Methods in

Molecular Biology 1394

Jörg Reinders Editor

Proteomics
in Systems
Biology
Methods and Protocols
METHODS IN MOLECULAR BIOLOGY

Series Editor
John M. Walker
School of Life and Medical Sciences
University of Hertfordshire
Hatfield, Hertfordshire, AL10 9AB, UK

For further volumes:

http://www.springer.com/series/7651
Proteomics in Systems Biology
Methods and Protocols

Edited by

Jörg Reinders
Institute of Functional Genomics, University of Regensburg, Regensburg, Germany
Editor
Jörg Reinders
Institute of Functional Genomics
University of Regensburg
Regensburg, Germany

ISSN 1064-3745 ISSN 1940-6029 (electronic)

Methods in Molecular Biology
ISBN 978-1-4939-3339-6 ISBN 978-1-4939-3341-9 (eBook)
DOI 10.1007/978-1-4939-3341-9

Library of Congress Control Number: 2015957400

Springer New York Heidelberg Dordrecht London

© Springer Science+Business Media New York 2016
This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is
concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction
on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation,
computer software, or by similar or dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not
imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and
regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to
be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty,
express or implied, with respect to the material contained herein or for any errors or omissions that may have been made.

Printed on acid-free paper

Humana Press is a brand of Springer

Springer Science+Business Media LLC New York is part of Springer Science+Business Media (www.springer.com)
Preface

Proteomics techniques have constantly been developed further through the last decade and
have been applied successfully for all kinds of samples and biological or medical questions.
Nowadays, they are established methods in many research labs, and proteomic studies can
be accomplished with good reliability and coverage on a routine basis. Therefore,
proteomics can be used as a powerful tool in functional genomics and systems biology
studies. Current challenges are thus the implementation of proteomic analyses in these
comprehensive studies. This applies for both sample generation and preparation to ensure
consistency over several levels of analyses like genomics, transcriptomics, and metabolomics
and integration of the multilevel data to generate biological knowledge. This book gives an
overview of contemporary quantitative proteomics methods and data interpretation
approaches and also gives examples of how to implement proteomics into systems biology.

Regensburg, Germany Jörg Reinders

v
Contents

Preface. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v
Contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix

1 Multiplexed Quantitative Proteomics for High-Throughput

Comprehensive Proteome Comparisons of Human Cell Lines . . . . . . . . . . . . . 1
Amanda Edwards and Wilhelm Haas
2 Sample Preparation Approaches for iTRAQ Labeling and Quantitative
Proteomic Analyses in Systems Biology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Christos Spanos and J. Bernadette Moore
3 Two Birds with One Stone: Parallel Quantification of Proteome
and Phosphoproteome Using iTRAQ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
Fiorella A. Solari, Laxmikanth Kollipara, Albert Sickmann,
and René P. Zahedi
4 Selected Reaction Monitoring to Measure Proteins of Interest
in Complex Samples: A Practical Guide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
Yuehan Feng and Paola Picotti
5 Monitoring PPARG-Induced Changes in Glycolysis by Selected Reaction
Monitoring Mass Spectrometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
Andreas Hentschel and Robert Ahrends
6 A Targeted MRM Approach for Tempo-Spatial Proteomics Analyses . . . . . . . . 75
Annie Moradian, Tanya R. Porras-Yakushi, Michael J. Sweredoski,
and Sonja Hess
7 Targeted Phosphoproteome Analysis Using Selected/Multiple Reaction
Monitoring (SRM/MRM) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
Jun Adachi, Ryohei Narumi, and Takeshi Tomonaga
8 Testing Suitability of Cell Cultures for SILAC-Experiments
Using SWATH-Mass Spectrometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
Yvonne Reinders, Daniel Völler, Anja-K. Bosserhoff, Peter J. Oefner,
and Jörg Reinders
9 Combining Amine-Reactive Cross-Linkers and Photo-Reactive Amino Acids
for 3D-Structure Analysis of Proteins and Protein Complexes . . . . . . . . . . . . . 109
Philip Lössl and Andrea Sinz
10 Tissue MALDI Mass Spectrometry Imaging (MALDI MSI) of Peptides . . . . . 129
Birte Beine, Hanna C. Diehl, Helmut E. Meyer, and Corinna Henkel
11 Ethyl Esterification for MALDI-MS Analysis of Protein Glycosylation . . . . . . . 151
Karli R. Reiding, Emanuela Lonardi, Agnes L. Hipgrave Ederveen,
and Manfred Wuhrer
12 Characterization of Protein N-Glycosylation by Analysis
of ZIC-HILIC-Enriched Intact Proteolytic Glycopeptides. . . . . . . . . . . . . . . . 163
Gottfried Pohlentz, Kristina Marx, and Michael Mormann

vii
viii Contents

13 Simple and Effective Affinity Purification Procedures for Mass

Spectrometry-Based Identification of Protein-Protein Interactions
in Cell Signaling Pathways. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
Julian H.M. Kwan and Andrew Emili
14 A Systems Approach to Understand Antigen Presentation
and the Immune Response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
Nadine L. Dudek, Nathan P. Croft, Ralf B. Schittenhelm,
Sri H. Ramarathinam, and Anthony W. Purcell
15 Profiling of Small Molecules by Chemical Proteomics . . . . . . . . . . . . . . . . . . . 211
Kilian V.M. Huber and Giulio Superti-Furga
16 Generating Sample-Specific Databases for Mass Spectrometry-Based
Proteomic Analysis by Using RNA Sequencing . . . . . . . . . . . . . . . . . . . . . . . . 219
Toni Luge and Sascha Sauer
17 A Proteomic Workflow Using High-Throughput De Novo Sequencing
Towards Complementation of Genome Information for Improved
Comparative Crop Science . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
Reinhard Turetschek, David Lyon, Getinet Desalegn, Hans-Peter Kaul,
and Stefanie Wienkoop
18 From Phosphoproteome to Modeling of Plant Signaling Pathways . . . . . . . . . 245
Maksim Zakhartsev, Heidi Pertl-Obermeyer, and Waltraud X. Schulze
19 Interpretation of Quantitative Shotgun Proteomic Data . . . . . . . . . . . . . . . . . 261
Elise Aasebø, Frode S. Berven, Frode Selheim, Harald Barsnes,
and Marc Vaudel
20 A Simple Workflow for Large Scale Shotgun Glycoproteomics. . . . . . . . . . . . . 275
Astrid Guldbrandsen, Harald Barsnes, Ann Cathrine Kroksveen,
Frode S. Berven, and Marc Vaudel
21 Systemic Analysis of Regulated Functional Networks . . . . . . . . . . . . . . . . . . . . 287
Luis Francisco Hernández Sánchez, Elise Aasebø, Frode Selheim,
Frode S. Berven, Helge Ræde, Harald Barsnes, and Marc Vaudel

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311
Contributors

ELISE AASEBØ • Proteomics Unit, Department of Biomedicine, University of Bergen,

Bergen, Norway
JUN ADACHI • Laboratory of Proteome Research, National Institute of Biomedical
Innovation, Health and Nutrition, Osaka, Japan
ROBERT AHRENDS • Leibniz-Institut für Analytische Wissenschaften—ISAS—e.V.,
Dortmund, Germany
HARALD BARSNES • Proteomics Unit, Department of Biomedicine, University of Bergen,
Bergen, Norway; KG Jebsen Center for Diabetes Research, Department of Clinical
Science, University of Bergen, Bergen, Norway
BIRTE BEINE • Leibniz-Institut für Analytische Wissenschaften—ISAS—e.V., Dortmund,
Germany
FRODE S. BERVEN • Proteomics Unit, Department of Biomedicine, University of Bergen,
Bergen, Norway; KG Jebsen Centre for Multiple Sclerosis Research, Department of
Clinical Medicine, University of Bergen, Bergen, Norway; Norwegian Multiple Sclerosis
Competence Centre, Department of Neurology, Haukeland University Hospital,
Bergen, Norway
ANJA-K. BOSSERHOFF • Institute of Pathology, University of Regensburg, Regensburg,
Germany; Institute of Biochemistry, Emil-Fischer-Zentrum, Friedrich-Alexander-University
Erlangen-Nürnberg, Erlangen, Germany
NATHAN P. CROFT • Department of Biochemistry and Molecular Biology, School of Biomedical
Sciences, Monash University, Clayton, VIC, Australia
GETINET DESALEGN • Department of Crop Sciences, University of Natural Resources
and Life Sciences, Vienna, Austria
HANNA C. DIEHL • Clinical Proteomics, Medizinisches Proteome-Center, Ruhr-University,
Bochum, Germany
NADINE L. DUDEK • Department of Biochemistry and Molecular Biology, School of Biomedical
Sciences, Monash University, Clayton, VIC, Australia
AMANDA EDWARDS • Center for Cancer Research, Massachusetts General Hospital, Charlestown,
MA, USA; Department of Medicine, Harvard Medical School, Boston, MA, USA
ANDREW EMILI • Department of Molecular Genetics, Donnelly Centre for Cellular
and Biomolecular Research, University of Toronto, Toronto, ON, Canada
YUEHAN FENG • Department of Biology, Institute of Biochemistry, ETH Zurich, Zurich,
Switzerland
ASTRID GULDBRANDSEN • Proteomics Unit, Department of Biomedicine, University
of Bergen, Bergen, Norway; KG Jebsen Centre for Multiple Sclerosis Research,
Department of Clinical Medicine, University of Bergen, Bergen, Norway
WILHELM HAAS • Center for Cancer Research, Massachusetts General Hospital, Charlestown,
MA, USA; Department of Medicine, Harvard Medical School, Boston, MA, USA
CORINNA HENKEL • Leibniz-Institut für Analytische Wissenschaften—ISAS—e.V.,
Dortmund, Germany
ANDREAS HENTSCHEL • Leibniz-Institut für Analytische Wissenschaften—ISAS—e.V.,
Dortmund, Germany

ix
x Contributors

LUIS FRANCISCO HERNÁNDEZ SÁNCHEZ • Graduate Program in Optimization, Universidad

Autónoma Metropolitana Azcapotzalco, Mexico City, Mexico
SONJA HESS • Proteome Exploration Laboratory, California Institute of Technology,
Pasadena, CA, USA
AGNES L. HIPGRAVE EDERVEEN • Center for Proteomics and Metabolomics, Leiden
University Medical Center, Leiden, The Netherlands
KILIAN V.M. HUBER • Structural Genomics Consortium, University of Oxford, Oxford, UK;
Target Discovery Institute, University of Oxford, Oxford, UK
HANS-PETER KAUL • Department of Crop Sciences, University of Natural Resources
and Life Sciences, Vienna, Austria
LAXMIKANTH KOLLIPARA • Leibniz-Institut für Analytische Wissenschaften—ISAS—e.V.,
Dortmund, Germany
ANN CATHRINE KROKSVEEN • Proteomics Unit, Department of Biomedicine, University
of Bergen, Bergen, Norway; KG Jebsen Centre for Multiple Sclerosis Research,
Department of Clinical Medicine, University of Bergen, Bergen, Norway
JULIAN H.M. KWAN • Department of Molecular Genetics, Donnelly Centre for Cellular
and Biomolecular Research, University of Toronto, Toronto, ON, Canada
EMANUELA LONARDI • Center for Proteomics and Metabolomics, Leiden University Medical
Center, Leiden, The Netherlands
PHILIP LÖSSL • Department of Pharmaceutical Chemistry and Bioanalytics, Institute
of Pharmacy, Martin-Luther University Halle-Wittenberg, Halle/Saale, Germany;
Biomolecular Mass Spectrometry and Proteomics, Netherlands Proteomics Center, Bijvoet
Center for Biomolecular Research and Utrecht Institute for Pharmaceutical Sciences,
Utrecht University, Utrecht, The Netherlands
TONI LUGE • Otto Warburg Laboratory, Max Planck Institute for Molecular Genetics,
Berlin, Germany
DAVID LYON • Department of Ecogenomics and Systems Biology, University of Vienna,
Vienna, Austria
KRISTINA MARX • Bruker Daltonik GmbH, Bremen, Germany
HELMUT E. MEYER • Leibniz-Institut für Analytische Wissenschaften—ISAS—e.V.,
Dortmund, Germany
J. BERNADETTE MOORE • Department of Nutritional Sciences, Faculty of Health
and Medical Sciences, University of Surrey, Guildford Surrey, UK
ANNIE MORADIAN • Proteome Exploration Laboratory, California Institute of Technology,
Pasadena, CA, USA
MICHAEL MORMANN • Institute for Hygiene, University of Münster, Münster, Germany
RYOHEI NARUMI • Laboratory for Synthetic Biology, RIKEN Quantitative Biology Center,
Kobe, Japan
PETER J. OEFNER • Institute of Functional Genomics, University of Regensburg, Regensburg,
Germany
HEIDI PERTL-OBERMEYER • Plant Systems Biology, Plant Physiology, University of Hohenheim,
Stuttgart, Germany
PAOLA PICOTTI • Department of Biology, Institute of Biochemistry, ETH Zurich, Zurich,
Switzerland
GOTTFRIED POHLENTZ • Institute for Hygiene, University of Münster, Münster, Germany
 Contributors xi

TANYA R. PORRAS-YAKUSHI • Proteome Exploration Laboratory, California Institute

of Technology, Pasadena, CA, USA
ANTHONY W. PURCELL • Department of Biochemistry and Molecular Biology,
School of Biomedical Sciences, Monash University, Clayton, VIC, Australia;
The Department of Biochemistry and Molecular Biology, The Bio21 Molecular Science and
Biotechnology Institute, University of Melbourne, Melbourne, VIC, Australia
HELGE RÆDER • KG Jebsen Center for Diabetes Research, Department of Clinical Science,
University of Bergen, Bergen, Norway; Department of Pediatrics, Haukeland University
Hospital, Bergen, Norway
SRI H. RAMARATHINAM • Department of Biochemistry and Molecular Biology,
School of Biomedical Sciences, Monash University, Clayton, VIC, Australia
KARLI R. REIDING • Center for Proteomics and Metabolomics, Leiden University Medical
Center, Leiden, The Netherlands
JÖRG REINDERS • Institute of Functional Genomics, University of Regensburg, Regensburg,
Germany
YVONNE REINDERS • Department of Biochemistry I, University of Regensburg, Regensburg,
Germany
SASCHA SAUER • Otto Warburg Laboratory, Max Planck Institute for Molecular Genetics,
Berlin, Germany
RALF B. SCHITTENHELM • Department of Biochemistry and Molecular Biology,
School of Biomedical Sciences, Monash University, Clayton, VIC, Australia
WALTRAUD X. SCHULZE • Plant Systems Biology, Plant Physiology, University of Hohenheim,
Stuttgart, Germany
FRODE SELHEIM • Proteomics Unit, Department of Biomedicine, University of Bergen,
Bergen, Norway
ALBERT SICKMANN • Leibniz-Institut für Analytische Wissenschaften—ISAS—e.V.,
Dortmund, Germany; Department of Chemistry, College of Physical Sciences,
University of Aberdeen, Aberdeen, Scotland, UK
ANDREA SINZ • Department of Pharmaceutical Chemistry and Bioanalytics, Institute
of Pharmacy, Martin-Luther University Halle-Wittenberg, Halle/Saale, Germany
FIORELLA A. SOLARI • Leibniz-Institut für Analytische Wissenschaften—ISAS—e.V.,
Dortmund, Germany
CHRISTOS SPANOS • Department of Nutritional Sciences, Faculty of Health and Medical
Sciences, University of Surrey, Guildford Surrey, UK
GIULIO SUPERTI-FURGA • CeMM Research Center for Molecular Medicine of the Austrian
Academy of Sciences, Vienna, Austria
MICHAEL J. SWEREDOSKI • Proteome Exploration Laboratory, California Institute
of Technology, Pasadena, CA, USA
TAKESHI TOMONAGA • Laboratory of Proteome Research, National Institute of Biomedical
Innovation, Health and Nutrition, Osaka, Japan
REINHARD TURETSCHEK • Department of Ecogenomics and Systems Biology, University
of Vienna, Vienna, Austria
MARC VAUDEL • Proteomics Unit, Department of Biomedicine, University of Bergen,
Bergen, Norway
DANIEL VÖLLER • Institute of Pathology, University of Regensburg, Regensburg, Germany
STEFANIE WIENKOOP • Department of Ecogenomics and Systems Biology, University
of Vienne, Vienna, Austria
xii Contributors

MANFRED WUHRER • Center for Proteomics and Metabolomics, Leiden University Medical
Center, Leiden, The Netherlands; Division of BioAnalytical Chemistry, VU University
Amsterdam, Amsterdam, The Netherlands
RENÉ P. ZAHEDI • Leibniz-Institut für Analytische Wissenschaften—ISAS—e.V.,
Dortmund, Germany
MAKSIM ZAKHARTSEV • Plant Systems Biology, Plant Physiology, University of Hohenheim,
Stuttgart, Germany
 Chapter 1

Multiplexed Quantitative Proteomics

for High-Throughput Comprehensive Proteome
Comparisons of Human Cell Lines
Amanda Edwards and Wilhelm Haas

Abstract
The proteome is the functional entity of the cell, and perturbations of a cellular system almost always cause
changes in the proteome. These changes are a molecular fingerprint, allowing characterization and a
greater understanding of the effect of the perturbation on the cell as a whole. Monitoring these changes
has therefore given great insight into cellular responses to stress and disease states, and analytical platforms
to comprehensively analyze the proteome are thus extremely important tools in biological research. Mass
spectrometry has evolved as the most relevant technology to characterize proteomes in a comprehensive
way. However, due to a lack of throughput capacity of mass spectrometry-based proteomics, researchers
frequently use measurement of mRNA levels to approximate proteome changes. Growing evidence of
substantial differences between mRNA and protein levels as well as recent improvements in mass
spectrometry-based proteomics are heralding an increased use of mass spectrometry for comprehensive
proteome mapping. Here we describe the use of multiplexed quantitative proteomics using isobaric label-
ing with tandem mass tags (TMT) for the simultaneous quantitative analysis of five cancer cell proteomes
in biological duplicates in one mass spectrometry experiment.

Key words Quantitative proteomics, Multiplexing, Isobaric labels, TMT

1 Introduction

Proteins are the primary functional units of the cell, and as such,
information about their abundance, interaction partners, and mod-
ifications is critical for understanding both healthy and abnormal
cellular function. Traditionally, such work has been accomplished
on a protein-by-protein basis through genetic or biochemical tech-
niques. More recently, large-scale approaches attempting to moni-
tor an entire proteome—all proteins expressed in a cell or tissue—in
one step have become accessible [1, 2]. Such a holistic approach
allows identification of proteome imbalances and changes in func-
tional networks, enabling us to study and probe the state of a cell
in an unbiased and rapid fashion.

Jörg Reinders (ed.), Proteomics in Systems Biology: Methods and Protocols, Methods in Molecular Biology, vol. 1394,
DOI 10.1007/978-1-4939-3341-9_1, © Springer Science+Business Media New York 2016

1
2 Amanda Edwards and Wilhelm Haas

Mass spectrometry has emerged as the leading platform to

rapidly characterize whole proteomes, primarily driven by
improvements in both MS sensitivity and throughput in the last
15 years. However, the technology has traditionally lagged behind
in throughput capacity when compared to genomics platforms,
such as DNA-microarrays or next-generation sequencing technol-
ogy, to study cellular expression profiles. Therefore, mRNA expres-
sion profiles are still the main source for estimations of protein-level
changes for most researchers [3, 4]. Yet evidence is accumulating
that significant differences exist between mRNA- and protein-level
changes in different cell or tissue states [1, 5–9]. There is thus an
enormous need for improved mass spectrometry-based proteomics
technology to enable direct protein-level measurements, with a
throughput comparable to that provided by genomics
technologies.
Mass spectrometry-based proteomics has been used as a quan-
titative tool since the late 1990s with the introduction of accurate
relative quantification using stable isotopes. One of the earliest
approaches was the employment of isotope-coded affinity tags
(ICAT). ICAT enables a chemical incorporation of differential sta-
ble isotopes into two different samples and, in parallel, a reduction
of proteome complexity by enrichment of cysteine-containing pep-
tides [10]. An approach to incorporate stable isotopes metaboli-
cally through heavy isotope-labeled amino acids—stable isotope
labeling in cell culture (SILAC)—was first described in 2002 [11]
and is still widely used. The commercial availability of high-
performance mass spectrometers [12, 13] optimized for use in
combination with liquid chromatography further contributed to
the propagation of quantitative proteomics, and new strategies for
chemical incorporation of stable isotopes into peptides, such as
reductive dimethylation [14, 15], arose. Each of the above meth-
ods relies on the use of full MS data from intact peptide ions to
perform quantitative analysis. Each peptide in the two samples of
interest is detected in its light and heavy form, leading to an
increase in the signal complexity in the full MS spectra. This
increase in signal complexity necessarily decreases the overall sensi-
tivity of the approach and complicates the quantitative analysis of
individual peptides. Consequently, although more is theoreti-
cally possible [16], the number of samples routinely compared
simultaneously using these methods is limited to three [17].
A very elegant strategy to remove this roadblock in multiplex-
ing MS proteomics was first described in 2003 through the use of
isobaric tags to incorporate stable isotopes into proteomics samples
[18]. These tags consist of three regions: a mass reporter ion, a
mass balancer region, and a reactive terminal amino group. To
quantify different protein levels in different biological samples, pep-
tide mixtures are labeled with different forms of the tag by allowing
the tag to react with amino groups at the N-terminus or lysine
 Multiplexed Quantitative Proteomics for High-Throughput Comprehensive Proteome… 3

residues of a peptide. Importantly, each tag has the same mass, as

the chemical structures only differ in the distribution of heavy stable
isotopes between reporter and balancer regions. Thus differentially
labeled peptides migrate together through the chromatographic
separation and are indistinguishable in MS1 scans. However, during
MS2 fragmentation, the mass reporter ions (with a unique mass for
each tag) separate from the parent tag, and their relative intensities
represent the relative abundance of the original peptides in the
measured samples. There are two commercial sources for isobaric
tags: isobaric tags for relative and absolute quantitation (ITRAQ)
reagents (Sciex) that allow the analysis of up to eight samples [19,
20], and tandem mass tag (TMT) reagents (Pierce) that enable a
simultaneous quantification of up to ten samples [21, 22]. An early
caveat of the isobaric labeling strategy was a limitation in the achiev-
able accuracy and reproducibility of quantitative results due to co-
isolation and fragmentation of contaminant ions with the ions
targeted for identification and quantification. Solutions to over-
come this limitation were presented in the form of applying ion-ion
chemistry for removing contaminant ions [23] or by separating
identification and quantification of a peptide ion, performing the
identification based on MS2 data but shifting the quantification to
an MS3 experiment as an additional gas-phase enrichment and frag-
mentation step [24]. The MS3 method was further optimized to
increase sensitivity and throughput, and this MultiNotch MS3
method [25] is now implemented as a synchronous precursor
selection (SPS)-supported MS3 method on the Orbitrap Fusion
mass spectrometer (Thermo Fisher Scientific). We believe that
multiplexed quantitative proteomics is a tool that will prove to be
indispensable in studying complex biological systems and disease
states requiring the analysis of many samples.
This chapter describes the workflow for using 10-plex tandem
mass tag (TMT) reagents for isobaric labeling-based multiplexed
quantitative proteomics to comprehensively map proteomes of
human cell lines. We routinely apply this protocol to quantify
approximately 8000 proteins simultaneously in ten samples, occu-
pying 36 h of mass spectrometry time, or less than 4 h to quantita-
tively characterize the proteome of a human cell line.

2 Materials

2.1 Cell Culture 1. Cell lines: This protocol is applicable for the proteomic analysis
of any adherent human cell line. Detached cell lines can also be
used, with modifications to the cell culture protocols.
2. Cell media: Use culture media appropriate for the chosen cell lines.
3. 1× sterile phosphate-buffered saline (PBS).
4. 0.25 % Trypsin.
4 Amanda Edwards and Wilhelm Haas

2.2 Cell Lysis 1. Lysis buffer: 75 mM NaCl, 50 mM HEPES (pH 8.5), 10 mM

sodium pyrophosphate, 10 mM NaF, 10 mM
β-glycerophosphate, 10 mM sodium orthovanadate, 10 mM
phenylmethanesulfonylfluoride (PMSF), Roche Complete
Protease Inhibitor EDTA-free tablet, 3 % SDS.

2.3 Sample 1. HPLC-grade methanol.

Preparation for Mass 2. HPLC-grade chloroform.
Spectrometry
3. HPLC-grade water.
4. HPLC-grade acetonitrile (ACN).
5. HPLC-grade acetone.
6. 1 M Dithiothreitol (DTT) in 50 mM HEPES (pH 8.5).
7. 1 M Iodoacetamide (IAA) in 50 mM HEPES (pH 8.5).
8. Digestion buffer: 1 M urea, 50 mM HEPES (pH 8.5).
9. Lysyl endopeptidase (LysC) (Wako Chemicals, 10 AU, resus-
pended in 2 mL HPLC-grade water, stored at −80 °C).
10. Trypsin (Promega, sequencing grade, 0.4 μg/μL, stored at
−80 °C).
11. 1 % and 10 % trifluoroacetic acid (TFA), 99.5 % purity.
12. 0.5 % Acetic acid.
13. 40 % ACN, 0.5 % acetic acid.
14. 80 % ACN, 0.5 % acetic acid.
15. 5 % ACN, 5 % formic acid.
16. 50 % ACN, 5 % formic acid.
17. Bicinchoninic acid (BCA) protein assay kit (Pierce).
18. Bovine serum albumin.
19. Tandem Mass Tag™ 10-plex reagent set (Pierce).
20. 200 mM HEPES (pH 8.5), 30 % anhydrous ACN.
21. 200 mM HEPES (pH 8.5), 5 % hydroxylamine.

2.4 High-pH 1. HpRP buffer A: 5 % ACN, 10 mM ammonium bicarbonate.

Reversed-Phase 2. HpRP buffer B: 90 % ACN, 10 mM ammonium bicarbonate.
High-Pressure Liquid
Chromatography

2.5 Mass 1. MS buffer A: 3 % ACN, 0.125 % formic acid.

Spectrometry 2. MS buffer B: 0.125 % formic acid in ACN.

2.6 Equipment 1. Minicentrifuge.

2. 1 cc syringes.
3. 21-gauge needles.
 Multiplexed Quantitative Proteomics for High-Throughput Comprehensive Proteome… 5

4. Vacuum manifold.
5. SepPak 1 cc (50 mg) C18 Cartridges (Waters).
6. Savant SC100 SpeedVac Concentrator.
7. High-pressure liquid chromatography system (ex: Agilent
1260 Infinity Quaternary LC System).
8. Agilent ZORBAX Extend-C18 column (4.6 mm × 250 mm, 5
μm particle size).
9. Deep-well 96-well plates.
10. Orbitrap Fusion (Thermo Fisher Scientific).
11. Easy-nLC 1000 (Thermo Fisher Scientific).
12. Resins: Magic C4 resin (5 μm, 100 Å, Michrom Bioresources),
Maccel C18AQ resin (3 μm, 200 Å, Nest Group), and GP-C18
(1.8 μm, 120 Å, Sepax Technologies).

3 Methods

3.1 Cell Culture 1. Grow each of the five cell lines to 90 % confluence in duplicate
in 10 cm2 dishes (a total of ten samples).
2. Prior to collecting the cells, wash gently twice with 2.5 mL
pre-warmed sterile 1× PBS (see Note 1).
3. Add 2 mL pre-warmed trypsin to each 10 cm2 dish, covering the
cell layer completely. Incubate for 5 min at 37 °C. Add 3 mL
pre-warmed media to each 10 cm2 dish, and collect each cell
mixture in 15 mL Falcon tubes.
4. Pellet cells by centrifuging at 500 × g for 5 min. Discard the
supernatant. Wash the cell pellet once with sterile 1× PBS (see
Note 2).

3.2 Cell Lysis Note: All subsequent steps should be performed at room temperature,
as the SDS in the lysis buffer will precipitate at cold temperatures.
1. Resuspend each cell pellet in 0.5 mL lysis buffer, pipetting up
and down to disrupt the cell pellet (see Note 3).
2. Lyse the cells by passing the resuspended cells ten times
through a 21-gauge needle (see Note 4). Transfer the suspen-
sion to 1.5 mL Eppendorf tubes.
3. Clear away cellular debris by centrifuging at 16,000 × g for
5 min. Collect the supernatant (see Note 5).

3.3 Reduction, 1. Add 2.5 μL of 1 M DTT to each sample (final concentration of

Alkylation, DTT = 5 mM), and vortex thoroughly. Centrifuge briefly at
and Precipitation 3000 × g to bring all contents to the bottom of the tube.
of Proteins Incubate at 56 °C for 30 min (see Note 6).
6 Amanda Edwards and Wilhelm Haas

2. Cool the tubes on ice for 3 min.

3. Add 7.5 μL of 1 M IAA to each sample (final concentration of
IAA = 15 mM), and vortex thoroughly. Centrifuge briefly at
3000 × g. Incubate in the dark for 20 min (see Note 7).
4. Add 2.5 μL of 1 M DTT to each sample to quench the reac-
tion, and vortex thoroughly. Centrifuge briefly at 3000 × g.
Incubate in the dark for 15 min.
5. Transfer each sample to a 15 mL Falcon tube.
6. Begin protein precipitation by adding 2.0 mL methanol to
each tube (4 × the initial lysate volume) (see Note 8). Vortex,
and centrifuge at 1300 × g for 3 min.
7. Add 0.5 mL chloroform to each tube (1 × the initial lysate vol-
ume), and vortex. Make sure to disrupt any pellet fully (see
Note 9). Centrifuge at 1300 × g for 3 min.
8. Add 1.5 mL water to each tube (3 × the initial lysate volume),
and vortex. Again, make sure that pellet is fully disrupted.
Centrifuge at 1300 × g for 3 min.
9. At this stage, precipitated proteins should form a white disk
between the aqueous and organic phases. Carefully remove all
aqueous and organic supernatant.
10. Wash the protein pellet with 2.0 mL methanol. Centrifuge at
1300 × g for 3 min.
11. Remove the supernatant, and place the protein pellet on ice.
Add 1.5 mL ice-cold acetone to each pellet. Disrupt the pellet,
and centrifuge at 1300 × g at 4 °C for 3 min. Remove the
supernatant, and repeat once (see Note 10).
12. Dry the precipitated protein in open tubes at 56 °C for 15 min
or at 37 °C for 60 min, until pellets are completely dry. Cool
the pellet on ice for 5 min (see Note 11).

3.4 Digestion 1. Resuspend the protein pellet in 0.5 mL digestion buffer (see
of Proteins Note 12).
2. Add 2.5 μL LysC stock (5 μg) to each pellet. Vortex. Centrifuge
briefly at 3000 × g. Incubate overnight at room temperature,
agitating gently on a tabletop vortexer.
3. Add 12.5 μL trypsin stock (5 μg) to each tube. Vortex.
Centrifuge briefly at 3000 × g. Incubate at 37 °C for 6 h.
4. Acidify the reaction with 25 μL 10 % TFA (final concentration
TFA = 0.5 %). Vortex. Centrifuge at 16,000 × g for 5 min, and
collect the supernatant (see Note 13).

3.5 Cleanup 1. Place 50 mg C18 SepPak columns on the vacuum manifold,

of Sample Using and wash with 5 × 1 mL ACN.
SepPak Columns (See 2. Wash with 5 × 1 mL 80 % ACN, 0.5 % acetic acid.
Notes 14 and 15)
3. Wash with 5 × 1 mL 0.1 % TFA.
 Multiplexed Quantitative Proteomics for High-Throughput Comprehensive Proteome… 7

4. Apply sample to column, and pull over column at a slow speed

(see Note 16).
5. Wash with 5 × 1 mL 0.1 % TFA.
6. Wash with 1 mL 0.5 % acetic acid, and allow the column to go
completely dry.
7. Remove SepPak columns from the vacuum manifold, and place
in clean 2 mL Eppendorf tubes. Add 0.75 mL of 40 % ACN,
0.5 % acetic acid to each column. Using a 1 mL syringe, push
solution through the column slowly. Add 0.75 mL of 80 %
ACN, 0.5 % acetic acid to each column, and push the solution
through the column, allowing the column to go completely
dry (see Note 17).
8. Dry the eluted peptides in a SpeedVac.

3.6 Quantify 1. Resuspend each sample in 0.5 mL 50 % ACN, 5 % formic acid.

and Aliquot Digested Vortex. Centrifuge briefly at 3000 × g. Sonicate for 5 min.
Peptides 2. (See Note 18) Prepare standard bovine serum albumin (BSA)
stocks ranging from 25 to 2000 μg/mL, with 50 % ACN, 5 %
formic acid as the buffer.
3. Add 10 μL of BSA standard or sample into 96-well plate wells
(see Note 19). Add 200 μL of working reagent to each well,
and mix thoroughly.
4. Cover the plate with plastic wrap, and incubate at 37 °C for
30 min. Remove the plate, and allow cooling to room tem-
perature for 5 min.
5. Measure the absorbance at 562 nM, and use the standard
curve to determine the protein concentration of each sample.
6. Prepare 50 μg aliquots of each sample, and dry the peptides in
a SpeedVac.

3.7 TMT Labeling 1. Resuspend the TMT reagent according to the manufacturer’s
of Peptides instructions in anhydrous acetonitrile (see Note 20).
2. Resuspend the peptides in 50 μL 200 mM HEPES (pH 8.5),
30 % anhydrous ACN (see Note 21). Vortex. Centrifuge briefly
at 3000 × g. Sonicate for 5 min.
3. Add 5 μL of TMT reagent to each peptide solution, with 1
TMT label used for each of the ten samples (126, 127n, 127c,
128n, 128c, 129n, 129c, 130n, 130c, and 131).
4. Incubate the reaction mixtures at room temperature for 1 h.
5. Quench the reaction by adding 6 μL of 200 mM HEPES
(pH 8.5), 5 % hydroxylamine. Incubate at room temperature
for 15 min.
6. Acidify the mixture by adding 50 μL of 1 % TFA. Combine all
ten samples into one sample, as they are now all distinctly
labeled.
8 Amanda Edwards and Wilhelm Haas

7. De-salt the mixture over a 50 mg C18 SepPak column (see

Subheading 3.5 above for details).
8. Dry the peptides in a SpeedVac.

3.8 Fractionation 1. Resuspend the peptides in 0.5 mL 5 % ACN, 5 % formic acid.

of Peptides Vortex. Centrifuge briefly at 3000 × g. Sonicate for 5 min.
2. Fractionate the sample by high pH reverse-phase high-pressure
liquid chromatography (HpRP) using a two-buffer gradient
system at a flow rate of 500 μL/min. Load the sample in 0 %
HpRP buffer B for 2 min, and then separate the peptides using
a linear gradient from 20 to 35 % HpRP buffer B over 60 min.
Wash the column with 100 % HpRP buffer B for 5 min, and
re-equilibrate the column with 100 % HpRP buffer A for
10 min. Monitor peptide elution by UV absorption at a wave-
length of 220 nm and collect a total of 96 fractions in a deep-
well 96-well plate from 11.5 to 69.5 min (see Note 22).
3. Combine the 96 fractions into 12 fractions, based on the sche-
matic in Fig. 1.
4. Dry the fractions in a SpeedVac.

3.9 Mass The details of the LC-MS2/MS3 methods will depend on the
Spectrometry Analysis instrumentation available. Here, we describe a method using an
Easy-nLC 1000 (Thermo Fisher Scientific) with chilled autosam-
pler and an Orbitrap Fusion mass spectrometer (Thermo Fisher
Scientific)
1. Sample preparation: Resuspend each fraction in 8 μL 5 % ACN,
5 % formic acid and sonicate to ensure full suspension of all
peptides. Inject 3 μL of each sample for chromatographic sepa-
ration and mass spectrometry analysis.
2. Nanospray liquid chromatography method: Separate peptides
over a 100 μm inner diameter microcapillary column, packed
in-house with 0.5 cm of Magic C4 resin, 0.5 cm of Maccell
C18 resin, and 29 cm of GP-C18 resin. Use a 6–25 % gradient
of MS buffer B over 165 min at 300 nL/min to elute the pep-
tides. End the gradient with a 10-min wash with 100 % MS
buffer B to clear all remaining peptides off the column, and
re-equilibrate the column with 9 μL of 100 % MS buffer A to
prepare the column for subsequent runs.
3. Mass spectrometry method: Begin acquisition with a full MS1
spectrum acquired in the Orbitrap, and use synchronous pre-
cursor selection to isolate the ten highest intensity peptides for
MS2 analysis. Following CID fragmentation of these peptides,
perform MS2 scans in the linear ion trap. Once again, use syn-
chronous precursor selection to isolate the ten highest inten-
sity peptides for MultiNotch MS3 analysis [25]. Following
HCD fragmentation of the peptides, perform MS3 scans in the
Orbitrap for maximum sensitivity (see Note 23).
 Multiplexed Quantitative Proteomics for High-Throughput Comprehensive Proteome… 9

Fig. 1 An overview of the workflow of a multiplexed quantitative proteomics measurement, from cell culture to
mass spectrometer

3.10 Data Analysis As above, the details of the data analysis will depend on the specific
search algorithms and software used. While we use Sequest [26] to
match peptide spectra to sequences, a variety of other options are
available (e.g., Mascot, X!Tandem). However, some parameters
should be universally applied.
1. Specific search parameters include digestion enzyme, static
peptide modifications, variable peptide modifications, and pre-
cursor ion tolerance. In this case, select trypsin as the enzyme,
10 Amanda Edwards and Wilhelm Haas

requiring all matching peptides to have termini consistent with

tryptic cleavage, allowing at most two missed cleavages. Static
modifications include the TMT label on the N-terminus and
lysine residues (229.162932 Da) as well as carbamidomethyl-
ation (57.021464 Da) on cysteine residues. Oxidation of
methionine (15.994915 Da) should be set as a variable modi-
fication. Set the precursor m/z ion tolerance to 50 ppm.
2. Several online servers provide complete or near-complete pro-
tein databases for a variety of species, including UniProt,
Ensembl, and RefSeq, against which MS2 spectra can be
searched. We use UniProt databases, and we apply a target-
decoy database search strategy to accurately estimate the false
discovery rate of peptide and protein identifications [27, 28].
This requires compiling a concatenated database with a target
component including the organism-specific protein sequence
database as well as that of known contaminants such as porcine
trypsin or other proteases used in the sample preparation. The
second and so-called decoy component includes the same
sequences but in reversed—or flipped—order, where every
protein sequence starts with the original C-terminus of the
original sequence. A practical protocol for the use of this data-
base for estimating the FDR of a proteomics dataset is described
elsewhere [29]. Filter the final results of peptides as well as
protein identifications to an FDR of at most 1 %. Several algo-
rithms are used for filtering proteomics data. We use a linear
discriminant analysis to assess the FDR of MS2 spectra assign-
ments to peptide sequences (peptide-spectral matches, PSMs).
For a protein identification FDR filter, we also rely on the tar-
get-decoy database search strategy by using a posterior error
histogram with protein FDR estimations that are based on
combining probabilities of correct assignments for each PSM
for all peptides matching a protein sequence [30].
3. In order to accurately quantify a peptide, signal-to-noise values
and isolation specificity must exceed a background threshold.
These values will depend on the instrument being used. To
quantify a protein level, sum all reporter ion intensities from all
peptides assigned to that protein.
4. Normalization of the data allows correction for slight prepara-
tion errors or MS anomalies. We recommend a two-step nor-
malization procedure. Begin by normalizing all protein
intensities to the ratio of the average intensity of that protein to
all median protein intensities. This will bring all protein values
closer to one another, allowing for more unbiased downstream
statistical testing. Secondly, normalize all protein intensities to
the ratio of the median protein intensities for a given TMT
channel to the median of all protein intensities. This will account
for any slight mixing errors from each sample.
 Multiplexed Quantitative Proteomics for High-Throughput Comprehensive Proteome… 11

4 Notes

1. Different cell types will adhere with different strengths, so one

must be careful not to dislodge the cells prematurely, depending
on cell type.
2. The cell pellets can be frozen here at −80 °C for future use.
3. A 5:1 ratio of lysis buffer:cell pellet or greater is helpful here in
order to make the lysate less viscous.
4. It works best to do this slowly so as not to create excess
bubbles.
5. Depending on the viscosity of the lysate, there may or may not
be a visible cell pellet that clearly separates from the supernatant.
If there is not, move forward with the entire mixture.
6. The goal of the DTT is to reduce all disulfide bonds.
7. The goal of the IAA is to alkylate free thiols.
8. This precipitation technique requires enough protein to visualize
the protein pellet. If the protein output is too low, a TCA
precipitation is preferred.
9. If the pellet is difficult to disrupt, rake the tube against an
Eppendorf tube rack until it is dislodged.
10. While loss in the previous precipitation steps will be unbiased,
loss at this stage will be biased towards hydrophobic peptides
and should be carefully avoided.
11. The protein pellets can be frozen here at −80 °C for future use.
12. Depending on the size, it may be difficult to fully resuspend
the pellet. Use a small pestle to grind the pellet and disrupt as
best as possible.
13. Any pellet is undigested protein. The pellet can be stored at
−20 °C and further digested in the future in the case of low
peptide yield.
14. SepPak columns are used for larger peptide amounts. For low
amounts (<10 μg), use StageTips (packed with C18 solid-phase
extraction disks [Empore]) instead to minimize peptide loss.
15. Do not let column go completely dry until indicated.
16. It may be easier to open other ports on the vacuum manifold
to allow a slower draw on the column.
17. Make sure to add the 40 % ACN solution first, followed by the
80 % ACN solution, to prevent too many peptides from eluting
at once and clogging the column.
18. Prepare the protein quantification assay of your choice. We use
the BCA assay from Pierce.
19. More concentrated samples may be diluted 1:3–1:10 to fall
into the standard curve range.
12 Amanda Edwards and Wilhelm Haas

20. It is very important here that the ACN used is anhydrous, as

hydrated ACN will reduce the labeling efficiency.
21. The manufacturer recommends performing TMT labeling
using a triethylammonium bicarbonate (TEAB) buffer.
However, it has been shown previously [24] that using this
buffer produces unidentified and unwanted site reaction prod-
ucts (in particular, singly charged ions with m/z of 303.26,
317.26, and 331.29) of high intensity in the LC-MS2/MS3
chromatograms. The formation of these products is avoided by
using the described buffer conditions.
22. The given retention time range is based on the described sys-
tem. This range may differ slightly between different HPLC
systems and the fraction collection retention time frame should
be adjusted accordingly.
23. The Orbitrap Fusion allows acquisition of high-resolution,
MultiNotch MS3 data [25]. As demonstrated previously [24,
25], this approach reduces the observed interference effect in
quantitation at the MS2 level. However, not all instrumenta-
tion allows for this approach. Other approaches to decrease
interference in the quantitation include TMTC quantitation
[31] and ion-ion chemistry for removing contaminant ions
[23]. If these approaches are untenable, one must use the
resulting quantitative data with appropriate caution.

References
1. Kim M, Pinto S, Getnet D et al (2014) A draft 8. Wu Y, Williams E, Dubuis S et al (2014)
map of the human proteome. Nature Multilayered genetic and omics dissection of
509:575–581 mitochondrial activity in a mouse reference
2. Wilhelm M, Schlegl J, Hahne H et al (2014) population. Cell 158:1415–1430
Mass-spectrometry-based draft of the human 9. Zhang B, Wang J, Wang X et al (2014)
proteome. Nature 509:582–587 Proteogenomic characterization of human
3. Wang Z, Gerstein M, Snyder M (2009) RNA- colon and rectal cancer. Nature 513:382–387
Seq: a revolutionary tool for transcriptomics. 10. Gygi S, Rist B, Gerber S et al (1999)
Nat Rev Genet 10:57–63 Quantitative analysis of complex protein mix-
4. Li G, Burkhardt D, Gross C et al (2014) tures using isotope-coded affinity tags. Nat
Quantifying absolute protein synthesis rates Biotechnol 17:994–999
reveals principles underlying allocation of cel- 11. Ong S, Blagoev B, Kratchmarova I et al (2002)
lular resources. Cell 157:624–635 Stable isotope labeling by amino acids in cell
5. Torres E, Dephoure N, Panneerselvam A et al culture, SILAC, as a simple and accurate
(2010) Identification of aneuploidy-tolerating approach to expression proteomics. Mol Cell
mutations. Cell 143:71–83 Proteomics 1:376–386
6. Stingele S, Stoehr G, Peplowska K et al (2012) 12. Syka J, Marto J, Bai D et al (2004) Novel linear
Global analysis of genome, transcriptome and quadrupole ion trap/FT mass spectrometer:
proteome reveals the response to aneuploidy in performance characterization and use in the
human cells. Mol Syst Biol 8:608 comparative analysis of histone H3 post-
7. Dephoure N, Hwang S, O’Sullivan C et al translational modification. J Proteome Res
(2014) Quantitative proteomic analysis reveals 3:621–626
posttranslational responses to aneuploidy in 13. Olsen J, de Godoy L, Li G et al (2005) Parts
yeast. Elife 3, e03023 per million mass accuracy on an Orbitrap mass
 Multiplexed Quantitative Proteomics for High-Throughput Comprehensive Proteome… 13

spectrometer via lock mass injection into a investigate host-pathogen interaction. Cell
C-trap. Mol Cell Proteomics 12:2010–2021 157:1460–1472
14. Hsu J, Huang S, Chow N et al (2003) Stable- 23. Wenger C, Lee M, Hebert A et al (2011)
isotope dimethyl labeling for quantitative pro- Gas-phase purification enables accurate, multi-
teomics. Anal Chem 75:6843–6852 plexed proteome quantification with isobaric
15. Wilson-Grady J, Haas W, Gygi S (2013) tagging. Nat Methods 8:933–935
Quantitative comparison of the fasted and re- 24. Ting L, Rad R, Gygi S et al (2011) MS3 elimi-
fed mouse liver phosphoproteomes using lower nates ratio distortion in isobaric multiplexed
pH reductive dimethylation. Methods 61: quantitative proteomics. Nat Methods 8:
277–286 937–940
16. Wu Y, Wang F, Liu Z et al (2014) Five-plex 25. McAlister G, Nusinow D, Jedrychowski M
isotope dimethyl labeling for quantitative et al (2014) MultiNotch MS3 enables accurate,
proteomics. Chem Commun (Camb) 50: sensitive, and multiplexed detection of differ-
1708–1710 ential expression across cancer cell line pro-
17. Blagoev B, Ong S, Kratchmarova I et al (2004) teomes. Anal Chem 86:7150–7158
Temporal analysis of phosphotyrosine- 26. Eng J, McCormack A, Yates J (1994) An
dependent signaling networks by quantitative approach to correlate tandem mass spectral
proteomics. Nat Biotechnol 22:1139–1145 data of peptides with amino acid sequences in a
18. Thompson A, Schäfer J, Kuhn K et al (2003) protein database. J Am Soc Mass Spectrom
Tandem mass tags: a novel quantification strat- 5:976–989
egy for comparative analysis of complex pro- 27. Peng J, Elias J, Thoreen C et al (2003)
tein mixtures by MS/MS. Anal Chem 75: Evaluation of multidimensional chromatogra-
1895–1904 phy coupled with tandem mass spectrometry
19. Ross P, Huang Y, Marchese J et al (2004) (LC/LC-MS/MS) for large-scale protein anal-
Multiplexed protein quantitation in ysis: the yeast proteome. J Proteome Res
Saccharomyces cerevisiae using amine-reactive 1:43–50
isobaric tagging reagents. Mol Cell Proteomics 28. Elias J, Gygi S (2007) Target-decoy search
3:1154–1169 strategy for increased confidence in large-scale
20. Choe L, D’Ascenzo M, Relkin N et al (2007) protein identifications by mass spectrometry.
8‐Plex quantitation of changes in cerebrospi- Nat Methods 4:207–214
nal fluid protein expression in subjects under- 29. Elias J, Gygi S (2009) Target-decoy search
going intravenous immunoglobulin treatment strategy for mass spectrometry-based pro-
for Alzheimer’s disease. Proteomics teomics. Methods Mol Biol 604:55–71
7:3651–3660 30. Huttlin E, Jedrychowski M, Elias J et al (2010) A
21. McAlister G, Huttlin E, Haas W et al (2012) tissue-specific atlas of mouse protein phosphoryla-
Increasing the multiplexing capacity of TMTs tion and expression. Cell 143:1174–1189
using reporter ion isotopologues with isobaric 31. Wühr M, Haas W, McAlister G et al (2012)
masses. Anal Chem 84:7469–7478 Accurate multiplexed proteomics at the MS2
22. Weekes M, Tomasec P, Huttlin E et al (2014) level using the complement reporter ion cluster.
Quantitative temporal viromics: an approach to Anal Chem 84:9214–9221
 Chapter 2

Sample Preparation Approaches for iTRAQ Labeling

and Quantitative Proteomic Analyses in Systems Biology
Christos Spanos and J. Bernadette Moore

Abstract
Among a variety of global quantification strategies utilized in mass spectrometry (MS)-based proteomics,
isobaric tags for relative and absolute quantitation (iTRAQ) are an attractive option for examining the rela-
tive amounts of proteins in different samples. The inherent complexity of mammalian proteomes and the
diversity of protein physicochemical properties mean that complete proteome coverage is still unlikely
from a single analytical method. Numerous options exist for reducing protein sample complexity and
resolving digested peptides prior to MS analysis. Indeed, the reliability and efficiency of protein identifica-
tion and quantitation from an iTRAQ workflow strongly depend on sample preparation upstream of
MS. Here we describe our methods for: (1) total protein extraction from immortalized cells; (2) subcel-
lular fractionation of murine tissue; (3) protein sample desalting, digestion, and iTRAQ labeling; (4)
peptide separation by strong cation-exchange high-performance liquid chromatography; and (5) peptide
separation by isoelectric focusing.

Key words Proteomics, Mass spectrometry, iTRAQ, Subcellular fractionation, High-performance

liquid chromatography, Isoelectric focusing

1 Introduction

Quantitative analysis of protein expression, function, and subcellular

localization is fundamental to network biology. Mass spectrometry
(MS)-based quantitative proteomic approaches have evolved rap-
idly in the last 15 years and are generating datasets essential for
systems biology and the modeling of biological networks [1].
Discovery applications in MS-based proteomics have largely
employed untargeted strategies where proteins in one or more
samples (diseased or treated) are quantified relative to the amount
of proteins in a separate sample (normal or control). Quantification
of the peptides/proteins can either be done by comparative analysis
of spectral features in a “label-free” workflow or alternatively be
accomplished through isotopic labeling or the incorporation of a
differential mass tag. Isobaric tags for relative and absolute

Jörg Reinders (ed.), Proteomics in Systems Biology: Methods and Protocols, Methods in Molecular Biology, vol. 1394,
DOI 10.1007/978-1-4939-3341-9_2, © Springer Science+Business Media New York 2016

15
16 Christos Spanos and J. Bernadette Moore

quantitation (iTRAQ) are widely used amine-specific, stable-

isotope reagents which can be used to label the N terminus and ε
side chain of lysines of peptides generated by tryptic digestion of
extracted proteins. The reagents were designed for “multiplex-
ing,” and four or eight separate iTRAQ labels are available permit-
ting the simultaneous analysis of multiple samples. The labels
consist of a low-mass reporter group, a balance group, and an
amine-reactive group; they have isobaric masses in MS mode (145
and 305 Da for 4-plex or 8-plex reagents), but upon fragmentation
release low-mass reporter ions (m/z values of 114.1, 115.1, 116.1,
117.1 for 4-plex, plus 113.1, 118.1, 119.1, 121.1 for 8-plex)
allowing quantification at the MS/MS level. Unlike metabolic
incorporation of stable isotopes, a key advantage to iTRAQ labels
is that samples from any source, including patient material, can be
chemically labeled. This fact, in combination with the ability to
simultaneously analyze multiple samples, has likely contributed to
the widespread use of these reagents [2, 3]. It should be noted that
quantitation by iTRAQ is not perfect; contamination during pre-
cursor ion selection, specific to MS/MS quantitation, results in
compression of the iTRAQ ratio and underestimation of relative
protein abundance estimates; and variance is higher for low-inten-
sity signals [4, 5]. Data processing and instrument-specific
approaches aimed at addressing issues related to the precision and
accuracy of iTRAQ quantitation continue to evolve and have been
reviewed elsewhere [2, 3].
Technological advances in high-resolution MS instrumenta-
tion means that almost complete coverage of unicellular organisms
such as yeast is now possible [6] and comprehensive analysis of
mammalian proteomes is envisaged as feasible in the near future
[7]. However, the required technology is not widely available and
currently most researchers will find that proteome coverage is
dependent on reducing sample complexity and their choice of mul-
tiple sample preparation steps including protein separation, diges-
tion, and peptide fractionation steps. Proteomic workflows can be
complex and the role of error propagation through multiple han-
dling steps should not be underestimated [8]. Critical to the suc-
cess of an iTRAQ experiment is the use of equal sample amounts,
reproducible protein digestion, and efficient peptide labeling.
Protein samples may be pre-fractionated either by subcellular frac-
tionation or based on size by polyacrylamide gel electrophoresis.
Proteins are most typically digested using trypsin, although other
proteases can be used [9], and digested peptides are separated
by either high-performance liquid chromatography (HPLC) or
isoelectric focusing.
We have used iTRAQ reagents to examine differential protein
expression in fatty acid-treated hepatocarcinoma cells and liver
tissue mice fed a high-fat diet. Here we describe our methods for:
(1) total protein extraction from immortalized cells; (2) subcellular
 Sample Preparation for iTRAQ-based Proteomics 17

fractionation of murine liver tissue; (3) protein sample desalting,

digestion, and iTRAQ labeling; (4) peptide separation by strong
cation-exchange (SCX) HPLC; and (5) peptide separation by
isoelectric focusing.

