the

Fig. 9.23 A Drosophila female/male genetic mosaic. The left side of the fly is composed of
XX cells and develops as a female, whereas the right side is composed of X cells and develops
as a male. The male fly has smaller wings, a special structure, the sex comb, on its first pair of female male
legs, and different genitalia at the end of the abdomen (not shown).
352 Chapter 9 Germ cells, fertilization, and sex

Fig. 9.24 Outline of the sex-determination

pathway in Drosophila. The number of Sex determination in Drosophila
X chromosomes is the primary sex-
Stable binary Transducers of Effectors of
determining signal, and in females the Primary signal sexual phenotype
genetic switch the sexual state
presence of two X chromosomes activates
the gene Sex-lethal (Sxl ). This produces dsx f Female
Sex-lethal protein, whereas no Sex-lethal
protein is made in males, which have only X chromosomes Sxl e.g. tra splicing
dsx
one X chromosome. The activity of Sex-lethal
is transduced via the transformer gene (tra) dsx m Male
and causes sex-specific splicing of doublesex
RNA (dsx f ), such that the cells follow a XX ON ON dsx f
female developmental pathway. In the
X OFF OFF dsx m
absence of Sex-lethal protein, the splicing
of doublesex RNA to give dsxm RNA leads to
male development. male form of the protein is the default pathway, but in the presence of Sxl, tra mRNA
undergoes RNA splicing, and this, together with the actions of the ­Transformer-2
protein, leads to the female form of the Dsx protein being made, resulting in develop-
ment as a female.
The Sex-lethal (Sxl) gene is only turned on in females with two X chromosomes.
In the absence of early Sxl expression, male development occurs. Once Sxl is activated
in females, it remains activated through an autoregulatory mechanism, which results in
the Sxl protein being synthesized throughout female development. Early expression of
Sxl in females occurs through activation of a promoter, Pe, at about the time of syncytial
blastoderm formation. Sxl protein is synthesized and accumulates in the blastoderm
of female embryos. At the cellular blastoderm stage, another promoter for Sxl, Pm,
becomes active in both males and females and Pe is shut off, but the sex is already deter-
mined. Splicing of the RNA transcribed from Pm into functional Sxl mRNA needs some
Sxl protein to be present already, and so can only happen in females (Fig. 9.25).
How does the number of X chromosomes control these key sex-determining genes?
In Drosophila, the mechanism involves interactions between the products of so-called
‘numerator’ genes on the X chromosome and of genes on the autosomes, as well as
maternally specified factors. Essentially, in females, the double dose of numerator pro-
teins activates Sxl by binding to sites in the Pe promoter (see Fig. 9.25), overcoming
the repression of Sxl by autosomally encoded proteins that would occur in males.

XX X
Fig. 9.25 Production of Sex-lethal protein
in Drosophila sex determination. When Sex-lethal establishment Sex-lethal maintenance Sex-lethal establishment Sex-lethal maintenance
two X chromosomes are present, the early promoter (SxlPe ) promoter (SxlPm ) promoter (SxlPe) promoter (SxlPm)
establishment promoter (Pe) of the Sex- ON ON OFF ON
lethal (Sxl ) gene is activated at the syncytial no early Sxl
blastoderm stage in future females, but not protein
in males. This results in the production of Sxl
early Sxl late Sxl
protein. Later, at the blastoderm stage, the protein protein
maintenance promoter (Pm) of Sxl becomes
active in both females and males, and Pe
productive nonproductive
is turned off. The Sxl RNA is only correctly splicing splicing
RNA RNA
spliced if Sxl protein is already present, which
is only in females. A positive feedback loop
for Sxl protein production is thus established trigger
in females. The continued presence of Sxl to start
feedback loop
protein initiates a cascade of gene activity no late Sxl protein
leading to female development. If no Sxl
protein is present, male development ensues. Female Male
After Cline, T.W.: 1993.
 Determination of the sexual phenotype 353

oocyte
ovary ovary
mouth anus

sperm stored in spermatheca

Hermaphrodite (XX)
vulva eggs and developing embryos in uterus
Fig. 9.26 Hermaphrodite and male
Caenorhabditis elegans. The hermaphrodite
testis has a ‘two-armed’ gonad and initially
makes sperm, which is then stored in the
Male (XO) vas deferens
spermotheca. It then switches to make eggs.
sperm copulatory organ
The eggs are fertilized internally. The male
makes sperm only.

The pathway outlined in Fig. 9.24 is an oversimplification, as dsx does not con-
trol all aspects of somatic sexual differentiation in Drosophila. There is an additional
branch of the sex-differentiation pathway downstream of tra, which controls sexually
dimorphic aspects of the nervous system and sexual behavior. This branch includes
the gene fruitless, whose activity has been shown to be necessary for male sexual
behavior.
In other dipteran insects, the same general strategy for sex determination is used,
but there are marked differences at the molecular level. Only dsx has been found in
dipterans distantly related to Drosophila; Sxl has been found in other dipterans but is
not involved in sex determination there.

9.15 Somatic sexual development in Caenorhabditis is determined by the

number of X chromosomes

In the nematode C. elegans, the two sexes are self-fertilizing hermaphrodite (essen-
tially a modified female) and male (Fig. 9.26), although in other nematodes they
are male and female. Hermaphrodites produce a limited amount of sperm early in
development, with the remainder of the germ cells developing into oocytes. Sex in
C. elegans (and other nematodes) is determined by the number of X chromosomes:
the hermaphrodite (XX) has two X chromosomes, whereas the presence of just one
X chromosome leads to development as a male (XO). One of the primary sex sig-
nals for hermaphrodite development is the SEX-1 protein, which is encoded on the­
X chromosome and is a key element in ‘counting’ the number of X chromosomes pres-
ent. SEX-1 is a nuclear hormone receptor that represses the sex-determining gene XO
lethal (xol-1), also present on the X chromosome. In the presence of a double dose of
SEX-1 produced from the two X chromosomes, xol-1 expression is inhibited, resulting
in the development of a hermaphrodite. When only one X chromosome is present,
xol-1 is expressed at a high level and the embryo develops as a male.
A cascade of gene activity converts the level of xol-1 expression into the somatic sex-
ual phenotype (Fig. 9.27). Unlike Drosophila, sex determination in C. elegans requires
intercellular interactions, as at least one of the genes involved encodes a secreted pro-
tein. At the end of the cascade is the gene transformer-1 (tra-1), which encodes a tran-
scription factor. Expression of TRA-1 protein is both necessary and sufficient to direct
all aspects of hermaphrodite (XX) somatic cell development, as a gain-of-function
mutation in tra-1 leads to hermaphrodite development in an XO animal, irrespective
of the state of any of the regulatory genes that normally control its activity. Mutations
that inactivate tra-1 lead to complete masculinization of XX hermaphrodites.
Before considering how the sex of the germ cells is determined and how the imbal-
ance of X-linked genes between the sexes are dealt with in animals, we will digress
briefly to touch upon sex determination in flowering plants.
354 Chapter 9 Germ cells, fertilization, and sex

Fig. 9.27 Outline of the somatic sex

determination pathway in Caenorhabditis Sex determination in Caenorhabditis elegans
elegans. The primary signal for sex
Binary Transducers of Effectors of
determination is set by the number of Primary signal sexual phenotype
genetic switch the sexual state
X chromosomes. When two X chromosomes
are present, the expression of the sdc-1 fem-1 Hermaphrodite
gene XO lethal (xol-1) is low, leading to (inhibition) tra-2
X chromosomes xol-1 sdc-2 her-1 fem-2 tra-1
hermaphrodite development, whereas xol-1 tra-3
sdc-3 fem-3
is expressed at a high level in males. There is Male
a cascade of gene expression starting from
xol-1 that leads to the gene transformer-1 XX Low High Low High Low High
(tra-1), which codes for a transcription X High Low High Low High Low
factor. If tra-1 is active, development as a
hermaphrodite occurs, but if it is expressed at
low levels, males develop. The product of the
hermaphrodite-1 (her-1) gene is a secreted 9.16 Most flowering plants are hermaphrodites, but some produce
protein, which probably binds to a receptor unisexual flowers
encoded by transformer-2 (tra-2), inhibiting
Unlike animals, plants do not set aside germ cells in the embryo, and germ cells are
its function.
only specified when a flower develops. Any meristem cell can, in principle, give rise
to a germ cell of either sex, and there are no sex chromosomes. The great majority of
flowering plants (angiosperms) give rise to flowers that contain both male and female
sexual organs, in which meiosis occurs. The male sexual organs are the stamens; they
produce pollen, which contains the male gamete nuclei corresponding to the sperm of
animals. The female reproductive structures are the carpels, which are either free or
are fused to form a compound ovary (see Box 7A, p. 259). Carpels are the site of ovule
formation, and each ovule produces an egg cell.
Angiosperms, such as Arabidopsis, undergo so-called ‘double fertilization.’ Each
pollen grain contains two sperm nuclei, which are delivered into the ovule by the
growth of a vegetative pollen tube. One nucleus fuses with the haploid egg cell,
forming the zygote that develops into the embryo. The second nucleus fuses with the
diploid ‘central cell’ in the ovule, forming a triploid cell that proliferates and devel-
ops into a storage tissue—the endosperm. In some plants, such as Arabidopsis, the
endosperm nourishes the growth of the embryo, while in others it is broken down at
germination to produce nutrients for the seedling.
Most flowering plants are hermaphrodite, bearing flowers with functional male and
female sexual organs, as in Arabidopsis. In about 10% of flowering plant species, flowers
of just one sex are produced. Flowers of different sexes may occur on the same plant, or be
confined to different plants. The development of male or female flowers usually involves
the selective resorption of either the stamens or pistil after they have been specified and
have started to grow. In maize, for example, male and female flowers develop at particular
sites on the shoot. The tassel at the tip of the main stem only bears flowers with stamens;
the ‘ears’ at the ends of the lateral branches bear female flowers containing pistils. Sex
determination becomes visible when the flower is still small, with the stamen primordia
being larger in males and the pistil longer in females. The smaller organs eventually
degenerate. The plant hormone gibberellic acid may be involved in sex determination, as
differences in gibberellin concentration are associated with the different sexual organs. In
the maize tassel, gibberellin concentration is 100-fold lower than in the developing ears. If
the concentration of gibberellic acid is increased in the tassel, pistils can develop.
Genomic imprinting (see Section 9.7) occurs in flowering plants, but has only been
detected in the endosperm and not in the embryo. The imprinting events take place in
the central-cell genome and the sperm genome. In plants, as in mammals, imprinting
involves gene silencing due to DNA methylation, histone modifications, Polycomb
proteins and non-coding RNAs. Unlike mammals, however, the imprinting does not
have to be removed to produce a new generation, as the endosperm is a temporary
tissue that does not contribute cells to the embryo. As well as the silencing of paternal
 Determination of the sexual phenotype 355

genes in sperm by DNA methylation and histone modification, a feature of angiosperm

imprinting is the DNA demethylation of previously silenced maternal genes in the migrating
germ cell
central cell, which then enables their expression. The activated genes include mem-
bers of the Polycomb group, which then maintain the repression of incoming silenced
paternal alleles and silence the maternal alleles of certain other genes. Imprinting Embryonic Embryonic
in angiosperms could have evolved together with double fertilization as a means of testis ovary
preventing endosperm proliferation in the absence of fertilization.

9.17 Determination of germ-cell sex depends on both genetic constitution Block in mitotic Meiotic
cell division prophase
and intercellular signals in G1 stage arrest

Determination of the sex of animal germ cells—that is, whether they will develop into
immature oocytes
eggs or sperm—is strongly influenced by the signals they receive when they become part spermatogonia
of a gonad. In the mouse, for example, their future development is determined largely
by the sex of the gonad in which they reside, and not by their own chromosomal con-
stitution. In reality, chromosomal constitution and gonadal signals will almost always
coincide, but it has been shown that germ cells from male mouse embryos can develop
into oocytes rather than sperm if grafted into female embryonic gonads and vice versa.
There is a distinct difference in the timing of meiosis in male and female mammals. After
birth
In male mouse embryos, the diploid germ cells stop dividing when they enter the
gonad, becoming arrested in the G1 phase of the mitotic cell cycle. They start dividing
mitotically again after birth and enter meiosis some 7 to 8 days after birth. In female
mouse embryos, the primordial diploid germ cells continue to proliferate for a few Enter and 1st meiotic
complete division
days after entering the genital ridge (see Sections 9.4 and 9.5). Diploid germ cells in meiosis completed
the gonad first undergo a few rounds of mitotic division and then enter prophase of
the first meiotic division; they then arrest at this primary oocyte stage until the mouse sperm
becomes a sexually mature female, about 6 weeks after birth, when at each reproduc-
tive cycle, selected oocytes complete the first meiotic division and start the second
meiotic division (Fig. 9.28). Meiosis is only completed after fertilization.
Although the development of germ cells is normally heavily influenced by the envi-
ronment in which they find themselves, if the usual external cues are absent, germ
cells seem to follow an intrinsic pathway of differentiation. All mouse germ cells that Oocytes released
Haploid sperm Start 2nd meiotic division
enter meiosis before birth develop as eggs, whereas those not entering meiosis until released Complete meiosis
after birth develop as sperm. Germ cells, whether XX or XY, that fail to enter the after fertilization
genital ridge and instead end up in adjacent tissues such as the embryonic adrenal
gland or mesonephros, enter meiosis and begin developing as oocytes in both male Fig. 9.28 The timing of meiosis in the
and female embryos; thus, the default germ-cell sex appears to be female. XX/XY chi- germ cells differs considerably between
meric mouse embryos can be made by combining male and female four-cell embryos males and females in mammals. Top panel:
and, because of the presence of one Y chromosome, can develop testes. In these XX/ Migrating germ cells, whether XX or XY, enter
XY embryos, XX germ cells that are surrounded by testis cells start to develop along meiotic prophase and start developing as
oocytes unless they enter a testis. In the
the spermatogenesis pathway, but the later development of the gametes is abnormal.
testis, the germ cells receive an inhibitory
In a particular strain of Y/XX mice that develops ovaries rather than testes, XY germ
signal that blocks mitotic division and
cells appear to develop quite normally as oocytes but fail to develop further after fer- prevents them entering meiotic prophase.
tilization as the spindle does not assembly properly for the second meiotic division. Bottom panel: In male mice after birth,
Experiments transferring XY oocyte nuclei into normal XX oocytes produced healthy immature diploid spermatogonia in the testis
offpring, indicating that the fault lies in the XY cytoplasm. enter and complete meiosis in the testis to
In Drosophila, the difference in the behavior of XY and XX germ cells depends ini- produce haploid germ cells which eventually
tially on the number of X chromosomes, as in somatic cells; the Sxl gene again plays mature into sperm (left). In female mice after
an important role, although most other elements in the sex-determination pathway birth, oocytes complete their first meiotic
may differ from those in somatic cells. Both chromosomal constitution and cell inter- division in the ovary, but do not enter the
actions are involved in the development of germ-cell sexual phenotype. Transplanta- second meiotic division meiosis until after
they are released from the ovary. The egg
tion of genetically marked pole cells (see Section 9.1) into a Drosophila embryo of the
only completes meiosis after fertilization.
opposite sex shows that male XY germ cells in a female XX embryo become integrated
into the ovary and begin to develop as sperm; that is, their behavior is autonomous
356 Chapter 9 Germ cells, fertilization, and sex

distal tip cells—signal

Larva

Fig. 9.29 Determination of germ-cell

sperm
sex in the hermaphrodite nematode
gonad. Top: during the larval stage, germ
cells in a zone close to the distal tips of the
gonad multiply; when they leave this zone
Adult
in the larval stage they enter meiosis and
develop into sperm. Bottom: in the adult, cells
that leave the proliferative zone develop into oocytes uterus spermatheca
oocytes. The eggs are fertilized as they pass proliferating germ cells
into the uterus. germ cells in meiotic pathway
After Clifford, R., et al.: 1994.

with respect to their genetic constitution. By contrast, XX germ cells in a testis attempt
to develop as sperm, showing a role for environmental signals. In neither case, how-
ever, are functional sperm produced.
The hermaphrodite of C. elegans provides a particularly interesting example of
germ-cell differentiation, as both sperm and eggs develop within the same gonad.
­Unlike the somatic cells, which have a fixed lineage and number (see Section 6.1), the
number of germ cells in an adult nematode is indeterminate, with about 1000 germ
cells in each ‘arm’ of the gonad. At hatching of the first-stage larva, there are just two
founder germ cells, which proliferate to produce the germ cells. The germ cells are
flanked on each side by cells called distal tip cells, and their proliferation is controlled
by a signal from the distal tip cells. This signal is the protein LAG-2, which is homolo-
gous to the Notch ligand Delta. The receptor for LAG-2 on the germ cells is GLP-1,
which is similar both to nematode LIN-12, which is involved in vulva formation (see
Section 6.6) and to Notch. We have already met GLP-1 acting to determine cell fate in
the early embryo (see Section 6.3).
In C. elegans, entry of germ cells into meiosis from the third larval stage onward is
controlled by the distal tip signal. In the presence of this signal, the cells proliferate,
but as they move away from it, they enter meiosis and develop as sperm (Fig. 9.29,
top). In the hermaphrodite gonad, all the cells that are initially outside the range of
the distal tip signal develop as sperm, but cells that later leave the proliferative zone
and enter meiosis develop as oocytes (see Fig. 9.29, bottom). The eggs are fertilized
by stored sperm as they pass into the uterus. The male gonad has similar proliferative
meiotic regions, but all the germ cells develop as sperm.
Sex determination of the nematode germ cells is somewhat similar to that of the
somatic cells, in that the chromosomal complement is the primary sex-determining
factor and many of the same genes are involved in the subsequent cascades of gene
expression. The terminal regulator genes required for spermatogenesis are called fem
and fog. In hermaphrodites, there must be a mechanism for activating the fem genes
in some of the XX germ cells, so allowing them to develop as sperm.

9.18 Various strategies are used for dosage compensation of X-linked genes

In all the animals we have considered in this chapter there is an imbalance of X-linked
genes between the sexes. One sex has two X chromosomes, whereas the other has
one. This imbalance has to be corrected to ensure that the level of expression of genes
carried on the X chromosome is the same in both sexes. The mechanism by which the
imbalance in X-linked genes is dealt with is known as dosage compensation. Failure
 Determination of the sexual phenotype 357

Fig. 9.30 Mechanisms of dosage

Mammals Drosophila Caenorhabditis elegans
compensation. In mammals, Drosophila,
Female Male Female Male Hermaphrodite Male and Caenorhabditis elegans there are two
X chromosomes in one sex and only one in
Ratio of X chromosomes the other. Mammals inactivate one of the
2 : 1 2 : 1 2 : 1 X chromosomes in females; in Drosophila
males there is an increase in transcription from
X X X X X X X X X the single X chromosome; and in C. elegans
there is a decrease in transcription from the
X chromosomes in hermaphrodites. The result
of these different dosage compensation
mechanisms is that the level of X chromosome
X-chromosome Increase in Decrease in
transcription transcription transcripts is approximately the same in males
inactivation
and females.
Transcripts

cell in early embryo

paternal X
chromosome (Xp)

1 : 1 1 : 1 1 : 1 maternal X
chromosome (Xm)
Ratio of X transcripts

Inactivation and
condensation of a randomly
to correct the imbalance leads to abnormalities and arrested development. Different selected X chromosome
animals deal with the problem of dosage compensation in different ways (Fig. 9.30). Xp Xm
Mammals, such as mice and humans, achieve dosage compensation in females
by inactivating one X chromosome in each cell after the blastocyst has implanted in
the uterine wall. Once an X chromosome has been inactivated in an embryonic cell,
this chromosome is maintained in the inactive state in all the resulting somatic cells,
and inactivation persists throughout the life of the organism (Fig. 9.31). The inactive
Direct inheritance of the pattern
X chromosome is replicated at each cell division but remains transcriptionally inactive of chromosome condensation
and in a different physical state from the other chromosomes during the rest of the cell
cycle. At mitosis all chromosomes become highly condensed. During the interphase
of the cell cycle—the period between successive mitoses—all the other chromosomes
decondense into extended threads of chromatin, the complex of DNA and proteins of
which chromosomes are made, and are no longer visible under the light microscope.
The inactive X chromosome, however, remains condensed and is visible in human
Direct inheritance of the pattern
cells as the Barr body (Fig. 9.32). Which X chromosome in a cell is inactivated seems of chromosome condensation
to be random, and so female mammals are a mosaic of cells with different X chromo-
somes inactivated.
The mosaic effect of X inactivation is sometimes visible in the coats of female mam-
mals. Female mice heterozygous for a coat pigment gene carried on the X chromosome
have patches of color on their coat, produced by clones of epidermal cells that express
the X chromosome carrying a functional pigment gene. The rest of the epidermis is
only Xm active
composed of cells in which that X chromosome has been inactivated. The coat pattern
of tortoiseshell cats is also due to X-linked mosaicism. They carry two co-dominant
Fig. 9.31 Inheritance of an inactivated X
alleles, orange (XO) and black (XB), of the main coat color gene, which is located on
chromosome. In early mammalian female
the X chromosome. These alleles produce orange and black ­pigments, ­respectively.
embryos one of the two X chromosomes,
Pigment genes are expressed in melanocytes that derive from neural crest cells that either the paternal X (Xp) or the maternal X
migrate to the skin. Cells in which the chromosome carrying X O is inactivated express (Xm), is randomly inactivated. In the figure,
the XB allele and vice versa. In bi-colored tortoiseshell cats (Fig. 9.33), the two cell Xp is inactivated and this inactivation is
types are intermingled, producing the characteristic brindled appearance. In calico maintained through many cell divisions. The
cats, which have distinct patches of white, orange, and black fur, there has been less inactivated chromosome becomes highly
intermingling and larger clones of cells of one color develop, along with white patches condensed.
with no pigment. After Alberts, B., et al.: 2002.
358 Chapter 9 Germ cells, fertilization, and sex

Fig. 9.32 Inactivated X chromosome (the Barr body). The photograph shows the Barr body
(arrow) in the interphase nucleus of a female human buccal cell.
Photograph courtesy of J. Delhanty.

How cells count and choose chromosomes for inactivation is not yet fully under-
stood. But the finding that tetraploid cells have two active X chromosomes has led to
the following proposal, which depends on the levels of a signal produced by the X chro-
mosomes themselves and invokes a negative-feedback mechanism between autosomes
and the X chromosomes. Before inactivation, the X chromosomes are producing a signal
(whether a protein or an RNA is not yet known) that binds to the other chromosomes
in the cell. When all the binding sites are occupied, the autosomal chromosomes in
turn produce a signal that can bind to and inactivate X chromosomes. The first X chro-
mosome to acquire a critical mass of bound signal initiates its own inactivation. The
output of the X-chromosome signal is therefore reduced, the autosomes stop producing
the X-inactivating signal, with the result that just one X chromosome is inactivated
per diploid genome. This type of mechanism can also accommodate the long-standing
­observation that in diploid female cells with an ­abnormal sex-chromosome constitu-
tion, such as XXY or XXXY, there is only one active X chromosome per somatic cell,
with all the other X chromosomes inactivated.
X inactivation is dependent on a small region of the X chromosome, the inactivating
center, which contains a pair of overlapping genes for two non-coding RNAs, Xist and
Tsix, whose competing activities regulate inactivation. Before inactivation, both RNAs
are transcribed at low levels from both X chromosomes. Xist expression then dramati-
cally increases on one or other of the X chromosomes and ceases on the other. What
causes this to happen is still unknown, but a candidate might be the still-hypothetical
inactivating signal from the autosomes noted above. The Xist transcripts spread from
their site of synthesis to coat the whole X chromosome, blocking transcription, includ-
ing that of Tsix, and inactivating the chromosome. On the chromosome that is not inac-
tivated, Tsix remains active for a little time but soon after X inactivation is complete,
both Tsix and Xist are switched off on the active chromosome. Xist RNA continues
to keep the inactivated chromosome inactive through subsequent cell divisions. The
inactive X chromosome has a pattern of DNA methylation that differs from that of the

Fig. 9.33 The coloring of tortoiseshell cats

is due to mosaicism of X-linked alleles due
to X inactivation.
Foxfire, photograph courtesy of Bruce Goatly.
 Determination of the sexual phenotype 359

active X, and this methylation is likely to help keep it inactive. DNA methylation is one
of the mechanisms used in mammals for long-term silencing of genes, and we have
already seen its effects in the differential imprinting of maternal and paternal genes in
the germline (see Section 9.8). DNA methylation and other epigenetic mechanisms for
controlling gene expression are discussed in more detail in Box 10A (p. 374).
Dosage compensation in Drosophila works in a different way to that in mice and
humans (see Fig. 9.30). Instead of repression of the ‘extra’ X activity in ­females,
transcription of the X chromosome in males is increased nearly twofold. A set of
male-specific genes, the MSL complex, controls most dosage compensation, and these
are repressed in females by Sxl protein, thereby preventing excessive ­transcription
of the X chromosome. The increased activity in males is regulated by the primary
­sex-determining signal, which results in the dosage compensation mechanism operat-
ing when Sxl is ‘off’. In females, where Sxl is ‘on’, it turns off the dosage-compensation
mechanism. As in the mouse, this regulatory mechanism involves non-coding RNA.
In C. elegans, dosage compensation is achieved by reducing the level of
X chromosome expression in XX animals to that of the single X chromosome in
XO males (see Fig. 9.30). The number of X chromosomes is communicated by a set of
X-linked genes that can repress the master gene xol-1. A key event in initiating nema-
tode dosage compensation is the expression of the protein SDC-2, which occurs only
in hermaphrodites. It forms a complex specifically with the X chromosome and trig-
gers assembly of a protein complex called the dosage compensation complex, which
binds to a specific region on the X chromosome and reduces transcription.

Summary

The development of early embryos of both sexes is very similar. A primary sex-
­determining signal sets off development toward one or the other sex, and in mammals,
Drosophila, and C. elegans, this signal is determined by the chromosomal complement
of the fertilized egg. In mammals, the Sry gene on the Y chromosome is responsible for
the embryonic gonad developing into a testis and producing hormones that determine
male sexual characteristics. The sexual phenotype of the somatic cells is determined
by the gonadal hormones. In C. elegans and Drosophila, the primary sex-determining
signal is the number of X chromosomes. In Drosophila, the gene Sex-lethal is turned on
in females but not in males, in response to this signal. In both cases, this results in fur-
ther gene activity in which sex-specific RNA splicing is involved. In C. elegans, the gene
XO lethal is turned off in hermaphrodites and on in males, eventually leading to sex-
­specific ­expression of the gene transformer-1, which determines the sexual pheno-
type. Somatic sexual differentiation in Drosophila is cell autonomous and is controlled
by the number of X chromosomes; in C. elegans, cell–cell interactions are also involved.
In mammals, signals from the gonads determine whether the germ cells develop into
oocytes or sperm. Male germ cells in Drosophila develop along the sperm pathway even
in an ovary, but female germ cells develop along the sperm pathway when placed in a
testis. Most C. elegans adults are hermaphrodites, and produce both sperm and eggs
from the same gonad.
Various strategies of dosage compensation are used to correct the imbalance of
X chromosomes between males and females. In female mammals, one of the X chro-
mosomes is inactivated; in Drosophila males the activity of the single X chromosome is
upregulated; and in C. elegans the activity of the X chromosomes in XX hermaphrodites
is downregulated to match that from the single X chromosome in males. Plants do not
have a germline as any cell can, in principle, give rise to ova and sperm. Most plants
produce flowers with hermaphrodite flowers, but some produce unisexual flowers,
either on the same plant or on two separate plants. Plants also have a form of genomic
imprinting involving sperm and endosperm cells.
360 Chapter 9 Germ cells, fertilization, and sex

Summary: determination of sexual phenotype

Mammals

Female Male
XX XY
one X inactivated Sry active on Y
via Xist chromosome
gonad gonad

Wolffian duct degenerates; Wolffian duct forms

Müllerian duct forms ovary testis vas deferens; Müllerian
oviduct and uterus duct degenerates

female secondary no testosterone; testosterone male secondary

sexual characteristics female gonadal sexual characteristics
hormones

undifferentiated genital region

female external male external

genitalia genitalia

Drosophila C. elegans

Female Male increased Hermaphrodite Male

XX YX transcription XX XO
reduced
transcription
Sxl on Sxl off XO lethal off XO lethal on

sex-specific splicing fruitless transformer-1 high transformer-1 low

of doublesex RNA
sexual
behavior
development of female development of male mab-3 off; development of male
reproductive organs reproductive organs development of reproductive organs
and somatic sexual and somatic sexual hermaphrodite and somatic sexual
phenotype phenotype reproductive organs phenotype
and somatic sexual
phenotype

Summary to Chapter 9

In many animals, the future germ cells are specified by localized cytoplasmic determi-
nants in the egg. Germ cells in mammals are unusual, as they are specified entirely by
cell–cell interactions. In animals, the development of germ cells into sperm or egg depends
both on chromosomal constitution and interactions with the cells of the gonad. At fertiliza-
tion, fusion of sperm and egg initiates development, and there are mechanisms to ensure
that only one sperm enters an egg. Both maternal and paternal genomes are required for
­normal mammalian development, as some genes are imprinted; for such genes, whether
they are expressed or not during development depends on whether they are derived
from the sperm or the egg. In many animals, the chromosomal constitution of the embryo
­determines which sex will develop. In mammals, the Y chromosome is male-determining;
it specifies the ­development of a testis, and the hormones produced by the testis cause
the development of male sexual characteristics. In the absence of a Y chromosome, the
 Determination of the sexual phenotype 361

embryo develops as a female. In Drosophila and C. elegans, sexual development is initially

determined by the number of X chromosomes, which sets in train a cascade of gene acti­
vity. In nematodes, the somatic sexual phenotype is determined by cell–cell interactions;
in flies, somatic cell sexual differentiation is cell autonomous. Animals use a variety of
strategies of dosage compensation to correct the imbalance in the number of X chromo-
somes in males and females.

n End of chapter questions 12. How does SRY trigger male development in mammals? Include:
Sox9, Sertoli cells, Leydig cells, Müllerian inhibiting substance and
Long answer (concept questions) testosterone in your answer.
1. Distinguish between the germline, gametes, and somatic cells. 13. Summarize the development of the gonads and related struc-
What key functions are the responsibility of the germ cells? tures in mammals, focusing on the Wolffian and Müllerian ducts.
2. Summarize the evidence for the existence of a special germ 14. Summarize the information presented in Figs 9.24 and 9.25 on
plasm in Drosophila and Xenopus. Is the presence of germ plasm a how cell-autonomous sexual differentiation occurs in Drosophila.
universal feature of animal development? Do not just copy the two figures, but instead incorporate the infor-
mation into one comprehensive schematic.
3. Draw a schematic of two generations of Drosophila: start with a
male and a female, each of which is heterozygous for a loss-of-func- 15. What is different about the sexes in Caenorhabditis elegans
tion recessive oskar mutation. Mate them together to get homozy- compared with the sexes in our other model organisms? How does
gous oskar males and females. Mate these oskar mutants to normal this affect its method of reproduction?
(wild-type) flies. How do these matings illustrate why such muta-
tions are referred to as ‘grandchildless’. Does it matter whether the 16. Briefly compare and contrast the sex-determination pathways
homozygous oskar mutant is a male or a female? Explain. in Drosophila and C. elegans. Are there similarities in the strategies
used? Are there differences in the mechanisms involved?
4. Contrast the roles of SDF-1/CXCR4 and STEEL/KIT in germ cell
migration. 17. Say how X-chromosome inactivation will achieve dosage com-
pensation in humans with the chromosomal compositions: XY, XX,
5. Discuss the following aspects of meiosis: (1) Why is a ‘reduction XXY, or XXX?
division’ (meiosis I) important? (2) What is the significance of recom-
bination in sexual reproduction? (3)Why do four sperm cells result 18. How does Xist control X-chromosome inactivation in mam-
from each cell undergoing meiosis during spermatogenesis, yet only mals? Speculate why the other gene involved in this process, Tsix,
one egg cell results during oogenesis—where do the other ‘eggs’ go? might have been given this name.

6. When in human development do primary oocytes form? What Multiple choice (factual recall questions)
happens to a primary oocyte between this time and its maturation NB There is only one correct answer to each question.
into an egg ready to be fertilized? At what point in meiosis is the
oocyte when fertilization occurs? 1. The germ plasm in Drosophila is specified by
a) bicoid
7. What is meant by an ‘epigenetic’ mechanism? What is meant by
‘genomic imprinting’? What is the connection between the two? b) Hox genes
c) oskar
8. Describe the process by which the sperm nucleus enters the egg.
d) the point of sperm entry
Include SED1, ZP3, and acrosomal enzymes in your answer.
2. In mice, germ cells can be first identified
9. Fertilization of a sea-urchin egg leads to a wave of calcium that
a) in prospective mesoderm of the proximal epiblast, as Blimp1-
sweeps across the egg. Is this calcium intracellular or extracellular in
expressing cells
origin? What are two consequences of this increase in calcium?
b) in the mesoderm that will form gonadal tissue
10. What is MPF? Summarize the role of MPF-like complexes in the c) in the posterior tip of the embryo, as cells containing pole plasm
control of the various stages of the cell cycle.
d) in the genital ridge, as Oct-4-expressing cells
11. Explain how chromosomal abnormalities can be used as evi- 3. Which feature of meiosis, as compared with mitosis, is most
dence for the following statements. (1) Determination of the male sex important for the production of gametes?
in humans is controlled by the presence of the Y chromosome, not
a) Homologous chromosomes separate in meiosis I, whereas sister
two X chromosomes. (2) Determination of female sex in Drosophila is
chromatids separate in mitosis.
controlled by the presence of two X chromosomes and not the pres-
ence of a Y chromosome. (3) One X chromosome in ­humans is suffi- b) Meiosis leads to the production of haploid products, whereas
cient to determine the female sex in humans. (4) One X chromosome ­mitosis leads to diploid daughter cells.
is sufficient to determine the male sex in Drosophila. (5) SRY is the c) Meiosis provides an opportunity for recombination, whereas
gene responsible for male sex determination in humans. ­mitosis does not.
362 Chapter 9 Germ cells, fertilization, and sex

d) Typically, three of the four products do not become oocytes ­during c) The cortical reaction is the fusion of the egg cortex with the egg
meiosis in female organisms, whereas mitosis will ­produce four plasma membrane, which allows the sperm to enter.
functional products after two rounds of cell ­division. d) The cortical reaction is the release of the cortical granules after
4. In what stage of which type of cell division, is the mammalian sperm entry, which converts the vitelline membrane into the fer-
oocyte at the birth of the animal? tilization membrane, which blocks polyspermy.
a) The oocyte is in G1 of the mitotic cell cycle. 10. The C. elegans GLP-1 protein is similar to which mammalian
b) The oocyte is in metaphase of meiosis II. signaling protein?
c) The oocyte is in prophase of meiosis I. a) BMPs
d) The oocyte is in prophase of mitosis. b) Delta
c) Notch
5. What is the zona pellucida of the mammalian egg?
d) testosterone receptor
a) a hardened membrane that forms a physical block to polyspermy,
formed from the vitelline membrane and the contents of the cor-
Multiple choice answer key
tical granules
1: c, 2: a, 3: b, 4: c, 5: c, 6: b, 7: c, 8: b, 9: d, 10: c.
b) a layer of follicle-derived cells called cumulus cells
c) an extracellular layer of glycoproteins
d) the plasma membrane of the egg n General further reading
6. The mammalian oviduct will form from the
Chadwick, D., Goode, J.: The Genetics and Biology of Sex Determina-
a) mesonephros tion 2002. Novartis Foundation Symposium 244. New York: John
b) Müllerian duct Wiley, 2002.
c) ureter Cinalli, R.M., Rangan, P., Lehman, R.: Germ cells are forever. Cell
2008, 132: 559–562.
d) Wolffian duct
Crews, D.: Animal sexuality. Sci. Am. 1994, 270: 109–114.
7. The molecular activity of the Sxl protein of Drosophila is Zarkower, D.: Establishing sexual dimorphism: conservation amidst
a) as a ‘numerator’, counting the number of X chromosomes diversity? Nat. Rev. Genet. 2001, 2: 175–185.
b) as a transcription factor
c) to control RNA splicing n Section further reading
d) to signal to the Tra receptor
9.1 Germ-cell fate is specified in some embryos by a distinct
8. The human congenital disorder Angelman syndrome is due to
germ plasm in the egg
the inheritance of a maternal chromosome 15 that has a small dele-
tion of a specific region. Why does this deletion not behave as a Extavour, C.G., Akam, M.: Mechanisms of germ cell specification
recessive allele; that is, why is its loss not made up for by the intact across the metazoans: epigenesis and preformation. Develop-
region on the unmutated paternal chromosome 15? ment 2003, 130: 5869–5884.
Matova, N., Cooley, L.: Comparative aspects of animal oogenesis.
a) Chromosomes with deletions do not go through mitosis correctly,
Dev. Biol. 2001, 231: 291–320.
so cell divisions in the embryo result in cells with abnormal num-
Mello, C.C., Schubert, C., Draper, B., Zhang, W., Lobel, R., Priess,
bers of chromosomes, and these cells do not contribute properly
J.R.: The PIE-1 protein and germline specification in C. elegans
to the development of the organism.
embryos. Nature 1996, 382: 710–712.
b) The father’s copy of chromosome 15 has genes in the region of Micklem, D.R., Adams, J., Grunert, S., St. Johnston, D.: Distinct
the deletion that are imprinted, and thus inactive; in the absence roles of two conserved Staufen domains in oskar mRNA locali-
of any active copies of these genes, development cannot proceed sation and translation. EMBO J. 2000, 19: 1366–1377.
normally. Ray, E.: Primordial germ-cell development: the zebrafish perspec-
c) The genes in this portion of chromosome 15 are special in that tive. Nat. Rev. Genet. 2003, 4: 690–700.
they are required in two copies for normal development, and so Williamson, A., Lehmann, R.: Germ cell development in Drosophila.
the loss of one set does not allow normal development. Annu. Rev. Cell Dev. Biol. 1996, 12: 365–391.
d) Two copies of every gene in the genome are required for devel-
opment, so loss of one of the copies from this region perturbs 9.2 In mammals germ cells are induced by cell–cell
development. interactions during development
Kurimoto, K., Yamaji, M., Seki, Y., Saitou, M.: Specification of
9. What is the cortical reaction, and why is it important?
the germ cell lineage in mice: a process orchestrated by the
a) The cortical reaction is the depolarization of the plasma mem- PR-domain proteins, Blimp1 and Prdm14. Cell Cycle 2008, 7:
brane after sperm entry, which helps to block polyspermy. 3514–3518.
b) The cortical reaction is the entry of Ca2+ ions into the egg via the McLaren, A.: Primordial germ cells in the mouse. Dev. Biol. 2003,
cortex, which initiates development. 262: 1–15.
 Determination of the sexual phenotype 363

Ohinata, Y., Payer, B., O’Carroll, D., Ancelin, K., Ono, Y., Sano, 9.6 Oocyte development can involve gene amplification and
M., Barton, S.C., Obukhanych, T., Nussenzweig, M., Tara- contributions from other cells
khovsky, A., Saitou, M., Surani, M.A.: Blimp1 is a critical Browder, L.W.: Oogenesis. New York: Plenum Press, 1985.
determinant of the germ cell lineage in mice. Nature 2005, Choo, S., Heinrich, B., Betley, J.N., Chen, A., Deshler, J.O.: Evi-
436: 207–213. dence for common machinery utilized by the early and late
Saitou, M., Barton, S.C., Surani, M.A.: A molecular programme RNA localization pathways in Xenopus oocytes. Dev. Biol. 2004,
for the specification of germ cell fate in mice. Nature 2002, 418: 278: 103–117.
293–300. de Rooij, D.G, Grootegoed, J.A.: Spermatogonial stem cells. Curr.
Saitou, M., Payer, B., Lange, U.C., Erhardt, S., Barton, S.C., Surani, Opin. Cell Biol. 1998, 10: 694–701.
M.A.: Specification of germ cell fate in mice. Philos. Trans. R.
Soc. Lond. B Biol. Sci. 2003, 358: 1363–1370. 9.7 Factors in the cytoplasm maintain the totipotent
potential of the egg
9.3 Germ cells migrate from their site of origin to the gonad
Gurdon, J.B.: Nuclear transplantation in eggs and oocytes. J. Cell
& 9.4 Germ cells are guided to their final destination by
Sci. Suppl. 1986, 4: 287–318.
chemical signals
Deshpande, G., Godishala, A., Schedl, P.: Gγ1, a downstream target 9.8 In mammals some genes controlling embryonic growth
for the hmgcr-isoprenoid biosynthetic pathway, is required for are ‘imprinted’
releasing the Hedgehog ligand and directing germ cell migra-
Ideraabdullah, F.Y., Vigneau, S., Bartolomei, M.S.: Genomic
tion. PLoS Genet. 2009, 5: e1000333.
imprinting mechanisms in mammals. Mutat. Res. 2008, 647:
Doitsidou, M., Reichman-Fried, M., Stebler, J., Koprunner, M.,
77–85.
­Dorries, J., Meyer, D., Esguerra, C.V., Leung, T., Raz, E.: Guidance
Morison, I.M., Ramsay, J.P., Spencer, H.G.: A census of mamma-
of primordial germ cell migration by the chemokine SDF-1.
lian imprinting. Trends Genet. 2005, 21: 457–465.
Cell 2002, 111: 647–659.
Reik, W., Walter, J.: Genomic imprinting: parental influence on
Knaut, H., Werz, C., Geisler, R., Nusslein-Volhard, C., Tubingen 2000
the genome. Nat. Rev. Genet. 2001, 2: 21–32.
Screen Consortium: A zebrafish homologue of the chemokine
Wood, A.J., Oakey, R.J.: Genomic imprinting in mammals: emerg-
receptor Cxcr4 is a germ-cell guidance receptor. Nature 2003,
ing themes and established theories. PLoS Genet. 2006, 2: e147.
421: 279–282.
Wu, H-A., Bernstein, E.: Partners in imprinting: noncoding RNA
Molyneaux, K., Wylie, C.: Primordial germ cell migration. Int.
and Polycomb group proteins. Dev. Cell 2008, 15: 637–638.
J. Dev. Biol. 2004, 48: 537–544.
Pelegri F.: Maternal factors in zebrafish development. Dev. Dyn.
9.9 Fertilization involves cell-surface interactions between
2003, 228: 535–554.
egg and sperm
Raz, E.: Guidance of primordial germ cell migration. Curr. Opin.
Cell Biol. 2004, 16:169–173. Ensslin, M.A., Lyng, R., Raymond, A., Copland, S., Shur, B.D.: ­Novel
Santos, A.C., Lehmann, R.: Germ cell specification and migration gamete receptors that facilitate sperm adhesion to the egg coat.
in Drosophila and beyond. Curr. Biol. 2004, 14: R578–R589. Soc. Reprod. Fertil. Suppl. 2007, 63: 367–383.
Weidinger, G., Wolke, U., Köprunner, M., Thisse, C., Thisse, B., Jungnickel, M.K., Sutton, K.A., Florman, H.M.: In the beginning:
Raz, E.: Regulation of zebrafish primordial germ cell migra- lessons from fertilization in mice and worms. Cell 2003, 114:
tion by attraction towards an intermediate target. Development 401–404.
2002, 129: 25–36. Ohelndieck, K., Lennarz, W.J.: Role of the sea-urchin egg receptor
Wylie, C.: Germ cells. Cell 1999, 96: 165–174. for sperm in gamete interactions. Trends Biochem. Sci. 1995, 20:
29–33.
9.5 Germ-cell differentiation involves a halving of Rubinstein, E., Ziyyat, A., Wolf, J.P., Le Naour, F., Boucheix, C.: The
chromosome number by meiosis molecular players of sperm-egg fusion in mammals. Semin. Cell
De Rooij, D.G., Grootegoed, J.A.: Spermatogonial stem cells. Curr. Dev. Biol. 2006, 17: 254–263.
Opin. Cell Biol. 1998, 10: 694–701. Shur, B.D., Ensslin, M.A., Rodeheffer, C.: SED1 function during
Hultén, M.A., Patel, S.D., Tankimanova, M., Westgren, M., mammalian sperm-egg adhesion. Curr. Opin. Cell Biol. 2004, 16:
Papadogiannakis, N., Jonsson, A.M., Iwarsson, E.: The origins of 477–485.
trisomy 21 Down syndrome. Mol. Cytogenet. 2008, 1: 21–31. Singson, A.: Every sperm is sacred: fertilization in Caenorhabditis
Mehlmann, L.M.: Stops and starts in mammalian oocytes: recent elegans. Dev. Biol. 2001, 230: 101–109.
advances in understanding the regulation of meiotic arrest and
oocyte maturation. Reproduction 2005, 130: 791–799. 9.10 Changes in the egg envelope at fertilization block
Pacchierottia, F., Adler, I.-D., Eichenlaub-Ritter, U., Mailhes, J.B.: polyspermy
Gender effects on the incidence of aneuploidy in mammalian Gardner, A.J., Evans, J.P.: Mammalian membrane block to
germ cells. Environ. Res. 2007, 104: 46–69. polyspermy: new insights into how mammalian eggs prevent
Vogta, E., Kirsch-Volders, M., Parry, J., Eichenlaub-Rittera, U.: fertilization by multiple sperm. Reprod. Fertil. Dev. 2005, 18:
Spindle formation, chromosome segregation and the spindle 53–61.
checkpoint in mammalian oocytes and susceptibility to mei- Hedrick, J.L.: A comparative analysis of molecular mechanisms
otic ­error. Mutat. Res. 2008, 651: 14–29. for blocking polyspermy: identification of a lectin-ligand
364 Chapter 9 Germ cells, fertilization, and sex

­ inding reaction in mammalian eggs. Soc. Reprod. Fertil. Suppl.

b 9.15 Somatic sexual development in Caenorhabditis is
2007, 63: 409–419. determined by the number of X chromosomes
Tian, J., Gong, H., Thomsen, G.H., Lennarz, W.J.: Xenopus laevis Carmi, I., Meyer, B.J.: The primary sex determination signal of
sperm-egg adhesion is regulated by modifications in the sperm Caenorhabditis elegans. Genetics 152: 999–1015.
receptor and the egg vitelline envelope. Dev. Biol. 1997, 187: Meyer, B. J.: X-chromosome dosage compensation. In WormBook
143–153. (June 25, 2005) (edited by The C. elegans Research Community).
Wong, J.L., Wessel, G.M.: Defending the zygote: search for the doi/10.1895/wormbook.1.8.1, http://www.wormbook.org (date
ancestral animal block to polyspermy. Curr. Top. Dev. Biol. accessed 21 May 2010).
2006, 72: 1–151. Raymond, C.S., Shamu, C.E., Shen, M.M., Seifert, K.J., Hirsch, B.,
Hodgkin, J., Zarkower, D.: Evidence for evolutionary conserva-
9.11 Sperm–egg fusion causes a calcium wave that results in
tion of sex-determining genes. Nature 1998, 391: 691–695.
egg activation
Knott, J.G., Gardner, A.J., Madgwick, S., Jones, K.T., Williams, C.J., 9.16 Most flowering plants are hermaphrodite, but some
Schultz, R.M.: Calmodulin-dependent protein kinase II triggers produce unisexual flowers
mouse egg activation and embryo development in the absence Irisa, E.N.: Regulation of sex determination in maize. BioEssays
of Ca2+ oscillations. Dev. Biol. 2006, 296: 388–395. 1996, 18: 363–369.
Swann, K., Yu, Y.: The dynamics of calcium oscillations that acti- Huh, J.H., Bauer, M.J., Hsieh, T.-F., Fischer, R.L.: Cellular program-
vate mammalian eggs. Int. J. Dev. Biol. 2008, 52: 585–594. ming of plant gene imprinting. Cell 2008, 132: 735–744.
Whitaker, M.: Calcium at fertilization and in early development.
Physiol. Rev. 2006, 86: 25–88. 9.17 Determination of germ-cell sex depends on both genetic
Yu, Y., Halet, G., Lai, F.A., Swann, K.: Regulation of diacylglycerol constitution and intercellular signals
production and protein kinase C stimulation during sperm- Childs, A.J., Saunders, P.T.K., Anderson, R.A.: Modelling germ cell
and PLCzeta-mediated mouse egg activation. Biol. Cell 2008, development in vitro. Mol. Hum. Reprod. 2008, 14: 501–511.
100: 633–643. McLaren, A.: Signaling for germ cells. Genes Dev. 1999, 13: 373–376.
Obata, Y., Villemure, M., Kono, T., Taketo, T.: Transmission of
9.12 The primary sex-determining gene in mammals is on
Y chromosomes from XY female mice was made possible by
the Y chromosome
the replacement of cytoplasm during oocyte maturation. Proc.
Capel, B.: The battle of the sexes. Mech. Dev. 2000, 92: 89–103.
Natl Acad. Sci. USA 2008, 105: 13918–13923.
Koopman, P.: The genetics and biology of vertebrate sex determi-
Seydoux, G., Strome, S.: Launching the germline in Caenorhabdi-
nation. Cell 2001, 105: 843–847.
tis elegans: regulation of gene expression in early germ cells.
Schafer, A.J., Goodfellow, P.N.: Sex determination in humans.
­Development 1999, 126: 3275–3283.
BioEssays 1996, 18: 955–963.
9.18 Various strategies are used for dosage compensation of
9.13 Mammalian sexual phenotype is regulated by gonadal
X-linked genes
hormones
Akhtar, A.: Dosage compensation: an intertwined world of RNA and
Swain, A., Lovell-Badge, R.: Mammalian sex determination: a
chromatin remodeling. Curr. Opin. Genet. Dev. 2003, 13: 161–169.
­molecular drama. Genes Dev. 1999, 13: 755–767.
Avner, P., Heard, E.: X-chromosomes inactivation: counting, choice
Vainio, S., Heikkila, M., Kispert, A., Chin, N., McMahon, A.P.:
and initiation. Nat. Rev. Genet. 2001, 2: 59–67.
­Female development in mammals is regulated by Wnt-4 sig-
Csankovszki, G., McDonel, P., Meyer, B.J.: Recruitment and spread-
nalling. Nature 1999, 397: 405–409.
ing of the C. elegans dosage compensation complex along
9.14 The primary sex-determining signal in Drosophila is the X chromosomes. Science 2004, 303: 1182–1185.
number of X chromosomes, and is cell autonomous Latham, K.E.: X chromosome imprinting and inactivation in
preimplantation mammalian embryos. Trends Genet. 2005, 21:
Brennan, J., Capel, B.: One tissue, two fates: molecular genetic
120–127.
events that underlie testis versus ovary development. Nat. Rev.
Panning, B.: X-chromosome inactivation: the molecular basis of
Genet. 2004, 5: 509–520.
silencing. J. Biol. 2008, 7: 30.
Hodgkin, J.: Sex determination compared in Drosophila and
Starmer, J., Magnuson, T.: New model for random X chromosome
Caenorhabditis. Nature 1990, 344: 721–728.
inactivation. Development 2009, 136: 1–10.
MacLaughlin, D.T., Donahoe, M.D.: Sex determination and differ-
entiation. New Engl. J. Med. 2004, 350: 367–378.
 10
Cell differentiation and
stem cells
nn The control of gene nn Models of cell nn The plasticity of gene
expression differentiation expression

The differentiation of unspecialized cells into many different cell types occurs first
in the developing embryo and continues after birth and throughout adulthood. The
character of specialized cells, such as blood or muscle cells, is the result of a ­particular
pattern of gene activity, which determines which proteins are synthesized. Thus, how
this particular pattern of gene activity develops is a central question in cell differenti-
ation, and is discussed in this chapter in relation to some well-studied model systems,
including muscle, blood, and neural crest cell differentiation. We shall see that gene
expression is under a complex set of controls that include the actions of transcription
factors, chemical modification of DNA, and post-translational modifications to chroma-
tin proteins. External signals play a key role in differentiation by triggering intracellular
signaling pathways that affect gene expression. Another key question in develop-
ment is the plasticity of differentiated cells—that is, their capacity to change their
state of differentiation. As we shall see, although one cell type rarely changes into
another under normal circumstances, such changes are possible. It is well-­established
that a nucleus from a differentiated cell can be reprogrammed when transferred into
an enucleated egg, and become able to direct embryonic development, the process
of cloning. Pluripotent embryonic stem cells and multipotent adult stem cells provide
insights into the mechanisms that underlie the flexibility and reversibility of a cell’s
differentiated state, and have potential uses in regenerative medicine.

Embryonic cells start out morphologically similar to each other but eventually become
different, acquiring distinct identities and specialized functions. As we have seen in
previous chapters, many developmental processes, such as the early specification
of the germ layers, involve transient changes in cell form, patterns of gene activity,
and the proteins synthesized. Cell differentiation, in contrast, involves the gradual
emergence of cell types that have a clear-cut identity in the adult, such as nerve cells,
red blood cells, and fat cells (Fig. 10.1). There are more than 200 clearly recognizable
differentiated cell types in mammals. Initially, embryonic cells fated to become differ-
ent cell types differ primarily from each other only in their pattern of gene activity,
and thus in the proteins present. Differentiation occurs over successive cell genera-
tions, the cells gradually acquiring new structural features while their potential fates
become more and more restricted.
366 Chapter 10 Cell differentiation and stem cells

Epithelial cell Fat cell Nerve cell

Olfactory neuron Red blood cell Retinal rod

Fig. 10.1 Differentiated cell types.

Mammalian cell types come in various shapes
and sizes. Scale bars: epithelial cell, 15 µm;
fat cell, 100 µm; nerve cell, 100 µm–1 m;
olfactory neuron, 8 µm; red blood cell, 8 µm;
retinal rod, 20 µm.

Differentiated cells serve specialized functions and have achieved a terminal and
usually stable state, in contrast to many of the transitory differences characteristic of
earlier stages in development. The precursors of cartilage and muscle cells have no
obvious structural differences from each other and so look the same and might be
described as undifferentiated; nevertheless, they will differentiate as cartilage and
muscle, respectively, when cultured under appropriate conditions. Similarly, at an
early stage in their differentiation, the precursors of white blood cells are structurally
indistinguishable from those of red blood cells, but are distinct in the proteins they
express. The differences between the precursor cells at these early stages reflect dif-
ferences in gene activity and thus the proteins they contain, which control their future
development.
As with earlier processes in development, the central feature of cell differentiation
is a change in gene expression. The genes expressed in a differentiated cell include
not only those for a wide range of ‘housekeeping’ proteins, such as the glycolytic
enzymes involved in energy metabolism, but also genes encoding cell-specific pro-
teins that characterize a fully differentiated cell: hemoglobin in red blood cells, keratin
in skin epidermal cells, and muscle-specific actins and myosins. It is important to
realize how many different genes are active in any given cell in the embryo at any
one time—it is of the order of several thousand, only a small number of which may
be involved in specifying cell fate or differentiation. All the genes that are being
expressed in a particular tissue or at a particular stage of development can be detected
by DNA microarray techniques (see Box 3A, p. 115).
A differentiated cell is characterized by the proteins it contains. The presence of
different proteins can result in considerable structural changes; mature mammalian
red blood cells lose their nucleus and become biconcave discs stuffed with hemoglo-
bin, whereas neutrophils, a type of white blood cell, develop a multi-lobed nucleus
 The control of gene expression 367

and a cytoplasm filled with secretory granules. Yet both have a similar early lineage.
Expression of a single protein can change a cell’s differentiated state. If the gene
myoD is introduced into fibroblasts, for example, they will develop into muscle cells,
as myoD encodes a master transcriptional regulator of muscle differentiation. As we
shall see later in this chapter, a very dramatic effect is obtained when genes encoding
four of the transcription factors expressed in embryonic stem cells (ES cells) are intro-
duced into fibroblasts. The fibroblasts become pluripotent, which means that, like ES
cells, they can be induced to differentiate into cells belonging to tissues derived from
all three germ layers.
In the earliest stages of differentiation, differences between cells are not easily
detected and probably consist of subtle alterations caused by a change in activ-
ity of a few genes. At this early stage, cells become determined, or committed,
with respect to their developmental potential (see Section 1.12). The mesoderm of
somites, for example, gives rise during normal development to muscle, cartilage,
dermis, and vascular tissue, but not to other cell types. Cells that are determined
with respect to their eventual fate will form only those appropriate cell types when
grafted to a different site in the embryo; they retain their identity (see Fig. 1.21).
Once determined for a particular fate, cells pass on that determined state to all
their progeny.
Cell differentiation is known to be controlled by a wide range of external signals,
from cell-surface proteins to secreted polypeptide cytokines and molecules of the
extracellular matrix. Examples of all of these signals will be encountered in this chap-
ter, but it is important to remember that, while the external signals that stimulate dif-
ferentiation are often referred to as being ‘instructive,’ they are in general ‘selective,’ in
the sense that, in practice, the number of developmental options open to a cell at any
given time is limited. These options are set by the cell’s internal state, which in turn
reflects its developmental history. External signals cannot, for example, convert an
endodermal cell into a muscle or nerve cell. Nevertheless the same external signals,
such as Wnts, FGFs, and members of the TGF-β family, can be used again and again
in different circumstances without upsetting previous development, as we have seen
in early vertebrate development (see Box 4A, p. 131).
There can be a conflict between cell division and cell differentiation, with a ces-
sation of cell proliferation being necessary for full cell-differentiation to occur. Cell
proliferation is most evident before the terminal stage of differentiation, and most
terminally differentiated cells divide only rarely. Some cells, such as skeletal muscle
and nerve cells, do not divide at all after they have become fully differentiated. In
cells that do divide after differentiation, the differentiated state, like the determined
state, is passed on through all subsequent cell divisions. As the pattern of gene
activity is the key feature in cell differentiation, this raises the question of how a
particular pattern of gene activity is first established, and then how it is passed on
to daughter cells.
We start this chapter with a consideration of the general mechanisms by which
a pattern of gene activity can be established, maintained, and inherited at cell divi-
sion. We then turn to a major question, namely the molecular basis of the specificity
of cell differentiation, with muscle cells, blood cells, and neural crest cells providing
our main examples. Finally, we look at the reversibility and plasticity of the differen-
tiated state, particularly the special properties of stem cells, and the reprogramming
of differentiated nuclei in oocytes that makes it possible to clone some animals—
that is, to make an animal that is genetically identical to the animal from which the
nucleus came. We shall also discuss the exciting recent discovery that pluripotency
can be induced in fully differentiated cells taken from adults. This chapter concen-
trates on differentiation in animal cells. As we have seen in Chapter 7, plant cells do
not have a permanently determined state, and a single somatic cell can give rise to
a whole plant.
368 Chapter 10 Cell differentiation and stem cells

The control of gene expression

Every nucleus in the body of a multicellular organism is derived from the single
­zygotic nucleus in the fertilized egg. But patterns of gene activity in differentiated
cells vary enormously from one cell type to another. The egg itself has a pattern of
gene activity that is different from that of the embryonic cells to which it gives rise.
This raises the question of how the particular pattern of gene activity in a differenti-
ated cell is specified and how it is inherited.
To understand the molecular basis of cell differentiation, we first need to know
how a gene can be expressed in a cell-specific manner. Why does a certain gene get
switched on in one cell and not in another? We focus here on the regulation of tran-
scription, which is the first (and generally the most significant) step in the expression
of a gene (see ­Section 1.10). Control of protein synthesis can, however, also occur
after transcription; for example, at the translation of mRNA to produce protein. We
have seen an example of translational regulation in early Drosophila development,
where the Nanos protein controls the production of Hunchback protein by preventing
translation of maternal hunchback RNA (see Section 2.9), and in Caenorhabditis ele-
gans, where a ­microRNA prevents the translation of lin-14 RNA to give LIN-14 protein
(see Section 6.5). In Xenopus, large numbers of maternal mRNAs are laid down in the
egg but many are not translated until after the egg is fertilized (see Section 4.1).

10.1 Control of transcription involves both general and tissue-specific

transcriptional regulators

Most of the key genes in development are initially in an inactive state and require
activating transcription factors, or activators, to turn them on. These activators bind
to specific cis-regulatory control regions in the DNA surrounding the gene, which
are often called enhancers. The importance of these regions in tissue-specific gene
expression was clearly demonstrated by experiments in which the control region of
Fig. 10.2 Tissue-specific gene expression one tissue-specific gene was replaced by the control region of another (Fig. 10.2). We
is controlled by regulatory regions. The have already seen how such replacements are routinely used in experimental devel-
control region of the mouse elastase I gene
opmental biology to force the expression or misexpression of a gene in a particular
is joined to a DNA sequence coding for
tissue, and to mark cells expressing a particular gene in order to investigate its devel-
human growth hormone. This DNA construct
is injected into the nucleus of a fertilized
opmental function (see Box 1D, p. 20 and Box 2D, p. 61).
mouse egg, where it becomes integrated In eukaryotic cells, most regulated protein-coding genes are transcribed by RNA
into the genome. When the mouse develops, polymerase II. The binding of RNA polymerase to the promoter region of a gene in
human growth hormone is made in the order to start transcription requires the cooperation of a set of so-called general tran-
pancreas. A 213-base-pair stretch of DNA scription factors, which form an initiation complex with the polymerase at the pro-
that includes the elastase I promoter and moter and help to activate the polymerase to start transcription at the correct place.
other control sites is sufficient to drive For highly regulated genes, such as those involved in development, this ­complex
growth hormone expression in the pancreas. cannot be formed and activated without additional help. The binding of activator pro-
Normally, growth hormone is made only teins to sites in the cis-regulatory control regions of the gene is needed to help ­attract
in the pituitary, and elastase I only in the
pancreas.

Elastase I gene from mouse and DNA construct engineered, containing

control region for elastase I and the gene Fertilized mouse egg is injected Human growth hormone is made
growth hormone gene from human with the DNA construct in the mouse pancreas
are identified and isolated for human growth hormone

control mouse elastase I

region gene

control human growth

region hormone gene
 The control of gene expression 369

repressor
activator
enhancer
enhancer
activators
co-activators
DNA

mediator general transcription factors

RNA
TFIID polymerase II
TATA box coding region

promoter RNA transcript

Fig. 10.3 Gene expression is regulated by the coordinated promoter and also further upstream, and sometimes downstream, of
action of gene-regulatory proteins that bind to control regions in the coding region. The control regions of all highly regulated genes
DNA. Most protein-coding genes in eukaryotic cells, including all highly also include sites for repressor proteins, whose binding prevents
regulated developmental genes, are transcribed by RNA polymerase II. transcription, for example, in the wrong tissue or at the wrong time
The transcription machinery itself, composed of the polymerase and during development. Control sites can be many kilobases away from
its associated general transcription factors (the core transcriptional the start point of transcription. The proteins bound to these distant
machinery), binds to the promoter, which contains some binding sites, sites are brought into contact with the transcription machinery by
such as the TATA box, that are common to many genes transcribed looping of the DNA. The Mediator protein complex is a large multi-
by RNA polymerase II. The TATA box is recognized by the TATA-box protein complex that seems to be present in many RNA polymerase II
binding protein (TBP), a component of the general transcription factor initiation complexes and its role is to link the gene-regulatory proteins
TFIID. Other control sites known as enhancers must be occupied by to the core transcriptional machinery. The fully operational initiation
specific activator proteins and their associated co-activators before complex activates the RNA polymerase and releases it to start
transcription can be initiated. These sites are located adjacent to the transcription.

the general transcription factors and polymerase and position them correctly on the
promoter (Fig. 10.3).
For any given gene, the specificity of its activation is due to particular combinations
of gene-regulatory proteins binding to individual sites in the control regions. These
sites are typically 7–10 nucleotides long. At least 1000 different transcription factors
are encoded in the genomes of Drosophila and C. elegans, and as many as 3000 in the
human genome. On average, around five different transcription factors act together
at a control region, and in some cases considerably more. This combinatorial action
of gene-regulatory proteins is a key principle in the control of gene expression and
is fundamental to the exquisite control and complexity of the gene expression that
drives development. As well as sites that bind activators, control regions may con-
tain sites that bind repressors, proteins that inhibit gene expression; these prevent
a gene being expressed at the wrong time or in the wrong place. Examples of such
complex control regions that regulate the temporal and spatial expression of a gene
during embryonic development are shown in Fig. 2.34 for the Drosophila even-skipped
(eve) pair-rule gene and in Fig. 6.26 for the sea urchin Endo-16 gene, which encodes
a calcium-binding extracellular matrix protein. The spatially delimited expression of
eve in the second of seven transverse stripes in the embryo involves four different
gene-regulatory proteins, which include activators and repressors, acting at 11 distinct
sites. The correct expression of Endo-16 throughout sea-urchin embryo development
requires the participation of 13 gene-regulatory proteins acting at a potential 56 regu-
latory motifs within a control region that spans 2.3 kilobases.
Some control sites are located within the promoter region and are present in similar
positions in most protein-coding genes. These sites, such as the TATA box, bind the
general transcription factors and the RNA polymerase (see Fig. 10.3). Tissue-specific or
developmental-stage expression is controlled by sites outside the immediate promoter
370 Chapter 10 Cell differentiation and stem cells

region, such as those at which tissue-specific gene-regulatory proteins act. These sites
vary greatly in type and position from gene to gene, and may be thousands of base
pairs away from the start point of transcription. These distant sites are able to control
gene activity because DNA can form loops, thus bringing the sites and their bound
proteins into close proximity to the promoter. Proteins bound at distant sites can thus
make contact with proteins at the promoter, forming a fully active transcription initia-
tion complex (see Fig. 10.3).
Another important class of regulatory proteins comprises the co-activators and
co-repressors, which do not bind DNA themselves but function to link the transcrip-
tional machinery to the DNA-binding activator and repressor proteins (see Fig. 10.3).
One co-activator we have already encountered is β-catenin, the end product of the
Wnt signaling pathway, which binds transcription factors of the TCF/LEF family (see
Box 1E, p. 26). In the absence of Wnt signaling, TCFs bound to control elements in
genes that are targets of Wnt signaling are bound by a co-repressor, called Groucho in
Drosophila and TLE in humans, and transcription is repressed. β-Catenin accumulat-
ing in the nucleus as the result of Wnt signaling displaces Groucho and binds to TCF,
turning it into a transcriptional activator (see Box 1E, p. 26).
Transcriptional regulators fall into two main groups: those that are required for the
transcription of a wide range of genes, and which are found in many different cell
types; and those that are required for a particular gene or set of genes whose expres-
sion is tissue restricted, or tissue-specific, and which are only found in one or a very
few cell types. We shall encounter both types of transcription factor in the following
sections, when we look at the expression of globin genes in red blood cell precursors
and the regulation of muscle genes in muscle precursor cells. In general, it can be
assumed that activation of each gene involves a unique combination of transcription
factors. It should also be noted that while differentiation may involve a relatively small
number of genes, the number is not easy to establish, as each cell has several thousand
active genes, most of which are probably concerned with housekeeping functions.

10.2 External signals can activate gene expression

Most of the molecules that act as signals between cells during development are pro-
teins, and in the case of developmentally important signals their effect is usually to
induce a change in gene expression. Protein signals bind to receptors in the cell mem-
brane and the signal is relayed to the cell nucleus by intracellular signaling pathways
such as the Wnt, Hedgehog, TGF-β, and FGF pathways we have already encountered
(see Box 1E, p. 26, Fig. 2.40, Fig. 4.33, and Box 5B, p. 184). Non-protein, lipid-soluble
signal molecules, such as retinoic acid, cross the plasma membrane unaided and also
stimulate an intracellular signal pathway (see Box 5D, pp. 192–193). Once inside the
cell, retinoic acid binds to receptor proteins, and the complex of retinoic acid and recep-
tor is then able to act as a transcriptional regulator, binding directly to control sites
in the DNA, known as retinoic acid response elements, to activate (or in some cases
repress) transcription. Other lipid-soluble signal molecules that act through similar
intracellular receptors include the steroid hormones, such as testosterone and estrogen
in mammals. As we saw in Chapter 9, testosterone produced in the testis is respon-
sible for the secondary sexual characteristics that make male mammals different from
females. In insects, the steroid hormone ecdysone is responsible for metamorphosis
(see Chapter 13) and induces the differentiation of a wide range of cell types. In all
these cases, the hormone turns a whole set of genes on or off by acting, via its intracel-
lular receptors, at a control element that is present in all these genes.
Estrogen is also involved in a classic example of tissue-specific gene expression, the
induction of chick oviduct cells to produce the protein ovalbumin, a major ­component
of egg white. The continued presence of estrogen is required for transcription of
the ovalbumin gene; when estrogen is withdrawn, ovalbumin mRNA and protein
 The control of gene expression 371

­ isappear. The circulating estrogen activates expression of the ovalbumin gene only
d
in the cells of the oviduct, whereas the ovalbumin gene in other cells, such as those
of the liver, is unaffected. Indeed, the expression of the ovalbumin gene is among the
most restricted known, but the precise mechanisms that lead to such a tissue-specific
response are still elusive after decades of research. There is evidence that estrogen
affects the structure of the chromosomal region containing the ovalbumin gene in
oviduct cells but not others, thus allowing the hormone–receptor complexes access
to the regulatory DNA of the gene. This tissue-specific effect of estrogen implies that
other tissue-specific activators, repressors, and co-regulators exist that allow ovalbu-
min gene expression in the oviduct when stimulated with estrogen and prevent its
expression in all other tissues.

10.3 The maintenance and inheritance of patterns of gene activity depend

on chemical and structural modifications to chromatin, as well as on gene-
regulatory proteins

A central feature of development in general, and cell differentiation in particular, is

that some genes are maintained in an active state, while others are repressed and inac-
tive. Many differentiated cells, including fibroblasts, liver cells, and myoblasts (cells
that give rise to muscle cells), continue dividing after differentiation, and the particu-
lar pattern of gene activity characterizing the differentiated state is reliably transmit-
ted through many cell divisions. Developmental genes generally have complex control
regions, containing binding sites for a range of activating or inhibitory transcription
factors, and whether or not a gene is activated depends on the precise combinations
and levels of these transcription factors and their associated co-­activators and co-
repressors. A given gene-regulatory protein can also affect the activity of several, often
many, other genes, activating some and repressing others, which provides an efficient
means of coordinating the expression of large numbers of genes (Fig. 10.4).

gene X

mRNA

5’ 3’

transcription factor

Fig. 10.4 A gene can regulate the

gene A gene B gene C gene D gene E activity of other genes when it codes
ON ON ON OFF ON
for a transcription factor. Transcription
factors can activate or repress the activity
of genes by binding to their control regions.
no protein D In the illustration, the activation of gene X,
encoding a transcription factor, eventually
leads to the production of four new proteins
(A, B, C, and E), and the repression of
protein A protein B protein C protein E protein D production.
After Alberts, B., et al.: 2002.
372 Chapter 10 Cell differentiation and stem cells

Protein product of gene A enters nucleus and binds control region for gene A in a positive feedback
loop. Protein A also represses gene B and activates gene C

gene A gene B gene C

Fig. 10.5 Continued expression of gene

regulatory proteins could maintain a
pattern of differentiated gene activity.
Top panel: transcription factor A is produced
by gene A, and acts as a positive gene-
Transcription of gene A, Following cell division none Protein A enters the nucleus
regulatory protein for its own control region. repression of gene B and and the original pattern of
of the genes are active
Once activated, therefore, gene A remains activation of gene C gene activity is established
switched on and the cell always contains A.
Transcription factor A also acts on the control
regions of genes B and C to repress and
activate them, respectively, setting up a cell-
specific pattern of gene expression. Bottom
panels: after cell division, the cytoplasm
of both daughter cells contains sufficient
amounts of protein A to reactivate gene A,
and thus maintain the pattern of expression
of genes B and C.

One way of maintaining a particular pattern of gene activity is to ensure the contin-
ued production of these specific gene-regulatory proteins in the differentiated cell and
its progeny. If the gene product itself acts as the positive regulatory protein, all that is
then required for the pattern of gene activity to be maintained is the initial event that
first activates the gene; once switched on the gene remains active (Fig. 10.5). This type
of positive-feedback control of gene expression occurs in muscle-cell differentiation,
where the MyoD protein acts as an activator of myoD gene expression. In Drosophila,
the selector gene engrailed remains active in the posterior compartment of segments
throughout embryonic and larval development and into the adult (see Chapter 2). Its
expression is at least partly maintained by a similar positive feedback of the Engrailed
transcription factor on its own gene, long after the pair-rule transcription factors that
initiated expression of engrailed in the embryo have disappeared.
Another mechanism for maintaining and inheriting a pattern of active and inactive
genes relies on chemical and structural changes in chromatin—the complex of DNA,
histones, and other proteins of which chromosomes are made. Whether a stretch of
chromosome can be transcribed depends on whether or not it is packaged into a chro-
matin structure that is accessible to transcription factors, RNA polymerase, and other
necessary proteins. The compaction of chromatin that occurs at mitosis, for example,
when the condensed chromosomes become visible under the light microscope, ren-
ders the chromosomes transcriptionally inactive. Localized structural alterations of
this sort that persist throughout the cell cycle and are reinstated after cell division
provide a means of shutting down, or silencing, genes in the long term. Indeed, this
is one way in which expression of the engrailed gene is prevented in those regions of
the fly in which it is not required. Chromatin that is packaged into a structure that
cannot be transcribed is called heterochromatin.
Evidence that changes in the packing state of chromatin can be inherited through
many cell divisions and can keep genes inactive over a long period was first pro-
vided by the phenomenon of X-chromosome inactivation in female mammals. One of
the two X chromosomes in each somatic cell in females becomes highly condensed
and inactive early in embryogenesis and is maintained in this state throughout the
 The control of gene expression 373

i­ndividual’s lifetime. X inactivation is a dosage-compensation mechanism to equalize

the level of gene expression from the X chromosome in males and females. The silenc-
ing mechanism involves the production of a non-coding RNA that coats the chromo-
some and is described in Section 9.17. Inactivation is not irreversible, however, as the
inactive X is reactivated in the germline during the formation of egg cells, which each
end up containing a single X chromosome.
Localized changes in chromatin structure and protein composition in chromosomes
are thought to provide a general mechanism for long-term gene inactivation. A reflec-
tion of the difference in chromatin structure between active and inactive genes is
that some active genes show a heightened sensitivity to digestion with the enzyme
DNase I at sites known as hypersensitive sites. This is just one indication that in
actively transcribed genes, the chromatin is packaged into a more ‘open’ structure
that allows enzymes and other proteins, including the transcription apparatus, access
to the DNA. When the chromatin is packed into heterochromatin, on the other hand,
the DNA is inaccessible to DNase I and is protected from digestion.
Underlying these changes in the transcriptional properties of chromatin are chemi-
cal modifications to the DNA itself and to the associated histone proteins (Box 10A,
p. 374). In vertebrates, methylation of cytosine at certain sites in the DNA is correlated
with the absence of transcription in those regions. Moreover, the pattern of methyla-
tion can be faithfully inherited when the DNA replicates, thus providing a mechanism
for passing on a pattern of gene activity to daughter cells (Fig. 10.6). After DNA
replication, DNA methylase recognizes the presence of a methyl group on a cytosine Fig. 10.6 Inheritance of the pattern of
and methylates the corresponding group on the other strand. DNA methylation is one methylation on DNA. Many cytosine bases
occurring in cytosine-guanine (CG) pairs in
way in which certain genes are rendered inactive in the maternal genome but not in
vertebrate DNA are methylated (a CH3 group
the paternal genome in the phenomenon of genomic imprinting (see Section 9.8). The
is added). When the DNA is replicated, the
inactive X chromosome also has a pattern of DNA methylation that differs from that of
new complementary strand lacks methylation
the active X, and this methylation is likely to play a part in keeping it inactive. but the pattern can be reinstated on the new
The post-translational modification of histone proteins is also closely linked to reg- strand by methylase enzymes that recognize
ulation of gene activity. The histones in chromatin can undergo modifications such as the methylated cytosine of the CG pair on the
methylation, acetylation, or phosphorylation on specific amino-acid residues, and this old DNA strand and methylate the cytosine of
can change the properties of the chromatin (see Box 10A, p. 374). the corresponding CG pair on the new strand.
After Alberts, B., et al.: 2002.

Daughter DNA helices lack Parental pattern of DNA methylation

methyl groups on new strands is inherited in both daughter helices

In vertebrate DNA a large CH3 CH3

proportion of cytosine bases in
CG pairs are methylated A C G T A T C G T A C G T A T C G T
5’ 3’ 5’ 3’
methylated Methylation
cytosine
CH3 3’ 5’ 3’ 5’
A C G T A T C G T T G C A T A G C A T G C A T A G C A
5’ 3’ CH3
DNA recognized by
not methyl-directed
replication recognized methylating enzyme
3’ 5’ CH3
T G C A T A G C A New DNA
strands shown A C G T A T C G T A C G T A T C G T
CH3 in yellow 5’ 3’ 5’ 3’
unmethylated
cytosine Methylation
3’ 5’ 3’ 5’
T G C A T A G C A T G C A T A G C A
CH3 CH3
374 Chapter 10 Cell differentiation and stem cells

Box 10A Histones and Hox genes

The post-translational modification

of histone proteins is very closely Expression of Hox genes over time
linked to regulation of gene activ- H3 K27 methylation
H3 acetylation
ity. The histones in chromatin can
undergo modifications, such as E8.5
tailbuds
methylation, acetylation, or phos-
phorylation, on specific amino-acid
residues, particularly lysine (K), and d13 d12 d11 d10 d9 d8
this can change the transcriptional
properties of the chromatin. Histone
acetylation, for example, tends to E9.5
be associated with regions of chro- tailbuds
matin that are being transcribed,
whereas methylation of particular
d13 d12 d11 d10 d9 d8
lysine residues is more characteris-
tic of transcriptionally silent hetero-
chromatin, whereas methylation at
other residues allows transcription. Changes such as DNA methy- cluster has been followed in excised mouse tail buds at suc-
lation, histone acetylation, and histone methylation are all thought cessive stages of development using chromatin immunopre-
to exert their effect on chromatin structure and gene expression cipitation (ChIP) to detected the modified histone proteins and
through the proteins they recruit to the altered sites. ChIP-chip to detect the target genes (see Section 3.11). At stage
Changes in chromatin status are often self-propagating as E8.5, the genes Hoxd1–9 are already expressed and their chro-
histone modifications, such as acetylation/deacetylation or matin is heavily acetylated, but there are few acetylation marks
methylation/demethylation, recruit the corresponding chromatin- on Hoxd10–13, which are not being expressed at this stage,
­modifying enzymes to the chromatin, causing a ‘domino effect’ although acetylation is starting to appear on Hoxd10. Twelve
that automatically propagates the change in chromatin structure hours later, Hoxd10 is more heavily acetylated and is expressed.
along the chromosome. At stage E9.5, histone acetylation has spread over the remain-
This type of effect is likely to underlie the typical sequential ing Hoxd genes, which are all now starting to be expressed. The
activation of Hox genes in the order they occur in the gene clus- opposite is seen when the progress of repressive methylation on
ter (see Box 5E, p. 194 and Fig. 5.24). The relationship between H3 lysine 27 is tracked, with less methylation on active genes
histone modification and expression of the genes of the Hoxd and strong methylation on those not yet expressed.

Summary

Transcriptional control is a key feature of cell differentiation. Tissue-specific expression

of a eukaryotic gene depends on regulatory sites located in the cis-regulatory regions
flanking the gene. These sites comprise both the promoter sequences adjacent to the
start site of transcription, which bind RNA polymerase, and more distant sites that con-
trol the tissue-specific or developmental-stage—specific expression of the gene. The
combination of different regulatory proteins bound to sites in the cis-regulatory regions
determines whether a gene is active or not, and small differences in regulatory regions
between genes can have significant effects on their pattern of expression. In some
cases, cell-specific expression is due to the correct combination of regulatory proteins
only being present in that cell type; in other cases, gene expression may be prevented
by the gene being packed away in a state in which it is inaccessible to transcription fac-
tors and RNA polymerase. Gene-regulatory proteins interact not only with DNA but also
with each other, forming multi-component transcription complexes that are responsible
for initiating transcription by RNA polymerase. Extracellular signals, such as protein
 Models of cell differentiation 375

growth factors, do not themselves enter the cell but can specifically influence gene
expression by acting at cell-surface receptors to stimulate intracellular signal transduc-
tion pathways. In contrast, signal molecules that can diffuse through cell membranes,
such as retinoic acid or steroid hormones, form a complex with an intracellular receptor
protein. The hormone–receptor complex then binds to the DNA at specific regulatory
sites and acts as a transcription factor to influence gene expression.
Once established, the pattern of gene activity in a differentiated cell can be main-
tained over long periods of time and is transmitted to the cell’s progeny. Mechanisms
that maintain the pattern include the continued action and production of gene regula-
tory proteins, and long-term localized changes in the structure and properties of chro-
matin that are due to chemical modifications to DNA and to histones.

Summary: control of specific gene expression by external signals

signal transduction steroid hormones

modification of a combination activation of intracellular receptor

of transcription factors which acts as a transcription factor

transcription factors bind

to cis-control regions in DNA

gene activity switched on or off

pattern of gene activity is passed on to daughter cells by various mechanisms

positive feedback control DNA methylation changes in chromatin

of gene expression and state (e.g. in inactivated
interaction with other genes X chromosome)

Models of cell differentiation

Cellular differentiation is by no means confined to the embryo. A number of adult tis-

sues, such as blood, skin, and the gut epithelial lining, are continually being renewed
from undifferentiated stem cells, and are much easier to investigate than embryos.
They are therefore among the best-studied examples of cell differentiation. We shall
also consider skeletal muscle, which, although mainly differentiating in the embryo,
can be renewed if necessary from stem cells in the adult. Embryonic neural crest cells,
which we discussed in Chapter 8 in regard to their migration, give rise to a wide range
of different structures, and we shall look here at some aspects of their differentiation.
Finally, we consider apoptosis, or programmed cell death, which is the fate of very
large numbers of developing cells, both in embryonic development and in the adult.

10.4 All blood cells are derived from multipotent stem cells

Hematopoiesis, or blood formation, is a particularly well-studied example of differen-

tiation, partly because cells at all stages are relatively accessible in adult animals and
can be grown in culture, and also because of its medical importance. In hematopoi-
esis, we are looking at the differentiation of a multipotent stem cell—a stem cell that
can give rise to a limited range of differentiated cell types (see Section 1.17). We first
look at the general process of hematopoiesis and at the external signals that influence
376 Chapter 10 Cell differentiation and stem cells

Bone marrow

hematopoietic stem cell

Bone marrow

megakaryocyte/
common common myeloid granulocyte/macrophage erythrocyte
lymphoid progenitor progenitor progenitor progenitor megakaryocyte erythroblast

Blood

granulocytes (or polymorphonuclear leukocytes)

unknown
B cell T cell neutrophil eosinophil basophil precursor monocyte platelets erythrocyte

Effector cells Tissues

activated
plasma cell T cell mast cell osteoclast macrophage

Fig. 10.7 The derivation of blood cells

from multipotent stem cells. Multipotent
it, and then consider the control of gene expression in one cell type—the developing
stem cells in the bone marrow give rise to
red blood cell.
all of the blood cells, as well as to tissue
macrophages, mast cells, and osteoclasts All the blood cells in an adult mammal originate from hematopoietic stem cells
in bone. As well as renewing themselves, located in the bone marrow. Hematopoietic stem cells originate from mesoderm in
the stem cells are thought to give rise to the yolk-sac blood islands of the early embryo; later, self-renewing hematopoietic
uncommitted progenitors, which then split stem cells are found in a region of the aorta and then in a number of definitive blood-
into more committed myeloid, erythroid, and forming sites, including the fetal liver and adult bone marrow. The stem cells are
lymphoid precursors. These then give rise self-renewing and are the precursors of progenitor cells that become committed to one
to the different blood cell lineages (not all of the hematopoietic lineages at a later stage. Thus, hematopoiesis is, in effect, a com-
final differentiated cell types are shown). plete developmental system in miniature, in which a single cell—the multipotent stem
Hematopoietic growth factors influence
cell—gives rise to numerous different blood cell types. There is a continual turnover
the proliferation and differentiation of the
of blood cells, so that hematopoiesis must continue throughout life. As a measure of
different lineages.
the complexity of hematopoiesis, hematopoietic stem cells from fetal liver have been
After Janeway, C.A., et al.: 2005.
found to express at least 200 transcription factors, a similar number of membrane-
associated proteins, and around 150 signaling molecules.
Mammalian blood cells comprise numerous fully differentiated cell types, together
with immature cells at various stages of differentiation (Fig. 10.7). The cell types are
derived from three main lineages—the erythroid, lymphoid, and myeloid ­lineages. The
 Models of cell differentiation 377

first yields the red blood cells, or erythrocytes, and the megakaryocytes, which give
rise to blood platelets. The lymphoid lineage produces the lymphocytes, which include
the two antigen-specific cell types of the immune system, the B and T lymphocytes.
In mammals, B lymphocytes develop in the bone marrow, whereas T lymphocytes
develop in the thymus, from precursors that originate from the multipotent stem cells
in the bone marrow and migrate from the bloodstream into the thymus. Both B and
T lymphocytes undergo a further terminal differentiation after they encounter anti-
gen. Terminally differentiated B lymphocytes become antibody-secreting plasma cells,
while T cells undergo terminal differentiation into at least three functionally distinct
effector T cells. The B and T lymphocytes of the immune system are the only known
examples in vertebrates in which the DNA in the differentiated cell becomes irrevers-
ibly altered. The myeloid lineage gives rise to the rest of the white blood cells, or
leukocytes. These are the eosinophils, neutrophils, and basophils (collectively known
as granulocytes or polymorphonuclear leukocytes), mast cells, and monocytes. Mono-
cytes differentiate into macrophages after they have left the bone marrow and entered
tissues. Mast cells also reside in tissues.
Within the bone marrow, the different types of blood cells and their precursors are
intimately mixed up with connective-tissue cells—the bone marrow stromal cells.
The ability of stem cells both to self-renew and to differentiate into different cell types
is tightly regulated by the cells and molecules in their immediate environment—the
‘niche’ that they inhabit. In the case of hematopoiesis, the stem-cell niche is provided
by the bone marrow stromal cells. Secreted signal proteins of the Wnt and BMP fami-
lies (see Box 4A, p. 131), together with retinoic acid, prostaglandin E2 and Notch sig-
naling, have been implicated in maintaining stem-cell proliferation and self-renewal.
The existence of the multipotent stem cell can be inferred from the ability of bone
marrow to reconstitute a complete blood and immune system when transplanted into
individuals whose own bone marrow has been destroyed, a property that is exploited
therapeutically in the use of bone marrow transplantation to treat diseases of the
blood and immune system. The key experiment that showed this property was the
transfusion of a suspension of bone marrow cells into a mouse that had received a
potentially lethal dose of X-irradiation. The mouse would normally have died as a
result of lack of blood cells, which, like other proliferating cells, are particularly sensi-
tive to radiation, but transfusion of bone marrow cells reconstituted the hematopoietic
system and the mouse recovered. Multipotent hematopoietic stem cells can be grown
in culture, where they also proliferate extensively and differentiate into blood cells.
The multipotent stem cell generates progenitor cells that become irreversibly com-
mitted to one or other of the lineages leading to the different blood cell types. There
is early commitment to either the myeloid or lymphoid lineage, for example, followed
by later commitment to the individual lineages leading to the differentiated cell types.
The hematopoietic system can thus be considered as a hierarchical system with the
multipotent hematopoietic stem cell at the top (see Fig. 10.7). This stem cell gener-
ates uncommitted precursors that then become committed to the main lineages, after
which they undergo further rounds of proliferation and further commitment, finally
undergoing overt differentiation into the different cell types.
All this activity occurs in the microenvironment of the bone marrow stroma, and
is regulated by external signals that are provided by the stromal cells in the form of
hematopoietic growth factors and other cytokines. In this way, the numbers of the
different types of blood cells can be regulated according to the individual’s physi-
ological need. Loss of blood results in an increase in red cell production, for example,
whereas infection leads to an increase in the production of lymphocytes and other
white blood cells.
A central question in considering the behavior of stem cells is how a single stem
cell can divide to produce two daughter cells, one of which remains a stem cell,
while the other gives rise to a lineage of differentiating cells. One possibility is that
378 Chapter 10 Cell differentiation and stem cells

there is an intrinsic difference between the two daughter cells because the stem-cell
division is an asymmetric division that results in the two cells acquiring a different
­complement of proteins. A clear example of this mechanism is neuroblast develop-
ment in Drosophila, which is considered in Chapter 12. The second possibility is that
external signals make the daughter cells different: a daughter cell that remains in
the stem-cell niche continues to renew itself, whereas one that ends up outside the
niche differentiates. In the case of hematopoiesis, both mechanisms may operate. On
the one hand, four proteins have been found to segregate asymmetrically at cell divi-
sion when human hematopoietic cells are cultured without the bone marrow stroma
that provides the stem-cell niche in vivo. On the other hand, factors produced by bone
marrow stromal cells that control stem-cell self-renewal have also been identified.

10.5 Colony-stimulating factors and intrinsic changes control differentiation

of the hematopoietic lineages

Hematopoiesis can be characterized by the activities of a hierarchy of transcription fac-

tors, whose overlapping patterns of expression specify the various cell lineages. Some
are only expressed in immature cells and are not lineage specific—the ­proto-oncogene
product c-Myb falls into this class. Somewhat suprisingly, the hematopoietic stem
cells express low levels of some of the genes that will be expressed later in the myeloid
and erythroid lineages. Thus, gene repression, as well as activation, must occur when
a cell becomes committed to a particular lineage. Some of the transcription-factor
interactions that occur in hematopoietic stem-cell differentiation into the erythroid
and myeloid lineages are illustrated in Fig. 10.8, and include activation, inhibition,
and more complex interactions.
In blood-cell precursors, the transcription factor GATA1, for example, is necessary
for the differentiation of the eythroid lineage, whereas another key transcription fac-
tor, PU.1, drives differentiation of the myeloid lineage. GATA1 and PU.1 bind to each
other, thus inhibiting each other’s ability to regulate the transcription of lineage-
­specific genes. The relative concentration of the two proteins determines cell fate.
When the concentration of GATA1 is low, PU.1 will activate the myeloid program, and

C-JUN

Fig. 10.8 Some transcription

factors of importance in blood cell
differentiation. Some of the gene-regulatory FOG1 GATA1 PU.1 CEBPA
interactions involved in the differentiation of
the erythroid and myeloid lineages of blood
cells. Macrophages, eosinophils, basophils,
monocytes, and neutrophils belong to the
myeloid lineage; the erythrocyte belongs
to the erythroid lineage. Arrows indicate
that the transcription factor stimulates
differentiation of that lineage, whereas erythrocyte macrophage eosinophil basophil monocyte neutrophil
barred lines indicate inhibition. For example,
the differentiation of neutrophils requires CEBPB
activation of the transcription factors
CEBPE, PU.1, and CEBPA, and inhibition of
CEBPE CUTL1
CUTL1. c-Jun and the CEPBs are ubiquitous
transcription factors, whereas PU.1 is more
restricted to white-blood-cell differentiation.
 Models of cell differentiation 379

Some hematopoietic growth factors and their target cells

Type of growth factor Responding hematopoietic cells

Erythropoietin (EPO) Erythroid progenitors

Granulocyte colony-stimulating factor (G-CSF) Granulocytes, neutrophils
Granulocyte-macrophage colony-stimulating factor (GM-CSF) Granulocytes, macrophages
Interleukin-3 Multipotent precursor cells
Stem cell factor (SCF) Stem cells
Macrophage colony-stimulating factor (M-CSF) Macrophages, granulocytes Fig. 10.9 Hematopoietic factors and their
target cells.

when the concentration of PU.1 is low, GATA1 will activate the erythroid program. As
with all cell types, it is the combination of transcription factors, and not any particular
one, that is responsible for the specific pattern of gene expression that leads to dif-
ferentiation into a given cell type.
How then is the activity of all these factors controlled? Signals delivered by extra-
cellular protein growth factors and differentiation factors have a key role. Studies on
blood-cell differentiation in culture have identified at least 20 extracellular proteins,
known generally as colony-stimulating factors or hematopoietic growth factors, that
can affect cell proliferation and cell differentiation at various points in hematopoiesis
(Fig. 10.9). Both stimulators and inhibitors have been identified. Not all the factors
controlling blood-cell production are made by blood cells or stromal cells. The pro-
tein erythropoietin, for example, which induces the differentiation of committed red-
stem cells
blood-cell precursors, is mainly produced by the kidney in response to physiological
signals that indicate red-blood-cell depletion.
Despite the well-established role of these factors in blood-cell proliferation and dif- GM-CSF
IL-3
ferentiation, there is evidence that the commitment of a cell to one or other pathway
in the myeloid lineage might be a chance event, with the growth factors simply pro- granulocyte–macrophage precursor cells
moting the survival of particular cell lineages. Evidence for this comes from experi-
ments where the two daughters of a single early progenitor cell (which can give rise
to a range of different blood-cell types) are cultured separately but under the same
G-CSF M-CSF
conditions. Both daughter cells usually give rise to the same combination of cell
GM-CSF GM-CSF
types, but in about 20% of cases they give rise to dissimilar combinations. IL-3 IL-3
Among the plethora of growth factors, it is difficult to distinguish those that might
be exerting specific effects on differentiation from those that may be required for the
survival and proliferation of a lineage or group of lineages. However, the contrasting
functions of three growth factors—granulocyte-macrophage colony-stimulating factor
(GM-CSF), macrophage colony-stimulating factor (M-CSF), and granulocyte colony-
stimulating factor (G-CSF)—are well established. GM-CSF is generally required for the
development of most myeloid cells from the earliest progenitors that can be identified.
In combination with G-CSF, however, it tends to stimulate only granulocyte (princi-
neutrophils monocytes/macrophages
pally neutrophil) formation from the common granulocyte-macrophage progenitor.
This is in contrast to M-CSF, which, in combination with GM-CSF, tends to stimulate
Fig. 10.10 Colony-stimulating factors can
differentiation of monocytes (macrophages) from the same progenitors (Fig. 10.10).
direct the differentiation of neutrophils
These growth and differentiation factors have no strict specificity of activity, in the
and macrophages. Neutrophils and
sense of each factor exerting a specific effect on one type of target cell only; rather, macrophages are derived from a common
they act in different combinations on target cells to produce different results, the out- granulocyte-macrophage precursor cell.
come also depending on the developmental history of their target. The choice of differentiation pathway can
be determined by for example, the growth
10.6 Developmentally regulated globin gene expression is controlled by factors G-CSF and M-CSF. The growth
factors GM-CSF and IL-3 are also required,
regulatory sequences far distant from the coding regions
in combination, to promote the survival and
We now focus on one type of blood cell, the red blood cell or erythrocyte, to con- proliferation of cells of the myeloid lineage
sider the transcription factors that are active in these cells during their differentia- generally.
tion, and that control expression of their cell-specific genes. The main feature of After Metcalf, D.: 1991.
380 Chapter 10 Cell differentiation and stem cells

red-blood-cell differentiation is the synthesis of large amounts of the oxygen-carrying

protein hemoglobin, which involves the coordinated regulation of two different sets
of globin genes. All the hemoglobin contained in a fully differentiated red blood cell
is produced before its terminal differentiation. Terminally differentiated mammalian
red blood cells lose their nucleus, whereas those of birds retain the nucleus but it
becomes transcriptionally inactive. The program of erythropoiesis is primarily driven
by the DNA-binding transcription factors GATA2 and GATA1, which bind to sites in
the globin genes and recruit other gene-regulatory proteins.
Vertebrate hemoglobin is a tetramer of two identical α-type and two identical β-type
globin chains. The α-globin and β-globin genes belong to different multigene families.
Each family consists of a cluster of genes and the two families are located on dif-
ferent chromosomes. In mammals, different members of each family are expressed
at various stages of development so that distinct hemoglobins are produced during
embryonic, fetal, and adult life. Hemoglobin switching is a mammalian adaptation
to differing requirements for oxygen transport at different stages of life; human fetal
hemoglobin, for example, has a higher affinity for oxygen than adult hemoglobin, and
so is able to efficiently take up the oxygen that has been released by maternal hemo-
globin in the placenta. In turn, the oxygen affinity of adult hemoglobin is sufficient for
it to pick up oxygen in the relatively oxygen-rich environment of the lungs. Here, we
look at the control of expression of the β-globin genes, as examples of genes that are
not only uniquely cell specific but are also developmentally regulated.
The human β-globin gene cluster contains five genes—ε, γG, γA, δ, and β (Fig. 10.11).
These are expressed at different times during development: is expressed in the early
embryo in the embryonic yolk sac; the two γ genes, which encode proteins differing
by only one amino acid, are expressed in the fetal liver; and the δ and β genes are
expressed in erythroid precursors in adult bone marrow. The protein products of all
these genes combine with globins encoded by the α-globin complex to form physi-
ologically different hemoglobins at each of these three stages of development.
Mutations in the globin genes are the cause of several relatively common inherited
blood diseases. One of these, sickle-cell anemia, is caused by a point mutation in
the gene coding for β-globin. In people homozygous for the mutation, the abnormal
hemoglobin molecules aggregate into fibers, forcing the cells into the characteristic
sickle shape. These cells cannot pass easily through fine blood vessels and tend to
block them, causing many of the symptoms of the disease. They also have a much
shorter lifetime than normal blood cells. Together these effects cause anemia—a lack

�-globin genes
regulatory
region � � �
Fig. 10.11 The human globin
Chromosome 16
genes. Human hemoglobin is made up of
two identical α-type globin subunits and two �-LCR
identical β-type globin subunits, which are
Hemoglobin types
encoded by α-family and β-family genes on
chromosomes 16 and 11, respectively. The embryo fetus adult
composition of human hemoglobin changes �2�2 �2�2 �2�2
during development. In the embryo it is made
up of ζ (α-type) and ε (β-type) subunits; in �-LCR
� γG γA � �
the fetal liver, of α and γ subunits; and in the
adult most of the hemoglobin is made up of Chromosome 11
α and β subunits, but a small proportion is regulatory
made up of α and δ subunits. The regulatory regions
regions α-LCR and β-LCR are locus control 80 kb
regions involved in switching on the �-globin genes
hemoglobin genes at different stages.
 Models of cell differentiation 381

�-LCR
� �G �A � �

Chromosome 11

Fig. 10.12 The control regions of the

β-globin gene. The β-globin gene is part of a
complex of other β-type globin genes and is
5’ control 3’ control only expressed in the adult. Binding sites for
region region the relatively erythroid-specific transcription
exon intron factor GATA1, as well as for other non-tissue-
specific transcription factors, such as NF1
and CP1, are located in the control regions.
Upstream of the whole β-globin cluster, in
the locus control region (LCR), are additional
binding NF1 CP1 PU.1 TATA control regions required for high-level
sites GATA1 GATA1 CP1 GATA1 GATA1 GATA1 GATA1 expression and full developmental regulation
of the β-globin genes.

of functional red blood cells. Sickle-cell anemia is one of the few genetic diseases
where the link between a mutation and the subsequent developmental effects on
health is fully understood.
The control regions that regulate expression of the β-globin gene cluster are com-
plex and extensive. Each gene has a promoter and control sites immediately upstream
(to the 5′ side) of the transcription starting point, and there is also an enhancer
downstream (to the 3′ side) of the β-globin gene, which is the last gene in the clus-
ter (Fig. 10.12). But these local control sequences, which contain binding sites for
transcription factors specific for erythroid cells, as well as for other more widespread
transcriptional activators, are not sufficient to provide properly regulated expression
of the β-globin genes.
The successive switching of different globin genes on and off during development
is an intriguing feature of globin gene expression. This developmentally regulated
expression of the β-globin gene cluster depends on a region a considerable distance
upstream from the ε gene. This is the locus control region (LCR; see Fig. 10.12),
which stretches over some 10,000 base pairs at around 5000 base pairs from the
5′ end of the ε gene. The LCR confers high levels of expression on any β-family gene
linked to it, and is also involved in directing the developmentally correct sequence of
expression of the whole β-globin gene cluster, even though the β-globin gene itself,
for example, is around 50,000 base pairs away from the extreme 5′ end of the LCR.
A similar control region has been found upstream of the α-globin gene cluster.
An attractive model for the control of globin-gene switching envisages an interac-
tion of LCR-bound proteins with proteins bound to the promoters of successive globin
genes (Fig. 10.13). The DNA between the LCR region and the globin genes is thought

Embryonic (yolk sac) Fetal liver Adult bone marrow

Fig. 10.13 A possible mechanism for the

� ON � � successive activation of β-family globin
genes by the LCR during development.
LCR

LCR

LCR

� � ON � The LCR is thought to make contact with

the promoter of each gene in succession,
� � � ON
at different stages of development, thus
controlling their temporal expression.
DNA
After Crossley, M., et al.: 1993.
382 Chapter 10 Cell differentiation and stem cells

to loop in such a way that proteins binding to the LCR can physically interact with
proteins bound to the globin gene promoters. Thus, in erythrocyte precursors in the
embryonic yolk sac the LCR would interact with the ε promoter, in fetal liver it would
interact with the two γ promoters, and in the adult bone marrow with the β-gene
promoter. There is some evidence that, for the β-globin locus at least, the LCR acts by
promoting the transition from initiation of transcription to transcriptional elongation
by RNA polymerase.

10.7 The epithelia of adult mammalian skin and gut are continually replaced
by derivatives of stem cells

Mammalian skin is composed of two cellular layers—the dermis, which mainly con-
tains fibroblasts, and the protective outer epidermis, which contains mainly keratin-
filled cells called keratinocytes. A basal lamina or basement membrane composed
of extracellular matrix separates the epithelial epidermis from the dermis. The epider-
mis, which we consider here, comprises a multilayered epithelium of keratinocytes
(Fig. 10.14) and associated hair follicles and sebaceous and sweat glands. The epi-
dermis between the hairs is known as the interfollicular epidermis. Because of the
protective function of the epidermis, cells are continuously lost from its outer surface
and must be replaced—every four weeks we have a brand new epidermis—and it is
maintained throughout life by stem cells that reside in the basal layer of the epidermis
in contact with the basement membrane, and at the base of the follicles.
Basal layer cells depend on stimuli received from fibroblasts in the dermis and the
extracellular molecules of the basement membrane to remain proliferative. Once a cell
leaves this stem-cell niche, it becomes committed to differentiate. Several hypotheses
have been proposed to explain how asymmetrical divisions of basal cells could lead
to one daughter remaining in the basal layer and the other becoming committed to
differentiation (Fig. 10.15). After becoming committed, the cells differentiate further
after leaving the basal layer. One terminal differentiation pathway culminates in the
production of keratinocytes, which form the outermost, cornified epidermal layers that
provide the skin’s protective covering. In the interfollicular epidermis, differentiating

interfollicular
epidermis epidermis
basement
membrane

sebaceous
gland

dermis
hair follicle

Fig. 10.14 Section through human skin sweat follicle bulge

gland region
showing epidermal structures. The skin is
composed of an outer epidermis (pale orange) subcutaneous
separated from the underlying dermis by a fat
basement membrane. Structures that are
epidermal in origin are shown in color here;
dermal structures are in gray.
 Models of cell differentiation 383

outer layer Fig. 10.15 Differentiation of keratinocytes

of dead cells
in human epidermis. Descendants of
basal-layer cells, which will become the
keratinocytes of the epidermis, detach from
the basement membrane, divide several
dividing and
differentiating high Notch activity inhibits times, and leave the basal layer before
keratinocytes integrin gene expression beginning to differentiate. Basal-layer cells
express the intermediate filament proteins
keratin 14 and keratin 5. Differentiation
spinous cells into keratinocytes involves a change in the
expressing keratin genes expressed and the production
keratins 1 and 10 of large amounts of keratin 1 and keratin 10.
In the intermediate layers, the cells are still
Asymmetric division promotes the switch large and metabolically active, whereas in
basal-layer from basal-layer cell to spinous cell the outer layers of the epidermis, the cells
cells expressing lose their nuclei, become filled with keratin
keratins 5 and 14
EGFR activity inhibits Notch filaments, and their membranes become
basement membrane insoluble owing to deposition of the protein
(laminin, collagen) involucrin. The dead cells are eventually shed
from the skin surface.

keratinocytes move through the skin as they mature, until by the time they reach the
outermost layer they are completely filled with the fibrous protein keratin and their
membranes are strengthened with the protein involucrin. The dead cells eventually
flake off the surface (see Fig. 10.15). The terminally differentiated cells of the seba-
ceous glands, by contrast, are lipid-filled sebocytes that eventually burst, releasing
their contents onto the surface of the skin. Differentiation of a hair follicle is even more
complex, comprising eight different cell lineages.
In living epidermal cells, keratins are an essential cytoskeletal component, forming
the intermediate filaments through which epidermal cells are attached to each other,
and giving skin its resilience and elasticity. Basal layer cells express keratin types 5
and 14 and mutations in the genes encoding these keratins are found in most patients
with epidermolysis bullosa, a disease in which the epidermis is very fragile and can
easily blister. Spinous cells expressing keratins 1 and 10 are located in the suprabasal
layer of the epidermis and the switch from expression of keratin types 5 and 14 to
expression of keratins 1 and 10 is a reliable indicator that cells have become commit-
ted to terminal differentiation. Several signals are now known to regulate this switch.
Activation of Notch signaling promotes differentiation and is inhibited in basal-layer
cells by various mechanisms, including signaling through the epidermal growth fac-
tor receptor (EGFR), which, as a consequence, contributes to maintaining basal cell
fate (see Fig. 10.15). Cell-adhesion molecules also play a key role. The basal cell
layer is attached to the basement membrane by specialized cell junctions that include
integrin-containing hemidesmosomes and focal contacts (see Box 8B, p. 293). Basal
cells express the integrin α6β4, which is required for hemidesmosome formation and
interacts with collagen and laminin in the extracellular matrix, and also express high
levels of integrin α3β1, which is a receptor for laminin. Mutations in the collagen
associated with epidermal hemidesmosomes have been found in a subset of patients
with epidermolysis bullosa blistering disease. Detachment of a cell from the basement
membrane is accompanied by decreased amounts of integrins on the cell surface as a
result of activation of Notch signaling.
One of the unresolved questions in epidermal cell biology is the location of the
stem cells in the basal layer and whether all the cells within the basal layer are
stem cells. Basal epidermal cells are heterogeneous, with some cells proliferating
384 Chapter 10 Cell differentiation and stem cells

r­ apidly and others dividing far less frequently. It has been suggested that the slowly
­cycling cells are stem cells and that the surrounding proliferating basal cells, known
as transit-­amplifying cells, are committed to terminal differentiation. Other investi-
gations tracking the fate of individual marked basal epidermal cells suggest, however,
that all basal cells could be equivalent in their developmental potential. In the hair
follicle, multipotent stem cells have been located in a region known as the bulge.
­Lineage-tracing experiments in transgenic mice in which, for example, green fluores-
cent protein (GFP) is expressed under the control of the keratin 5 promoter, showed
that the cells that give rise to new hair follicles originate in the bulge, and under
certain conditions, bulge cells can also contribute to sebaceous gland and epidermis.
epithelial shedding villi
cells cells The descendants of single follicle bulge cells from mouse epidermis cultured in the
laboratory can give rise to hair follicles, sebaceous glands, and epidermis when com-
bined with dermal fibroblasts and used as skin grafts.
The epithelial cells lining the gut are also continuously replaced from stem cells. In
the small intestine the epithelial cells form a single-layered epithelium that is folded
to form villi that project into the gut lumen and crypts that penetrate the underlying
connective tissue. Cells are continuously shed from the tips of the villi, while the
stem cells that produce their replacements lie in the base of the crypts. The new cells
generated by the stem cells move upward toward the tips of the villi. En route, they
multiply in the lower half of the crypt and constitute a transit-amplifying population
crypt (Fig. 10.16). The turnover of cells in the mammalian small intestine is about 4 days.
A single crypt in the small intestine of the mouse contains about 250 cells compris-
ing four main differentiated cell types. Paneth cells, goblet cells and enteroendrocrine
cells shedding from villus cells are secretory cells, while the enterocyte, the predominant cell type in the epi-
thelium of the small intestine, absorbs nutrients from the gut contents. About 150 of
the crypt cells are proliferative, dividing about twice a day so that the crypt produces
300 new cells each day; about 12 cells move from the crypt into the villus each hour.
Canonical Wnt signaling leading to activation of the transcription factor TCF by
loose its co-activator β-catenin (see Box 1E, p. 26) is essential for stem-cell proliferation,
connective proliferation of the transit-amplifying population, and for the terminal differentiation
tissue
of Paneth cells. Two types of stem cell have been identified in the mouse small intes-
tine in the stem-cell niche at the base of each crypt. One type, originally identified
cross-section
of villus by its slow rate of division, is located a short distance above the bottom of the crypt,
at the edge of the stem-cell niche. The other stem cells, which are considerably more
proliferative, are slender columnar cells—between four to six cells per crypt—that are
Direction of interspersed with Paneth cells at the bottom of the crypt (see Fig. 10.16). Both sets
cell movement
of stem cells have distinctive patterns of gene expression. The slowly dividing cells
express high levels of Bmi1, a component of the Polycomb complex, and long-term
goblet lineage tracing of cells expressing Bmi1 in transgenic mice using LacZ as a marker
enterocyte cell resulted in the labeling of entire crypts and villi after 12 months. The other popula-
tion of stem cells expresses a Wnt target, the gene encoding Lgr5, a G-protein-coupled
receptor whose ligand is still unknown. Experiments in which cells expressing Lgr5

nondividing Fig. 10.16 Epithelial cells lining the mammalian gut are continually replaced. The upper
differentiated cells panel shows the villi—epithelial extensions covering the wall of the small intestine. The lower
panel is a detail of a crypt and villus. The stem cells from which the intestinal epithelium is
renewed are found in the base of the crypts. One type of stem cell (yellow) is characterized by a
transit-amplifying cells
slow rate of division, and is found at some distance from the bottom of the crypt, while a second
type of stem cell (red) is interspersed between the Paneth cells in the bases of the crypts.
These stem cells divide more frequently than the first type and are the ones mostly involved
slowly dividing stem cell
frequently dividing stem cell in routinely renewing the epithelium. The stem cells give rise to cells that proliferate and move
upward. By the time they have left the crypt they have differentiated into epithelial cell types.
Paneth cell They continue to move upwards on the projecting villus, and are finally shed from the tip. The
total time of transit is about 4 days.
 Models of cell differentiation 385

a b c
Fig. 10.17 Cell lineage tracing of
Lgr5-expressing cells in the mouse
small intestine. Transgenic mice which
expressed LacZ from the Lgr5 promoter
were used to trace the origin and movement
of Lgr-expressing cells. The photos show
histological analysis of LacZ activity (blue-
staining cells) in small intestine: (a) 1 day
after induction; (b) 5 days after induction;
and (c) 60 days after induction.
From Barker, N., et al.: 2007.

and their progeny were marked irreversibly by expression of LacZ showed that these
columnar cells could give rise to entire crypts and to ribbons of cells on adjacent villi
after 2 months (Fig. 10.17). Thus, like the slowly cycling cells, the Lgr5-expressing
columnar cells are indeed multipotent stem cells. It is not yet clear whether these
two stem-cell populations serve different functions: one could perhaps be involved
in normal tissue homeostasis, while the other provides a reserve of stem cells for
regeneration.
The multipotency of the Lgr5-expressing stem cells was confirmed by isolating
them and growing them in culture. When cultured under the right conditions, a single
Lgr5-expressing cell can give rise to organized intestinal villus-crypt tissue containing
all four differentiated cell types. Another epithelial structure, the mammary gland, has
also been shown to be generated from stem cells. A complete mouse mammary gland,
capable of producing milk, has been grown from a single stem cell.
Specification as either absorptive or secretory cells in the crypts of the small intes-
tine also depends on Notch signaling. Under conditions in which Notch signaling is
reduced, proliferative crypt cells are completely converted to goblet cells, whereas
when Notch signaling is activated, there is increased proliferation of stem cells and
a severe reduction in differentiated secretory cells (goblet cells, enteroendocrine and
Paneth cells). Details of some of the transcription factors that operate in each of the
four cell lineages are beginning to be revealed.

10.8 The MyoD family of genes determines differentiation into muscle

In Chapter 4 we considered the signals involved in specifying vertebrate muscle cells

in the somite. Here we focus on the differentiation of cells that are already specified to
become skeletal muscle; differentiation of cardiac muscle involves different transcrip-
tion factors. Differentiation of vertebrate skeletal striated muscle can be studied in cell
culture and provides a valuable model system for cell differentiation. Skeletal muscle
cells derive from the myotome of the somites. Myoblasts, cells that are committed to
forming muscle, can be isolated from chick or mouse embryos and mouse cells can be
cloned to produce cell cultures derived from a single myoblast. These cells will prolif-
erate until growth factors are removed from the culture medium, whereupon prolifera-
tion ceases and overt differentiation into muscle cells begins. The cells then begin to
synthesize muscle-specific proteins such as actin, myosin II, and tropomyosin, which
are part of the contractile apparatus, and muscle-specific enzymes, such as creatine
phosphate kinase. Myoblasts also undergo structural change during differentiation.
They first become bipolar in shape as a result of reorganization of the microtubule
cytoskeleton, and then fuse to form multinucleate myotubes (Fig. 10.18). Fusion in
vivo requires the presence of integrin β1 on the cell membrane. Within about 20 hours
of withdrawal of growth factors, typical striated muscle fibers can be seen.
The gene myoD is a member of a family of basic helix-loop-helix transcription-
­factor genes that are expressed only in muscle precursors and muscle cells. Genes
386 Chapter 10 Cell differentiation and stem cells

Dividing myoblasts Cell alignment Cell fusion Muscle fiber

Fig. 10.18 Differentiation of striated
growth factors
muscle in culture. Myoblasts are cells that
are committed to becoming muscle but have
yet to show signs of differentiation. In the
presence of growth factors they continue
to multiply but do not differentiate. When
growth factors are removed the myoblasts
stop dividing; they align and fuse into
multinucleate myotubes, which develop into Appearance of Spontaneous
Cell multiplication Cell multiplication ceases muscle-specific proteins contractions begin
myofibers that contract spontaneously.

in this family can be considered as key controlling genes for muscle differentiation.
­Activation of these genes leads to the switching on of muscle-specific genes and dif-
ferentiation. myoD will even induce differentiation into muscle if transfected into
fibroblasts, which do not normally express either the MyoD protein or the muscle-
specific structural proteins and enzymes. As well as myoD, the family includes three
other genes—mrf4, myf5, and myogenin—which can all induce muscle differentia-
tion in fibroblasts and other non-muscle cells. mrf4, myf5, and myoD are the first
genes to be switched on in muscle precursors in mammals and they act as muscle-
­determination genes (Fig. 10.19). They are also activated by the transcription fac-
tor Pax3, which becomes restricted to muscle-cell precursors (see ­Section 5.9). Once
turned on, these factors maintain their own expression by means of positive-feedback
loops. Mrf4, Myf5, and MyoD in turn activate the myogenin gene, which is involved in
muscle differentiation and maturation. myoD and myf5 are expressed in proliferating,
undifferentiated myogenic cells, whereas myogenin is only expressed in non-dividing
differentiating cells; mrf4 also has a second role in myofiber maturation.
Despite the demonstrated muscle-determining effects of MyoD and Myf5, gene knock-
out experiments in transgenic mice (see Section 3.9 and Box 3B, p. 118) show that mice
lacking MyoD make apparently normal skeletal muscle if myf5 is active, suggesting
redundancy in the system. Apparent redundancy is not uncommon in development,
and has probably evolved to serve as a back-up mechanism to ensure the continuation
of development in the face of fluctuations in gene expression (see ­Section 1.19). In
mice lacking myoD, the level of myf5 expression is raised, implying that the expression
of myoD normally inhibits myf5 expression, and that Myf5 protein can compensate
for the absence of MyoD. Mice lacking Myf5 also make skeletal muscle, although they
have other defects; the most obvious abnormality is a shortening of the ribs. Mice in
which both Myf5 and MyoD are lacking, however, die before birth, although if such
embryos still express mrf4, they make some muscle. Such experiments show that these
proteins are acting as myogenic determination factors. Indeed, in ­embryos in which

Fig. 10.19 Key features of the mesodermal myoblasts multinucleate muscle

progenitor myotube fiber
differentiation of vertebrate skeletal
muscle. External signals initiate muscle
differentiation by activating the genes mrf4,
myoD and myf5. One of these genes tends
to be expressed preferentially (depending
on the species) and their activity is pax3
mutually inhibitory and self-sustaining. The Determination myoD Differentiation Maturation
muscle-
transcription factors encoded by mrf4, myf5, external mrf4
myogenin specific
signals mrf4 proteins
and myoD activate muscle-differentiation myf5
genes such as myogenin, which in turn mrf4
activate expression of muscle-specific inhibition pax3
growth
proteins, such as the proteins that make up activation factors
the contractile apparatus in muscle cells.
 Models of cell differentiation 387

mutant myf5 and myoD genes have been activated but which do not make functional
Myf5 or MyoD protein, prospective muscle cells mislocate, and may be integrated
into other differentiation pathways, such as cartilage and bone. Mice in which the
myogenin gene has been knocked out, on the other hand, lack most skeletal muscle
but precursor myoblasts are present. Myogenin is thus considered to be a myogenic
differentiation factor rather than a protein essential for muscle determination.
The MyoD family of transcription factors activates transcription of muscle-specific
genes by binding to a nucleotide sequence called the E-box, which is present in the
control regions or promoter regions of these genes. MyoD forms a heterodimer with
products of the E2.2 gene such as E12, and this binds to the E-box. As the activity of
MyoD and its relatives is essential for muscle-cell differentiation, we must now look
at how the activity of MyoD could be regulated.