2 Materials

All reagents should be of analytical grade and solvents either HPLC

or LC-MS grade. All solutions should be prepared with distilled,
deionized water (ddH2O), typically 18.2 MΩ·cm at 25 °C, except
for LC-MS/MS buffers which require LC-MS-grade H2O.

2.1 Protein 1. Phosphate-buffered saline (PBS; 1×): 137 mM NaCl, 2.7 mM

Extraction from Cells KCl, 10 mM Na2HPO4.
2. Radioimmunoprecipitation assay buffer (RIPA): 150 mM
NaCl, 1.0 % IGEPAL® CA-630, 0.5 % sodium deoxycholate,
0.1 % SDS, and 50 mM Tris, pH 8.0.
3. Protease inhibitor cocktail-EDTA free (PI): Mixture of several
protease inhibitors inhibiting serine, cysteine, but not
metalloproteases.
4. Low-speed swinging bucket centrifuge.
5. Refrigerated table top microcentrifuge.
6. QIAshredder columns (QIAGEN, UK).
7. BCA assay.

2.2 Protein 1. HEPES, EDTA, mannitol (HEM) buffer: 20 mM HEPES,

Extraction from Liver 1 mM EDTA, 300 mM mannitol.
Tissue 2. Protease inhibitor cocktail EDTA free (PI): Mixture of several
protease inhibitors inhibiting serine, cysteine, but not
metalloproteases.
3. 10 ml glass Dounce homogenizer with pestle attached to
electric drill.
4. Refrigerated swinging bucket centrifuge.
5. Ultracentrifuge.
6. Appropriate polyallomer tubes for ultracentrifugation.
7. Amicon Ultra-15 Centrifugation Filter Device (NMWL of 3
kDa; Millipore, USA).
8. BCA assay (Pierce Scientific, UK).

2.3 Protein 1. 2 ml ZEBA columns (Pierce Scientific, UK; see Note 6).
Desalting, Digestion, 2. RIPA buffer and PI as before.
and iTRAQ Labeling
3. Vacuum centrifuge (Eppendorf, UK).
4. Low-bind microcentrifuge tubes (see Note 7).
18 Christos Spanos and J. Bernadette Moore

5. Bovine serum albumin (BSA) as control to monitor digestion

and future LC separation efficiency.
6. Triethylammonium bicarbonate (TEAB) buffer, 1 M pH 8.5
(Sigma Aldrich, UK).
7. iTRAQ Reagents Multiplex Kit (ABSciex, UK).
8. Shaking dry heat block (Eppendorf, UK).
9. Mass spectrometry-grade trypsin gold (Promega, UK).
10. Sonication water bath.

2.4 Peptide 1. HPLC instrument (Hewlett Packard 1100 series) with autos-
Separation by Strong ampler (Agilent Technologies 1200).
Cation-Exchange HPLC 2. Polysulfoethyl A chromatography column (100 × 94 2.1
mm—300 Å; Hichrom Ltd, UK).
3. Guard cartridge (Hichrom Ltd, UK; see Note 9).
4. Buffer A: 10 mM KH2PO4 (pH 2.75), 25 % acetonitrile
(see Note 10).
5. Buffer B: 10 mM KH2PO4, 1 M KCl (pH 2.75), 25 % acetonitrile
(see Note 10).
6. Vacuum centrifuge (Eppendorf, UK).

2.5 Peptide 1. 3100 OFFGEL Fractionator (Agilent Technologies, UK).

Separation 2. OFFGEL High Resolution Kit, pH 3–10, 12 samples, 24
by Isoelectric fractions (Agilent Technologies, UK).
Focusing
3. Vacuum centrifuge (Eppendorf, UK).

2.6 Sample Cleanup 1. 0.1 % (v/v) trifluoroacetic acid (TFA; Sigma Aldrich, UK) in
Prior to Mass HPLC-grade H2O.
Spectrometry 2. ZipTips C18 with 0.6 μl bed of chromatography media
(Millipore, USA).
3. “Wet solution”: 25 % acetonitrile in HPLC-grade H2O.
4. “Equilibration and wash solution”: 0.1 % TFA in HPLC-
grade H2O.
5. “Elution solution”: 50 % acetonitrile/0.1 % TFA in HPLC-
grade H2O.

3 Methods

3.1 Total Protein 1. Add protease inhibitor cocktail to RIPA buffer to a final con-
Extraction centration of 1.5×; keep on ice. Bring a microcentrifuge for the
from Hepatocellular 10,000 × g spin to 4 °C.
Carcinoma Cell Line 2. Following your preferred cell treatment, detach cells from flask
surface using trypsin and centrifuge cell suspension for 5 min
at 200 × g RT.
 Sample Preparation for iTRAQ-based Proteomics 19

3. Wash cell pellets with 1× PBS and centrifuge again for 5 min at
1200 × g.
4. Resuspend cell pellet in 350 μl RIPA buffer containing PI and
leave on ice for 10 min.
5. Vortex lysed cells, then place on QIAshredder column, and
centrifuge for 2 min at 10,000 × g.
6. Measure protein concentration of resulting eluate using
BCA assay.

3.2 Membrane 1. Bring both an ultracentrifuge and swinging bucket centrifuge

and Cytosol Protein to 4 °C.
Extraction from Mouse 2. Add protease inhibitor cocktail to of ice-cold HEM buffer
Liver Tissue (see Note 1) to a final concentration of 1.5×; keep on ice.
3. From frozen liver lobes (see Note 2) excise approximately
50 mg of liver with a razor blade on dry ice. Place into Dounce
homogenizer with 6 ml of ice-cold HEM buffer and homog-
enize using electric drill (see Note 3).
4. Transfer sample to sterile 15 ml conical tube and keep on ice
while repeating for all biological replicates. Rinse pestle and
container in between each sample with ddH2O.
5. Centrifuge samples in swinging bucket centrifuge at 2000 × g
for 10 min at 4 °C.
6. Transfer supernatant (containing membrane and cytosol pro-
teins) to polyallomer ultracentrifuge tubes on ice; add ~4 ml of
HEM buffer with PI and balance tubes within 1/10th of a
gram of each other. Discard pellet (see Note 4).
7. Centrifuge at 100,000 × g for 30 min at 4 °C.
8. Transfer the supernatant, representing the cytosolic fraction,
into Amicon Ultra-15 Centrifugation Filter Device and centrifuge
for 1 h at 4000 × g at 4 °C to collect concentrated cytosolic
protein (>3 kDa).
9. Re-homogenize (see Note 5) the membrane-enriched pellet in
0.5–1.0 ml of fresh HEM buffer with PI.
10. Determine protein concentration using BCA assay before
aliquoting samples and storing at −80 °C.

3.3 Protein 1. Remove storage solution from 2 ml ZEBA columns (see Note 6)
Desalting/Buffer by centrifuging at 2000 × g for 2 min. For tissue samples use
Exchange, Digestion, RIPA for buffer exchange; for cells use ddH2O for desalting.
and iTRAQ Labeling Add 1 ml RIPA buffer with PI or ddH2O to column resin and
centrifuge for 2 min at 2000 × g; repeat three times. Then add
samples to column resin and centrifuge for 2 min at 2000 × g.
Process BSA control alongside protein samples up until iTRAQ
labeling (step 9).
20 Christos Spanos and J. Bernadette Moore

2. Collect the resulting desalted protein eluate and quantify protein

concentration by BCA assay.
3. Aliquot 100 μg of protein into low-bind microcentrifuge tubes
(see Note 7) from each treatment and dry in vacuum centri-
fuge at 45 °C; dried sample can be stored at −80 °C prior to
digestion.
4. If stored at −80 °C bring dried protein samples to RT, and
then add 40 μl of 1 M, pH 8.5, TEAB buffer (see Note 8) and
2 μl of 2 % sodium dodecyl sulfate (SDS; denaturing agent
from iTRAQ kit). Vortex thoroughly to solubilize, and then
pulse spin to bring the sample to the bottom of the tube.
5. Add 4 μl of tris(2-carboxyethyl)phosphine (TCEP; reducing
agent from iTRAQ kit); vortex thoroughly, then pulse spin,
and incubate for 1 h at 60 °C.
6. Add 1 μl of methyl methanethiosulfonate (MMTS; cysteine
blocking reagent from iTRAQ kit); vortex thoroughly, then
pulse spin, and incubate for 10 min at RT.
7. Add 10 μl of 1 μg/μl trypsin solution (reconstituted in 50 mM
acetic acid) vortex thoroughly, then pulse spin, and incubate,
shaking, at 37 °C for 12–16 h.
8. Post-digestion, dry down samples by vacuum centrifugation at
45 °C and prepare 100 μl/sample 7:3 (v/v) ethanol (100 %):
TEAB (0.5 M) solution.
9. Resuspend peptide digests in 80 μl of ethanol:TEAB by vor-
texing and then sonicate for 5 min at RT in sonicating water
bath. Repeat to ensure complete solubilization.
10. Bring required iTRAQ reagent vials (114, 115, 116, 117) to
RT; pulse spin, add each peptide digest to the appropriate
iTRAQ reagent vial, and vortex thoroughly.
11. Rinse each sample tube with the additional 20 μl of ethanol:TEAB
solution and transfer to the correct iTRAQ reagent vials. Vortex,
pulse spin, and then incubate for 1 h at RT.
12. Combine iTRAQ-labeled samples into single tube; rinse empty
tubes with one aliquot of 100 μl of 50 % acetonitrile and add
to combined tube.
13. Dry down pooled sample via vacuum centrifugation at 45 °C.

3.4 Peptide 1. The dried, digested, and iTRAQ-labeled sample (one tube)
Separation by Strong was resuspended in 100 μl of buffer A.
Cation-Exchange HPLC 2. Set up guard cartridge (see Note 9) before HPLC column.
3. Condition the HPLC column by introducing buffer A (see
Note 10) for 30 min at a flow rate of 1.5 ml/min.
4. Use digested BSA sample to check both the efficiency of the
tryptic digestion and the separation efficiency of the column
using the stepped gradient (Table 1).
 Sample Preparation for iTRAQ-based Proteomics 21

Table 1
The stepped gradient that was followed during the SCX approach. The
pressure was set to 400 bar

Time (min) % Buffer B Flow rate (ml/min)

0.00 0.00 1.00
0.01 0.00 0.80
10.00 0.00 0.80
10.01 0.10 0.80
10.02 6.00 0.80
12.02 6.00 0.80
12.03 9.00 0.80
14.03 9.00 0.80
14.04 11.00 0.80
16.04 11.00 0.80
16.05 13.00 0.80
18.05 13.00 0.80
18.06 15.00 0.80
20.06 15.00 0.80
20.07 17.00 0.80
22.07 17.00 0.80
22.08 19.00 0.80
24.08 19.00 0.80
24.09 23.00 0.80
26.09 23.00 0.80
26.10 27.00 0.80
28.10 27.00 0.80
28.11 33.00 0.80
30.12 33.00 0.80
30.13 40.00 0.80
32.13 40.00 0.80
33.14 50.00 0.80
38.14 50.00 0.80
39.15 100.00 0.80
42.16 100.00 0.80

(continued)
22 Christos Spanos and J. Bernadette Moore

Table 1
(continued)

Time (min) % Buffer B Flow rate (ml/min)

42.17 6.00 0.80
43.17 6.00 0.80
43.18 0.00 0.50
50.01 0.00 0.10

5. Inject the entire volume of sample with a 100 μl injection

volume and collect fractions in 2-min time slices; use LC results
to assess how many fractions contain sample (see Note 11).
6. Dry down collected fractions by vacuum centrifugation at 45 °C.

3.5 Peptide 1. Make up “peptide OFFGEL stock solution (1.25×)” and

Separation “peptide-immobilized pH gradient (IPG) strip rehydration
by Isoelectric solution” with the appropriate ampholytes depending on
Focusing which IPG strips you are using according to kit instructions
(see Note 12). Assemble the IPG strips, frames, and electrodes
according to the manufacturer’s protocol.
2. Resuspend by vortexing the dried peptide samples in 3.6 ml 1×
OFFGEL stock solution (e.g., 2.88 ml 1.25× OFFGEL stock
solution + 0.72 ml dH2O). Load 150 μl of this sample in each
of the 24 wells.
3. Samples should be focused using a maximum current of 50 μA
and the focusing is stopped after the total voltage reaches 50
kVh (60 h). During the focusing, pads on the electrodes should
be exchanged to prevent evaporation (see Note 13).
4. After focusing, 50–150 μl of peptide fractions should be recov-
ered for each well and transferred to individual microfuge
tubes. To recover as much as possible of the focused peptides,
150 μl of HPLC-grade methanol should be added to each
well, incubated for 15 min without voltage, and then added to
the appropriate tube.
5. Dry down collected fractions by vacuum centrifugation at 45 °C.

3.6 Sample Cleanup 1. Resuspend peptide samples (fractionated by either SCX or

Prior to Mass IEF) in 120 μl 0.1 % TFA.
Spectrometry 2. Equilibrate the ZipTip by first aspirating 10 μl of the “wet
solution” and discarding to waste, repeat once; then aspirate
10 μl of the “equilibration and wash solution” and discarding
to waste, repeating three times.
3. Bind peptides to ZipTip by cycling (aspirate-dispense-aspirate-
dispense) ten times with your sample.
 Sample Preparation for iTRAQ-based Proteomics 23

4. Wash sample by aspirating 10 μl of the “equilibration and wash

solution” and discarding to waste; repeat three times.
5. Elute sample by aspirating 10 μl of the “elution solution” and
collecting in microfuge tubes; repeat six times until 70 μl is
collected in all.
6. Dry down collected cleaned up fractions by vacuum centrifu-
gation at 45 °C. Resuspend in 10 μl formic acid just prior to
nLC-ESI-MS/MS analysis.

4 Notes

1. An alternative buffer for tissue lysis may be applied as deemed

appropriate.
2. We recommend at the time of study termination excising the
same liver lobe from all animals, immediate snap freezing in
liquid nitrogen, and then keeping on dry ice until transfer to
−80 °C.
3. In our case we used the lowest speed setting on a very power-
ful bench-top electric drill and used ten, slow, up-and-down
strokes for each sample. This may vary depending on your
setup; just keep consistent for all samples.
4. Pellet contains nuclear proteins and cellular debris; you may
wish to use differential detergent fractionation here to isolate
nuclear proteins as well.
5. We use pestle manually here in addition to pipetting.
6. Methanol-chloroform and other precipitation methods may be
used, but we have experienced better recovery and reproduc-
ibility using Zeba columns for buffer exchange and sample
concentration.
7. Drying down by vacuum centrifuge may result in sample
adhering on the walls of the tube. To minimize any protein loss
and to make resuspension easier, low-bind tubes should be
used in all drying, by vacuum centrifuge, steps.
8. The TEAB buffer that is included in the iTRAQ kit is 0.5 M. We
have had better recovery of the sample from the dry phase
using 1 M TEAB for resuspension of the dried sample before
the iTRAQ labeling process begins. For the next step and
whenever required, the TEAB buffer provided by the iTRAQ
kit should be used.
9. Although not necessary, it is advised to use a guard cartridge
attached in front of the HPLC column. In case of blockage
due to sample impurities the column remains unaffected and
the guard cartridge is cheaper to replace.
24 Christos Spanos and J. Bernadette Moore

10. For strong cation exchange, the pH of the buffers should be

tightly controlled below or equal to 3. Whenever buffers need
to be replaced, the pH levels should be exactly the same as
previously.
11. In our case, separating peptide digests of total protein from
HuH7 cells, we had sample in first 17 fractions.
12. The OFFGEL stock solution contains glycerol; this can make
later sample resuspension difficult and interfere with the subse-
quent performance of the mass spectrometer by blocking the
nLC lines. It has been suggested that the glycerol can be omitted
or reduced but we have not tested this.
13. The total run for the OFFGEL fractionation is around 60 h.
In order to prevent water evaporation from the pads that will lead
to insufficient peptide separation, upper electrode pads should be
changed every 15 h with fresh pads wetted in ddH2O.

Acknowledgements

This work was supported by the Biotechnology and Biological

Sciences Research Council, with a studentship grants (BB/
D526853/1) to C.S. and a project grant (BB/I008195/1) to
J.B.M.

References
1. Moore JB, Weeks ME (2011) Proteomics and 6. Nagaraj N, Kulak NA, Cox J et al (2012)
systems biology: current and future applications System-wide perturbation analysis with nearly
in the nutritional sciences. Adv Nutr 2:355–364 complete coverage of the yeast proteome by
2. Evans C, Noirel J, Ow SY et al (2012) An insight single-shot ultra HPLC runs on a bench top
into iTRAQ: where do we stand now? Anal Orbitrap. Mol Cell Proteomics 11:M111.
Bioanal Chem 404:1011–1027 013722
3. Christoforou AL, Lilley KS (2012) Isobaric tag- 7. Mann M, Kulak NA, Nagaraj N et al (2013) The
ging approaches in quantitative proteomics: the coming age of complete, accurate, and ubiqui-
ups and downs. Anal Bioanal Chem tous proteomes. Mol Cell 49:583–590
404:1029–1037 8. Kruger T, Lehmann T, Rhode H (2013) Effect
4. Ow SY, Salim M, Noirel J et al (2009) iTRAQ of quality characteristics of single sample prepa-
underestimation in simple and complex mixtures: ration steps in the precision and coverage of pro-
“the good, the bad and the ugly”. J Proteome teomic studies—a review. Anal Chim Acta
Res 8:5347–5355 776:1–10
5. Karp NA, Huber W, Sadowski PG et al (2010) 9. Vandermarliere E, Mueller M, Martens L (2013)
Addressing accuracy and precision issues in Getting intimate with trypsin, the leading prote-
iTRAQ quantitation. Mol Cell Proteomics ase in proteomics. Mass Spectrom Rev
9:1885–1897 32:453–465
 Chapter 3

Two Birds with One Stone: Parallel Quantification

of Proteome and Phosphoproteome Using iTRAQ
Fiorella A. Solari, Laxmikanth Kollipara, Albert Sickmann,
and René P. Zahedi

Abstract
Altered and abnormal levels of proteins and their phosphorylation states are associated with many disorders.
Detection and quantification of such perturbations may provide a better understanding of pathological
conditions and help finding candidates for treatment or biomarkers. Over the years, isobaric mass tags for
relative quantification of proteins and protein phosphorylation by mass spectrometry have become increas-
ingly popular. One of the most commonly used isobaric chemical tags is iTRAQ (isobaric tag for relative
and absolute quantitation). In a typical iTRAQ-8plex experiment, a multiplexed sample amounts for up to
800 μg of peptides. Using state-of-the-art LC-MS approaches, only a fraction (~5 %) of such a sample is
required to generate comprehensive quantitative data on the global proteome level, so that the bulk of the
sample can be simultaneously used for quantitative phosphoproteomics. Here, we provide a simple and
straightforward protocol to perform quantitative analyses of both proteome and phosphoproteome from
the same sample using iTRAQ.

Key words iTRAQ, Phosphopeptide enrichment, Quantitative proteomics, Quantitative

phosphoproteomics

1 Introduction

Quantitative analysis of proteins and their phosphorylation sites

can provide valuable insights into altered molecular mechanisms
and signaling pathways and thus can be used to study abnormal
behavior, such as in cancer [1]. For almost a decade, isobaric mass
tags have been used extensively for relative quantification in pro-
teomics [2–8] and are particularly relevant for biological samples
that are not accessible to metabolic labeling, such as human tissues
and body fluids. Chemical labeling with iTRAQ reagents [9]
enables relative quantification of up to eight different multiplexed
conditions/samples in a single LC-MS run, and can provide
sample amounts (up to 0.8 mg) that are considerably higher than
what is required for conventional proteomics experiments using

Jörg Reinders (ed.), Proteomics in Systems Biology: Methods and Protocols, Methods in Molecular Biology, vol. 1394,
DOI 10.1007/978-1-4939-3341-9_3, © Springer Science+Business Media New York 2016

25
26 Fiorella A. Solari et al.

state-of-the-art equipment and workflows (even with fractionation

no more than 50 μg). Therefore, the remaining >90 % of the mul-
tiplexed sample can be effectively used to analyze low-abundance
posttranslational modifications (PTM), such as phosphorylation.
Here, we describe a protocol based on iTRAQ-8plex labeling that
allows for both global and phosphoproteome analyses. For the global
proteome analysis, 5 % of the multiplexed sample is fractionated using
high-pH reversed-phase chromatography [10], whereas the rest of
the sample is used for phosphopeptide enrichment with titanium
dioxide (TiO2) beads [11] followed by hydrophilic interaction liquid
chromatography (HILIC) [12] for fractionation of phosphopeptides
prior to LC-MS. Thus, with as little as 40 μg for proteome and 760
μg for phosphoproteome analysis, it is possible to quantify thousands
of proteins and phosphopeptides from the same sample.

2 Materials

All solutions and buffers should be prepared using ultrapure

deionized water. Alternatively, LC-MS-grade water (Biosolve)
can be used.

2.1 Cell Lysis 1. Biological sample obtained from primary cells or tissue, e.g.,
of Biological Samples 100 μg of fibroblasts.
2. Lysis buffer: 50 mM Tris–HCl (pH 7.8, adjust with HCl), 150
mM sodium chloride (NaCl) containing 1 % (w/v) sodium
dodecyl sulfate (SDS). Add one tablet of phosphatase inhibitor
cocktail phosSTOP (Roche Diagnostics) to 10 mL of buffer.
3. Benzonase (Novagen) and 1 M magnesium chloride (MgCl2)
solution.
4. Refrigerated benchtop centrifuge (Eppendorf) and a vortex mixer.
5. Determination of protein concentration: Bicinchoninic acid
(BCA) protein assay.

2.2 Carbamido- 1. Reducing agent: 2 M dithiothreitol (DTT) solution. Stock can

methylation, Protein be stored at −40 °C.
Precipitation, 2. Alkylating agent: 1 M iodoacetamide (IAA) solution. Freshly
and Enzymatic Protein prepared. See Note 1.
Digestion 3. Organic solvents: Ethanol and acetone. Store both solvents
at −40 °C prior to usage.
4. Centrifuge: Refrigerated benchtop centrifuge (Eppendorf).
5. Enzyme: Sequencing Grade Modified Trypsin (Promega).
Dissolve the lyophilized trypsin in 50 mM ammonium bicar-
bonate (NH4HCO3), pH 7.8 to get a final concentration of
1 mg/mL.
 Two Birds with One Stone: Parallel Quantification of Proteome and Phosphoproteome… 27

6. Digestion buffer: 0.2 M Guanidine hydrochloride (GuHCl),

5 % acetonitrile (ACN), 2 mM calcium chloride (CaCl 2),
trypsin solution [1 mg/mL], and 50 mM NH4HCO3
(pH 7.8).
7. Stop digestion: 10 % trifluoroacetic acid (TFA).

2.3 Digestion Control 1. HPLC: UltiMate 3000 rapid separation liquid chromatography
by Monolithic (RSLC) (Thermo Fisher Scientific) or similar HPLC system.
Reversed-Phase 2. HPLC column: PepSwift monolithic trap column 200
Chromatography μm × 5 mm and PepSwift monolithic capillary column 200
μm × 5 cm (both Thermo Fisher Scientific).
3. HPLC buffer A: 0.1 % TFA.
4. HPLC buffer B: 0.08 % TFA, 84 % ACN.

2.4 Solid-Phase 1. Solid-phase extraction cartridge (SPEC) sorbent material: C18

Extraction Cartridges AR 4 mg columns (Agilent).
for Sample Cleanup 2. Vacuum manifold system and vacuum centrifuge (SpeedVac).
3. Wetting/activating buffer: 100 % ACN.
4. Equilibration and wash buffer: 0.1 % TFA.
5. Elution buffer: 60 % ACN in 0.1 % TFA.

2.5 iTRAQ Labeling 1. iTRAQ reagents: 8 plex kit (113–119, 121) from AB SCIEX.
2. Dissolution buffer: 0.5 M triethylammonium bicarbonate
(TEAB), pH 8.5.
3. Reagent dilution solvent: 100 % isopropanol, LC-MS grade.
4. Thermomixer.

2.6 Phosphopeptide 1. Metal oxide affinity chromatography (MOAC) material:

Enrichment Titanium dioxide (TiO2) beads, 5 μm diameter (Titansphere,
with Titanium Dioxide GL Sciences).
(TiO2) Beads 2. Loading buffer 1: 80 % ACN, 5 % TFA, and 1 M glycolic acid
[13]. See Note 2.
3. Washing buffer 1: 80 % ACN, 1 % TFA.
4. Washing buffer 2: 10 % ACN, 0.1 % TFA.
5. Elution buffer: 1 % ammonium hydroxide (NH4OH), pH 11.3.
6. Loading buffer 2: 70 % ACN, 2 % TFA.
7. Washing buffer 3: 50 % ACN, 0.1 % TFA.
8. Phosphopeptide purification: C8 Empore Disc (3 M).
9. Acidifying buffer: 100 % formic acid (FA) and 10 % TFA.
10. C18 SPE: Oligo R3 beads (Applied Biosystems) and C18
Empore Disc (3 M).
11. Air-filled syringe.
28 Fiorella A. Solari et al.

2.7 Sample 1. HPLC: UltiMate 3000 liquid chromatography (Thermo

Fractionation Using Fisher Scientific) or similar HPLC system.
Chromatography 2. HPLC column: RP C18, 0.5 mm × 15 cm, 5 μm (Biobasic,
2.7.1 High-pH Reversed- Thermo Scientific).
Phase Chromatography 3. HPLC buffer A: 10 mM ammonium formate (NH4HCO2),
pH 8.0.
4. HPLC buffer B: 84 % (v/v) in 10 mM NH4HCO2, pH 8.0.

2.7.2 Hydrophilic 1. HPLC: UltiMate 3000 rapid separation liquid chromatography

Interaction (RSLC) (Thermo Fisher Scientific) or similar HPLC system.
Chromatography 2. HPLC column: Polar-phase TSK gel, 150 μm × 15 cm, 5 μm,
self-packed.
3. HPLC solvent A: 0.1 % TFA, 98 % ACN.
4. HPLC solvent B: 0.1 % TFA.

2.8 LC-MS/MS 1. HPLC: UltiMate 3000 nano RSLC system (Thermo Fisher
Scientific) with nano-UV cell for quality control or a similar
HPLC system.
2. HPLC column: Acclaim PepMap100 C18 trap column 100
μm × 2 cm, and Acclaim PepMap100 C18 main column 75
μm × 50 cm (both Thermo Fisher Scientific).
3. HPLC loading buffer: 0.1 % TFA.
4. HPLC solvent A: 0.1 % FA.
5. HPLC solvent B: 0.1 % FA, 84 % ACN.
6. Mass spectrometer: Q-Exactive mass spectrometer (Thermo
Fisher Scientific) or other MS systems that can provide high
mass accuracy and high resolution for both MS and MS/MS
scans, as well as beam-type CID fragmentation (or HCD in
Thermo instruments).

2.9 Data Analysis 1. Data analysis software: Proteome Discoverer (version 1.4,
Thermo Scientific) using the following nodes.
(a) Search algorithms: Mascot [14] (version 2.4.1, Matrix
Science), Sequest [15], and MS Amanda [16].
(b) Quantification: Reporter ion quantifier.
(c) Phosphorylation site assignment: PhosphoRS [17].
(d) False discovery rate (FDR) estimator: Peptide validator.

3 Methods

Make sure that you treat all samples completely the same way during
all mentioned steps until labeling, in order to keep the technical
variation as low as possible. Any differences during sample treatment
may affect the quantitative results.
 Two Birds with One Stone: Parallel Quantification of Proteome and Phosphoproteome… 29

3.1 Cell Lysis 1. The volume of the lysis buffer depends on the number of cells.
and Determination Typically, 100 μL of buffer is required to solubilize for, e.g.,
of Protein 106 HeLa cells. Promote homogenization by mechanical/
Concentration shearing forces, induced by pipetting and ultrasonication.
2. To hydrolyze the DNA, add 3 μL of benzonase per 200 μL of
cell lysate plus 2 mM MgCl2 and incubate at 37 °C for 30 min.
3. Clarify the lysates by centrifugation. Use a precooled (4 °C)
centrifuge and spin down the sample tubes at 18,000 rcf for
30 min.
4. Collect the supernatant in a new LoBind Eppendorf tube and
determine the protein concentration using BCA assay accord-
ing to the manufacturer’s instructions. See Note 3.

3.2 Carbamido- 1. Reduce cysteines by incubating the cell lysates with 10 mM of

methylation, Sample DTT at 56 °C for 30 min. After incubation, cool down the
Cleanup, tubes to room temperature (RT).
and Proteolytic 2. Without further delay, add the alkylating agent (IAA) to a final
Digestion concentration of 30 mM and incubate at RT for 30 min in the
dark.
3. Organic solvent protein precipitation: See Note 4.
(a) Take an aliquot of the cell lysate corresponding to 100 μg
of protein and dilute tenfold with cold (−40 °C) ethanol,
i.e., one part of sample plus nine parts of ethanol, and
vortex briefly.
(b) Store the sample at −40 °C for 1 h. In the meantime, cool
the centrifuge to 4 °C.
(c) Place the samples in the precooled (4 °C) centrifuge and
spin down the tubes at 12,000 rcf for 30 min. Discard the
supernatant.
(d) Place 100 μL of cold acetone onto the protein pellet, briefly
vortex, and centrifuge as mentioned above for 10 min.
(e) Remove the supernatant and allow the protein pellet to dry.
See Note 5.
4. Digestion:
(a) Resolubilize the protein pellet with 6 M GuHCl prepared
in 50 mM NH4HCO3, pH 7.8. For 100 μg of protein, use
20 μL of 6 M GuHCl buffer.
(b) Add 100 % ACN and 1 M CaCl2 solution to get final
concentrations of 5 % and 2 mM, respectively.
(c) Vortex the tubes to solubilize the pellet completely.
(d) Dilute GuHCl concentration from 6 M to 0.2 M with
50 mM NH4HCO3, pH 7.8.
(e) Take an aliquot of 1 μg of protein and label as “before
digest.”
30 Fiorella A. Solari et al.

(f) Add trypsin solution [1 mg/mL] to get 1:20 (w/w) ratio

of enzyme to protein and incubate the sample tubes at
37 °C for 14 h [18]. See Note 6.
(g) Stop the enzyme activity by adding 10 % TFA to a final
concentration of 1 %.

3.3 Digestion Control 1. Take an aliquot of 1 μg of peptides. Label as “after digest.”

with Monolithic 2. For all samples measure “before digest” and “after digest” on an
Reversed-Phase RP monolithic HPLC system to evaluate the digestion efficiency
Chromatography and reproducibility. See Note 7 [19].
3. If the samples look well digested and reproducible, then
continue with Subheading 3.5; otherwise consider repeating
the sample preparation.

3.4 SPEC Sample 1. To improve the reproducibility of this step, use a vacuum
Cleanup (Desalting) manifold system for peptide desalting. Use C18 AR 4 mg
material (Agilent) tips.
(a) Activation: Three times with 100 μL of 100 % ACN.
(b) Equilibration: Three times with 100 μL of 0.1 % TFA.
(c) Sample loading: Place the sample on the material and
reload the flow through three times.
(d) Washing: Three times with 100 μL of 0.1 % TFA.
(e) Peptides elution: Two times with 100 μL of 60 % ACN in
0.1 % TFA
2. To control the reproducibility of the desalting procedure, take
a small but equal volume (e.g., 2 μL) of eluate from each sam-
ple and dry it completely in the SpeedVac. Snap freeze the
remaining volume using liquid nitrogen and store the frozen
eluates at −80 °C until further use. See Note 8.
3. Measure the samples (2 μL) using nano-LC with UV detection
(214 nm) or LC-MS. If they look reproducible, take 100 μg
aliquots from each sample, dry them in the SpeedVac, and
proceed with iTRAQ labeling.

3.5 iTRAQ Labeling 1. Resolubilize each peptide pellet (~100 μg) in 30 μL of dissolution
buffer (0.5 M TEAB, pH 8.5).
2. Perform the labeling reaction as per the manufacturer’s instruc-
tions (iTRAQ-8plex, AB SCIEX).
3. After incubation, combine all the eight differentially labeled
samples in a new 1.5 mL LoBind Eppendorf tube.
4. Divide the multiplexed sample into two parts. Use one part
corresponding to ~5 % (~40 μg) for complete (global)
proteome analysis and the second part (~95 % or 760 μg) for
the enrichment of phosphopeptides.
5. Dry both parts completely in the SpeedVac and store the pellets
at −80 °C until further use.
 Two Birds with One Stone: Parallel Quantification of Proteome and Phosphoproteome… 31

3.6 Fractionation Pre-fractionation can be conducted at pH 10 [20] albeit high pH

at High-pH RP conditions affect the column stability due to the hydrolysis of silox-
for Complete ane groups in the fused silica [21, 22]. Therefore, we recommend
Proteome Analysis performing the RP fractionation at pH 8.0; nevertheless, pH 10
yields slightly higher orthogonality.
1. Solubilize the dried multiplexed sample (~40 μg) in buffer A.
2. Perform peptide fractionation on a Biobasic C18, 0.5
mm × 15 cm column using an UltiMate 3000 liquid chroma-
tography (LC) system (Thermo Fisher Scientific) with the
following gradient:
0–3 % B in 15 min, 3–50 % B in 65 min, 50–60 % B in 10
min, 95 % B hold for 5 min, 95 %–3 % B in 5 min, and finally
re-equilibrate the column with 3 % B for 20 min.
3. Collect 16 fractions at 1-min interval from 15 to 70 min in a
concatenation mode (Fig. 1). To minimize sample losses,
collect fractions directly in HPLC sample inlets.
4. Finally, dry the collected fractions in the SpeedVac.

Fig. 1 Pre-fractionation of an iTRAQ-labeled sample on a high-pH C18 reversed-phase (RP) system. The pep-
tides are separated using a binary gradient (buffer A: 10 mM NH4HCO2, buffer B: 10 mM, NH4HCO2, 84 % ACN,
both pH 8.0) ranging from 3 to 50 % buffer B in 65 min. In total, 16 fractions are collected at 1-min intervals
using a concatenation approach. Separation of the peptides on high-pH RP columns and subsequent concat-
enation of the eluted fractions reduce sample complexity, improve selectivity, and increase proteome coverage
[23]. See Note 10
32 Fiorella A. Solari et al.

3.7 Phosphopeptide The following protocol for the enrichment of phosphopeptides is

Enrichment Using TiO2 based on the workflow published by Larsen and co-workers [24].
Beads
1. Resolubilize the multiplexed sample (~760 μg) in 1 mL of
loading buffer 1.
2. Adjust the amount of TiO2 beads according to the starting
material. Use a bead-to-peptide ratio of 6:1, e.g., weigh in
4.56 mg of TiO2 beads for ~760 μg of sample. Wash the beads
with 200 μL of loading buffer 1. Pellet the beads and discard
the supernatant.
3. To the washed beads, add 200 μL of loading buffer 1 and vor-
tex to obtain a homogenous suspension.
4. Add appropriate amount of beads to the peptides and incubate
the peptide-bead suspension for 10 min on a vortex mixer at low
speed. Afterwards, centrifuge at 3000 rcf for 1 min to pellet the
beads. Collect the supernatant in a LoBind Eppendorf tube.
5. Repeat the enrichment process in a stepwise manner (Fig. 2)
by treating the supernatant with freshly prepared suspension of

Fig. 2 Serial enrichment of phosphopeptides from an iTRAQ-labeled sample using TiO2 beads. It is recom-
mendable to perform a second round of enrichment (from Step 2) to enhance the phosphoproteome coverage.
The supernatants (1–3), which contain mostly non-phosphorylated peptides, can be combined and used for,
e.g., global proteome analysis. See Note 9
Two Birds with One Stone: Parallel Quantification of Proteome and Phosphoproteome… 33

TiO2 beads. However, use bead-to-peptide ratios of 3:1 and

1.5:1 for the subsequent enrichment steps, i.e., 2.28 mg and
1.14 mg of TiO2 beads, respectively. Collect the supernatant in
a LoBind Eppendorf tube.
6. Pool the beads from all the three enrichment steps in one
LoBind Eppendorf tube using 100 μL of loading buffer 1.
Centrifuge as above to pellet the beads.
7. Collect the supernatant and combine with the one obtained
from step 5.
8. Wash the beads with 100 μL of TiO2 washing buffer 1, vortex,
and centrifuge as in step 4 to pellet the beads. This step is
important to remove non-phosphorylated peptides, which
bind in a HILIC mode unspecifically to TiO2.
9. Afterwards, perform another washing step using 100 μL of
washing buffer 2 and pellet the beads as mentioned above.
Collect the supernatant and combine with the one obtained
from step 7. See Note 9.
10. Finally, dry the beads completely in the SpeedVac.
11. Resuspend the beads in 100 μL of elution buffer, briefly vortex,
and incubate the sample on a vortex mixer at low speed for
10 min to elute the phosphopeptides from the beads.
12. Transfer the eluate containing the phosphopeptides into a new
LoBind Eppendorf tube and acidify the sample with 2 μL of 10
% TFA and 8 μL of 100 % FA.
13. Dry the sample completely in the SpeedVac.
14. To increase the phosphoproteome coverage, a second serial
enrichment step with TiO2 beads should be performed. However,
use loading buffer 2 instead of 1 and repeat steps 2–5.
15. Next, wash the beads with washing buffer 3, centrifuge to pel-
let the beads, and remove the supernatant. Dry the beads in
the SpeedVac.
16. Afterwards, elute the phosphopeptides from the beads using
100 μL of TiO2 elution buffer. Vortex and incubate the sam-
ples on a vortex mixer at low speed for 10 min.
17. In the meantime prepare a C8 stage tip; cut 5–6 mm from the
bottom of a 2–200 μL pipette tip. Use it as a stamp to excise a
small piece from the C8 Disc Empore. Place this piece on the
top of a new 2–200 μL pipette tip. Use a gel loader tip to push
down the piece of C8 material to the bottom of the tip and
finally make sure that the C8 material is properly fixed [25].
18. After 10 min, pellet the beads by centrifugation at 3000 rcf for
1 min and transfer the eluate onto the self-made C8 stage tip
(see above) to remove possible residues of beads.
19. For the following steps, use an air-filled syringe to allow the
passage of the liquid through the material.
34 Fiorella A. Solari et al.

20. Recover the flow-through containing the phosphopeptides

into a LoBind Eppendorf tube.
21. To further elute the phosphopeptides that might be still
attached to the beads, add 30 μL of TiO2 elution buffer to the
beads, vortex, and centrifuge to pellet the beads at 3000 rcf for
1 min. Collect the eluate and transfer onto the previously used
C8 stage tip and recover the flow-through in the same LoBind
Eppendorf tube (step 20).
22. For maximum recovery of the phosphopeptides, place 2–3 μL
of 30 % ACN over the C8 tip and collect the flow through in
the LoBind Eppendorf tube (step 21).
23. Finally, prior to the SPE acidify the eluate with 2 μL of 10 % TFA
and 8 μL of 100 % FA. Make sure that the pH is around 2.0.

3.8 SPE Cleanup 1. Prepare a C18 stage tip following the instructions from
of the step 17, but use C18 material instead of C8 material.
Phosphopeptide- 2. Weigh in 5 mg of Oligo R3 material and add 200 μL of 70 %
Enriched Sample ACN in 0.1 % TFA.
3. Fill the tip with 10 μL of R3 material (the height of the R3
material should be between 2 and 3 mm).
4. For the following steps, use an air-filled syringe to allow the
passage of the liquid through the material.
5. Activate the material with 50 μL of 100 % ACN. Repeat this
step twice.
6. Equilibrate the material with 50 μL of 0.1 % TFA. Repeat this
step twice.
7. Load the acidified phosphopeptides (from step 23 of
Subheading 3.8) onto the stage tip. Reload the flow-through
once more.
8. Wash the material with 50 μL of 0.1 % TFA. Repeat this step
twice.
9. Finally, elute the phosphopeptides from the material with 50
μL of 98 % ACN in 0.1 % TFA, in an HPLC vial, and directly
proceed with HILIC fractionation.

3.9 HILIC 1. Perform separation of the TiO2-enriched phosphopeptide

Fractionation sample on a polar-phase column (TSK gel) 150 μm × 15 cm, 5
μm using an UltiMate 3000 liquid chromatography (LC) sys-
tem (Thermo Fisher Scientific) with the following gradient:
1 % B for 20 min, 1–15 % B in 1.5 min, 15–40 % B in 37.5
min, 40–80 % B in 5 min, 80 % B hold for 5 min, and finally
80–1 % B for 6 min.
2. Collect ten fractions at 3.5-min intervals, as depicted in Fig. 3.
3. Dry the collected fractions completely in the SpeedVac.
 Two Birds with One Stone: Parallel Quantification of Proteome and Phosphoproteome… 35

Fig. 3 An off-line-HILIC fractionation of phosphopeptides. The peptides (previously enriched using TiO2 beads)
are separated using a binary gradient (solvent A: 0.1 % TFA, 98 % ACN, solvent B: 0.1 % TFA) ranging from 15
to 40 % of solvent B in 37.5 min. HILIC fractionation of a pre-enriched sample improves the specificity and
increases the phosphoproteome coverage [26]

3.10 LC-MS/MS 1. LC conditions.

Analysis (a) Complete proteome: Dissolve each peptide fraction (from
Subheading 3.7) in an appropriate volume of HPLC load-
ing buffer (0.1 % TFA). Load each fraction with 0.1 % TFA
and a flow rate of 20 μL/min for 10 min onto the trap
column, followed by separation of peptides on the main
column using a binary gradient ranging from 3 to 42 % B
in 140 min at a flow rate of 250 nL/min at 60 °C.
(b) Phosphoproteome: Use similar HPLC conditions as above,
i.e., columns, buffers, and flow rates. Owing to the reduced
complexity as compared to the global proteome analysis,
analyze each fraction using a 55-min gradient, ranging
from 3 to 42 % B.
2. MS conditions
(a) Operate the Q-Exactive mass spectrometer in data-
dependent acquisition mode.
(b) Complete proteome: Acquire MS survey scans at a resolu-
tion of 70,000 with a target value of 1 × 106 ions and maxi-
mum injection time of 120 ms. Acquire MS/MS scans of
15 most abundant ions (Top 15) using 2 × 105 ions as tar-
get value and a maximum fill time of 120 ms. Use a nor-
malized collision energy (NCE) of 30 and a dynamic
exclusion of 20 s. Set the first fixed mass 100 m/z to allow
a good signal intensity of the first reporter ion (i.e., 113)
and select only precursor ions with charge state between
+2 and +5 for MS/MS fragmentation (Fig. 4).
(c) Phosphoproteome analysis: Same as above, but using an
NCE of 35 and a dynamic exclusion of 12 s. See Note 10.
 36
Fiorella A. Solari et al.

Fig. 4 MS/MS spectrum of the phosphopeptide iTRAQ-QPGLRQPsPSHDGSLSPLQDR (s = phosphorylated serine) acquired at 17,500 resolution in the Orbitrap mass
analyzer. The low-mass region contains the iTRAQ 8plex reporter ions that represent the relative abundances of the same peptide in eight different biological
samples
 Two Birds with One Stone: Parallel Quantification of Proteome and Phosphoproteome… 37

3.11 Data Analysis 1. Search the mass spectrometry (MS) raw data against a human
Uniprot database (see Note 11) with Proteome Discoverer
(PD) software. Use a mere “target” database as PD generates
random decoy hits on the fly.
2. In PD, use the spectrum selector node for precursor ion selec-
tion to process MS raw data with default settings.
3. Use the reporter ion quantifier node and select iTRAQ-8plex
as the quantification method. See Note 12.
4. To maximize the number of peptide spectrum matches (PSMs),
include different search algorithms (Mascot, SEQUEST, and
MS Amanda) using the same set of parameters, i.e., precursor
and fragment ion tolerances of 10 ppm and 0.02 Da for MS
and MS/MS, respectively; trypsin as enzyme with a maximum
of two missed cleavages; carbamidomethylation of Cys,
iTRAQ-8plex on N-terminus, and Lys as fixed modifications;
oxidation of Met, and phosphorylation (only for the phospho-
proteome data) of Ser, Thr, and Tyr as variable modifications.
5. Use the peptide validator node to filter the data with a false
discovery rate (FDR) of 1 %.
6. For the phosphoproteome data analysis, add the PhosphoRS
[17] (version 3.1) node to score localization probabilities for
the identified phosphorylation sites.

3.12 Data Evaluation 1. Complete proteome:

(a) Export the list of unique proteins from PD to a Microsoft
Excel spread sheet. See Note 13.
(b) To improve the reliability of the relative protein quantifica-
tion, select only those proteins that are identified with at
least two unique peptides.
(c) PD only provides ratios; that is, only seven values are given
in relation to a selected reference (here 113) sample. To
allow a statistic comparison in case of comparing four
against four samples, create an artificial 113/113 ratio of
1.0 for each protein.
(d) For each protein, calculate the median by taking the ratios
from different iTRAQ channels (113/113, …, 121/113).
(e) Then, divide the iTRAQ ratio of each protein by the previ-
ously calculated median to obtain relative abundance per
individual channel.
(f) Use a Student’s t-test to determine p-values for the respec-
tive sample groups/conditions.
(g) Calculate average ratios for the replicates and determine
regulated proteins, e.g., by selecting those that have
p-values < 0.05 and fold changes >2.
38 Fiorella A. Solari et al.

2. Phosphoproteome:
(a) Export the phosphopeptide PSM list from PD to Microsoft
Excel. See Note 13.
(b) Select only unique phospho-PSMs that were used for
quantification.
(c) Calculate the normalization factor as described previously
(from Subheading 3.12) and normalize each iTRAQ channel
(see Subheading 3.12, steps 4(a)–(d)).
(d) With the help of ready-to-use Excel macro provided
by Mechtler lab (http://ms.imp.ac.at/?goto=phosphors),
determine the confident phosphorylation sites for each
peptide, as well as the position of the phosphorylation
within the protein sequence. (The analysis should be done
in the same Excel worksheet.)
(e) In Excel concatenate: (1) peptide sequence, (2) protein
accession, and (3) phospho-RS phosphorylation site
(considering only those with probabilities ≥95 %) to define
distinguishable phosphopeptide PSMs. See Note 14.
(f) Sort the data according to the concatenated row to group
phospho-PSMs that belong together. For those, determine
the median normalized abundance values per iTRAQ
channel, as well as the relative standard deviation.
(g) If the experiment consists of replicates, determine student’s
t-test and significantly change phosphopeptides as done
for the global proteome.

4 Notes

1. IAA solution is unstable and light sensitive. Therefore, prepare

the stock solution freshly and use it within 1 h after preparation.
In addition, buffers that either contain sulfhydryls or are not
slightly alkaline (pH 7.5–8.0) should be avoided. Moreover,
excess of IAA or non-buffered IAA reagent can lead to the alkyla-
tion of amines (lysine, N-termini), thioethers (methionine),
imidazoles (histidine), and carboxylates (aspartate, glutamate).
2. Glycolic acid improves the selectivity to enrich phosphopeptides
using TiO2, by reducing unspecific binding from non-
phosphorylatedpeptides [13].
3. Besides BCA, other calorimetric assays such as Bradford or
modified-Lowry could be used for the determination of protein
concentration. Irrespective of the method, the assay should
provide comparably accurate protein concentrations as it is
decisive for calculating the amount of trypsin (for digestion)
and TiO2 beads (for phosphopeptide enrichment).
Two Birds with One Stone: Parallel Quantification of Proteome and Phosphoproteome… 39

4. The use of cold organic solvents (ethanol and acetone) for

protein precipitation is a fast and easy procedure that yields
reproducible results. Nevertheless, membrane filter-based
sample preparation protocols such as filter-aided sample prepa-
ration (FASP) [27, 28] are widely used nowadays. However,
FASP involves urea, which could lead to undesired side reac-
tions, especially carbamylation of primary amines on N-terminus
and Lys side chains [29].
5. Avoid contamination with keratin at this step and do not dry
the pellet in an oven or a thermomixer. Instead, air-dry the
protein pellet under the laminar flow hood.
6. It has been shown that the presence of phosphoamino acids
(Ser and Thr) near to the cleavage site of trypsin could impair
cleavage. Moreover, 50 mM TEAB buffer also shows slightly
reduced cleavage efficiency, compared to NH4HCO3 buffer.
Therefore, for enhancing digestion efficiency, thus improve
the coverage of the phosphoproteome, and achieve better
reproducibility, it is recommended to use 1:20 (w/w)
trypsin-to-protein ratio and perform the proteolytic digestion
in 50 mM NH4HCO3 buffer [18].
7. It is most important to quality control the enzymatic digests as
this step is critical for the subsequent quantitative analysis [19].
The monolithic HPLC system provides a rapid and direct
comparison of the samples as it can be used to measure pro-
teins and peptides. Additionally, monolithic columns are more
robust and sensitive in contrast to other techniques that are
used for digestion control such as SDS-PAGE followed by
Coomassie or silver staining.
8. Resolubilize the dried peptides in HPLC loading buffer (0.1 %
TFA) and analyze each sample on a nanoLC system with UV
detection (214 nm) or if possible MS-coupling, using a 90-min
gradient ranging from 3 to 42 % of HPLC buffer B (84 % ACN
in 0.1 % FA). Compare UV and/or total ion chromatogram
(TICs) intensities of all samples and if necessary correct the
amounts to compensate for systematic errors such that each
sample has identical starting material before labeling with
iTRAQ reagents.
9. The supernatant of the enrichment steps contains mostly non-
phosphorylated peptides. It is possible to keep it and combine
it with the supernatant from the washing steps; therefore, after
SPE this sample can again be used for global proteome analysis.
In general, it is possible to use the entire multiplexed sample
for phosphopeptide enrichment, and just use the loading
and washing supernatants for the global proteome analysis
(after cleaning by SPE C18). However, we recommend using
the completely “unbiased” sample prior to TiO2 incubation
40 Fiorella A. Solari et al.

and take into account losing 5 % of the sample amount for

phosphopeptide enrichment instead as the loss in sensitivity
might be negligible.
10. Due to the depletion of unmodified peptides and the HILIC
fractionation, the resulting peptide fractions are not highly
complex. Therefore, set the dynamic exclusion to 12 s on the
Q-Exactive mass spectrometer (Thermo Fisher Scientific) and
reduce gradient length.
11. Note down the date and the source of the download, as well as
the number of target (forward) sequences.
12. Select the vendor-provided isotope correction factors for
iTRAQ-8plex reagents in the reporter ion node of Proteome
Discoverer software.
13. Before exporting the unique protein list from the Proteome
Discoverer, apply the data reduction filters such as high confi-
dence corresponding to an FDR <1 % on the PSM level and
peptide search engine rank of 1.
14. This step is necessary to allow separating phospho-PSMs in
clearly distinguishable groups, such that PSMs that are identi-
fied with the same sequence but different positions of the
phosphorylation site are differentiated.

References
1. Macek B, Mann M, Olsen JV (2009) Global protein phosphorylation. J Am Soc Mass
and site-specific quantitative phosphopro- Spectrom 18(11):1932–1944
teomics: principles and applications. Annu Rev 7. Beck F, Geiger J, Gambaryan S et al (2013)
Pharmacol Toxicol 49(1):199–221 Time-resolved characterization of cAMP/
2. Batalha IL, Lowe CR, Roque ACA (2012) PKA-dependent signaling reveals that platelet
Platforms for enrichment of phosphorylated inhibition is a concerted process involving
proteins and peptides in proteomics. Trends multiple signaling pathways. Blood 123(5):
Biotechnol 30(2):100–110 e1–e10
3. Boja ES, Phillips D, French SA et al (2009) 8. Wu Y-B, Dai J, Yang XL et al (2009) Concurrent
Quantitative mitochondrial phosphoproteomics quantification of proteome and phosphopro-
using iTRAQ on an LTQ-Orbitrap with high teome to reveal system-wide association of
energy collision dissociation. J Proteome Res protein phosphorylation and gene expression.
8(10):4665–4675 Mol Cell Proteomics 8(12):2809–2826
4. Jones AE, Nühse T (2011) Phosphoproteomics 9. Ross PL, Huang YN, Marchese JN et al (2004)
using iTRAQ. In: Dissmeyer N, Schnittger A Multiplexed protein quantitation in
(eds) Plant kinases. Humana Press, New York, Saccharomyces cerevisiae using amine-reactive
pp 287–302 isobaric tagging reagents. Mol Cell Proteomics
5. Mertins P, Udeahi ND, Clauser KR et al 3(12):1154–1169
(2012) iTRAQ labeling is superior to mTRAQ 10. Gilar M, Olivova P, Daly AE et al (2005) Two-
for quantitative global proteomics and phos- dimensional separation of peptides using
phoproteomics. Mol Cell Proteomics RP-RP-HPLC system with different pH in first
11(6):M111.014423 and second separation dimensions. J Sep Sci
6. Liang X, Fonnum G, Hajivandi M et al (2007) 28(14):1694–1703
Quantitative comparison of IMAC and TiO2 11. Thingholm TE, Jorgensen TJ, Jensen ON et al
surfaces used in the study of regulated, dynamic (2006) Highly selective enrichment of phos-
 Two Birds with One Stone: Parallel Quantification of Proteome and Phosphoproteome… 41

phorylated peptides using titanium dioxide. 21. Batth TS, Francavilla C, Olsen JV (2014)
Nat Protoc 1(4):1929–1935 Off-line high-pH reversed-phase fractionation
12. Alpert AJ (1990) Hydrophilic-interaction for in-depth phosphoproteomics. J Proteome
chromatography for the separation of peptides, Res 13(12):6176–6186
nucleic acids and other polar compounds. 22. Castillo A, Roig-Navarro AF, Pozo OJ (2007)
J Chromatogr 499:177–196 Secondary interactions, an unexpected problem
13. Thingholm TE, Larsen MR (2009) The use of emerged between hydroxyl containing analytes
titanium dioxide micro-columns to selectively and fused silica capillaries in anion-exchange
isolate phosphopeptides from proteolytic micro-liquid chromatography. J Chromatogr A
digests. Methods Mol Biol 527:57–66 1172(2):179–185
14. Perkins DN, Pappin DJ, Creasy DM et al 23. Yang F, Shen Y, Camp DG et al (2012) High
(1994) Probability-based protein identification pH reversed-phase chromatography with
by searching sequence databases using mass fraction concatenation for two-dimensional
spectrometry data. Electrophoresis 20(18): proteomic analysis. Expert Rev Proteomics
3551–3567 9(2):129–134
15. Geer LY, Markey SP, Kowalak JA et al (2004) 24. Larsen MR, Thingholm TE, Jensen ON et al
Open mass spectrometry search algorithm. (2005) Highly selective enrichment of phos-
J Proteome Res 3(5):958–964 phorylated peptides from peptide mixtures
16. Eng JK, McCormack AL, Yates JR (1994) An using titanium dioxide microcolumns. Mol
approach to correlate tandem mass spectral Cell Proteomics 4(7):873–886
data of peptides with amino acid sequences in a 25. Dickhut C, Radau S, Zahedi RP (2014)
protein database. J Am Soc Mass Spectrom Fast, efficient, and quality-controlled phos-
5(11):976–989 phopeptide enrichment from minute sample
17. Taus T, Kocher T, Pichler P et al (2011) amounts using titanium dioxide. In: Martins-
Universal and confident phosphorylation site de-Souza D (ed) Shotgun proteomics.
localization using phosphoRS. J Proteome Res Springer, New York, pp 417–430
10(12):5354–5362 26. Engholm-Keller K, Birck P, Storling J et al
18. Dickhut C, Feldmann I, Lambert J et al (2014) (2012) TiSH—a robust and sensitive global
Impact of digestion conditions on phosphopro- phosphoproteomics strategy employing a
teomics. J Proteome Res 13(6):2761–2770 combination of TiO2, SIMAC, and HILIC.
J Proteomics 75(18):5749–5761
19. Burkhart JM, Schumbrutzki C, Wortelkamp S
et al (2012) Systematic and quantitative com- 27. Wisniewski JR, Zougman A, Nagaraj N et al
parison of digest efficiency and specificity (2009) Universal sample preparation method
reveals the impact of trypsin quality on for proteome analysis. Nat Methods 6(5):
MS-based proteomics. J Proteomics 75(4): 359–362
1454–1462 28. Manza LL, Stamer SL, Ham AJ et al (2005)
20. Toll H, Oberacher H, Swart R et al (2005) Sample preparation and digestion for pro-
Separation, detection, and identification of teomic analyses using spin filters. Proteomics
peptides by ion-pair reversed-phase high- 5(7):1742–1745
performance liquid chromatography-electrospray 29. Kollipara L, Zahedi RP (2013) Protein car-
ionization mass spectrometry at high and low bamylation: in vivo modification or in vitro
pH. J Chromatogr A 1079(1–2):274–286 artefact? Proteomics 13(6):941–944
 Chapter 4

Selected Reaction Monitoring to Measure Proteins

of Interest in Complex Samples: A Practical Guide
Yuehan Feng and Paola Picotti

Abstract
Biology and especially systems biology projects increasingly require the capability to detect and quantify
specific sets of proteins across multiple samples, for example the components of a biological pathway
through a set of perturbation-response experiments. Targeted proteomics based on selected reaction mon-
itoring (SRM) has emerged as an ideal tool to this purpose, and complements the discovery capabilities of
shotgun proteomics methods. SRM experiments rely on the development of specific, quantitative mass
spectrometric assays for each protein of interest and their application to the quantification of the protein
set in various biological samples. SRM measurements are multiplexed, namely, multiple proteins can be
quantified simultaneously, and are characterized by a high reproducibility and a broad dynamic range. We
provide here a practical guide to the development of SRM assays targeting a set of proteins of interest and
to their application to complex biological samples.

Key words Selected reaction monitoring, Targeted proteomics, Protein quantitation, Assay design,
Assay validation

1 Introduction

In the last two decades mass spectrometry (MS)-based proteomics

has evolved into a powerful technique for large-scale protein analysis,
and is now routinely applied to identify and quantify proteins in a
variety of biological samples [1]. Classical, unbiased proteomics
approaches, commonly referred to as shotgun proteomics, are used
in discovery-based projects to generate an inventory of the protein
content of a sample and to identify proteins with varying abun-
dance across different perturbing conditions. In these workflows,
proteins are extracted from biological samples, denatured, and
proteolytically cleaved with a specific enzyme (typically trypsin).
The resulting peptide mixtures are subjected to liquid chromatog-
raphy (LC) separation, at the end of which peptide ions are trans-
ferred to the gas phase and enter the mass spectrometer. In shotgun
approaches, peptide ions are then subjected to data-dependent MS

Jörg Reinders (ed.), Proteomics in Systems Biology: Methods and Protocols, Methods in Molecular Biology, vol. 1394,
DOI 10.1007/978-1-4939-3341-9_4, © Springer Science+Business Media New York 2016

43
44 Yuehan Feng and Paola Picotti

acquisition, where fragmentation spectra are generated from

selected peptide precursors based on their measured intensity.
Peptide identifications from MS spectra are validated by statistical
models and translated into protein identifications. While extremely
powerful in the large-scale characterization of the protein content
of biological samples, shotgun approaches are not ideal when spe-
cific proteins need to be measured across various samples, as their
intrinsic semi-stochastic nature might affect the detection and con-
sistent quantification of the proteins of interest across the sample
set. To address this limitation, a parallel proteomics workflow
involving selected reaction monitoring (SRM) mass spectrometry
was proposed and has quickly developed into the gold standard for
the targeted detection and quantification of specific proteins in
complex biological matrices [2].
SRM-based proteomics experiments share an identical sample
preparation workflow with shotgun approaches, until the mass
spectrometric analysis. An SRM measurement starts with the defi-
nition of a set of proteins of interest, for example all the compo-
nents of a protein complex or a biological pathway, a set of
biomarker candidate proteins, or, in general, proteins that are rel-
evant to address a given biological question. For each of these pro-
teins then specific, sensitive, and quantitative mass spectrometric
assays based on SRM are developed and subsequently applied to
the measurement of the target proteins in biological samples. SRM
assays consist of MS coordinates that enable the selective measure-
ment on a triple-quadrupole mass spectrometer (QqQ) of each
target protein, through measurement of some of its representative
peptides. These coordinates consist of pairs of mass-to-charge
(m/z) values (so-called SRM transitions) that are selected with the
first (Q1) and second (Q3) analyzer of a QqQ, to isolate a peptide
ion and corresponding fragment ions generated upon fragmenta-
tion of the peptide in a collision cell (q2). The detector placed at
the end of the quadrupole series counts ions matching the defined
m/z values and returns a signal intensity over the chromatographic
time. The area under the resulting (SRM) peak is proportional to
the amount of the protein initially contained in the sample. SRM
measurements are characterized by a high reproducibility and a
broad dynamic range (up to 4.5 orders of magnitude). SRM data
acquisition is also multiplexed; that is, several SRM transitions can
be sequentially measured within an MS duty cycle, thus allowing
for the concurrent quantification of multiple peptides and proteins.
When targeting many transitions, however, the time spent measur-
ing each transition (dwell time) will be reduced, resulting in a
lower signal-to-noise of the associated peaks, and thus lower sensi-
tivity for the target peptides. To address this issue, the scheduled
acquisition of SRM transitions was devised, where transitions for a
 Selected Reaction Monitoring to Measure Proteins of Interest in Complex Samples… 45

given peptide are monitored only for a short time interval, centered
around the known elution time of the peptide. This decreases the
number of concurrent SRM transitions measured at a given chro-
matographic time, allowing optimizing cycle time and dwell time
while measuring larger numbers of peptides per MS analysis. Using
this approach, up to 150 proteins or several hundred peptides [3]
can be simultaneously monitored in a 30-min LC-SRM run. Once
developed, SRM assays are stable and applicable to measuring the
target proteins in a variety of samples at high throughput. The
application of SRM in biological and biomedical projects has been
further promoted by a growing number of open-access databases
[4] providing experimentally validated SRM coordinates and by
automated or semiautomated data analysis tools [5, 6].
In the following, we provide a basic protocol for developing
SRM assays and conducting SRM measurements, assuming that a
list of proteins of interest has been defined.

2 Materials

2.1 Sample 1. Denaturation buffer: 8 M Urea, 0.1 M ammonium bicarbonate

Preparation (AmBic), pH 8.0.
2. Estimation of total protein amount: Bicinchoninic acid (BCA)
Protein Assay Kit (Thermo Scientific, Rockford, IL, USA).
3. Reduction buffer: 1 M Tris(2-carboxyethyl)phosphine hydro-
chloride (TCEP–HCl).
4. Alkylation buffer: 1 M Iodoacetamide (IAA).
5. Proteases: Endoproteinase Lys-C from Lysobacter enzymogenes
and porcine trypsin (both sequencing grade).
6. Peptide desalting: Acetonitrile; 0.1 % formic acid; 60 % aceto-
nitrile; Sep-Pak tC18 Vac Cartridges (Waters, Milford, MA,
USA).

2.2 LC-SRM/MS 1. Buffer A: HPLC-grade water with 0.1 % (v/v) formic acid.
2. Buffer B: HPLC-grade acetonitrile with 0.1 % (v/v) formic
acid.
3. A QqQ spectrometer equipped with a nano-electrospray ion
source and interfaced with a liquid chromatography (LC) system
operating in the nanoliter/min flow rate range.
4. A chromatographic column for nano-LC separation, packed
with C18 resin (20 cm length × 75 μm diameter).
5. Optional
(a) iRT Kit (Biognosys, Switzerland).
46 Yuehan Feng and Paola Picotti

3 Methods

3.1 Sample 1. Depending on the sample type under investigation (e.g., bac-
Preparation teria, yeast, mammalian cells, human tissue, or plasma) differ-
ent protein extraction procedures are used. After protein
extraction, estimate the concentration of the protein solution
using the BCA assay. Typical protein extraction buffers may
contain reagents that interfere with the BCA assay (e.g., DTT)
or with the subsequent MS analysis (e.g., detergents such as
sodium dodecyl sulfate or NP-40) and should thus be omitted
or diluted prior to the BCA assay.
2. The extraction buffer should contain a chaotropic agent at
high concentration, such as 8 M urea, to achieve complete pro-
tein denaturation. If the protein extract derives from a protein
precipitation step (for example using cold acetone), add the
denaturation buffer to the protein pellet to a final protein con-
centration of 2–3 mg/ml.
3. Adjust the pH of the protein extract to ~8.
4. Add the reduction buffer to a final concentration of 5 mM
TCEP–HCl, followed by incubation for 30 min at 37 °C to
reduce disulfide bridges.
5. Add the alkylation buffer to a final IAA concentration of 40
mM, followed by incubation at 25 °C in the dark for 45 min to
alkylate free cysteine residues.
6. Dilute the protein solution with 0.1 M AmBiC to a final con-
centration of 6 M urea to prevent denaturation of the protease
used in the first digestion step.
7. To achieve completeness of the digestion reaction, two proteases
are sequentially applied, Lys-C (cleaving at the C-terminus of
lysine) followed by trypsin (cleaving at the C-terminus of lysine
and arginine). Add first Lys-C to the protein solution to an
enzyme-to-substrate (E:S) ratio of 1:100, and incubate the
reaction mixture at 37 °C for 4 h. Lys-C tolerates harsher con-
ditions (e.g., higher denaturant concentration) compared to
trypsin. Therefore, dilute the proteolytic mixture again prior
to addition of trypsin to a final concentration of 2 M urea.
Add then trypsin to an E:S ratio of 1:100 and incubate the
proteolytic mixture at 37 °C overnight.
8. Stop the tryptic digestion by addition of formic acid to pH <3.
9. Desalt the peptide mixture using Sep-Pak tC18 Vac cartridges,
according to the manufacturer’s instruction, and using acetoni-
trile for the initial wash of the cartridges, 0.1 % formic acid for
cartridge equilibration and peptide desalting, and 60–80 %
acetonitrile for peptide elution. Evaporate the peptide mixture
eluted in 60–80 % acetonitrile in a vacuum centrifuge to dryness.
 Selected Reaction Monitoring to Measure Proteins of Interest in Complex Samples… 47

10. Resuspend the peptide pellet in buffer A to a concentration of

1 mg peptides per ml, based on the protein amount estimated
using the BCA assay at step 1.

3.2 SRM A protein SRM assay includes SRM transitions for at least one
Assay Design unique peptide generated upon digestion of the target protein, and
ideally comprises transitions for multiple peptides from that pro-
tein to improve reliability in the protein detection and quantifica-
tion steps. An SRM assay for a peptide in turn minimally includes
the mass-to-charge ratio (m/z) of a peptide precursor ion and
those of a set of fragments generated upon collision-induced dis-
sociation of the precursor. The design of protein SRM assays
involves selection of optimal peptide and fragment ion coordinates
and is followed by refinement and validation of the assays, after
which the assays can be applied to the quantification of the target
proteins. The information required to design SRM assays can be
obtained from in silico analyses, databases, or shotgun proteomics
data, as described below (Fig. 1).
In silico design: Ideal peptides for SRM uniquely map to the target
protein and show good ionization and fragmentation proper-
ties, resulting in intense SRM peaks. If the number of target
proteins is limited (for example, below 20), ideal peptides and
their fragments can simply be selected by testing the MS
performance of all unique peptides generated upon tryptic
digestion of each protein [7]. This involves retrieval of the target

Peak Statistical
Validation evaluation analysis
SRM assays

Refinement

Fig. 1 Workflow of SRM-based targeted proteomic analyses. Coordinates of protein SRM assays are generated
by in silico design, retrieved from Web-based repositories or derived from shotgun proteomics experiments.
The assay coordinates are validated using MS/MS spectra filtered to an acceptable FDR or synthetic peptides.
The assay coordinates are measured by SRM for assay refinement and the final SRM transitions are applied to
perform quantitative SRM measurements across the samples of interest. The SRM peak data are scrutinized
using tools such as Skyline and statistically evaluated
48 Yuehan Feng and Paola Picotti

protein sequences from protein databases, their in silico tryptic

digestion, prediction of all possible (b- and y-) ion fragments
for each peptide precursor, and analysis of the intensity and
specificity of the generated transitions by SRM, using samples
that contain the protein of interest. Testing large numbers of
transitions, even for a small set of proteins, might require con-
siderable instrument time. Therefore, prioritization criteria are
typically applied to analyze only transitions with the highest
likelihood of resulting in detectable and reliable SRM peaks.
For example, peptides which have been previously detected in
proteomics databases with a large number of observations
across different experiments should be prioritized (see below).
Further prioritization criteria are in Note 1. When the number
of target proteins is large, the number of peptides to test by
SRM can be further reduced by computational prediction of
the best responding peptides (those that can be detected with
intense MS signals), using tools such as PeptideSieve [8], the
ESP predictor [9], PeptideRank [10], or PeptidePicker [11].
The assays generated by in silico design need to be additionally
validated (see below).
Databases: Web-based repositories of SRM assays, such as the
SRMAtlas [12] (www.srmatlas.org), the CPTAC (Clinical pro-
teomic Tumor Analysis Consortium) assay portal [13] (assays.
cancer.gov), or Panorama [14] (panoramaweb.org/labkey/
project/home/begin.view), have recently emerged (see Note 2).
Coordinates for SRM assays can be retrieved from these data-
bases, paying attention to the considerations reported in
Note 3. Typically, the assays stored in such databases have
already been validated. However, to grant reliability of the
resulting SRM assay, care should be taken to ensure that the
detected peaks match reported retention time constraints
(upon retention time realignment to the local system setup, see
Note 4) and reported fragment ion relative intensities.
Shotgun proteomics experiments: SRM coordinates can be selected
from existing shotgun (LC-MS/MS) proteomics measure-
ments performed on the same proteome and filtered to a rea-
sonable (e.g., 1 %) protein false discovery rate (FDR). Precursor
ions from unique peptides with the highest MS signal or the
largest number of spectral counts should be prioritized.
Fragment ions can be selected from shotgun MS/MS spectra
and then tested by SRM analysis (see Note 5). We recommend
testing the ten most intense b- and y-fragment ions selected
from MS/MS spectra and depending on the combination of
instruments used, prioritizing y-ions might be advisable. Assays
generated from shotgun data are automatically validated by
the existence of the corresponding MS/MS spectra, when
Selected Reaction Monitoring to Measure Proteins of Interest in Complex Samples… 49

these have been assigned to peptide sequences using broadly

accepted confidence thresholds. However, care should be
taken in ensuring that the detected SRM peaks match the
retention times at which the MS/MS spectra were acquired,
upon RT realignment to the applied LC-SRM setup.
We will now exemplify SRM assay design using the open-
source software tool Skyline [5] (http://skyline.maccosslab.org),
assuming that no assays are available in public databases for the
proteins of interest. Here, we only provide a simplistic guide to the
use of Skyline and recommend consulting the tutorials and videos
on the Skyline website for more detailed instructions.
1. Import the amino acid sequences of the target proteins in
FASTA format into Skyline. In silico digestion is performed by
Skyline automatically according to the conditions specified
under Peptide and Transition Settings.
2. Under Peptide Settings choose trypsin KR/P as the digestion
protease, allow for no missed cleavages, and build a background
proteome database from the FASTA file for your species of
interest to check for peptide unicity in the proteome (proteo-
typicity). Choose carbamidomethylation of cysteines as a static
modification and keep the other default settings. If shotgun
datasets are available, build the corresponding spectral library
and include it in the document using the Library tab, to guide
the peptide and fragment selection.
3. Under Transition Settings, choose calculation of collision
energies according to your instrument model, optionally apply
the fragment prioritization criteria described in Note 1, using
the Filter and Instrument tabs, and choose to pick ten fragment
ions from library spectra, if available.
4. Skyline will generate a list of SRM assays based on the defined
criteria. Instrument-dependent collision energies (CE) are
calculated based on the peptide precursor m/z value. If
needed, CEs can be optimized in CE ramping experiments
(see Note 6).
5. Export the assay coordinates as a transition list which essen-
tially contains the Q1 and Q3 m/z values, the CE, and the
transition identifier (ID, i.e., peptide sequence, precursor
charge, and fragment ion information). Measure the transition
list by SRM in one or multiple runs, depending on its length
(see below), and use biological samples or synthetic peptides to
evaluate transition performance. It is advisable not to monitor
more than 100 transitions within the same unscheduled SRM
measurement, due to sensitivity considerations. Analyze the
data as described in Subheading 3.3.
50 Yuehan Feng and Paola Picotti

3.3 SRM Assay The coordinates of an SRM assay require validation [2]; that is, it
Validation should be confirmed that the chosen transitions and associated
and Refinement retention times detect the target peptide and not a false-positive
peak, based on an acceptable error rate. SRM assays from databases
or shotgun data are typically already validated by statistically fil-
tered LC-MS/MS spectra or other approaches, and can in theory
be directly applied to protein quantification with the precautions
described above. SRM assays from in silico design and computa-
tional prediction require validation by alternative approaches. One
option is the acquisition of MS/MS spectra for each target peptide
from the biological sample containing the protein of interest, their
assignment to peptide sequences via database search, and statistical
filtering to the chosen error rate. Another simple and cost-efficient
possibility is the (MS/)MS analysis of either crude synthetic peptide
analogues [15] of the targeted peptides (see Note 7) or tryptic
digests of recombinant proteins [16]. These options are particu-
larly useful when the targeted proteins are of low abundance and
thus difficult to detect in the available biological samples or if the
samples of interest are precious (e.g., from patients) and can only
be used in the final quantification step.
For the refinement of SRM assays, the SRM coordinates derived
from Subheading 3.2 are measured in SRM mode using the biologi-
cal samples of interest. The resulting peaks are inspected to select a
minimal set of suitable transitions and associated retention time (for
scheduled SRM measurements) that defines the final quantitative
assay for the protein. The three to five most intense transitions for
the best ionizing peptides are typically chosen in this step. Shouldered
peaks, likely resulting from multiple co-elutingpeptides sharing the
same transition(s), should be discarded. Retention times of SRM
peaks at the peak apex are annotated, if the aim is a scheduled SRM
experiment. Note that the extracted retention times are valid only
if the chromatographic settings remain the same in the quantifica-
tion step (see Note 4).
Once the assays are validated as described above, SRM assay
refinement can be performed with the tool Skyline.
1. First, open the Skyline (.sky) file from which the assays were
derived, to enable matching of the measured transitions to the
associated identifiers.
2. Import the raw SRM data as Results.
3. Inspect the peaks associated to each peptide transition group.
Important criteria to evaluate if a peptide was identified and
where in the chromatogram include (1) co-elution and shape
similarity of all transitions for that peptide; (2) retention time
of the selected peak matching the RT extracted during the
validation step; and (3) relative intensities of fragment ions
matching those extracted from the reference spectral library
 Selected Reaction Monitoring to Measure Proteins of Interest in Complex Samples… 51

(evaluated by the “dot product” score in Skyline) or measured

from synthetic peptides.
4. To facilitate the peak inspection process, use the Skyline View
option and display normalized peak areas (normalized to total
area). The relative intensities of different transitions will be
visualized by bar plots. In case you judge that a wrong peak has
been selected by the software, correction can be applied by re-
selecting the retention time frame (clicking and dragging the
cursor beneath the x(RT)-axis) where the correct peak is
located.
5. Transitions associated to shouldered peaks can be manually
discarded by removing the corresponding fragment ions from
the drop-down peptide/transition list of the Skyline interface.
Similarly, you can here select to retain only the most intense
three to five transitions per peptide by removing all the others
in the same way, since Skyline displays peak area rankings in
brackets (e.g., [1], [2], [3] …). Selection of most intense tran-
sition peaks may also be performed in bulk, using the
Edit > Refine > Advanced option. Export the transition list with
measured retention times for each peptide from Skyline as
above.

3.4 LC-SRM/ A QqQ mass spectrometer operated with a nano-electrospray ion

MS Setup source and interfaced with a nano-LC system is required for SRM
analysis, according to the setup described in Subheading 2.
1. Separate the peptide mixture using a linear gradient of 30 min,
from 5 to 35 % buffer B.
2. We recommended the use of an MS cycle time of 2–2.5 s, to
ensure acquisition of a sufficient number of data points per
peak, with the described chromatographic setup, while maxi-
mizing the available dwell time per transition. For unscheduled
SRM measurements, in one such LC-SRM the total number of
transitions measured should be below 100, to ensure a dwell
time of at least 20 ms per transition (see Note 8). For sched-
uled SRM measurements, we recommend a retention time
window of 3–5 min and measurement of a maximum number
of 1000–1200 SRM transitions.

3.5 Final Validated and refined SRM coordinates constitute the final SRM
Quantitative Analysis assay for a protein. Final SRM assays are directly applied to mea-
sure the target proteins using scheduled or unscheduled SRM
acquisition. The resulting data are manually inspected or processed
automatically using software tools such as the mProphet [6], to
evaluate detection and (relative) amount of the protein in each
sample. Manual data evaluation should include application of the
constraints mentioned above in Subheading 3.3 to minimize the
52 Yuehan Feng and Paola Picotti

likelihood of false positives. In addition, if heavy isotope-labeled

peptides are spiked into the sample as quantitation standards, elution
time and relative fragment ion intensities of the heavy and light
transitions should correspond. Tools for the automated analysis of
SRM data such as the mProphet apply most of the constraints
described above to SRM data and combine the resulting scores
into a statistical model, enabling also calculation of an error rate for
the SRM dataset. Manual inspection of such data should still be
performed when statistical significance estimates differ for the same
peptide measured in different samples to avoid missing data.
Once peptide detection is confirmed, SRM peaks are inte-
grated, the corresponding areas or heights are extracted, and the
quantitative data from multiple peptides and multiple transitions
are statistically evaluated using tools such as MSStats [17]. MSstats
(www.msstats.org) is an open-access R-based tool enabling the
application of linear mixed-effects models to the statistical analysis
of quantitative SRM data.
The software Skyline can be used for the semiautomated analysis
of quantitative SRM data. An automated peak-picking function is
implemented into Skyline, after which peak assignments are manu-
ally validated, as described below. Recent updates of the software
have enabled users to apply the statistically calibrated scoring
model of mProphet (see step-by-step procedure to include
mProphet in the Skyline-based data analysis in one of the Skyline
tutorials available online).
1. To semiautomatically evaluate quantitative SRM data with
Skyline, open first the .sky file from which the final validated
assays were derived and import the raw SRM data.
2. Skyline will automatically apply its default peak-picking algorithm
to all imported measurements.
3. Manually inspect SRM peaks, using the validated peaks from
Subheading 3.3 as reference to judge the correctness of the
peak-picking step. Apply the same procedure as described in
Subheading 3.3 to make corrections where needed.
4. Results can be exported at this stage in .csv format using the
“Export Report” function. Skyline offers the possibility to cus-
tomize the type of information that will be included in the
exported data sheet (e.g., peptide sequence, length, transition
area or height, retention time, and more—a full reference is
available at the end of the Live Reports tutorial on the Skyline
website). Data can be further processed using spreadsheet
applications such as Microsoft Excel and statistical tools such
as R. For example, a normalization factor can be applied that
accounts for the variability in the total peptide amount across
the sample set (see Note 9).
5. To perform further statistical analysis of the quantitative dataset,
install the MSstats tool using the Tools > Tool Store option in
 Selected Reaction Monitoring to Measure Proteins of Interest in Complex Samples… 53

Skyline. For extended processing with MSstats, choose the

“MSstats Input” option when exporting a report. The exported
file can be directly processed by MSstats, in an R programming
environment, following the user manual provided in www.
msstats.org. The MSstats tool includes also a total-peptide
amount normalization step (see Note 10).

4 Notes

1. When selecting peptides and fragment ions for SRM measure-

ments, the following considerations apply. Peptides with a
length below 6 and above 25 amino acids should be discarded,
due to specificity and MS detectability issues, respectively.
Avoiding methionine-containing peptides excludes the possi-
bility of quantification artifacts due to methionine oxidation.
Half-tryptic peptides and peptides embedding missed-cleavage
sites should be avoided, especially in experiments aiming at the
absolute quantification of the target protein. Focusing only on
doubly and triply charged peptide precursors for peptides of
average length, and only on doubly charged precursor ions if
peptides do not contain histidines, -KP-, or -RP- bonds, is
typically reasonable. Similarly, singly charged y-ions should be
preferred, due to their high likelihood of detection. Fragment
or precursor ions with m/z below 350–400 should be avoided,
as in this m/z region typical environmental contaminant ions
are detected. To further maximize specificity, fragment ions
with low m/z, for example below that of the precursor, could
be avoided. M/z values of a precursor and its fragment ion
used in an SRM transition should be at least 5 Da apart, to
avoid spurious signals from unfragmented precursors.
2. Currently, the SRMAtlas contains assays with high proteome
coverage from four different organisms (Mycobacterium tuber-
culosis, Saccharomyces cerevisiae, Mus musculus, and Homo
sapiens). The CPTAC database instead focuses specifically on
the collection of assays for human cancer-related proteins.
3. The SRM assays deposited in publicly accessible databases have
been generated using different types of instruments and assay
transferability may vary depending on the system used. For
example, the geometry of the collision cell may vary across MS
systems, affecting the conservation of peptide fragmentation
patterns. In addition, different equations to calculate peptide
collision energies are suggested by the different instrument
vendors (see MS or acquisition software user manual).
4. Peptide retention times depend on the specific chromato-
graphic setup used, which includes parameters such as gradient
coordinates, column dimension, and packing material or
length of the transport capillaries, and may change slightly
54 Yuehan Feng and Paola Picotti

over time with aging of the chromatographic column. Thus,

RTs should be updated when changing chromatographic setup
or upon fluctuating environmental conditions, e.g., when
reproducing SRM assays from Web-based databases, or
between the SRM assay validationphase and the final quantita-
tive analyses, when these are conducted several days apart. This
can be achieved using sets of synthetic peptides eluting over
the whole chromatographic space and spiked into the sample
(e.g., the iRT peptides [18]) and serving as normalization
anchors. An equation that linearly correlates “old” and “new”
retention times is typically extracted based on the anchor pep-
tides and can be applied to predict the new RTs of the target
peptides. Accurate retention times are also essential in sched-
uled SRM measurement, conducted typically with narrow RT
windows to increase the throughput (e.g., 3–5 min). Therefore,
even a slight deviation in RT could lead to a missing peak,
resulting in misinterpretation of the data.
5. Shotgun proteomics instruments with q2-fragmentation (e.g.,
Q-TOFs) or employing higher energy collisional dissociation
(HCD) [19] are preferable to guide selection of optimal frag-
ment ions for SRM, as fragmentation patterns typically resemble
those produced during SRM. However, also MS/MS spectra
from instruments such as linear ion traps are a reasonable starting
point to guide the selection of fragment ions for SRM, provided
that several intense fragment y-ions (the intensity of b-ions is less
conserved between QqQ and trapping instruments) are selected
from those spectra and tested using SRM.
6. The sensitivity of an SRM assay can be further increased by
optimizing specific peptide and MS-specific parameters, such
as, most commonly, the collision energy. Experimental proce-
dures for CE optimization, optionally using the Skyline plat-
form, have been described [5, 20]. Using optimized collision
energies instead of CEs calculated using equations provided by
instrument vendors results typically in a signal increase below
threefold [15]. We recommend performing CE optimization
only when the target proteins are difficult to detect.
7. Synthetic peptides (optionally heavy labeled and spiked into
the sample of interest) can also be measured in SRM mode to
confirm the validity of the chosen transitions, but this option
does not allow assignment of an error rate to the analysis.
8. The dwell time is the time the mass spectrometer spends on
measuring a given SRM transition. The sum of dwell times for
all the transitions measured is the cycle time of an SRM mea-
surement, which in turn determines the number of data points
collected along the chromatographic elution profile of a
peptide. For example 100 SRM transitions measured with a
dwell time of 20 ms each will result in a cycle time of about 2 s,
 Selected Reaction Monitoring to Measure Proteins of Interest in Complex Samples… 55

translating in the acquisition of a data point every 2 s for a

given transition peak. Longer dwell times result in a higher
signal-to-noise and thus sensitivity for the target analyte.
However, increasing the number of transitions leads to a lon-
ger cycle time and less data points collected along a chromato-
graphic peak, which can affect the quantitative precision.
9. Relative quantification results from SRM measurements will be
affected by variations in the total peptide amount injected for
the different samples measured. For example, let us assume that
sample A is two times more concentrated than sample B due to
sample processing issues and that the same volume is injected
into the LC-SRM system for the two peptide samples. Albeit
peptide P is present in the same amount in both samples, SRM
measurements will report a twofold higher amount for peptide
P in sample A. To address this issue, it is highly recommended
to perform a data normalization step based on the total peptide
amount in each sample (even if BCA assays are performed after
protein extraction). One can measure by SRM a set of well-
characterized (e.g., housekeeping) proteins whose abundance is
assumed to remain constant across the different conditions and
use their average abundance change as a normalization factor.
Similar estimations of the relative total peptide amount can be
extracted from absorbance measurements of the peptide mix-
tures at 280 nm before MS injection or from the total ion chro-
matogram (TIC) intensity of the samples measured in shotgun
LC-MS/MS mode.
10. The MSstats package performs normalization of the total
peptide amount loaded onto the LC-MS/MS system based
on the comparison of the median intensity of all target ana-
lytes (or a reference set of analytes) between different runs.
For samples where the majority of the measured targets
change abundance, this normalization approach may not be
suitable, and should be replaced by one of the methods men-
tioned in Note 9.

Acknowledgements

We thank Paul J. Boersema and Martin Soste (ETH Zurich) for

insightful discussions. We thank Brendan MacLean (University of
Washington) for critical reading of the manuscript. P.P. is sup-
ported by an EU-FP7 ERC Starting Grant (FP7-ERC-
StG-337965), by a FP7-Reintegration grant (FP7-PEOPLE-
2010-RG-277147), by a Professorship grant from the Swiss
National Science Foundation (grant PP00P3_133670), and by a
Promedica Stiftung (grant 2-70669-11). Y.F. is supported by an
ETH Research Grant (grant 4412-1).
56 Yuehan Feng and Paola Picotti

References
1. Aebersold R, Mann M (2003) Mass 11. Mohammed Y, Domanski D, Jackson AM et al
spectrometry-based proteomics. Nature 422: (2014) PeptidePicker: a scientific workflow
198–207 with web interface for selecting appropriate
2. Picotti P, Aebersold R (2012) Selected reaction peptides for targeted proteomics experiments.
monitoring-based proteomics: workflows, J Proteomics 106:151–161
potential, pitfalls and future directions. Nat 12. Deutsch EW, Lam H, Aebersold R (2008)
Methods 9:555–566 PeptideAtlas: a resource for target selection for
3. Soste M, Hrabakova R, Wanka S et al (2014) A emerging targeted proteomics workflows.
sentinel protein assay for simultaneously quan- EMBO Rep 9:429–434
tifying cellular processes. Nat Methods 13. Whiteaker JR, Halusa GN, Hoofnagle AN et al
11:1045–1048 (2014) CPTAC assay portal: a repository of
4. Kusebauch U, Deutsch EW, Campbell DS et al targeted proteomic assays. Nat Methods 11:
(2014) Using PeptideAtlas, SRMAtlas, and 703–704
PASSEL: comprehensive resources for discov- 14. Sharma V, Eckels J, Taylor GK et al (2014)
ery and targeted proteomics. Curr Protoc Panorama: a targeted proteomics knowledge
Bioinformatics 46:13.25.11–13.25.28 base. J Proteome Res 13:4505–4510
5. MacLean B, Tomazela DM, Shulman N et al 15. Picotti P, Rinner O, Stallmach R et al (2010)
(2010) Skyline: an open source document High-throughput generation of selected
editor for creating and analyzing targeted reaction-monitoring assays for proteins and
proteomics experiments. Bioinformatics 26: proteomes. Nat Methods 7:43–46
966–968 16. Stergachis AB, MacLean B, Lee K et al (2011)
6. Reiter L, Rinner O, Picotti P et al (2011) Rapid empirical discovery of optimal peptides
mProphet: automated data processing and sta- for targeted proteomics. Nat Methods
tistical validation for large-scale SRM experi- 8:1041–1043
ments. Nat Methods 8:430–435 17. Choi M, Chang CY, Clough T et al (2014)
7. Bereman MS, MacLean B, Tomazela DM et al MSstats: an R package for statistical analysis of
(2012) The development of selected reaction quantitative mass spectrometry-based proteomic
monitoring methods for targeted proteomics experiments. Bioinformatics 30:2524–2526
via empirical refinement. Proteomics 18. Escher C, Reiter L, MacLean B et al (2012)
12:1134–1141 Using iRT, a normalized retention time for
8. Mallick P, Schirle M, Chen SS et al (2007) more targeted measurement of peptides.
Computational prediction of proteotypic pep- Proteomics 12:1111–1121
tides for quantitative proteomics. Nat 19. de Graaf EL, Altelaar AF, van Breukelen B et al
Biotechnol 25:125–131 (2011) Improving SRM assay development: a
9. Fusaro VA, Mani DR, Mesirov JP et al (2009) global comparison between triple quadrupole,
Prediction of high-responding peptides for tar- ion trap, and higher energy CID peptide frag-
geted protein assays by mass spectrometry. Nat mentation spectra. J Proteome Res
Biotechnol 27:190–198 10:4334–4341
10. Qeli E, Omasits U, Goetze S et al (2014) 20. Maclean B, Tomazella DM, Abatiello SE et al
Improved prediction of peptide detectability for (2010) Effect of collision energy optimization
targeted proteomics using a rank-based algo- on the measurement of peptides by selected
rithm and organism-specific data. J Proteomics reaction monitoring (SRM) mass spectrome-
108:269–283 try. Anal Chem 82:10116–10124
 Chapter 5

Monitoring PPARG-Induced Changes in Glycolysis

by Selected Reaction Monitoring Mass Spectrometry
Andreas Hentschel and Robert Ahrends

Abstract
As cells develop and differentiate, they change in function and morphology, which often precede earlier
changes in signaling and metabolic control. Here we present a selected reaction monitoring (SRM) approach
which allows for the parallel quantification of metabolic regulators and their downstream targets.
In particular we explain and describe how to monitor abundance changes of glycolytic enzymes upon
PPARγ activation by using a label-free or a stable isotope-labeled standard peptide (SIS peptides) approach
applying triple-quadrupole mass spectrometry. We further outline how to fractionate the cell lysate into
cytosolic and nuclear fractions to enhance the sensitivity of the measurements and to investigate the
dynamic concentration changes in those compartments.

Key words Metabolic control, Adipocytes, Selected reaction monitoring, SIS peptides, Quantification,
Mass spectrometry, Proteomics

1 Introduction

In the past adipose tissue was just seen as a storage depot for free
fatty acids. Recent studies and models replaced this with the idea
that adipose tissue is a complex, essential, and highly active meta-
bolic and endocrine organ [1, 2]. Adipocytes also play a central
role in lipid and glucose metabolism and produce a large number
of hormones and cytokines, whose dysregulation is known to be
involved in metabolic syndrome, diabetes mellitus, and vascular
diseases. Therefore it affects the entire set of organ systems in our
body [3, 4].
The increasing rate of obesity in our society has led to intense
interest in understanding the mechanisms underlying the formation
of adipose tissue and its capacity to store fat [5–7]. The nuclear
receptor and lipid-binding protein PPARγ is a master regulator of
the formation and function of mature fat cells [8]. PPARγ is
expressed and activated during adipocyte differentiation, and
artificial induction of PPARγ in cells with adipogenic differentiation

Jörg Reinders (ed.), Proteomics in Systems Biology: Methods and Protocols, Methods in Molecular Biology, vol. 1394,
DOI 10.1007/978-1-4939-3341-9_5, © Springer Science+Business Media New York 2016

57
58 Andreas Hentschel and Robert Ahrends

potential can convert them into mature adipocytes [9]. In vitro

studies suggest that PPARγ is the ultimate regulator of adipogenesis
in a transcriptional cascade that also involves members of the C/EBP
transcription factor family and an interconnected feedback loop
system [10–14]. Additionally, PPARγ knockout mice fail to develop
adipose tissue [10, 15, 16]. Consistent with these findings, humans
with dominant-negative mutations in a single allele of PPARγ
(the gene encoding PPARγ) have partial lipodystrophy and insulin
resistance [17–19]. Adipose tissue also has a key role in directing
whole-body glucose homeostasis. The realization that a fatty acid
sensor like PPARγ might be an important regulator of glucose
metabolism arose from the discovery that the insulin-sensitizing
thiazolidinediones (TZDs) such as rosiglitazone are potent agonists
for PPARγ [20, 21] and therefore activators of the glucose trans-
porter GLUT4. In addition to glucose uptake regulation, PPARγ
regulates many glycolytic enzymes by binding to their transcrip-
tional promoters [22]. The activity of the pathway can be regulated
at different key steps to ensure that glucose consumption and
energy production match the needs of the cell. The key regulatory
enzyme in mammalian glycolysis is phosphofructokinase, the rate-
limiting catalyst.
The glycolytic pathway is extremely ancient in evolution and is
common to nearly all living organisms. This pathway can be broken
down into three stages: (1) the conversion of glucose into fructose
1,6-bisphosphate, (2) the cleavage of fructose 1,6-bisphosphate
into two three-carbon fragments, and (3) the generating of ATP
after the three-carbon fragments are oxidized to pyruvate [23].
Though the glycolytic pathway was one of the first to be discov-
ered, its importance in modern biology is still being elucidated. Its
metabolites and intermediates serve as entry points for many other
important pathways and dysregulation is involved in multiple dis-
eases. Because of the regulatory function of PPARγ in glucose
metabolism, it is of great interest to monitor the changes in glycolysis
upon its activation.
In this protocol we describe how to induce and analyze
abundance changes of PPARγ and its downstream targets with a
particular focus on glycolysis enzymes.
To estimate protein abundance of single proteins over time
and/or activation ranges, immunological methods such as the
enzyme-linked immunosorbent assay (ELISA) or Western blot
analysis can be carried out. But those approaches are not suited to
analyze entire pathways since they are expensive and difficult to
assess regarding their specificity and their linear dynamic range of
detection which is often very limited.
Mass spectrometry has significant advantages that overcome
the limitations of antibody-based assays. It allows direct detection
and quantification of peptides unique to a protein of interest.
Quantification can be improved by using SIS peptides, allowing
 Monitoring PPARG-Induced Changes in Glycolysis 59

Fig. 1 Selected reaction monitoring (SRM) total ion chromatogram of peptides derived from glycolysis proteins
out of an OP9 cytosolic protein extraction. The total ion chromatogram showing measurements of 50 proteins
(150 peptides and 900 transitions total) during an 80-min experiment on a single sample consisting of 1 μg of
total cytosolic protein digest. The inset panels display chromatograms of individual monitored peptides derived
from 1 to 8: L-lactate dehydrogenase A, tubulin beta 5 (control), glyceraldehyde-3-phosphate dehydrogenase,
fructose-bisphosphate aldolase A, pyruvate kinase, triosephosphate isomerase, phosphoglycerate mutase and
alpha-enolase

the absolute quantification of targeted proteins. Another advantage

of using mass spectrometry is the multiplexing capability, which
allows for hundreds of proteins to be measured in parallel (Fig. 1).
Further, this method is easy to automate, allowing the analysis of
many samples in a row without loading and detecting each sample
manually.
By using an LC/MS-based method in conjunction with a
triple-quadrupole mass spectrometer, a selected reaction monitor-
ing (SRM) experiment can be carried out. SRM mode works like a
double-mass filter, which drastically reduces noise and increases
selectivity.
To apply this approach, first, the proteins are digested into
short peptides (5–50 amino acids) with a hydrolytic enzyme such
as trypsin. After sample preparation the digest is subjected to nano
reverse-phase liquid chromatography and ionized by nano electro-
spray ionization (NSI).
The ionized peptides enter the mass spectrometer and the
desired precursor peptide ion is selected by mass-to-charge ratio at the
60 Andreas Hentschel and Robert Ahrends

first quadrupole according to the developed method. Other ions

generated in the ion source that have a different m/z ratio cannot
pass the quadrupole (Q1) at this time. The selected precursor ion
enters the second quadrupole (Q2) and is fragmented by collision-
induced dissociation (CID). The generated fragments are trans-
ferred to the third quadrupole and one specific fragment is selected
again and transferred to the detector where the abundance over
the peptide retention time is measured.
To obtain the best sensitivity usually fragment ions are selected
that have been found to give the best signal-to-noise ratio in previ-
ous experiments. The precursor ion and one of its corresponding
fragment ions are known as a transition. To precisely monitor a
peptide, generally two to three transitions are required. Furthermore
at least two to three unique peptides per protein are needed, mak-
ing it necessary to monitor six to nine transitions per protein. If
SIS peptides are used for validation and quantification up to 18
transitions should be monitored per protein.
There are also challenges of using SRM mass spectrometry on
triple-quadrupole instruments. Since the precision of mass mea-
surement is limited to a mass accuracy of ±0.3 Da and a resolution
of 7500 (at m/z of 508) the triple quadrupole is not among the
instruments with the highest mass accuracy and resolution. This
may lead to false positives when similar MS/MS fragmentation
patterns are detected simultaneously or at a different time point
during analysis. To solve this issue the use of SIS is highly recom-
mended. Another handicap is the limited number of total fragment
ions that can be detected in each sample. This can be overcome by
performing a scheduled analysis (each peptide, separate analysis
window) to enhance the limited number of fragment ions
significantly.
For validation and quantification of the chosen proteins SIS
peptides should be used. A suitable reference peptide has to be
selected based on the peptide sequence, retention time behavior,
ionization efficiency, and its fragmentation pattern. The chosen
peptide is then synthesized by solid-phase synthesis using light
amino acids and one isotopically coded “heavy” amino acid. The
result is a chemically identical peptide homolog that only differs in
mass (8–10 Da) and can therefore be easily distinguished by
SRM. SIS-peptides can be spiked in to the tryptic digest of the
sample at a known concentration. Since the reference peptides
elute at the same time and have the same fragmentation pattern as
the endogenous peptides, the concentration of endogenous pep-
tides can be directly measured as a ratio between the added isotopi-
cally coded SIS and the endogenous peptides (Fig. 2).
Here we describe a method and SRM assay to analyze and
quantify specific glycolytic proteins and their regulator PPARγ
after its activation. This method is also easily applicable to other
signaling or metabolic pathways where dynamics occur.
Fig. 2 Selected reaction monitoring (SRM) chromatogram of aldolase A and tubulin beta 5, monitoring the changes in abundance, after PPARγ activation. (a)
Quantitative analysis of aldolase A (upper panel) using a label-free approach for relative quantification, lower panel displays the same peptide in an absolute quan-
tification approach using internal isotope-coded standard peptides (SIS) after treating the OP9 cells with three different concentrations of rosiglitazone (1 nM,
Monitoring PPARG-Induced Changes in Glycolysis

112 nM, and 250 nM). (b) Transitions of the control protein tubulin beta 5 derived from same sample and analysis. The SIS peptides for absolute quantification were
added in an amount of 500 fmol each, whereby every sample contained 1 μg of total cytosolic protein digest
61
62 Andreas Hentschel and Robert Ahrends

2 Materials

All solutions have to be prepared with ultrapure water (prepared by

purifying deionized water to attain a sensitivity of 18 MΩ cm at
25 °C) and analytical grade reagents. Prepare and store all reagents
at 4 °C (unless indicated otherwise). All % values are given in v/v
if not otherwise indicated.

2.1 Cell Culture 1. Cells: OP9, a stromal cell line from mouse bone marrow, was
of OP9 Cells purchased from the Tokyo Metropolitan Institute of Medical
Science.
2. High-serum culture media: MEMα containing 20 % fetal
bovine serum (FBS) (Biochrome), 1 % penicillin/streptomycin
(PSG) (GIBCO).
3. Low-serum culture media: MEMα containing 10 % fetal bovine
serum (FBS) (Biochrome), 1 % penicillin/streptomycin (PSG)
(GIBCO).
4. Cell culture dishes: Cell culture flasks (T75) (Sarstedt), dispos-
able filter units, pore size 0.2 μm (Thermo) sterile serological
pipettes 5, 10, 25 ml (VWR).
5. Rosiglitazone stock solution: 1 mg/ml Rosiglitazone in
DMSO (Sigma), store at −80 °C.
6. Differentiation: Low-serum media containing 1 μM
rosiglitazone.

2.2 Cell Harvesting 1. Buffer: 1× Phosphate-buffered saline (PBS) pH 7.4 (GIBCO),

and Lysis warm up to 37 °C (water bath).
2. Trypsin: 0.05 % Trypsin-EDTA (GIBCO) with phenol-red
indicator.
3. Trypsin stopping solution: High-serum culture media.
4. Cell culture dishes: 15 ml Falcon tubes (Sarstedt), sterile sero-
logical pipettes 5, 10, 25 ml (VWR), Neubauer improved
counting chamber (Marienfeld).
5. Centrifuge: Eppendorf centrifuge 5804.

2.3 Protein 1. Buffer A: 10 mM HEPES pH 7.9 (Sigma, cell culture tested),

Extraction 1.5 mM MgCl2 (Sigma), 10 mM KCl (Sigma), 1× complete
protease inhibitor cocktail (Roche), 1 mM DTT, 0.1 % digitonin
(Sigma), please check for purity and protein contamination.
2. Buffer B: Same as buffer A but without digitonin.
3. Buffer C: 20 mM HEPES pH 7.9 (Sigma, cell culture tested),
1.5 mM MgCl2 (Sigma), 450 mM KCl (Sigma), 25 % v/v
glycerol, 0.2 EDTA (titriplex II Sigma).
4. BCA protein assay kit (Pierce).
 Monitoring PPARG-Induced Changes in Glycolysis 63

5. Additional chemicals: Acetone (Sigma, Chromasolv plus for

HPLC) ice-cold, 1 M NH4HPO3 (Sigma) pH 8, 6 M urea,
(Sigma), 0.5 M TCEP (Sigma, pH 7), 0.5 M Iodoacetamide
(Sigma), trypsin (Promega, sequencing grade), 99 % formic
acid (FA) (Biosolve ULC/MS grade).
6. Centrifuge: Eppendorf centrifuge 5810R.
7. Additional equipment: 25, 30 gauge needle + syringe (luer-
lock system), Thermomixer (Eppendorf) for 1.5 ml Eppendorf
tubes.

2.4 Peptide Cleanup 1. Buffer A: 0.05 % Trifluoroacetic acid (TFA) (pH 3.0); can also
Procedure use 5 % acetonitrile, 0.2 % formic acid, but can lose small
peptides.
2. Buffer B: 80 % Acetonitrile (ACN), 0.2 % formic acid (FA)
(pH below 4.0).
3. 100 % Methanol (Biosolve, ULC/MS).
4. Filters: SepPak C18 (100 mg, 1 cc), Waters #WAT023590,
1 mg capacity.

2.5 HPLC/MS 1. Standard peptides: Heavy arginine ([13C6] [15N4])- and lysine
Conditions ([13C6])-labeled peptides can be purchased from multiple ven-
and Materials dors (see Note 6). We use our own synthesized heavy peptides
and store them in 30 % ACN and 0.1 % TFA solution. Combine
all the heavy peptides needed for the experiment to a final con-
centration of 1.6 μM and store the peptide mix, as well as the
remainder of the concentrated individual peptide solutions,
aliquoted at −80 °C.
2. HPLC buffers: A: 0.1 % FA in HPLC-grade water, B: 84 %
ACN in 0.1 % FA.
3. Nano-HPLC instrument: Dionex UltiMate 3000 UHPLC
equipped with an Acclaim PepMap RSLC c18 reversed-phase
main column (Thermo) with 75 μm × 25 cm and 2 μm, 100 Å
particles, and an Acclaim PepMap 100 c18 reversed-phase pre
column (Thermo) with 100 μm × 2 cm and 5 μm, 100 Å
particles.
4. MS instrument: TSQ Vantage (Thermo Fisher Scientific)
triple-quadrupole mass spectrometer.

2.6 Data Analysis 1. Xcalibur 2.2.44, data analysis and instrument control.
Software 2. Skyline 2.5 software suite—available for free from the MacCoss
Lab website: https://brendanx-uw1.gs.washington.edu/lab-
key/project/home/software/Skyline/begin.view.
3. Microsoft Excel.
4. Origin 9.1.
64 Andreas Hentschel and Robert Ahrends

3 Methods

3.1 Cell Culture Each step must be carried out under sterile conditions.
1. Culture OP9 cells in high-serum culture media in cell culture
flasks as needed. Seed them out at 10 % of confluence (1E + 5 in
T15, 5E + 5 in T75, 1E + 6 in T175). Change the media every
2 days. After 4 days the cells are almost confluent (see Note 1).
2. For the experiment prepare 6 × T75 cell culture flasks with 10 %
confluence and incubate them for 3 days (should contain three
to four million cells after this period). Change the media on
third day to low-serum culture media and add rosiglitazone to
three of the flasks (end concentration 0.25 μM). Treat the
other ones just with DMSO with the same concentration.
Incubate all flasks for 48 h at 37 °C with 5 % CO2 in humidi-
fied atmosphere.