10.9 The differentiation of muscle cells involves withdrawal from the cell
cycle, but is reversible

The proliferation and differentiation of myoblasts are mutually exclusive activities.

Skeletal myoblasts that are multiplying in culture do not differentiate. Only when pro-
liferation ceases does differentiation begin. In the presence of protein growth factors,
myoblasts that are expressing both MyoD and Myf5 proteins continue to proliferate
and do not differentiate into muscle. This means that the presence of MyoD and Myf5 Proliferating muscle
is not in itself sufficient for muscle differentiation and that an additional signal or precursors
signals are required. In culture, this stimulus to differentiate can be supplied by the
removal of growth factors from the medium. The myoblasts then withdraw from the
cell cycle, cell fusion occurs, and differentiation takes place.
There is an intimate relationship between the proteins that control progression RNA
TFIID polymerase II
through the cell cycle and the muscle determination and differentiation factors.
MyoD and Myf5 are phosphorylated by cyclin-dependent protein kinases, which
become activated at key points in the cell cycle (see Fig. 13.2). Phosphorylation target gene
affects the activity of MyoD and Myf5 by making them more likely to be degraded.
Thus, in actively proliferating cells, the levels of the myogenic factors will be kept low.
Differentiating
A number of other proteins also interact with myogenic factors or with their active myotubes
partners such as E12, and inhibit their activity. One interacting protein is the transcrip-
tion factor Id, which is present at high concentrations in proliferating cells, and which
probably helps to prevent premature activation of muscle-specific genes.
The myogenic factors themselves actively interfere with the cell cycle to cause cell- RNA
cycle arrest. Myogenin and MyoD activate transcription of members of the p21 family TRF3 TAF3 polymerase II
of proteins, which block cell-cycle progression, causing the cell to withdraw from the
cell cycle and to differentiate. Another key protein in preventing progression through
the cell cycle is the retinoblastoma protein RB. RB is phosphorylated in proliferating myogenin
cells by the cyclin-dependent kinase 4 (Cdk4) and dephosphorylation of RB is essen-
tial both for withdrawal from the cell cycle and for muscle differentiation. MyoD can Fig. 10.20 The cessation of myoblast
interact strongly and specifically with Cdk4 to inhibit RB phosphorylation, which proliferation and the onset of muscle cell
would also help to lead to cell-cycle arrest. differentiation is accompanied by a change
A more general mechanism for shutting down genes involved in cell proliferation in the basal transcriptional complex.
has also been found to operate in muscle cells. Rather unexpectedly, this involves In myoblasts, as in most other cells studied,
the basal transcription complex contains
changing components of the core transcriptional machinery. One of the general tran-
the general transcription factor TFIID,
scription factors in the basic promoter–recognition complex described in Fig. 10.3 is
which recognizes the TATA box in gene
TFIID, which is a multisubunit protein that recognizes the TATA box. TFIID is used in
promoters. In differentiating myotubes, TFIID
the promoter-recognition complex that controls the expression of genes involved in is dismantled and is thought to be replaced
progression through the cell cycle and cell division. When the muscle cells undergo by a complex containing the proteins TRF3
differentiation, however, TFIID is dismantled and is replaced by a different protein and TAF3. Along with specific activators,
complex, which controls the expression of differentiation genes such as myogenin, this complex recognizes the myogenin gene
among others (Fig. 10.20). This switch selectively turns on one transcriptional ­program promoter and initiates transcription.
388 Chapter 10 Cell differentiation and stem cells

in the cell while effectively shutting down another, and could be a general mechanism
for shutting down cell proliferation and initiating differentiation.
The opposition between cell proliferation and terminal differentiation makes devel-
opmental sense. In those tissues where fully differentiated cells do not divide further,
it is essential that sufficient cells are produced to make a functional structure, such
as a muscle, before differentiation begins. The reliance on external signals to induce
differentiation, or to permit a cell already poised for differentiation to start differ-
entiating, is a way of ensuring that differentiation only occurs in the appropriate
­conditions.
Despite the apparently terminal nature of muscle-cell differentiation, muscle cells
provide an example of a differentiated cell that can undergo dedifferentiation—the
loss of differentiated characteristics—re-enter the cell cycle and even give rise to other
cell types when manipulated in culture—a phenomeon called transdifferentiation.
The gene Msx1 encodes a homeodomain-containing transcriptional repressor, which
is expressed in determined but undifferentiated myoblasts during limb development
and in other undifferentiated cells at the tip of the limb bud. When mouse muscle
myotubes in culture were transfected with Msx1, 5–10% of the myotubes dediffer-
entiated, cleaving to form both smaller multinucleate myotubes and, more signifi-
cantly, mononuclear cells that can undergo cell division. When dividing mononuclear
cells were cultured in the appropriate conditions, some progeny of a single myotube
expressed markers for other cell types, such as cartilage or fat cells. It was previously
thought that a similar transdifferentiation of muscle cells to produce other cell types
might occur in the regeneration of amphibian limbs, but this has been shown not to
be the case in the axolotl (see Chapter 14).

Before birth
10.10 Skeletal muscle and neural cells can be renewed from stem cells
in adults
Dorsal
neural tube dermomyotome Once formed, skeletal muscle cells can enlarge by cell growth but do not divide. They
can, however, be replaced. In adult mammalian muscle there are cells called satellite
cells that can proliferate and differentiate into new muscle cells if the muscle is dam-
aged. When satellite cells are activated, for example, by injury to the muscle, they
proliferate and develop into myoblasts.
somite Satellite cells are the muscle stem cells. They lie between the basement membrane
and the plasma membrane of mature muscle fibers and can be recognized by their
limb precursor expression of characteristic marker proteins, such as the cell-surface protein CD34
muscle cells
Ventral expressing Pax3 and the transcription factor Pax7 (Fig. 10.21). Under favorable conditions, the satellite
cells associated with one myofiber (about 7 satellite cells per myofiber), grafted into
an irradiated muscle in a mouse, can generate more than 100 new myofibers and will
After birth
also undergo self-renewal. The irradiation kills the mouse’s own satellite cells, thus
limb skeletal muscle satellite cell
making it impossible that the muscle could have regenerated on its own. Even a single
expressing Pax7 muscle stem cell transplanted into irradiated mouse muscle can produce both muscle
basement
fibers and more satellite cells.
membrane The function of Pax7 seems to overlap with that of the closely related Pax3. As dis-
cussed earlier, Pax3 is essential for early determination of muscle-cell precursors. In
nucleus
contrast, Pax7 is required for the maintenance of the satellite-cell population at the time
of birth and is expressed in adult muscle. If, however, Pax7 is deleted in mice 2–3 weeks
after birth, limb skeletal muscle growth and regeneration is, surprisingly, unaffected and
Fig. 10.21 Muscle satellite cells. During
the satellite cells still make new muscle even though they no longer express the gene.
embryonic development, Pax3-expressing
muscle cell precursors leave the
We will see in Chapter 12 how neurons and other nervous system cells develop
dermomyotome and migrate into the limbs. from embryonic neural stem cells. Once fully differentiated, however, neurons in the
In the limb muscle after birth, muscle stem mammalian brain do not divide, and for many years it was thought that no new brain
cells remain associated with the muscle as neurons at all were produced in the adult mammal. Relatively recently, however, the
satellite cells but now express Pax7. production of new neurons, or neurogenesis, has been demonstrated as a normal
 Models of cell differentiation 389

­ ccurrence in the adult mammalian brain, and neural stem cells have been identi-
o
fied in adult mammals that can generate neurons, astrocytes, and oligodendrocytes olfactory lateral
bulb ventricle subgranular
in culture. Neurogenesis in adult mammals occurs in just two regions: the subven- zone
tricular zone of the lateral ventricle, and the subgranular zone of the dentate gyrus
in the hippocampus. New neurons generated in the subventricular zone migrate to dentate gyrus
give rise to interneurons in the olfactory bulb, whereas the stem cells in the hip-
subventricular
pocampus give rise to new neurons of the dentate granule layer of the hippocampus. zone
Slowly proliferating radial glial cells that express astrocyte-specific genes act as the
adult neural stem cells. In the subventricular zone of the lateral ventricle, these stem
cells give rise to proliferating transit-amplifying cells before maturing into neural pro-
genitors (Fig. 10.22), whereas in the hippocampus, radial glial cells can give rise to
neurons directly. Other adjacent glial cells in these regions provide a stem-cell niche, lateral
ventricle
producing signals that control neural stem-cell behavior. Wnt signaling through the
β-catenin pathway in neural stem cells has been shown to play an important role in neuronal
regulating adult neurogenesis. Notch signaling determines the number of stem cells progenitor
and regulates Sonic hedgehog signaling, which promotes their survival. Other sources
of signals in the adult brain include the ependymal cells lining the ventricle, the blood transit-amplifying
precursors
vessels, and blood components, and some of the signaling molecules involved have stem cell/
been identified. astrocyte
ependymal
cell

10.11 Embryonic neural crest cells differentiate into a wide range of different
Fig. 10.22 Sites of neurogenesis in the
cell types
adult mammalian brain. Neurogenesis
We now turn to the embryonic neural crest, which gives rise to a remarkable number occurs mainly in the subgranular layer of
of different cell types. As described in Chapter 5, neural crest cells arise from the dor- the dentate gyrus in the hippocampus and
sal neural tube and are recognizable as discrete cells when they emerge from the neu- the anterior part of the subventricular zone,
which is shown in more detail in this figure.
ral epithelium. In the cranial region of mammals, neural crest cells begin to emerge
Astrocyte-like neural stem cells give rise to
from the elevating neural folds before they fuse. The specification of the neural crest
rapidly dividing transit-amplifying cells, which
from uncommitted precursor cells at the margin of the neural plate is the result of differentiate into neurons. New neuronal cells
inductive interactions between the neural plate and adjacent tissues—the non-neural generated in the dentate gyrus migrate to
ectoderm and the underlying mesenchyme. The classical model was that neural crest the granular layer, while new neuronal cells
induction could be explained by the response of the ectoderm to intermediate levels generated in the subventricular zone migrate
of BMP signaling (see Chapter 5). It is now clear from work in many species that, to the olfactory bulb and differentiate there
in addition, Wnt and FGF signaling is critical for neural crest induction. Once neural into interneurons. Endothelial cells of nearby
crest cells have detached from the neural tube, they migrate to many sites, as dis- blood vessels also provide signals for stem-
cussed in Chapter 8. cell differentiation (not shown).
Here we shall discuss their potential for differentiation. Neural crest cells give rise
to the neurons and glia of the peripheral nervous system (which includes the sensory
and autonomic nervous systems), endocrine cells, such as the chromaffin cells of the
adrenal medulla, and melanocytes, which are pigmented cells found in the skin and
other tissues (Fig. 10.23). In the head of mammalian and avian embryos, the neural

uncommitted Fig. 10.23 Derivatives of the neural

neural crest cell crest. Neural crest cells give rise to a wide
Migration and
determination range of cell types that include melanocytes,
cartilage, glia, and various types of
melanocyte glia sensory sympatho- parasympathetic enteric cartilage
adrenal neurons distinguished by their functional
specializations and the neurotransmitters
non- Schwann cholinergic peptidergic they produce. Cholinergic neurons use
Terminally and
differentiated
myelinating cell neurons acetylcholine as their neurotransmitter,
glia serotonergic
derivatives neurons adrenergic neurons principally use
adrenergic adrenergic and norepinephrine (noradrenaline), and
chromaffin cholinergic
cell sympathetic peptidergic and serotonergic neurons produce
neurons peptide neurotransmitters and serotonin,
respectively.
390 Chapter 10 Cell differentiation and stem cells

Diagrammatic representation of a 28-somite chick embryo

Fig. 10.24 Fate and developmental
Cervical Thoracic Lumbosacral
potential of neural crest cells. The fate map Cephalic (head) region spinal cord spinal cord spinal cord
of the neural crest (bottom panel) shows
that there is a general correspondence
between the position of the neural crest
cells along the antero-posterior axis and
the position of the structure to which these somites
cells give rise (top panel). For example, the
crest that gives rise to the mesectoderm of Fate map of neural crest
the head comes from the anterior region.
The developmental potential of neural crest Mesectoderm
Parasympathetic ganglia Enteric ganglia
cells, with the exception of the presumptive
Sympathetic ganglia
mesectoderm, is, however, very much greater Sensory ganglia
than their presumptive fate.

crest also gives rise to mesodermal types of tissues in the form of the facial bones,
cartilage, and dermis. Neural crest with potential to form mesodermal as well as ecto-
dermal derivatives is called the mesectoderm.
The tremendous differentiation potential of the neural crest was originally dis-
covered by removing the neural folds from amphibian embryos and noting which
structures failed to develop. This was later followed by an extensive series of experi-
ments using quail neural crest transplanted into chick embryos. Quail cells contain
a distinctive nucleus that distinguishes them from chick cells (see Fig. 5.6). It is
therefore possible to track the transplanted quail cells and see which tissues they
contribute to and what cell types they differentiate into. To construct a fate map of
the neural crest, small regions of quail dorsal neural tube were grafted to equivalent
positions in chick embryos of the same age, at a stage before neural crest migra-
tion. The fate map shows a general correspondence between the position of crest
cells along the antero-posterior axis and the axial position of the cells and tissues
they will give rise to (Fig. 10.24). For example, the neural crest cells that give rise
to tissues in the facial and pharyngeal regions are derived from crest anterior to
somite 5, whereas ganglion cells of the autonomic sympathetic system come from
crest posterior to somite 5.
To what extent is the fate of neural crest cells fixed before migration? Some crest
cells are unquestionably multipotent. Single neural crest cells injected with a tracer
shortly after they have left the neural tube can be seen to give rise to a number of
different cell types, both neuronal and non-neuronal. Also, by changing the position
of neural crest before cells start to migrate, it has been shown that the developmental
potential of these cells is greater than their normal fate would suggest. Multipotent
stem cells have been isolated from neural crest that can give rise to neurons, glia, and
smooth muscle. Similar multipotent cells have even been isolated from the peripheral
nerves of mammalian embryos days after neural crest migration. Almost all of the cell
types to which the neural crest can give rise will differentiate in tissue culture, and
the developmental potential of single neural crest cells has been studied in this way.
Most of the clones that develop from neural crest cells taken at the time of migration
contain more than one cell type, showing that the cell was multipotent at the time it
was isolated. As neural crest cells migrate, however, their potential decreases progres-
sively: the size of cultured clones gets smaller and the diversity of cell types they con-
tain is less. Thus, shortly after the beginning of migration, the neural crest is a mixed
population of multipotent cells and cells whose potential is already restricted.
Numerous extracellular signaling proteins that affect the differentiation and pro-
liferation of neural crest cells have been identified from mutations in mice and
 Models of cell differentiation 391

e­ xperiments with cultured cells. For example, melanocyte survival and differentiation Skin fibroblast provides signal for
is stimulated by Wnts and by the cytokines endothelin and stem-cell factor (SCF), melanoblast to develop into a pigment cell
gliogenesis by neuregulin (also called glial growth factor), while BMPs are required
for the development of autonomic neurons. Activation of the canonical Wnt/β-catenin Steel
factor
Kit
pathway in the neural tube promotes the formation of sensory neurons by neural crest
at the expense of virtually all other derivatives, and brain-derived neurotrophic factor
(BDNF) produced by the neural tube has been implicated in the survival of sensory
neurons. A number of different transcription factors have been identified that are fibroblast melanoblast
expressed at the neural plate border and in pre-migratory and early migrating neural
crest progenitors. These include members of the SoxE family of transcription factors,
Mutant white spotting
such as Sox9, which has been shown to be important for neural crest survival in both (Kit) Mutant Steel
mouse and zebrafish embryos, and Sox10, which is expressed in migrating neural
crest cells (see Box 1D, p. 20) and then regulates the terminal differentiation of neu-
rons and melanocytes.
Melanocytes are derived from neural crest cells that follow a dorsal migration
route under the ectoderm. Mice homozygous for either the white spotting (W) or
Steel ­mutations (genes originally identified by unusual coat color) lack melanocytes
in the skin and other tissues. Although mutations in the two different genes appar-
No signal for melanoblast to develop
ently have the same gross effect, the genes code for two quite different proteins,
but which nevertheless function in the same pathway. white spotting codes for the
Fig. 10.25 The receptor Kit and its
cell-surface protein Kit, which is the receptor for SCF. It is expressed on developing
ligand the Steel factor are involved in
melanoblasts, which are the undifferentiated precursors of skin melanocytes. Steel,
melanoblast differentiation. When either
in contrast, codes for SCF, which is produced by the fibroblasts and keratinocytes the gene encoding Kit (the gene white
that surround the melanoblast. Mutations in either gene block melanocyte differ- spotting, or W) or the Steel gene is mutant,
entiation, showing that the signal generated by the interaction of SCF and Kit is melanoblasts fail to differentiate into
essential for differentiation (Fig. 10.25). There is a very rare human condition known melanocytes.
as piebaldism in which melanocytes fail to develop in very localized regions of the
body, including the scalp and forehead—giving rise to a white forelock—and the
ventral trunk and limbs. It is now known that this condition is due to mutations in
the human KIT gene.
Neural crest cells give rise to both the sensory dorsal root ganglia and the auto-
nomic ganglia of the peripheral nervous system. These ganglia contain both neurons
and glia, the neurons differentiating first. So what determines whether a neural crest
cell gives rise to a neuron or to glia? There is evidence that activation of Notch in
neural crest stem cells inhibits neuronal formation and promotes differentiation into
glia. There is also evidence that Notch ligands, such as Delta, are expressed on neu-
roblasts, the precursors of neurons. Thus, once a neuroblast has embarked on the
neurogenic differentiation pathway under the influence of neurogenic signals such
as BMP-2, it may provide a feedback signal that prevents the surrounding cells from
developing further as neuroblasts and promotes formation of glia instead. Wnt signal-
ing biases neural crest cells towards the sensory neuron lineage, whereas BMP signals
influence them to acquire properties of sympathetic neurons.
In mammal and bird embryos, the neural crest in the head, which migrates to the
branchial arches (see Fig. 5.37), appears to give rise not only to the usual ectodermal
derivatives, but also to cells that form facial bones, cartilage, and dermis. Recently,
however, the presence of non-neural cells has been shown in the neural folds imme-
diately adjacent to the neural crest in the heads of mouse embryos. A non-neural
lateral epithelial domain characterized by the expression of various proteins normally
associated with connective tissues can be distinguished from the adjacent neural crest
region. It has therefore been suggested that this lateral epithelium might be the source
of mesectodermal cells and that these are distinct from the neural crest. However,
this intriguing idea has not yet been confirmed by the demonstration that these cells
migrate to the branchial arches.
392 Chapter 10 Cell differentiation and stem cells

Hox gene expression in the cranial neural crest (see Section 5.13) is a determinant
cell of crest potential. For example, no Hox genes are expressed in the anterior region of
apoptotic programmed
signal to die
the head neural folds, which give rise to the cells that form the bones of the face,
whereas anterior Hox genes are expressed in more posterior parts of the head neural
crest. If the genes Hoxa2, Hoxa3, or Hoxb4 are ectopically expressed in the anterior
Induction
of head neural crest of chick embryos, no facial skeletal structures differentiate. And in
apoptosis transgenic mouse embryos in which the entire Hoxa gene cluster has been inactivated
altogether in the posterior cranial neural crest, these cells now give rise to anterior
structures normally produced by cells that do not express Hox genes.
apoptotic
cell
10.12 Programmed cell death is under genetic control

Engulfment by Selective cell death is a normal part of development. While not strictly cell differen-
phagocytosis tiation, it is convenient to consider it here in these terms. It is involved, for example,
and
degradation in the morphogenesis of the vertebrate limb, where interdigital cell death is essential
for separating the digits (see Chapter 11), and the development of the vertebrate ner-
vous system involves the death of large numbers of neurons. Programmed cell death
scavenger cell is particularly important in the development of the nematode (see Chapter 6): 959
somatic cells come from the egg and 131 die during development. In all these cases,
the dying cell undergoes a type of cell suicide known as apoptosis, which requires
both RNA and protein synthesis and is quite distinct from the death that occurs as a
result of pathological damage. In apoptosis, the cytosolic calcium concentration rises
and leads to the activation of an endonuclease, which fragments the chromatin. The
cell contents remain bounded by a cell membrane throughout, even though the cell
breaks up into fragments. The dying cell is finally phagocytosed by scavenger cells
(Fig. 10.26). These features distinguish apoptosis from cell death due to damage,
Fig. 10.26 Programmed cell death. During where the whole cell tends to swell and eventually burst open (lyze); this type of cell
development many cells undergo death is known as necrosis.
programmed cell death or apoptosis. Studies of apoptosis in the nematode have shown that, although many different
Apoptosis may be triggered by an external types of cells die, their death is initiated through a common mechanism centered on
signal (as shown here) or cells may be the actions of CED-3, which is a type of protease called a caspase. CED-3 triggers the
programmed to die unless they receive a
cellular changes that lead to apoptotic cell death. The nematode caspase-activation
specific survival signal.
pathway is shown in Fig. 10.27 (left panel). CED-4 in this pathway is an adaptor protein
that activates CED-3. In nematodes with mutations that inactivate either ced-3 or ced-4,
none of the 131 cells that normally die do so, but appear instead to differentiate in the
same way as their sister cells. Such worms appear to survive well and die at the usual
age of several weeks.
The nematode’s cell-death program is under the control of another protein, CED-9,
which acts as a brake on the process by preventing the activation of CED-3 by CED-4.
If ced-9 is inactivated by mutation, many cells that normally do not die will die. And if
the ced-9 gene is made abnormally active by mutation, no cell deaths occur. Apoptosis
occurs when the actions of the protein EGL-1 remove inhibition by CED-9. EGL-1 is
produced in response to pro-apoptotic signals. A similar pathway is present in Droso-
phila and mammals (Fig. 10.27, middle and right panels). The mammalian homologs
of ced-9 are the Bcl-2 family of genes (Fig. 10.27, right panel). The sequence homolo-
gies between the gene Bcl-2 and ced-9 are so strong that Bcl-2 introduced into nema-
todes will function in place of ced-9. Mammals also possess caspases homologous to
CED-3, which initiate the proteolytic cascade that leads to cell death.
Programmed cell death in animals is far from being a rare phenomenon. The
cells in all tissues are intrinsically programmed to undergo cell death, and are
only prevented from dying by positive control signals from neighboring cells. Cell
death also plays a key role both in controlling growth and preventing cancer (see
Chapter 13).
 Models of cell differentiation 393

Fig. 10.27 Apoptotic pathways

Nematodes Drosophila Mammals
in nematodes, Drosophila, and
Apoptotic stimuli Apoptotic stimuli Apoptotic stimuli mammals. Left panel: in nematodes,
programmed cell death during development
requires the activation of a pathway that
BID,
mitochondrion ends in the activation of the caspase CED-3,
EGL-1 Apaf1 BIM
Reaper a type of protease, by the protein CED-4.
Hid CED-3 triggers the cellular changes that
Grim
Sickle Bcl-2 lead to cell death. The actions of CED-3 and
CED-9 Dronc CED-4 are inhibited by CED-9. If CED-9 is
Cyt c/dATP
SMAC, active, the cell will not undergo apoptosis.
DIABLO Inhibition by CED-9 is lifted by the actions
CED-4 Diap1 Drice APAF1 of EGL-1, which is expressed in response
to pro-apoptotic signals. Middle panel: a
similar pathway is present in Drosophila,
CED-3 Cellular targets Caspase-9 with the addition of a second caspase
(Drice). Diap1 is a negative regulator of the
caspases that is not present in nematodes.
IAPs Its inhibition is lifted by the actions of
Apoptosis Apoptosis Caspase-3, -7
proteins such as Reaper and Sickle. Right
panel: the apoptotic pathway in mammals
Cellular targets is even more complex, but uses the same
main components. Apoptotic signals are
transduced through the EGL-1 homologs
BIM and BID to cause the release of pro-
Apoptosis
apoptotic factors from mitochondria. One of
these is cytochrome c, which triggers the
caspase-activation pathway. In the absence
of apoptotic signals, cell death is inhibited
by BCL-2, which is homologous to CED-9. In
Summary all three diagrams, the proteins are color-
coded to show homology.
Several adult tissues, such as the blood and the skin, are continuously being replaced
by the differentiation of new cells from stem cells. Hematopoiesis illustrates how a
single progenitor cell, the multipotent stem cell, can be both self-renewing and gener-
ate a range of different cell lineages. The differentiation of the various lineages of
blood cells seems to depend on external signals, although there is some evidence for
an intrinsic tendency to diversification. External factors are, however, essential for the
survival and proliferation of the specific cell types. The epithelial tissues of adult mam-
malian skin and gut are also replaced from self-renewing stem cells. The derivation of
blood cells from the multipotent stem cell involves a successive restriction of develop-
mental potential as cells become committed to the different lineages. Similar processes
occur in the diversification of the embryonic neural crest. Before they leave the crest,
some crest cells still have a broad developmental potential, and environmental signals
both direct the pathways of neural crest differentiation and promote the survival of
particular cell types.
In general, cell differentiation is controlled by complex combinations of transcrip-
tion factors, whose expression and activity are influenced by external signals. Some
transcription factors are common to many cell types, but others have a very restricted
pattern of expression. Muscle differentiation, for example, can be brought about by
the expression of transcription factors specific to the muscle lineage, such as MyoD,
which initiate the differentiation program and bind to the control regions of muscle-
specific genes. Programmed cell death is a common fate for cells, especially during
development, and there is evidence that positively acting signals are generally required
to ­prevent programmed cell death and allow cells to survive.
394 Chapter 10 Cell differentiation and stem cells

Summary: models of cell differentiation

Muscle Blood cells

muscle precursor cell multipotent stem cell

MyoD family of transcription factors
activation of muscle-specific genes
hematopoietic
growth factors
myoblasts
all blood cell types

myotubes
Neural crest cells
neural crest cells
signals

melanocytes glia cartilage neurons adrenal cells

The plasticity of gene expression

In earlier parts of this chapter we have looked at examples of genes being turned on
and off during development and have considered some of the mechanisms involved.
These chiefly involve the binding of different combinations of gene-regulatory pro-
teins to the control regions of genes. How reversible are these changes, and thus how
reversible is cell differentiation? We have already seen several examples of mutipotent
stem cells that both self-renew and give rise to a range of different cell types. If such
stem cells could be produced reliably and in sufficient numbers, it might be possible
to use them to replace cells that have been damaged or lost by disease or injury. This
is one of the main aims of the field of regenerative medicine. The therapeutic use of
stem cells will depend on understanding precisely how gene activity can be controlled
in stem cells to give the desired cell type, and just how plastic stem cells are. Could,
for example, blood stem cells give rise to neurons under appropriate conditions? How
easy is it to alter the pattern of gene activity? Is it possible to make pluripotent stem
cells from differentiated cells?
We will start the discussion of cell plasticity by examining the extent to which the
pattern of gene activity in differentiated cells can revert to that found in the fertil-
ized egg. One way of finding out whether it can be reversed in practice is to place
the nucleus of a differentiated cell in a different cytoplasmic environment, one that
contains a different set of gene-regulatory proteins.

10.13 Nuclei of differentiated cells can support development

The most dramatic experiments addressing the reversibility of differentiation have

investigated the ability of diploid nuclei from cells at different stages of development
to replace the nucleus of an egg and to support normal development. If they can
do this, it would indicate that no irreversible changes have occurred to the genome
during differentiation. It would also show that a particular pattern of nuclear gene
activity is determined by whatever transcription factors and other regulatory proteins
are being synthesized in the cytoplasm of the cell. Such experiments were first car-
ried out using the eggs of amphibians, which are particularly robust for experimental
manipulation.
 The plasticity of gene expression 395

Cultured adult skin cells UV radiation destroys

DNA in nucleus of
unfertilized egg

Transfer of nucleus
into egg Tadpole develops

Tadpole gut epithelial cells Removal of nucleus from

either cell source

Fig. 10.28 Nuclear transplantation. The

haploid nucleus of an unfertilized Xenopus
egg is rendered non-functional by treatment
with ultraviolet radiation. A diploid nucleus
In the unfertilized eggs of Xenopus, the nucleus lies directly below the surface at taken from an epithelial cell lining the
the animal pole. A dose of ultraviolet radiation directed at the animal pole destroys tadpole gut or from a cultured adult skin cell
the DNA within the nucleus, and thus effectively removes all nuclear function. These is transferred into the enucleated egg, and
enucleated eggs can then be injected with a diploid nucleus taken from a cell at a can support the development of an embryo
later stage of development to see whether it can function in place of the inactivated until at least the tadpole stage.
nucleus. The results are striking: nuclei from early embryos and from some types
of differentiated cells of larvae and adults, such as gut and skin epithelial cells, can
replace the egg nucleus and support the development of an embryo up to the tadpole
stage (Fig. 10.28), and in a small number of cases even into an adult. The organisms
that result are a clone of the animal from which the somatic cell nucleus was taken, as
they have the identical genetic constitution. The process by which they are produced
is called cloning. Several nuclei removed from blastula
Nuclei from adult skin, kidney, heart, and lung cells, as well as from the intestinal
cells and myotome of Xenopus tadpoles, can support development when transplanted
into enucleated eggs. This technique is known as somatic cell nuclear transfer. How-
ever, the success rate with nuclei from somatic cells of adults is very low, with only a
small percentage of nuclear transplants developing past the cleavage stage. In general,
the later the developmental stage the nuclei come from, the less likely they are to be
able to support development. Transplantation of nuclei taken from blastula cells is
much more successful; by transplanting nuclei from the same blastula into several
Nuclei implanted in enucleated eggs
enucleated eggs, clones of genetically identical frogs can be obtained (Fig. 10.29). An
even greater success rate is achieved if the process is repeated with nuclei from the
blastula stage of a cloned embryo. These results show that, at least in those differenti-
ated cell types tested, the genes required for development are not irreversibly altered.
More importantly, when exposed to the egg’s cytoplasmic factors, they behave as the
genes in the zygotic nucleus of a fertilized egg would. So in this sense at least, many
of the nuclei of the embryo and the adult are equivalent, and their behavior is deter-
mined entirely by factors present in the cell.
What about organisms other than Xenopus? Similar results have been obtained Genetically identical frogs develop
in insects, where nuclei from the blastoderm stage can, when returned to the egg,
participate in the formation of a wide range of tissue types. In ascidians, the genetic
equivalence of nuclei at different developmental stages can also be demonstrated. In

Fig. 10.29 Cloning by nuclear transfer. Nuclei from cells of the same blastula are transferred
into enucleated, unfertilized Xenopus eggs. The frogs that develop are clones as they all have
the same genetic constitution.
396 Chapter 10 Cell differentiation and stem cells

plants, single somatic cells can give rise to fertile adult plants, demonstrating com-
plete reversibility of the differentiated state at the cellular level (see Fig. 7.9). The first
mammal to be cloned by somatic cell nuclear transfer was a lamb, the famous Dolly.
In this case, the nucleus was taken from a cell line derived from the udder. Mice have
been similarly cloned using, for example, nuclei from the somatic cumulus cells that
surround the recently ovulated egg, and several generations of these mice have been
produced by repeating the procedure.
In general, the success rate of cloning by somatic cell nuclear transfer in mammals
is extremely low, and the reasons for this are not yet well understood. Most cloned
mammals derived from nuclear transplantation die before birth, and those that sur-
vive are usually abnormal in some way. The most likely cause of failure is incomplete
reprogramming of the donor nucleus to remove all the epigenetic modifications, such
as DNA methylation and histone modification, that are involved in determining and
maintaining the differentiated cell state, as discussed earlier, and thus a failure to
restore the DNA to a state resembling that of a newly fertilized oocyte. A related
cause of abnormality may be that the reprogrammed genes have not gone through the
normal imprinting process that occurs during germ-cell development, where different
genes are silenced in the male and female parents (see Section 9.8). The abnormalities
in adults that do develop from cloned embryos include early death, limb deformities
and hypertension in cattle, and immune impairment in mice, and all these defects are
thought to be due to abnormalities of gene expression that arise from the cloning pro-
cess. Dolly the sheep developed arthritis at 5 years old and was humanely put down
at the age of 6 years (relatively young for sheep) after developing a progressive lung
disease, although there is no evidence that cloning was a factor in Dolly contract-
ing the disease. Studies have shown that some 5% of the genes in cloned mice are
not correctly expressed and that almost half of the imprinted genes are incorrectly
expressed.
The generation of cloned blastocysts following the transfer of an adult somatic
cell nucleus has more recently been achieved in primates and humans by modifying
the protocols used for nuclear transfer in other animals. In both cases, nuclei from
adult skin cells have been transferred into enucleated oocytes, many of which then
underwent cleavage, with a small number eventually forming blastocysts. One of the
reasons for attempting to clone mice and other mammals is to generate embryonic
stem cells (ES cells) from the cells of the inner cell mass of a cloned blastocyst (see
Section 3.9). ES cells have been successfully derived from cloned monkey blastocysts
but this has not yet been achieved for human blastocysts. Cloning of an adult primate
by nuclear transfer has not yet been accomplished and cloning a human being by
these means, even if it could be achieved, is banned in most countries on both practi-
cal and ethical grounds. Despite reports in the media that humans have been cloned,
none of these reports has been verified.

10.14 Patterns of gene activity in differentiated cells can be changed

by cell fusion

Nuclear transplantation into eggs, particularly frog eggs, is facilitated by their size and
the large amount of cytoplasm. With other cell types, particularly differentiated cells,
injection of nuclei into foreign cytoplasm is not possible. It is possible, however, to
expose the nucleus of one cell type to the cytoplasm of another type by fusing the two
cells together. In this process, which can be brought about by treatment with certain
chemicals or viruses, the plasma membranes fuse and nuclei from different cells come
to share a common cytoplasm. A cell-division inhibitor is then used to ensure that the
two nuclei remain separate.
A striking example of the reversibility of gene activity after cell fusion is provided
by the fusion of chick red blood cells with human cancer cells in culture. Unlike
 The plasticity of gene expression 397

mammalian red blood cells, mature chick red blood cells have a nucleus. Gene activ-
Human Mouse muscle cell
ity in the nucleus is, however, completely turned off and it thus produces no mRNA. liver cell
When a chick red blood cell is fused with a cell from a human cancer cell line, gene
expression in the red blood cell nucleus is reactivated, and chick-specific proteins are
expressed. This expression of chick-specific proteins shows that human cells contain
cytoplasmic factors capable of initiating transcription in chick nuclei.
Fusion of differentiated cells with striated muscle cells from a different species
provides further evidence for reversibility of gene expression in differentiated cells
(Fig. 10.30). Multinucleate striated muscle cells are good partners for cell-fusion stud- Human liver cell and mouse muscle cell fused
ies as they are large, and muscle-specific proteins can easily be identified. Differenti-
ated human cells representative of each of the three germ layers have been fused
with mouse multinucleate muscle cells, so that the human cell nuclei are exposed to
the mouse muscle cytoplasm. This results in muscle-specific gene expression being
switched on in the human nuclei. For example, human liver cell nuclei in mouse
muscle cytoplasm no longer express liver-specific genes; instead, their muscle-specific
Activation of human muscle gene expression.
genes are activated and human muscle proteins are made. Furthermore, expression Mouse and human muscle proteins are produced
of genes such as MyoD, which as we have seen can initiate muscle differentiation, is
activated, thus showing that reprogramming of the liver-cell nuclei in mouse muscle
cytoplasm appears to proceed through the same steps as those taken when a muscle
cell differentiates.
These results clearly show that patterns of gene expression in differentiated cells
can be changed, and that gene expression can be controlled by factors present in Fig. 10.30 Cell fusion shows the
the cytoplasm. At least some of these factors are transcription factors, as we will see reversibility of gene inactivation during
below, and this leads to the conclusion that the differentiated state may be at least differentiation. A human liver cell is fused
partly maintained by the continuous action of transcription factors on their target with a mouse muscle cell. The exposure
genes (see Section 10.3). of the human nucleus (red) to the mouse
muscle cytoplasm results in the activation
of muscle-specific genes and repression of
10.15 The differentiated state of a cell can change by transdifferentiation liver-specific genes in the human nucleus.
Human muscle proteins are made together
A fully differentiated cell is generally stable, which is essential if it is to serve a par-
with mouse muscle proteins. This shows that
ticular function in the mature animal. If it has retained the ability to divide, the cell
the inactivation of muscle-specific genes in
passes its differentiated state on to all its descendants. In some long-lived cells, such as human liver cells is not irreversible.
neurons, which do not divide once they are differentiated, the differentiated state must
remain stable for many years. Plant cells maintain their differentiated state while in the
plant, but do not maintain it when placed in cell culture (see Chapter 7). Under cer-
tain conditions differentiated animal cells are not stable, and this provides yet another
demonstration of the potential reversibility of patterns of gene activity. The change of
one differentiated cell type into another is known as transdifferentiation. The allied
phenomenon in which there is a switch of committed but not yet differentiated pro-
genitor cells into a different lineage is known as transdetermination. The best stud-
ied examples of transdetermination occur in regenerating Drosophila imaginal discs,
where rare transdetermination events result in the homeotic transformation of one type
of adult structure into another.
Transdifferentiation occurs very rarely as part of normal development, and in verte-
brates is more commonly seen in regeneration (discussed in Chapter 14) and certain
pathological conditions. There are some examples of transdifferentation in normal
invertebrate development, however. In Caenorhabitis elegans, in which the fate of
every cell can be followed, a few cells appear to undergo transdifferentiation. An epi-
thelial cell (known as Y), which forms part of the rectum, is generated in the embryo
and has the morphological and molecular hallmarks of a differentiated epithelial cell.
In the second larval stage, this cell leaves the rectum epithelium and migrates to
form a motor neuron (known as PDA), which now possesses an axonal process and
makes synaptic connections. Thus a fully differentiated epithelial cell with no detect-
able neuronal features undergoes transdifferentiation into a neuron with no residual
398 Chapter 10 Cell differentiation and stem cells

epithelial characteristics. It is not yet clear whether the Y cell dedifferentiates and
then redifferentiates as the PDA cell or whether dedifferentiation and redifferentiation
occur at the same time.
A role for the extracellular matrix in determining the state of cell differentiation
is suggested by a particularly interesting example of transdifferentiation in jellyfish
striated muscle. This can be made to transdifferentiate into two different types in
succession. When a small patch of striated muscle and its associated extracellular
matrix (mesogloea) is cultured, the striated muscle state is maintained. However,
if the cultured tissue is treated with enzymes that degrade the extracellular matrix,
the cells form an aggregate, and within 1–2 days some have transdifferentiated into
smooth muscle cells, which have a different cellular morphology. This is followed by
the appearance of a second cell type—nerve cells. This example indicates a role for
the extracellular matrix in maintaining the differentiated state of the striated muscle.
The classic example of transdifferentiation in vertebrates is lens regeneration in the
adult newt eye (Fig. 10.31). When the lens is completely removed by surgery (lentec-
tomy), a new lens vesicle starts to form around 10 days post-lentectomy from the dorsal
region of the pigmented epithelium of the iris. The iris cells first dedifferentiate—they
lose their pigmentation and alter their shape from a flat to a columnar epithelium—and
start to proliferate. They then redifferentiate to form a new lens. The lens of the adult
mammalian or avian eye cannot regenerate in vivo, but the pigmented epithelial cells
of the embryonic chick retina can be induced to transdifferentiate in culture. A single
pigmented cell from the embryonic retina can be grown in culture to produce a mono-
layer of pigmented cells. On further culture in the presence of hyaluronidase, serum,
and phenylthiourea, the cells lose their pigment and retinal cell characteristics. If cul-
tured at a high density with ascorbic acid, the cells then start to take on the structural
characteristics of lens cells, and to produce the lens-specific protein crystallin. It should
be noted that, in both the newt and the chick, transdifferentiation occurs to a develop-
mentally related cell type. Retinal and iris pigment cells and lens cells are all derived
from the ectoderm (vertebrate eye development is discussed in Chapter 11).
Transdifferentiation could also underlie well-documented examples in humans, usu-
Fig. 10.31 Lens regeneration in adult ally associated with tissue damage and repair, in which rare differentiated cells of one
newt. Top panels: Removal of the lens from
the eye of a newt results in regeneration
of a new lens from the dorsal pigmented Lens removed Regeneration of lens from iris
epithelium of the iris. Bottom panels: sections
through an intact and a regenerating newt lens iris
eye viewed by scanning electron microscopy.
(a) A section through an intact eye. Note that
the aqueous chamber (a) is clear, whereas
the vitreous chamber (v) is dominated by
compacted layers of extracellular matrix
(pink). Lens is gray, cornea is shown as light
blue, the retina as purple and the ligaments
and matrix associated with the attachment
to the lens is colored brown. (b) A section
through an eye 5 days post-lentectomy.
The aqueous chamber has been invaded
by extracellular matrix and the iris has
become thickened. (c) A section through an
eye 25 days post-lentectomy. The aqueous
chamber is gradually clearing, the lens, which
developed from the dorsal rim of the iris, is
now attached to the ventral iris (arrow), and
extracellular matrix in the vitreous chamber
is becoming reorganized. Magnification × 50.
Micrographs from Tsonis, P.A., et al.: 2004. a b c
 The plasticity of gene expression 399

type appear in an inappropriate location in another tissue. Such examples include the
appearance of hepatic cells in the pancreas and vice versa, and gastric-type epithe-
lium in the esophagus in place of stratified squamous epithelium. In the latter exam-
ple, known as Barrett’s metaplasia, the change predisposes the patient to developing
esophageal cancer, and illustrates the clinical importance of transdifferentiation. More
optimistically, it might be possible to harness the process of transdifferentiation to pro-
duce selected differentiated cell types for regenerative medicine, as discussed later.

10.16 Embryonic stem cells can proliferate and differentiate into many cell
types in culture

The multipotent stem cells that are responsible for tissue renewal in adults are limited in
the range of different cell types they give rise to, and require a specific stem-cell niche to
maintain their stem-cell state, such as the bone marrow stroma that supports hematopoi-
etic stem cells (see Section 10.4). In contrast, pluripotent early embryonic cells can dif-
ferentiate into cell types of all three germ layers and do not require a special niche.
The primary examples of pluripotent stem cells in mammals are the embryonic
stem cells (ES cells) derived from the inner cell mass of the blastocyst. Mouse ES cells
have been studied intensively. They can be maintained in culture for long periods,
apparently indefinitely, but if injected into a blastocyst that is then returned to the
uterus, they can contribute to all the types of cells in that embryo, but not to extraem-
bryonic structures. One way of testing the pluripotent nature of ES cells is to return
them to a blastocyst formed of tetraploid cells, which themselves can only form pla-
cental tissues (Box 10B). The ES cells then form the entire embryo.

Box 10B Testing ES cell potential in tetraploid blastocysts

One reason for cloning mice is to generate embryonic stem

cells (ES cells) from the cells of the inner cell mass of a cloned
2n 2n diploid
blastocyst. These cells can be grown in culture to provide large blastomere
numbers of pluripotent stem cells of a known genetic constitu-
tion for further experiments; for example, in mouse models of
human diseases. To determine whether cultured ES cells were
truly pluripotent and could support the complete development
4n tetraploid
of a fetus, they were injected into tetraploid host blastocysts blastomere
derived from the fusion of blastomeres from two-cell diploid
embryos in culture (see figure). The chimeric blastocyst was ES cells
then replaced in the uterus of a surrogate mother mouse. The
tetraploid cells are only able to form the placenta, and there-
fore injecting ES cells into such embryos provides a way of
confirming and studying their full developmental potential.
Many of the embryos developed into normal mice, and all the
tetraploid cultured diploid
tissues in the subsequent embryos came from the injected ES blastocyst ES cells
cells.
400 Chapter 10 Cell differentiation and stem cells

To maintain mouse ES cells in a pluripotent state in cell culture, the cells must
express a particular combination of transcription factors (most notably Oct 3/4, Sox2,
c-Myc and Kfl4), whose expression together is restricted to pluripotent stem cells. It
was found that transcription of these genes could be maintained when the cytokines
leukemia inhibitory factor (LIF) and BMP-4 were added to the culture medium. More
recently it has been shown that small-molecule drugs that block differentiation by
inhibiting mitogen-activated kinase (MAPK) signaling pathways, and which promote
survival by inhibiting GSK3 signaling, can be used to derive and propagate ES cells
from mouse and rat blastocysts. This may turn out to be a universal approach to
deriving pluripotent ES cells.
The first human ES cells were produced in 1998. They express the same transcrip-
tion factors as do mouse ES cells, but the signaling pathways that support self-renewal
in cell culture are different. FGF and activin/Nodal signaling but not LIF are required
for maintaining human ES cells, while BMP signaling promotes differentiation, the
opposite effect to that in mouse ES cells. Several hundred human ES cell lines have
now been established around the world.
ES cells can be made to differentiate into a particular cell type by manipulating the
culture conditions, particularly in respect of the growth factors present. If grown in sus-
pension, ES cells spontaneously form aggregates called embryoid bodies, which can dif-
ferentiate into many cell types. When cultured in the absence of serum, which contains
various growth and differentiation factors, many of which are still unknown, ES cells dif-
ferentiate into neural cells. Under other treatments, ES cells can differentiate into heart
muscle, blood islands, neurons, pigmented cells, epithelia, fat cells, macrophages, and
even germ cells. For example, aggregates of ES cells successively treated with particular
combinations of FGF-2, EGF and platelet-derived growth factor continue to proliferate as
long as these growth factors are present. But when they are withdrawn, the cells differ-
entiate into either of two glial cell types—astrocytes or oligodendrocytes. Cell shape can
also regulate stem-cell commitment. Human mesenchymal stem cells allowed to flatten
in culture differentiate as osteocytes (bone cells), whereas if they remain rounded they
become adipocytes (fat cells). In regard to the potential use of neural stem cells, it is
encouraging that mouse ES cells made to differentiate specifically into neural progeni-
tors by treatment with retinoic acid can, when transplanted into the neural tube of a
mouse embryo, develop into neurons appropriate to their new location.

10.17 Stem cells could be a key to regenerative medicine

The goal of regenerative medicine is to restore the structure and function of damaged
or diseased tissues. As stem cells can proliferate and differentiate into a wide range of
cell types, they are strong candidates for use in cell-replacement therapy, the restora-
tion of tissue function by the introduction of new healthy cells. This type of therapy
might eventually offer an alternative to conventional organ transplantation from a
donor, with its attendant problems of rejection and shortage of organs, and might also
be able to restore function to tissues such as brain and nerves. Both ES cells and adult
stem cells have been studied with cell-replacement therapy in mind. Transdifferentia-
tion is also emerging as an alternative route to generating selected cell types, but the
recent development of induced pluripotent stem cells (iPS cells) (Box 10C, p. 401)
offers the most exciting new opportunities.
ES cells have the advantage over adult stem cells in that they can differentiate into
a wide range of different cell types and could, in theory, be used to repair any tis-
sue. But stem cells from an unrelated donor embryo will still cause immune rejection
reactions, and because of the proliferative capacity and pluripotency of ES cells, the
risk of embryonal tumors—tumors derived from embryonic cells—is high. Cultured
ES cells introduced into another embryo will develop normally, but when ES cells
are introduced under the skin of a genetically identical adult mouse they give rise to
 The plasticity of gene expression 401

Box 10C Induced pluripotent stem cells

The most dramatic example of reprogramming of adult differenti-

Skin fibroblasts grown from transgenic mouse
ated cells is the production of embryonic stem (ES) cell-like cells,
known as induced pluripotent stem cells (iPS cells) by the intro-
duction of the genes for four transcription factors associated
with pluripotency (see figure). It had previously been shown that
when adult differentiated somatic cells are fused with ES cells
they are transformed into pluripotent cells, suggesting that cyto-
plasmic factors in the ES cells might be able to confer ES cell-like
Retroviral transfection of pluripotency genes
properties on differentiated cells. Nevertheless it was surprising
that so few factors are needed to induce pluripotency.
The first iPS cells were produced from skin fibroblasts iso-
lated from transgenic mice that carried a drug-selectable marker
linked to expression of a gene associated with pluripotency,
either Oct3/4 or Nanog. The genes encoding four transcription
factors associated with pluripotency in ES cells—Oct3/4, Sox2, Antibiotic selection and growth in culture
Kfl4, and c-Myc—were introduced into the adult fibroblasts using
iPS cell line
retroviral vectors. The transfected cells were then treated with
the appropriate drugs to select for the rare iPS cells. Most trans-
fected cells did not express either Oct3/4 or Nanog and there-
fore had no drug resistance and died. The remaining colonies of
cells that survived resembled ES cells and could be isolated and
expanded in culture. The inserted genes in the iPS cells shut Inject into normal blastocyst
themselves off once the reprogramming process was completed,
and the cells reactivated their own pluripotency genes to main-
tain the pluripotent state. The same set of transcription factor
genes was very soon found to be able reprogram adult human
fibroblasts, but the conditions needed to expand human iPS cell
populations are different, with mouse cells requiring LIF and
human cells FGF (see Section 10.16). Resulting chimeric mouse crossed with normal mouse

The pluripotency of mouse iPS cells was tested by injecting

them into normal mouse blastocysts, which were then replaced
in a surrogate mother. The iPS cells contributed to all cell types
in the resulting embryos, including the germline. When these iPS
chimeric mice were crossed with normal mice, iPS-derived gam-
etes successfully participated in fertilization and live mice were
Offspring
born. More recently, fertile adult mice generated entirely from
iPS cells have been derived from tetraploid blastocyts into which iPS-derived mouse with
fibroblast-derived iPS cells were injected (see Box 10B, p. 399, one gamete of iPS origin
for a description of this definitive procedure for testing pluripo-
tency). The efficiency of the generation of healthy mice from iPS
cells by this procedure is still very low, however, and many of the
pups produced died soon after birth.
Advances have been made over the past few years in inducing
pluripotency without the need to introduce a persisting viral vec-
tor. This will be a prerequisite for any medical use for iPS cells in An equal effort is going into reducing the number of genes
cell replacement. Transposon-based vectors have been produced that need to be introduced. Adult neural stem cells have been
that will snip themselves and the genes they carry cleanly out of reprogrammed by introducing just one of the transcription fac-
the genome after the transformation to pluripotency is complete. tors, Oct4. Adult neural stem cells already express high levels of
Other approaches involve searching for small-molecule drugs Sox2 and c-Myc, however, and this is why introduction of Oct4
that will act as co-activators for a cell’s own pluripotency genes. alone is sufficient to induce pluripotency.
402 Chapter 10 Cell differentiation and stem cells

ES cells in culture

ES cells injected into inner cell ES cells injected under the skin of adult mouse
mass of normal blastocyst

Fig. 10.32 Embryonic stem cells (ES cells)

can develop normally or into a tumor,
Chimeric mouse develops normally Mouse develops teratocarcinoma
depending on the environmental signals
they receive. Cultured ES cells originally
obtained from an inner cell mass of a mouse
contribute to a healthy chimeric mouse when
injected into an early embryo (bottom left
panels), but if injected under the skin of an
adult mouse, the same cells will develop into
a teratocarcinoma (bottom right panels).

a tumor known as a teratocarcinoma (Fig. 10.32). These unusual tumors contain a

mixture of differentiated cells.
To generate ES cells of the patient’s own tissue type, it would be a great advantage to
be able to make use of therapeutic cloning by somatic cell nuclear transfer (see Sec-
tion 10.13). A nucleus from a somatic cell from the patient would be introduced into
an enucleated human oocyte which would then be allowed to develop to the blastocyst
stage. The blastocyst could then be used as a source of ES cells that could be maintained
in culture, and which would not cause immune rejection when implanted into the
patient. But although human blastocysts have now been made by somatic cell nuclear
transfer, the success rate is very low and no ES cells have yet been derived from them.
There are claimed to be ethical issues associated with the use of stem cells and thera-
peutic cloning. To make human ES cells, the blastocyst must be destroyed, and there
are those who believe that this is a destruction of a human life. There is good evidence
that the blastocyst does not necessarily represent an individual at this very early stage,
as it is still capable of giving rise to twins at a later stage. And in practice, many early
embryos are lost during assisted reproduction involving in vitro fertilization (IVF), a
widely accepted medical intervention. Acceptance of IVF and rejection of the use of ES
cells could be seen as a contradiction. New alternative approaches, such as transdiffer-
entiation and the use of induced pluripotent stem cells, could produce cells of a patient’s
own tissue type but avoid the ethical issues associated with therapeutic cloning.
Adult stem cells have also been considered for use in cell-replacement therapy.
Adult hematopoietic stem-cell transplants in the form of bone marrow transplants from
donors have been successfully used for many years as therapy for certain immunologi-
cal diseases and cancers, despite associated problems such as graft-versus-host rejec-
tion. The advantage of adult stem cells in certain other conditions is that they could
 The plasticity of gene expression 403

be taken from the patient’s own undamaged tissues and re-implanted where needed,
without any problems of tissue rejection. This is already done in the case of skin grafts,
which use the patient’s own skin. But the range of cell types and tissues that could be
repaired using a patient’s own stem cells is limited. The isolation of stem cells from
the brain, for example, would not be feasible. There have been reports over the years
that adult hematopoietic stem cells will differentiate into cell types quite different from
their normal range when transplanted into another type of tissue. However, it was sub-
sequently shown that hematopoietic stem cells had not been able to give rise to cells
outside their normal repertoire but instead had fused with existing cells.

10.18 Various approaches can be used to generate differentiated cells for

cell-replacement therapies

One of the challenges facing cell-replacement therapy is devising differentiation pro-

tocols that will generate the particular cell types needed for repair, whether from ES
cells, adult stem cells, or by transdifferentiation of pre-existing cell types. The gen-
eration of insulin-producing pancreatic β cells to replace those destroyed in type 1
diabetes is a prime medical target, and we shall use it to illustrate the experimental
approaches that are being developed for cell-replacement therapy generally.
In type 1 diabetes, the insulin-producing β cells of the endocrine pancreas (the pan-
creatic islets) are destroyed by an autoimmune reaction, resulting in non-production
of insulin and the potentially lethal build-up of sugar in the blood. Unlike blood or
muscle, the pancreas does not contain dedicated stem cells from which it can regener-
ate, and people with type 1 diabetes face a lifetime of dependence on insulin injec-
tions. Pancreas or islet-cell transplants from cadaver donors have had some success,
but tissue for transplantation is scarce and treatment with powerful immunosuppres-
sant drugs is necessary to prevent rejection. Transplantation of insulin-producing pan-
creatic cells derived from the patient’s own cells, along with the induction of immune
tolerance to these cells, might eventually provide one long-term solution.
Treatments that direct the differentiation of ES cells towards making endoderm
derivatives such as pancreatic cells have been particularly difficult to find. And to be
useful for cell replacement, the insulin-producing cells generated must also be respon-
sive to the signals, such as glucose, that switch on insulin production as required.
Nevertheless, using knowledge of the signals that induce endoderm and pancreas
development in mouse embryos, progress has been made in devising methods for
differentiating human ES cells into pancreatic progenitor cells. Human ES cells put
through a 9-day laboratory protocol involving stepwise treatments with different sig-
nal molecules differentiate into endoderm, expressing the endoderm-specific tran-
scription factor FoxA2, but only a few of the cells in these cultures (around 5%)
express Pdx1, a transcription factor essential for pancreas development. The next step
was therefore to find ways of boosting the differentiation of pancreatic progenitor
cells. A library of 5000 organic compounds was screened and one called indolactam
V was found to stimulate the differentiation of Pdx1-positive cells. The percentage of
Pdx1-expressing cells increased to 45% when endoderm cultures derived from the
human ES cells were treated with indolactam V together with FGF-10, one of the sig-
nals involved in embryonic pancreatic development. When the Pdx1-expressing cells
were transplanted under the kidney capsule in a mouse, a site that allows continued
differentiation, they differentiated into insulin-secreting cells (Fig. 10.33).
Induced pluripotent stem cells (iPS cells) offer the greatest opportunities for gener-
ating pluripotent cells of a patient’s own tissue type for cell replacement. As described
in Box 10C (p. 401), iPS cells were first derived from mouse fibroblasts by introduc-
ing and expressing genes for four transcription factors, Oct 3/4, Sox2, c-Myc and
Kfl4, which are associated with pluripotency in ES cells. In regard to diabetes, iPS
cells have been derived from fibroblasts taken from patients with type 1 diabetes.
404 Chapter 10 Cell differentiation and stem cells

Fig. 10.33 Three experimental routes

for producing insulin-producing cells Somatic cell nuclear transfer Skin fibroblast Induced pluripotency
by nuclear reprogramming. Two routes
start with a skin fibroblast. In somatic
cell nuclear transfer, the nucleus of the
blastocyst Transfection with
fibroblast is introduced into an unfertilized
OCT4, SOX2, cMYC and KLF4
egg and embryonic stem cells (ES cells)
are isolated from the resulting blastocyst. unfertilized egg
Alternatively, cultured fibroblasts can
be transfected with the genes for the
pluripotency transcription factors (Oct4,
Sox2, c-Myc, and Kfl4) to produce induced
pluripotent cells (iPS cells). The ES cells
and iPS cells can then be differentiated
into insulin-producing cells in vitro. A third
possible route is the transdifferentiation of
adult exocrine pancreatic cells or liver cells ES cells iPS cells
into insulin-producing cells in vivo by the
targeted transfection of these organs with Transdifferentiation in vivo Cells put through differentiation protocol
with growth factors and small molecules
genes (including that for the transcription
Adult exocrine pancreas
factor Pdx1) required for differentiation of Liver
endocrine pancreatic cells. Steps in red have
been accomplished for human cells. Note that
Pdx1-expressing cells
insulin-producing glucose-responsive cells
can also be made starting with human ES Transfection of organs with Pdx1 and
cells derived from a normal blastocyst and three other transfection factors or with Insulin-producing glucose-responsive cells
activated Pdx1
differentiated in vitro.
Adapted from Gurdon, J.B., Melton, D.A.: 2008.

The fibroblasts were reprogrammed using the human genes OCT4, SOX2 and KLF4,
and were subsequently redifferentiated into insulin-producing cells (see Fig. 10.33). In
the immediate future, such cells will be used to investigate the cellular and molecular
defects underlying type 1 diabetes, which are still largely unknown.
Transdifferentiation may be yet another way of generating replacement cells, although
this approach is likely to be limited to developmentally related cell types. The feasibility
of this approach for generating insulin-producing endocrine cells from liver cells and
from pancreatic exocrine cells has been demonstrated experimentally. The liver and the
pancreas both derive from endoderm, and arise from adjacent regions of endoderm in
the embryo, but the transcription factor Pdx1, which is essential for pancreatic develop-
ment, is not expressed in the liver. When an activated form of Pdx1 is expressed tran-
siently in the liver in transgenic Xenopus tadpoles under the control of a liver-specific
promoter, however, part or all of the liver is converted to both exocrine and endocrine
pancreatic tissue, while proteins characteristic of differentiated liver cells disappear from
the cells in the converted regions. In another example of transdifferentiation, progenitor
liver cells in adult mice have also been induced to develop into insulin-secreting cells
resembling pancreatic islet cells by infecting the liver with viral vectors expressing the
transcription factor neurogenin, which is another pancreatic transcription factor.
Transdifferentiation of pancreatic exocrine cells into endocrine β cells has also been
achieved in adult mice. A combination of three transcription factors required for endo-
crine pancreatic cell differentiation, including Pdx1, were delivered into the pancreas
using genetically modified adenoviruses. Endocrine cells were induced and, although
they were not found in islets, these induced cells closely resembled their normal
counterparts and had the ability to secrete insulin. Transdifferentiation was efficient
(around 20%) and relatively fast, with the first insulin-positive cells appearing on day
3 after treatment.
 The plasticity of gene expression 405

There are, however, many obstacles to overcome before the potential therapeutic ben-
efits of technologies based on ES cells, iPS cells, or transdifferentiation can be realized.
The most effective current procedures for making iPS cells or causing transdifferentia-
tion depend on using viral vectors to introduce the pluripotency genes. Viral vectors can
integrate randomly into the genome and might activate genes that cause cancer, and
so alternative ways of inducing expression of the necessary transcription factors will
need to be found before medical applications are possible. Fortunately, it is possible
to remove the vectors along with the introduced genes after successful conversion to
pluripotency and still maintain a stable iPS cell phenotype. This is because once pluri-
potency is induced by the introduced genes, the cell’s own pluripotency genes become
activated. In addition, there has been some success in obtaining iPS cells by introducing
the four pluripotency proteins into mouse embryonic fibroblasts rather than the genes.
But even if non-viral methods of inducing pluripotency are perfected, it will not
eliminate the risk of tumor induction in patients undergoing cell-replacement ther-
apy with ES cells or iPS cells. If undifferentiated pluripotent cells remain after the
differentiation process and are introduced into the patient they could cause tumors.
Only stringent selection procedures that ensure no undifferentiated cells are present
in the transplanted cell population will overcome this problem. And it is not yet clear
how stable differentiated ES cells and iPS cells will be in the long term. A final, but
not insignificant, obstacle to be overcome is finding ways of generating enough cells
for an effective treatment. At present, ES cells are easier to expand to large numbers
in culture than are adult stem cells or even iPS cells.
We have focused here on just one medical target, the generation of pancreatic insulin-
producing cells to treat type 1 diabetes. Similar strategies could also be used in other
diseases. The neurodegenerative Parkinson’s disease is another medical target and iPS
cells have successfully been made from people with this disease. In the near future,
iPS cells derived from patients will probably be most useful for investigating disease
processes in the context of human cells, as animal models do not always mimic human
disease completely. For example, iPS cells have been derived from skin fibroblasts from a
child with genetically determined spinal muscular atrophy, and motor neurons have been
produced in culture from these cells. Neuron production in the child’s iPS cell cultures
differed from neuron production in cultures of iPS cells made from the child’s healthy
mother, with the number and size of motor neurons produced decreasing over time in
cultures derived from the child. Such cultures could be used to test the effects of drugs,
which could lead to new clinical treatments. More generally, human iPS cells could be
useful for screening compounds for toxicity and for teratogenicity in drug development.

Summary

Transplantation of nuclei from differentiated animal cells into fertilized eggs, and cell-
fusion studies, show that the pattern of gene expression in the nucleus of a differenti-
ated cell can often be reversed, implying that it is determined by factors supplied by
the cytoplasm and that no genetic material has been lost. Although the differentiated
state of an animal cell in vivo is usually extremely stable, some cases of differentiation
are reversible. Transdifferentiation of one differentiated cell type into another has been
shown to occur during regeneration and in cultured cells. Both amphibians and some
mammals can be cloned by nuclear transplantation from an adult somatic cell into an
enucleated egg. Stem cells are self-renewing and can differentiate into a wide range of
cell types. They are being investigated with a view to being used in cell-replacement
therapy. It has been found that the introduction of four transcription factors expressed
in embryonic stem cells into adult differentiated cells induces these cells to become
pluripotent. These induced pluripotent stem cells together with transdifferentiation
provide alternative approaches for producing cells for cell-replacement therapy.
406 Chapter 10 Cell differentiation and stem cells

Summary to Chapter 10

Cell differentiation leads to distinguishable cell types whose specialized characters and
properties are determined by their pattern of gene activity and thus the proteins they pro-
duce. The pattern of gene activity in a differentiated cell can be maintained over long peri-
ods of time and can be transmitted to the cell’s progeny. Maintenance and inheritance of a
pattern of gene activity probably involves a combination of several mechanisms, including
the continued action of gene-regulatory proteins, changes in the packing state of chromatin,
and chemical modifications to DNA.
Differentiation reflects the establishment of a particular pattern of gene expression rather
than the loss of genetic material. Lymphocytes are an exception as their DNA has undergone
irreversible changes during differentiation. The differentiated state of an animal cell in vivo
is usually extremely stable, but in some cases transdifferentiation of one differentiated cell
type into another can occur, for example, during tissue regeneration. The cloning of amphib-
ians and some mammals by somatic cell nuclear transfer into enucleated eggs also shows
that the nuclei of differentiated cells can be reprogrammed and restored to a state in which
they are able to direct embryonic development. Adult somatic cells can also be induced to
develop into pluripotent cells by expression of four transcription factors associated with
pluripotency in embryonic stem cells. Pluripotent cells that can proliferate and differentiate
into cell types from all three germ layers could be the key to regenerative medicine.
Apart from the use of universal mechanisms of gene regulation, there may be little
similarity between the detailed mechanisms of differentiation in different cell types. The
understanding of any given differentiation pathway lies in the detail of the external sig-
nals, internal signal transduction pathways, gene-regulatory proteins, and gene products
at work in each particular case. Induction of cell differentiation by signaling molecules such
as growth factors requires the execution of a complex program of intracellular events. Dif-
ferent stimuli, or even the same stimulus at different developmental stages, can activate
the same internal signaling pathway, yet activate different genes in different cells, because
of the cells’ own developmental histories. Signal transduction at the cell membrane sets in
motion a cascade of events within the cell, which results, for example, in the phosphoryla-
tion and activation of transcription factors and the consequent switching on and off of gene
expression. Programmed cell death is a common fate of cells during development, and there
is evidence that positively acting signals are generally required to prevent it and to allow
cells to survive.

n End of chapter questions possible (include the role of both cis-regulatory DNA sequences and
transcription factors)?
Long answer (concept questions) 4. Create two lists: one of peptide signaling molecules, and one
1. Use the Glossary to review the terms fate, specification, and of lipid-soluble signaling molecules. How do their mechanisms of
determination. How do these cellular states differ from the topic of action differ? What generalization(s) can you make regarding the
this chapter, differentiation? way they activate a transcription factor?
2. Contrast the concept of ‘housekeeping’ genes (and proteins) from 5. Cell division entails the separation of the two strands of the DNA
the cell-specific genes (and proteins) that characterize, and define, a during DNA replication, which disrupts patterns of gene expression
differentiated cell type. Propose a microarray experiment that would that have been previously established in the cell. Nonetheless, the
distinguish the tissue-specific gene expression from housekeeping two daughter cells must somehow ‘remember’ the differentiated
gene expression in muscle cells and neurons. (The principles behind state of the parental cell. Discuss how the differentiated state of a
a DNA microarray analysis are described in Box 3A, p.115.) cell is remembered, using transcription factors, chromatin structure,
and DNA methylation.
3. Gene function is often studied by misexpression of the protein in
an abnormal site in the embryo or abnormal tissue in the adult body 6. Hematopoiesis involves the development of lymphocytes (B and
to see what effects the protein has. What are the essential features of T cells), red blood cells, macrophages, and osteoclasts (among other
the transcriptional apparatus that make misexpression experimentally blood-cell types) from cells in the bone marrow. Outline the general
 The plasticity of gene expression 407

stages that lead to each of the cell types named, giving the name of 19. ES cells and iPS cells are both pluripotent stem cells but are
the cell from which they all derive. obtained in different ways. Describe how each is obtained.

7. What is meant by ‘stem-cell niche’ and what is its role? For one 20. What are the potential advantages of iPS cells over ES cells
of the stem-cell niches described in the text give examples of the for therapeutic purposes? What are the disadvantages? What other
signaling molecules that enable it to carry out this role. approaches for obtaining stem-like cells are possible, and what
problems does each of them circumvent?
8. Differentiation of the erythroid lineage requires signaling by
erythropoietin (Epo), a peptide signaling molecule made by the kid-
ney in response to red blood cell deficiency. One target of Epo is
Multiple choice (factual recall questions)
the GATA-1 transcription factor, and Epo is itself a target of GATA-1.
β-globin is also a target of GATA-1. Using these facts, propose a path- NB There is only one correct answer to each question.
way from a physiological need for more red blood cells, through the 1. The many different cell types in the blood of an adult mammal
multipotent stem cell, to the globin-expressing red blood cell. are derived from
9. Describe the choice between differentiation as a neutrophil and a) single type of multipotent stem cell found in the bone marrow
differentiation as a macrophage at the level of the transcription fac- b) differentiated cells that first developed in the fetal liver, but now
tors involved. Describe the same choice at the level of the signaling reside in the bone marrow
molecules involved.
c) division of differentiated cell types while circulating in the
10. The mouse gene encoding M-CSF is named op, as mice mutant blood
for this gene develop the bone disease osteopetrosis, in which the
d) stem cells that became committed to each specific blood cell type
bones become abnormally hard and brittle. Review the lineages of the
during embryonic development
blood cell types and explain why this mutation would affect bone.
2. What might be the result of producing a mouse knock-out for
11. Why are different globins required at different stages of devel-
G-CSF?
opment? How is the switching between globin types thought to be
controlled? a) The mouse would be unable to make granulocytes.
b) The mouse would be unable to make macrophages.
12. Describe the process by which epithelial cells of the intestine
are replaced from a stem-cell population. What is the role of Wnt c) The mouse would lack red blood cells.
signaling in this process? d) The mouse would lose its hematopoietic stem cell population
completely.
13. Despite its ability to act as a master regulator of muscle dif-
ferentiation, mouse knockouts of myoD are viable. Refer to Fig. 3. The stem cells responsible for renewal of the keratinocytes of the
10.19, and propose a reason that myoD is not essential. In contrast, skin are found in which layer of the epidermis?
knock-out of both myoD and myf5 is lethal; knock-out of mrf4 is a) basal lamina
also lethal; how would you interpret this result?
b) basal layer
14. The differentiation of muscle cells requires withdrawal from c) dermis
the cell cycle. Elaborate on the following points regarding growth
versus differentiation of myoblasts: (1) Removal of growth factors d) keratinocyte layer
causes muscle precursor cells to differentiate in culture. (2) MyoD 4. The introduction of myoD into fibroblasts will cause
activates expression of p21. (3) Loss of Rb will allow myoblasts to
a) apoptosis
re-enter the cell cycle.
b) commitment to the myeloid lineage
15. Referring to Figure 10.23, fill in the extracellular signaling pro-
c) differentiation into muscle cells
teins that promote the various fates of neural crest cells, much as
was done for blood cells in Fig. 10.9. d) formation of red blood cells

16. What is stem-cell factor (SCF)? Why would defects in SCF sig- 5. Differentiation, as opposed to determination, of muscle cells is
naling by loss of its receptor, Kit, lead to a mouse phenotype called dependent on
‘white spotting’? a) Mrf4
17. Contrast apoptosis with necrosis. What is the role of caspases b) Myf5
in apoptosis? c) MyoD
18. Describe the process referred to as cloning by nuclear trans- d) myogenin
plantation. How does the success of this process illustrate the impor-
6. What are satellite cells?
tance of the transcription factor complement present in any given
cell? In contrast, how does the difficulty of the process illustrate the a) Satellite cells are circulating cells ‘orbiting’ in the blood.
importance of epigenetic mechanisms in development? b) Satellite cells are muscle stem cells.
408 Chapter 10 Cell differentiation and stem cells

c) Satellite cells are muscle-associated support cells, much as glial 10.3 The maintenance and inheritance of patterns of gene
cells are neuron-associated support cells. activity depend on chemical and structural modifications to
d) Satellite cells are neuronal stem cells. chromatin, as well as on gene-regulatory proteins
de Laat, W., Grosveld, F.: Spatial organization of gene expression:
7. Neural crest cells can give rise to
the active chromatin hub. Chromosome Res. 2003, 11: 447–459.
a) cartilage of the face Ho, L., Crabtree, G.R.: Chromatin remodelling during develop-
b) parasympathetic ganglia ment. Nature 2010, 463: 474–484.
c) sensory ganglia 10.4 All blood cells are derived from multipotent stem cells
d) all of these Ema, H., Nakauchi, H.: Self-renewal and lineage restriction of
8. The human anti-apoptotic protein that is homologous to Ced-9 hematopoietic stem cells. Curr. Opin. Genet. Dev. 2003, 13:
from C. elegans is ­508–512.
Phillips, R.L., Ernst, R.E., Brunk, B., Ivanova, N., Mahan, M.A.,
a) Bcl-2
Deanehan, J.K., Moore, K.A., Overton, G.C., Lemischka, I.R.: The
b) Caspase-9 genetic program of hematopoietic stem cells. Science 2000, 288:
c) Ced-3 1635–1640.
d) Diablo Zon, L.I.: Intrinsic and extrinsic control of haematopoietic stem-
cell renewal. Nature 2008, 453: 306–313.
9. If the lens of the eye is removed in the adult, which of the fol-
lowing organisms can produce a new lens by transdifferentiation of 10.5 Colony-stimulating factors and intrinsic changes control
iris cells? differentiation of the hematopoietic lineages
a) chicken Anguita, E., Hughes, J., Heyworth, C., Blobel, G.A., Wood, W.G.,
Higgs, D.R.: Globin gene activation during haemopoiesis is
b) human
driven by protein complexes nucleated by GATA-1 and GATA-2.
c) mouse EMBO J. 2004, 23: 2841–2852.
d) newt Kluger, Y., Lian, Z., Zhang, X., Newburger, P.E., Weissman, S.M.:
A panorama of lineage-specific transcription in hematopoiesis.
10. ES cells are
BioEssays 2004, 26: 1276–1287.
a) cells isolated from teratocarcinomas Metcalf, D.: Control of granulocytes and macrophages: molecular,
b) embryonic stem cells derived from the inner cell mass of the cellular, and clinical aspects. Science 1991, 254: 529–533.
mammalian embryo Orkin, S.H.: Diversification of haematopoietic stem cells to spe-
c) epidermal stem cells found in the basal layer of the epidermis cific lineages. Nat Rev. Genet. 2000, 1: 57–64.
d) human stem cells generated by treatment of mature cell types 10.6 Developmentally regulated globin gene expression
with a specific combination of transcription factors is controlled by regulatory sequences far distant from the
coding regions
Multiple choice answer key
Engel, J.D., Tanimoto, K.: Looping, linking, and chromatin activ-
1: a, 2: a, 3: b, 4: c, 5: d, 6: b, 7: d, 8: a, 9: d, 10: b. ity: new insights into β-globin locus regulation. Cell 2000, 100:
499–502.
Tolhuis, B., Palstra, R.J., Splinter, E., Grosveld, F., de Laat, W.: Loop-
n Section further reading ing and interaction between hypersensitive sites in the active
beta-globin locus. Mol. Cell 2002, 10: 1453–1465.
10.1 Control of transcription involves both general and
tissue-specific transcriptional regulators 10.7 The epithelia of adult mammalian skin and gut are
Cheung, W.L., Briggs, S.D., Allis, C.D.: Acetylation and chromo- continually replaced by derivatives of stem cells
somal functions. Curr. Opin. Cell Biol. 2000, 12: 328–333. Barker, N., van Es, J.H., Kuipers, J., Kujala, P., van den Born, M.,
Levine, M., Tjian, R.: Transcription regulation and animal diver- ­Cozijnsen, M., Haegebarth, Andrea, Korving, J., Begthel, H.,
sity. Nature 2003, 424: 147–151. ­Peters, P.J., Clevers, H.: Identification of stem cells in small
Mannervik, M., Nibur, Y., Zhang, H., Levine, M.: Transcriptional intestine and colon by marker gene Lgr5. Nature 2007, 449:
coregulators in development. Science, 1999, 284: 606–609. 1003–1008.
Coulombe, P.A., Kerns, M. L., Fuchs, E.: Epidermolysis bullosa
10.2 External signals can activate genes simplex: a paradigm for disorders of tissue fragility. J. Clin.
Brivanlou, A.H., Darnell, J.E.: Signal transduction and control of ­Invest. 2009, 119: 1784–1793.
gene expression. Science 2002, 295: 813–818. Fuchs, E.: The tortoise and the hair: slow-cycling cells in the stem
Dougherty, D.C., Park, H.M., Sanders, M.M.: Interferon regulatory cell race. Cell 2009, 137: 811–819.
factors (IRFs) repress transcription of the chicken ovalbumin Fuchs, E.: Finding one’s niche in the skin. Cell Stem Cell 2009, 4:
gene. Gene 2009, 439: 63–70. 499–502.
 The plasticity of gene expression 409

Fuchs, E., Nowak, J.A.: Building epithelial tissues from skin stem Relaix, F., Rocancourt, D., Mansouri, A., Buckingham, M.A.: pax3/
cells. Cold Spring Harb. Symp. Quant. Biol. 2008, 73: 333–350. pax7 dependent population of skeletal muscle progenitor cells.
Sato, T., Vries, R.G., Snippert, H.J., van de Wetering, M., Barker, N., Nature 2005, 435: 948–953.
Stange, D.E., van Es, J.H., Abo, A., Kujala, P., Peters, P.J., Clevers,
H.: Single Lgr5 stem cells build crypt villus structures in vitro 10.11 Embryonic neural crest cells differentiate into a wide
without a mesenchymal cell niche. Nature 2009, 459: 262–265. range of different cell types
Shackleton, M., Vaillant, F., Simpson, K.J., Stingl, J., Smyth, G.K., Anderson, D.J.: Genes, lineages and the neural crest. Proc. R. Soc.
Asselin-Labat, M-L., Wu, L., Lindeman, G.J., Visvader, J.E.: Lond. B 2000, 355: 953–964.
­Generation of a functional mammary gland from a single stem Breau M.A., Pietri, T., Stemmler, M.P., Thiery, J.P., Weston, J.A.:
cell. Nature 2006, 439: 84–88. A non-neural epithelial domain. Proc. Natl Acad. Sci. USA 2008,
Van der Flier, L.G., Clevers, H.: Stem cells self-renewal and differ- 105: 7750–7757.
entiation in the intestinal epithelium. Annu. Rev. Physiol. 2009, Bronner-Fraser, M.: Making sense of the sensory lineage. Science
71: 241–260. 2004, 303: 966–968.
Dorsky, R.I., Moon, R.T., Raible, D.W.: Environmental signals and
10.8 The MyoD family of genes determines differentiation cell fate specification in premigratory neural crest. BioEssays
into muscle 2000, 22: 708–716.
Kassar-Duchossoy, L., Gayraud-Morel, B., Gomes, D., Rocancourt, Le Douarin, N.M., Creuzet, S., Couly, G., Dupin, E.: Neural crest cell
D., Buckingham, M., Shinin, V., Tajbakhsh, S.: Mrf4 determines plasticity and its limits. Development 2004, 131: 4637–4650.
skeletal muscle identity in Myf5: Myod double-mutant mice. Lee, H.Y., Kleber, M., Hari, L., Brault, V., Suter, U., Taketo, M.M.,
Nature 2004, 431: 466–471. Kemler, R., Sommer, L.: Instructive role of Wnt/beta-catenin
Tapscott, S.J.: The circuitry of a master switch: Myod and the in sensory fate specification in neural crest stem cells. Science
regulation of skeletal muscle transcription. Development 2005, 2004, 303: 1020–1023.
132: 2685–2695. Meier, P., Finch, A., Evan, G.: Apoptosis in development. Nature
2000, 407: 796–801.
10.9 The differentiation of muscle cells involves withdrawal Minoux, M., Antonarakis, G.S., Kmita, M., Duboule, D., Rijli, F.M.:
from the cell cycle, but is reversible Rostral and caudal pharyngeal arches share a common neural
Deato, M.D., Tjian, R.: An unexpected role of TAFs and TRFs in crest ground pattern. Development 2009, 136: 637-634.
skeletal muscle differentiation; switching of core promoter Morrison, S.J., White, P.M., Zock, C., Anderson, D.J.: Prospective
complexes. Cold Spring Harb. Symp. 2008, 73: 217–225. identification, isolation by flow cytometry, and in vivo self-re-
Novitch, B.G., Mulligan, G.J., Jacks, T., Lassar, A.B.: Skeletal mus- newal of multipotent mammalian neural crest stem cells. Cell
cle cells lacking the retinoblastoma protein display defects in 1999, 96: 737–749.
muscle gene expression and accumulate in S and G2 phases of Morrison, S.J., Perez, S.E., Qiao, Z., Verdi, J.M., Hicks, C., Weinmas-
the cell cycle. J. Cell Biol. 1996, 135: 441–456. ter, G., Anderson, D.J.: Transient Notch activation initiates an
Odelberg, S.J., Kolhoff, A., Keating, M.: Dedifferentiation of mam- irreversible switch from neurogenesis to gliogenesis by neural
malian myotubes induced by msx1. Cell 2000, 103: 1099–1109. crest stem cells. Cell 2000, 101: 499–510.
Sauka–Spengler, T., Bronner-Fraser, M.: A gene regulatory network
10.10 Skeletal muscle and neural cells can be renewed from orchestrates neural crest formation. Nat. Rev. Mol. Cell Biol.
stem cells in adults 2008, 9: 557–568.
Collins, C.A., Partridge T.A.: Self-renewal of the adult skeletal
10.12 Programmed cell death is under genetic control
muscle satellite cell. Cell Cycle 2005, 4: 1338–1341.
Dhawan, J., Rando, T.A.: Stem cells in postnatal myogenesis: Jacobson, M.D., Weil, M., Raff, M.C.: Programmed cell death in
­molecular mechanisms of satellite cell quiescence, activation animal development. Cell 1997, 88: 347–354.
and replenishment. Trends Cell Biol. 2005, 15: 663–673. Raff, M.: Cell suicide for beginners. Nature 1998, 396: 119–122.
Lepper, C., Conway, S.J., Fan, C.M.: Adult satellite cells and embry- Riedl, S.J., Shi, Y: Molecular mechanisms of caspase regulation
onic muscle progenitors have distinct genetic requirements. during apoptosis. Nat. Rev. Mol. Cell Biol. 2004, 5: 897–907.
Nature 2009, 460: 627–631. Sancho, E., Batlle, E., Clevers, H.: Live and let die in the intestinal
Lie, D-C., Colamarino, S.A., Song, H-J., Desire, L., Mira, H., Con- epithelium. Curr. Opin. Cell Biol. 2003, 15: 763–770.
siglio, A., Lein, E.S., Jessberger, S., Lansford, H., Dearier, A.R., Vaux, D.L., Korsmeyer, S.J.: Cell death in development. Cell 1999,
Gage, F.H.: Wnt signalling regulates adult hippocampal neuro- 96: 245–254.
genesis. Nature 2005, 437: 1370–1375.
Ninkovic, J., Gotz, M.: Signaling in adult neurogenesis: from stem 10.13 Nuclei of differentiated cells can support development
cell niche to neuronal networks. Curr. Opin. Neurobiol. 2007, Eggan, K., Baldwin, K., Tackett, M., Osborne, J., Gogos, J., Chess,
17: 338–344. A., Axel, R., Jaenisch, R.: Mice cloned from olfactory sensory
Sacco, A., Doyonnas, R., Kraft, P., Vitorovic, S., Blau, H.M.: Self- neurons. Nature 2004, 428: 44–49.
­renewal and expansion of single transplanted muscle stem Gurdon, J.B.: Nuclear transplantation in eggs and oocytes. J. Cell
cells. Nature 2008, 456: 502–506. Sci. Suppl. 1986, 4: 287–318.
410 Chapter 10 Cell differentiation and stem cells

Gurdon, J.B., Melton, D.A.: Nuclear reprogramming in cells. Sci- Molofsky, A.V., Pardal, R., Morrison, S.J.: Diverse mechanisms
ence 2008, 322: 1811–1815. regulate stem cell self-renewal. Curr. Opin. Cell Biol. 2004, 16:
Humpherys, D., Eggan, K., Akutsu, H., Friedman, A., Hochedlinger, 700–707.
K., Yanagimachi, R., Lander, E.S., Golub, T.R., Jaenisch, R.: Plachta, N., Bibel, M., Tucker, K.L., Barde, Y.A.: Developmental
­Abnormal gene expression in cloned mice derived from embry- potential of defined neural progenitors derived from mouse
onic stem cell and cumulus cell nuclei. Proc. Natl Acad. Sci. USA embryonic stem cells. Development 2004, 131: 5449–5456.
2002, 99: 12889–12894. West, J.A., Daley, G.Q.: In vitro gametogenesis from embryonic
Rhind, S.M., Taylor, J.E., De Sousa, P.A., King, T.J., McGarry, M., stem cells. Curr. Opin. Cell Biol. 2004, 16: 688–692.
Wilmut, I.: Human cloning: can it be made safe? Nat. Rev. Genet. Ying, Q.L., Nichols, J., Chambers, I., Smith, A.: BMP induction of
2003, 4: 855–864. Id proteins suppresses differentiation and sustains embryonic
Wilmut, I., Taylor, J.: Primates join the club. Nature 2007, 450: stem cell self-renewal in collaboration with STAT3. Cell 2003,
485–486. 115: 281–292.

10.14 Patterns of gene activity in differentiated cells can be 10.17 Stem cells could be a key to regenerative medicine
changed by cell fusion Jaenisch, R.: Human cloning—the science and ethics of nuclear
Blau, H.M.: How fixed is the differentiated state? Lessons from transplantation. N. Engl. J. Med. 2004, 351: 2787–2792.
heterokaryons. Trends Genet. 1989, 5: 268–272. McClaren, A.: Ethical and social considerations of stem cell
Blau, H.M., Baltimore, D.: Differentiation requires continuous research. Nature 2001, 414: 129–131.
regulation. J. Cell Biol. 1991, 112: 781–783. Okita, I., Ichisaka, T., Yamanka, S.: Generation of germline-compe-
Blau, H.M., Blakely, B.T.: Plasticity of cell fates: insights from het- tent induced pluripotent stem cells. Nature 2007, 448: 313–317.
erokaryons. Cell Dev. Biol. 1999, 10: 267–272. Pera, M.F., Trounson, A.O.: Human embryonic stem cells: pros-
Pomerantz, J.H., Mukherjee, S., Palermo, A.T., Blau, H.M.: Repro- pects for development. Development 2004, 131: 5515–5525.
gramming to a muscle fate by fusion recapitulates differentia- Pomerantz, J., Blau, H.M.: Nuclear reprogramming: a key to stem
tion. J. Cell Sci. 2009, 122: 1045–1053. cell function in regenerative medicine. Nat. Cell Biol. 2004, 6:
810–816.
10.15 The differentiated state of a cell can change by Rolletschek, A., Wobus, A.M.: Induced human pluripotent stem
transdifferentiation cells: promises and open questions. Biol. Chem. 2009, 390:
Horb, M.E., Shen, C.N., Tosh, D., Slack, J.M.: Experimental conver- 845–849.
sion of liver to pancreas. Curr. Biol. 2003, 13: 105–115. Wurmer, A.E., Palmer, T.D., Gage, F.H.: Cellular interactions in the
Jarriault, S., Shwab, Y., Greenwald, I.A.: Caenorhabitis elegans stem cell niche. Science 2004, 304: 1253–1255.
model for epithelial-neuronal transdifferentiation. Proc. Natl Yechoor, V., Liu, V., Espiritu, C., Paul, A, Oka, K., Kojima, H., Chan,
Acad. Sci. USA 2008, 105: 3790–3795. L.: Neurogenin3 is sufficient for transdetermination of hepatic
Slack, J.M.W.: Metaplasia and transdifferentiation from pure biol- progenitor cells into neo-islets in vivo but not transdifferentia-
ogy to the clinic. Nat. Rev. Mol. Cell Biol. 2007, 8: 369–378. tion of hepatocytes. Dev. Cell 2009, 16: 358–373.
Tsonis, P.A., Madhavan, M., Tancous, E.E., Del Rio-Tsonis, K.:
A newt’s eye view of lens regeneration. Int. J. Dev. Biol. 2004, 10.18 Various approaches can be used to generate
48: 975–980. differentiated cells for cell-replacement therapies
Chen, S., Borowiak, M., Fox, J.L., Maehr, R., Osafune, K., Davidow,
10.16 Embryonic stem cells can proliferate and differentiate L., Lam, K., Peng, L.F., Schreiber, S.L., Rubin, L.L., Melton, D.:
into many cell types in culture A small molecule that directs differentiation of human ESCs
Ho, A.D.: Kinetics and symmetry of divisions of hematopoietic into the pancreatic lineage. Nat. Chem. Biol. 2009, 5: 258–265.
stem cells. Exp. Hematol. 2005, 33: 1–8. Maehr, R., Chen, S., Snitow, M., Ludwig, T., Yagasaki, L., Goland,
Loebel, D.A., Watson, C.M., De Young, R.A., Tam, P.P.: Lineage R., Leibel, R.L., Melton, D.A.: Generation of pluripotent stem
choice and differentiation in mouse embryos and embryonic cells from patients with type 1 diabetes. Proc. Natl Acad. Sci.
stem cells. Dev. Biol. 2003, 264: 1–14. USA 2009, 106: 15768–15773.
McBeath, R., Pirone, D.M., Nelson, C.M., Bhadriraju, K., Chen, C.S.: Zhou, Q., Brown, J., Kanarek, A., Rajagopal, J., Melton, D.A.: In vivo
Cell shape, cytoskeletal tension, and RhoA regulate stem cell reprogramming of adult pancreatic exocrine cells to beta-cells.
lineage commitment. Dev. Cell 2004, 6: 483–495. Nature 2008, 455: 627–632.
 11
Organogenesis
nn The vertebrate limb nn Vertebrate and insect nn Internal organs: tracheal
nn Insect wings and legs eyes system, lungs, kidneys,
blood vessels, heart, and
teeth

Once the basic animal body plan has been laid down, the development of organs as
varied as insect wings and vertebrate eyes begins. One question is whether quite
new developmental mechanisms and principles are brought into play, or whether
­organogenesis depends on the same basic mechanisms as those used in early devel-
opment. Organ development involves very large numbers of genes and, because of
this complexity, general principles can be quite difficult to distinguish. Nevertheless,
many of the mechanisms used in organogenesis, such as positional information, are
similar to those of earlier development, and certain signals are used again and again.

So far, we have concentrated almost entirely on the aspects of development involved in

laying down the basic body plan in various organisms, and on early morphogenesis and
cell differentiation. We now turn to the development of specific organs and structures—
organogenesis—which is a crucial phase of development that will ultimately lead to the
embryo at last becoming a fully functioning organism, capable of independent survival.
The development of certain organs has been studied in great detail and they provide
excellent models for looking at developmental processes, such as pattern formation,
the specification of positional information, induction, change in form, and cellular dif-
ferentiation. In this chapter, we first consider the development of some of these classic
model systems—the vertebrate limb, the legs and wings of Drosophila, and vertebrate
and insect eyes. We then consider some internal organs. The structure of many inter-
nal organs is based on tubules—for example, the lungs, the vascular system, and the
kidneys—and we look at the different ways in which tubules can be formed. We then
look briefly at the initial patterning and regionalization of the heart, which develops
from a tube of mesoderm, and finally at the induction and patterning of teeth. The
development of other major organs, such as the gut, liver, and pancreas, involves
no new principles and we will not consider them here. The liver is interesting as an
example of a mammalian organ that can regenerate, and some aspects of liver growth
and regeneration are discussed in Chapters 13 and 14.
The cellular mechanisms involved in organogenesis are essentially similar to those
encountered in earlier stages of development; they are merely employed in different
spatial and temporal patterns. Many of the genes and signaling molecules involved will
be familiar from earlier chapters. We shall, however, see that the mechanisms involved
are much more complex and that, while one can identify some general principles, such
412 Chapter 11 Organogenesis

as the use of positional information, there is much additional detail for which there are
no unifying principles—it just works that way in that particular o­ rgan.

The vertebrate limb

The vertebrate embryonic limb is a particularly good system in which to study pat-
tern formation, as the basic pattern is initially quite simple, and limbs have distinct
elements of pattern along all three axes that can be easily recognized. The limb is
also a good model for studying cellular interactions within a structure containing a
wing bud large number of cells, and for elucidating the role of intercellular signaling in devel-
opment. Mice are used to study some aspects of limb development, mainly through
apical
ectodermal spontaneous mutants and artificial gene knock-outs, but the basic principles of limb
somites ridge
patterning have been most extensively investigated in chick embryos, because here
the developing limbs themselves are easily accessible for microsurgical manipulation.
leg bud A window is made in the egg shell and the limb buds manipulated while the embryo
remains in the egg (see Fig. 3.29). After surgery, the window can be re-sealed with
adhesive tape and the embryo allowed to continue development, so that the effects of
the procedures on limb development can be assessed.
In chick embryos, the first signs of limb development can be seen around the
Fig. 11.1 The limb buds of the chick third day after the egg is laid, when the structures of the main body axis are already
embryo. Limb buds appear on the flanks of well-established. Small protrusions—the limb buds—arise from the body wall of the
the embryo on the third day of incubation embryo (Fig. 11.1). By 10 days, the main features of the limbs are well-developed.
(only the limb buds on the right side ­Figure 11.2 shows the pattern of the elements that make up the skeleton; they are first
are shown here). They are composed of formed as cartilage and are later replaced by bone (how bones grow is described in
mesoderm, with an outer covering of
Section 13.8). Structures in the surface epithelium, such as feather buds, can also be
ectoderm. Along the tip of each runs a
seen. The limb at this stage also has muscles and tendons. The limb has three devel-
thickened ridge of ectoderm, the apical
ectodermal ridge.
opmental axes: the proximo-distal axis runs from the base of the limb to the tip; the
antero-posterior axis runs parallel with the body axis (in the human hand it goes
from the thumb (anterior) to the little finger (posterior), and in the chick wing from
digit 2 to digit 4); the dorso-ventral axis is the third axis—in the human hand it runs
from the back of the hand to the palm.

11.1 The vertebrate limb develops from a limb bud

The early limb bud has two major components—a core of loose mesenchymal cells
derived from the lateral plate mesoderm, and an outer layer of ectodermal epithelial

Anterior
Fig. 11.2 The embryonic chick wing. The
photograph shows a stained whole mount Dorsal
of the wing of a chick embryo 10 days after Proximal Distal
the egg has been laid. By this time, the main Ventral
cartilaginous elements (e.g. humerus, radius,
and ulna) have been laid down. They later Posterior

become ossified to form bone. The muscles

humerus
and tendons are also well developed at carpals
this stage but cannot be seen in this type radius
of preparation. Feather buds can be seen digit 2
particularly along the posterior margin of
the limb. The three developmental axes of digit 3
the limb are proximo-distal, antero-posterior,
and dorso-ventral, as shown in the top panel.
Note that the chick wing has only three digit 4
ulna
digits, which have been called 2, 3, and 4.
Scale bar = 1 mm.
 The vertebrate limb 413

cells (Fig. 11.3). The skeletal elements and connective tissues of the limb develop from Dorsal presumptive muscle zone of
these mesenchyme cells, but the mesodermal cells that give rise to the limb muscles proliferating
undifferentiated
have a separate lineage. They derive from the somites and migrate into the limb bud cells
(see Section 5.9). The limb vasculature is, at least in part, also formed by cells derived
condensing cartilage
from the somites. At the very tip of the limb bud is a thickening in the ectoderm—the of humerus
apical ectodermal ridge or apical ridge, which runs along the dorso-ventral boundary
(Fig. 11.4). Directly beneath the apical ectodermal ridge lies a region that is composed
of rapidly proliferating undifferentiated mesenchymal cells. It is only when cells leave
this zone that they begin to differentiate. As the bud grows, the cells start to differenti- ectoderm
ate and cartilaginous structures begin to appear in the mesenchyme. The proximal part Ventral apical ectodermal ridge
of the limb—that is, the part nearest to the body—is the first to differentiate, and dif-
ferentiation proceeds distally as the limb extends. Of the structures found in the devel- Fig. 11.3 Cross-section through a
oping limb, the patterning of the cartilage has been the best studied, as cartilage can developing limb bud. The thickened apical
be stained and seen easily in whole mounts of the embryonic limb (see Fig. 11.2). The ectodermal ridge is at the tip. Beneath
disposition of muscles and tendons is more intricate, and although this can be studied the ridge is a region of proliferating
in whole mounts by staining with antibodies for tissue-specific proteins, histological undifferentiated cells. Proximal to this region,
examination of serial sections through the limb may be needed. mesenchyme cells condense and differentiate
The first clear sign of cartilage differentiation is the increased packing of groups into cartilage. Presumptive muscle cells
of cells, a process known as condensation. The cartilage elements are laid down in migrate into the limb from the adjacent
somites and form dorsal and ventral muscle
a proximo-distal sequence in the chick wing—the humerus, radius, and ulna, the
masses. Scale bar = 0.1 mm.
wrist elements (carpals), and then three easily distinguishable digits, 2, 3, and 4 (see
Fig. 11.2). Figure 11.5 compares this sequence in the development of a chick wing
with the similar sequence in the development of a mouse forelimb.
The chick limb bud at 3 days is about 1 mm wide by 1 mm long, but by 10 days
it has grown around 10-fold, mostly in length. The basic pattern has been laid down
well before then. But even at 10 days, the limb is still small compared with the size of
the limb when the chick hatches. The apical ridge disappears as soon as all the basic
elements of the limb are in place, and growth occupies most of the subsequent devel-
opment of the limb, both before and after hatching. During the later growth phase,
the cartilaginous elements are largely replaced by bone. Nerves only enter the limb
Ventral Dorsal
after the cartilage has been laid down, at around 4.5 days, and we shall discuss this in
Chapter 12. The problem of limb patterning is to understand how the basic pattern of
apical
cartilage, muscle, and tendons is formed in the right places, and how they make the ectodermal
ridge
right connections with each other. The first consideration, however, is how limb buds
develop at the appropriate positions on the body.

11.2 Genes expressed in the lateral plate mesoderm are involved in

specifying the position and type of limb

The forelimbs and hindlimbs of vertebrates arise at precise positions along the antero-
posterior axis of the body. Transplantation experiments have shown that the lateral plate
mesoderm that gives rise to the limb-bud mesenchyme becomes determined to form
limbs in these exact positions, long before limb buds are visible. Like the pre-somitic
mesoderm discussed in Chapter 5, the lateral plate mesoderm of the trunk becomes
regionalized by Hox gene expression along the antero-posterior axis, and so it is pos-
sible that Hox gene coding could determine prospective forelimb and hindlimb regions,
thus specifying the position and type of a limb bud. This is analogous to the way
in which the different thoracic segments and their appendages are specified in insect
development (see Chapter 2). The homeodomain transcription factor Pitx1 (which is
related to the Otx homeodomain proteins) is expressed in the hindlimb region and plays
a key role in determining the differences between hindlimb and forelimb. Fig. 11.4 Scanning electron micrograph of
Proteins of the fibroblast growth factor (FGF) family appear to be key signaling mol- a chick limb bud at 4.5 days after laying,
ecules in initiating both forelimb and hindlimb development in vertebrates. Localized showing the apical ectodermal ridge.
application of FGF-4 to the flank of a chick embryo, between the wing and leg buds, Scale bar = 0.1 mm.
414 Chapter 11 Organogenesis

Chick wing Mouse forelimb

4 days 10.5 days

4.5 days 11 days

5.5 days 12 days

6.5 days 13.5 days

8 days 14.5 days

Fig. 11.5 The development of the chick

wing and mouse forelimb is similar. The
skeletal elements are laid down, as cartilage,
in a proximo-distal sequence as the limb
bud grows outward. The cartilage of the
humerus is laid down first, followed by the
radius and ulna, wrist elements, and digits.
Scale bar = 1 mm.

induces an ectopic limb bud. Wing buds develop from application of FGF to the ante-
rior flank, whereas leg buds develop from application to the posterior flank (Fig. 11.6).
The T-box transcription factors, Tbx5 and Tbx4, which are related to the mesoder-
mal marker Brachyury (see Section 4.18), are expressed in forelimbs and hindlimbs,
respectively, and are required for limb initiation via FGF. In humans, mutations in the
related T-box gene Tbx5 are associated with the Holt–Oram syndrome, which is most
commonly characterized by abnormalities of the thumbs and heart defects.
FGF is involved in establishing and maintaining the two main organizing regions of
a limb. These are the apical ectodermal ridge, which is essential for limb outgrowth
and for correct patterning along the proximo-distal axis of the limb, and a mesoder-
mal polarizing region, which is situated on the posterior side of the limb bud and
is crucial for determining pattern along the antero-posterior limb axis, as discussed
later in the chapter. Cells of the polarizing region express the signaling protein Sonic
hedgehog (Shh), and staining for Shh mRNA has been used to reveal the polarizing
regions in the normal and the ectopic limb buds in the embryo shown in Fig. 11.6. The
 The vertebrate limb 415

gene Fgf10 is expressed in limb-forming regions, and in the mouse embryo, in which,
Anterior
unlike chick, transgenic knock-outs are possible, knock-out of either Fgf10 or the gene
encoding its receptor results in embryos lacking limb buds.
In chick embryos, Wnt proteins produced and secreted by the mesoderm play a key wing
role in determining where FGFs are expressed and maintained. A Hox gene coding in
the mesoderm could therefore determine where limbs develop by specifying where
Wnts, and thus FGFs, are expressed, and so where an apical ridge and a polarizing
region develop. Evidence for the combined actions of Hox genes in determining the
position of the forelimb bud along the antero-posterior axis comes from knock-out
Sonic
mice lacking Hoxb5 expression, in which the forelimbs develop at a more anterior hedgehog
induced expression
level. The different patterns of Hox gene expression at different positions along the leg
antero-posterior axis of the body could also be involved in specifying whether the
limb will be a forelimb or a hindlimb.

11.3 The apical ectodermal ridge is required for limb outgrowth

We shall next look at how the limb buds start to grow and develop. One crucial signal-
ing region in the limb bud is the apical ectodermal ridge, whose presence is essential
for limbs to grow and develop. The apical ectodermal ridge consists of closely packed
leg
columnar epithelial cells, which are directly connected to each other by gap junc-
tions. Their tight packing gives the ridge a mechanical strength that probably is what Posterior
keeps the limb flattened along its dorso-ventral axis. The initial outgrowth of the
limb bud depends on proliferating undifferentiated mesenchymal cells lying beneath Fig. 11.6 Result of local application of
the apical ectodermal ridge. Perhaps surprisingly, the establishment of the early bud FGF-4 to the flank of a chick embryo. FGF-4
does not reflect a local increase in cell division in the limb-bud region, but rather a is applied to a site close to the leg bud,
decrease from a previously high rate of cell proliferation along the rest of the flank of where it induces outgrowth of an additional
the embryo. leg bud. The dark regions indicate Sonic
The experiment that showed the importance of the apical ectodermal ridge involved hedgehog transcripts. Sonic hedgehog is
removing the ridge from a chick limb bud by microsurgery. This resulted in a sig- expressed at the anterior of the additional
leg bud and the leg that subsequently
nificant reduction in outgrowth and the limb was anatomically truncated, with distal
develops has an opposite polarity to that of
parts missing. The proximo-distal level at which the limb is truncated depends on the
the normal limbs. It is not known why Sonic
time at which the ridge is removed (Fig. 11.7). The earlier the ridge is removed, the hedgehog is expressed in this way in the
greater the effect; removal at a late stage only results in loss of the ends of the digits, additional leg bud. Scale bar = 1 mm.
whereas when the ridge is removed at an early stage most of the limb is missing. Fol- Photograph courtesy of J-C. Izpisúa-Belmonte,
lowing apical-ridge removal, proliferation of cells at the tip of the limb bud is greatly from Cohn, M.J., et al.: 1995.
reduced and there is also cell death in that region. The importance of signaling by the

Fig. 11.7 The apical ectodermal

At 3 days At 3.5 days At 4 days
ridge is required for proximo-distal
development. Limbs develop in a proximo-
distal sequence. Removal of the ridge from
Removal a developing limb bud leads to truncation of
of apical the limb; the later the ridge is removed, the
ectodermal more complete the resulting limb.
ridge

Normal limb

Resulting
wing
416 Chapter 11 Organogenesis

Apical ectodermal ridge removed, followed by apical ectodermal ridge in promoting outgrowth has also been shown by grafting an
implantation of beads soaked in FGF-4 isolated ridge to the dorsal surface of an early chick wing bud. The result is an ectopic
outgrowth, which may even develop cartilaginous elements and digits.
humerus A key signal from the ridge is provided by FGFs. The gene for FGF-8 is expressed
throughout the ridge and genes for FGF-4 and two other FGFs are expressed in the
digit 2 posterior region. Experiments in chick wing buds showed that FGF-8 can act as a
radius
functional substitute for an apical ridge. If the ridge is removed and beads that release
the growth factor are grafted into the tip of the wing bud in its place, more-or-less
ulna
digits 3/4 normal outgrowth of the limb continues. If sufficient FGF-8 is provided to the out-
Limb develops almost normally growing cells, a fairly normal limb develops. Similar results are obtained with FGF-4
(Fig. 11.8). Experiments in transgenic mice in which Fgf genes are knocked-out spe-
Fig. 11.8 The growth factor FGF-4 can cifically in the apical ectodermal ridge, also show that FGF signaling is required for
substitute for the apical ectodermal limb-bud outgrowth and normal development. It is tricky to delete FGF signaling com-
ridge. After the ridge is removed, grafting pletely using transgenic approaches because four different Fgf genes are expressed in
heparin beads that release FGF-4 into the the apical ectodermal ridge. Complete limb truncations are produced only when three
limb bud results in almost normal limb of these Fgf genes are knocked out. All this evidence together confirmed that apical-
development. ridge signaling is essential for normal vertebrate limb development.

11.4 Patterning of the limb bud involves positional information

We have just seen how the apical ridge controls the outgrowth of the limb, but how
are the skeletal and other elements of the limb laid down in the correct positions?
Some aspects of vertebrate limb development fit very well with a model of pattern
formation based on positional information. The developing chick limb bud behaves
as if its future development is determined by the positions of its cells with respect to
the main limb axes while the cells are in the undifferentiated region at the tip of the
limb bud (Fig. 11.9).
Central to the idea of positional information is the distinction between positional
specification and interpretation. Cells acquire positional information first and then
interpret these positional values according to their developmental history. It is the dif-
ference in developmental history that makes wings and legs different, for positional
information is signaled in wing and leg buds in the same way. An attractive hypoth-
esis for limb patterning is that a single three-dimensional positional field controls
the development of the cells that give rise to all the mesodermal limb elements—
cartilage, muscle, and tendons. The specification of the three axes—proximo-distal,
antero-­posterior, and dorso-ventral—is, as we shall see, linked by molecular signals
emanating from different parts of the bud.

11.5 How position along the proximo-distal axis of the limb bud is specified is
still a matter of debate

Fig. 11.9 Cells acquire positional values We will start by discussing patterning along the limb’s all-important proximo-distal
along the antero-posterior and proximo- axis, which runs from the shoulder to the tips of the digits. The cartilage elements
distal axes. In the early bud there are two
signaling regions—the polarizing region at
the posterior margin (red), and the apical Anterior Anterior
cartilage of
ectodermal ridge (blue). The polarizing undifferentiated developing humerus
proliferating cells humerus
region specifies position along the antero- radius
posterior axis. How position is specified Proximal Distal
along the proximo-distal axis is still under apical digit 2
debate (see Section 11.5). Cartilage begins to ectodermal digit 3
ridge
differentiate first in the most proximal region, digit 4
and the cartilaginous elements are laid down polarizing
region
in a proximo-distal sequence (carpals not ulna
Posterior Posterior Proximal Distal
shown).
 The vertebrate limb 417

are laid down in a proximal to distal sequence, with the proximal elements begin-
ning to differentiate first. The mechanisms that pattern the proximo-distal axis are,
however, still a matter of some debate, with several different hypotheses currently
being proposed. For simplicity, in this section we will discuss mainly the longest-
standing model, which derives from work on chick embryo limbs and proposes that
proximo-distal patterning is specified by the length of time cells spend in the zone of
undifferentiated cells at the tip of the limb. This region has therefore been called the
progress zone, and we shall refer to this model as the ‘timing model.’ An alternative
mechanism for proximo-distal patterning has been proposed more recently, based on
newer experiments in chick and mouse, and we shall refer to this as the ‘two-signal’
model. There is evidence in support of both models, but neither is entirely consistent
with all the known evidence, so it is clear that there is more to be learned about how
the limb is patterned in the proximo-distal direction.
As the limb bud grows, cells are continually leaving the zone of undifferentiated
cells at the distal end. Because the limb bud extends from proximal to distal, the
earliest cells leaving the zone develop into the humerus and those leaving last form
the tips of the digits. The timing model proposes that, if cells could measure the time
they spend in the undifferentiated zone, by counting cell divisions, for example, this
would give them their positional value along the proximo-distal axis (Fig. 11.10, upper
panel). The length of time cells spend in the somite stem-cell region has similarly
been suggested to account for the specification of the position of a somite along the
antero-posterior axis (see Section 5.8). A timing mechanism of this sort for the limb is
consistent with the experimental observation that removal of the apical ridge results
in a distally truncated limb (see Fig. 11.7). according to the timing theory, removal of

Progress-zone model of proximo–distal patterning

progress
zone

Fig. 11.10 A cell’s proximo-distal

AER positional value may depend on the time
it spends in the ‘progress zone’ or be
specified by opposing signal gradients in
the early limb bud. The upper two sets of
Time in panels show the ‘timing’ model. Cells are
progress continually leaving the progress zone. If the
zone
cells could measure how long they spend in
the zone, this could specify their position
along the proximo-distal axis. Cells that leave
the zone early (red) form proximal structures
Proximal Distal whereas cells that leave it last (blue) form
the tips of the digits. The lower panel shows
Two-signal model of proximo–distal patterning the ‘two-signal’ model. Cells in the early
limb bud acquire positional values from the
presence of two opposing signal gradients,
one of retinoic acid from the most proximal
region (gray) specifying proximal (red) and
retinoic acid

one of FGF from the apical ectodermal ridge

FGF

specifying distal (turquoise and later blue).

As the limb bud grows, cells that are out of
range of both signaling sources acquire an
‘intermediate’ positional value (green).
Adapted from Zeller, R., et al.: 2009.
418 Chapter 11 Organogenesis

the apical ridge means that cells in the progress zone will no longer proliferate, and
so the more distal structures cannot form.
Another line of evidence put forward in support of a timing mechanism is that
killing or blocking the proliferation of cells in the progress zone of a chick wing bud
at an early stage (by X-irradiation, for example), results in the absence of proximal
structures, whereas distal ones are present and can be almost normal. Because many
cells in the irradiated progress zone do not divide, fewer cells than usual will leave
the zone during each unit of time. As the bud grows out, the progress zone becomes
repopulated by the proliferation of surviving cells, which would spend a longer time
than usual in the progress zone and therefore acquire a distal positional value and
form normal distal structures, such as digits. According to other models, proximal
structures suffer most because, although the few cells that left soon after irradiation
will be molecularly specified as proximal, there will not be enough of them to form
a cartilage element. Similar arguments could be made to account for the absence
of proximal limb structures in patients born in the late 1950s and early 60s whose
mothers took the drug thalidomide to ease morning sickness. Thalidomide is known
to interfere with the development of blood vessels, and this could have resulted in
extensive cell death throughout the early limb bud. The effects of thalidomide may,
it has been suggested, be explained by their simply eliminating cartilage precursor
cells at a time when proximal regions are forming but distal differentiation has not
yet commenced.
We have already seen that FGFs produced by the apical ridge are key signals in
limb development, but what specific part do they play in proximo-distal pattern-
ing? According to the timing model, the function of FGF signaling is to maintain
the progress zone, and cells might measure time by the length of time they are
exposed to FGF signals. There is, however, also evidence that retinoic acid, which is
synthesized by cells at the base of the limb buds and in the adjoining trunk, might
contribute to specifying proximal identity in the limb. Retinoic acid targets include
genes of the Meis family, which encode homeodomain transcription factors, and
misexpression of Meis genes in the chick wing leads to distal abnormalities or trun-
cations. Retinoic acid signaling and Meis expression are normally restricted to the
proximal region of the limb bud by FGF signaling from the apical ridge, while FGF
signaling is restricted to the distal region of the limb bud. This interplay between
FGF and retinoic acid signaling is not unique to the developing limb. Opposing
gradients of retinoic acid and FGF have been proposed to operate during growth
and antero-posterior patterning of the main body axis and to control differentiation
of the spinal cord (see Section 5.5). For the limb, it has been proposed that these
two signals may specify a complete proximo-distal axis at an early stage, rather
than the axis being constructed progressively over time as proposed by the timing
model, and this newer model is known as the ‘two-signal’ model (see Fig. 11.10,
lower panel).
Recent genetic data from the mouse has been interpreted in terms of the two-signal
model, and is based on limb patterns that result when FGF signaling in the apical
ridge of mouse limb buds is incrementally reduced by genetically deleting Fgf genes
in the apical ridge one by one. The proposal is that proximal and distal components
of the limb pattern are first specified by retinoic acid and FGF signaling, respectively,
and that subsequent outgrowth of the limb, controlled by FGFs produced by the
apical ridge, leads to the emergence of a region of cells outside the range of both
distal and proximal signals. As a consequence, this region adopts an ‘intermediate’
fate—and forms the radius and ulna. It is not clear, however, whether this suggested
mechanism for generating intermediate fates is universally applicable. When the tip
of an old chick wing bud, which will form distal limb components, is grafted on to a
stump of a young wing bud, which will form proximal limb components, subsequent
outgrowth and development generates a wing that lacks radius and ulna. This failure
 The vertebrate limb 419

of such composite chick wing buds to form intermediate structures is difficult to Anterior
reconcile with the model.