3.2 Harvesting Cells 1. Aspirate the media from the flasks.

2. Add 7.5 ml of PBS (37 °C) to the cells. Move the flasks a bit
to wash the remaining media and aspirate the PBS.
3. Add 7.5 ml of trypsin (RT) to the flasks and incubate them for
5 min in the incubator at 37 °C, 5 % CO2.
4. Check the flasks under the microscope to ensure that all the
cells are detached from the surface. If this is not the case, knock
the flask carefully against your hand and check again.
5. Add 7.5 ml of high-serum culture media to the flasks to stop
the trypsin reaction.
6. Use a 5 or 10 ml serological pipette to transfer everything of
one flask into one Falcon tube. Rinse the surface with the
suspension thoroughly to collect as much cells as possible.
7. Centrifuge the Falcon tubes at 200 × g for 5 min (RT) to pellet
the cells.
8. Aspirate the media and resuspend the pellet in 10 ml of PBS
(37 °C).
9. Use a small amount (10 μl) of the cell suspension to count the
cells with the Neubauer counting chamber.
10. Centrifuge again at 200 × g for 5 min (RT) to pellet the cells
and aspirate the PBS.
11. You can now store the cells by snap freezing them in liquid
nitrogen and store them at −80 °C or go on with the cell lysis.

3.3 Cell Lysis 1. Resuspend the pellets in 100 μl lysis buffer (buffer A) per million
and Fractionated cells; incubate on ice for 10 min, flicking the tube every minute.
Extraction Can add more lysis buffer if needed.
 Monitoring PPARG-Induced Changes in Glycolysis 65

2. Transfer the suspension into protein low-bind Eppendorf tube.

Use low-bind Eppendorf tube for every following step.
3. Dounce cells by passing the suspension through a 25-gauge
needle/syringe 5× and through a 30 g needle 3× (avoid bubbles
while douncing and make sure to check for breakage of mem-
branes by looking under a microscope; should only see nuclei and
membrane remnants, not whole cells anymore) (see Note 2).
4. Centrifuge suspension for 10 min at 2300 × g and 4 °C to pellet
nuclei. Collect the supernatant.
5. Resuspend pellet in 30 μl per million cells of buffer A (no digi-
tonin). Centrifuge suspension for 10 min at 2300 × g and 4 °C.
6. Add supernatant to that collected in the previous step. The
collected supernatants are the cytosol/membrane fraction.
The pellet from this step should contain an enriched prepara-
tion of nuclei. Call this pellet #1 (see Note 3).
7. Resuspend pellet #1 (the nuclei) in high-salt solution (buffer
C) to destroy the nuclei and strip most of the proteins from
DNA. Use 50 μl of buffer C for every one million OP9 cells.
8. Shake vigorously to resuspend the pellet. Incubate the suspen-
sion at 4 °C for 15 min, and then at 5-min intervals for another
15 min, flick to resuspend, and place back on ice. Under these
conditions most nuclear proteins will be extracted (see Note 4).
9. Centrifuge for 10 min at 4 °C at highest speed (20,000 × g).
10. Transfer the supernatant into clean Eppendorf tubes. This is
your nuclear extract and the pellet is your histone fraction
(see Note 5).
11. Do an acetone precipitation on the nuclear and cytosolic
fractions: add at least 3 volumes of ice-cold acetone (precooled
to −20 °C). Incubate proteins overnight at −20 °C.

3.4 Protein Digest 1. Centrifuge all fractions at max speed with the centrifuge
(20,000 × g) for 20 min at 4 °C. Carefully remove the superna-
tant leaving the protein pellet intact.
2. Allow the acetone to evaporate from the uncapped tube at
room temperature for 20 min. Do not over-dry pellet, or it
may not dissolve properly.
3. Resuspend the nuclear and cytosolic pellet in 6 M urea to
solubilize (10 μl per million cells at room temperature). Leave
for at least 1 h in urea, vortexing often. If the pellet is not dis-
solving you can use sonication in a water bath for 3–5 min.
4. Dilute samples to less than or equal to 2 M urea using 10 mM
NH4HCO3. Measure protein concentration with BCA kit. Make
a duplicate assay (standard and each sample). 1:5 dilution of each
sample is recommended to stay within the standard range.
66 Andreas Hentschel and Robert Ahrends

5. Add the internal isotope-labeled standard peptide mix (final

concentration should be 50–500 fmol/1 μg total proteins)
(see Note 6).
6. Add TCEP (final concentration 10 mM). Shake for 30 min at
37 °C (Thermomixer).
7. Cool sample to room temperature. Add 0.5 M iodoacetamide
(30 μl per 1 ml) and shake for 30 min in the dark at room tem-
perature (Thermomixer with aluminum foil). Iodoacetamide
has to be freshly made before use.
8. Add trypsin in a ratio of trypsin to protein (1:50).
9. Incubate shaking for over 12 h at 37 °C (Thermomixer).
10. After digestion stop the reaction by adding about 10 μl of 99 %
FA (total volume 500 μl) to lower the pH to 3 or lower. Check
with pH indicator paper.

3.5 Peptide Cleanup 1. Wash cartridges (SepPak C18) with 3× volumes (1 volume is
Procedure equivalent of 1 ml) of methanol (100 %), fast flow (with vac-
uum applied). This gets rid of contaminants in the cartridge.
Make sure not to run the filter dry. Stop when meniscus is just
above it.
2. Equilibrate with 3× volumes of buffer A, fast flow (with vac-
uum applied).
3. Load sample (pH should be <3), slow flow, best by gravity.
4. Wash and desalt with 3× volumes of buffer A, fast flow (with
vacuum applied).
5. Elute peptides with 1 ml buffer B, slow flow (gravity). Elute the
remaining fluid by vacuum past filter (after gravity elution).
6. Vacuum centrifuge sample to dryness and resuspend in 2 %
acetonitrile and 0.1 % FA (want a concentration of 1 μg/μl as
stock). This can take up to 5 h, depending on the vacuum
centrifuge. Best to use a high-performance solvent-resistant
vacuum centrifuge connected to a vacuum pump of sufficient
power.

3.6 Setting Before starting with the measurements you have to set up the right
Up the SRM Method method containing the proteins and peptides of interest including
in Skyline the transitions for the SIS peptides.
1. First, set up the Skyline document with the following
preferences:
Peptide settings
(a) Digestion: Trypsin [KR|P], missed cleavages 0
(b) Filter: Min length 8, max length 25, exclude N-terminal
AAs 25, you can also exclude all peptides that contain specific
AAs like Cys or Met, as needed
(c) Modifications: Carbamidomethyl (C)
 Monitoring PPARG-Induced Changes in Glycolysis 67

Transition settings
(a) Filter: Precursor charges 2 and 3, ion charges 1, ion types y
(b) Product ions: 6 ions
2. After those preparations are done, load a spectral library con-
taining the proteome of the species of interest (see Note 7).
3. Copy the amino acid sequences of each protein from Table 1
and paste it directly into the new Skyline document. The pro-
gram will automatically rank the best transitions for each peptide
according to the loaded spectral library (see Note 8).
4. Export the method (File/Export/Transition list) and choose
multiple methods. For max concurrent transitions choose 100
or 120 and method type: standard.
5. You can now import all the method files into the TSQ Vantage
program to search for all the peptides with their transitions and
get the right retention times.

3.7 Selected 1. Set up the following program for the separation/quantitation

Reaction Monitoring of peptides:
Mass Spectrometry
Separation: Linear gradient from 3 to 35 % ACN over 60 min
with a flow rate of 350 nl/min for separation and a 20 μl/
min flow rate for sample loading.
Polarity: Positive (tune file).
Resolution for Q1 and Q3: 0.7u FWHM (method file).
Emitter voltage: 1200–1500 V (tune file).
Temperature of the transfer capillary: 270 °C (tune file).
Declustering voltage: 10.
2. Unscheduled SRM: To obtain the right retention times for
scheduled SRM you first have to run an unscheduled method
with the internal standard peptides. Retention times for spe-
cific transitions may vary slightly from run to run depending
on the composition of the sample and the performance of the
instrument.
3. Scheduled SRM: Maximum time window 5 min, cycle time
1 s, average dwell time 20 ms.
4. Inject a volume of 5–15 μl with a total peptide amount of 1 μg
of the sample and 50–500 fmol of total amount of your SIS
peptides and start the measurement. Repeat each analysis at
least twice to account for run-to-run variability of the system.
Be sure that at least one blank is in between each sample
measurement.
5. To verify proper operating performance of the system and to
determine if the LC/MS needs cleaning or calibration,
perform SRM on a standard peptide mix at regular intervals
between the sample measurements and at least once per day.
Table 1
68

List of proteins, peptides, and transitions for monitoring PPARγ-induced glycolytic changes. The transition list for PPARγ and other nuclear proteins
can be found in Ahrends et al. (2014, Science, DOI: 10.1126/science.1252079)

Q8VDL4 ADP-dependent Adpgk LGPAPVPVGPLSPESR 786.9408++ 791.9450++ y12 1234.68 1244.69 1

glucokinase y10 1038.56 1048.57 1
y8 842.44 852.44 1
VAGTQACATETIDTNR 854.4020++ 859.4061++ y10 1180.53 1190.53 1
y9 1020.50 1030.50 1
y8 949.46 959.47 1
P05064 Fructose-bisphosphate Aldoa GILAADESTGSIAK 666.8539++ 670.8610++ y11 1049.51 1057.53 1
aldolase A y10 978.47 986.49 1
y7 663.37 671.38 1
QLLLTADDR 522.7878++ 527.7920++ y7 803.43 813.43 1
y6 690.34 700.35 1
y5 577.26 587.27 1
Andreas Hentschel and Robert Ahrends

ADDGRPFPQVIK 671.8593++ 675.8664++ y8 984.60 992.61 1

y7 828.50 836.51 1
y5 584.38 592.39 1
P17182 Alpha-enolase Eno1 TIAPALVSK 450.2817++ 454.2888++ y7 685.42 693.44 1
y6 614.39 622.40 1
y5 517.33 525.35 1
YITPDQLADLYK 720.3745++ 724.3816++ y10 1163.59 1171.61 1
y9 1062.55 1070.56 1
y5 609.32 617.34 1
SCNCLLLK 504.2543++ 508.2614++ y6 760.44 768.45 1
y5 646.40 654.41 1
y4 486.36 494.38 1
P21550 Beta-enolase Eno3 SGETEDTFIADLVVGLCTGQIK 1177.0832++ 1181.0903++ y13 1373.75 1381.76 1
y12 1302.71 1310.72 1
y11 1187.68 1195.70 1
AAVPSGASTGIYEALELR 902.9756++ 907.9797++ y10 1164.63 1174.63 1
y9 1063.58 1073.59 1
y8 1006.56 1016.57 1
P16858 Glyceraldehyde-3- Gapdh IVSNASCTTNCLAPLAK 910.4557++ 914.4628++ y12 1335.64 1343.65 1
phosphate dehydrogenase y10 1088.58 1096.59 1
y5 499.32 507.34 1
GAAQNIIPASTGAAK 685.3753++ 689.3824++ y11 1042.59 1050.60 1
y9 815.46 823.48 1
y8 702.38 710.39 1
LISWYDNEYGYSNR 890.4023++ 895.4064++ y9 1117.45 1127.46 1
y6 759.34 769.35 1
y5 596.28 606.29 1
P06745 Glucose-6-phosphate Gpi ELFEADPER 553.2617++ 558.2658++ y6 716.32 726.32 1
isomerase y5 587.28 597.28 1
y3 401.21 411.22 1
TFTTQETITNAETAK 828.4098++ 832.4169++ y10 1077.54 1085.55 1
y9 948.50 956.51 1
y7 734.37 742.38 1
HFVALSTNTAK 594.8222++ 598.8293++ y10 1051.58 1059.59 1
y9 904.51 912.52 1
y8 805.44 813.45 1
Q9WUA3 6-Phosphofructokinase Pfkp LGITNLCVIGGDGSLTGANLFR 1124.5914++ 1129.5955++ y14 1377.71 1387.72 1
type C y13 1264.63 1274.63 1
y10 1035.56 1045.56 1
VTILGHVQR 511.8089++ 516.8130++ y6 709.41 719.41 1
y5 596.33 606.33 1
y4 539.30 549.31 1
EIGWADVGGWTGQGGSILGTK 1045.0211++ 1049.0282++ y14 1318.67 1326.68 1
y11 1018.55 1026.56 1
y8 732.43 740.43 1
P52480 Pyruvate kinase Pkm NTGIICTIGPASR 680.3561++ 685.3602++ y9 974.51 984.51 1
y8 861.42 871.43 1
y7 701.39 711.40 1
GIFPVLCK 467.2650++ 471.2721++ y6 763.42 771.43 1
Monitoring PPARG-Induced Changes in Glycolysis

y5 646.35 624.36 1
y4 519.30 527.31 1
DAVLNAWAEDVDLR 793.8941++ 798.8982++ y10 1188.56 1198.57 1
y9 1074.52 1084.52 1
69

y8 1003.48 1013.49 1
(continued)
Table 1
70

(continued)

P09411 Phosphoglycerate kinase 1 Pgk1 YSLEPVAAELK 610.3321++ 614.3392++ y8 856.48 864.49 1

y7 727.43 735.44 1
y6 630.38 638.39 1
ITLPVDFVTADK 659.8663++ 663.8734++ y9 991.51 999.52 1
y7 795.39 803.40 1
y6 680.36 688.37 1
GCITIIGGGDTATCCAK 877.8971++ 881.9042++ y12 1210.52 1218.53 1
y11 1097.44 1105.44 1
y5 639.26 647.27 1
P06151 L-Lactate dehydrogenase Ldha LVIITAGAR 457.2951++ 462.2992++ y7 701.43 711.43 1
y6 588.35 598.35 1
y5 475.26 485.27 1
FIIPNIVK 472.3024++ 476.3095++ y7 796.53 804.54 1
y6 683.45 691.45 1
Andreas Hentschel and Robert Ahrends

y5 570.36 578.37 1
VTLTPEEEAR 572.7959++ 577.8000++ y8 944.47 954.47 1
y7 831.38 841.39 1
y6 730.34 740.34 1
P17751 Triosephosphate isomerase Tpi1 IAVAAQNCYK 569.2897++ 573.2968++ y8 953.45 961.46 1
y7 854.38 862.39 1
y6 783.35 791.35 1
DLGATWVVLGHSER 770.3993++ 775.4035++ y8 896.49 906.50 1
y7 797.43 807.43 1
y5 585.27 595.28 1
VVLAYEPVWAIGTGK 801.9481++ 805.9552++ y9 928.53 936.53 1
y7 732.40 740.41 1
y6 546.32 554.33 1
O08528 Hexokinase-2 Hk2 LPLGFTFSFPCHQTK 890.4480++ 894.4551++ y8 1004.46 1012.47 1
y6 770.36 778.37 1
y5 673.31 681.32 1
NILIDFTK 482.2791++ 486.2862++ y6 736.42 744.43 1
y5 623.34 631.35 1
y4 510.26 518.27 1
P17710 Hexokinase-1 Hk1 LSDEILIDILTR 700.9034++ 705.9075++ y7 843.53 853.53 1
y6 730.45 740.45 1
y5 617.36 627.37 1
GDFIALDLGGSSFR 727.8673++ 732.8715++ y10 1022.53 1032.53 1
y9 951.49 961.49 1
y6 610.29 620.30 1
SANLVAATLGAILNR 742.4332++ 747.4373++ y11 1098.66 1108.67 1
y10 999.59 1009.60 1
y6 643.39 653.39 1
Q9DBJ1 Phosphoglycerate mutase 1 Pgam1 VLIAAHGNSLR 575.8382++ 580.8423++ y8 825.43 835.44 1
y7 754.3955 764.40 1
y6 683.35 693.36 1
YADLTEDQLPSCESLK 934.9328++ 938.9399++ y10 1176.56 1184.57 1
y8 933.47 941.48 1
y7 820.39 828.40 1
ALPFWNEEIVPQIK 842.4589++ 846.4660++ y9 1069.59 1077.60 1
y5 584.38 592.39 1
y4 485.31 493.32 1
Q9QXG4 Acetyl-coenzyme A Acss2 VAFYWEGNEPGETTK 864.3992++ 868.4063++ y10 1061.47 1069.48 1
synthetase y9 932.43 940.44 1
y6 632.32 640.33 1
AELGMNDSPSQSPPVK 828.8985++ 832.9056++ y11 1155.56 1163.57 1
y9 926.49 934.50 1
y8 839.46 847.47 1
IGPIATPDYIQNAPGLPK 933.0120++ 937.0191++ y12 1312.69 1320.70 1
y7 696.40 704.41 1
y6 582.36 590.37 1
P99024 Tubulin beta-5 chain Tubb5 ISVYYNEATGGK 651.3222++ 655.3293++ y9 1002.45 1010.46 1
y8 839.39 847.40 1
y7 676.33 684.34 1
Monitoring PPARG-Induced Changes in Glycolysis

YLTVAAVFR 520.3004++ 525.3045++ y7 763.45 773.45 1

y6 662.40 672.40 1
y5 563.33 573.33 1
NSSYFVEWIPNNVK 848.9201++ 852.9272++ y8 999.53 1007.54 1
71

y6 684.40 692.41 1
y5 571.32 579.33 1
72 Andreas Hentschel and Robert Ahrends

3.8 Data Analysis 1. Import your raw files into Skyline.

2. Review the chromatograms of each peptide and check for
selection of the correct peaks. After your unscheduled/sched-
uled SRM analysis and data refinement your dataset should
contain at least three unique peptides for each protein with
three transitions each.
3. Use the export function of Skyline to export a report containing
transition results to create a .csv file containing the following
data: PeptideSequence, ProteinName, ReplicateName, Peptide
RetentionTime, and Total Area (for label free) or Ratio
ToStandard (for SIS).
4. To compare the different samples (OP9 control with DMSO
against PPARγ activated cells) and calculate the fold change of
each peptide do the following:
(a) Label free: Divide every total area by the control (DMSO)
to obtain the fold change over the treatment experiment.
(b) SIS: The ratio of the control sample is set to one and all
other ratios to standard values are divided by the ratio of
the control.
(c) If the exact amount of the injected SIS peptide is known
also the absolute amount of the endogenous protein can
be estimated by the peptide-to-standard ratio.

4 Notes

1. Do not allow the cells to reach full confluence until right before
the experiment. It is best to split the cells every 2 days to keep
them between 10 and 70 % confluence. OP9 cells are prone to
spontaneous differentiation if they grow too confluent.
2. At this point you can stain with trypan blue. Pipette approxi-
mately 2–5 μl from your sample onto an object slide, add 1–2 μl
of the trypan blue solution, and mix gently with the pipette.
Check under the microscope for the destroyed cells. If the
mechanical lysis was successful, you should only see blue stained
cell remnants. Everything that is not stained was not lysed.
3. This is a time-critical step because you can lose nuclear proteins
by diffusion if you wait too long to take the supernatant.
4. Do not exceed 500 mM KCl or you will extract histones.
5. For the further steps the histone fraction is not needed.
6. SIS peptides that are used in this protocol were synthesized
in-house without any modifications such as carbamidomethyl-
ation. SIS peptides without those modifications have to be
added to the digest before adding TCEP and IAA treatment.
If it is not possible to synthesize the peptides on your own
 Monitoring PPARG-Induced Changes in Glycolysis 73

there it is possible to buy them from different vendors (Thermo

Scientific, JPT, and Sigma Aldrich). SIS peptides that are
already carbamidomethylated can be added to the digest after
the sample has been treated with IAA and TCEP.
7. Spectral libraries for SRM method design and for data analysis
can be either directly added to a Skyline document or a custom
library can be built within Skyline. For the latter, data-dependent
acquisition (shotgun) as well as SRM-triggered MS 2 measure-
ments searched with one of the common search engines can be
used. Here we used our own spectral libraries obtained from
data-dependent acquisition measurements.
8. Even if Skyline chooses the transitions according to the transi-
tion settings and the spectral library you should check every
peptide manually for the right transitions. At this optimization
stage using a triple-quadrupole instrument, it is recommended
to try five to six transitions for each peptide ranging from the
m/z value of around 500–1200 for the unscheduled measure-
ments. The lower the m/z value gets, the more unspecific the
transition is. After the unscheduled measurements you can refine
the transitions by choosing the three most intense transitions
per peptide for the actual experiments in order to maximize the
dwell time and to increase the sensitivity. Since the heavy-labeled
standard peptides are very expensive, they should only be
ordered once the best precursor peptides and transitions have
been selected based on the sample of interest. If you are able to
produce your own standard peptides measuring the precursors
of them directly with an unscheduled method to get the right
retention times is possible. If the retention time or fragmenta-
tion pattern differs between the heavy and the light peptide, you
are analyzing a false-positive signal. In such case, the design of
the peptide transitions should be started from scratch.

References
1. Kershaw EE, Flier JS (2004) Adipose tissue as 5. Van Gaal LF, Mertens IL, De Block CE (2006)
an endocrine organ. J Clin Endocrinol Metab Mechanisms linking obesity with cardiovascu-
89(6):2548–2556. doi:10.1210/jc.2004-0395 lar disease. Nature 444(7121):875–880.
2. Rosen ED, Spiegelman BM (2006) Adipocytes doi:10.1038/nature05487
as regulators of energy balance and glucose 6. Kahn SE, Hull RL, Utzschneider KM (2006)
homeostasis. Nature 444(7121):847–853. Mechanisms linking obesity to insulin resistance
doi:10.1038/nature05483 and type 2 diabetes. Nature 444(7121):840–
3. Tilg H, Moschen AR (2008) Inflammatory 846. doi:10.1038/nature05482
mechanisms in the regulation of insulin resis- 7. Olshansky SJ, Passaro DJ, Hershow RC et al
tance. Mol Med 14(3–4):222–231. (2005) A potential decline in life expectancy in
doi:10.2119/2007-00119.Tilg the United States in the 21st century. N Engl
4. Hajer GR, van Haeften TW, Visseren FL J Med 352(11):1138–1145. doi:10.1056/
(2008) Adipose tissue dysfunction in obesity, NEJMsr043743
diabetes, and vascular diseases. Eur Heart 8. Rosen ED, Walkey CJ, Puigserver P et al (2000)
J 29(24):2959–2971. doi:10.1093/eurheartj/ Transcriptional regulation of adipogenesis.
ehn387 Genes Dev 14(11):1293–1307
74 Andreas Hentschel and Robert Ahrends

9. Tontonoz P, Hu E, Spiegelman BM (1994) adipocyte hypertrophy and insulin resistance.

Stimulation of adipogenesis in fibroblasts by Mol Cell 4(4):597–609
PPAR gamma 2, a lipid-activated transcription 17. Agarwal AK, Garg A (2002) A novel heterozy-
factor. Cell 79(7):1147–1156 gous mutation in peroxisome proliferator-activated
10. Rosen ED, Sarraf P, Troy AE et al (1999) receptor-gamma gene in a patient with familial
PPAR gamma is required for the differentiation partial lipodystrophy. J Clin Endocrinol Metab
of adipose tissue in vivo and in vitro. Mol Cell 87(1):408–411. doi:10.1210/jcem.87.1.8290
4(4):611–617 18. Hegele RA, Cao H, Frankowski C et al (2002)
11. Wu Z, Rosen ED, Brun R et al (1999) Cross- PPARG F388L, a transactivation-deficient
regulation of C/EBP alpha and PPAR gamma mutant, in familial partial lipodystrophy.
controls the transcriptional pathway of adipogene- Diabetes 51(12):3586–3590
sis and insulin sensitivity. Mol Cell 3(2):151–158 19. Savage DB, Tan GD, Acerini CL et al (2003)
12. Rosen ED, Hsu CH, Wang X et al (2002) C/ Human metabolic syndrome resulting from
EBPalpha induces adipogenesis through dominant-negative mutations in the nuclear
PPARgamma: a unified pathway. Genes Dev receptor peroxisome proliferator-activated
16(1):22–26. doi:10.1101/gad.948702 receptor-gamma. Diabetes 52(4):910–917
13. Park BO, Ahrends R, Teruel MN (2012) 20. Forman BM, Tontonoz P, Chen J et al (1995)
Consecutive positive feedback loops create a 15-deoxy-delta 12, 14-prostaglandin J2 is a
bistable switch that controls preadipocyte- ligand for the adipocyte determination factor
to- adipocyte conversion. Cell Rep PPAR gamma. Cell 83(5):803–812
2(4):976–990. doi:10.1016/j.celrep.2012. 21. Lehmann JM, Moore LB, Smith-Oliver TA
08.038 et al (1995) An antidiabetic thiazolidinedione
14. Ahrends R, Ota A, Kovary KM et al (2014) is a high affinity ligand for peroxisome
Controlling low rates of cell differentiation proliferator-activated receptor gamma (PPAR
through noise and ultrahigh feedback. Science gamma). J Biol Chem 270(22):12953–12956
344(6190):1384–1389. doi:10.1126/science. 22. Mikkelsen TS, Xu Z, Zhang X et al (2010)
1252079 Comparative epigenomic analysis of murine
15. Barak Y, Nelson MC, Ong ES et al (1999) and human adipogenesis. Cell 143(1):156–
PPAR gamma is required for placental, cardiac, 169. doi:10.1016/j.cell.2010.09.006
and adipose tissue development. Mol Cell 23. Masters C, Reid S, Don M (1987) Glycolysis—
4(4):585–595 new concepts in an old pathway. Mol Cell
16. Kubota N, Terauchi Y, Miki H et al (1999) Biochem 76(1):3–14. doi:10.1007/
PPAR gamma mediates high-fat diet-induced BF00219393
 Chapter 6

A Targeted MRM Approach for Tempo-Spatial

Proteomics Analyses
Annie Moradian, Tanya R. Porras-Yakushi, Michael J. Sweredoski,
and Sonja Hess

Abstract
When deciding to perform a quantitative proteomics analysis, selectivity, sensitivity, and reproducibility are
important criteria to consider. The use of multiple reaction monitoring (MRM) has emerged as a powerful
proteomics technique in that regard since it avoids many of the problems typically observed in discovery-
based analyses. A prerequisite for such a targeted approach is that the protein targets are known, either as
a result of previous global proteomics experiments or because a specific hypothesis is to be tested. When
guidelines that have been established in the pharmaceutical industry many decades ago are taken into
account, setting up an MRM assay is relatively straightforward. Typically, proteotypic peptides with favor-
able mass spectrometric properties are synthesized with a heavy isotope for each protein that is to be moni-
tored. Retention times and calibration curves are determined using triple-quadrupole mass spectrometers.
The use of iRT peptide standards is both recommended and fully integrated into the bioinformatics pipe-
line. Digested biological samples are mixed with the heavy and iRT standards and quantified. Here we
present a generic protocol for the development of an MRM assay.

Key words MRM, Quadrupole mass spectrometry, Quantitation

1 Introduction

The ultimate goal of a quantitative proteomics experiment is to

identify and quantify protein changes in time and/or space.
Particularly, when monitoring the changes of specific proteins over
many time and/or space points, the use of multiple reaction moni-
toring (MRM) has emerged as a powerful technique because of its
high quantitative precision and sensitivity resulting in low detec-
tion limits [1–5]. A prerequisite for such a targeted approach is
that the protein targets are known, either as a result of previous
global proteomics experiments or because a specific hypothesis is
to be tested.
Typically, proteotypic peptides with favorable mass spectro-
metric properties are selected for each protein and synthesized

Jörg Reinders (ed.), Proteomics in Systems Biology: Methods and Protocols, Methods in Molecular Biology, vol. 1394,
DOI 10.1007/978-1-4939-3341-9_6, © Springer Science+Business Media New York 2016

75
76 Annie Moradian et al.

Fig. 1 The two operational modes of a QTRAP instrument. (a) QTRAP mode. Selected ions of separated
peptides pass through Q1 and are fragmented in Q2, all fragment ions pass through Q3, and a full MS/MS
spectrum is recorded. This mode is particularly useful in the early stages of assay development when identities
of peptides need to be confirmed. (b) QQQ mode. Selected ions of separated peptides pass through Q1 and are
fragmented in Q2, and all selected fragment ions pass through Q3. Co-eluting transition ion pairs are recorded.
This mode is particularly useful in maximizing sensitivity of known peptides

with a heavy isotope. Retention times and calibration curves are

determined using predominantly triple-quadrupole mass
spectrometers (Fig. 1), although quadrupole time-of-flight and
Orbitrap mass spectrometers can also be used. We preferentially
use the QTRAP in the trap mode (Fig. 1a) during the initial stages
of assay development, to establish retention time of the peptides,
select the best suited three to five fragment ions, and optimize
scheduling. Once this is accomplished, scheduled analysis is per-
formed in the more sensitive QQQ mode (Fig. 1b). To determine
realistic limits of detection (LOD) and quantitation (LOQ), it is
recommended to analyze the heavy standards in a complex
biological background. This prevents unwanted absorption of the
peptides and helps to identify interferences of the peptides with the
background at an early stage of the method development. Thus,
digested biological samples are mixed with the heavy and retention
time iRT standards and quantified over a dynamic range [6].
 Targeted Tempo-Spatial Proteomics 77

3.0

VEA
2.5

YIL
2.0

AGG
Intensity (10Ÿ6)

1.5

TPV...T...
TPV...S...
1.0

DAV
GDL

TGF

GTF
0.5
LGG

FLL
0.0
10 20 30 40 50
Retention Time

Fig. 2 Elution profile of 11 iRT peptides in a 45-min gradient. Peptides are LGGNETQVR (LGG), AGGSSEPVTGLADK
(AGG), VEATFGVDESANK (VEA), YILAGVESNK (YIL), TPVISGGPYYER (TPV…S...), TPVITGAPYYER (TPV…T...), GDLD
AASYYAPVR (GLD), DAVTPADFSEWSK (DAV), TGFIIDPGGVIR (TGF), GTFIIDPAAIVR (GTF), FLLQFGAQGSPLFK (FLL)

A typical elution profile of the iRT peptides is shown in Fig. 2.

The use of iRT peptide standards is recommended to monitor and
ensure consistent chromatographic performance throughout the
assay [6]. Additionally, the iRT peptide standards can be fully inte-
grated into the bioinformatics pipeline using Skyline [7]. Figure 3
shows an example of a calibration curve of standard peptides and
their transition ion responses from 30 to 500,000 attomol, mea-
sured in a complex background, e.g., HeLa lysate if a HeLa cell
culture is to be investigated. The calibration curves of the heavy
standards can be used to determine targeted peptide amounts. As
shown in Fig. 3, a response of 100,000 AU for a targeted peptide
could be associated with an amount of 550 amol, in comparison to
the calibration curve of the heavy standard shown (see Note 1).
The most accurate quantitative results are usually achieved when
the heavy standards are in the same range as the expected light
samples. For this, it might be necessary to do a preliminary analysis
first, followed by a full analysis with heavy peptides spiked in
according to the preliminary results. Here, we provide a generic
guideline for the development of an MRM assay.
78 Annie Moradian et al.

Fig. 3 Example of a calibration curve for a standard peptide covering a dynamic range from 30 to 500,000
attomol. This curve can be used to quantify unknown samples. If an unknown sample has a response of
100,000 AU, the sample contains 550 attomol of that particular peptide

2 Materials

2.1 Chemicals 1. Solvent A: 0.2 % Formic acid. Add 2 mL of formic acid (for
mass spectrometry; 98 %) to 998 mL of water (HPLC gradient
2.1.1 LC and LC
grade quality).
Samples
2. Solvent B: Acetonitrile containing 0.2 % formic acid. Add 2
mL of formic acid (for mass spectrometry; 98 %) to 998 mL of
acetonitrile (HPLC gradient grade quality).
3. Synthetic heavy standard peptides (either prepared in-house or
commercially).
4. Synthetic iRT peptides (either prepared in-house or
commercially).
5. Standard solvent (30 % acetonitrile, 70 % of 0.2 % formic acid).
Add 30 mL of acetonitrile to a 69.8 mL of water and 0.2 mL
of formic acid.
6. All reagents should be of the highest purity available.

2.1.2 FASP 1. Lysis buffer. Add 5 g sodium dodecyl sulfate (SDS), 1.58 g
(100 mM) Tris/HCl (pH 7.6; adjust with HCl as necessary),
1.54 g (100 mM) D,L-dithiothreitol (DTT), supplemented
with 1× protease inhibitor cocktail (complete, EDTA free,
Roche) and 17.42 mg (1 mM) phenylmethanesulfonyl fluoride
(PMSF) brought up to 100 mL with ddH2O.
 Targeted Tempo-Spatial Proteomics 79

2. Urea A (UA) buffer. Add 48.05 g (8 M) urea and 1.58 g (100

mM) Tris/HCl (pH 8.5, adjust with HCl as necessary) to a
total volume of 100 mL of ddH2O; prepare fresh on the day of
the digestion.
3. Urea B (UB) buffer. Add 48.05 g (8 M) urea and 1.58 g (100
mM) Tris/HCl (pH 8.0, adjust with HCl as necessary) to a
total volume of 100 mL of ddH2O; prepare fresh on the day of
the digestion.
4. 27 mM Tris(2-carboxyethyl)phosphine hydrochloride (TCEP)
solution. Add 14.33 g (500 mM) TCEP to 1.58 g (100 mM)
Tris/HCl (pH 8.5, adjust with HCl as necessary) in a total
volume of 100 mL of ddH2O and store at −20 °C. Right before
reduction step, thaw aliquot and dilute 5.4 μL of 500 mM
TCEP by adding 94.6 μL of UA buffer.
5. Alkylating solution. Prepare fresh using 0.925 mg (50 mM)
iodoacetamide (IAA) in 1 mL of UA buffer.
6. 50 mM Ammonium bicarbonate (NH4HCO3) solution. Add
0.40 g of NH4HCO3 to 100 mL of H2O.
7. 100 mM Calcium chloride (CaCl2) solution. Add 14.69 mg of
calcium chloride dehydrate to 1 mL of water.

2.1.3 Preparation 1. Heavy standard proteotypic peptides of target proteins are syn-
of Heavy Standard Peptide thesized, either commercially or if available in-house (see Note 2).
Solutions 2. Prepare solutions of peptide standards at a nominal concentra-
tion of 1 pmol/μL using 30 % acetonitrile and 70 % of 0.2 %
formic acid (see Note 3).
3. Eppendorf tubes; pipettes.

2.1.4 Preparation of iRT 1. Eleven iRT peptides (LGGNETQVR (LGG), AGGSSEPVT

Peptide Solutions GLADK (AGG), VEATFGVDESANK (VEA), YILAGVESNK
(YIL), TPVISGGPYYER (TPV…S), TPVITGAPYYER (TPV…T),
GDLDAASYYAPVR (GLD)) are synthesized, either commer-
cially or if available in-house (see Note 4).
2. Prepare a solution containing all iRT peptide standards at a
nominal concentration of 1 pmol/μL using 30 % acetonitrile
and 70 % of 0.2 % formic acid.
3. Eppendorf tubes; pipettes.

2.2 Liquid 1. QTRAP 6500 (AB Sciex, Framingham, MA) with an Eksigent
Chromatography- ekspert nanoLC 425 pump, ekspert nanoLC400 autosampler,
Mass Spectrometry ekspert cHiPLC, and Analyst software. The LC system is coupled
(LC-MS/MS) to a NanoSpray III Source and Heated Interface (AB/Sciex).
2. CHiPLC Chrom XP C18-CL 3 μm trap column, 120 Å (200
μm × 0.5 mm).
80 Annie Moradian et al.

3. CHiPLC Chrom XP C18-CL 3 μm column, 120 Å (75

μm × 150 mm).
4. Sonicate solvent A and B prior to use (see Note 5).

2.3 Data Analysis Skyline software (see Note 6).

3 Methods

3.1 Preparation 1. Resuspend cells or tissue in a 1:10 sample-to-buffer ratio and

of HeLa Tryptic incubate at 95 °C for 5 min.
Peptide Solutions 2. Lyse cells or homogenize tissue using standard methods.
Using a Modified Clarify lysate by centrifugation at 16,000 × g for 5 min at 20 °C
Filter-Assisted Sample and determine protein concentration (see Note 7).
Preparation (FASP)
Procedure [8]
3.1.1 Lysate Preparation

3.1.2 FASP (Double 1. Combine up to 200 μg of lysate with 200 μL of UA buffer in

Digestion) a 30 K microcon filter unit and centrifuge at 14,000 × g for
15 min.
2. Discard flow-through from the collection tube, add an addi-
tional 200 μL of UA buffer to the filter unit, and centrifuge at
14,000 × g for 15 min.
3. Repeat step 2 at least two additional times (see Note 8).
4. Discard the flow-through, add 100 μL of UA buffer contain-
ing 27 mM TCEP to the filter unit, and mix at 600 rpm in a
thermo-mixer for 1 min. Then incubate for 20 min at room
temperature without shaking (see Note 9).
5. After the incubation with TCEP is complete, centrifuge the
filter unit at 14,000 × g for 10 min.
6. Discard the flow-through, add an additional 100 μL of UA buffer
to the filter unit, and then centrifuge at 14,000 × g for 15 min.
7. Discard the flow-through, add 100 μL of alkylating solution,
and mix at 600 rpm in a thermo-mixer for 1 min. Then incu-
bate for 20 min at room temperature without shaking, in the
dark (see Note 10).
8. After the incubation with the alkylation solution is complete,
centrifuge the filter unit at 14,000 × g for 10 min.
9. Discard the flow-through, add 100 μL of UB buffer to the
filter unit, and centrifuge at 14,000 × g for 15 min.
10. Repeat step 9 two additional times.
11. Transfer the filter unit to a new collection tube, add 40 μL of
UB buffer containing Lys-C (enzyme-to-protein ratio 1:50),
 Targeted Tempo-Spatial Proteomics 81

and mix at 600 rpm for 1 min in the thermo-mixer, followed

by a 4-h incubation at room temperature without shaking in
the dark.
12. Then add 120 μL of trypsin dissolved in 50 mM NH4HCO3
(enzyme-to-protein ratio 1:100), along with 1.6 μL of 100
mM CaCl2 and mix for 1 min at 600 rpm in a thermo-mixer
(see Note 11).
13. Incubate the units at room temperature for 14 h, in the dark.
14. Centrifuge the filter unit at 14,000 × g for 15 min (see Note 12).
15. To increase recovery of peptides, add 100 μL of 50 mM
NH4HCO3 to the filter unit and centrifuge at 14,000 × g for
15 min (see Note 12).
16. Repeat step 15 two additional times and combine all flow-
through (see Note 12).
17. Add enough formic acid to bring up the concentration to 5 %.
18. Lyophilize the peptide solution.
19. Resuspend the lyophilized peptides in 100 μL of 0.2 % formic
acid and desalt by HPLC, e.g., with a C8 peptide macrotrap
(3 × 8 mm) (200 μg maximum capacity) (see Note 13).
20. Lyophilize the desalted peptides and store at −20 °C until
ready for mass spec analysis (see Note 14).
21. Prior to analysis, resuspend HeLa digest in 0.2 % formic acid at
a nominal concentration of 4 μg/μL.

3.2 Preparation 1. To 1.5 μL of trypsin-digested HeLa lysate (4 μg/μL), add 3

of Serial Dilutions μL of heavy standard solution (1 pmol/μL), 0.6 μL of iRT
of Heavy Standard peptide solution (1 pmol/μL), and 0.9 μL of 0.2 % formic
Peptide and iRT acid: solution A, 500 fmol/μL of heavy standard.
Peptide Solutions 2. Keep 1 μL of solution A (500 fmol/μL) for injection (see
in Complex Note 15).
Background 3. Dilute 1.5 μL of solution A (500 fmol/μL) in 1.125 μL trypsin-
digested HeLa lysate, 0.45 μL iRT peptide solution (1 pmol/
μL), and 2.925 μL of 0.2 % formic acid: solution B, 125 fmol/
μL of heavy standard.
4. Keep 1 μL of solution B (125 fmol/μL) for injection.
5. Dilute 1.5 μL of solution B (125 fmol/μL) in 1.125 μL
trypsin-digested HeLa lysate, 0.45 μL iRT peptide solution (1
pmol/μL), and 2.925 μL of 0.2 % formic acid: solution C,
31.3 fmol/μL of heavy standard.
6. Keep 1 μL of solution C (31.3 fmol/μL) for injection.
7. Dilute 1.5 μL of solution C (31.3 fmol/μL) in 1.125 μL
trypsin-digested HeLa lysate, 0.45 μL iRT peptide solution
(1 pmol/μL), and 2.925 μL of 0.2 % formic acid: solution D,
7.8 fmol/μL of heavy standard.
82 Annie Moradian et al.

8. Keep 1 μL of solution D (7.8 fmol/μL) for injection.

9. Dilute 1.5 μL of solution D (7.8 fmol/μL) in 1.125 μL trypsin-
digested HeLa lysate, 0.45 μL iRT peptide solution (1 pmol/
μL), and 2.925 μL of 0.2 % formic acid: solution E, 1.95 fmol/
μL of heavy standard.
10. Keep 1 μL of solution E (1.95 fmol/μL) for injection.
11. Dilute 1.5 μL of solution E (1.95 fmol/μL) in 1.125 μL
trypsin-digested HeLa lysate, 0.45 μL iRT peptide solution
(1 pmol/μL), and 2.925 μL of 0.2 % formic acid: solution F,
488 amol/μL of heavy standard.
12. Keep 1 μL of solution F (488 amol/μL) for injection.
13. Dilute 1.5 μL of solution F (488 amol/μL) in 1.125 μL
trypsin-digested HeLa lysate, 0.45 μL iRT peptide solution
(1 pmol/μL), and 2.925 μL of 0.2 % formic acid: solution G,
122 amol/μL of heavy standard.
14. Keep 1 μL of solution G (122 amol/μL) for injection.
15. Dilute 1.5 μL of solution G (122 amol/μL) in 1.125 μL
trypsin-digested HeLa lysate, 0.45 μL iRT peptide solution
(1 pmol/μL), and 2.925 μL of 0.2 % formic acid: solution H,
30.5 amol/μL of heavy standard.
16. Keep 1 μL of solution H (30.5 amol/μL) for injection.
17. Dilute 1.5 μL of solution H (30.5 amol/μL) in 1.125 μL
trypsin-digested HeLa lysate, 0.45 μL iRT peptide solution
(1 pmol/μL), and 2.925 μL of 0.2 % formic acid: solution I,
7.6 amol/μL of heavy standard.
18. Keep 1 μL of solution I (7.6 amol/μL) for injection (see Note 16).
19. Repeat steps 1–17 two additional times so that three samples
are created for each solution.
20. Calculate 2–3 transitions per iRT, heavy standard, and light
sample peptide.

3.3 LC-MRM 1. For the separation and analysis of the samples, a nanoLC
of Heavy Proteotypic QTRAP 6500 (AB Sciex, Framingham, MA) can be used
Standard Peptide (see Note 5).
Solutions 2. The peptides are separated using a CHiPLC Chrom XP
C18-CL 3 μm column, 120 Å (75 μm × 150 mm) equipped
with a CHiPLC Chrom XP C18-CL 3 μm trap column, 120 Å
(200 μm × 0.5 mm) at a column temperature of 45 °C and a
flow rate of 300 nL/min. Solvent A is 0.2 % formic acid and
solvent B is 98.8 % acetonitrile containing 0.2 % formic acid.
Linear gradients from 5 to 30 % B are applied within 45 min,
30–90 % B in 2 min, followed by 100 % B for 10 min.
Preliminary mass spectra are recorded in positive ion mode
acquiring data from the transition lists, initially in the QTRAP
 Targeted Tempo-Spatial Proteomics 83

mode (Fig. 1a) to confirm correct peak identification.

Optimization of declustering potentials and collision energy is
done automatically in Skyline. Subsequent analyses are
performed using MRM scheduling in QQQ mode (Fig. 1b)
(see Note 17).
3. Inject 1 μL of solution I through A (see Note 18).
4. Inject 1 μL of 0.2 % formic acid solution (blank).
5. Repeat steps 5 and 6 two more times.
6. After an initial data analysis, spike in the heavy standard
concentration to your sample that is most appropriate, i.e., at
roughly the same concentration range.
7. Inject 1 μL of this adjusted solution.
8. Repeat step 7 two additional times for triplicate measurements.

3.4 Data Analysis Import the wiff (or appropriate raw) files of all measurements into
Skyline software (see Note 19).

4 Notes

1. In addition to the calibration curves and dynamic range, ana-

lytical validation is achieved by testing repeatability (technical
replicates, injecting the same sample ten times), reproducibility
(biological replicates, injecting ten samples that have been pre-
pared the same way ten times), limit of detection (best approx-
imated by using ca. three times noise), and limit of quantitation
(best approximated by using ca. ten times noise), similar to
common practice in the pharmaceutical field [2, 9–12].
2. Generally, it is sufficient to have one heavy-labeled amino acid
such as [13C6]Lys or [13C6]Arg at the C-terminal side of the
tryptic peptide. It may be necessary to purify and/or desalt the
peptides after synthesis. Peptides containing Cys, Met, His,
N-terminal Glu or Gln, glycosylation site motifs (NXS/T), or
Pro following Lys should be avoided if possible.
3. If the stock solutions are to be stored in the freezer, we recom-
mend using higher concentrations (nmol/μL) for storage to
avoid sample adsorption to the storage vials.
4. iRT peptides are used to assess chromatographic reproducibil-
ity. They also enable users to schedule their analysis in smaller
windows and help verify transitions in case of interference.
Finally, they aid in transferring methods from one lab to
another lab.
5. The use of this setup is not mandatory. Other QQQ mass spec-
trometers and quadrupole-based mass spectrometers such as
84 Annie Moradian et al.

Q-TOF or Q-Exactive from other vendors with nanoLC setup

are equally suited for this purpose.
6. This software tool has been developed by the MacCoss lab and
is freely available at proteome.gs.washington.edu/software/
skyline/. Extensive documentation is provided on the Skyline
website for data analysis and interpretation. Commercial packages
may serve the same purpose.
7. Depending on the cell or tissue type used lysis methods
will vary.
8. The purpose of steps 1 and 2 is to wash away the SDS; since
proteins are extended and denatured they will remain in the
filter unit. It may be necessary to repeat this wash step multiple
times. We have found five washes to be sufficient for most
samples.
9. The final concentration of TCEP once the 100 μL is added
to the 30 μL of filtrate that remains in the filter unit will be 20
mM. The purpose of this step is to ensure complete reduction
of all disulfide bridges.
10. The purpose of this step is to alkylate the cysteine residues to
prevent disulfide bridges from reforming.
11. Adding CaCl2 will enhance the activity of trypsin.
12. The flow-through contains the trypsin-digested peptides.
Keep!
13. The FASP procedure is optimized for the recovery of purified
tryptic peptides from intact cells or tissue. Loss of sample will
be greater than 50 %; therefore adjust the amount of starting
material accordingly. A total of 100 μg will be sufficient for a
comprehensive assay development.
14. Determine peptide concentration, e.g., through UV response
during the desalting step.
15. The resulting 1 μL contains 500 fmol of each heavy standard
and 100 fmol of each iRT peptide.
16. The serial dilutions are prepared to determine linearity, dynamic
range, and LOD/LOQ values.
17. While theoretically all samples can be analyzed without MRM
scheduling, it is highly recommended to take advantage of the
MRM scheduling. Once retention times have been recorded in
a preliminary analysis, transitions in the following analysis
according to the predetermined retention times should be
scheduled. This ensures higher sensitivity and more data points
across the peaks for smoother peak shapes and better accuracy
in quantification.
18. Start with the lowest concentration to avoid carryover.
 Targeted Tempo-Spatial Proteomics 85

19. The files are imported and show target masses, individual tran-
sitions per peptide, peak areas, and retention times over the
multiple analysis. Considerable manual intervention is needed
to confirm assignment of peaks.

Acknowledgement

This work was supported by the Beckman Institute, the Gordon

and Betty Moore Foundation through Grant GBMF775, and the
NIH through grant 1S10OD010788-01A1.

References
1. Picotti P, Aebersold R (2012) Selected reaction Proteomics 12(8):1111–1121. doi:10.1002/
monitoring-based proteomics: workflows, pmic.201100463
potential, pitfalls and future directions. Nat 7. MacLean B, Tomazela DM, Shulman N et al
Methods 9(6):555–566. doi:10.1038/ (2010) Skyline: an open source document editor
nmeth.2015 for creating and analyzing targeted proteomics
2. Carr SA, Abbatiello SE, Ackermann BL et al experiments. Bioinformatics 26(7):966–968.
(2014) Targeted peptide measurements in biology doi:10.1093/bioinformatics/btq054
and medicine: best practices for mass spectrometry- 8. Wisniewski JR, Zougman A, Nagaraj N et al
based assay development using a fit-for-purpose (2009) Universal sample preparation method
approach. Mol Cell Proteomics 13(3):907–917. for proteome analysis. Nat Methods 6(5):359–
doi:10.1074/mcp.M113.036095 362. doi:10.1038/nmeth.1322
3. Huttenhain R, Soste M, Selevsek N et al (2012) 9. Hess S, Akermann M, Ropte D et al (2001)
Reproducible quantification of cancer-associated Rapid and sensitive LC separation of new
proteins in body fluids using targeted pro- impurities in trimethoprim. J Pharm Biomed
teomics. Sci Transl Med 4(142):142ra194. Anal 25(3–4):531–538
doi:10.1126/scitranslmed.3003989 10. Hess S, Dolker M, Haferburg D et al (2001)
4. Kuzyk MA, Smith D, Yang J et al (2009) Separation, analyses and syntheses of trime-
Multiple reaction monitoring-based, multi- thoprim impurities. Pharmazie 56(4):306–310
plexed, absolute quantitation of 45 proteins in 11. Hess S, Muller CE, Frobenius W et al (2000)
human plasma. Mol Cell Proteomics 8(8):1860– 7-deazaadenines bearing polar substituents:
1877. doi:10.1074/mcp.M800540-MCP200 structure-activity relationships of new A(1) and
5. Picotti P, Rinner O, Stallmach R et al (2010) A(3) adenosine receptor antagonists. J Med
High-throughput generation of selected Chem 43(24):4636–4646. doi:10.1021/
reaction-monitoring assays for proteins and Jm000967d
proteomes. Nat Methods 7(1):43–46. 12. Hess S, Teubert U, Ortwein J et al (2001)
doi:10.1038/nmeth.1408 Profiling indomethacin impurities using high-
6. Escher C, Reiter L, MacLean B et al (2012) performance liquid chromatography and
Using iRT, a normalized retention time for nuclear magnetic resonance. Eur J Pharm Sci
more targeted measurement of peptides. 14(4):301–311
 Chapter 7

Targeted Phosphoproteome Analysis Using Selected/

Multiple Reaction Monitoring (SRM/MRM)
Jun Adachi, Ryohei Narumi, and Takeshi Tomonaga

Abstract
Mass spectrometry-based phosphoproteomics has been rapidly spread based on the advancement of mass
spectrometry and development of efficient enrichment techniques for phosphorylated proteins or peptides.
Non-targeted approach has been employed in most of the studies for phosphoproteome analysis. However,
targeted approach using selected/multiple reaction monitoring (SRM/MRM) is an indispensible tech-
nique used for the quantitation of known targets especially when we have many samples to quantitate
phosphorylation events on proteins in biological or clinical samples. We herein describe the application of
a large-scale phosphoproteome analysis and SRM-based quantitation for the systematic discovery and
validation of biomarkers.

Key words Phosphoproteomics, Selected/multiple reaction monitoring, Targeted proteomics,

Biomarker, IMAC

1 Introduction

Mass spectrometry (MS) is a powerful tool to identify protein

phosphorylation sites, and it is possible to identify over 50,000
phosphorylation sites using the state-of-the-art proteomic analysis
platform [1]. Combined with a quantification method such as met-
abolic labeling method (e.g., SILAC), chemical labeling method
(e.g., TMT, iTRAQ), or label-free quantification method, it is pos-
sible to perform systematic quantification of protein phosphoryla-
tion. These non-targeted analyses do not require previous
knowledge of the target proteins; thus it is suitable to use at discov-
ery phase in order to identify key phosphorylation events or bio-
marker candidates and create a new hypothesis. Chemical labeling
techniques are particularly useful for quantitatively comparing
proteomes between clinical samples such as tissues or plasma/
serum. For example, a large-scale phosphoproteome analysis
can be performed by combining chemical labeling with

Jörg Reinders (ed.), Proteomics in Systems Biology: Methods and Protocols, Methods in Molecular Biology, vol. 1394,
DOI 10.1007/978-1-4939-3341-9_7, © Springer Science+Business Media New York 2016

87
88 Jun Adachi et al.

phosphopeptide-enrichment technique, and has recently been

applied to the discovery of biomarker candidates using tissue sam-
ples [2].
Another approach for MS-based phosphopeptide quantifica-
tion is selected reaction monitoring (SRM, also called as MRM,
multiple reaction monitoring). SRM is a targeted approach to
quantitate specific peptides by monitoring specific precursor-to-
product ion transitions during liquid chromatography (LC)-MS/
MS in a triple-quadrupole instrument. Thus, SRM is useful for
an extensive validation for tens or hundreds of biomarker candi-
dates identified by a phosphoproteome analysis at discovery
phase. Antibody-based validation is also common; however,
available number of phosphorylation-specific antibody is limited
in the case of protein phosphorylation. Thus MS-based phos-
phoproteomic targeted analysis such as SRM has an importance
at the validation phase.
Here, we describe the application of a large-scale phosphopro-
teome analysis and SRM-based quantification to develop a strategy
for the systematic discovery and validation of biomarkers using tis-
sue samples or cultured cells. At first we identify differentially mod-
ulated phosphopeptides using immobilized metal ion affinity
chromatography (IMAC) coupled with non-targeted quantitative
proteomic analysis. Identified phosphopeptide candidates are then
validated by the SRM analysis. This systematic approach has enor-
mous potential for the discovery of bona fide disease-specific bio-
markers (Fig. 1).

2 Materials

2.1 Reagents 1. Phase-transfer surfactants A (PTS-A) buffer: 50 mM ammo-

and Equipment nium bicarbonate.
for Sample 2. Phase-transfer surfactants B (PTS-B) buffer: 50 mM ammo-
Preparation nium bicarbonate, 12 mM sodium deoxycholate, 12 mM
and Enzymatic sodium N-lauroyl sarcosinate.
Digestion (See Note 1)
3. Lysis buffer: PhosSTOP phosphatase inhibitor cocktail (Roche
Diagnostics, Manheim, Germany) dissolved in PTS-B buffer.
Prepare the lysis buffer just before the use.
4. LysC stock solution: 0.004 unit/μL lysyl endopeptidase (LysC)
(Wako Pure Chemical Industries, Osaka, Japan). Dissolve
lyophilized LysC in water (see Note 2). Store at −80 °C.
5. Trypsin stock solution: 1 μg/μL Trypsin (proteomics grade;
Roche Diagnostics). Dissolve lyophilized trypsin in 10 mM
HCl (see Note 3). Store at −80 °C.
6. DTT solution (10×): 100 mM dithiothreitol (DTT). Add 65
μL of PTS-A buffer to 1 mg of DTT and dissolve immediately
prior to use.
 Targeted-Phosphoproteomics by SRM/MRM Approach 89

Tissues Cells
Non-targeted analysis Targeted analysis

Lysates Lysates
2 mg/sample 0.5 mg/sample
PTS method
Stable isotope
Peptides Peptides labeled peptides
Fe-IMAC

Phosphopeptides Phosphopeptides

SCX fractions LC-SRM/MS

LC-MS/MS Data analysis

Data analysis
Target selection

Fig. 1 Workflows of non-targeted analysis and targeted (SRM) analysis. In non-targeted analysis, proteins are
denatured and digested using phase transfer surfactant (PTS) method. Phosphopeptides are enriched using
Fe-IMAC, followed by SCX off-line fractionation and LC-MS/MS. In the SRM analysis, the individual sample is
processed by PTS method and spiked with a mixture of the isotope-labeled peptides. The resulting sample is
applied to Fe-IMAC to enrich the phosphopeptides and analyzed by SRM analysis

7. Iodoacetamide (IAA) solution (10×): 500 mM Iodoacetamide.

Add 10.8 μL of PTS-A buffer to 1 mg of IAA and dissolve
immediately prior to use.
8. DC Protein Assay Kit (Bio-Rad, Hercules, CA).
9. Tissue grinder.
10. Liquid nitrogen.
11. Phosphate-buffered saline (PBS) buffer.
12. Bovine serum albumin (BSA), e.g., Pierce BSA Protein Assay
Standards (Thermo Scientific, Rockford, IL).
13. Benchtop centrifuge.
14. Sonicator, e.g., Bioruptor-UCD-250 (Cosmo Bio, Tokyo,
Japan).
15. Speed Vac.
16. Ethylacetate.
17. Trifluoroacetic acid (TFA) (HPLC grade).
90 Jun Adachi et al.

2.2 Reagent 1. Empore™ C18 47 mm disk (3 M, St. Paul, MN).

for Desalting Peptide 2. 200 μL pipet tips.
Mixtures by C18-Stage
3. Methanol (LC-MS grade)
Tip (See Note 4)
4. 80 % Acetonitrile, 0.1 % TFA: Mix acetonitrile (LC-MS grade),
distilled water (LC-MS grade), and TFA.
5. 2 % Acetonitrile, 0.1 % TFA.
6. 60 % Acetonitrile, 0.1 % TFA.

2.3 Reagents 1. ProBond™ Nickel-Chelating Resin (Life Technology,

for IMAC Carlsbad, CA).
2. 50 mM EDTA.
3. 0.1 % Acetic acid.
4. 100 mM FeCl3, 0.1 % acetic acid.
5. 2 % Acetonitrile, 0.1 % TFA.
6. 60 % Acetonitrile, 0.1 % TFA.
7. 1 % Phosphate: Dilute phosphoric acid (HPLC grade, 85 %)
85 times with water.

2.4 Strong Cation- 1. Buffer A: 25 % Acetonitrile, 10 mM H3PO4 (pH 3). Mix 250
Exchange mL of acetonitrile, approximately 650 mL of water, and 685
Chromatography μL of phosphoric acid. Adjust to pH 3.0 by adding KOH solu-
tion and to 1 L with water.
2. Buffer B: 25 % Acetonitrile, 10 mM H3PO4 (pH 3), 1 M KCl.
Mix 250 mL of acetonitrile, approximately 650 mL of water,
685 μL of phosphoric acid, and 74.55 g of KCl. Adjust to pH
3.0 by adding KOH solution and to 1 L with water.
3. HPLC system, e.g., Prominence UFLC (Shimadzu, Kyoto,
Japan).
4. Strong cation-exchange (SCX) column, e.g., 50 mm × 2.1 mm,
5 μm, 300 Å, ZORBAX 300SCX (Agilent Technology, Santa
Clara, CA).

2.5 Non-targeted 1. Buffer-A: 0.1 % Formic acid, 2 % acetonitrile.

LC-MS/MS Analysis 2. Buffer-B: 0.1 % Formic acid, 90 % acetonitrile.
3. Sample vial or sample plate for LC-MS/MS analysis.
4. Mass spectrometer for LC-MS/MS analysis, e.g., LTQ-
Orbitrap Velos mass spectrometer (Thermo Scientific).
5. Nano-electrospray ion source.
6. Nano-LC system, e.g., paradigm system (Michrom Biosciences,
Auburn, CA, USA).
7. Analytical column, e.g., a self-packed ESI column (see Note 5).
 Targeted-Phosphoproteomics by SRM/MRM Approach 91

8. Trap column, e.g., L-column2 ODS (Chemicals Evaluation

and Research Institute, Tokyo, Japan).
9. Software for non-targeted proteomic data analysis, e.g.,
Proteome Discoverer 1.3 (Thermo Scientific) connected to a
search engine Mascot server 2.4 (Matrix Science) (see Note 6).
10. 2 % Acetonitrile, 0.1 % TFA.

2.6 Targeted (SRM/ Basically same reagents and equipment are used except for those
MRM-Based) LC-MS/ listed below.
MS Analysis 1. Stable isotope-labeled peptides (SI peptides) (crude grade)
(Thermo Fisher Scientific) (see Note 7).
2. Mass spectrometer for SRM analysis, e.g., TSQ Vantage triple-
quadruple mass spectrometer (Thermo Scientific).
3. Software for targeted (SRM/MRM) proteomic analysis, e.g.,
Pinpoint 1.2 (Thermo Scientific), which is a software to quan-
titate the peak areas (quantitative data of targeted peptides)
from the raw data of SRM analysis as well as to develop the
SRM methods.
4. 2 % Acetonitrile, 0.1 % TFA.
5. 2 % Acetonitrile, 0.1 % TFA, 25 μg/mL EDTA.
6. 1 pmol/μL BSA digest solution.

3 Methods

3.1 Sample 1. Chill the stainless tissue pulverizer in liquid nitrogen. Place a
Preparation piece of frozen tissue in the chilled device and pulverize the
for Tissue Sample tissue by striking the device with a mallet several times.
2. Check the size of crushed particles. Keep pulverizing the par-
ticles until there are no large pieces left in the particles.
3. Transfer the grinded tissue into the chilled tube. Store at −80
°C.
4. Take the required amount of pulverized tissue sample into a
microcentrifuge tube (see Note 8). If a high level of blood
contamination is predicted, wash the sample by an appropriate
volume of PBS.
5. Add cold lysis buffer, approximately 15 μL per 1 mg of tissue
(see Note 9). Suspend the tissue by pipetting.
6. Immediately place the sample tubes into an ice-cold water bath
in the Bioruptor-UCD-250 sonicator. Homogenize the sam-
ple by the sonication for 10 min (10 cycles of 30 s on/30 s off)
with the amplitude set to 250 W (see Note 10).
7. Centrifuge the sample at 100,000 × g for 30 min at 4 °C. Collect
the supernatant into a new tube. Place a small amount of the
92 Jun Adachi et al.

sample into another tube to determine the protein concentra-

tion. Store the remainder at −80 °C.
8. The protein concentration is determined by a DC protein assay
kit using BSA as the standard.

3.2 Sample 1. Wash cells in the tissue culture dish by directly adding cold
Preparation PBS and rocking gently. Aspirate PBS and repeat. Keep tissue
for Cultured Cells culture dish on ice throughout.
2. Add appropriate volume of ice-cold lysis buffer to the dish,
approximately 1 mL for a 100 mm tissue culture dish.
3. Scrape cells from the surface using a rubber spatula. Transfer
the cells into the chilled tube.
4. Sonicate the samples in an ice-cold water bath using the
Bioruptor-UCD-250 sonicator. Homogenize the sample by
the sonication for 10 min (10 cycles of 30 s on/30 s off) with
the amplitude set to 250 W. Repeat the sonication if the sam-
ple is still viscose.
5. Take a small amount of the sample into another tube to deter-
mine the protein concentration. Store the remainder at −80 °C.
6. The protein concentration is determined by a DC protein assay
kit using BSA as the standard.

3.3 Protein Digestion 1. Add the homogenate to a new tube. Dilute the homogenate
with the lysis buffer to a concentration that is constant across
all samples (see Note 11).
2. Reduce cysteine residues with 10 mM dithiothreitol (DTT) for
30 min.
3. Alkylate the residues with 50 mM iodoacetamide (IAA) for
30 min in the dark.
4. Dilute the sample five times with PTS-A buffer.
5. Digest the sample by 1:100 (w/w) trypsin for 12 h at 37 °C.
6. Add an equal volume of ethyl acetate. Acidify the sample by
adding 1/200 volume of trifluoroacetic acid (TFA) in order to
transfer the detergents from the water layer into the ethyl ace-
tate layer while most of the peptides remain in the water layer.
Mix the ethyl acetate and water layers well by vortexing the
tube. Centrifuge the tube at 10,000 × g for 10 min at room
temperature to separate the sample into two layers. Discard the
upper ethyl acetate layer.
7. Dry the aqueous layer using Speed Vac and store at −80 °C.

3.4 Preparation 1. Transfer 2 mL slurry (1 mL resin) of Probond™ nickel-

of Fe-IMAC Resin chelating resin to empty spin columns (see Note 12). Centrifuge
the resin at 150 × g for 2 min to discard the flow through.
 Targeted-Phosphoproteomics by SRM/MRM Approach 93

2. In order to remove nickel ions from resin, wash the resin by 50

mM EDTA solution (3 mL of the solution per 1 mL of the
resin) and centrifuge at 150 × g for 2 min. Repeat this step until
the color of the resin turns white (see Note 13).
3. Add water (3 mL of water per 1 mL of the resin) to the resin
and centrifuge at 150 × g for 2 min. Discard the flow through.
4. Add 1 % acetate solution (3 mL of the solution per 1 mL of the
resin) to the resin in the column, centrifuge at 150 × g for 2
min, and discard the flow through. Repeat this step once more.
5. Add 100 mM FeCl3 in 0.1 % acetic acid (2 mL of the solution
per 1 mL of the resin) to the resin and centrifuge at 150 × g for
2 min to chelate iron ions to the resin. Repeat this step once
more.
6. Wash the resin by 1 % acetate solution (3 mL of the solution
per 1 mL of the resin) and centrifuge at 150 × g for 2 min.
Repeat this step twice more.
7. Add 60 % acetonitrile and 0.1 % TFA solution (3 mL of the
solution per 1 mL of the resin) to the resin in the column,
centrifuge at 150 × g for 2 min, and discard the flow through.
Repeat this step once more (see Note 14).

3.5 Phosphopeptide 1. Dissolve the tryptic digests prepared from the tissue samples or
Enrichment cultured cells in 60 % acetonitrile and 0.1 % TFA solution.
by Fe-IMAC for Non- 2. Load the digests to the Fe-IMAC resin. Centrifuge at 150 × g
targeted for 2 min and discard the flow through.
Phosphoproteome 3. Wash the resin by 60 % acetonitrile and 0.1 % TFA solution (3
Analysis mL of the solution per 1 mL of the resin) and centrifuge at
150 × g for 2 min to wash off the non-phosphopeptides. Repeat
this step twice more.
4. Add 2 % acetonitrile and 0.1 % TFA solution (3 mL of the
solution per 1 mL of the resin) to the resin. Centrifuge at
150 × g for 2 min and discard the flow through.
5. Elute phosphopeptides by 1 % phosphate solution (1 mL of
the solution per 1 mL of the resin) and centrifuge at 150 × g for
2 min. Collect the elute into a tube. Repeat this step once
more and then collect the second elute into the same tube.
6. Desalt the elute with a disposable solid-phase extraction (SPE)
device such as Sep-Pak C18 cartridge.
7. Dry the sample using Speed Vac and store at −80 °C.

3.6 Strong Cation- 1. Dissolve the sample in mobile buffer A (see Note 15).
Exchange 2. Apply the sample to an HPLC and separate on an SCX column
Chromatography using a linear gradient of mobile buffer A and B and sequen-
for Non-targeted tially collect eluate every 1 min (see Note 16).
Phosphoproteome
Analysis
94 Jun Adachi et al.

3. Adjust the number of the fractions for the subsequent MS

analysis by combining the fractions based on the peak intensity
on the HPLC chromatogram (see Note 17).
4. Evaporate acetonitrile in the fractions using Speed Vac.
5. Desalt the combined fractions with a disposable SPE device
such as C18-Stage Tip (see Note 4).
6. Elute the sample into a sample vial for MS analysis and then
dry it using Speed Vac.
7. Store at −80 °C until MS analysis.

3.7 LC-MS/MS 1. Add 10 μL of 2 % acetonitrile and 0.1 % TFA to each sample vial.
for Non-targeted 2. Vortex each vial for 1 min and then spin down.
Phosphoproteome 3. Set the operating parameters of the mass spectrometer (see
Analysis Note 18).
4. Analyze each sample by LC-MS/MS (see Note 19).
5. Apply the acquired raw file to data analysis software with search
engine to identify and quantify the phosphopeptides (see Note 20)
6. Select the phosphopeptide that has to be validated in the sub-
sequent SRM analysis.

3.8 Sample 1. Prepare a homogenate as previously described in Subheading

Preparation 3.1 or 3.2.
for Targeted Analysis 2. Digest the homogenate according to steps 1–5 in Subheading
3.3.
3. Add all stable isotope-labeled (SI) peptides to each sample (see
Notes 21 and 22).
4. Extract the peptides from the sample as described in steps 6–7
in Subheading 3.3.
5. Desalt the resulting sample with C18-Stage Tip.
6. Prepare the micro-scale IMAC column using C18-Stage Tip.
Pack two disks of C18 disk at the end of a 200 μL pipet tip and
then load 50 μL of Fe-IMAC resin. Centrifugation at 800 × g
for 2 min.
7. Load the desalted sample in 60 % acetonitrile and 0.1 % TFA
to the IMAC-C18-Stage Tip and then centrifuge at 600 × g for
5 min.
8. Wash the column by 200 μL of 60 % acetonitrile and 0.1 %
TFA and then centrifuge at 800 × g for 2 min. Repeat this step
twice more.
9. Add 200 μL of 0.1 % TFA to the IMAC-C18-Stage Tip and
then centrifuge at 800 × g for 2 min to equilibrate the C18
resin under the IMAC resin.
 Targeted-Phosphoproteomics by SRM/MRM Approach 95

10. Elute phosphopeptides by 100 μL of 1 % phosphate and then

centrifuge at 800 × g for 2 min. Eluted peptides are trapped on
the C18 disk. Repeat this step once more.
11. Add 200 μL of 0.1 % TFA to the IMAC-C18-Stage Tip and
then centrifuge at 2300 × g for 2 min to wash the C18 disk.
12. Elute the phosphopeptides bound to the C18 disk to a sample
tube using 60 μL of 60 % acetonitrile and 0.1 % TFA.
13. Dry the sample using Speed Vac.
14. Store at −80 °C until MS analysis.

3.9 Targeted Analysis 1. In order to acquire target information (retention time, mass of
by LC-SRM/MS target ion, and transition ions) prior to the analysis, prepare a
mixture of the stable isotope-labeled peptide (SI peptides),
which has the same sequence as the phosphopeptide selected
from the results of non-targeted analysis. Analyze the mixture
by LC-MS/MS using data-dependent mode (see Note 23).
2. Create a primary method for the subsequent SRM analysis by
analyzing the acquired MS data and selecting the precursor
ions of each target observed with a strong signal intensity
(doubly, triply, or higher charged ions) and the product ions
generated from the precursor ion with a strong signal intensity
(see Note 24).
3. Optimize the parameters (m/z of product ions and collision
energy (CE)) of the SRM method (see Note 25).
4. Add 10 μL of 2 % acetonitrile, 0.1 % TFA, and 25 μg/mL
EDTA to each sample.
5. Vortex each vial for 1 min to dissolve the peptides and then
spin down.
6. Set the optimized SRM method and other operating parame-
ters for the SRM analysis (see Note 26).
7. Analyze each sample by LC-SRM/MS (see Note 27)
8. Analyze the acquired raw data by the software (see Note 28).
Target peptides are compared across the samples by extracted
ion chromatogram (XIC) intensity of each SRM transition and
then normalize the values of the endogenous targeted peptides
to those of the corresponding SI peptides.

4 Notes

1. The procedures used for homogenizing samples and enzy-

matic digestion are based on phase transfer surfactant (PTS)-
aided trypsin digestion as described in a previous study [3].
2. Add 0.5 mL of water to a bottle containing 2.0 unit of lyophi-
lized LysC.
96 Jun Adachi et al.

3. Add 100 μL of 10 mM HCl to a bottle containing 100 μg of

lyophilized trypsin.
4. Peptide mixtures are desalted using C18-Stage Tip or other
solid-phase extraction (SPE) devices such as Oasis
HLB. Desalting by C18-Stage Tip is performed as described
in a previous study [4]. Briefly, a small 47 mm Empore™ C18
disk is stamped out using a blunt-ended syringe needle (16 G),
and then the layers are placed in a 200 μL pipet tip by pushing
them from the top of the tip using a plunger. C18-Stage Tip is
preconditioned by methanol (for swelling), 80 % acetonitrile,
0.1 % TFA (for washing), and 2 % acetonitrile and 0.1 % TFA
(for equilibrating). After the sample is applied to the C18-
Stage Tip, the tip is washed by 2 % acetonitrile and 0.1 %
TFA. Elution of the peptides is performed by 60 % acetonitrile
and 0.1 % TFA. The volume of all solutions is 20 μL per layer
of C18 disk.
5. We make use of 20 cm long nano HPLC columns with an
inner diameter of 100 μm packed in-house with 3 μm C18
beads (L-column ODS, CERI, Tokyo, Japan).
6. We obtain protein and peptide lists and qualitative data includ-
ing Mascot ion score and local probability of phosphorylation
sites using Proteome Discoverer 1.3.
7. We mostly replace lysine or arginine at the C-terminal of target
peptides with isotope-labeled lysine (13C6, 15N2) or arginine
(13C6, 15N4) in order to make y-ion fragments heavier than
those of endogenous peptides. When the amino acid at the
C-terminal is not lysine or arginine (e.g., the C-terminal of a
protein), we replace the other amino acids (e.g., alanine) at or
near the C-terminal with the other isotope-labeled one (e.g.,
Alanine-13C3,15N1).
8. Two milligrams protein for non-targeted phosphoproteome
analysis and 0.5 mg protein for SRM analysis are required in
our study. To obtain enough amount of protein, we use more
than 40 mg of tissue or a 15 cm dish (in the case of Hela cells)
if possible.
9. By adding the buffer to samples at this ratio, we can generally
obtain a solution containing 5–15 mg of proteins per mL.
10. After several rounds of sonication, we check whether the resid-
ual pieces of the tissue are left or not for the tissue sample. If
the tissues are completely dissolved, we stop the sonication. If
not, a few rounds of sonication are additionally performed
until the samples are solved uniformly. At this stage, we con-
sider the proteins to be sufficiently extracted from the tissue
and stop the sonication.
 Targeted-Phosphoproteomics by SRM/MRM Approach 97

11. We use 2 mg of protein for non-targeted phosphoproteome

analysis and 0.5 mg protein per individual sample for SRM
analysis. As shown in Fig. 1, we optimized these initial protein
amounts to fit the loading capacity of our LC-MS system (2–3
μg peptides per injection).
12. The ProBond resin is initially provided as a 50 % slurry in 20
% ethanol. We use 1 mL of the resin (2 mL of the suspension)
for up to 2 mg of proteins.
13. When nickel ions are released from the resin by EDTA, the
color of the resins turns from blue to white.
14. We store the Fe-IMAC resin as a 50 % slurry at 4 °C and use it
within 1 week.
15. The volume of SCX buffer A to dissolve the sample depends
on the HPLC systems. We dissolve the sample in 110 μL of
SCX buffer A according to the maximum injection volume
(100 μL) of the autosampler in our HPLC system. 100 μL of
the sample is loaded onto the HPLC equipment.
16. We use a flow rate of 200 μL/min and four-step linear gradi-
ent for the separation, as follows: 0 % B for 30 min, 0–10 % B
in 15 min, 10–25 % B in 10 min, 25–40 % B in 5 min, and
40–100 % B in 5 min, and 100 % B for 10 min.
17. We usually combine 75 fractions into 30 fractions. The flow-
through fraction is not combined, because polymer-like con-
taminants are eluted in it. We combined fractions to make
peak area of each combined fraction as equal as possible.
18. For example, when we perform non-labeling or metabolic
labeling analysis (e.g., SILAC) using the LTQ-Orbitrap Velos
mass spectrometer, the operating parameters are set as follows:
full MS scans are performed in the Orbitrap mass analyzer
(scan range 350–1500 m/z, with 30 K FWHM resolution at
400 m/z). The eight most intense precursor ions are selected
for the MS/MS scans. MS/MS scans are performed using
collision-induced dissociation (CID). Collision energy is set to
35 %. A dynamic exclusion option is implemented with a
repeat count of 1 and exclusion duration of 60 s. The values of
automated gain control (AGC) are set to 5.00e + 05 for full
MS, 1.00e + 04 for CID MS/MS.
When we perform chemical labeling analysis (e.g., iTRAQ)
using the LTQ-Orbitrap Velos mass spectrometer, the operat-
ing parameters are set non-labeling or metabolic labeling anal-
ysis except for the following parameters: The five most intense
precursor ions are selected for the MS/MS scans. MS/MS
scans are performed using collision-induced dissociation
(CID) and higher energy collision-induced dissociation
(HCD, 7500 FWHM resolution at 400 m/z) for each precur-
98 Jun Adachi et al.

sor ion. Collision energy is set to 35 % for CID and 50 % for

HCD. The values of automated gain control (AGC) are set to
5.00e + 04 for HCD MS/MS.
19. Analytical column is self-made by packing C18 particles
(L-column2 ODS, 3 μm) into a self-pulled needle (200 mm
length × 100 μm for the inner diameter). The mobile phases
consist of buffers A (0.1 % formic acid, 2 % acetonitrile) and
B (0.1 % formic acid, 90 % acetonitrile). Samples are loaded
onto the trap column. The nano LC gradient is delivered at
500 nL/min and consists of a linear gradient of buffer B
developed from 5 to 30 % B in 135 min. A spray voltage of
2000 V is applied.
20. To identify the phosphopeptides, the CID and/or HCD raw
spectra are extracted and searched separately against the
Uniprot or IPI database using Proteome Discoverer 1.3 and
Mascot v2.4. The precursor mass tolerance is set to 7 ppm
and a fragment ion mass tolerance is set to 0.6 Da for CID
and 0.01 Da for HCD. The search parameters allow for one
missed cleavage for trypsin, fixed modification (carbamido-
methylation at cysteine), and variable modifications (oxida-
tion at methionine, phosphorylation at serine, threonine, and
tyrosine). Furthermore, when we employ SILAC quantita-
tion, set the appropriate SILAC plex at the precursor ions
quantifier node of Proteome Discoverer. In the case of iTRAQ
quantitation, iTRAQ labeling at lysine and the N-terminal
residue are added to the fixed modification and iTRAQ label-
ing at tyrosine are added as variable modifications. The score
threshold for peptide identification is set at 1 % false discovery
rates (FDR).
21. In the case of crude SI peptides, the purities are very different
between the products especially for phosphopeptides. In addi-
tion, the ionization efficiency depends on peptide sequences.
As a result, the signal intensities of SI peptides can be very
different. To maintain the robustness of the experimental sys-
tem, the signal intensity of each SI peptide should be checked
by LC-MS/MS before mixing and then the amount of each SI
peptide should be adjusted.
22. When we perform serial dilutions of the SI-peptides for the
addition of small amounts of peptides, we perform the dilu-
tion using a 1 pmol/μL BSA digest as a matrix to prevent
adsorption.
23. In our case, the SI peptide mixture is analyzed by LC-MS/MS
using LTQ-Orbitrap XL (CID mode) and obtained data is
analyzed by Proteome Discoverer.
 Targeted-Phosphoproteomics by SRM/MRM Approach 99

24. Msf file generated by Proteome Discoverer is opened with

Pinpoint software (version 2.3.0, Thermo Scientific) and the
list of MS/MS fragment ions derived from SI-peptides is
generated. A total of multiple product ions (four to ten
product ions) are selected for the SRM transitions of each
target peptide based on the following criteria: y-ion series,
strong ion intensity, at least two amino acids in length, and
no neutral loss fragment. When phosphopeptides contain
more than two amino acids of serine, threonine, or tyrosine,
it is important to consider the possibility of sequence iso-
mers which have the same amino acid composition, same
number of phosphorylation sites, and, however, different
phosphorylation sites. In that case, we select site-specific
fragments as transitions. Finally, we create SRM method
that consists of SRM transitions, which means pairs of m/z
of the precursor/product ions, the collision energies (CEs),
and retention time.
25. At first, we optimize collision energy (CE) for every SRM
transition around the theoretical value calculated according to
the formulas CE = 0.044 × m/z + 5.5 for doubly charged pre-
cursor ions, and CE = 0.051 × m/z + 0.55 for triply charged
precursor ions. In cases which the theoretical value is over 35
eV, the value is set to 35 eV. After this optimization, four most
intense transitions are selected for each target peptide.
26. In addition to the SRM method (SRM transitions, CE, and
the retention time for each target peptide), the parameters
of the instrument are set as follows: a scan width of
0.002 m/z, Q1 and Q3 resolution of 0.7 FWHM, cycle
time of 1 s, and gas pressure of 1.8 mTorr. Data are acquired
in the time-scheduled SRM mode (retention time window:
8 min).
27. We use the TSQ-Vantage triple-quadruple mass spectrometer
equipped with the LC system described above. The nanoLC
gradient is delivered at 300 nL/min and consists of a linear
gradient of mobile phase B developed from 5 to 23 % B in
45 min. A spray voltage of 1800 V is applied.
28. We use Pinpoint software for the analysis of SRM data. SRM
transitions with more than 1 × 103 ion intensity at the peak are
used for quantitation. We check that the ratios among the
peak areas of individual SRM transitions for each targeted
phosphopeptide are comparable to those of the corresponding
SI peptide.
100 Jun Adachi et al.

References

1. Sharma K, D’Souza RC, Tyanova S et al (2014) 3. Masuda T, Tomita M, Ishihama Y (2008) Phase
Ultradeep human phosphoproteome reveals a transfer surfactant-aided trypsin digestion for
distinct regulatory nature of tyr and ser/thr- membrane proteome analysis. J Proteome Res
based signaling. Cell Rep 8(5):1583–1594. 7(2):731–740. doi:10.1021/pr700658q
doi:10.1016/j.celrep.2014.07.036 4. Rappsilber J, Mann M, Ishihama Y (2007)
2. Narumi R, Murakami T, Kuga T et al (2012) A Protocol for micro-purification, enrichment,
strategy for large-scale phosphoproteomics and pre-fractionation and storage of peptides for
SRM-based validation of human breast cancer proteomics using StageTips. Nat Protoc
tissue samples. J Proteome Res 11(11):5311– 2(8):1896–1906. doi:10.1038/nprot.2007.26
5322. doi:10.1021/pr3005474 1,nprot.2007.261 [pii]
 Chapter 8

Testing Suitability of Cell Cultures for SILAC-Experiments

Using SWATH-Mass Spectrometry
Yvonne Reinders, Daniel Völler, Anja-K. Bosserhoff, Peter J. Oefner,
and Jörg Reinders

Abstract
Precise quantification is a major issue in contemporary proteomics. Both stable-isotope-labeling and label-
free methods have been established for differential protein quantification and both approaches have differ-
ent advantages and disadvantages. The present protocol uses the superior precision of label-free
SWATH-mass spectrometry to test for suitability of cell lines for a SILAC-labeling approach as systematic
regulations may be introduced upon incorporation of the “heavy” amino acids. The SILAC-labeled cell
cultures can afterwards be used for further analyses where stable-isotope-labeling is mandatory or has
substantial advantages over label-free approaches such as pulse-chase-experiments and differential protein
interaction analyses based on co-immunoprecipitation. As SWATH-mass spectrometry avoids the missing-
value-problem typically caused by undersampling in highly complex samples and shows superior precision
for the quantification, it is better suited for the detection of systematic changes caused by the SILAC-
labeling and thus, can serve as a useful tool to test cell lines for changes upon SILAC-labeling.

Key words Label-free proteomics, SWATH-MS, SILAC, Melanoma cells

1 Introduction

Proteomic analyses are most frequently performed for the differen-

tial quantification of proteins from highly complex mixtures such
as total cell lysates. Particularly, stable-isotope labeling methods
such as ICAT and iTRAQ have been used with huge success for
more than a decade [1]. Incorporation of stable-isotope labels by
metabolic labeling is even more suited for quantification primarily
in cell cultures, as labeling occurs at the earliest time-point possi-
ble. Thereby, all subsequent steps may be conducted with pooled
samples minimizing sample-specific losses and eliminating run-to-
run-variations. Ong et al. [2] introduced the SILAC-technique
(Stable Isotope Labeling by Amino acids in Cell culture) using
stable-isotope-coded, essential amino acids instead of native amino

Jörg Reinders (ed.), Proteomics in Systems Biology: Methods and Protocols, Methods in Molecular Biology, vol. 1394,
DOI 10.1007/978-1-4939-3341-9_8, © Springer Science+Business Media New York 2016

101
102 Yvonne Reinders et al.

acids within the cell culture medium. Thereby, these “heavy”

amino acids replace all the respective “light” amino acids in the
proteins over time. Proteins may be analyzed by different separa-
tion methods (using top-down or bottom-up proteomics) coupled
to mass spectrometry by pooling “heavy” and “light” samples and
measuring them together in a single run. Relative quantification is
accomplished by integrating the signals and comparing the respec-
tive areas for the “heavy” and “light” species. The major advan-
tages of this method are the compatibility to downstream processing
methods, elimination of run-to-run-variations, minimization of
sample-specific losses, multiplexing capability and -if label-
switching [3] is used- easy identification of contaminating pro-
teins. However, the method is also afflicted with several
disadvantages such as high cost for stable-isotopes, potential intro-
duction of artificial differences by the stable-isotope-labeled amino
acids, conversion of these amino acids to other amino acids and the
increased sample complexity as combining “heavy” and “light”
samples doubles the amount of analytes in the sample (in case of
higher multiplexing even more). Moreover, the higher complexity
further increases the probability of overlapping signals in the
Full-MS which are used for quantification. Thereby, potentially
false quantification might occur. Systematic errors can also occur if
incorporation of the “heavy” amino acids has an impact on the
abundance of proteins in a particular cell line. Therefore, such an
effect on the cell line at hand has to be tested before conducting
subsequent SILAC-experiments.
Mainly the high cost of stable-isotope-labeling experiments
(SILE) has given rise to the development of label-free quantifica-
tion methods [4]. SWATH (Sequential Window Acquisition of all
THeoretical fragment-ion spectra) mass spectrometry introduced
by Gillet et al. [5] is a method of particular interest as it yields a
comprehensive view of the analyzed proteome and eliminates the
missing-value-problem that typically afflicts information-
dependent-acquisition (IDA) workflows [6]. Furthermore, as
quantification is done in a selected reaction monitoring-like fash-
ion [7], the quantitative data is obtained at the MS/MS-level and,
thus, less prone to interference of fragment ions derived from dif-
ferent precursors. In contrast to SILE, all samples have to be mea-
sured individually (no multiplexing) and the so-called “library” has
to be generated by a preferably comprehensive IDA-based analysis.
Therefore, the needed measurement period on the mass spectrom-
eter for such analyses is typically much longer than for correspond-
ing SILAC-experiments. Furthermore, the demands for
reproducibility of all upstream methods are much higher for the
SWATH-MS-based approach.
The researcher has to decide which methods to apply for the
task at hand. Here, we do not only provide a procedure for com-
parison of SILAC- and SWATH-MS-based quantification, but also
Testing Suitability of Cell Cultures for SILAC-Experiments Using SWATH-Mass… 103

Fig. 1 Distribution of the coefficients of variation (CV) for the SILAC- and the
SWATH-based quantification of the same HMB2-cell lysates. Both methods
were accomplished in triplicate. Only CVs for proteins identified and quantified
in all of the three SILAC-runs were included in this graph. Due to the undersam-
pling in the SILAC-based approach, substantially fewer proteins (719) are con-
tained in the SILAC-violin plot than in the SWATH-violin plot (2202) because the
SWATH-methodology eliminates undersampling. Furthermore, the precision of
the SWATH-technique (most frequent CV ~8 %) is far superior to the SILAC-
method (most frequent CV ~25 %)

a protocol to test for suitability of cells for SILAC-labeling using

SWATH-MS, which provides superior quantification results in
comparison to typical IDA-based methods.
Our lab usually prefers SWATH-MS over SILAC-labeling for
protein quantification from total cell culture lysates. This is not
primarily due to the higher cost of SILE, but to the avoidance of
missing-values and the superior precision of the quantification
(Fig. 1). The protein regulation values from the SWATH- and the
SILAC-based approach, however, are very similar (see Note 1).
Therefore, the presented protocol is well suited to test for suitabil-
ity of cell cultures for SILAC-labeling, e.g., for SILAC-pulse-chase
experiments [8] or protein interaction studies based on co-
immunoprecipitation [9] that are often poorly reproducible. The
superior precision of the SWATH-MS-approach can reveal system-
atic regulations induced by incorporation of the stable-isotope-
label, which might flaw the planned SILAC-based analysis (see
Note 2).
For the testing, the respective cell line is grown on “light” and
“heavy” SILAC medium at least in triplicate and analyzed using
the SWATH-methodology and pair-wise t-tests for the proteins in
the “light” and the “heavy” group are done. The results are cor-
rected for multiple testing using the procedure of Benjamini and
Hochberg [11]. The resulting proteins with a significant p-value
are hence regulated by the incorporation of the stable-isotope-
labeled amino acids.
104 Yvonne Reinders et al.

2 Materials

2.1 Cell Culture All items are given for the cell line tested in our lab and should be
for SILAC-Labeling adapted to the respective cell line to be tested (see Note 3).
1. HMB2 cell line.
2. Dulbecco’s Modified Eagle Medium (DMEM) without argi-
nine and lysine.
3. Penicillin and streptomycin.
4. L-glutamine.
13
5. C6,15N2-lysine/13C6,15N4-arginine and the respective unla-
beled lysine and arginine.
6. Dialyzed fetal calf serum (FCS).
7. Phosphate buffered saline (PBS).
8. Trypsin.
9. Lysis buffer: 20 mM sodium phosphate, pH 7.4, 0.2 % SDS.
10. Sonicator.
11. Kit for determination of total protein amount or material for
your method of choice.

2.2 Filter-Aided 1. Ultrafiltration devices (30 kDa, 500 μL).

Sample Preparation 2. 100 mM dithiothreitol (DTT).
[10]
3. 300 mM iodoacetamide (IAA).
4. Washing buffer: 8 M urea in 100 mM Tris–HCl, pH 8.5.
5. Digestion buffer: 50 mM ammonium bicarbonate, pH 7.8.
6. Trypsin Gold, mass spectrometry grade, Promega.
7. 5 % formic acid.

2.3 Nano-LC-MS/ 1. HPLC-solvents: A: 0.1 % formic acid; B: 0.1 % formic acid in

MS-Analysis acetonitrile.
2. Nano-HPLC-system with precolumn concentration.
3. Nano-ESI-QTOF-mass spectrometer of the TripleTOF-series.

2.4 Data Analysis 1. Protein Pilot-software (Sciex).

2. PeakView-software with SWATH-plugin (Sciex).
3. MarkerView-software (Sciex).

3 Methods

3.1 Cell Culture 1. Seed the cells into “light” and “heavy” SILAC DMEM media
for SILAC-Labeling supplemented with 10 % dialyzed FCS, 400 U/mL penicillin,
50 μg/mL streptomycin, and 300 μg/mL glutamine at the
required density (see Note 4).
 Testing Suitability of Cell Cultures for SILAC-Experiments Using SWATH-Mass… 105

2. Cultivate the cells for at least five cell doubling times with fresh
medium every other day to ensure almost complete labeling of
all proteins.
3. Check for equal growth of the cells in both media regularly.
4. Wash the cells three times with PBS before harvesting the cells.
5. Detach adherent cells using 500 μg/L trypsin.
6. Lyse the cells by sonication in 1 mL lysis buffer.
7. Determine protein concentrations by your method of choice.

3.2 Filter-Aided 1. Add 10 μL DTT solution to 100 μg protein lysate (1 μg/μL)

Sample Preparation and incubate for 30 min.
[10] (See Note 5) 2. Add 10 μL IAA solution and incubate at room temperature for
20 min in the dark.
3. Add 5 μL DTT solution and incubate at room temperature for
15 min.
4. Add 100 μL washing buffer and transfer the solution to an
ultrafiltration device.
5. Spin through and wash three times with 100 μL washing buf-
fer and additional three times with 100 μL digestion buffer.
6. Add 80 μL digestion buffer and 20 μL trypsin solution (0.1
μg/μL) and incubate at 37 °C over night.
7. Spin through and wash with 100 μL ultrapure water.
8. Acidify the eluate with 5 μL formic acid.

3.3 Nano-LC-MS/ HPLC settings can be changed according to the depth of the
MS-Analysis analysis that shall be achieved; in our lab 2–4 h runs are typically
accomplished. The HPLC conditions are the same for the SILAC-
runs with combined samples (also used for library generation) and
the SWATH-runs of the individual samples (see Note 6).
1. Load 1 μg of protein digest on a 2 cm, 100 μm I.D. C18-trapping
column (particle size 5 μm) with a flow rate of 5 μL/min.
2. Separate the samples on a 25 cm long 75 μm I.D. C18 column
(3 μm particle size) at a flow rate of 300 nL/min using a
212 min long gradient from 4 to 40 % B.
3. Operate the mass spectrometer at a TOP25 method (0.25 s for
TOF-MS from 350 to 1250 m/z, 0.1 s per MS/MS from 100
to 1500 m/z) for the SILAC-runs (see Note 7).
4. For the SWATH runs a 50 ms TOF-MS scan is followed by 37
windows of 25 m/z (100 ms per window) spanning the mass
range of 350–1250 m/z.

3.4 Data Analysis 1. Search the IDA-data using Protein Pilot against the
Uniprot-database.
2. Apply a 1 % false discovery rate (FDR) for the identifications.
106 Yvonne Reinders et al.

3. Use the SWATH-plugin in the PeakView-software to generate

a library and extract the quantitative SWATH-data using up to
six unique peptides per protein, six transitions per peptide and
a retention time window of 10 min.
4. Import the data into MarkerView and normalize to “Total
Area Sum”.
5. Generate two groups consisting of the “heavy” and the “light”
samples, respectively.
6. Do a pairwise-t-test and correct for multiple testing, e.g., by the
method of Benjamini and Hochberg [11] or Bonferroni [12].

4 Notes

1. Contaminating proteins, e.g., from residual fetal calf serum or

faulty sample handling, are more easily spotted in the SILAC-
approach using label switching which is strongly recommended
in all cases. Contaminations are not regulated in opposite direc-
tion upon switch of the labels and can thus be easily identified.
2. In principle, a test for changes introduced by the SILAC-
labeling can also be accomplished using a comprehensive
SILAC-analysis including a label switch. However, elimination
of undersampling issues by SWATH-mass spectrometry,
thereby avoiding the missing value problem and the increased
quantification precision for highly complex mixtures, result in
a higher sensitivity of the SWATH-based approach for detec-
tion of stable-isotope-labeling-induced alterations.
3. The presented protocol uses a melanoma cell line (HMB2) and
stable-isotope-coded lysine and arginine with a mass difference
of 8 Da and 10 Da, respectively, for duplex labeling. It is pos-
sible to use higher multiplexing by using for example 13C6-
lysine and 13C6-arginine for a further sample. Deuterated labels
may allow even higher multiplexing than 4-plex but deuterated
compounds show different retention times in reversed-phase
HPLC and are therefore not recommended.
It is also possible to use only one isotope-coded amino acid
for SILAC or use other amino acids than lysine and/or arginine,
e.g., leucine. However, a high degree of labeling is only obtained
with amino acids that are essential or only produced in negligi-
ble amounts by the cell. One has also to keep in mind that when
using only one amino acid for labeling not all peptides may be
used for the quantification, resulting in less precision.
Several SILAC-media are commercially available for differ-
ent cell types but assure that they are deficient of the amino
acid carrying the stable isotope-label, also assure the use of
dialyzed FCS to avoid incomplete labeling. Furthermore, do
not assume that proliferation rates are similar in standard
 Testing Suitability of Cell Cultures for SILAC-Experiments Using SWATH-Mass… 107

medium and SILAC medium as they might be substantially

different and should be determined for the SILAC media to
assure sufficient labeling efficiency.
4. The cells should be cultivated in at least three replicates per
label. Monitor also possible amino acid interconversions (e.g.,
arginine to proline) that lead to unexpected labeling of other
amino acids. Also other (essential) amino acids than lysine and
arginine maybe used for labeling but this type of labeling—in
combination with subsequent tryptic digestion—yields a pep-
tide mixture where virtually all peptides can be used for the
quantification.
5. Alternatively, also other sample preparation protocols can be
used. Gel separations for the generation of the library are also
possible and can increase the size of the library enormously
due to the enhanced separation. However, additional measure-
ment time is needed for this approach.
6. We used the IDA-runs of the combined SILAC-samples for
generation of the library because we wanted to compare the
regulation factors obtained from the IDA- and the SWATH-
measurements. The size of the library will typically increase if
non-combined samples are used for library generation as the
complexities of the non-combined samples are lower.
Furthermore, combination of the results of different IDA-runs
may increase library size. Moreover, newer versions of the
PeakView-software (> V2.0) are capable of aligning retention
times of the library and the SWATH-runs. Therefore, there is
no need to use the same gradient for the library generation and
the SWATH-measurements, although slightly wider retention
time windows may be necessary in the extraction.
7. Reduction of the total mass range of the SWATH-analysis that
is spanned by the different windows can be used to prolong the
accumulation time per window (thereby increasing signal-to-
noise-ratio) and/or reduction of duty cycle length (thereby
increasing the number of points across an LC peak). However,
distinct peptides will be lost if the mass range is restricted,
although the vast majority of signals are in the range of 400–
1000 m/z. Therefore, only very few proteins will escape quan-
tification by these means.

Acknowledgements

This work was supported by the “Deutsche Forschungsgemeinschaft

DFG” by the joint research programs “Klinische Forschergruppe
262” and “Sonderforschungsbereich 960,” the Wilhelm-Sander
foundation, the German Cancer Aid, and Bavarian Systems-Biology
Network (BioSysNet).
108 Yvonne Reinders et al.

References
1. Putz S, Reinders J, Reinders Y et al (2005) targeted proteomics. Curr Top Med Chem
Mass spectrometry-based peptide quantifica- 14(3):344–350
tion: applications and limitations. Expert Rev 7. Kiyonami R, Domon B (2010) Selected reac-
Proteomics 2(3):381–392 tion monitoring applied to quantitative pro-
2. Ong SE, Blagoev B, Kratchmarova I et al teomics. Methods Mol Biol 658:155–166
(2002) Stable isotope labeling by amino acids 8. Fierro-Monti I, Racle J, Hernandez C et al
in cell culture, SILAC, as a simple and accurate (2013) A novel pulse-chase SILAC strategy mea-
approach to expression proteomics. Mol Cell sures changes in protein decay and synthesis rates
Proteomics 1(5):376–386 induced by perturbation of proteostasis with an
3. Ong S, Mann M (2006) A practical recipe for Hsp90 inhibitor. PLoS One 8(11):e80423.
stable isotope labeling by amino acids in cell doi:10.1371/journal.pone.0080423
culture (SILAC). Nat Protoc 1(6):2650–2660. 9. Takahashi Y (2015) Co-immunoprecipitation
doi:10.1038/nprot.2006.427 from transfected cells. Methods Mol Biol
4. Neilson KA, Ali NA, Muralidharan S et al 1278:381–389. doi:10.1007/978-1-4939-
(2011) Less label, more free: approaches in 2425-7_25
label-free quantitative mass spectrometry. 10. Wisniewski JR, Zougman A, Nagaraj N et al
Proteomics 11(4):535–553. doi:10.1002/ (2009) Universal sample preparation method
pmic.201000553 for proteome analysis. Nat Methods 6(5):359–
5. Gillet LC, Navarro P, Tate S et al (2012) Targeted 362. doi:10.1038/nmeth.1322
data extraction of the MS/MS spectra generated 11. Benjamini Y, Hochberg Y (1995) Controlling
by data-independent acquisition: a new concept the false discovery rate: a practical and powerful
for consistent and accurate proteome analysis. approach to multiple testing. J Roy Statist Soc
Mol Cell Proteomics 11(6):O111.016717. Ser B 57(1):289–300. doi:10.2307/2346101
doi:10.1074/mcp.O111.016717 12. Dunn OJ (1961) Multiple comparisons among
6. Gomez YR, Gallien S, Huerta V et al (2014) means. J Am Stat Assoc 56(293):52–64.
Characterization of protein complexes using doi:10.1080/01621459.1961.10482090
 Chapter 9

Combining Amine-Reactive Cross-Linkers and Photo-

Reactive Amino Acids for 3D-Structure Analysis of Proteins
and Protein Complexes
Philip Lössl and Andrea Sinz

Abstract
During the last 15 years, the combination of chemical cross-linking and high-resolution mass spectrometry
(MS) has matured into an alternative approach for analyzing 3D-structures of proteins and protein com-
plexes. Using the distance constraints imposed by the cross-links, models of the protein or protein complex
under investigation can be created. The majority of cross-linking studies are currently conducted with
homobifunctional amine-reactive cross-linkers. We extend this “traditional” cross-linking/MS strategy by
adding complementary photo-cross-linking data. For this, the diazirine-containing unnatural amino acids
photo-leucine and photo-methionine are incorporated into the proteins and cross-link formation is
induced by UV-A irradiation. The advantage of the photo-cross-linking strategy is that it is not restricted
to lysine residues and that hydrophobic regions in proteins can be targeted, which is advantageous for
investigating membrane proteins. We consider the strategy of combining cross-linkers with orthogonal
reactivities and distances to be ideally suited for maximizing the amount of structural information that can
be gained from a cross-linking experiment.

Key words Chemical cross-linking, Photo-cross-linking, Photo-affinity labeling, Mass spectrometry,

Protein 3D-structure, Protein interactions, Unnatural amino acids

1 Introduction

In the era of systems biology one of the most important tasks

becomes the elucidation of protein networks in living cells.
Protein–protein interactions in living cells are usually identified
using affinity-based methods, such as co-immunoprecipitation or
tandem-affinity purification experiments [1]. Yet, as these methods
use cell lysates as starting material, the detection of false-positives
resulting from a disruption of protein complexes during the lysis
procedure presents a serious problem. Moreover, weakly binding
substances might get lost during the washing procedures. As an
alternative, chemical cross-linking can be employed to covalently
fix the interaction partners. Chemical cross-linking relies on the

Jörg Reinders (ed.), Proteomics in Systems Biology: Methods and Protocols, Methods in Molecular Biology, vol. 1394,
DOI 10.1007/978-1-4939-3341-9_9, © Springer Science+Business Media New York 2016

109
110 Philip Lössl and Andrea Sinz

introduction of a covalent bond between functional groups of

amino acids within a protein or between different interaction
partners by a chemical reagent [2–5]. After the cross-linking reac-
tion, the proteins of interest are usually enzymatically digested and
the resulting peptide mixtures are analyzed by high-resolution
mass spectrometry (MS) in a so-called “bottom-up” approach [6–8].
Analysis of cross-linked peptides makes use of several advantages
associated with MS analysis. First, the mass of the protein or the
protein complex under investigation is theoretically unlimited
because proteolytic peptide mixtures are analyzed. Second, analysis
is generally fast and, third, requires low amounts of protein. The
greatest challenge for using the cross-linking method is posed by
the high complexity of the created reaction mixtures requiring
high-resolution MS techniques for analyzing the cross-linked
products.
The functional groups of cross-linking reagents that are
commonly used for this technique are amine-reactive
N-hydroxysuccinimide (NHS) esters (Fig. 1). Also, photo-reactive

Fig. 1 Amine- and photo-reactive cross-linking of proteins. (a) Amine-reactive NHS esters typically connect
lysine residues (shown in blue and red). NHS esters are usually employed as equimolar mixture of non-
deuterated and deuterated derivatives to facilitate the identification of cross-linked products in the mass
spectra based on their characteristic isotope patterns. The cross-linker BS2G (black ) possesses a spacer
length of 7.7 Å, but Cα–Cα distances of up to 29 Å can be bridged [14]. (b) Photo-Leu (red-black ) as an
example for photo-reactive diazirines. Upon irradiation with UV-A light (~365 nm) a reactive carbene (free
electron pair shown in black ) is created that can insert into NH or CH groups (blue ). The carbene can also
create an alkene or react with water, resulting in the formation of an alcohol [30] (see Fig. 3)
 Cross-Linking/MS for Protein 3D-Structure Analysis 111

cross-linkers are increasingly used for photo-cross-linking (or

photo-affinity labeling) experiments [9]. The reliable identification
of cross-linked products is in most cases achieved by using isotope-
labeled, i.e., deuterated cross-linkers, such as BS3-D0/D4, BS2G-
D0/D4, or DSS-D0/D12 [10–13], that can bridge lysines with
Cα–Cα distances up to ca. 29 Å [14]. Photo-activatable cross-
linkers are particularly attractive because they allow cross-linking
to proceed in time- and location-specific ways. Here, a covalent
linkage is created between a protein and its binding partner upon
irradiation by UV-A light. The requirements for the ideal photoaf-
finity label include its chemical stability prior to photoactivation, its
photolysis at wavelengths, which do not cause photochemical
damage to the protein, as well as a high reactivity with C-H groups
and with nucleophilic X-H bonds. Reproducible UV-induced
labeling of target proteins is achieved by phenyl azides, diazirines,
and benzophenone photophores [15, 16].
One has to discriminate between three defined goals of cross-
linking studies: The first aim is to identify the complex-constituting
components by peptide mass fingerprint analysis, which is employed
for deriving information of protein interaction networks in cells.
The second goal goes beyond a mere identification of interaction
partners by pinpointing the interface regions within the protein
complexes based on the detected cross-linked amino acids. The
third goal is to employ the obtained distance constraint informa-
tion for creating structural models of the protein complexes with
computational methods, such as Rosetta [17, 18].
In this work, we extend the cross-linking tool box from the
exclusive use of amine-reactive cross-linkers towards the incorpora-
tion of the photo-reactive diazirine amino acids photo-methionine
(photo-Met) and photo-leucine (photo-Leu) that can deliver valu-
able short-distance information [19]. Combined with a mass spec-
trometric analysis of the created cross-links and computational
modeling, detailed insights can be obtained revealing details of the
interaction mechanisms between proteins. We exemplify this
approach of using cross-linkers with orthogonal reactivities and
distances for investigating the interactions between the basement
membrane proteins nidogen-1 and laminin γ1 [20, 21].

2 Materials

Prepare all solutions using ultrapure water, i.e., Milli-Q water, and
use reagents at the highest available purity. Amine-reactive chemi-
cal cross-linkers (NHS esters) should be stored dry under inert gas
(helium or nitrogen) and stock solutions have to be prepared in
dimethylsulfoxide (DMSO) directly before use. The photo-reactive
diazirine-containing amino acids L-photo-methionine and
L-photo-leucine can be commercially obtained from Thermo
Fisher Scientific and have to be stored under light exclusion.
112 Philip Lössl and Andrea Sinz

2.1 Chemical 1. Cross-linking reagent: BS2G-D0, and BS2G-D4 (Thermo

Cross-Linking Fisher Scientific; see Note 1).
2. Neat DMSO (see Note 2).
3. Cross-linking buffer: 20 mM HEPES, pH 7.4 (see Notes 3
and 4).
4. Quenching solution: 1 M ammonium bicarbonate.
5. Recombinant proteins are prepared in cross-linking buffer, the
optimum protein concentration should be between 1 and 10
μM (see Note 4).