11.6 The polarizing region specifies position along the limb’s

antero-posterior axis

The specification of pattern along the antero-posterior axis of the limb bud is bet-
ter understood and the subject of less contention. Patterning along this axis is what
specifies the individual digits and gives them their identities. In this and the next two
­sections we shall look at how a standard set of digits is specified, the signaling mol-
ecules that operate in antero-posterior patterning, and, finally, how each digit is given
its unique identity. The organizing region for the antero-posterior axis is a mesoder-
mal region at the posterior margin of the limb bud, known as the polarizing region
or the zone of polarizing activity (ZPA) (Fig. 11.11). Posterior
The polarizing region of a vertebrate limb bud has organizing properties that are
almost as striking as those of the Spemann organizer in amphibians. When the polar- Fig. 11.11 The polarizing region or zone
izing region from an early chick wing bud is grafted to the anterior margin of another of polarizing activity (ZPA) of a chick limb
early chick wing bud, a wing with a mirror-image pattern develops: instead of the bud is located at the posterior margin
normal pattern of digits—2 3 4—the pattern 4 3 2 2 3 4 develops (Fig. 11.12, center of the bud. The polarizing region can be
panels). The pattern of muscles and tendons in the limb shows similar mirror-image detected by its expression of Sonic hedgehog
changes. A polarizing-region graft can also specify another ulna anteriorly, showing (Shh) RNA (blue stain). Scale bar = 0.1 mm.
that the polarizing region influences antero-posterior pattern throughout the region Photograph courtesy of C. Tabin.
distal to the elbow or the knee.
The additional digits come from the host limb bud and not from the graft, showing
that the grafted polarizing region has altered the developmental fate of the host cells in
the anterior region of the limb bud. The limb bud widens in response to the polarizing
graft, which enables the additional digits to be accommodated. Widening of a limb bud
is associated with an increase in the extent of the apical ectodermal ridge.
One way that the polarizing region could specify position along the antero-posterior
axis is by producing a morphogen that forms a posterior to anterior gradient (Box 11A,
p. 421). The concentration of morphogen could specify the position of cells along the
antero-posterior axis with respect to the polarizing region located at the posterior
margin of the limb. Cells could then interpret their positional values by developing
specific structures at particular threshold concentrations of morphogen. Digit 4, for
example, would develop at a high concentration, digit 3 at a lower one, and digit 2
at an even lower one (see Fig. 11.12, top panels). According to this model, a graft of
an additional polarizing region to the anterior margin would set up a mirror-image
gradient of morphogen, which would result in the 4 3 2 2 3 4 pattern of digits that is
observed (see Fig. 11.12, center panels).
If the action of the polarizing region in specifying the character of a digit is due to
the level of the signal, then when the signal is weakened, the pattern of digits should
be altered in a predictable manner. Grafting small numbers of polarizing region cells
to a limb bud results in the development of only an additional digit 2 (see Fig. 11.12,
bottom panels). An analogous result can be obtained by leaving the polarizing region
graft in place for a short time and then removing it. When left for 15 hours, an addi-
tional digit 2 develops, whereas the graft has to be in place for up to 24 hours for a
digit 3 to develop.
The limb buds of a number of other vertebrates, including mouse, pig, ferret, and
turtles, and even human limb buds, have been shown to have a polarizing region.
When the posterior margin of a limb bud from an embryo of these species is cut out
and grafted to the anterior margin of a chick wing bud, additional digits are produced.
The additional digits induced are chick wing digits, which shows that, although the
polarizing-region signal is conserved among vertebrates, its interpretation depends on
the responding cells.
420 Chapter 11 Organogenesis

Normal limb
Anterior

Decreasing
apical morphogen Proximal
ectodermal concentration 2
ridge Anterior
Limb bud 2
thresholds
3
Distal
3

polarizing 4
region
Posterior 4
Posterior

Additional polarizing region grafted to anterior margin

Anterior

4
3

2 4
3
2 2

3 2
3
4
Posterior 4

Small number of polarizing region cells grafted to anterior margin

Anterior

2 2 2
3

3
4
4

Posterior

Fig. 11.12 The polarizing region can specify pattern along the center panel of each row and the digits in the right-hand panels are
antero-posterior axis. If the polarizing region is the source of a graded color-coded to match. Grafting an additional polarizing region to the
morphogen, the different digits could be specified at different threshold anterior margin of a limb bud (center row of panels) would result in a
concentrations of signal, as shown for a normal limb bud in the top mirror-image gradient of signal, and thus the observed mirror-image
panels panels, with digit 4 developing where morphogen concentration duplication of digits. A small number of polarizing region cells grafted to
is high and digit 2 where it is lowest. Threshold concentrations of the anterior margin of the limb bud produces only a weak signal and so
morphogen that give rise to the different digits are indicated in the only an extra digit 2 is produced (bottom row of panels).

11.7 Sonic hedgehog produced by the polarizing region is likely to be the

primary morphogen patterning the antero-posterior axis of the limb

A persuasive case can be made for the secreted protein Sonic hedgehog (Shh) acting
as a morphogen and a gradient of Shh protein being a key component of the natu-
ral polarizing signal. The Sonic hedgehog gene (Shh) is expressed in the polarizing
region of the limb buds of both chick and mouse embryos and the embryos of many
other vertebrates that have been studied (see Fig. 11.11), and by the use of anti-Shh
 The vertebrate limb 421

a­ ntibodies, a graded distribution of Shh protein has been detected in the posterior
region of the limb bud at a distance from the cells that produce it. Shh protein is
involved in numerous patterning processes elsewhere in vertebrate development; for
example, in somite patterning (see Section 5.9), in the establishment of internal left–
right asymmetry (see Section 4.7), and in the patterning of the neural tube (discussed
in ­Chapter 12). The Drosophila Hedgehog protein is a key signaling molecule in the
patterning of the segments in the Drosophila embryo (see Section 2.24), as well as of
Drosophila legs and wings, as we shall see later in this chapter.
Cultured chick fibroblasts transfected with a retrovirus containing the Shh gene
acquire the properties of a polarizing region; when grafted to the anterior margin of a
wing bud they cause the development of a mirror-image wing. A bead soaked in Shh
protein will give the same result. The bead must be left in place for 16 to 24 hours
for extra digits to be specified, and the pattern of digits depends on the concentration

Box 11A Positional information and morphogen gradients

Pattern formation in development can be specified by positional information, which

Simple diffusion
is in turn specified by a gradient in some property. The basic idea, related to the
French Flag problem (see Fig. 1.25), is that a specialized group of cells at one bound-
ary of the area to be patterned secrete a molecule whose concentration decreases
with distance from the source, thus forming a gradient of information. Any cell along responsive
producing cell cells
the gradient then ‘reads’ the concentration at that particular point, and interprets
this to respond in a manner appropriate to its position by switching on a particular
pattern of gene expression, for example. Molecules that can induce such changes
in cell fate are known as morphogens. Gradients of positional information are used
for various tasks during development: regionalization of the antero-­posterior and
dorso-ventral axes and patterning of segments and imaginal discs in insects; meso- Endocytosis
derm patterning in vertebrates; vertebrate limb patterning; and patterning along the
dorso-ventral axis of the vertebrate neural tube, among others.
It is still unclear how gradients of positional information are formed, particularly
the relative roles of morphogen diffusion and cell–cell interactions, although there
is evidence for gradients of morphogen molecules in many cases. There are several
questions about positional gradients still to be solved. How is positional informa-
tion specified over distances of up to 20 cell diameters? Is it the extracellular or
intracellular concentration of a morphogen that specifies positional identity? Does
endocytosis play a key role? Transport by proteoglycans
Morphogens could travel by simple diffusion through extracellular spaces or, as proteoglycans
described for the Decapentaplegic protein in Drosophila dorso-ventral patterning (see
Section 2.20), a gradient can be formed by the interaction of the morphogen with
­various other proteins, both secreted extracellular proteins and cell-surface recep-
tors. Morphogens may also be transported through cells by endocytic vesicles (see
figure). In other cases, receptor-mediated endocytosis followed rapidly by degrada-
tion has been proposed as a way of producing a sharp gradient in protein activity. It
is most likely that a cell measures a particular morphogen concentration according
Direction of movement
to how many receptors for the morphogen are occupied, and thus by the strength
of the signal that is transmitted from the receptors. Despite years of work on the
topic, however, it is still not clear whether mechanisms based on secreted diffusible
morphogens will be sufficiently reliable for pattern formation. Moreover, the concentration of a diffusible morphogen in the extracel-
lular space is continually changing over time. An alternative to simple diffusion would be that positional information is specified by
a mechanism involving the generation of a gradient of activity based on short-range interactions between cells, such as occur in the
generation of planar cell polarity (see Box 8C, p. 310).
422 Chapter 11 Organogenesis

of Shh protein in the bead. A higher concentration is required for a digit 4 to develop
than for a digit 2, for example. It has also been shown that Shh has a direct effect on
mesenchymal cell proliferation by controlling expression of genes encoding cell-cycle
regulators, thus revealing a mechanism that contributes to the widening of the limb
bud following a polarizing graft.
The signaling molecule retinoic acid (see Box 5D, p. 192–193) was the first defined
chemical that was found to mimic the signaling of the polarizing region of the chick
wing and can induce mirror-image duplications of the digits when applied to the ­anterior
margin of a chick wing bud. But it was later shown that retinoic acid induces Shh
expression in anterior tissue and this explains the digit duplications. We have already
discussed the evidence that retinoic acid signaling may pattern the proximal part of the
limb but how this is related to its ability to induce Shh expression is not clear.
Mutations in mice and other animals, including humans, have been discovered that
cause preaxial polydactyly. ‘Polydactyly’ denotes limbs with extra digits and ‘pre-
axial’ that the extra digits form anteriorly, characteristically forming an extra ‘thumb.’
These mutations affect the polarizing region and provide more evidence for the cru-
cial role of Shh in specifying the pattern of digits (Box 11B, p. 423).
The next question is how Shh exerts its effects on limb development. The intrac-
ellular signaling pathway that is activated in response to Shh involves the Gli tran-
scription factors Gli1, Gli2, and Gli3, which are homologous to Ci in the Hedgehog
signaling pathway in Drosophila (see Fig. 2.40). The vertebrate Shh signaling pathway
is shown in Box 11C, pp. 424–425. In the presence of Shh signaling, the Gli proteins
function as transcriptional activators, while in its absence, Gli2 and Gli3 are processed
to short forms and Gli3 functions as a transcriptional repressor. Analysis of mice
mutant for the Gli genes has shown that Gli3 is the key regulator of limb patterning,
and several mutations in human Gli3 have also been found to be associated with
limb abnormalities. Gli3 knock-out mice have polydactyly and the digits are all of the
Gli3 same, unpatterned, type. This indicates that in normal circumstances the repressor
form of Gli3 is ­preventing digit formation in the anterior part of the limb bud, where
there is no Shh protein, and that in the posterior part of a normal limb bud, a Shh
protein gradient will produce different ratios of Gli3 activator to Gli3 repressor along
the antero-posterior axis and in this way will determine digit pattern.
In the limb buds of Shh knock-out mice, in contrast, high levels of repressor Gli3
are produced throughout the bud and prevent digit formation altogether. But when
Gli3 is knocked out as well as Shh, the limbs are similar to those of the Gli3 knock-
out mouse embryos. Taken together, these experiments show that the main function
of Shh signaling in the limb bud is to suppress repressor Gli3 function in the poste-
rior part of the bud. Some mouse mutants with reduced levels of repressor Gli and
polydactyly do not have defects in the genes encoding Shh or components of the
signaling pathway, but instead have mutations in genes essential for the formation of
a cell structure called the primary cilium. This immotile microtubule-based structure
is essential for Shh signaling and the processing of Gli3, as described in Box 11C,
pp. 424–425. However, the fact that a limb with many digits, but no antero-posterior
pattern, can develop when both Shh and Gli3 function are lacking implies the exis-
tence of some self-organizing mechanism in the limb bud that is capable of setting
up a regular set of digit-like elements but cannot give them distinct identities. Such a
mechanism could, for example, be based on the principle of reaction–diffusion (see
Box 11D, p. 432), which has been proposed to be responsible for patterns such as the
stripes on angelfish.

11.8 Transcription factors might specify digit identity

Several transcription factors have been implicated in digit formation and in conferring
digit identity, including Hoxd13 and Hoxa13 (as we shall see later in the chapter),
 The vertebrate limb 423

Box 11B Too many fingers: mutations that affect antero-posterior patterning can cause polydactyly

Mutations have been discovered in mice that affect the function embryos in which the Shh gene has been knocked-out lack dis-
of the polarizing region and cause preaxial polydactyly. ‘Poly- tal structures, although proximal structures are still present. At
dactyly’ denotes limbs with extra digits and ‘preaxial’ the fact best, one very small digit develops in the hindlimb. Shh knock-out
that the extra digits form anteriorly. These mouse mutants, such mouse embryos also have other defects, including craniofacial
as the Sasquatch mutant, provide more evidence for the crucial defects such as holoprosencephaly (see Box 1F, p. 28) and pat-
role of Shh in specifying the pattern of digits, as it has been terning defects in the neural tube. Complete deletion of the ZRS
shown that formation of the extra anterior digit in Sasquatch is in mouse embryos also leads to loss of posterior Shh expression
due to additional anterior expression of the Shh gene in the limb in the limb buds and results in limb truncations resembling those
buds. The Sasquatch mutation maps to in Shh knock-out embryos. In embryos
a long-range cis-­regulatory region in with the ZRS deletion, however, no
the mouse genome, which has become other structures are affected because
known as the ZRS (zone of polarizing Shh function is deleted just in the limb
activity regulatory sequence). The ZRS bud. Loss of distal limb structures is also
governs limb-specific Shh expression seen in the very rare inherited human
even though it is located 1 Mb away condition, acheiropodia, in which struc-
from the Shh gene. This example of tures below the elbow or knee fail to
long-range control is similar to the devel- develop but no other defects are pres-
opmental control of b-globin expression ent. Acheiropodia has been shown to
in red blood cells by the locus control be due to a chromosomal deletion near
region (discussed in Section 10.6). the ZRS, suggesting that there may be
Mutations in the ZRS have also been a cis-regulatory region in this position in
found in polydactylous cats, with extra the human genome that regulates SHH
anterior digits, including the celebrated expression in the polarizing region.
tribe of cats that inhabit the novelist Just because a morphogen gradient
Ernest Hemingway’s old home in Key model is consistent with many experi-
West, Florida (see photo). The tribe is mental results in the chick wing, how-
descended from a polydactylous cat ever, this is not sufficient to prove that
given to Hemingway by a sea captain, it is the mechanism that specifies all of
and all the polydactylous individuals have the same single-nu- the digits in the limbs of other vertebrates. In the mouse limb,
cleotide substitution in the ZRS. There are also inherited human for example, cell-lineage analysis of cells that express Shh show
conditions based on mutations and/or duplications of the ZRS. that the two most posterior digits arise from cells of the polar-
Human patients with preaxial polydactyly also have mutations izing region itself, and it is suggested that the identity of these
in the ZRS upstream of the human SHH gene, and several other two digits is determined by the length of time these cells spend
polydactylous conditions in human patients are associated with in the polarizing region.
microduplications within this regulatory sequence.
In contrast to the extra digits that develop from limb buds Photograph from Lettice, L.A., et al.: 2008.
with additional anterior Shh gene expression, limbs of mouse

and Sall and Tbx genes. Each of the four toes of the chick leg has a different num-
ber of bones (the phalanges) and so the different digits can be easily distinguished.
This makes it an excellent system for studying how digit identity is specified. In the
chick leg, expression of Tbx2 and Tbx3 in the most posterior interdigital region may
contribute to specifying the identity of digit 4 (the most posterior toe), while expres-
sion of Tbx3 alone in the next interdigital region may specify the identity of digit 3.
This conclusion comes from experiments in which Tbx2 and Tbx3 were misexpressed
in chick leg buds and changed one digit into the other. Limb defects are seen in
human syndromes associated with mutations in TBX3 (mammary ulnar syndrome)
and SALL1 (Townes–Brockes syndrome). The Tbx and Sall genes are homologs of the
Omb and Spalt genes, respectively, which are involved in antero-posterior patterning
of the Drosophila wing (discussed later in this chapter).
424 Chapter 11 Organogenesis

Box 11C Sonic hedgehog signaling and the primary cilium

The intercellular signaling protein Sonic hedgehog is a verte- The repressor Gli3 and Gli2 then enter the nucleus and prevent
brate homolog of the Hedgehog protein that plays a major role the transcription of Shh target genes. When Shh is present, it
in the patterning of the Drosophila embryo (see Chapter 2). Mice binds to Patched and activates Smoothened. Hedgehog ligand
and humans have three genes for hedgehog proteins—Sonic bound to Patched is internalized in membrane vesicles and plays
hedgehog (Shh), Indian hedgehog (Ihh), and Desert hedgehog no further part in signaling. Membrane vesicles containing the
(Dhh); zebrafish have a fourth hedgehog gene, Tiggywinkle, as activated Smoothened are targeted to the cilium membrane,
a result of genome duplication. The signaling pathway stimu- fuse with it, and deliver Smoothened into the membrane. At the
lated by Shh and the other vertebrate hedgehog proteins (see tip of the cilium, Smoothened antagonizes the inhibition of Gli
figure) has many similarities to the Hedgehog signaling pathway proteins by their binding proteins and prevents phosphorylation
in Drosophila (see Fig. 2.40). There are vertebrate homologs of of Gli3 and Gli2. The activated Gli proteins are then transported
the membrane proteins Patched and Smoothened, which sense along microtubules down the cilium to the cell body, where they
the presence or absence of a hedgehog signal, and the verte- enter the nucleus and activate transcription of target genes.
brate counterparts of the Drosophila transcription factor Cubitus The first clues to the importance of the primary cilium in Shh
interruptus (Ci) are the Gli transcription factors Gli1, Gli2, and signaling came from a genetic screen in mice for embryonic pat-
Gli3. The name Gli comes from the initial identification of one of terning mutants following mutagenesis with ethylnitrosourea
these genes as being amplified and highly expressed in a human (ENU) (see Fig. 3.30). Two mutants were identified with defects
glioma. Like Ci, Gli3 and Gli2 can be processed so that they act as in Shh signaling in the neural tube, one of which survived long
transcriptional activators in the presence of a hedgehog signal enough for limb polydactyly to be observed. Most unexpectedly,
and a transcriptional repressor in its absence. Proteins homol- a search against the databases revealed that the genes mutated
ogous with Cos2 and Su(fu) in Drosophila (see Fig. 2.40) form in these embryos encoded proteins similar to those that trans-
complexes with the Gli proteins and keep them in an inactive port material within flagella (a longer version of cilia), and which
state. were known to be essential for the growth and maintenance of
A major difference between a vertebrate hedgehog signaling flagella in the unicellular alga Chlamydomonas. Subsequently,
pathway and Hedgehog signaling in Drosophila, is that in ver- mutations in genes encoding subunits of intraflagellar dynein
tebrates Gli processing takes place at the tip of a cell structure and kinesin motor proteins were also found to result in defective
called the primary cilium. Most vertebrate cells have a primary Shh signaling, providing confirmation of the essential role of cilia
cilium, which, like all cilia, is based on microtubules and extends in vertebrate hedgehog signaling.
from a basal body derived from a centrosome. Unlike most other All known mutations that prevent formation of the primary cil-
types of cilia, the primary cilium is non-motile. ium cause patterning defects in the limb and the neural tube. In
In the absence of a hedgehog ligand, such as Shh, Patched the absence of the primary cilium, both Gli repressor and Gli acti-
represses Smoothened activity and moves into the cilium vator functions fail. Because of the importance of Gli ­repressor
­membrane whereas Smoothened does not. The Gli3 and Gli2 pro- function in antero-posterior limb patterning, this leads to poly-
teins are phosphorylated, which targets them for cleavage when dactyly, whereas the neural tube is dorsalized, which is due to
they encounter the proteasome at the base of the cilium. Cleav- the importance of Gli activator ­function in its dorso-ventral pat-
age produces short forms that act as transcriptional repressors. terning (see Section 12.6).

Sal1, Tbx2, and Hoxd13 are among the genes recently identified by a genome-wide
analysis of gene expression in mouse limb buds that looked for genes regulated by
Gli3 repressor. A Gli3 repressor tagged with a Flag label (which can be detected by
specific antibodies) was expressed in the developing limbs of transgenic mice and
the genomic sites occupied by the tagged Gli3 repressor were identified by chromatin
immunoprecipitation (see Section 3.11). These results, together with transcriptional
profiling by DNA microarrays of anterior and posterior regions of normal developing
mouse limbs and limbs from mouse mutants with defective Shh signaling, identi-
fied more than 200 genes that are putative targets for the Gli3 repressor during limb
development. From these data, a gene regulatory network mediated by Gli3 in the
developing limb is being constructed (for other examples of gene regulatory networks
see Box 4E, p. 162, and Section 6.12). So far, this network for the mouse limb has
linked up about 10 genes, but many more will eventually be added. This is just the
 The vertebrate limb 425

Shh absent Shh present

Smoothened

modifying
factor
Su(fu)
Patched 1
P
Gli1 Gli2 Gli3

primary
ciliium Shh

Patched 1 P

Cytoplasm
Cytoplasm P

P
Smoothened Gli3
repressor cleavage in
proteasome

Gli2 Nucleus
repress activate
Nucleus target repressor
target
genes genes

e.g. Hoxd13 e.g. Gli1, Ptc1

Defects in the formation of cilia and basal bodies have been disease, and hearing loss. Mutations in any of 11 genes can
linked to several dozen human congenital disorders, collectively cause BBS, and some of these genes encode proteins that are
known as ciliopathies. For example, patients with Bardet–Beidl part of the basal body and/or the centrosome, thus providing an
syndrome (BBS), in which the primary cila are affected, have a explanation for defects in ciliogenesis.
range of defects that can include polydactyly, polycystic kidney

beginning of unraveling the molecular basis for digit patterning downstream of Shh
signaling.
Another Gli-regulated gene identified by microarray profiling was Bmp2, which
encodes a secreted signaling protein of the TGF-b family and is expressed in the pos-
terior of the early limb bud. BMPs are also expressed later in limb development in
interdigital regions, the regions between the forming digits. Experiments on chick
embryo leg indicate that BMP signaling in interdigital tissue might regulate digit iden-
tity even at late stages in digit formation. Removal of interdigital tissue or the implan-
tation into interdigital regions of beads coated with the BMP antagonist Noggin (see
Section 4.17), induce changes in the number of phalanges in the adjacent digit that
are consistent with a more anterior identity. Each forming digit is characterized by a
specific level of activity of Smad transcription factors, which provide the read-out of
the levels of BMP signaling (see Fig. 4.33) and would activate genes involved in the
subsequent stages of digit patterning.
426 Chapter 11 Organogenesis

11.9 The dorso-ventral axis of the limb is controlled by the ectoderm

The chick wing has a well-defined pattern along the dorso-ventral axis: large feath-
ers are only present on the dorsal surface, and muscles and tendons have a complex
dorso-ventral organization. Dorsal and ventral compartments of cell-lineage restric-
tion (see Section 2.23 for an explanation of compartments) are specified in the trunk
ectoderm of early chick embryos, including the ectoderm overlying the presumptive
limb-bud regions, before any buds are visible; the apical ridge develops at the bound-
ary between the compartments. Similar dorso-ventral compartments in the ectoderm
have also been found in mouse limb buds.
The development of pattern along the dorso-ventral axis in the mesoderm has been
studied by recombining ectoderm taken from left limb buds with mesoderm from right
limb buds in such a way that the dorso-ventral axis of the ectoderm is reversed with
respect to the underlying mesoderm, but the antero-posterior axis is unchanged. The
right and left wing buds are removed from a chick embryo and, after treatment with cold
trypsin, the ectoderm can be pulled off from the mesodermal core just like taking off a
glove. Ectoderm from a left bud is then recombined with mesoderm from a right bud so
that the dorso-ventral axis of the ectoderm runs the opposite way to that of the meso-
derm—the equivalent of putting a left-hand glove on the right hand (Fig. 11.13). The
reconstituted limb buds are grafted to the flank of a host embryo and allowed to develop.
In general, the proximal regions of the limbs that develop have dorso-ventral polarity cor-
responding to that of the mesoderm, whereas the distal regions, in particular the ‘hands’,
have a reversed dorso-ventral axis, with the pattern of muscles and tendons reversed and
corresponding to the dorso-ventral polarity of the ectoderm. The ectoderm covering the
sides of the limb bud can therefore specify dorso-ventral pattern in the limb.
Genes controlling the dorso-ventral axis in vertebrate limbs have been identified
from mutations in mice, as a dorso-ventral pattern is easily visible in mouse limbs,
where the ventral surface of the paw normally has no fur, whereas the dorsal surface
does. The dorsal surfaces of the digits also have claws (nails in humans). Mutations
that inactivate the gene Wnt-7a, for example, result in limbs in which many of the
dorsal tissues adopt ventral fates to give a double ventral limb, the two halves being
mirror images. The Wnt-7a gene is expressed in the dorsal ectoderm (Fig. 11.14),
which suggests that the ventral pattern may be the ground state and is modified
dorsally by the dorsal ectoderm, with the secreted Wnt-7a protein diffusing into, and
playing a key role in, patterning the dorsal mesoderm. The ventral ectoderm is charac-
terized by expression of the gene Engrailed-1 (En-1), which encodes a homeodomain
transcription factor (see Section 2.23). Mutations that destroy En-1 function result in
Wnt-7a being expressed ventrally as well as dorsally, giving a double dorsal limb. En-1
expression is induced by BMP signaling in the ventral ectoderm.
One function of Wnt-7a is to induce expression of the gene encoding the LIM
homeobox transcription factor Lmx1b in the mesenchyme underlying the dorsal

Dorsal and ventral surfaces of a left-hand glove Putting left-hand glove on right hand reverses D–V axis

Fig. 11.13 Reversing the dorso-ventral

axis of limb bud ectoderm. The only way a
left-hand glove (representing the ectoderm
from a left-hand limb bud) can fit on a right
D V
hand (the mesoderm of a right hand) is by
turning it over so that the former ventral Left hand Right hand
side (patterned) is now uppermost. The
dorso-ventral polarity of the glove is now
reversed in relation to the dorso-ventral axis
of the hand. The antero-posterior relationship Left-hand glove Left-hand glove Left-hand glove Left-hand glove
remains the same.
 The vertebrate limb 427

ectoderm (see Fig. 11.14). Clonal analysis of marked cells in mouse limb buds has
Transverse section through tip of a limb bud
recently revealed dorso-ventral compartments in the mesenchyme, with the progeny
of dorsal cells being restricted to the dorsal half of the limb bud and the progeny of D
Lmx1b Wnt-7a
ventral cells to the ventral half. This is the first time that cell-lineage restrictions have
been shown in mesenchymal tissue; all the other compartments discovered so far
A P
in both vertebrates and invertebrates are in epithelia—for example, the Drosophila
epidermis (see Chapter 2) and the rhombomeres of the vertebrate hindbrain (see
Chapter 5). The compartmentalization of the mesenchyme may serve to control gene engrailed polarizing
V region
expression; Lmx1b, for example, is expressed exactly in the dorsal compartment. In
chick embryos, Lmx1b has been shown to specify a dorsal pattern in the mesoderm.
Fig. 11.14 The ectoderm controls dorso-
Ectopic expression of Lmx1b in the ventral mesoderm of the chick limb results in the
ventral pattern in the developing limb. The
cells adopting a dorsal fate, resulting in a mirror-image dorsal limb. Loss-of-function gene for the secreted signal Wnt-7a is
mutations in human LMX1B cause an inherited disease called nail-patella syndrome, expressed in the dorsal ectoderm and the
in which these dorsal structures (nails and kneecaps) are malformed or absent. gene engrailed is expressed in the ventral
ectoderm. Expression of the gene for the
transcription factor Lmx1b is induced in the
11.10 Development of the limb is integrated by interactions between
dorsal mesoderm compartment by Wnt-7a
signaling centers signaling and Lmx1b is involved in specifying
It is crucial that patterning along all three limb axes is integrated so that the correct dorsal structures.
anatomy of the limb is generated. Patterning is integrated by interactions between the
signals; Wnt-7a from dorsal ectoderm, FGFs from the apical ectodermal ridge, and
Shh from the polarizing region. In many mice mutant for loss of Wnt-7a function,
posterior digits are missing, suggesting that Wnt-7a is also required for normal antero-
posterior patterning, and Shh expression in the polarizing region is also reduced in
these mice. Similar results are observed in chick embryos when the dorsal ectoderm
of the wing bud is removed.
A positive-feedback loop has been discovered between the polarizing region and the
apical ectodermal ridge that links development along the proximo-distal and antero-
posterior axes. Shh signaling maintains the level of Fgf gene expression in the ridge,
while FGF signaling in turn maintains Shh gene expression in the polarizing region.
This feedback loop also involves BMP, another of the proteins required to pattern the
antero-posterior axis (see Section 11.7). A minimum level of BMP signaling is required
to maintain outgrowth of the limb bud, but if the level rises too high the apical ridge
regresses. The level of BMP signaling in the early limb bud is controlled by the BMP
antagonist, Gremlin. Gremlin expression is first induced in the limb bud mesoderm
by BMP-4 signaling and a negative-feedback loop is rapidly set up in which Gremlin
protein antagonizes BMP activity, keeping it low. Fgf expression is consequently main-
tained in the apical ridge and, in turn, maintains Shh expression in the polarizing region
in the earliest stages of limb-bud development. Shh signaling subsequently maintains
Gremlin expression and thus low levels of BMP activity in the limb-bud mesoderm,
reinforcing the positive-feedback loop between apical ridge FGF signaling and polar-
izing region Shh signaling. The epithelial–mesenchymal positive-feedback loop takes
about 12 hours to complete, and is linked to the rapid negative-feedback loop between
BMP-4 and Gremlin. This self-regulating system is an excellent example of the feedback
control mechanisms that make development so robust and reliable (see Section 1.19).

11.11 Different interpretations of the same positional signals give

different limbs

The signals controlling limb-bud outgrowth and patterning, such as FGF and Shh,
are the same in chick wing and leg, but they are interpreted differently. A polarizing
region from a wing bud, for example, will specify additional mirror-image digits if
grafted into the anterior margin of a leg bud. However, these digits develop as toes,
not wing digits, as the signal is being interpreted by leg cells. In a similar ­manner,
428 Chapter 11 Organogenesis

Wing bud

Resulting wing with clawed toes

toes
Fig. 11.15 Chick proximal leg bud cells wing bud
grafted to a distal position in a wing bud
acquire distal positional values. Proximal
tissue from a leg bud that would normally
develop into a thigh is grafted to the tip of a leg bud
wing bud underneath the apical ectodermal
ridge. It acquires more distal positional polarizing
region
values and interprets these in terms of leg
structures, forming clawed toes at the tip of
Leg bud
the wing.

signals are conserved between different vertebrates; for example, a mouse apical
­ectodermal ridge grafted in place of a chick apical ectodermal ridge can provide the
appropriate signal for early chick limb development. Signals are, however, interpreted
according to the origin of the responding cells: when grafted into the anterior margin
of a chick wing bud, a polarizing region from a mouse limb bud specifies additional
wing structures, not mouse structures (see Section 11.6). Thus, the signals from the
different polarizing regions are the same, as are the signals from the different apical
ridges; the difference in the structures formed by the limbs is a consequence of how
the signals are interpreted and depends on the genetic constitution and developmen-
tal history of the responding cells in the limb bud.
A further demonstration of the interchangeability of positional signals comes from
grafting limb bud tissue to different positions along the proximo-distal axis of the bud.
If tissue that would normally give rise to the thigh is grafted from the proximal part of
an early chick leg bud to the tip of an early wing bud, it develops into toes with claws
(Fig. 11.15). The tissue has acquired a more distal positional value after transplanta-
tion, but interprets this value according to its own developmental program, which is
to make leg structures. One limb-specific protein is the transcription factor Pitx1. It is
expressed in the leg bud only and its ectopic expression in the chick wing results in
the development of leg-like distal structures, such as toes and claws.

11.12 Hox genes establish the polarizing region and also provide a code for
limb patterning

Hox genes specify position along the antero-posterior axis of the vertebrate body (see
Section 5.6). In the early limb bud, Hox genes establish the polarizing region and later
on in development they provide positional values for the limb pattern. At least 23 dif-
ferent Hox genes are expressed during mouse and chick limb development. Attention
has been largely focused on the genes of the Hoxa and Hoxd gene clusters, which are
related to the Drosophila Abdominal-B gene (see Box 5E, p. 194) and genes in the
5′ regions of these clusters are expressed in both forelimbs and hindlimbs.
The expression of Hoxa and Hoxd genes during limb development is very dynamic—­
that is, it changes as the limb develops. In later limb buds, the pattern of expression
changes in response to signaling by the polarizing region, in concert with signaling by
the apical ectodermal ridge. But early on, Hoxd9 and Hoxd10 genes are expressed in
the lateral plate mesoderm as it begins to thicken to become a limb bud. ­Expression
 The vertebrate limb 429

of Hoxd11, Hoxd12, and Hoxd13 genes is then initiated in rapid succession in pro- Anterior
gressively more restricted domains centered on the posterior distal region of both
Hoxd9 Hoxd9–13
forelimbs and hindlimbs, and these domains then extend along the limb bud as it
grows out, as shown for the chick wing bud in Fig. 11.16 (upper panel). Expression of
Hoxa9–13 is also initiated around the same time and gives rise to similar overlapping
Proximal Distal
domains of expression along the proximo-distal axis, but without a posterior bias,
and these domains also become more extended along the limb bud as it grows out.
At the stage shown in Fig. 11.16 (lower panel), Hoxa9 expression covers the complete
Hoxd9–11
region in which the Hoxa genes are expressed, and successive genes are expressed in Posterior
more and more distal regions. At later stages of limb development, Hoxa11 expression
becomes restricted to the intermediate region of the limb and Hoxd11 is also expressed Anterior
there; Hoxd13 and Hoxa13 are expressed throughout the distal region of the limb, Hoxa9 Hoxa9–13
where the digits are beginning to form, while Hoxd10–12 are also expressed distally
except in the most anterior part of the bud.
Sophisticated genetic experiments in mice have succeeded in identifying cis- Proximal Distal
­regulatory regions that drive Hoxd gene expression in the limb. An enhancer lying 3′
(downstream) to the cluster drives the nested pattern of Hoxd gene expression in early
limb buds and another enhancer, lying 100 kb 5′ (upstream) of the cluster, drives Hoxd Hoxa9–11
gene expression in the tissue that will form digits. The particularly strong activation Posterior
of Hoxd13 throughout the digital region (compared to the other Hoxd genes) may be
because it is the first gene that this distant control region comes into contact with. Fig. 11.16 Pattern of Hox gene
The early phase of Hox gene expression in the limb is involved in establishing the expression in the chick wing bud. The
polarizing region. Genetic deletion of all Hoxa and Hoxd genes in the mouse forelimb Hoxd genes (upper panel) are expressed in
a nested pattern centered on the posterior
bud results in the absence of Shh expression and severe limb truncations. Conversely,
tip of the wing bud, Hoxd13 being expressed
genetic manipulations leading to inversions of the Hoxd cluster or deletions of large
most posteriorly and distally. The Hoxa genes
parts of the downstream control region result in Hoxd genes in the 5′ part of the
(lower panel) are similarly expressed in a
cluster being expressed in the patterns of 3′ Hoxd genes, so that Hoxd13 is expressed nested pattern along the proximo-distal axis,
throughout the early bud instead of being restricted to the posterior. The unrestricted Hoxa13 being expressed most distally.
expression of Hoxd13 results in Shh expression both anteriorly and posteriorly, and
the development of duplicated limb patterns. Hoxb8, one of the few Hox genes out-
side the Hoxa and Hoxd clusters that is expressed in the forelimb, also seems to be
involved in establishing Shh expression.
Hox gene expression is maintained and elaborated as the limb bud develops in
response to combined Shh and FGF signaling. These later phases of Hox gene expres-
sion are involved in recording positional information in the limb and, in some cases,
are also involved in the growth of the cartilaginous elements once they have formed.
If the Hox genes are involved in recording positional information, then experimen-
tal manipulations that lead to changes in the pattern of the limb skeletal elements
should cause a corresponding change in domains of Hox expression. When a polar-
izing region is grafted to the anterior margin of a wing bud, there is indeed a change
to a mirror-image pattern of the expression of Hoxd genes (Fig. 11.17). The change
occurs within 24 hours of grafting, which is about the time required for the polarizing
region to exert its effect (see Section 11.6).
As outlined above, the proximo-distal sequence of expression of Hoxa and Hoxd
genes eventually comes to correspond generally with the three main proximo-distal Fig. 11.17 Change in expression of Hoxd
regions of the limb. Hoxa9 and Hoxd9 are expressed in the upper part of the ­forelimb, genes following a graft of the polarizing
region. A polarizing region grafted to the
where the humerus forms. Hoxa11 and Hoxd11 are expressed in the lower limb,
anterior margin of a limb bud results in mirror-
where the radius and ulna (or tibia and fibula) develop, and Hoxa13 and Hoxd13 are
image expression of the Hoxd13 genes and
expressed in the wrist (ankle) and digit-forming regions. There are some differences
a mirror-image duplication of the digits (main
in the pattern of expression of the Hox paralogous subgroups 9 to 13 in the forelimb panel). The inset shows the expression of
and hindlimb. Hox9 paralogs are not expressed in the hindlimb, but there are only Hoxd13 (arrows) at the posterior tip in a
relatively small variations in the pattern of expression of the other groups. normal limb bud (inset, left). In the operated
Gene function can only be inferred from expression patterns, and mutational experi- limb (inset, right) Hoxd13 is expressed at
ments are needed to prove function. We need to answer the question of whether both the anterior and posterior tips.
430 Chapter 11 Organogenesis

Fig. 11.18 Functional domains of Hox gene expression in the patterning of the mouse
Hox paralogous subgroup forelimb. Colored regions denote regions of Hox gene influence on patterning as determined
III IV 9 10 11 12 13 from gene knock-out experiments. Hox9 (red) and Hox10 (orange) paralogs function together
II to pattern the upper forelimb (humerus), Hox10 paralogs also affect the middle forelimb (radius
V and ulna), as represented by the lighter orange shading. Hox11 paralogs (yellow) are mostly
P3
P2
involved in patterning this middle region, with some influence on pattern in the wrist and ‘hand’
I P1 region (lighter yellow shading). Hox12 paralogs (green) predominantly pattern the wrist, while
MC Hox13 paralogs (purple) predominantly pattern the hand. MC = metacarpal, P = phalanx.
Adapted from Wellik, D.M. and Capecchi, M.R.: 2003.
Carpus

a­ ltering Hox gene expression in the limb leads to homeotic changes similar to those
Radius Ulna
seen in vertebrae after Hox gene knock-out (see Section 5.7). The results of Hox gene
knock-outs in mice are quite difficult to interpret in relation to the limb, but they do
show that Hox genes clearly influence the patterning of different regions of the mouse
limb along the proximo-distal axis (Fig. 11.18). This conclusion is based on the pheno-
types of mice mutant for genes in the same paralogous subgroup. For example, paralo-
gous subgroup Hox10 controls the development of proximal elements, whereas Hox11
Humerus patterns the radius and ulna in the forelimb (tibia and fibula in the hindlimb). Inactiva-
tion of genes of the Hox13 subgroup disrupts digit development, as one might expect
from the expression of Hoxd13 and Hoxa13 in the distal region of the limb. Because of
the dynamic nature of Hox gene expression as the limbs develop, it is difficult to inter-
pret these results in terms of either the progress zone model or the two-signal model
for proximo-distal patterning of the limb (see Section 11.5). The fact that the domain
of Hoxd13 at the tip of the limb is maintained by FGF signaling as the limb grows out
could be seen as consistent with the two-signal model, but it is not clear that the cells
that express Hoxd13 at early stages have already been specified to form digits.
Which target genes the Hox proteins subsequently act on to produce the pattern
along the proximo-distal axis is still largely unknown, but expression of the ephrin
receptors EphA7 and EphA3 has been shown to be regulated by Hoxd13 and Hoxa13.
Ephrins and their receptors are involved in various cell-biological processes, such as
the formation of regional boundaries (see Section 5.11), cell sorting, and guidance of
cell migration (see Chapter 12).
Mutations in Hox genes are known to underlie human conditions in which limbs
are abnormal. Mutations in Hoxd13 lead to fusion of digits and some types of polydac-
tyly, while mutations in Hoxa13 lead to hand-foot-genital syndrome, but the precise
phenotypes seen in patients are difficult to explain given our current knowledge.

11.13 Self-organization may be involved in the development of the limb bud

Signaling from the polarizing region might not be the only means by which a series
of cartilaginous elements can develop along the antero-posterior axis of the limb,
as it seems that the limb bud has some capacity for self-organization. Cells from a
disaggregated limb bud can be reaggregated and develop digits, even in the absence
of a polarizing region. Mesenchymal cells disaggregated from early chick limb buds
are thoroughly mixed to disperse the polarizing region, or the mesenchymal cells
are taken from just the anterior half of a limb bud, which lacks a polarizing region.
The disaggregated cells are placed in an ectodermal jacket (prepared as described
in Section 11.9), and grafted to a site where they can acquire a blood supply, but
which is ‘neutral’ in terms of providing polarizing signals, such as the dorsal surface
of an older limb. Limb-like structures develop from the grafted reaggregated limb
bud cells, even though they have developed in the absence of a polarizing region.
Several long cartilaginous elements may form in the more proximal regions of these
abnormal limbs, although none of the proximal elements can be easily identified with
normal structures. More distally, however, reaggregated hindlimb bud cells develop
 The vertebrate limb 431

Leg bud Mesodermal cells Mesodermal cells reaggregated, placed in

disaggregated ectodermal jacket, and grafted to neutral site Toes develop distally

mesoderm

ectoderm

Fig. 11.19 Reaggregated limb bud cells

i­dentifiable toes (Fig. 11.19). The fact that well-formed cartilaginous elements can
form digits in the absence of a localized
develop at all in the absence of a discrete polarizing region shows that the bud has polarizing region. Mesodermal cells from a
considerable capacity for self-organization. chick leg bud are disaggregated, mixed to
These results suggest that the limb bud might be able to generate a basic prepat- disperse all the cells, including the polarizing
tern consisting of equivalent cartilaginous elements. A prepattern is a basic organiza- region, and then reaggregated, placed in an
tion that is generated automatically in the embryo and can be identified in advance ectodermal jacket, and grafted to a neutral
of the later development of a similar pattern of structures. The pattern that eventually site. Well-formed toes develop distally.
develops does not have to follow the prepattern exactly but may be a modified form
of the prepattern. The initially equivalent cartilaginous elements of the limb prepat-
tern could then be given their identities and further refined in response to positional
information involving signals such as Shh and the activation of Hox genes and their
downstream targets (see Sections 11.7 and 11.12). The mechanism for generating the
prepattern could be based on a reaction–diffusion mechanism (Box 11D, p. 432).
In the wing, for example, a reaction–diffusion or related mechanism could result in dorsal muscle mass
a single peak in some morphogen forming in the proximal region of the limb, which ulna radius
would specify a prepattern for the humerus. More distally, alterations in the reaction–
diffusion conditions, due to changes in proximo-distal positional information, could
give rise to three peaks of the morphogen, giving the cartilaginous elements of the
three wing digits. These prepatterns could then be modified by the signals that specify
antero-posterior and dorso-ventral positional information.
In this reaction–diffusion model, polydactyly could simply result from a chance
widening of the limb bud. If a reaction–diffusion mechanism generates a periodic pat-
ventral muscle mass
tern of cartilage elements across the limb as the digits are forming, merely widening
the limb bud by some small developmental accident would enable a further digit to
develop.

11.14 Limb muscle is patterned by the connective tissue

Limb muscle cells have a different origin from that of the limb connective tissue cells,
including muscle-associated connective tissue cells. If quail somites are grafted into a
chick embryo at a site opposite where the wing bud will develop, the wing that forms
will have muscle cells of quail origin, but all its other cells will be of chick origin.
The cells that give rise to limb muscles migrate into the limb bud from the somites as
myoblasts at a very early stage (see Section 5.9). After migration, the myoblasts mul-
Fig. 11.20 Development of muscle in
tiply and ­initially form dorsal and ventral blocks of presumptive ­muscle (Fig. 11.20).
the chick limb. A cross-section through
These blocks undergo a series of divisions to give the individual muscles. The initial
the chick limb in the region of the radius
migration of myoblasts into the limb is restricted so that they enter either the dorsal
and ulna shortly after cartilage formation
or the ventral region; cells do not cross the dorsal–ventral compartment boundary, shows presumptive muscle cells present as
which is characterized by the dorsal expression of the transcription factor Lmx1b two blocks—the dorsal muscle mass and the
(see Fig. 11.14). Presumptive muscle cells, at least initially, do not acquire positional ventral muscle mass (upper panel). These
values in the same way as do cartilage and connective tissue cells, and are all equiva- blocks then undergo a series of divisions to
lent. However, Hoxa11 is expressed in the myoblasts when they enter the limb and is give rise to individual muscles (lower panel).
432 Chapter 11 Organogenesis

Box 11D Reaction–diffusion mechanisms

Spontaneous self organization Doubling of width

+
Activator
– +
Inhibitor

There are some self-organizing chemical systems that spontaneously generate spatial pat-
Increasing
concentration terns of concentration of some of their molecular components. The initial distribution of the
molecules is uniform, but over time the system forms wave-like patterns (see figure). The
essential feature of such a self-organizing system is the presence of two or more types of
diffusible molecules that interact with one another. For that reason, such a system is known
as a reaction–diffusion system. For example, if the system contains an activator molecule
that stimulates both its own synthesis and that of an inhibitor molecule, which in turn inhib-
its synthesis of the activator, a type of lateral inhibition will occur so that synthesis of activa-
tor is confined to one region (see upper diagram, top left).
Under appropriate conditions, which are determined by the reaction rates and diffusion
constants of the components, a closed system of a certain size can spontaneously develop
a spatial pattern of activator with a single concentration peak. If the size of the system is
increased, two peaks will eventually develop, and so on. Such a mechanism could thus in
theory generate periodic patterns such as the arrangement of digits or the sepals or petals
of flowers. When the chemicals can diffuse in two dimensions, the system can give rise to a
number of peaks that are somewhat irregularly spaced (see lower diagram, top left).
Such a system may underlie some of the patterns of pigmentation that are common through-
out the animal kingdom, such as the patterns of spots and stripes seen in the zebra and chee-
tah (see photos, left). How these patterns are generated is not yet known, but one possibility
is a reaction–diffusion mechanism. Assuming that pigment is synthesized in response to some
activator, and that synthesis only occurs at high activator concentrations, some animal color
patterns can be mimicked by a reaction–diffusion system in computer simulations.
A characteristic feature of reaction–diffusion patterns is that new intermediate peaks appear
as the system grows in size. The angelfish Pomocanthus semicirculatus provides a remarkable
example of striping that could
2 months 6 months 12 months be generated by a reaction–­
diffusion mechanism (see below).
Juvenile P. ­semicirculatus, less
than 2 cm long, have only three
dorso-­ventral stripes. As the fish
grow, the intervals between the
stripes increase until the fish is
around 4 cm long. New stripes
then appear between the origi-
1 cm nal stripes and the stripe inter-
vals revert to those present
at the 2 cm stage. As the fish
grows larger, the process is repeated. This type of dynamic patterning is what would be expected of a reaction–diffusion mechanism.
Computer simulations of reaction–diffusion mechanisms can also generate the patterns seen on a wide variety of mollusc shells.
­Nevertheless, there is as yet no direct evidence for a reaction–diffusion system patterning any developing organism.
Top figure after Meinhardt, H., et al.: 1974.
 The vertebrate limb 433

probably induced by the surrounding mesenchyme. Cells in the posterior region of

the dorsal and ventral masses express Hoxa13. In both cases, the pattern of Hox gene
expression in the myoblasts is different from that of the surrounding mesenchyme.
Evidence for the early equivalence of myoblasts comes from experiments where
somites are grafted from the future neck region of the early chick embryo to replace
the normal wing somites. The myoblasts that enter the wing therefore come from neck
somites, but a normal pattern of limb muscles still develops. This shows that the muscle
pattern is determined by the connective tissue into which the myoblasts migrate, rather
than by the myoblasts themselves. One way this could be achieved is if the prospective
muscle-associated connective tissue had surface or adhesive properties that the myo-
blasts recognized and were attracted to. If the pattern of adhesiveness in the connective
tissue changed over time, this could result in associated changes in the migration of pre-
sumptive muscle cells, and this could account for the splitting of the muscle masses.
The myoblasts proliferating in the limb express Pax3. When Pax3 is downregulated,
cell proliferation stops and the cells differentiate (see Section 10.10). Signals from the
overlying ectoderm, including BMP-4, prevent premature differentiation. The devel-
opment of tendons is marked by the expression of the transcription factor Scleraxis
(see Section 5.9).

11.15 The initial development of cartilage, muscles, and tendons

is autonomous

As we have seen, a graft of the polarizing region causes the development of a mirror-
image pattern of all the elements of the limb, not only the skeletal elements but also
muscles and tendons. This indicates that the patterns of cartilage, tendon, and muscle
in the limb are all specified by the same set of signals and there is little or no com-
munication between them—that is, the development of each element is autonomous.
If, for example, just the tip of an early chick wing bud is removed and grafted to the
flank of a host embryo, it initially develops into normal distal structures with a wrist
and three digits. The long tendon that normally runs along the ventral surface of
digit 3 starts to develop, even though both its proximal end and the muscle to which
it attaches are absent. The tendon does not continue to develop, however, because it
does not make the necessary connection to a muscle, and so is not put under tension.
The mechanism whereby the correct connections between tendons, muscles, and car-
tilage are established has still to be determined. It is clear, however, that there is little
or no specificity involved in making such connections; if the tip of a developing limb
is inverted dorso-ventrally, dorsal and ventral tendons can join up with inappropriate
muscles and tendons. They simply make connections with those muscles and tendons
nearest to their free ends—they are promiscuous.

11.16 Joint formation involves secreted signals and mechanical stimuli

The first event in joint formation is the formation of an ‘interzone’ at the site of
the prospective joint, in which the cartilage-producing cells become fibroblast-like
and cease producing the cartilage matrix, thus separating one cartilage element from
another. At this time, Wnt-14 and the BMP-related protein GDF-5 are expressed in the
presumptive joint region. Ectopic expression of Wnt-14 can induce the early steps in
joint formation, and mice lacking GDF-5 have some of their joints missing. A little
later, the musculature begins to function and muscle contraction occurs. At this point,
high levels of hyaluronan, a component of the proteoglycans that lubricate joint sur-
faces, are made by the cells of the interzone; hyaluronan is secreted in vesicles that
coalesce to form the joint cavity, which separates the articulating surfaces of the joint
from each other. Synthesis of hyaluronan by the cells lining these surfaces depends on
the mechanical stimulation that results from muscle contraction, and in the absence
434 Chapter 11 Organogenesis

Fig. 11.21 Cell death during leg

development in the chick. Programmed
cell death in the interdigital region results in
separation of the toes. Scale bar = 1 mm.
Photographs courtesy of V. Garcia-Martinez,
from Garcia-Martinez, V., et al.: 1993.

of this activity, the joints fuse. Cells isolated from the interzone region and grown in
culture respond to mechanical stimulation by increased secretion of hyaluronan when
the substratum to which they are attached is stretched.

11.17 Separation of the digits is the result of programmed cell death

Programmed cell death by apoptosis plays a key role in molding the form of chick and
mammalian limbs, especially the digits. The region where the digits form is initially
plate-like, as the limb is flattened along the dorso-ventral axis. The cartilaginous ele-
ments of the digits develop from the mesenchyme at specific positions within this
plate. Separation of the digits then depends on the death of the cells between these
cartilaginous elements (Fig. 11.21). BMPs may be involved in signaling this cell death:
if the functioning of BMP receptors is blocked in the developing chick leg, cell death
does not occur and the digits are webbed.
Programmed cell death is a normal part of patterning and cell differentiation (see
Chapter 10). The fact that ducks and other waterfowl have webbed feet, whereas
Ectoderm chickens do not, is simply the result of less cell death between the digits of waterfowl.
If chick-limb mesoderm is replaced with duck-limb mesoderm, cell death between
Chick Duck
the digits is reduced and the chick limb develops ‘webbed’ feet (Fig. 11.22). It is the
mesoderm that determines the patterns of cell death both within the mesoderm and
the overlying ectoderm. In amphibians, digit separation is not due to cell death, but
results from the digits growing more than the interdigital region.
Chick

Programmed cell death also occurs in other regions of the developing limb, such as the
anterior margin of the limb bud, between the radius and ulna, and in the wing polarizing
region at a later stage. Indeed, it was the investigation of cell death in this region of a
Mesoderm

chick wing bud by transplanting it to the anterior margin of another chick wing bud that
led to the discovery that it acts as a polarizing region. One suggestion for the role of cell
death in the polarizing region is that it could control the number of cells expressing Shh.
Duck

Summary

Positioning and patterning of the vertebrate limb is largely carried out by intercellu-
lar interactions that provide the cells with positional information. The position of the
limb buds along the antero-posterior axis of the body is probably related to Hox gene
Fig. 11.22 The mesoderm determines expression. There are two key signaling regions within the limb bud. One is the apical
the pattern of cell death. In the developing
ectodermal ridge, which is required for limb bud outgrowth; the second is the polar-
webbed feet of ducks and other water birds,
izing region at the posterior margin of the limb, which specifies pattern along the
there is less cell death between the digits
antero-posterior axis. The signals from the apical ridge are fibroblast growth factors
than in birds without webbed feet. When the
mesoderm and ectoderm of embryonic chick (FGFs) and are essential for limb outgrowth. Shh protein is produced by the polarizing
and duck limb buds are exchanged, webbing region and provides a graded positional signal across the antero-posterior axis of the
develops only when duck mesoderm is limb bud that is pivotal for determining digit number and pattern. Pattern across the
present.
 Insect wings and legs 435

­ orso-ventral axis is specified by the ectoderm. Hox genes are expressed in a well-
d
defined spatio-temporal pattern within the limb bud and may provide the molecular
basis of positional identity. The limb muscle cells are not generated within the limb but
migrate in from the somites and are patterned by the limb connective tissue. Pattern-
ing of the cartilaginous elements may involve a self-organizing mechanism. In birds and
mammals, separation of the digits is achieved by programmed cell death.

Summary: vertebrate limb development

expression of Hox genes, FGF, and Wnts, in flank initiates limb-bud development
signals from polarizing
region help maintain
apical ectodermal ridge apical ridge polarizing region ectoderm specifies
in limb bud in limb bud dorso-ventral axis
signals from apical
ridge help maintain
polarizing region
signals from signals from polarizing region (Shh)
apical ridge (FGF) and BMPs specify position
maintain region of along antero-posterior axis
undifferentiated cells
at tip of growing limb

proximo-distal and antero-posterior specification

cartilage elements produced
complex pattern of Hox gene expression
controls pattern of cartilage and muscle
cartilage elements and muscle masses
develop from cells originating in somites

Insect wings and legs

The adult organs and appendages of Drosophila, such as the wings, legs, eyes, and anten-
nae, develop from imaginal discs, which are excellent experimental systems for analyzing
pattern formation. The discs invaginate from the embryonic ectoderm as simple pouches
of epithelium during embryonic development and remain as such until metamorphosis
(see Fig. 2.3). Although all imaginal discs superficially look rather similar, they develop
according to the segment in which they are located. The specification and basic patterning
of the imaginal discs in the thoracic segments as wing or leg occur within the embryonic
epithelium, at the time when the segments are being patterned and given their identity.
The wing and leg discs are initially specified as clusters of 20–40 cells, and during larval
development they grow about 1000-fold. The adult Drosophila compound eyes develop
from imaginal discs at the anterior end of the embryo. A feature of imaginal discs is that
if cells are removed or destroyed, the disc will regrow to the correct size and develop
normally. The mechanism underlying this is not yet known: one proposal invokes the
gradient mechanism for determining organ size discussed in Box 13A (p. 514).
Although insect wings and legs differ so greatly in appearance, they are partially
homologous structures, and the strategy of their patterning is similar. Moreover, the
mechanisms by which they are patterned, and even the genes involved, show some
amazing similarities to the patterning of the vertebrate limb. We will start by discuss-
ing the patterning of the wing disc and its development into the adult wing and then
compare the patterning of the leg disc. The development of the Drosophila eye is
discussed later in the chapter.
436 Chapter 11 Organogenesis

D 11.18 Positional signals from compartment boundaries pattern the wing

imaginal disc
notum
blade Distal
Imaginal discs are essentially epithelial sheets, which, in the case of wing and leg
Ventral discs, will give rise to epidermis in the adult. The epidermis of body segments, and of
wing
surface the wings and legs that derive from them, is divided into anterior and posterior com-
A P partments—regions of cell lineage restriction (see Section 2.23). These correspond to
the compartment boundaries in the segments in which they develop. In the wing disc
V
epithelium, a second compartment boundary between the dorsal and ventral regions
Ventral
hinge develops during the second larval instar (Fig. 11.23). The proximal parts of the wing
and leg discs give rise to parts of the thoracic body wall.
The adult wing is a largely epidermal structure in which two epidermal layers—the
dorsal and ventral surfaces—are close together. At metamorphosis, by which time
patterning of the imaginal discs is largely complete, wing and leg discs undergo a
series of profound anatomical changes. Essentially, the invaginated epithelial pouch
is turned inside out, as its cells differentiate and change shape. In the case of the
notum
Proximal wing, the epithelial pouch extends and folds so that one surface comes to lie beneath
the other to form the double-layered wing structure (Fig. 11.24). The zinc-finger tran-
Fig. 11.23 Fate map of the wing imaginal scription factors Elbows and No-ocelli act in both the wing and leg imaginal discs to
disc of Drosophila. The wing disc before repress genes specifying the body wall, and thus permit the appendages to develop.
metamorphosis is an ovoid epithelial In the patterning of the wing disc, signaling regions are set up along the compart-
sheet bisected by the boundary between ment boundaries (Fig. 11.25). We shall look first at the antero-posterior patterning
the anterior (A) and the posterior (P) produced from the signaling center at the anterior–posterior compartment boundary.
compartments. There is also a compartment Cells at this boundary form a signaling region that specifies pattern along the antero-
boundary between the future ventral (V) posterior axis of the wing. A cascade of events sets up the antero-posterior signaling
and dorsal (D) surfaces of the wing. At
center. It begins with the expression of the engrailed gene in the posterior compart-
metamorphosis, the ventral surface will fold
ment of the disc (see Fig. 11.25), which reflects the pattern of gene expression in the
under the dorsal surface (see Fig. 11.24). The
disc also contains part of the dorsal region
embryonic parasegment from which the disc derives. We shall return to the question
of thoracic segment 2 in the region of wing of what gives wing and leg discs their segmental identity later in this chapter.
attachment. As we saw in relation to the setting up of compartments in the Drosophila embryo,
After French, V., et al.: 1994. the cells that express engrailed also express the gene hedgehog (see Section 2.24).
The maintenance of compartment boundaries in the wing depends partly on com-
munication between compartments. At the antero-posterior compartment boundary,
the ­secreted Hedgehog protein acts as a short-range morphogen over about 10 cell
diameters and induces adjacent cells in the anterior compartment to express the
­decapentaplegic (dpp) gene (see Fig. 11.25). As noted in Box 11A (p. 421), Hedgehog
protein may not move by simple diffusion or endocytosis. The Hedgehog signal acts
on the anterior cells through a receptor complex involving the proteins Patched and
Smoothened (see Fig. 2.40).
Decapentaplegic protein is secreted at the antero-posterior compartment boundary
(see Fig. 11.25) and is probably the positional signal for patterning both the anterior
and posterior compartments along the antero-posterior axis. Dpp appears to form a

Imaginal wing disc Longitudinal section The wing blade develops by the extension and
coming together of the dorsal and ventral surfaces

Fig. 11.24 Schematic representation of

Dorsal
the development of the wing blade from
the imaginal disc. Initially, the dorsal and Dorsal
ventral surfaces are in the same plane. At Ventral
metamorphosis, the invaginated pouch turns
Ventral
inside out and extends, so the dorsal and
ventral surfaces of the wing come together.
 Insect wings and legs 437

Fig. 11.25 Establishment of signaling

engrailed hedgehog decapentaplegic wingless
regions in the wing disc at the
Dorsal compartment boundaries. The disc
is divided into anterior and posterior
compartments and, later in development,
into dorsal and ventral compartments.
The gene engrailed is expressed in the
posterior compartment, where the cells also
express the gene hedgehog and secrete the
Hedgehog protein. Where Hedgehog protein
Ventral
interacts with anterior compartment cells,
anterior posterior apterous Notch signaling the gene decapentaplegic is activated, and
compartment compartment
Decapentaplegic protein is secreted into
both compartments, as indicated by the black
arrows. At the dorso-ventral compartment
concentration gradient across the wing disc that provides a long-range signal control- boundary, the wingless gene is activated
ling the localized expression of the genes for the transcription factors Spalt (Sal), Omb and the signaling molecule is the protein
and Brinker in the wing disc (Fig. 11.26), and they provide a further level of patterning Wingless.
in the wing. Decapentaplegic acts as a morphogen, specifying the expression of spalt
and omb when particular threshold concentrations are reached (Fig. 11.27); low levels
induce omb, whereas higher levels are required to induce spalt. As noted earlier, Omb
and Spalt are related to the Tbx and Sall genes, respectively, which are involved in
vertebrate limb patterning (see Section 11.8).
Decapentaplegic regulates its target genes by binding to the receptor protein, Thick
veins, and activating an intracellular signaling pathway that results in changes in
gene expression. One of the key genes regulated by Dpp in regard to wing pattern-
ing is brinker, which encodes a transcription factor. Dpp signaling downregulates
brinker expression, thus forming a gradient of intranuclear Brinker protein that is
the inverse of the Dpp gradient (see Fig. 11.26). Brinker represses some of the target
genes of Dpp, such as omb, and so repression of brinker by Dpp is required for Dpp
to be able to activate some of its target genes. These genes are then expressed in
patterns that reflect their sensitivity to activation by Dpp and ­repression by Brinker
(see Fig. 11.26). In addition, brinker reduces cell growth, while Dpp promotes it.
The evidence that Dpp is a morphogen in the wing disc comes from several types
of experiments. First, clones of cells unable to respond to a Dpp signal do not express
spalt or omb. Second, ectopic expression of the dpp gene in a region that does not nor-
mally express spalt and omb leads to the localized activation of these genes around
the dpp-expressing cells. Formation of the long-range Dpp gradient in the wing disc
is a complex process and is not yet fully understood. It appears to involve receptor-
mediated endocytosis, as blocking the action of dynamin, an intracellular protein
required for the internalization of endocytic vesicles, blocks gradient formation. The

a b c d e

Fig. 11.26 Domains of gene expression initiated by Dpp signaling therefore expressed in a wide strip over the antero-posterior border,
pattern the Drosophila wing disc. Panel (a) shows the expression of where brinker is not expressed. Panel (e) shows a composite overlay of
decapentaplegic (blue) at the antero-posterior compartment border of (b), (c), and (d).
the wing blade. Dpp signaling represses expression of the gene brinker (a) courtesy of K. Basler, from Nellen, D., et al.: 1996. (b–e) from
(green, panel (b)), which is therefore expressed only in two regions on Moser, M., Campbell, G.: 2005.
either side of the zone of dpp expression. In turn brinker represses the
expression of sal (red, panel (c)) and omb (blue, panel (d)), which are
438 Chapter 11 Organogenesis

glypican Dally, present on the cell surface, is also involved in both shaping the Dpp
Wing disc
decapentaplegic gradient and influencing signaling via the Dpp receptor Thick veins.
omb The pattern of veins on the adult wing blade is one of the ultimate outcomes of the
A P spalt antero-posterior patterning activity of Hedgehog and Dpp in the wing disc (Fig. 11.28).
This was shown by ectopically expressing the hedgehog gene in genetically marked
X Y
antero-posterior cell clones generated at random in wing discs. When such clones form in the posterior
compartment
boundary compartment they have little effect (as hedgehog is normally expressed throughout
threshold for this compartment in the disc) and development of the wing is more or less normal.
omb
expression With hedgehog-expressing clones in the anterior compartment, however, a wing with
threshold for
spalt a mirror-symmetric repeated pattern is produced (see Fig. 11.28). Ectopic expression
expression
of hedgehog has set up new sites of dpp expression in the anterior compartment of the
wing disc, which results in new gradients of Dpp protein being formed.
  There is no single or simple process by which the morphogens Hedgehog and Dpp
A P
specify the vein pattern: rather, the position of each vein is likely to be specified by
a unique combination of factors, including the level of Dpp and Hedgehog signaling,
Fig. 11.27 A model for patterning the
which compartment the cells are in, and the induction or repression of particular
antero-posterior axis of the wing disc. The
decapentaplegic protein is produced at the transcription factors. For example, the lateral vein L5 in the posterior compartment
compartment boundary. The schematic develops at the border between omb and brinker expression domains. It seems, there-
cross-section of the wing disc (X–Y) along fore, that Dpp and Hedgehog do not directly pattern the veins, but set up a series of
the antero-posterior axis shows how the cell–cell interactions that specify positional information across the wing disc, perhaps
presumed gradient in decapentaplegic in even to the level of single cells. It is striking how precise the pattern of veins is if one
both anterior and posterior compartments compares the wings on both sides of a fly.
activates the genes spalt and omb at their
threshold concentrations.
11.19 A signaling center at the boundary between dorsal and ventral
compartments patterns the Drosophila wing along the dorso-ventral axis

The wing disc epithelium is also subdivided into dorsal and ventral compartments
(see Fig. 11.25), which correspond to the dorsal and ventral surfaces of the adult wing
(see Fig. 11.23). These compartments develop after the wing disc is formed—during
the second larval instar. The dorsal and ventral compartments were originally identi-
fied by cell-lineage studies, but they are also defined by expression of the ­homeotic
selector gene apterous, which is confined to the dorsal compartment, and which
­defines the dorsal state. The Apterous protein is a structural relative of Lmx1b from
the mouse, which specifies a dorsal pattern in the mesoderm of the mouse limb bud
(see Section 11.9).
Like the anterior–posterior compartment boundary, the boundary between dorsal
and ventral compartments acts as an organizing center, and Wingless is the signaling
molecule at this boundary (see Fig. 11.25). Wingless is one member of the large Wnt
family of secreted signal proteins, and we have already seen it in action in Drosophila
embryonic development (see Section 2.24). Reduction of wingless function results

Fig. 11.28 Alteration of wing patterning

Anterior Anterior 2
due to ectopic expression of hedgehog
decapentaplegic ectopic hedgehog
and decapentaplegic. When hedgehog 3
hedgehog expression
is expressed in a clone of cells in the
anterior compartment, a new source of 1 3
Decapentaplegic protein is set up. Left panel: 1 2 11 2
in the normal wing, Decapentaplegic protein 2
is made at the compartment boundary,
where Hedgehog is expressed. Right panel: 3 3
the ectopic expression of hedgehog in the
4 4
anterior compartment results in a new source
of Decapentaplegic protein and a new wing 5 compartment 5
boundary
pattern develops in relation to this new
Posterior Posterior
source.
 Insect wings and legs 439

in loss of wing structures; the gene is named after this adult phenotype. Apterous
induces the expression of Serrate, a Notch ligand, in dorsal cells, and restricts the
expression of the other Notch ligand, Delta, to ventral cells. Notch signaling at the
compartment boundary induces expression of Wingless. Wingless in turn induces
expression Delta and Serrate at the boundary, thus maintaining its own expression.
At the early stages of wing-disc development, Apterous also induces expression of the
glycosyltransferase Fringe (see Box 5C, p. 190) in dorsal cells. Fringe promotes Notch
signaling in response to Delta, but inhibits it in response to Serrate, which helps to
localize Notch signaling, and thus Wingless expression, to the compartment bound-
ary (see Fig. 11.25). Apterous is also thought to induce the expression of cell-surface
proteins that, in conjunction with Fringe-modulated Notch signaling, maintain the
boundary of lineage restriction once Apterous itself is no longer expressed. Wingless
serves a role in the dorso-ventral patterning system analogous to Dpp in the ante-
ro-posterior patterning system, and activates expression of various genes at specific
thresholds, which further pattern the wing and are responsible for its outgrowth. One
of these genes is vestigial, which encodes a transcriptional co-activator essential for Fig. 11.29 Expression of wingless
wing development and outgrowth and which is induced in a broad stripe centered on and vestigial in the wing disc. wingless
the dorso-ventral boundary (Fig. 11.29). expression is shown in green, while vestigial
How the proximo-distal axis of the wing blade is patterned remains unclear in is red. Both wingless and vestigial are
Drosophila, but there is evidence for both short- and long-range activation of genes expressed at the dorso-ventral boundary, as
indicated by the yellow stripe.
in the wing blade by Wingless. We shall see later in the chapter how the colorful pat-
tern on butterfly wings is a manifestation of proximo-distal patterning. Photograph courtesy of K. Basler, from
Zecca, M., et al.: 1996.
While both Dpp and Wingless may act as morphogens, whose graded concentra-
tions provide cells with positional information, the formation of these gradients is not
due to simple diffusion. The gradient of activity can also be modified in other ways, as
both Dpp and Wingless can regulate the expression of their receptors. Dpp, for exam-
ple, represses the expression of one of its receptors, Thick veins, so that fewer recep-
tors are present where Dpp levels are high. Wing disc cells also have thin actin-based
extensions called cytonemes that project to the antero-posterior signaling boundary.
These projections may also be involved in long-range cell–cell communication.
As well as the general asymmetry of wing structure, which is due to patterning
along the various axes, the individual cells in the adult Drosophila wing epidermis
become polarized in a proximo-distal direction; this is reflected by the pattern of hairs
(trichomes) on the wing surface. Each of the approximately 30,000 cells in the wing
epidermis produces a single hair, which points distally (see Fig. 11.28, which shows
only those hairs that would be visible on the edge of the wing). This is an example
of planar cell polarity and what is known of the underlying mechanism is outlined in
Box 2F, p. 78.

11.20 The leg disc is patterned in a similar manner to the wing disc, except
for the proximo-distal axis

Insect legs are essentially jointed tubes of epidermis and thus have a quite dif-
ferent structure from those of vertebrates. The epidermal cells secrete the hard
­outer ­cuticle that forms the exoskeleton. Internally, there are muscles, nerves, and
­connective tissues. The easiest way of relating the leg imaginal disc to the adult leg
is to think of it as a collapsed cone. Looking down on the disc, one can imagine it
as a series of concentric rings, each of which will form a proximo-distal segment
of leg. A change in the shape of the epithelial cells is responsible for the outward
extension of the leg at metamorphosis. The process is rather like turning a sock
inside out, with the result that the center of the disc ends up as the distal end, or
tip, of the leg (Fig. 11.30). The outermost ring gives rise to the base of the leg, which
is attached to the body, and the rings nearer the center give rise to the more distal
structures (Fig. 11.31).
440 Chapter 11 Organogenesis

Third instar leg disc is a The disc begins to evert Eversion complete
folded sac

larval cuticle
Distal

Fig. 11.30 Drosophila leg disc extension

at metamorphosis. The disc epithelium,
which is an extension of the epithelium of Proximal
the body wall, is initially folded internally.
larval
At metamorphosis it extends outward, epidermis
rather like turning a sock inside out. The
red arrow in the first panel is the viewpoint
that produces the concentric rings shown in
Fig. 11.31.

The leg disc contains some 30 cells at its initial formation but grows to over
Proximal 10,000 cells by the third instar. The first steps in the patterning of the leg disc along
its antero-posterior axis are the same as in the wing. The engrailed gene is expressed
in the posterior compartment and induces expression of hedgehog. The Hedgehog
protein induces a signaling region at the antero-posterior compartment border. In the
dorsal region of the leg disc, the expression of decapentaplegic is induced, as it is in
the wing. In the ventral region, however, Hedgehog induces expression of wingless
instead of decapentaplegic along the border, and Wingless protein acts as the posi-
tional signal (Fig. 11.32).
Coxa
The patterning of the proximo-distal axis is far better understood in the leg disc
Thorax
than in the wing. It also involves interactions between Wingless and ­Decapentaplegic.
Trochanter The distal end of the proximo-distal axis is the point at which Wingless and Decapen-
taplegic expression meet, and is marked by expression of the homeodomain transcrip-
Femur tion factor Distal-less, which marks what will be the distal tip of the leg. Distal-less is
not used in Drosophila wing development.
Patterning along the proximo-distal axis takes place sequentially and produces
Tibia a two-dimensional pattern in the leg disc that will be converted at metamorphosis
into the three-dimensional tubular leg. Early expression of distal-less is governed by
Tarsus a particular module in the gene’s control region and lasts for just a few hours. Later

Anterior Posterior

engrailed hedgehog decapentaplegic

Dorsal
wingless
Distal Distal-less

Fig. 11.31 Fate map of the leg

imaginal disc of Drosophila. The disc is
a roughly circular epithelial sheet, which
becomes transformed into a tubular leg
at metamorphosis. The center of the disc Ventral proximo-distal
becomes the distal tip of the leg and the anterior posterior axis
circumference gives rise to the base of the compartment compartment
leg—this defines the proximo-distal axis. The
tarsus is divided into five tarsal segments. Fig. 11.32 Establishment of signaling centers in the antero-posterior compartment of
The presumptive regions of the adult leg, the leg disc, and the specification of the distal tip. The gene engrailed is expressed in the
such as the tibia, are thus arranged as a posterior compartment and induces expression of hedgehog. Where Hedgehog protein meets
series of circles with the future tip at the and signals to anterior compartment cells, the gene decapentaplegic is expressed in dorsal
center. A compartment boundary divides the regions and the gene wingless is expressed in ventral regions. Both of these genes encode
disc into anterior and posterior regions. secreted proteins. Expression of the gene Distal-less, which specifies the proximo-distal axis,
After Bryant, P.J.: 1993. is activated where the Wingless and Decapentaplegic proteins meet.
 Insect wings and legs 441

Fig. 11.33 Regional subdivision of the Drosophila leg along the proximo-distal axis. The
pattern of gene expression in the leg is shown on the right. The stages of gene expression are Dpp
shown in the two columns on the left, and are viewed as if looking down on the disc. Because Hth
Dpp coxa
of the way in which the leg extends from the disc, the center corresponds to the future tip DII Wg
of the leg, and the more proximal regions to successive rings around it. decapentaplegic
(dpp) and wingless (wg) are initially expressed in a graded manner along the antero-posterior Wg trochanter
compartment boundary and together induce Distal-less (Dll) in the center and repress
homothorax (hth), which is expressed in the outer region. They then induce dachshund (dac)
in a ring between Dll and hth. Further signaling leads to these domains overlapping. While hth
expression corresponds to proximal regions and Dll to distal regions, there is no simple relation femur
Dac
between the other genes and the leg segments. Dpp
Wg
After Milan, M. and Cohen, S.M.: 2000.
tibia

expression of distal-less in the future distal region is directed by other cis-regulatory

modules. Another homeodomain transcription factor, Homothorax, is expressed in
the peripheral region surrounding Distal-less, which is the future proximal region Hth/Dac/DII
(Fig. 11.33). Their actions lead to the expression of the gene dachshund, which also Dac/DII
encodes a transcription factor, in a ring between Distal-less and homothorax, which tarsus
eventually leads to some overlap of the expression of dachshund with Distal-less and
homothorax. Each of these genes is required for the formation of particular leg ­regions,
but the expression domains do not correspond precisely with leg segments. Expres-
sion of dachshund, for example, corresponds to femur, tibia, and proximal ­tarsus.
There is no evidence for a proximo-distal compartment boundary; cells ­expressing
homothorax and Distal-less are, however, prevented from mixing at the interface
­between their two territories.
Additional patterning of the leg involves a gradient of EGF receptor activity from
distal to proximal. At the third instar larval stage, the EGF receptor ligand Vein and
the signaling pathway component Rhomboid are expressed at the central point of
the disc. Genes coding for transcription factors Bric-a-brac and Bar are expressed at
different levels in the tarsal segments and may determine their identity. The level of
activity of these genes may be determined by the gradient of EGF receptor activity.
Leg joint formation requires Notch signaling. Delta and Serrate, ligands for Notch,
are expressed as a ring in each leg segment and their activation of Notch results in
specification of the joint-forming cells.

11.21 Butterfly wing markings are organized by additional positional fields

The variety of color markings on butterfly wings is remarkable: more than 17,000 spe-
cies can be distinguished. Many of these patterns are variations on a basic ‘ground
plan’ consisting of bands and concentric eyespots (Fig. 11.34). The wings are cov-
ered with overlapping cuticular scales, which are colored by pigment synthesized
and ­deposited by the underlying epidermal cells. How are these patterns specified?
­Butterfly wings develop from imaginal discs in the caterpillar in a similar way to
Drosophila wings. Surgical manipulation has shown that the eyespot is specified at a
late stage in the development of the wing disc, and that the pattern is dependent on a
signal emanating from the center of the spot. A number of the genes that control wing
development in Drosophila, such as apterous, are expressed in the butterfly in a spa-
tial and temporal pattern similar to that in the fly. Thus, as in Drosophila, the shape
and structure of the butterfly wing is patterned by a field of positional information. Fig. 11.34 Butterfly wing pattern. Ventral
The patterning of the pigmentation, however, involves the establishment of additional view of a female African butterfly Bicyclus
fields of positional information. anynana, showing wing color pattern and
Unlike Drosophila wings, development of butterfly wings does involve expression prominent eyespots. Scale bar = 5 mm.
of Distal-less, and it is this that provides the extra dimension of pattern. The expres- Photograph courtesy of V. French and
sion pattern of Distal-less in butterfly wing discs suggests that the mechanism used P. Brakefield.
442 Chapter 11 Organogenesis

to delineate the color pattern on the butterfly wing is similar to the mechanism that
specifies positional information along the proximo-distal axis of the insect leg. In
the butterfly wing, Distal-less is expressed in the center of the eyespot, whereas in the
Drosophila leg disc it is expressed in the central region corresponding to the future
tip of the limb. Thus, it is possible that eyespot development and distal leg patterning
involve similar mechanisms, although in the butterfly wing Notch expression pre-
cedes that of Distal-less. The eyespot may be thought of as a proximo-distal pattern
superimposed on the two-dimensional wing surface. The center of the eyespot may
represent the distal-most positional value, with the surrounding rings representing
progressively more proximal positions, as in the leg. The site of the eyespot may be
specified with reference to the primary wing pattern (that is, the anterior, posterior,
dorsal, and ventral compartments); a secondary coordinate system centered on the
eyespot would then be established, with expression of the Distal-less gene defining
Fig. 11.35 Scanning electron micrograph
the central focus.
of Drosophila carrying the Antennapedia
mutation. Flies with this mutation have 11.22 Different imaginal discs can have the same positional values
the antennae converted into legs (arrows).
Scale bar = 0.1 mm. Patterning of legs and wings involves similar signals, such as Decapentaplegic pro-
Photograph by D. Scharfe, from Science tein, yet the actual pattern that develops is very different. This implies that the
Photo Library. wing and leg discs interpret positional signals in different ways. This interpretation
is under the control of the Hox genes and can be illustrated in regard to the leg and
antenna. If the Hox gene Antennapedia, which is normally expressed in paraseg-
ments 4 and 5 (see Section 2.27) and specifies the discs for the second pair of
legs, is expressed in the head region, the antennae develop as legs (Fig. 11.35). No
homeotic genes are normally expressed in the antennal disc. Using the technique
of mitotic recombination (see Box 2E, p. 73), it is possible to generate a clone of
Antennapedia-expressing cells in a normal antennal disc. These cells develop as
leg cells, but exactly which type of leg cell depends on their position along the
proximo-distal axis; if, for example, they are at the tip, they form a claw. It is as if
the positional values of the cells in the antenna and leg are similar, and the differ-
ence between the two structures lies in the interpretation of these values, which is
governed by the expression or non-expression of the Antennapedia gene. This can
be illustrated with respect to the French flag and the Union Jack (Fig. 11.36)—cells
develop according to both their position and their developmental history, which
determines the genes that they are expressing at any given time. This principle
applies also to wing and haltere imaginal discs. Thus, we find a similarity in devel-
opmental strategy between insects and vertebrates; both use the same positional
information in appendages like legs and wings, and interpret it differently. Still to
be resolved is the question of how the expression of a single transcription factor like
Antennapedia can transform an antenna into a leg. This requires an understand-
ing of its downstream targets. So far, there is evidence that Antennapedia acts as a
repressor of antennal identity in the leg by, for example, preventing the coexpres-
sion of homothorax and Distal-less in the femur region, which occurs in the antenna
but not in the leg disc, where these two genes are expressed in adjacent and non-
overlapping domains (see Fig. 11.33). In evolutionary terms, the antennal state may
Fig. 11.36 Cells interpret their position be the ground state for limb development.
according to their developmental history The character of a disc and how positional information is interpreted is therefore
and genetic make-up. For example, if there
determined by the pattern of Hox gene expression (or the absence of Hox expression
are two flags that use the same positional
in the case of the antennal disc). Insect legs develop only on the three thoracic seg-
information to produce a different pattern,
then a graft from one to the other would
ments and not on the abdominal segments and, in Drosophila, wings develop only
result in the graft developing according to on the second thoracic segment. These adult structures are segment-specific because
its intrinsic pattern, but with the pattern the particular type of imaginal disc that gives rise to them is only formed by cer-
appropriate to its new position. This is what tain parasegments. There are no appendages on abdominal segments in Drosophila
happens in imaginal discs. because the genes required for formation of leg and wing discs are suppressed in the
 Insect wings and legs 443

abdomen. The type of disc formed in a particular thoracic segment is typically speci-
wing disc haltere disc D
fied by the action of one of the Hox genes expressed in the segment. For example,
expression of the genes Antennapedia and Ultrabithorax specifies the second and
third pair of legs, respectively. A P
The leg imaginal discs arise from small clusters of ectodermal cells in paraseg-
ments 3–6, which contribute to the future thoracic segments of the embryo (Fig. 11.37).
In the embryo, each disc initially contains around 25 cells and is formed during growth leg disc
T1 T2 T3
of the blastoderm. The discs arise at the parasegment boundaries, with anterior and thoracic segments V
posterior compartments of adjacent parasegments contributing to each disc. In the
future second thoracic segment, the leg disc splits off a second disc early in its devel- parasegment 4 parasegment 5 parasegment 6

opment, which becomes the wing disc. Similarly, an additional disc, which develops
into the haltere, a balancing organ, is formed in the future third thoracic segment.
Because imaginal discs are formed across parasegment boundaries, mutations in
Hox genes in Drosophila embryos can cause compartment-specific homeotic transfor-
mations of halteres into wings. In normal Drosophila adults, the wing is on the second
thoracic segment and the haltere on the third thoracic segment; they arise from ima-
ginal discs originating at the boundaries of parasegments 4/5 and 5/6, respectively
(see Fig. 11.37). In the normal embryo, the Ultrabithorax gene, one of the genes of
the bithorax complex (see Fig. 2.46), is expressed in parasegments 5 and 6 and is
involved in specifying their identity. The bithorax mutation (bx), which causes the T2 T3
Ultrabithorax gene to be misexpressed, can transform the anterior compartment of
the third thoracic segment, and thus of the haltere, into the corresponding anterior Fig. 11.37 The position in the late
compartment of the second segment: the anterior half of the haltere thus becomes Drosophila embryo of the imaginal
a wing (see Fig. 2.47). The postbithorax mutation (pbx), which affects a regulatory discs that give rise to the adult thoracic
region of the Ultrabithorax gene, transforms the posterior compartment of the haltere appendages. The imaginal discs for the
into a wing (Fig. 11.38). If both mutations are present in the same fly, the effect is legs, wings, and halteres lie across the
additive and the result is a fly that has four wings but cannot fly. Another mutation, parasegment boundaries in the thoracic
segments as shown for the wing and haltere
Haltere mimic, causes a homeotic transformation in the opposite direction: the wing
discs in the lower panel.
is transformed into a haltere.
As with the antenna and leg, it is possible to generate a mosaic haltere with a small
clone of cells containing an Ultrabithorax mutation (such as bithorax) in the disc; the
cells in the clone make wing structures, which correspond exactly to those that would
form in a similar position in a wing. It is as if the positional values in haltere and wing
discs are identical and all that has been altered in the mutant is how this positional
information is interpreted (see Fig. 11.38). In fact, other imaginal discs seem to have
similar positional fields.

Summary
postbithorax mutant
The legs and wings of Drosophila develop from epithelial sheets—imaginal discs—that
T3
are set aside in the embryo. Hox genes acting in the parasegments specify which sort
anterior compartment of haltere
of appendage will form and can direct the interpretation of positional information. The
leg and wing imaginal discs are divided at an early stage into anterior and posterior com-
partments. The boundary between the compartments is a pattern-organizing center and
a source of signals that pattern the disc. In the wing disc, expression of decapentaplegic
is activated by the Hedgehog protein at the anteroposterior compartment boundary and
the Decapentaplegic and Hedgehog proteins act as antero-posterior patterning signals. transformed posterior compartment
The wing disc also has dorsal and ventral compartments, with Wingless produced at the
dorso-ventral boundary acting as a patterning signal. In the leg disc, the antero-­posterior
Fig. 11.38 Effects of the postbithorax
compartment boundary is established in a very similar way, except that Hedgehog acti- mutation in Drosophila. The mutation
vates wingless instead of decapentaplegic in the ventral region and the Wingless protein postbithorax acts on the posterior
acts as the patterning signal in this region. There is no evidence of dorsal and ventral compartment of a haltere, converting it into
the posterior half of a wing.
444 Chapter 11 Organogenesis

compartments in the leg disc. The proximo-distal axis of the leg is specified by the inter-
action between Decapentaplegic and Wingless proteins, which activates genes such as
dachshund and homothorax in concentric domains corresponding to regions along the
leg. The colorful eyespots on butterfly wings may be patterned by a mechanism similar
to that used to organize the proximo-distal axis of the insect leg.

Summary: pattern formation in leg and wing imaginal discs of Drosophila

genes of the HOM complex expressed

character of each disc specified; similar positional values in each disc

Wing

Antero-posterior axis Dorso-ventral axis

anterior and posterior compartments dorsal and ventral compartments

delimited in embryo delimited in second larval instar

decapentaplegic activated by Hedgehog wing margin wingless

at compartment boundary forms at expressed at
compartment compartment
boundary boundary
Decapentaplegic and Hedgehog
proteins act as positional signals
for both compartments boundary controls
wing outgrowth

Leg

Antero-posterior axis Proximo-distal axis

anterior and posterior compartment distal tip of axis defined by Distal-less

delimited; decapentaplegic expressed expression at adjoining domains
in dorsal region, of wingless and decapentaplegic
wingless ventrally; expression; homothorax
Decapentaplegic and Wingless in proximal region
proteins pattern leg disc

Vertebrate and insect eyes

Structures as complex as the compound eyes of insects and the camera eyes of ver-
tebrates are a remarkable achievement of evolution (Fig. 11.39). Camera eyes and
­compound eyes do share some basic similarities. They all contain a lens to focus the
light, a retina composed of light-sensing photoreceptor cells, and a pigmented layer
that absorbs stray light and prevents it interfering with photoreceptor signaling. And
despite the great difference in the anatomy of the final eye, some of the same tran-
scription factors specify eye formation in insects and vertebrates. We shall begin here
with the development of the vertebrate camera eye, which is essentially an offshoot
of the forebrain. In the vertebrate eye, light enters through the pupil at the front of
the eye and passes through a convex transparent lens, which focuses the light on
 Vertebrate and insect eyes 445

Vertebrate eye

pigmented epithelium

cornea neural retina

lens
pupil

iris neurons pigmented

epithelium
photoreceptors

optic nerve

Insect compound eye

Fig. 11.39 Vertebrate and insect
eyes. Vertebrate eyes (upper panel) are
Individual ommatidium ‘camera’ eyes with a single lens that
crystalline focuses light on photoreceptor cells in
pseudocone the neural retina lining the back of the
eyeball. The photoreceptor cells connect
to retinal neurons, whose axons form the
optic nerve, which connects the eye to the
brain. The Drosophila compound eye (lower
panel) is composed of numerous individual
corneal photoreceptor pigment axons light-sensing structures called ommatidia;
lens cells cell a section through one is shown on the right.
Each ommatidium has a lens, through which
light passes to activate the photoreceptor
cells, which are surrounded by pigment cells.

the photosensitive retina lining the back of the eyeball (see Fig. 11.39). Photorecep-
tor cells in the retina—the rods and cones—register the incoming photons and pass
signals on to nerve cells, which transmit them via the optic nerve to the brain, where
they are decoded.

11.23 The vertebrate eye develops from the neural tube and the ectoderm
of the head

In developmental terms, the vertebrate eye is essentially an extension of the forebrain,

together with a contribution from the overlying ectoderm and migrating neural crest
cells. The development of an eye starts at E8.5 in the mouse and around 22 days in
human embryos with the formation of a bulge, or evagination, in the epithelial wall
of the posterior forebrain, in the region called the diencephalon. This evagination is
called the optic vesicle; an optic vesicle is formed on either side of the head and extends
to meet the surface ectoderm (Fig. 11.40). The optic vesicle interacts with the ecto-
derm to induce formation of the lens placode, which is a thickened region of ectoderm
from which the lens will develop. The lens placode is part of a larger region of head
ectoderm that gives rise to the epithelial placodes of some other sensory organs, includ-
ing the placode that gives rise to the semi-circular canals, cochlea, and endolymphatic
duct of the ear, and the placode that gives rise to the olfactory epithelium in the nose.
After induction of the lens placode, the tip of the optic vesicle invaginates to form
a two-layered cup, the optic cup, adjacent to the placode. The inner epithelial layer of
446 Chapter 11 Organogenesis

ectoderm

cavity of
forebrain

lens

presumptive retinal lens neural

pigmented epithelium placode retina

optic lens
vesicle

mesenchyme

surface presumptive presumptive

of ectoderm optic stalk neural retina
of head lens vesicle optic nerve cornea

Fig. 11.40 The main stages in the

development of the vertebrate eye. The
the optic vesicle will form the neural retina, while the outer layer will form the retinal
optic vesicle (blue) develops as an outgrowth pigment epithelium. The lens placode then invaginates and detaches from the surface
of the forebrain and induces the lens ectoderm to form a small hollow sphere of epithelium that will develop into the lens.
placode in the surface ectoderm of the The lens is formed by proliferation of cells of the epithelium on the anterior side of the
head. The optic vesicle invaginates to form lens vesicle, with the new cells moving into the center of the lens, where they start to
a two-layered cup, the optic cup, around the manufacture crystallin proteins. The cells eventually lose their nuclei, mitochondria,
developing lens. The inner layer of the optic and internal membranes to become completely transparent lens fibers filled with crys-
cup forms the neural retina and the outer tallin. Renewal of lens fibers continues after birth, but proceeds much more slowly
layer the pigmented epithelium underlying
than in the embryo. In the adult chicken, the transformation of an epithelial cell into
the retina. The lens vesicle detaches from the
a crystallin-filled fiber takes two years.
surface ectoderm to form the lens and the
The cornea is a transparent epithelium that seals the front of the eye. It is composed
remaining overlying ectoderm forms the
cornea. The iris develops from the rim of the
of inner and outer layers with different developmental origins. The inner layer is
optic cup. formed by mesenchymal neural crest cells that migrate into the anterior eye chamber
Adapted from Adler, R. and Valeria Canto-
to form a thin layer overlying the lens. Neural crest cells also contribute to other
Soler, M.: 2007. structures in the anterior part of the eye. The outer layer of the cornea is derived from
the surface ectoderm adjacent to the eye. Most of the iris develops from the rim of the
optic cup, with neural crest contributing to the anterior iris.
The neural retina in the vertebrate eye develops distinct layers of cells, with the
photoreceptor cells forming the innermost layer, underneath layers of ganglion and
bipolar cells. This is in contrast to the camera eyes of octopus and squid in which the
photoreceptor cells are on the surface of the neural retina. Visual signals are transmit-
ted from the eye to the brain via the optic nerve, which is formed by the axons of the
neural ganglion cells. In Chapter 12 we will see how the neurons of the optic nerve
connect with precise positions in the visual-processing centers in the brain in order to
produce a ‘map’ of the retina.
Although evagination of the optic vesicles does not occur until neural tube closure
is almost completed, the specification of cells as prospective eye cells occurs much
earlier, in the neural plate. Key conserved eye-specifying transcription factors, such
as Pax6, Six3, and Otx2, are expressed in anterior neural plate at late gastrula stages
 Vertebrate and insect eyes 447

as part of the initial antero-posterior patterning of the neural plate. They continue to
be expressed in the optic vesicle epithelium and the lens placode, as well as in other
precursors of sensory organs such as the olfactory placode. Retinal precursor cells
are first specified as a single ‘eye field’ across the center of the anterior neural plate.
The eye field eventually becomes separated into two lateral regions of cells that will
give rise to the optic vesicles after neural tube formation. Separation is achieved by
downregulating expression of Pax6 and the other eye-specifying transcription factors
along the midline. Failure of this separation could be one of the causes of cyclopia in
infants, in which a single abnormal central eye is produced. The normal downregula-
tion of the expression of Pax6 and other transcription factors in the forebrain midline
is likely to be mediated by Shh signaling, which is known to be required for the cor-
rect specification of ventral midline structures. As we saw in Chapter 1 (see Box 1F,
p. 28), a failure of Shh signaling can result in the condition of holoprosencephaly,
which in its most severe form results in failure of the forebrain to divide into right and
left hemispheres, and loss of midline facial structures, resulting in cyclopia.
The same set of transcription factors is essential for eye formation throughout the
animal kingdom. The classic example of a gene with a conserved basic function is
Pax6, which is required for the development of light-sensing structures in all bilate-
rian animals (animals with bilateral symmetry)—from the simple light-sensing organs
of planarians to the compound eyes of insects (described later in this chapter), and
the camera eyes of vertebrates and cephalopods. Pax6 was initially identified from the
genetic analysis of mutations causing abnormal eye development in mice and humans
and is homologous with the Drosophila gene eyeless. Mouse embryos with defective
Pax6 function have smaller eyes than normal, or no eyes at all. People heterozygous
for mutations in Pax6 have a variety of eye malformations collectively known as
aniridia, because of the partial or complete absence of the iris; these patients also
have cognitive defects, as Pax6 has a number of roles in brain development other than
eye formation. In Xenopus, injection of Pax6 mRNA into an animal pole blastomere
at the 16-cell stage leads to formation of ectopic eye-like structures, with well-formed
lenses and epithelial optic cups, in the tadople head. Even more amazingly, the ecto-
pic expression of Pax6 from mouse, Xenopus, ascidians, or squid in Drosophila ima-
ginal discs causes the development of Drosophila-type compound eye ommatidia on
adult structures such as antennae (Fig. 11.41).
Formation of the lens is a crucial step in vertebrate eye development. Among much
other evidence, this is nicely demonstrated by experiments in the teleost fish Astyanax
mexicanus, which has both surface-dwelling sighted forms and cave-dwelling blind
forms—called cavefish (see Fig. 15.18). A small optic primordium with an optic cup and
a lens is formed in cavefish embryos but the lens then undergoes massive apoptosis.
An expansion of Shh signaling along the embryonic midline in the cavefish has been

a b

Fig. 11.41 Pax6 is a master gene for

eye development. The ectopic expression
of mouse Pax6 in Drosophila antennal
disc results in compound eye structures
developing on the antenna.
Photographs from Gehring, W.J.: 2005.
448 Chapter 11 Organogenesis

i­mplicated in lens apoptosis. As a consequence of lens non-development, the cornea, the

cone cell
iris, and other eye structures at the anterior of the eye, fail to develop. Transplantation of
R3 R4
R2 R5
a lens from the surface-dwelling A. mexicanus into a cavefish optic cup during embryo-
R8
R1 R6
genesis reverses lens degeneration, indicating that the absence of a lens is at least partly
R7
responsible for the non-development of the cavefish eye. The effect of expanded Shh
signaling in eye development was investigated by increasing Shh expression in embryos
of surface-dwelling Astyanax, which normally develop into sighted adults. When Shh
mRNA was injected into one side of a cleavage-stage embryo, expression of the eye gene
Pax6 was downregulated unilaterally in the corresponding developing eye region, and
the larvae that developed lacked an eye on that side of the head.

11.24 Patterning of the Drosophila eye involves cell–cell interactions

The Drosophila compound eye is quite different in structure from the vertebrate eye. It
is composed of about 800 identical photoreceptor organs called ommatidia ­(singular
Fig. 11.42 Scanning electron
ommatidium) arranged in a hexagonal array of crystalline regularity (Fig. 11.42).
micrograph of the compound eye of adult
Drosophila. Each unit is an ommatidium. At
Each fully developed ommatidium is made up 20 cells: eight photoreceptor neurons
the third larval instar, the ommatidial cluster (R1–R8), together with four overlying cone cells (which secrete the lens), and eight
(inset) is made up of eight photoreceptor additional pigment cells (not shown in the inset in Fig. 11.42). The genetic analysis
neurons (R1–R8) and four cone cells. of ommatidium development has provided one of the best model systems for study-
Red line indicates the equator of eye. ing the patterning of a small group of cells. An important early finding from lineage
Scale bar = 50 mm. analysis was that the pattern of each ommatidium is specified by cell–cell interactions
and is not based on cell lineages.
The eye develops from the single-layered epithelial sheet of the eye imaginal disc,
located at the anterior end of the larva. Specification and patterning of the cells of
the ommatidia begins in the middle of the third larval instar. Patterning starts at the
posterior of the eye disc and progresses anteriorly, taking about 2 days, during which
time the disc grows eight times larger. One of the earliest events in eye differentiation
is the formation of a groove, the morphogenetic furrow, which sweeps across the
disc from posterior to anterior in response to a wave of signals that initiate develop-
ment of ommatidia from the eye disc cells. As the furrow moves across the epithelium
from posterior to anterior, clusters of cells that will give rise to the ommatidia appear
behind it, spaced in a hexagonal array (Fig. 11.43). The furrow moves slowly across
the disc, at a rate of 2 hours per row of ommatidial clusters. As the morphogenetic fur-
row moves forward, the cells behind it start to differentiate to form regularly spaced
ommatidia. The ommatidia are arranged in rows, with each row half an ommatidium
out of register with the previous one. This gives the characteristic hexagonal packing
arrangement. The first cells to differentiate are the R8 photoreceptor neurons. These
appear as regularly spaced cells in each row, separated from each other by about eight
cells. This separation sets the spacing pattern for the ommatidia.
The passage of the morphogenetic furrow is essential for differentiation of the
ommatidia, as mutations that block its progress also block the differentiation of new
rows of ommatidia, resulting in a fly with abnormally small eyes. Although there
are no anterior and posterior compartments in the eye discs, the cells just behind
the furrow can be regarded as resembling posterior cells, similar to the situation
in the wing-disc posterior compartment (see Section 11.18). They secrete Hedgehog
protein, which triggers the expression of Decapentaplegic, which in turn causes the
cells to become competent to form neural tissue. The gene Wingless also plays a part
in eye disc patterning. It is expressed at the lateral edges of the eye disc, and prevents
the furrow starting in these regions. We thus see that, even though imaginal discs give
rise to very diverse structures, the key signals involved in patterning the leg, wing,
and eye are similar, although they have different roles in each case.
Once specified, each R8 cell initiates a cascade of signals that eventually recruits
a cluster of 20 cells that form the mature ommatidium. The first cells to be recruited
 Vertebrate and insect eyes 449

Eye imaginal disc from third larval instar Differentiation of ommatidia

Cell division No cell division Cell division

Antenna Eye

Morphogenetic
furrow

Anterior Posterior

Prefurrow Furrow Preclusters R8, R2, and R5 R3 and R4 R1 and R6 R7

identifiable added added added
5 5 5 6 5 6 5 6
4 4 4 4
8 8 8 8 7 8 7
3 3 3 3
2 2 2 1 2 1 2 1

Fig. 11.43 Development of ommatidia

are the prospective photoreceptor cells, which are sensory neurons. R2 and R5 dif- in the Drosophila compound eye. The
ferentiate on either side of R8, to form two functionally identical neurons. R3 and R4, compound eye develops from the eye
which are a slightly different type of photoreceptor, are specified next. All these cells imaginal disc, which is part of a larger disc
become arranged in a semi-circle with R8 at the center. R1 and R6 differentiate next that also gives rise to an antenna. During the
third larval instar, the morphogenetic furrow
and almost complete the circle, which is finally closed by the differentiation of R7
develops in the eye disc and moves across it
adjacent to R8 (see Fig. 11.43).
in a posterior to anterior direction. Ommatidia
The clusters rotate 90°, so that R7 comes to be closest to the equator of the disc and develop behind the furrow, the photoreceptor
R3 furthest away; rotation is in the opposite direction in the dorsal and ventral halves neurons being specified in the order
of the eye. This means that the dorsal and ventral regions of the eye have distinct and shown, R8 developing first and R7 last. The
different polarities, with ommatidia on the dorsal and ventral sides of the equator individual ommatidia are regularly spaced in a
having mirror-image symmetry (see Fig. 11.42). The polarization of the ommatidia is hexagonal grid pattern.
yet another example of planar cell polarity (see Box 2F, p. 78) and involves a higher After Lawrence, P.: 1992.
level of Frizzled signaling in R3 than in R4.
The determination of the equator of the eye results from the specification of the dor-
sal region of the eye disc by members of the Iroquois gene complex, which is followed
by specification of the equator itself, involving the actions of Notch, its ligands Serrate
and Delta, and the secreted protein Fringe, in a similar way to the specification of the
dorso-ventral compartment boundary in the wing (see Section 11.19).
The regular spacing of the ommatidia within the eye involves a lateral inhibition
mechanism that spaces the R8 cells. All cells in the eye disc initially have the capacity
to differentiate as R8 cells, and as the morphogenetic furrow passes they start to do
so. But some inevitably gain a lead and are thus able to inhibit the differentiation of
another R8 cell over a range of around three cell diameters. Cells that will give rise to
R8 express the gene atonal. Inhibitors of atonal that space the R8 cells are the secreted
Scabrous protein and Notch.
In the eye, cell fate is specified and determined cell by cell, not in groups of cells.
Two proteins crucial in the patterning of an individual ommatidium are the EGF recep-
tor and its ligand Spitz. One model for patterning the ommatidium is based on both
EGF receptor activation and the age of the cells (Fig. 11.44). Spitz is produced by the
three earliest-specified and most centrally located cells—R8, R2, and R5. It activates
the EGF receptor on neighboring cells, and this recruits R3, R4, R1, R6, and R7 to a
photoreceptor fate. The responding cells also secrete the protein Argos. This diffuses
away and inhibits more distant cells from being activated by Spitz, so that no more
cells in the prospective ommatidial cluster develop as photoreceptors. The actual char-
acter of each photoreceptor may be determined by the age of the cell, with cells ­passing
450 Chapter 11 Organogenesis

through a series of ‘states’, each representing a potential fate. Other signals are also
c involved, however. The difference between R3 and R4, for example, involves Notch
3 4 signaling; the cell with the high level of Notch activity is inhibited from becoming R3,
2 5 and becomes R4. After the photoreceptors have differentiated, the four lens-producing
c 8 c
cone cells develop, and finally the surrounding ring of accessory cells.
1 6
The specification of R7 as a photoreceptor cell is one of the best understood cases
7
c
of the specification of cell fate on an individual cell basis. It requires expression of
the gene sevenless in the prospective R7 and bride-of-sevenless (boss) in R8. When
either gene is inactivated, the phenotype is the same: R7 does not develop and an
extra cone cell is formed. The Sevenless protein is a transmembrane receptor tyrosine
c kinase, and Boss is its ligand. Sevenless protein is produced not only by R7 but also
3 4
by other cells in the ommatidium, including lens cells. Thus, expression of Sevenless
2 5
is a necessary, but not sufficient, condition for R7 specification. Using genetic mosa-
c 8 c
ics it can be shown that for R7 to develop, only R8 need express Boss protein and
1 6
7 that this is the signal by which R8 induces R7. Boss is also an integral membrane
c protein; it is present on the apical surface of the R8 cell, where it makes contact with
R7. A second signal for R7 is provided by the R1/R6 pair, which must activate Notch
in the R7 cell.
Fig. 11.44 Sequential recruitment of
photoreceptors and cone cells during
ommatidial development. The three 11.25 Eye development in Drosophila is initiated by the actions of the same
earliest-specified and most centrally located transcription factors that specify eye-precursor cells in vertebrates
photoreceptor cells—R8, R2, and R5—produce
The Drosophila version of Pax6 is called eyeless, as mutations in this gene result in the
the protein Spitz. It activates the Drosophila
reduction or complete absence of the compound eye. eyeless is expressed in the region
EGF receptor DER on neighboring cells,
of the eye disc anterior to the morphogenetic furrow (see Fig. 11.43). Ectopic expres-
which recruits R3, R4, R1, R6, and R7 to a
photoreceptor fate (top). These then also sion of the eyeless gene in other imaginal discs results in the development of ectopic
produce Spitz, which interacts with DER on eyes, and these have been induced in this way in wings, legs, antennae, and halteres.
the prospective cone cells (c) to recruit them The fine structure of these ectopic eyes is remarkably normal, and distinct ommatidia
to the ommatidium. Once cells are determined are present, although they are not connected to the nervous system. It is estimated
they secrete the protein Argos. This diffuses that some 2000 genes are eventually activated as a result of eyeless activity, and all of
away and inhibits more distant cells from them are required for eye morphogenesis. The mode of action of eyeless in Drosophila
being activated by Spitz, so that no more cells may be similar to that of the Hox genes, as it appears to change the interpretation of
in the prospective ommatidial cluster develop positional information in discs in which it is ectopically expressed.
as photoreceptors.
As we saw earlier, the remarkable conservation of Pax6/eyeless function is dem-
Adapted from Freeman, M.: 1997. onstrated by the fact that Pax6 from the mouse can substitute for eyeless and induce
ectopic eye structures in Drosophila if expressed in the imaginal disc for the antenna
(see Fig. 11.41). Whatever the source of the Pax6 gene, the eye structures that form
are always Drosophila ommatidia, not eyes like those of the animal from which the
Pax6 gene came. Pax6 is acting as a master gene that turns on Drosophila’s own eye
development program.
The discovery of this underlying conservation of function now suggests to many
biologists that the very different eyes found in different groups of animals are exam-
ples of parallel evolution that has adapted the same genetic circuitry and certain cell
specializations that were present in the ancestors of all bilaterian animals, or even
earlier. Animals are thought to have evolved from unicellular cells, probably most
resembling flagellate protozoa (discussed in Section 15.2), which are proposed to have
had a light-sensitive organelle similar to the ‘eyespot’ in some modern flagellates. The
eyespot transmits signals to the flagella and enables the cell to move towards light.
The evolution of multicellularity and of differentiated cell types eventually led to
simple types of photoreceptor cells, such as those in present-day jellyfish larva, which
have individual photosensitive cells in the ectoderm. Jellyfish photoreceptor cells not
only use opsins as the photoreceptor proteins, as do all animals and some unicellular
organisms, but also contain the pigment melanin, which is the pigment in the retinal
pigmented epithelium of the vertebrate eye. The function of the pigment is to absorb
 Internal organs 451

any light that passes through the photoreceptor layer and so prevent it reflecting back
onto the photoreceptors and producing an imprecise signal.
The next step in the evolution of the eye would have been the differentiation of
unicellular light-sensing cells into two cell types: a sensory photoreceptor cell and a
pigment cell. The idea that the origin of the eye involved two such cells goes back to
Charles Darwin, who considered the evolution of the vertebrate eye one of the most
difficult problems in evolution. Simple two-celled ‘pigment cup’ light-sensitive organs
are present in the larvae of many bilaterian animals, while the more complex frontal
eye of the larval amphioxus (a cephalochordate) consists of a pigment cup and two
rows of photoreceptor cells connected by neurons to the central nervous system, fore-
shadowing the eyes of vertebrates. When and how Pax6 became such a crucial factor
in eye development is still unknown.

Summary: development of the ommatidia of the Drosophila eye

eyeless gene expression required for eye development. Morphogenetic furrow moves
across eye disc, associated with hedgehog and decapentaplegic signals

the eight photoreceptors of the future ommatidia begin to develop behind

the furrow, R8 developing first and R7 last.
Ommatidia spaced by lateral inhibition

photoreceptors specified by DER activation, other signals, and time;

R7 specification depends on a signal from R8

Summary

The vertebrate eye develops as an extension of the forebrain. Evagination of the fore-
brain lateral wall produces the optic vesicle, which in turn induces formation of the
lens from the surface ectoderm. Invagination of the tip of the optic vesicle produces
a two-layered optic cup, which surrounds the lens and develops into the eyeball, with
the inner epithelial layer forming the neural retina, and the outer epithelium develop-
ing into the pigmented epithelium at the back of the retina. Development of the lens is
essential for the eye to develop further. The transcription factor Pax6 is essential for
eye development in animals as diverse as vertebrates and Drosophila. The Drosophila
compound eye contains around 800 individual ommatidia, arranged in a regular hexago-
nal pattern, and develops from an imaginal disc. The regular spacing of the ommatidia
is achieved by lateral inhibition by the photoreceptor neuron R8. The patterning of the
eight photoreceptor neurons in each ommatidium is due to local cell–cell interactions, in
which the photoreceptor cells are specified and differentiate in a strict order, initiated
by R8.

Internal organs: insect tracheal system, vertebrate lungs,

kidneys, blood vessels, heart, and teeth

All the structures discussed so far are external, and this has made them relatively
easy to study. We now consider some, mainly vertebrate, internal organs. Their devel-
opment illustrates developmental phenomena we have not yet encountered, such
as the formation and branching of tubular epithelia and the organization of loosely
­organized mesenchyme into epithelium.
452 Chapter 11 Organogenesis

Epithelia are the most common type of tissue organization in animals, and many
organs, such as the lungs, kidneys, and blood vessels, are predominantly composed of
functionally specialized epithelia. Epithelial cells adhere tightly together to form a cell
sheet, which can be single layered, as in the endothelium of capillaries and kidney
tubules, or multilayered, as in skin. A common feature of the organs we shall discuss
first is that the epithelium forms branching tubes.

11.26 The Drosophila tracheal system is a model for branching

morphogenesis

The bodies of multicellular animals contain a multitude of tubes and tubules com-
posed of epithelia, including blood vessels, kidney tubules, and the branched airways
in the mammalian lung. Many of these tubular systems undergo extensive branch-
ing during their development, a phenomenon known as branching morphogenesis.
The development of the Drosophila tracheal system provides an excellent model for
branching morphogenesis and has identified genes controlling the process that also
act in the morphogenesis of the vertebrate lung.
Air enters the tracheal system of the Drosophila larva through openings in the body
wall called spiracles, and oxygen is delivered to the tissues by some 10,000 or so fine
tubules, which develop during embryogenesis from 20 ectodermal placodes (10 on
each side). Each placode produces the tracheal system for a metamere (one lateral
half of a complete segment) in the larva. At the start of germ-band retraction, towards
the end of embryonic development (see Section 2.2), the placode ectoderm invagi-
nates to form a hollow sac of around 80 cells, which gives rise through successive
branching to hundreds of fine terminal branches. Remarkably, the extension of the
sacs to form branched tubes does not involve any further cell proliferation but is all
achieved by directed cell migration, cell rearrangement by intercalation, and changes
in cell shape. As development proceeds, branches from different placodes fuse to form
a body-wide network of interconnected tubes.
The initial invagination of a tracheal placode involves constriction of apical cell
surfaces and changes in cell shape, very similar to the internalization of the Droso-
phila mesoderm (see Fig. 8.21). The first tubules are formed when some cells in the
wall of the sac develop filopodia, which enable them to migrate toward a source of
chemoattractant, drawing an elongating tube of cells behind them (Fig. 11.45). These
tubes then branch by a combination of cell intercalation and remodeling of intercel-
lular junctions to produce secondary tubules composed of single cells joined end to
end, with each cell wrapped around itself to form the tubule lumen. The chemoat-
tractant guiding tube extension is the protein Branchless, which is the Drosophila
version of mammalian FGF. It is expressed in clusters of mesenchymal cells in a pat-
tern determined by the previous antero-posterior and dorso-ventral patterning of the
embryo. Branchless acts via the Drosophila version of an FGF receptor called Breath-
less, which is expressed in the tracheal cells and which is also involved in initiating
branching. Genes such as breathless and branchless were identified by mutations that
disrupted tracheal morphology—hence their names. Another such gene is sprouty,
which is expressed by the tracheal cells. Sprouty protein prevents excess branching
by ­antagonizing Breathless signaling, which prevents the more proximal cells from
forming branches. In the absence of Sprouty, many more secondary branches form
than normal.
As the tubules extend into the larval body, low oxygen levels lead to local expres-
sion of Branchless in the mesenchyme. Breathless signaling in response induces cells
at the tips of the secondary tubules to sprout to form the fine terminal branches of
the tracheal system. Each tip cell forms a much-branched unicellular sprout, with the
lumen of the tubules formed within the cell itself (see Fig. 11.45). Excessive ­terminal
 Internal organs 453

Fig. 11.45 Branching in the Drosophila

Primary branching Seconday branching by Terminal branching by
from tracheal sac cell intercalation unicellular sprouting tracheal system is solely the result of cell
rearrangement and remodeling. In the first
stage of tracheal development, localized
secretion of Branchless protein (blue) causes
cells of the tracheal sac epithelium to form
filopodia and move towards the source of
Branchless, creating a primary branch (left
panel). A secondary branch is formed by
filopodia
cells intercalating between each other and
wrapping round themselves to form a tube of
single cells placed end to end, with a lumen
running through the center (center panel).
At the tip of the secondary branch, the gene
2 cells 1 cell sprouty is induced, and the Sprouty protein
(green) inhibits branching further away
tracheal from the source of Branchless. Tip cells that
sac
will undergo terminal branching are then
specified by Notch–Delta signaling. A tip cell
puts out fine cytoplasmic extensions that
develop a lumen and branch extensively
(right panel). Cells behind the tip are inhibited
by the Notch signaling (red barred lines) from
branching is prevented by Notch–Delta signaling, which assigns tip-cell fate and pre- adopting a tip-cell fate.
vents the more proximal cells becoming tip cells and being able to form terminal
branches.