2.2 Incorporation 1. Cell culture dishes and Triple Flasks (Thermo Fisher Scientific).
of Photo-reactive 2. Cell culture medium: Dulbecco’s Modified Eagle Medium
Amino Acids into DMEM/F12 (Gibco/Life Technologies), 10 % (v/v) FCS,
Proteins in HEK 293 1 % (v/v) penicillin/streptomycin solution and 1 μg/μL
Cells puromycin.
3. Dulbecco’s Modified Eagle Limiting Medium DMEM-LM,
leucine- and methionine-depleted (Gibco/Life Technologies).
4. Fetal calf serum (FCS, Biochrom).
5. Penicillin/streptomycin solution.
6. Puromycin.
7. Phosphate-buffered saline (PBS): 140 mM NaCl, 10 mM KCl,
8 mM Na2HPO4, 2 mM KHPO4, pH 7.4.
8. 0.05 % (w/v) trypsin–EDTA solution (T/E).
9. 2 M Tris–HCl, pH 8.0.
10. L-photo-leucine and L-photo-methionine (Thermo Fisher
Scientific).
11. HEK293 EBNA cells stably transfected with the protein-
coding DNA sequence (see Note 9).

2.3 Photochemical After incorporation of the photo-reactive amino acids photo-Met and
Cross-Linking photo-Leu, we conduct the photo-cross-linking reactions in a home-
built UV irradiation chamber [22]. The chamber contains fluores-
cence lamps (40 W; emission spectrum between 305 and 420 nm,
emission maximum at 360 nm (CLEO Performance R UV-A-
fluorescence lamp, Phillips)). The UV-A sensor for quantifying the
irradiation energy was purchased from Kühnast (see Note 7).

2.4 SDS– 1. Milli-Q H2O.

Polyacrylamide Gel 2. Mini-PROTEAN vertical gel electrophoresis system
Electrophoresis (Bio-Rad).
(SDS-PAGE)
3. Precise™ Protein Gel 4–20 % acrylamide (Thermo Fisher
Scientific).
4. Running buffer: 100 mM Tris–HCl, 100 mM HEPES, 0.1 %
SDS, pH 8.0.
 Cross-Linking/MS for Protein 3D-Structure Analysis 113

5. Laemmli buffer and 10 % (w/v) SDS solution (store at room

temperature, both Bio-Rad).
6. Ammonium persulfate (Sigma): Prepare a 10 % (w/v) solution
in Milli-Q H2O and immediately freeze in single-use (100 μL)
aliquots at −20 °C.
7. Page Ruler Protein Ladder (Fermentas).
8. Colloidal Coomassie staining solution: Solution A: 5 % (w/v)
Coomassie Brilliant Blue G250 suspended in Milli-Q
H2O. Solution B: 2 % (w/v) ortho-phosphoric acid, 10 %
(w/v) ammonium sulfate, dissolved in Milli-Q H2O.
9. Fixing solution: 40 % (v/v) methanol, 10 % (v/v) acetic acid.

2.5 Proteolytic In-Gel 1. Porcinetrypsin sequencing grade (Promega). Storage: dissolve

Digestion 25 μg lyophilized trypsin in 50 μL 1 mM HCl, prepare ali-
quots containing trypsin (1 μg/2 μL), quick-freeze in liquid
nitrogen and store at −80 °C.
2. GluC sequencing grade (Promega, see Note 14).
3. Chymotrypsin sequencing grade (Promega, see Note 14).
4. Acetonitrile (ACN) Hypersolv gradient grade (VWR, see
Note 15).
5. Destaining solutions: ACN, 50 % (v/v) ACN in 100 mM
ammonium bicarbonate (Sigma) (see Note 15).
6. Reducing buffer: freshly prepared 10 mM dithiothreitol (DTT)
in 100 mM ammonium bicarbonate.
7. Alkylating buffer: freshly prepared 55 mM iodoacetamide
(Sigma) in 100 mM ammonium bicarbonate.
8. Digestion buffer, freshly prepared: Dissolve one trypsin or
GluC aliquot in 80 μL 50 mM ammonium bicarbonate to
obtain a solution of 13 ng/μL, store on ice.
9. Extraction solution: 5 % (v/v) trifluoroacetic acid (TFA,
Sigma)–ACN 1:2 (v/v).

2.6 Proteolytic 1. Porcine trypsin sequencing grade (Promega). Storage: dissolve

In-Solution Digestion 25 μg lyophilized trypsin in 50 μL 1 mM HCl, prepare ali-
quots containing 1 μg/2 μL trypsin, quick-freeze in liquid
nitrogen and store at −80 °C.
2. GluC sequencing grade (Promega, see Note 14).
3. Chymotrypsin sequencing grade (Promega, see Note 14).
4. Denaturing solution: 1.6 M urea in 320 mM ammonium
bicarbonate.
5. Reducing buffer: freshly prepared 45 mM DTT in 400 mM
ammonium bicarbonate.
6. Alkylating buffer: freshly prepared 100 mM iodoacetamide
(Sigma) in 400 mM ammonium bicarbonate.
114 Philip Lössl and Andrea Sinz

7. TFA (Sigma).
8. Digestion buffer, freshly prepared: dissolve one trypsin, GluC
or chymotrypsin aliquot in 80 μL of 50 mM ammonium bicar-
bonate to obtain a solution of 13 ng/μL, store on ice.

2.7 Liquid In our lab, we use the following LC/MS setup, consisting of an
Chromatography/Mass Ultimate 3000 RSLC nano-HPLC system (Dionex) that is directly
Spectrometry (LC/MS) coupled to the nano-ESI source (EASY Spray source, Thermo
Fisher Scientific) of an Orbitrap Fusion Tribrid mass spectrometer
(Thermo Fisher Scientific) [23]. Alternatively, an LTQ-Orbitrap
XL mass spectrometer (Thermo Fisher Scientific) equipped with
nano-ESI source (Proxeon) is employed. Other nano-HPLC sys-
tems or mass spectrometers, such as FTICR or Q-TOF instru-
ments equipped with a nano-ESI source, can be used as well.

2.8 Software 1. Perform MS and MS/MS data acquisition and data analysis
for Data Acquisition with XCalibur 3.0 in combination with DCMS link 2.14.
and Analysis 2. Analyze cross-linked products with StavroX 3.4.5 (www.
StavroX.com) [24] that allows an automatic analysis and anno-
tation of MS and MS/MS data.
3. For obtaining optimum results, the following settings should be
used: Maximum mass deviation of 3 ppm between calculated and
experimental precursor masses and signal-to-noise ratio of ≥2.
4. Consider primary amino groups (Lys side chains and N-termini)
as well as hydroxyl groups (Thr, Ser, Tyr) as potential cross-
linking sites for BS2G [25, 26].
5. Consider all 20 amino acid residues as cross-linking sites for
UV-A-induced cross-linking of photo-Met and photo-Leu.
6. Oxidation of Met should be set as variable modification for all
cross-linked proteins.
7. Include carbamidomethylation as fixed modification for Cys.
8. Consider two missed cleavage sites for each amino acid (cleav-
age sites: Lys and Arg for trypsin; Tyr, Trp, and Phe for chy-
motrypsin; Glu and Asp for GluC).

3 Methods

3.1 Chemical 1. Prepare a mixture of your proteins of interest at a specific molar

Cross-Linking ratio (final concentrations ranging from 1 to 10 μM each) in
the cross-linking buffer (e.g., 20 mM HEPES). For equilibra-
tion of the protein complex, the mixture is incubated for ca.
15 min at room temperature using an incubation shaker. For
each cross-linking condition, a fresh Eppendorf tube should be
used. The minimum final volume should be 100 μL per reac-
tion condition.
 Cross-Linking/MS for Protein 3D-Structure Analysis 115

2. Freshly prepare cross-linker stock solutions (200 mM) in neat

DMSO (see Note 2).
3. Add 1:1 mixtures of non-deuterated (D0) and four times deu-
terated (D4) BS2G in 20- to 200-fold molar excess of cross-
linker over the protein. Allow the reaction to proceed between
5 and 60 min (see Notes 5 and 6).
4. For reaction quenching, add 1 M ammonium bicarbonate
solution (20 mM final concentration).
5. For storage, quick-freeze the samples in liquid nitrogen and
keep at −20 ° C.

3.2 Incorporation All steps that require manual handling of cultured cells must be
of Photo-reactive conducted under a laminar flow box. Solutions and materials have
Amino Acids into to be sterile-packaged or autoclaved prior to use.
Proteins 1. Add HEK293 EBNA cells to 10 mL of cell culture medium.
2. Culture the cells on a 57 cm2 cell culture dish in an incubator
with a water-saturated 5 % CO2 atmosphere at 37 °C.
3. Exchange the medium when the contained pH indicator turns
yellow indicating acidification.
4. Monitor the cell growth every other day by checking the cell
confluence under a microscope. Once a confluence of 80 % is
reached, passage the cells to new dishes:
(a) Wash the cells with 5 mL PBS.
(b) Add 1.5 mL T/E and incubate for 5 min at 37 °C to
cleave off the cells from the plate.
(c) Stop proteolysis by adding 8.5 mL of cell culture medium.
(d) Pellet the cells by centrifuging them for 5 min at 500 × g in
a 15-mL Falcon tube.
(e) Resuspend the cell pellet in 5 mL fresh cell culture
medium.
(f) Distribute the suspension on fresh cell culture dishes.
5. When the cells reach 80 % confluence again, replace the cell
culture medium with FCS-free medium (containing only
DMEM/F12, penicillin/streptomycin, and puromycin) and
incubate them for another 24 h (see Note 10).
6. Remove the FCS-free medium and add 10 mL of DMEM-LM
supplemented with 4 mM L-photo-Leu and 2 mM L-photo-
Met (see Notes 8 and 10).
7. Replace the medium after 2 days, storing the first batch of
supernatant at −20 °C.
8. Collect the second batch after 2 days of incubation.
9. Equilibrate the collected medium by adding Tris–HCl buffer
(pH 8) to a final concentration of 12 mM.
116 Philip Lössl and Andrea Sinz

10. Remove cells and other insoluble components by centrifuging

the equilibrated medium for 20 min at 14,000 × g and passing
the supernatant through a folded filter. Both steps should be
performed at 4 °C.
11. Add NaN3 to a final concentration of 0.02 % (w/v) to prevent
microbial growth in the supernatant.
12. Proceed with the chromatographic purification of the overex-
pressed protein (see Note 11).
13. Perform in-gel or in-solution digestion of the purified protein
(according to Subheading 3.4 or 3.5 or 3.6) and analyze the
peptide mixtures by LC/MS and LC/MS/MS. Estimate the
incorporation rate of the photo-reactive amino acids by com-
paring the MS intensities of the respective peptide signals con-
taining Leu/photo-Leu and Met/photo-Met (Fig. 2).

3.3 Photochemical 1. Prepare a mixture of your proteins of interest as described in

Cross-Linking Subheading 3.1, step 1. Samples should be shielded from
ambient light during incubation (see Note 8).
2. Transfer the reaction mixture into a transparent tube and expose
it to 8000 mJ/cm2 UV-A irradiation (see Notes 7 and 12).
3. For storage, quick-freeze the samples in liquid nitrogen and
keep at −20 °C.

3.4 SDS-PAGE 1. Load your samples on a Precise™ Protein Gel 4–20 % acryl-
amide gradient gel.
2. Perform the gel electrophoresis at a constant current of 25 mA.
3. After the electrophoresis is finished, keep the gel in fixing solu-
tion for 1 h.
4. Rinse the gel two times with Milli-Q H2O.
5. Stain the gel overnight in colloidal Coomassie staining
solution.
6. Destain the gel by shaking it in Milli-Q H2O.
7. Inspect the gel for signals that might represent the cross-linked
protein complex (Fig. 3).

3.5 In-Gel Digestion Working under a laminar flow box wearing gloves and sleeve pro-
tectors is recommended in order to avoid keratin and dust con-
tamination of samples. The laminar flow box should include an
incubation shaker suitable for Eppendorf tubes.

3.5.1 Excise Protein 1. Wash the Coomassie-stained gel twice for 10 min with Milli-Q
Bands from the SDS Gel H2O on a shaker.
2. Excise gel bands of interest using a sterile scalpel and transfer
gel bands to a clean glass plate. Cut the gel band into cubes of
ca. 1 mm3 and transfer them into an Eppendorf tube.
 Cross-Linking/MS for Protein 3D-Structure Analysis 117

Fig. 2 Incorporation of photo-reactive amino acids into nidogen-1 and laminin

γ1. (a) Met and Leu variants that were considered during MS analysis, including
the reaction products of the diazirine amino acids photo-Met and photo-Leu (1:
photo-Leu, 5: photo-Met, 2 and 6: alkene; 3 and 7: alcohol; 4: unmodified Leu; 8
and 9: unmodified and oxidized Met); (b and c) MS-based quantification of
photo-reactive amino acid incorporation. The pie charts show the number of
leucines (blue) and methionines (red) within nidogen-1 (b) and laminin γ1 short
arm (c) that remained unmodified (light shades) or were partially replaced by
their photo-reactive counterparts (dark shades). The bars represent the relative
abundance of isoforms of the partially modified peptides, containing the Leu
(dark blue) and Met (dark red) variants listed in (a)

3.5.2 Reduce 1. Add 100 μL Milli-Q H2O to the gel pieces, shake for 10 min.
and Alkylate Proteins 2. Add 500 μL ACN. Shake for 5 min and discard the liquid.
3. Add 50 μL reduction buffer. Incubate at 56 °C for 30 min,
then let the solution cool down to room temperature.
118 Philip Lössl and Andrea Sinz

Fig. 3 SDS-PAGE analysis of nidogen-1 and laminin γ1 short arm after cross-
linking with BS2G. Nidogen-1 and laminin γ1 short arm were cross-linked both
separately and after co-incubation at molar ratios of 1:1 and 1:2. M molecular
weight marker

4. Add 500 μL ACN. Shake for 5 min and discard the liquid.
5. Add 50 μL alkylation buffer. Incubate at room temperature in
the dark for 30 min.
6. Add 500 μL ACN. Shake for 5 min and discard the liquid.

3.5.3 Wash and Destain 1. Add 100 μL ACN/100 mM ammonium bicarbonate (1:1),
Gel Pieces shake for 10–30 min to destain the gel pieces.
2. Add 500 μL ACN. Shake for 5 min and discard the liquid.
3. Repeat steps 1 and 2 if necessary.

3.5.4 In-Gel Digestion 1. Bring the gel pieces to complete dryness before adding the
digestion buffer (protease in ammonium bicarbonate).
2. Per gel band, add 25 μL trypsin digestion buffer and incubate
gel pieces for 1 h on ice. In case a double digestion is per-
formed, add GluC (or another protease) first and incubate the
gel pieces for 1 h on ice, then add trypsin and incubate for
another 1 h on ice.
3. Add 50 mM ammonium bicarbonate. The gel pieces have to
be covered completely.
4. Incubate overnight at 37 °C.

3.5.5 Extract Peptides 1. Stop the digestion by adding the double volume of extraction
solution.
2. Shake for 15 min at 37 °C, remove the supernatant, and trans-
fer it to a new Eppendorf tube.
 Cross-Linking/MS for Protein 3D-Structure Analysis 119

3. Repeat steps 1 and 2 if the gel band is expected to contain a

low amount of protein.
4. Concentrate supernatants in a vacuum concentrator to a vol-
ume between 60 and 120 μL. Do not concentrate to complete
dryness in order to avoid sample loss.
5. Peptide mixtures are ready to be analyzed by nano-HPLC/
nano-ESI-Orbitrap mass spectrometry. Before MS analysis,
samples may be stored at −20 °C.

3.6 In-Solution 1. Incubate proteins for 1 h with a 20-fold excess (v/v) of pre-
Digestion cooled acetone to precipitate them from solution.
3.6.1 Denature Proteins 2. Centrifuge the sample for 30 min at 15,000 × g (4 °C).
3. Let the pellet dry at room temperature.
4. Solubilize the pellet in 25 μL denaturing solution.

3.6.2 Reduce 1. Add 7.5 μL of reduction buffer. Incubate at 50 °C for 15 min,

and Alkylate Proteins then allow the solution to cool down to room temperature.
2. Add 5 μL alkylation buffer. Incubate at room temperature in
the dark for 15 min.

3.6.3 In-Solution 1. Dilute the sample with 60 μL Milli-Q H2O to avoid denatur-
Digestion ation of the proteases by urea.
2. Add trypsin digestion buffer to an enzyme/protein ratio of
1:50 (v/v) and incubate the sample at 37 °C for 2 h. In case a
double digestion is performed, add GluC (or another prote-
ase) first and incubate for 1 h at room temperature, then add
trypsin and incubate at 37 °C for 2 h.
3. Acidify the sample to a pH value below 4 (check with pH
paper) and concentrate it in a vacuum concentrator to a vol-
ume between 60 and 120 μL.
4. Peptide mixtures are ready to be analyzed by nano-HPLC/
nano-ESI-Orbitrap mass spectrometry. Before MS analysis,
samples may be stored at −20 °C.

3.7 Liquid 1. For LC/MS analysis of the cross-linked peptide mixtures, we

Chromatography/Mass use the nano-HPLC/nano-ESI-Orbitrap-MS setup described
Spectrometry (LC/MS) in Subheading 2.7.
2. Separate the enzymatically digested cross-linked peptide mix-
ture by reversed-phase LC. For this, inject the peptide mixture
via an autosampler and load them onto a precolumn (Acclaim
PepMap, RP C18, 5 mm × 300 μm, 5 μm, 100 Å).
3. Concentrate and desalt samples during a 15-min washing step
with 0.1 % TFA.
120 Philip Lössl and Andrea Sinz

4. Separate the peptides by gradient elution (1 % to 40 % B; A:

0.1 % formic acid (FA), B: ACN, 0.08 % FA) over 90 min using
a PepMap RSLC C18 column (250 mm × 75 μm, 2 μm, 100
Å) at a flow rate of 300 nL/min.
5. Introduce the peptides into the nano-ESI source (EASY Spray
source) of the mass spectrometer. Use a fused-silica emitter at
a source voltage of 1.9 kV and set the temperature of the trans-
fer capillary to 275 °C.
6. Record the mass spectra automatically in data-dependent MS/
MS mode during gradient elution (Orbitrap Fusion instru-
ment): For each 5-s cycle, one mass spectrum (R = 120,000 at
m/z 200) is recorded in the m/z range 350–1500 in the
Orbitrap analyzer. The most abundant species are isolated in
the quadrupole (isolation window 2 Th or larger in case deu-
terated amine-reactive cross-linkers, such as BS2G-D0/D4 are
used) and fragmented with collision-induced dissociation
(CID) at normalized collision energies of 25 % and 35 %.
Detection of the fragment ions is performed in the Orbitrap
(R = 15,000 at m/z 200) using dynamic exclusion for 90 s.

3.8 Data Analysis Data analysis of cross-linked peptides is performed with the StavroX
software allowing lysines and protein N-termini as reaction sites of
NHS-ester based cross-linkers (Fig. 1). As NHS-esters have also
been found to react with serines, threonines, and tyrosines, these
reaction sites should also be taken into account [25, 26]. For
photo-cross-links induced by diazirines, all 20 amino acids were
considered as potential reaction sites (see Note 13, Fig. 1).
Please note that trypsin will cleave at modified lysines with low
frequency, therefore the number of missed cleavage sites has to be
set to a higher value for cross-link identification. Filtering of putative
cross-linked candidates is facilitated by the application of isotope-
labeled D0/D4 cross-linkers, which generate a characteristic isotope
pattern for cross-linked peptides. To identify this isotope pattern,
experimental mass lists obtained by high-resolution mass spectrom-
etry are filitered for monoisotopic deconvoluted masses (using the
Proteome Discoverer software, Thermo Fisher Scientific) that show
a mass difference of 4 amu. This reduced experimental mass list is
then compared to a theoretical mass list of all possibly cross-linked
peptides, allowing a maximum mass deviation of 3 ppm. This allows
automated identification of potentially cross-linked peptides, for
which MS/MS data have to be evaluated. Both the D0 and the D4
cross-linked species have to be isolated and fragmented, so the MS/
MS data can also be checked for the characteristic isotope patterns
(see Subheading 3.7, step 6). To further automate the MS/MS-based
identification of cross-linked species, the StavroX software calculates
the corresponding theoretical fragment ions and compares them to
the experimental mass spectra (Fig. 4). Subsequently, StavroX assigns
 Cross-Linking/MS for Protein 3D-Structure Analysis 121

Fig. 4 Reaction products of amine- and photo-reactive cross-linking analyzed by nano-HPLC/nano-ESI-LTQ-

Orbitrap MS/MS. (a) Two nidogen-1 peptides cross-linked by BS2G. The reaction product comprises the amino
acids 407–420 (α-peptide, red) and 939–949 (β-peptide, blue ), in which Lys-407 is connected to Lys-948/949.
(b) A photo-Leu (L*)-containing laminin γ1 peptide (amino acids 988–994, red ) cross-linked to a nidogen-1
peptide (amino acids 1033–1039, blue). Based on the detected b- and y-type ions that are created by CID, the
exact site of cross-linking can be mapped to Arg-1038 in nidogen-1
122 Philip Lössl and Andrea Sinz

a score for cross-linked products based on the quality of MS/MS

spectra and calculates false discovery rates (FDR). Nevertheless, it is
recommended to validate all MS and MS/MS data manually.

3.9 Use Cross-Links 1. Make sure that the numbering of the identified cross-links
as Distance complies with the atom numbering in the PDB files and/or
Constraints amino acid sequences used as input for the modeling process
for Rosetta-Based (see Note 16).
Computational 2. List all identified cross-links in a tab-separated constraint file,
Modeling as exemplified in Table 1 (see Note 17). Constraints obtained
from different cross-linking experiments may be listed in the
same file.
3. Enforce the specified distance constraints by the command line
flag “-constraints:cst_fa_file</path/to/constraint/file>”, if
working with high-resolution models where all amino acid side
chain atoms are modeled, or “-constraints:cst_file</path/to/co
nstraint/file>”, if working with low-resolution models where
amino acid side chains are represented by a centroid sphere.

Table 1
Overview of cross-links in the structurally solved region of the nidogen-1/laminin γ1 short arm
complex. Cα–Cα distances of cross-linked residues indicated in the crystal structure (PDB entry
1NPE) and in the best-scoring models created by Rosetta

Cα–Cα distances (Å)

Model (best atom-pair Model (best

Cross-linked lysines 1NPE constraint score) total score)
K-948 × K-953 10.4 10.9 11.1
K-1128 × K-1165 13.3 12.3 16.0
K-1072 × K-1128 16.7 19.1 16.2
K-948 × K-1144 17.9 16.4 17.6
K-850 (laminin) × K-1072 (nidogen-1) 20.9 17.5 16.9
K-948 × K-1152 22.2 21.2 22.2
K-1032 × K-1072 27.1 27.1 27.0
K-961 × K-1072 28.7 28.0 28.2
K-864 (laminin) × K-1152 (nidogen-1) 32.2 22.4 27.1
K-850 (laminin) × K-953 (nidogen-1) 33.0 29.5 29.4
K-1032 × K-1152 35.8 35.4 35.4
Photo-L-990 × R-1038 24.7 23.4 23.5
Photo-L-844 (laminin) × K-1072 (nidogen-1) 33.8 19.4 20.8
 Cross-Linking/MS for Protein 3D-Structure Analysis 123

Fig. 5 Incorporation of cross-linking distance constraints into X-ray structures

using the Rosetta Relax application. Zoom-in on the high-affinity binding region
of nidogen-1 and laminin γ1 short arm (PDB entry 1NPE). Two cross-links (black
dashed lines), one detected with BS2G (Cα–Cα distance 20.9 Å) and one detected
after photochemical cross-linking (Cα–Cα distance 33.8 Å), were mapped in the
crystal structure (gray cartoon representation ). The distance constraint imposed
by the photochemical cross-link is clearly violated. In the Rosetta model (blue
cartoon representation ), which was built based on the crystal structure and the
experimentally identified cross-links, the Cα–Cα distances of the cross-linked
residues are substantially reduced (red dashed lines ), while secondary structure
elements and disulfide bridges are retained as indicated in the crystal structure.
Thus, the model reflects a plausible in solution conformation of the protein com-
plex (see also Table 1)

4. Set the impact of constraint violation on the total score of

the created models by means of the command line flags
“-constraints:cst_fa_weight < number>” or “-constraints:cst_
weight<number>”, respectively (see Note 18).
5. Analyze the created models to check for steric clashes and
constraint violations. If necessary, adjust the constraint weight
(see Note 18, Fig. 5).

4 Notes

1. NHS ester cross-linkers are highly sensitive to hydrolysis.

Therefore, aliquots of the cross-linkers should be prepared and
stored in a desiccator under inert gas, e.g., helium or
nitrogen.
2. Neat DMSO is required to prepare the NHS cross-linker stock
solutions immediately before use in order to prevent hydrolysis
of the reagents.
124 Philip Lössl and Andrea Sinz

3. The pH value of the protein solution should range between ca.

7.0 and 8.5. Please note that an increased pH value will increase
the reactivity of amine groups towards NHS esters, but pH
values higher than 8.5 should be avoided due to potential sta-
bility problems of certain proteins.
4. Tris buffers interfere with amine-reactive cross-linking reagents
and should be avoided. Please note that amine-reactive NHS
esters also react with hydroxyl groups of serines, threonines,
and tyrosines, albeit with a lower frequency compared to amine
groups [25, 26]. Considering these amino acids as potential
cross-linking sites will increase the amount of information you
get from your cross-linking experiment.
5. Using equimolar mixtures of non-deuterated (D0) and deuter-
ated (D4) cross-linking reagents facilitates the identification of
cross-linking products in mass spectra by characteristic mass
shifts of 4 amu. Please note that deuterated and non-deuterated
species exhibit different retention times in reversed phase-LC,
so they may not coelute.
6. For chemical cross-linking with NHS esters, the following
parameters are recommended to be optimized for obtaining
maximum yields of cross-linked products:
(a) Molar excess of cross-linker (20- to 200-fold).
(b) Cross-linking time: Allow the cross-linking reaction to
proceed between 5 and 60 min at room temperature. If
the proteins of interest are not sufficiently stable at room
temperature, conduct the cross-linking reaction at 4 °C
for 2 h.
7. For photochemical cross-linking, a UV lamp should be used
with a filter blocking wavelengths lower than ca. 300 nm in
order to avoid photolytic damage of the protein. During UV-A
irradiation, the protein solution should be kept on ice.
8. Avoid the exposure of the samples to ambient light when con-
ducting the photo-cross-linking reactions, i.e., by using light-
protected reaction tubes or by wrapping aluminum foil around
the reaction tubes. Also, protein production and purification
should be conducted under low-light conditions.
9. Protein constructs must comprise an export sequence to facili-
tate their secretion into the cell culture medium.
10. Incorporation of photo-reactive amino acids is straightforward
since it is sufficient to add them to the leucine- and methionine-
depleted cell culture media [27]. While the incorporation effi-
ciency is acceptable for photo-Met, yields are comparably low
for photo-Leu (Fig. 2). To increase the incorporation effi-
ciency, cells can be temporarily cultured in FCS-free medium
before transferring them to photo-reactive amino acid-
containing medium, thereby ensuring full depletion of Met
 Cross-Linking/MS for Protein 3D-Structure Analysis 125

and Leu. However, a prolonged exposure to FCS-free medium

will negatively affect the cell survival rate.
11. To the best of our knowledge, all types of affinity purification
are compatible with this protocol. Details of the protein puri-
fication are not described herein as the purification procedure
applied varies for the respective protein construct.
12. The optimum UV irradiation energy might be different for
other protein systems. We recommend to optimize the
procedure by testing different exposure times, especially when
the irradiation device is not equipped with a UV-A sensor.
13. Analysis of the cross-linking datasets with the StavroX software
[24] can become quite time-consuming—especially when
using photo-reactive amino acids—as a high number of amino
acids have to be considered as potential reaction sites.
14. Trypsin cleaves with lower frequency at lysines that are modi-
fied by cross-linkers. Therefore, a second protease is required
to achieve high proteolytic digestion yields, e.g., GluC (cleaves
C-terminally of glutamic and also aspartic acid). As an alterna-
tive to trypsin, chymotrypsin (cleaves C-terminally of large and
hydrophobic amino acids) might be used, which is especially
beneficial for membrane proteins.
15. Organic solvents must not be kept in plastics due to a potential
contamination by polymers.
16. Constraint files are compatible with multiple Rosetta applica-
tions allowing, for example, protein-protein docking, compar-
ative modeling, and structural optimization based on
cross-linking distance constraints (Fig. 5, Table 1). It is beyond
the scope of this protocol to describe these procedures in
detail. A protocol on Rosetta-based comparative modeling has
been published recently [28]. The incorporation of cross-
linking constraints into Rosetta workflows has been described
elsewhere, including detailed command line flags [18, 19, 29].
For further information on the installation and usage of
Rosetta, please consult https://www.rosettacommons.org/
docs/latest/.
17. Atom-pair constraints allow specifying experimentally found
distances between two atoms, e.g., Cα–Cα distances derived
from cross-linking data. Models violating these constraints will
be penalized according to the applied function, i.e., their total
energy score will increase. The “flat_harmonic” function ren-
ders an energy penalty if the modeled Cα–Cα distance exceeds
the sum of x0 and the granted tolerance (Table 2). The respec-
tive penalty calculates to
C a - C a distance - x 0 - tolerance
atom-pair constraint score =
standard deviation
126 Philip Lössl and Andrea Sinz

Table 2
Information to be included in a Rosetta constraint file. The file should be tab-separated and must not
include any column headings. All listed information has to be in accordance with the input PDB file

Constraint Atom Res no 1/ Atom Res no. 2/

type name 1 chain ID1 name 2 chain ID2 Function type x0 SD Tolerance
Atom pair CA 1A CA 1B FLAT_HARMONIC 20.3 1.0 5.7
Res no residue number, chain ID chain identifier, x0 + tolerance maximum Cα–Cα distance, SD standard deviation

18. Increasing the constraint weight and decreasing the standard

deviation will lead to a higher energy penalty for constraint
violation, favoring models that fulfill all experimental con-
straints. Careful optimization of both settings is recom-
mended since too harsh penalties entail the danger that
realistic models, which marginally exceed the distance limits,
are overseen during analysis. Of note, forcing Rosetta to
comply with cross-links that turn out to be sterically impossible
may lead to a distortion of the model and a loss in secondary
structure [19].

Acknowledgements

This work is funded by the BMBF (Pronet-T3). We would like to

thank Dr. Christian Ihling for LC/MS analyses, Dr. Knut Kölbel
for fruitful discussions, and Dirk Tänzler for excellent technical
support. Prof. Jens Meiler and Dr. David Nanneman, Vanderbilt
University, are acknowledged for introducing P.L. into Rosetta,
Prof. Gunter Fischer and Dr. Cordelia Schiene-Fischer, Max-
Planck Forschungsstelle Halle, are acknowledged for generously
providing their cell culture facilities. We are indebted to Prof. Mats
Paulsson and Dr. Frank Zaucke; University of Cologne, for their
help with laminin and nidogen expression and purification.

References
1. Gavin A, Bösche M, Krause R et al (2002) 4. Tang X, Bruce JE (2010) A new cross-linking
Functional organization of the yeast proteome by strategy: protein interaction reporter (PIR) tech-
systematic analysis of protein complexes. Nature nology for protein-protein interaction studies. Mol
415(6868):141–147. doi:10.1038/415141a Biosyst 6(6):939–947. doi:10.1039/b920876c
2. Petrotchenko EV, Borchers CH (2010) 5. Calabrese AN, Pukala TL (2013) Chemical
Crosslinking combined with mass spectrome- cross-linking and mass spectrometry for the
try for structural proteomics. Mass Spectrom structural analysis of protein assemblies. Aust
Rev 29(6):862–876. doi:10.1002/mas.20293 J Chem 66(7):749. doi:10.1071/CH13164
3. Sinz A (2014) The advancement of chemical 6. Kalisman N, Adams CM, Levitt M (2012)
cross-linking and mass spectrometry: from sin- Subunit order of eukaryotic TRiC/CCT chap-
gle proteins to protein interaction networks. eronin by cross-linking, mass spectrometry,
Expert Rev Proteomics 11:733–743 and combinatorial homology modeling. Proc
 Cross-Linking/MS for Protein 3D-Structure Analysis 127

Natl Acad Sci U S A 109(8):2884–2889. 19. Lössl P, Kölbel K, Tänzler D et al (2014)

doi:10.1073/pnas.1119472109 Analysis of nidogen-1/laminin γ1 interaction
7. Chen ZA, Jawhari A, Fischer L et al (2010) by cross-linking, mass spectrometry, and com-
Architecture of the RNA polymerase II-TFIIF putational modeling reveals multiple binding
complex revealed by cross-linking and mass modes. PLoS One 9:e112886
spectrometry. EMBO J 29(4):717–726. 20. Paulsson M, Aumailley M, Deutzmann R et al
doi:10.1038/emboj.2009.401 (1987) Laminin-nidogen complex. Extraction
8. Leitner A, Joachimiak LA, Bracher A et al (2012) with chelating agents and structural character-
The molecular architecture of the eukaryotic ization. Eur J Biochem 166(1):11–19
chaperonin TRiC/CCT. Structure 20(5):814– 21. Takagi J, Yang Y, Liu J et al (2003) Complex
825. doi:10.1016/j.str.2012.03.007 between nidogen and laminin fragments
9. Tanaka Y, Bond MR, Kohler JJ (2008) reveals a paradigmatic beta-propeller interface.
Photocrosslinkers illuminate interactions in Nature 424(6951):969–974. doi:10.1038/
living cells. Mol Biosyst 4(6):473–480. nature01873
doi:10.1039/b803218a 22. Schaks S, Maucher D, Ihling CH et al (2012)
10. Müller DR, Schindler P, Towbin H et al (2001) Investigation of a calmodulin/peptide complex
Isotope-tagged cross-linking reagents. A new by chemical cross-linking and high-resolution
tool in mass spectrometric protein interaction mass spectrometry. In: Koenig S (ed)
analysis. Anal Chem 73(9):1927–1934 Biomacromolecular mass spectrometry: tips
11. Rinner O, Seebacher J, Walzthoeni T et al from the bench. Nova Science, Hauppauge,
(2008) Identification of cross-linked peptides NY, pp 1–18
from large sequence databases. Nat Methods 23. Senko MW, Remes PM, Canterbury JD et al
5(4):315–318. doi:10.1038/nmeth.1192 (2013) Novel parallelized quadrupole/linear
12. Schmidt A, Kalkhof S, Ihling C et al (2005) ion trap/Orbitrap tribrid mass spectrometer
Mapping protein interfaces by chemical cross- improving proteome coverage and peptide
linking and Fourier transform ion cyclotron identification rates. Anal Chem 85(24):11710–
resonance mass spectrometry: application to a 11714. doi:10.1021/ac403115c
calmodulin/adenylyl cyclase 8 peptide com- 24. Götze M, Pettelkau J, Schaks S et al (2012)
plex. Eur J Mass Spectrom (Chichester, Eng) StavroX—a software for analyzing crosslinked
11(5):525–534. doi:10.1255/ejms.748 products in protein interaction studies. J Am
13. Schulz DM, Kalkhof S, Schmidt A et al (2007) Soc Mass Spectrom 23(1):76–87. doi:10.1007/
Annexin A2/P11 interaction: new insights into s13361-011-0261-2
annexin A2 tetramer structure by chemical 25. Mädler S, Bich C, Touboul D et al (2009)
crosslinking, high-resolution mass spectrome- Chemical cross-linking with NHS esters: a sys-
try, and computational modeling. Proteins tematic study on amino acid reactivities. J Mass
69(2):254–269. doi:10.1002/prot.21445 Spectrom 44(5):694–706. doi:10.1002/
14. Merkley ED, Rysavy S, Kahraman A et al jms.1544
(2014) Distance restraints from crosslinking 26. Kalkhof S, Sinz A (2008) Chances and pitfalls
mass spectrometry: mining a molecular dynam- of chemical cross-linking with amine-reactive
ics simulation database to evaluate lysine-lysine N-hydroxysuccinimide esters. Anal Bioanal
distances. Protein Sci 23(6):747–759. Chem 392(1–2):305–312
doi:10.1002/pro.2458 27. Suchanek M, Radzikowska A, Thiele C (2005)
15. Brunner J (1993) New photolabeling and cross- Photo-leucine and photo-methionine allow
linking methods. Annu Rev Biochem 62:483– identification of protein-protein interactions in
514. doi:10.1146/annurev.bi.62.070193.002411 living cells. Nat Methods 2(4):261–267.
16. Weber PJ, Beck-Sickinger AG (1997) doi:10.1038/nmeth752
Comparison of the photochemical behavior of 28. Combs SA, Deluca SL, Deluca SH et al (2013)
four different photoactivatable probes. J Pept Small-molecule ligand docking into comparative
Res 49(5):375–383 models with Rosetta. Nat Protoc 8(7):1277–
17. Doberenz C, Zorn M, Falke D et al (2014) 1298. doi:10.1038/nprot.2013.074
Pyruvate formate-lyase interacts directly with 29. Kahraman A, Herzog F, Leitner A et al (2013)
the formate channel FocA to regulate formate Cross-link guided molecular modeling with
translocation. J Mol Biol 426(15):2827–2839. ROSETTA. PLoS One 8(9):e73411.
doi:10.1016/j.jmb.2014.05.023 doi:10.1371/journal.pone.0073411
18. Herzog F, Kahraman A, Boehringer D et al 30. Kölbel K, Ihling CH, Sinz A (2012) Analysis of
(2012) Structural probing of a protein phos- peptide secondary structures by photoactivat-
phatase 2A network by chemical cross-linking able amino acid analogues. Angew Chem Int
and mass spectrometry. Science 337(6100): Ed Engl 51(50):12602–12605. doi:10.1002/
1348–1352. doi:10.1126/science.1221483 anie.201205308
 Chapter 10

Tissue MALDI Mass Spectrometry Imaging (MALDI MSI)

of Peptides
Birte Beine*, Hanna C. Diehl*, Helmut E. Meyer, and Corinna Henkel

Abstract
Matrix assisted laser desorption/ionization mass spectrometry imaging (MALDI MSI) is a technique to
visualize molecular features of tissues based on mass detection. This chapter focuses on MALDI MSI of
peptides and provides detailed operational instructions for sample preparation of cryoconserved and
formalin-fixed paraffin-embedded (FFPE) tissue. Besides sample preparation we provide protocols for the
MALDI measurement, tissue staining, and data analysis. On-tissue digestion and matrix application are
described for two different commercially available and commonly used spraying devices: the SunCollect
(SunChrom) and the ImagePrep (Bruker Daltonik GmbH).

Key words MALDI imaging, MSI, Peptide, Trypsin, Matrix application, ImagePrep, SunCollect,
Formalin-fixed paraffin-embedded (FFPE), Cryoconserved

1 Introduction

Mass Spectroscopy Imaging (MSI)started off in 1962 with the

usage of secondary ion mass spectrometry (SIMS) as ion source
[1]. The term MSI generally encompasses SIMS, desorption elec-
trospray ionization (DESI) and matrix assisted laser desorption/
ionization (MALDI) mass spectrometry (MS) whereas all three
ionization techniques provide insights into the molecular content
of an intact piece of tissue while preserving the spatial resolution
[2]. In case of MALDI MSI a laser with a micron dimension
between 10 and 150 μm rasters over a predefined area of a tissue
covered with matrix. An individual mass spectrum consisting of
several mass-to-charge (m/z) ratios and intensities is generated at
each position. A color scale is used to represent the signal intensity
at each position. Basically three different mass analyzers (time of
flight (TOF), quadrupole ion trap and Fourier transform ion cyclo-
tron resonance (FT-ICR)) are used in combination with the

* The authors contributed equally to the manuscript.

Jörg Reinders (ed.), Proteomics in Systems Biology: Methods and Protocols, Methods in Molecular Biology, vol. 1394,
DOI 10.1007/978-1-4939-3341-9_10, © Springer Science+Business Media New York 2016

129
130 Birte Beine et al.

mentioned ion sources. Different classes of molecules have been

analyzed by MALDI MSI so far as for example proteins [3], pep-
tides [4], lipids [5], and metabolites [6]. The possibility to acquire
spatially resolved mass spectra directly is a big advantage in com-
parison to tedious laser microdissection of defined tissue areas of
interest. MALDI MSI is a complementary method to liquid chro-
matography (LC)-MS approaches and for example protein arrays
and can efficiently close the gap between histology and mass
spectrometry.
Several different workflows for MALDI MSI were established
in the last years and methods get more and more precise regarding
preparation techniques and laser focus [6, 7]. A perfect tissue prep-
aration for peptide MALDI MSI would consist of small trypsin
droplets and small matrix crystals to result in optimal spatial reso-
lution. Several papers in the beginning of the 2000th century
describe experiments with a raster width of 200 or even 300 μm
[8]. Over the years methods have changed and other lasers as for
example the Smartbeam laser were generated to achieve a better
performance with regard to spatial resolution. Additionally the way
of trypsin and matrix application significantly influences the limita-
tion of spatially resolved images, e.g., due to large matrix crystals
which lower the overall resolution or trypsin droplets bigger than
100 μm [9, 10] which in return limit the spatial resolution too.
Those problems can be avoided by using automatic sample prepa-
ration devices as for example the ImagePrep or SunCollect, which
both work with a defined nebulized spray of trypsin and matrix,
resulting in small trypsin droplets and matrix crystals. An alterna-
tive are spotting instruments with a raster width of about 100 μm
or more [9], whereas the resolution is mainly limited by the spotter
itself, because the distance from one spot to the next is quiet large.
Nevertheless the advantage is to deposit a large amount of trypsin
at one defined position and thus obtain better digestion efficiency.
Another option to apply the matrix is sublimation by means of dry
matrix application [11] but then trypsin still has to be sprayed or
spotted so that only trypsin deposition determines the spatial
resolution.
Most people in the community tend to work with automatic
spraying devices self-made or provided by a company. For peptide
MALDI MSI experiments the well-known matrices HCCA
(α-cyano-4-hydroxycinnamic acid) or DHB (2,5-dihydroxybenzoic
acid) are used. HCCA is slightly preferred because of its small crys-
tal size [7, 12].
There are numerous different peptide MALDI MSI protocols
published which makes it hard to define a standard procedure. The
workflow generally consists of different washing and drying steps
(or deparaffinization and antigen retrieval for FFPE tissue), trypsin
application and incubation under humid and warm conditions for
 Tissue MALDI Mass Spectrometry Imaging (MALDI MSI) of Peptides 131

digestion, matrix application, measurement and data analysis.

Ideally one would optimize all these different steps for every used
tissue type. Diehl and coworkers performed an extensive experi-
mental setup with 69 rat brains, to test different digestion times,
different matrices, and quality of matrix and enzyme [7]. According
to their results and other published data washing steps, type of
enzyme, digestion time, matrix application, and type of matrix are
major factors, which can essentially influence the outcome of your
experiment [13].
This chapter focuses on the use of MALDI MSI for the analysis
of peptides from fresh frozen and formalin-fixed paraffin-embedded
(FFPE) tissue samples and provides operation procedures from
sample preparation to measurement and data analysis. Moreover a
list of necessary material and reagents will be given and notes on
possible pitfalls to be avoided.

2 Materials

Prepare all solutions using ultrapure water (prepare by purifying

deionized water to obtain a sensitivity of 18 MΩ cm at 21 °C) and
analytical or mass spectrometry (MS) grade reagents. Prepare and
store all reagents at room temperature unless stated otherwise.

2.1 Instruments 1. Ultrasonic water bath.

2. Shandon Pathcentre Tissue Processor (Thermo Fisher
Scientific, Germany).
3. Microtome for sectioning FFPE tissue (e.g., RM 2155, Leica
Instruments).
4. Cryostat for sectioning cryoconserved tissue (e.g., HM550,
Thermo Fisher Scientific).
5. Water bath and flattening table for FFPE tissue sectioning.
6. Two small paint brushes.
7. Incubator (using at 37 °C and 60 °C).
8. Digital pH meter.
9. Microwave oven (e.g., LG MS-202VUT, Seoul, South Korea).
10. Glass staining jar for histology.
11. Flatbed scanner (e.g., Epson Scan Photo V600, Suwa, Japan).
12. Spraying devices ImagePrep (Bruker Daltonik GmbH,
Germany) and SunCollect (SunChrom Wissenschaftliche
Geräte GmbH, Germany).
13. MTP Slide Adapter II (#235380, Bruker Daltonik GmbH).
14. MALDI-MS, e.g., ultrafleXtreme (Bruker Daltonik GmbH).
132 Birte Beine et al.

2.2 Material 1. Liquid nitrogen or isopentane.

for Collection 2. Phosphate buffered saline (PBS).
and Sectioning
3. Acetone, 100 %, room temperature.
of Cryoconserved
Tissue 4. Optimal cutting temperature (OCT) compound.
5. Conductive indium tin oxide (ITO) coated glass slides (Bruker
Daltonik GmbH).

2.3 Material 1. Phosphate buffered saline (PBS).

for Tissue Fixation 2. Formalin (4 %) for fixation.
and Embedding
3. Ethanol for dehydration (70 %, 96 %, 100 %).
4. Xylene for dehydration (100 %).
5. Paraffin wax for embedding (Richard-Allan Scientific™
Histoplast Paraffin, Thermo Fisher Scientific).
6. Conductive indium tin oxide (ITO) coated glass slides (Bruker
Daltonik GmbH).

2.4 Solutions 1. Xylene, 100 %.

for Deparaffinization 2. Ethanol, 100 %.
and Antigen Retrieval
3. Ammonium bicarbonate (NH4HCO3), 0.01 M.
of FFPE Tissue (See
Note 1) 4. Citric acid monohydrate, 0.01 M, pH 6.0.

2.5 Solutions 1. NH4HCO3, 0.05 M.

for Trypsin Application
2. Acetonitrile (ACN).
3. Trifluoroacetic acid (TFA).
4. Methanol.

2.6 Materials 1. Liquid Tip Ex (water based) to set marks for geometric orien-
for Setting Reference tation of the tissue.
Points on the ITO Slide 2. Solvent-resistant pen to set fine marks for coregistration (e.g.,
laboratory marker, #6130603, Paul Marienfeld GmbH &
Co. KG).

2.7 Matrix Solution 1. For tissue preparations using α-cyano-4-hydroxycinnamic acid

(See Note 2) (HCCA): 7 mg/ml HCCA, 50 % ACN, 0.2 % trifluoroacetic
acid (TFA).
2. For tissue preparations using 2,5-dihydroxybenzoic acid
(DHB): 30 mg/ml DHB, 50 % methanol, 1 % TFA.

2.8 Histological 1. Ready to use Meyer’s hematoxylin stain (Sigma-Aldrich, St.

Staining: Louis, MO, USA).
Hematoxylin–Eosin 2. Eosin Y.
(HE) Staining
3. Tap water.
4. Distilled Water.
 Tissue MALDI Mass Spectrometry Imaging (MALDI MSI) of Peptides 133

5. Ethanol, 100 %.
6. Xylene, 100 %.
7. Mounting medium (e.g., Richard-Allan Scientific Cytoseal
XYL, Thermo Fisher Scientific).
8. Cover glass.

2.9 External 1. Peptide standard II (Bruker Daltonik GmbH).

Calibration Standard 2. 0.1 % TFA.
for the MALDI
Instrument

2.10 Data Analysis 1. FlexImaging, version 4.1 (Bruker Daltonik GmbH).

Software 2. FlexControl, version 3.7 (Bruker Daltonik GmbH).
3. SCiLS Lab, version 2014b (SCiLS GmbH, Bremen, Germany).

3 Methods

3.1 Fresh Frozen The following description is based on the assumption that all steps
Tissue Sample of tissue preparation need to be performed including the tissue
Preparation conservation process. If you work for example with clinical samples
this process is in most cases already done and you can skip
Subheading 3.1.1 and start directly with the sectioning of the tis-
sue (Subheading 3.1.2) (see Note 3).

3.1.1 Sample Collection Ideal samples are snap or flash frozen in the gas phase of either
and Snap Freezing liquid nitrogen or isopentane directly after dissection in order to
avoid autolysis and to preserve the molecular composition of the
tissue (see Note 4).
1. In case of nitrogen do not drop the sample directly into the
nitrogen to avoid cracks within the sample. Rather place the
sample if small enough in a tube or vial and cap it tightly or
wrap in aluminum foil and submerge in liquid nitrogen for
flash freezing (see Note 5).
2. Supervise the temperature with a thermometer when using
isopentane to make sure that it does not reach −78 °C (subli-
mation temperature of CO2) (see Note 6).
3. If samples are contaminated with blood it is advisable to briefly
wash the tissue in cold phosphate buffered saline (PBS) prior
to freezing.
4. Place frozen samples in pre-labeled and prechilled containers.
The samples can be stored at −80 °C or used for subsequent
cutting of tissue sections.
134 Birte Beine et al.

3.1.2 Fresh Frozen A cryostat is used to cut the samples into 10 μm thin tissue sections.
Tissue Sectioning
1. Remove the frozen sample from the freezer and allow it to
equilibrate in the cryostat chamber to the desired temperature
for approximately 30 min, depending on the size of the sample
(see Note 7).
2. Fix the sample on the metal holder using optimal cutting tem-
perature (OCT) compound (see Note 8). Complete embed-
ding of the sample in OCT should be avoided since it can lead
to suppressed ion formation and intensity during the
measurement.
3. Cut samples into 10 μm thin tissue slices using the glass anti-
roll plate to prevent upward curling of the cut section. Remove
the glass plate and directly pick up the section from the stage
onto a conductive indium tin oxide (ITO) glass slide (see Note
9). Static attraction will draw the section to adhere quickly to
the warm slide. Folding and curling of the section has to be
avoided during this “thaw mounting” process. Assure com-
plete adhesion of the tissue by placing the back of the slide
shortly on the back of the hand. Perform this step also inside
the cryostat (see Note 10).
4. Let the sample dry inside the cryostat (see Note 11). Then
place the dried sample into a precooled, labeled air tight con-
tainer. Samples can either be processed right away or in that
condition be stored at ultra-low temperature (−80 °C) until
analysis (see Note 12).

3.1.3 Drying 1. Samples taken from the −80 °C freezer are dried in a vacuum
and Washing of the Sample desiccator for about 30 min (see Note 13).
Prior to Enzyme Application 2. The slides are washed for 1 min per step in a series of ethanol
steps 70 %/70 %/100 %. Slightly moving of the slides while in
solution enhances the washing process. This step is primarily
done to remove salts, lipids and other contaminants.
3. After washing dry the tissue once again in the vacuum desicca-
tor for about 30 min.

3.1.4 Setting Reference Reference points for orientation and instrument settings need to
Points and Scanning be placed on the glass slide close to the tissue prior to enzyme and
matrix application. Figure 1 shows and example for reference point
placement.
1. Draw crosses with Tip Ex (water based) in some distance to
the tissue to prevent contamination (see Note 14).
2. Then write letters using a solvent-resistant pen very close to
the tissue (see Fig. 1).
3. Scan the tissue with 2400 dpi resolution using a flatbed scan-
ner. This scan is used for teaching in order to transfer the
 Tissue MALDI Mass Spectrometry Imaging (MALDI MSI) of Peptides 135

Fig. 1 Sample slide with reference points. The crosses are made with Tipp-Ex. The letters are written very close
to the tissue using a solvent-resistant pen

geometry of the tissue to the measurement software of the

MALDI instrument.
The tissue preparation continuous with the enzyme applica-
tion (Subheading 3.3).

3.2 FFPE Tissue The following description is based on the assumption that all steps
Sample Preparation of tissue preparation need to be performed including the fixation
and embedding process of the tissue. If you work for example with
clinical samples in most of the cases these processes are already
done in the clinic and you can skip reading Subheading 3.2.1 (see
Note 3).

3.2.1 Tissue Fixation, 1. After dissection of the tissue transfer the sample immediately
Dehydration and Paraffin into 4 % formaldehyde. To remove blood or other fluids per-
Embedding form a brief washing step with ice cold phosphate buffered
saline (PBS) prior to fixation.
2. A standardized routine fixation, dehydration, and paraffin
embedding in pathology department is performed overnight
in an automated manner (e.g., using the Shandon Pathcentre
Tissue Processor, Thermo Fisher Scientific). A typical program
for tissue fixation is displayed in Table 1 (see Note 15).

3.2.2 FFPE Tissue It is much easier to generate very thin sections (2–6 μm) from
Sectioning FFPE tissue than from cryoconserved tissue. Instead of using a
cryostat at subzero temperatures, FFPE sectioning is performed
using a microtome at room temperature.
1. Preheat the water bath and flattening table for tissue section-
ing to 39 °C.
2. Place the paraffin block with the tissue into the specimen
holder (see Note 16).
3. Cut the sample into 5 μm thin sections and transfer the sec-
tions into the water bath using small paint brushes. Wait until
136 Birte Beine et al.