11.27 The vertebrate lung also develops by branching of epithelial tubes

The paired lungs of vertebrates develop from two buds on each side of the end of
the embryonic trachea (the windpipe) that grow out to form the bronchi. The lung
is one of the internal organs that has left–right asymmetry (see Section 4.7) and this
becomes evident very early in its development. Each bud grows out as a tubular epi-
thelium, which will become a bronchus, and in humans it then makes one further
branch on the right and three on the left to form the bronchioles (Fig. 11.46). These
in turn make finer and finer branches, which terminate in thin-walled sacs of epithe-
lium, the alveoli, where gas exchange occurs—oxygen is taken up by the blood and
carbon dioxide released.
Unlike the tracheae of Drosophila, the outgrowth and branching of tubules in the
vertebrate lung is the result of cell proliferation at the tip of the advancing tube, rather
than cell migration. Nevertheless, sprouting and outgrowth of tubules from the main
tube depends, as in the fly, on interaction of the tubular epithelium with signals
from the surrounding mesenchyme, and many of these are the same as in the fly.
FGF-10 is secreted by the splanchnic mesoderm cells adjacent to the lung epithelium,
and interacts with the receptor FGFR2b, which is related to Drosophila Breathless,
expressed by the lung epithelial cells. As in the fly, activation of FGFR2b induces
expression of Sprouty in the lung cells, which acts as negative feedback to prevent the
more proximal cells in the main tube from forming branches.
Another signaling protein essential for lung development is Shh, which is expressed
in cells at the tips of the extending tubes. In mice with a homozygous loss of Shh
function, the trachea and esophagus fail to separate properly and the primary lung
buds fail to grow and branch, resulting in rudimentary sac-like lungs. Wnt signaling
via the planar polarity pathway (see Box 8C, p. 310) is involved in regulating the prox-
imo-distal polarity of the developing tubules. Mice lacking Wnt-5a have a ­truncated
trachea which branches excessively.
454 Chapter 11 Organogenesis

Branching morphogenesis in human lung

28 days 35 days 42 days 56 days

mesenchyme

trachea

Fig. 11.46 Outgrowth and branching

bronchial bud right main left main
of the bronchial buds of the mammalian bronchus bronchus
lung involve cell proliferation as well as
deformation of an epithelial layer. Top
panels: the lung epithelium develops by
successive branching of the bronchial buds
in response to signals from the surrounding
mesenchyme. Bottom panels: deformation
of a small section of an epithelial sheet
accompanied by cell proliferation initially
forms a pouch. The process is repeated to
form successive branches. Branching in the
developing mammalian lung is accompanied
by cell proliferation.

11.28 The development of kidney tubules involves reciprocal induction by the

ureteric bud and surrounding mesenchyme

The mammalian kidney is mainly composed of a set of convoluted tubules that

receive fluid and salts from the blood, balance its composition, and then convey the
final product—urine—to the ureter, which carries it out of the body. One end of each
tubule ends in a glomerulus; this is the site at which blood capillaries release fluid
and salts that are then taken up by the tubule. The other end of the tubule connects
to the ureter. The kidney is formed from two regions of mesoderm, the ureteric bud
and the metanephric mesenchyme (another name for the mammalian kidney is the
metanephros). The ureteric bud branches off the Wolffian duct, which develops from
a primitive form of kidney, the pronephros, which functions as the adult kidney in
amphibians and fish. In mammals, the Wolffian ducts develop into the vas deferens
in males (see Section 9.13).
The bud is induced to form at the posterior region of the Wolffian duct by the adja-
cent metanephric mesenchyme. Under the influence of the mesenchyme, the ureteric
bud branches to form the proximal portions of the urine-carrying tubules, which are
known as the collecting ducts. The ureteric bud induces the mesenchymal cells to
condense around it and form epithelial structures that develop into renal tubules or
nephrons. Each tubule elongates to form a glomerulus at one end, where filtration of
the blood will occur, while the other end fuses to the ureteric collecting duct connect-
ing to the ureter (Fig. 11.47). The intermediate portion of the tubule is convoluted and
its epithelial cells become specialized for the resorption of ions from urine. The mam-
malian kidney is a good developmental system to study as it can develop in organ
culture; explanted metanephric mesenchyme will form many glomeruli and tubules
over a period of about 6 days, even though a blood supply is absent.
The development of the ureteric bud and the mesenchyme depends on mutual
inductive interactions, neither being able to develop in the absence of the other. The
mesenchyme induces the ureteric bud to grow out by secreting glial-cell-derived neu-
rotrophic factor (GDNF), as well as two members of the FGF family. The site of
 Internal organs 455

pronephros
collecting
Embryo ducts
metanephric
blastema
gut

developing
ureter renal
tubules
ureteric bud

glomerulus
cellular
condensation
ureteric bud tubule formation duct fusion

renal tubule

Fig. 11.47 Kidney development is an

expression of GDNF is critical for positioning ureteric bud development and restricting example of mesenchymal-to-epithelial
it to a single site; positioning of GDNF expression is under the control of the proteins transition. The kidney develops from a loose
Slit2 and its receptor Robo2. As we shall see in Chapter 12, these proteins are also mass of mesenchyme, the metanephric
involved in the guidance of axons in nervous-system development. GDNF in the mes- blastema, which is induced to form tubules
enchyme is the main controller of ureteric bud branching, and its receptor, Ret, is by the ureteric bud. The ureteric bud is
present in the bud. The cellular basis for branching is similar to that described earlier itself induced by the mesenchyme to
for lung development (see Section 11.27). Activation of Ret activates the expression of grow and branch to form the collecting
ducts of the kidney that connect to the
sprouty, which, as in the lung, inhibits branching in the region behind the tip of the
ureter. The mesenchyme cells form cellular
tubule. Knock-out of the gene for either GDNF or Ret results in the absence of ureteric
condensations that become epithelial
bud outgrowth.
tubules, which open into the collecting
As the bud branches it induces the overlying mesenchyme at the tips to condense system formed from the ureteric bud. Each
and to form epithelium that then forms the nephrons. Among the growth factors tubule develops a glomerulus, through which
secreted by the bud and involved in this induction are FGF-2, leukemia inhibitory fac- waste products are filtered out of the blood.
tor (LIF), and TGF-b2. A key protein that must be expressed in the mesenchyme for
tubule formation to occur is the zinc-finger transcription factor WT1. Loss-of-function
mutations in the WT1 gene are associated with Wilm’s tumor, a cancer of the kidney
in children, and the gene is therefore known as the Wilm’s tumor suppressor gene.
A very early response to induction is the expression of the transcription factor
Pax2 in the mesenchyme, which is essential for subsequent tubule development,
probably through its stimulation of the expression of GDNF. The expression of WT1
then increases, the loose mass of mesenchyme cells condenses, and the cells start to
express a matrix glycoprotein, syndecan, on their surface. This condensing stage is
followed by the formation of distinct cellular aggregates, in which the mesenchymal
cells become polarized and acquire an epithelial character and secrete the signaling
protein Wnt-4. Each aggregate then forms an S-shaped tube, which elongates and dif-
ferentiates to form the functional unit of a renal tubule and glomerulus. During this
transition, the composition of the extracellular matrix secreted by the cells changes:
mesenchymal collagen I is replaced by basal lamina proteins, such as collagen IV and
laminin, that are typically secreted by epithelial cells. The adhesion molecules (see
456 Chapter 11 Organogenesis

Box 8B, p. 293) expressed by the cells also change; for example, the N-CAM expressed
by the mesenchymal cells is replaced by E-cadherin in the epithelial cells. Integrins
are also involved in the epithelial–mesenchyme interactions.
The number of nephrons per kidney varies widely, from around 230,000 to
1,800,000. Nephron number is determined by the number of bud branches, and how
branching is terminated is not well understood. Many different proteins appear to be
involved in determining nephron number.

11.29 The vascular system develops by vasculogenesis followed

by angiogenesis

Not surprisingly, the vascular system, including blood vessels and blood cells, is
among the first organ systems to develop in vertebrate embryos, so that oxygen and
nutrients can be delivered to the rapidly developing tissues. The defining cell type
of the vascular system is the endothelial cell, which forms the lining of the entire
circulatory system, including the heart, veins, and arteries. The development of blood
vessels starts with mesodermal cells called angioblasts; these are the precursors of
endothelial cells (Fig. 11.48). Angioblasts undergo a mesenchymal-to-epithelial tran-
sition and assemble into the main vessels of the vasculature, in a process called
­vasculogenesis. The initial vessels are then elaborated into a vascular system ramify-
ing throughout the body by the process of angiogenesis, in which vessels extend and
branch to form venules, arterioles, and extensive networks of capillaries.
Angioblast differentiation into endothelial cells requires the growth factor VEGF
(vascular endothelial growth factor) and its receptors. VEGF is also a potent mitogen
for endothelial cells, stimulating their proliferation. The primary blood vessels, such
as the dorsal aorta and the main veins, arise from angioblasts located in the lateral
mesoderm. VEGF is secreted by axial structures, such as somites, and its expression
is driven by Shh signaling in the notochord. Expression of the Vegf gene is induced
Fig. 11.48 Vasculogenesis and
by lack of oxygen or hypoxia, and thus an active organ using up oxygen promotes its
angiogenesis. First panel: angioblasts
own vascularization.
assemble into simple tubes of endothelium
during vasculogenesis. Second and third
New blood capillaries are formed by sprouting from pre-existing blood vessels (see
panels: the tubes are then elongated Fig. 11.48). The capillary grows by degradation of the extracellular matrix and prolif-
during angiogenesis, which involves the eration of cells at the tip of the sprout. Cells at the tip extend filopodia-like processes
development of filopodia on endothelial that guide and extend the sprout. The response of the tip cells to VEGF is to express the
cells (inset third panel), which enable Notch ligand Delta-like 4, which then activates Notch in adjacent proximal cells. Notch
the cell to move towards a source of a signaling blocks expression of the VEGF receptor and is one of the mechanisms that
chemomattractant (blue circles) resulting in
vessel extension, branching, and growth.

Vasculogenesis Angiogenesis
 Internal organs 457

confines outgrowth to the tip of the tube. This is reminiscent of the Delta–Notch feed-
back mechanism deployed in Drosophila tracheal development (see Section 11.26).
During their development, blood vessels navigate along specific paths towards their
targets, with the filopodia at the leading edge responding to both attractant and ­repellent
cues on other cells and in the extracellular matrix. Cues in the extracellular environ-
ment are provided by proteins of the netrin and semaphorin families, which can act on
the tips of growing blood vessels to block filopodial activity. Netrins and semaphorins
are also involved in guidance of axons (discussed in Chapter 12), and there is a striking
similarity in the signals and mechanisms that guide developing blood vessels and neu-
rons. Blood-vessel morphogenesis also requires modulation of adhesive interactions
between endothelial cells and the extracellular matrix and between the endothelial
cells themselves. Changes in integrin-mediated cell adhesion to the extracellular matrix
through focal contacts are particularly important in this context.
VEGF stimulates endothelial cells to degrade the surrounding basement membrane
matrix and to migrate and proliferate. In the skin of mice, embryonic nerves form a
template that directs the growth of arteries. The nerves secrete VEGF, which may both
attract blood vessels and specify them as arteries. Many solid tumors produce VEGF
and other growth factors that stimulate angiogenesis, and blocking new vessel forma-
tion is a means of reducing tumor growth. The monoclonal antibody bevacizumab
(Avastin), an anti-angiogenic agent that targets VEGF, has been approved for clinical
use in combination chemotherapy for colorectal cancer.
The first tubular structures of the vasculature are formed by endothelial cells and
these vessels are then covered by pericytes and smooth muscle cells. Arteries and
veins are defined by the direction of blood flow, as well as by structural and func-
tional differences. Evidence from lineage tracing shows that angioblasts are speci-
fied as arterial or venous before they form blood vessels; their identities are still
undetermined, however, and they can switch identity. Arterial and venous capillar-
ies join each other in capillary beds, which form the interchange sites between the
arterial and venous blood systems. The cell sorting and guidance molecule ephrin
B2 is expressed in ­arterial blood vessels, while its receptor EphB4 is expressed in
venous vessels, and interaction between them in primitive capillary networks may
be required for endothelial cell sorting out into distinct arterial and venous vessels.
Lymphatic vessels are also developmentally part of the vascular system and originate
by budding from veins. The expression of the homeobox gene Prox1 marks an early
stage of commitment to the lymphatic lineage.
Angiogenesis is not just a property of the embryo. It can occur throughout life and,
properly regulated, is the means of repairing damaged blood vessels. Excessive or
abnormal angiogenesis is, however, the hallmark of many human diseases. These
include cancer, obesity, psoriasis, atherosclerosis, and arthritis. Human disorders
­relating to the nervous system that involve abnormalities in the vasculature include
Alzheimer’s disease, where blockage of the vasculature is a major cause of disease
symptoms, and motor neuron disease.

11.30 The development of the vertebrate heart involves specification of a

mesodermal tube that is patterned along its long axis

The heart is one of the first large organs to form in the embryo. It is of mesodermal
origin and is first established as a single tube consisting of two epithelial layers—the
inner endocardium, which is sheet of endothelium, and the outer myocardium,
which will become the contractile cardiac muscle cells. During development, this
tube becomes divided longitudinally into two chambers: the atrial and ventricular
chambers. A two-chambered heart is the basic adult form in fish, but in higher ver-
tebrates, such as birds and mammals, looping and further partitioning give rise to
the four-chambered heart. In humans, about 1 in 100 live-born infants have some
458 Chapter 11 Organogenesis

congenital heart malformation, while in utero, heart malformation leading to death

of the embryo occurs in between 5 and 10% of conceptions (the different numbers
reflect different studies).
The lateral plate mesoderm is the major source of heart mesoderm. In all verte-
brates, precursor heart cells come to lie in two patches of lateral plate mesoderm
on either side of the midline. They then undergo a complex morphogenesis, move
towards the midline and fuse to form the heart tube (Fig. 11.49). A number of muta-
tions in zebrafish have been found that disrupt this process and result in two laterally
positioned hearts—a condition known as cardia bifida. One of the mutated genes is
called miles apart, and codes for a receptor that binds lysosphingolipids; sphingosine
1-phosphate is the likely ligand in this case. Miles apart is not expressed in the migrat-
ing heart cells themselves, but in cells on either side of the midline, and thus may be
involved in directing the migration of the presumptive heart cells.
As we saw in Chapter 4, heart mesoderm in Xenopus is initially specified by
signals from the organizer during the general process of mesoderm induction and
dorso-­ventral patterning. In the chick and mouse, cells that will eventually become
heart cells ingress through the primitive streak and become part of the lateral plate
mesoderm. In the chick, presumptive heart cells ingress in a roughly similar antero-
­posterior order to their eventual position in the heart, but at this point they are not
yet determined as heart cells nor is the antero-posterior patterning of the presumptive
heart mesoderm fixed. A few hours after ingression, the cells have become committed
to be heart cells and will differentiate into cardiac muscle cells if isolated. The antero-
posterior patterning of the prospective heart mesoderm appears to become set soon
afterwards, just before the formation of the heart tube.
There is some evidence that the heart develops as a modular organ in which each
anatomical region is controlled by a distinct transcriptional program. The regions are
the segments along the heart tube—atria, left ventricle, right ventricle, and ventricular
outflow tract in the mammalian heart (Fig. 11.50). The precursor cells for these regions
appear to acquire specific patterns of gene expression according to their position along
the antero-posterior axis. The homeodomain transcription factor Nkx2.5, for example,
is required for heart development and is one of the first markers of early heart cells;
mutations in the Nkx2.5 gene in mice and humans result in heart abnormalities. Nkx2.5
is expressed in the developing heart in a complex pattern. Seven ­different ­activating
regions and three repressor regions have been identified in the control region of the
Nkx2.5 gene. Activating regions include those that lead to its expression in the entire
heart tube, or just in the right ventricle and outflow tract. Retinoic acid is thought to
act as a signal for antero-posterior positional values in heart development.
In mice, experiments that traced the lineage of cells in the early heart region
revealed the existence of two distinct lineages of heart precursor cells that derived
Fig. 11.49 Early heart development
in the mouse. Left panel: schematic of a 7.5 days 8 days
transverse section through a mouse embryo
neural epithelium D
at the level of the heart at around 7.5 days. forming neural tube
Paired heart rudiments have formed from
lateral plate mesoderm on either side of the
ventral midline. Gut endoderm is shown in anterior
mesoderm
yellow. Right panel: by 8.5 days, the ventral
surface of the embryo has folded to form
blood vessel
the foregut and the body cavity, and the
two heart rudiments have fused to form a notochord foregut
single tubular structure composed of the pericardial
cavity
myocardium, which will form heart muscle, myocardium
and the epithelial endocardium lining the heart
rudiment endocardium
internal surface. The heart is surrounded by V
the pericardial cavity.
 Internal organs 459

Day 15 Day 21 Day 28 Day 50

aortic
sac aorta

pulmonary
vena artery
outflow tract cava
ventricle
right
ventricle

mitral
AVV valve

atrium
tricuspid
valve

left pulmonary
ventricle valve
right atrium left atrium
Right Left

Fig. 11.50 Schematic of human heart

development. By day 15 of human
from a common precursor very early in development. One lineage of cells gives rise embryonic development, cardiogenic
to the primitive left ventricle and outflow tract; all other heart regions are colonized precursors have formed a crescent, as shown
by both lineages. in the first panel, in which the main regions
The later development of the heart involves asymmetric looping of the heart tube. of the heart are already specified. The two
This is not well understood but is related to the left–right asymmetry of the embryo arms of the crescent fuse along the midline
(see Section 4.7). The extracellular matrix molecule plectin is expressed earlier on the to give a linear heart tube, which is patterned
left side compared with the right and this may be as the result of the expression of along the antero-posterior axis with the
regions and chambers of the mature heart
Pitx2 on the left side. The heart tube then forms the separate chambers—four in the
(second panel). After looping (third panel)
case of the mammalian heart. In addition, cells of neural crest origin contribute to
these regions are disposed approximately in
the outflow tracts; they are essential, for example, for the formation of the pulmonary
their eventual positions. Later development
artery and aorta from the single embryonic outflow tract (see Fig. 11.50). Defects in results in further patterning (fourth panel),
these regions account for 30% of congenital heart defects in humans, some of which and the formation of valves between, for
are the result of developmental perturbations in neural crest cell specification and example, the atria and ventricles. AVV,
migration. atrioventricular valve region.
There are remarkable similarities—although we should no longer be surprised— After Srivastava, D. and Olson, E.N.: 2000.
between the genes involved in heart development in Drosophila and in vertebrates.
The homeobox gene tinman is required for heart formation in Drosophila, and is a
homolog of vertebrate Nkx2.5. When vertebrate tinman-like genes are expressed in
Drosophila they can substitute for Drosophila tinman and rescue some of the abnor-
malities caused by lack of normal tinman functions. In Drosophila, dpp maintains
tinman expression in the dorsal mesoderm, while in vertebrates, BMPs (relatives
of Decapentaplegic) have been shown to induce Nkx2.5 and are also able to induce
cardiac cell differentiation. Nkx2.5 or Tinman act together with GATA transcription
factors to turn on cardiac-specific gene expression, although GATA transcription fac-
tors are not expressed exclusively in cardiac precursors.

11.31 A homeobox gene code specifies tooth identity

Teeth develop from a well-defined series of interactions between the oral epithelium
and the mesenchyme of the jaws, which is of neural crest origin. Teeth develop from
areas of thickened oral epithelium called the tooth placodes. The epithelium invagi-
nates and the surrounding mesenchyme condenses, forming a tooth primordium or
tooth germ (Fig. 11.51). In each primordium, a specialized group of cells—the enamel
knot—acts as a signaling center for later development, such as cusp formation. The
mesenchyme forms the dental papilla, which gives rise to the pulp and dentine, while
the enamel is secreted by the epithelial cells. Tooth primordia that give rise to different
460 Chapter 11 Organogenesis

Epithelial thickening Bud stage Cap stage Late cap stage

dental organ oral epithelium enamel knot oral epithelium

oral epithelial thickening

neural crest-derived
mesenchyme condensing
ectomesenchyme dental papilla dental papilla

Fig. 11.51 Schematic diagram of the

stages of tooth development.
types of teeth, such as incisors and molars, are specified at precise positions, and it is
this positioning that we will discuss here.
Early tooth development is controlled by reciprocal interactions, by way of secreted
signals, between the oral epithelium and the underlying mesenchyme. The sites of tooth
formation are determined initially by signals produced by the epithelium. ­BMP-4 is
expressed in the epithelium at all sites of tooth formation, and induces, or maintains,
expression of the homeodomain transcription factor Msx1 in the underlying mesen-
chyme. FGF signaling is also involved in a similar pattern. Shh is expressed in the epi-
thelium at sites of tooth formation throughout tooth development, whereas Wnt-7b is
expressed elsewhere in the oral epithelium except at sites of tooth formation. It is likely
that the formation of supernumerary teeth is the result of increased Shh signaling, which
prevents the death of any supernumerary tooth buds that form, and increased Wnt sig-
naling, which increases the number of tooth buds formed from the oral epithelium.
The reciprocal expression of these latter two signaling molecules establishes the
boundaries between the oral epithelium and the dental epithelium that invaginates
to form the tooth. As tooth-bud development proceeds, the direction of signaling
changes so that signals secreted by the mesenchyme direct the development of the
epithelium; for example, the development of the enamel knot.
In rodents, it has been established that the type of tooth that develops—incisor
or molar—is determined by the expression of homeobox genes in the mesenchyme.
These homeobox genes, which include Barx1, Dlx1, Dlx2, Msx1, and Msx2 among oth-
ers, provide a spatial code in the facial mesoderm (Fig. 11.52), analogous to the Hox

Barx1

Dil1

Msx1

Msx2

Fig. 11.52 The expression domains of

four homeobox genes in the mesoderm
of the mandible before initiation of tooth molar
primordia. The positions at which the
incisors and molars will develop are indicated
in black. incisor
After Ferguson, C.A., et al.: 2000.
 Internal organs 461

gene code along the antero-posterior body axis. Thus, Dlx2 and Barx1 are expressed
in mesenchymal cells that will form molars, while Msx1 and Msx2 are expressed in
cells that will form incisors. In mouse mutants lacking both Dlx2 and Dlx1 (a similar
gene that can substitute for Dlx2 in development), molars do not develop, but ectopic
cartilage structures form at the molar sites. The homeodomain transcription factor
Islet1, which we shall meet again in neural development, is expressed exclusively
in developing incisors. It has a positive regulatory interaction with BMP-4, which
induces Msx1 expression in the incisor region.

Summary

Many internal organs, such as lungs, kidneys, and blood vessels, are formed of tubu-
lar epithelial structures. Tubular structures can be formed de novo by two basic
mechanisms: invagination and remodeling of a pre-existing epithelium, as in the
Drosophila trachea and the vertebrate lung, or by the aggregation of mesenchymal
cells to form a tubular epithelium, as in the initial formation of kidney tubules and
blood vessels. Branching of the Drosophila tracheal system and the airways of the
vertebrate lung are controlled by conserved signals produced by the surrounding
mesenchyme. The vascular system develops from mesenchymal endothelial cells
that assemble during vasculogenesis into tubular epithelial structures that are the
precursors of the large blood vessels. The tubules then extend and branch into finer
vessels in the process of angiogenesis. The development of the mammalian kidney
involves both mesenchymal-to-epithelial transitions and reciprocal induction. The
ureteric bud is induced by the prospective kidney mesenchyme to grow and branch,
while the ureteric bud induces mesenchymal cells to condense together and form
tubules. The vertebrate heart initially forms as a linear tube made up of an inner
endocardium and an outer myocardium. Its later development may be modular and
is controlled by genes similar to those that regulate heart formation in Drosophila.
Teeth develop from an invagination of the oral epithelium and the underlying mes-
enchyme and their positioning and further development is controlled by reciprocal
signaling between epithelium and mesenchyme. A homeobox gene code in the mes-
enchyme specifies tooth identity.

Summary to Chapter 11

The development of organs involves similar processes and, in some cases, the same genes, as
those used earlier in development. The pattern of the vertebrate limb is specified along three
axes, at right angles to each other. Signaling molecules provide positional information, and
interpretation of this information involves Hox genes. The development of Drosophila wings
and legs from imaginal discs depends on positional signals generated at compartment bound-
aries, and some of these signal proteins are similar to those used in vertebrate limb pattern-
ing. The development of internal vertebrate organs such as the vascular system, lungs, and
teeth involves epithelial morphogenesis and, in some cases, branching of tubular structures.

n End of chapter questions e­ xpression of FGFs? Propose a linear gene expression pathway
from Hox gene expression to limb bud formation based on these
Long answer (concept questions) facts.

1. Describe an experiment that showed that FGFs play a ­critical 2. Describe the location and extent of the apical ectodermal ridge
role in the initiation of limb bud formation. What causes the with respect to the three main axes of the limb.
462 Chapter 11 Organogenesis

3. What is the consequence of removing the ridge at different times Multiple choice (factual recall questions)
during limb formation? How do you interpret these results? Cite evi- NB There is only one correct answer to each question.
dence for your interpretation.
1. Cartilage (bone-forming tissue) and connective tissue of the ver-
4. Describe the proposed relationship between FGF and retinoic tebrate limb originate from
acid signaling in specifying identity along the proximo-distal axis in
a) mesodermal cells that migrate into the limb bud from the somites
the developing limb.
b) mesodermal mesenchyme of the limb bud
5. Where in the limb bud is the polarizing region (also known as
the zone of polarizing activity or ZPA)? What properties of this c) the apical ectodermal ridge
region make it an organizing region (see Fig. 11.12)? Through which d) the progress zone
signaling molecule does it exert its effects? What is the evidence for
2. Removal of the apical ectodermal ridge of a limb bud leads to
the importance of this molecule?
a) continued development of the proximal structures, but no forma-
6. What is Gli3? What is its role in specifying digit identity, and tion of new distal structures
through what downstream genes does it appear to exert this role?
b) degeneration of the limb bud
7. What is the consequence of grafting the ectoderm from a left limb
c) regeneration of a new apical ridge from adjacent epidermal
bud onto a right mesodermal core, such that the dorso-ventral axis is
­tissue
reversed without altering the antero-posterior axis? What conclusion
is drawn from this experiment? How are Wnt and Engrailed involved d) regeneration of the entire limb bud from underlying mesoderm
in this patterning, and how do we know? 3. Grafting of a second polarizing region into the anterior region of
8. What are the signals from the dorsal ectoderm and the apical a limb bud results in
ectodermal ridge that are required to maintain Shh expression in the a) formation of additional digits in mirror symmetry with the nor-
polarizing region? In turn, how does Shh maintain signaling from mal digits
the apical ectodermal ridge? b) formation of a second limb bud at the site of the graft
9. What is the origin of the muscle cells in a limb? What is the evi- c) formation of new digits that lack any specific identity
dence that the limb mesoderm controls the pattern of muscles that d) formation of new digits, but only those of anterior identity
forms, so that the musculature is appropriate to the limb?
4. The programmed cell death that separates the digits is dependent
10. How is the identity of the different Drosophila imaginal discs
upon which signaling pathway?
established in the embryo—for example, how is it determined
whether a disc will form a wing, a leg, or some other structure? a) BMP
How does misexpression of Antennapedia in the head region illus- b) FGF
trate this? c) Shh
11. The gene engrailed is expressed in the posterior compartment of d) Wnt
the wing disc in the larva and in the posterior compartment of seg-
ments in the embryo. Compare and contrast the signals ­downstream 5. The Drosophila apterous gene is related to the mouse Lmx1b
of posterior compartment engrailed expression in the wing disc and gene, and both are involved in
in segmentation (refer back to Chapter 2 for information about seg- a) controlling segmentation along the antero-posterior axis
mentation). b) determining whether or not wings are made
12. The Drosophila genes wingless, apterous (Pteron is Greek for c) specifying dorsal identity in their respective appendages
wing), and vestigial can all reduce or eliminate the formation of the d) specifying what structure is made from a given segment
wing when mutated. Based on the roles of these genes in the wing
imaginal disc, propose mechanisms by which these mutations cause 6. The formation of the eyespot on butterfly wings is dependent on
the wing defects. the expression of _____ in the disc, which is not expressed in the
Drosophila wing disc.
13. What type of protein does the gene Pax6 encode and discuss
its role in eye development? What human conditions result from a) Distal-less
defects in the normal function or expression of Pax6? b) Engrailed
14. Describe the developmental events leading to formation of the c) Hedgehog
lens of the eye in vertebrate embryos. What have experiments in d) Wingless
the teleost genus Astyanax contributed to our understanding of the
importance of the lens in eye development? 7. The engineered expression of the mouse Pax6 gene in the
­Drosophila leg disc
15. What is the role of signaling through the EGF receptor, DER, in
a) has no effect, since mouse genes cannot work in insects
development of the Drosophila ommatidial cell cluster?
b) can cause formation of mouse-type eye structures on the leg
16. Summarize the general processes involved in blood-vessel
­formation in vertebrates. What is the role of VEGF in blood-vessel c) will transform the leg disc into a wing disc
formation? d) can cause formation of Drosophila eye structures on the leg
 Internal organs 463

8. Drugs that block VEGF are used for 11.4 Patterning of the limb involves positional information
a) correcting birth defects Niswander, L.: Pattern formation: old models out on a limb. Nat.
b) preventing limb development in mice Rev. Genet. 2003, 4: 133–143.
Towers, M., Tickle, C.: Growing models of vertebrate limb devel-
c) studying eye development in Drosophila
opment. Development 2009, 136: 179–190.
d) treating cancer
11.5 Position along the proximo-distal axis of the limb bud
9. The heart develops from
may be specified is still a matter of debate
a) foregut endoderm
Galloway, J.L., Delgado, I., Ros, M.A., Tabin, C.J.: A reevaluation of
b) invaginations of the ectodermal epithelium X-irradiation-induced phocomelia and proximodistal limb pat-
c) lateral plate mesoderm terning. Nature 2009, 460: 400–404.
d) mesodermal angioblasts Summerbell, D., Lewis, J.H., Wolpert, L.: Positional information in
chick limb morphogenesis. Nature 1973, 244: 492–496.
10. The type of tooth that will form in different regions of the jaw Tabin, C., Wolpert, L.: Rethinking the proximodistal axis of the
is controlled by vertebrate limb in the molecular era. Genes Dev. 2007, 21: 1433–
a) BMP expression 1442.
Therapontos, C., Erskine, L., Gardner, E.R., Figg, W.D., Vargesson,
b) homeobox gene expression
N.: Thalidomide induces limb defects by preventing angio-
c) lateral inhibition genic outgrowth during early limb formation. Proc. Natl Acad.
d) previous patterning in the mesoderm Sci. USA 2009, 106: 8573–8578.
Wolpert, L., Tickle, C., Sampford, M.: The effect of cell killing by
X-irradiation on pattern formation in the chick limb. J. Embryol.
Multiple choice answer key
Exp. Morph. 1979, 50: 175–193.
1: b, 2: a, 3: a, 4: a, 5: c, 6: a, 7: d, 8: d, 9: c, 10: b. Zeller, R., Lopez-Rios, J., Zuniga, A.: Vertebrate limb bud develop-
ment: moving towards integrative analysis of organogenesis.
Nat. Rev. Genet. 2009, 10: 845–858.
n Section further reading
11.6 The polarizing region specifies position along the limb’s
11.1 The vertebrate limb develops from a limb bud antero-posterior axis
Delaurier, A., Burton, N., Bennett, M., Baldock, R., Davidson, D., Litingtung, Y., Dahn, R.D., Li, Y., Fallon, J.F., Chiang, C.: Shh and
Mohun, T.J., Logan, M.P.: The Mouse Limb Anatomy Atlas: an Gli3 are dispensable for limb skeleton formation but regulate
interactive 3D tool for studying embryonic limb patterning. digit number and identity. Nature 2002, 418: 979–983.
BMC Dev. Biol. 2008, 8: 83–89. Riddle, R.D., Johnson, R.L., Laufer, E., Tabin, C.: Sonic hedgehog
Tickle, C.: The contribution of chicken embryology to the under- mediates polarizing activity of the ZPA. Cell 1993, 75: 1401–
standing of vertebrate limb development. Mech. Dev. 2004, 121: 1416.
1019–1029. te Welscher, P., Zuniga, A., Kuijper, S., Drenth, T., Goedemans, H.J.,
Meijlink, F., Zeller, R.: Progression of vertebrate limb develop-
11.2 Genes expressed in the lateral plate mesoderm are ment through SHH-mediated counteraction of GLI3. Science
involved in specifying the position and type of limb 2002, 298: 827–830.
Kawakami, Y., Capdevila, J., Buscher, D., Itoh, T., Rodriguez Esteban, Tickle, C.: Making digit patterns in the vertebrate limb. Nat. Rev.
C., Izpisua Belmonte, J.C.: WNT signals control FGF-­dependent Mol. Cell Biol. 2006, 7: 1–9.
limb initiation and AER induction in the chick embryo. Cell Towers, M., Mahood, R., Yin, Y., Tickle, C.: Integration of growth
2001, 104: 891–900. and specification in chick wing digit-patterning. Nature 2008,
Minguillon, C., Buono, J.D., Logan, M.P.: Tbx5 and Tbx4 are not suf- 452: 882–886.
ficient to determine limb-specific morphologies but have com-
mon roles in initiating limb outgrowth. Dev. Cell 2005, 8: 75–84. Box 11A Positional information and morphogen gradients
Kerszberg, M., Wolpert, L.: Specifying positional information in the
11.3 The apical ectodermal ridge is required for limb embryo: looking beyond morphogens. Cell 2007, 130: 205–209.
outgrowth
Fernandez-Teran, M., Ros, M.A.: The apical ectodermal ridge: 11.7 Sonic hedgehog produced by the polarizing region is
morphological aspects and signaling pathways. Int. J. Dev. Biol. likely to be the primary morphogen patterning the antero-
2008, 52: 857–871. posterior axis of the limb
Mariani, F.V., Ahn, C.P., Martin, G.R.: Genetic evidence that FGFs Riddle, R.D., Johnson, R.L., Laufer, E., Tabin, C.: Sonic hedgehog
have an instructive role in limb proximal-distal patterning. mediates the polarizing activity of the ZPA. Cell 1993, 75: 1401–
Nature 2008, 453: 401–405. 1416.
Niswander, L., Tickle, C., Vogel, A., Booth, I., Martin, G.R.: FGF-4 Yang, Y., Drossopoulou, G., Chuang, P.T., Duprez, D., Marti, E., Bum-
replaces the apical ectodermal ridge and directs outgrowth and crot, D., Vargesson, N., Clarke, J., Niswander, L., ­McMahon, A.,
patterning of the limb. Cell 1993, 75: 579–587. Tickle, C.: Relationship between dose, distance and time
464 Chapter 11 Organogenesis

in Sonic Hedgehog-mediated regulation of anteroposterior 11.10 Development of the limb is integrated by interactions
­polarity in the chick limb. Development 1997, 124: 4393–4404. between signaling centers
Bénazet, J.D., Bischofberger, M., Tiecke, E., Gonçalves, A., Martin,
Box 11B Too many fingers: mutations that affect antero- J.F., Zuniga, A., Naef, F., Zeller, R.: A self-regulatory system of
posterior patterning can cause polydactyly interlinked signaling feedback loops controls mouse limb pat-
Ahn, S., Joyner, A.L.: Dynamic changes in the response of cells terning. Science 2009, 323: 1050–1053.
to positive hedgehog signaling during mouse limb patterning. Niswander, L., Jeffrey, S., Martin, G.R., Tickle, C.: A positive feed-
Cell 2004, 118: 505–516. back loop coordinates growth and patterning in the vertebrate
Harfe, B.D., Scherz, P.J., Nissim, S., Tian, H., McMahon, A.P., Tabin, limb. Nature 1994, 371: 609–612.
C.J.: Evidence for an expansion-based temporal Shh gradient in Pizette, S., Niswander, L.: BMPs negatively regulate structure and
specifying vertebrate digit identities. Cell 2004, 118: 517–528. function of the limb apical ectodermal ridge. Development 1999,
Hill, R.E., Heaney, S.J., Lettice, L.A.: Shh: restricted expression 126: 883–894.
and limb dysmorphologies. J. Anat. 2003, 202: 13–20. Zeller, R., Lopz-Rios, J., Zuniga, A.: Vertebrate limb development:
Lettice, L.A., Hill, A.E., Devenney, P.S., Hill, R.E.: Point mutations moving towards integrative analysis of organogenesis. Nature
in a distant Shh cis-regulator generate a variable regulatory Rev. Genet. 2009, 10: 845–855.
output responsible for preaxial polydactyly. Hum. Mol. Genet.
2008, 17: 978–985. 11.11 Different interpretations of the same positional
signals give different limbs

Box 11C Sonic hedgehog signaling and the primary cilium Krabbenhoft, K.M., Fallon, J.F.: The formation of leg or wing spe-
cific structures by leg bud cells grafted to the wing bud is influ-
Eggenschwiler, J.T., Anderson, K.V.: Cilia and developmental sig-
enced by proximity to the apical ridge. Dev. Biol. 1989, 131:
naling. Annu. Rev. Cell Dev. Biol. 2007, 23: 345–373.
373–382.
Huangfu, D., Anderson, K.V.: Cilia and Hedgehog responsiveness
Logan, M.: Finger or toe: the molecular basis of limb identity.
in the mouse. Proc. Natl. Acad. Sci. USA 2005, 102: 11325–11330.
Development 2003, 130: 6401–6410.
Huangfu, D., Liu, A., Rakeman, A.S., Murcia, N.S., Niswander, L.,
Logan, M., Tabin, C.J.: Role of Pitx1 upstream of Tbx4 in specifica-
Anderson, K.V.: Hedgehog signalling in the mouse requires
tion of hindlimb identity. Science 1999, 283: 1736–1739.
intraflagellar transport proteins. Nature 2003, 426: 83–87.

11.12 Hox genes establish the polarizing region and also

11.8 Transcription factors might specify digit identity provide a code values for limb patterning
Dahn, R.D., Fallon, J.F.: Interdigital regulation of digit identity Charité, J., De Graaff, W., Shen, S., Deschamps, J.: Ectopic expres-
and homeotic transformation by modulated BMP signaling. Sci- sion of Hoxb-8 causes duplication of the ZPA in the forelimb
ence 2000, 289: 438–441. and homeotic transformation of axial structures. Cell 1994, 78:
Suzuki, T., Takeuchi, J., Koshiba-Takeuchi, K., Ogura, T.: Tbx genes 589–601.
specify posterior digit identity through Shh and BMP signal- Goodman, F.R.: Limb malformations and the human HOX genes.
ing. Dev. Cell 2004, 6: 43–53. Am. J. Med. Genet. 2002, 112: 256–265
Suzuki, T., Hasso, S.M., Fallon, J.F.: Unique SMAD1/5/8 activity Kmita, M., Tarchini, B., Zàkàny, J., Logan, M., Tabin, C.J., Duboule, D.:
at the phalanx-forming region determines digit identity. Proc. Early developmental arrest of mammalian limbs lacking HoxA/
Natl. Acad. Sci. USA 2008, 105: 4185–4190. HoxD gene function. Nature 2005, 435: 1113–1116.
Vokes, S.A., Ji, H., Wong, W.H., McMahon, A.P.: A genome-scale Nelson, C.E., Morgan, B.A., Burke, A.C., Laufer, E., DiMambro, E.,
analysis of the cis-regulatory circuitry underlying Shh-medi- Muytaugh, L.C., Gonzales, E., Tessarollo, L., Parada, L.F., Tabin,
ated patterning of the mammalian limb. Genes Dev. 2008, 22: C.: Analysis of Hox gene expression in the chick limb bud.
2651–2663. ­Development 1996, 122: 1449–1466.
Wellik, D.M., Capecchi, M.R.: Hox10 and Hox11 genes are required
11.9 The dorso-ventral axis of the limb bud is controlled by to globally pattern the mammalian skeleton. Science 2003, 301:
the ectoderm 363–367.
Altabef, M., Clarke, J.D.W., Tickle, C: Dorso-ventral ectodermal Zakany, J., Kmita, M., Duboule, D.: A dual role for Hox genes in
compartments and origin of apical ectodermal ridge in devel- limb anterior-posterior asymmetry. Science 2004, 304: 1669–
oping chick limb. Development 1997, 124: 4547–4556. 1672.
Arques, C.G., Doohan, R., Sharpe, J., Torres, M.: Cell tracing reveals Zakany, J., Duboule, D.: The role of Hox genes during verte-
a dorsoventral lineage restriction plane in the mouse limb bud brate limb development. Curr. Opin. Genet. Dev. 2007, 17:
mesenchyme. Development 2007, 134: 3713–3722. 359–366.
Geduspan, J.S., MacCabe, J.A.: The ectodermal control of meso-
dermal patterns of differentiation in the developing chick 11.13 Self-organization may be involved in the development
wing. Dev. Biol. 1987, 124: 398–408. of the limb bud
Riddle, R.D., Ensini, M., Nelson, C., Tsuchida, T., Jessell, T.M., Hardy, A., Richardson, M.K., Francis-West, P.N., Rodriguez, C.,
Tabin, C.: Induction of the LIM homeobox gene Lmx1 by Izpisúa-Belmonte, J.C., Duprez, D., Wolpert, L.: Gene expression,
­Wnt-7a establishes dorsoventral pattern in the vertebrate limb. polarising activity and skeletal patterning in reaggregated hind
Cell 1995, 83: 631–640. limb mesenchyme. Development 1995, 121: 4329–4337.
 Internal organs 465

Box 11D Reaction–diffusion mechanisms c­ ooperative action of Frizzled and heparan sulfate proteoglycan
Kondo, S., Asai, R.: A reaction–diffusion wave on the skin of the receptors. Dev. Biol. 2004, 276: 89–100.
marine angelfish Pomocanthus. Nature 1995, 376: 765–768. Blair, S.S.: Notch signaling: Fringe really is a glycosyltransferase.
Meinhardt, H., Gierer, A.: Pattern formation by local self-activation Curr. Biol. 2000, 10: R608–R612.
and lateral inhibition. BioEssays 2000, 22: 753–760. Crozatier, M., Glise, B., Vincent, A.: Patterns in evolution: veins of
Murray, J.D.: How the leopard gets its spots. Sci. Am. 1988, 258: the Drosophila wing. Trends Genet. 2004, 20: 498–505.
80–87. Entchev, E.V., Schwabedissen, A., González-Gaitán, M.: Gradient
formation of the TGF-b homolog Dpp. Cell 2000, 103: 981–991.
11.14 Limb muscle is patterned by the connective tissue Fujise, M., Takeo, S., Kamimura, K., Matsuo, T., Aigaki, T., Izumi, S.,
Nakato, H.: Dally regulates Dpp morphogen gradient formation
Amthor, H., Christ, B., Patel, K.: A molecular mechanism enabling
in the Drosophila wing. Development 2003, 130: 1515–1522.
continuous embryonic muscle growth—a balance between
Han, C., Belenkaya, T.Y., Wang, B., Lin, X.: Drosophila glypicans
proliferation and differentiation. Development 1999, 126: 1041–
control the cell-to-cell movement of Hedgehog by a dynamin-
1053.
independent process. Development 2004, 131: 601–611.
Hashimoto, K., Yokouchi, Y., Yamamoto, M., Kuroiwa, A.: Distinct
Kruse, K., Pantazis, P., Bollenbach, T., Julicher, F., González-Gáitan,
signaling molecules control Hoxa-11 and Hoxa-13 expression in
M.: Dpp gradient formation by dynamin-dependent endocyto-
the muscle precursor and mesenchyme of the chick limb bud.
sis: receptor trafficking and the diffusion model. Development
Development 1999, 126: 2771–2783.
2004, 131: 4843–4856.
Robson, L.G., Kara, T., Crawley, A., Tickle, C.: Tissue and cellu-
Lawrence, P.: Morphogens: how big is the picture? Nat. Cell Biol
lar patterning of the musculature in chick wings. Development
2001, 3: E151–E154.
1994, 120: 1265–1276.
Lawrence, P.A., Struhl G.: Morphogens, compartments, and pat-
Schweiger, H., Johnson, R.L., Brand-Sabin, B.: Characterization of
tern: lessons from Drosophila? Cell 1996, 85: 951–961.
migration behaviour of myogenic precursor cells in the limb
Matakatsu, H., Blair, S.S.: Interactions between Fat and Daschous
bud with respect to Lmx1b expression. Anat. Embryol. 2004,
and the regulation of planar cell polarity in the Drosophila
208: 7–18.
wing. Development 2004, 131: 3785–3794.
Milán, M., Cohen, S.M.: A re-evaluation of the contributions of
11.15 The initial development of cartilage, muscles, and
­Apterous and Notch to the dorsoventral lineage restriction
tendons is autonomous
boundary in the Drosophila wing. Development 2003, 130:
Kardon, G.: Muscle and tendon morphogenesis in the avian hind ­553–562.
limb. Development 1998, 125: 4019–4032. Morata, G.: How Drosophila appendages develop. Nature Rev. Mol.
Ros, M.A., Rivero, F.B., Hinchliffe, J.R., Hurle, J.M.: Immunohisto- Cell Biol. 2001, 2: 89–97.
logical and ultrastructural study of the developing tendons of Moser, M., Campbell, G.: Generating and interpreting the Brinker
the avian foot. Anat. Embryol. 1995, 192: 483–496. gradient in the Drosophila wing. Dev. Biol. 2005, 286: 647–658.
Muller, B., Hartmann, B., Pyrowolakis, G., Affolter, M., Basler, K.:
11.16 Joint formation involves secreted signals and Conversion of an extracellular Dpp/BMP morphogen gradient
mechanical stimuli into an inverse transcriptional gradient. Cell 2003, 113: 221–233.
Khan, I.M., Redman, S.N., Williams, R., Dowthwaite, G.P., Oldfield, Piddini, E., Vincent, J.P.: Interpretation of the Wingless gradi-
S.F., Archer, C.W.: The development of synovial joints. Curr. Top. ent requires signaling-induced self-inhibition. Cell 2009, 136:
Dev. Biol. 2007, 79: 1–36. 296–307.
Spitz, F., Duboule, D.: The art of making a joint. Science 2001, 291: Rauskolb, C., Correia, T., Irvine, K.D.: Fringe-dependent separation
1713–1714. of dorsal and ventral cells in the Drosophila wing. Nature 1999,
401: 476–480.
11.17 Separation of the digits is the result of programmed Strutt, D.I.: Asymmetric localization of Frizzled and the establish-
cell death ment of cell polarity in the Drosophila wing. Mol. Cell 2001, 7:
Garcia-Martinez, V., Macias, D., Gañan, Y., Garcia-Lobo, J.M., 367–375.
­Francia, M.V., Fernandez-Teran, M.A., Hurle, J.M.: Internu- Tabata, T.: Genetics of morphogen gradients. Nature Rev. Genet.
cleosomal DNA fragmentation and programmed cell death 2001, 2: 620–630.
(apoptosis) in the interdigital tissue of the embryonic chick leg Teleman, A.A., Strigini, M., Cohen, S.M.: Shaping morphogen gra-
bud. J. Cell. Sci. 1993, 106: 201–208. dients. Cell 2001, 105: 559–562.
Zuzarte-Luis, V., Hurle, J.M.: Programmed cell death in the embry- Zecca, M., Basler, K., Struhl, G.: Sequential organizing activities
onic vertebrate limb. Semin. Cell Dev. Biol. 2005, 16: 261–269. of engrailed, hedgehog and decapentaplegic in the Drosophila
wing. Development 1995, 121: 2265–2278.
11.18 Positional signals from compartment boundaries
pattern the wing imaginal disc & 11.19 A signaling center 11.20 The leg disc is patterned in a similar manner to the
at the boundary between dorsal and ventral compartments wing disc, except for the proximo-distal axis
patterns the Drosophila wing along the dorso-ventral axis Emerald, B.S., Cohen, S.M.: Spatial and temporal regulation of the
Baeg, G.H., Selva, E.M., Goodman, R.M., Dasgupta, R., Perrimon, homeotic selector gene Antennapedia is required for the establish-
N.: The Wingless morphogen gradient is established by the ment of leg identity in Drosophila. Dev. Biol. 2004, 267: 462–472.
466 Chapter 11 Organogenesis

Kojima, T.: The mechanism of Drosophila leg development Strutt, H., Strutt, D.: Polarity determination in the Drosophila eye.
along the proximodistal axis. Dev. Growth Differ. 2004, 46: Curr. Opin. Genet. Dev. 1999, 9: 442–446.
115–129. Tomlinson, A., Struhl, G.: Decoding vectorial information from a
McKay, D.J., Estella, C., Mann, R.S.: The origins of the Drosophila gradient: sequential roles of the receptors Frizzled and Notch
leg revealed by the cis-regulatory architecture of the Distalless in establishing planar polarity in the Drosophila eye. Develop-
gene. Development 2009, 136: 61–71. ment 1999, 126: 5725–5738.
Tomlinson, A., Struhl, G.: Delta/Notch and Boss/Sevenless signals
11.21 Butterfly wing markings are organized by additional act combinatorially to specify the Drosophila R7 photoreceptor.
positional fields Mol. Cell 2001, 7: 487–495.
Brakefield, P.M., French, V.: Butterfly wings: the evolution of
development of colour patterns. BioEssays 1999, 21: 391–401. 11.25 Eye development in Drosophila is initiated by the
French, V., Brakefield, P.M.: Pattern formation: a focus on Notch in actions of the same transcription factors that specify eye-
butterfly spots. Curr. Biol. 2004, 14: R663–R665. precursor cells in vertebrates
Gehring, W.J., Kazuho, I.: Mastering eye morphogenesis and eye
11.22 Different imaginal discs can have the same positional evolution. Trends Genet. 1999, 15: 371–377.
values Halder, G., Callaerts, P., Gehring, W.J.: Induction of ectopic eyes by
Carroll, S.B.: Homeotic genes and the evolution of arthropods and targeted expression of the eyeless gene in Drosophila. Science
chordates. Nature 1995, 376: 479–485. 1995, 267: 1788–1792.
Morata, G.: How Drosophila appendages develop. Nature Rev. Mol.
Cell Biol. 2001, 2: 89–97. 11.26 The Drosophila tracheal system is a model for
Si Dong, P.D., Chu, J., Panganiban, G.: Coexpression of the homeo- branching morphogenesis
box genes Distal-less and homothorax determines Drosophila Affolter, M., Caussinus, E.: Branching morphogenesis in Droso-
antennal identity. Development 2000, 127: 209–216. phila: new insights into cell behaviour and organ architecture.
Development 2008, 135: 2055–2064.
11.23 The vertebrate eye develops from the neural tube and Ribeiro, C., Neumann, M., Affolter, M.: Genetic control of cell
the ectoderm of the head intercalation during tracheal morphogenesis in Drosophila.
Chow, R.L., Lang, R.A.: Early eye development in vertebrates. Cell 2004, 14: 2197–2207.
Annu. Rev. Cell Dev. Biol. 2001, 17: 255–296.
Donner, A.L., Lachke, S.A., Maas, R.L.: Lens induction in verte- 11.27 The vertebrate lung also develops by branching of
brates: Variations on a conserved theme of signaling events. epithelial tubes
Semin. Cell Dev. Biol. 2006, 17: 676–685. Chuang, P-T., McMahon, A.P.: Branching morphogenesis of the
Gehring, W.J.: New perspectives on eye development and the lung: new molecular insights into an old problem. Trends Cell
evolution of eyes and photoreceptors. J. Hered. 2005, 96: Biol. 2003, 13: 86–91.
­171–184. Hogan, B.L.M., Kolodziej, P.A.: Molecular mechanisms of tubulo-
Streit, A.: The preplacodal region: an ectodermal domain with genesis. Nat Rev. Genet. 2002, 3: 513–523.
multipotential progenitors that contribute to sense organs and Horowitz, A., Simons, M.: Branching morphogenesis. Circ. Res.
cranial sensory ganglia. Int. J. Dev. Biol. 2007, 51: 447–461. 2008, 103: 784–795.
Takahashi, S., Asashima, M., Kurata, S., Gehring, W.J.: Conserva- Pepicelli, C.V., Lewis, P.M., McMahon, A.P.: Sonic hedgehog regu-
tion of Pax 6 function and upstream activation by Notch signal- lates branching morphogenesis in the mammalian lung. Curr.
ing in eye development of frogs and flies. Proc. Natl Acad. Sci. Biol. 1998, 8: 1083–1086.
USA 2002, 99: 2020–2025.
11.28 The development of kidney tubules involves reciprocal
11.24 Patterning of the Drosophila eye involves cell–cell induction by the ureteric bud and surrounding mesenchyme
interactions Barasch, J., Yang, J., Ware, C.B., Taga, T., Yoshida, K., Erdjument-
Baonza, A., Casci, T., Freeman, M.: A primary role for the epi- Bromage, H., Tempst, P., Parravicini, E., Malach, S., Aranoff, T.,
dermal growth factor receptor in ommatidial spacing in the Oliver, J.A.: Mesenchymal to epithelial conversion in rat meta-
Drosophila eye. Curr. Biol. 2001, 11: 396–404. nephros is induced by LIF. Cell 1999, 99: 377–386.
Bonini, N.M., Choi, K.W.: Early decisions in Drosophila eye mor- Gao, X., Chen, X., Taglienti, M., Rumballe, B., Little, M.H., Kreidberg,
phogenesis. Curr. Opin. Genet. Dev. 1995, 5: 507–515. J.A.: Angioblast-mesenchyme induction of early kidney devel-
Chou, W.H., Huber, A., Bentrop, J., Schulz, S., Schwab, K., Chadwell, opment is mediated by Wt1 and Vegfa. Development 2005, 132:
L.V., Paulsen, R., Britt, S.G.: Patterning of the R7 and R8 photo- 5437–5449.
receptor cells of Drosophila: evidence for induced and default Grieshammer, U., Le, Ma, Plump, A.S., Wang, F., Tessier-Lavigne,
cell-fate specification. Development 1999, 126: 607–616. M., Martin, G.R.: SLIT2-mediated ROBO2 signaling restricts
Frankfort, B.J., Mardon, G.: R8 development in the Drosophila eye: kidney induction to a single site. Dev. Cell 2004, 6: 709–717.
a paradigm for neural selection and differentiation. Develop- Kuure, S., Vuolteenaho, R., Vainio, S.: Kidney morphogenesis: cel-
ment 2002, 129: 1295–1306. lular and molecular regulation. Mech. Dev. 2000, 94: 47–56.
Freeman, M.: Cell determination strategies in the Drosophila eye. Schedl, A., Hastie, N.D.: Cross-talk in kidney development. Curr.
Development 1997, 124: 261–270. Opin. Genet. Dev. 2000, 10: 543–549.
 Internal organs 467

Shah, M.M., Sampogna, R.V., Sakurai, H., Bush, K.T., Nigam, S.K.: 11.30 The development of the vertebrate heart involves
Branching morphogenesis and kidney disease. Development specification of a mesodermal tube that is patterned along its
2004, 131: 1449–1462. long axis
Shakya, R., Watanabe, T., Costantini, F.: The role of GDNF/Ret sig- Brand, T.: Heart development: molecular insights into cardiac
naling in ureteric bud cell fate and branching morphogenesis. specification and early morphogenesis. Dev. Biol. 2003, 288:
Dev. Cell 2005, 8: 65–74. 1–19.
Vainio, S., Muller, U.: Inductive tissue interactions, cell signal- Carmeliet, P.: Angiogenesis in health and disease. Nat Med. 2003,
ing, and the control of kidney organogenesis. Cell 1997, 90: 9: 653–660.
975–978. Driever, W.: Bringing two hearts together. Nature 2000, 406: 141–
142.
11.29 The vascular system develops by vasculogenesis
Harvey, R.P.: Patterning the vertebrate heart. Nat Rev. Genet. 2002,
followed by angiogenesis
3: 544–556.
Carmeliet, P.: Angiogenesis in health and disease. Nat Med. 2003, Kelly, R.G., Buckingham, M.E.: The anterior heart-forming field:
9: 653–660. voyage to the arterial pole of the heart. Trends Genet. 2002, 18:
Carmeliet, P., Tessier-Lavigne, M.: Common mechanisms of nerve 210–216.
and blood vessel wiring. Nature 2005, 436: 193–200. Linask, K.K., Yu, X., Chen, Y., Han, M.D.: Directionality of heart
Coultas, L., Chawengsaksophak, K., Rossant, J.: Endothelial cells looping: effects of Pitx2c misexpression on flectin asymmetry
and VEGF in vascular development. Nature 2005, 438: 937– and midline structures. Dev. Biol. 2002, 246: 407–417.
945. Meilhac, S.M., Esner, M., Kelly, R.G., Nicolas, J.F., Buckingham,
Harvey, N.L., Oliver, G.: Choose your fate: artery, vein or lym- M.E.: The clonal origin of myocardial cells in different regions
phatic vessel? Curr. Opin. Genet. Dev. 2004, 14: 499–505. of the embryonic mouse heart. Dev. Cell 2004, 6: 685–698.
Hogan, B.L.M., Kolodziej, P.A.: Molecular mechanisms of tubulo- Moorman, A.F., Christoffels, V.M.: Cardiac chamber formation:
genesis. Nat Rev. Genet. 2002, 3: 513–523. development, genes and evolution. Physiol Rev. 2003, 83: 1223–
Jain, R.K.: Molecular recognition of vessel maturation. Nat Med. 1267.
2003, 9: 685–692. Rosenthal, N., Xavier-Neto, J.: From the bottom of the heart:
Larrivée, B., Freitas, C., Suchting, S., Brunet, I., Eichmann, A.: Guid- anteroposterior decisions in cardiac muscle differentiation.
ance of vascular development: lessons from the nervous sys- Curr. Opin. Cell Biol. 2000, 12: 742–746.
tem. Circ. Res. 2009,104: 428–441. Srivastava, D., Olson, E.N.: A genetic blueprint for cardiac devel-
Lu, X., Le Noble, F., Yuan, L., Jiang, Q., De Lafarge, B., Sugiyama, opment. Nature 2000, 407: 221–226.
D., Breant, C., Claes, F., De Smet, F., Thomas, J.L., Autiero, M., Stainier, D.Y.R.: Zebrafish genetics and vertebrate heart forma-
Carmeliet, P., Tessier-Lavigne, M., Eichmann, A.: The netrin tion. Nature Rev. Genet. 2001, 2: 39–48.
receptor UNC5B mediates guidance events controlling morpho-
genesis of the vascular system. Nature 2004, 432: 179–186. 11.31 A homeobox gene code specifies tooth identity
Nelson, K.S., Beitel, G.J.: More than a pipe dream: uncovering Cobourne, M.T., Sharpe, P.T.: Making up the numbers: the molec-
mechanisms of vascular lumen formation. Dev. Cell 17: 435– ular control of mammalian dental formula. Semin. Cell Dev.
436. Biol. 2010, 21: 314–324.
Rafii, S., Lyden, D: Therapeutic stem and progenitor cell trans- Ferguson, C.A., Hardcastle, Z., Sharpe, P.T.: Development and pat-
plantation for organ vascularization and regeneration. Nat terning of the dentition. Linn. Soc. Symp. 2000, 20: 188–201.
Med. 2003, 9: 702–712. Mitsiadis, T.A., Angeli, I., James, C., Lendahl, U., Sharpe, P.T.: Role
Reese, D.E., Hall, C.E., Mikawa, T.: Negative regulation of mid- of Islet1 in the patterning of murine dentition. Development
line vascular development by the notochord. Dev. Cell 2004, 6: 2003, 130: 4451–4460.
699–708. Sarkar, L., Cobourne, M., Naylor, S., Smalley, M., Dale, T., Sharpe,
Roman, B.L., Weinstein, B.M.: Building the vertebrate vasculature: P.T.: Wnt/Shh interactions regulate ectodermal boundary for-
research is going swimmingly. BioEssays 2000, 22: 882–893. mation during mammalian tooth development. Proc. Natl. Acad.
Rossant, J., Hirashima, M.: Vascular development and pattern- Sci. USA 2000, 97: 4520–4524.
ing: making the right choices. Curr. Opin. Genet. Dev. 2003, 13: Tucker, A., Sharpe, P.: The cutting-edge of mammalian develop-
408–412. ment: how the embryo makes teeth. Nat. Rev. Genet. 2004, 5:
Strilić , B., Kucera, T., Eglinger, J., Hughes, M.R., McNagny, K.M., 499–508.
Tsukita, S., Dejana, E., Ferrara, N., Lammert, E.: The molecular
bases of vascular lumen formation in the developing mouse
aorta. Dev. Cell 2009, 17: 505–515.
Wang, H.U., Chen, Z-F., Anderson, D.J.: Molecular distinction and
angiogenic interaction between embryonic arteries and veins
revealed by ephrin-B2 and its receptor Eph-B4. Cell 1998, 93:
741–753.
12
Development of the
nervous system
nn Specification of cell nn Axon navigation nn Synapse formation and
identity in the nervous refinement
system

The nervous system is the most complex of all the organ systems in the animal
­embryo. In mammals, for example, billions of nerve cells, or neurons, develop a highly
organized pattern of connections, creating the neuronal network that makes up the
functioning brain and the rest of the nervous system. There are many hundreds of
different types of neurons, differing in identity and the connections they make, even
though they may look quite similar. Understanding the developmental processes that
shape the nervous system is not simple, as they involve neuronal cell differentiation,
morphogenesis, and migration. Neurons send out long processes from the cell body
and these must be guided to find their targets. The picture is further complicated
by the activity of the nerve cells, which can refine the pattern of connections they
make.

All nervous systems, vertebrate and invertebrate, provide a system of communica-

tion through a network of electrically excitable cells—the nerve cells or neurons—of
varying sizes, shapes, and functions (Fig. 12.1). The nervous system also contains
non-neuronal cells known collectively as glia, which have a variety of supporting
roles. The Schwann cells in the peripheral nervous system are glial cells, as are the
oligodendrocytes and astrocytes of the central nervous system. Neurons connect with
each other and with other target cells, such as muscle, at specialized junctions known
as synapses. A neuron receives input from other neurons through its highly branched
dendrites and, if the signals are strong enough to activate the neuron, it generates
a new electrical signal—a nerve impulse, or action potential—at the cell body. This
electrical signal is then conducted along the axon to the axon terminal, or nerve
ending, which makes a synapse with the dendrites or cell body of another neuron or
with the surface of a muscle cell. The dendrites and axon terminals of individual neu-
rons can be extensively branched, and a single neuron in the central nervous system
can receive as many as 100,000 different inputs. At a synapse, the electrical signal is
converted into a chemical signal, in the form of a chemical neurotransmitter, which is
released from the nerve ending and acts on receptors in the membrane of the oppos-
ing target cell to generate or suppress a new electrical signal. The nervous system can
only function properly if the neurons are correctly connected to one another, and thus
 Specification of cell identity in the nervous system 469

a central question surrounding nervous-system development is how the connections

between neurons develop with the appropriate specificity. The number of neurons in
the human brain seems generally to be estimated at around 100 billion. How many of
them have unique or similar identities is not known.
In this chapter, we shall use examples from both invertebrate and vertebrate cen-
tral nervous systems and this raises the question: do the central nervous systems
of vertebrates and invertebrates trace back to a common precursor or are they of
independent evolutionary origin? Anatomically, a central nervous system is a spa-
tially delimited nervous tissue composed of neurons with specialized functions that
are interconnected by highly organized tracts of axons. The central nervous system
receives sensory inputs from the body, processes them, and integrates them centrally,
and produces outputs that guide behavior, such as movement. Such an organization
broadly defines the central nervous systems of animals as far apart evolutionarily as
annelid worms and mammals, and is in contrast to the diffuse nervous system, or
nerve net, of cnidarians (Hydra and sea anemones), which receives sensory input and
processes output without central integration.
One important aspect of central nervous-system development in both vertebrates
and invertebrates is the early segregation of the ectoderm into a ‘non-neural’ epider-
mal portion and a ‘neural’ portion—the neuroectoderm (see Section 5.2). In the Bila-
teria, that is, all animals with bilateral symmetry, the neuroectoderm is located at the
anterior end of the embryo, where the brain and associated sensory organs develop.
Similarities can be traced between the embryonic central nervous systems of simple
invertebrates and those of vertebrates. In the annelid worm Platynereis, for example,
the neuroectoderm is subdivided longitudinally into partially overlapping domains of
progenitor neurons that correspond to similar domains in the vertebrate neural tube
axon
and give rise to similar cell types as those of vertebrates. In vertebrates, as discussed
later in this chapter, the genes that pattern the central nervous system are sensitive
to signaling by members of the bone morphogenetic protein (BMP) family, and this is cell body
also the case in Platynereis. All this evidence suggests that a central nervous system axon
dendrites terminal
architecture was present in the last common ancestor of the bilaterians, and supports
a common origin for the nervous system in Bilateria.
Fig. 12.1 Neurons come in many shapes
For all its complexity, the nervous system is the product of the same kind of cel-
and sizes. Upper panel: a montage of the
lular and developmental processes as those involved in the development of other
various types of neurons found in the optic
organs. The overall process of nervous-system development can be divided into lobe of the avian brain. In the brain there are
four major stages: the specification of neural cell (neuron or glial cell) identity; the many more nervous connections that are not
migration of neurons and the outgrowth of axons to their targets; the formation of shown here. Lower panel: a single neuron
synapses with target cells, which can be other neurons, muscle or gland cells; and consists of a small, rounded cell body, which
the refinement of synaptic connections through the elimination of axon branches contains the nucleus, and from which extend
and cell death. a single axon and a ‘tree’ of much-branched
We have already considered some aspects of the development of the vertebrate dendrites. Signals from other neurons are
nervous system, including neural induction (Chapter 5), the formation of the neural received at the dendrites, processed, and
tube (Chapter 8), the segmental arrangement of sensory dorsal root ganglia (Chap- integrated in the cell body, and an outgoing
signal—a nerve impulse—is generated and
ter 8), and the differentiation of neurons from neural crest cells (Chapter 10). We will
sent down the axon. We shall consider
revisit some of these events when we consider the specification of neural-cell identity.
in more detail how neurons signal and
A short chapter, such as this one, cannot cover all aspects of nervous-system develop- communicate with each other in the last part
ment, and the general focus here is on the mechanisms that control the identity of of the chapter.
cells in the nervous system and that pattern their synaptic connections.

Specification of cell identity in the nervous system

Prospective neural tissue—the neuroectoderm—is distinguished from the rest of

the ectoderm as part of the early patterning of invertebrate and vertebrate embryos,
as discussed in previous chapters (Chapters 2 and 5). We will start this chapter by
470 Chapter 12 Development of the nervous system

epidermis
c­ onsidering the next stage in the process of generating a nervous system: how indi-
Dorsal
vidual cells in the neuroectoderm become specified first as neural stem cells, or
­neuroblasts, and subsequently give rise to neurons and glia (see Chapter 10). We will
first consider Drosophila, which has revealed some key developmental processes in
neurogenesis, or the formation of neurons. We then turn to the more complex pro-
cess of neurogenesis in the vertebrate neural tube. As the epithelial cells of the neural
mesoderm
tube proliferate and differentiate into neurons, the wall of the tube develops a multi-
layered structure, with the neurons migrating outwards to form the different layers.
We shall look at the formation of this layered structure in relation to the mammalian
cortex, which develops from the forebrain region of the neural tube.
The nervous systems of both invertebrates and vertebrates also contain enormous
numbers of glial cells as well as neurons. These also develop from the neural stem
neuroectoderm cells of the neuroectoderm, but we shall not consider the specification and develop-
Ventral ventral midline ment of glial cells in any detail here.

Fig. 12.2 Cross-section of a Drosophila

early gastrula. At around 3 hours after 12.1 Neurons in Drosophila arise from proneural clusters
fertilization, the neuroectoderm (blue) is
In Drosophila, the cells that will give rise to the central nervous system of the larva are
located on either side of the ventral midline.
The mesoderm (red), which originally lies
specified at an early stage of embryonic development, as part of the patterning that
along the ventral midline, has already been divides the embryo into different regions along the dorso-ventral and antero-posterior
internalized. axes (see Section 2.19). In insects, the main nerve cord runs ventrally, rather than dor-
sally as in vertebrates, and the future central nervous system is specified as two lon-
gitudinal regions of neuroectoderm in the ventral half of the embryo, just above the
mesoderm. The neuroectoderm, or neurogenic zone, comprises ectodermal cells that
are fated to form either neural cells (neurons or glia) or epidermis. After gastrulation
and internalization of the mesoderm, the neuroectoderm remains on the outside of
the embryo on either side of the ventral midline (Fig. 12.2). Once individual cells are
specified as neuroblasts in the neuroectoderm they move from the surface into the
interior of the embryo, where they divide further and differentiate into neurons and
glia, forming two tracts of axons running longitudinally on either side of the ventral
midline, connected at intervals by neurons that cross the midline.
Despite very great differences between the insect central nervous system and that of
vertebrates, there are intriguing parallels between the genes that pattern both systems
at a very early stage. Drosophila neuroblasts form three longitudinal columns of cells
on each side of the ventral midline, and the identity of these columns is specified while
the cells are still in the neuroectoderm (Fig. 12.3). The homeodomain transcription
factor genes msh, ind, and vnd are expressed in dorsal to ventral order in the Droso-
phila neuroectoderm, in response to the actions of earlier dorso-ventral patterning
genes, such as rhomboid (see Section 2.19), and the activation of the Drosophila EGF
signaling pathway. The developing vertebrate neural plate also has three longitudinal
domains of prospective neural cells (see Fig. 12.3), which express genes homologous
to those in the fly—Msx, Gsh, and Nkx2, respectively. This is a striking example of the
evolutionary conservation of a mechanism for regional specification.
In response to previous patterning, the Drosophila neuroectoderm becomes sub-
divided along the antero-posterior and dorso-ventral axes into a precise orthogonal
pattern of proneural clusters, each composed of three to five cells, which can be dis-
tinguished by the expression of genes known as proneural genes (Fig. 12.4). Initially,
all the cells within a proneural cluster are capable of becoming neural precursor cells,
or neuroblasts, but one cell, through an apparently random event, produces a signal
that promotes its own development as a neuroblast and prevents the adjacent cells
from following that fate. This cell will become the single neuroblast in the cluster, and
can be distinguished, for example, by a high level of expression of a proneural gene
such as achaete, whereas the surrounding cells no longer express achaete and will
develop as epidermis (see Fig. 12.4).
 Specification of cell identity in the nervous system 471

Drosophila Mouse

Dorsal Dorsal

Neural plate
Msx Msx
Gsh Gsh
Nkx2.1 Nkx2.1
msh msh
ind ind Fig. 12.3 The neuroepthelium of both
vnd vnd Ventral
Ventral Drosophila and vertebrates is organized
into three columns of neural precursors on
either side of the midline. The developing
Neural precursor delamination Neurulation Drosophila embryonic central nervous system
(left) and the vertebrate embryonic neural
plate (right) are both organized into three
longitudinal domains (columns) of gene
expression on either side of the midline.
The cells of each column express a specific
Msx Msx homeobox gene and, as is evident before
neural neurulation (upper panels), similar genes are
precursors Gsh Gsh
expressed in the same medial–lateral order
msh msh Nkx2.1 Nkx2.1 in Drosophila and in vertebrates, with vnd/
ind ind Nkx2.1 expressed most medially (nearest the
vnd vnd Neural tube midline), ind/Gsh in an intermediate position,
and msh/Msx expressed most laterally.

Fig. 12.4 Proneural clusters in the

neuroectoderm of the Drosophila
embryo. At the stage of development shown
here, there are eight proneural clusters
within each future segment, marked here by
expression of the proneural gene achaete
(black). Half of the clusters are formed in
the region of engrailed expression (purple)
at the posterior end of the segment, and
half in a more anterior region. Some of the
A P
clusters have resolved into a single neural
cell, the prospective neuroblast. The embryo
is viewed here from the ventral surface; the
ventral midline runs horizontally along the
segment middle of the photo.
Photograph from Skeath, J.B., et al.: 1996.

The singling out of the prospective neuroblast is an example of lateral inhibition

(Fig. 12.5) and is due to signaling between the prospective neuroblast and its neigh-
bors via the transmembrane protein Notch and its transmembrane ligand Delta (see
Box 5C, p. 190 for the Notch–Delta signaling pathway). All the cells in a proneural
cluster initially can express both Notch and Delta. One cell will, however, start to
express Delta sooner or more strongly than the others. Through the interaction of
Delta with the Notch receptors on its less-advanced neighbors, this cell inhibits their
further development as neural cells, and also suppresses their expression of Delta
472 Chapter 12 Development of the nervous system

Proneural cluster Lateral inhibition Neuronal precursor

Fig. 12.5 The role of Notch signaling in

lateral inhibition in neural development
in Drosophila. As shown in the top row,
the proneural cluster gives rise to a single
neuronal precursor cell, the neuroblast
or sensory organ precursor, by means of
lateral inhibition. The rest of the cells in the
cluster become epidermal or support cells. As
shown in the bottom panels, Notch signaling
between cells of the proneural cluster is
Enhancer-of-split proteins bHLH proteins
initially similar and keeps the cells in the
proneural state. An imbalance develops when
Achaete–Scute Surpressor- Achaete–Scute Achaete–Scute
one cell begins to express higher levels of proteins of-hairless proteins proteins
Delta, the ligand for Notch, and thus activates
Notch to a higher level in neighboring cells.
This initial imbalance is rapidly amplified by a Delta Delta
feedback pathway involving two transcription Delta Notch
factors, Suppressor-of-hairless and Enhancer-
Notch Delta
of-split. Their activities repress the production Notch Notch
of Achaete and Scute proteins and Delta
protein in the affected cells. This prevents
them from proceeding along the pathway of Achaete-Scute Achaete–Scute Surpressor- Achaete–Scute Surpressor-
neuronal development and from delivering proteins proteins of-hairless proteins of-hairless
inhibitory Delta signals to the prospective
neuroblast. bHLH, basic helix–loop–helix. Enhancer-of-split proteins Enhancer-of-split proteins
After Kandel, E.R., et al.: 2000.

(see Fig. 12.5, ­bottom row). In this way, the initially equivalent cells of the proneural
cluster are resolved by lateral inhibition into one neuroblast per cluster. The other
cells of the proneural cluster cease expression of Delta and of the proneural genes
and go on to become epidermis. In the Drosophila ectoderm, if either Notch or Delta
function is inactivated, the neuroectoderm makes many more neuroblasts and fewer
epidermal cells than normal.
After specification, neuroblasts enlarge, leave the epithelium and move into the
interior of the embryo, where they divide repeatedly and give rise to neurons and glia,
as described in the next section. In Drosophila, neuroblasts are specified and exit the
neuroectoderm in five waves over a period of around 90 minutes, to form an invariant
pattern of approximately 30 neuroblasts in each hemisegment (a lateral half-segment)
arranged in three longitudinal columns. Each neuroblast has a unique identity based
on the particular embryonic segment it is part of, its position within the segment, and
the time of its formation, and gives rise to a distinct set of neurons and glial cells in
the larva.

12.2 The development of neurons in Drosophila involves asymmetric cell

divisions and timed changes in gene expression

Before it leaves the neuroectoderm, a neuroblast becomes polarized in an apical–basal

direction, with the apical and basal ends marked by the accumulation of specific sets
of cytoplasmic determinants. The determinants at the apical end help set the plane
of cell division by influencing the orientation of the mitotic spindle, and also direct
the correct positioning of the basal determinants, which are crucial for determining
neural cell fate. Once it has left the surface of the embryo, the neuroblast behaves
 Specification of cell identity in the nervous system 473

Fig. 12.6 The formation of neuronal

Delamination of Localization of Orientation of the Asymmetric
embryonic neuroblast determinants mitotic spindle cell division cells from neuroblasts in Drosophila
Apical Neuroectoderm
by asymmetric cell division. Generation
of ganglion mother cells of the central
nervous system by the asymmetric division
of neuroblasts in the Drosophila embryo.
Once specified, neuroblasts move out of the
neuroectodermal epithelium and then behave
Basal as neural stem cells. Each neuroblast divides
neuroblast asymmetrically to give an apical cell, which
remains a neuroblast stem cell, and a smaller
basal cell—the ganglion mother cell. The
Bazooka orientation of cell division and cell fate are
ganglion
Insc/Pins mother specified by localized protein determinants.
Numb/Prospero/Miranda cell Numb protein is localized to one end of the
neuroblast before division and the daughter
that receives it becomes the ganglion
as a stem cell, dividing asymmetrically to give a larger apical cell, which remains a
mother cell that will give rise to neurons.
neural stem cell, and a smaller basal cell, the ganglion mother cell, which will divide
Prospero and Miranda are required for the
once more and whose daughter cells will differentiate into neurons or glia (Fig. 12.6). proper localization of Numb, and Inscuteable
Among the basal cytoplasmic determinants distributed to the ganglion mother cell by (Insc) and Pins are required for the correct
this asymmetric division is the transcription factor Prospero, which has been shown orientation of cell division.
by DNA microarray profiling to regulate hundreds of target genes involved in sup-
pressing neural stem cell behavior and promoting differentiation of the neuroblast.
The neuroblasts at the different locations along the body continue to divide accord-
ing to the fly’s developmental program, giving rise to a ganglion mother cell and a
neural stem cell at each division. As a neuroblast generates more ganglion mother
cells, the older cells are pushed further into the embryo, giving the final nerve cord
a layered structure. As neurogenesis proceeds, the neuroblast undergoes successive
changes in the expression of a particular set of transcription factors, which give the
neurons produced at different times their different identities.
Because neurons with different functions must develop in the correct positions
within the ventral nerve cord, the precise timing of the transition from expression of
one transcription factor to the next in the neuroblasts is crucial for the correct forma-
tion of the nerve cord. The timing of the first transition seems to be linked to neuro-
blast division, whereas later transitions are linked to an intrinsic timing mechanism
in the neuroblast that is independent of cell division.
The embryonic phase of neuroblast division and neurogenesis provides a func-
tional central nervous system for the first instar Drosophila larva, but these neurons
comprise only about 10% of the neurons that are eventually required by the adult fly.
A further extended period of neurogenesis, involving most of the original embryonic
neurons, occurs before metamorphosis and produces the remaining 90%.
Another classical system of neurogenesis in Drosophila is the generation of sen-
sory bristles in the adult fly from neural cells that are specified in the imaginal discs
(Box 12A, p. 474).

12.3 Specification of vertebrate neuronal precursors also involves

lateral inhibition

The mechanisms involved in early neural specification in vertebrates are in many

ways remarkably similar to those operating in Drosophila. As we saw in Chapter 5,
the cells of the vertebrate nervous system derive from the neural plate, a region of
columnar epithelium induced from the ectoderm on the dorsal surface of the embryo
during gastrulation, and from sensory placodes in the head region that contribute to
the cranial sensory ganglia and produce the sensory receptor cells of the eyes, ears,
and nose (see Chapter 11). Towards the end of gastrulation, the neural plate begins
474 Chapter 12 Development of the nervous system

Box 12A Specification of the sensory organs of adult Drosophila

The sensory organs of adult

bristle
­Drosophila have provided a classic
bristle cell (trichogen)
system for investigating both the
socket
patterning of the epithelia in which
socket cell
they arise, and the role of asym-
sheath cell
metric localization of cytoplasmic sensory
neuron sensory neuron
determinants and asymmetric cell
division to produce a structure
composed of a neuron and non-
neuronal cells. Adult flies have two
main types of sensory organ: external sensory organs in the epidermis, in the form of sensory bristles (see top figure) and other
structures that act as mechanoreceptors and chemoreceptors, and internal chordotonal organs that monitor stretching of tissues.
Each sensory organ is made up of four cells, one of which is a sensory neuron, and each organ arises from a single sensory organ
precursor (SOP) cell that is initially specified in the epidermis and in the larval imaginal discs in a constant pattern from fly to fly.
Unlike the neuroblasts in the
Localization of determinants and Asymmetric cell division
neuroectoderm (see Fig. 12.6), divi- Sensory organ precursor orientation of mitotic spindle in plane of epithelium
sion of an SOP is oriented in such a
Anterior Posterior
way that both daughter cells stay
within the epidermis (see figure on
right), and this is reflected in the
location of the Pins protein, which
Pins
is involved in orienting the plane Numb
SOP IIb IIa
of cell division (compare Fig. 12.6).
The Numb protein is also asymmetrically located in the SOP, such that only one daughter cell ­receives it. This cell (IIb) will give rise to
a sensory neuron and a sheath cell at the next division, with the daughter cell that receives Numb protein becoming the neuron. The
SOP daughter cell that did not receive Numb protein (IIa) gives rise at the next division to a bristle cell and a socket cell, which are non-
neuronal cells. Notch signaling is also required for neuronal differentiation; if it is absent, a sheath cell develops instead of the neuron.
The pattern of sensory bristles in the wing epidermis is already present in the imaginal disc in the form of a pattern of sensory
organ precursors, each of which will give rise to a sensory organ. As in the embryonic neuroectoderm (see Section 12.1), proneural
clusters are first selected by expression of ­proneural genes—in this case genes of the achaete–scute complex—in a precise spatial pat-
tern. ­Exhaustive genetic analysis revealed that this pattern is controlled by the use of different cis-­regulatory modules in the achaete–
scute complex DNA to determine gene expression at specific locations in the imaginal disc. The achaete–scute complex is an excellent
example of how a spatial pattern of gene expression can be precisely controlled by highly complex control regions (see figure ­below).
The top panel of the figure shows the position of various control modules that determine the position-­specific expression of the
achaete and scute genes. These ­modules are located both upstream and downstream of the achaete and scute genes themselves
(shown in black). The bottom panel
shows the location of the proneural Simplified map of the achaete–scute complex showing control modules specifying gene expression
at specific locations in the wing disc
clusters in the imaginal disc (left)
achaete scute
and of the corresponding sensory
bristles in the adult wing and adja-
cent thoracic segments (right),
with colors corresponding to the
Imaginal disc Adult wing
enhancer responsible for the site-
specific expression of the genes.

Photograph courtesy of Y. Jan.

Diagram after Campuzano, S., et al.:

1992
proneural clusters sensory organs
 Specification of cell identity in the nervous system 475

to fold and form the neural tube (see Fig. 5.17). Neural crest cells migrate away from
the dorsal part of the tube and give rise to both the sensory neurons of the peripheral
nervous system and to the autonomic nervous system, and to the glial Schwann cells
of the peripheral nervous system (neural crest migration is discussed in Chapter 8).
The cells that remain within the neural tube give rise to the brain and spinal cord,
which together comprise the central nervous system.
Neuronal precursors are not generated simultaneously throughout the neural plate
and its successor neural tube, but, as in Drosophila, are initially confined to three lon-
gitudinal stripes in the neural plate on each side of the dorsal midline (see Fig. 12.3).
In the Xenopus spinal cord, the normal fate of the medial stripe (the one nearest the
midline), which becomes the ventral region of the spinal cord, is to give rise to motor
neurons. The more lateral stripes, which become the lateral and dorsal regions of the
spinal cord, give rise to interneurons and sensory neurons.
The initial specification of the stripes in the neural plate ectoderm involves lateral
inhibition, in which Delta–Notch signaling between adjacent cells restricts the expres-
sion of neuroblast-specifying proneural genes in the same way as it does in Drosophila
(see Fig. 12.5). Prospective neuroblasts in the three stripes come to express Delta
more strongly than adjacent cells, and the feedback loop set up by Notch activa-
tion in the adjacent cells shuts down proneural gene expression, leaving only the
prospective neuroblasts expressing proneural genes. Vertebrate proneural proteins,
such as the transcription factor neurogenin are homologs of Drosophila proneural
transcription factors, while the proteins that inhibit specification as neural cells are
vertebrate ­homologs of the Notch-activated transcription factor Su(H) and its target
gene Enhancer of split (see Fig. 12.5). Cells that express neurogenin go on to express
genes characteristic of differentiation into neurons, such as NeuroD. Confirmation of
the role of Notch and Delta in vertebrates comes from the transgenic overexpression
of Delta in the neural plate or expression of a mutant Notch that signals continuously,
which both result in a reduction in the number of neurons formed. Inhibition of Delta,
on the other hand, leads to an increase in the number of neurons, as does overexpres-
sion of neurogenin.
The neural tube is formed from the whole of the neural plate and will contain both
committed neural precursors expressing neurogenin and cells in which the neuro-
genin gene is repressed. The latter form a pool of undifferentiated potential neural
precursors in the neural tube, which can be brought into play later, as the first wave
of neuroblasts to be formed stop dividing and start to differentiate. This ensures that
neurogenesis can continue over a relatively long period of time.

12.4 Neurons are formed in the proliferative zone of the vertebrate

neural tube and migrate outwards

In vertebrates, all the neurons and glia of the brain and spinal cord arise from a prolif-
erative layer of epithelial cells that lines the lumen of the neural tube and is called the
ventricular zone. As we saw in Chapters 5 and 8, the neural tube is initially made up
of a single layer of epithelium. As the cells of the neural tube divide, the ventricular
zone develops as a zone of rounded-up dividing cells on the inner surface of the tube.
Once formed, neurons migrate outward from this zone and build up the neural tis-
sue in concentric layers. In this way, the hollow neural tube eventually develops into
the hollow spinal cord, and into the brain with its fluid-filled ventricles that represent
the remains of the neural tube lumen.
The neural progenitor cells in the ventricular zone are multipotent neural stem
cells, which give rise to many different types of neurons and to glia. In the ventricular
zone of the dorsal forebrain, for example, cells known as radial glial cells can act
as neural stem cells, generating both radial glia, whose role is described below, and
476 Chapter 12 Development of the nervous system

some types of neurons. As with all stem cells, neural stem cells must generate both
neural precursors and more stem cells. Two ways of achieving this seem to operate
in the development of neurons in the developing nervous system. The stem cell can
either divide ‘vertically’ and give rise to two similar cells—both stem cells or both
neurons—or it can divide horizontally, parallel to the ventricular surface, with one
daughter remaining a stem cell and the other developing into a neuron.
Once specified as neuroblasts, the cells migrate from the inner surface of the neu-
ral tube towards the outer surface. Here we shall look at the process linking neuron
production and neuron migration with reference to the mammalian cerebral cortex,
which develops from the dorsal part of the embryonic forebrain. The cortex is orga-
nized into six layers (I–VI), numbered from the cortical surface inward, and each layer
contains neurons with distinctive shapes and connections. For example, large pyra-
midal cells are concentrated in layer V, and smaller stellate neurons predominate in
layer IV. All of these neurons have their origin in the ventricular zone and migrate out
to their final positions along radial glial cells—greatly elongated cells that extend
across the developing neural tube (Fig. 12.7). The three-dimensional representation
of radial fibers and migrating neurons was painstakingly reconstructed in the 1960s
and 70s from electron-microscopic images of successive sections of the monkey fetal
cortex, and the reconstruction drawn by hand in great detail.
The layer to which neurons migrate after their birth in the proliferative layer is
related to the time when the neuron is born, and the identity of a cortical neuron is

cortical neural tube radial glial cell

pia leading process

of neuron

Outer surface
migrating neuron
cortical
plate

nucleus

process of
radial glial cell

Fig. 12.7 Cortical neurons migrate along

radial glial cells. Neurons generated in
the ventricular zone migrate to their final trailing process
radial
locations in the cortex along radial glial cells glial cell of neuron
that extend right across the wall of the ventricular
zone
neural tube.
inner surface
After Rakic, P.: 1972.
 Specification of cell identity in the nervous system 477

thought to be specified before it begins to migrate. The neurons of the mammalian

central nervous system do not divide once they have become specified, which is why
newly formed neurons are often called post-mitotic neurons. A neuron’s time of
birth is defined by the last mitotic division that its progenitor cell underwent, and the
experiments that originally established the relation between the birth time of a neu-
ron and the layer in which it ended up are described in Box 12B (p. 478). The newly
formed neuron is still immature; later it extends an axon and dendrites (see Fig. 12.1),
and assumes the morphology of a mature neuron.
Neurons born at early stages of cortical development migrate to layers closest to
their site of birth, whereas those born later end up further away, in more superficial
layers (Fig. 12.8). The younger neurons must migrate past the older ones on their
way to their correct position, giving rise to layers and columns of cell bodies. The
exception is the first-formed layer, which remains as the outer layer. Mouse muta-
tions that disrupt cortical neuronal migration have provided considerable insight
into the process. The reeler mouse, as its name implies, has very poor motor coordi-
nation, which is hardly surprising as its cortical layers are arranged in reverse order.
The reeler gene encodes an extracellular matrix molecule, reelin, that is normally
expressed in the first-formed, outermost layer of the cortex. Mutations in reeler that
cause a loss of reelin protein disrupt radial neuronal migration, with the successive
waves of neurons being unable to bypass their predecessors. As one might expect,
the migration of neurons in the developing brain is in fact much more complicated
than described here; in addition to radial migration, brain neurons also migrate
tangentially within the wall of the neural tube in a process that is independent of
radial glial fibers.
The inside-to-outside migration of neurons in the developing mammalian cerebral
cortex, means that the cortex has most of its cell bodies (the gray matter of the brain)
located at the outer surface, while the axons (the white matter) extend through the
intermediate zone. Compared with a strict outside-to-inside stacking up of succes-
sively formed neurons into serial layers of cell bodies and axons, this arrangement
gives the mammalian cortex great potential for expansion during evolution.

embryonic rodent brain pia Fig. 12.8 The generation of neuronal

layers in the mammalian cortex. Left:
development of the cortex begins with the
migration of the first wave of post-mitotic
neurons (green) towards the outer surface of
the neural tube to form the pre-plate above
superficial the ventricular zone. Middle: the second wave
marginal zone
of newly formed neurons (blue) migrates past
cortical
plate neurons the first layer of the pre-plate to form the
subplate first layer of so-called cortical plate neurons.
pre-plate intermediate The upper layer of the pre-plate neurons
zone
becomes the superficial marginal zone and
will become cortical layer I (cortex layers are
ventricular
zone numbered from the outside inwards). Right:
a third wave of neurons (purple) migrates
to form a second layer of cortical plate
radial
glial cell neurons above the first. Subsequent waves
of neurons will each form a more superficial
ventricle
layer of cortical plate. The intermediate zone
Time contains migrating neurons (light blue).
Adapted from Honda, T., et al.: 2003.
478 Chapter 12 Development of the nervous system

Box 12B Timing the birth of cortical neurons

The time at which a neuron is born Time of [3H]thymidine injection

Cortical
determines what cortical layer it layers
becomes part of. A neuron’s time I
of birth is determined as the time II

Distribution of [3H]thymidine neurons

that the precursor cell stops divid-
ing and leaves the mitotic cell III
cycle. Some of the earliest exper­ IVA
iments, done in the 1960s and IVB
1970s, used radiolabeled isotopes
IVC
to label cells so that their migra-
V
tion could be followed. If neuronal
VI
precursors are given a short pulse
of a labeled compound that can be
Proliferative ventricular zone
incorporated into DNA, those cells
that undergo their last round of 30 40 50 60 70 80 90 100 170
replication immediately after tak- Embryonic age (days) (Birth)
ing up the compound will have the
most heavily labeled DNA. If they continue to divide and repli- ­ istribution of heavily labeled neurons found after each injection.
d
cate their DNA, the label becomes diluted among their progeny. Neurons born earliest remain closest to their site of birth, the
Cells that stop dividing immediately after taking up the label can proliferative ventricular zone. Neurons born later become part of
therefore be distinguished and their final destinations detected. successively higher cortical zones. Cortical zones are numbered
The figure shows the results of an experiment of this type that from the surface of the neural tube inward; that is, layer V1 is
originally determined the order in which neurons reached their the layer nearest to the ventricular zone. For this type of experi-
final layers in the monkey visual cortex. Radioactively labeled ment nowadays, radioisotopic labeling has been superseded by
thymidine ([3H]thymidine), which becomes incorporated into the labeling of cells with, for example, replication-incompetent
replicating DNA, was injected into the developing neural tube at retroviruses carrying a marker gene such as lacZ, whose product
different times during development. The red bars represent the can be detected histochemically.

For many years it was thought that no new neurons could be generated in the adult
epidermis mammalian brain, although adult neurogenesis was known to occur in the brains of
roof plate songbirds, reptiles, amphibians and fish. Neural stem cells are now known to exist in
C C limited regions of the adult mammalian brain, and can give rise to new neurons and
glial cells, as described in Section 10.10. These adult neural stem cells are a population
of slowly dividing radial glial cells, like those in the embryonic cortex.

M M 12.5 The pattern of differentiation of cells along the dorso-ventral axis of the
V3 V3 neural spinal cord depends on ventral and dorsal signals
crest cells
Like the brain, the spinal cord initially develops by the radial migration of neurons,
floor plate notochord
although in this case the final structure is of a central core of cell bodies (gray matter)
surrounded by an outer layer of axon tracts (white matter). A properly functioning
Fig. 12.9 Dorso-ventral organization in nervous system depends on neurons of particular types and functions differentiating
embryonic spinal neural tube. In neural tube in the right positions in the developing neural tube, and the developing spinal cord is
that develops into spinal cord, a floor plate of
a particularly good system in which to study this aspect of development. It has a par-
non-neuronal cells develops along the ventral
ticularly clear dorso-ventral pattern of functionally distinct neurons, which is similar
midline, and a roof plate of non-neuronal
cells along the dorsal midline. Commissural
along its length (Fig. 12.9). Future motor neurons and their associated interneurons
neurons (C) differentiate in the dorsal region, are located ventrally, and form the ventral roots of the spinal cord, whereas commis-
near the roof plate. Motor neurons (M) and V3 sural neurons, sensory neurons, and their associated interneurons differentiate primar-
interneurons differentiate close to the floor ily in the dorsal region. Each cell type is present symmetrically on either side of the
plate (see Fig. 12.12). midline. While motor and sensory neurons differentiating in the right and left halves
 Specification of cell identity in the nervous system 479

Fig. 12.10 Formation of sensory

Chick spinal cord (3 days) Chick spinal cord (4 days)
and motor neurons in the chick spinal
cord. Three days after an egg has been laid,
motor neurons are beginning to form within
the chick embryo neural tube. A day later, the
motor neurons send at axons to form the
ventral root. The sensory neurons that
migrate from the dorsal part of the spinal
cord are derived from neural crest cells and
form the segmental dorsal root ganglia (see
Section 10.11). The commissural neurons
send axons towards and across the ventral
midline.

ventral
root

sensory neurons Dorsal

commissural neurons
motor neurons neural tube

of the spinal cord will remain on the side of the body in which they originated, the
axons of commissural neurons will cross the midline of the spinal cord, providing a
link ­between the two sides of the body at the level of the spinal cord (Fig. 12.10). The
neurons of the sensory nervous system develop from neural crest cells that are located Ventral
notochord
laterally and dorsally in the neural tube, and migrate away from the surface of the tube
to form the dorsal root ganglia that lie at regular intervals on each side of the spinal
cord (see Fig. 12.10 and Figs 8.36 and 8.37). As in the brain, the neurons and glia of the
spinal cord are generated in a proliferative zone at the inner face of the neural tube.
We shall now consider how the dorso-ventral organization of the spinal cord is pro-
duced by signals received by the early neural tube, which induce the differentiation of
particular types of neurons in the correct positions. In additional to the future nerve-
cell precursors, a group of non-neural cells forms a region called the floor plate in
the ventral midline of the neural tube, while another group of non-neural cells forms
the roof plate in the dorsal midline. These two regions produce signals that pattern the
neural tube in a dorso-ventral direction.
The floor plate is induced by signals secreted by the mesodermal notochord, which
lies immediately below the neural tube. The patterning activity of the notochord on
the neural tube can be shown by grafting a segment of notochord to a site lateral or Shh BMPs
dorsal to the neural tube (see Fig. 5.30, where the experimental set-up is described
in regard to its effect on somite patterning). A second floor-plate is induced in the Fig. 12.11 Dorso-ventral patterning of the
neural tube in direct contact with the transplanted notochord. Molecular markers of spinal cord involves both dorsal and ventral
signals. As the neural tube forms (top
dorsal-cell differentiation are suppressed in the nearby neural cells, and additional
panel), dorsal and ventral signaling occurs.
motor neurons, which normally only develop in the ventral region, are generated.
The ventral signal from the floor plate is the
In situ staining for gene expression and misexpression experiments identified the
Shh protein (red), whereas the dorsal signal
inductive signal secreted by the notochord as the protein Sonic hedgehog (Shh). Shh (blue) includes BMP-4, which is induced
induces the floor-plate cells themselves to produce Shh, which then forms a gradient during neural tube formation by the adjacent
of activity from ventral to dorsal in the neural tube, and acts as the ventral patterning ectoderm. Ventral and dorsal signals oppose
signal (Fig. 12.11). each other’s action.
480 Chapter 12 Development of the nervous system

The conversion of neuroepithelial neuronal precursors in the ventral neural tube to

committed motor-neuron progenitor cells is driven by Shh from the floor plate and by
retinoic acid secreted by the adjacent mesoderm. Control of the proliferation of neural
progenitor cells by mitogenic Wnt signals is also important in regulating the number
of neurons produced. In line with the role of Shh and retinoic acid in neural differen-
tiation in vivo, they can direct the development of embryonic stem cells in culture into
motor neurons and interneurons.
The roof plate is initially specified by signals from the dorsal epidermal ectoderm,
and signals from the roof plate pattern the dorsal half of the neural tube. From genetic
knock-outs and cell-fate mapping, it has been shown that selective removal of the roof
plate from the neural tube of the mouse embryo results in the loss of the most dorsal
interneurons. The dorsalization signal includes BMPs: BMPs from the ectoderm (see
Chapter 5) specify the roof plate, and BMPs produced in the roof plate then pattern
the dorsal region of the neural tube. The BMP signals apparently ensure dorsal dif-
ferentiation by opposing the action of signals from the ventral region, thus together
patterning the dorso-ventral axis of the neural tube (see Fig. 12.11).
The spatial pattern of expression of BMPs in the dorsal neural tube is strongly influ-
enced by the ventral signals, and this system bears a strong similarity to the mecha-
nism for patterning the dorso-ventral body axis in early Drosophila development. In
that case, the expression of the gene decapentaplegic is confined to dorsal regions
because its expression is repressed in ventral regions by the transcription factor Dorsal
(see Section 2.12). Like BMPs, Decapentaplegic is a member of the TGF- family of
secreted signals. Similarly, in the spinal cord, Shh and BMPs provide positional sig-
nals with opposing actions emanating from the two ends of the dorso-ventral axis.

12.6 Neuronal subtypes in the ventral spinal cord are specified by the ventral
to dorsal gradient of Shh

The ventral spinal cord produces five different classes of neurons—motor neurons and
four classes of interneurons, each distinguished by the expression of specific genes.
Experiments in vitro have shown that these neuronal subtypes can be specified in a
dose-dependent manner by Shh, the signal produced by the notochord and floor plate;
the different neuronal subtypes can be generated from chick neural-plate explants
in response to a two- to threefold change in concentration of Shh. The incremental
changes in Shh concentration are mimicked by smaller changes in the activity of the
transcription factor Gli3, which is activated by Shh signaling (see Box 11C, pp. 424–
Fig. 12.12 Specification of neuronal
425), and which provides the key intracellular signal in a graded form (Fig. 12.12).
subtypes in the ventral neural
tube. Different neural subtypes (shown
as different colors) are specified along the Dorsal Pax7 Dbx1 Dbx2 Irx3 Pax6
dorso-ventral axis of the ventral neural tube
in response to a gradient of Sonic hedgehog
(Shh) with its source in the floor plate.
p0 V0
The Shh gradient results in an intracellular
gradient of active transcription factor Gli3.
Shh signaling mediates the repression of p1 V1
the so-called Class I homeodomain protein
genes (Pax7, Pax6, Dbx1, Dbx2, and Irx3), p2 V2
whereas Class II homeodomain protein
genes (Nkx2.2 and Nkx6.1), are activated.
pMN MN
Interactions between Class I and Class II
genes pattern this expression further. Five Gli Shh
neuronal subtypes are generated from the p3 V3
five domains. MN, motor neurons.
Ventral Nkx6.1 Nkx2.2
Adapted from Jessell, T.M.: 2000.
 Specification of cell identity in the nervous system 481

A number of homeobox transcription factor genes are regulated in ventral pro-

genitor cells in response to Shh signaling, and these genes can be divided into two
classes—one (class I) containing genes that are generally repressed by Shh, and one
(class II) containing genes that are activated (see Fig. 12.12). The threshold levels of
Shh that allow expression are different for each gene, so that overall, the responses
of cells to the Shh gradient divides the dorso-ventral axis of the spinal cord into at
least five regions of different types of progenitor neurons. Boundaries between each
region are sharpened by interactions between the class I and class II proteins. The
detailed mechanism by which Shh specifies the Gli gradient in the progenitor neu-
rons in vivo is not yet completely worked out. As discussed elsewhere for morpho-
gens in general (see Box 11A, p. 421), processes other than simple diffusion must be
involved in forming the Shh gradient. Diffusion alone would not be a reliable way of
setting up such a gradient, as the extracellular concentration of a diffusing molecule
at any point changes over time, and its movement is affected by other extracellular
molecules.

12.7 Spinal cord motor neurons at different dorso-ventral positions project to

different trunk and limb muscles

The motor neurons developing in the ventral region of the spinal cord can be further
classified on the basis of the position of their cell bodies along the cord dorso-ventral
axis, and the muscles that their axons innervate. In the chick embryo, motor neurons
are subdivided into longitudinal columns on each side of the midline—a median col-
umn nearest the midline, which runs the whole length of the spinal cord, and a ­lateral
motor column (LMC), which is only present in the brachial and lumbar regions,
where the forelimbs and hindlimbs, respectively, develop (Fig. 12.13). Motor neurons
in the median motor column send axons to axial and body-wall muscles, whereas
neurons of the LMC innervate the limb. The LMC is divided into two divisions:lateral
and medial.
Once designated, each motor neuron subtype expresses a different combination
of transcription factors of the LIM homeodomain family, which provides a neuron
with its positional identity within the spinal cord. In the LMC, for example, neurons
expressing Lim1 settle in the lateral division, whereas those expressing Isl1 settle in
the medial division. Experiments tracing the outgrowth of axons from LMC motor
neurons show that those in the lateral LMC send axons into the dorsal muscle masses

lumbar
Caudal
Fig. 12.13 The organization of the motor
thoracic
columns in the developing chick spinal
brachial cord. The organization of motor columns
along the chick embryo spinal cord at
cervical stages 28–29. The median motor column is in
blue, the medial division of the lateral column
in red and the lateral division of the lateral
motor column in green. The lateral motor
columns are only present in the brachial and
Rostral lumbar regions, in register with the position
of limb formation.
482 Chapter 12 Development of the nervous system

motor
axon
lsl1

Motor neuron
lsl1 generation
lsl2
limb bud

Median motor column Medial lateral column Lateral lateral column

lsl1 lsl1 LIM1 Axonal
Fig. 12.14 Different motor neurons in lsl2 lsl2 lsl2 projections
the chick spinal cord express different LIM Lhx3
homeodomain proteins. The photograph Lhx4
(inset) shows a transverse section through a
chick embryo at the level of the developing
wing. Motor neurons can be seen entering
the limb bud. Initially, all developing motor
neurons express the LIM proteins Isl1 and
Isl2. At the time of axon extension, motor
neurons that innervate different muscle
regions express different combinations of
Isl1, Isl2, Lim1, Lhx3, and Lhx4.
Axial muscle Ventral limb muscle Dorsal limb muscle
Micrograph courtesy of K. Tosney.

in the limb bud, whereas medial LMC neurons send their axons ventrally into the
limb bud (Fig. 12.14). We shall look in more detail at the mechanisms that guide the
growing axon in the next part of the chapter.

12.8 Antero-posterior pattern in the spinal cord is determined in response to

secreted signals from the node and adjacent mesoderm

As well as being organized along the dorso-ventral axis, neurons at different positions
along the antero-posterior axis of the spinal cord become specified to serve different
functions. Antero-posterior specification of neuronal function in the spinal cord was
dramatically illustrated some 40 years ago by experiments in which a section of the
spinal cord, which would normally innervate wing muscles, was transplanted from
one chick embryo into the lumbar region, which normally serves the legs, of another
embryo. Chicks developing from the grafted embryos spontaneously activated both
legs together as though they were trying to flap their wings, rather than activating
each leg alternately as if walking. These studies showed that motor neurons generated
at a given antero-posterior level in the spinal cord had intrinsic properties character-
istic of that position.
Neuronal fate along the antero-posterior axis of the spinal cord is determined
initially by signal molecules secreted by the organizer (or node) and later by the
adjacent paraxial mesoderm (see Chapter 5). This was first shown by experiments
in which quail paraxial mesoderm was grafted to the thoracic level in chick embryos
at the time of neural-tube closure. The result was that presumptive chick thoracic
neurons became specified as brachial neurons. The signals from the node and the
mesoderm form gradients along the antero-posterior axis of the embryo that activate
patterns of Hox gene expression in post-mitotic neurons, giving groups of neurons
a positional identity that determines their fate. The signals are composed of FGF,
 Specification of cell identity in the nervous system 483

growth/differentiation factor (GDF)—a member of the TGF-β family—and retinoic

acid (see Fig. 5.21). FGF is graded from posterior to anterior, and there is a shallow Hox6 Hox9
Hox5 Hox8
retinoic acid gradient at the anterior end of the spinal cord and a similar one of GDF
Hox4
at the posterior end. Hox7
The spinal cord becomes demarcated into different regions along the antero-­posterior
axis by combinations of expressed Hox genes. For example, Hoxc6 is expressed in Spinal cord
the motor neurons in the brachial region—adjacent to the forelimb—while Hoxc9 is
expressed in thoracic motor neurons. In some neuronal subtypes, different concentra-
tions of the transcription factor FoxP1 act together with the Hox proteins to specify
LMC
different neuronal identities. This is in addition to the gene expression specifying
dorso-ventral identity described in the previous section.
A typical vertebrate limb contains more than 50 muscle groups with which neurons
must connect in a precise pattern. Individual neurons express particular combina-
tions of Hox genes, which determines which muscle they will innervate. As well as Sca Pec
being organized into divisions within the LMC with ventral and dorsal trajectories
ALD FCU
into the limb (see Fig. 12.14), motor neurons serving the limbs are further subdivided Wing muscles
into so-called ‘pools’ along the dorso-ventral and antero-posterior axes. The number
of pools of motor neurons corresponds to the number of muscles in the limb, with the
neurons in each pool directing their axons to a particular target muscle. The identity Fig. 12.15 Pools of motor neurons in the
of the motor neurons in each pool is specified by the combinations of Hox genes developing spinal cord make connections
expressed. For example, in the chick, the posterior limit of the region containing bra- with specific limb muscles. The lateral
chial motor neurons maps to the posterior boundary of the expression of Hox6 genes, columns of motor neurons in the brachial
while the anterior boundary is defined by the anterior limit of expression of Hoxc6. region of the developing chick spinal cord are
organized into a number of groups, or pools,
Interactions between genes of the Hox8, Hox7, Hox6, Hox5, and Hox4 paralogous
that each make connections to a specific
subgroups specify the identities of pools of motor neurons that innervate particular
muscle group in the developing wing. The
muscles in the chick wing (Fig. 12.15).
identity of each pool of neurons is specified
So, all together, expression of genes resulting from dorso-ventral position, together by interactions between the Hox genes
with those resulting from antero-posterior position, confer a virtually unique identity expressed in the spinal cord in this region.
on functionally distinct sets of neurons in the spinal cord. LMC, lateral motor column (see Fig. 12.13).
Muscles: Sca, scapulohumeralis posterior; Pec,
pectoralis; ALD, anterior latissimus dorsi; FCU,
Summary flexor carpi ulnaris.
Adapted from Dasen, J.S., et al.: 2005.
In Drosophila, prospective neural tissue is specified early in development as a band of
neuroectoderm along the dorso-ventral axis. Within the neuroectoderm, expression of
proneural genes distinguishes clusters of neuroectoderm cells with the potential to
form neural cells. As a result of lateral inhibition involving Notch and Delta, only one cell
of the cluster finally gives rise to a neuroblast. Neuroblasts that give rise to the central
nervous system behave as stem cells, undergoing repeated asymmetric division, each
giving rise to a neuroblast and a daughter cell that can give rise to neurons.
The vertebrate nervous system is derived from the neural plate, which is specified
during gastrulation. It contains the cells that form the brain and spinal cord, as well as
the neural crest cells that contribute to the peripheral nervous system. Prospective
neural cells are initially specified within the neural plate by a mechanism of lateral
inhibition, similar to that in Drosophila. Neurons are born in the proliferative zone on
the inner surface of the neural tube and then migrate to different locations within the
developing brain and spinal cord. Patterning of neuronal cell types along the dorso-
ventral axis of the spinal cord involves signals from both the ventral and dorsal regions.
Secretion of Shh by the notochord and floor plate provides a graded signal that leads
to specification of different neuronal subtypes in the ventral region of the spinal cord.
Neuronal identity along the antero-posterior axis is specified by the combinatorial
expression of Hox genes.
484 Chapter 12 Development of the nervous system

Summary: specification of neuronal identity

Axon navigation

We shall now look at a feature of development that is unique to the nervous system—
the outgrowth and guidance of axons to their final targets. This stage of neuronal dif-
ferentiation occurs mainly after immature post-mitotic neurons have migrated to their
eventual locations. Each neuronal cell body extends an axon and dendrites, through
Fig. 12.16 Neurons make precise which it will, respectively, send and receive signals. The working of the nervous sys-
connections with their targets. Neurons tem depends on the formation of discrete neuronal circuits, in which neurons make
(green) and their target cells usually develop numerous and precise connections with each other (Fig. 12.16). In the rest of this
in different locations. Connections between
chapter we will look at how these connections can be set up in various situations.
them are established by axonal outgrowth,
What controls the growth of axons and the contacts they make with other cells, and
guided by the movement of the axon tip
(growth cone). Often, the initial set of
how are such precise connections achieved? In this part of the chapter we will first
relatively nonspecific synaptic connections consider axon outgrowth and the evidence that the growing axon is guided specifi-
is then refined to produce a more precise cally to its target.
pattern of connectivity.
After Alberts, B., et al.: 2002.

Birth of neurons Outgrowth of axons and dendrites Synaptic connections made Refinement of synaptic connections
growth cone
 Axon navigation 485

12.9 The growth cone controls the path taken by a growing axon

An early event in the differentiation of a neuron is the extension of its axon by the
growth and migration of the growth cone located at the tip of the axon (Fig. 12.17).
The growth cone is specialized for both movement and for sensing its environment
for guidance cues. As in other cells capable of migration, such as the primary mesen-
chyme of the sea-urchin embryo (see Section 8.7), the growth cone can continually
extend and retract filopodia at its leading edge, making and breaking connections
with the underlying substratum to pull the axon tip forward. Between the filopodia,
the edge of the growth cone forms thin ruffles—lamellipodia—similar to those on a
moving fibroblast (see Box 8A, p. 291). Indeed, in its ultrastructure and mechanism
of movement, the growth cone closely resembles the leading edge of a fibroblast
crawling over a surface. Unlike a moving fibroblast, however, the extending axon also
grows in size, with an accompanying increase in the total surface area of the neuron’s
plasma membrane. The additional membrane is provided by intracellular vesicles,
which fuse with the plasma membrane.
The growth cone guides axon outgrowth, and is influenced by the contacts the
filopodia make with other cells and with the extracellular matrix. In general, the
growth cone moves in the direction in which its filopodia make the most stable con-
tacts. In addition, its direction of migration can be influenced by diffusible extracellu-
lar signal molecules that bind to receptors on the growth-cone surface. Some of these
extracellular cues promote axon extension, whereas others inhibit it by causing the
retraction of the filopodia and the ‘collapse’ of the growth cone. The extension and
retraction of filopodia and lamellipodia involve the assembly and disassembly of the
actin cytoskeleton, and many of the extracellular signals that affect axon guidance are
known to have effects on the actin cytoskeleton through the Ras-related GTPase fam-
ily of intracellular signaling proteins. But precisely how growth cones transduce extra-
cellular signals so as to extend or collapse filopodia is not yet fully understood.
Axon growth cones are guided by two main types of cue—attractive and repulsive.
In addition, cues can act either at long or short range, thus giving four ways in which
the growth cone can be guided (Fig. 12.18). Long-range attraction involves diffusible
chemoattractant molecules released from the target cells. In contrast, chemorepel-
lents are molecules that repel migrating cells or axons. Both chemoattractant and
chemorepellent proteins have been identified in the developing nervous system.
Short-range guidance is mediated by contact-dependent mechanisms involving mol-
ecules bound to other cells or to the extracellular matrix; again, such interactions can
be either attractive or repulsive. Some signals can function as either an attractant or a
repellent, depending on the cellular and developmental context.

filopodia axon cell body

Fig. 12.17 A developing axon and growth

cone. The axon growing out from a neuron’s
cell body ends in a motile structure called the
growth cone. Many filopodia are continually
extended and retracted from the growth
cone to explore the surrounding environment.
Scale bar = 10 µm.
growth cone
Photograph courtesy of P. Gordon-Weeks.
486 Chapter 12 Development of the nervous system

Long-range cues Short-range cues

Chemoattraction Chemorepulsion Contact attraction Contact repulsion

e.g. Netrins e.g. Semaphorins e.g. Cadherins e.g. Ephrins

Fig. 12.18 Axon-guidance

mechanisms. Four general types of
mechanism can contribute to controlling the
direction in which a growth cone moves: long-
range attraction, long-range repulsion, short-
range attraction, and short-range repulsion.
Long-range cues are provided by secreted
molecules (such as netrins and semaphorins)
that form gradients in the extracellular
matrix. Short-range cues are often provided
by transmembrane proteins (such as the
ephrins and their Eph receptors, and the
cadherins, which bind to cadherins).

A number of different extracellular molecules that guide axons have been identi-
fied. The semaphorins are among the most prominent of the conserved families of
axon-guidance molecules. There are seven different subfamilies, three of which are
secreted, whereas the others are associated with the cell membrane. There are two
classes of neuronal receptors for semaphorins—the plexins and neuropilins. Axon
repulsion is a major theme of semaphorin action. Some, however, can either attract or
repel growth cones; which they do will depend on the nature of the neuron and on the
type of semaphorin receptor present in the neuron membrane. Different regions of the
same neuron can respond in opposite ways to the same signal—the apical dendrites
of pyramidal neurons in the cortex grow towards a source of semaphorin 3A, whereas
their axons are repelled, which enables the receiving end of the neuron and the trans-
mitting end of the neuron to be oriented in opposite directions by the same signal.
This difference in response appears to be due to the presence of guanylate cyclase, a
component of the pathway that transduces the semaphorin signal, in the dendrites
but not in the axon.
Other classes of neuronal-guidance proteins, such as the secreted netrins, are also
bifunctional, in that they attract some neurons and repel others. The attractant func-
tion of netrins is mediated by the DCC (deleted in colorectal cancer) family of recep-
tors on growth cones. The Slit proteins are secreted glycoproteins with a repellent
function. They repel a variety of axon classes by acting on receptors of the Robo fam-
ily on neurons, but can also stimulate elongation and branching in sensory axons.
Guidance molecules that act by short-range cell–cell contact include the transmem-
brane ephrins and their Eph receptors (see Box 5F, p. 205), which in the nervous
system mediate mainly repulsive interactions, and the cadherins, which bind to
­cadherins on adjacent cells (see Box 8B, p. 293), and which can act as short-range
­attractors of axons. As we shall see in the next section, ephrins are involved in guiding
the axons of spinal cord motor neurons to their correct target muscles in the limb.