Table 1
Program for tissue fixation using the Shandon Pathcentre. The protocol was kindly provided by the
pathology department of the University Hospital RWTH Aachen (Germany) directed by Prof. Dr. med.
R. Knüchel-Clarke

Step Solution Concentration (%) Incubation (min)

1 Formalin 4 80
2 Formalin 4 55
3 Ethanol 70 80
4 Ethanol 96 55
5 Ethanol 100 55
6 Ethanol 100 55
7 Ethanol 100 55
8 Xylene 100 55
9 Xylene 100 55
10 Xylene 100 55
11 Histoplast 100 45
12 Histoplast 100 45
13 Histoplast 100 45
14 Histoplast 100 45

the tissue is smoothed then use a conductive ITO slide to pick

up the whole tissue section (see Note 17).
4. Place the ITO slide onto the preheated flattening table and
leave it there until the surplus water is evaporated.
5. For optimal adhesion of the tissue onto the ITO slide incuba-
tion at 37 °C for 12 h is advisable.
6. The slides and the tissue block can be stored for several weeks,
airtight and dry at 4 °C.

3.2.3 Tissue Before performing antigen retrieval it is necessary to remove the

Deparaffinization paraffin wax and rehydrate the tissue effectively.
and Rehydration
1. Incubate the slide for 1 h at 60 °C prior to paraffin wax removal
to assure optimal adhesion of the tissue onto the glass slide.
2. Transfer the slide into a glass staining jar filled with 100 %
xylene and incubate the sample for 5 min (see Note 18).
3. Place the slide in a second glass jar with fresh 100 % xylene and
incubate another 5 min.
 Tissue MALDI Mass Spectrometry Imaging (MALDI MSI) of Peptides 137

4. Incubate the slide twice in 100 % ethanol for 2 min using

another staining glass jar. Use fresh solution for the second
step. Afterwards allow the tissue to dry completely.
5. For rehydration of the tissue place the slide in a glass jar filled
with 0.01 M NH4HCO3 and leave it for 3 min. Repeat this
step once again with fresh 0.01 M NH4HCO3 solution.

3.2.4 Antigen Retrieval In this protocol antigen retrieval is achieved using a heat induced
citric acid antigen retrieval method [14].
1. To prevent bubble formation during the antigen retrieval pro-
cedure prepare a histology glass jar with a rack filled up with
dummy glass slides leaving one space free for your sample.
2. Place the slide into 0.01 M citric acid buffer (pH 6) and put
the glass jar into the microwave. Start the 600 W program for
3 min. Afterwards reduce the power to 90 W for 10 more min-
utes (see Note 19). Avoid strong boiling.
3. To proceed with the antigen retrieval prepare a heat plate to
150 °C and place the glass jar onto the heat plate and incubate
for another 30 min directly after the incubation in the micro-
wave. Let the sample cool down for 15 min on the bench top.
4. To optimize the pH for the subsequent enzymatic digestion,
incubate the sample twice in 0.01 M NH4HCO3 solution once
the citric acid buffer has cooled down. Prepare a histology
glass jar with the solution. Use fresh solution for the second
1 min incubation.

3.2.5 Setting Reference Apply the reference points and scan the tissue as described in
Points and Scanning Subheading 3.1.4.
The tissue preparation continuous with the enzyme applica-
tion and digest of the tissue (see Subheading “Protease Application
and Digest—SunCollect/SunDigest” or “Protease Application
and Digest—ImagePrep”).

3.3 Enzyme This protocol provides the experimental procedures of enzyme and
and Matrix Application matrix application with two different instruments which are the
for Peptide MALDI MSI most common commercially available devices for these applica-
tions; the SunCollect (SunChrom) and the ImagePrep (Bruker
Daltonik GmbH).
There are two main differences comparing both instruments
regarding their spraying mechanism.
The SunCollect device uses pneumatic atomization to spray
enzyme and matrix solution onto the sample. The spray head is
flexible and moves quickly and continuously during the prepara-
tion process over the tissue. The spaying process is regulated by
software and individual spraying sessions can be programmed.
Parameters which can be modified are for example the distance of
138 Birte Beine et al.

the spaying head to the sample, the speed of the spay head while
moving over the tissue and the liquid flow per minute. Those are
only some of the parameters which can be modified for optimizing
the spraying result. For detailed information please read the manu-
factures’ manual.
TheImagePrep device is an automatic spraying device, which
uses vibrational vaporization for spray production. The spray head
is fixed and the spray must be optimized to reach the sample which
must be positioned in a defined distance towards the spray head.
The ImagePrep software is preset with some default methods for
protease and matrix applications and is controlled by a sensor
which is positioned near the tissue. Each method consists of one or
more phases each with individual spraying, incubating or drying
conditions. Parameters of the different phases can be individually
modified as well as the number of spraying phases can be changed.
As before for detailed information please read the manufactures’
manual. Figure 2 provides an overview of the two instruments
together with a detailed view of the actual spraying device.

3.3.1 Protease Enzyme application is a crucial step in tissue preparation for pep-
Application and Digest tide MALDI MSI. There are several parameters which need to be
considered: optimal digestion of the proteins requires a humid
environment but delocalization due to diffusion needs to be mini-
mized to remain the spatial resolution of the peptides. Therefore
the spray for enzyme application needs to produce very small drop-
lets and at the same time provide sufficient humidity for enzyme
activity. In the following section we describe procedures that fulfill
these requirements either using the SunCollect (Subheading
“Protease Application and Digest—SunCollect/SunDigest”) or
the ImagePrep (Subheading “Protease Application and Digest—
ImagePrep”) spraying device.
The enzymatic digestion requires an environment with high
relative humidity (RH) and temperature. When using trypsin, tem-
peratures of 37 °C are the lowest temperature level for sufficient
digestion. In the MALDI MSI community a lot of self-made
instrumentations (digestion chambers) for efficient enzyme diges-
tion are used. In this protocol we provide two different types of
digestion chambers: the SunDigest, a sensor regulated and pro-
grammable automatic device and a self-made digestion chamber
using pipette boxes.

Protease Application 1. Prechill a 250 μl syringe on ice (see Note 20).

and Digest— 2. Prepare 20 μg of trypsin in 200 μl cold buffer (0.05 M
SunCollect/SunDigest NH4HCO3 and 5 % ACN) and fill the syringe without
bubbles.
3. Place the syringe into the syringe pump of the SunCollect,
connect it with the capillary and place a cold pack on top of
the syringe to prevent enzyme activity.
 Tissue MALDI Mass Spectrometry Imaging (MALDI MSI) of Peptides 139

Fig. 2 SunCollect and ImagePrep instrument. In addition to an overview of the two instruments are detailed
views of the spraying devices given. A1–A3 show the SunCollect device. B1–B3 illustrates the ImagePrep
device

4. After placing the sample into the instrument it is necessary to

insert the coordinates of the tissue position into the software.
For that follow the instructions of the manual.
5. Use the software to design a spraying protocol for enzyme
application of your sample (see Note 21).
6. Before starting the program place an empty glass slide below the
spray and check for a continuous and fine spray (see Note 22).
7. Remove the glass slide and start the spray. The spray head will
move over the entire predefined area in fixed lines thus covering
140 Birte Beine et al.

the whole sample with enzyme solution. It is necessary to apply

several layers of enzyme solution (see Note 23).
8. Prepare the SunDigest before continuing with the transfer of
the sample into the digestion chamber. Apply approximately
5 ml water onto the felt pads (see Note 24).
9. Choose the parameters for digestion (temperature, relative
humidity (RH) and time). The instrument regulates the RH
and prevents condensation by active heating of the lid. Two
sensors regulate the heat inside the instrument; one is placed
above the other below the sample inside the heating block (see
Note 25). The measured data are recorded and graphically
displayed as quality control for the user.
10. Check the graphical summary of all parameters and wait until
the instrument executed the chosen settings (Chamber tem-
perature, base temperature, RH and lid temperature) (see
Note 26).
11. After digestion (see Note 27) remove the slide from the
SunDigest and start the matrix application (see Subheading
3.4.1).

Protease Application 1. Test the spray performance by loading 200 μl 0.05 M

and Digest— NH4HCO3 buffer directly into the spray head. Hold the head
ImagePrep in a horizontal position while pipetting. Afterwards seal the
head with the empty glass bottle in the metal surrounding and
set the power to 38 % and the modulation to 0 % in the adjust-
ment menu of the ImagePrep. Stop the time while spraying.
The spray should reach the sensor at the round glass plate in
the middle of the spraying chamber thereby evenly distributing
the solution. The entire volume should be finished within a
time frame of 20 s. If not adjust the spray power and repeat the
test.
2. Wipe the inside of the chamber to remove the moist and clean
the area of the sensor in the middle of the chamber carefully.
3. Put the ITO slide on the placement area for the slide above the
light-scattering sensor without covering the sensor itself with
the tissue.
4. Load the prepared trypsin solution (200 μl) in the same man-
ner as the test solution and start the Bruker standard program
for digest: trypsin_deposition_nsh01.
5. Start the NIVI-Logger (part of the ImagePrep Software that
runs on a computer) to monitor the spray cycles. The program
should reach about 20 spray cycles.
6. Upon completion transfer the slide to a humidity chamber
and let the digestion take place at 37 °C for as long as desired
(see Note 28).
 Tissue MALDI Mass Spectrometry Imaging (MALDI MSI) of Peptides 141

3.4 Matrix Matrix application is another critical step of the sample preparation
Application for MALDI MSI. During matrix application it is necessary to use
solvents and water to extract the peptides from the tissue and to
induce the cocrystallization of the matrix and the analyte. Crystal
size formation and extraction efficiency have a large impact on the
MSI result. Thus it is the overall goal during matrix application to
achieve efficient peptide extraction while maintaining good spatial
resolution.

3.4.1 Matrix 1. Place the tissue in the SunCollect instrument. When teaching
Application—SunCollect was applied for enzyme application no further teaching is nec-
essary. Otherwise teach the instrument (see Subheading
“Protease Application and Digest—SunCollect/SunDigest”).
2. Use a 2500 μl syringe for matrix application and fill it with
matrix solution. Place the syringe into the pump and connect
it with the capillary (see Note 29).
3. Use the software to design a spraying program for matrix appli-
cation of your sample (see Note 30). The program for matrix
application differs considerably to the one for enzyme applica-
tion. To initiate crystal formation flow rates of 10–20 μl/min
should be applied. The flow rate per minute increases with the
number of layers. The maximal flow rate to prevent diffusion
should be 40 μl/min.
4. Hold an empty glass slide below the spray and check for a con-
tinuous and fine spray.
5. Remove the glass slide and start the matrix application
program.
6. After matrix application fix the slide into the MTP Slide
Adapter II. Apply a peptide standard in the middle of the steel
frame of the adapter for external calibration: according to the
dried droplet method, mix equal volumes (1 μl) of calibration
standard sample with matrix solution and dry at room
temperature.
7. Clean the capillary by filling the syringe with 70 % ACN. Use
a flow rate of 40 μl/min and let it run for at least 15 min.
While flushing the capillary with ACN clean the spray tip
with the same solvent using a pipette. Let several drops run
down the capillary tip without touching the end of the capil-
lary with the pipette tip.
Continue with the MALDI MSI measurement (see Subheading 4).

3.4.2 Matrix 1. Start off with testing the sprayer. Fill the glass vial 2/3 with
Application—ImagePrep methanol and choose in the display of the ImagePrep Spray
(See Note 31) Head → Adjust → Test Spray. First test 100 % and then 10 %. With
the first setting the spray should reach the round plate on the
opposite site of the chamber whereas the 10 % should at least
142 Birte Beine et al.

reach the end of the placement area (with the sensor in the mid-
dle) for the slide. If not adjust the offset from for example 4–5 %.
2. Remove the methanol residues by wiping the inside of the
chamber with a paper towel.
3. Adjust the power and modulation for the matrix. Load the
prepared matrix into the glass vial and choose adjustment in
the display of the instrument. By default is the power set to 25
and the modulation to 30 for DHB and 15 and 40 for
HCCA. The matrix spray should reach the end of the place-
ment area of the slide. If necessary adjust the power or modu-
lation (see Note 32).
4. Once again clean the inside of the chamber especially the area
above the sensor. Afterwards place the sample slide on top
without covering the sensor with the tissue (see Note 33). If
more than one tissue is on the slide, position the sensor in
between the two samples. Lay a cover glass on top of the slide,
above the sensor.
5. Start the NIVI-Logger (part of the ImagePrep software that
runs on a computer) to monitor the spray cycles.
6. Start the run using the Bruker standard program
DHB_for_digest_nsh01 or HCCA_nsh04.
7. Wait at least until the first 3 cycles of the second phase are
finished and control the profile of the logger. An increase of
the measured curve should be visible if not adjust the spray
power boost (see Note 34).
8. Once the program has finished, stop the logger, take out the
slide and remove the matrix from the short edges of the slides
with 100 % ethanol before fixing the slide into the MTP Slide
Adapter II (see Note 35).
9. Apply a peptide standard in the middle of the steel frame of
the adapter for external calibration: according to the dried
droplet method, mix equal volumes (1 μl) of calibration stan-
dard sample with matrix solution and dry at room
temperature.
10. Clean the inside of the ImagePrep with methanol by running
the corresponding clean program. Afterwards dry the inside of
the chamber with a paper towel.
Continue with the MALDI MSI measurement.

3.5 MALDI MSI The following section describes the peptide MALDI MSI measure-
Measurement ment using a MALDI-TOF mass spectrometer (ultrafleXtreme,
Bruker Daltonik GmbH) controlled by the Bruker software
FlexControl and FlexImaging.
1. Open FlexControl.
 Tissue MALDI Mass Spectrometry Imaging (MALDI MSI) of Peptides 143

2. Load the MTP Slide Adapter II with the fixed ITO slide into
the instrument.
3. Open FlexImaging and create a file for your sample, chose the
desired autoXecute method and raster width. Use the Tip Ex
crosses on the slide to coordinate the sample and correlate it to
the previously made scan of the slide.
4. Define the measurement region on the tissue.
5. In FlexControl: choose reflector mode with a set mass range of
m/z 600–4000. Save those settings in a FlexControl method.
The method should also contain a statistical recalibration of
the spectra. For this purpose integrate the flexAnalysis method
ImageID_StatRecal.FAMSMethod [15].
6. Load the generated FlexControl method and measure the
external calibrants on the target frame. The error of each pep-
tide should not exceed 20 ppm. Apply the calibration and save
the method.
7. Finally determine the necessary laser power for the tissue. Save
the method and start the run using FlexImaging.

3.6 HE-Staining The unique characteristic of MALDI MSI compared to other mass
of the Analyzed Tissue spectrometry based analyses is the ability to maintain the spatial
resolution of the analyzed molecules. It is thus possible to compare
the MS data directly to the histological features, which are visible
after staining the tissue. Standard histological staining protocols
can be used to stain the tissue after the measurement. Check which
kind of staining is the best to answer your questions. In the follow-
ing we provide a fast hematoxylin–eosin (HE) staining which is
suitable for many tissue types. Hematoxylin stains nuclei blue,
while eosin results in a pink staining of the connective tissue.
1. Remove the matrix by washing the tissue in 70 % ethanol. For
this purpose move the slide up and down for several times.
2. Incubate the tissue in Meyers’ hematoxylin solution for 1 min
(see Note 36).
3. Incubate the tissue in tab water for 4 min. This step causes a
pH shift and intensifies the staining of hematoxylin.
4. Dip the slide twice into 0.25 % eosin solution.
5. Dip the slide twice into distilled water.
6. Repeat dipping the slide in 70 % ethanol.
7. Incubate the slide for 20 s each in 96 %, 100 %, and again 100
% ethanol.
8. For complete dehydration incubate the slide 5 min in 100 %
xylene. Then dry the slide completely.
9. Put one drop of mounting medium on the tissue and place a
cover slip on top. Try to avoid air bubbles since they will dete-
144 Birte Beine et al.

riorate the quality of subsequent microscopy. To remove bub-

bles press slightly onto the glass. Surplus mounting media can
be removed after complete drying (2 h at room temperature)
by scraping it off the surface using another glass slide.

3.7 Data Analysis For the analysis of the MALDI MSI data we use SCiLS Lab. This
Using SCiLS Lab software allows you to analyze several tissues simultaneously, which
is important when comparing different samples in one study or
comparing different methods.
1. Import the mis-files from FlexImaging into SCiLS Lab. You
can arrange your samples in the way you want them to be dis-
played (rotation is possible as well). SCiLS Lab converts the
data into a so-called h5-file. The generated h5-file contains all
the data you need for further analysis (see Note 37).
2. In SCiLS Lab you have several options for data analysis. To
gain a first impression it is a good starting point to run the
“Segmentation pipeline” and let the software perform all neces-
sary steps automatically. This pipeline contains preprocessing
steps such as baseline removal, normalization and the genera-
tion of an overview spectrum for all the samples in your data
set. Furthermore peak picking, alignment, spatial denoising
and segmentation are performed.
3. To evaluate different data sets statistical and structure analysis
are suitable tools. The software provides a large amount of
analysis tools so that only some selected examples can be pro-
vided. For statistical analysis two different component analyses
can be performed in SCiLS Lab: PCA (principal component
analysis) and pLSA (probabilistic latent semantic analysis)
whereby the latter one can also be seen as a form of structure
analysis (see Note 38).
4. When you use a model tissue like rat brain it is possible to
judge the quality of your different methods using an additional
strategy: counting the number of m/z values, which show a
structured image for a well-known and highly abundant pro-
tein, e.g., myelin basic protein (MBP) in rat brain (see Note
39). For this strategy you have to first perform an in-silico
digest of MBP with trypsin using for example MS-Digest from
the ProteinProspector website (http://prospector.ucsf.edu/
prospector/cgi-bin/msform.cgi?form=msdigest) to obtain a
list of MBP specific peptides (see Note 40). For the in-silico
digest defined settings must be used: maximal number of
missed cleavages 1, oxidation of methionine as variable modi-
fication, peptide mass range 800–4000 Da and MALDI-TOF/
TOF as instrument type. In case of MBP this results in a list of
45 peptides respectively 45 monoisotopic m/z values.
 Tissue MALDI Mass Spectrometry Imaging (MALDI MSI) of Peptides 145

This list can be imported into the h5 data set (see Note 41). It
is then possible to visualize the images of each individual m/z
value allowing for example a peak shift of ±0.3 Da. It is now
possible to count the number of visible MPB structures and
compare them between you different experiments.

4 Notes

1. All solutions used for deparaffinization besides ammonium

bicarbonate can be used several times. When you change the
tissue type you should also change the solutions. The citric
acid buffer can be stored for 1 week, check pH prior to use.
2. Matrix solution should be sonicated for approximately 15 min
prior to use. There should be no visible matrix particles left.
The matrix solution can be used up to 7 days but should be
kept in the dark in an air tight glass bottle at 4 °C to avoid
volatilization of the solvent.
3. When you contribute in clinical studies you usually get sam-
ples that are already sectioned. If that is the case it is important
to note that the people preparing you samples stick to the
rules which need to be fulfilled when working with MALDI
MSI samples like section thickness and awareness of tempera-
tures during the preparation process.
4. For beginners in the field of MALDI MSI we recommend to
start with commercially available rat brain samples which are
the most common model sample in MALDI MSI. In this way
errors during sample collection can be avoided. The sample is
fairly easy to cut and the user can focus on MALDI MSI prep-
aration parameter optimization.
5. For a rat brain a 100 ml beaker will be sufficient for freezing.
6. Independent of the freezing method be careful and avoid
morphological distortion or damage of the sample. Freezing
artifacts in the middle of a tissue occur more frequently when
large objects are flash frozen. Guideline: at least one dimen-
sion of the sample must be below 10 mm. Whole rat brains
can be successfully frozen.
7. The optimal temperature for sectioning depends mainly on
the type and nature of the tissue. Reference charts for tem-
perature settings are generally provided in the manual of the
producer of the cryostat.
8. Tissue Tek® is a useful OCT material.
9. Avoid placing more than one tissue sample on the same slide
when using the ImagePrep device for subsequent spraying.
146 Birte Beine et al.

10. It has been reported that ITO slides modified with poly-L-
lysine improved the adherence of the cells to the substrate
[16].
11. Do not place the slide on the metal ground of the cryostat
while letting the cut tissue section dry inside the instrument.
The direct contact with the metal will lead to freeze artifacts.
We therefore recommend placing a vacuum formed slide
holder inside the cryostat to overcome this problem.
12. In general it is advisable to use flash frozen samples right away
for imaging rather than storing them long term. The same
applies for the cut tissue sections on ITO glass slides. In case
of long-term storage we recommend to put each slide in an
individual container even though there might be space for
more but by doing so we avoid temperature changes when
opening the box to remove single slides. We found
LockMailer™ with screw caps to be the best storage device for
that purpose.
13. Drying of the samples is also possible in a regular desiccator.
14. Use only little amount of Tip Ex to maintain an even surface
otherwise you will have problems to cover the subsequently
HE stained sample with a cover glass after MALDI MSI
measurement.
15. Fixation time can be reduced to ~30 min in case of small
biopsy samples whereas some bloody or fatty tissues as well as
some fetal tissues require longer fixation times. Make sure to
check the optimal conditions for your tissue sample prior to
fixation.
16. We recommend removing surplus paraffin wax at the edges of
the block to minimize the sectioning area.
17. If the flattening of the tissue is not sufficient after 5–10 min,
temperature needs to be optimized.
18. We recommend using a histology glass jar with molded glass
cover in which the slides are in an upright position.
19. During this process it is necessary to constantly check for bub-
ble formation and to remove bubbles by simply pulling the
slide holder upwards.
20. Put the syringe into a plastic bag to avoid contamination.
21. To find the perfect enzyme spray settings for your tissue you
should test one to three slides before you prepare a real sam-
ple. Test enzyme spray settings and digestion settings and
compare the MALDI MSI results in regard to intensities, spa-
tial resolution and digest efficiency.
22. We spray for at least 1 min before starting the program.
Tissue MALDI Mass Spectrometry Imaging (MALDI MSI) of Peptides 147

23. Clean the capillary flow by inserting a syringe with 70 % ACN

after every use. Use a flow rate of 40 μl/min for 5 min.
24. Estimate the amount of water on the felt pads of the SunDigest.
Your glove should be wet after touching the pads.
25. Fill both racks with the same number of slides to ensure
homogenous distribution of heat and humidity. We use the
program at 50 °C and 95 % RH, 2 h for cryoconserved tissue,
1 h for FFPE tissue.
26. For quality control of the digest it is reasonable to apply for
example bovine serum albumin standard onto a small part of
the tissue. You can use the same settings for the application of
standard as for trypsin digest (we advise to use only one layer
of standard and therefore use a suitable amount of protein in
the solution).
27. Clean the capillary flow by inserting a syringe with 70 % ACN
after every use. Use a flow rate of 40 μl/min for 5 min.
28. An empty pipette box can be used as a humidity chamber, by
filling the lower part of the box with water (fill height ~ 1 cm),
placing the slide on top of the perforated top and closing the
lid before placing the box into an incubator for digestion.
29. Before spraying the matrix we run 70 % ACN for at least
5 min.
30. Before starting with your sample of interest you need to pre-
pare some test slides. Use these slides to verify different param-
eters like height of the spray head (Z position), flow rate,
speed and number of layers. Check the crystal shape and size
under a microscope and chose the one which fits best for your
purpose (examples given in Fig. 3). In general the crystals
should be very small and show a homogenous distribution
over the whole sample. In Fig. 3 we provide examples of
matrix crystals. Be aware that you should check the resulting
MALDI MSI image as well. Based on the results shown in Fig.
3 we decided to use the z position 24. We think using z posi-
tion 24 in this case results in the best density, distribution and
crystal shape of HCCA. Comparing the matrix distribution in
different magnitudes is imported to gain a sufficient overview.
It is for example easier to identify clogging when using a lower
magnitude. We advise to regularly check matrix distribution
patterns, crystal size and shape under the microscope as qual-
ity control.
31. Based on our observations the adjustment of the spray settings
strongly varies between different persons in the sense of what
is considered to be a good spray. Consequently we strongly
advise that all spraying operations within a project should be
carried out by the same person.
148 Birte Beine et al.

Fig. 3 Overview of different crystal shapes and distributions using the SunCollect spraying device. HCCA (7
mg/ml, 50 % ACN, 0.2 % TFA) matrix was applied onto rat brain tissue sections only the parameter of the z
position during matrix spray was changed (see Note 29). (a) and (b) display images with different magnitude
(×20 and ×40)

32. When adjusting the settings (power and modulation) it can be

helpful to place an empty glass slide on the intended position
to control the crystal size and density of the sprayed matrix.
33. Best results were obtained when placing the tissue directly in
front of the sensor.
34. It is advisable to observe from time to time the actual spraying
process inside the chamber since the logger alone is sometimes
not sufficient for judgment.
35. Add an additional fifth phase if the matrix appears still uneven
and not dense enough.
36. You can intensify the staining of hematoxylin in two ways:
increase the incubation time for hematoxylin and/or increase
the time for the following incubation step in tap water.
37. Please check the system requirements for running SCiLS Lab
before you start.
38. Principle component analysis (PCA) is a widely used compo-
nent analysis method to assess the overall structure in a data
set. The components resulting from a PCA are statistically
 Tissue MALDI Mass Spectrometry Imaging (MALDI MSI) of Peptides 149

uncorrelated and arranged so that the first principal compo-

nent has the largest variance within the data set. Each subse-
quent component has the largest variance within the remaining
orthogonal dimension of the data set. Thus a PCA is able to
uncover and visualize variability within the data. Scaling the
data before calculating the PCA is often useful; variances can
be scaled before calculating the PCA for spatial and spectral
features. Use the aligned peaks of the preprocessed datasets
for computation of PCA for individual and mean spectra [7].
Using the probabilistic latent semantic analysis (pLSA) both,
score images and loadings can be interpreted in terms of mass
spectra intensities [17]. They are non-negative for pLSA in
comparison to PCA. The results of the pLSA can therefore be
interpreted as spatial tissue components and their correspond-
ing mass distribution in the tissue. Before starting the pLSA
estimate the optimal number of pLSA components. Therefore
it is helpful to know the anatomical features of your used sam-
ple (for rat brain the Allen mouse brain atlas is a good source
(http://mouse.brain-map.org/)). When you try to establish a
protocol in your lab keep in mind, that you should produce
technical replicates (e.g., triplicates) of each tested condition
and measure the samples in an alternating order.
39. MBP is a suitable protein for such analysis because it is one of
the most abundant proteins of the myelin membrane of the
central nervous system (CNS). Thus it is guaranteed that the
protein is present in all the samples for comparison and sec-
ondly that the strong expression would facilitate detection of
MBP during MALDI MSI measurement. Information about
the anatomical distribution of the protein in the brain can be
obtained from for example the Allen mouse brain atlas.
40. The necessary protein sequence information (FASTA format)
can be obtained from the UniProt/Swiss-Prot data base in this
case one would use Rattus norvegicus as taxonomy.
41. All you need for the generation of the peak list into your h5
file is an excel csv.-file with the m/z of interest. Copy the col-
umn with the m/z values and paste them during the import
process into the import window. To do so open in SCiLS:
File → Import MZ Range from CSV and paste the copied
column.

Acknowledgement

We would like to thank Julian Elm for his skilful technical assis-
tance during the method development process. We thank Dennis
Trede for excellent support dealing with SCiLS Lab. Furthermore
150 Birte Beine et al.

we like to thank Prof. Dr. med. Ruth Knüchel-Clarke for providing

the protocol for the fixation process of FFPE tissue. H.D. was
financially supported by PURE (Protein research Unit Ruhr
within Europe).

References

1. Castaing R, Slodzian G (1962) Microanalyse 10. Delvolve AM, Woods AS (2011) Optimization
par emission ionique secondaire. J Microscopie of automated matrix deposition for biomolec-
1:395–410 ular mapping using a spotter. J Mass Spectrom
2. Trim PJ, Djidja MC, Muharib T et al (2012) 46(10):1046–1050. doi:10.1002/jms.1986
Instrumentation and software for mass spec- 11. Yang J, Caprioli RM (2011) Matrix sublima-
trometry imaging—making the most of what tion/recrystallization for imaging proteins by
you’ve got. J Proteomics 75(16):4931–4940. mass spectrometry at high spatial resolution.
doi:10.1016/j.jprot.2012.07.016 Anal Chem 83(14):5728–5734. doi:10.1021/
3. Groseclose MR, Andersson M, Hardesty WM ac200998a
et al (2007) Identification of proteins directly 12. Jaskolla TW, Karas M, Roth U et al (2009)
from tissue: in situ tryptic digestions coupled Comparison between vacuum sublimed matri-
with imaging mass spectrometry. J Mass ces and conventional dried droplet preparation
Spectrom 42(2):254–262. doi:10.1002/ in MALDI-TOF mass spectrometry. J Am Soc
jms.1177 Mass Spectrom 20(6):1104–1114.
4. Monroe EB, Annangudi SP, Hatcher NG et al doi:10.1016/j.jasms.2009.02.010
(2008) SIMS and MALDI MS imaging of the 13. Ait-Belkacem R, Sellami L, Villard C et al
spinal cord. Proteomics 8(18):3746–3754. (2012) Mass spectrometry imaging is moving
doi:10.1002/pmic.200800127 toward drug protein co-localization. Trends
5. Murphy RC, Hankin JA, Barkley RM (2009) Biotechnol 30(9):466–474. doi:10.1016/j.
Imaging of lipid species by MALDI mass spec- tibtech.2012.05.006
trometry. J Lipid Res 50 Suppl:S317–S322. 14. Gustafsson OJ, Eddes JS, Meding S et al
doi:10.1194/jlr.R800051-JLR200 (2013) Matrix-assisted laser desorption/ion-
6. Khatib-Shahidi S, Andersson M, Herman JL ization imaging protocol for in situ character-
et al (2006) Direct molecular analysis of whole- ization of tryptic peptide identity and
body animal tissue sections by imaging MALDI distribution in formalin-fixed tissue. Rapid
mass spectrometry. Anal Chem 78(18):6448– Commun Mass Spectrom 27(6):655–670.
6456. doi:10.1021/ac060788p doi:10.1002/rcm.6488
7. Diehl HC, Beine B, Elm J et al (2015) The 15. Mann M, Hojrup P, Roepstorff P (1993) Use
challenge of on-tissue digestion for MALDI- of mass spectrometric molecular weight infor-
MSI—a comparison of different protocols to mation to identify proteins in sequence data-
improve imaging experiments. Anal Bioanal bases. Biol Mass Spectrom 22(6):338–345.
Chem. doi:10.1007/s00216-014-8345-z doi:10.1002/bms.1200220605
8. Oezdemir RF, Gaisa NT, Lindemann-Docter K 16. Zavalin A, Todd EM, Rawhouser PD et al
et al (2012) Proteomic tissue profiling for the (2012) Direct imaging of single cells and tissue
improvement of grading of noninvasive papillary at sub-cellular spatial resolution using transmis-
urothelial neoplasia. Clin Biochem 45(1–2):7–11. sion geometry MALDI MS. J Mass Spectrom
doi:10.1016/j.clinbiochem.2011.09.013 47(11):1473–1481. doi:10.1002/jms.3108
9. Casadonte R, Caprioli RM (2011) Proteomic 17. Hanselmann M, Kirchner M, Renard BY et al
analysis of formalin-fixed paraffin-embedded (2008) Concise representation of mass spec-
tissue by MALDI imaging mass spectrometry. trometry images by probabilistic latent seman-
Nat Protoc 6(11):1695–1709. doi:10.1038/ tic analysis. Anal Chem 80(24):9649–9658.
nprot.2011.388 doi:10.1021/ac801303x
 Chapter 11

Ethyl Esterification for MALDI-MS Analysis

of Protein Glycosylation
Karli R. Reiding, Emanuela Lonardi, Agnes L. Hipgrave Ederveen,
and Manfred Wuhrer

Abstract
Ethyl esterification is a technique for the chemical modification of sialylated glycans, leading to enhanced
stability when performing matrix-assisted laser desorption/ionization (MALDI)-mass spectrometry (MS),
as well as allowing the efficient detection of both sialylated and non-sialylated glycans in positive ion mode.
In addition, the method shows specific reaction products for α2,3- and α2,6-linked sialic acids, leading to
an MS distinguishable mass difference. Here, we describe the ethyl esterification protocol for 96 glycan
samples, including enzymatic N-glycan release, the aforementioned ethyl esterification, glycan enrichment,
MALDI target preparation, and the MS(/MS) measurement.

Key words Sialic acid (N-acetylneuraminic acid), Stabilization, Linkage-specific, Peptide-N-

glycosidase F (PNGase F), N-glycan release, Ethyl esterification (EE), Solid-phase extraction (SPE),
Cotton, Sepharose, Hydrophilic interaction liquid chromatography (HILIC), Matrix-assisted laser
desorption/ionization (MALDI), Mass spectrometry (MS)

1 Introduction

Protein glycosylation stands for a group of important and highly

complex post-translational modifications, which can have a pro-
found effect on many characteristics of the conjugate, including
folding, solubility, plasma half-life, and receptor binding activity
[1–3]. Changes in glycosylation depend on genetic, physiological
and environmental factors, and are known to occur under influ-
ence of biological processes like aging, adolescence, and pregnancy,
as well as in neoplastic or inflammatory diseases [4–8]. As such, the
study of glycosylation has the potential of pinpointing relevant bio-
markers (for a particular disease or physiological state), but also to
assist personalized medicine by separating patients according to
disease etiology [9, 10]. Adding to this the importance of biophar-
maceutical glycosylation, the need for high-throughput method-
ologies for glycan analysis becomes apparent [11].

Jörg Reinders (ed.), Proteomics in Systems Biology: Methods and Protocols, Methods in Molecular Biology, vol. 1394,
DOI 10.1007/978-1-4939-3341-9_11, © Springer Science+Business Media New York 2016

151
152 Karli R. Reiding et al.

Of the methods to study glycosylation, mass spectrometry

(MS) has proven to be extremely fast and efficient for achieving
compositional assignment [12, 13]. It has, however, the downside
of not providing information on monosaccharide linkage posi-
tions, and is hampered by the innate instability and charge of gly-
cans carrying a sialic acid [13, 14]. This is evident in matrix-assisted
laser desorption/ionization (MALDI)-MS, where sialylated glycan
species are affected by ionization bias, neutral salt adduction and
loss of sialic acid residues. Several strategies can be employed to
overcome these problems, among which the chemical modification
of the causative sialic acid carboxyl groups into esters or amides
[15–17]. In some cases these reactions can make use of the specific
chemical characteristics of α2,3- and α2,6-linked sialic acids, as
α2,3-linked variants may undergo an intramolecular lactonization,
while the α2,6-linked variants are susceptible to intermolecular
reactions with alcohols or amines [18, 19]. The respective water
loss and ester/amide formation introduce a mass difference to the
isomers, which can then be used for MS assignment.
Recently we published a rapid and robust method for linkage-
specific stabilization of sialylated glycans, using the combination
of 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC) and
1-hydroxybenzotriazole (HOBt) in an environment of etha-
nol [20]. Full reaction specificity was demonstrated between
α2,3- and α2,6-linked sialylation isomers, and the products could
be stably analyzed by reflectron mode MALDI-time-of-flight
(TOF)-MS. Notably, the protocol was shown resistant to impuri-
ties, making it suitable for the analysis of free glycans from a wide
range of sources, and has proven informative for the study of gly-
cans obtained from plasma as well as from specific proteins like
immunoglobulin G [20, 21].
Here the steps required in the protocol will be thoroughly
discussed, including N-glycan release by peptide-N-glycosidase F
(PNGase F), ethyl esterification, purification by hydrophilic inter-
action liquid chromatography (HILIC) solid phase extraction
(SPE), MALDI target spotting, and concluding with a section on
MS(/MS) measurement and quality control. The focus will be on
a high-throughput 96-well variant of the protocol, describing
Sepharose beads for the HILIC enrichment. However, when per-
forming the protocol for a low number of samples, the use of cot-
ton HILIC microtips (using the same solutions) is easier and faster,
while producing highly comparable results [22].

2 Materials

Ultrapure deionized water is required for the preparation of all

solutions (≥18 MΩ at 25 °C). The reagent volumes here described
are sufficient for 100 samples, and can be up- or downscaled unless
 Ethyl Esterification for MALDI-MS Analysis of Protein Glycosylation 153

otherwise specified. Take note when performing the protocol on

very low sample quantities, as pipetting imprecision and ethanol
evaporation start playing a larger role. Importantly, make sure that
all glassware is clean before use or risk polymer contamination in
the resulting mass spectra.

2.1 PNGase F 1. 5× phosphate buffered saline (PBS): Order directly, or prepare

N-Glycan Release as 50 mM Na2HPO4 9 mM KH2PO4 685 mM NaCl 13.5 mM
KCl, pH 7.4. 5× PBS can be stored at room temperature.
2. 2 % (w/v) sodium dodecyl sulfate (SDS) in water. Weigh
200 mg SDS on an analytical balance and transfer to a 15 mL
tube (see Note 1). Add 9.8 mL water. Store at room
temperature.
3. 4 % (w/v) Nonidet P-40 (NP-40) substitute (Sigma-Aldrich,
Steinheim, Germany) in water. Pipette 400 mg NP-40 substi-
tute to a 15 mL tube on an analytical balance (see Note 2).
Add 9.6 mL water. Store at room temperature.
4. Release solution: In a 1.5 mL tube, thoroughly mix 400 μL
4 % NP-40 substitute, 400 μL 5× PBS, and 80 μL PNGase F
(Roche Diagnostics, Mannheim, Germany) (see Note 3).
5. Polypropylene (PP) 96-well plate.
6. Adhesive tape for plate sealing.
7. Ovens at 37 °C and 60 °C.

2.2 Ethyl 1. Esterification reagent: 250 mM EDC hydrochloride

Esterification (Fluorochem, Hadfield, UK) 250 mM HOBt monohydrate
(Sigma-Aldrich) in ethanol. Weigh 67.56 mg HOBt and
transfer to a 2 mL tube. Weigh 95.85 mg EDC and transfer to
the same tube (see Note 4). Add 2 mL ethanol and mix thor-
oughly (see Note 5). The reagent can be stored at −20 °C (see
Note 6).
2. 96-well plate (PP).
3. Adhesive tape for plate sealing.
4. An oven at 37 °C.

2.3 Sepharose HILIC 1. Sepharose bead slurry: 25 % Cl-4B Sepharose beads (45–
Solid Phase Extraction 165 μm, GE Healthcare, Uppsala, Sweden) in 20 % ethanol.
Transfer 500 μL beads to a 2 mL tube. Spin down the beads
and remove the supernatant. Add 1.5 mL 20 % ethanol. Make
fresh before using.
2. 85 % acetonitrile (ACN). Measure 150 mL water in a 50 mL
graduated cylinder and add to a 1 L screw-cap glass bottle.
Using a 1 L graduated cylinder, measure 850 mL ACN and
add to the glass bottle. Store at 4 °C, but bring to room
temperature before usage.
154 Karli R. Reiding et al.

3. 85 % ACN 0.1 % trifluoroacetic acid (TFA). Prepare as 2, add

1 mL TFA (use a chemical fume hood to contain and exhaust
volatile dangerous chemicals). Store at 4 °C, but bring to room
temperature before usage.
4. 96-well filter plate (0.7 mL/well, PE frit, Orochem,
Naperville, IL).
5. 2× 96-well plate (PP).
6. Adhesive tape for plate sealing.
7. Lint-free paper.
8. Vacuum manifold equipped with liquid collection box.
9. Shaking platform.
10. Centrifuge equipped with plate-holder inserts.

2.4 Target Plate 1. Matrix solution: 5 mg/mL 2,5-dihydroxybenzoic acid (2,5-

Spotting DHB) (Bruker Daltonics, Bremen, Germany) 1 mM NaOH in
50 % ACN. Weigh 5 mg 2,5-DHB on an analytical balance,
and dissolve in 1 mL 50 % ACN. Prepare a 100 mM NaOH
solution by adding 5 μL 50 % NaOH to 947 μL water and
mixing thoroughly. Of this, add 10 μL to the matrix solution
(see Note 7). Can be stored at −20 °C.
2. Ethanol.
3. MTP AnchorChip 800/384 TF MALDI target (Bruker
Daltonics).

3 Methods

All steps can be performed at room temperature unless otherwise

noted.

3.1 PNGase F 1. Add 4 μL 2 % SDS to each well of a 96-well plate (see Note 8).
N-Glycan Release 2. Add 2 μL glycoprotein sample to each well and mix briefly by
pipetting up and down (see Notes 9 and 10).
3. Cover the plate with adhesive tape to prevent evaporation (see
Note 11).
4. Incubate the plate for 10 min in a 60 °C oven to denature the
proteins (see Note 12).
5. Remove the plate from the oven, and allow it to return to
room temperature (±5 min).
6. Carefully remove adhesive tape, making sure any condensation
on the tape does not cause cross-contamination of the
samples.
7. Add 4 μL release solution to each well (see Note 8).
 Ethyl Esterification for MALDI-MS Analysis of Protein Glycosylation 155

8. Cover the plate with new adhesive tape to prevent

evaporation.
9. Incubate the plate for 16 h (overnight) in a 37 °C oven. After
this, the PNGase F-released N-glycan mixture may be stored at
−20 °C until further treatment.

3.2 Ethyl 1. Add 20 μL esterification reagent to each well of a 96-well plate

Esterification (see Note 8).
2. Add 1 μL of PNGase F-released N-glycan sample to each well,
and mix briefly by pipetting (see Notes 9 and 13).
3. Cover the plate with adhesive tape to prevent evaporation.
4. Incubate the plate for 1 h in a 37 °C oven (see Note 14).
5. Proceed with HILIC SPE on the now ethyl esterified samples.
Do not store the samples under esterification reaction
conditions.

3.3 Sepharose HILIC 1. Carefully remove the adhesive tape and discard it (see Note 10).
Solid Phase Extraction 2. Add 20 μL pure ACN to each sample and mix briefly by pipet-
ting up and down (see Notes 9 and 15).
3. Place a 96-well filter plate on a vacuum manifold equipped
with a liquid collection box.
4. Transfer 20 μL Sepharose bead slurry to each well of the
96-well filter plate. The Sepharose will sediment rapidly, keep
it thoroughly suspended by mixing the solution before each
pipetting step (see Note 9).
5. Apply low vacuum until the liquid has flown through (see Note
16). However, make sure the filters and beads remain moist
(see Note 17). Any flow-through building up in the collection
box can best be discarded after steps 8 and 15 when the filter
plate is removed from the vacuum manifold.
6. Add 100 μL water to each well, then apply low vacuum until
the liquid has flown through (see Note 9). Repeat two addi-
tional times.
7. Add 100 μL of 85 % ACN to each well, then apply low vacuum
until the liquid has flown through. Repeat two additional times
(see Note 9).
8. Press the filter plate briefly on lint-free paper to remove any
solution adhering to the bottom, then place the filter plate on
an empty 96-well plate (see Note 18).
9. Briefly mix the ethyl esterified samples, then transfer them to
the filter plate (±38 μL). Apply the samples directly to the
beads (see Note 9).
156 Karli R. Reiding et al.

10. Place the filter plate (on the 96-well plate) on a shaking
platform, and incubate for 5 min at maximum velocity (see
Note 19).
11. Transfer the filter plate back to the vacuum manifold and apply
low vacuum until the liquid has flown through.
12. Apply 100 μL 85 % ACN 0.1 % TFA to each well, then apply
low vacuum until the liquid has flown through. Repeat two
additional times (see Notes 9 and 20).
13. Apply 100 μL 85 % ACN to each well, then apply low vacuum
until the liquid has flown through. Repeat two additional times
(see Note 9).
14. Apply additional vacuum to remove residual fluid from the fil-
ter plate.
15. Press the filter plate briefly on lint-free paper to remove solution
adhering to the bottom, then place it on a clean (new) 96-well
plate (see Note 21).
16. Add 30 μL water to each well for elution (see Note 9).
17. Place the stacked plates (filter plate on the 96-well plate) on a
shaking platform, and incubate for 5 min at maximum velocity
(see Note 22).
18. Transfer the stacked plates to a centrifuge equipped with plate-
holder inserts. Use a counter-weight if only one plate will be
centrifuged (see Note 23).
19. Centrifuge for 1 min at 200 × g (see Note 24).
20. The enriched ethyl esterified N-glycans are now in the flow-through.
Do not store the samples before measurement, as this may over
time lead to degradation of the lactonized reaction products.

3.4 MALDI Target 1. Spot 1 μL matrix solution on an AnchorChip MALDI target

Spotting (see Note 9).
2. Immediately add 1 μL purified ethyl esterified N-glycan sample
on the same spot (see Notes 9 and 25).
3. Allow the spots to dry by air.
4. Recrystallize the spots by briefly tapping them with a pipette
tip containing 0.2 μL ethanol. No pressure is required to trans-
fer the ethanol, and a new tip has to be used for each spot (see
Notes 9, 26 and 27).

3.5 MALDI-MS 1. Measure the samples in positive ion mode (see Note 28). When
Measurement preforming TOF-MS the resolution can, due to the derivatiza-
tion step, be increased by reflectron mode measurement (as
compared to linear mode measurement) without leading to
visible metastable decay of the sialylated glycan species.
2. Before sample measurement, use a standard for calibration (see
Note 29).
 Ethyl Esterification for MALDI-MS Analysis of Protein Glycosylation 157

3. Assess laser power requirements for the specific matrix/sample

combination (see Note 30).
4. Record the mass spectra using a random walking algorithm in
order to sample a large part of the spot and limit measurement
variation.
5. When analyzing the spectra, take into account the changes in
sialic acid mass due to the ethyl esterification. The sialic acid
residues have undergone linkage-specific modification, with
the carboxyl groups of α2,6-linked sialic acids having formed
ethyl esters, while α2,3-linked sialic acids are now lactonized.
Consequentially, while unmodified N-acetylneuraminic acid
residues have a mass increment of 291.10 Da, after the reac-
tion the α2,3-linked and α2,6-linked variants will have incre-
ment masses of 273.08 Da and 319.13 Da respectively
(Δ46.04 Da).
6. Be sure to perform quality control on the spectra. While the
protocol should give only minor side reactions, a number of
potential signals have proven to be informative for trouble-
shooting purposes and are listed in Table 1.
7. Fragmentation spectra can be obtained by laser- or collision-
induced dissociation MS/MS. A list of the most common mass
differences suitable for structural assignment is provided in
Table 2.

4 Notes

1. Care should be taken when weighing the SDS powder. Avoid

inhalation, as SDS may cause irritation of the respiratory tract.
2. NP-40 (substitute) is a very viscous solution, and is not easy to
pipette. The tip of a pipette tip can be cut off to widen it and
improve the flow.
3. PNGase F is best removed from the freezer just prior to adding
it to the release solution. It should be returned to −20 °C
immediately afterwards to protect its integrity.
4. Optionally use a flow cabinet for EDC handling. While only of
mild toxicity, the chemical produces a strong odor.
5. When preparing EDC/HOBt in larger volumes, store it in a
glass bottle. Be sure to rinse the bottle beforehand with etha-
nol to prevent contamination of the reagent.
6. The stability of the reagent was tested at various conditions,
and while reactivity will remain secured for over three months
when stored at −20 °C, repeated exposure to room tempera-
ture will remove reactivity within a day. When preparing a large
quantity of reagent that needs storage, be sure to aliquot prior
to storage.
158 Karli R. Reiding et al.

Table 1
Potential satellite signals relative to expected glycan peaks. Whereas amidation and incomplete
esterification affect only the sialic acids, and therefore change the relative distribution of the
analyzed glycans, the reducing end losses and salt variation have an impact on all species and
preserve the relative distribution. In addition to the displayed masses, the presence of peaks in the
spectrum with significantly lower resolution (metastable peaks) may also indicate incomplete
esterification or too high laser power settings

Δmass (Da) Explanation Preventive measure

−367.15 Loss of fucosylated reducing end Lower MALDI laser power
N-acetylglucosamine
−221.09 Loss of reducing end N-acetylglucosamine Lower MALDI laser power
when not fucosylated
0,2
−101.05 A crossring fragment of reducing Lower MALDI laser power
end N-acetylglucosamine
−29.02 Amidation of α2,6-linked sialic acid Limit ammonium-based buffering
rather than ethyl esterification during sample preparation or perform
desalting before ethyl esterification
−6.05 Incomplete ethyl esterification of α2,6-linked Increase amount of reagent relative to
sialic acids with neutral proton to sodium sample
exchange at the carboxylic acid
15.97 [M + K]+ ionization rather than [M + Na]+ Increase NaOH concentration in matrix
ionization or dilute sample before spotting
17.03 Amidation of α2,3-linked sialic acid Limit ammonium-based buffering
rather than lactonization during sample preparation or perform
desalting before ethyl esterification
39.99 Incomplete lactonization of α2,3-linked Increase amount of reagent relative to
sialic acids with neutral proton to sodium sample
exchange at the carboxylic acid
132.19 Unidentified ionization variant Increase NaOH concentration in matrix
rather than [M + Na]+ ionization or dilute sample before spotting
357.18 Unidentified reducing end modification Perform glycan enrichment before ethyl
esterification

7. NaOH is added to the matrix to promote [M + Na]+ ioniza-

tion, particularly limiting the formation of [M + K]+ ions.
8. This step can rapidly be performed with a repeating pipette.
Do not touch the liquid at the bottom of the wells, but rather
pipette to the side of wells taking a new side for each new solu-
tion added to the mix.
9. This step can rapidly be performed with a multichannel pipette
(8 or 12-channel). Samples can directly be transferred from
one 96-well plate to the next, while stock solutions can be put
into reservoirs before pipetting.
 Ethyl Esterification for MALDI-MS Analysis of Protein Glycosylation 159

Table 2
The most commonly observed structurally informative mass differences from a given precursor ion.
Unless multiple fucoses are present on a structure, assigning a fucose to an antenna requires a
−221.09 signal to prove the lack of core fucosylation. Observed losses can be verified at the lower
m/z range of the spectrum as [M + Na]+ ions

Δmass (Da) Explanation

−221.09 Reducing end N-acetylglucosamine when not fucosylated
−319.13 α2,6-linked N-acetylneuraminic acid
−365.13 N-acetyllactosamine (LacNAc) (unsialylated antenna)
−367.15 Reducing end N-acetylglucosamine when fucosylated
−511.19 Fucosylated LacNAc (Lewis A or X)
−638.22 Antenna carrying an α2,3-linked N-acetylneuraminic acid
−684.26 Antenna carrying an α2,6-linked N-acetylneuraminic acid
−784.28 Antenna carrying both an α2,3-linked N-acetylneuraminic acid and a fucose
(sialyl Lewis A or X)

10. Upscaling the volume of the PNGase F release is not a prob-

lem, but downscaling leads to issues with evaporation—keep
10 μL as a minimum total volume unless using smaller wells
(e.g., a 384-well plate).
11. Adhesive tape is a source of contamination when not handled
correctly. Condensation and droplets on the tape can bring
polymers into a sample, while imprecision in replacement can
lead to cross-contamination. Rather use new tape for every
step in the protocol.
12. Use a heating source that has homogeneous temperature
distribution (oven), or a source featuring top-heating (PCR
machine). Bottom-heating-only will cause significant con-
densation on the plate seal, increasing the chance of
contamination.
13. 1:10 sample/reagent still yields complete ethyl esterification
under all tested conditions, and 1:20 has been chosen to have
a margin of error. Lowering the relative amount of reagent
may be attractive when starting with high sample volumes, but
may result in incomplete reactions (most likely due to increased
water content).
14. The temperature and time requirements of the ethyl esterifica-
tion are lenient, as half an hour at room temperature has
already shown to yield reaction completeness for the standards
tested. 1 h and 37 °C were chosen to be controllable condi-
tions suitable for a wide range of samples.
160 Karli R. Reiding et al.