12.10 Motor neuron axons in the chick limb are guided

by ephrin–Eph interactions

An animal’s muscular system enables it to carry out an enormous variety of move-

ments, which all depend on motor neurons from the spinal cord having taken the cor-
rect path and made the correct connections with muscles. The pattern of innervation
of the chick wing, for example, is precise and always the same (Fig. 12.19). As we
 Axon navigation 487

Fig. 12.19 Innervation of the embryonic

chick wing. The nerves are stained brown
and include both sensory and motor neurons.
Scale bar = 1 mm.
Photograph courtesy of J. Lewis.

saw in Sections 12.7 and 12.8, particular ‘pools’ of neurons that will innervate specific
muscles in the limb are already distinguished by the expression of particular combina-
tions of LIM and Hox genes. To reach their correct targets, however, the growing axons
require local guidance cues and these are provided by the limb tissue.
In the chick embryo, the motor neuron axons enter the developing limb bud at
around 4.5 days, after cartilage elements, and dorsal and ventral muscle masses, have
started to form. When the motor axons growing out from the spinal cord first reach
the base of the limb bud, they are all mixed up in a single bundle. At the base of the
limb bud, however, the axons separate out and form new nerves containing only
those axons that will make connections with particular muscles.
Evidence that local cues in the limb are responsible for this sorting-out comes from
experiments in chick embryos in which a section of the spinal cord whose motor neu-
rons will innervate the hindlimb was inverted before the axons grew out. After inver-
sion along the antero-posterior axis, motor axons from lumbo-sacral segments 1–3,
which normally enter anterior parts of the limb, initially grew into posterior limb
regions, but then took novel paths to innervate the correct muscles. So, even when
the axon bundles entered in reverse order, the correct relationship between motor
neurons and muscles was achieved. Motor axons can thus clearly find their appropri-
ate muscles when their displacement relative to the limb is small. But with a large
displacement, this is not the case. A complete dorso-ventral inversion of the limb bud
resulted in the axons failing to find their appropriate targets. In such cases, the axons
followed paths normally taken by other neurons. The mesoderm of the limb thus pro-
vides local cues for the motor axons and the axons themselves have an identity that
allows them to choose the correct pathway.
These local cues include signals such as the transmembrane ephrins. Motor neu-
rons in the medial division of the LMC (see Fig. 12.14) innervate ventral muscle
masses in the limb, while those in the lateral division innervate dorsal limb muscles.
EphA4 signaling in the lateral set of motor axons is a major determinant of the dorsal
route they take into the limb. EphA4 is thought to mediate a repulsive interaction with
its ligand ephrin A, which is restricted to the ventral limb mesenchyme. Migration of
medial motor axons into the ventral mesenchyme appears to be regulated by another
set of receptors that similarly respond to repulsive signals from the dorsal limb mes-
enchyme. The expression of EphA4 is determined by the LIM homeodomain proteins
(see Section 12.7), with Lim1 expression in the lateral LMC motor neurons leading to
raised levels of EphA4, whereas Isl1 expression in the medial LMC axons leads to a
reduction in EphA4. The concentration of ephrin A in the dorsal limb mesenchyme is
apparently kept at a low level by the expression there of the LIM protein Lmx1b (see
Fig. 11.14).
488 Chapter 12 Development of the nervous system

12.11 Axons crossing the midline are both attracted and repelled

Although the development of the nervous system is largely symmetrical about the
midline of the body, and proceeds independently on either side, one half of the body
needs to know what the other half is doing. So, although some neurons must remain
entirely within the lateral half of the body in which they were formed, many axons
must cross the ventral midline of the embryo. Chemoattractant and chemorepellent
molecules are involved in these decisions.
In the vertebrate spinal cord, commissural neurons provide the connections
between the two sides of the cord. The cell bodies of commissural neurons are located
in the dorsal region but they send their axons ventrally along the lateral margin of the
cord. Signals from the roof plate prevent the commissural axons extending dorsally.
About halfway along the margin of the spinal cord, the axons make an abrupt turn
and project to the floor plate in the ventral midline, bypassing the motor neurons
(Fig. 12.20; also see Fig. 12.10). After crossing the floor plate to the other side of the
midline, commissural axons make another sharp turn and project anteriorly along the
spinal cord towards the head; they do not re-cross the midline. Chemoattractant and
chemorepellent molecules present in the ventral midline are involved in both enabling
the commissural neurons to cross the midline and in preventing them from crossing
back. Netrin-1 is a key attractant produced in the floor plate and in the midline, and
knock-out of the mouse netrin-1 gene, or the gene for its receptor DCC, results in
abnormal commissural axon pathways (Fig. 12.21).
Axons approaching the midline of the spinal cord provide good examples of the long-
range attraction and repulsion of growth cones. The axons of vertebrate commissural
neurons are attracted to the floor plate at the ventral midline by the ­chemoattractant
Netrin-1, which signals attraction via the netrin-1 receptor DCC. In both Drosophila
and vertebrates, the mechanism that repels axons from the midline involves the extra-
cellular protein Slit and its Robo receptors on the axons’ growth cones. Slit is present
at the midline and, when it binds to receptors called Robo1 and Robo2 on the growth
cone, it exerts a chemorepellent effect. For vertebrate commissural axons approach-
ing the midline, however, the chemorepellent effect of Slit is overcome by the Robo
receptor-related protein Robo3 expressed on the growth cone, which interferes with
Slit signaling through Robo1 and Robo2 (Fig. 12.22). Once the commissural axons
cross the midline, Robo3 ceases to be expressed and the growth cone can be repelled
from the midline again by Slit proteins, which prevents it crossing back. Mammalian
commissural neurons make a sharp anterior turn after crossing the midline, which is
thought to be due to a gradient of Wnt proteins in the floor plate with the high point
anterior.

Commissural neurons extend ventrally then

project to floor plate cells at the ventral midline

roof plate

Fig. 12.20 Chemotaxis can guide

commissural axons in the spinal cord. In the dorsal
vertebrate spinal cord, the axons of spinal spinal
cord commissural
cord commissural neurons extend ventrally, neuron
and then toward the floor-plate cells. They
cross the midline at the floor plate and then
grow anteriorly along the spinal cord. The
photograph shows a section through a rat
spinal cord with the axons of the commissural
neurons (those inside the spinal cord) stained
green and the cell bodies red. floor plate motor neurons
Photograph from Samantha Butler.
 Axon navigation 489

Fig. 12.21 Effect of netrin-1 gene knock-

normal: netrin-1 expressed netrin-1 knock-out results in failure of commissural
axon crossing at the floor plate out in mice. In mice lacking netrin-1, the
commissural axons do not migrate toward
the floor plate (compare photograph in
Fig. 12.20). Scale bar = 0.1 mm.
Photographs courtesy of M. Tessier-Lavigne,
from Serafini, T., et al.: 1996.

Ipsilateral Contralateral
Midline
Commissural axon
12.12 Neurons from the retina make ordered connections with visual centers
in the brain Slit
Robo1/2
An arguably even more complex task for the developing nervous system is to link up DCC Netrin
the sensory receptors that receive signals from the outside world with the neuronal
targets in the brain that enable us to make sense of these signals. A characteristic fea- Growth of axon in
ture of the vertebrate brain is the presence of topographic maps. That is, the neurons nervous system
from one region of the nervous system project in an ordered manner to one region of Robo3
the brain, so that nearest-neighbor relations are maintained. The vertebrate visual sys-
tem has been the subject of intense investigation for many years, and the highly orga- chemoattractant
nized projection of neurons from the eye via the optic nerve to the brain is one of the
best models we have to show how topographic neural projections are made. There are
Axon commits to
around 126 ­million individual photoreceptor cells in a human retina, and some animals cross midline
with excellent night vision, such as owls, have many more. Each of those photoreceptor
cells is continuously recording a minute part of the eye’s visual field and the signals
must be sent to the brain in an orderly manner so that they can be assembled into a
coherent picture. Small sets of adjacent photoreceptor cells indirectly activate individual
neurons—the retinal ganglion cells—whose axons are bundled together and exit the eye
as the optic nerve (see Section 11.23 for a brief overview of the development of the eye chemorepellent
itself). The optic nerve from each human eye is made up of over a million axons.
In birds and amphibians, the optic nerve connects the retina to a region of the
Axon continues to grow
midbrain called the optic tectum, where visual signals are processed. In birds and in on other side of midline
larval amphibians, the optic nerve from the right eye makes connections with the left
optic tectum, while the nerve from the left eye makes connections to the right-side of

Fig. 12.22 Competing chemoattractant and chemorepellent signals both enable

commissural axons to cross the midline and prevent them crossing back. Commissural axons
are attracted to the midline by the chemoattractant protein Netrin (green circle) acting through Non-commissural axon
its receptor DCC (green). As they grow toward the midline, the level of Robo1/2 (yellow), the
receptors for the chemorepellent protein Slit (yellow diamond), are kept low. Commissural
axon growth cones also start to express the cell-surface receptor Robo3 (orange) as they
approach the midline, which inhibits the chemorepellent effect of Slit acting at Robo1/2.
Once commissural axons reach and cross the midline, alternative splicing of Robo3 produces chemorepellent
a cell-surface receptor that is now responsive to Slit and does not inhibit Robo1/2, and so
repulsion by Slit can now prevent the axons from re-crossing the midline. Increased expression
of Robo1/2 occurs after crossing and helps block attraction to the midline exerted by Netrin.
Non-commissural axons do not express Robo3 and so are repelled from the midline by Slit.
490 Chapter 12 Development of the nervous system

Fig. 12.23 Comparison of amphibian and

Frog tadpole Mammal
mammalian visual systems. Left panel: in
the frog tadpole, neurons from the retina N N
project directly to the optic tectum, neurons
from the left retina connecting to the right retina retina
T T
optic tectum and those from the right retina
connecting to the left tectum, crossing over optic
optic nerve optic
at the optic chiasm. Right panel: in mammals chiasm chiasm
with good binocular and stereoscopic vision, A
the main visual pathway is to the lateral lateral
geniculate nucleus (LGN), from which neurons optic tectum
geniculate
L M nucleus
project to the visual cortex. Neurons from
one half of the retina project to the LGN on
the same side as the eye, and those from P
the other half cross over to the opposite LGN
at the optic chiasm. For the sake of clarity, visual
cortex
the neurons from only one half of the retina
are shown. N, nasal; T, temporal, L, lateral,
M, medial. A, anterior, P, posterior.
the brain, the axons crossing over each other in a structure called the optic chiasm
After Goodman, C.S. and Shatz, C.J.: 1993.
(Fig. 12.23, left panel). In mammals, the main target for retinal axons are the paired
lateral geniculate nuclei (LGN) in the forebrain, from which other neurons then con-
vey the signals to the visual cortex, in which most visual processing occurs. A minor
pathway in mammals goes from the retina to the superior colliculus in the midbrain
(which is analogous to the avian and amphibian optic tectum), which is involved
mainly in aiding the orientation of the eyes and head to a variety of sensory stimuli.
As this pathway provides a simpler map than that to the lateral geniculate nucleus, it
is often studied as the mammalian counterpart to the optic tetum.
In most mammals (and in frogs after metamorphosis), retinal axons from one eye
do not all go to the same LGN. Instead, axons from the ventro-temporal region of the
retina project to the LGN on the same side as the eye (ipsilateral), while the remain-
ing retinal axons project to the LGN on the opposite side (contralateral) (Fig. 12.23,
right panel). This separation of outputs occurs at the optic chiasm and enables excel-
lent binocular vision in mammals with forward-facing eyes, such as primates and
humans, by providing continuity of representation in the visual-processing structures
in the brain. Birds, many of which also have excellent binocular vision, integrate the
visual signals from each eye in other ways.
At the optic chiasm, retinal ganglion cell axons must make a decision to either
avoid or traverse the midline of the brain. How this is achieved is not yet clear but it
D M
is likely that short-range guidance molecules, such as ephrins and their Eph recep-
tors, are involved. In mice, for example, the ipsilateral retinal projection arises from
neurons in the peripheral ventro-temporal crescent of the retina (see Fig. 12.23, right
N T A P panel). These neurons express the guidance receptor EphB1 (see Box 5F, p. 205),
which interacts with ephrin B2 on radial glia cells at the optic chiasm to repel the
axons away from the midline and into the ipsilateral optic tract.
V L
Retinal neurons map in a highly ordered manner on to their target structures, with
Retina Tectum
a point-to-point correspondence between a position on the retina and one on the
tectum. Figure 12.23 (left panel) shows how neurons in different positions along the
Fig. 12.24 The retina maps on to the naso-temporal (antero-posterior) axis of the tadpole retina, map in a reverse direction
tectum. In the amphibian tectum dorsal (D)
to points along the antero-posterior axis of the optic tectum. A similar reversed polar-
retinal neurons connect to the lateral (L) side
ity is seen in the mapping of neurons along the dorso-ventral axis of the retina: dorsal
of the tectum and ventral (V) retinal neurons
retinal neurons project to the lateral region of the tectum and neurons from the ventral
connect to the medial (M) tectum. Similarly,
temporal (T) (or posterior) retinal neurons region of the retina project to the medial region of the tectum (Fig. 12.24).
connect to anterior (A) tectum and nasal (N) Remarkably, in some lower vertebrates, such as fish and amphibians, the pattern of
(or anterior) retinal neurons connect to the connections can be re-established with precision when the optic nerve is cut. The ends
posterior (P) tectum. of the axons distal to the cut die, new growth cones form, and axon outgrowth reforms
 Axon navigation 491

The neurons in each sector of the

retina send their axons to a Optic nerve severed, Neurons regenerate and
corresponding sector in the eye rotated 180o re-establish original
connections direction of
opposite tectum target (fly)

left retina right retina

D V
V
T N T N
N T
V D D
direction of
attack
a a a
l m l m l m
p p p

left tectum right tectum

Fig. 12.25 Retino-tectal connections in amphibians are respectively). Center and right panels: if one optic nerve of a frog is cut
re-established in the original arrangement after severance of the and the eye rotated dorso-ventrally through 180°, the severed ends
optic nerve and rotation of the eye. First panel: neurons in the optic of the axons degenerate. When the neurons regenerate, they make
nerve from the left eye mainly connect to right optic tectum and those connections with their original sites of contact in the tectum. However,
from the right eye to the left optic tectum. There is a point-to-point because the eye has been rotated, the image falling on the tectum is
correspondence between neurons from different regions in the retina upside down compared with normal. When the frog sees a fly above its
(nasal (N), temporal (T), dorsal (D), and ventral (V)) and their connections head, it thinks the fly is below it, and moves its head downward to try
in the tectum (posterior (p), anterior (a), lateral (l), and medial (m), to catch it (fourth panel).

connections to the tectum. In frogs, even if the eye is inverted through 180°, the axons
still find their way back to their original sites of contact (Fig. 12.25). However, the
animals subsequently behave as if their visual world has been turned upside down: if
a visual stimulus, such as a fly, is presented above the inverted eye, the frog moves its
head downward instead of upward, and can never learn to correct this error.
From such experiments it was suggested that each retinal neuron carries a chemical
label that enables it to connect reliably with an appropriately chemically labeled cell
in the tectum. This is known as the chemoaffinity hypothesis of connectivity. In the
retino-tectal projection, it is thought that graded spatial distributions of a relatively
small number of factors on the tectum provide positional information, which can be
detected by the retinal axons. The spatially graded expression of another set of factors
on the retinal axons would provide them with their own positional information. The
development of the retino-tectal projection could thus, in principle, result from the
interaction between these two gradients. Cultured temporal retinal ganglion cells grow only
onto anterior tectal tissue
Such gradients were first found in the developing visual system of the chick embryo.
An axon-guiding activity based on repulsion has been detected along the antero-pos- Tectal membranes
terior axis of the chick tectum. Normally, the temporal (posterior) half of the retina Posterior
Temporal retinal ganglion cells

projects to the anterior part of the tectum and the nasal (anterior) retina projects to
the posterior tectum. When offered a choice between growing on posterior or ante- Anterior
rior tectal cells, temporal axons from an explanted chick retina show a preference for
Posterior
anterior tectal cells (Fig. 12.26). This choice is mediated by repulsion of the axons—as
shown by the collapse of their growth cones—by a factor located on the surface of Anterior
posterior tectal cells. Ephrins and their receptors are the likely molecules mediating
this repulsion as they are expressed in reciprocal gradients in retina and tectum. Posterior

In the mouse, the role of ephrins and Ephs in creating a map has been investigated
in the retino-collicular pathway, and repulsive interactions between the EphA family of
receptors and their ephrin A ligands have been found to pattern this map in the antero-
Fig. 12.26 Choice of targets by retinal
posterior dimension. Mapping of the retina on to the superior colliculus is similar to axons. If pieces of temporal retina are placed
that from the retina to the optic tectum: retinal neurons from the nasal region map to next to a ‘carpet’ of alternating stripes
the posterior colliculus and neurons from the temporal retina map to the anterior col- (90 µm wide) of anterior and posterior tectal
liculus. EphAs are expressed in a gradient in the mouse retina, low in the nasal region, membranes, the temporal retinal cells only
high in the temporal region, whereas ephrin As are graded in the superior colliculus extend axons on to anterior tectal membranes.
492 Chapter 12 Development of the nervous system

from high posterior to low anterior. Thus retinal neurons with the most EphA receptors
Retina Superior colliculus
connect to collicular targets with the fewest ephrin As. Retinal axons bearing low levels
Nasal Temporal Anterior Posterior
of the receptor move into areas containing high levels of the ligands, whereas retinal
axons bearing high levels of receptor stop before they reach areas with high levels of
ephrin A (Fig. 12.27). A possible explanation for this behavior is that binding of ligand
to the receptor sends a repulsive signal, and the axon ceases to extend when this signal
reaches a threshold value. The strength of the signal received by the migrating axon
Expression of: EphA ephrin As will be proportional to the product of the concentrations of receptor and ligand. Thus,
the threshold will be reached when there is a high level of both receptor and ligand,
Fig. 12.27 Complementary expression but not when there is a low level of receptor, even when ligand level is high.
of the ephrins and their receptors in the Mapping from the dorso-ventral dimension of the retina to the medial-lateral dimen-
mouse retino-collicular projection. The sion of the superior colliculus or optic tectum also involves graded expression of Ephs
superior colliculus is a structure in the and ephrins. In this case, EphB is graded in the retinal neurons from high-ventral to
mammalian midbrain to which some retinal low-dorsal and ephrin B in the tectum from high-medial to low-lateral. EphB and
neurons project and which is analogous to ephrin B mediate adhesive and attractive interactions, rather than repulsion, and this
the tectum of amphibians and birds. The poses a conceptual problem. If their interaction were the only factor at work, all the
receptor EphA is a receptor tyrosine kinase
retinal neurons would be attracted to the area of highest ephrin B, rather than map-
and ephrin As are its ligands. Temporal retinal
ping across the whole tectum. In this case, the necessary repulsive interactions in
neurons, which express high levels of EphA
these dimensions appear to be provided by gradients of the signaling protein Wnt3 in
on their surface, are repulsed by the posterior
tectum, where high levels of ephrin A are the tectum and a receptor, Ryk, on the retinal axons.
present, but can make connections with the
anterior tectum. Nasal retinal neurons make
Summary
the best contacts in the posterior tectum, as
they express low levels of EphA and thus are
Growth cones at the tip of the extending axon guide it to its destination. Filopodial
repulsed only where the levels of ephrins are
activity at the growth cone is influenced by environmental factors, such as contact
highest.
with the substratum and with other cells, and can also be guided by chemotaxis. Guid-
ance involves both attraction and repulsion. In the development of motor neurons that
innervate vertebrate limb muscles, the growth cones guide the axons so as to make the
correct muscle-specific connections, even when their normal site of entry into the limb
is disturbed. Combinations of transcription factors give each motor neuron an identity
that determines its pathway. Attraction and repulsion control axons crossing the mid-
line in both Drosophila and vertebrates. Gradients in diffusible molecules are probably
responsible for the directional growth of commissural axons in the spinal cord. Neuronal
guidance of axons from the retina to make the correct connections with the optic tec-
tum involves gradients in cell-surface molecules both on the tectal neurons and on the
retinal axons that can promote or repel growth cone approach, and there is competition
between neurons for sites.

Summary: axonal guidance

axonal outgrowth led by growth cone

attractive and repulsive cues are both short range and long range

motor neurons take attraction and repulsion retinal axons connect with
distinct pathways in limb control midline crossing tectal neurons to make
in Drosophila and vertebrates retino-tectal map

trajectory in limb connections guided by

controlled by EphA signaling gradients; EphA on tectum
and ephrins on retinal cells
 Synapse formation and refinement 493

Synapse formation and refinement

Having seen the outcome of axonal navigation, we now focus on the specialized
connections—the synapses—that the axons make with their targets and which are
essential for signaling between neurons and their target cells. Formation of synapses
in the correct pattern is a basic requirement of any developing nervous system. The
connections may be made with other nerve cells, with muscles, and also with certain
glandular tissues. Here, we focus mainly on the development and stabilization of syn-
apses at the junctions between motor neurons and muscle cells in vertebrates, which
are called neuromuscular junctions. There are very many neuronal cell-types—
hundreds if not thousands—and how they match up to make synapses reliably is a
central problem. The molecular mechanisms are only beginning to be known. Studies
on the vertebrate neuromuscular junction have guided much of the understanding of
synapse formation.
Setting up the organization of a complex nervous system in vertebrates involves
refining an initially rather imprecise organization by extensive programmed cell death
(see Section 10.12). The establishment of a connection between a neuron and its
target appears to be essential not only for the functioning of the nervous system, but
also for the very survival of many neurons. Neuronal death is very common in the
developing vertebrate nervous system; too many neurons are produced initially and
only those that make appropriate connections survive. Survival depends on the neu-
ron receiving neurotrophic factors, such as nerve growth factor, which are produced
by the target tissue and for which neurons compete.
A special feature of nervous-system development is that fine-tuning of synaptic
connections depends on the interaction of the organism with its environment and the
consequent neuronal activity. This is particularly true of the vertebrate visual system,
where sensory input from the retina in a period immediately after birth modifies syn-
aptic connections so that the animal can perceive fine detail. Again, this refinement
seems to involve competition for neurotrophic factors. We return to this topic after
first considering synapse formation.

12.13 Synapse formation involves reciprocal interactions

We will start our look at synapse formation with the neuromuscular junction, as this
is one of the most intensively studied and best-understood types of synapse. The
­mature neuromuscular junction is a complex structure involving extensive modifica-
tion of the nerve ending and the muscle cell membrane (Fig. 12.28). Just before the
junction, the axon terminal branches into a network-like arrangement. Each branch
ends in a swelling, which is in contact with a special endplate region on the muscle
fiber. The axon’s plasma membrane is separated from the muscle-cell’s plasma mem-
brane by a narrow cleft (the synaptic cleft) filled with extracellular material secreted
by both the neuron and the muscle cell. A basal lamina of extracellular material
forms around the muscle-cell surface. The structure comprising axon plasma mem-
brane, the opposing muscle-cell plasma membrane, and the cleft between them, is
a synapse.
Electrical signals cannot pass across the synaptic cleft, and for the neuron to signal
to the muscle, the electrical impulse propagated down the axon is converted at the
terminal into a chemical signal. This is a chemical neurotransmitter that is released
into the synaptic cleft from synaptic vesicles in the axon terminal. Molecules of the
neurotransmitter diffuse across the cleft and interact with receptors on the muscle-cell
membrane, causing the muscle fiber to contract. The neurotransmitter used by motor
neurons connecting with skeletal muscles is acetylcholine. Because the signal travels
from nerve to muscle, the axon terminal is called the pre-synaptic part of the junction
and the muscle cell is the post-synaptic partner (see Fig. 12.28).
494 Chapter 12 Development of the nervous system

neuron

myelinated axon
Schwann cell

neuromuscular
junction
muscle fiber

Schwann cell
axon

Endplate region

pre-synaptic
terminal
synaptic cleft

Fig. 12.28 Structure of the vertebrate mitochondrion

neuromuscular junction. The motor
neuron axon, which is covered in a myelin
sheath produced by the glial Schwann synaptic vesicle
containing acetylcholine
cell, innervates a muscle fiber. At the
neuromuscular junction, the axon branches pre-synaptic membrane
in the endplate region, and makes synaptic synaptic cleft
post-synaptic membrane
connections with the muscle membrane.
Communication between nerve and muscle basal lamina
is by release of the neurotransmitter
acetylcholine receptors
acetylcholine from synaptic vesicles into the
synaptic cleft. Acetylcholine diffuses across junctional fold
the cleft and binds to acetylcholine receptors
on the muscle cell membrane.
After Kandel, E.R., et al.: 1991.

Development of a neuromuscular junction is progressive. Before the arrival of

the axon there is already some pre-patterning on the muscle, with acetylcholine
receptors concentrated in the central region of the muscle fiber. When the motor
neuron terminal arrives, it releases the proteoglycan Agrin into the basal lamina,
which binds to, and activates, the transmembrane muscle-specific kinase (MuSK),
whose activity is required to maintain the clustering of acetylcholine receptors and
the specialization of the muscle-cell surface at the site (Fig. 12.29). Signals from the
muscle in turn induce differentiation of the pre-synaptic zone on the axon terminal
 Synapse formation and refinement 495

Growth cone of axon approaches Growth cone forms contact with Pre-synaptic terminal Fig. 12.29 Development of the
muscle fiber. Pre-pattern on muscle fiber. differentiates as a result of signals neuromuscular junction. Before contact
muscle fiber Post-synaptic induction from muscle cell by an axon, acetylcholine receptors (AChR)
are gathered in a central region of the
axon terminal synaptic vesicles muscle fiber. When a motor axon growth
cone contacts a muscle fiber it releases the
Agrin proteoglycan Agrin (blue arrow) into the
extracellular material. Agrin binds to and
activates the muscle-cell kinase MuSK, which
AChR in turn maintains clustering of AChR in the
muscle-cell membrane and specialization of
the post-synaptic surface. A signal (as yet
MuSK
muscle fiber unknown) from the muscle cell (purple arrow)
then helps to induce the differentiation of
the pre-synaptic terminal, with the formation
and align this zone with the post-synaptic area on the muscle. These signals are not of synaptic vesicles. Synaptic signaling by
acetylcholine (small red circles) released from
yet clearly identified, but the glycoprotein laminin β2 in the basal lamina is ­required
the synaptic vesicles can then begin.
for correct alignment and full pre-synaptic differentiation. The growth factor neu-
regulin-1, which is related to epidermal growth factor and acts through the ErbB
receptors, is required for the formation of some synapses, including neuromuscular
junctions, where it helps mediate the expression of acetylcholine receptors on the
muscle cell.
Neuregulin-1 is also involved in the development of glial cells, including the
Schwann cells of the peripheral nervous system. In vertebrates, nerve impulses often
have to travel long distances along axons in peripheral nerves—up to several meters
in a large animal—and rapid propagation of nerve impulses over such distances is
achieved by myelination of axons by glial cells. Schwann cells in the peripheral ner-
vous system and oligodendrocytes in the central nervous system wrap themselves
around axons so that their cell membranes form a fatty, insulating layer of myelin.
Depolarization of the neuron membrane then only occurs at the small areas of bare
neuronal membrane between one insulating cell and the next, and the nerve impulse
is able to ‘jump’ over these areas without loss of signal. The progressive loss of neural
function in the human disease multiple sclerosis is due to the destruction of myelin,
and eventually leads to severe muscle weakness due to the reduction in signals from
nerve to muscle.
The neuromuscular junction is a highly specialized type of synapse, but synapse
formation within the vast network of neurons that form the central nervous system
follows similar principles. Synapses are made between the axon terminal of one
neuron and a dendrite, cell body, or more rarely the axon, of another neuron. An
individual motor neuron in the spinal cord, for example, will receive input from
thousands of other neurons on its dendrites and cell body. As with neuromuscular
junctions, interneuronal synapse-formation is likely to involve reciprocal signal-
ing between axon and dendrite. Synapses between neurons can form quite rap-
idly, on a time scale of about an hour, and observations suggest that filopodia on
the dendrite initiate synapse formation by reaching out to the axon (Fig. 12.30).
Signals from the axon, including release of neurotransmitter, may initially guide
the dendrite filopodia. Many other factors have been shown to promote synapse
formation, including cell-adhesion molecules and Wnt signaling. Cell-adhesion
molecules implicated in synapse formation include the neuroligins, which are
post-synaptic membrane proteins that bind to proteins called β-neurexins on the
pre-synaptic axon surface, causing the neurexins to cluster, which induces differen-
tiation in the pre-synaptic terminal. Neuroligins promote the function of excitatory
synapses rather than inhibitory synapses. Cadherins, protocadherins, and members
of the immunoglobulin superfamily (see Box 8B, p. 293) have also been found
localized to synapses.
496 Chapter 12 Development of the nervous system

0 minutes 40 minutes 60 minutes Maturation

mitochondrion

axon

actin dendritic dendritic

filopodium spine

dendrite

active zone precursor complex active zone proteins synaptic vesicle precursor complex synaptic vesicle neurotransmitter receptor scaffold protein cell-adhesion molecule

Fig. 12.30 Interneuronal synapse

formation. A dendrite sends out filopodia In Drosophila, the cell-adhesion molecule Dscam, a member of the immunoglobulin
that contact the axon growth cone. superfamily, is required both for axon guidance in the embryonic nervous system and
Reciprocal signaling (red arrows) initiates in specifying particular neuronal connections. It initiates a signaling pathway that
synapse formation aided by cell-adhesion leads to activation of Rho-family GTPases that affect axonal migration by causing
molecules. In the axon, proteins that compose changes in the actin cytoskeleton. The Dscam gene is exceptional in that alternative
the pre-synaptic active zone gather at the splicing could produce an estimated 38,000 distinct forms of cell-surface receptor,
site of contact along with synaptic vesicle thus raising the possibility that different forms of the protein might be produced in
precursors. In the dendrite, neurotransmitter different neurons.
receptors are synthesized and inserted into
the post-synaptic membrane along with post-
synaptic scaffold proteins. The post-synaptic 12.14 Many motor neurons die during normal development
connection develops into typical dendritic
spines. During development, some 20,000 motor neurons are formed in the segment of spinal
cord that provides innervation to a chick leg, but about half of them die soon after
they are formed (Fig. 12.31). Cell death occurs after the axons have grown out from
Number of 22 the cell bodies and entered the limb, at about the time that the axon terminals are
neurons
(thousands) reaching their potential targets—the skeletal muscles of the limb. The role of the
20 target muscles in preventing cell death is suggested by two experiments. If the leg
bud is removed, the number of surviving motor neurons decreases sharply. When an
18 additional limb bud is grafted at the same level as the leg, providing additional targets
for the axons, the number of surviving motor neurons increases.
16 Survival of a motor neuron depends on its establishing contacts with a muscle cell.
Once a contact is established, the neuron can activate the muscle, and this is followed
by the death of a proportion of the other motor neurons that are approaching the
14
muscle cell. Muscle activation by neurons can be blocked by the drug curare, which
prevents neuromuscular signal transmission, and this block results in a large increase
12
in the number of motor neurons that survive.
Incubation (days) 5 10 15 18 Even after neuromuscular connections have been made, some are subsequently
eliminated. At early stages of development, single muscle fibers are innervated by
Developmental 28 31 35 38 43
stage axon terminals from several different motor neurons. With time, most of these con-
Hatching
nections are eliminated, until each muscle fiber is innervated by the axon terminals
from just one motor neuron (Fig. 12.32). This is due to competition between the
Fig. 12.31 Motor neuron death is a
synapses, with the most powerful input to the target cell destabilizing the less power-
normal part of development in the chick
ful inputs to the same target. This elimination process and the maturation of a stable
spinal cord. The number of spinal cord motor
neuromuscular junction take up to 3 weeks in the rat.
neurons innervating a chick limb decreases
by about half before hatching, as a result of The well-established matching of the number of motor neurons to the number of
programmed cell death during development. appropriate targets by these mechanisms suggests that a general mechanism for the
Most of the neurons die over a period of development of nervous-system connectivity in all parts of the vertebrate nervous
4 days. system is that excess neurons are generated and only those that make the required
 Synapse formation and refinement 497

connections are selected for survival. This mechanism is well-suited to regulating

Embryo
cell numbers by matching the size of the neuronal population to its targets. We
now look in more detail at the neurotrophic factors that promote neuronal survival,
and then consider the role of neural activity in the elimination of neuromuscular
synapses.

12.15 Neuronal cell death and survival involve both intrinsic

muscle fiber within a single muscle
and extrinsic factors

Neuronal death during development takes place as a result of the activation of the
Adult
apoptotic pathway (see Section 10.12). Apoptosis can be triggered or prevented either
by an intrinsic developmental program or by external factors. In the development of
the nematode, for example, particular neurons and other cells are intrinsically pro-
grammed to die (see Chapter 6), and the specificity of cell death can be shown to be
under genetic control.
As we saw in Section 10.12, members of the Bcl-2 protein family control the apop-
totic pathway in vertebrates—some acting to promote apoptosis and some to prevent
it. A key family member promoting neuronal apoptosis during development is Bax;
mice mutant for Bax have increased numbers of motor neurons in the face. A protein Fig. 12.32 Refinement of muscle
called Survivin is known to inhibit apoptosis during early neurogenesis. Mutant mice innervation by neural activity. Initially,
in which Survivin is deleted from embryonic day 12.5 onwards are born with much several motor neurons innervate the same
smaller brains than normal and numerous foci of apoptosis throughout the brain and muscle fiber. Elimination of synapses means
spinal cord and die soon after birth. that each fiber is eventually innervated by
In many developmental situations it seems that cells are programmed to die unless only one neuron.
they receive specific survival signals. Extracellular factors play a key role in neu- After Goodman, C.S. and Shatz, C.J.: 1993.
ronal survival in vertebrates. The first one identified was nerve growth factor (NGF),
whose discovery was due to the serendipitous observation that a mouse tumor
implanted into a chick embryo evoked extensive growth of nerve fibers towards the
tumor. This suggested that the tumor was producing a factor that promoted axonal
outgrowth. The factor was eventually identified as a protein, NGF, using axon out-
growth in culture as an assay. NGF is necessary for the survival of a number of types
of neurons, particularly those of the sensory and sympathetic nervous systems.
NGF is a member of a family of proteins known as the neurotrophins. Different
types of neuron require different neurotrophins for their survival, and the require-
ment for certain neurotrophins also changes during development. The neurotrophin
GDNF, for example, prevents the death of facial motor neurons, and is also expressed
in developing limb buds. There is increasing evidence that individual neurotrophins
act on a variety of neural cell types.

12.16 The map from eye to brain is refined by neural activity

The development and function of the nervous system depends not only on synapse
formation, but also on the regulated disassembly of previously functional synaptic
connections. We have already considered how axons from the retina make connec-
tions with the tectum so that a retino-tectal map is established. This map is initially
rather coarse-grained, in that axons from neighboring cells in the retina make contacts
over a large area of the tectum. This area is much larger early on than at later stages
of development, when the retino-tectal map is more finely tuned. Fine-tuning of the
map results, as in muscle, from the withdrawal of axon terminals from most of the
initial contacts, and requires neural activity. This requirement is seen particularly
clearly in the development of visual connections in mammals. If a mammal, including
a human, is deprived of vision during a critical period, then it will suffer from poor
visual acuity and will not be able to resolve fine detail.
498 Chapter 12 Development of the nervous system

In mammals with good binocular and stereoscopic vision, such as monkeys and
Birth
humans, visual-system axons from the retina first connect to the lateral geniculate
nucleus, on to which they map in an ordered manner. Input from half of each eye goes
to the opposite side of the brain, whereas input from the other half of each eye goes
to the same side of the brain (see Fig. 12.23, right panel).
2 weeks
In the lateral geniculate nucleus, as in the cortex, the neurons are arranged in lay-
ers. Each layer receives input from retinal axons from either the right or the left eye,
but not both. Thus, inputs from left and right eyes are kept separate. Neurons from
the lateral geniculate nucleus then send axons to the visual cortex (see Fig. 12.23,
3 weeks
right panel). When there is a visual stimulus, the inputs from the retinal axons acti-
vate neurons in the lateral geniculate nucleus, which then activate neurons in the
corresponding region of the visual cortex. The adult visual cortex consists of six cell
layers, but we need only focus here on layer 4, which is where many axons from the
6 weeks
lateral geniculate nucleus make connections. Layers in the lateral geniculate nucleus
that receive left or right inputs both make connections with layer 4 of the visual cor-
tex. This means that input from corresponding positions in both eyes, that is from the
same part of the visual field, finally arrives at the same location in the cortex (see
Fig. 12.23, right panel).
At birth, the nerve endings from the two eyes overlap and are mixed at their
common final location, but with time the inputs from left and right eyes become
separated into blocks of cortical cells about 0.5 mm wide, which are known as
ocular dominance columns (Fig. 12.33). Adjacent columns respond to the same
stimulus in the visual field; one column responding to signals from the left eye,
and the next to signals from the right eye. This arrangement enables good binocu-
lar and stereoscopic vision. The columns can be detected and mapped by making
Fig. 12.33 Visualization of ocular electrophysiological recordings. They can also be directly observed by injecting
dominance columns in the monkey visual a tracer, such as radioactive proline, into one eye. The tracer is taken up by reti-
cortex. A radioactive tracer is injected into nal neurons, transported by the optic nerve to the lateral geniculate, and from
one eye, from where it is transported to the there to the visual cortex, where its pattern can be detected by autoradiography.
visual cortex through the neurons. Tracer
This reveals a striking array of stripes representing the input from one eye (see
injected at birth is broadly distributed in the
Fig. 12.33).
cortex. Tracer injected at later times becomes
Neural activity and visual input are essential for the development and maintenance
confined to alternating columns of cortical
cells (brighter stripes), representing the of the ocular dominance columns. While sensory input is important, spontaneous
ocular dominance columns for that eye, as activity also plays a key role. Initial formation of the stripes in non-human primates
seen in the photograph. Scale = 1 mm. occurs before visual experience, and involves spontaneously generated waves of ­action
After Kandel, E.R., et al.: 1995. potentials in the mammalian retina. These waves may cause their effects through the
release of neurotrophins that can remodel synaptic connections. If neural activity is
blocked during development by the injection of tetrodotoxin, ocular dominance col-
umns do not develop, and the inputs from the two eyes into the visual cortex remain
mixed. If the input from one eye is blocked, the territory in the visual cortex occupied
by the other eye’s input expands at the expense of the blocked eye. The release of
the inhibitory neurotransmitter γ-aminobutyric acid (GABA) is required to initiate the
critical period.
The favored explanation for the formation of ocular dominance columns is based
on competition between incoming axons (Fig. 12.34). Because of the initial connec-
tions that establish the map, individual cortical neurons can initially receive input
from both eyes. Within a particular region, there will, therefore, be overlap of stimuli
originating from the two eyes, and this overlap has to be resolved. Neighboring cells
carrying input from the same eye tend to fire simultaneously in response to a visual
stimulus; if they both innervate the same target cell, they can thus cooperate to excite
it. As in muscle, stimulation of electrical activity in the target cell tends to strengthen
the active synapses and suppress those that are not active at the time—cells that
 Synapse formation and refinement 499

fire together, wire together. As there is competition between neurons for targets, this
Initial projection
could generate discrete regions of cortical cells that respond only to one eye or the
other, and so form the ocular dominance columns. Such a mechanism explains why layers of visual cortex
experimental exposure of animals to continuous strobe lighting after birth, which 1
2
causes simultaneous firing of neurons in both eyes, prevents the formation of ocular 3
dominance columns. 4
5
In frogs the optic tecta do not normally receive input from both eyes until after mor- 6
phogenesis and ocular dominance columns do not develop in the tecta under normal
circumstances. Tecta that receive inputs from two eyes throughout development can, left eye right eye
axon axon
however, be produced by implanting a third eye primordium into the embryo. Retinal
axons from the two eyes then segregate within the tectum and form alternating stripes
lateral
that resemble the pattern of ocular dominance columns in cats and primates. In these geniculate
nucleus
tecta, the dendrites of some types of tectal neurons seem to respond to the stripe
boundaries, as evidenced by marked changes in their normal direction of migration or
by their termination at a stripe border. The behavior of these neurons in this artificial
situation is further evidence that in all vertebrates, not simply mammals, highly local- Adult
ized interactions between incoming axon terminals and the post-synaptic dendrites
with which they connect are critical in producing the patterning of a segregated,
topographically organized cortex.
One possible mechanism for refining connections in response to neuronal activity
involves local release of neurotrophins. A certain level of activation, or activation
by two axons simultaneously, could induce the release of neurotrophins from the ocular
dominance
target cells, and only those axons that have been recently active might be able to column
respond to them. When neural activity in developing chick or amphibian embryos
is blocked with drugs that prevent neuronal firing, no fine-grained retino-tectal map
develops.

Summary Fig. 12.34 Development of ocular

dominance columns. Initially, neurons from
Neurons communicate with each other and with target cells, such as muscles, by the lateral geniculate nucleus, representing
means of specialized junctions called synapses. The formation of a neuromuscular projections from both eyes and stimulated
junction involves changes in both pre-synaptic (neuronal) and post-synaptic (muscle by the same visual stimulus, project to the
cell) membranes once contact is made, and depends on reciprocal signaling between same region of the visual cortex. (Only
muscle and nerve cell. Reciprocal interactions also occur in the formation of synapses projection to layer 4 is shown here.) With
between axons and dendrites, and cell-adhesion molecules are likely to be involved. visual stimulation, the neuronal connections
Initially, most mammalian muscle fibers are innervated by two or more motor axons, separate out into columns, each representing
but nervous activity results in competition between synapses so that a single fiber is innervation from only one eye. If stimulation
by vision is blocked, ocular dominance
eventually innervated by only one motor neuron. Many neurons produced in the devel-
columns do not form.
oping nervous system die. About half of the motor neurons that initially innervate
After Goodman, C.S. and Shatz, C.J.: 1993.
the vertebrate limb undergo cell death; those that survive do so because they make
functional connections with muscles. Many neurons depend on neurotrophins, such
as nerve growth factor, for their survival, with different classes of neurons requiring
different neurotrophins. Brain function is based not only on the assembly of synapses,
but also on their regulated disassembly. Neural activity has a major role in refining the
connections between the eye and the brain. In mammals, input from the left and right
eyes is required for the development of ocular dominance columns in the visual cortex.
These are adjacent columns of cells responding to the same stimulus from left and
right eyes, respectively. Formation of the columns, which are essential for binocular
vision, is a result of competition for cortical targets by axons carrying visual input from
different eyes.
500 Chapter 12 Development of the nervous system

Summary: synapse formation and refinement

Neuromuscular junction Neuron–neuron synapse

axon of motor neuron releases dendrite extends filopodia which

Agrin at junction with muscle contact axon terminal;
cell-adhesion molecules form initial
contact between dendrite and axon

acetylcholine receptors cluster at pre-synaptic active zone assembles;

junction and local synthesis of neurotransmitter receptors cluster on
receptors occurs pre-synaptic membrane

electrical activity of muscle

reduces receptor synthesis
elsewhere

neurotrophins are required for retino-tectal map refined by neuronal

neuronal survival—about 50% activity, leading to ocular dominance
of limb motor neurons die columns

Summary to Chapter 12

The processes involved in the development of the nervous system, with its numerous con-
nections, are similar to those found in other developmental systems. Presumptive neural
tissue is specified early in development—during gastrulation in vertebrates, and when the
dorso-ventral axis is patterned in Drosophila. Within the neuroectodermal tissue, the speci-
fication of cells that give rise to neural cells involves lateral inhibition. The further devel-
opment of neurons from neuronal precursors involves both asymmetric cell divisions and
cell–cell signaling. The patterning of different types of neurons within the vertebrate spinal
cord is due to both ventral and dorsal signals. As they develop, neurons extend axons and
dendrites. The axons are guided to their destination by growth cones at their tips. Guid-
ance is due to the growth cone’s response to attractive and repulsive signals, which may
be diffusible or bound to the substratum. Gradients in such molecules can guide the axons
to their destination, as in the retino-tectal system of vertebrates. The functioning of the
nervous system depends on the establishment of specific synapses between axons and
their targets. Specificity appears to be achieved by an initial overproduction of neurons that
compete for targets, with many neurons dying during development. Refinement of synaptic
connections involves further competition. Neural activity plays a major role in refining con-
nections, such as those between the eye and the brain.

n End of chapter questions 3. How do Numb and Pins proteins collaborate to cause one daugh-
ter cell to become a neuron after (a) division of the neuroblast in the
Long answer (concept questions) Drosophila embryo? (b) After division of the sensory organ precur-
sor in the Drosophila adult?
1. In Drosophila, the central nervous system is ventral, whereas
in vertebrates, the central nervous system is dorsal. Despite these 4. What are radial glial cells? Include in your answer: the embryonic
differences, what similarities exist that would indicate a common structure in which they are found, their location in that structure, and
evolutionary origin? their roles in development of the nervous system (name two roles).
2. Outline the mechanism of lateral inhibition in the proneural clus- 5. Although the adult mammalian brain cannot in general produce
ter of Drosophila, in which the expression of Delta in one cell leads new neurons, the wall of the lateral ventricle is an exception. Relate
to loss of Delta in a neighboring cell. Compare the Drosophila path- this observation to the embryological origin of neurons in the neural
way to that in vertebrates. tube.
 Synapse formation and refinement 501

6. The formation of motor neurons in the developing spinal cord c) neural crest cells that migrate through the anterior half of the
depends on a gradient of Sonic hedgehog (Shh) protein. What is the somites
source of this Shh gradient? What Class I and Class II homeodomain d) somitic mesoderm
protein genes are active in response to the precise level of this gradi-
ent required for motor neuron progenitors (pMN) to become speci- 3. What would be the consequence of reducing Delta expression in
fied as motor neurons (MN)? What role do the LIM-homeodomain the developing vertebrate neural tube?
proteins play in motor neuron identity? a) Excess neurons are produced, because reduced Delta prevents
7. Compare and contrast netrins, semaphorins, cadherins, and eph- the inhibition of neurogenesis in neighboring cells.
rins: in what way are netrins and cadherins similar in effect, and b) Excess neurons are produced, because the reduction in Delta
in what way are they different? In what way are semaphorins and expression increases neurogenin expression.
ephrins similar, and in what way do they differ?
c) Fewer neurons are produced, because Delta signaling is required
8. What is meant by the phrase ‘the retina maps onto the tectum’ for neurogenesis.
(see Fig. 12.26)? How does the chemoaffinity hypothesis attempt to d) There is no change in neurogenesis unless Delta expression is
explain this phenomenon? How do the ephrins and their receptors eliminated completely.
fit into this hypothesis?
4. Which of the following is a determined neuronal cell type?
9. Describe the results of the experiment in which the optic nerve
to a frog’s eye is cut and the eye rotated by 180°. How is the subse- a) premigratory neural crest cells
quent response of the frog to a visual stimulus explained? b) floor plate cells
10. What is the role of neurotrophic factors, such as nerve growth c) commissural cells
factor, in refining the connections between neurons and their tar- d) roof plate cells
gets? Give a specific example of such action.
5. Cortical neurons migrate to their final positions in the cortex along
11. Draw a schematic diagram of a synapse, labeling the presyn-
aptic cell and the post-synaptic cell and indicating the direction of a) astrocytes
signal transmission. Label the synaptic cleft and say why it prevents b) commissural neurons
neurons from passing electrical signals directly to each other? How
c) radial glia
do neurons communicate across the synaptic cleft?
d) roof-plate cells
12. Describe the steps involved in the formation of the neuromus-
cular junction, starting with the agrin signal from the neuronal 6. Axons of spinal cord neurons from the lateral motor column in
growth cone. the chick embryo innervate
13. Cell death (apoptosis) in the developing nervous system is a) muscles of the body wall
under the control of both intrinsic and extrinsic influences. Give b) muscles of the limb buds
examples of intrinsic (i.e. intracellular) and extrinsic (i.e. cell–cell
c) muscles attached to the spine
signaling) proteins that control apoptosis in the nervous system dur-
ing development. d) muscles in the gut wall

14. Contrast the retinal inputs to the lateral geniculate nucleus in the 7. Which of these is a cell-surface adhesion molecule that mainly
mammalian brain with those to the tectum in the amphibian brain. attracts axons during their outgrowth?
What are ocular dominance columns, and why are they important?
a) neurotrophins
What is meant by ‘cells that fire together, wire together’?
b) cadherins
Multiple choice (factual recall questions) c) netrins
NB There is only one correct answer to each question. d) semaphorins

1. The genes of the achaete–scute complex encode proteins of 8. Knock-outs of the mouse netrin-1 gene have a phenotype similar
which type? to that of knock-outs of what other gene?
a) cell–cell adhesion molecules a) the DCC receptor
b) cytoskeletal proteins b) the Robo1 receptor
c) membrane transport proteins c) the semaphorin signaling molecule
d) transcription factors d) the Slit signaling molecule

2. The vertebrate central nervous system (brain and spinal cord) is 9. Neurons in the anterior (nasal) region of the right retina will
derived from project axons to
a) cells of the neural tube a) the anterior portion of the left tectum
b) neural crest cells that migrate along the dorso-lateral route b) the anterior portion of the right tectum
502 Chapter 12 Development of the nervous system

c) the posterior portion of the left tectum nervous system: the intermediate neuroblasts defective homeo-
d) the posterior portion of the right tectum box gene specifies intermediate column identity. Genes Dev.
1998, 12: 3591–3602.
10. The synapses formed between motor neurons and muscles are
called 12.2 The development of neurons in Drosophila involves
asymmetric cell divisions and timed changes in gene
a) myotomes
expression
b) neuromuscular junctions Brody, T., Odenwald, W.: Regulation of temporal identities dur-
c) Nieuwkoop centers ing Drosophila neuroblast lineage development. Curr. Opin. Cell
d) retino-tectal projections Biol. 2005, 17: 672–675.
Carmena, A.: Signaling networks during development: the case
11. Schwann cells of asymmetric cell division in the Drosophila nervous system.
a) are glial cells Dev. Biol. 2008, 321: 1–17.
Grosskortenhaus, R., Pearson, B.J., Marusich, A., Doe, C.Q.: Regula-
b) are responsible for myelination of axons in the peripheral ner-
tion of temporal identity transitions in Drosophila neuroblasts.
vous system
Dev. Cell 2005, 8: 193–202.
c) are related to the oligodendrocytes of the central nervous system Jan, Y-N., Jan, L.Y.: Polarity in cell division: what frames thy fear-
d) are all of the above ful asymmetry? Cell 2000, 100: 599–602.
Karcavich, R.E.: Generating neuronal diversity in the Drosophila
12. The neurotransmitter used in the communication between central nervous system: a view from the ganglion mother cells.
­motor neurons and muscles is Dev. Dyn. 2005, 232: 609–616.
a) serotonin
b) dopamine Box 12A Specification of the sensory organs of adult
c) epinephrine Drosophila
d) acetylcholine Gómez-Skarmeta, J.L., Campuzano, S., Modolell, J.: Half a century
of neural prepatterning: the story of a few bristles and many
genes. Nat. Rev. Neurosci. 2003, 4: 587–598.
Multiple choice answer key
Knoblich, J.A.: Asymmetric cell division during animal develop-
1: d, 2: a, 3: a, 4: c, 5: c, 6: b, 7: b, 8: a, 9: c, 10: b, 11: d, 12: d. ment. Nat. Rev. Mol. Cell. Biol. 2001, 2: 11–20.
Modolell, J., Campuzano, S.: The achaete-scute complex as an in-
tegrating device. Int. J. Dev. Biol. 1998, 42: 275–282.
n General further reading
12.3 Specification of vertebrate neuronal precursors also
Denes, A.S., Jékely, G., Steinmetz, P.R., Raible, F., Snyman, H., involves lateral inhibition
Prud’homme, B., Ferrier, D.E., Balavoine, G., Arendt, D.: Molecu- Chitins, A., Henrique, D., Lewis, J., Ish-Horowitcz, D., Kintner, C.:
lar architecture of annelid nerve cord supports common origin Primary neurogenesis in Xenopus embryos regulated by a
of nervous system centralization in bilateria. Cell 2007, 129: ­homologue of the Drosophila neurogenic gene Delta. Nature
277–288. 1995, 375: 761–766.
Kandel, E.R., Schwartz, J.H., Jessell, T.H.: Principles of Neural Sci- Ma, Q., Kintner, C., Anderson, D.J.: Identification of neurogenin,
ence (4th edn). New York: McGraw–Hill, 2000. a vertebrate neuronal determination gene. Cell 1996, 87:
Kerszberg, M.: Genes, neurons and codes: remarks on biological ­43–52.
communication. BioEssays 2003, 25: 699–708.
12.4 Neurons are formed in the proliferative zone of the
neural tube and migrate outwards
n Section further reading D’Arcangelo, G., Curran, T.: Reeler: new tales on an old mutant
mouse. BioEssays 1998, 20: 235–244.
12.1 Neurons in Drosophila arise from proneural clusters Gage, F.H.: Mammalian neural stem cells. Science 2000, 287: 1433–
Cornell, R.A., Ohlen, T.V.: Vnd/nkx, ind/gsh, and msh/msx: con- 1438.
served regulators of dorsoventral neural patterning? Curr. Opin. Kriegstein, A., Alvarez-Buylla, A.: The glial nature of embryonic
Neurobiol. 2000, 10: 63–71. and adult neural stem cells. Annu. Rev. Neurosci. 2009, 32:
Gibert, J.M., Simpson, P.: Evolution of cis-regulation of the proneu- 149–184.
ral genes. Int. J. Dev. Biol. 2003, 47: 643–651. Lee, S.K., Lee, B., Ruiz, E.C., Pfaff, S.L.: Olig2 and Ngn2 function
Skeath, J.B.: At the nexus between pattern formation and cell- in opposition to modulate gene expression in motor neuron
type specification: the generation of individual neuroblast fates progenitor cells. Genes Dev. 2005, 19: 282–294.
in the Drosophila embryonic central nervous system. BioEssays Lyuksyutova, A.I., Lu, C.C., Milanesio, N., King, L.A., Guo, N.,
1999, 21: 922–931. Wang, Y., Nathans, J., Tessier-Lavigne, M., Zou, Y.: Anterior-
Weiss, J.B., Von Ohlen, T., Mellerick, D.M., Dressler, G., Doe, C.Q., ­posterior guidance of commissural axons by Wnt-frizzled sig-
Scott, M.P.: Dorsoventral patterning in the Drosophila central naling. ­Science 2003, 302: 1984–1988.
 Synapse formation and refinement 503

Maricich, S.M., Gilmore, E.C., Herrup, K.: The role of tangential 12.9 The growth cone controls the path taken by the growing
migration in the establishment of the mammalian cortex. axon
­Neuron 2001, 31: 175–178. Carmeliet, P., Tessier-Lavigne, M.: Common mechanisms of nerve
Sauvageot, C.M., Stiles, C.D.: Molecular mechanisms controlling and blood vessel wiring. Nature 2005, 436: 193–200.
cortical gliogenesis. Curr. Opin. Neurobiol. 2002, 12: 244–249. Lowery, L.A., Van Vactor, D.: The trip of the tip: understanding
Wichterle, H., Lieberam, I., Porter, J.A., Jessell, T.M.: Directed dif- the growth cone machinery. Nat. Rev. Mol. Cell. Biol. 2009, 10:
ferentiation of embryonic stem cells into motor neurons. Cell 332–343.
2002, 110: 385–397. Tear, G.: Neuronal guidance: a genetic perspective. Trends Genet.
1999, 15: 113–118.
Tessier-Lavigne, M., Goodman, C.S.: The molecular biology of
Box 12B Timing the birth of cortical neurons
axon guidance. Science 1996, 274: 1123–1133.
Rakic, P.: Neurons in rhesus monkey visual cortex: systematic
­relation between time of origin and eventual disposition. 12.10 Motor neuron axons in the chick limb are guided by
­Science 1974, 183: 425–427. ephrin–Eph interactions
Dasen, J.S., Tice, B.C., Brenner-Morton, S., Jessell, T.M.: A Hox reg-
12.5 The pattern of differentiation of cells along the dorso-
ulatory network establishes motor neuron pool identity and
ventral axis of the spinal cord depends on ventral and dorsal
target-muscle connectivity. Cell 2005, 123: 477–491.
signals
Lance-Jones, C., Landmesser, L.: Pathway selection by embryonic
Jacob, J., Hacker, A., Guthrie, S.: Mechanisms and molecules in chick motoneurons in an experimentally altered environment.
motor neuron specification and axon pathfinding. BioEssays Proc. R. Soc. Lond. B 1981, 214: 19–52.
2001, 23: 582–595. Tosney, K.W, Hotary, K.B., Lance-Jones, C.: Specifying the target
Jessell, T.M.: Neuronal specification in the spinal cord: induc- identity of motoneurons. BioEssays 1995, 17: 379–382.
tive signals and transcriptional controls. Nat. Rev. Genet. 2000,
1: 20–29. 12.11 Axons crossing the midline are both attracted and
Megason, S.G., McMahon, A.P.: A mitogen gradient of dorsal mid- repelled
line Wnts organizes growth in the CNS. Development 2002, 129: Giger, R.J., Kolodkin, A.L.: Silencing the siren: guidance cue hier-
2087–2098. archies at the CNS midline. Cell 2001, 105: 1–4.
Long, H., Sabatier, C., Ma, L., Plump, A., Yuan, W., Ornitz, D.M.,
12.6 Neuronal subtypes in the ventral spinal cord are Tamada, A., Murakami, F., Goodman, C.S., Tessier-Lavigne, M.:
specified by the ventral to dorsal gradient of Shh Conserved roles for Slit and Robo proteins in midline commis-
Dessaud, E., McMahon, A.P., Briscoe, J.: Pattern formation in the sural axon guidance. Neuron 2004, 42: 213–223.
vertebrate neural tube: a sonic hedgehog morphogen-regulated Simpson, J.H., Bland, K.S., Fetter, R.D., Goodman, C.S.: Short-range
transcriptional network. Development 2008, 135: 2489–2503. and long-range guidance by Slit and its Robo receptors: a com-
Stamataki, D., Ulloa, F., Tsoni, S.V., Mynett, A., Briscoe, J.: A gradi- binatorial code of Robo receptors controls lateral position. Cell
ent of Gli activity mediates graded Sonic Hedgehog signaling in 2000, 103: 1019–1032.
the neural tube. Genes Dev. 2005, 19: 626–641. Williams, S.E., Mason, C.A., Herrera, E.: The optic chiasm as a
midline choice point. Curr. Opin. Neurobiol. 2004, 14: 51–60.
12.7 Spinal cord motor neurons at different dorso-ventral Woods, C.G.: Crossing the midline. Science 2004, 304: 1455–1456.
positions project to different trunk and limb muscles Zou, Y., Lyuksyutova, A.I.: Morphogens as conserved axon guid-
Kania, A., Jessell, T.M.: Topographic motor projections in the limb ance cues. Curr. Opin. Neurobiol. 2007, 17: 22–28.
imposed by LIM homeodomain protein regulation of ephrin-A:
EphA interactions. Neuron 2003, 38: 581–596. 12.12 Neurons from the retina make ordered connections
Kania, A., Johnson, R.L., Jessell, T.M.: Coordinate roles for LIM with visual centers in the brain
­homeobox genes in directing the dorsoventral trajectory of Feldheim, D.A., Kim, Y-I., Bergemann, A.D., Frisen, J., Barbacid, M.,
­motor axons in the vertebrate limb. Cell 2000, 102: 161–173. Flanagan, J.G.: Genetic analysis of ephrin A2 and ephrin A5
shows their requirement in multiple aspects of retinocollicular
12.8 Antero-posterior pattern in the spinal cord is mapping. Neuron 2000, 25: 563–574.
determined in response to secreted signals from the node Hansen, M.J., Dallal, G.E., Flanagan, J.G.: Retinoic axon response
and adjacent mesoderm to ephrin-As shows a graded concentration-dependent transi-
Dasen, J.S., Jessell, T.M.: Hox networks and the origins of motor tion from growth promotion to inhibition. Neuron 2004, 42:
neuron diversity. Curr. Top. Dev. Biol. 2009, 88: 169–200. 707–730.
Ensini M., Tsuchida T.N., Belting, H.G., Jessell, T.M.: The control Klein, R.: Eph/ephrin signaling in morphogenesis, neural devel-
of rostrocaudal pattern in the developing spinal cord: specifi- opment and plasticity. Curr. Opin. Cell Biol. 2004, 16: 580–589.
cation of motor neuron subtype identity is initiated by signals Löschinger, J., Weth, F., Bonhoeffer, F.: Reading of concentration
from paraxial mesoderm. Development 1998, 125: 969–982. gradients by axonal growth cones. Phil. Trans. R. Soc. Lond. B
Sockanathan, S., Perlmann, T., Jessell, T.M.: Retinoid receptor sig- 2000, 355: 971–982.
naling in postmitotic motor neurons regulates rostrocaudal McLaughlin, T., Hindges, R., O’Leary, D.D.: Regulation of axial pat-
­positional identity and axonal projection pattern. Neuron 2003, terning of the retina and its topographic mapping in the brain.
40: 97–111. Curr. Opin. Neurobiol. 2003, 13: 57–69.
504 Chapter 12 Development of the nervous system

Petros, T.J., Shrestha, B.R., Mason, C.: Specificity and sufficiency axon guidance receptor exhibiting extraordinary molecular
of EphB1 in driving the ipsilateral retinal projection. J. Neuro- ­diversity. Cell 2000, 101: 671–684.
sci. 2009, 29: 3463–3474.
Reber, M., Bursold, P., Lemke, G.: A relative signalling model for 12.14 Many motor neurons die during normal development
the formation of a topographic neural map. Nature 2004, 431: Oppenheim, R.W.: Cell death during development of the nervous
847–853. system. Annu. Rev. Neurosci. 1991, 14: 453–501.
Schmitt, A.M., Shi, J., Wolf, A.M., Lu, C-C., King, L.A., Zou, Y.:
Wnt-Ryk signalling mediates medial-lateral retinotectal topo- 12.15 Neuronal cell death and survival involve both intrinsic
graphic mapping. Nature 2006, 439: 31–37. and extrinsic factors
Wilkinson, D.G.: Topographic mapping: organising by repulsion Birling, M.C., Price, J.: Influence of growth factors on neuronal
and competition? Curr. Biol. 2000, 10: R447–R451. differentiation. Curr. Opin. Cell Biol. 1995, 7: 878–884.
Burden, S.J.: Wnts as retrograde signals for axon and growth cone
12.13 Synapse formation involves reciprocal interactions differentiation. Cell 2000, 100: 495–497.
Buffelli, M., Burgess, R.W., Feng, G., Lobe, C.G., Lichtman, J.W., Davies, A.M.: Neurotrophic factors. Switching neurotrophin
Sanes, J.R.: Genetic evidence that relative synaptic efficacy dependence. Curr. Biol. 1994, 4: 273–276.
biases the outcome of synaptic competition. Nature 2003, 424: Jiang, Y., de Bruin, A., Caldas, H., Fangusaro, J., Hayes, J., Conway,
430–434. E.M., Robinson, M.L., Altura, R.A.: Essential role for survivin in
Chess, A.: Monoallelic expression of protocadherin genes. Nat. early brain development. J. Neurosci. 2005, 25: 6962–6970.
Genet. 2005, 37: 120–121. Pettmann, B., Henderson, C.E.: Neuronal cell death. Neuron 1998,
Goda, Y., Davis, G.W.: Mechanisms of synapse assembly and dis- 20: 633–647.
assembly. Neuron 2003, 40: 243–264. Serafini, T.: Finding a partner in a crowd: neuronal diversity and
Hua, J.Y, Smith, S.J.: Neural activity and the dynamics of central synaptogenesis. Cell 1999, 98: 133–136.
nervous-system development. Nat. Neurosci. 2004, 7: 327–332. Šestan, N., Artavanis-Tsakonas, S., Rakic, P.: Contact-dependent
Jan, Y-N., Jan, L.Y.: The control of dendritic development. Neuron inhibition of cortical neurite growth mediated by Notch signal-
2003, 40: 229–242. ing. Science 1999, 286: 741–746.
Katz, L.C., Constantine-Paton, M.: Relationships between segre-
gated afferents and postsynaptic neurones in the optic tectum 12.16 The map from eye to brain is refined by neural activity
of three-eyed frogs. J. Neurosci. 1988, 8: 3160–3180. Del Rio, T., Feller, M.B.: Early retinal activity and visual circuit
Kummer, T.T., Misgled, T., Sanes, J.R.: Assembly of the postsynap- development. Neuron 2006, 52: 221–222.
tic membrane at the neuromuscular junction. Curr. Opin. Neu- Horton, J.C., Adams, D.L.: The cortical column: a structure with-
robiol. 2006, 16: 74–82. out a function. Philos. Trans. R. Soc. Lond. B Biol. Sci. 2005, 360:
Levinson, J.N., El-Husseini, A.: Building excitatory and inhibitory 837–862.
synapses: balancing neuroligin partnerships. Neuron 2005, 48: Katz, L.C., Crowley, J.C.: Development of cortical circuits: lessons
171–174. from ocular dominance columns. Nat. Rev. Neurosci. 2002, 3:
Li, Z., Sheng, M.: Some assembly required: the development of 34–42.
neuronal synapses. Nat. Rev. Mol. Cell Biol. 2003, 4: 833–841. Katz, L.C., Shatz, C.J.: Synaptic activity and the construction of
Schmucker, D., Clemens, J.C., Shu, H., Worby, C.A., Xiao, J., cortical circuits. Science 1996, 274: 1133–1138.
Muda, M., Dixon, J.E., Zipursky, S.L.: Drosophila Dscam is an
 13
Growth and post-embryonic
development
nn Growth nn Molting and nn Aging and senescence
­metamorphosis

Patterning of the embryo occurs on a small scale and is followed by growth. The control
of growth, and therefore size, is a key problem in development, involving at its core
the control of cell proliferation. Such control is also crucial in adult life in preventing
cancer—the result of uncontrolled cell proliferation. In this chapter we look at the bal-
ance between intrinsic programs of cell proliferation during embryonic development,
and growth that is stimulated by circulating factors such as hormones. Another aspect
of post-embryonic development in many invertebrates is metamorphosis, in which the
form of the animal is completely changed. Finally, the individual ages, a process that is
not programmed genetically but which is due to cellular damage.

Development does not stop once the embryonic phase is complete. Most, but by no
means all, of the growth in animals and plants occurs in the post-embryonic period,
when the basic form and pattern of the organism has already been established. The
basic patterning is done on a small scale, over dimensions of less than a millimeter.
In many animals, the embryonic phase is immediately succeeded by a free-living
larval or immature adult stage. In others, such as mammals, considerable growth
occurs during a late embryonic or fetal period, while the embryo is still dependent on
maternal resources. Growth then continues after birth. Growth is a central aspect of
all ­developing systems, determining the final size and shape of the organism and its
parts. Animals with a larval stage not only grow in size, but may undergo metamor-
phosis, in which the larva is transformed into the adult form. Metamorphosis often
involves a radical change in form and the development of new organs.
We first consider the roles of intrinsic growth programs and of factors such as
growth hormones in controlling both embryonic and post-embryonic growth. We shall
also touch on the disturbance of growth control that leads to cancer. This is followed
by a discussion of metamorphosis in insects and amphibians. Finally, we look at what
might be considered an abnormal aspect of post-embryonic development—aging.

Growth

Growth is defined as an increase in the mass or overall size of a tissue or organism;

this increase may result from cell proliferation, cell enlargement without division, or
506 Chapter 13 Growth and post-embryonic development

Cell Cell Growth by by accretion of extracellular material, such as bone matrix or even water (Fig. 13.1).
proliferation enlargement accretion In animals, the basic body pattern is laid down when the embryo is still small. The
growth program—that is, how much an organism or an individual organ grows and
responds to factors like hormones—may also be specified at an early stage in develop-
ment. Overall growth of the organism mainly occurs in the period after the basic pat-
tern of the embryo has been established; there are, however, many examples where
earlier organogenesis involves localized growth, as in the vertebrate limb bud (see
Chapter 11), and the developing nervous system (see Chapter 12). Different rates of
growth in different parts of the body, or at different times, during early development
matrix profoundly affect the shape of organs and the organism.
Unlike the situation in animals, where the embryo is essentially a miniature version
of the free-living larva or adult, plant embryos bear little resemblance to the mature
plant. Most of the adult plant structures are generated after germination by the shoot
or root meristems, which have a capacity for continual growth (see Chapter 7). In
Fig. 13.1 The three main strategies for
woody plants, the cambial layer in the trunk, branches, and roots also retains prolif-
growth in vertebrates. The most common
erative capacity. It can give rise to the main tissues of the plant axis, enabling trees to
mechanism is cell proliferation—cell growth
followed by division. A second strategy is cell
increase in girth year after year.
enlargement, in which cells increase their
size without dividing. The third strategy is to 13.1 Tissues can grow by cell proliferation, cell enlargement, or accretion
increase size by accretionary growth, such as
secretion of a matrix. Although growth often occurs through cell proliferation, which increases the ­number
of cells by mitotic cell division, this is only one of three main ways by which growth in
size of an organ or body structure can occur (see Fig. 13.1). There is also a significant
amount of programmed cell death in many growing tissues, and the overall growth
rate is determined by the rates of cell death and cell proliferation.
A second strategy is growth by cell enlargement—that is, by individual cells
increasing their mass and getting bigger. This is the case in Drosophila larval growth,
for example, except for the imaginal discs, in which cells proliferate during the larval
stages. And differences in size between closely related species of Drosophila are partly
the result of the larger species having larger cells. In mammals, skeletal and heart
muscle cells and neurons never divide again once differentiated, although they do
increase in size. Neurons grow by the extension and growth of axons and dendrites,
whereas muscle growth involves an increase in mass, as well as the fusion of satellite
cells to pre-existing muscle fibers, providing additional nuclei to support the large cell
mass. Much growth is through a combination of cell proliferation and cell enlarge-
ment. For example, the cells in the lens of the eye are produced by cell division from a
proliferative zone for an extended period, whereas their differentiation involves con-
siderable enlargement. Finally, cell enlargement is one of the main means by which
plants grow in size.
The third growth strategy, accretionary growth, involves an increase in the vol-
ume of the extracellular space, which is achieved by secretion of large quantities of
extracellular matrix by cells. This occurs in both cartilage and bone, where most of
the tissue mass is extracellular.
Certain vertebrate tissues, including the blood (see Section 10.4) and epithelia (see
Section 10.7), are continually renewed throughout an animal’s lifetime by cell division
and differentiation from a population of stem cells. The end products of this type of
proliferative system, such as mature red blood cells and keratinocytes, are themselves
incapable of division and eventually die.

13.2 Cell proliferation is controlled by regulating entry into the cell cycle

When a eukaryotic cell duplicates itself by mitosis it goes through a fixed sequence
of stages called the cell cycle, which were described in Box 1B (p. 7). The cell grows
 Growth 507

in size, the DNA is replicated (S phase), and the replicated chromosomes undergo
mitosis and become segregated into two daughter nuclei. The cell then divides
(cyto­kinesis) to give two daughter cells. The cell cycle shown in Box 1B is the stan-
dard cell cycle of a dividing somatic animal cell. At different stages of development,
or in specialized cell types, cell cycles vary in the presence or absence of particular
stages, or the length of time taken to complete them. During the cleavage of the
fertilized amphibian egg, for example, the growth phases G1 and G2 are virtually
absent and so cells do not increase in size between each division (see Chapter 4).
They therefore get smaller at each division. In Drosophila larval salivary glands
there is no M phase, as the DNA replicates repeatedly without mitosis or cell divi-
sion, leading to the formation of giant chromosomes. These are known as polytene
chromosomes, as they are composed of large numbers of identical DNA molecules
aligned side by side. Fig. 13.2 Progression through the
The timing of events in the cell cycle is controlled by a set of ‘central’ timing cell cycle is regulated by the levels of
mechanisms (Fig. 13.2). Proteins known as cyclins control the passage through cyclins. Different cyclins (G1/S, S and M)
regulate progression through different
key transition points in the cell cycle. Cyclin concentrations oscillate during the cell
phases of the cell cycle. The expression of
cycle, and these oscillations correlate with transitions from one phase of the cycle
each cyclin rises and falls throughout the
to the next. Cyclins act by forming complexes with, and helping to activate, protein
cell cycle. As cyclin protein levels increase,
kinases known as cyclin-dependent kinases (Cdks). These kinases phosphorylate the cyclins bind to their corresponding
proteins that trigger the events of each phase, such as DNA replication in S phase or inactive cyclin-dependent kinases (Cdks)
mitosis in M phase. (not shown) forming active enzymes. These
Once a cell has entered the cycle and progressed beyond a point commonly known active kinases are ultimately responsible for
as ‘Start’, which occurs in mid- to late-G1, it will continue through and complete the driving cell-cycle events such as the initiation
cycle without needing any further external signal. Transitions into successive phases of DNA replication (S phase) and the entry
are marked by cell-cycle checkpoints at which the cell monitors progress to ensure, into mitosis (M phase). There are a number
for example, that an appropriate size has been reached, that DNA replication is com- of different checkpoints in the cell cycle (red
bars) which ensure that the cell does not
plete, and that any DNA damage has been repaired. If such criteria are not met,
progress to the next stage until the previous
progress into the next stage is delayed until all the necessary processes have been
stage has been completed successfully.
completed. If the cell has suffered some damage that cannot be repaired, the cell cycle
At the end of mitosis, existing cyclins are
will be arrested and the cell will usually undergo apoptosis. These mechanisms are targeted to the proteasome and destroyed,
intrinsic to all normal eukaryotic cells and, for example, prevent them from continu- so that the new cells formed at cell division
ing to divide with damaged DNA. are re-set to early G1 phase.
Adapted from Morgan, D.O.: 2007.

Cdk inactive Cdk active Cdk inactive

G1/S cyclin S cyclin M cyclin

paternal
homolog chromatid

maternal
homolog

G1 S G2 Mitosis G1

Start Checkpoint Checkpoint Cell

checkpoint division
508 Chapter 13 Growth and post-embryonic development

Extracellular factors control cell proliferation by stimulating entry into the cell cycle.
Studies of animal cells in culture show that growth factors are essential for cells to
multiply, with the particular growth factor or factors required depending on the cell
type. When somatic cells are not stimulated to proliferate they are usually in a state
known as G0, into which they withdraw after mitosis (see Box 1B, p. 7). Growth fac-
tors enable the cell to proceed out of G0 and progress through the cell cycle. Most cells
in adult organisms are not actively proliferating and so need an external signal before
they can re-enter the cell cycle. Numerous extracellular signaling proteins that can
stimulate or inhibit cell proliferation have been discovered. Many of the extracellular
signal molecules we met in earlier chapters in their roles as patterning agents, such
as fibroblast growth factors (FGF) and members of the TGF-β family, were initially
discovered through their ability to stimulate cell proliferation. They can also act in
this way during embryonic and fetal growth, and in the control of cell proliferation in
the adult. Other growth factors have specialized roles in directing the proliferation of
particular tissues. One example is erythropoietin, which promotes the proliferation
of red blood cell precursors (see Section 10.5).
A striking discovery over the past few decades has been the recognition that cells
must receive signals, such as growth factors, not only for them to divide, but also sim-
ply to survive. In the absence of all growth factors, cells commit suicide by apoptosis,
as a result of activation of an internal cell death program (see Section 10.12). There
is a significant amount of cell death in all growing tissues, so that overall growth rate
depends on the rates of both cell death and cell proliferation.

13.3 Cell division in early development can be controlled by an intrinsic

developmental program

Compared with cells in adult organisms, embryonic cells proliferate much more freely,
and in many organisms the pattern of cell division in the earliest embryonic stages
is controlled by a cell-autonomous developmental program that does not depend on
stimuli such as growth factors. One well-understood example comes from Drosophila,
18 20 16 where the early cell-cycles are controlled by the proteins that pattern the embryo;
3 4 these proteins exert their effects by interacting with components of the cell-cycle
1 5
23 6 control system.
8 9 11 The first cell-cycles in the Drosophila embryo are represented by rapid and synchro-
2
15 17 nous nuclear divisions, without any accompanying cell divisions, which create the
21 25
14 syncytial blastoderm (see Fig. 2.2). There are virtually no G phases, just alternations
of DNA synthesis (S phase) and mitosis. But at cycle 14, there is a major transition to
Fig. 13.3 Domains of mitosis in the a different type of cell cycle, a transition similar to the mid-blastula transition in frogs
Drosophila blastoderm. Areas composed (see Section 4.13). A well-defined G2 phase is seen in cycle 14, and the blastoderm
of cells that divide at the same time are ­becomes cellularized. A G1 phase is seen at the 16th cycle. After the 17th or 18th
indicated by the various colors, and the cycle, cells in the epidermis and mesoderm stop dividing and differentiate. The ces-
numbers indicate the order in which these sation of cell proliferation is caused by the exhaustion of maternal cyclin E reserves
domains undergo mitosis at the 14th cell originally laid down in the egg.
cycle, when zygotic String protein is first At the 14th cell cycle, distinct spatial domains with different cell-cycle times can
expressed. The schematic illustrates a lateral be seen in the Drosophila blastoderm (Fig. 13.3). This patterning of cell cycles is pro-
view of the embryo corresponding to a stage
duced by a change in the synthesis and distribution of a protein phosphatase called
after mesoderm has been internalized and
String, which exerts control on the cell cycle by dephosphorylating and activating a
segmentation has begun (segments are
indicated by the black marks along the lower
cyclin-dependent kinase. In the fertilized egg, the String protein is of maternal origin
surface). Anterior is to the left and dorsal is and is uniformly distributed. It therefore produces a synchronized pattern of nuclear
up. The gray region on the dorsal surface is division throughout the embryo. After cycle 13, the maternal String protein disappears
the amnioserosa. The internalized mesoderm and zygotic String protein becomes the controlling factor.
and some other domains are not visible in Zygotic string gene transcription occurs in a complex spatial and temporal pattern.
this view. Only cells in which the string gene is expressed enter mitosis. This results in variation
After Edgar, B.A., et al.: 1994. in the rate of cell division in different parts of the blastoderm, which ensures that the
 Growth 509

correct number of cells is generated in different tissues. The pattern of zygotic string
gene expression is controlled by transcription factors encoded by the early patterning
genes, such as the gap and pair-rule genes and those patterning the dorso-ventral
axis. One exception to the rule that expression of string leads to cell proliferation is
the presumptive mesoderm, which is the first domain in which string is expressed but
the tenth to start cell division. The delay in cell division is due to the expression of the
gene tribbles in this region, as Tribbles protein degrades String. The delay is necessary
to allow ventral furrow formation and thus mesoderm invagination (see Section 8.8),
as cell division inhibits ventral furrow formation. The prospective mesodermal cells
start to proliferate after they have been internalized. Thus, the early cell cycles in
Drosophila development provide a good example of how a pattern of cell division can
be under genetic control. In contrast to this intrinsic embryonic program, the cell pro-
liferation and growth that occurs in the imaginal discs is modulated by extracellular
signals produced by the previous patterning process, and involves the coordination of
cell growth, cell proliferation, and cell death.
In early mammalian embryos, cell division times vary with developmental time and
place. The first two cleavage cycles in the mouse last about 24 hours, and subsequent
cycles take about 10 hours each. After implantation, the cells of the epiblast prolifer-
ate rapidly; cells in front of the primitive streak have a cycle time of just 3 hours.

13.4 Organ size can be controlled by both intrinsic growth programs and
extracellular signals

Organ size in vertebrates is determined by both intrinsic developmental programs and

extracellular factors that stimulate or inhibit growth, but the relative importance of
these two mechanisms in different organs varies a good deal. The liver, for example,
which is the main detoxifying organ in mammals, has excellent powers of regenera-
tion in both embryo and adult, whereas the pancreas does not. Experiments to inves-
tigate these properties in the embryo make use of the fact that organ precursor cells
can be specifically destroyed in embryonic mice using transgenic techniques. A strain
of mice is generated that carries an introduced gene for diphtheria toxin under the
control of a tetracycline-repressible promoter. Part of this promoter comes from a
gene that is only expressed in the particular organ precursor cells targeted, and so
only directs expression of the toxin in those cells. If a pregnant mouse of this strain
is continuously fed tetracycline, the toxin is not expressed at all in the embryos she
carries. If tetracycline is withdrawn, the toxin is produced in the targeted precursor
cells, killing them but leaving other tissues in the body untouched.
To investigate the regenerative powers of the liver, tetracycline is withdrawn for a
short period at the appropriate time during gestation, with the result that a proportion
of the liver precursor cells in the embryos are destroyed. The longer the period with-
out tetracycline, the greater the number of precursor cells lost. After such treatment,
the embryonic liver grows back to a normal size, indicating that it does not arise from
a fixed number of progenitor cells. Even in adult mammals, the liver will regrow to
regain its normal mass after as much as two-thirds of it has been surgically removed.
In contrast, if some of the progenitor cells of the pancreas in a mouse embryo are
destroyed (by the technique described above) once the pancreatic ‘bud’ has formed,
a smaller than normal pancreas develops. The size of the embryonic pancreas seems,
therefore, to be largely under intrinsic control. Another organ with intrinsic growth
control is the thymus gland. If multiple fetal thymus glands are transplanted into a
developing mouse embryo, each one grows to full size. If the same experiment is done
with spleens, however, each spleen grows much smaller than normal, so that the final
total mass of the spleens is equivalent to one normal spleen.
The liver and the spleen regulate their final size by a negative feedback mechanism.
The developing or regenerating liver secretes some factors that stimulate cell proliferation
510 Chapter 13 Growth and post-embryonic development

and others that potentially inhibit growth, and these have been studied in detail in
regard to the regenerating adult liver. Cell division is relatively rare in a normal adult
liver but if a substantial part of it is surgically removed, a process known as resection,
95% of the remaining cells re-enter the cell cycle. In humans, as much as 70% of the
liver can be removed, to cut out a tumor for example, and the liver will grow back to a
normal size; however, if more than 75–80% is removed, it cannot regenerate.
In the regenerating human liver, hepatocytes start to divide a day after surgery,
and the release of growth factors and cytokines by the remaining liver tissue is
required for regrowth. The key growth factors that initiate regeneration and promote
growth include hepatocyte growth factor (HGF) and epidermal growth factor (EGF),
and other growth-promoting cytokines include tumor necrosis factor α (TNF-α) and
interleukin-6. The serotonin produced by blood platelets seems also to stimulate liver
regrowth by in turn stimulating the production of HGF by liver cells. Transforming
growth factor β (TGF-β) acts as an inhibitor of liver growth, although it seems to be
required for the proper organization of liver tissue towards the end of the regenera-
tion period. The expression of TGF-β receptors by liver tissue temporarily decreases
immediately after liver resection, preventing inhibition by TGF-β during this crucial
period. As the liver regrows, TGF-β eventually reaches a level that blocks further
growth.
When the liver gets to a certain size, the concentration of inhibitory factors in
the circulation is sufficient to stop further growth—an example of negative feedback
determining the size of an organ. The size of the spleen is determined in a similar
way. In the experiment described above, the total mass of spleen tissue is being moni-
tored by the embryo according to the level of growth-inhibitory proteins produced by
the spleen tissue and circulating in the blood.
In muscle, the growth inhibitor myostatin, which is produced by the myoblasts
(immature muscle cells) themselves, helps to regulate muscle size and cell numbers.
Mutation in the myostatin gene in mice leads to a significant increase in muscle mass;
the number of muscle fibers and their size are both increased. There is now some
evidence for a similar negative signal controlling cell numbers in neural tissue. And in
butterflies, growth-controlling interactions occur between imaginal discs; the removal
of hindwing discs results in larger forewings and forelegs.
Even when they are capable of division, the cells of many adult tissues do not
Length 9
(mm) divide, or divide infrequently, under normal circumstances. Adult cells can, however,
8 be induced to divide by injury or other stimuli, such as a reduction in the mass of the
7 humerus tissue, as in the case of the liver. Removal of a kidney leads to an increase in size of
6 the remaining kidney, but in this case the growth is mainly the result of cell enlarge-
ulna
ment rather than cell proliferation. And in Chapter 14 we shall see how some adult
5
amphibians and insects can regenerate whole limbs after amputation, complete with
4 all the different types of tissue in their correct place.
3 Patterning of the embryo occurs while the organs are still very small. For example,
2
human limbs have their basic pattern established when they are less than 1 cm long.
wrist Over the years, the limb grows to be at least one hundred times longer. How is this
1
growth controlled? It appears that each of the cartilaginous elements in the embryonic
0 limb has its own growth program. In the chick embryonic wing, the cartilaginous ele-
3 5 7 9 11 13 15 17
ments representing the long bones—the humerus and the ulna—are initially similar
Time after initial formation (days)
in size to the elements in the wrist (Fig. 13.4). Yet, with growth, the humerus and
ulna increase many times in length compared with the wrist bones. These growth
Fig. 13.4 Comparative growth of the
programs are specified when the elements are initially patterned and involve both cell
cartilaginous elements in the embryonic
chick wing. When first laid down, the
multiplication and matrix secretion (accretion). Each skeletal element follows its own
cartilaginous elements of the humerus, ulna, growth program even when grafted to a developmentally neutral site, provided that a
and wrist are the same size, but the humerus good blood supply is established.
and ulna then grow much more than the wrist A classic illustration of an intrinsic growth program comes from grafting limb buds
element. between large and small species of salamanders of the genus Ambystoma. A limb
 Growth 511

bud from the larger species grafted to the smaller species initially grows slowly, but
limb from grafted bud
eventually ends up at its normal size, which is much larger than any of the limbs
of the host (Fig. 13.5). This indicates that whatever the circulatory factors, such as
hormones, that influence growth, the intrinsic response of a tissue to them is crucial.
Even though circulating homones are required for growth in many cases, such intrin-
sically patterned differential growth can affect the overall form of an organism consid-
erably, as can be seen in Fig. 13.6. This figure shows that growth of the different parts
Fig. 13.5 The size of limbs is genetically
of the body is not uniform, and different organs grow at different rates. At 9 weeks
programmed in salamanders. An embryonic
of development, the head of a human embryo is more than a third of the length of
limb bud from a large species of salamander,
the whole embryo, whereas at birth it is only about a quarter. After birth, the rest of
Ambystoma tigrinum, grafted to the embryo
the body grows much more than the head, which is only about an eighth of the body of a smaller species, Ambystoma punctatum,
length in the adult (see Fig. 13.6). grows much larger than the host limbs—to
the size it would have grown in Ambystoma
tigrinum.
13.5 The amount of nourishment an embryo receives can have profound
Photograph from Harrison, R.G.: 1969.
effects in later life

Whatever the type of growth program, no animal will reach its full potential size if
inadequately nourished as an embryo and during the post-natal growth period. In
mammals, inadequate nutrition or poor nutrition of the embryo not only has direct
effects on embryonic and fetal growth, but can have serious, and in some ways unex-
pected, effects in adult life. Population studies in developed countries, such as the
United Kingdom, have associated smaller size or relative thinness at birth (due to
either maternal undernutrition or premature birth) and during early infancy with an
increased risk of developing coronary heart disease, stroke, or type 2 diabetes in adult
life. When undernutrition during early development is followed by improved nutri-
tion later, whether later in gestation or in the early post-natal period, the period of
‘catch-up’ growth may carry a cost, as the body’s resources are diverted from ‘repair
and maintenance’ to growth. Premature birth itself, independent of size for gestational
age, has been associated with insulin intolerance and glucose intolerance in pre-pu- Fig. 13.6 Different parts of the human
bertal children that may continue into young adulthood and may be accompanied by body grow at different rates. At 9 weeks of
development the head is relatively large but,
high blood pressure. Catch-up growth seems to predispose to overweight and even
with time, other parts of the body grow much
obesity, and this may partly explain the effects on health in later life. One proposal
more than the head.
is that in response to undernutrition in utero the fetus lays down more fat cells as a
After Gray, H.: 1995.

9 16 newborn 2 5 15 adult
Weeks of development Age in years
512 Chapter 13 Growth and post-embryonic development

precaution. That is a good strategy if conditions after birth are indeed hard and food is
short, but in conditions of plentiful food, this leads to undesirable consequences.
Experiments in animals support the effects of undernutrition and an unbalanced
maternal diet observed in humans. The embryos of pregnant rats fed a low-protein,
but otherwise calorie-sufficient, diet during the preimplantation period (0 to 4.5 days)
showed altered development in multiple organ systems. If the pregnancy was allowed
to go to full term, the offspring had low birth weight, increased post-natal growth, or
adult-onset high blood pressure.
Obesity is associated with numerous diseases in later life, including type 2 diabetes
and heart disease. Although much obesity in children and adults is due to overeating
and lack of exercise, early developmental nutritional experience and genetic back-
ground can contribute. Human fatty tissue comprises some 40 billion adipose cells,
with most of the bulk stashed under the skin, and obesity represents both greater
numbers of adipose cells compared with lean people and excessive deposition of fat
in these cells, which increases cell size. Humans are born with a certain number of
adipose cells, with females generally having more than males. The number of adipose
cells increases throughout late childhood and early puberty, and after this remains
fairly constant. However, the number of adipose cells increases more rapidly in genu-
inely obese children than in lean children, meaning that they end up with more adipose
cells. Once fat cells develop in the body, they remain there for life and they seldom die,
although there is some turnover. Each month about 1% of adipose cells in the human
body die and are replaced. Thus adult obesity is often linked to childhood obesity.
Overweight people with extra fat cells can shrink the size of the cells and lose weight
by burning the excess fat by careful dieting and exercise, but the fat cells themselves
do not disappear, and are only too ready to start accumulating excess fat again.

13.6 Determination of organ size involves coordination of cell growth,

cell division, and cell death

How do cells know to stop growing and dividing when an organ or body structure
reaches its final size? Evidence that organ size and body size can be determined by
monitoring overall dimensions, rather than the absolute number of cells or cell divi-
sions, comes from a number of different sources. Species with either fewer or more than
the diploid number of chromosomes—that is, haploid and polyploid organisms, respec-
tively—provide one line of evidence. For a given cell type, cell size is usually propor-
tional to ploidy, so haploid cells are half the volume of diploid cells and tetraploid cells
twice the volume. Some salamander species are naturally tetraploid; they grow to the
same size as their diploid relatives but have only half the number of cells. Artificially
produced tetraploid mouse embryos also compensate for having larger cells by having
fewer of them, but usually die before birth. In Drosophila, the final size of animals that
are a mosaic of haploid and diploid cells is normal, with the haploid regions contain-
ing smaller but more numerous cells. Polyploid species of plants are, in fact, generally
larger overall than their diploid relatives, but in plant mosaics of diploid and polyploid
cells it is possible to find the same sort of size compensation as seen in animals.
The Drosophila wing imaginal disc has proved an interesting model system for
studying how organ size might be determined. At its formation, the wing disc is ini-
tially composed of about 40 cells, and normally grows in the larva to about 50,000
cells. Cell division occurs throughout the disc, and then ceases uniformly when the
correct size is reached. The adult wing is produced at metamorphosis by changes
in cell shape without further cell division (see Fig. 11.24). Early experiments inves-
tigating wing growth showed that the final size of the wing does not depend on the
imaginal disc undergoing a fixed number of cell divisions, or attaining a particular
number of cells. Instead, final size seems to be controlled by some mechanism that
monitors the overall size of the developing wing disc and adjusts cell division and
 Growth 513

cell size accordingly. Experiments show that there is no restriction on how much of
the wing a given cell can make; the clonal descendants of a single cell can contribute
from a tenth to as much as a half of the wing. Wing growth can also be experimentally
decoupled from cell proliferation, as demonstrated by the fact that if cell division is
blocked in either the anterior or posterior compartment in the larval wing disc, the
final size of the wing is normal but the individual cells are larger.
In the same way, the size of a region or organ need not be determined by the rate
of cell division. This can be shown by making mosaics of wild-type cells and slower-
growing Minute cells (see Box 2E, p. 73). Within a compartment, the faster-dividing
wild-type cells contribute more cells than the slow-growing Minute cells, but the size
of the compartment remains the same as if all the cells were wild type.
In normal growth, the final size of the wing is likely to be achieved by a balance
between cell proliferation and apoptosis. The recently discovered Hippo signaling
pathway (Fig. 13.7), is likely to be an important coordinator of cell growth, cell divi-
sion, and apoptosis in Drosophila and in other animals, suppressing cell proliferation
and promoting apoptosis of extra, unwanted cells. The Hippo signaling network was
discovered in Drosophila and has since been found in mammals and other verte-
brates. It seems to be involved generally in controlling and coordinating cell prolif-
eration and apoptosis in growing tissues, and components of the pathway have been
found mutated in human cancers—yet another example of a pathway first identified Hippo pathway Hippo pathway
in Drosophila development proving of importance to human disease. The Hippo path- inactive active
way involves a cascade of protein kinases, of which Hippo is one, and the outcome
Exp Mer Exp Mer
of activation of the pathway is the inactivation of a transcriptional co-activator called
Yorkie (Yki), which has a direct effect on cell proliferation (see Fig. 13.7). Loss-of-
function mutations in Hippo or overexpression of Yki in Drosophila wing discs results
in a dramatic increase in cell proliferation and a decrease in apoptosis, resulting in Hpo Hpo
wing discs growing to almost eight times the normal size. When the pathway is acti-
vated, Yki is directly inhibited by phosphorylation by the protein kinase Warts (Wts), P
which is itself activated by phosphorylation by Hippo. The inactivation of Yki by the Wts Wts
Hippo signaling network is therefore a mechanism for suppressing growth, and the
Hippo pathway is a strong candidate for one means of terminating growth when an P
Yki
organ reaches its required size.
One target of the Hippo pathway in Drosophila is a microRNA (miRNA) called
bantam, which is known both to promote cell proliferation and suppress apoptosis by
Yki
suppressing production of the pro-apoptotic protein Hid. The Hippo pathway down-
regulates the production of bantam miRNA, and this could be one of the mechanisms
target genes target genes
by which the pathway promotes apoptosis and suppresses cell proliferation.
Because the Hippo pathway suppresses growth, Hippo, Warts, and associated Cell growth and Cell growth and
proliferation proliferation suppressed
components of the pathway are characterized as tumor-suppressor proteins. Tumor- Apoptosis suppressed Apoptosis promoted
suppressor genes are a class of genes defined by the fact that their absence or inac-
tivation predisposes to uncontrolled cell proliferation and the development of cancer Fig. 13.7 The Hippo signaling
(see Section 13.10). In contrast, Yki is classed as an oncogene, as its overexpression pathway. The outcome of activation of the
promotes uncontrolled growth. In Drosophila, the Hippo pathway can be activated by Hippo pathway is the inactivation of the
signals from the atypical cadherin Fat. Whether it is activated by extracellular signals transcriptional co-activator Yorkie (Yki). In the
in mammalian tissues, what those might be, and how the pathway might be linked absence of signaling (left panel) Yki helps
to the determination of mammalian organ size, are still unknown. In Drosophila, the activate genes that promote cell growth
Hippo pathway has been linked to a possible gradient mechanism for determining and proliferation and suppress apoptosis.
When the pathway is activated, the adaptor
adult organ size (see Box 13A, p. 514).
proteins Expanded (Exp) and Merlin (Mer)
activate the protein kinase Hippo (Hpo).
13.7 Body size is also controlled by the neuroendocrine system in both Hippo phosphorylates and activates the
insects and mammals protein kinase Warts (Wts), which in turn
phosphorylates and inactivates Yki, and
Body size depends not only on the rate of growth, but on how long growth continues. so the growth-promoting and apoptosis-
This is particularly evident in insects that undergo metamorphosis, in which growth suppressing genes are not expressed.
514 Chapter 13 Growth and post-embryonic development

Box 13A Gradients of signaling molecules could determine organ size

Evidence from various sources has suggested for some time that
the final size of Drosophila imag­inal discs, and thus of the adult
organs, might be determined by a molecular gradient across the molecular gradient
disc that could be formed as a result of previous patterning. The
basic idea is that when the disc is small, the gradient is steep and
the steepness of the gradient in some way promotes growth. As
the organ grows, the gradient flattens out (see ­figure). Growth Growth
declines and finally terminates. The discovery of the Hippo path-
way as a mechanism for regulating growth, and the discovery
that it could be regulated by the cadherin Fat, turned attention
to how the Hippo pathway might interact with molecular gra-
dients of Fat-associated proteins present along the proximo-
distal axes of imaginal discs. The gradients are those of the
atypical cadherin Dachsous (Ds) and the transmembrane serine/­
threonine kinase Four-jointed (Fj), which modulates the interac-
tion of Ds with Fat.
Experiments in which Ds and Fj were overexpressed in distinct
clones of cells in otherwise wild-type imaginal discs showed that
the Hippo pathway was suppressed (which would allow cell pro-
liferation) only where there was a significant difference in the
levels of Ds/Fj expressed by neighboring cells. This occurs in cells disc reaches the required size, because the gradient has flat-
on the boundary of the clone, where Ds/Fj-overexpressing cells tened out and the difference in Ds/Fj levels between cells is no
are adjacent to wild-type cells. If a patch of cells all expresses longer sufficient to suppress the Hippo pathway. It is thought
the same level of Ds/Fj, no matter whether it is a low or a high that these effects are being mediated in some way through the
level, the Hippo pathway is activated in these cells (which would interaction of Ds/Fj with Fat, but precisely how is not yet known.
suppress cell proliferation). Suppression or activation of the Ds/Fj cannot simply be acting as conventional ligands that either
pathway was detected by looking at whether target genes of activate or suppress Fat signaling. This proposed mechanism is
the pathway were transcribed or not. reminiscent of the one for planar cell polarity in Drosophila (see
The argument then goes that an initial steep gradient of Box 2F, p. 78), which also involves discontinuities in levels of
Ds/Fj across a disc would allow it to grow, as there would be Ds/Fj between cells. It will be fascinating to see how the same
sharp discontinuities in the level of Ds/Fj between adjacent gradient of information can be read out to give instructions as to
cells all across the disc. Growth will presumably stop as the both organ size and cell directionality.

occurs only in the larval stages and the size of the adult is determined by the size
the larva reaches before pupation. In normal circumstances the larva stops feeding
and pupates when it reaches a size typical for the species, but if the larval stage is
prolonged or shortened experimentally, adults that are, respectively, larger or smaller
than usual can be produced. As discussed in more detail later in this chapter, pupa-
tion and metamorphosis in insects are initiated by a pulse of the steroid hormone
ecdysone, which is released by the prothoracic gland under the stimulation of the
neurohormone prothoracicotropic hormone (PTTH).
How body size is monitored by insect larvae and how the timing of the ecdysone
signal is determined are still not entirely understood, and the mechanisms are dif-
ferent in different groups of insects. In Drosophila, for example, the larva must
attain a minimum critical size, as otherwise it will not survive metamorphosis. In
normal circumstances this minimum size is attained about half-way through the
last instar and the larva continues to feed and grow until pupation (Fig. 13.8). If
the larva is starved after it reaches the critical size, it still undergoes metamorphosis
but produces smaller adults than normal. Experiments in Drosophila indicate that
signaling via the insulin and target of rapamycin (TOR) pathways in the prothoracic
 Growth 515

Drosophila

Body
size

normal size adult normal size adult

TGP TGP
critical
size

Time

Manduca Fig. 13.8 Nutrition-dependent regulation

of insect body size. The level of nutrition
Body available regulates adult body size in
size
different ways in different insects. Upper
panel: in Drosophila, once the larva has
reached a critical size, if nutrients are
normal size adult small size adult scarce the subsequent time the larva
spends feeding (the terminal growth period,
TGP) is simply prolonged until the larva
reaches a normal size. It then undergoes
TGP TGP metamorphosis, producing a normal-sized
adult. Lower panel: in the tobacco hornworm
(in which the adult is a moth), the larval
terminal growth period (after reaching a
critical size) remains the same whether
nutrients are plentiful or scarce. This means
Time
that when nutrients are restricted a smaller
adult is produced.
high nutrition, rapid growth low nutrition, slow growth final size TGP = terminal growth period
Adapted from Nijhout, H.F.: 2008.

gland in response to ­nutritional status plays an important part in determining the

level of ecdysone production and thus the timing of metamorphosis. In larvae put
on a low-nutrient diet after they had reached the critical size, pupation was delayed,
enabling the larvae to grow to nearly normal size. This delay was accompanied
by downregulation of the TOR pathway in the prothoracic gland, which had the
effect of keeping ecdysone production at a low level. As the larvae did eventually
pupate and metamorphose, it seems that in Drosophila, exposure to low levels of
ecdysone over a period can have the same effect as the distinct pulse of ecdysone
that is the usual signal for pupation.
Larval growth rate is governed ultimately by the level and quality of nutrients avail-
able, and, as in other animals, insulin-like peptide growth factors are involved in the
direct control of cell growth. The ecdysone and PTTH mentioned earlier have no direct
effect on growth rate—they act by affecting the length of the feeding period. Insulin-
like peptides are also thought to be involved in coordinating growth with the availabil-
ity of nutrients and with the hormonal signals that terminate the feeding period.
Human growth during the embryonic, fetal, and post-natal periods is typical of
the different phases of growth in mammals. The human embryo increases in length
from 150 mm at implantation to about 50 cm over the 9 months of gestation. During
516 Chapter 13 Growth and post-embryonic development

Average growth after birth Male and female comparative growth rates

Height Change
(cm) 180 in height 10 Boys
(cm/year) Girls
160
8
140
6
120

100 4

Fig. 13.9 Normal human growth. Left 80

2
panel: an average growth curve for a human
60
male after birth. Right panel: comparative
0
growth rates of boys and girls. There is a 2 4 6 8 10 12 14 16 18 8 9 10 11 12 13 14 15 16 17 18
growth spurt at puberty in both sexes, which Age (years) Age (years)
occurs earlier in girls.

the first 8 weeks after conception, the embryonic body does not increase greatly in
size, but the basic human form is laid down in miniature. After 10 weeks the human
embryo is technically known as a fetus. The greatest rate of growth occurs at about
4 months, when the fetus grows as much as 10 cm per month. Growth after birth
follows a well-defined pattern (Fig. 13.9, left panel). During the first year after birth,
growth occurs at a rate of about 2 cm per month. The growth rate then declines
steadily until the start of a characteristic adolescent growth spurt at puberty at about
11 years in girls and 13 years in boys (Fig. 13.9, right panel). In pygmies, sexual matu-
ration at puberty is not accompanied by this adolescent growth spurt, hence their
characteristic short stature.
The maternal environment plays an important role in controlling fetal growth. This
is well illustrated by crossing a large shire horse with a much smaller Shetland pony.
When the mother is a shire mare, the newborn foal is similar in size to a normal shire
foal, but when the mother is a Shetland, the newborn is much smaller. However, with
growth after birth, the offspring of both these crosses become similar in size, and
achieve a final size intermediate between shires and Shetlands.
In mammals, early embryonic growth is controlled by growth factors secreted
into the local extracellular environment. For example, we saw in Chapter 11 that
FGFs control cell proliferation at the tip of the developing chick limb, and are
required for proliferation and outgrowth of the bud. Evidence for the roles of indi-
vidual growth factors in embryonic growth comes from the technique of gene knock-
out (see Box 3B, p. 118). For example, the insulin-like growth factors 1 and 2
(IGF-1 and IGF-2) have a key role in growth during embryonic development. They
are single-chain protein growth factors that closely resemble insulin and each other in
their amino-acid sequence and produce their effects on cells through a shared insulin-
signaling pathway. Newborn mice lacking a functional Igf2 gene develop relatively
normally, but weigh only 60% of the normal newborn body weight. Mice in which
the Igf1 gene has been inactivated are also growth retarded. Both the factors and their
receptors can be detected as early as the eight-cell stage of mouse development. Igf2
is one of the genes that are imprinted in mammals (see Section 9.8); it is inactivated
in maternal germ cells and is only expressed in the embryo from the paternal genome.
Like many growth factors, the IGFs and FGFs also have important roles in post-natal
growth and in controlling cell proliferation in adult mammals.
Growth hormone is a circulating protein hormone produced by the pituitary gland
that is essential for the growth of humans and other mammals after birth. It is also
synthesized in embryonic and fetal tissues, where it is thought to act directly on cells
as an extracellular growth factor. Within the first year of birth, the pituitary gland
­begins to secrete growth hormone. A child with insufficient growth hormone grows
 Growth 517

less than normal, but if growth hormone is given regularly, normal growth is restored.
In this case, there is a catch-up phenomenon, with a rapid initial response that tends Hypothalamus
to restore the growth curve to its original trajectory. Growth hormone- Somatostatin
Production of growth hormone in the pituitary is under the control of two hor- releasing hormone
mones produced in the hypothalamus: growth hormone-releasing hormone, which
promotes growth hormone synthesis and secretion, and somatostatin, which inhibits + – – +
its production and release. Growth hormone produces many of its effects by inducing
the synthesis of IGF-1 (Fig. 13.10) and, to a lesser extent, IGF-2, which act directly Pituitary
on cells to promote proliferation. Post-natal growth, as well as embryonic growth, is Growth hormone
largely due to the actions of the insulin-like growth factors, and complex hormonal
regulatory circuits control their production.
Puberty is initiated by the activity of the hypothalamic neuronal network, which
governs the intermittent release of gonadotropin-releasing hormone (GnRH) by IGF-1
hypothalamic neurons. The mechanism determining this timing is not known. One synthesis
of the results of a pulse of GnRH is a sharp increase in the secretion of gonadotro-
pins (luteinizing hormone and follicle-stimulating hormone) by the pituitary; these
cause increased production of the steroid sex hormones—the estrogens and andro-
Cell multiplication
gens. These in turn stimulate the production of pulses of growth hormone, which are
responsible for the growth spurt at puberty.
Fig. 13.10 Growth hormone production
is under the control of the hypothalamic
13.8 Growth of the long bones occurs in the growth plates hormones. Growth hormone is made in
the pituitary gland and is secreted. Growth
An important aspect of post-embryonic vertebrate growth is the growth of the long
hormone-releasing hormone from the
bones of the limbs (humerus, femur, radius, ulna, tibia, and fibula). The long bones hypothalamus promotes growth hormone
are initially laid down as cartilaginous elements (see Section 11.1) and then become synthesis, while somatostatin inhibits it.
ossified. The early growth of these elements involves both cell proliferation and matrix Growth hormone controls its own release
secretion in a well-defined pattern. The end of a long bone is known as the epiphysis by negative feedback signals to the
and the central region as the diaphysis. In both fetal and post-natal growth, the carti- hypothalamus. Growth hormone causes the
lage is replaced by bone in a process known as endochondral ossification, in which synthesis of the insulin-like growth factor
ossification starts in the centers of the long bones and spreads outward (Fig. 13.11). IGF-1 and this promotes the production of
Secondary ossification centers then develop at each end of the bone. The adult long growth hormone.
bones thus have a bony shaft with cartilage confined to the articulating surfaces at
each end, and to two internal regions near each end—the growth plates—in which
growth occurs. In the growth plates, the cartilage cells, or chondrocytes, are usually
arranged in columns, and various zones can be identified. Just next to the bony epi-
physis is a narrow germinal zone, which contains stem cells. Next is a proliferative
zone of cell division, followed by a zone of maturation, and a hypertrophic zone, in
which the cartilage cells increase in size. Finally, there is a zone in which the cartilage
cells die and are replaced by bone laid down by cells called osteoblasts, which dif-
ferentiate from cells that form the perichondrium surrounding the cartilage. This stage
involves Wnt signaling. There is a strong similarity to the development of skin, where
basal stem cells give rise to dividing cells, which differentiate into keratinocytes and
finally die (see Section 10.7).
The proliferation of chondrocytes at the ends of the long bones, and later in the
growth plate, is controlled so that at a given distance from the end of the bone they
stop dividing and enlarge to form the scaffolding for bone, as appropriate. In mice, the
proliferation of chondrocytes is controlled by the secreted signaling proteins parathy-
roid-hormone-related protein (PHRP) and Indian hedgehog. Indian hedgehog belongs
to the large family of Hedgehog signaling proteins. PHRP is secreted by the chon-
drocytes and perichondrial cells at the ends of the prospective bones and stimulates
chondrocytes to proliferate, which prevents them from expressing Indian hedgehog.
Once chondrocytes move out of the zone of influence of PHRP they stop proliferat-
ing, start to express Indian hedgehog, and become hypertrophic (Fig. 13.12). Indian
hedgehog diffuses back into the pool of proliferating chondrocytes, where it increases
518 Chapter 13 Growth and post-embryonic development

secondary
ossification
center germinal zone
Epiphysis containing stem cells

Growth plate proliferative zone

zone of maturation

Diaphysis

bone

hypertrophic zone

Growth plate
perichondrium

cartilage
Epiphysis invading capillary
secondary osteoblast
ossification
center bone

Fig. 13.11 Growth plates and endochondral ossification in the long multiply in the proliferative zone, then mature and undergo hypertrophy
bone of a vertebrate. The long bones of vertebrate limbs increase in (cell enlargement). They are then replaced by bone, which is laid down by
length by growth from cartilaginous growth plates. The growth plates specialized cells called osteoblasts. These derive both from perichondrial
are cartilaginous regions that lie between the epiphysis of the future cells and from cells that invade the bone along with the blood vessels.
joint and the central region of the bone, the diaphysis. In the figure, bone Secondary sites of ossification are located within the epiphyses.
has already replaced cartilage in the diaphysis, and more bone is being After Wallis, G.A.: 1993.
added at the growth plates. Within the growth plates, cartilage cells

their rate of proliferation, and it also, by some mechanism as yet unknown, stimulates
production of PHRP by the cells at the ends of the bone. It also acts on the adja-
cent perichondrial cells to form bone-producing osteoblasts. In the absence of Indian
PHRP hedgehog in mice there is an accelerated rate of hypertrophy and differentiation of
chondrocytes to osteoblasts, ­resulting in short, stubby limbs. Cell proliferation is also
proliferative
chondrocytes
Fig. 13.12 Indian hedgehog (Ihh) and parathyroid-hormone-related protein (PHRP)
form feedback loops that maintain chondrocyte proliferation and growth at the ends
of developing bones. PHRP secreted by the chondrocytes at the ends of developing long
bones acts on proliferating chondrocytes (blue) to maintain proliferation and thus prevents the
hypertrophic expression of Indian hedgehog. Chondrocytes farther away from the end of the bone (orange)
chondrocytes escape the influence of PHRP and express Ihh. This acts on adjacent chondrocytes to increase
Ihh
the rate of proliferation and also acts on perichondrial cells to form osteoblasts. In some way
that is not yet understood, production of Ihh also stimulates PHRP synthesis by the chondrocytes
perichondrium at the end of the bone, thus forming a positive feedback loop that maintains PHRP production.
Adapted from Kronenberg, H.M.: 2003.
 Growth 519

suppressed, and Indian hedgehog production increased, by FGFs that are produced in
the perichondrium.
Growth hormone affects bone growth by acting on the growth plates. The cells
in the germinal zone have receptors for growth hormone, and growth hormone is
probably directly responsible for stimulating these stem cells to proliferate. Further
growth, however, is probably mediated by IGF-1, whose production in the growth
plate is stimulated by growth hormone. Thyroid hormones are also necessary for
optimal bone growth; they act both by increasing the secretion of growth hormone
and IGF-1, and by stimulating hypertrophy of the cartilage cells. FGF is also important
for bone growth; the genetic defect that gives rise to achondroplasia (short-limbed
dwarfism) is a dominant mutation in the FGF receptor-3, whose normal function is
to limit, rather than promote, bone formation. When it is mutated, it limits growth
abnormally in response to FGF.
The rate of increase in the length of a long bone is equal to the rate of new cell
­production per column multiplied by the mean height of an enlarged cell. The rate
of new cell production depends both on the time cells take to complete a cycle in
the proliferative zone, and the size of this zone. Different bones grow at different
rates, and this can reflect the size of the proliferative zone, the rate of proliferation,
and the degree of cell enlargement in the growth plate. If bone growth is prevented,
for example by heavy loading, then there is some catch-up growth when the load
is removed.
In view of the complexity of the growth plate, it is remarkable that human bones in
limbs on opposite sides of the body can grow for some 15 years independently of each
other, and yet eventually match to an accuracy of about 0.2%. This may be achieved
by having many columns of cells in each plate, so that growth variation between
cells is averaged out. When growth of a bone ceases the growth plate ossifies, and
this occurs at different times for different bones. Ossification of growth plates occurs
in a strict order in different bones and can therefore be used to provide a measure of
physiological age. The timing of growth cessation in the growth plate appears to be
instrinsic to the plate itself rather than to hormonal influence. The cessation of growth
is due to the cessation of cell proliferation and this timing may be programmed in
the cells. Cell senescence, and thus growth cessation, may be due to the chondrocyte
stem cells having only a finite potential for division. 3
Sarcomere number (thousands)

Normal development
13.9 Growth of vertebrate striated muscle is dependent on tension 2
Recovery
The number of striated (skeletal) muscle fibers in vertebrates is determined during
embryonic development. Once differentiated, striated muscle cells lose the ability to
1
divide. Post-embryonic growth of muscle tissue results from an increase in individual plaster cast removed
fiber size, both in length and girth, during which the number of myofibrils within the young muscles immobilized
enlarged muscle fiber can increase more than 10-fold. Additional nuclei to support the 0
functioning of the much-enlarged cell are provided by the fusion of satellite cells with 0 2 4 6 8 10 12 14
the fiber. Satellite cells, which are undifferentiated cells lying adjacent to the differen- Time (weeks)
tiated muscle, also act as a reserve population of stem cells that can replace damaged
muscle (see Section 10.10). Fig. 13.13 The growth in length of the
The increase in the length of a muscle fiber is associated with an increase in the muscles attached to the long bones of
number of sarcomeres—the functional contractile units—it contains. For example, in the mouse leg depends on the tension
provided by long bone growth. The length
the soleus muscle of the mouse leg, as the muscle increases in length, the number of
of a muscle is related to the number of
sarcomeres increases from 700 to 2300 at 3 weeks after birth. This increase in num-
sarcomeres—the basic contractile unit. If
ber seems to depend on the growth of the long bones putting tension on the muscle the limb is immobilized by a plaster cast in
through its tendons. If the soleus muscle is immobilized by placing the leg in a plaster a position in which there is no tension on
cast at birth, sarcomere number increases slowly over the next 8 weeks, but then the muscle, there is little increase in muscle
increases rapidly when the cast is removed (Fig. 13.13). One can, therefore, see how length until the cast is removed. There is
bone and muscle growth are mechanically coordinated. then a rapid increase in length.
520 Chapter 13 Growth and post-embryonic development

13.10 Cancer can result from mutations in genes that control cell
multiplication and differentiation

Cancer can be regarded as a major perturbation of normal cellular behavior that results
from certain mutations in somatic cells. Creating and maintaining tissue organization
requires strict controls on cell division, differentiation, and growth. In cancer, cells
escape from these normal controls, and proceed along a path of uncontrolled growth
and migration that can kill the organism. There is usually a progression from a benign
localized growth to malignancy in which the cells metastasize—migrate to many parts
of the body where they continue to grow. Most cancers derive from a single abnormal
cell that has acquired a number of mutations. Progression of a mutant cell to becoming a
tumor-producing cell, a process known as tumor progression, is an evolutionary process,
involving both further mutations and selection of those cells best able to proliferate.
The cells most likely to give rise to cancer are those that are undergoing continual
division, such as stem cells. Because they replicate their DNA frequently, they are
more likely than other cells to accumulate mutations that arise from errors in DNA
replication. In almost all cancers, the cancer cells are found to have a mutation in one
or more, and usually many, genes. An average of 63 mutations are found in a pancre-
atic cancer cell. There is a major rewiring of genetic circuits in cancer cells. Particular
genes in which mutation can contribute to cancer formation have been identified in
humans and other mammals. These genes are known as proto-oncogenes; when such
a gene undergoes mutation it becomes an oncogene. In some cases, the presence of
a single oncogene is capable of making a cell cancerous. At least 70 proto-oncogenes
have been identified in mammals.
There is another group of genes in which mutations can also lead to cancer. These
are the tumor-suppressor genes, in which inactivation or deletion of both copies of
the gene is required for a cell to become cancerous. We have already seen why some
of the components of the Hippo pathway are considered as tumor suppressors (see Sec-
tion 13.6). The classic human example of a tumor caused by the loss of such a gene
is the childhood tumor retinoblastoma, which is a tumor of retinal cells. Although
retinoblastoma is normally very rare, some families have an inherited predisposition
to it. This predisposition is determined by a single gene, and the affected families were
the means of identifying this gene. The inherited defect in some of the families turned
out to be a deletion of a particular region on one of the two copies of chromosome 13.
This on its own does not cause the cells to be cancerous. However, if a retinal cell
also acquires a deletion of the same region on the other copy of chromosome 13, a
retinal tumor develops. The gene in this region that is responsible for susceptibility to
retinoblastoma is known as the retinoblastoma (RB) gene. Both copies of the RB gene
must be lost or inactivated for a cell to become cancerous (Fig. 13.14), and hence RB
is regarded as a tumor-suppressor gene. The RB gene encodes a protein, RB, which is
involved in the regulation of the cell cycle.
The tumor-suppressor gene p53 plays a key role in many cancers; about half of all
human tumors contain a mutated form of p53. This gene is not required for develop-
ment per se, but when cells are exposed to agents that damage DNA, then p53 is acti-
vated and arrests the cell cycle, giving the cell time to repair the DNA. The p53 protein
thus prevents the cell from replicating damaged DNA and giving rise to mutant cells.
Instead, p53 will cause the cell to die by apoptosis if the damage is too severe to be
repaired. The mutant forms of p53, found in many cancers, do not promote apoptosis,
and so the affected cells are more likely to accumulate mutations.
A major feature of cancer is the failure of tumor cells to differentiate properly. The
majority of cancers—over 85%—occur in epithelia. This is not surprising when it is
recalled that many epithelia (such as the epidermis and the lining of the gut) are con-
stantly being renewed by division and differentiation of stem cells (see Section 10.7).
In normal epithelia, cells generated by stem cells continue to divide for a little time
 Growth 521

Normal individual Hereditary retinoblastoma

inherited
mutant
Rb gene

Fig. 13.14 The retinoblastoma (RB)

gene is a tumor-suppressor gene.
Occasional deletion of one Occasional inactivation of other
of the two Rb genes functional Rb gene copy If only one copy of the RB gene is lost or
inactivated, no tumor develops (left panel).
In individuals already carrying an inherited
mutant RB gene, if the other copy of the
Excessive cell gene is lost or inactivated in a cell, that cell
proliferation, leading will generate a retinal tumor (right panel).
to retinoblastoma Such individuals are thus at a much greater
risk of developing retinoblastoma, and usually
No tumor formation Tumor formation
do so at a young age.

until they undergo differentiation, when they stop dividing. By contrast, cancerous
epithelial cells continue to divide, although not necessarily more rapidly, and usually
fail to differentiate. Another feature of cancer cells, unlike developing cells, is that
when they divide, they are genetically unstable; the gain or loss of chromosomes is
common in solid tumors.
This failure of cancer cells to differentiate is also clearly seen in certain leukemias—
cancers of white blood cells. All blood cells are continually renewed from a multipo-
tent stem cell in the bone marrow, by a process in which steps in differentiation are
interspersed with phases of cell proliferation. The pathway eventually culminates in
cell differentiation and a cessation of cell division (see Section 10.4). Several types of
leukemia are caused by cells continuing to proliferate instead of differentiating. These
cells become stuck at a particular immature stage in their normal development, a
stage that can be identified by the molecules expressed on their cell surfaces.
A number of developmental genes that we have considered elsewhere are also
involved in cancer. One view is that many cancers arise from tissue stem cells and
that the Hedgehog- and Wnt-signaling pathways may promote cancer growth by pro-
moting stem-cell renewal. For example, the first known mammalian member of the
Wnt family was an oncogene (mouse Int-1, which is involved in brain development),
and abnormal expression of Wnts can block cell differentiation. The Wnt pathway
is crucial to the proliferation of intestinal cells, and overactivation of this pathway is
known to be involved in colorectal cancer, most commonly through mutations in the
gene for APC (adenomatous polyposis coli). Normally, APC helps to keep the pathway
inhibited in the absence of Wnt signaling (see Box 1E, p. 26) and thus acts as a tumor
suppressor. Mutations that inactivate APC lead to the constitutive activation of the
pathway and thus to unregulated cell proliferation. The Hedgehog-signaling pathway
is thought to control Wnt signaling in normal intestinal epithelia, confining Wnt activ-
ity to stem and progenitor cells, and mutations in Hedgehog pathway components
that disrupt this function can lead to tumors. A link between Hedgehog signaling and
cancer was first recognized by the finding that mutations in the Hedgehog receptor
Patched (see Fig. 2.40) were a cause of the rare, inherited Gorlin’s syndrome, one
feature of which is multiple cancers of epidermal basal cells. Abnormal activation
of the Hedgehog pathway is also present in nearly all pancreatic cancers, where it
is thought to maintain cancer stem cells and to be involved in cancer progression.
Mutations in the Notch pathway can also lead to a block in differentiation, and can
522 Chapter 13 Growth and post-embryonic development

therefore also result in a cancer, whereas members of the TGF-β family are involved in
tumor suppression. Mutations in microRNAs (miRNAs) (which act as gene-regulatory
RNAs) have also recently been shown to be involved in some human cancers. Some
of the miRNAs involved can act as either tumor suppressors or oncogenes in different
circumstances.
Most cancer-related deaths result from tumors that have spread from their site
of origin to other tissues, the process known as metastasis. As most cancers have
their origin in epithelial tissue, the ability of tumor cells to undergo an epithelial-
to-mesenchymal transition is central to metastasis, as the tumor cells must lose their tis-
sue organization and migrate as mesenchymal cells. Metastasis involves the loss of the
­E-cadherin-mediated cell–cell adhesion present in epithelia, allowing the cancer cells
to migrate out of the epithelium and invade the underlying tissue. If the migrating cells
enter the blood stream, they can be carried a long way from their origin. Epithelial-
to-mesenchymal transitions are a feature of many stages of embryonic development
and the control of this transition is discussed in more detail in Section 8.7.
Rarely, cancers can develop without any detectable alteration in the cell’s genetic
material. The clearest examples are teratocarcinomas, solid tumors that arise spon-
taneously from germ cells. Teratocarcinomas are unusual tumors, in that they can
contain a bizarre mixture of differentiated cell types. Spontaneous teratocarcinomas
usually occur in the ovary or testis, and are derived from the germ cells. In the mouse
ovary, accidental activation of an unfertilized egg results in its development in situ to
the stage of epiblast formation; the epiblast then gives rise to a tumor. Similarly, if the
epiblast of an early mouse embryo is transplanted to any site in the body of an adult
mouse in which it receives a good blood supply, it gives rise to a teratocarcinoma.
Various lines of evidence indicate that teratocarcinomas are not caused by genetic
alterations. A mouse inner cell mass placed in culture gives rise to embryonic stem
cells (ES cells), which can grow indefinitely in culture (discussed in Chapter 10).
When put back into the inner cell mass of another embryo, ES cells contribute nor-
mally to many different tissues, including the germline, to produce a chimeric mouse.
However, when the same ES cells are placed under the skin of an adult mouse, they
develop into a teratocarcinoma (see Fig. 10.32). Transgenic mice containing tissues
derived from ES cells do not have an increased probability of forming tumors, yet
those same cells consistently form tumors when grafted into adult mice. This indi-
cates that the teratocarcinoma must be the result of the ES cells receiving the wrong
developmental signals, and not of a genetic change, and raises a warning in relation
to the therapeutic use of ES cells, as discussed in Chapter 10.

13.11 Hormones control many features of plant growth

Plant growth is achieved by cell division in meristems and organ primordia, followed
by irreversible cell enlargement, which achieves most of the increase in size (see Chap-
ter 7 and Section 8.16). Unlike the protein growth hormones of animals, plant hor-
mones are typically small organic molecules. Auxin (indole-3-acetic acid) is one of the
main regulators of plant growth, and is implicated in a large number of developmental
processes, including embryonic development (see Section 7.3), growth towards the
light, tissue polarity, vascular tissue differentiation, and apical dominance, which is
the suppression of the growth of lateral buds immediately below the apical bud. Api-
cal dominance is caused by a diffusible inhibitor of bud outgrowth produced by the
shoot apex, as shown by putting an excised apex in contact with an agar block, which
can then, on its own, inhibit lateral growth. This inhibitor is auxin. It is produced by
the apical bud, transported down the stem, and suppresses the outgrowth of buds that
fall within its sphere of influence. If the apical bud is removed, apical dominance is
also removed, and lateral buds start to grow (Fig. 13.15). Application of auxin to the
cut tip replaces the suppressive effect of the apical bud.
 Molting and metamorphosis 523

When the upper part of a shoot is removed the lateral buds If the main apex is removed and placed on an agar block, auxin diffuses
beneath the cut start to grow into the block, which can then inhibit lateral bud growth

cut cut

lateral
buds

agar block

Fig. 13.15 Apical dominance in

Another family of plant hormones, the gibberellins, regulate stem elongation and
plants. Left panels: the growth of lateral
have some similar effects to auxin. Cytokinins, which can stimulate cell proliferation
buds is inhibited by the apical meristem
in culture, are derivatives of adenine. The chemical nature of plant hormones is very above them. If the upper part of the stem is
different from those of animals; nevertheless, they are thought to act through specific removed, lateral buds start growing. Right
hormone-binding receptors and to stimulate intracellular signal transduction. panels: an experiment to show that apical
Plant growth is affected by a variety of environmental factors, such as temperature, dominance is due to inhibition of lateral bud
humidity, and light. A seedling grown in the dark takes on a characteristic etiolated form outgrowth by a substance secreted by the
in which chloroplasts do not develop, there is extensive elongation of internodes, and apical region. This substance is the plant
the leaves do not expand. The effect of light on plant growth (photomorphogenesis) is hormone auxin.
mediated through a family of intracellular receptor proteins called phytochromes, which
respond to red light and regulate many aspects of plant development and growth.

Summary

Growth in both animals and plants mainly occurs after the basic body plan has been
laid down and the organs are still very small. Final organ size can be controlled both
by external signals and by intrinsic growth programs. In some cases, organ size can
be determined by monitoring the dimensions of the growing organ rather than being
set by cell number or cell size. In animals, growth can occur by cell multiplication, cell
enlargement, and secretion of large amounts of extracellular matrix. A plant’s growth in
size is by cell division usually followed by considerable cell enlargement. In mammals,
insulin-like growth factors are required for normal embryonic growth, and they also
mediate the effects of growth hormone after birth. Human post-natal growth is largely
controlled by growth hormone, which is made in the pituitary gland. Growth in the long
bones occurs in response to growth hormone stimulation of the cartilaginous growth
plates at either end of the bone. Cancer is the result of the loss of growth control
and differentiation. Plant growth is due to cell division in meristems followed by cell
enlargement, and is dependent on auxins, gibberellins, and other growth hormones.

Molting and metamorphosis

Many animals do not develop directly from an embryo into an ‘adult’ form, but into a
larva from which the adult eventually develops by metamorphosis. The changes that
occur at metamorphosis can be rapid and dramatic, the classic examples being the
metamorphosis of a caterpillar into a butterfly, a maggot into a fly, and a tadpole into
a frog. Another striking example of metamorphosis is the transformation of the pluteus
524 Chapter 13 Growth and post-embryonic development

larva of the sea urchin into the adult (see Fig. 6.18). In some cases it is hard to see any
4
resemblance between the animal before and after metamorphosis. The adult fly does
not resemble the larva at all, because adult structures develop from the imaginal discs
3
and so are completely absent from the larval stages (see Chapters 2 and 11). In frogs, the
most obvious external changes at metamorphosis are the regression of the tadpole’s tail
2
and the development of limbs, although many other structural changes occur. In some
1
insects, the entire body plan is transformed, with most larval tissues undergoing cell
death as the adult tissues develop from the imaginal discs and histoblasts. In arthropods
and nematodes, increase in size in larval and pre-adult stages requires shedding of the
5 external cuticle, which is rigid, a process known as molting.
A number of features distinguish early embryogenesis from molting, metamorpho-
sis, and other aspects of post-embryonic development. Whereas the signal molecules in
Fig. 13.16 Growth and molting of the
early development act over a short range and are typically protein growth factors, many
caterpillar of the tobacco hawkmoth
(Manduca sexta). The caterpillar, known as signals in post-embryonic development are produced by specialized endocrine cells,
the tobacco hornworm, goes through a series and include both protein and non-protein hormones. The synthesis of these hormones
of molts. The tiny hatchling (1)—indicated by is orchestrated by the central nervous system in response to environmental cues, and
the arrow—molts to become a caterpillar (2), there is complex feedback between the endocrine glands and their secretions.
and then undergoes three further molts
(3, 4, and 5). The increase in size between
13.12 Arthropods have to molt in order to grow
molts is about twofold. The caterpillars are
sitting on a lump of caterpillar food. Scale Arthropods have a rigid outer skeleton, the cuticle, which is secreted by the epider-
bar = 1 cm. mis. This makes it impossible for the animals to increase in size gradually. Instead,
Photograph courtesy of S.E. Reynolds. increase in body size takes place in steps, associated with the loss of the old outer
skeleton and the deposition of a new larger one. This process is known as ecdysis
or molting. The stages between molts are known as instars. Drosophila larvae have
three instars and molts. The increase in overall size between molts can be striking, as
illustrated in Fig. 13.16 for the tobacco hornworm (Manduca).
At the start of a molt, the epidermis separates from the cuticle in a process known
as apolysis, and a fluid (molting fluid) is secreted into the space between the two
(Fig. 13.17). The epidermis then increases in area by cell multiplication or cell enlarge-
ment, and becomes folded. It begins to secrete a new cuticle, and the old cuticle is
partly digested away, eventually splits, and is shed.
Molting is under hormonal control. Stretch receptors that monitor body size are
activated as the animal grows, and this results in the brain secreting prothoracico-
tropic hormone (PTTH). This activates the prothoracic gland to release the steroid
hormone ecdysone, which is the hormone that causes molting. A similar hormonal
Fig. 13.17 Molting and growth of the circuit controls metamorphosis, and is described more fully in the next section.
epidermis in arthropods. The cuticle is
secreted by the epidermis. At the start of
13.13 Metamorphosis is under environmental and hormonal control
molting, the cuticle separates from the
epidermis—apolysis—and a fluid is secreted When an insect larva has reached a particular stage, it does not grow and molt any
between them. The epidermis grows, further but undergoes a more radical metamorphosis into the adult form. Metamor-
becomes folded, and begins to secrete a new phosis occurs in many animal groups other than arthropods, including amphibians. In
cuticle. Enzymes weaken the old cuticle,
both insects and amphibians, environmental cues, such as nutrition, temperature, and
which is shed.

Apolysis: separation of Secretion of fluid. Activation of enzymes

Intermolt stage epidermis from cuticle Growth of epidermis Secretion of new cuticle in molting fluid Shedding of old cuticle

cuticle

epidermis
 Molting and metamorphosis 525

Environmental cues brain

e.g. nutrition,
temperature,
light PTTH

corpus prothoracic gland

allatum

Juvenile hormone Ecdysone

(prevents metamorphosis) (promotes metamorphosis)
Fig. 13.18 Insect metamorphosis. The
corpus allatum of a butterfly larva
secretes juvenile hormone, which inhibits
metamorphosis. In response to environmental
changes, such as an increase in light
and temperature, the corpus allatum of
the final instar larva begins to secrete
metamorphosis prothoracicotropic hormone (PTTH). This
acts on the prothoracic gland to stimulate
the secretion of ecdysone, the hormone that
overcomes the inhibition by juvenile hormone
Embryo Larva Pupa Adult and causes metamorphosis.
After Tata, J.R.: 1998.

light, as well as the animal’s internal developmental program, control metamorphosis

through their effects on neurosecretory cells in the brain. There are two groups of hor-
mone-producing cells, one of which promotes metamorphosis, and the other inhibits it.
Metamorphosis occurs when the inhibition, which is predominant in the larval stage,
is overcome in response to environmental cues. The signals produced by the two sets
of endocrine cells control the development of all the cells involved in metamorphosis.
Larval tissues, such as gut, salivary glands, and certain muscles, undergo programmed
cell death. The imaginal discs now develop into rudimentary adult appendages, such
as wings, legs, and antennae. The nervous system is also remodeled.
In insects, temperature and light cues stimulate neurosecretory cells in the lar-
va’s central nervous system to release signals that act on a neurosecretory release
site behind the brain, which then secretes PTTH. This acts on the prothoracic gland
to stimulate the production of the steroid hormone ecdysone. It is ecdysone that
promotes metamorphosis, as demonstrated by its ability to induce premature meta-
morphosis in Drosophila larvae. The action of ecdysone is counteracted by another
hormone—juvenile hormone—produced by the corpus allatum, an endocrine gland
located just behind the brain. As its name implies, juvenile hormone maintains the
larval state. In butterflies, a pulse of ecdysone in the final instar larva triggers the
beginning of pupa formation, and another pulse some days later initiates the later
stages of metamorphosis (Fig. 13.18).
Ecdysone crosses the plasma membrane, where it interacts with intracellular ecdy-
sone receptors that belong to the steroid hormone receptor superfamily. These recep-
tors are gene-regulatory proteins, which are activated by binding their hormone ligand
(see Box 5D, pp. 192–193). The hormone–receptor complex binds to the regulatory
regions of a number of different genes, inducing a new pattern of gene activity char-
acteristic of metamorphosis.
In amphibians, in response to nutritional status and environmental cues, such as
temperature and light, the neurosecretory cells of the hypothalamus release corticotro-
pin-releasing hormone, which acts on the pituitary gland, causing it to release thyroid-
stimulating hormone (thyrotropin). (This action of corticotropin-releasing hormone is
peculiar to non-mammalian vertebrates, and in Xenopus only occurs at the tadpole
526 Chapter 13 Growth and post-embryonic development

Environmental cues
e.g. nutrition, temperature, light hypothalamus +

CRH
pituitary +

TSH
thyroid

Prolactin T4 T3 Positive feedback

(delays metamorphosis) (promote metamorphosis)
Fig. 13.19 Amphibian
metamorphosis. Changes in the
environment, such as an increase in
nutritional levels, cause the secretion of
corticotropin-releasing hormone (CRH) from
the larval hypothalamus, which acts on
the pituitary to release thyroid-stimulating
hormone (TSH). This in turn acts on the
thyroid glands to stimulate secretion
of the thyroid hormones thyroxine (T4)
and tri-iodothyronine (T3), which cause
metamorphosis. The thyroid hormones also
act on the hypothalamus and pituitary to
maintain synthesis of CRH and TSH. Embryo Tadpole Adult
After Tata, J.R.: 1998.

stage; in adult frogs, as in mammals, the release of thyroid-stimulating hormone is

due to thyrotropin-releasing hormone.) Thyroid-stimulating hormone in turn acts on
the thyroid gland to stimulate the secretion of the thyroid hormones that bring about
metamorphosis (Fig. 13.19). The thyroid hormones are the iodo-amino acids thyroxine
(T4), and tri-iodothyronine (T3). They are signaling molecules of ancient origin, occur-
ring even in plants. Although very different in chemical structure to ecdysone, they
too pass through the plasma membrane and interact with intracellular receptors that
belong to the same superfamily as the receptors for retinoic acid and steroid hormones
(see Box 5D, pp. 192–193). The pituitary also produces the protein hormone prolactin,
which was originally thought to be an inhibitor of metamorphosis; however, overex-
pression of prolactin does not prolong tadpole life but reduces tail resorption.
A striking feature of the hormones that stimulate metamorphosis is that, as well as
affecting a wide variety of tissues, they affect different tissues in different ways, their
effects varying from the subtle to the gross. In the tadpole limb, for example, thyroid
hormones promote development and growth, whereas they cause cell death and degen-
eration in the tail. Fast muscles are the first to go, and later the notochord collapses. Yet
in all these cases, the hormone produces these very different effects by binding to the
same intracellular receptor. The difference in outcomes is due to the hormone–receptor
complex switching on or off different sets of genes in different tissues, as a result of the
different developmental histories of the regions. Each tissue has its own response to the
hormones that cause metamorphosis, and some of these effects can be reproduced in
culture. When excised Xenopus tadpole tails are exposed to thyroid hormones in cul-
ture, for example, they cause cell death and complete tissue regression. Metamorphosis
also leads to changes in the responsiveness of cells to other signals; for example, in
Xenopus, estrogen can only induce the synthesis of vitellogenin, a protein required for
the yolk of the egg, after metamorphosis and the attainment of sexual maturity.
There are alterations in the expression of many genes—several hundred at least—
during Drosophila metamorphosis. In Drosophila, because of the special characteristics
of polytene chromosomes, changes in gene activity that occur during metamorphosis are
 Aging and senescence 527

actually visible. Cells in some larval tissues (such as the salivary glands) grow and repeat-
edly pass through S phase without undergoing mitosis and cell division. The cells become
very large and can have several thousand times the normal complement of DNA. In sali-
vary gland cells, many copies of each chromosome are packed side by side to form giant
polytene chromosomes. When a gene is active, the chromosome at that site expands into
a large localized ‘puff’, which is easily visible (Fig. 13.20). The puff represents the unfold-
ing of chromatin and the associated transcriptional activity. When the gene is no longer
active, the puff disappears. During the last days of larval life, a large number of puffs are
formed in a precise sequence, a pattern that is under the direct influence of ecdysone.

Summary

Arthropod larvae grow by undergoing a series of molts in which the rigid cuticle is shed.
Metamorphosis during the post-embryonic period can result in a dramatic change in the
form of an organism. In insects, it is hard to see any resemblance between the animal
before and after metamorphosis, whereas in amphibians, the change is somewhat less
dramatic. Environmental and hormonal factors control metamorphosis. In both insects
and amphibians there are two sets of hormonal signals, one promoting, the other delay-
ing, metamorphosis. Thyroid hormones cause metamorphosis in amphibians, and ecdy-
sone does the same in insects. In Drosophila, gene activity during metamorphosis can
be monitored by localized puffing on the giant polytene chromosome.

Fig. 13.20 Gene activity seen as

Summary: molting and metamorphosis puffs on the polytene chromosomes of
Drosophila. A region of a chromosome is
Molting in arthropods Metamorphosis in amphibians shown from a young third instar (left), and
from an older larva (right), after ecdysone has
growth of arthropods stretches cuticle environmental cues induced puffs at three loci (arrowed).
brain prothoracicotropic Photograph courtesy of M. Ashburner.
hormone (PTTH)
acts on prothoracic gland acts on hypothalamus
corticotropin-
releases ecdysone
releasing hormone
molting; old cuticle destroyed, pituitary gland
new one formed thyroid-stimulating
hormone
thyroid
Metamorphosis in arthropods thyroxine

environmental and internal factors metamorphosis

lead to secretion of PTTH

ecdysone released

metamorphosis juvenile hormone inhibits

Aging and senescence

Organisms are not immortal, even if they escape disease or accidents. With aging—
the passage of time—comes an increasing impairment of physiological functions,
which reduces the body’s ability to deal with a variety of stresses, and an increased
susceptibility to disease. This age-related decline in function is known as senescence.
528 Chapter 13 Growth and post-embryonic development

Aging is observed in most multicellular animals, but there are notable exceptions,
% 100
survival such as cnidarians (sea anemones, Hydra, and their relations) and planarians. All
sexually reproducing animals age, whereas most asexual animals do not. Hydra,
75 whose capacity for regeneration is discussed in the next chapter, does not age unless
it undergoes sexual differentiation. The phenomenon of senescence raises many
50
questions as to the underlying mechanisms, and these are still largely unanswered,
but we can at least consider some general questions, such as whether senescence
is part of an organism’s post-embryonic developmental program or whether it is
25 simply the result of wear and tear. Germ cells do not age; if they did, the species
would die out.
0 Although individuals may vary in the time at which particular aspects of aging
0 25 50 75 100 appear, the overall effect is summed up as an increased probability of dying in most
Age (days)
animals, including humans, with increased age. This life pattern, which is illustrated
in relation to Drosophila (Fig. 13.21), is typical of many animals, but there is little evi-
Fig. 13.21 Aging in Drosophila. The
dence that aging contributes to mortality in the wild; more than 90% of wild mice die
probability of dying increases rapidly at older
ages.
during their first year. There are, however, exceptions, such as the Pacific salmon, in
which death does not come after a process of gradual aging, but is linked to a certain
stage in the life cycle, in this case to spawning.
One view of aging is that it is due to an accumulation of damage that eventually
outstrips the ability of the body to repair itself, and so leads to the loss of essen-
tial functions. For example, some old elephants die of starvation because their teeth
have worn out. The nematode Caenorhabditis elegans lives for an average or around
20 days and the major cellular change that occurs with age is the progressive deterio-
ration of muscle. This deterioration has a random effect on the life span, which varies
from 10 to 30 days. Nevertheless, there is clear evidence that senescence is under
genetic control, as different species age at vastly different rates, as shown by their
different life spans (Fig. 13.22). An elephant, for example, is born after 21 months’
embryonic development, and at that point shows few, if any, signs of aging, whereas
a 21-month-old mouse is already well into middle age and beginning to show signs
of senescence.
The genetic control of aging can be understood in terms of the ‘disposable soma’
theory, which puts it into the context of evolution. The disposable soma theory

Longevity and time to attain reproductive maturity at puberty for various mammals
Maximum Length Age at
lifespan (months) of gestation (months) puberty (months)

Human 1440 9 144

Finback whale 960 12 –
Indian elephant 840 21 156
Horse 744 11 12
Chimpanzee 534 8 120
Brown bear 442 7 72
Dog 408 2 7
Cattle 360 9 6
Rhesus monkey 348 5.5 36
Cat 336 2 15
Pig 324 4 4
Squirrel monkey 252 5 36
Sheep 240 5 7
Gray squirrel 180 1.5 12
European rabbit 156 1 12
Guinea-pig 90 2 2
House rat 56 0.7 2
Fig. 13.22 Table showing life span, Golden hamster 48 0.5 2
length of gestation, and age at puberty for Mouse 42 0.7 1.5
various mammals.
 Aging and senescence 529

­ roposes that natural selection tunes the life history of the organism so that sufficient
p
­resources are invested in maintaining the repair mechanisms that prevent aging, at
least until the organism has reproduced and cared for its young. Thus mice, which
start reproducing when just a few months old, need to maintain their repair mecha-
nisms for much less time than do elephants, which only start to reproduce when
around 13 years old. In most species of animals in the wild, few individuals live long
enough to show obvious signs of senescence and senescence need only be delayed
until reproduction is complete. Cells have numerous mechanisms to delay aging,
which are quite similar to the mechanisms used to prevent malignant transforma-
tion. These cellular mechanisms protect the cell from internal damage by reactive
chemicals and routinely repair damage to DNA, which is occurring continually in
living cells even when they are not actively dividing. They are particularly active in
germ cells.

13.14 Genes can alter the timing of senescence

The maximum recorded life spans of animals show dramatic differences (see
Fig. 13.22). Humans can live as long as 120 years, some owls 68 years, cats 28 years,
Xenopus 15 years, mice 3.5 years, and the nematode about 25 days. Mutations in
genes that affect life span have been identified in C. elegans, Drosophila, mice, and
humans, and may give clues to the mechanisms involved; the ability to resist damage
to DNA and the effects of oxygen radicals are important.
A C. elegans worm that hatches as a first instar larva in an uncrowded environ-
ment with ample food grows to adulthood and can survive for 25 days. In crowded
conditions and when food is short, however, the animal enters a quiescent third-instar
larval state known as the dauer state, where it neither eats nor grows until food
becomes available again. When conditions become favorable, the dauer larva molts
and becomes a fourth instar larva. The dauer state can last for 60 days and has no
effect on the post-dauer life span: it is therefore considered to be a state in which the
Calorie Absence
larva does not age. An insulin/IGF-1 signaling system plays an important role both in restriction of
controlling entry into the dauer state in response to stress and in the overall control (Sir2) germline
of fertility, life span, and metabolism in C. elegans and in Drosophila. Reduction in
insulin/IGF-I signaling increases longevity (Fig. 13.23).
In C. elegans, mutations that cause a strong reduction in the expression of daf-2,
which encodes a receptor in the insulin-signaling pathway, arrest development in the Longevity
dauer state. The normal role of DAF-2 is to antagonize the activity of DAF-16, a tran-
scription factor whose activity lengthens life span and increases resistance to some
types of stress. A partial loss of DAF-2 function results in longer adult life-span after
the dauer state but with reduced fertility and viability of the progeny, while a further Insulin/IGF signaling via DAF-2
(inhibition of DAF-16/Fox0)
reduction in DAF-2 at the larval stage by RNA interference increases life span even
more, without shortening the dauer state. The presence of a germline seems to have
a negative effect on life-span extension. Removal of germline precursor cells in daf-2 Fig. 13.23 Potential inputs to aging
mutant embryos resulted in their having a mean life-span of 125 days and remaining in Caenorhabditis elegans and
quite healthy—in humans, this would equate to a life span of 500 years. But when Drosophila. Calorie restriction and the
daf-2 mutants were cultured alongside wild-type worms, the mutants became extinct absence of a germline promote an extended
in just a few generations, partly due to their reduced fertility. Microarray analysis of life-span in both animals. In Drosophila, the
the effects of DAF-16 show that it activates stress-response and antimicrobial genes. calorie-restriction effect has been shown to
require the histone deacetylase Sir2, which
In laboratory conditions it seems that aging animals in which DAF-16 is inhibited are
is likely to act on gene expression through
killed by the bacteria on which they feed.
its ability to modifiy chromatin (see Box 10A,
A similar system regulates aging in Drosophila. Mutations disabling the insulin/ p. 374). In C. elegans, insulin/IGF signaling
IGF-1 pathway almost double the life span in this animal, and calorie restriction also has been shown to regulate life span via
extends life span. The mutant flies, rather like the dauer larvae, enter a state of repro- its inhibitory effect on expression of the
ductive diapause or quiescence. These effects on life span may be due to resistance to transcription factor DAF-16 (a member of the
oxidative stress. Aging in mice can be retarded by reduction in pituitary activity, and FoxO family of transcription factors).
530 Chapter 13 Growth and post-embryonic development

there is also evidence that female mice with a mutation in the IGF-1 receptor (which
is the receptor activated by IGF-1 and IGF-2) can live 33% longer than usual.
There is evidence that oxidative damage accelerates aging and that reactive oxygen
radicals are key players in causing cell damage. Reduction in food intake increases life
span in other animals; rats on a minimal diet live about 40% longer than rats allowed
to eat as much as they like. This is thought to be partly due to a reduced exposure to
free radicals, which are formed during the oxidative breakdown of food. Free radicals
are highly reactive, and can damage both DNA and proteins. A long-lived rodent spe-
cies generates less reactive oxygen than the laboratory mouse. Oxidative stress also
exerts its effects by damaging mitochondria.
Humans who are homozygous for the recessive gene defect known as Werner
syndrome show striking effects of premature aging. There is growth retardation at
puberty, and by their early twenties those affected by this syndrome have gray hair
and suffer from a variety of illnesses, such as heart disease, that are typical of old age.
Most people affected die before the age of 50. The gene affected in Werner syndrome
has been identified, and is thought to encode a protein involved in unwinding DNA.
Such unwinding is required for DNA replication, DNA repair, and gene expression.
The inability to carry out DNA repair properly in Werner-syndrome patients could
subject the genetic material to a much higher level of damage than normal. The link
between Werner syndrome and DNA thus fits with the possibility that aging is linked
to the accumulation of damage in DNA, but it may also be related to cell senescence.
Hutchinson–Gilford progeria syndrome (progeria means premature aging) is even
more severe in its effects. The gene affected in this syndrome encodes the intranuclear
protein lamin A, and the primary cellular defect is an instability in the structure of the
nuclear envelope, which means cells are more likely to die prematurely.

13.15 Cell senescence blocks cell multiplication

Fig. 13.24 Vertebrate fibroblasts can only
One might think that when cells are isolated from an animal, placed in culture, and
go through a limited number of divisions
provided with adequate medium and growth factors, they would continue to prolifer-
in culture. Fibroblasts placed in culture are
ate almost indefinitely. But this is not the case. For example, mammalian fibroblasts—
subcultured until they stop growing (top
panels). The number of cell doublings in connective tissue cells—will only go through a limited number of cell doublings in
culture before they stop dividing is related culture; the cells then stop dividing, however long they are cultured (Fig. 13.24). For
to their maximum age, as indicated by the normal fibroblasts, the number of cell doublings depends both on the species and the
figures in brackets on the graph (bottom age of the animal from which they are taken. Fibroblasts taken from a human fetus
panel).

Cells divide until they completely cover the dish and continue to divide After a finite number of cell multiplications,
when placed in fresh culture medium cells stop dividing
cells stop dividing

Galapagos
tortoise (175 years)

Human (120 years)

Chicken (30 years)

Mouse (3.5 years)

0 25 50 75 100 125
number of population doublings of cells in culture
 Aging and senescence 531

go through about 60 doublings, those from an 80-year-old about 30, and those from
an adult mouse about 12–15 doublings. When the cells stop dividing, they appear to
be healthy, but are stuck at some point in the cell cycle, often G0; this phenomenon
is known as cell senescence. Cells taken from patients with Werner’s syndrome, who
show an acceleration of many features of normal aging (see Section 13.14), make
significantly fewer divisions in culture than normal cells. However, it is far from clear
how this behavior of cells in culture is significant for the aging of the organism and to
what extent it reflects the way the cells are cultured.
A feature shared by senescent cells in culture and in vivo is shortening of the
­telomeres. Telomeres are the repetitive DNA sequences at the ends of the chromo-
somes that preserve chromosome integrity and ensure that chromosomes replicate
themselves completely without loss of information-encoding DNA at the ends. The
length of the telomeres is reduced in older human cells. Telomere length is found to
decrease at each DNA replication, suggesting that they are not completely replicated
at each cell division and this may be related to senescence. If the enzyme telom-
erase, which maintains telomere length and is normally absent from cells in culture,
is expressed in those cells, senescence in culture does not occur. However, it is not
yet clear whether telomere shortening is a major cause of aging in somatic cells, as
certain rodent cells such as Schwann cells can, under appropriate conditions, prolifer-
ate indefinitely and their telomeres do not control replication. ES cells also express
telomerase and proliferate indefinitely.
Senescent cells that have stopped dividing have changes in the expression of proteins
that are involved in cell-cycle control. The proteins p21 and p16 are often expressed;
they are part of the tumor-suppressor pathways governed by p53 and RB in response
to DNA damage (see Section 13.10). They act as inhibitors of cyclin-dependent kinases
and prevent a cell with damaged DNA from entering the cell cycle. This helps to pre-
vent senescent cells, with their DNA damage, from becoming cancerous.

Summary

Aging is largely caused by damage to cells, particularly by reactive oxygen, but is also
under genetic control. Many normal cells in culture can only undergo a limited number
of cell divisions, which correlate with their age at isolation and the normal life-span of
the animal from which they came. Genes that can increase life span have been identi-
fied in C. elegans and Drosophila and may act by increasing the animals’ resistance to
oxidative stress.

Summary to Chapter 13

The form of many animals is laid down in miniature during embryonic development, and they
then grow in size, keeping the basic body form, although different regions grow at different
rates. Growth may involve cell multiplication, cell enlargement, and the laying down of extra-
cellular material. In vertebrates, structures have an intrinsic growth program, which is under
hormonal control. Arthropod larvae grow by molting and both they and other animals, such
as frogs, undergo metamorphosis, which is under hormonal control. Cancer can be viewed
as an aberration of growth, as it usually results from mutations that lead to excessive cell
proliferation and failure of cells to differentiate. The symptoms of aging appear mainly to
be caused by damage to cells that accumulates over time, and aging is also under genetic
control. In plants, all adult structures are derived by growth from specialized regions called
meristems.
532 Chapter 13 Growth and post-embryonic development

n End of chapter questions Multiple choice (factual recall questions)

NB There is only one correct answer to each question.
Long answer (concept questions)
1. The maternally supplied Drosophila protein string is a
1. Define ‘growth.’ What are the main mechanisms of growth? How
does cell death fit into a discussion of growth? a) kinase
b) phosphatase
2. Why is it inaccurate to say that the cell cycle starts when cytoki-
nesis produces two cells after M phase? According to molecular anal- c) signaling molecule
ysis of the cell cycle, each round of the cycle ‘starts’ at what point in d) transcription factor
the cycle? What are the molecular events that define this start point?
2. What will be the result of grafting a limb bud from a large species
(The cell-cycle diagram in Box 1B might also be helpful.)
of the salamander Ambystoma onto a smaller species?
3. Most cell types in adult vertebrates are not actively dividing; where a) The grafted bud will be unable to grow in a smaller animal, and
in the cell cycle are they considered to be? What is required for them will be lost.
to re-enter the cell cycle? Name some adult differentiated cell types
that are able to divide and some that are incapable of cell division. b) The grafted bud will grow to a size appropriate to its host, and
will thus be of the same size as the other limbs.
4. In what way is the standard cell cycle modified during the first
c) The grafted bud will grow to become a larger limb than the
13 pre-cellular divisions in Drosophila embryogenesis, to allow very
­others, reflecting its origin in the larger species.
rapid cell divisions? What is the maternally supplied protein that
drives these early divisions; and what is this protein’s function? d) The grafted bud will initially grow larger than the others, but
with time will regulate and become a limb of the same size as
5. The final size of some organs, such as liver and spleen, is regu- the other limbs of its host.
lated by signaling between cells, whereas the size of other organs,
such as pancreas and thymus, is under the control of a cell-intrinsic 3. Growth is stimulated by which signaling molecules?
developmental program. Describe the experiments that reveal these a) insulin
two different mechanisms of growth control. b) IGF-1 and -2
6. What are some of the consequences of poor maternal nutrition c) FGF
on the offspring? d) all of these may stimulate growth
7. In Drosophila, the Hippo protein and the bantam microRNA have
4. Growth in the length of long bones in vertebrates occurs in the
opposing functions, as their names imply. What are the functions of
these two regulators of growth? Outline the pathway from the cell- a) diaphysis
surface protein that initiates signaling to Hippo, and signalling to the b) epiphysis
bantam gene. c) growth plate
8. Address the following questions about human growth hormone: d) ossification centers
What is growth hormone? Where is it produced? How is its synthe-
5. Normal postembryonic growth of striated muscle fibers occurs
sis and secretion controlled? What other growth factor is a major
through
mediator of growth hormone’s effect on cellular proliferation?
a) an increase in muscle fiber size, but not the number of muscle
9. Contrast the roles of chondrocytes and osteoblasts in growth of fibers
the long bones of vertebrates.
b) an increase in the number of muscle fibers
10. Summarize the effects of each of the following signaling mol-
c) differentiation of stem cells to form new muscle fibers
ecules on chondrocytes: PHRP, Indian hedgehog, thyroid hormones,
and FGF. d) division of the muscle cells to increase the total number of
muscle cells
11. Review the various signaling molecules and signaling pathways
discussed in the chapter, and indicate which could be classified as 6. Which of the following statements is true?
oncogenes, and which as tumor suppressors. a) The genes involved in cancer are all classed as oncogenes.
12. Elucidate the many connections found between Wnt, APC, b) All genes involved in cancer are of a class called tumor suppres-
β-catenin, and cadherins in the context of the progression of normal sor genes.
cells to metastatic cancer cells. c) Both oncogenes and tumor suppressor genes are involved in
­cancer.
13. In one experiment, a frog tadpoleis exposed to thyroxine in the
water in which it is being reared. In another experiment, the thyroid d) No unifying classifications can be applied to the genes involved
gland is removed from a tadpole. What results would you predict in in cancer.
each case. Justify your predictions. 7. The majority of cancers occur in epithelia because
14. Discuss the evidence for and against the statement: ‘The a) Epithelial tissues are more likely than other tissue types to
changes that occur with aging are an inevitable part of an animal’s be constantly renewed by division and differentiation of stem
genetically determined postembryonic developmental program.’ cells.
 Aging and senescence 533

b) Growth regulatory pathways are more common in epithelia than 13.3 Cell division in early development can be controlled by
in other tissue types, and hence are more likely to be deregulated an intrinsic developmental program
in epithelia than in other tissues. Edgar, B., Lehner, C.F.: Developmental control of cell cycle regula-
c) The cell adhesion contacts between cells in epithelia predispose tors: a fly’s perspective. Science 1996, 274: 1646–1652.
these cells to abnormal growth. Follette, P.J., O’Farrell, P.H.: Connecting cell behavior to pattern-
d) The differentiation of epithelia is complex, and more likely to go ing: lessons from the cell cycle. Cell 1997, 88: 309–314.
wrong than other tissues, resulting in cancer.
13.4 Organ size can be controlled by both intrinsic growth
8. Molting in insects and other arthropods is triggered by programs and extracellular signals
a) auxin Amthor, H., Huang, R., McKinnell, I., Christ, B., Kambadur, R.,
Sharma, M., Patel, K.: The regulation and action of myostatin
b) ecdysone
as a negative regulator of muscle development during avian
c) Hedgehog embryogenesis. Dev. Biol. 2002, 251: 241–257.
d) juvenile hormone Shingleton, A.W.: Body-size regulation: combining genetics and
physiology. Curr. Biol. 2005, 15: R825–R827.
9. Which signaling pathway seems to play a role in aging in several
Stanger, B.Z.: The biology of organ size determination. Diabetes
different animals, from worms to mice?
Obes. Metab. 2008, 10: 16–26.
a) ecdysone Stanger, B.Z., Tanaka, A.J., Melton, D.A.: Organ size is limited by
b) FGF the number of embryonic progenitor cells in the pancreas but
c) Hippo not in the liver. Nature 2007, 445: 886–891.
Taub, R.: Liver regeneration: from myth to mechanism. Nat. Rev.
d) insulin/IGF-1
Mol. Cell Biol. 2004, 5: 836–847.
10. Which of the following is consistent with a model for aging in
which stresses leading to DNA damage cause senescence? 13.5 The amount of nourishment an embryo receives can
have profound effects in later life
a) Werner’s syndrome is a premature aging illness, possibly caused
by a defect in DNA repair. Barker, D.J.: The Wellcome Foundation Lecture. 1994: The fetal ori-
gins of adult disease. Proc. R. Soc. Lond. 1995, 262: 37–43.
b) Dietary restriction in mammals reduces the production of DNA-
Gluckman, P.D., Hanson, M.A., Cooper, C., Thornburg, K.L.: Effect
damaging free radicals in the mitochondria.
of in utero and early-life conditions on adult health and dis-
c) Senescent cells in culture that have stopped dividing often express ease. N. Engl. J. Med. 2008, 359: 61–73.
p21 and p16, proteins that are normally activated in response to
DNA damage. 13.6 Determination of organ size involves coordination of
cell growth, cell division, and cell death
d) All of these facts are consistent with a model for aging based on
DNA damage. de la Cova, C., Abril, M., Bellosta, P., Gallant, P., Johnston, L.A.:
Drosophila myc regulates organ size by inducing cell competi-
Multiple choice answer key tion. Cell 2004, 117: 107–116.
Martin, F.A., Herrera, S.C., Morata, G.: Cell competition, growth
1: b, 2: c, 3: d, 4: c, 5: a, 6: c, 7: a, 8: b, 9: d, 10: d.
and size control in the Drosophila wing imaginal disc. Develop-
ment 2009, 136: 3747–3756.
n General further reading Pan, D.: Hippo signaling in organ size control. Genes Dev. 2007,
21: 886–897.
Kirkwood, T.B.L.: Understanding the odd science of aging. Cell
2005, 120: 437–447. Box 13A Gradients of signaling molecules could determine
Nature Insight: Cell division and cancer. Nature 2004, 432: 293–341. organ size
Partridge, L.: The new biology of ageing. Phil. Trans R. Soc. B 2010, Lawrence, P.A., Struhl, G., Casal, J.: Do the protocadherins Fat and
365: 147–154. Daschous link up to determine both planar cell polarity and
the dimensions of organs? Nat. Cell Biol. 2008, 10: 1379–1382.
Rogulja, D., Irvine, K.D.: Regulation of cell proliferation by a mor-
n Section further reading phogen gradient. Cell 2005, 123: 449–461.
Willecke, M., Hamaratoglu, F., Sansores-Garcia, L., Tao, C., Halder, G.:
13.1 Tissues can grow by cell proliferation, cell enlargement, Boundaries of Dachsous cadherin activity modulate the Hippo
or accretion signaling pathway to induce cell proliferation. Proc. Natl Acad.
Goss, R.J.: The Physiology of Growth. New York: Academic Press, Sci USA 2008, 105: 14897–14902.
1978.
13.7 Body size is also controlled by the neuroendocrine
13.2 Cell proliferation is controlled by regulating entry into system in both insects and mammals
the cell cycle Mirth, C.K., Riddiford, L.M.: Size assessment and growth control:
Morgan, D.O.: The Cell Cycle: Principles of Control. Oxford University how adult size is determined in insects. BioEssays 2007, 29:
Press 2007. 344–355.
534 Chapter 13 Growth and post-embryonic development

Nijhout, H.F.: Size matters (but so does time), and it’s OK to be De Loof, A.: Ecdysteroids, juvenile hormone and insect neuropep-
different. Dev. Cell 2008, 15: 491–492. tides: Recent successes and remaining major challenges. Gen.
Sanders, E.J., Harvey S.: Growth hormone as an early embryonic Comp. Endocrinol. 2008, 155: 3–13.
growth and differentiation factor. Anat. Embryol. 2004, 209: 1–9. Huang, H., Brown, D.D.: Prolactin is not juvenile hormone in
Stern, D.: Body-size control: how an insect knows it has grown Xenopus laevis metamorphosis. Proc. Natl Acad. Sci. USA 2000,
enough. Curr. Biol. 2003, 13: R267–R269. 97: 195–199.
Nijhout, H.F.: Size matters (but so does time), and it’s OK to be
13.8 Growth of the long bones occurs in the growth plates different. Dev. Cell 2008, 15: 491–492.
Kember, N.F.: Cell kinetics and the control of bone growth. Acta Tata, J.R.: Hormonal Signaling and Postembryonic Development.
Paediatr. Suppl. 1993, 391: 61–65. Heidelberg: Springer, 1998.
Kronenberg, H.M.: Developmental regulation of the growth plate. Thummel, C.S.: Flies on steroids—Drosophila metamorphosis and
Nature 2003, 423: 332–336. the mechanisms of steroid hormone action. Trends Genet. 1996,
Nilsson, O., Baron, J.: Fundamental limits on longitudinal bone 12: 306–310.
growth: growth plate senescence and epiphyseal function. White, K.P., Rifkin, S.A., Hurban, P., Hogness, D.S.: Microarray
Trends Endocrinol. Metab. 2004, 8: 370–374. analysis of Drosophila development during metamorphosis.
Roush, W.: Putting the brakes on bone growth. Science 1996, 273: Science 1999, 286: 2179–2184.
579. Wu, H.H., Ivkovic, S., Murray, R.C., Jaramillo, S., Lyons, K.M.,
Johnson, J.E., Calof, A.L.: Autoregulation of neurogenesis by
13.9 Growth of vertebrate striated muscle is dependent on GDF11. Neuron 2003, 37: 197–207.
tension
Schultz, E.: Satellite cell proliferative compartments in growing 13.14 Genes can alter the timing of senescence
skeletal muscles. Dev. Biol. 1996, 175: 84–94. Arantes-Oliviera, N., Berman, J.R., Kenyon, C.: Healthy animals
Williams, P.E., Goldspink, G.: Changes in sarcomere length and with extreme longevity. Science 2003, 302: 611.
physiological properties in immobilized muscle. J. Anat. 1978, Campisi, J., d’Adda di Fagagna, F.: Cellular senescence: when bad
127: 450–468. things happen to good cells. Nat. Rev. Mol. Cell Biol. 2007, 8:
729–740.
13.10 Cancer can result from mutations in genes that control Finkel, T., Serrano, M., Blasco, M.A.: The common biology of can-
cell multiplication and differentiation cer and ageing. Nature 2007, 448: 767–774.
Beachy, P.A., Karhadkar, S.S., Berman, D.M.: Tissue repair and stem Harper, M.E., Bevilacqua, L., Hagopian, K., Weindruch, R., Ramsey,
cell renewal in carcinogenesis. Nature 2004, 432: 324–331. J.J.: Ageing, oxidative stress, and mitochondrial uncoupling.
Hunter, T.: Oncoprotein networks. Cell 1997, 88: 333–346. Acta Physiol. Scand. 2004, 182: 321–331.
Jones, S., Zhang, X., Parsons, D.W., Lin, J.C., Leary, R.J., Angenendt, Kenyon, C.: The plasticity of aging: insights from long-lived
P., Mankoo, P., Carter, H., Kamiyama, H., Jimeno, A., et al.: Core mutants. Cell 2005, 120: 449–460.
signaling pathways in human pancreatic cancers revealed by Kipling, D., Davis, T., Ostler, E.L., Faragher, R.G.: What can prog-
global genomic analyses. Science 2008, 321: 1801–1806. eroid syndromes tell us about human aging? Science 2004, 305:
Shilo, B.Z.: Tumor suppressors. Dispatches from patched. Nature 1426–1431.
1996, 382: 115–116. Kudlow, B.A., Kennedy, B.K., Monnat, R.J.: Werner and Hutchin-
Taipale, J., Beachy, P.A.: The hedgehog and Wnt signalling path- son-Gilford progeria syndromes: mechanistic basis of progerial
ways in cancer. Nature 2001, 422: 349–353. diseases. Nat. Rev. Mol. Cell Biol. 2007, 8: 394–404.
Van Dyke, T.: p53 and tumor suppression. N. Engl. J. Med. 2007, Martinez, D.E.: Mortality patterns suggest lack of senescence in
356: 79–92. hydra. Exp. Gerontol. 1998, 33: 217–225.
Vernon, A.E., LaBonne, C.: Tumor metastasis: a new Twist on epithe- Murphy, C.T., Partridge, L., Gems, D.: Mechanisms of ageing: pub-
lial-mesenchymal transitions. Curr. Biol. 2004, 14: R719–R721. lic or private. Nat Rev. Genet. 2002, 3: 165–175.
Partridge, L.: Some highlights of research on aging with inverte-
13.11 Hormones control many features of plant growth brates. Aging Cell 2008, 7: 605–608.
Cosgrove, D.J.: Plant cell enlargement and the action of expansins. Tatar, M., Bartke, A., Antebi, A.: The endocrine regulation of aging
BioEssays 1996, 18: 533–540. by insulin-like signals. Science 2003, 299: 1346–1351.
Hauser, M.T., Morikami, A., Benfey, P.N.: Conditional root expan- Weindruch, R.: Caloric restriction and aging. Science 1996, 274:
sion mutants of Arabidopsis. Development 1995, 121: 1237–1252. 46–52.
Raven, P.H., Evert, R.F., Eichhorn, S.E.: The Biology of Plants Yoshida, K., Fujisawa, T., Hwang, J.S., Ikeo, K., Gojobori, T.: Degen-
(7th edn). New York: W. H. Freeman, 2005. eration after sexual differentiation in hydra and its relevance to
Sablowski, R.: Root development: the embryo within. Curr. Biol. the evolution of aging. Gene 2006, 385: 64–70.
2004, 14: R1054–R1055.
13.15 Cell senescence blocks cell multiplication
13.12 Arthropods have to molt in order to grow & 13.13 Shay, J.W., Wright, W.E.: When do telomeres matter? Science 2001,
Metamorphosis is under environmental and hormonal control 291: 839–840.
Brown, D.D., Cai, L.: Amphibian metamorphosis. Dev. Biol. 2007, Sherr, C.J., DePinho, R.A.: Cellular senescence: mitotic clock or
306: 20–33. culture shock. Cell 2000, 102: 407–410.
 14
Regeneration
nn Limb and organ nn Regeneration in Hydra
regeneration

Some animals, such as newts, can regenerate limbs, tails, and the eye lens. In the
limb, this involves dedifferentiation of differentiated cells to progenitor cells and
subsequent distal regrowth, with correct development guided by the generation of
new distal positional values. Hydra, a small cnidarian, regenerates its head without
growth and this involves gradients in inhibition and positional information set up by
the head.

Many of the cells in the adult body, such as heart muscle and most neurons of the
central nervous system, are the same cells that were originally generated during
embryonic development. But some tissues, such as blood and epithelia, are con-
tinually being replaced by stem cells, and others, such as skeletal muscle, can be
regenerated from quiescent stem cells if they are damaged (see, for example, Sec-
tion 10.10). We have also seen many examples of the capacity of the embryo to self-
regulate when parts of it are removed or rearranged (see, for example, Section 4.10).
Here we look at the related phenomenon of regeneration in adult organisms. Regen-
eration is the ability of the fully developed organism to replace tissues, organs, and
appendages by growth or repatterning of somatic tissue. Plants have remarkable
powers of regeneration: a single somatic plant cell can give rise to a complete new
plant (see Section 7.4). Some animals also show great ability to regenerate: small
fragments of animals such as starfish, planarians (flatworms), and Hydra can give
rise to a whole animal (Fig. 14.1). The ability of these animals to regenerate may

Planarian Hydra Starfish

Fig. 14.1 Regeneration in some

invertebrate animals. A planarian, Hydra,
and a starfish all show remarkable powers of
regeneration. When parts are removed or a
small fragment isolated, a whole animal can
be regenerated.
536 Chapter 14 Regeneration

be related to their ability to reproduce asexually; that is, to produce a complete new
individual by budding or fission.
1
Some insects and other arthropods can regenerate lost appendages, such as legs,
3 4 and among vertebrates, newts and other urodele amphibians show a remarkable
capacity for regeneration, being able to regenerate complete new tails and limbs, as
5
well as some internal tissues (Fig. 14.2). The lens in the newt eye, for example, regen-
erates from the pigmented epithelium of the iris, an example of transdifferentiation
2
(see Fig. 10.31). Another striking case of regeneration in vertebrates is the zebrafish,
which can regenerate the heart after removal of part of the ventricle. The regenerative
powers of mammals are much more restricted. Mammals cannot regenerate the heart,
for example, and so this ability in the genetically tractable zebrafish has implications
Fig. 14.2 The capacity for regeneration for medical research. The mammalian liver can regrow if a part of it is removed, as
in urodele amphibians. The emperor newt liver cells (hepatocytes) retain the ability to divide (see Section 13.4), the antlers of
can regenerate its dorsal crest (1), limbs
male deer regrow each year, and fractured bones mend by a regenerative process. But
(2), retina and lens (3 and 4), jaw (5), and tail
mammals cannot regenerate lost limbs, although they do have a limited capacity to
(not shown).
replace the ends of digits.
The issue of regeneration raises several major questions. Why are some animals
able to regenerate and others not? What is the origin of the cells that give rise to the
regenerated structures? What mechanisms pattern the regenerated tissue and how are
these related to the patterning processes that occur in embryonic development? The
Intact French flag system
lack of regenerative powers is not necessarily associated with increasing complexity:
X animals with no powers of regeneration at all include nematodes and rotifers. We will
mainly focus in this chapter on two systems in which regeneration has been inten-
X sively studied: regeneration of limbs in amphibians (newts and axolotls) and insects;
Positional
value and regeneration of a whole animal—the freshwater cnidarian Hydra. Understanding
regeneration in these systems could lead to progress in stimulating it in mammalian
cut
tissues and so help the development of medical ways of repairing tissues such as
the heart and the spinal cord. There is also the question of whether stem cells are
involved in regeneration, and so whether stem cells might be used to help repair dam-
Left half removed
aged tissues (see Chapter 10). We will also look briefly at regeneration of heart muscle
in zebrafish and the regeneration of peripheral nerves in mammals.
Regeneration requires, at the minimum, the production of a population of cells
whose growth, differentiation, and patterning is regulated. A distinction has been
drawn between two types of regeneration. In one—morphallaxis—there is little
new cell division and growth, and regeneration of structure occurs mainly by the
repatterning of existing tissue and the re-establishment of boundaries. Regeneration
in Hydra is a good example of morphallaxis. By contrast, regeneration of a newt
Morphallaxis Epimorphosis
limb depends on the growth of completely new, correctly patterned structures, and
X X this is known as epimorphosis. Both types of regeneration can be illustrated with
reference to the French flag pattern (Fig. 14.3). In morphallaxis, new boundary
X new X
positional regions are first established and new positional values are specified in relation to
value them; in epimorphosis, new positional values are linked to growth from the cut
growth surface. In both cases, the origin of the progenitor cells for the regenerated tissue
is a key issue.

Fig. 14.3 Morphallaxis and epimorphosis.

A pattern such as the French flag may be
Limb and organ regeneration
specified by a gradient in positional value (see
Fig. 1.25). If the system is cut in half, it can
Urodele amphibians, such as newts and axolotls, show a remarkable capacity for
regenerate in one of two ways. In regeneration
by morphallaxis, a new boundary is established
regenerating body structures such as tails, limbs, jaws, and the lens of the eye (see
at the cut and the positional values are Fig. 14.2). Regeneration of these structures involves cell proliferation and new growth,
changed throughout. In regeneration by and is therefore of the epimorphic type. Regeneration of a structure such as an adult
epimorphosis, new positional values are linked vertebrate limb, which contains a variety of fully differentiated cell types in a highly
to growth from the cut surface. organized arrangement, raises the question of the origin of the cells that give rise to
 Limb and organ regeneration 537

the regenerated structure. Do the differentiated cells dedifferentiate and start dividing?
Are there special reserve cells? Do existing cells change their character after dedif-
ferentiation? Damaged insect legs and other appendages can also regenerate. Insect
regeneration has been studied mainly in the larval cockroach, which has relatively
large legs that are easy to manipulate, and we shall also briefly discuss this system.

14.1 Amphibian limb regeneration involves cell dedifferentiation and

new growth

Amputation of a newt limb is followed by a rapid migration of epidermal cells from

the edges of the wound to form a covering over the wound surface. This process
is essential for subsequent regeneration; if it is prevented by suturing the edges of
the stump together, there is no regeneration. A mass of undifferentiated cells called
the blastema then forms under the epidermal cap and this is what gives rise to the
regenerated limb (Fig. 14.4). The blastema is formed from cells beneath the wound

Distal amputation Proximal amputation

Original limbs

After amputation

blastema

After 7 days

After 21 days

After 25 days

After 32 days

Fig. 14.4 Regeneration of the forelimb

in the red-spotted newt Notophthalmus
After 42 days viridescens. The left panel shows the
regeneration of a forelimb after amputation
at a distal (mid-radius/ulna) site. The right
panel shows regeneration after amputation
at a proximal (mid-humerus) site. At the top,
the limbs are shown before amputation.
Successive photographs were taken at the
After 70 days times shown after amputation. Note that the
blastema gives rise to structures distal to the
cut. Scale bar = 1 mm.
538 Chapter 14 Regeneration

epidermis that lose their differentiated character and start to divide, eventually form-
ing an elongated cone. As the limb regenerates over a period of weeks, the blastema
cells differentiate into cartilage, muscle, and connective tissue. The blastema cells are
derived locally from the mesenchymal tissues of the stump, close to the site of ampu-
tation. They come in particular from the dermis, but have also been shown to derive
from cartilage and muscle.
The ability of newt skeletal muscle cells to dedifferentiate and return to a proliferat-
ing state is particularly intriguing, as vertebrate skeletal muscle cells do not normally
divide after becoming fully differentiated. A general feature of vertebrate muscle dif-
ferentiation is the withdrawal of muscle precursor cells from the cell cycle after myo-
blast fusion to produce myotubes. This withdrawal involves the dephosphorylation
of the cell-cycle control protein Rb (retinoblastoma protein) (see Section 10.9), and
cultured mouse muscle cells lacking Rb can re-enter the cell cycle. In the former
muscle cells in the regenerating newt limb, Rb protein is inactivated by phosphoryla-
tion, allowing the cells to re-enter the cell cycle and divide. Re-entry into the cell cycle
is also associated with the activation of thrombin in the blastema. Indeed, thrombin
could be a key factor in regeneration in a variety of systems, as the regeneration of
the newt lens from the dorsal margin of the iris also correlates with thrombin activity
in that region.
Thrombin is a proteolytic enzyme that is more familiar as part of the blood-
clotting cascade, but appears also to be involved in providing an environment for
dedifferentiation. Multinucleate post-mitotic newt muscle cells can be converted
to dividing mononuclear cells in culture in the presence of thrombin in the culture
medium. Dedifferentiation of the muscle cells also involves renewed expression
of the homeodomain transcription factor Msx1. This multifunctional transcrip-
tional factor is known to prevent myogenic differentiation in mammalian cells (see
Section 10.9), and its expression is characteristic of undifferentiated mesenchy-
mal cells that can undergo regeneration. The question of the origin of regenerated
muscle in newts was further complicated when cells functionally similar to the
satellite stem cells that regenerate skeletal muscle fibers in adult mammals (see
Section 10.10) were identified in adult newts, which had previously been thought
to lack such cells.
The next question is whether the cells differentiating into cartilage and muscle in
the regenerating newt limb are remaining true to type, or whether muscle satellite
cells or dedifferentiated muscle cells can re-differentiate into another type of cell
entirely—the process of transdifferentiation (see Section 10.15). Can dedifferentiated
skeletal muscle cells in the stump re-differentiate as cartilage, for example? This ques-
tion was originally addressed by experiments in which cultured newt limb muscle
myotubes, which are multinucleate and have stopped dividing, were labeled in cul-
ture with a retrovirus expressing alkaline phosphatase, and then introduced into the
blastema of a regenerating limb. Strongly labeled mononuclear cells were observed
in the blastema a week after, and the majority of the introduced myotubes gave rise
to mononuclear cells. After proliferating, the labeled mononuclear cells did appear to
give rise to cartilage, as well as to new muscle, suggesting that transdifferentation had
occurred. Altogether, these observations on regenerating newt limbs suggested that
the blastema cells were likely to have reverted to a homogeneous multipotent stem-
cell state, from which they re-differentiated anew.
However, more recent experiments in the regenerating limb of the axolotl Ambys-
toma mexicanum give a very different picture. These experiments showed that the
blastema cells did not revert to a multipotent state but retained a restricted develop-
mental potential related to their origin. The fate of individual tissue types in a regener-
ating limb was traced by an ingenious technique. The first step was to produce trans-
genic axolotls that expressed green fluorescent protein (GFP) in all their cells. Next,
a patch of tissue of a particular type, such as muscle or epidermis, was transplanted
 Limb and organ regeneration 539

Cell fate?
Positional identity?

Graft Amputation Blastema Regenerate

from these animals into the forelimbs of non-transgenic juveniles of the same species. Fig. 14.5 Cells that regenerate the
The forelimb of the juvenile axolotl was amputated across the site of the transplant, axolotl limb have restricted developmental
potential. The experimental procedure is
and the fate of the glowing green transplanted cells could then be traced as the limb
shown in the top row and photographs of
regenerated (Fig. 14.5). The tissue type of the fluorescent cells after re-differentiation
the limb at various stages are shown in the
was then identified by testing for tissue-specific markers. bottom row. Cells of a particular tissue type
These experiments showed that transplanted muscle tissue produced blastema (e.g. dermis as shown here) and position in
cells that could only give rise to muscle. The experimental technique could not, the limb of a juvenile axolotl transgenic for
however, resolve whether the regenerated muscle derived from satellite cells or from green fluorescent protein, were transplanted
dedifferentiation of muscle cells or from both, as both cell types would have been into the limb of a non-transgenic animal
present in the transplanted tissue. Cartilage and epidermal cells were also restricted and the limb amputated across the graft.
in their further development, as were the glial Schwann cells, which provide the In the regenerating limb that grew from
myelin coating of ­peripheral nerve fibers. One exception was the cells of the dermis, the blastema, the transplanted cells and
which could contribute to both new dermis and new cartilage skeleton, and the their progeny can be detected by their
green fluorescence. These cells could then
potential of fibroblasts to re-differentiate remains unclear. This investigation also
be tested for their tissue type. In the case
found that cartilage-derived blastema cells retained a ‘memory’ of their original posi-
illustrated, the transplanted dermis cells
tional identity along the proximo-distal axis of the limb. Cartilage cells transplanted could give rise to new dermis and to the
from a distal position in the original limb would not give rise to more proximal cartilage skeleton, but not muscle. When
regions in the regenerating limb. muscle cells were transplanted, they could
Growth of the blastema is dependent both on its nerve supply (Fig. 14.6) and on the only give rise to muscle. Cells taken from a
overlying wound epidermis, which may play a role similar to the apical ectodermal proximal position in the limb could give rise
ridge in limb development (see Section 11.3). In amphibian limbs in which the nerves to distal structures (as seen here by the
have been cut before amputation, a blastema forms but fails to grow. The nerves have labeled cells in the digits), but transplanted
no influence on the character or pattern of the regenerated structure; it is the amount cartilage cells would not give rise to
more-proximal structures.
of innervation, not the type of nerve that matters. In regenerating adult salamander
limbs, the nerves have been shown to provide an essential growth factor called ante- Photographs from Kragl, M., et al.: 2009.
rior gradient protein (nAG), which is secreted first by the glial Schwann cells of the
incoming nerves and later by the glandular cells of the wound epidermis. Treatment
of the blastema of a denervated limb with nAG is sufficient to enable it to regenerate
completely.
A striking instance of the influence of nerves on regeneration of amphibian limbs
is that if a major peripheral nerve, such as the sciatic nerve, is cut and the branch
inserted into a wound on a limb or on the surface of the adjacent flank, a supernu-
merary limb develops at that site. This experimental system provides an opportunity
to study limb regeneration in the absence of the considerable cell damage caused by
amputation. It was found, for example, that expression of the zinc finger transcription
factor Sp9 by epidermal keratinocytes in the region of the wound was necessary for
their dedifferentiation into an epithelium that could support limb regeneration.
540 Chapter 14 Regeneration

Normal limb Denervated limb Aneurogenic limb

nerve

Amputate limb

Fig. 14.6 Innervation and limb

regeneration. Normal limbs require a nerve
supply to regenerate (left panels). Limbs
denervated prior to amputation will not
regenerate (center panels). However, limbs
that have never been innervated, because
the nerve was removed during development,
can regenerate normally in the absence of
Regeneration No regeneration Regeneration
innervation (right panels).

An interesting phenomenon, as yet unexplained, is that if embryonic newt limbs

are denervated very early in their development, and so do not become exposed to the
influence of nerves, they can regenerate in the complete absence of any nerve sup-
ply (see Fig. 14.6, right panel). If an aneurogenic limb is subsequently innervated,
however, it rapidly becomes dependent on nerves for regeneration. This suggests that
dependence on innervation is imposed on the limb only after the nerves grow into it.
The Xenopus laevis tadpole can regenerate amputated limbs but, after metamorpho-
sis, the young frog cannot, and if a limb is amputated, just a spike of tissue develops.
Experiments investigating the epigenetic regulation of key developmental genes in the
regenerating tissue have found that the ZRS (zone of polarizing activity regulatory
­sequence) of the Sonic hedgehog gene (Shh) is highly methylated in the froglet but not
in the tadpole (see Box 11B, p. 423). The DNA methylation presumably leads to silenc-
ing of Shh expression via this enhancer in the froglet, and points to Shh as a key player
in vertebrate limb regeneration, as it is in normal limb development (see Chapter 11).
Although mammals cannot regenerate whole limbs, many mammals, including
young children, can regenerate the tips of their digits, provided the nail-generating
tissue—the nail organ—is still present. In mice and children, the level from which
digits are able to regenerate is limited to the base of the claw or nail, respectively.
This probably reflects the presence of connective tissue cells under the nail organ that
express the transcription factor Msx1, which regulates BMP-4 expression and is asso-
ciated with undifferentiated mesenchymal cells, as noted earlier. In mammalian limb
development, Msx1 is expressed at the tip of the embryonic limb bud, and in mice it
continues to be expressed in the tips of the digits even after birth. The region in which
digit regeneration in mice can occur corresponds with the expression of Msx1.

14.2 The limb blastema gives rise to structures with positional values distal
to the site of amputation

Although proteins active in normal limb development also play a part in limb regenera-
tion, regeneration is unlikely to occur by precisely the same mechanisms as in embry-
onic development. Regeneration always proceeds in a direction distal to the cut surface,
 Limb and organ regeneration 541

Limb is inserted into belly, vascular Regeneration starts from proximal Both proximal and distal surfaces
Hand amputated connections are established, and distal surfaces of humerus regenerate distal structures
then humerus is cut

Fig. 14.7 Limb regeneration is always

enabling replacement of the lost part. If the hand is amputated at the wrist, only the car- in the distal direction. The distal end of
pals and digits are regenerated, whereas if the limb is amputated through the middle of a limb is amputated and the limb inserted
the humerus, everything distal to the cut (including the distal humerus) is regenerated. into the belly. Once vascular connections
Positional value along the proximo-distal axis is therefore of great importance, and is at are established, a cut is made through the
least partly retained in the blastema (see Section 14.1). The blastema has considerable humerus. Both cut surfaces regenerate the
morphogenetic autonomy. If it is transplanted to a neutral location that permits growth, same distal structures even though, in the
such as the dorsal crest of a newt larva or even the anterior chamber of the eye, it gives case of one of the regenerating limbs, distal
rise to a regenerated structure appropriate to the position from which it was taken. structures are already present.
The growth of the blastema and the nature of the structures it gives rise to depend
on the site of the amputation and not on the nature of the more proximal tissues. The
Donor and host limbs amputated
limb is not, however, simply ‘trying’ to replace missing parts. This was shown in a and blastemas form
classic experiment in which the distal end of a newt limb that had been amputated
at the wrist was inserted into the belly of the same animal, so as to establish a blood Proximal Distal
supply to it. The limb was then cut mid-humerus. Both surfaces regenerated distally, blastema blastema
even though the part attached to the belly already had a radius and ulna (Fig. 14.7).
The blastema is much larger than the embryonic limb bud—it can be ten times larger
in terms of cell numbers. This size makes it most unlikely that there are signals that
diffuse right across the blastema. Instead, regeneration can best be understood in terms
of an adult limb having a set of positional values along its proximo-distal axis, which
have been set up during embryonic development (see Section 11.5). The regenerating
limb in some way reads the positional value at the site of the amputation and then
regenerates all positional values distal to it. Epimorphic regeneration also involves the
retention of embryonic processes, such as the ability to specify new positional values. Proximal blastema removed,
The ability of cells to recognize a discontinuity in positional values is illustrated by distal blastema grafted to proximal stump
grafting a distal blastema to a proximal stump. In this experiment, the forelimb’s stump
and blastema have different positional values, corresponding to shoulder and wrist,
respectively. The result is a normal limb in which structures between the shoulder and
wrist have been generated by intercalary growth, predominantly from the proximal
stump, while the cells from the wrist blastema mostly give rise to the hand (Fig. 14.8).
A fundamental question in any discussion of pattern formation is the molecular basis All proximal structures in regenerated limb
of the proposed positional information. A major advance on this front has been the mostly come from proximal stump tissue
identification of a cell-surface protein that is expressed in a graded manner along the
proximo-distal axis of the newt limb. Proximo-distal identity in the blastema can be
altered by treatment with retinoic acid, as discussed in more detail in the next section,
and this property was used to compare gene expression in blastemas whose proximo-
distal identity had been altered. An almost twofold difference in the level of expression of
the gene for a cell-surface protein called Prod1 (the newt equivalent of the ­mammalian

Fig. 14.8 Proximo-distal intercalation in limb regeneration. A distal blastema grafted to

a proximal stump results in intercalation of all the structures proximal to the distal blastema.
Almost all of the intercalated region comes from the proximal stump.
542 Chapter 14 Regeneration

distal cell-surface protein CD59), was found between ‘proximal’ and ‘distal’ blastemas. Prod1
blastema might therefore be used by the cells to determine positional identity by short-range cell–
cell interactions. The growth factor nAG, which is essential for limb regeneration (see
distal Section 14.1), is a ligand for Prod1. It is thought that nAG has no input into generating
blastema
proximal from cut wrist proximo-distal pattern, but acts through Prod1 to promote cell division.
blastema
The gradient in Prod1 fits very well with the observation that cell-surface properties
are involved in regeneration. When mesenchyme from two blastemas from different
proximo-distal sites are confronted in culture, the more proximal mesenchyme engulfs
the distal (Fig. 14.9, left panels), whereas mesenchyme from two blastemas from
similar sites maintains a stable boundary. This behavior is suggestive of a graded dif-
ference in cell adhesiveness along the axis, with adhesiveness being highest distally.
The cells in the distal explant remain more tightly bound to each other than do the
cells from the proximal blastema, which therefore spread to a greater extent. A dif-
ference in adhesiveness is also suggested by the behavior of a distal blastema when
grafted to the dorsal surface of a proximal blastema so that their mesenchymal cells
Fig. 14.9 Cell-surface properties vary
along the proximo-distal axis. Left panels:
are in contact. Under these conditions, the distal blastema moves during limb regen-
when mesenchyme from distal and proximal eration to end up at the site from which it originated (see Fig. 14.9, right panels). This
blastemas is placed in contact in culture, the suggests that its cells adhere more strongly to the regenerated wrist region than they
proximal mesenchyme engulfs the distal do to the proximal region to which it was transplanted. Transplantation of a shoulder-
mesenchyme, which has greater adhesion level blastema to a shoulder stump does not mobilize the stump tissue, but leads to
between its cells. Right panels: if a distal a normal distal outgrowth from the shoulder blastema. All these experiments suggest
blastema (in this case from a cut wrist) that proximo-distal positional values in urodele limb regeneration are encoded as a
is grafted to the dorsal surface of a more graded property, probably in part at the cell surface, and that cell behavior—growth,
proximal blastema, the regenerating wrist movement, and adhesion—relevant to axial specification is a function of the expres-
blastema will move distally to a position on
sion of this property, relative to neighboring cells.
the host limb that corresponds to its original
Maintaining the continuity of positional values by intercalation is a fundamental
level and regenerates a hand.
property of regenerating epimorphic systems and later in the chapter we will consider
it in relation to the cockroach leg. Even normal regeneration by outgrowth from a
blastema could be considered the result of intercalation between the cells at the level
of amputation and those with the most distal positional values, as specified by the
wound epidermis. It is not clear to what extent all blastemal cells inherit a particular
positional value from their differentiated precursors, and to what extent they are sub-
ject to signals that induce the appropriate expression of positional value. In the axolotl
experiment outlined in Fig. 14.5, cartilage cells retained their original positional iden-
tity whereas Schwann cells did not.
The precise relationship between Hox gene expression and positional identity is
not understood, either for limb embryonic development or regeneration. During limb
development in the embryo, the Hoxa genes are expressed with temporal and spatial
co-linearity along the proximo-distal axis, as we saw in Chapter 11. In the regenerat-
ing axolotl limb, however, two Hoxa genes from the 3′ and 5′ ends of the complex are
expressed in the stump cells at the same time—24 to 48 hours after amputation. This
suggests that the most distal region of the blastema is specified first and that regen-
eration involves intercalation of positional values between this distal region and the
stump. Grafting experiments also show that distal identity is established early, and
that by 4 days the blastema is already divided into distinct zones that will give rise to
the proximo-distal regions of the limb.

14.3 Retinoic acid can change proximo-distal positional values in

regenerating limbs

We have already seen that retinoic acid is present in developing vertebrate limbs
and how experimental treatment with retinoic acid can alter positional values in the
developing chick limb (see Section 11.7). It also has striking effects on regenerating
amphibian limbs.
 Limb and organ regeneration 543

Fig. 14.10 Retinoic acid can proximalize

positional values. A forelimb amputated at
the level of the hand, as indicated by the
dotted line, and then treated with retinoic
acid, regenerates structures corresponding
to a cut at the proximal end of the humerus.
Scale bar = 1 mm.

Exposing a regenerating limb to retinoic acid results in the blastema becoming proxi-
malized; that is, the limb regenerates as if it had originally been amputated at a more
proximal site. For example, if a limb is amputated through the radius and ulna, treat-
ment with retinoic acid will result not only in the regeneration of the elements distal to
the cut, but also in the production of an extra complete radius and ulna. The effect of
retinoic acid is dose-dependent, and with a high dose it is possible to regenerate a whole
extra limb, including part of the shoulder girdle, on a limb from which only the hand has
been amputated (Fig. 14.10). Retinoic acid, therefore, alters the proximo-distal positional
value of the blastema, making it more proximal. It probably does this by affecting vari-
ous pathways for specifying positional information, in particular increasing the expres-
sion of Prod1 (see Section 14.2), as overexpression of Prod1 in distal blastema cells also
causes them to translocate to more proximal positions. Retinoic acid also appears to shift
positional values in the blastema in a proximal direction through the activation of meis
homeobox genes, which are involved in specifying proximal identity and are normally
repressed in distal regions. Retinoic acid can also, under some experimental conditions,
shift positional values along the antero-posterior axis in a posterior direction.
Retinoic acid can act through a variety of nuclear receptors, but only one (RAR δ2)
is involved in changes in proximo-distal positional value in the limb. By construct-
ing a chimeric receptor from a RAR δ2 DNA-binding portion and a thyroxine recep-
tor ­hormone-binding portion, the retinoic acid receptor can be selectively activated
by thyroxine, which enables the effects of its activation to be studied experimentally
(Fig. 14.11). Cells in the distal blastema are first transfected with the chimeric receptor
gene and then grafted to a proximal stump and treated with thyroxine. The transfected

Newt retinoic acid Distal blastema Blastema containing

receptor and Xenopus transfected by transfected cells grafted Transfected cells
thyroxine receptor chimeric receptor to proximal stump move proximally

retinoic chimeric Thyroxine

N C acid receptor Fig. 14.11 Retinoic acid proximalizes the
receptor added
(�1) positional value of individual cells. Some
of the cells of a newt distal blastema are
N C thyroxine
receptor transfected by a chimeric receptor, through
which retinoic acid receptor function can
be activated by thyroxine. This blastema is
grafted to a proximal stump and treated with
thyroxine. During intercalary growth, the
transfected cells, which have been labeled,
move proximally because their positional
values have been proximalized by the
activation of retinoic acid receptor function.
The photographs illustrate proximalization of
the transfected cells. Scale bar = 0.5 mm.
Photographs from Pecorino, L.T., et al.: 1996.
544 Chapter 14 Regeneration

Fig. 14.12 Retinoic acid can induce

additional limbs in the regenerating tail
of a frog tadpole. After amputation, the
regenerating tail stump of a tadpole of
Rana temporaria is treated with retinoic
acid at the time when hindlimbs are normal limb
developing. This results in the appearance
of additional hindlimbs in place of a new tail.
Scale bar = 5 mm.
Photograph courtesy of M. Maden.

cells behave as if they have been treated with retinoic acid. The result is a movement
of the transfected blastemal cells to more proximal regions in the intercalating regen-
erating blastema, which shows that activation of the retinoic-acid pathway can proxi-
malize the positional values of the cells, which then respond by movement to more
proximal sites. Prod1 is a very strong candidate for mediating this change, as it too can
cause cells to move proximally when its concentration is increased.
Another remarkable effect of retinoic acid is its ability to bring about a homeotic
transformation of tails into limbs in tadpoles of the frog Rana temporaria. If the tail of a
tadpole is removed, it will regenerate. But treatment of regenerating tails with ­retinoic
acid at the same time as the hindlimbs are developing results in the appearance of
additional hindlimbs in place of a regenerated tail (Fig. 14.12). There is, as yet, no
satisfactory explanation for this result, but it has been speculated that the retinoic acid
alters the antero-posterior positional value of the regenerating tail blastema to that of
the site along the antero-posterior axis where hindlimbs would normally develop.

14.4 Insect limbs intercalate positional values by both proximo-distal and

circumferential growth

The legs of some insects, such as the cockroach and cricket, can regenerate. Unlike
Drosophila, these insects do not undergo a complete metamorphosis, and the juvenile
stages (nymphs) are more similar to miniature adults. When the tibia of a third-instar
nymph of the cricket Gryllus bimaculatus is amputated, it takes about 40 days to leg
restore the adult. Wingless and Decapentaple­gic signaling activate epidermal growth
factor receptor (EGFR) signaling in a distal-to-proximal gradient in the cricket blast-
ema, which patterns the regenerating distal leg.
In Drosophila, as we saw in Chapter 13, signaling through the cadherins Fat and
Dachsous and the associated Hippo pathway coordinates cell proliferation and apopto-
sis in imaginal discs, determining their final size and thus the size of adult wings and
legs (see Section 13.6). Experiments using RNA interference to investigate the role of
Fat, Dachsous, and the Hippo pathway in cricket limb regeneration provide evidence
that this seems also to be the case for an insect whose legs develop directly. Knockdown
of either Fat or Dachsous resulted in a shorter than normal regenerated limb, whereas
knockdown of another set of genes in the Hippo pathway produced a longer leg than
normal. It is proposed that a Dachsous/Fat gradient exists along the limb and that the
steepness of the gradient could control cell proliferation, reminiscent of the gradient
model for determining organ size suggested for Drosophila (see Box 13A, p. 514).
 Limb and organ regeneration 545

Regenerating insect legs follow an epimorphic process of blastema formation and

outgrowth that intercalates missing positional values, as described for the proximo-
distal axis of the amphibian limb (see Section 14.2), even though the structure of the
insect leg is different in many ways from those of vertebrates, especially in having an
ectodermally derived exoskeleton. Intercalation of positional values, therefore, seems
to be a general property of epimorphically regenerating systems. When cells with dis-
parate positional values are placed next to one another, intercalary growth occurs in
order to regenerate the missing positional values. Intercalation is particularly clearly
illustrated by limb regeneration in the cockroach, and also occurs in regenerating
Drosophila leg and wing imaginal discs after amputation.
A cockroach leg is made up of a number of distinct segments, arranged along the
proximo-distal axis in the order: coxa, femur, tibia, tarsus. Each segment seems to con-
tain a similar set of proximo-distal and circumferential positional values, and will inter-
calate missing positional values during regrowth. When a distally amputated tibia is
grafted onto a host tibia that has been cut at a more proximal site, localized growth
occurs at the junction between graft and host, and the missing central regions of the
tibia are intercalated (Fig. 14.13, left panels). In contrast to amphibian regeneration,
there is a predominant contribution from the distal piece. As in the amphibian, however,
regeneration is a local phenomenon and the cells are indifferent to the overall pattern of
the tibia. Thus, when a proximally cut tibia is grafted onto a more distal site, making an
abnormally long tibia, regenerative intercalation again restores the missing positional
values, making the tibia even longer (Fig. 14.13, right panels). The regenerated portion
is in the reverse orientation to the rest of the limb, as indicated by the direction in which
the bristles point, suggesting that the gradient in positional values also specifies cell
polarity, as in insect body segments (see ­Chapter 2). These results also show that when

Donor and host limbs amputated

1 2 3 4 5 1 2 3 4 5

positional value gradient

1 2 3 4 5 1 2 3 4 5

Fig. 14.13 Intercalation of positional

values by growth in the regenerating
cockroach leg. Left panels: when a distally
amputated tibia (5) is grafted to a proximally
Limb pieces grafted
amputated host (1), intercalation of the
1 5 1 2 3 4 2 3 4 5 positional values 2–4 occurs, irrespective of
the proximo-distal orientation of the grafts,
and a normal tibia is regenerated. Right
panels: when a proximally amputated tibia
(1) is grafted to a distally amputated host
(4), however, the regenerated tibia is longer
than normal and the regenerated portion is in
Intercalation
the reverse orientation to normal, as judged
by the orientation of surface bristles. The
1 2 3 4 5 1 2 3 4 4 3 2 2 3 4 5
reversed orientation of regeneration is due
to the reversal in positional value gradient.
The proposed gradient in positional value is
shown under each figure.
After French, V., et al.: 1976.
546 Chapter 14 Regeneration

Removal of strip of cockroach leg Cut edges heal together Intercalary regeneration
Numbers indicate arbitrary
positional values
around the
circumference 12/0
12/0 12/0
11 1 11 1 11 1
2 10 2 2
Fig. 14.14 Circumferential intercalation 10 10
in the cockroach leg. The leg is seen in
transverse section. When a piece of cockroach 9
9 3 3 9 3
ventral epidermis is removed (left panel),
the cut edges heal together (center panel).
When the insect molts and the cuticle
regrows, circumferential positional values 4 8 4
8 64 8
are intercalated (right panel). The positional longitudinal 7
7 6 5 strip of 7 6 5
values are arranged around the circumference epidermis
removed
of the leg, rather like the hours on a clock face. intercalated
region
After French, V., et al.: 1976.

cells with non-adjacent positional values are placed next to each other, the missing val-
ues are intercalated by growth to provide a set of continuous positional values.
A similar set of positional values is present in each segment of the limb. Thus, a
mid-tibia amputation, when grafted to the mid-femur of a host, will heal without
intercalation. But grafting a distally amputated femur onto a proximally amputated
host tibia results in intercalation, largely femur in type. There must be other factors
making each segment different, rather like the segments of the insect larva.
Intercalary regeneration also occurs in a circumferential direction. When a longitu-
dinal strip of epidermis is removed from the leg of a cockroach, normally non-adjacent
cells come into contact with one another, and intercalation in a circumferential direc-
tion occurs after molting (Fig. 14.14). Cell division occurs preferentially at sites of mis-
match around the circumference. One can treat positional values in the circumferential
direction as a clock face, with values going continuously 12, 1, 2, 3 … 6 … 9 … 11. As
in the proximo-distal axis, there is intercalation of the missing positional values.

14.5 Heart regeneration in the zebrafish involves the resumption of cell

division by cardiomyocytes

One tissue that does not renew itself in mammals is the cardiac muscle or myo-
cardium. Thus, damaged hearts do not repair themselves like other muscles do. If
mammalian heart muscle is damaged, it forms a fibrous scar. However, the cells that
make up heart muscle—the cardiomyocytes—are, in principle, capable of resuming
cell division, and some vertebrates do have the ability to regenerate the myocardium.
If 20% of the ventricle of the adult zebrafish heart is removed, it will regrow. This
phenomenon is also seen in newts, but is of particular interest in the zebrafish as in
this model organism the regeneration process can be studied genetically.
Unlike blastema formation in amphibian limb regeneration, zebrafish heart muscle
regeneration seems not to involve dedifferentiation of the cardiomyocytes. Instead,
cardiomyocytes nearest to the wound surface start to divide and the new myocytes
are displaced inwards to regenerate the ventricle wall. An analysis of gene expres-
sion during heart regeneration in zebrafish showed that genes were expressed that
are also involved in the normal development of cardiac muscle, such as Nkx2.5, an
early marker of prospective heart cells (see Section 11.30). This gene was expressed
only at a low level in the regenerating heart muscle, however, suggesting that regen-
eration was not using the same processes as embryonic heart development, which
involves differentiation of cardiomyocytes from undifferentiated precursors. In addi-
tion, ­expression of the zebrafish homeodomain-containing transcription factors MsxB
 Limb and organ regeneration 547

and MsxC (related to Msx1) was increased during regeneration, whereas these factors
are not expressed during embryonic heart development.

14.6 The mammalian peripheral nervous system can regenerate

Complete regeneration of major organs is more limited in adult mammals than in fish
or amphibians. One organ that can regrow functional tissue is the liver, as we saw in
Chapter 13. Another mammalian system with considerable powers of regeneration in
adults is the peripheral nervous system. But, unlike in the liver, the regeneration of
peripheral nerves involves the regrowth of axons, and not the replacement of the cells
themselves by cell division. If a neuronal cell body is destroyed, the neuron cannot
regrow and cannot be replaced.
Axons of peripheral neurons in adult vertebrates, such as the motor and sensory axons
running between the spinal cord and the ends of the limbs, can be up to hundreds of cen-
timeters long. When such an axon is severed, a new growth cone forms at the cut surface
and grows down the pathway of the original nerve trunk to make functional connections,
leading to an almost complete recovery of function. In the case of motor neurons, the
axon terminal finds the site of the original synapse on the muscle cell by recognizing the
basal lamina that fills the synaptic cleft. Axon regrowth is promoted by the presence of
the glial Schwann cells, which form the myelin covering peripheral neurons.
In contrast, the central nervous system of adult birds and mammals is not able to
regenerate correct axonal and dendritic connections after damage to neurons. This
is not due to an intrinsic deficit in the neurons themselves but rather is due to the
damaged environment, which does not support, and even actively prevents, axon
regeneration. Although the remaining cell body produces new outgrowths, these do
not develop further. It seems that the failure of axons to regenerate is partly the result
of inhibition by the glial cells of the central nervous system. Culturing immature
neurons with oligodendrocytes, glial cells that are the source of myelin in the central
nervous system, can induce growth-cone collapse, and astrocytes, another type of
central nervous system glial cell, can be equally inhibitory in damaged grey matter.
By contrast, when Schwann cells, the myelin-producing glial cells of the peripheral
nervous system, are implanted into the central nervous system, they can still pro-
mote axon regrowth, with the regenerating axons migrating along the Schwann cells.
Another reason for the lack of regrowth in a damaged central nervous system is that
debris is cleared much more slowly than in the peripheral nervous system. It is not
difficult to think of reasons why new axonal sprouting in the mammalian central
nervous system, with its highly complex architecture, is generally inhibited. If axons
were allowed to sprout unchecked, that would be very likely to disrupt the elaborate
pattern of neuronal connections and destroy brain function.
Stem cells have great promise as sources for introducing new neurons and glia
into damaged regions of the central nervous system. Embryonic stem cells (see
Section 10.16), for example, can differentiate into neurons when transplanted into
the mammalian brain. While such experiments are encouraging, it still needs to be
shown that this approach can restore normal function after damage.

Summary

Urodele amphibians can regenerate amputated limbs and tails. Stump tissue at the site
of amputation first dedifferentiates to form a blastema, which then grows and gives rise
to a regenerated structure. The dedifferentiated cells of the blastema are the progenitors
of the various cell types within the regenerated limbs. Regeneration is usually depen-
dent on the presence of nerves, but limbs that have never been innervated can regener-
ate. Regeneration always gives rise to structures with positional values more distal than
548 Chapter 14 Regeneration

those at the site of amputation. When a blastema is grafted to a stump with different
positional values, proximo-distal intercalation of the missing positional values occurs.
Positional values may be related to a proximo-distal gradient in a cell-surface protein.
Retinoic acid changes the positional values of the cells of the blastema, giving them more
proximal values. Insect limbs can also regenerate, with intercalation of missing positional
values occurring in both proximo-distal and circumferential directions. Among vertebrate
internal organs, liver regenerates in mammals but heart muscle does not, although it can
do so in amphibians and fish. Central nervous system neurons do not regenerate in adult
mammals and birds after damage, but peripheral nerves can regrow axons and reform
connections, although if the cell body is damaged the neuron cannot be replaced.

Summary: regeneration of an amphibian limb

amputation of a newt limb

local dedifferentiation of stump tissue to form a blastema

the blastema grows to form distal structures

provided nerves are present

graft of distal blastema to proximal stump

intercalation of missing proximo-distal positional values from stump tissue growth

Regeneration in Hydra

Hydra is a freshwater cnidarian consisting of a hollow tubular body about 0.5 cm

long, with a head region at one end (the distal end) and a basal region at the other
(proximal) end, by which it sticks to a surface (Fig. 14.15). The head consists of a
small conical hypostome where the mouth opens, surrounded by a set of tentacles
that are used for catching the small animals on which Hydra feeds. Unlike most of the
animals discussed in this book, which have three germ layers, Hydra has only two. The
body wall is composed of an outer epithelium, which corresponds to the ectoderm,
and an inner epithelium lining the gut cavity, which corresponds to the endoderm.
These two layers are separated by a gel-like basement membrane or mesoglea. There
are about 20 different cell types in Hydra, which include highly specialized types such
as neurons, secretory cells, and the nematocytes that are used to capture prey. Stem
cells called interstitial cells give rise to these specialized cells in normal growth. Hydra
has no central nervous system but has a network of neurons distributed throughout the
body. The reason for looking at patterning and regeneration in Hydra is that it provides
insight into an organizer region and developmental gradients that arose early in the
evolution of animal development. It is likely that the more complex body patterns of
other animals evolved from a simple body plan like that of Hydra.

14.7 Hydra grows continuously but regeneration does not require growth

Well-fed Hydra are in a dynamic state of continuous growth and pattern formation. Asex-
ual reproduction is by budding, but in hard times Hydra can reproduce sexually. The cells
of both epithelial layers proliferate steadily and, as the tissues grow, cells are displaced
 Regeneration in Hydra 549

mouth
tentacles H (hypostome)

tentacle
head
outer epithelium
1 (ectoderm) body column
2 inner epithelium
gastric (endoderm)
region 3
4
budding B
region bud
5
6
basal disc basal disc

Fig. 14.15 Hydra. This freshwater cnidarian

along the body column toward the head or foot. In order for an adult Hydra to maintain has a head with tentacles and mouth at
a constant size, excess cells must be continually lost. Cell loss occurs at the tips of the one end and a sticky foot at the other
tentacles and at the basal disc of the foot bud. Most of the excess cell production is taken (photograph, left panel). It can reproduce
by budding. For the purposes of grafting
up by the asexual budding of new Hydra from the body column. Budding occurs about
experiments, the body column is divided
two-thirds of the way down the body (see Fig. 14.15); the body wall evaginates by a mor-
into a series of regions, as indicated in the
phogenetic change in cell shape, generated locally by the cells in the bud region, to form
center panel. The body wall is made up of
a new column that develops a head at the end and then detaches as a small new Hydra. two epithelial layers, corresponding to the
Continuous growth in Hydra means that cells are continually changing their relative ectoderm and endoderm found in other
positions and are forming new structures as they move up or down the body column. organisms featured in this book (right panel).
There must, therefore, be mechanisms for repatterning cells during this dynamic pro- Scale bar = 1 mm.
cess. It is these mechanisms that give Hydra its remarkable capacity for regeneration. Photograph courtesy of W. Müller from Müller,
If the body column of a Hydra is cut transversely, the lower piece will regenerate a W.A.: 1989.
head and the upper piece will regenerate a foot. Thus, what structure the cells regen-
erate at a cut surface depends on their relative position within the regenerating piece.
The cut surface nearest the original head end forms a head—this shows that Hydra
has a well-defined overall polarity. This polarity is maintained even in small pieces of
the body and this can be seen when a short piece is cut out of the body column: the
distal end regenerates a head while the proximal end becomes the basal disc.
Regeneration in Hydra does not require cell division and new growth and is there-
fore an example of morphallactic regeneration. When a short fragment of the column
regenerates, there is no initial increase in size and the regenerated animal will be a
small Hydra. Only after feeding will the animal return to a normal size. The lack of
a growth requirement for regeneration is shown in heavily irradiated Hydra; no cell A piece of hypostome induces a secondary
divisions occur in these animals but they can still regenerate more or less normally. axis when grafted to an intact Hydra
Hydra lacking interstitial cells also regenerate normally. Even more dramatically,
a mixture of dissociated Hydra cells will form an animal.
The basis for the enormous regenerative capacity of Hydra may reside in the stem-
cell-like nature of its epithelial cells. These cells are continually self-renewing and
do not seem to age, and they are the only cells that can stimulate regeneration. The
epithelial cells also give rise to various cell types, such as those of the basal disc and
the gut lining, with the type of cell produced being determined by positional values
along the body axis.

Fig. 14.16 The hypostome can induce

14.8 The head region of Hydra acts both as an organizing region and as an a new head and body in Hydra. When an
inhibitor of inappropriate head formation excised fragment of hypostome is grafted
into the gastric region of another, intact
At the beginning of the twentieth century, it was found that grafting a small frag- Hydra, it can induce the formation of a
ment of the hypostome region of a Hydra into the gastric region of another Hydra complete new secondary axis with a head
induced a new head, complete with tentacles, and a body axis (Fig. 14.16). Similarly, and tentacles.
550 Chapter 14 Regeneration

Fig. 14.17 The head region of Hydra produces an inhibitory signal that falls off with
A piece of region 1 fails to induce a secondary
axis when grafted into an intact Hydra distance. Region 1 does not induce a head when grafted into the gastric region of an intact
Hydra (top panel), indicating the presence of an inhibitory signal in the host. If the host’s head
is removed and a piece of region 1 is then grafted into the host, a secondary axis is induced
(middle panel), indicating that the head region is the source of the inhibitory signal. Region 1
can induce a new axis in the foot region of an intact Hydra because the inhibitory signal grows
weaker further away from the head (bottom panel).

t­ransplantation of a fragment of the basal region induced a new body column with
a basal disc at its end. Hydra, therefore, has two organizing region, one at each end,
which give the animal its overall polarity. The hypostome and the basal disc act as
When the host’s head is removed, grafting organizing regions (like the Spemann organizer in amphibians and the polarizing
a region 1 induces a secondary axis
region in vertebrate limb buds). The organizing function of the head region is due to
two signals it produces which act in graded fashion down the body column: one sig-
nal inhibits head formation and the other specifies a gradient in positional value.
Grafting experiments originally showed that the hypostome produces an inhibi-
tor of head formation (Fig. 14.17). For example, when a body piece from just below
the head (region 1 in Fig. 14.15) is grafted into the gastric region, it rarely induces a
new head and is usually simply absorbed into the body. But if the head of the host is
removed at the time of grafting, the graft will then induce a new axis and head, sug-
gesting that the removal of the head results in the loss of some factor that is inhibiting
A piece of region 1 successfully induces a head formation. The inhibitory effect falls off with distance from the head: when a
secondary axis when grafted further away from region 1 is grafted to near the foot it can induce a new head even when the original
the head of an intact host
head is still in place (see Fig. 14.17, bottom panel). These experiments suggested that
the inappropriate formation of extra heads is normally prevented in Hydra by a lateral
inhibition mechanism (see Section 1.16) acting through a gradient of inhibitory signal
with its highest concentration at the head end. An opposing gradient inhibiting foot
regeneration seems to be produced by the basal disc. These gradients are dynamic
and must be continually under adjustment as the animal grows; this is shown by the
fact that when the head is removed, the concentration of the inhibitor falls.
A simple model for the two gradients emanating from the head assumes that the head
inhibitor is a secreted factor, made by the head, which diffuses down the body column
and is degraded at the basal end. The gradient in positional value, on the other hand, is
assumed to be an intrinsic property of cells. In this model, both gradients are linear—their
values decrease at a constant rate with distance from the head. We further assume that,
provided the level of inhibitor is greater than the threshold set by the positional value,
head regeneration is inhibited. Removal of the head results in the concentration of inhibi-
tor falling, as the inhibitor is degraded and cannot be replaced. The decrease in inhibitor
concentration is greatest at the cut end, and when the inhibitor falls below the threshold
concentration set by the local positional value, the positional value increases to that of the
Fig. 14.18 A simplified model for head head end (Fig. 14.18). Thus, the first key step in this morphallactic regeneration, when
regeneration in Hydra. This model assumes the head region is removed, is the specification of a new head region at the cut surface.
that there are two gradients running from
When the positional value has increased to that of a normal head region, the cells
head to foot. One is a diffusible molecule (I)
start to make inhibitor and so prevent head formation in other body regions. The level of
that inhibits head formation and is produced
by the head. The other is a gradient of
inhibitor will always first fall below the threshold for head inhibition where the ­positional
positional value (P) that is an intrinsic
property of the cells. When the head is 1 2 3 4 5
Positional value P ( )

removed, the concentration of inhibitor falls

and inhibitor I ( )

at the cut surface (graph 2) until a threshold I

is reached at which the positional value P
increases to that of the head region (graph 3).
This then re-establishes the inhibitor gradient
(graph 4). The overall gradient in positional
value takes a much longer time to return to
normal (graph 5).
 Regeneration in Hydra 551

value is highest, thus maintaining polarity. Once a new head has been specified and
the inhibitory gradient re-established, the gradient in positional value also returns to
normal, but this can take more than 24 hours. Morphallactic regeneration results in a
smaller Hydra which, after feeding, will eventually grow back to a normal size.

14.9 Genes controlling regeneration in Hydra are similar to those expressed

in vertebrate embryos

Several novel peptides have been identified in Hydra that are linked to epithelial cell
positional differentiation along the body axis and which can induce a head or a basal
disc. In addition, genes identified as important in vertebrate embryo development
seem to be involved in development in Hydra. The ubiquitous canonical Wnt path-
way leading to the activation of β-catenin (see Box 1E, p. 26) has an important role
in Hydra development. In situ hybridization experiments revealed that expression of
HyWnt, the Hydra Wnt homolog, in adult animals is restricted to the most apical tip
of the body axis, the head organizer (Fig. 14.19). Within an hour after amputation
of this region, both Hydra β-catenin (Hyβ-Cat) and Wnt genes are expressed at the
regenerating tip. During budding also, expression of Hyβ-Cat is significantly upregu-
lated in a ring-like domain in the prospective budding zone before evagination starts,
and the high level of expression is maintained until the bud detaches. A dramatic
demonstration of the importance of the Wnt signaling pathway comes from blocking
the activity of glycogen synthase kinase-3β, a component of the pathway (see Box 1E,
p. 26). The block causes a rise in the concentration of Hyβ-Cat along the body axis
and all regions acquired the characteristics of the head organizer, as shown by graft-
ing individual regions to another organism. These results show that Wnt signaling is
involved in axis formation in Hydra and support the idea that it played a key part in
the evolution of axial differentiation in early multicellular animals. Hyβ-Cat shows a
high degree of conservation, as injection of Hyβ-Cat mRNA into ventral Xenopus blas-
tomeres at the eight-cell stage is able to induce a complete Xenopus secondary body
axis. This contains the anterior-most structures, such as eyes and cement gland, and
is indistinguishable from the secondary axis induced by Xenopus β-catenin.
Various genes expressed in the vertebrate organizer are also expressed in the Hydra
organizing region. A protein related to vertebrate Chordin is expressed at the site
of regeneration or budding in Hydra, and might block BMP signaling, as it does in
dorso-ventral specification in vertebrates. A Hydra homolog of goosecoid, another
organizer-specific gene, is expressed just above where tentacles will appear. When

Fig. 14.19 Wnt expression in adult Hydra

and in a regenerating tip. Left panel:
adult Hydra stained for Wnt expression
(black), shows expression only the tip of
the hypostome. Right panel: at 1 hour after
hypostome amputation, Wnt is expressed
throughout the site of amputation. After
48 hours, expression is again restricted to
the very tip of the regenerating head (not
shown).
Photographs courtesy of T.W. Holstein.
552 Chapter 14 Regeneration

injected into an early Xenopus embryo, the Hydra version of goosecoid can induce a
partial secondary axis. This suggests that the roles of those genes in organizer tissues
have been retained over many hundreds of millions of years of evolution.

Summary

Hydra grows continually, losing cells from its ends and through the formation of buds. Two
organizing regions, one at the head end and one at the basal end, pattern the body and
maintain polarity. If transplanted to another site, a head region can induce a new body axis
and head. If the body of an intact Hydra is severed in two, it regenerates by morphallaxis,
which does not require new growth. Regeneration initially involves the respecification of
cells at the cut end as ‘head,’ leading to the establishment of an organizing region. The
head region produces an inhibitor that prevents other regions forming a head. The con-
centration of inhibitor decreases with distance from the head. There is also a gradient in
positional value specified by the head that determines the threshold at which head inhibi-
tion occurs. When the head is removed, the inhibitor level in the rest of the body falls, and
a new head region develops where the positional value is highest, thus maintaining polar-
ity. The expression of genes in the Hydra head organizer homologous with those in the
organizer region of vertebrates suggests an ancient evolutionary origin for the organizer.

Summary to Chapter 14

Regeneration is the ability of an adult organism to replace a lost part of its body. The capac-
ity to regenerate varies greatly among different groups of organisms and even between dif-
ferent species within a group. Mammals have very limited powers of regeneration, whereas
newts can regenerate limbs, jaws, and the lens of the eye. Regeneration of amphibian limbs
involves dedifferentiation of cells and the formation of a blastema, which grows and forms
the regenerate. Regeneration requires the presence of nerves. Evidence exists for a prox-
imo-distal gradient in positional values as intercalation of positional values occurs when
normally non-adjacent values are placed next to each other in amphibian and insect limbs.
The zebrafish heart can regenerate, as can the mammalian liver and the peripheral nervous
system. Regeneration in the cnidarian Hydra does not require growth. The head end pro-
duces an inhibitory signal for head formation and a signal that specifies positional value
along the body. When the head is removed, the inhibitor concentration falls and a new head
forms. Hydra may provide a model for an evolutionarily early developmental system.

Summary: regeneration in Hydra

Hydra grows continuously and loses cells from its ends and by budding

the head produces a gradient of inhibitor that prevents other heads forming;
the head also specifies a gradient in positional values

local reduction in concentration of inhibitor

local increase in positional value

formation of new head without new growth

 Regeneration in Hydra 553

n End of chapter questions c) the reinitiation of division in existing cells and new growth, fol-
lowed by patterning, as occurs in salamanders
Long answer (concept questions) d) the reinitiation of division in existing cells and new growth, fol-
1. Contrast morphallaxis and epimorphosis. Give an example of each. lowed by patterning, as occurs in Hydra

2. What is the effect of denervation on regeneration of a salaman- 2. Which statement is true about regeneration in the newt eye?
der limb? What is the role of the nerve supply? What protein can a) The entire eye of a newt can be removed and will regenerate from
substitute for the nerves in enabling regeneration? the stump of the optic nerve.
3. Refer to Fig. 14.7. Describe in your own words what the experi- b) The lens can be removed, and will regenerate from the overlying
ment illustrated was intended to investigate, what was observed, and cornea.
what interpretation can be drawn from results of the experiment. c) The lens can be removed, and will regenerate via transdifferen-
tiation from the iris.
4. The dedifferentiated cells of the salamander blastema will regen-
erate only structures distal to the site of the amputation, which d) The newt eye does not show any power of regeneration.
requires some kind of memory of proximo-distal position in the 3. Msx1 is
intact limb. Summarize the information that Prod1 may be involved
a) a growth factor that acts on the blastema to stimulate regenera-
in registering proximo-distal position; be sure to include the results
tion
of retinoic acid treatment on Prod1 expression.
b) a proteolytic enzyme involved in the blood-clotting cascade
5. Describe the effect of retinoic acid on regeneration; for example,
c) a regulator of the cell cycle
if a limb is cut at the wrist level and treated with retinoic acid, how
will its regeneration be altered, compared to regeneration without d) a transcription factor expressed in undifferentiated mesenchymal
such treatment? cells

6. In the regeneration of the cockroach leg, proximo-distal values 4. If the nerve supply to a newt limb is severed before amputation,
will be restored through intercalation if two pieces of leg are ­apposed how will this affect regeneration?
by grafting (see Fig. 14.13). How does this process of intercalation a) A blastema will form but will not grow, and regeneration will fail.
explain the results of the experiments shown in the figure? b) No regeneration occurs, and the stump heals over as it would in
7. Describe what is meant by ‘circumferential intercalation’ in cock- a mammal.
roach limb regeneration. How does this differ from proximo-distal c) Outgrowth will occur, but the identity of the limb will be lost and
intercalation? normal proximo-distal patterning will not occur.

8. Discuss the importance to biomedical science of the ability of d) Regeneration of most tissues will occur normally, but regenera-
portions of the adult zebrafish heart to regenerate (you may wish to tion of the nerves will not occur.
review some of the advantages of the zebrafish as a model organism 5. Which is the result of grafting a blastema derived from a distal
from Chapter 3). amputation onto a stump that has been amputated at a proximal
position?
9. What are the limits of regeneration in the adult human liver—
that is, how much can be removed or damaged and the liver will a) The distal blastema is unable to properly interpret its positional
still regrow? What signals promote liver regeneration? What signal values in this circumstance, and regeneration fails.
limits further growth after the liver reaches a size appropriate to the b) The distal blastema regenerates the entire limb, supplying all the
organism? cells required for this regeneration.
10. Contrast the effect of glial Schwann cells on regeneration in the c) The distal blastema regenerates the limb regions removed in
peripheral nervous system with the effect of the oligodendrocyte the distal amputation, resulting in a shortened limb missing the
glial cells on regeneration in the central nervous system. regions between the proximal and distal amputation sites.
d) The limb regions between the distal and proximal amputation
11. In what way do the hypostome and basal disc regions of a
sites are filled in by intercalary growth from the proximal stump,
hydra qualify as organizers? Describe an experiment that illustrates
and the regions distal to the distal amputation are regenerated
this property.
from the distal blastema.
12. Describe the two-gradient model for head regeneration in Hydra.
6. In the amputated newt limb, what is the effect of retinoic acid
treatment of the regenerating stump on the blastema cells’ posi-
Multiple choice (factual recall questions) tional information?
NB There is only one correct answer to each question. a) Proximal cells will behave as if they are more distal than they
1. Epimorphosis is regeneration through were before the amputation.
a) repatterning of existing cells without a requirement for new b) Retinoic acid will have no effect on regeneration.
growth, as occurs in Hydra c) The cells will become respecified to a more proximal position.
b) repatterning of existing cells without a requirement for new d) The cells will lose their positional identity and remain dediffer-
growth, as occurs in salamanders entiated in the blastema.
554 Chapter 14 Regeneration

7. How does regeneration in the peripheral nervous system differ Brockes, J.P., Kumar, A.: Principles of appendage regeneration in
from that in the liver? adult vertebrates and their implications for regenerative medi-
a) In neuronal regeneration, axons can regrow, but the neuron itself cine. Science 2005, 310: 1919–1923.
cannot be replaced by cell division of remaining cells, as can Brockes, J.P., Kumar, A.: Comparative aspects of animal regenera-
liver cells. tion. Annu. Rev. Cell Dev. Biol. 2008, 24: 525–549.
Galliot, B., Tanaka, E., Simon, A.: Regeneration and tissue repair.
b) In regeneration in the peripheral nervous system, cells that are
Cell. Mol. Life Sci. 2008, 65: 3–7.
removed or damaged are replaced from stem cell populations,
whereas in the liver cells are replaced from differentiated liver
cell populations.
c) In the liver, growth factors stimulate regeneration, whereas in the
n Section further reading
peripheral nervous system no growth factors are required since
14.1 Amphibian limb regeneration involves cell
the neurons are continually dividing.
dedifferentiation and new growth
d) Regeneration cannot occur in the peripheral nervous system, in
Brockes, J.P., Kumar, A.: Plasticity and reprogramming of differen-
contrast to the ability of the liver to regenerate when damaged.
tiated cells in amphibian regeneration. Nat. Rev. Mol. Cell Biol.
8. Which statement correctly characterizes regeneration in the 2002, 3: 566–574.
mammalian nervous system? Han, M., Yang, X., Farrington, J.E., Muneoka, K.: Digit regenera-
a) Cells of the brain can regenerate if damaged, but the peripheral tion is regulated by Msx1 and BMP4 in fetal mice. Development
nervous system is incapable of regeneration. 2003, 130: 5123–5132.
Imokawa, Y., Simon, A., Brockes, J.P.: A critical role for thrombin
b) Killed cells of the peripheral nervous system are replaced by divi-
in vertebrate lens regeneration. Phil. Trans. R. Soc. Lond. B Biol.
sion of stem cells, which will extend axons to restore connectivity
Sci. 2004, 359: 765–776.
with targets.
Kragl, M., Knapp, D., Nacu, E., Khattak, S., Maden, M., Epperlein,
c) Peripheral nerves can regenerate an axon if cut, and make correct H.H., Tanaka, E.M.: Cells keep a memory of their tissue origin
connections with muscles and other targets in some cases. during axolotl limb regeneration. Nature 2009, 460: 60–65.
d) The glial cells of the central nervous system support axon out- Kumar, A., Godwin, J.W., Gates, P.B., Garza-Garcia, A.A., Brockes,
growth and regeneration, but the glial cells of the peripheral ner- J.P.: Molecular basis for the nerve dependence of limb regen-
vous system seem to actually suppress nerve regeneration. eration in an adult vertebrate. Science 2007, 318: 772–777.
Kumar, A., Velloso, C.P., Imokawa, Y., Brockes, J.P.: The regenera-
9. Why is Hydra a useful animal for studying development and
tive plasticity of isolated urodele myofibers and its dependence
regeneration?
on MSX1. PLoS Biol. 2004, 2: E218.
a) An extensive set of developmental mutants has been developed Lee, H., Habas, R., Abate-Shen, C.: Msx1 cooperates with histone
in Hydra, making possible a genetic analysis of development and H1b for inhibition of transcription and myogenesis. Science
regeneration. 2004, 304: 1675–1678.
b) Hydra gives insights into mechanisms that arose very early in the Satoh, A., Graham, G.M.C., Bryant, S.V., Gardiner, D.M.: Neu-
evolution of animals. rotrophic regulation of epidermal dedifferentiation during
c) Hydra illustrates unique mechanisms, such as the existence of an wound healing and limb regeneration in the axolotl (Ambys-
organizer region, and the use of morphogenetic gradients, which toma mexicanum). Dev. Biol. 2008, 319: 321–355.
are different from any other animal. Tanaka, E.M., Drechel, D.N., Brockes, J.P.: Thrombin regulates
­S-phase re-entry by cultured newt myotubes. Curr. Biol. 1999,
d) Hydra is a simple organism that nonetheless has all the tissues
9: 792–799.
and cell types of more complex organisms such as mammals.
Yakushiji, N., Suzuki, M., Satoh, A., Sagai, T., Shiroishi, T.,
10. A highly conserved signaling pathway involved in Hydra axis ­Kobayashi, H., Sasaki, H., Ide, H., Tamura, K.: Correlation
formation is between Shh expression and DNA methylation status of the
a) the FGF pathway limb-specific Shh enhancer region during limb regeneration
in amphibians. Dev. Biol. 2007, 312: 171–182.
b) the insulin pathway
c) the retinoic acid pathway 14.2 The limb blastema gives rise to structures with
d) the Wnt/β-catenin pathway positional values distal to the site of amputation
Da Silva, S., Gates, P.B., Brockes, J.P.: New ortholog of CD59 is
Multiple choice answer key
implicated in proximodistal identity during amphibian limb
1: c, 2: c, 3: d, 4: a, 5: d, 6: c, 7: a, 8: c, 9: b, 10: d. regeneration. Dev. Cell 2002, 3: 547–551.
Echeverri, K., Tanaka, E.M.: Proximodistal patterning during limb
regeneration. Dev. Biol. 2005, 279: 391–401.
n General further reading Garza-Garcia, A.A., Driscoll, P.C., Brockes, J.P.: Evidence for the
local evolution of mechanisms underlying limb regenera-
Birnbaum, K.D., Sánchez Alvarado, A.: Slicing across kingdoms: tion in salamanders. Integr. Comp. Biol. 2010, doi:10.1093/icb/
regeneration in plants and animals. Cell 2008, 132: 697–710. icq022.
 Regeneration in Hydra 555

Han, M., Yang, X., Lee, J., Allan, C.H., Muneoka, K.: Development Clavien, P.A.: Liver regeneration: a spotlight on the novel role of
and regeneration of the neonatal digit tip in mice. Dev. Biol. platelets and serotonin. Swiss Med. Wkly. 2008, 138: 361–370.
2008, 315: 125–135. Johnson, E.O., Zoubos, A.B., Soucacos, P.N.: Regeneration and
Torok, M.A., Gardiner, D.M., Shubin, N.H., Bryant, S.V.: Expression repair of peripheral nerves. Injury 2005, 36: S24–S29.
of HoxD genes in developing and regenerating axolotl limbs. Taub, R.: Liver regeneration: from myth to mechanism. Nat. Rev.
Dev. Biol. 1998, 200: 225–233. Mol. Cell Biol. 2004, 5: 836–847.

14.3 Retinoic acid can change proximo-distal positional 14.7 Hydra grows continuously but regeneration does not
values in regenerating limbs require growth
Maden, M.: The homeotic transformation of tails into limbs in Bosch, C.G.: Why polyps regenerate and we don’t: towards a cel-
Rana temporaria by retinoids. Dev. Biol. 1993, 159: 379–391. lular and molecular framework for Hydra regeneration. Dev.
Mercader, N., Tanaka, E.M., Torres, M.: Proximodistal identity dur- Biol. 2007, 303: 421–433.
ing limb regeneration is regulated by Meis homeodomain pro- Hicklin, J., Wolpert, L.: Positional information and pattern regula-
teins. Development 2005, 132: 4131–4142. tion in Hydra: the effect of gamma-radiation. J. Embryol. Exp.
Nakamura, T., Mito, T., Miyawaki, K., Ohuchi, H., Noji, S.: EGFR Morph. 1973, 30: 741–752.
signaling is required for re-establishing the proximodistal axis Otto, J.J., Campbell, R.D.: Tissue economics of Hydra: regulation
during distal leg regeneration in the cricket Gryllus bimacula- of cell cycle, animal size and development by controlled feed-
tus nymph. Dev. Biol. 2008, 319: 46–55. ing rates. J. Cell Sci. 1977, 28: 117–132.
Pecorino, L.T., Entwistle, A., Brockes, J.P.: Activation of a single
14.8 The head region of Hydra acts both as an organizing
retinoic acid receptor isoform mediates proximo-distal respeci-
region and as an inhibitor of inappropriate head formation
fication. Curr. Biol. 1996, 6: 563–569.
Scadding, S.R., Maden, M.: Retinoic acid gradients during limb Bosch, T.C.G., Fujisawa, T.: Polyps, peptides and patterning.
regeneration. Dev. Biol. 1994, 162: 608–617. BioEssays 2001, 23: 420–427.
Brown, M., Bode, H.R.: Characterization of the head organizer in
14.4 Insect limbs intercalate positional values by both hydra. Development 2002, 129: 875–884.
proximo-distal and circumferential growth Müller, W.A.: Pattern formation in the immortal Hydra. Trends
Genet. 1996, 12: 91–96.
Bando, T., Mito, T., Maeda, Y., Nakamura, T., Ito, F., Watanabe, T.,
Wolpert, L., Hornbruch, A., Clarke, M.R.B.: Positional information
Ohuchi, H., Noji, S.: Regulation of leg size and shape by the
and positional signaling in Hydra. Am. Zool. 1974, 14: 647–663.
Dachsous/Fat signalling pathway during regeneration. Devel-
opment 2009, 136: 2235–2245.
14.9 Genes controlling regeneration in Hydra are similar to
French, V.: Pattern regulation and regeneration. Phil. Trans. R. Soc.
those expressed in vertebrate embryos
Lond. B Biol. Sci. 1981, 295: 601–617.
Broun, M., Gee, L., Reinhardt, B., Bode, H.R.: Formation of the
14.5 Heart regeneration in the zebrafish involves the head organizer in Hydra involves the canonical Wnt pathway.
resumption of cell division by cardiomyocytes Development 2005, 132: 2907–2916.
Galliot, B.: Conserved and divergent genes in apex and axis devel-
Poss, K.D., Wilson, L.G., Keating, M.T.: Heart regeneration in
opment in cnidarians. Curr. Opin. Genet. Dev. 2000, 19: 629–637.
­zebrafish. Science 2002, 298: 2188–2190.
Rentzsch, F., Guder, C., Vocke, D., Hobmayer, B., Holstein, T.W.:
Raya, A., Consiglio, A., Kawakami, Y., Rodriguez-Esteban, C.,
An ancient chordin-like gene in organizer formation of Hydra.
­Izpisúa-Belmonte, J.C.: The zebrafish as a model of heart regen-
Proc. Natl Acad. Sci. USA 2007, 104: 3249–3254.
eration. Cloning Stem Cells 2004, 6: 345–351.
Takahashi, T., Fujisawa, T.: Important roles for epithelial cell pep-
14.6 The mammalian peripheral nervous system can tides in hydra development. BioEssays 2009, 31: 610–619.
regenerate
Case, L.C., Tessier-Lavigne, M.: Regeneration of the adult central
nervous system. Curr. Biol. 2005, 15: R749–R753.
15
Evolution and development
nn The evolution of nn The evolutionary nn Changes in the timing of
development modification of embryonic developmental processes
development

The evolution of multicellular organisms is fundamentally linked to embryonic devel-

opment, for it is through development that genetic changes cause changes in body
form that can be passed on to future generations. This raises important questions as
to how evolutionary and developmental processes are linked. How did development
itself evolve? How has embryonic development been modified during animal evolu-
tion? How important are changes in the timing of developmental processes? In this
chapter, we shall discuss these general questions using particular examples from
animal evolution.

From paleontological, molecular, and cellular evidence, the multicellular animals—

the Metazoa—are presumed to descend from a common ancestor that was itself mul-
ticellular and that had, in turn, evolved from a unicellular organism. A simplified view
of the evolution of animals and plants is shown in Fig. 15.1. The history of evolution
is a long one, taking place over thousands of millions of years, and it can only be
accessed indirectly—through the study of fossils and by comparisons between living
organisms. As Charles Darwin was the first to realize, evolution is the result of heri-
table changes in life forms and the selection of those that are best adapted to their
environment (Box 15A, p. 558).
Development is a fundamental process in evolution. The evolution of multicellular
forms of life is the result of changes in embryonic development, and these in turn are
entirely due to heritable changes in genes that control cell behavior in the embryo. It
is also true, however, as the evolutionary biologist Theodosius Dobzhansky once said,
that nothing in biology makes sense unless viewed in the light of evolution. Certainly,
it would be very difficult to make sense of many aspects of development without an
evolutionary perspective. For example, in our consideration of vertebrate develop-
ment we have seen how, despite different modes of very early development, all verte-
brate embryos develop through a rather similar stage, after which their ­development
diverges again (see Fig. 3.2). This shared stage, which is the embryonic stage after
neurulation and the formation of the somites, is a stage through which some distant
ancestor of the vertebrates passed. It has persisted ever since, to become a fundamen-
tal characteristic of the development of all vertebrates, whereas the stages before and
after this stage have evolved differently in different organisms.
Genetically based changes in development that generated more successful modes of
reproduction or adult forms better adapted to their environment, have been selected
 The evolution of development 557

Major geological, paleontological, and cellular events in animal and plant evolution

Geological events Paleontology Molecular, cellular,

developmental biology
0 0
Quaternary glaciation earliest hominids chimp/gorilla/human
Cenzoic 24
radiations of mammals
Tertiary birds and insects
66
warm Earth Mass extinction
100 Cretaceous
opening of Atlantic
144 earliest fossil angiosperms
earliest birds
Mesozoic Jurassic dinosaurs dominant
200 213
mass extinction
Triassic glaciation earliest mammals,
first dinosaurs
248
Permian mass extinction
286 radiation of reptiles
forests of vascular plants monocot/dicot
300 earliest reptiles (amniotes) angiosperm divergence
Carboniferous
mass extinction scales (keratin)
earliest tetrapods and insects
360
bony fish diversity
Devonian invasion of land by plants
Paleozoic and arthropods
400
408 earliest jawed fish hemoglobin duplication
Silurian glaciation mass extinction into � and � chains
438 radiation of planktonic neural crest
Ordovician protists fourfold increase in
Burgess Shale vertebrate DNA
500 widespread phylotypic organization
Millions of years ago

505
Cambrian biominerization of metazoans
543 segmentation,
Vendian increasing O2 Hox clusters
Proterozoic breakup of abundant Ediacaran fossils
600 supercontinent
ice ages evolution of collagen

1000 red algae

major eukaryotic
radiation
Proterozoic ? introns

? chromatin
glaciation earliest eukaryotes
2000 accretion of ? eukaryotic
continental crust endosymbiosis

? bacterial radiation
including cyanobacteria

3000
Archean ? earliest fossils
Fig. 15.1 Major geological, paleontological,
earliest reasonable ? earliest prokaryotes
evidence for life and cellular events in the evolution of
? origin of life animals and plants.
4000 Hadean
Modified from Gerhart, J. and Kirschner, M.:
Earth formed 1997.

for during evolution. Changes in vertebrate development occurring before neurula-

tion are more often associated with changes in reproduction; those occurring after
neurulation are more associated with the evolution of animal form. Genetic variability
resulting from mutations, sexual reproduction, and genetic recombination is present
in the populations of all the organisms we have looked at in this book, and provides
new phenotypes upon which selection can act. Changes in the control regions of
genes that alter the level, the timing, or the tissue-specificity of their expression, are
thought to be particularly important in evolution.
Both changes in the regulation of gene expression and changes in protein structure
that generate novel protein functions have played a part in evolution. There is ample
evidence for changes in protein structure having a direct impact on animal ­biochemistry,
558 Chapter 15 Evolution and development

Box 15A ‘Darwin’s finches’

‘Darwin’s finches’ are a good example of the evolutionary role of

development and of changes in gene expression. Charles Darwin
visited the Galapagos Islands in 1835 and collected a group of
finches, distinguishing at the time 13 closely related species.
What he found particularly striking was the variation in their
beaks. He wrote in The Voyage of the Beagle: ‘It is very remark-
able that a nearly perfect gradation of structure in this one
group can be traced in the form of the beak, from one exceeding
in dimensions that of the largest gros-beak, to another differ-
ing but little from that of a warbler.’ The shapes of the beaks
reflected differences in the birds’ diets and how they got their
food. And Darwin later commented: ‘Seeing this gradation and
diversity of structure in one small, intimately related group of which BMP-4 was injected into the developing beak of a chick
birds, one might really fancy that from an original paucity of embryo resulted in the beak growing broader and deeper. A sub-
birds in this archipelago, one species had been taken and modi- sequent DNA microarray analysis of differences in the levels of
fied for different ends.’ The figure shows the great variation in all the mRNAs expressed in the different finch beaks revealed
the shape and size of their beaks of closely related Galapagos another key player in shaping beak form. The calcium-binding
finches as a result of adaptation to different sources of food. protein calmodulin, which is a regulatory subunit for many pro-
The drawings are by the English ornithologist John Gould from teins, is expressed at higher levels in the long, pointed beaks
specimens collected by Darwin on the Galapagos Islands. than in the short, broad ones, and was also able to cause the
Modern investigations have uncovered clues to the develop- upper beak of a chick embryo to elongate. So beaks are indeed
mental basis of these differences in beak form. The species with an excellent example of how basic developmental processes
broader, deeper beaks relative to length express higher levels of and pathways can be varied, producing differences in form and
the bone morphogenetic protein BMP-4 in the beak growth zone ­function on which natural selection can act.
compared with species with long, pointed beaks. Experiments in

and playing an important role in producing key physiological differences between

species. And changes in gene-control regions are clearly linked to changes in the
activity of genes involved in pattern formation, as we have seen in the fly embryo
(see Chapter 2).
Throughout this book we have emphasized the conservation of some developmen-
tal mechanisms at the cellular and molecular level among distantly related organ-
isms. The widespread use of the Hox gene complexes and of the same few families
of protein-signaling molecules provide excellent examples of this. It is these basic
similarities in molecular mechanisms that have made developmental biology so excit-
ing in recent years; it has meant that discoveries of genes in one animal have had
important implications for understanding development in other animals. It seems that
when a useful developmental mechanism evolved, it was retained and redeployed in
very different organisms, and at different times and places in the same organism. The
Wnt (Wingless) signaling pathway, for example, is already present in simple multicel-
lular animals, such as the cnidarians, which arose early in animal evolution. A major
technical advance over the past decade has been the determination of the genome
sequences of a wide range of animals, including non-bilaterians such as the cni-
darians Hydra magnipapillata and the starlet sea anemone (Nematostella vectensis).
These genome sequences are filling gaps in our knowledge of the gene complements
of these developmentally simple organisms and throwing light on the evolution of
development in multicellular animals in general.
The animal kingdom is conventionally divided into three main groups in terms of
basic body structure—the bilaterally symmetrical Bilateria, the radially symmetrical
Cnidaria (and the Ctenophora), and the Parazoa (the sponges). The Cnidaria and
 The evolution of development 559

the Bilateria are ‘sister groups’—that is, both groups are considered to arise from
a ­common ancestor—and, together with the very simple Placozoa, they constitute
Arthropods
the Metazoa (Fig. 15.2). The Parazoa (the sponges) are considered to be a quite
separate group from the rest of the multicellular animals and we shall not consider
them here. The largest group of animals is the Bilateria, which have bilateral sym-
metry across the main body axis in at least some stage of development, and have a Nematodes
characteristic pattern of Hox gene expression. The Bilateria includes vertebrates and

Bilaterians (triploblasts)
other chordates, echinoderms (despite the apparent radial symmetry of the adults),
arthropods, annelids, molluscs, and nematodes. These animals are triploblasts, as Annelids
they have the three germ layers—endoderm, mesoderm, and ectoderm. The Cnidaria
(corals, jellyfish, hydras and their relatives) are diploblasts, as are the Ctenophora

Urbilateria
(the comb-jellies). They have only two germ layers (ectoderm and endoderm), and
Molluscs
generally have radial symmetry. Some cnidarians, such as Nematostella, do however
show evidence of bilateral symmetry, and studies of Hox gene expression in these
animals have suggested to some researchers that the common ancestor of the Bilateria
and the Cnidaria might have had bilateral symmetry. Platyhelminths
In the rest of this book we have looked at the early development of quite a wide vari-
ety of organisms and found some similarities, as well as a number of differences. In this
chapter we mainly confine our attention to two phyla—the chordates (which include Echinoderms
the vertebrates) and the arthropods (which include the insects and crustaceans). We
focus on those differences that distinguish the members of a large group of related
animals, such as the vertebrates or the insects, from each other. We will look at the
Amphioxus
relationship between the development of the individual organism (its ­ontogeny) and
the evolutionary history of the species or group (the phylogeny): why, for example,
do mammalian embryos pass through an apparently fish-like stage that has structures
resembling gill slits? We discuss the many variations that occur on the theme of a basic, Vertebrates
segmented, body plan: what determines the different numbers and positions of paired
appendages, such as legs and wings, in different groups of segmented organisms? We
consider the timing of developmental events, and how variations in growth can have Cnidaria

Diploblasts
major effects on the shape and form of an organism. In all cases, we ultimately want to
understand the changes in the developmental processes and the genes controlling them
that have resulted in the extraordinary variety of multicellular animals. This is an excit-
Ctenophora
ing area of study in which many problems remain to be solved. But first we will briefly
consider how development itself might have evolved. What is the origin of the egg, and
how might processes such as pattern formation and gastrulation have evolved?
Placozoa

The evolution of development

Fig. 15.2 A metazoan family tree. The tree
To look for the origins of development we must consider what we know about the shown here is based on ribosomal DNA and
Hox gene sequences and on comparisons of
ancestors of multicellular animals. The origin of the ancestor of animals is a tough
whole-genome sequences. The position of
problem. From genomic evidence and studies of the simplest animals, the last com-
the Urbilateria is represented here by a red
mon ancestors of the bilaterians—the Urbilateria—must already have been complex
circle.
creatures, possessing most of the developmental gene pathways used by existing
bilaterians. What Urbilateria might have been like and how they might have been
constructed are now key questions in the evolution–development (evo-devo) field.
Another central challenge is to explain how conserved gene networks already ­present
in the archetypal ancestor were modified to generate the wonderful diversity of ani-
mal life on Earth today.

15.1 Genomic evidence is throwing light on the origin of metazoans

About 35 different animal phyla with distinct body plans currently exist, and almost
30 of these phyla are bilaterians. Bilaterians are traditionally subdivided into the
560 Chapter 15 Evolution and development

­ rotostomes and the deuterostomes. The protostomes develop the mouth close to
p
the blastopore and have a ventral nerve cord traversed by the foregut, and dorsal
­ganglia that serve as a ‘brain.’ Of the animals covered in this book, the arthropods,
such as Drosophila (see Chapter 2), and the nematodes, such as Caenorhabditis
(see Chapter 6), are protostomes. Deuterostomes develop the anus close to the blas-
topore and have a dorsal central nervous system. They include the chordates, such
as the vertebrates (see Chapters 3–5) and the ascidians, and the echinoderms, such
as the sea urchin (ascidian and sea-urchin development are described in Chapter 6).
­Fossils of adult bilaterians make their appearance suddenly during the Cambrian
­period, at a time between 535 to 525 million years ago or even earlier—a time called
the ‘Cambrian explosion’ because of the sudden appearance of an enormous variety
of animal life.
The remaining, non-bilaterian, metazoans mostly belong to the phylum Cnidaria; they
are the simplest and oldest known animals possessing a gut. The genome of the sea
anemone Nematostella reveals an extensive gene repertoire more similar to that of verte-
brates than to flies or nematodes, implying that the genome of the ancestor of all animals
was similarly complex. Nearly one-fifth of the inferred genes in the Nematostella genome
are novelties—not known from other animal groups—and the genome is enriched in
genes for animal functions such as cell–cell signaling, cell adhesion, and synaptic
transmission.
Taking all the evidence from bilaterians and cnidarians together, we can speculate that
the ancestor of all extant animals lived perhaps 700 million years ago and would have
had flagellate sperm, development through a process of gastrulation, multiple germ
layers, true epithelia lying on a basement membrane, a lined gut, neuromuscular and
sensory systems, and fixed body axes. Nematostella’s genome contains about 18,000
protein-coding genes, a similar number to Caenorhabditis. Genes representing 56 fami-
lies of homeodomain gene-regulatory proteins are present in Nematostella and include
representatives of four different classes of Pax genes, genes that in bilaterians are
involved in many different aspects of development such as eye development, segmen-
tation, and neural patterning. More remarkably for an animal with only two germ
layers, at least seven genes associated with the formation of mesoderm in bilaterians
are expressed in the developing endoderm of Nematostella. Thus the machinery for
germ-layer specification was already present in the ancestors of both bilaterians and
cnidarians.
There is one very simple and primitive free-living marine animal called Trichoplax,
which is so unusual it has been placed in a phylum of its own, the Placozoa. Its rela-
tionship to other animals is not yet clear, but genomic comparisons place it as a very
primitive member of the Metazoa. Trichoplax is composed of two layers of epithelium,
which form a flat disc with no gut, and it has just four different cell-types. It mainly
reproduces by fission. Nevertheless, in line with the genomes of other animals, the
Trichoplax genome contains an estimated 11,500 protein-coding genes, which encode
a rich array of transcription factors and signaling proteins, including homeodomain-
containing proteins, Tbox genes similar to Brachyury, and components of the Wnt/b-
catenin signaling pathway.

15.2 Multicellular organisms evolved from single-celled ancestors

If recognizable animals had evolved by some 700 million years ago, the question still
remains of how they evolved from a unicellular ancestor, which from genomic and
other evidence is thought to have been an organism rather similar to the modern cho-
anoflagellates, unicellular and colonial protozoa with a single flagellum surrounded
by a ‘collar’ of villi, which are very similar in morphology to the collared cells (cho-
anocytes) of sponges. What had to be invented for the transition to multicellularity?
And how did embryonic development from an egg evolve? The key requirements for
 The evolution of development 561

embryonic development, as we have seen, are a program of gene activity, cell differ-
entiation, signal transduction (so cells can communicate with each other), and cell
motility and cohesion (so that overall form can change).
Judging by modern unicellular eukaryotes, such as the choanoflagellates, the
single-celled organism ancestral to animals would have possessed all these features
in primitive form, and little new would have had to be invented, although much
had to evolve further and acquire different functions for multicellularity to evolve.
The genome sequence of the modern choanoflagellate Monosiga brevicollis has been
determined and shows evidence of some cell adhesion and signaling proteins that
are otherwise only present in metazoans. Unicellular eukaryotes have the ability to
move, to adhere, and to respond to signals. At a minimum, the ancestor of multi-
cellular organisms would have already possessed the characteristic eukaryotic cell
cycle, which entails a complex program of gene activity, the ability to undergo cel-
lular differentiation, signaling by receptor tyrosine kinases, and cadherin-mediated
cell adhesion.
What then was the origin of multicellularity and the embryo? One possibility, and
it is highly speculative, is that mutations resulted in the progeny of a single-celled
organism not separating after cell division, leading to a loose colony of identical
cells that occasionally fragmented to give new ‘individuals.’ One advantage of a col-
ony might originally have been that when food was in short supply, the cells could
feed off each other, and so the colony survived. This could have been the origin of
both multicellularity and the requirement for cell death in multicellular organisms.
The egg may subsequently have evolved as the cell fed by other cells; in sponges,
for example, the egg phagocytoses neighboring cells. The evolutionary advantage
of an organism developing from a single cell—the egg—is that all the cells of the
organism will have the same genes. This is a prerequisite for the development of
complex patterns and body plans, as patterning requires communication between
cells and that can only be reliable if the cells obey the same rules as a result of hav-
ing the same genetic instructions. There are ‘multicellular’ organisms that develop
from more than one cell, such as slime molds, but they have never evolved numer-
ous complex forms.
Once multicellularity evolved, it opened up all sorts of new possibilities, such as
cell specialization for different functions. Some cells could specialize in providing
motility, for example, and others in feeding, as we see in the endoderm of Hydra.
The origin of pattern formation—that is, cell differentiation in a spatially organized
arrangement—is not known, but may have depended on gradients set up by external
influences such as inside–outside differences.
How gastrulation evolved is also unknown, but it is not implausible to consider
a scenario in which a hollow sphere of cells, the common ancestor of all multicel-
lular animals, changed its form to assist feeding (Fig. 15.3). This ancestor may, for
­example, have sat on the ocean floor, ingesting food particles by phagocytosis. A small Fig. 15.3 A possible scenario for the
invagination developing in the body wall could have promoted feeding by forming a development of the gastrula. A hollow
primitive gut. Movement of cilia could have swept food particles more efficiently into multicellular sphere of cells, perhaps derived
this region, where they would be phagocytosed. Once the invagination formed, it is from a colonial protozoan, could have settled
not too difficult to imagine how it could eventually extend right across the sphere to on the sea bottom and developed a gut-like
invagination to aid feeding.
fuse with the other side and form a continuous gut, which would be the endoderm.
Based on Jägersten, G.: 1956.
562 Chapter 15 Evolution and development

At a later stage in evolution, cells migrating inside, between the gut and the outer
epithelium, would give rise to the mesoderm.
Gastrulation provides a good example of developmental change during evolution.
While there is considerable similarity in the process of gastrulation in many different
animals, there are also significant differences. But how these evolved and what could
have been the adaptive nature of the intermediate forms is a hard problem.

Summary

The embryo arose during the evolution of multicellular organisms from single-celled
organisms, which have most of the cellular properties required by embryonic develop-
ment. Multicellularity, having arisen, might have persisted originally because a colony
of cells could provide a source of food for some of the members in times of scarcity.
The egg might have evolved as the cell that is fed by other cells and provides the basis
for the development of complex structures. The development of an embryo provides
insights into the evolutionary origin of the animal.

The evolutionary modification of embryonic development

Comparisons of embryos of related species have suggested an important generaliza-

tion about development: the more general characteristics of a group of animals (that
is, those shared by all members of the group) usually appear earlier in their embryos
than the more specialized ones, and arose earlier in evolution. In the vertebrates,
a good example of a general characteristic would be the notochord, which is common
to all vertebrates and is also found in other chordate embryos. The phylum Chordata
comprises three subphyla—the vertebrates; the cephalochordates, such as amphioxus
(lancelet) (Fig. 15.4); and the urochordates, such as the ascidians (see Chapter 6).
They all have, at some embryonic stage, a notochord, flanked by muscle, and a dorsal
neural tube. The amphioxus in particular has many features of a primitive vertebrate,
even when adult: a dorsal hollow nerve cord, in this case protected by the persis-
tent notochord, rather than a bony spine, and segmental muscles that derive from
somites. Paired appendages, such as limbs, which develop later, are special characters
that develop only in the vertebrates, and differ in form among different vertebrates.
Fig. 15.4 The cephalochordate amphioxus
compared to a hypothetical primitive
vertebrate similar to a present-day Vertebrates
lamprey. The overall construction of the two
organisms is very similar, with a dorsal nerve
cord, a more ventral axial skeleton known as
the notochord, and a ventral digestive tract.
Both animals have gills in the pharyngeal
region, structures that are designed to
capture food and also take in oxygen mouth pharyngeal gills digestive tract notochord anus nerve cord
from the water. The notochord extends
right to the anterior end of amphioxus,
but the vertebrate has a prominent head
at the anterior end, extending beyond the
notochord. In more advanced vertebrates,
the notochord is present only in the embryo,
being replaced by the vertebrae.
Amphioxus
Modified from Finnerty, J.R.: 2000.
 The evolutionary modification of embryonic development 563

All vertebrate embryos pass through a phylotypic stage at which they all more or Number
less resemble each other and show the specific embryonic features of the phylum to of
somites
which they belong. After the phylotypic stage, vertebrate embryos diversify and give
rise to the diverse forms of the different vertebrate classes. The phylotypic stage itself 40 lamprey
is somewhat variable, however, as can be seen in Fig. 15.5, which shows that the
number of somites can be very different, and that some organs, such as limb buds, 44 ray
are at different stages of development at this time. The development of the different
vertebrate classes before the phylotypic stage is also highly divergent, because of their
very different modes of reproduction; some developmental features that precede the
phylotypic stage are evolutionarily highly advanced, such as the formation of a tro- 38 lungfish
phoblast and inner cell mass in mammals. This is an example of a special character
that developed late in vertebrate evolution, and is related to the nutrition of the em-
bryo through a placenta, rather than a yolky egg.
The reason that vertebrate embryos pass through a phylotypic stage may be related
to that being the stage at which the Hox genes, which control general pattern along 15 zebrafish
the main body axis, are expressed. Also, it is the stage at which structures common to
all vertebrates are developed, such as somites, notochord, and neural tube.
We will now consider the modifications that have occurred to a variety of embry-
15 frog
onic structures during evolution, including the basic body plan and the limbs.

15.3 Hox gene complexes have evolved through gene duplication

Hox genes play a key role in the development of both vertebrates and insects—and in 35 quail
most other animals. By comparing the organization and structure of the Hox genes in
­insects and vertebrates, we can determine how one set of important developmental
genes has changed during evolution. A major general mechanism of evolutionary
change has been gene duplication and divergence. Tandem duplication of a gene,
which can occur by a variety of mechanisms during DNA replication, provides the
embryo with an additional copy of the gene. This copy can diverge in both its nucle- 33 mouse
otide sequence and its regulatory region, thus changing its pattern of expression and
its downstream targets without depriving the organism of the function of the original
gene (Fig. 15.6). The process of gene duplication has been fundamental in the evolu-
tion of new proteins and new patterns of gene expression; it is clear, for example, that Fig. 15.5 Vertebrate embryos at the
the different hemoglobins in humans (see Fig. 10.11) have arisen as a result of gene phylotypic (tailbud) stage show wide
duplication. variation in somite number. Not to scale.
The Hox gene complexes provide one of the clearest examples of the importance Modified from Richardson, M.K., et al.: 1997.
of gene duplication in developmental evolution. The Hox genes specify pattern along
the antero-posterior axis, as we have seen in previous chapters, and are among the
trademark genes that characterize the Metazoa. They are members of the homeobox
gene superfamily, which is characterized by a short 180 base pair motif, the homeo-
box, which encodes a helix–turn–helix domain that is involved in transcriptional
regulation (see Box 5E, p. 194). Two features characterize all vertebrate Hox genes:
the individual genes are organized into one or more gene clusters or complexes, and
the order of expression of individual genes along the antero-posterior axis is usually
the same as their sequential order in the gene complex (see Section 5.6).
The Hox genes are assumed to have evolved by duplication of a single ancestral
gene. The simplest Hox gene complexes are found in invertebrates, and comprise a
small number of sequence-related genes carried on one chromosome. Vertebrates
typically have four sets of Hox genes, carried on four different chromosomes,
­suggesting two rounds of wholesale duplication of an ancestral Hox gene complex,
in line with the generally accepted idea that large-scale duplications of the genome
have occurred during vertebrate evolution. There has also been further duplication of
Hox genes within each complex. In several fish there has been one further round of
564 Chapter 15 Evolution and development

genome duplication: the zebrafish has seven Hox clusters, the result of three rounds
of duplication and the subsequent loss of one cluster. The Japanese puffer fish Fugu
rubripes has four Hox clusters, but many genes have been lost, and one cluster bears
duplication little relation to any in other vertebrates. It appears that in the ancestral chordates
that gave rise to the vertebrates, the spatial co-linearity between the order of the
Hox genes along the chromosome and their expression along the main body axis
was confined to the ectoderm-derived neural tube, and possibly the epidermis, and
that it was only later extended to other tissues, thus allowing increased complexity
of body plan.
change in change in Comparisons of Hox genes of various arthropod and chordate species indicate that
coding promoter
region the common ancestor of arthropods and vertebrates was most likely to have had
a simple Hox complex of seven genes (Fig. 15.7). As we get nearer to the verte-
brates, the cephalochordate amphioxus has just one Hox gene cluster, containing
14 genes, and one can think of this as most closely resembling the ancestor of the four
­vertebrate Hox gene complexes: Hoxa, Hoxb, Hoxc, and Hoxd (see Fig. 15.7). Possible
change in
promoter ways in which both the vertebrate and Drosophila Hox complexes could have evolved
from a simpler ancestral complex by gene duplication have been reconstructed. In
Drosophila, for example, successive tandem duplications of an ancestral Antp-like
gene could have given rise to the genes abdominal-A (abd-A), Ultrabithorax (Ubx),
and Antennapedia (Antp) (see Fig. 15.7). An Antennapedia-like gene has been impli-
cated as the ‘original’ Hox gene by genetic experiments that suggest that the thoracic
Fig. 15.6 Gene duplication and
segment T2 of insects, such as Drosophila, which is specified by Antp, represents the
diversification. Once a gene has become
‘ground state’ segmental type, from which other body segments have diversified, both
duplicated, the second copy can evolve new
functions and/or new expression patterns.
anteriorly and posteriorly, as a result of Hox gene duplication, mutational changes,
Illustrated is a hypothetical example of and consequent changes in spatial and temporal patterns of Hox gene expression
a gene (purple) with two cis-regulatory imposed on this ground state. If the later-evolved posteriorly expressed Hox genes
modules (blue and green) that confer of the bithorax complex—namely Ubx, abd-A and Abd-B—are deleted in Drosophila,
expression in two different tissues. Left all the thoracic and abdominal segments posterior to T2 revert to the T2 type, and
pathway: after duplication of the complete express Antp (see Fig. 2.48, second panel).
gene and its regulatory regions, mutation In vertebrates, the Hox genes are arranged in four separate clusters, each of which
in the coding region of one of the copies is on a different chromosome. The separate gene complexes probably arose from
can generate a gene with a new and useful duplications of whole chromosomal regions, in which new Hox genes had already
function that is selected for. Both copies will
been generated by tandem duplication. For example, sequence comparisons sug-
be retained in the genome as they both fulfill
gest that the multiple mouse Hox genes most similar in sequence to the Drosophila
useful functions. Subsequent mutation in
a regulatory module could lead to the new
Abd-B gene do not have direct homologs within the Drosophila HOM-C complex (see
gene acquiring a new expression pattern. Fig. 15.7); they probably arose by tandem duplication from an ancestral gene after the
Right pathway: alternatively, a mutation in split of the insect and vertebrate lineages, but before duplication of the whole cluster
one regulatory region in one copy and the in vertebrates. The advantage of duplication was that the embryo had more Hox genes
other regulatory region in the other copy to control downstream targets and so could make a more complicated body. We now
can generate two copies of the gene that consider the role of these genes in evolution of the axial body plan.
retain the original function but are now only
expressed in a single tissue each. Both will
be retained in the genome as they are both 15.4 Changes in Hox genes generated the elaboration of vertebrate and
now needed to fulfill the function of the arthropod body plans
original gene. Examples of all these types
The broad similarities in the pattern of Hox gene expression in vertebrates and
of duplication and divergence are found in
animal genomes.
­arthropods, whose evolution diverged hundreds of millions of years ago, are taken as
good supporting evidence for the idea that all multicellular animals descend from a
common ancestor. Hox genes are key genes in the control of development and are
expressed regionally along the antero-posterior axis of the bilaterian embryo (see
Chapters 2 and 5). The apparently conserved nature of the Hox genes (and of certain
other developmentally important genes) in animal development has led to the con-
cept of the zootype. This defines a general pattern of expression of these key genes
that is present in all animal embryos.
 The evolutionary modification of embryonic development 565

Antennapedia complex bithorax complex

Drosophila lab pb bcd zen z2 Dfd Scr Antp Ubx abd-A Abd-B Fig. 15.7 Gene duplication and Hox
HOM-C gene evolution. A suggested evolutionary
relationship between the Hox genes of
hypothetical a hypothetical common ancestor and
common
ancestor Drosophila (an arthropod), amphioxus
(a cephalochordate), and the mouse
amphioxus (a vertebrate). Duplications of genes of
Hox cluster 1 2 3 4 5 6 7 8 9 10 the ancestral set (red) could have given
rise to the additional genes in Drosophila
and amphioxus. Two duplications of the
mouse Hoxa
whole cluster in a chordate ancestor of
a1 a2 a3 a4 a5 a6 a7 a9 a10 a11 a13
the vertebrates could have given rise to
mouse Hoxb the four separate Hox gene complexes in
b1 b2 b3 b4 b5 b6 b7 b8 b9 b13 vertebrates. There has also been a loss of
some of the duplicated genes in vertebrates.
mouse Hoxc
c4 c5 c6 c8 c9 c10 c11 c12 c13
In Drosophila, the two Hox gene clusters
comprising the HOM-C complex are called
mouse Hoxd the Antennapedia complex and the bithorax
d1 d3 d4 d8 d9 d10 d11 d12 d13 complex. Posterior four genes not shown for
amphioxus.

The Hox genes exert their influence by controlling the activity of other genes that
determine, for example, how the cells in a region develop into segments and append-
ages. Thus, changes in these target genes, as well as changes in the Hox genes them-
selves, can be a major source of change in evolution; in large part, the targets of
the Hox genes still have to be identified. In addition, changes in the pattern of Hox
gene expression along the body can have important consequences. An example of
this is a relatively minor modification of the body plan that has taken place within
vertebrates. One easily distinguishable feature of pattern along the antero-posterior
axis in ­vertebrates is the number and type of vertebrae in the main anatomical
regions—cervical (neck), thoracic, lumbar, sacral, and caudal (see Fig. 5.26). The
number of vertebrae in a particular region varies considerably among the different
vertebrate classes—mammals, with rare exceptions, have seven cervical vertebrae,
whereas birds can have between 13 and 15. How does this difference arise? A com-
parison between the mouse and the chick shows that the domains of Hox gene expres-
sion have shifted in parallel with the change in number of vertebrae in the different
­regions (see Fig. 5.26). For example, the anterior boundary of Hoxc6 expression in the
mesoderm in mice and chicks is always at the boundary of the cervical and thoracic
regions, irrespective of the number of cervical vertebrae. Moreover, the Hoxc6 expres-
sion boundary is also at the cervical–thoracic boundary in geese, which have three
more cervical vertebrae than chickens. The changes in the spatial expression of Hoxc6
thus correlate specifically with the number of cervical vertebrae rather than with a
fixed number of vertebrae. Other Hox genes are involved in the patterning of the
antero-posterior axis, and their boundaries also shift with a change in anatomy.
Changes in the targets of master regulatory genes, such as Hox genes, probably
play a key role in determining the differences between animal species. There may be
many—perhaps very many—downstream targets of the Hox genes. In Drosophila, for
example, it is expression of the Hox gene Ubx that prevents the rudimentary haltere
from growing and so makes it different from its neighboring appendage, the wing.
Tests have shown that six of the twelve genes known to be involved in patterning
the wing are under the control of Ubx and are repressed in the haltere. But because
566 Chapter 15 Evolution and development

of changes in their cis-regulatory regions, only some of these genes are repressed
by Ubx in butterflies, resulting in the development of a hind wing rather than a
haltere.
Another example of how subtle variations in body form can be controlled by Hox
genes comes from insects. There are around 150,000 species of flies and all use
roughly the same set of developmental genes. Diversity in form can reflect different
patterns of gene activity, but the fascinating question in this case is how the same sets
of proteins can generate such subtle morphological diversity. In the genus Drosophila,
different species have different patterns of non-sensory bristles—trichomes—on the
femur of the second leg. Drosophila melanogaster has a small naked patch, whereas
D. simulans has a large one and D. virilis no naked patch at all (Fig. 15.8). Ubx
represses trichome development, and the extent or presence of the naked patch is
determined by subtle differences in the regulation of Ubx expression in this region.
These differences are due to small sequence differences in the control region of Ubx
in the different species, and this variation provides the evolutionary basis for the
variation in pattern.
Changes in patterns of Hox gene expression can also help explain the evolution
of arthropod body plans. Insects and crustaceans are distinct groups of arthropods
that have evolved from a common arthropod ancestor, which probably had a body
composed of more or less uniform segments. A comparison of Hox gene expression
patterns in the embryos of a grasshopper (an insect) and of the brine shrimp Artemia
(a crustacean), indicates how the different body plans might have evolved from an
ancestral body plan. This comparison shows that both the pattern of Hox gene expres-
sion and the adult body regions to which particular Hox genes relate have changed
during the evolution of these two groups (Fig. 15.9).

D. melanogaster

D. simulans

D. virilis
Fig. 15.8 Trichome patterns on the
posterior second femur vary among
Drosophila species. Top panel: Drosophila
melanogaster. Middle panel: D. simulans.
Bottom panel: D. virilis. The arrow and
arrowhead in the top two panels indicate the
extent of the patch of naked cuticle.
From Stern, D.L.: 1998.
 The evolutionary modification of embryonic development 567

Fig. 15.9 A comparison of body plans and

genital
thorax segments abdomen Hox gene expression in two arthropods. A
comparison of Hox gene expression during
embryogenesis in an insect, the grasshopper
(bottom), with that in a crustacean, the brine
shrimp Artemia (top), shows that both the
pattern of Hox gene expression and the adult
body regions to which particular Hox genes
thorax abdomen relate have changed during the evolution of
these two groups of arthropods from their
common ancestor. In Artemia, the three Hox
Antennapedia genes Antennapedia, Ultrabithorax, and
Ultrabithorax abdominal-A are all expressed throughout
Ultrabithorax and abdominal-A the thorax, where most of the segments
Abdominal-B are similar. However, the expression of
these genes in the thorax and abdomen
of the grasshopper is different. They
have overlapping and distinct patterns
of expression that reflect the regional
The grasshopper has a pattern of Hox gene expression similar to that of Drosophila. differences in the insect thorax. The gene
The Hox genes Antp, Ubx, and abd-A specify distinct segment types in the thorax Abdominal-B is expressed in the genital
and abdomen, but are expressed in overlapping domains, and segment types are regions of both animals, indicating that these
defined by combinatorial expression (see Section 2.27). In the brine shrimp, how- two regions are homologous.
ever, these genes are all expressed together in a thoracic region that is composed of After Akam, M.: 1995.
uniform segments. This suggests that the thorax of Artemia might be homologous
to the whole insect thorax and much of the insect abdomen. The term homologous
is used here to describe structures that share a common ancestry or developmental
derivation, even if they no longer serve the same function. In the evolution of the
grasshopper and the brine shrimp from their common ancestor, the changes in
body plan have resulted in part from spatial changes in the expression of particular
Hox genes, but key changes in their downstream targets have also been involved.
Comparisons between species clearly show that Hox genes do not specify particular
structures but simply provide a regional identity. How that identity is interpreted to
produce a particular morphology is the role of genes acting downstream from the
Hox genes.
Genes other than Hox genes are involved in the basic specification and pattern-
ing of the arthropod head. The head regions of different types of arthropods are
very ­different, and indeed spiders do not seem to have a head at all. Their body is
divided into a front and a back region, the front end having segments that bear the
spider’s fangs, pedipalps, and four pairs of walking legs; the front region thus acts
as a head and thorax combined. In spite of these differences, typical ‘head’ genes
are expressed in the anterior-most region in both spiders and insects, including the
homeobox gene orthodenticle, which is involved in specifying the larval head and
anterior brain in Drosophila (see Section 2.31), and the Hox gene labial, which in
Drosophila is expressed in the imaginal discs specifying the labial palps in the adult
fly. This comparison shows how much morphology can change as a result of changes
in the downstream targets of genes.

15.5 The position and number of paired appendages in insects is dependent

on Hox gene expression

Good illustrations of the role of Hox genes in regional specification are provided by
arthropod appendages. Insect fossils display a variety of patterns in the position and
number of their paired appendages—principally the legs and wings. Some insect
568 Chapter 15 Evolution and development

f­ossils have legs on every segment, whereas others only have legs in a distinct thoracic
region. The number of abdominal segments bearing legs varies, as does the size and
shape of the legs. Wings arose later than legs in insect evolution. Wing-like append-
ages are present on all the thoracic and abdominal segments of some insect fossils,
but are restricted to the thorax in others. To understand how these different patterns
of appendages arose during evolution we need to look at how the different patterns of
appendages develop in two orders of modern insects, the Lepidoptera (butterflies and
moths) and the Diptera (flies, including Drosophila).
The basic pattern of Hox gene expression along the antero-posterior axis is the same
in all present-day insect species that have been studied. Yet the larvae of Lepidoptera
have legs on the abdomen as well as on the thorax, and the adults have two pairs of
wings, whereas the Diptera, which evolved later, have no legs on the abdomen in the
larva or adult, and only one pair of wings, with the second pair of wings having been
modified into halteres. How are these differences related to differences in Hox gene
activity in the two groups of insects?
In Drosophila, Hox genes of the bithorax complex (BX-C) suppress appendage for-
mation in the abdomen by repressing the expression of the Distal-less gene. This
suggests that the potential for appendage development is present in every segment,
even in flies, and is actively repressed in the fly abdomen. It thus seems likely that
the ancestral arthropod from which insects evolved had appendages on all its seg-
ments. During the embryonic development of Lepidoptera, the genes Ubx and Abd-B
are turned off in the ventral parts of the abdominal segments; this results both in
Distal-less being expressed and in legs developing on the abdomen in the larva. The
presence or absence of legs on the abdomen is thus determined by whether or not a
particular Hox gene is expressed there, which shows that changes in the pattern of
Hox gene expression have played a key role in evolution. But, and this is fundamental,
downstream targets are even more important: in insects, Ubx and abd-A expressed
­together repress limb development in the abdomen, whereas in crustaceans these
genes are expressed together in the thorax yet limbs still develop there. The difference
must lie in the genes they target.
The Hox genes can also determine the nature of an appendage: we have seen
how mutations can convert legs into antenna-like structures and an antenna into a
leg that still retains the same proximo-distal positional values (see Section 11.22). The
coding sequences outside the DNA-binding homeodomain in Hox genes have changed
during evolution, and this has significant effects on the specification of segment iden-
tity and appendage development in different species.

15.6 The basic body plan of arthropods and vertebrates is similar, but the
dorso-ventral axis is inverted

A comparison between the body plans of arthropods and chordates reveals an intrigu-
ing difference. In spite of many similarities in their basic body plan—both have an
anterior head, a nerve cord running anterior to posterior, a gut, and appendages—the
dorso-ventral axis of vertebrates is inverted when compared to that of arthropods.
The most obvious manifestation of this is that the main nerve cord runs ventrally in
arthropods and dorsally in vertebrates (Fig. 15.10, left panels).
One explanation for this, first proposed in the nineteenth century, is that during
the evolution of the vertebrates from their common ancestor with the arthropods, the
­dorso-ventral axis was turned upside-down, so that the ventral nerve cord of the
ancestor became dorsal. This startling idea has recently found some support from
molecular evidence showing that the same genes are expressed along the dorso-
­ventral axis in both insects and vertebrates, but in inverse directions. This inversion
 The evolutionary modification of embryonic development 569

Signals specifying dorso-ventral

Idealized arthropod axis in Drosophila
Dorsal Dpp Dorsal

Sog Sog Fig. 15.10 The vertebrate and Drosophila

Ventral
dorso-ventral axes are related but
mouth
Ventral inverted. In arthropods, the nerve cord
Transverse section of blastoderm
is ventral, whereas in vertebrates it is
dorsal—dorsal and ventral being defined by
Signals specifying dorso-ventral
Idealized vertebrate axis in Xenopus the position of the mouth. In Drosophila and
Xenopus, the signals specifying the dorso-
Dorsal BMP-4 Chordin
ventral axis are similar, but are expressed
in inverted positions. The protein Chordin, a
Ventral Dorsal dorsal specifier in vertebrates, is related to
Sog, which is a ventral specifier in Drosophila,
mouth and the vertebrate ventral specifier BMP-4 is
Ventral Surface view of blastula
related to Drosophila Decapentaplegic (Dpp),
which specifies dorsal.
nerve cord gut circulatory system
After Ferguson, E.L.: 1996.

may have been dictated by the position of the mouth. The mouth defines the ventral
side, and a change in the position of the mouth away from the side of the nerve cord
would have resulted in the reversal of the dorso-ventral axis in relation to the mouth.
The position of the mouth is specified during gastrulation, and it is not difficult to
imagine how its position could have moved, but other changes in body structure and
the nervous system were also involved.
We have seen in Chapters 2 and 4 that the patterning of the dorso-ventral axis
in insects and vertebrates involves intercellular signaling. In Xenopus, the protein
Chordin is one of the signals that specifies the dorsal region, whereas the growth
factor BMP-4 specifies a ventral fate. In Drosophila, the pattern of gene expression is
reversed: the protein Decapentaplegic, which is closely related to BMP-4, is the dorsal
signal, and the protein short gastrulation (Sog), which is related to Chordin, is the
ventral signal (see Fig. 15.10, right panels). These signaling molecules are experimen-
tally interchangeable between insects and frogs. Chordin can promote ventral devel-
opment in Drosophila, and Decapentaplegic protein promotes ventral development in
Xenopus. The molecules and mechanisms that set up the dorso-ventral axes in the
two groups of animals are thus homologous, strongly suggesting that the divergence
in the body plans of the arthropods and vertebrates involved an inversion of this axis,
by movement of the mouth during the evolution of the vertebrates. Thus our image
of the common ancestor of chordates and arthropods is an animal in which the estab-
lishment of the two major body axes, antero-posterior and dorso-ventral, was similar
to that in present-day animals.

15.7 Limbs evolved from fins

The limbs of tetrapod vertebrates are special characters that develop after the phy-
lotypic stage. Amphibians, reptiles, birds, and mammals have limbs, whereas fish
have fins. The limbs of the first land vertebrates evolved from the pelvic and pec-
toral fins of their fish-like ancestors. The basic limb pattern is highly conserved in
both the forelimbs and hindlimbs of all tetrapods, although there are differences
both between forelimbs and hindlimbs, and between different vertebrates. Limbs
evolved from fins, but how fins evolved is far less clear, and is a complex problem.
570 Chapter 15 Evolution and development

Nevertheless, the development of these appendages made use of signaling mol-

ecules, like Sonic hedgehog and fibroblast growth factor (FGF), and of transcription
factors, such as the Hox proteins, which were already being used to pattern the
body.
The fossil record suggests that the transition from fins to limbs occurred in the
Devonian period, between 400 and 360 million years ago. The transition probably
occurred when the fish ancestors of the tetrapod vertebrates living in shallow waters
moved onto the land. The fins of Devonian lobe-finned fishes, such as Panderichthys,
are probably ancestral to tetrapod limbs, an early example of which is the limb of
the Devonian tetrapod Tulerpeton (Fig. 15.11). The proximal skeletal elements cor-
responding to the humerus, radius, and ulna of the tetrapod limb are present in the
ancestral fish, and a recent analysis of a fossil Panderichthys using CT scanning has
shown that, contrary to previous ideas, the distal region of the pectoral fin contains
separate skeletal elements, and so fingers are not a tetrapod novelty, as previously
thought.
To gain insights into the transition from fin to limb, researchers have turned to
a modern fish, the zebrafish, in which fin development can be followed in detail
and the genes involved can be identified. The fin buds of the zebrafish embryo are
initially similar to tetrapod limb buds, but important differences soon arise during
development. The proximal part of the fin bud gives rise to skeletal elements, which
may be homologous to the proximal skeletal elements of the tetrapod limb. There
are four main proximal skeletal elements in a zebrafish fin, which arise from the
subdivision of a cartilaginous sheet (Fig. 15.12). The essential difference between fin
and limb development is in the distal skeletal elements. In the zebrafish fin bud, an
ectodermal fin fold develops at the distal end of the bud and fine bony fin rays are
formed within it.
As in the tetrapod limb bud (see Chapter 11), the key gene Shh is expressed at the
posterior margin of the zebrafish fin buds and the expression pattern of Hoxd and
Hoxa genes is similar to that in tetrapods. As in limb development, the expression of
the Hoxa and Hoxd genes in zebrafish pectoral fins occurs in three distinct phases,
in which the most distal, the third phase, is correlated with the ­development of the

Distal

autopodium

radius ulna zeugopodium

Fig. 15.11 The fin-to-limb transition. radius ulna
In the lobe-like fin of the Devonian fish Anterior Posterior
Panderichthys, there were proximal
elements corresponding to the humerus
stylopodium
(the stylopodium), radius and ulna (the humerus
zeugopodium), and distal elements (the
autopodium). The Devonian tetrapod
Tulerpeton has similar proximal elements,
and the distal elements have developed into Panderichthys Tulerpeton
recognizable digits.
Proximal
Adapted from: Boisvert, C.A., et al.: 2008.
 The evolutionary modification of embryonic development 571

proximal cartilaginous element Fig. 15.12 The development of the

pectoral fin of the zebrafish Danio. Left
panel: the pectoral girdle and fin fold. Middle
fin ray panel: four proximal cartilaginous elements
and distal fin rays. Right panel: four proximal
fin fold bony elements supporting the distal fin rays
proximal bony element
in the adult fish.
Photographs courtesy of D. Duboule, from
most distal fin structure—the fin blade. Hox gene expression in the pectoral fin dur-
Sordino, P., et al.: 1995.
ing this phase is dependent on Shh signaling, as in the limb, which indicates that
the regulatory mechanisms underlying the three phases of Hox gene expression in
fins and limbs have remained relatively unchanged during evolution. Although sim-
pler in organization than limbs, the fins of teleost fishes do have a distal structure
that might be considered comparable to the autopod region of the vertebrate limb.
If zebrafish fin development reflects that of the primitive ancestor, then more distal
cartilage elements may have evolved from the distal recruitment of the same develop-
mental mechanisms and processes to generate the radius and ulna, wrist, and digits.
There are, as discussed in Chapter 11, mechanisms in the limb for generating periodic
cartilaginous structures such as digits. It is likely that such a mechanism was involved
in the evolution of digits by an extension of the region in which the embryonic carti-
laginous elements form, together with the establishment of a new pattern of Hox gene
expression in the more distal region.
Differences in fore- and hindlimb development are related to differences in Hox
genes, which are involved in establishing positional differences in the lateral plate
mesoderm from which the limb develops. Limb loss is quite common, and the python
provides a good example of the developmental changes involved. Snakes have hun-
dreds of similar vertebrae in their backbones, as can be seen in the skeleton of a
python embryo in Fig. 15.13 (left). Snakes have no forelimbs, but pythons have a
pair of hindlimb rudiments at the junction between the rib-bearing thoracic verte-
brae, and vertebrae with shorter, forked ribs. In four-limbed vertebrates, expression of
Hoxb5 and Hoxc8 is confined to a short trunk region. In the python embryo, however,
these genes are expressed along the whole body as far as the pelvic rudiment (see
Fig. 15.13). The expansion of these Hox-expression domains is thought to underlie
the expansion of rib-bearing vertebrae and the loss of forelimbs in snake evolution.
Hindlimb buds begin to develop, but the signals associated with apical ridge and
polarizing region signaling (see Chapter 11) are not activated. However, there is evi-
dence that the pattern of Hox gene expression is more complex and might involve a Fig. 15.13 Comparison of Hox gene
different downstream interpretation of the Hox code. expression in python and chick
embryos. The photograph shows a skeleton
of a python embryo at 24 days’ incubation
stained with Alcian blue and Alizarin red.
Chick Python The arrow marks the position of hindlimb
rudiments, which have been removed in
this preparation. Note the similarity of the
cervical forelimb
vertebrae anterior to the arrow. The right
panel shows a schematic comparison of
flank flank domains of expression of Hoxb5 (green),
thoracic thoracic
Hoxc8 (blue) and Hoxc6 (red) in chick and
python embryos. The expansion of the Hoxc8
and Hoxc6 domains in the python correlates
with the expansion of thoracic identity in the
lumbar
axial skeleton and flank identity in the lateral
hindlimb hindlimb
plate mesoderm.
Adapted from Cohn, M.J. and Tickle, C: 1999.
572 Chapter 15 Evolution and development

The great range of anatomical specializations in the limbs of mammals (Fig. 15.14)
is due to changes both in limb patterning and in the differential growth of parts of
the limbs during embryonic development, but the basic underlying pattern of skeletal
­elements is maintained. This is an excellent example of the modularity of the skel-
etal elements. If one compares the forelimb of a bat and a horse, one can see that,
although both retain the basic pattern of limb bones, it has been modified to provide
a specialized function in each. In the bat, the limb is adapted for flying: the digits are
greatly lengthened to support a membranous wing. The homeobox transcription fac-
tor Prx1 is upregulated in the bat wing compared to the mouse forelimb. In the horse,
the limb is adapted for running: in the forelimb, lateral digits are reduced, the central
metacarpal (a hand bone in humans) is lengthened, and the radius and ulna are fused
for greater strength. The role of differential growth rates and the loss of skeletal ele-
ments in the evolution of the horse’s limb, and cases of limb reduction, are considered
later. In some cases, differences in size of fore- and hindlimbs and the length of the
skeletal elements are present from a very early embryonic stage, as in the develop-
ment of the flightless kiwi and in bats (Fig. 15.15).
Many of the changes that occur during evolution reflect changes in the relative
dimensions of parts of the body. We have seen how growth can alter the proportions
of the human baby after birth, as the head grows much less than the rest of the body
(see Fig. 13.6). The variety of face shapes in the different breeds of dogs, which
are all members of the same species, descended from the gray wolf, also provides a
good example of the effects of differential growth after birth. All dogs are born with
rounded faces; some keep this shape, but in others the nasal regions and jaws elon-
gate during growth. The elongated face of the baboon is also the result of growth of
this region after birth.
Because individual structures, such as bones, can grow at different rates, the over-
all shape of an organism can be changed substantially during evolution by heritable
changes in the duration of growth, which also leads to an increase in overall size of
the organism. In the horse, for example, the central digit of the ancestral horse grew
faster than the digits on either side, so that it ended up longer than the lateral digits
(Fig. 15.16). As horses continued to increase in overall size during evolution, this
discrepancy in growth rates resulted in the relatively smaller lateral digits no longer
touching the ground because of the much greater length of the central digit. At a later
stage in evolution, the now-redundant lateral digits became reduced even further in
size because of a separate genetic change.

Human Bat Horse

radius
humerus
humerus
Fig. 15.14 Diversification of mammalian
limbs. The basic pattern of bones in the humerus ulna-radius
forelimb is conserved throughout the
mammals, but there are changes in the
proportions of the different bones, as well
as fusion and loss of bones. This is seen
particularly in the horse limb, in which the radius
radius and ulna have become fused into a ulna
single bone, and the central metacarpal (a ulna metacarpals
hand bone in humans) is greatly elongated.
In addition, there has been loss and reduction metacarpals
of the digits in the horse. In the bat wing,
by contrast, the digits have become greatly
elongated to support the membranous wing.
 The evolutionary modification of embryonic development 573

Kiwi (Apteryx australis)

Bat (Rousettus amplexicaudatus)

Fig. 15.15 Developmental expression
of adult limb size. Top row: the flightless
kiwi (Apteryx. australis) has relatively
short forelimbs and large hindlimbs at all
stages of development, even in the early
embryo. Bottom row: in the bat Rousettus
amplexicaudatus, which has large forelimbs
and smaller hindlimbs when adult, the
forelimb bud is large at all stages, relative to
the hindlimb.
Adapted from Richardson, M.K.: 1999.

lateral
digit

Fig. 15.16 Evolution of the forelimb

in horses. Hyracotherium, the first true
equid, was about the size of a large dog. Its
lateral forefeet had four digits, of which one (the
digit
third digit in anatomical terms) was slightly
longer, as a result of a faster growth rate.
All digits were in contact with the ground.
As equids increased in size, the lateral digits
lost contact with the ground, as the result
of the relatively greater increase in length
of metacarpal 3 (the hand bone of the third
Hyracotherium Mesohippus Parahippus Pliohippus Equus (domestic horse) digit). At a later stage, the lateral digits
became even shorter, because of a separate
55 50 45 40 35 30 25 20 15 10 5 0
genetic change.
Time (millions of years from present)
After Gregory, W.K.: 1957.

A feature of limb evolution is that, while reduction in digit number is common

(there are only three in the chick wing and reduction is common in lizards), ­species
with more than five digits are very rare. It seems that there is a developmental
constraint on evolving more than five different kinds of digits. This may be due
574 Chapter 15 Evolution and development

to the Hox genes in the distal part of the limb providing only five discrete genetic
programs for giving a digit an identity. In limbs with polydactyly, at least two of the
digits are the same (see Box 11B, p. 423), such that there are still only five ­different
kinds of digits. This may be the reason why in an animal with an additional distinc-
tive digit-like element, such as the giant panda’s ‘thumb,’ this digit is in fact a modi-
fied wrist bone.

15.8 Vertebrate and insect wings make use of evolutionarily conserved

developmental mechanisms

Vertebrate and insect wings are not homologous but have some superficial similari-
ties; they have similar functions yet are very different in structure. The insect wing is a
double-layered epithelial structure, whereas the vertebrate limb develops mainly from
a mesenchymal core surrounded by ectoderm. Despite these great anatomical differ-
ences, however, there are striking similarities in the genes and signaling molecules
involved in setting up the axes and so patterning insect legs, insect wings, and verte-
brate limbs (Fig. 15.17, and see Chapter 11). Patterning along the antero-posterior axis
of all these appendages uses Hedgehog-related signals and members of the TGF-β fam-
ily, such as Decapentaplegic (in insects) and BMP-2 (in vertebrates). It is remarkable
that the dorsal surface of the insect wing is characterized by expression of the gene
apterous, whereas the related gene Lmx1 is expressed in the dorsal mesenchyme of
the vertebrate limb. A fringe-like gene is involved in the specification of the boundary
between dorsal and ventral regions in both insect and bird wings.
Another example illustrating the conservation of the genetic machinery for insect
appendages is provided by the gene Distal-less, which is also expressed along the prox-
imo-distal axis of a wide variety of developing appendages in other animals, ­including
annelid parapodia and the tube feet of sea urchins. It is also expressed during vertebrate
limb outgrowth.

Chick
wing bud
fringe apical ridge Wnt-7a Lmx1 Sonic hedgehog BMP-2

Drosophila
imaginal
wing disc

fringe wingless apterous hedgehog decapentaplegic

Fig. 15.17 Comparison of developmental signals in the chick wing double lines represent the dorso-ventral compartment border. Second
bud and Drosophila wing imaginal disc. First column: the chick wing column: the dorsal region of the chick wing is specified by Wnt-7a in
bud (top) is shown with the distal end facing. The double line bisecting the ectoderm, whereas wingless is expressed at the dorso-ventral
it represents the apical ridge. The chick apical ectodermal ridge forms insect wing margin. Third column: the gene Lmx1 is expressed in the
at the boundary between the dorsal cells, which express radical fringe, dorsal region of chick wing bud mesoderm, whereas the Drosophila
and the ventral cells, which do not. The dorso-ventral boundary in the gene apterous, to which it is structurally related, specifies the dorsal
insect wing disc also forms at the boundary between fringe and non- region of the insect wing. Fourth and fifth columns: Sonic hedgehog in
fringe cells (bottom). In the insect wing imaginal disc, the future dorsal the chick wing and hedgehog in Drosophila are expressed in posterior
and ventral regions are in the same plane. The vertical double lines regions and both induce expression of genes of the TGF-β family—
represent the antero-posterior compartment border, and the horizontal BMP-2 and decapentaplegic, respectively.
 The evolutionary modification of embryonic development 575

All this emphasizes how evolution has used a wonderfully modifiable system for
setting up axes and specifying positional values.

15.9 The evolution of developmental differences can be based on changes in

just a few genes

Because of the existence of the two very different morphological forms within the
same species, the teleost fish Astyanax mexicanus is a good model system in which
to study the evolution of developmental mechanisms. Astyanax mexicanus is a single
species consisting of two radically different forms: a sighted, pigmented, surface-
dwelling form and a blind, unpigmented, cave-dwelling form known as a cavefish
Fig. 15.18 The two different forms of
(Fig. 15.18). At least 30 different populations of Astyanax cavefish are present in
Astyanax mexicanus. Surface-living fish
limestone caverns in Mexico, having been isolated from their surface relations for
have eyes and are pigmented (upper photo).
the past few million years. Although functional eyes are lacking in cavefish adults, In cave-dwelling fish (cavefish) the eyes do
the embryos begin to develop eye primordia, which subsequently degenerate. The not develop and pigmentation has been lost.
major cause of eye degeneration is apoptosis of the developing lens cells, which then Photographs courtesy of A. Strickle,
prevents the growth of other optic tissues, including the retina. The non-development Y. Yamamoto, and W. Jeffery.
of the eye is, in turn, the cause of other craniofacial differences between cavefish
and surface fish. Lens apoptosis is induced by increased activity of the Shh signaling
system along the cavefish embryonic midline compared to the embryos of surface-
dwelling Astyanax (see Section 11.23).
Somewhat surprisingly, there is no evidence that cell proliferation stops in the
degenerating cavefish eye. The reason there is no net growth is that the new cells
are quickly removed by apoptosis, which persists during cavefish larval develop-
ment and into adult life. If a lens from a surface-dwelling form of Astyanax is
transplanted into a cavefish embryo, it does not undergo apoptosis and there is a
dramatic restoration of eye development. The role of the increased Shh pathway sig-
naling in cavefish eye development was investigated by increasing Shh expression
in embryos of surface-dwelling Astyanax embryos, which normally develop into
sighted adults. When Shh mRNA was injected into one side of a cleaving embryo,
expression of the ‘eye’ gene pax6 was downregulated unilaterally in the correspond-
ing developing eye region, and the larvae that developed from the embryos overex-
pressing Shh lacked an eye on that side of the head. In other words, the phenotype
of the blind cavefish had been phenocopied by manipulating Shh signals to cause
eye degeneration.
Eyes are initially formed and then degraded during larval or adult development in
all sightless cave-dwelling vertebrates. It is surprising that, eyes begin to develop at
all. This may be because early steps in eye development are required for other essen-
tial steps in development, and the elimination of these steps would be fatal. It is thus
unlikely that cave vertebrates lacking embryonic eyes will be discovered because of
this strong developmental constraint. But why have eyes been lost in cavefish? The
answer is not known, but it may be due to selection based on energy conservation
once eyes were no longer adaptive.
Another intriguing example of developmental evolution is the reduction or loss
Fig. 15.19 Loss of the pelvic spines in
of the pelvic, fin or spine in stickleback populations that became ­separated when
three-spine sticklebacks (Gasterosteus
they colonized new lakes and streams formed as the glaciers melted after the ice
aculeatus). Upper panel: most populations
ages about 20,000 years ago. The paired pelvic spines are present in all marine and of three-spine sticklebacks have a pair of
most freshwater populations of three-spine and nine-spine sticklebacks (those num- pelvic spines (arrow). Lower photo: in some
bers refer to the spines on the back of the fish), but the pelvic spine has been lost freshwater populations, the spines have been
repeatedly in several freshwater populations, as a result of reduction in the pelvic lost due to a mutation in the Pitx1 gene.
fin (Fig. 15.19). Genetic crosses between sticklebacks from different ­populations Photographs courtesy of Mike Shapiro and
enabled the cause of pelvic reduction to be identified as mutations in the control David Kingsley.
576 Chapter 15 Evolution and development

region of the gene for the homeodomain transcription factor Pitx1, with different
D. melanogaster mutations in the different populations. Pitx1 plays a key role in vertebrate hindlimb
development (see Section 11.2) and when Pitx1 and the related Pitx2 gene are
knocked-out in mice, hindlimb development is reduced. These adaptive mutations
in sticklebacks cause loss of expression of Pitx1 exclusively in the pelvic region,
preserving the essential developmental role of this gene in other parts of the body.
Pelvic spine loss in sticklebacks is an interesting example of convergent evolution,
15 Myr because similar cis-regulatory mutations have been selected independently in the
D. biarmipes different stickleback populations.
Another example of evolution of a pattern as a result of changes in cis-regulatory
regions is provided by the pigment pattern in the wing of male Drosophila ­biarmipes,
which is closely related to our standard model organism, D. melanogaster, although
separated by some 15 million years of evolution. Drosophila biarmipes males have
25 Myr a conspicuous dark spot at the anterior tip of the wing that is lacking in male
D. melanogaster (Fig. 15.20). This difference is due to differences in the expression of
the gene yellow, which is required for production of the black pigment, in the wings
D. pseudoobscura
of the two species. In D. melanogaster, yellow is expressed at a low level throughout
the wing. The spot in D. biarmipes appears to have evolved as a result of changes
in the cis-regulatory region of the yellow gene, such that it is now activated in a
particular spatial pattern by transcription factors that are involved in patterning the
wing. For example, the protein Engrailed is expressed in the posterior wing compart-
ment (see Section 11.18) and it activates yellow in the distal region of this compart-
ment. Other conserved gene regulatory proteins have also been co-opted to control
Fig. 15.20 Evolution of differences in the yellow gene by the evolution of binding sites for them within the gene’s control
pigment pattern in closely related Drosophila
region.
species. A spot of dark pigment on the tip
of the wing of male Drosophila biarmipes
is a relatively recently evolved trait among 15.10 Embryonic structures have acquired new functions during evolution
Drosophila species. It arises from a change in
the regulatory region of the yellow gene, which If two groups of animals that differ greatly in their adult structure and habits (such as
is required for pigment production. Myr, millions fishes and mammals) pass through a very similar embryonic stage, this could indicate
of years ago. that they are descended from a common ancestor and, in evolutionary terms, are
From Gompel, N. et al.: 2005. closely related. Thus, an embryo’s development reflects the evolutionary history of
its ancestors. Division of the body into segments, which then diverge from each other
in structure and function, is a common feature in the evolution of both vertebrates
and arthropods. In the vertebrates, one example of segmented structure is the bran-
chial arches and clefts that are present in all vertebrate embryos, including humans,
located just behind the head on either side (see Fig. 5.33). These structures are not
the relics of the gill arches and gill slits of an adult fish-like ancestor, but represent
structures that would have been present in the embryo of the fish-like ancestor of ver-
tebrates as developmental precursors to gill slits and gill arches. During evolution, the
branchial arches gave rise both to the gill arches of the primitive jawless fishes and,
in a later modification, to gills and jaw elements in later-evolved fishes (Fig. 15.21).
With time the arches became further modified, and in mammals they now give rise to
various structures in the face and neck (Fig. 15.22), many of which are derived from
the neural crest cells that migrate into the branchial arches early in development (see
­Section 5.13). The cleft between the first and second branchial arches provides the
opening for the Eustachian tube, and endodermal cells in the clefts give rise to various
glands, such as the thyroid and thymus.
Evolution rarely, if ever, generates a completely novel structure out of the blue. New
anatomical features usually arise from modification of an existing structure. One can
therefore think of much of evolution as a ‘tinkering’ with existing structures, which
 The evolutionary modification of embryonic development 577

Fig. 15.21 Modification of the branchial arches during the evolution of jaws in
Hypothetical ancestral jawless vertebrate
vertebrates. The ancestral jawless fish had a series of at least seven gill slits—branchial
clefts—supported by cartilaginous or bony arches. Jaws developed from a modification of the branchial cleft—gill slit
first arch to give the mandibular arch, with the mandibular cartilage of the lower jaw and the
hyoid arch behind it.

gradually fashions something different. It is possible because many structures are

modular: that is, animals have anatomically distinct parts that can evolve indepen- I II III IV V VI VII

dently. Vertebrae are modules, for example, and can evolve independently of each
branchial arch
other; so too are limbs.
A nice example of a modification of an existing structure is provided by the evolu-
tion of the mammalian middle ear. This is made up of three bones—malleus, incus, Jawed vertebrate
and stapes—that transmit sound from the eardrum (the tympanic membrane) to the
spiracle
inner ear. In the reptilian ancestors of mammals, the joint between the skull and
the lower jaw was between the quadrate bone of the skull and the articular bone
I
of the lower jaw, which were also involved in transmitting sound (Fig. 15.23). The
vertebrate lower jaw was originally composed of several bones, but during mamma- II III IV V VI VII
lian evolution one of these bones, the dentary, increased in size and came to comprise
the whole lower jaw, and the other bones were lost. The articular bone was no longer mandibular arch hyoid arch
attached to the lower jaw, and by changes in their development, the articular and the mandibular cartilage
quadrate in mammals were modified into two bones, the malleus and incus, respec-
tively, whose function was now to help transmit sound from the tympanic membrane.
The quadrate is evolutionarily and developmentally homologous with the dorsal car-
tilage of the first branchial arch in the ancestral vertebrate, and the stapes with the
dorsal cartilage of the second.
Evolution can adapt the same proteins for quite different purposes. The eye lenses
of both cephalopods (octopuses and squids) and vertebrates consist of cells packed
with crystallin proteins, which give the lens its transparency (see Section 11.23).
The crystallins were originally thought to be unique to the lens and to have evolved
for this special function, but more recent research indicates that they are co-opted
­proteins that are not structurally specialized for lens function, and in other contexts
act as enzymes.

Embryo 20 weeks

alisphenoid
mandibular
arch malleus incus
stapes
Meckel’s
cartilage
Reichert’s Fig. 15.22 Fate of branchial arch cartilage
I cartilage in humans. In the embryo, cartilage develops
II III styloid
ligament in the branchial arches, which gives rise to
IV
remnants of elements of the three auditory ossicles, the
hyoid arch cardiac Meckel’s cartilage thyroid
bulge cartilage hyoid, and the pharyngeal skeleton. The fate
of the various elements is shown by the color
body of hyoid cricoid coding.
cartilage
After Larsen, W.J.: 1993.
578 Chapter 15 Evolution and development

dentary

dentary
Fig. 15.23 Evolution of the bones of the inner
ear
mammalian middle ear. The articular and incus
quadrate bones of ancestral reptiles (left
stapes inner malleus
panel) were part of the lower jaw articulation. ear
Sound was transmitted to the inner ear via
these bones and their connection to the
stapes. When the lower jaw of mammals outer
became a single bone (the dentary), the quadrate ear
articular bone became the malleus, and the
articular
quadrate bone the incus of the middle ear, eustachian eustachian
acquiring a new function in transmitting tube tube tympanic
stapes membrane
sound to the inner ear from the tympanic
membrane (right panel). The eustachian tube
forms between branchial arches I and II.
After Romer, A.S.: 1949.

These examples provide evidence of a key relationship between evolution and

development, the gradual change of a structure into a different form. In many cases,
however, we do not understand how intermediate forms were adaptive and gave a
selective advantage to the animal. Consider, for example, the intermediate forms in
the transition of the first branchial arch to jaws; what was the adaptive advantage?
We do not know, and because of the passage of time and our current ignorance of the
ecology of ancient organisms we may never know.

Summary

Groups of animals that pass through a similar embryonic stage are descended from
a common ancestor and have evolved modifications of gene-regulatory circuits. The
basic body plan of all animals is defined by patterns of Hox gene expression that
provide positional identity, the interpretation of which has changed in evolution.
The Hox genes themselves have undergone considerable evolution by gene dupli-
cation and divergence and are a good example of Darwin’s description of evolution
as ‘descent with modification.’ The limbs of tetrapod vertebrates evolved from fins,
with the digits as a novel feature. The development of vertebrate and insect limbs
involves the same set of pattern-establishing genes, reflecting the evolution of limb
development from an ancestral mechanism for specifying body appendages. Compari-
son of patterns of dorso-ventral gene expression suggests that during the evolution
of the vertebrates, the dorso-ventral axis of an invertebrate ancestor was inverted.
During evolution, the development of structures can be altered so that they acquire
new functions, as has happened in the evolution of the mammalian middle ear from
a reptilian jaw bone, which itself evolved from a skeletal element in an ancestral
branchial arch.
 Changes in the timing of developmental processes 579

Summary: evolution of structures by development modification

articular and quadrate bones in reptile jaw fins of ancestral fish

malleus and incus bones of limbs of tetrapods, with

mammalian middle ear digits as novel structures

wings of chick and Drosophila use similar signals to set up axes:

Sonic hedgehog–Hedgehog; Lmx1–Apterous; Wnt-7a–Wingless; BMP-2–Decapentaplegic

Hox gene duplication, diversification, and interpretation

different body plans

signals for Xenopus and Drosophila dorso-ventral axes are homologous but axes are inverted:
BMP-4–Decapentaplegic; Chordin–Short gastrulation

Terrestrial species
Changes in the timing of developmental processes

In the previous part of the chapter we focused on changes in spatial patterning that
have occurred during evolution. But changes in the timing of developmental process-
es can also have major effects. In this part of the chapter, we look at some examples
of how changes in the timing of growth and sexual maturation can affect animal form
and behavior.

15.11 Evolution can be due to changes in the timing of developmental events

Differences among species in respect of the time at which developmental processes

occur relative to one another, and relative to their timing in an ancestor, can have dra-
matic effects on both morphology and behavior. Differences in the feet of members of a
genus of tropical salamanders illustrate the effect of a change in developmental timing
on both the morphology and ecology of different species. Many species of the salaman-
der genus Bolitoglossa are arboreal (tree living), rather than typically terrestrial, and
their feet are modified for climbing on smooth surfaces. The feet of the arboreal spe- Arboreal species
cies are smaller and more webbed than those of terrestrial species, and their digits are
shorter (Fig. 15.24). These differences seem to be mainly the result of the development
and growth of the foot ceasing at an earlier stage in the arboreal species than in the ter-
restrial species. The term used to describe such differences in timing is heterochrony.
Some of the clearest examples of heterochrony come from alterations in the timing
of onset of sexual maturity in organisms with larval stages. The acquisition of sexual
maturity by an animal while still in the larval stage is a process that goes under the
name neoteny. The development of the animal, although not its growth, is retarded in
relation to the maturation of the reproductive organs. This occurs in the Mexican axo-
lotl, a type of salamander; the larva grows in size and matures sexually, but does not
undergo metamorphosis. The sexually mature form remains aquatic and looks like an
Fig. 15.24 Heterochrony in
overgrown larva. However, the axolotl can be induced to undergo metamorphosis by
salamanders. In terrestrial species of the
treatment with the hormone thyroxine (Fig. 15.25).
salamander Bolitoglossa (top panel), the
Many animals have evolved free-swimming larval forms that have an advantage when
foot is larger, has longer digits, and is less
it comes to dispersal and feeding, and then undergo a dramatic change in morphol- markedly webbed than in those that live in
ogy to reach the adult state at metamorphosis. The essence of development is gradual trees (bottom panel). This difference can
change; yet at metamorphosis there is no gradual continuity between larva and adult. be accounted for by foot growth ceasing
Metamorphosis makes more evolutionary sense, however, if it is assumed that all larval at an earlier stage in the arboreal species.
forms evolved by the insertion of the larval stage into the pre-existing ­development Scale bar = 1 mm.
580 Chapter 15 Evolution and development

Aquatic Terrestrial

thyroxine
Metamorphosis

larval gills

Fig. 15.25 Neoteny in salamanders. The

sexually mature Mexican axolotl retains larval ­ rogram of a directly developing animal, possibly as a result of heterochrony. In many
p
features, such as gills, and remains aquatic. invertebrates, the larva initially resembles the late gastrula stage, which could have
This neotenic form metamorphoses into a given rise to the free-swimming larval form. Larval stages could not have been the
typical terrestrial adult salamander if treated original state because the metamorphosis that brings a larva back to the developmental
with thyroxine, which it does not produce, program into a sexually mature adult is a highly complex process, and it is difficult to
and which is the hormone that causes see how it could have evolved other than to re-enter the developmental pathway.
metamorphosis in other amphibians. In vertebrates that undergo metamorphosis, such as frogs, if we assume that frog
ancestors developed directly into adults, a change in the timing of events in the post-
neurula stages, such as a delay in hindlimb development, could have led to acquisi-
tion of mobility and evolution of the feeding tadpole stage. Becoming mobile would
have aided dispersal, but to become a reproducing adult, the larva had to return to the
normal developmental program. That, in essence is what metamorphosis is for.
Larval stages can be lost as well as acquired during evolution. Some modern frogs
have re-evolved direct development to the adult by a loss of the tadpole stage and the
acceleration of the development of adult features. Frogs of the genus Eleutherodactylus,
unlike the more typical amphibians Rana and Xenopus, develop directly into an adult
frog and there is no aquatic tadpole stage, the eggs being laid on land. Typical tadpole
features, such as gills and cement glands, do not develop, and prominent limb buds
appear shortly after the formation of the neural tube (Fig. 15.26). In the embryo, the tail
is modified into a respiratory organ. Such direct development requires a large supply of
yolk to the egg in order to support development through to an adult without a tadpole
feeding stage. This increase in yolk may itself be an example of heterochrony, involving
a longer or more rapid period of yolk synthesis in the development of the egg.
Most sea urchins have a larval stage that takes a month or more to become adult.
But there are some species in which direct development has evolved, so that they no
longer go through a functional larval stage. Such species have large eggs and, as a
result of their very rapid development, become juvenile urchins within four days. This
has involved changes in early development such that the directly developing embryo
gives rise to a ‘larval’ stage that lacks a gut and cannot feed, and which metamorpho-
ses rapidly into the adult form.
Fig. 15.26 Development of the frog
Eleutherodactylus. Typical frogs, such as
Xenopus or Rana, lay their eggs in water and
develop through an aquatic tadpole stage,
which undergoes metamorphosis into the
hatching
adult. Frogs of the genus Eleutherodactylus
lay their eggs on land and the frog hatches
from the egg as a miniature adult, without
going through an aquatic free-living larval respiratory
stage. The embryonic tail is modified as a organ
respiratory organ.
 Changes in the timing of developmental processes 581

15.12 The evolution of life histories has implications for development

Animals and plants have very diverse life histories: small birds breed in the spring
following their birth, and continue to do so each year until their death; Pacific salmon
breed in a suicidal burst at 3 years of age; and oak trees require 30 years of growth
before producing acorns, which they eventually produce by the thousand. In order to
understand the evolution of such life histories, evolutionary ecologists consider them
in terms of probabilities of survival, rates of reproduction, and optimization of repro-
ductive effort. These factors have important implications for the evolution of develop-
mental strategies, particularly in relation to the rate of development. For example, as
we have just seen, a characteristic feature of many animal life histories is the presence
of a free-living larval stage that is distinct from, and usually simpler in form than, the
sexually mature adult, and which feeds in a different way.
Life histories help us to understand the evolution of long germ-band insects, such
as Drosophila, which are of more recent origin than short germ-band insects, such
as grasshoppers (see Section 2.25). Unlike Drosophila, short germ-band insects do
not have a larval stage and develop directly into small, immature adult-like forms. In
contrast, Drosophila develops very rapidly into a feeding larva, taking only 24 hours
to start feeding, compared with the grasshopper’s 5–6 days. It is easy to imagine
conditions in which there would have been a selective advantage to insects whose
larvae begin to feed as quickly as possible. It is also likely that embryos are more
vulnerable than adults and so there would be selection for making the embryonic
stage shorter. Thus it is likely that the complex developmental mechanisms of long-
band insects—the system for setting up the complete antero-posterior axis in the
egg—evolved as a result of selection pressure for rapid development. In the case
of fruit flies like Drosophila, this was perhaps related to the ability to feed on fast-
disappearing fruit.
Egg size can also be best understood within the context of life histories. If we
assume that the parent has limited energy resources to put into reproduction, the
question is how should these resources best be invested in making gametes, particu-
larly eggs; is it more advantageous to make lots of small eggs or a few large ones? In
general, it seems that the larger the egg, and thus the larger the offspring at birth, the
better are the chances of the offspring surviving. This would seem to suggest that in
most circumstances an embryo needs to give rise to a hatchling as large as possible.
Why then, do some species lay many small eggs capable of rapid development? One
possible answer is that a parental investment in large eggs may reduce the parents’
own chance of surviving and laying more eggs. This strategy may be especially suc-
cessful in variable environmental conditions, where populations can suddenly crash.
The rapid development of a larval stage enables early feeding and dispersal to new
sites in such circumstances.

Summary

Changes in the timing of developmental processes that have occurred during evolution
can alter the form of the body, for if different regions grow faster than others, their size
is proportionately increased if the animal gets larger. The decrease in size of the lateral
digits of horses is partly due to this type of developmental change. Speed of develop-
ment and egg size also have important evolutionary implications. Changing the time
at which an animal becomes sexually mature can result in adults with larval character-
istics. Some animals, such as frogs and sea urchins, which usually have a larval form,
have evolved species that develop directly into the adult without a larval stage. The life
histories of animals also have implications for evolution and development.
582 Chapter 15 Evolution and development

Summary: changes in timing of development processes

increase in size early sexual maturity

faster growing regions get proportionately larger neoteny

e.g. third digit in the horse

elimination of larval stage

rapid development to adult

Summary to Chapter 15

Many developmental processes have been conserved during evolution. While many ques-
tions relating to evolution and development remain unanswered, it is clear that develop-
ment reflects the evolutionary history of ancestral embryos. The origins of multicellularity
and the embryo from a single-celled ancestor are still highly speculative. All vertebrate
embryos pass through a conserved phylotypic developmental stage, although there can
be considerable divergence both earlier and later in development. The signals involved in
patterning of vertebrate and arthropod appendages, as well as of the dorso-ventral axis,
show remarkable similarity and conservation. The pattern of Hox gene expression along
the body axis of vertebrates and arthropods is conserved, and major changes in body plan
reflect changes in both Hox gene expression and their downstream targets. Changes in
gene-­regulatory regions were crucial. Alterations in the timing of developmental events
have played an important role in evolution. Such changes can alter the overall form of an
organism as a result of differences in growth rates of different structures, and can also result
in sexual maturation at larval stages.

n End of chapter questions 6. Compare the patterns of Hox gene expression in Fig. 15.9; how
would you explain these observations in words? (As a challenge,
Long answer (concept questions) based solely on these observations, predict the consequence of mis-
expressing Antp, Ubx, and abd-A throughout the thorax and abdo-
1. Trace the possible steps in the evolution of the Metazoa from a
men of a grasshopper (or fly).) After this, refer back to Fig. 2.48:
single-celled ancestor. What were the key innovations that led to the
what would be the actual result of abd-A expression throughout the
evolution of the Metazoa, and especially of bilaterally symmetric
thorax? How might the evolution of the Hox gene targets explain the
animals?
differences between Artemia and Drosophila?)
2. Why has the phylotypic stage been so strongly conserved during
7. Drosophila larvae have no legs, yet lepidopteran larvae have legs
evolution? What events and processes are occurring at this point
on each thoracic and abdominal segment. Explain how Hox genes
in development that make this stage so central to animal develop-
and the Distal-less gene mediate this difference.
ment?
8. How could changes in Hox gene expression in lateral plate meso-
3. Give an example of a change in protein structure and function
derm explain the absence of limbs on snakes?
that occurred after gene duplication. Give an example of a change in
gene expression that occurred after gene duplication. 9. How do regulatory changes in yellow explain the pigmentation
differences in Drosophila biarmipes compared to D. melanogaster?
4. Which genes of the Drosophila Hox complex would you expect to
be most closely related, at the level of their protein sequences, based 10. Synthesize the information in Figs 15.21, 15.22, and 15.23 and
on their evolutionary history (see Fig. 15.7)? provide a narrative for the evolution of the incus and malleus bones
in the inner ear of mammals.
5. Genetic experiments in Drosophila have led to a hypothesis that
Antp is an ancestral gene, and that diversification of the segments 11. Describe what is meant by the term heterochrony and give an
posterior to T2 is a later evolutionary addition; what are these exper- example of an evolutionary change that can be described as a het-
iments, and how do they support this hypothesis? erochrony.
 Changes in the timing of developmental processes 583

12. Using examples from the text, discuss how probabilities of sur- b) embryonic precursors to gills, which do not continue to develop
vival, rates of reproduction, and optimization of reproductive effort in humans
can drive evolutionary changes in development. c) embryonic precursors to structures predominantly in the jaw
d) embryonic structures that develop into the ribs
Multiple choice (factual recall questions)
8. The evolutionary acquisition of sexual maturity by an animal
NB There is only one correct answer to each question.
while still in a larval stage is called
1. Ontogeny is a) apoptosis
a) the development of an individual organism b) metamorphosis
b) the development of the ear c) neoteny
c) the evolutionary history of a species or group d) ontogeny
d) the study of cancer
9. Which is true?
2. An example of a triploblast is a) Bat wings evolved from bird wings.
a) Hydra b) Ears evolved from jaws.
b) sponges c) Insects evolved from crustaceans.
c) Xenopus d) Limbs evolved from fins.
d) triploid plants such as seedless watermelons
10. Hedgehog family members are expressed in the _____ position
3. Considering the ancient evolutionary split between protostomes in both vertebrate and insect wings.
and deuterostomes, which of the following pairs of animals is most a) compartment boundary
closely related evolutionarily?
b) distal
a) Caenorhabditis elegans and amphioxus
c) dorsal
b) humans and Drosophila
d) posterior
c) sea urchins and humans
d) snails and starfish Multiple choice answer key
4. The evolution of the vertebrate Hox clusters probably arose 1: a, 2: c, 3: c, 4: c, 5: b, 6: a, 7: c, 8: c, 9: d, 10: d.
through
a) duplications of an ancestral cluster to give four ancestral clus-
ters, followed by tandem duplications which have apparently
n General further reading
occurred differently in each of the four clusters
Carroll, S.B.: Evo-devo and an expanding evolutionary synthesis:
b) random changes in DNA sequences followed by natural selec- a genetic theory of morphological evolution. Cell 2008, 134:
tion, leading to the independent evolution of each of the Hox 25–36.
genes from unrelated sequences Carroll, S.B., Grenier, J.K., Weatherbee, S.D.: From DNA to Diversity,
c) tandem duplications of an ancestral gene, followed by duplica- 2nd edn. Malden, MA: Blackwell Science, 2005.
tions of the whole cluster Finnerty, J.R., Pang, K., Burton, P., Paulson, D., Martindale, M.Q.:
d) the generation of four ancestral Hox genes on separate chromo- Origins of bilateral symmetry: Hox and dpp expression in a sea
somes, followed by tandem duplications to give rise to the four anemone. Science 2004, 304: 1335–1337.
Hox clusters Hoekstra, H.E., Coyne, J.A.: The locus of evolution: evo devo and
the genetics of adaptation. Evolution 2007, 61: 995–1016.
5. An animal’s ventral surface is defined by Kirschner, M., Gerhart, J.: Evolvability. Proc. Natl. Acad. Sci. USA
a) the expression of BMP family members 1998, 95: 8420–8427.
b) the position of the mouth Kirschner, M., Gerhart, J.: The Plausibility of Life. Yale University
Press, 2005.
c) the side of the animal that faces the ground
Raff, R.A.: The Shape of Life. University of Chicago Press, 1996.
d) the side opposite the central nervous system Richardson, M.K.: Vertebrate evolution: the developmental ­origins
6. The horse’s hoof is homologous to human of adult variation. BioEssays 1999, 21: 604–613.
Wray, G.A.: The evolutionary significance of cis-regulatory muta-
a) fingers or toes tions. Nat. Rev. Genet. 2007, 8: 206–216.
b) humerus
c) radius and ulna Box 15A ‘Darwin’s finches’
d) wrist bones Abzhanov, A., Kuo, W.P., Hartmann, C., Grant, B.R., Grant, P.R.,
Tabin, C.J.: The calmodulin pathway and evolution of elon-
7. In humans, the branchial arches are gated beak morphology in Darwin’s finches. Nature 2006, 442:
a) embryonic gill slits, reflecting our evolution from fish 563–567.
584 Chapter 15 Evolution and development

n Section further reading Averof, M.: Origin of the spider’s head. Nature 1998, 395:
436–437.
15.1 Genomic evidence is throwing light on the origin of Duboule, D.: A Hox by any other name. Nature 2000, 403:
metazoans 607–610.
Galant, R., Carroll, S.B.: Evolution of a transcriptional repression
Putnam, N.H., Srivastava, M., Hellsten, U., Dirks, B., Chapman, J.,
domain in an insect Hox protein. Nature 2003, 415: 910–913.
Salamov, A., Terry, A., Shapiro, H., Lindquist, E., Kapitonov, V.V.,
Pavlopoulos, A., Averof, M.: Developmental evolution: Hox pro-
et al.: Sea anemone genome reveals ancestral eumetazoan gene
teins ring the changes. Curr. Biol. 2002, 12: R291–R293.
repertoire and genomic organization. Science 2007, 317: 86–94.
Stern, D.L.: A role of Ultrabithorax in morphological differences
Srivastava, M., Begovic, E., Chapman, J., Putnam, N.H., Hellsten, U.,
between Drosophila species. Nature 1998, 396: 463–466.
Kawashima, T., Kuo, A., Mitros, T., Salamov, A., Carpenter, M.L.,
et al.: The Trichoplax genome and the nature of placozoans.
Nature 2008, 454: 955–960. 15.5 The position and number of paired appendages in
insects is dependent on Hox gene expression
15.2 Multicellular organisms evolved from single-celled Carroll, S.B., Weatherbee, S.D., Langeland, J.A.: Homeotic genes
ancestors and the regulation and evolution of insect wing number. ­Nature
Brooke, N.M., Holland, P.W.H.: The evolution of multicellular- 1995, 375: 58–61.
ity and early animal genomes. Curr. Opin. Genet. Dev. 2003, Levine, M.: How insects lose their wings. Nature 2002, 415:
6: 599–603. 848–849.
Jaegerstern, G.: The early phylogeny of the metazoa. The bilatero- Weatherbee, S.D., Carroll, S.: Selector genes and limb identity in
gastrea theory. Zool. Bidrag. (Uppsala) 1956, 30: 321–354. arthropods and vertebrates. Cell 1999, 97: 283–286.
King, N.: The unicellular ancestry of animal development. Dev. Weatherbee, S.D., Nijhout, H.F, Grunnert, L.W., Halder, G., Galant, R.,
Biol. 2004, 7: 313–325. Selegue, J., Carroll, S.: Ultrabithorax function in butterfly wings
King, N., Westbrook, M.J., Young, S.L., Kuo, A., Abedin, M., Chapman, and the evolution of insect wing patterns. Curr. Biol. 1999, 9:
J., Fairclough, S., Hellsten, U., Isogai, Y., Letunic, I., et al.: The 109–115.
genome of the choanoflagellate Monosiga brevicollis and the
origin of metazoans. Nature 2008, 451: 783–788. 15.6 The basic body plan of arthropods and vertebrates is
Miller, D.J., Ball, E.E.: Animal evolution: the enigmatic phylum similar, but the dorso-ventral axis is inverted
Placozoa revisited. Curr. Biol. 2005, 15: R26–R28. Arendt, D., Nübler-Jung, K.: Dorsal or ventral: similarities in fate
Rudel, D., Sommer, R.J.: The evolution of developmental mecha- maps and gastrulation patterns in annelids, arthropods and
nisms. Dev. Biol. 2003, 264: 15–37. chordates. Mech. Dev. 1997, 61: 7–21.
Szathmary, E., Wolpert, L.: The transition from single cells to mul- Davis, G.K., Patel, N.H.: The origin and evolution of segmenta-
ticellularity. In Genetic and Cultural Evolution of Cooperation (ed. tion. Trends Biochem. Sci. 1999, 24: M68–M72.
Hammersteen, P.) 271–289. Cambridge, MA: MIT Press, 2004. Gerhart, J., Lowe, C., Kirschner, M.: Hemichordates and the ori-
Wolpert, L.: Gastrulation and the evolution of development. Devel- gins of chordates. Curr. Opin. Genet. Dev. 2005, 15: 461–467.
opment (Suppl.) 1992, 7–13. Holley, S.A., Jackson, P.D., Sasai, Y., Lu, B., De Robertis, E.,
Wolpert, L., Szathmary, E.: Evolution and the egg. Nature 2002, Hoffman, F.M., Ferguson, E.L.: A conserved system for dorso-
420: 745. ventral patterning in insects and vertebrates involving sog and
chordin. Nature 1995, 376: 249–253.
15.3 Hox gene complexes have evolved through gene
duplication 15.7 Limbs evolved from fins
Brooke, N.M., Gacia-Fernandez, J., Holland, P.W.H.: The ParaHox Ahn, D., Ho, R.K.: Tri-phasic expression of posterior Hox genes
gene cluster is an evolutionary sister of the Hox gene cluster. during development of pectoral fins in zebrafish: implications
Nature 1998, 392: 920–922. for the evolution of vertebrate paired appendages. Dev. Biol.
Duboute, D.: The rise and fall of Hox gene clusters. Development 2008, 322: 220–233.
134: 2549–2560. Boisvert, C.A., Mark-Kurik, E., Ahlberg, P.E.: The pectoral fin
Ferrier, D.E.K., Holland, P.W.H.: Ancient origin of the Hox gene of Panderichthys and the origin of digits. Nature 2008, 456:
cluster. Nat. Rev. Genet. 2001, 2: 33–34. 636–638.
Prince, V.E., Pickett, F.B.: Splitting pairs: the diverging fates of Capdevila, J., Izpisua Belmonte, J.C.: Perspectives on the evolu-
duplicated genes. Nat. Rev. Genet. 2002, 3: 827–837. tionary origin of tetrapod limbs. J. Exp. Zool. (Mol. Dev. Evol.)
Valentine, J.W., Erwin, D.H., Jablonski, D.: Developmental evo- 2000, 288: 287–303.
lution of metazoan bodyplans: the fossil evidence. Dev. Biol. Coates, M.I., Jeffery, J.E., Rut, M.: Fins to limbs: what the fossils
1996, 173: 373–381. say. Evol. Dev. 2002, 4: 390–401.
Cohn, M.J., Tickle, C.: Developmental basis of limblessness and
15.4 Changes in Hox genes generated the elaboration of axial patterning in snakes. Nature 1999, 399: 474–479.
vertebrate and arthropod body plans Ruvinsky, I., Gibson-Brown, J.J.: Genetic and developmental bases
Akam, M.: Hox genes and the evolution of diverse body plans. of serial homology in vertebrate limb evolution. Development
Phil. Trans. R. Soc. Lond. B 1995, 349: 313–319. 2000, 127: 5211–5244.
 Changes in the timing of developmental processes 585

Sordino, P., van der Hoeven, F., Duboule, D.: Hox gene expression genetic architecture of skeletal convergence and sex determina-
in teleost fins and the origin of vertebrate digits. Nature 1995, tion in ninespine sticklebacks. Curr. Biol. 2009, 19: 1140–1145.
375: 678–681. Yamamoto, Y., Stock, D.W., Jeffery, W.R.: Hedgehog signaling
Woltering, J.M., Vonk, F.J., Müller, H., Bardine, N., Tuduce, I.L., controls eye degeneration in blind cavefish. Nature 2004, 431:
de Bakker, M.A., Knöchel, W., Sirbu, I.O., Durston, A.J., Richardson, 844–847.
M.K.: Axial patterning in snakes and caecilians: evidence for
an alternative interpretation of the Hox code. Dev. Biol. 2009, 15.10 Embryonic structures have acquired new functions
332: 82–89. during evolution
Cerny, R., Lwigale, P., Ericsson, R., Meulemans, D., Epperlein, H.H.
15.8 Vertebrate and insect wings make use of evolutionarily Bronner-Fraser, M.: Developmental origins and evolution of
conserved developmental mechanisms jaws: new interpretation of ‘maxillary’ and ‘mandibular’. Dev
Panganiban, G., Irvine, S.M., Lowe, C., Roehl, H., Corley, L.S., Biol. 2004, 276: 225–236.
­Sherbon, B., Grenier, J.K., Fallon, J.F., Kimble, J., Walker, M., Cohn, M.J.: Lamprey Hox genes and the origin of jaws. Nature
Wray, G.A., Swalla, B.J., Martindale, M.Q., Carroll, S.B.: The ori- 2002, 416: 386–387.
gin and evolution of animal appendages. Proc. Natl. Acad. Sci. De Robertis, E.M.: Evo-devo: variations on ancestral themes. Cell
USA 1997, 94: 5162–5166. 2008, 132: 185–195.
Erwin, D.H.: Early origin of the bilaterian developmental toolkit.
15.9 The evolution of developmental differences can be Philos. Trans. R. Soc. Lond. B Biol. Sci. 2009, 364: 2253–2261.
based on changes in just a few genes Romer, A.S.: The Vertebrate Body. Philadelphia: W.B. Saunders,
Chan, Y.F., Marks, M.E., Jones, F.C., Villarreal, G. Jr., Shapiro, M.D., 1949.
Brady, S.D., Southwick, A.M., Absher, D.M., Grimwood, J.,
Schmutz, J., et al.: Adaptive evolution of pelvic reduction in 15.11 Evolution can be due to changes in the timing of
sticklebacks by recurrent deletion of a Pitx1 enhancer. Science developmental events
2010, 327: 302–305. Alberch, P., Alberch, J.: Heterochronic mechanisms of morphologi-
Gompel, N., Carrol, S.B.: Genetic mechanisms and constraints cal diversification and evolutionary change in the neotropical sal-
governing the evolution of correlated traits in drosophilid flies. amander Bolitoglossa occidentales (Amphibia: Plethodontidae).
Nature 2003, 424: 931–935. J. Morphol. 1981, 167: 249–264.
Gompel, N., Prud’homme, B., Wittkopp, P.J., Kassner, V.A., Carroll, Lande, R.: Evolutionary mechanisms of limb loss in tetrapods.
S.B.: Chance caught on the wing: cis-regulatory evolution and Evolution 1978, 32: 73–92.
the origin of pigment patterns in Drosophila. Nature 2005, 433: Raynaud, A.: Developmental mechanism involved in the embry-
481–487. onic reduction of limbs in reptiles. Int. J. Dev Biol. 1990, 34:
Jeffery, W.R.: Evolution and development in the cavefish Astya- 233–243.
nax. Curr. Top. Dev. Biol. 2009, 86: 191–221. Wray, G.A., Raff, R.A.: The evolution of developmental strategy in
Shapiro, M.D., Marks, M.E., Peichel, C.L., Blackman, B.K., Nereng, marine invertebrates. Trends Evol. Ecol. 1991, 6: 45–56.
K.S., Jonsson, B., Schluter, D., Kingsley, D.M.: Genetic and devel-
opmental basis of evolutionary pelvic reduction in threespine 15.12 The evolution of life histories has implications for
sticklebacks. Nature 2004, 428: 717–723. development
Shapiro, M.D., Summers, B.R., Balabhadra, S., Aldenhoven, J.T., Partridge, L., Harvey, P.: The ecological context of life history evo-
Miller, A.L., Cunningham, C.B., Bell, M.A., Kingsley, D.M.: The lution. Science 1988, 241: 1449–1455.
This page intentionally left blank
Glossary

In insects the body is divided into three distinct parts, the head at the The amnioserosa of the Drosophila embryo is an extra-embryonic
anterior end, followed by the thorax and the posterior abdomen. membrane on the dorsal side of the embryo.

Abembryonic pole see embryonic–abembryonic axis. Amniotes are vertebrates whose embryos have an amnion. They
comprise the mammals, birds, and reptiles.
Accretionary growth is an increase in mass due to secretion of large
quantities of extracellular matrix by cells, which enlarges the volume Anamniotes are vertebrates whose embryos do not have an amnion.
of extracellular space. It occurs, for example, in cartilage and bone, They comprise the fish and amphibians.
where most of the tissue mass is extracellular.
An androgenetic embryo is an embryo in which the two sets of
Acheiropodia is a rare, inherited human condition in which homologous chromosomes are both paternal in origin.
structures below the elbow or knee do not develop; but there is no Angioblasts are the mesodermal precursor cells that will give rise to
other defect. blood vessels.
Achondroplasia is a form of dwarfism in which the limbs are short Angiogenesis is the process by which small blood vessels sprout
in relation to the rest of the body. from the larger vessels.
The acron is a specialized structure associated with the most anterior The animal region of an egg is the end of the egg where the nucleus
region of the Drosophila embryo. resides, usually away from the yolk. The most terminal part of this
The acrosomal reaction is the release of enzymes and other proteins region is the animal pole, which is directly opposite the vegetal pole
from the acrosomal vesicle or acrosome of the sperm head that at the other end of the egg. In Xenopus the pigmented animal half is
occurs once a sperm has bound to the outer surface of the egg. It called the animal cap.
helps the sperm to penetrate the outer layers of the egg. The animal–vegetal axis runs from the animal pole to the vegetal
pole in an egg or early embryo.
Actin filaments or microfilaments are one of the three principal
protein filaments of the cytoskeleton. They are involved in cell Aniridia is the absence of an iris in the eye.
movement and changes in cell shape. Actin filaments are also part
The antennapedia complex comprises one part of the Hox gene
of the contractile apparatus of muscle cells.
complex in Drosophila
Gene-regulatory proteins that turn genes on are known as activators.
The anterior visceral ectoderm (AVE) is an extra-embryonic tissue
Actomyosin is an assembly of actin filaments and the motor protein in the early mouse embryo that is involved in inducing anterior
myosin that can undergo contraction. regions of the embryo.

In plants, the adaxial–abaxial axis runs from the center of the plant The antero-posterior axis defines which is the ‘head’ end and
stem to the circumference. In plant leaves it runs from the upper which is the ‘tail’ end of an animal. The head is anterior and the tail
surface to the lower surface (dorsal to ventral). posterior. In the vertebrate limb, this axis runs from the thumb to
little finger.
Adherens junctions are a type of adhesive cell junction in which
the adhesion molecules linking the two cells together are cadherins Anticlinal cell divisions are divisions in planes at right angles to the
that are linked intracellularly to the actin cytoskeleton. outer surface of a tissue.

The allantois is a set of extra-embryonic membranes that develops Antisense RNA is an RNA complementary in sequence to an mRNA,
in many vertebrate embryos. In bird and reptile embryos it acts as a which blocks expression of a protein by binding to its mRNA and
respiratory surface, while in mammals its blood vessels carry blood blocking translation.
to and from the placenta. The apical–basal axis of a plant is the axis running from shoot tip to
root tip.
An allele is a particular version of a gene. In diploid organisms,
two alleles of each gene are present (one on each chromosome of a The fact that buds behind the tip of a plant shoot will not form side
homologous pair), which may or may not be identical. stems when the tip is intact, is known as apical dominance. It is due
to the production of the hormone auxin by the shoot tip.
The amnion is an extra-embryonic membrane in birds, reptiles, and
mammals, which forms a fluid-filled sac that encloses and protects The apical ectodermal ridge or apical ridge is a thickening of the
the embryo. It is derived from extra-embryonic ectoderm and ectoderm at the distal end of the developing chick and mammalian
mesoderm. limb bud.
588 Glossary

An apical meristem is the region of dividing cells at the tip of a In amphibian limb regeneration, a blastema is formed from the
growing shoot or root. dedifferentiation and proliferation of cells beneath the wound
epidermis, and gives rise to the regenerated limb.
Apoptosis or programmed cell death is a type of cell death that
occurs widely during development. In programmed cell death, a cell The blastocoel is the fluid-filled cavity that develops in the interior of
is induced to commit ‘suicide,’ which involves fragmentation of the a blastula.
DNA and shrinkage of the cell. These apoptotic cells are removed by
The blastocyst stage of a mammalian embryo corresponds in form to
the body’s scavenger cells and, unlike necrosis, their death does not
the blastula stage of other animal embryos, and is the stage at which
cause damage to surrounding cells.
the embryo implants in the uterine wall.
The archenteron is the cavity formed inside the embryo when the
A blastoderm is a post-cleavage embryo composed of a solid layer
endoderm and mesoderm invaginate during gastrulation. It forms
of cells rather than a spherical blastula, as in early chick, zebrafish,
the gut.
and Drosophila embryos. The chick blastoderm is also known as the
The area opaca is the outer dark area of the chick blastoderm. blastodisc.

The area pellucida is the central clear area of the chick blastoderm. Blastomeres are the cells derived from cleavage of the early embryo.

Asymmetric cell divisions or asymmetric divisions are cell The blastopore is the slit-like or circular invagination on the surface
divisions in which the daughter cells are different from each other of amphibian and sea-urchin embryos, where the mesoderm and
because some cytoplasmic determinant(s) have been distributed endoderm move inside the embryo at gastrulation.
unequally between them. The blastula stage in animal development is the outcome of cleavage.
A developmental process is said to be proceeding autonomously The blastula is a hollow ball of cells, composed of an epithelial layer
when it can continue without a requirement for extracellular signals of small cells enclosing a fluid-filled cavity—the blastocoel.
to be continuously present. See also cell autonomous. Blastula organizer see Nieuwkoop center.
Aux/IAA proteins are transcriptional repressors that block the The block to polyspermy is the mechanism used by many animal
expression of auxin-responsive genes in the absence of the plant eggs to prevent fertilization by more than one sperm. In sea urchins
hormone auxin. and Xenopus it is composed of two stages—an immediate rapid block
The plant hormone auxin is a small organic moelcule that is to polyspermy and a subsequent slow block to polyspermy.
important in almost all aspects of plant development. It acts by The body plan describes the overall organization of an organism;
regulating the expression of auxin-responsive genes by means of for example, the position of the head and tail, and the plane of
its effects on Aux/IAA proteins. bilateral symmetry, where it exists. The body plan of most animals
Auxin-response factors are gene regulatory proteins that are is organized around two main axes: the antero-posterior axis and the
responsive to the plant hormone auxin. dorso-ventral axis.

AVE see anterior visceral endoderm. The brachial region of a vertebrate embryo is the region that includes
the forelimb or that gives rise to the structures of the forelimb.
Axial structures are those that form along the main axis of the
body, such as the notochord, vertebral column, and neural tube in The branchial arches are structures that develop on each side of the
vertebrates. embryonic head and give rise to the gill arches in fishes, and to the
jaws and other facial structures in other vertebrates.
Axons are long cell processes of neurons that conduct nerve impulses
Branching morphogenesis describes the development and growth
away from the cell body. The end of an axon, the axon terminal,
of stuctures such as blood vessels or the bronchi and bronchioles
forms contacts (synapses) with other neurons, muscle cells, or
of the lung, which develop by the successive branching of a tube of
glandular cells.
epithelium.
The protein β-catenin functions both as a transcriptional co-activator
The cadherins are a family of cell-adhesion molecules with important
and as one of the proteins present at cell junctions. In its role as a
roles in development.
transcriptional co-activator it is activated in early development in
many vertebrates as the end result of a Wnt signaling pathway. The cambium in plants is a ring of meristem in the stem that gives
rise to new stem tissue, which increases the diameter of the stem.
The basal lamina or basement membrane is a sheet of extracellular
matrix that separates an epithelial layer from the underlying tissues. The canonical Wnt/β-catenin pathway is an intracellular signaling
For example, the epidermis of the skin is separated from the dermis pathway stimulated by members of the Wnt family of signal proteins
by a basal lamina. that leads to the stabilization of β-catenin and its entry into nuclei,
where it acts as a transcriptional co-activator.
Animals in which the only axis of symmetry is the central axis
running from head to tail are said to possess bilateral symmetry. The functional maturation of sperm after they have been deposited in
The two sides of the body are mirror images of each other. the female reproductive tract is known as capacitation.

The bithorax complex comprises one part of the Hox gene complex Caspases are intracellular proteases, some of which are involved in
in Drosophila. apoptosis.
 Glossary 589

Cell-adhesion molecules bind cells to each other and to the A chimeric organism or tissue (a chimera) is made up of cells
extracellular matrix. The main classes of adhesion molecules from two or more different sources, and thus of different genetic
important in development are the cadherins, the immunoglobulin constitutions.
superfamily, and the integrins.
ChIP-chip and ChIP-seq are techniques for determining which sites
Cell adhesiveness is the property of cells that causes them to stick in chromosomal DNA are bound by particular proteins in vivo. They
together. involve immunoprecipitation of chromatin fragments by antibodies
against the protein of interest followed by analysis of the bound DNA
The effects of a gene are cell-autonomous if they only affect the cell
by microarrays (ChIP-chip) or DNA sequencing (ChIP-seq).
the gene is expressed in.
Chondrocytes are differentiated cartilage cells.
The cell body is the part of a neuron that contains the nucleus, and
from which the axon and dendrites extend. The chorion is the outermost of the extra-embryonic membranes
in birds, reptiles, and mammals. It is involved in respiratory gas
Cell–cell interaction and cell–cell signaling are general terms
exchange. In birds and reptiles it lies just beneath the shell. In
to describe the various types of intercellular communication by
mammals it is part of the placenta and is also involved in nutrition and
which one cell influences the behavior of another cell. Cells can
waste removal. The chorion of insect eggs has a different structure.
communicate with each other via cell contact or by the secretion of
signaling molecules that influence the behavior of other cells nearby Chromatin is the material of which chromosomes are made. It is
or at a distance. composed of DNA and protein. Enzyme complexes called chromatin-
remodeling complexes can act on chromatin to modify it and alter
The cell cycle is the sequence of events by which a cell duplicates
the ability of the DNA to be transcribed.
itself and divides in two.
The circadian clock is an internal 24-hour timer present in living
During cell differentiation, cells become functionally and structurally organisms that causes many metabolic and physiological processes,
different from one another and become distinct cell types, such as including the expression of some genes, to vary in a regular manner
muscle or blood cells. throughout the day.
Growth can occur by cell enlargement, which is an increase in The cis-regulatory control region of a gene comprises the sequences
cell size without division. This type of growth is common in plants flanking the gene and containing sites at which the expression
especially. of that gene can be controlled. Many control regions contain a
Cell-lineage restriction occurs when all the descendants of a variety of different cis-regulatory modules, which are short regions
particular group of cells remain within a ‘boundary’ and never mix containing multiple binding sites for different transcription factors;
with an adjacent group of cells of a different lineage. Compartment the combination of factors bound determines whether the gene is
boundaries in insect development are boundaries of lineage switched on or off.
restriction. Cleavage occurs after fertilization and is a series of rapid cell
Cell motility describes the ability of cells to move, to contract or to divisions without growth that divides the embryo up into a number
change shape. of small cells.

Growth, or an increase in mass, can occur by cell proliferation—the A clone is a collection of genetically identical cells derived from
growth and division of cells to form new cells. a single cell by repeated cell division, or the genetically identical
offspring of a single individual produced by asexual reproduction or
Cell-replacement therapy aims to correct disease by replacing artifical cloning techniques.
damaged cells by new healthy cells, including stem cells.
Cloning is the procedure by which an individual genetically identical
Cell senescence see senescence. to a ‘parent’ is produced by transplantation of a parental somatic cell
The centrosome of a cell is the organizing center for microtubule nucleus into an unfertilized oocyte.
growth. It duplicates before mitosis or meiosis, and each centrosome A co-activator is a gene-regulatory protein that promotes gene
forms one end of the microtubule spindle. It consists of a pair of expression but does not bind to DNA itself. It binds to and influences
centrioles, one of which gives rise to the basal body of the primary the behavior of other DNA-binding transcription factors.
cilium in many types of differentiated cells.
The coding region of a gene is that part of the DNA that encodes a
The checkpoints that occur during the cell cycle enable cells to polypeptide or functional RNA.
monitor progress through the cycle and ensure that a previous stage
The correspondence between the order of Hox genes on a
has been completed before embarking on the next.
chromosome and their temporal and spatial order of expression in the
The chemoaffinity hypothesis proposes that each retinal neuron embryo is known as co-linearity.
carries a chemical label that enables it to connect reliably with an
Colony-stimulating factors are proteins that drive the differentiation
appropriately labeled cell in the optic tectum.
of blood cells.
A chemoattractant is a molecule that attracts cells to move towards it.
Induction of cell differentiation in some tissues depends on a
A chemorepellent is a molecule that repels cells, causing them to community effect, in that there have to be a sufficient number of
move away from it. responding cells present for differentiation to occur.
590 Glossary

Compaction of the mouse embryo occurs during early cleavage. The Cytoplasmic determinants are cytoplasmic proteins that influence
blastomeres flatten against each other and microvilli become confined development.
to the outer surface of the ball of cells.
Cytoplasmic localization is the non-uniform distribution of some
Compartments are discrete areas of an embryo that contain all the factor or determinant in a cell’s cytoplasm, so that when the cell
descendants of a small group of founder cells and which show cell- divides, the determinant is unequally distributed to the daughter cells.
lineage restriction. Cells in compartments respect the compartment
A cell’s cytoskeleton is the network of protein filaments—
boundary and do not cross over into an adjacent compartment.
microfilament, microtubules, and intermediate filaments—that gives
Compartments tend to act as discrete developmental units.
cells their shape, enables them to move, and provides tracks for
Competence is the ability of a tissue to respond to an inducing signal. transport of materials in the cell.
Embryonic tissues only remain competent to respond to a particular
The long-lived dauer larvae of Caenorhabditis elegans are a response
signal for a limited period of time.
to starvation conditions and neither eat nor grow until food is again
Condensation describes the increased packing together of cells to available.
form structures such as cartilages (precursors of bones).
Dedifferentiation is loss of the structural characteristics of a
Conditional mutations are mutations whose effects only become differentiated cell, which may result in the cell then differentiating
manifest in a particular condition, e.g. at a higher temperature than into a new cell type.
normal.
The deep layer of the zebrafish blastoderm is the layer several cells
Contralateral refers to the opposite side of the body. deep under the outer enveloping layer and gives rise to the embryo.

A control region of a gene is a region to which regulatory proteins Delamination is the process in which epithelial cells leave an
bind and so determine whether or not the gene is transcribed. epithelium as individual cells. It occurs, for example, in the primitive
streak and in the movement of neural crest cells out of the neural tube.
Cooperativity describes the phenomenon in which the binding of
a molecule to one site on another molecule (such as the binding Dendrites are extensions from the body of a nerve cell that receive
of a transcription factor to a site in DNA) makes the binding of stimuli from other nerve cells.
subsequent molecules to other sites on the same molecule more likely
Denticles are small tooth-like outgrowths of the cuticle on insect
to occur.
larvae.
Convergent extension is the process by which a sheet of cells
The dermatome is the region of the somite that will give rise to the
changes shape by extending in one direction and narrowing—
dermis.
converging—in a direction at right angles to the extension, caused
by the cells intercalating between each other. The dermis of the skin is the connective tissue beneath the
epidermis, from which it is separated by a basal lamina.
A co-repressor is a gene-regulatory protein that suppresses gene
expression but does not bind to DNA itself. It binds to and influences The dermomyotome is the region of the somite that will give rise to
the behavior of other DNA-binding transcription factors. both muscle and dermis.

Cortical granules are granules present in the cortex of the some eggs, Determinants are cytoplasmic factors (e.g. proteins and RNAs) in the
which release their contents on fertilization to form the fertilization egg and in embryonic cells that can be asymmetrically distributed at
membrane. cell division and so influence how the daughter cells develop.

Cortical rotation occurs immediately after an amphibian egg is Determination implies a stable change in the internal state of a cell
fertilized. The egg cortex, an actin-rich layer of cytoplasm lying such that its fate is now fixed, or determined. A determined cell will
immediately below the surface, rotates with respect to the underlying follow that fate when grafted into other regions of the embryo.
cytoplasm, toward the point of sperm entry.
Deuterostomes are those animals, such as chordates and
The cortex of a plant stem or root is the tissue between the epidermis echinoderms, that have radial cleavage of the egg, and in which the
and central vascular tissue. primary invagination of the gut at gastrulation forms the anus, with
the mouth developing independently.
A cotyledon is the part of the plant embryo that acts as a food
storage organ. Developmental biology comprises the study of the development of
a multicellular organism from its origin in a single cell—the fertilized
The Cre/loxP system is a transgenic modification of mice that
egg—until it is fully adult.
enables a gene to be expressed in a particular tissue or at a particular
time in development. A development constraint is a pre-existing developmental process
that constrains the evolution of new forms.
Cumulus cells are somatic cells that surround the developing
mammalian oocyte and are shed with it at ovulation. Genes that specifically control a developmental process are known as
developmental genes.
Cyclins are proteins that periodically rise and fall in concentration
during the cell cycle and are involved in controlling progression The differential adhesion hypothesis explains the movement of cells
through the cycle. They act by binding to and activating in developing embryos, e.g. at gastrulation, in terms of the differences
cyclin-dependent kinases (Cdks). in the strength of adhesion of different types of cells for each other.
 Glossary 591

Differential gene expression describes the turning on and off Embryogenesis is the process of development of the embryo from the
of different genes in different cells in a multicellular organism, fertilized egg.
thus generating cells with different developmental and functional
Embryology is the study of the development of an embryo.
properties.
The embryonic–abembryonic axis in the mammalian blastocyst runs
Diploblasts are animals with two germ layers (endoderm and
from the site of attachment of the inner cell mass—the embryonic
ectoderm) only and include cinidarians such as Hydra and jellyfish.
pole—to the opposite pole, the abembryonic pole.
Diploid cells contain two sets of homologous chromosomes, one from
The embryonic ectoderm is the name given to the mouse epiblast
each parent, and thus two copies of each gene.
once it has developed into an epithelial sheet.
Directed dilation is the extension of a tube-like structure at each
Embryonic endoderm is the layer of cells in the mammalian embryo
end due to hydrostatic pressure, the direction of extension reflecting
that gives rise to the endoderm of the embryo.
greater circumferential resistance to expansion.
The embryonic organizer is an alternative name for the Spemann
The distal end of a structure such as a limb is the end furthest away
organizer in amphibians and similar organizing regions in
from the point of attachment to the body.
other vertebrates (such as the node in chick) that can direct the
The distal visceral endoderm (DVE) is the endoderm located at the development of a complete embryo.
distal end of the cup-shaped mouse epiblast, which moves away from
Embryonic stem cells (ES cells) are derived from the inner cell mass
the posterior side of the embryo upon which it becomes the anterior
of a mammalian embryo, usually mouse, and can be indefinitely
visceral endoderm.
maintained in culture. When injected into another blastocyst, they
DNA microarrays, also known as DNA chips, are arrays of combine with the inner cell mass and can potentially contribute to all
oligonucleotides that are used to detect and measure the expression the tissues of the embryo.
of large numbers of genes simultaneously, by hybridization of cellular In the chick embryo, the hypoblast underlying the epiblast is replaced
RNA or cDNA. by a layer of cells called the endoblast that grows out from the
A dominant allele is one that determines the phenotype even when posterior marginal zone prior to primitive streak formation.
present in only a single copy. The endocardium is the inner endothelial layer of the developing
A dominant-negative mutation inactivates a particular cellular heart.
function by the production of a defective RNA or protein molecule Endochondral ossification is the replacement of cartilage with bone
that blocks the normal function of the gene product. in the growth plates of vertebrate embryonic skeletal elements, such
Dorsalized embryos develop greatly increased dorsal regions at the as the long bones of the limbs.
expense of ventral regions. The endoderm is the germ layer that gives rise to the gut and
Dorsalizing factors in vertebrate embryos are proteins that promote associated organs, such as the lungs and liver in vertebrates.
the formation of dorsal structures. The endodermis is a tissue layer in plant roots interior to the cortex
The dorso-ventral axis defines the relation of the upper surface and outside the vascular tissue.
or back (dorsal) to the under surface (ventral) of an organism or Endomesoderm is a tissue that can give rise to both mesoderm and
structure. The mouth is always on the ventral side. endoderm.
Dosage compensation is the mechanism that ensures that, although The endosperm in higher plant seeds is a nutritive tissue that serves
the number of X chromosomes in males and females is different, as a source of food for the embryo.
the level of expression of X-chromosome genes is the same in both
sexes. Mammals, insects, and nematodes all have different dosage The endothelium is the epithelium lining blood vessels.
compensation mechanisms. Enhancers are sites in gene control regions to which activating
proteins bind to switch on the gene, especially in respect of highly
Ecdysis is a type of molting in arthropods in which the external
regulated tissue-specific genes.
cuticle is shed to allow for growth.
The enhancer-trap technique is used in Drosophila to turn on
Ecdysone is a steroid hormone in insects that is responsible
the expression of a specific gene in a particular tissue or stage in
for initiating molting and also the transition to pupation and
development.
metamorphosis.
Ephrins and their receptors, Eph receptors, are cell-surface
The ectoderm is the germ layer that gives rise to the epidermis and
molecules involved in delimiting compartments in rhombomeres and
the nervous system.
in axonal guidance. Interactions of ephrins and their receptors can
An egg chamber in a female Drosophila is the structure within which cause repulsion of cells or attraction and adhesion.
an oocyte develops surrounded by its nurse cells and follicle cells.
The epiblast of mouse and chick embryos is a group of cells within
The egg cylinder in early post-implantation mouse embryogenesis is the blastocyst or blastoderm, respectively, that gives rise to the
the cylindrical structure comprising the epiblast covered by visceral embryo proper. In the mouse, it develops from cells of the inner cell
endoderm. mass.
592 Glossary

Epiboly is the process during gastrulation in which the ectoderm Forward genetics describes the type of genetic experiment in which a
extends to cover the whole of the embryo. mutant organism is first identified by its unusual phenotype and then
genetic experiments are done to discover which gene is responsible
The epidermis in vertebrates, insects, and plants is the outer layer
for the mutant phenotype.
of cells that forms the interface between the organism and its
environment. Its structure is quite different in the different organisms. Plant organs such as flowers and leaves each develop from a small
number of founder cells that derive from the apical meristem.
Epigenetic mechanisms of gene regulation involve the modification
of chromatin by, e.g. DNA methylation, histone methylation, and The gametes are the cells that carry the genes into the next
histone acetylation. generation—in animals they are the eggs and sperm.
Epimorphosis is a type of regeneration in which the regenerated A ganglion mother cell is formed by division of a neuroblast in
structures are formed by new growth. Drosophila and gives rise to neurons.
Epithelial cells can undergo an epithelial-to-mesenchymal transition Gap genes are zygotic genes coding for transcription factors
in which they lose adhesiveness and detach from the epithelium as expressed in early Drosophila development that subdivide the embryo
single cells. into regions along the antero-posterior axis.
ES cells see embryonic stem cells. The gastrula is the stage in animal development when prospective
Euchromatin is chromatin that is in a state in which the DNA in it endodermal and mesodermal cells of the blastula or blastoderm move
can be transcribed. inside the embryo.

An epithelium is said to evaginate when it forms a tubular outgrowth Gastrulation is the process in animal embryos in which prospective
from the surface. endodermal and mesodermal cells move from the outer surface of the
embryo to the inside, where they give rise to internal organs.
Evo-devo is an informal term for the study of the evolution of
development. Gene knockdown see gene silencing

Extra-embryonic ectoderm in mammals contributes to the formation Gene knock-in refers to the introduction of a new functional
of the placenta. gene into the genome using techniques involving homologous
recombination and transgenesis.
Extra-embryonic membranes are membranes external to the
embryo proper that are involved in the protection and nutrition of Gene knock-out refers to the complete and permanent inactivation of
the embryo. In mammals they include the amnion, chorion, and a particular gene in an organism by means of genetic manipulation.
placental tissues. Gene-regulatory proteins are proteins that bind to control regions in
The fate of cells describes what they will normally develop into. By DNA and help to switch genes on and off.
marking cells in the embryo, a fate map of embryonic regions can be When genes are switched off by microRNAs, RNA interference, or
constructed. Having a particular fate does not, however, imply that a modifications to chromatin, this is known as gene silencing. Unlike
cell could not develop differently if placed in a different environment. gene knock-out, it does not affect the structure of the gene itself but
Fertilization is the fusion of sperm and egg to form the zygote. prevents its transcription or the translation of the mRNA.

The fertilization membrane is formed by the eggs of some species General transcription factors are gene-regulatory proteins common
after fertilization to prevent entry of further sperm. to the expression of many different genes; they bind to RNA
polymerase and position it in the correct place of the DNA to start
A file of cells in a plant root is a vertical column of cells that
transcription.
originates from a single initial in the root meristem.
The genome of an organism contains a generative program rather
The floor plate is a small region of the developing neural tube at the
than a ‘blueprint’ for development. This means that it contains
ventral midline that is composed of non-neural cells. It is involved in
instructions for determining when and where the proteins that control
patterning the ventral part of the neural tube.
cell behavior, and thus development, are made.
A floral meristem is a region of dividing cells at the tip of a shoot
Genetic equivalence describes the fact that all the somatic cells in
that gives rise to a flower.
the body of a multicellular organism contain the same set of genes.
The individual parts of a flower develop from floral organ primordia
The genital ridge in vertebrates is the region of mesoderm lining the
generated by the floral meristem and are given their individual
abdominal cavity from which the gonads develop.
identities by the expression of floral organ identity genes.
Genomic imprinting see imprinting.
The follicle is the structure in the ovary that contains the egg cell
and its supporting somatic cells. In Drosophila, follicle cells are The genotype is a description of the exact genetic constitution of
somatic cells that surround the oocyte and nurse cells during egg a cell or organism in terms of the alleles it possesses for any given
development. gene.

The forebrain or prosencephalon is the anterior part of the embryonic The germarium in the adult female Drosophila is a reproductive
vertebrate brain that will give rise to the cerebral hemispheres, and structure containing stem cells that give rise to a succession of egg
the thalamus and the hypothalamus. chambers, each containing an oocyte.
 Glossary 593

The germ band is the name given to the ventral blastoderm of The anterior end of the notochord that projects into the head of
the early Drosophila embryo, from which most of the embryo will mammalian and avian embryos is called the head process.
eventually develop.
The heart stage is a stage in embryogenesis in dicotyledonous plants
Germ cells are those cells in an animal that give rise to eggs and sperm. in which the cotyledons and embryonic root are starting to form,
giving a heart-shaped embryo.
The germ layers refer to the regions of the early animal embryo that
will give rise to distinct types of tissue. Most animals have three germ The Hedgehog-signaling protein of Drosophila is a member of an
layers—ectoderm, mesoderm, and endoderm. important family of developmental signaling proteins that includes
Sonic hedgehog in vertebrates.
The germline cells give rise to the gametes—eggs and sperm.
Hematopoiesis is the process by which all the blood cells are derived
A germline cyst is the 16-cell structure in adult female Drosophila that
from a pluripotent stem cell. This occurs mainly in the bone marrow
contains an oocyte precursor and nurse cells, before meiosis occurs.
and is controlled by proteins called hematopoietic growth factors,
Germplasm is the special cytoplasm in some animal eggs, such as which induce differentiation into the various blood cell types.
those of Drosophila, that is involved in the specification of germ cells.
A hemisegment in Drosophila is the lateral half of a segment, one
Glia or glial cells are the non-neuronal supporting cells of the on each side of the midline, and is the developmental unit for the
nervous system, such as the Schwann cells of the peripheral nervous nervous system.
system and the astrocytes of the brain.
Hensen’s node is a condensation of cells at the anterior end of the
The globular stage of a plant embryo is a ball of around 32 cells. primitive streak in chick and mouse embryos. It corresponds to the
Glycosylation is the addition of carbohydrate side-chains to a protein, Spemann organizer in amphibians. Cells of the node give rise to
a common modification in membrane and extracellular proteins that the prechordal plate and the notochord in chick embryos and to the
occurs after their translation. notochord in mammals.

Gonadotropin-releasing hormone (GnRH) is a protein hormone An hermaphrodite is an organism that possesses both male and
secreted by the hypothalamus and acts on the pituitary to stimulate female gonads and produces both male and female gametes.
the release of gonadotropins, which in turn increase the production of Heterochromatin is the state of chromatin in which transcription of
the steroid sex hormones at puberty. the DNA is not possible.
The gonads are the reproductive organs of animals. Heterochrony is an evolutionary change in the timing of
Growth is an increase in size, which occurs by cell multiplication, developmental events. A mutation that changes the timing of a
increase in cell size and deposition of extracellular material. developmental event is called a heterochronic mutation.

Axons of developing neurons extend by means of a growth cone at A diploid individual is heterozygous for a given gene when it carries
their tip. The growth cone both crawls forward on the substrate and two different alleles of that gene.
senses its environment by means of filopodia. The hindbrain or rhomboencephalon is the most posterior part of the
Growth hormone is a protein hormone produced by the pituitary embryonic brain and gives rise to the cerebellum, the pons, and the
gland that is essential for the post-embryonic growth of humans medulla oblongata.
and other mammals. It is produced as a result of stimulation of the The homeobox is a region of DNA in homeotic genes that encodes a
pituitary by growth hormone-releasing hormone, which is produced DNA-binding domain called the homeodomain. Genes containing this
by the hypothalamus. motif are known generally as homeobox genes. The homeodomain is
Growth of vertebrate long bones occurs at the cartilaginous growth present in a large number of transcription factors that are important in
plates. The cartilage grows and is eventually replaced by bone by the development, such as the products of the Hox genes and the Pax genes.
process of endochondral ossification.
Homeosis is the phenomenon in which one structure is transformed
A gynogenetic embryo is one in which the two sets of homologous into another, homologous, structure. An example of a homeotic
chromosomes are both maternal in origin. transformation is the development of legs in place of antennae in
Drosophila as a result of mutation in a homeotic gene.
Haploid cells are derived from diploid cells by meiosis and
contain only one set of chromosomes (half the diploid number of Homeotic selector genes in Drosophila are genes that specify
chromosomes), and thus contain only one copy of each gene. In most the identity and developmental pathway of a group of cells. They
animals the only haploid cells are the gametes—the sperm or egg. encode homeodomain transcription factors and act by controlling the
expression of other genes. Their expression is required throughout
The head of a bilaterally symmetrical animal, such as an insect or
development. The Drosophila gene engrailed is an example of a
a vertebrate, is located at the anterior end of the body and typically
homeotic selector gene.
houses the brain (or equivalent), various sense organs, and the
mouth. Homologous genes share significant similarity in their nucleotide
sequence and are derived from a common ancestral gene.
The head fold is an infolding of the three germ layers in the head
region of the gastrula in chick and mammalian embryos that indicates Homologous recombination is the recombination of two DNA
the start of the pharynx and the foregut. molecules at a specific site of sequence similarity.
594 Glossary

Homology refers to morphological or structural similarity due to The inner cell mass of the early mammalian embryo is derived from
common ancestry. the inner cells of the morula, which form a discrete mass of cells in
the blastocyst. Some of the cells of the inner cell mass give rise to the
A diploid individual is homozygous for a given gene when it carries
embryo proper.
two identical alleles of that gene.
In situ hybridization is a technique used to detect where in the
Hox genes are a family of homeobox-containing genes that are
embryo particular genes are being expressed. The mRNA that is
present in all animals (as far as is known) and are involved in
being transcribed is detected by its hybridization to a labeled single-
patterning the antero-posterior axis. They can be clustered on
stranded complementary DNA probe.
the chromosomes in one or more gene complexes. Combinatorial
expression of different Hox genes characterizes different regions or In animals in which the larva goes through successive phases of
structures along the axis and these position-specific combinations are growth and molting before developing into an adult, the phase
sometimes called the Hox code. between each molt is known as an instar.
The hypoblast in the early chick embryo is a sheet of cells that covers In an instructive induction, the cells respond differently to different
the yolk under the blastoderm and gives rise to extra-embryonic concentrations of the inducing signal.
structures such as the stalk of the yolk sac.
Insulin-like growth factors (IGF) are polypeptide growth factors that
The hypocotyl is the seedling stem that develops from the region mediate many of the effects of growth hormone and are essential for
between the embryonic root and the future shoot. post-natal growth in mammals.
The hypophysis is a cell in some plant embryos that is recruited from Integrins are a class of cell-adhesion molecules by which cells attach
the suspensor and contributes to the embryonic root meristem and to the extracellular matrix.
root cap.
Intercalary growth can occur in animals capable of epimorphic
IGF see insulin-like growth factor. regeneration when two pieces of tissue with different positional
Imaginal discs are small sacs of epithelium present in the larva of values are placed next to each other. The intercalary growth replaces
Drosophila and other insects, which at metamorphosis give rise to the intermediate positional values.
adult structures such as wings, legs, antennae, eyes, and genitalia.
Interdigital regions in the developing hand or foot are the regions of
Some cell-adhesion molecules, such as N-CAM, are members of the tissues between the forming digits.
immunoglobulin superfamily (which also contains many proteins
Intermediate filaments are one of the three principal protein
that are not cell-adhesion molecules).
filaments of the cytoskeleton. They are involved in strengthening
A gene is said to be imprinted when it is expressed differently (either tissues such as epithelia.
active or inactive) in the embryo depending on whether it is derived
An internode is that portion of a plant stem between two nodes (sites
from the mother or father. This genomic or parental imprinting
at which a leaf or leaves form).
occurs during gamete formation.
Intracellular signaling describes the process by which a signal
Plants with indeterminate growth do not make a fixed number of
received at a cell’s surface is relayed onward to its final destination
leaves or flowers.
inside the cell by a series of proteins that interact with each other.
Somatic cells can be converted into pluripotent stem cells called
Invagination is the local inward deformation of a sheet of embryonic
induced pluripotent stem cells (iPS cells) by the introduction and
epithelial cells to form a bulge-like structure, as in early gastrulation
expression of a few specific genes.
in the sea-urchin embryo.
Induction is the process whereby one group of cells signals to
another group of cells in the embryo and so affects how they will Involution is a type of cell movement that occurs at the beginning of
develop. amphibian gastrulation, when a sheet of cells enters the interior of
the embryo by rolling in under itself.
An inflorescence in plants is a flowerhead—a flowering shoot.
Shoots that can bear flowers develop as a result of the conversion iPS cells see induced pluripotent stem cells.
of a vegetative apical meristem into an inflorescence meristem. Ipsilateral refers to the same side of the body.
Ingression is the movement of individual cells from the outside of The isthmus is a signaling center located at the midbrain–hindbrain
the embryo into the interior during gastrulation. boundary in the embryonic brain that helps to pattern the brain along
Initials are cells in the meristems of plants that are able to divide the antero-posterior axis.
continuously, giving rise both to dividing cells that stay within
Keratinocytes are differentiated epidermal skin cells that produce
the meristem and to cells that leave the meristem and go on to
keratin, eventually die, and are shed from the skin surface.
differentiate.
Knock-out see gene knock-out.
The initiation of transcription of a gene is a crucial step in its
expression and in most developmental genes is under tight control Koller’s sickle is a crescent-shaped region of small cells lying at the
so that genes are not expressed at the wrong time and place. front of the posterior marginal zone in the chick blastoderm.
 Glossary 595

The lateral geniculate nucleus (LGN) is the main region of the brain Meristems are groups of undifferentiated, dividing cells that persist at
in mammals where the axons from the retina terminate. the growing tips of plants. They give rise to all the adult structures—
shoots, leaves, flowers, and roots. The meristem identity genes specify
Lateral inhibition is the mechanism by which cells inhibit
whether a meristem is a vegetative or an inflorescence meristem.
neighboring cells from developing in a similar way to themselves.
Mesectoderm is composed of cells that may give rise to both
The lateral plate mesoderm in vertebrate embryos lies lateral and
ectoderm and mesoderm.
ventral to the somites and gives rise to the tissues of the heart,
kidney, gonads, and blood. Mesenchyme describes loose connective tissue, usually of
mesodermal origin, whose cells are capable of migration; some
Lateral shoot meristems arise from the apical shoot meristem and
epithelia of ectodermal origin, such as the neural crest, undergo an
give rise to lateral shoots.
epithelial to mesenchymal transition.
A plant leaf develops from a small set of cells called a leaf
primordium at the edge of the apical meristem. A mesenchyme-to-epithelium transition occurs when loose
mesenchyme cells aggregate and then form an epithelium like a tube,
The bilateral asymmetry of the arrangement and structure of most as in kidney development.
internal organs in vertebrates is known as left–right asymmetry. In
mice and humans, for example, the heart is on the left side, the right Mesendoderm is composed of cells that may give rise to both
lung has more lobes than the left, and the stomach and spleen lie to endoderm and mesoderm.
the left. The mesoderm is the germ layer that gives rise to the skeleto-
The life history of an organism or species is its life cycle viewed muscular system, connective tissues, the blood, and internal organs
in terms of its reproductive strategy and its unique ecology or such as the kidney and heart.
interaction with the environment. The mesonephros in mammals is an embryonic kidney that
The small embryonic structures that give rise to the limbs of contributes to the male and female reproductive organs.
vertebrates are called limb buds. Messenger RNA (mRNA) is the RNA molecule that specifies the
A cell’s lineage is the sequence of cell divisions that give rise to sequence of amino acids in a protein. It is produced by transcription
that cell. from DNA.

Lineage restriction see cell-lineage restriction. Metamorphosis is the process by which a larva is transformed into an
adult. It often involves a radical change in form, and the development
Locus control regions are gene regulatory sites that control the
of new organs, such as wings in butterflies and limbs in frogs.
expression of a complete locus, such as the globin gene loci, and are
located quite distant from the genes that they control. Metastasis is the movement of cancer cells from their site of origin
to invade underlying tissues and to spread to other parts of the body.
In long-germ development the blastoderm gives rise to the whole
Such cells are said to metastasize.
of the future embryo, as in Drosophila.
Microfilaments see actin filaments.
Macromeres are the larger of the cells that result from unequal
cleavage in certain embryos, such as those of sea urchins. Micromeres are small cells that result from unequal cleavage during
The marginal zone of an amphibian embryo is the belt-like region early animal development.
of presumptive mesoderm at the equator of the late blastula. MicroRNAs (miRNAs) are small RNAs that suppress the expression
Maternal-effect mutations are mutations in genes of the mother of specific genes.
that affect the development of the egg and later the embryo. Genes Microtubules are one of the three principal protein filaments of the
affected by such mutations are called maternal-effect genes. cytoskeleton. They are involved in the transport of proteins and RNAs
Maternal factors are proteins and RNAs that are deposited in the egg within cells.
by the mother during oogenesis. The production of these maternal The mid-blastula transition in amphibian embryos is when the
proteins and RNAs is under the control of so-called maternal genes. embryo’s own genes begin to be transcribed, cleavages become
The medio-lateral axis in vertebrates runs from the midline to the asynchronous, and the cells of the blastula become motile.
periphery. The midbrain or mesencephalon is the middle section of the
Medio-lateral intercalation of cells occurs during convergent embryonic vertebrate brain and gives rise to the tectum (in
extension in amphibian gastrulation. The sheet of cells narrows and amphibians and birds) and similar structures in mammals, which are
elongates by cells pushing in sideways between their neighbors. the sites of integration and relay centers for signals coming to and
from the hindbrain, and also for inputs from the sensory organs.
Meiosis is a special type of cell division that occurs during formation
of sperm and eggs, and in which the number of chromosomes is miRNAs see microRNAs.
halved from diploid to haploid.
Mitosis is the nuclear division that occurs during the proliferation
In mericlinal chimeras a genetically marked cell gives rise to a sector of somatic diploid cells and results in both daughter cells having the
of an organ or of a whole plant. same diploid complement of chromosomes as the parent cell.
596 Glossary

The small numbers of species that are commonly studied in The neural crest cells of vertebrates are derived from the edge
developmental biology are known as model organisms. of the neural plate. They migrate to different regions of the body
and give rise to a wide variety of tissues, including the autonomic
Molting is the shedding of an external cuticle when arthropods grow,
nervous system, the sensory nervous system, pigment cells, and some
and its replacement with a new one.
cartilage of the head.
Morphallaxis is a type of regeneration that involves repatterning
Neural folds, neural plate, neural tube see neurulation.
of existing tissues without growth.
β-Neurexins are transmembrane proteins on the pre-synaptic
A morphogen is any substance active in pattern formation whose
membranes of some synapses, which interact with neuroligins of the
spatial concentration varies and to which cells respond differently
post-synaptic membrane to promote the functional differentiation of
at different threshold concentrations.
the synapse.
Morphogenesis refers to the processes involved in bringing about
The neuroectoderm is embryonic ectoderm with the potential to
changes in form in the developing embryo.
form neural cells and epidermis.
The morphogenetic furrow in Drosophila eye development moves
A neuroblast is an embryonic cell that will give rise to neural tissue
across the eye disc and initiates the development of the ommatidia.
(neurons and glia).
Morpholino antisense RNA is a type of antisense RNA composed
Neurogenesis is the formation of neurons from their precursor cells.
of stable morpholino nucleotide analogs.
Neuroligins see under β-neurexins.
A morula is the very early stage in a mammalian embryo when
cleavage has resulted in a solid ball of cells. The neuromuscular junction is the specialized area of contact
between a motor neuron and a muscle fiber, where the neuron can
Mosaic development was a term used historically to describe the stimulate muscle activity.
development of organisms that appeared to develop mainly by
distribution of localized cytoplasmic determinants. Neurons or nerve cells are the electrically excitable cells of the nervous
system, which convey information in the form of electrical signals.
The Müllerian duct runs adjacent to the Wolffian duct in the
mammalian embryo and becomes the oviduct in females. Neurotrophins are proteins that are necessary for neuronal survival,
such as nerve growth factor.
Müllerian-inhibiting substance is secreted by the developing testis
and induces regression of the Müllerian ducts in males. A neurula is the stage of vertebrate development at the end of
gastrulation when the neural tube is forming.
A multipotent cell is one that can give rise to many different types,
but not all types, of differentiated cell. Neurulation in vertebrates is the process in which the ectoderm of
the future brain and spinal cord—the neural plate—develops folds
The mural trophectoderm of the mammalian blastocyst is that (neural folds) that come together to form the neural tube.
trophectoderm not in contact with the cells of the inner cell mass.
The Nieuwkoop center is a signaling center on the dorsal side of the
A myoblast is a committed but undifferentiated muscle cell. In early Xenopus embryo. It forms in the dorsal vegetal region of the
developing skeletal muscle, it will first develop into a multinucleate blastula as a result of cortical rotation.
myotube and then into a fully differentiated muscle fiber.
Nodal and Nodal-related proteins comprise a subfamily of the TGF-β
The myocardium is the outer contractile layer of the developing family of signaling proteins of vetebrates. They are involved in all
heart. stages of development, but particularly in early mesoderm induction
Myoplasm is special cytoplasm in ascidian eggs involved in the and patterning.
specification of muscle cells. A node in a plant is that part of the stem at which leaves and lateral
buds form. In avian and mammalian embryos, the node is the
The myotome is that part of the somite that gives rise to muscle.
embryonic organizing center analogous to the Spemann organizer of
Necrosis is a type of cell death due to pathological damage in which amphibians. It is also known as Hensen’s node in birds.
cells break up, releasing their contents.
The effects of a gene are non-cell-autonomous or non-autonomous if
Negative feedback is a type of regulation in which the end-product they affect cells other than the cell in which the gene is expressed.
of a pathway or process inhibits an earlier stage.
The notochord in vertebrate embryos is a stiff rod-like cellular
Neoteny is the phenomenon in which an animal acquires sexual structure that runs from head to tail and lies centrally beneath the
maturity while still in larval form. future central nervous system. It is derived from mesoderm.

Netrins are secreted proteins that act as guidance molecules for The DNA in chromosomes is packaged into a series of nucleosomes,
neurons in the nervous system; they can be either attractant or which each consist of a core of histone proteins with DNA wrapped
repellent. round it.
 Glossary 597

Nurse cells surround the developing oocyte in Drosophila and The pair-rule genes in Drosophila are involved in delimiting
synthesize proteins and RNAs that are to be deposited in it. parasegments. They are expressed in transverse stripes in the
blastoderm, each pair-rule gene being expressed in alternate
The eight-cell stage of a plant embryo is called the octant stage.
parasegments.
Alternate ocular dominance columns in the visual cortex are
Genes within a species that have arisen by duplication and
columns of neurons that respond to the same visual stimulus from
divergence are called paralogs. Examples are the Hox genes in
either the left or right eye.
vertebrates, which comprise several paralogous subgroups made
Insect compound eyes are composed of hundreds of individual up of paralogous genes.
photoreceptor organs, the ommatidia (singular ommatidium).
Parasegments in the developing Drosophila embryo are independent
Many of the genes involved in cell regulation can be mutated into developmental units that give rise to the segments of the larva and
oncogenes, which cause cells to become cancerous. adult.

Ontogeny refers to the development of an individual organism. The mesoderm lying on either side of the midline and forming the
somites is sometimes called the paraxial mesoderm.
An oocyte is an immature egg.
PAR proteins (partitioning proteins) are a group of proteins initially
Oogenesis is the process of egg formation in the female.
discovered in Caenorhabditis elegans that are required for the correct
Oogonia are diploid germ cells that divide by mitosis within the positioning and orientation of mitotic spindles in the early cleavages
ovary before entering meiosis to produce the oocytes. to ensure division at the required place in the cell and in the required
plane.
The optic chiasm or optic chiasma is the site at which the retinal
nerve fibers from the right eye cross over fibers from the left eye Eggs that are able to develop into an embryo without fertilization are
on their way to opposite sides of the optic tectum (chicks and said to develop parthenogenetically.
amphibians). In mammals, the optic chiasm is the site at which the
nerve fibers from an eye divide, some taking a contralateral course Pattern formation is the process by which cells in a developing
to the opposite side of the lateral geniculate nucleus (LGN), whereas embryo acquire identities that lead to a well ordered spatial pattern
some connect with the ipsilateral side of the LGN. of cell activities.

The optic tectum is the region of the brain in amphibians and birds Pax genes encode transcriptional regulatory proteins that contain
where the axons from the retina terminate. both a homeodomain and another protein motif, the paired motif.

The optic vesicle is the precursor of the retina of the vertebrate eye P elements are transposable DNA elements found in Drosophila.
and is derived from the wall of the forebrain. They are short sequences of DNA that can become inserted in
different positions within a chromosome and can also move to
The oral–aboral axis in sea urchins and other radially symmetrical other chromosomes. This property is exploited in the technique of
organisms runs from the centrally situated mouth to the opposite side P-element-mediated transformation for making transgenic flies.
of the body.
Periclinal cell divisions are divisions in a plane parallel to the surface
An organizer, organizing region, or organizing center is a signaling of the tissue.
center that directs the development of the whole embryo or of part
of the embryo, such as a limb. In amphibians, the organizer usually In periclinal chimeras in plants, one of the three meristem layers has
refers to the Spemann organizer. The organizing center in plants a genetic marker, which distinguishes it from the other two.
refers to the cells underlying the central zone of the meristem, which The perivitelline space in the fertilized eggs and early embryos
maintains the stem cells of the central zone. of insects and other animals is the space between the vitelline
Organogenesis is the development of specific organs such as limbs, membrane lining the egg case and the egg plasma membrane.
eyes, and heart.
Permissive inductions occur when a cell makes only one kind of
Osteoblasts are the precursors from which differentiated bone cells response to an inducing signal, and makes it when a given level of
are formed. signal is reached.

The ovary is the internal reproductive structure in female animals P granules are granules that become localized to the posterior end of
and produces the female germ cells, the oocytes. the fertilized egg of Caenorhabditis elegans.

The oviduct in female birds and mammals transports the eggs from The phenotype is the observable or measurable characters and
the ovaries to the uterus. features of a cell or an organism.

An ovule is the structure in plants that contains an egg cell. The response of an organism to relative day length is known as
photoperiodism, and in plants is responsible for promoting flowering
The outer enveloping layer on the outer surface of the zebrafish
as days become longer.
blastoderm is a single-cell layer that eventually disappears as the
embryo develops. Phyllotaxy is the way the leaves are arranged along a shoot.
598 Glossary

Phylogeny is the evolutionary history of a species or group. information. The cells then interpret this positional value according
to their genetic constitution and developmental history, and develop
Vertebrate embryos pass through a developmental stage known
accordingly.
as the phylotypic stage at which the embryos of the different
vertebrate groups closely resemble each other. This is the stage at Positive feedback is a type of regulation in which the end-product of
which the embryo possesses a distinct head, a neural tube, and a pathway or process can activate an earlier stage.
somites.
Posterior dominance or posterior prevalence is the process whereby
Mammalian embryos (with the exception of the monotremes such as the more posteriorly expressed Hox genes can inhibit the action of
the egg-laying duck-billed platypus and echidna) are nourished by the more anteriorly expressed Hox genes, when they are expressed in the
mother by the passage of nutrients through the placenta, a structure same region.
that forms in the uterine wall where the blood systems of mother and
The posterior marginal zone of the chick embryo is a dense region
embryo form an interface with each other.
of cells at the edge of the blastoderm that will give rise to the
Placodes are regions of thickened epithelium, usually on the surface primitive streak.
of the embryo, that give rise to specific structures. An example is the
Most neurons do not divide further once they are formed, when they
lens placode, which gives rise to the lens of the eye.
are known as post-mitotic neurons.
Planar cell polarity is the situation in which cells are polarized in
The post-synaptic side of a synapse is the part that receives the
the plane of the tissue, as in the epidermis of insect wings, in which
signal.
wing hairs all point in the same direction.
Post-translational modification of a protein involves changes in the
A plasmid is a small circular DNA, derived from a bacterium or yeast
protein after it has been synthesized. The protein can, for example,
cell, that is often used as a vector or carrier to introduce genes into
be enzymatically cleaved, glycosylated, or acetylated.
cells.
Preaxial polydactyly is the occurrence of extra anterior digits on
Plasmodesmata (singular plasmodesma) are the threads of
hands or feet, for example, an extra thumb or an extra big toe.
cytoplasm that run through the cell wall and interconnect adjacent
plant cells. Prechordal plate mesoderm is the anterior-most mesoderm in the
vertebrate embryo, located anterior to the notochord. It gives rise to
A pluripotent stem cell, such as an embryonic stem cell, is one that
various ventral tissues of the head.
can give rise to all types of cells in the body.
A basic pattern generated automatically in a structure is known as a
The pluteus is the larval stage of the sea urchin.
prepattern. It may subsequently be modified during development.
Polar bodies are formed during meiosis in the developing egg. They
The pre-somitic mesoderm is the unsegmented mesoderm between
are small cells and take no part in embryonic development.
the node (in chick and mouse) and the already formed somites.
When one end of a cell, structure, or organism is different from the It will form somites from its anterior end.
other end it is said to have polarity or to be polarized.
The pre-synaptic side of a synapse is the part that generates the
In the developing chick and mouse limb buds, the polarizing region signal.
or zone of polarizing activity (ZPA) at the posterior margin of the
The primary cilium is an immotile microtubule-based structure
bud produces a signal specifying position along the antero-posterior
found on most vertebrate cells; it is the site of the Sonic hedgehog
axis.
signaling.
The polar trophectoderm of the mammalian blastocyst is the
Primary embryonic induction is the induction of the whole body
trophectoderm in contact with the inner cell mass.
axis, as demonstrated by transplantation of the Spemann organizer in
Pole cells give rise to the germ cells in Drosophila and are formed at amphibians.
the posterior end of the blastoderm.
The primary oocytes are female germ cells that have entered meiosis.
Pole plasm is the cytoplasm at the posterior end of the Drosophila
The primary trophoblast is the layer of giant cells that invade the
egg that is involved in specifying germ cells.
uterus wall at the implantation of the mammalian blastocyst.
Polydactyly is the occurrence of extra digits on hands or feet.
The primitive ectoderm or epiblast is the part of the inner cell mass
Polyspermy is the entry of more than one sperm into the egg. in the mammalian blastocyst that gives rise to the embryo proper.

Polytene chromosomes are giant chromosomes that are formed by The primitive endoderm or primary endoderm in mammalian
repeated DNA replication in the absence of cell division. embryos is that part of the inner cell mass that contributes to extra-
embryonic membranes.
Positional information in the form, for example, of a gradient of an
extracellular signaling molecule, can provide the basis for pattern The primitive streak of the chick and mouse embryo is the site
formation. Cells acquire a positional value that is related to their of gastrulation and the forerunner of the antero-posterior axis. It
position with respect to the boundaries of the given field of positional is a strip of ingressing cells that extends into the epiblast from the
 Glossary 599

posterior margin. Epiblast cells move through the streak into the Radial glial cells are elongated glial cells that span the whole width
interior of the embryo to form mesoderm and endoderm. of the neural tube wall and provide tracks for migrating neurons.

Minute undifferentiated growths that will give rise to a structure Radial intercalation occurs in a multilayered ectoderm of an
such as a tooth, leaf, flower or floral organ are known as primordia amphibian gastrula when cells intercalate in a direction perpendicular
(singular primordium). to the surface, so thinning and extending the cell sheet.

Primordial germ cells are precursor cells in the early embryo that Rapid block to polyspermy see block to polyspermy.
represent the germline and will produce the germ cells.
Reaction–diffusion mechanisms produce self-organizing patterns of
The two-celled stage in plants is called the proembryo. chemical concentrations which could underlie periodic patterns.

Programmed cell death see apoptosis. A recessive mutation is a mutation in a gene that only changes the
phenotype when both copies of the gene carry the mutation.
In chick and mouse limb buds the timing model proposes that cells
at the tip constitute a progress zone where they acquire positional The process of genetic recombination occurs during meiosis and
values. shuffles parental genes into new combinations in the haploid
gametes.
The promeristem is the central region of the meristem that contains
cells capable of continued division—the initials. Redundancy refers to an apparent absence of an effect when a
gene that is normally active during development is inactivated. It is
The promoter is a region of DNA close to the coding sequence to assumed that other pathways exist that can substitute for the missing
which RNA polymerase binds to begin transcription of a gene. gene action.

Proneural clusters are small clusters of cells within the Regeneration is the ability of a fully developed organism to replace
neuroectoderm in which one cell will eventually become a lost parts.
neuroblast.
Regenerative medicine aims to use stem cells and their derivatives
Genes that promote a neural fate in neuroectoderm cells are called to replace diseased tissue with healthy tissue.
proneural genes.
Regulation is the ability of the embryo to develop normally even
A pronucleus is the haploid nucleus of sperm or egg after fertilization when parts are removed or rearranged. Embryos that can regulate are
but before nuclear fusion and the first mitotic division. called regulative.

Prothoracicotropic hormone (PTTH) is a protein hormone Repressors are gene-regulatory proteins that act to suppress gene
secreted by the insect brain that causes the secretion of the steroid activity when they bind to specific sites in the gene control regions.
hormone ecdysone, and the initiation of molting or pupation and
Reverse genetics describes an approach that starts with the
metamorphosis.
nucleotide sequence of a gene or amino acid sequence of a protein,
A proto-oncogene is a gene that is involved in regulation of cell and then uses that information to determine the gene’s function.
proliferation and that can cause cancer when mutated into an The rhombomeres are a sequence of compartments of cell-lineage
oncogene or expressed under abnormal control. restriction in the hindbrain of chick and mice embryos.
Protostomes are those animals, such as insects, in which cleavage of RNA interference (RNAi) is a means of suppressing gene expression
the zygote is not radial and in which gastrulation primarily forms the by promoting the destruction of a specific mRNA by targeting it with
mouth. short complementary RNAs called short interfering RNAs (siRNAs).
The proximo-distal axis of a limb or other appendage (such as a leaf RNA processing is the process in eukaryotic cells in which newly
in a plant) runs from the point of attachment to the body or stem transcribed RNAs are modified in various ways to make a functional
(proximal) to the tip of the appendage (distal). messenger RNA or structural RNA. It includes RNA splicing, which
removes introns from the transcript to leave a continuous coding
The pupa in Drosophila and similar insects is a stage following the
messenger RNA or a functional structural RNA.
larval stages in which the organism can remain dormant for long
periods and in which metamorphosis occurs. The roof plate of the developing neural tube is composed of non-
neural cells and is involved in patterning the dorsal part of the tube.
The quiescent center in a plant root tip meristem is a central group
of cells that divide rarely but are essential for meristem function. Skeletal muscle can be renewed from undifferentiated stem cells
called satellite cells.
The radial axis of a structure is the axis running from the center to
the circumference. Cylindrical structures such as plant stems and The sclerotome is that part of a somite that will give rise to the
roots that are completely symmetrical around a central axis are said cartilage of the vertebrae.
to have radial symmetry.
Segmentation is the division of the body of an organism along the
Radial cleavage is a type of cleavage that occurs at right angles to the antero-posterior axis into a series of morphologically similar units or
egg surface and produces blastomeres sitting directly over each other. segments.
600 Glossary

Segmentation genes in Drosophila are involved in patterning the Somites in vertebrate embryos are segmented blocks of mesoderm
parasegments and segments. lying on either side of the notochord. They give rise to trunk and limb
muscles, the vertebral column and ribs, and the dermis.
Selector genes in Drosophila determine the activity of a group of
cells, and their continued expression is required to maintain that A specification map shows how the tissues of an embryo will
activity. develop when placed in a simple culture medium.

Semaphorins are secreted molecules that act as guidance cues for A group of cells is called specified if when isolated and cultured in a
neurons. neutral medium they develop according to their normal fate.

A semi-dominant mutation is a mutation that affects the phenotype The Spemann organizer or Spemann–Mangold organizer is a
when just one allele carries the mutation but where the effect on the signaling center on the dorsal side of the amphibian embryo that
phenotype is much greater when both alleles carry the mutation. acts as the main embryonic organizer. Signals from this center can
organize new antero-posterior and dorso-ventral axes.
Senescence is the impairment of function associated with aging.
Spermatogenesis is the production of sperm, which are the haploid
A sensory organ precursor (SOP) is an ectodermal cell that will give
male gametes in animals.
rise to a sensory bristle in the adult Drosophila epidermis.
The sphere stage in zebrafish embryos comprises a hemispherical
The septate junction in invertebrate cells is a type of cell junction
blastoderm of around 1000 cells lying over a spherical yolk.
with a similar function to the vertebrate tight junction.

Sex determination is the genetic and developmental process by Spiral cleavage is a type of cleavage typical of molluscs and annelid
which an organism’s sex is specified. In many organisms, sexual worms, in which the plane of cell division is at a slight angle to the
phenotype is determined by specific chromosomes called sex egg surface and blastomeres end up in a spiral arrangement.
chromosomes. A stem cell is a type of undifferentiated cell that is both self renewing
The sex-determining region of the Y chromosome (SRY) determines and also gives rise to differentiated cell types. They are found in some
maleness by specifying the gonad as a testis. adult tissues. They are maintained in microenvironments known as
stem-cell niches. See also embryonic stem cells.
The shield stage in zebrafish embryos is the stage in which the
organizer, known as the shield, has been formed on one side of the The subgerminal space is the cavity that develops under the area
blastoderm. pellucida in the early chick blastoderm.

Short-germ development characterizes those insects in which most The superior colliculus in mammals is a region of the brain to which
of the segments are formed sequentially by growth. The blastoderm some retinal neurons project. It corresponds to the optic tectum of
itself only gives rise to the anterior segments of the embryo. amphibians and birds.

Short interfering RNA (siRNA) see RNA interference. The suspensor attaches the embryo to maternal tissue and is a source
of nutrients.
A signaling center is a localized region of the embryo that exerts a
special influence on surrounding cells and thus determines how they A synapse is the specialized point of contact where a neuron
develop. communicates with another neuron or a muscle cell. Neurotransmitter
is produced in synaptic vesicles in the presynaptic neuron and is
Signal transduction is the process by which a cell converts an released into the synaptic cleft separating the two cells.
extracellular signal received in one form, such as the binding of a
substance to a cell-surface receptor, into an intracellular signal of a A syncytium is a cell with many nuclei in a common cytoplasm.
different form, such as the phosphorylation of a cytoplasmic protein. Cell walls do not develop during nuclear division within the very
early Drosophila embryo. This gives rise to the syncytial blastoderm
Silencing see gene silencing.
in which the nuclei are arranged around the periphery of the
In the rare condition situs inversus in humans there is complete embryo.
mirror-image reversal of the position of the internal organs.
The tailbud is the structure at the posterior end of vertebrate
Slit proteins are secreted proteins that act as guidance cues to repel embryos containing stem cells that give rise to the post-anal tail.
growing axons in the developing nervous system.
Telomeres are structures composed of non-coding DNA at the ends of
Somatic cells are any cells other than germ cells. In most animals, chromosomes that prevent chromosomes sticking to each other and
the somatic cells are diploid. which prevent gene loss at DNA replication. They typically become
shorter at each cell division.
Cloning of an animal by the transfer of a somatic cell nucleus into an
enucleated egg is known as somatic cell nuclear transfer. The telson is a distinctive structure at the posterior end of the
Drosophila embryo.
Somatostatin is a protein hormone produced by the hypothalamus
that inhibits the production and release of growth hormone by the Temperature-sensitive mutations are mutations that only cause a
pituitary. change in phenotype at a different temperature than normal, most
 Glossary 601

commonly a higher temperature. They are often mutations that cause Transgenic techniques describes various techniques that can be
a protein to become more unstable than normal. used to change the genetic make-up of an organism by the deliberate
introduction of new DNA; for example, by introducing new genes,
Teratocarcinomas are solid tumors that arise from germ cells and
or introducing DNAs that cause the inactivation of specific genes.
which can contain a mixture of differentiated cell types.
DNAs introduced by transgenesis are known as transgenes.
The testis is the internal male reproductive organ in animals and
Transient transgenesis describes the temporary introduction into,
produces the sperm.
and expression of, new or altered genes in somatic tissues, rather
A tetraploid cell is one that contains four sets of chromosomes. than the germline. Also known as somatic transgenesis.

Therapeutic cloning is the potential use of somatic cell transfer to In continually renewing tissues, such as gut epithelium and
generate embryos from which ES cells can be derived. The goal is to epidermis, the transit-amplifying cells are rapidly dividing cells
create cells that exactly match those of a patient and can be used to that are produced from the stem cells and will differentiate into the
alleviate disease. specialized cell types of the tissue.

In insects the body is divided into three distinct parts, the head at the The process by which messenger RNA directs the order of amino
anterior end, followed by the thorax and the posterior abdomen. The acids in a protein during protein synthesis at ribosomes is known as
thoracic segments of Drosophila are the segments that carry the legs translation. The messenger RNA is said to be translated into protein.
and wings in the adult. In vertebrates, the thorax is the chest region.
A transposon is a DNA sequence that can become inserted into a
A threshold concentration is that concentration of a chemical different site on the chromosome, either by the insertion of a copy of
signal or morphogen that can elicit a particular response from a cell. the original sequence or by excision and reinsertion of the original
A specific response to a chemical signal that only occurs above or sequence.
below a particular threshold concentration of the signal is known
as a threshold effect. A trichome is a hair-bearing cell in plant epidermis.

Tight junctions are a type of adhesive cell junction that binds A triploblast is an animal with three germ layers—endoderm,
epithelial cells very tightly together to form an epithelium and that mesoderm, and ectoderm.
seals off the environment on one side of the epithelium from the
Abnormal segregation of chromosomes during meiosis can give rise
other.
to gametes with abnormal numbers of a particular chromosome, such
TILLING (targeting-induced local lesions in genomes) is a technique as three copies rather than two—a trisomy.
for detecting mutations by hybridizing the mutated DNA against
The trophectoderm is the outer layer of cells of the early mammalian
unmutated DNA and detecting mismatched bases, which indicate
embryo. It gives rise to extra-embryonic structures such as the
the site of a mutation.
placenta.
The expression of a gene only in a particular tissue or cell type is
Tumor suppressor genes are genes that can cause a cell to become
known as tissue-specific expression.
cancerous when both copies of the gene have been inactivated.
Totipotency is the capacity of a cell to develop into a new organism.
Vasculogenesis is the initial stages in the formation of blood vessels,
Such a cell is called totipotent.
comprising the condensation of angioblasts to form a tubular vessel.
The tracheal system of insects is a system of fine tubules that deliver
air (and thus oxygen) to the tissues. The vas deferens is the duct that connects the sperm-producing testis
to the penis.
When a gene is active its DNA sequence is copied, or transcribed,
into a complementary RNA sequence, a process known as The vegetal region of an amphibian egg is the most yolky region,
transcription. and is the region from which the endoderm will develop. The most
terminal part of this region is called the vegetal pole, and is directly
A transcription factor is a regulatory protein required to initiate or opposite the animal pole.
regulate the transcription of a gene into RNA. Transcription factors
act within the nucleus of a cell by binding to specific regulatory Ventral closure is the closing together of the sides of the chick or
regions in the DNA. mammalian embryo on the ventral side of the body to form the gut.

Transdetermination is the process by which a differentiated cell can Embryos that are ventralized are deficient in dorsal regions and have
become re-determined as a different cell type. much increased ventral regions.

Transdifferentiation is the process by which a differentiated cell can The ventricular zone is a layer of proliferating cells lining the lumen
differentiate into a different cell type, such as pigment cell to lens. of the vertebrate neural tube, from which neurons and glia are
formed.
Transfection is the technique by which mammalian and other animal
cells are induced to take up foreign DNA molecules. The introduced The phenomenon by which flowering is accelerated after the plant
DNA sometimes becomes inserted permanently into the host cell’s has been exposed to a long period of cold temperature is known as
DNA. vernalization.
602 Glossary

The vertebral column is the backbone or spine of vertebrates, The yolk syncytial layer in zebrafish forms a continuous layer of
composed of a succession of vertebrae. multinucleate non-yolky cytoplasm underlying the blastoderm.

The visceral endoderm is derived from the primitive endoderm Zinc-finger nucleases are used in a technique to specifically cleave
that develops on the surface of the egg cylinder in the mammalian and disrupt genes in vivo in a targeted fashion.
blastocyst.
The zona limitans intrathalamica is a signaling center located in
The visual cortex is that part of the mammalian cerebral cortex the forebrain of the vertebrate embryo, which helps pattern the brain
where visual signals are sent after the LGN and are processed to along the antero-posterior axis.
produce a visual perception.
The zona pellucida is a layer of glycoprotein surrounding the
The vitelline membrane or vitelline envelope is an extracellular mammalian egg that serves to prevent polyspermy.
layer surrounding the eggs of animals. In the sea urchin it gives rise
The zone of polarizing activity (ZPA) see polarizing region.
to the fertilization membrane.
The zootype refers to a pattern of expression of Hox genes and
The Wnt family of secreted signaling proteins is important in
certain other genes along the antero-posterior axis of the embryo that
many aspects of development. It includes the Wingless protein in
is characteristic of all animal embryos.
Drosophila. See also canonical Wnt/β-catenin pathway and planar
cell polarity. The zygote is the fertilized egg. It is diploid and contains chromosomes
from both the male and female parents.
Wolffian ducts are ducts associated with the mesonephros in
mammalian embryos. They become the vas deferens in males. Zygotic genes are those present in the fertilized egg and which are
expressed in the embryo itself.
The yolk sac is an extra-embryonic membrane in birds and
mammals. In the chick embryo it surrounds the yolk.
Index

A androgenetic embryos 340 mammals 392, 393

ABC model of floral organ identity 278–9 Angelman syndrome 341–2 nematodes 392, 393
abdominal-A (abd-A) gene and product angioblasts 456 neuronal cell death 497
82–3, 564, 567, 568 angiogenesis 456–7 apterous gene and product 438–9, 441, 574
abdominal-B (abd-B) gene and product angiosperms 258, 259 APX-1 protein 223, 224
82–3, 564 angustifolia mutation 322, 323 Arabidopsis thaliana 3, 9, 256–8
abembryonic pole 139 animal pole 3, 97, 129 embryonic development 258–64
accretionary growth 506 animal–vegetal axis fate map 261
acerebellar mutant 203 ascidian 244–6 fertilization 257, 354
acetylcholine 389, 493–5 sea urchin 234, 235 flower development 276–7, 280, 284
receptor 494–5 Xenopus 97, 129–30 flowering control 282–3
achaete gene 470, 474 zebrafish 129, 130 homeotic gene role 277–9
acheiropodia 423 aniridia 447 sex determination 354
acrosomal reaction 343, 344 Antennapedia (Antp) gene 84, 442–3, germination 258
actin filaments 291 564, 567 leaf axes 270–1
activators 368 Antennapedia homeotic gene complex 82, leaf directed dilation 322–3
activin 157, 163 84–6, 194 life cycle 257
receptor 158, 159 anterior gradient protein (nAG) 539, 542 meristems 265
actomyosin 291 anterior visceral endoderm (AVE) 140–1, cell fate 267–8, 269–70
adaxial-abaxial axis, leaves 270–1 164, 165 stem cells 265–7
adenomatous polyposis coli (APC) 521 antero-posterior axis 14 organizer center 266–7
adherens junctions 292, 293 ascidian 244–6 root tissues 273–5
adhesive cell junctions 292 Caenorhabditis 218–21, 226 see also plant development
AGAMOUS (AG) gene and product 278, Drosophila 43–8 archenteron 99
279, 282 egg 52–4 area opaca 104, 106
aging 527–9 spinal cord 482–3 area pellucida 104, 106
cell senescence 530–1 vertebrates 93 Argos protein 449
genetic influences 529–30 chick 136–8 Aristotle 4–5
Agrobacterium tumefaciens 263, 264 limb axes 412, 419–22 Artemia 566–7
allantois 108 mouse 138–41 arteries 457
alleles 11, 336 somite formation 187–91 ascidians 215, 244–9
alx1 gene and product 239 somite identity 191–7 body axes 244–6
Ambystoma mexicanum 510–11, 538–9 see also body axes; pattern formation cell fate 246
amnion 108 anticlinal divisions 260, 261 cleavage 244–5, 246
amnioserosa, Drosophila 41 Antirrhinum 276, 277, 279–81 egg 244–5
amniotes 96 antisense RNA 19, 220 genome 244
amphibians APETALA genes and products 277, 278, 279 life cycle 245
fate map 146–7 apical dominance 522, 523 mosaic development 23
gastrulation 99, 312–13 apical ectodermal ridge 413 muscle specification 246–7
limb regeneration 537–44 role in vertebrate limb development 415–16 notochord 247–9
nerve influence 539–40 apical–basal axis 14 Astyanax mexicanus 447–8, 575
metamorphosis 525–6 Arabidopsis 260–2 asymmetric cell division 27–9, 30, 215
neurulation 100 see also pattern formation ascidians 247–8
visual system 490 apoptosis 301 Caenorhabditis 218–21, 223–6
eye determination 23 digit separation role 434 neurogenesis 472–3
see also Xenopus Drosophila 392, 393 sea urchin 233
amphioxus 562, 564 eye loss in Astyanax 575 stem cells 29, 30, 31, 377–8
anamniotes 96 genetic control 392 atonal gene 449
604 Index

Aux/IAA proteins see auxin body axes, establishment of 14, 15 C

auxin 522 ascidian 244–6 Cactus protein 50
gradients 260–2 Caenorhabditis 218–23, 232 cadherins 293, 294–5, 486
leaf arrangement on stem and 272–3 double-gradient model 179 E-cadherin 52, 294–5, 300
root development and 274–5 Drosophila 40–51, 52–4 N-cadherin 294, 295
signaling pathway 262 wing 438–9 P-cadherin 294
auxin response factors (ARFs) 261, 262 leg disc 439–41 Caenorhabditis 3, 9, 216–32, 560
axes see body axes vertebrates 129–45 aging 528, 529
axolotl 510–11, 538–9, 580 chick 136–8, 177–8 anatomy 217
axons 468 limb axes 412, 416–22, 426–7 body axes 218–23, 232
growth cone 485–6 mouse 138–41 antero-posterior axis 218–21, 226
navigation 484–92 Xenopus 129–36 dorso-ventral axis 221–3
guidance 485–9 zebrafish 129–36 positional identity 226
midline crossing 488 see also pattern formation cell fate 218, 219
regeneration 547 body plan specification 223–6
evolutionary changes 564–7 cell lineage 218, 219, 227
B vertebrates 94, 565 cleavage 217
bantam microRNA 513 versus arthropods 568–9 dauer state 529
Bardet-Beidl syndrome (BBS) 425 Bolitoglossa 579 developmental genes 18–19
Barr body 357, 358 bone growth 517–19 dosage compensation 357, 359
Barx1 gene 460–1 bone marrow 377 genome sequence 218
bat limb 572, 573 transplants 402–3 Hox genes 226
Bax gene 497 bone morphogenetic proteins (BMP) 157 larva 217–18
Bcl-2 gene and product 392, 497 antagonism 160, 179–81, 185, 200 life cycle 216
β-catenin 130–3, 135–6, 370 BMP-2/4 243, 574 mosaic development 23
Hydra 551 BMP-4 63–4, 569 sex determination 353–4, 356
sea urchin 238, 243 chordin interaction 65, 160 timing of developmental events
see also canonical Wnt/ β -catenin mesoderm patterning 159–60 226–9
pathway; Wnt/ β -catenin asymmetry somite patterning 200, 201 transdifferentiation 397–8
pathway limb development and 425, 427, 434 ventral closure 308–9
β -globin gene expression 380–2 nervous system development and 480 vulval development 229–31
β -neurexins 495 bottle cells 309, 311 calcium wave, egg activation 346–8
bicoid gene and product 44, 51, 53–5, 57–8 Brachyury gene 161–3, 248, 311, 312 cancer 520–2
Bicoid protein 44–6, 45 semi-dominant mutation, mouse 12 cancer cells 2, 520–1
bilateral symmetry 558–9 brain development 202–9 canonical Wnt/ β -catenin pathway 26, 76,
Bilateria 558–60 cortical neurons 477–8 132–3, 185
birds 103 hindbrain rhombomeres 203–6 Hydra development role 551
see also chick Hox gene role 206–7 see also β -catenin; Wnt protein signaling
bithorax homeotic gene complex 82–6, 194 signaling centers 202, 203 capillaries 457
bithorax mutation 443 visual centers 489–92, 497–9 CAPRICE protein 275
blastema 537–42, 545 branchial arches 203 Capsella 259
blastocoel 3, 98 evolutionary changes 576–7 cardiomyocytes 546–7
fluid accumulation 300–1 Hox gene expression 207, 208 cartilage 433
blastocyst 110, 139–40 neural crest cell migration to 207–8 caspase 392
blastoderm 101, 104, 148 Xenopus 99 catenins 293
antero-posterior axes 136–8 branching morphogenesis 452–4 cat, X inactivation 357, 358
blastomere 98 branchless gene and product 452–3 caudal gene and product 47
microsurgical removal 115 breathless gene and product 452–3 cavity formation 301
specification 138 Brenner, Sydney 216–17 CED proteins 392, 393
blastopore 8, 98, 309–11 bride-of-sevenless (boss) gene and ceh-13 gene 226
blastula 3, 98, 129, 298 protein 450 cell adhesion 289–90, 292–5
fate map 146–7 brine shrimp 566–7 cell-adhesion molecules 17, 292, 293
formation of 298–302 brinker gene and product 437 neurulation and 317
Blimp1 protein 333 budding, Hydra 549 synapse formation and 495–6
block to polyspermy 345–6 butterfly differential adhesion hypothesis 290
blood cells 375–82 metamorphosis 525 regeneration and 542
blood vessels 456–7 wing development 566 tissue variation 292–4
B lymphocytes 377 markings 441–2 cell-autonomous gene effects 150
Bmi1 expression 384 buttonhead gene and product 85 cell complexity 30–1
 Index 605

cell cycle 7, 506–8 cell-replacement therapy 400, 402–3 Chordin protein 160, 179, 180, 569
checkpoints 507 cell generation 403–5 BMP-4 interaction 65, 160
regulation 506–9 cell senescence 530–1 chorion 108
cell death see apoptosis; necrosis cell specification 21, 22 chromatin 182
cell determination 21, 22 cell theory 5–6 gene expression regulation 372–4
cell differentiation 16, 365–7 cell–cell interactions 13 immunoprecipitation techniques 124
Caenorhabditis 224–5 Caenorhabditis 221–6 chromatin-remodelling complexes 182
cell fusion effects 396–7 cell fate specification 223–6 chromosome-attracting body (CAB) 246
evolution 561 vulval development 229–31 Churchill gene and product 181
models 375–94 cell–cell signaling 17 ciliopathies 425
blood cells 375–82 germ cell development 333 Ciona 244
epithelia 382–5 centrosome 297 genome 244
muscle 385–9 cephalochordates 562, 564 life cycle 245
neural crest cells 389–92 cephalopods 577 notochord 247
reversibility 394–6 Cerberus protein 160, 179, 180, 185 see also ascidians
dedifferentiation 388, 537–40 c-hairy genes 188 CLAVATA (CLV) genes 266, 267
senescence effect 530–1 chemoaffinity hypothesis of connectivity 491 cleavage 3, 6, 14, 295–8
spinal cord 478–80 chick 103–8 ascidian 244–5, 246
transdifferentiation 388, 397–9, body axes 136–8 Caenorhabditis 217
538–9 brain development 202 Drosophila 37
replacement cell generation 404 hindbrain rhombomeres 203–4 furrow 296, 297
vertebrate limb 413 signaling centers 202 plane of 297–8
cell division cleavage 104, 106 radial 296
anticlinal 260, 261 egg 104 sea urchin 233, 234
asymmetric 27–9, 30, 215 embryo 4, 9, 95 body axis and 241–2
ascidians 247–8 extra-embryonic structures 108 spiral 296
Caenorhabditis 218–21, 223–6 fate map 117, 148 vertebrates 94
neurogenesis 472–3 gastrulation 103, 104, 314–15 chick 104, 106
sea urchin 233 germ cell development 334 mouse 109–10, 138
coordination 512–13 heart 458 Xenopus 14, 98
differentiation and 367 Hox genes 196 zebrafish 101, 102
periclinal 260, 261 left-right asymmetry 142–4 cloning 395
see also meiosis; mitosis life cycle 105 mammals 396
cell enlargement 506 limb development see vertebrate limb somatic cell nuclear transfer 395–6
cell fate 21, 22 development therapeutic 402
ascidian 246 mesoderm induction and patterning Cnidaria 558–9, 560
Caenorhabditis 219 163–4 co-activators 370
specification 223–6 microsurgical manipulation 116 cockroach limb regeneration 545–6
germ cells 331–3 neural crest migration 318–20 colony-stimulating factors 379
neural crest cells 390 neural induction 179–80, 181 colorectal cancer 521
plant meristems 267–70 neuronal development 486–7 community effect 152–3, 154
sea urchin 236 neuronal cell death 496–7 compaction 299–300
somite cell fate determination 199–201 neurulation 107, 317 compartments 71–4, 76–7
see also fate maps organizer region see Hensen’s node complexity 30–1
cell fusion 396–7 somite formation 187, 188–9 compound eye 444, 445, 448–51
cell lineage patterning within somite 199–201 patterning 448–50
Caenorhabditis 218, 219, 227 spinal cord 479, 481, 482 planar cell polarity 77
restriction 71–2 transgenic technique 121 condensation 413
hindbrain development 203–6 transient transgenesis studies 123 congenital abnormalities 2
tracing 117–18 wing development 574 CONSTANS (CO) gene and product 282, 283
cell migration 318 see also vertebrate limb development constriction 305, 307
germ cells 334 chimera 150 contraction 291, 305
chemical signals 334–5 mericlinal 268–9 hypodermal cells 321–2
neural crest cells 186, 207–8, 318–20 periclinal 268 convergent extension
sea urchin gastrulation 303–6 ChIP-chip technique 124 gastrulation 305, 309–13, 315
cell motility 290 ChIP-seq technique 124 neurulation 316
cell packing 298 choanoflagellates 560–1 co-repressors 370
cell proliferation 506 chondrocytes 517–18 cornea 446
control of 506–8 chordamesoderm 313–14 corticotropin-releasing hormone 525
606 Index

craniorachischisis 316 evolutionary changes 562–4 gastrulation 38, 39, 306–9

Cre/loxP system 118 identity specification 422–6 gene expression regulation 369, 372
cricket limb regeneration 544 polydactyly 422, 423, 431 germ band extension 38, 39, 307–8
crystallins 577 programmed cell death role in germ cell development 331, 335
cumulus cells 343 separation 434 growth 506, 514–15
CXCR4 receptor 335 diploblasts 559 head region 85, 567
cyclin-dependent kinases (Cdks) 507 diploidy 6, 11, 335 heart 459
Cdk4 387 directed dilation 320–3 Hox genes 81–6, 194, 442–3, 564
cyclins 507 leaves 322–3 targets 565–6
CYCLOIDEA gene 281 nematode embryo elongation 321–2 imaginal discs 39, 72–4, 435, 439–43, 444
cyclopia 28 notochord 321 size determination 512–13, 514
Cyp26a1 gene 199 Dishevelled (Dsh) protein 132 intercellular signals 53
cytokinins 275, 523 disposable soma theory 528–9 larva 38–9
cytoplasmic determinants 27–30 distal-less gene and product 441–2, 568, 574 leg development 435, 439–41, 444, 568
distal visceral endoderm (DVE) 140–1 life cycle 36–40, 581
D Dlx2 gene 460–1 maternal genes 40, 41–51
Dachsous (Ds) cadherin 514, 544 DNA mutation effects 44
daf-2 gene and product 529 control regions 18 mesoderm 38, 40
Darwin, Charles 556, 558 methylation 341, 358–9, 373 invagination 306–7
Darwin’s finches 558 microarray analysis 114, 115 metamorphosis 39, 436, 440, 525–6
daschund gene 441 mitochondrial 346 misexpression screening 60, 61
dauer state 529 repair 529, 530 neurogenesis 470–3
DCC receptor 488 dorsal closure, Drosophila 308 oogenesis 51–65
decapentaplegic (dpp) gene and product dorsal convergence 312 antero-posterior polarity 52–4, 56
62–5, 436–8, 440, 480, 569, 574 dorsal gene and product 49–50, 59–63 dorso-ventral polarity 54–5, 56
dedifferentiation 388 dorsalized embryos 49, 62 oocyte cyotskeleton reorganization 53–5
limb regeneration and 537–40 dorsalizing factors 132 pattern formation 56–65
Deformed gene 84 dorsal root ganglia 319, 391 denticle patterns 74–7
Delta-Notch signaling pathway 52, 189, 190, dorso-ventral axis main genes involved 87
238 Caenorhabditis 221–3 planar cell polarity 77–9
see also Delta protein; Notch protein Drosophila 48–50 pole cells 36
Delta protein 131, 238 egg 54–5 segmentation 38, 39
compound eye development and 449 wing 438–9 compartments 71–4, 76–7
leg development and 439, 441 spinal cord 478–82 parasegment boundary stabilization 74–7
neurogenesis role 471–2, 475 vertebrates 93 parasegment establishment 66–70
dendrites 468 chick 136–8 segmentation gene expression 70–81
denticle pattern, Drosophila 74–7 limb axes 412, 426–7 segment identity specification 81–6
dermomyotome 200, 201 versus arthropods 568–9 sensory bristles 473, 474
Derriere protein 157, 158 Xenopus 130–3 sex determination 351–3, 355–6
desmosomes 293 zebrafish 133 synapse formation 496
determinants 6 see also body axes; pattern formation syncytial blastoderm 36, 41
cytoplasmic localization 27–9 dosage compensation 356–9, 373 targeted gene expression 61
determination 21, 22, 149, 367 doublesex (dsx) gene and product 351–2 terminal region specification 47–8
amphibian eye region 23 Down syndrome 338 tracheal system 452–3
by neurula stage 208–9 Driesch, Hans 7 transgenic 60, 61
deuterostomes 232, 560 Drosophila 3–4, 8, 9, 35–6, 560 trichome patterns 566
developmental constraint 573–4 aging 528, 529–30 wing development 77, 435, 565–6, 568
developmental genes 18–19 apoptosis 392, 393 axis establishment 438–9
control of expression 19–21 body axes 40–51 compartment boundaries 72–4, 436–9
cis-regulatory modules 19 antero-posterior axis 43–8, 52–4 evolutionary comparisons 574–5
Drosophila 18–19 dorso-ventral axis 48–50, 54–5 pigment pattern in D. biarmipes 576
genetic screening 39–40, 42 cell cycle regulation 508–9 positional signals 436–8
see also specific genes cleavage 37 Dscam cell-adhesion molecule 496
developmental stages 96 compound eye 448–51
Dharma transcription factor 136 cytoplasmic determinants 29 E
diabetes 403–4 developmental genes 18–19 early signal (ES), sea urchin 238
Dickkopf1 179, 180, 185 genetic screening 39–40, 42 ecdysone 370, 514, 515, 525
differentiation see cell differentiation dorsal closure 308 echinoderms 232
digits dosage compensation 357, 359 see also sea urchins
 Index 607

ectoderm 3, 15 Engrailed-1 (En-1) gene 426 fate see cell fate; fate maps
Drosophila 38 engrailed gene and product 67, 70, 75–7, 80, fate maps 117–18
limb development and 426–7 308, 440 Arabidopsis 261
mouse 110 compartment boundaries and 70–4, 436, ascidian 246
sea urchin 242–3 437 Caenorhabditis 218, 225
vertebrates 94, 103, 105 expression regulation 372 Drosophila imaginal discs 436, 440
Xenopus 153–4 ENHANCER OF TRYPTYCHON protein 275 neural crest cells 390
Ectodermin 153, 160 enhancers 368 plants 268–70, 274
EGF see epidermal growth factor (EGF) enhancer-trap technique 61 sea urchin 236
egg 330, 336–9, 342–3 Eph receptors 205, 486–7, 491–2 somites 200
ascidian 244–5 ephrins 131, 205, 430, 486–7, 491–2 vertebrates 147–8, 149
Caenorhabditis 219–20 epiblast 103–6, 111, 140, 163–4, 314–15 amphibians 117, 146–7, 174
changes at fertilization 345–8 cavity formation 301 chick 117, 148
activation 346–8 polonaise movements 314, 315 mouse 148, 149
polyspermy prevention 345–6 epiboly 309, 312 zebrafish 147, 148
chicken 104 Xenopus 99 see also cell fate
cleavage and 296 zebrafish 102, 312, 313 fem gene 356
Drosophila 51 epidermal growth factor (EGF) 131, 510 fertilization 5, 342–8
evolution of 561 epidermal growth factor receptor (EGFR) Arabidopsis 257, 354
germ plasm 331–3 383, 441, 449, 544 cell-surface interactions 343–4
mouse 109, 138 epidermis 382–4 egg activation 346–8
sea urchin 235 epigenesis 4–5, 339 polyspermy prevention 345–6
size 338–9, 581 epimorphosis 536 sea urchin 345, 346
totipotent potential 339 epithelia 451–2, 520–1 Xenopus 97, 130–2, 345, 347
Xenopus 97, 129 epithelial-to-mesenchymal transition 303–4 fertilization membrane 345
zebrafish 101 ErbB receptors 495 fibroblast growth factor (FGF) 131, 164,
see also fertilization; oogenesis erythrocytes 377, 379–82 519, 570
egl genes and products 226, 392 estrogen 370–1 FGF-8 203
Elbows transcription factor 436 ets1 gene and product 239 FGF-9 350
electroporation 117 euchromatin 182 FGFA 304–5
Eleutherodactylus 580 even-skipped (eve) gene and product 67, neural development and 180–1,
embryogenesis 2–3 68–9, 71, 80 482–3
generative program 29–30 regulation of expression 369 notochord induction 248
human embryo shape changes 16 evolution 31, 556–9 receptors (FGFR) 453
nutrition effects in later life 511–12 of development 559–62 signaling pathway 184
phylotypic stage 563 developmental modification 562–79 somite formation and 189–90
see also specific organisms functional changes 576–8 vertebrate limb development and
embryological manipulation see ­microsurgical gene duplication 563–4, 565 413–16, 418, 427, 429
manipulation Hox gene changes 563–8 fibroblasts 367
embryology 4 limbs 569–74 senescence effect 530–1
embryonic pole 139 wings 574–5 FILAMENTOUS FLOWER (FIL) gene 271
embryonic stem cells (ES) 29, 121–3, 139, gastrulation 561–2 filopodia
399–400 multicellular ogranisms 560–2 Caenorhabditis 309
cancer and 522 timing of developmental processes 579–82 Drosophila 308
human 403 eye 444–5 sea urchin 304–6
empty spiracles gene and product 85 determination 23 synapse formation and 495
Endo-16 gene and product 240–1, 242 development 445–8 fins 569–70
regulation of expression 369 non-development in Astyanax 575 zebrafish 570–1
endocardium 457 lens 577 floor plate 479–80
endochondral ossification 517 formation 447–8 floral organ identity genes 278
endoderm 3, 15 placode 445–6 FLORICAULA (FLO) gene 276, 281
Caenorhabditis 224 regeneration, newt 398, 536 FLOWERING LOCUS C (FC) gene and
Drosophila 38 see also compound eye; retina ­product 283
vertebrates 94, 103 eyeless gene 450 FLOWERING LOCUS D (FD) gene and
chick 104 ­product 282–3
mouse 110–11, 140 F FLOWERING LOCUS T (FT) gene and
Xenopus 153, 311–12 fasciated mutation 281 product 282
endomesoderm specification, sea urchin 237, fass mutation 260, 273–4 flowers 276–7
238, 239, 241 Fat cadherin 514, 544 development 277–81, 284
608 Index

flowering control 281–3 Cre/loxP system 118 goosecoid gene and product 161, 163, 178,
sex determination 354–5 Hox genes 197–8, 208 243, 552–3
fog gene 356 genes 8–9, 13 Gorlin’s syndrome 521
follicle cells 51, 336 control of development 13, 17–19 gpa-16 gene and product 222–3
Follistatin protein 160, 179, 180 homologous 11 granulocyte colony-stimulating factor
forebrain 202 identification 10–11 (G-CSF) 379
see also brain development spontaneous mutations 11–13 granulocyte-macrophage colony-stimulating
Foxl transcription factors 154, 243, 403 see also developmental genes; maternal factor (GM-CSF) 379
Fragilis protein 333 genes; specific genes grasshopper 566–7, 581
Fringe protein 190, 439 gene silencing 13, 123–4, 220, 372 green fluorescent protein (GFP) 146, 384
Frizzled (Fz) receptor 78 by microRNAs 228 Gremlin gene and product 427
Frizzled-related protein (Frzb) 160, 185 genetic equivalence 19 Groucho 370
fruit fly see Drosophila melanogaster genetic mosaics 73 growth 16, 505–23
fruitless gene 353 genetics 8–9 coordination 512–13
Fugu rubripes 564 reverse 13 head region 572
fushi tarazu gene and product 67, 69, 71 genetic screening Hydra 548–9
developmental genes, Drosophila 39–40, 42 intercalary 541–2, 544–6
G misexpression screening 60, 61 long bones 517–19
Gal4 transcription factor 60, 61 mutagenesis screen 119, 120 molting 524, 527
ganglion mother cell 473 preimplantation 153 muscle 510, 519
gap genes genitalia 351 neuroendocrine control 513–17
expression 57–9, 68–9, 80 genital ridge 332, 334 nutrition effects 515–16
head gap genes 85 genome 10–11, 558 embryo nutrition 511–12
gastrula 98 ascidian 244 organ size control 509–11, 512–13, 514
gastrulation 3, 16, 173–4, 175, 302–15 Caenorhabditis 218 plants 522–3
Drosophila 38, 39, 306–9 genomic imprinting 339–42 tissue growth 506
evolution 561–2 plants 354–5 growth factors 379, 508
movements 302–15 genotype 9 hematopoietic 379
organizer inductive capacity changes germ band extension, Drosophila 38, 39, see also specific factors
175–9 307–8 growth hormone 516–17, 519
sea urchin 16, 233–4, 303–6, 316 germ cells 5, 329 growth hormone-releasing hormone 517
vertebrates 94, 309–15 development of 330–42 growth plates 517–19
chick 103, 104, 314–15 cell-cell interactions 333 Gryllus bimaculatus 544
mouse 103, 111–13, 148, 314 cell fate specification 331–3, 342 gurken gene and product 52–5
Xenopus 98–9, 147, 309–12, 313–14, 316 meiosis 335–8 gynogenetic embryos 340
zebrafish 102, 312, 313, 316 migration to gonad 334
GATA1 transcription factor 378–9, 380 chemical signals 334–5 H
GATA2 transcription factor 380 primordial 330 H19 gene 341
Gbx2 transcription factor 203 see also sex determination Halocynthia roretzi 247
gene duplication 563–4, 565 germ layers 3, 15–16, 145–65 see also ascidians
gene expression 17 Xenopus 150–4 Haltere mimic mutation 443
control of 17–21, 368–75 see also ectoderm; endoderm; mesoderm handedness see left-right asymmetry
chromatin changes 371–4 germline cyst 51 haploidy 6, 335
external signals 370–1, 375 germ plasm 330, 331–3 head
feedback loops 21 giant gene and product 57, 59 arthropods 567
transcription regulators 368–70 gibberellins 523 Drosophila 85, 567
differential 19 giraffe mutation 199 growth variation 572
DNA microarray analysis 114, 115 glial-cell-derived neurotrophic factor (GDNF) Hydra 549–51
evolutionary changes 557–8 454–5, 497 Xenopus 179
hindbrain 207 glial cells 468, 495 head gap genes 85
misexpression screening 60, 61 radial 475–6 head organizer 176
organizer region 178 Gli genes and products 422, 424–5 heart 457–9
plasticity 394–405 globin gene expression 379–82 regeneration, zebrafish 536, 546–7
protein synthesis 17–19 locus control region (LCR) 381–2 hedgehog gene and product 75–7, 131,
targeted 61 GLP-1 protein 223–4 436–8, 440
tracking in embryos 20 glycosylation 18 Hedgehog signaling pathway 75
see also specific genes gonadotropin-releasing hormone cancer and 521
gene knockdown 13, 123 (GnRH) 517 hematopoiesis 375–9
gene knock-outs 13, 109, 121 gonads 330, 350 hemidesmosomes 293, 383
 Index 609

hemoglobin 380–2 order of expression 84–5, 193–6 insects

Hensen’s node 105 somite identity and 191–7 compound eye 77, 444, 445, 448–51
inductive capacity changes during targets 565–6 limb regeneration 544–6
­gastrulation 177–8 tooth identity specification 459–61 paired appendages 567–8
regression 105, 106 zebrafish 193, 564, 570–1 patterning mechanisms 79–80
transplantation studies 177–8, 181–3 see also homeobox genes; homeotic genes planar cell polarity 77–9
hepatocyte growth factor (HGF) 510 human see also Drosophila
hermaphrodites 217, 354 embryo development 1–2, 95 in situ hybridization 20
hesC gene and product 239 shape changes 16 insulin-like growth factor-1 (IGF-1) 516,
heterochromatin 182, 372 embryonic stem (ES) cells 403 517, 519
heterochronic mutations 227 growth 511, 515–17 insulin-like growth factor-2 (IGF-2) 341,
heterochrony 579, 580 sex determination 349, 351 516, 517
hindbrain 202 hunchback gene and product 46–7, 57–9, insulin-producing cell generation 403–4
gene expression 207 68–9 integrins 293
Hox gene role 206–7 Hutchinson–Gilford progeria syndrome 530 intercalary growth 541–2, 544–6
rhombomeres 203–6 hyaline layer 298 intercalation 308
see also brain development hyaluronan 433–4 circumferential 546
Hippo signaling pathway 513, 514, 544 Hydra 469, 548, 558 medio-lateral 310
histones 374 budding 549 proximo-distal 541–2, 544–6
holoprosencephaly 28, 447 growth 548–9 radial 310, 312
homeobox genes 194 regeneration 535–6, 548–52 intercellular signals
nervous system development 480, 481 genetic control 551–2 Drosophila 53
plants 262 hypostome region 549–51 vertebrates 131
see also homeotic genes; Hox genes hydrostatic pressure 320–1 interleukin-6 510
homeodomain 194 hypoblast 104, 137–8 intermediate filaments 291
homeosis 82 intestinal epithelia 384–5
homeotic genes 81–6 I intracellular signaling 24, 26
flower development role 277–9 identical twins 152 intracytoplasmic sperm injection (ICSI) 344
see also homeobox genes; Hox genes imaginal discs, Drosophila 39, 72–4, 435, 444 invagination
homeotic transformation 197–8 leg 439–41 Drosophila 306–7
homologous genes 11 positional values 442–3 gastrulation evolution and 561–2
homologous recombination 122, 123 size determination 512–13, 514 sea urchin 305
homology 567 wing 436–8 in vitro fertilization (IVF) 153, 344
homothorax gene 441, 442 fate map 436 involution 98, 309, 311
homunculus 5 immunoglobulin superfamily 293 Iroquois gene complex 449
hormones 370–1, 514–17 imprinting 339–42 isthmus 202, 203
see also growth factors plants 354–5
horse limbs 572, 573 indeterminate growth 269–70 J
housekeeping activities 17 Indian hedgehog protein 517–19 JAK-STAT pathway 52
Hox code 197 indoleacetic acid (IAA) see auxin jellyfish, transdifferentiation 398
Hox genes 81–6, 194, 374, 558 induced pluripotent stem cells (iPS cells) Johannsen, Wilhelm 9
activation pattern 198–9 400, 401, 403–4, 405 joint formation 433–4
axial patterning 195–9 induction 8, 23–4 Jun N-terminal kinase (JNK) 308
branchial arches 207, 208 ascidians 247–9 juvenile hormone 525
Caenorhabditis 226 competence 24, 25
chick 196 mesoderm K
deletion/overexpression studies 197–8 chick 163–4 keratinocytes 382–3
Drosophila 81–6, 194, 442–3, 564, 565–6 mouse 164–5 keratins 383
evolutionary changes 563–8 Xenopus 150–3, 155–9, 166 kidney tubules 454–6
body plan elaboration 564–7 zebrafish 160–1 KIT gene and product 391
gene duplication 563–4, 565 neural plate 179–85 kiwi 572, 573
insect paired appendages 567–8 permissive/instructive 24 Klinefelter syndrome 349
limb development roles 413, 415, 428–30, primary embryonic induction 175–6 knirps gene and product 57, 59
431–3, 542 response to 25 knock-outs see gene knock-outs
limb regeneration 542 sea urchin 236–7 Knot protein 162
mouse 194, 195–7, 374 vulval development, Caenorhabditis KNOTTED-1 gene and product 268
nervous system development 483 229–31 Koller’s sickle 104, 106, 136, 164
hindbrain 206–7 ingression 105 Krox20 gene and product 206–7
neural crest cells 392 inner cell mass, mouse 138–9 Krüppel gene and product 57, 58–9, 80
610 Index

L insect leg development 435, 439–41, Meis gene family 418

labial gene 567 444, 568 melanocytes 391
LAG-2 protein 356 regeneration 25, 536–46 Mendel, Gregor 6
laminin 495 amphibians 537–44 meristems 256, 265–76
larva 579–80 insects 544–6 cell fate 267–70
Caenorhabditis 217–18 see also vertebrate limb development floral meristem 265, 276, 278, 281–3
Drosophila 38–9 LIM protein 487 inflorescence meristem 276
sea urchin 233 lineage see cell lineage lateral shoot meristem 265
lateral geniculate nuclei (LGN) 490, 498 lin genes and products root apical meristem 265, 273–5
lateral inhibition 27, 29 lin-3 230, 231 shoot apical meristem 265–71
compound eye 449 lin-4 227 transition to floral meristem 281–3
neural development 471–2, 473–5 lin-14 227–9 stem cells 265–7
root hairs 275 lin-39 226, 229 mesectoderm 390
lateral motor column (LMC) 481–2, 487 liver 509–10 mesenchymal-to-epithelial transition 455
lateral plate mesoderm 99, 458 regeneration 509–10, 536 mesenchyme 105
LEAFY (LFY) gene and product 276, 282 Lmx1 genes and products 426–7, 438, 487, 574 ascidians 248–9
leaves locus control region (LCR) 381–2 primary 234, 303–5
arrangement on stem 272–3 longevity 528, 529 mesendoderm 102, 147
axes 270–1 long-germ development 79–80, 581 mesoderm 3, 15
directed dilation 322–3 Lunatic fringe enzyme 189 Caenorhabditis 224
phyllotaxis 271 lung development 453–4 Drosophila 38, 40
primordia 270–1, 272 lymphatic vessels 457 invagination 306–7
left-right asymmetry lymphocytes 377 induction see mesoderm induction
Caenorhabditis 221–3 nervous system patterning and 185–6
chick 142–4 M patterning of 156–7, 158–65
mouse 141–5 mab-5 gene 226 threshold responses 161–3
Xenopus 142–4 macho-1 gene and product 246, 247, 249 vertebrates 94, 103
Lefty protein 243 macrophage colony-stimulating factor chick 104, 105, 107, 148, 163–4
lens 577 (M-CSF) 379 heart 458
formation 447–8 MADS box 278, 279 lateral plate mesoderm 99, 458
placode 445–6 maize 269, 270, 354 mouse 164–5
regeneration, newt 398, 536 Malpighi, Marcello 4–5 paraxial mesoderm 187
Lepidoptera see butterfly mammals 103 prechordal plate mesoderm 105
let-23 gene and product 230, 231 apoptosis 392, 393 pre-somitic mesoderm 187, 191
leukemia inhibitory factor (LIF) 400, 455 cloning 396 Xenopus 147, 150–63, 311–12
leukocytes 377 dosage compensation 357–8, 373 zebrafish 160–1
Lewis, Edward 82 growth 516 mesoderm induction
Lgr5 gene and product 384–5 imprinted genes 339–42 chick 163–4
life cycle middle ear 577, 578 mouse 164–5
Arabidopsis 257 nervous system regeneration 547 Xenopus 150–3, 155–9, 166
ascidian 245 sex determination 349–51 timing 154
Caenorhabditis 216 visual system 490 zebrafish 160–1
Drosophila 36–40, 581 see also human; mouse mesonephros 350
evolution of 581 Manduca sexta 524 Mesp2 gene and product 199
sea urchin 233, 580 Mangold, Hilde 8 message transport organizer region (METRO)
vertebrates 94–6 marginal zone 147 130
chicken 105 maternal factors 40 messenger RNA (mRNA) 18
mouse 109 Drosophila 40–65 metamorphosis 505, 513–15, 523–4, 579–80
Xenopus 97 body axis establishment 40, 41–51 control of 524–7
zebrafish 101 maternal determinant localization during Drosophila 39, 436, 440
see also metamorphosis oogenesis 51–65 metastasis 520, 522
limb buds 412–13, 416 Xenopus 129–30, 133 Metazoa 559
patterning 416 zebrafish 129–30 origin of 559–60
self-organization 430–1 maternal genes 40 microfilaments 291
see also vertebrate limb development maturation-promoting factor (MPF) 347–8 microRNAs (miRNA) 18, 228–9, 271
limbs MEF2 transcription factor 278 cancer and 522
evolution 31, 568, 569–74 meiosis 6, 330, 335–8 microsurgical manipulation
limb loss 571 timing 355–6 chick 116
 Index 611

mouse 116–17 Hox genes 194, 195–7, 374 N

suitability of different vertebrate intestinal epithelia 384–5 nanos gene and product 44, 46–7,
embryos 116 left-right asymmetry 141–5 51, 54
Xenopus 114–16 life cycle 109 Nasonia 80
microtubules 291 limb development 414, 418, 422, 430 necrosis 392
reorganization in Drosophila oocyte 53–5 lineage tracing 117–18 negative feedback 21, 30
mid-blastula transition 155–6 mesoderm induction and patterning nematodes 215–32
midbrain 202 164–5 apoptosis 392
see also brain development microsurgical manipulation 116–17 directed dilation 321–2
middle ear 577, 578 muscle growth 519 germ cell development 331
miles apart gene 458 mutagenesis screen 119 see also Caenorhabditis
Minute technique 72, 73, 513 neurogenesis 477 Nematostella 558, 560
misexpression screening 60, 61 neurulation 112, 113 neotony 579, 580
mitochondrial DNA 346 polarity 299–300 nephrons 456
mitogen-activated protein kinase (MAPK) postimplantation development 110–11 nerve growth factor (NGF) 497
181, 400 sex determination 349, 355 nervous system 468–9
mitosis 6, 508 somite patterning 199 axon navigation 484–92
see also cell cycle specification 138, 139 cell identity specification 469–84
mitotic recombination 73 transgenic techniques 117–18, 121–3 Drosophila 470–3, 484
mitotic spindle 297–8 turning of the embryo 113 vertebrates 473–84
model organisms 9–11 mrf4 gene and product 386 regeneration in mammals 547
phylogenetic tree 10 MSL genes 359 see also brain development; neurons;
molting 524, 527 Msx1 gene and product 388, 460–1, ­neurulation; spinal cord
MOM-2 protein 224, 225 538, 540 netrins 457, 486, 488–9
MOM-5 protein 224, 225 Müllerian ducts 350 neural crest cells 99, 186
MONOPTEROS (MP) gene 262 Müllerian-inhibiting substance 350 differentiation 389–92
Monosiga brevicollis 561 muscle 385–9 Hox gene expression 392
monozygotic twinning 152 gene expression, Xenopus 154–5 migration 186, 207–8, 318–20, 391, 475
Morgan, Thomas Hunt 8 growth 510, 519 neural folds 99, 174, 316, 317
morphallaxis 536 limb muscle patterning 431–3 neural plate 99, 107, 175, 473–5
morphogenesis 16, 289 specification, ascidians 246–7 cell shape changes 317
branching 452–4 muscle-specific kinase (MuSK) 494–5 induction 179–85
morphogenetic furrow 448 mutagenesis patterning 185–6
morphogens 25–7 Drosophila 42 neural tube 3, 95, 107, 174, 316, 475
gradients 421 mouse 118 formation see neurulation
threshold concentrations 27, 46 mutagenesis screens 119 neuregulin-1 496
see also specific morphogens mouse 119 neuroblasts 470, 472–3
morpholino antisense RNA 123 zebrafish 119, 120 neuroectoderm 469–70
morula 110, 299–300 mutations 11–13 Drosophila 40
mosaic development 6, 22–3, 215 conditional 12 neuroepithelium 471
X inactivation effect 357–8 dominant 11–12 neurogenesis 388–9, 470
motor neurons 481–2, 483 heterochronic 227 Drosophila 470–3
axonal guidance 486–7 maternal-effect 40 vertebrates 475–8
neuronal cell death 496–7 recessive 12 neurogenin 475
mouse 9–10, 95, 109–13 semi-dominant 11–12 neuroligins 495
body axes 138–41 spontaneous 119 neuromuscular junctions 493, 494, 495
patterning 195–6 temperature-sensitive 12 development 495
body plan 94 see also specific genes and organisms neurons 468, 469
chimeric 150 myf5 gene and product 386–7 migration 476–7
cleavage 109–10, 138 myoblasts 385, 386, 431–3 motor neurons 481–2, 483
compaction 299–300 myocardium 457 neuronal cell death 496–7
egg 109, 138 myoD gene and product 367, 372, post-mitotic 477
embryonic stem (ES) cells 399–400 385–7, 397 regeneration 547
fate map 148, 149 myogenin gene and product 386, 387 retinal 489–92
gastrulation 103, 111–13, 148, 314 myoplasm 246–7 see also axons; neurogenesis
gene expression regulation 368 myostatin gene and product 510 neurotrophins 497
germ cell development 333, 334, 335 myotome formation 200 neurula 99
heart 458–9 myotubes 385, 386 patterning 208–9
612 Index

neurulation 3, 17, 94, 174–5, 316–17 optic chiasm 489–90 leaves 270–3
chick 107, 317 optic nerve 489–90 roots 273–5
mouse 112, 113 optic tectum 489, 499 positional information 25–7, 29
Xenopus 99–100, 174 optic vesicle 445–7 somites 187–97
zebrafish 103 oral-aboral axis, sea urchin 234, 241–3 cell fate determination 199–201
newt organizer region 175 vertebrate limb development
lens regeneration 398 genes expressed 178 digit separation 434
limb regeneration 537, 538 Hydra hypostome region 549–51 Hox gene role 428–30, 431–3
Nieuwkoop center 133–5 inductive capacity changes during limb bud 416–20
Nkx2.5 gene and product 458, 459 ­gastrulation 175–9 muscle 431–3
Nodal protein 142, 179, 242–3 plants 265–7, 273 self-organization 430–1
signaling 143, 164–5 sea urchin 236–7, 242–3 Sonic hedgehog (Shh) role 420–4, 429
Nodal-related proteins (Ndr) 135–6, 160–1, 163 see also Hensen’s node; Spemann ­organizer see also segmentation; specific organisms
see also Xenopus Nodal-related (Xnr) genes organogenesis 3, 411 Pax genes 194, 200, 201, 388, 560
and products organ size control 509–11, 512–13, 514 compound eye development and 447,
Noggin gene and product 160, 179, 180, Xenopus 99–100 450–1
185, 200 see also specific organs kidney tubule development and 455
non-cell-autonomous gene effects 150 orthodenticle gene and product 85, 207, 567 vertebrate limb development and 433
non-coding RNAs (ncRNA) 341 oskar gene 51, 53–4, 55, 331, 332 Pdx1 transcription factor 403, 404
No-ocelli transcription factor 436 osteoblasts 517 P elements 60, 61
No tail (Ntl) gene and product 161 Otx2 gene and product 203, 207 periclinal divisions 260, 261
targets 162 ovary 330 perivitelline space, Drosophila 48
Notch-Delta signaling pathway see oviducts 350 P granules 219, 331, 332
­Delta-Notch signaling pathway oxidative damage 530 pha-4 gene 226
Notch protein 144 PHABULOSA (PHAB) gene 271
cancer and 521–2 P PHAVOLUTA (PHAV) gene 271
compound eye development and 449 p53 tumor-suppressor gene 520 phenotype 9
leg development and 439, 441 pair-rule gene expression 66–70, 80 phenotypic suppression 84
neural crest cell differentiation role 391 pancreatic cancer 521 photoperiodism 282
neurogenesis role 471–2, 475 Panderichthys 570 phyllotaxis 271, 272, 273
signaling pathway 190, 383 Paneth cells 384 phylogeny 559
notochord 95–6, 312–13 paralogs 194 phylotypic stage 563
ascidians 247–9 parasegments, Drosophila pie-1 gene and product 331, 332
chick 105–7, 177 boundary stabilization 74–7 PIN1 protein 272, 274
directed dilation 321 establishment 66–70 PIN7 protein 261–2
somitic cell fate and 200–1 parathyroid-hormone-related protein (PHRP) pipe gene and product 48
Xenopus 174, 313–14 517–18 PISTILLATA gene and products 277, 278, 279
Notophthalmus viridescens 537 Parazoa 558–9 Pitx1 transcription factor 413, 576
nucleosomes 182 PAR (partitioning) proteins 53, 220–1, 246 Pitx2 transcription factor 243, 459
nurse cells 51, 54 pattern formation 14–16, 209, 510 placenta 96
butterfly wing markings 441–2 mouse 109–10
O Drosophila Placozoa 560
obesity 512 compound eye 448–50 planar cell polarity 15, 77–9
Oct4 transcription factor 333 imaginal discs 436–41, 444 loss of 79
ocular dominance columns 498, 499 leg 439–41, 444 plant development 255–6
omb gene and product 437 wing 438–9 embryonic development 258–64
ommatidia 448–50 genes involved 167 flower development 276–83
oncogenes 513, 520 Hox genes in axial patterning 195–9 homeotic gene role 277–9
Oncopeltus 79 lateral inhibition 27, 29 sex determination 354–5
ontogey 559 mesoderm genomic imprinting 354–5
oocyte 336–8 chick 163–4 growth 522–3
decline in numbers with age 338 mouse 164–5 leaf arrangement on stem 272–3
primary 336 threshold responses 161–3 meristems 265–71, 273–6
see also egg Xenopus 156–7, 158–63 cell fate 267–70
oogenesis 336–8 zebrafish 160–1 stem cells 265–7
Drosophila 51–65 nervous system 185–6 regeneration capacity 263
antero-posterior polarity 52–4, 56 neurula stage 208–9 root hairs 275
dorso-ventral polarity 54–5, 56 plants root tissues 273–5
oocyte cytoskeleton reorganization 53–5 flowers 277–81 see also Arabidopsis thaliana
 Index 613

Platynereis 469 gene regulatory proteins 19 retino-tectal map 489–92

PLETHORA gene and products 274 genetic control 17–19 refinement 497–9
pmarl1 gene and product 238, 239 post-translational modification 18 Ret receptor 455
polar body 97, 336, 338 tissue-specific 17 reverse genetics 13
polarity 14–15 prothoracicotropic hormone (PTTH) 514, 515, REVOLUTA (REV) gene 271
ascidian egg 244–5 524, 525 rhomboid gene and product 62, 441
Caenorhabditis egg 219 proto-oncogenes 520 rhombomeres 203–6
Hydra 549 protostomes 232, 560 ribosomal RNA (rRNA) 18
mouse morula 299–300 proximo-distal axis RNA 18
planar cell polarity 77–9 Drosophila leg disc 439–41 antisense 19, 220
sea urchin egg and blastula 235, 298–9 leaves 270–1 morpholino 123
vertebrate limb development 419–20 vertebrate limb 412, 416–19 interference (RNAi) 19, 123–4, 220
see also body axes regeneration 541–4 microRNAs 18, 228–9, 271
pole cells 331 PU.1 transcription factor 378–9 cancer and 522
pole plasm 331 python, limb loss 571 non-coding RNAs (ncRNA) 341
polonaise movements 314, 315 processing 18
Polycomb gene family 84 Q RNA polymerase II 368
polydactyly 422, 423, 431 quail 177, 390 Robo2 receptor 455
preaxial 422, 423 quiescent center 273 roof plate 479, 480
polyspermy prevention 345–6 root hairs 275
polytene chromosomes 507, 527 R root tissues 273–5
Pomocanthus semicirculatus 432 radial glial cells 475–6 rotundifolia mutation 322, 323
POP-1 protein 224–5 radial symmetry 558–9 Roux, Wilhelm 6, 135
positional identity 192, 193–5 Rana temporaria 544
Caenorhabditis 226 RAR δ-2 receptor 543 S
plant leaves 271 reaction-diffusion mechanism Sall genes 423–4
positional information 25–7, 29 431, 432 Sasquatch mutation 423
imaginal discs 442–3, 444 reactive oxygen radicals 530 satellite cells 388, 519
morphogen gradients 421 recombination 336 SCARECROW gene 273, 275
root hairs 275 redundancy 30 Schleiden, Matthias 5
vertebrate limb development 416–22, reeler mutation 477 Schwann, Theodor 5
426–8 regeneration 535–6 Schwann cells 547
Sonic hedgehog (Shh) role 420–4, 429 Hydra 535–6, 548–52 Scleraxis transcription factor 201
see also body axes, establishment of; limbs 25, 536–46 sclerotome formation 200–1
­intercalation; positional identity amphibians 537–44 SCRAMBLED protein 275
positive feedback 21, 372 nerve influence 539–40 scute gene 474
postbithorax mutation 443 liver 509–10, 536 SDC-2 protein 359
posterior dominance 84, 197 mammalian nervous system 547 SDF-1 protein 335
posterior marginal zone 136, 137, 164 newt lens 398, 536 sea squirts see ascidians
Prader-Willi syndrome 341 plants 263 sea urchins 232–44
preaxial polydactyly 422, 423 zebrafish heart 536, 546–7 body axes 234–5, 241–3
preformation 4–5 regenerative medicine 2, 394 cleavage 233, 234, 241–2
preimplantation genetic screening 153 stem cell potential 400–3 egg 235
primary cilium 422, 424 regulation 7–8, 149–50 endomesoderm specification 237, 238,
primary embryonic induction 175–6 capacity at neurula stage 208–9 239, 241
primitive streak 103 chick 150 fate map 236
chick 105, 106, 136–8, 148, 163–4, 314–15 Xenopus 149 fertilization 345, 346
formation 314–15 regulative development 22–3, 215 gastrulation 16, 233–4, 303–6, 316
mouse 111, 148, 164–5, 314 sea urchin 236–7 gene expression regulation 369
primordial germ cells 330 see also regulation life cycle 233, 580
primordium 265 reporter gene 20 micromere specification 240
leaf 270–1, 272 repressors 369 organizer regions 236–7, 242–3
proamniotic cavity 140 retina 446 polarity 235, 298–9
Prod1 gene and product 541–2, 543–4 neural connections 489–92 skeletogenic pathway genetic control
programmed cell death see apoptosis refinement 497–9 238–41
proneural clusters 470–2 retinoblastoma (RB) gene and product 387, SED1 protein 343–4
proneural genes 470 520, 521, 538 segmentation, Drosophila 38, 39
proteins 17–19 retinoic acid 190–3, 199, 370, 422, 458 parasegment boundary stabilization 74–7
evolutionary changes 557–8 limb regeneration and 542–4 parasegment establishment 66–70
614 Index

segmentation gene expression 70–81 evolutionary variation 57 stickleback spines 575–6

segment identity specification 81–6 limb regeneration and 540 string gene and product 508–9
semaphorins 319, 457, 486 lung development and 453 Strongylocentrotus purpuratus 232
senescence nervous system development and 479–81 see also sea urchins
cell senescence 530–1 SoxE transcription factors 391 Styela 246
genetic influences 529–30 Sp9 transcription factor 539 see also ascidians
sensory bristles, Drosophila 473, 474 spalt gene and product 437 superior colliculus 490
sensory organ precursor (SOP) 474 Spätzle protein 48 SUPERMAN gene 279
SEPALLATA (SEP) genes and products 278–9 specification 21–2 Survivin protein 497
septate junctions 299 germ cells 331–3, 342 Sxl gene and product 359
Serrate protein 131, 144, 439, 449 neuronal identity 469–84 synapses 468, 493
sevenless gene and product 450 Spemann, Hans 8 formation 493–500
SEX-1 protein 353 Spemann organizer 8, 17, 23, 98, 134–5 reciprocal interactions 493–6
Sex combs reduced gene 84 genes expressed 178 synaptic cleft 493
sex determination 348–60 inductive capacity changes during synaptic vesicles 493
Caenorhabditis 353–4, 356 ­gastrulation 175–9 syncytial blastoderm 36, 41
Drosophila 351–3, 355–6 mesoderm induction and patterning 156–7, syncytium 36, 40–1
flowering plants 354–5 159–61, 163 SYS-1 protein 225
mammals 349–51, 355 transplantation study 157
Y chromosome 349 sperm 342–3 T
Sex-lethal (Sxl) protein 351, 352 acrosomal reaction 343, 344 tailbud, Xenopus 99, 100
SHOOT MERISTEMLESS (STM) gene 262, capacitation 343 tailless gene and product 57, 59
263, 266 see also fertilization Talerpeton 570
Short gastrulation (Sog) protein 65 spermatogenesis 337, 338 target of rapamycin (TOR) pathway 514–15
short-germ development 79–80, 581 spinal cord 183–5 TATA box 369, 387
short interfering RNA (siRNA) 220 antero-posterior pattern 482–3 Tbx genes and products 414, 423–4
SHORT-ROOT (SHR) protein 275 cell differentiation 478–80 TCF transcription factors 26, 370
siamois gene 135 chick 183 teeth 459–61
signaling centers 133–6 dorso-ventral pattern 478–82 telomeres 531
signal transduction 24, 26 motor neurons 481–2, 483 tendons 433
situs inversus 141 mouse 183 teratocarcinoma 402, 522
skin 382 Xenopus 183, 475 testis 330, 350
skn-1 gene and product 224 zebrafish 183–5 testosterone 350–1, 370
Slit proteins 455, 486 spiracles 452 tetraploidy 512
Smad signaling proteins 158, 159, 161 Spitz protein 449 TFIID transcription factor 387
small interfering RNAs (siRNAs) 19 spleen 509, 510 thale-cress see Arabidopsis thaliana
snail gene and product 62, 186, 303–4, 307 Splotch mutants 201 therapeutic cloning 402
snakes, limb loss 571 Sprouty gene and product 452–3, 455 thrombin 538
somatic cell nuclear transfer 395–6 Sry gene 349, 350 thyroid hormones 526
somatic cells 5 Staufen RNA-binding protein 54, 55 thyroid-stimulating hormone 525–6
plants, regeneration potential 263 Steel gene 391 tight junctions 293, 300–1
somatic transgenesis 123 Stella protein 333 TILLING (targeting-induced local lesions in
somatostatin 517 stem-cell factor (SCF) 391 genomes) 120
somites 95, 173 stem cells 2 tinman gene 459
cell fate determination 199–201 asymmetric division 29, 30, 31, 377–8 tissue growth 506
chick 105–8, 187, 188–9 blood cell formation 375–8 T lymphocytes 377
formation 187–91 embryonic (ES) 29, 121–3, 139, toll gene and product 48, 49–50
sequence of 188–90 399–400 tolloid gene and product 62, 65
identity 191–7 cancer and 522 Toll signaling pathway 50
mouse 113 epidermal 383–4 tooth identity 459–61
neural crest cells 319–20 induced pluripotent stem cells (iPS cells) topless-1 gene and product 262
patterning within 199–201 400, 401, 403–4, 405 torso gene and product 44, 47–8, 55
Xenopus 187 intestinal epithelia 384–5 Torso-like protein 48, 51, 55
zebrafish 103 meristem 265–7 totipotency
Sonic hedgehog (Shh) 28, 144, 201 multipotent 29, 375–8 egg 339
eye development and 447–8 muscle 388 plants 256
non-development in Astyanax 575 neural 388–9 trachea
limb development and 414, 415, 420–4, pluripotent 29, 399–400 Drosophila 452–3
427, 429 regenerative medicine potential 400–3 vertebrates 453
 Index 615

transcription 18, 368 Urbilateria 559 WEREWOLF protein 275

control of 368–70 uteric bud 454–6 Werner syndrome 530, 531
transcription factors 19, 368–70, 371–2 white spotting (W) gene 391
blood cell differentiation 378–9 V wing development
general 369–70 vasa mRNA 333 butterfly 566, 568
tissue-specific 370 vascular epithelian growth factor (VEGF) wing markings 441–2
Xenopus blastula 161–2 304–5, 456–7 chick
see also specific factors vascular system 456–7 evolutionary comparisons 574–5
transdifferentiation 388, 397–9, 538–9 vasculogenesis 456 see also vertebrate limb development
replacement cell generation 404 vas deferens 350 Drosophila 77, 435, 565–6
transfer RNA (tRNA) 18 vegetal pole 3, 97, 129 axis establishment 438–9
transformer (tra) genes 351, 353 VegT transcription factor 130, 135, 153 compartment boundaries 72–4, 436–9
transforming growth factor-β (TGF-β) family mesoderm induction and patterning role pigment pattern in D. biarmipes 576
131, 178–9, 455, 574 158–9 positional signals 436–8
liver growth and 510 veins 457 evolutionary comparisons 574–5
mesoderm induction 157–8 ventral closure 108, 308–9 wingless gene and product 74–7, 80, 131,
see also bone morphogenetic proteins ventralized embryos 50, 62 438–40, 558
(BMP); Nodal protein ventricular zone 475 see also Wnt protein signaling
transgenic techniques 20, 109, 120–3 vernalization 282 Wnt/β-catenin asymmetry pathway 225
chick 121 vertebrae 565 Wnt protein signaling 26, 75–6, 130, 131,
Drosophila 60, 61 vertebrate limb development 412–35 179, 558
mouse 117–18, 121–3 apical ectodermal ridge role 415–16 antero-posterior gradient 179
plants 263, 264 axes 412 cancer and 521
transient transgenesis 123 positional information 416–22, 426–8 cell differentiation and 370, 389, 391
translation 18, 368 digits 422–6 convergent extension role 310, 316
transplantation studies 115–16, 149 polydactyly 422, 423, 431 Hydra 551
Hensen’s node 177–8, 181–3 growth regulation 510–11 limb development 426–7, 433
Hydra hypostome region 549–51 Hox gene roles 413, 415, 428–30, 431–3 lung development 453
somatic cell nuclear transfer 395–6 integration 427 mesoderm induction and patterning
Spemann organizer 157 joint formation 433–4 158–9, 164
transposons 60 long bones 517–19 somite patterning 201
tribbles gene and product 509 neuronal guidance 486–7 see also canonical Wnt/β-catenin ­pathway;
Tribolium 79–80 patterning 416–22, 428–30 wingless gene and product
trichome patterns, Drosophila 566 muscle 431–3 Wolffian ducts 350
Trichoplax 560 self-organization 430–1 WOX2 gene 262
triploblasts 15, 559 polarizing region 419–20, 428–9 WOX8 gene 262
trisomy 338 position and type of limb 413–15 WRM-1 protein 225
Trithorax gene family 84 programmed cell death role 434 WT1 gene and product 455
trophectoderm 110, 138–9 progress zone 417 wunen gene 335
trunk organizer 176 Sonic hedgehog (Shh) role 414, 415, 420–4, WUSCHEL (WUS) gene 266
Trunk protein 48, 55 427, 429
TRYPTYCHON protein 275 timing model 417–18 X
Tsix gene 358 two signal model 418–19 X chromosomes 349, 350, 351–3
tumor necrosis factor α (TNF-α) 510 vestigial gene 12, 439 X inactivation 357–9, 372–3
tumor suppressor genes and products Vg-1 signaling protein 130, 136–7, 150, Xenopus 3, 96–100
513, 520 157, 164 body axes 129–36
Turner syndrome 349 visual centers 489–92, 497–9 cleavage 14, 98
twinning 150 visual cortex 490, 498 developmental stages 2, 3
monozygotic 152 vitelline envelope, Drosophila 48–9 egg 97, 129
Twisted gastrulation (Tsg) protein 65 vitelline membrane embryo development 95
twist gene and product 62, 63, 307 Drosophila 48 fate map 117, 146–7, 174
type 1 diabetes 403–4 sea urchin 345 fertilization 97, 130–2, 345, 347
Xenopus 97 gastrulation 98–9, 147, 309–12, 313–14, 316
U vulval development, Caenorhabditis 218, germ cell development 332
Ultrabithorax (Ubx) gene and product 82–3, 229–31 germ layers 147, 150–4
84, 443, 564, 568 convergent extension 311–12
evolutionary changes 565–7 W mesoderm induction and patterning
UNUSUAL FLORAL ORGANS (UFO) gene Warts (Wts) protein kinase 513 150–63, 166
and product 279 Weismann, August 5, 6 head formation 179
616 Index

heart 458 mesoderm induction and patterning role germ cell development 333, 334, 335
left-right asymmetry 142–4 158–9 heart 458
life cycle 97 Xist gene 358 regeneration 536, 546–7
limb regeneration 540 X-linked genes 356–9 Hox genes 193, 564
metamorphosis 525–6 xol-1 gene 353, 359 life cycle 101
microsurgical manipulation 114–16 Xwnt-8 protein 159–60 mesoderm induction and patterning
mid-blastula transition 156 Xwnt-11 protein 130, 132, 133 160–1, 162
neural induction 179–86 mid-blastula transition 156
default model 180 Y mutagenesis screen 119, 120
neurula stage patterning 208–9 Y chromosome 349 neurulation 103
neurulation 99–100, 174 yellow gene 576 zerknüllt gene and product 62
notochord 174, 313–14 yolk sac 108 zinc-finger nucleases (ZFNs) 120
directed dilation 321 Yorkie (Yki) transcription factor 513 zona limitans intrathalamica 202, 203
organizer region see Spemann organizer zona pellucida 109, 343
organogenesis 99–100 Z zone of polarizing activity (ZPA) 419–20
regulation 149 zebrafish 9, 10, 101–3 zone of polarizing activity regulatory
somatic cell nuclear transfer 395–6 body axes 129–36 ­sequence (ZRS) 423, 540
somite formation 187 cleavage 101, 102 zootype 564–5
spinal cord 183, 475 embryo development 95 ZP3 glycoprotein 344
transient transgenesis studies 123 fate map 147, 148 zygote 5
Xenopus Frizzled 7 (Xfz7) receptor 133 fin development 570–1 zygotic genes
Xenopus Nodal-related (Xnr) genes and gastrulation 102, 312, 313, 316 Drosophila 42, 43, 56–66
­products 136, 155, 157 gene expression detection 20 timing of expression, Xenopus 155–6