15. The ethyl esterified glycans are retained on Sepharose (or

cotton) directly from a 50:50 ACN/ethanol solution.
16. The vacuum can be just enough to cause flow, which will lead
to maximum interaction time. The vacuum can even be so low
as to not even register on the pressure gauge.
17. Take care not to let the beads dry, as it will decrease the effi-
ciency of all interactions in the purification protocol, including
the sample binding and elution, as well as cause clogging.
18. This empty 96-well plate can be reused across experiments.
19. Sample binding is a critical step in the purification process, so
be sure to allow for at least 5 min of incubation.
20. While the ion pairing agent TFA does increase the overall
purity of the recovered glycan mixture, the acidic conditions it
causes can potentially lead to hydrolysis of the esters formed by
the ethyl esterification. While we did not observe this break-
down, be sure to limit the exposure time of the samples to the
acidic conditions, and rapidly proceed towards the washing
steps without TFA.
21. Droplets of organic solution remaining at the bottom of the
plate may cause cross-contamination, or mix with the eluent,
having a deleterious effect on the glycan profiles obtained.
22. Elution is a critical step in the purification process, be sure to
allow for at least 5 min of incubation.
23. A counterweight plate can be prepared by stacking a filter plate
and elution plate, and adding 30 μL of water as with the actual
samples. This counterweight can be reused for future
experiments.
24. Should water remain in the filter plate (which may happen if
those wells had fallen dry before) try spinning for an additional
1 min or at higher g-force value.
25. The organics in the matrix will start evaporating rapidly when
on the plate, changing the eventual crystallization conditions.
It is recommended to always spot the sample straight after
spotting the matrix, and only then moving on to the next
sample.
26. Recrystallization increases the homogeneity of the sample, and
facilitates automatic measurement. Recrystallization with etha-
nol will not work on a MALDI target without having a hydro-
philic patch within a hydrophobic layer (like an AnchorChip
target, but unlike a polished steel target) as the ethanol will not
be contained on the spot.
27. Next to recrystallization, alternative matrices can be used to create
a homogeneous matrix layer. One example would be super-DHB
(a 9:1 mixture of 2,5-DHB and 2-hydroxy-5-methoxybenzoic acid)
 Ethyl Esterification for MALDI-MS Analysis of Protein Glycosylation 161

(Sigma-Aldrich) which leads to a similar increase in shot-to-shot

repeatability. As downside, super-DHB is less stable in solution
than regular 2,5-DHB, and therefore has to be prepared freshly
more often.
28. While native sialylated glycans preferentially ionize as [M − H]−
and require measurement in negative ion mode, after ethyl
esterification positive ion mode can be used for both sialylated
and unsialylated species. In addition, the increased stability
after modification allows the use of a reflectron for enhanced
resolution.
29. One example of a standard would be peptide calibration stan-
dard (Bruker Daltonics), but any sample can be used that is
amenable to ionization and contains known masses. Preparing
an ethyl esterified glycan standard would increase the similarity
to the actual sample in both the crystallization and ionization
conditions, and may be preferential.
30. For ethyl esterified samples the laser power generally needs to
be higher than for underivatized or reducing end-labeled
samples.

Acknowledgements

This work was supported by the European Union (Seventh

Framework Programme HighGlycan project, grant number
278535).

References
1. Wormald MR, Dwek RA (1999) Glycoproteins: of plasma protein fucosylation. PLoS
glycan presentation and protein-fold stability. Genet 6(12), e1001256. doi:10.1371/jour-
Structure 7(7):R155–R160 nal.pgen.1001256
2. Crocker PR, Paulson JC, Varki A (2007) 6. Vanhooren V, Desmyter L, Liu XE et al (2007)
Siglecs and their roles in the immune N-glycomic changes in serum proteins during
system. Nat Rev Immunol 7(4):255–266. doi: human aging. Rejuvenation Res 10(4):521–
10.1038/nri2056 531a. doi:10.1089/rej.2007.0556
3. Alessandri L, Ouellette D, Acquah A et al 7. Reis CA, Osorio H, Silva L et al (2010)
(2012) Increased serum clearance of oligo- Alterations in glycosylation as biomarkers for
mannose species present on a human IgG1 cancer detection. J Clin Pathol 63(4):322–
molecule. MAbs 4(4):509–520. doi:10.4161/ 329. doi:10.1136/jcp.2009.071035
mabs.20450 8. Knezevic A, Gornik O, Polasek O et al (2010)
4. Moremen KW, Tiemeyer M, Nairn AV (2012) Effects of aging, body mass index, plasma lipid
Vertebrate protein glycosylation: diversity, syn- profiles, and smoking on human plasma
thesis and function. Nat Rev Mol Cell Biol N-glycans. Glycobiology 20(8):959–969.
13(7):448–462. doi:10.1038/nrm3383 doi:10.1093/glycob/cwq051
5. Lauc G, Essafi A, Huffman JE et al (2010) 9. Adamczyk B, Tharmalingam T, Rudd PM
Genomics meets glycomics—the first (2012) Glycans as cancer biomarkers. Biochim
GWAS study of human N-Glycome iden- Biophys Acta 1820(9):1347–1353. doi:10.
tifies HNF1alpha as a master regulator 1016/j.bbagen.2011.12.001
162 Karli R. Reiding et al.

10. Albrecht S, Unwin L, Muniyappa M et al construction. Anal Chem 79(10):3894–3900.

(2014) Glycosylation as a marker for inflam- doi:10.1021/ac070091j
matory arthritis. Cancer Biomark 14(1):17– 17. Liu X, Qiu H, Lee RK et al (2010)
28. doi:10.3233/CBM-130373 Methylamidation for sialoglycomics by
11. Jefferis R (2009) Glycosylation as a strategy to MALDI-MS: a facile derivatization strategy for
improve antibody-based therapeutics. Nat Rev both alpha2,3- and alpha2,6-linked sialic acids.
Drug Discov 8(3):226–234. doi:10.1038/ Anal Chem 82(19):8300–8306. doi:10.1021/
nrd2804 ac101831t
12. Huffman JE, Pucic-Bakovic M, Klaric L et al 18. Wheeler SF, Domann P, Harvey DJ (2009)
(2014) Comparative performance of four Derivatization of sialic acids for stabilization in
methods for high-throughput glycosylation matrix-assisted laser desorption/ionization
analysis of immunoglobulin G in genetic and mass spectrometry and concomitant differen-
epidemiological research. Mol Cell Proteomics tiation of alpha(2 → 3)- and alpha(2 → 6)-iso-
13(6):1598–1610. doi:10.1074/mcp.M113. mers. Rapid Commun Mass Spectrom
037465 23(2):303–312. doi:10.1002/rcm.3867
13. Harvey DJ (1999) Matrix-assisted laser 19. Alley WR Jr, Novotny MV (2010) Glycomic
desorption/ionization mass spectrometry of analysis of sialic acid linkages in glycans derived
carbohydrates. Mass Spectrom Rev 18(6):349– from blood serum glycoproteins. J Proteome
450. doi:10.1002/(SICI)1098-2787(1999) Res 9(6):3062–3072. doi:10.1021/pr901210r
18:6<349::AID-MAS1>3.0.CO;2-H 20. Reiding KR, Blank D, Kuijper DM et al (2014)
14. Powell AK, Harvey DJ (1996) Stabilization of High-throughput profiling of protein
sialic acids in N-linked oligosaccharides and N-glycosylation by MALDI-TOF-MS employ-
gangliosides for analysis by positive ion matrix- ing linkage-specific sialic acid esterification.
assisted laser desorption/ionization mass spec- Anal Chem 86(12):5784–5793. doi:10.1021/
trometry. Rapid Commun Mass Spectrom ac500335t
10(9):1027–1032. doi:10.1002/(SICI)1097- 21. Bondt A, Rombouts Y, Selman MH et al
0231(19960715)10:9<1027::AID-RCM634> (2014) IgG Fab glycosylation analysis using a
3.0.CO;2-Y new mass spectrometric high-throughput pro-
15. Miura Y, Shinohara Y, Furukawa J et al (2007) filing method reveals pregnancy-associated
Rapid and simple solid-phase esterification of changes. Mol Cell Proteomics. doi:10.1074/
sialic acid residues for quantitative glycomics mcp.M114.039537
by mass spectrometry. Chemistry 13(17): 22. Selman MH, Hemayatkar M, Deelder AM
4797–4804. doi:10.1002/chem.200601872 et al (2011) Cotton HILIC SPE microtips for
16. Liu X, Li X, Chan K et al (2007) “One-pot” microscale purification and enrichment of gly-
methylation in glycomics application: esterifi- cans and glycopeptides. Anal Chem
cation of sialic acids and permanent charge 83(7):2492–2499. doi:10.1021/ac1027116
 Chapter 12

Characterization of Protein N-Glycosylation by Analysis

of ZIC-HILIC-Enriched Intact Proteolytic Glycopeptides
Gottfried Pohlentz*, Kristina Marx*, and Michael Mormann

Abstract
Zwitterionic hydrophilic interaction chromatography (ZIC-HILIC) solid-phase extraction (SPE) com-
bined with direct-infusion nanoESI mass spectrometry (MS) and tandem MS/MS is a well-suited method
for the analysis of protein N-glycosylation. A site-specific characterization of N-glycopeptides is achieved
by the combination of proteolytic digestions employing unspecific proteases, glycopeptide enrichment by
use of ZIC-HILIC SPE, and subsequent mass spectrometric analysis. The use of thermolysin or a mixture
of trypsin and chymotrypsin leads per se to a mass-based separation, that is, small nonglycosylated peptides
and almost exclusively glycopeptides at higher m/z values. As a result of their higher hydrophilicity
N-glycopeptides comprising short peptide backbones are preferably accumulated by the ZIC-HILIC-
based separation procedure. By employing this approach complications associated with low ionization
efficiencies of N-glycopeptides resulting from signal suppression in the presence of highly abundant non-
glycosylated peptides can be largely reduced. Here, we describe a simple protocol aimed at the enrichment
of N-glycopeptides derived from in-solution and in-gel digestions of SDS-PAGE-separated glycoproteins
preceding mass spectrometric analysis.

Key words N-glycosylation, Glycopeptides, In-solution digestion, In-gel digestion, NanoESI MS,
CID

1 Introduction

Glycosylation is found in over 50 % of all eucaryotic proteins and is

described as the most complex form of posttranslational modifica-
tion leading to a heterogeneous expression of glycoproteins as
mixtures of glycoforms. The biological role of glycans is highly
variable since glycoproteins have a ubiquitous and complex nature.
They occur inside cells and in extracellular fluids and are embedded
in cell membranes [1]. The glycan moieties are involved in deter-
mination of the physicochemical properties of glycoproteins, e.g.,
charge, size, accessibility, structure, and solubility. Therefore, they

*The first two authors (Pohlentz and Marx) contributed equally and share the first authorship.

Jörg Reinders (ed.), Proteomics in Systems Biology: Methods and Protocols, Methods in Molecular Biology, vol. 1394,
DOI 10.1007/978-1-4939-3341-9_12, © Springer Science+Business Media New York 2016

163
164 Gottfried Pohlentz et al.

can influence structural and modulatory functions which are

important for enzymes, structural proteins, or hormones or are
directly involved in specific recognition of glycans serving as recep-
tors [2, 3]. Thus, glycoproteins are crucial for the develop-
ment, growth, function, or survival of an organism. Hence, their
relationship of structure, location, and function is still an impor-
tant feature in life sciences [4].
The main types of glycosylation are N-glycosylation—where
the glycan is covalently attached to asparagine residues within the
consensus-sequence NXS/T with X as any amino acid except
proline—and O-glycosylation where the glycan is linked covalently
to serine and/or threonine residues of eucaryotic glycoproteins [3].
N-glycosylation is subdivided into three types, viz. complex
type, high mannose type, and hybrid type, with all sharing a com-
mon trimannose-chitobiose core whereas O-glycans show a wide
variety of core structures [2].
Techniques and methodologies aimed at glycosylation analysis
require high sensitivity to provide the detection and separation of
large molecules containing very small structural differences.
Various analytical techniques have been described to determine
protein N-glycosylation [5–9]. Many protocols involve a deglyco-
sylation step using specific glycosidases (e.g., PNGase F) prior to
proteolytic digestion. However, this approach limits the structural
information since evidence on the glycan-protein linkage with
respect to site specificity is lost [10, 11]. This problem can be com-
passed by direct inspection of intact glycopeptide ions derived
from proteolytic digestions by use of nanoESI mass spectrometry
(MS) which can provide information on glycan structure and spe-
cific glycan-protein linkage site [12]. However, direct mass spec-
trometric analysis may still be hampered by low ionization
efficiencies of glycopeptides, signal suppression as a result of highly
abundant nonglycosylated peptides, and a lower overall abundance
of individual glycosylated peptides due to their high structural het-
erogeneity [13]. These problems can be overcome either by use of
high-performance liquid chromatography-mass spectrometry
(HPLC-MS) [14–16] or by enrichment of glycoproteins and
glycopeptides employing solid-phase extraction (SPE) [17, 18].
Recently, zwitterionic hydrophilic interaction liquid chromatogra-
phy (ZIC-HILIC) has been introduced for the selective enrich-
ment of hydrophilic analytes such as glycans or glycopeptides based
on the strong interaction with zwitterionic sulfobetaine moieties
present at the surface of the stationary phase. The first reports on
SPE of tryptic glycopeptides by use of ZIC-HILIC described their
selective enrichment followed by enzymatic deglycosylation and
subsequent separate mass spectrometric analysis of glycans and
deglycosylated peptides. The data obtained allowed for an identifi-
cation of N-glycosylation sites and glycan characterization though
a site-specific assignment of glycosylation was not possible [19–21].
 ZIC-HILIC Enrichment of Proteolytic N-Glycopeptides 165

Recently, we have shown that the combination of less specific

proteases like thermolysin, elastase, or the use of a mixture of
trypsin and chymotrypsin and ZIC-HILIC SPE leads to specific
enrichment of glycopeptides obtained by in-solution digestions
[13, 22–24]. Shorter peptide backbones lead to higher hydrophi-
licity of the analytes and therefore stronger interactions with the
stationary phase. However, this protocol cannot solve the problem
of analyzing complex glycoprotein mixtures. Separation of glyco-
proteins in mixtures prior to their enzymatic degradation, followed
by enrichment of glycopeptides and characterization by mass spec-
trometry, is a prerequisite to obtain an entire and unambiguous
structural characterization. Sodium dodecyl sulfate-polyacrylamide
gel electrophoresis (SDS-PAGE) is a commonly used method for
separation of proteins and also applied to glycoproteins. Separated
(glyco)proteins can be submitted to in-gel digestion and glycopep-
tides can be selectively enriched by use of ZIC-HILIC [25, 26]. A
significant improvement of the method can be achieved if HILIC
SPE is combined with a desalting step using C18-reversed-phase
(RP) SPE (vide infra).
Here, we present a simple protocol for in-solution and in-gel
digestion of glycoproteins followed by selective glycopeptide
enrichment employing ZIC-HILIC SPE. An overview of the
workflow is depicted in Fig. 1. We have chosen bovine IgG as a
representative example. The nanoESI mass spectra obtained
from in-solution and in-gel tryptic/chymotryptic digestions of
bovine IgG (in-gel digestion of the heavy chain) obtained after
different steps of SPE using ZIC-HILIC and/or C18-RP as sta-
tionary phase are depicted in Fig. 2. All detected glycopeptides
are listed in Table 1. Table 2 gives a synopsis of the different
glycan structures at specific glycosylation sites independent of
the length of the peptide backbone. For a simplified assignment
of different glycopeptide species the following notation has been
used: The first number of the code represents the glycan struc-
ture listed in Table 2 (column 1). The following letter refers to
the glycosylation site (cf. Table 1, columns 1 and 2) and the
indices correspond to the different peptide backbones (cf. Table 1,
columns 1 and 3).
The three most abundant glycopeptides (No. 9, 10, 11; cf.
Table 1 and Fig. 2a) can be detected in their ionized form imme-
diately after in-solution digest or in-gel digest with C18-RP SPE
(ZipTip C18) purification without any glycopeptide enrichment.
Figure 2b shows the advantages of ZIC-HILIC separation after
in-solution digest (B1) compared to in-solution digest without
further preparation (B2). Glycopeptides give rise to intense signals
and the number of detected glycoforms increases significantly after
ZIC-HILIC SPE separation (cf. Table 2, column 4 and 5). The
best results for in-gel digests of IgG heavy chains could be observed
after combined purification, i.e., desalting with C18-RP SPE
166 Gottfried Pohlentz et al.

Fig. 1 Strategy for analysis of glycopeptides derived from in-solution and in-gel
digestions of glycoproteins
 ZIC-HILIC Enrichment of Proteolytic N-Glycopeptides 167

Fig. 2 Direct-infusion nanoESI mass spectra obtained from tryptic/chymotryptic digestions of bovine IgG prep-
arations analyzed by different methods: IS = In-solution digest, IG = In-gel digest, ZT = ZipTip desalting,
ZH = ZIC-HILIC enrichment. The code used for peak annotation corresponds to the glycan structures listed in
Table 2 (column 1)—first number, the glycosylation site (see Table 1, columns 1 and 2)—letter, and the peptide
backbone (see Table 1, columns 1 and 3)—indices
Table 1
168

Synopsis of detected glycopeptides derived from in-solution and in-gel tryptic/chymotryptic digestions of bovine IgG (in-gel digestion of heavy chain)
after enrichment (for details see Table 2)

Peak Glycosylation m/z m/z m/z m/ z m/ z m/ z

assignmenta site Peptideb HexNAc Hex Fuc (calc+2) (exp+2) (calc+3) (exp+3) (calc+4) (exp+4)
Bovine IgG1
1a1 N183 EEQFN#STYR 3 3 0 1134.96 1135.02
2a1 N183 EEQFN#STYR 3 4 0 1215.99 1216.06
4a1 N183 EEQFN#STYR 4 4 0 1317.53 1317.46
Gottfried Pohlentz et al.

5a1 N183 EEQFN#STYR 4 5 0 1398.55 1398.48

6a1 N183 EEQFN#STYR 3 2 1 1126.96 1127.15
7a1 N183 EEQFN#STYR 3 3 1 1207.99 1207.90
8a1 N183 EEQFN#STYR 3 4 1 1289.02 1288.96
9a1 N183 EEQFN#STYR 4 3 1 1309.53 1309.45
10a1 N183 EEQFN#STYR 4 4 1 1390.56 1390.49 927.37 927.36
11a1 N183 EEQFN#STYR 4 5 1 1471.58 1471.55 981.39 981.37
13a1 N183 EEQFN#STYR 5 4 1 1492.10 1491.98
14a1 N183 EEQFN#STYR 5 5 1 1573.12 1573.05 1049.08 1049.08
3a2 N183 TRPKEEQFN#STY 4 3 0 933.40 933.38
4a2 N183 TRPKEEQFN#STY 4 4 0 987.42 987.41
5a2 N183 TRPKEEQFN#STY 4 5 0 1041.44 1041.43
7a2 N183 TRPKEEQFN#STY 3 3 1 1371.09 1371.03 914.39 914.38
9a2 N183 TRPKEEQFN#STY 4 3 1 982.09 982.07
10a2 N183 TRPKEEQFN#STY 4 4 1 1036.10 1036.10
11a2 N183 TRPKEEQFN#STY 4 5 1 1090.12 1090.13
12a2 N183 TRPKEEQFN#STY 5 3 1 1049.78 1049.76
13a2 N183 TRPKEEQFN#STY 5 4 1 1103.80 1103.79
14a2 N183 TRPKEEQFN#STY 5 5 1 1157.82 1157.81
1a3 N183 TRPKEEQFN#STYR 3 3 0 917.74 917.71
2a3 N183 TRPKEEQFN#STYR 3 4 0 971.76 971.73
6a3 N183 TRPKEEQFN#STYR 3 2 1 912.41 912.38
7a3 N183 TRPKEEQFN#STYR 3 3 1 966.43 966.43
8a3 N183 TRPKEEQFN#STYR 3 4 1 1020.45 1020.44
9a3 N183 TRPKEEQFN#STYR 4 3 1 1034.12 1034.12 775.84 775.83
 10a3 N183 TRPKEEQFN#STYR 4 4 1 1088.14 1088.14 816.36 816.34
11a3 N183 TRPKEEQFN#STYR 4 5 1 1142.16 1142.17 856.87 856.86
13a3 N183 TRPKEEQFN#STYR 5 4 1 867.13 867.09
14a3 N183 TRPKEEQFN#STYR 5 5 1 907.64 907.66
Bovine IgG2
7b2 N183 TRPNEEQFN#STY 3 3 1 1364.06 1363.95
8b2 N183 TRPNEEQFN#STY 3 4 1 1445.09 1445.01
9b2 N183 TRPNEEQFN#STY 4 3 1 1465.60 1465.53 977.40 977.41
10b2 N183 TRPNEEQFN#STY 4 4 1 1546.63 1546.62 1031.42 1031.42
11b2 N183 TRPNEEQFN#STY 4 5 1 1627.65 1627.57 1085.44 1085.45
12b2 N183 TRPNEEQFN#STY 5 3 1 1045.10 1045.10
13b2 N183 TRPNEEQFN#STY 5 4 1 1099.11 1099.12
14b2 N183 TRPNEEQFN#STY 5 5 1 1153.13 1153.15
7b3 N183 TRPNEEQFN#STYR 3 3 1 961.74 961.73
8b3 N183 TRPNEEQFN#STYR 3 4 1 1015.76 1015.77
9b3 N183 TRPNEEQFN#STYR 4 3 1 1029.44 1029.45
10b3 N183 TRPNEEQFN#STYR 4 4 1 1083.45 1083.47 812.84 812.84
11b3 N183 TRPNEEQFN#STYR 4 5 1 853.36 853.37
14b3 N183 TRPNEEQFN#STYR 5 5 1 904.13 904.13
a
Annotation code of glycopeptides: First number indicates glycan structure (see Table 2), letter denotes glycosylation site, suffix indicates peptide backbone
b
# indicates occupied glycosylation site
ZIC-HILIC Enrichment of Proteolytic N-Glycopeptides
169
170 Gottfried Pohlentz et al.

Table 2
Detected glycan structures in glycopeptides derived from in-solution and in-gel tryptic/chymotryptic
digestions of bovine IgG (in-gel digestion of heavy chain)

Bovine IgG
No./glycosylation sitea Glycan IS IS + ZH IG + ZT IG + ZH IG + ZT + ZH
1
1 N183 − + − + +
2
N183 − − − − −
1
2 N183 − + − − +
2
N183 − − − − −
1
3 N183 − + − − −
2
N183 − − − − −

1
4 N183 − + − − +
2
N183 − − − − −

1
5 N183 − + − − +
2
N183 − − − − −

1
6 N183 − + − + +
2
N183 − − − − −

1
7 N183 + + − + +
2
N183 − + − − +
1
8 N183 + + + + +
2
N183 − + − − +
1
9 N183 + + + + +
2
N183 − + − + +

1
10 N183 + + + + +
2
N183 + + + + +
1
11 N183 + + + + +
2
N183 − + − − +

1
12 N183 − + − − +
2
N183 − + − − −

1
13 N183 − + − − +
2
N183 − + − − −

1
14 N183 − + − − +
2
N183 − + − − +

Sample preparation techniques: IS = In-solution, IG = In-gel, ZT = ZipTip, ZH = ZIC-HILIC. Glycan structures

depicted using the recommendation of the Consortium for Functional Glycomics [30]. Blue square: N-acetylglucosamine,
green circle: mannose, yellow circle: galactose, red triangle: fucose
a
Prefix indicates IgG 1 and IgG 2, respectively; suffix denotes glycosylation site
 ZIC-HILIC Enrichment of Proteolytic N-Glycopeptides 171

preceding ZIC-HILIC SPE separation (cf. Fig. 2c/d; Table 2, col-

umn 6–8). The direct separation with ZIC-HILIC after digests
seems to be hampered caused by electrostatic (ionic) interactions
which occur in the presence of high concentrations of salt ions in
the sample after gel separation.
Eventually, the application of the presented methods in combi-
nation with direct-infusion nanoESI MS enables multiple CID
experiments and therefore facilitates the analysis of N-glycans of
individual glycoproteins separated by SDS-PAGE. Henceforward,
these methods may also support the analysis of unknown glycopro-
teins in complex glycoprotein mixtures.

2 Materials

All aqueous solutions should be prepared using ultrapure water

(18.2 MΩ × cm at 25 °C, Merck Millipore Synergy Ultrapure Water
Systems, Billerica, MA, USA) and analytical grade solvents and
reagents should be employed. All solutions should be prepared
freshly prior to use at ambient temperature.

2.1 Reduction 1. Reduction and alkylation buffer: 6.0 M guanidinium hydro-

and Alkylation chloride, 250 mM trizma base/HCl, and 65 mM dithiothrei-
tol, pH 8.6.
2. Iodoacetamide.

2.2 Size-Exclusion 1. Sephadex G-25 size-exclusion chromatography columns (illus-

Chromatography tra NAP™ 5 columns, volume 0.5 ml, GE Healthcare Europe
GmbH, Freiburg, Germany). Alternatively, polyacrylamide gel
columns (Micro Bio-Spin Biogel P-6 chromatography col-
umns, Bio-Rad Laboratories GmbH, München, Germany) can
be used (see Note 1).
2. Column equilibration and elution buffer: 50 mM Ammonium
bicarbonate.

2.3 In-Solution 1. Digestion buffer: 10 mM ammonium bicarbonate.

Digestion 2. Proteases: Trypsin (0.1 μg/μl in 1 mM HCl), sequencing
grade (Roche Diagnostics, Mannheim, Germany);
chymotrypsin (0.1 μg/μl in 1 mM HCl), sequencing grade
(Roche Diagnostics, Mannheim, Germany);
thermolysin (0.5 μg/μl in H2O) (Sigma-Aldrich Chemie
GmbH, Taufkirchen, Germany) (see Note 2).

2.4 In-Gel Digestion 1. Destaining solutions: 100 mM Ammonium bicarbonate/ace-

tonitrile (50/50) and neat acetonitrile.
2. Digestion buffer: 10 mM Ammonium bicarbonate.
172 Gottfried Pohlentz et al.

3. Proteases: Trypsin (0.1 μg/μl in 1 mM HCl), sequencing

grade (Roche Diagnostics, Mannheim, Germany);
chymotrypsin (0.1 μg/μl in 1 mM HCl), sequencing grade
(Roche Diagnostics, Mannheim, Germany);
thermolysin (0.5 μg/μl in H2O) (Sigma-Aldrich Chemie
GmbH, Taufkirchen, Germany) (see Note 2).
4. Extraction buffers: Acetonitrile/water + formic acid
(50/50 + 5), acetonitrile/water + formic acid (80/20 + 5),
acetonitrile.

2.5 Desalting 1. Solid-phase extraction C18-RP tips: ZipTip C18, Tip Size
of Extracted P10, (Merck Chemicals GmbH, Schwalbach, Germany).
Proteolytic Peptides 2. Wetting solution: Acetonitrile.
3. Binding solution: Trifluoroacetic acid (0.1 %).
4. Equilibration solution: Trifluoroacetic acid (0.1 %).
5. Wash solution: Trifluoroacetic acid (0.1 %).
6. Elution solution 1: Trifluoroacetic acid (0.1 %)/acetonitrile
(50/50).
7. Elution solution 2: Trifluoroacetic acid (0.1 %)/acetonitrile
(20/80).
8. Elution solution 3: Acetonitrile.

2.6 ZIC-HILIC 1. Solid-phase extraction ZIC-HILIC tips: ZIC-HILIC

Enrichment ProteaTip, 10–200 μl (dichrom GmbH, Marl, Germany).
of Proteolytic 2. Binding solution: Acetonitrile/water + formic acid (80/20 + 2).
N-glycopeptides
3. Equilibration solution: Acetonitrile/water + formic acid
(80/20 + 2).
4. Wash solution: Acetonitrile/water + formic acid (80/20 + 2).
5. Elution solution: Water + formic acid (98/2).

2.7 Nano 1. Sample buffer: Acetonitrile/water + formic acid (50/50 + 2).

Electrospray Mass 2. Mass spectrometer: Quadrupole time-of-flight (Q-TOF) mass
Spectrometry spectrometer (Micromass, Manchester, UK) equipped with a
(NanoESI MS) Z-spray source in the positive ion mode. Typical source
parameters: source temperature: 80 °C, desolvation gas (N2)
flow rate: 75 l/h, capillary voltage: 1.1 kV, cone voltage: 30
V. Low energy CID parameters: collision gas (Ar) pressure:
3.0 × 10−3 Pa, collision energies: 20–40 eV (Elab).

3 Methods

All procedures should be performed at ambient temperature unless

otherwise specified.
 ZIC-HILIC Enrichment of Proteolytic N-Glycopeptides 173

3.1 Reduction 1. For in-solution reduction and alkylation 1 nmol of glycopro-

and Alkylation tein is dissolved in 200 μl of a mixture of 6.0 M guanidinium
of Glycoproteins hydrochlorid, 250 mM trizma base/HCl, and 65 mM
dithiothreitol.
2. The resulting mixture is incubated for 1 h at 56 °C.
3. 2.5 mg Iodoacetamide is added and the resulting mixture is
incubated for 45 min under exclusion of light.

3.2 Desalting 1. Allow excess packing buffer to drain from the NAP™ 5 column
by Size-Exclusion by gravity to the top of the gel bed.
Chromatography 2. Apply 2.5 ml of 50 mM ammonium bicarbonate and allow the
(Sephadex G-25) buffer to drain out by gravity to the top of the gel bed. Repeat
this step four times.
3. Carefully add the sample obtained under Subheading 3.1 to
the column and allow the sample to penetrate the gel bed
completely.
4. Apply 500 μl of 50 mM ammonium bicarbonate and allow the
buffer to drain out by gravity to the top of the gel bed. The
flow through is dicarded.
5. Place an appropriate collection tube under the column, apply
500 μl of 50 mM ammonium bicarbonate, and allow the buf-
fer to drain out by gravity to the top of the gel bed.
6. The sample is dried by removal of the solvent in vacuo by use
of a centrifugal evaporator.

3.3 Desalting The experimental procedure follows the manufacturer’s instructions.

by Size-Exclusion
1. Resuspend settled gel by sharply inverting the Micro Bio-Spin
Chromatography
Biogel P-6 chromatography column several times and remove
(Biogel P-6) residual air bubbles by tapping the column.
2. Snap off the tip of the column and place in a 2 ml collection tube.
3. Remove the cap and allow excess packing buffer to drain from
column by gravity to the top of the gel bed. If the buffer flow
does not start immediately push back the cap on the column
and remove again to initiate the buffer flow.
4. The packing buffer is discarded and the column is placed back
in the collection tube. The column is centrifuged in a micro-
centrifuge for 2 min at 1000 × g. Subsequently, the flow
through is discarded and the column is placed back in the col-
lection tube.
5. Apply 500 μl of 50 mM ammonium bicarbonate, drain buffer by
centrifugation for 1 min at 1000 × g, discard buffer, and place
back column in the collection tube. Repeat this step four times.
6. Place column in an appropriate collection tube (1.5 ml),
carefully add the sample obtained under Subheading 3.1 (20–
174 Gottfried Pohlentz et al.

75 μl) to the column, and allow the sample to penetrate the gel
bed completely. Centrifuge the assembly for 4 min at 1000 × g.
7. The sample is dried by removal of the solvent in vacuo by use
of a centrifugal evaporator.

3.4 In-Solution 1. For in-solution digestion 100 pmol of either the reduced and
Digestion alkylated or the untreated glycoprotein is dissolved in 20 μl 10
mM ammonium bicarbonate (see Note 3).
2. Shake the mixture for 7 min at 95 °C (see Note 4) and chill
tubes to room temperature.
3. Add 1 μl of the protease solution and incubate the reaction
mixture overnight at 37 °C in a shaker (750 rpm) (see Note 5).
If thermolysin is employed as protease incubate at 65 °C in a
shaker (750 rpm).
4. The sample is dried by removal of the solvent in vacuo by use
of a centrifugal evaporator.
5. Add 50 μl water and dry in vacuo by use of a centrifugal evapo-
rator. Repeat this step at least once.

3.5 In-Gel Digestion If reduced and alkylated (glyco-)proteins have been separated by
use of one-dimensional or two-dimensional sodium dodecyl
sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and visu-
alized by either silver or Coomassie® brilliant blue staining glyco-
peptides are formed by in-gel digestion (see Note 3). The in-gel
digestion procedure follows the method described by Shevchenko
et al. [27].
1. Excise protein bands or spots by use of a clean scalpel.
2. Cut bands into small pieces (see Note 6) and transfer gel cubes
into a reaction tube of appropriate volume (0.5 or 1.5 ml).
3. Add 100 μl of 100 mM ammonium bicarbonate/acetonitrile
(50/50) and shake for 30 min. Subsequently, remove destain-
ing solution.
4. Add 500 μl of acetonitrile and shake for 10 min until gel pieces
shrink and become opaque. Remove supernatant and dry gel
pieces in vacuo by use of a centrifugal evaporator.
5. Add 30 μl of 10 mM ammonium bicarbonate and 10 μl of pro-
tease solution and incubate gel pieces for 30 min on an ice bath.
6. If digestion buffer solution is absorbed entirely add an appro-
priate amount of 10 mM ammonium bicarbonate and protease
solution (3/1, v/v) to cover gel pieces completely. Incubate
reaction mixture for 30 min on an ice bath (see Note 7).
7. Add an appropriate amount of 10 mM ammonium bicarbonate
to cover gel pieces completely and incubate the reaction mixture
overnight at 37 °C in a shaker (750 rpm). If thermolysin is
employed as protease incubate at 65 °C in a shaker (750 rpm).
 ZIC-HILIC Enrichment of Proteolytic N-Glycopeptides 175

8. Allow reaction mixture to chill to ambient room temperature

and spin down gel pieces and solution.
9. Withdraw supernatant and collect the solution in an appropri-
ate collection tube.
10. Add 100 μl acetonitrile/water (50/50) containing 5 % formic
acid and shake for 15 min at 37 °C. Withdraw supernatant and
collect the extract (see Note 8).
11. Add 100 μl acetonitrile/water (80/20) containing 5 % formic
acid and shake for 15 min at 37 °C. Withdraw supernatant and
collect the extract (see Note 8).
12. Add 100 μl of neat acetonitrile and shake for 15 min at 37
°C. Withdraw supernatant and collect the extract (see Note 8).
13. Combine extracts and supernatant obtained under Subheading
3.5, step 9, and remove solvents in vacuo by use of a centrifu-
gal evaporator.

3.6 Desalting Proteolytic peptides obtained from in-gel-digestions are desalted

of Extracted prior to glycopeptide enrichment by reversed-phase solid-phase
Proteolytic Peptides extraction employing pipette tips containing an immobilized C18
resin, viz. C18 ZipTip pipette tips. The experimental procedure
follows the manufacturer’s instructions.
1. Samples are dissolved in 10 μl of trifluoroacetic acid (0.1 %).
2. Attach C18 ZipTip pipette tip to a compatible 10 μl pipettor,
aspirate 10 μl wetting solution, and dispense to waste. Repeat
this step three times.
3. Aspirate three times 10 μl equilibration solution and dispense
to waste.
4. Load sample by aspirating and dispensing the sample solution
at least ten times.
5. Aspirate 10 μl wash solution and dispense into a fresh reaction
tube of appropriate volume. Repeat this step three times
( see Note 9).
6. Peptides are released by consecutively aspirating and dispensing
five times 10 μl of elution solution 1, 2, and 3. Eluates are collected
and combined in a fresh reaction tube of appropriate volume and
finally dried in vacuo by use of a centrifugal evaporator.

3.7 ZIC-HILIC The following procedure can be directly applied to N-glycopeptides

Enrichment derived from in-solution digestion of glycoproteins (Subheading
of Proteolytic 3.4) or to N-glycopeptides obtained by in-gel digestions purified
N-glycopeptides by reversed-phase solid-phase extraction.
1. Samples are dissolved in 15 μl of binding solution.
2. Attach ZIC-HILIC ProteaTip pipette tip to a compatible
pipettor, aspirate 15 μl equilibration solution, and dispense to
waste. Repeat this step five times.
176 Gottfried Pohlentz et al.

3. Load sample by aspirating and dispensing the sample solution

(15 μl) at least 20 times.
4. Aspirate 15 μl wash solution and dispense in a fresh reaction
tube of appropriate volume. Repeat this step five times (see
Note 9).
5. Glycopeptides are released by consecutively aspirating and dis-
pensing ten times 15 μl of the elution solution. Eluates are
collected in a fresh reaction tube of appropriate volume and
finally dried in vacuo by use of a centrifugal evaporator.

3.8 Nano 1. Mass spectra of glycopeptides enriched by ZIC-HILIC obtained

Electrospray Mass under electrospray conditions in positive ion mode typically
Spectrometry exhibit the analytes in charge states from +2 to +4. Analytes are
(NanoESI MS) mainly found in their corresponding protonated form or as ionic
species with one or two protons replaced by sodium ions.
Subsequent to the application of a mass deconvolution glycosyl-
ated species are noticed straightforward by identifying series of
ions harboring the same peptide backbone but different glycan
isoforms linked to the same glycosylation site. These analyte mol-
ecules differ by typical mass increments characteristic for glycan
building blocks, i.e., 132.042 Da (pentose), 146.058 (deoxyhex-
ose), 162.053 Da (hexose), 203.079 Da (N-acetylhexosamine),
291.095 Da (neuraminic acid), or combinations of these masses.
Collisional activation of selected glycopeptide precursor ions
mainly gives rise to fragment ions originating from cleavage of
glycosidic bonds. Typically, B- and Y-type fragment ions lead to
intense signals in the CID spectra of glycopeptide ions (nomen-
clature according to Domon and Costello [28]). While the latter
comprise the reducing end and thus the peptide moiety B-type
oxonium ions harbor the non-reducing end of the glycan and
exhibit characteristic m/z-values: m/z([HexNAc − H2O + H]+):
204.087, m/z([NeuAc − H2O + H]+): 292.103, m/z([HexNAc-
Hex − H2O + H]+): 366.139, m/z([HexNAc-Hex2 − H2O + H]+):
528.192, etc. N-glycan biosynthesis leads to a common core
sugar sequence attached to a highly conserved sequon, viz.
Manα1–6(Man α1–3)Manβ1–4GlcNAcβ1–4GlcNAcβ1–Asn–X–
Ser/Thr
(X = any amino acid except proline) which is typically
extended to finally yield either high-mannose-, complex-, or
hybrid-type glycans. Combining this blueprint with the infor-
mation deduced from the appearance of specific B- and Y-type
fragment ions finally leads to a structural assignment of the
N-glycan structure. Loss of the entire oligosaccharide chain
liberates the deglycosylated peptide residue that might be
identified by its exact mass if the amino acid sequence of the
protein under inspection is known, thus giving rise to the
glycosylation site (see Note 10).
 ZIC-HILIC Enrichment of Proteolytic N-Glycopeptides 177

4 Notes

1. The use of micro Bio-Spin Chromatography columns packed

with polyacrylamide gel (Bio-Gel P-6) for removal of salts and
other low-molecular-weight compounds by size-exclusion
chromatography is recommended for the purification of lectins
and other carbohydrate-binding proteins. These analytes are
often prone to strong binding to the Sephadex G-25 gel matrix
and (glyco-) protein recovery is typically very low.
2. Protease solutions may be stored at −20 °C for 1 month without
significant loss of catalytic activity.
3. If the current study also aims at characterization and identifica-
tion of disulfide bridges in glycoproteins one aliquot of protein
should be submitted to proteolytic digestion without a preced-
ing reduction and alkylation step. Under nanoESI-CID condi-
tions fragmentation of intra- and inter-peptide disulfide bonds
of proteolytic peptides provides sufficient information for their
determination. Collisional activation of proteolytic peptides
comprising a disulfide bridge gives rise to a set b- and of y-type
fragment ions which typically allow the determination of the
sequence of the amino acids located outside the disulfide loop.
Additionally, fragment ion spectra reveal the presence of low-
abundance fragment ions formed by the cleavage of peptide
bonds within the disulfide loop. These fragmentations are pre-
ceded by asymmetric cleavage of the disulfide bridge, giving
rise to a modified cysteine containing a disulfohydryl substitu-
ent and a dehydroalanine residue on the remote cleavage site
[22–24, 29].
4. Thermal (glyco-)protein denaturation typically increases acces-
sibility of cleavage sites and thus improves efficiency of proteo-
lytic digestion.
5. The selection of an appropriate protease strongly depends on
the primary structure of the (glyco-)protein under inspection.
Since the use of trypsin as protease usually gives rise to very
large glycopeptide species digestions using unspecific proteases
such as thermolysin or a 1:1 mixture of trypsin and chymotryp-
sin furnish rather short nonglycosylated peptides and glycopep-
tides of significantly higher molecular mass. Owing to the high
hydrophilicity of N-glycopeptides harboring a large glycan
moiety and a short peptide stretch an efficient enrichment by
ZIC-HILIC can be achieved [13].
6. If by accident gel bands/spots were divided into very small
pieces leading to clogging of regular pipette tips microloader
tips (Eppendorf AG, Hamburg, Germany) can be used instead.
178 Gottfried Pohlentz et al.

7. If – after 30 min – protease solution is not absorbed entirely

supernatant should be removed by use of either a regular
pipette tip or a microloader tip (see Note 6), discard superna-
tant. According to our experience this step largely reduces the
occurrence of autoproteolytic peptide ions in the resulting
mass spectra.
8. Notwithstanding the protocol described by Shevchenko et al.
[27] proteolytic peptides are extracted by use of solvents with
decreasing polarity.
9. Wash solutions should be retained until the actual sample has
been analyzed to avoid unintended sample loss.
10. Most low-energy CID spectra of N-glycopeptide ions also
contain fragment ions formed by cleavage of the peptide back-
bone. These ionic species can be used to deduce amino acid
sequence stretches that lead to an unambiguous identification
of the glycosylation site (for an example refer to ref. 22).

References

1. Rudd PM, Dwek RA (1997) Glycosylation: 11. Künneken K, Pohlentz G, Schmidt-Hederich

heterogeneity and the 3D structure of pro- A et al (2004) Recombinant human laminin-5
teins. Crit Rev Biochem Mol Biol 32:1–100 domains. Effects of heterotrimerization, pro-
2. Dwek RA (1996) Glycobiology: toward under- teolytic processing, and N-glycosylation on
standing the function of sugars. Chem Rev alpha3beta1 integrin binding. J Biol Chem
96:683–720 279:5184–5193
3. Varki A, Cummings RD, Esko JD et al (2009) 12. Henning S, Peter-Katalinić J, Pohlentz G
Essentials of glycobiology, 2nd edn. Cold (2007) Structure elucidation of glycoproteins
Spring Harbor Laboratory Press, Cold Spring by direct nanoESI MS and MS/MS analysis of
Harbor, NY proteolytic glycopeptides. J Mass Spectrom
4. Hart GW, Copeland RJ (2010) Glycomics hits 42:1415–1421
the big time. Cell 143:672–676 13. Neue K, Mormann M, Peter-Katalinić J et al
5. Geyer H, Geyer R (2006) Strategies for analy- (2011) Elucidation of glycoprotein structures
sis of glycoprotein glycosylation. Biochim by unspecific proteolysis and direct nanoESI
Biophys Acta 1764:1853–1869 mass spectrometric analysis of ZIC-HILIC-
enriched glycopeptides. J Proteome Res
6. Marino K, Bones J, Kattla JJ et al (2010) A 10:2248–2260
systematic approach to protein glycosylation
analysis: a path through the maze. Nat Chem 14. Wuhrer M, Deelder AM, Hokke CH (2005)
Biol 6:713–723 Protein glycosylation analysis by liquid
chromatography-mass spectrometry.
7. Zauner G, Selman MHJ, Bondt A et al (2013) J Chromatogr B 825:124–133
Glycoproteomic analysis of antibodies. Mol
Cell Proteomics 12:856–865 15. Zauner G, Koeleman CAM, Deelder AM et al
(2013) Nano-HPLC-MS of glycopeptides
8. Desaire H (2013) Glycopeptide analysis, recent obtained after non-specific proteolysis. In:
developments and applications. Mol Cell Kohler JJ, Patrie SM (eds) Mass spectrometry
Proteomics 12:893–901 of glycoproteins, vol 951, Methods in molecu-
9. Alley WR Jr, Mann BF, Novotny MV (2013) lar biology. Springer, Heidelberg, pp 113–127
High-sensitivity analytical approaches for the 16. Wuhrer M, de Boer AR, Deelder AM (2009)
structural characterization of glycoproteins. Structural glycomics using hydrophilic interac-
Chem Rev 113:2668–2732 tion chromatography (HILIC) with mass spec-
10. Haslam SM, North SJ, Dell A (2006) Mass trometry. Mass Spectrom Rev 28:192–206
spectrometric analysis of N- and O-glycosylation 17. Chen CC, Su WC, Huang BY et al (2014)
of tissues and cells. Curr Opin Struct Biol Interaction modes and approaches to glyco-
16:584–591
 ZIC-HILIC Enrichment of Proteolytic N-Glycopeptides 179

peptide and glycoprotein enrichment. Analyst 24. Reissner C, Stahn J, Breuer D et al (2014)
139:688–704 Dystroglycan binding to α-neurexin competes
18. Huang BY, Yang CK, Liu CP et al (2014) with neurexophilin-1 and neuroligin in the
Stationary phases for the enrichment of glyco- brain. J Biol Chem 289:27585–27603
proteins and glycopeptides. Electrophoresis 25. Gnanesh Kumar BS, Pohlentz G, Mormann M
35:2091–2107 et al (2013) Characterization of α-mannosidase
19. Hägglund P, Bunkenborg J, Elortza F et al from Dolichos lablab seeds: partial amino acid
(2004) A new strategy for identification of sequencing and N-glycan analysis. Protein
N-glycosylated proteins and unambiguous Expr Purif 89:7–15
assignment of their glycosylation sites using 26. Gnanesh Kumar BS, Pohlentz G, Schulte M
HILIC enrichment and partial deglycosylation. et al (2014) N-glycan analysis of mannose/
J Proteome Res 3:556–566 glucose specific lectin from Dolichos lablab
20. Thaysen-Andersen M, Højrup P (2006) seeds. Int J Biol Macromol 69:400–407
Enrichment and characterization of glycopep- 27. Shevchenko A, Tomas H, Havlis J et al (2006)
tides from gel-separated glycoproteins. Am In-gel digestion for mass spectrometric charac-
Biotechnol Lab 24:14–17 terization of proteins and proteomes. Nat
21. Picariello G, Ferranti P, Mamone G et al (2008) Protoc 1:2856–2860
Identification of N-linked glycoproteins in 28. Domon B, Costello CE (1988) A systematic
human milk by hydrophilic interaction liquid nomenclature for carbohydrate fragmentations
chromatography and mass spectrometry. in FAB-MS MS spectra of glycoconjugates.
Proteomics 8:3833–3847 Glycoconjugate J 5:397–409
22. Gnanesh Kumar BS, Pohlentz G, Schulte M 29. Mormann M, Eble J, Schwoeppe C et al (2008)
et al (2014) Jack bean α-mannosidase: amino Fragmentation of intra-peptide and inter-
acid sequencing and N-glycosylation analysis of peptide disulfide bonds of proteolytic peptides
a valuable glycomics tool. Glycobiology by nanoESI collision-induced dissociation.
24:252–261 Anal Bioanal Chem 392:831–838
23. Sharma A, Pohlentz G, Bobbili KB et al (2013) 30. Consortium for Functional Glycomics. CFG
The sequence and structure of snake gourd functionalglycomicsgateway. http://www.
(Trichosanthes anguina) seed lectin, a three- functionalglycomics.org/static/consortium/
chain nontoxic homologue of type II RIPs. Acta CFGnomenclature.pdf. Accessed 20 July
Crystallogr D Biol Crystallogr 69:1493–1503 2012
 Chapter 13

Simple and Effective Affinity Purification Procedures

for Mass Spectrometry-Based Identification of Protein-
Protein Interactions in Cell Signaling Pathways
Julian H.M. Kwan and Andrew Emili

Abstract
Identification of protein-protein interactions can be a critical step in understanding the function and
regulation of a particular protein and for exploring intracellular signaling cascades. Affinity purification
coupled to mass spectrometry (APMS) is a powerful method for isolating and characterizing protein
complexes. This approach involves the tagging and subsequent enrichment of a protein of interest along
with any stably associated proteins that bind to it, followed by the identification of the interacting proteins
using mass spectrometry. The protocol described here offers a quick and simple method for routine sample
preparation for APMS analysis of suitably tagged human cell lines.

Key words Affinity purification-mass spectrometry, Cell signaling pathway, Liquid chromatography-mass
spectrometry

1 Introduction

The identification and mapping of protein-protein interactions,

which underpin many fundamental biological processes including
intracellular signaling pathways, is an important aspect of under-
standing protein function and regulation. Affinity purification-
mass spectrometry (APMS) is a powerful tool for identifying the
components of multi-protein complexes [1–8]. The fundamental
basis of APMS is the use of one or more high-affinity recognition
reagents (such as an antibody) to selectively enrich a particular tar-
get protein, together with physically associated factors, relative to
the myriad of functionally unrelated proteins present in a cell
extract or tissue lysate.
The endogenous form of the protein of interest may be
enriched using a specific antibody. However this approach may not
always be the most expedient since the generation of antibodies of
sufficient specificity and affinity can be difficult and expensive.

Jörg Reinders (ed.), Proteomics in Systems Biology: Methods and Protocols, Methods in Molecular Biology, vol. 1394,
DOI 10.1007/978-1-4939-3341-9_13, © Springer Science+Business Media New York 2016

181
182 Julian H.M. Kwan and Andrew Emili

A popular and effective alternative is to use molecular cloning

techniques to introduce an affinity tag (e.g., epitope or small protein
like GFP) as a fusion to the open reading frame corresponding to
the protein of interest. Using a tag/antibody pair that has been
established to exhibit sufficient affinity and specificity avoids the
need to generate a specific antibody for each target protein, and
can lead to more efficient and consistent APMS results [8]. In this
protocol, we suggest the use of FLAG or GFP epitope tags.
There are many tools, options, and caveats for the generation of
suitable cell lines or tissue samples expressing an affinity-tagged pro-
tein of interest. These might range from endogenous tagging of loci
in transgenic animals [9] to transient transfection of plasmids
over-expressing a particular target in cell culture [10]. The protocol
reported here provides a simple procedure that we have found to be
effective for APMS analysis of tagged proteins expressed in the com-
mon tissue culture cell line HEK 293T. In principle the protocol can
be easily adapted to other cell lines or biological samples with two key
considerations. The first requirement is that the protein of interest
must be solubilized under gentle lysis conditions, which must not be
too harsh so as to disrupt physiologically relevant protein-protein
interactions. The second requirement is that the sample remains
compatible with subsequent mass spectrometry analysis; this might
mean taking steps to reduce the levels of potential contaminants,
such as detergents, which are detrimental to protein identification.
When performed properly, affinity purification will enrich pro-
teins (i.e., prey) that are physically bound to a protein of interest
(i.e., the bait); these are the interacting proteins one intends to
identify. However, many other nonspecific proteins will also bind
(i.e., artifacts). The inclusion of control purifications (e.g., no
tag/bait, or irrelevant bait) is important for distinguishing genuine
protein-protein interactions from nonspecific proteins. In the sim-
plest interpretation of APMS data, proteins identified together
with the bait are considered to interact unless they are also detected
with the negative control(s). There are two key considerations for
why this interpretation might not be sufficient. First, the bait may
specifically interact with a prey protein that is also found in the
control, but at a significantly lower level. In such a scenario,
considering the prey’s enrichment relative to the control is impor-
tant. The second consideration takes into account the inconsistent
recovery and identification of some proteins by APMS due to vari-
ations in experimental and biological conditions. The same baits
processed repeatedly over several different experiments may each
identify preys that are unique to the run. This is true of both the
experimental and control samples. To address this source of spurious
results (i.e., potential false positives), one must consider detection
reproducibility across runs. This variability also underlines the
importance of performing independent biological replicates, such
as by performing two sets of APMS experiments starting from
different cell culture batches on different days.
 Simple and Effective Affinity Purification Procedures for Mass Spectrometry-Based… 183

Different algorithms can be used to assign a score to each

potential protein interaction, which represents how likely a prey
protein is to genuinely interact with the bait. Popular algorithms,
such as compPASS [6] and SAINT [11, 12], take into account
both the reproducibility and the enrichment relative to controls to
systematically assign a score which indicates the likelihood that a
particular candidate interactor is specific (i.e., a genuinely interacting
prey protein). Such algorithms are particularly useful for analysis of
larger APMS datasets, with multiple baits, since they can use data
from each other purification as additional controls between baits
and thus generate a more robust census of nonspecifically adsorbed
proteins versus biologically relevant interactors captured by your
experimental procedure.
The steps described in the protocol that follows cover generat-
ing soluble protein cell lysate, capturing the protein of interest on
a solid support, washing to remove unbound proteins, and pro-
cessing bound proteins for mass spectrometry identification. Since
the methods of data acquisition and final format of results will vary
based on the instrumentation, protocols, and software favored by
the mass spectrometry (MS) facility, this protocol will not cover
the details of MS procedures and data analysis.

2 Materials

1. Sample: 2 × 150 mm dishes per sample of HEK 293T cells

(80–90 % confluent) expressing GFP or FLAG epitope-tagged
bait protein of interest.
2. Lysis buffer: TBS (30 mM Tris–HCl, pH 7.5, 150 mM NaCl)
with 0.5 % Nonidet P40 and protease and phosphatase
inhibitors.
3. Affinity media: Antibody (Life Technologies anti-GFP
#G10362, or anti-FLAG #F1804 as appropriate for affinity
tag), Dynabeads Protein G (Life Technologies #10003D).
4. Wash buffer: TBS (30 mM Tris–HCl, pH 7.5, 150 mM NaCl)
without detergent and protease and phosphatase inhibitors
(see Note 1).
5. Trypsin digestion buffer: 50 mM ammonium bicarbonate
(see Note 2).
6. Proteolytic digestion: Trypsin sequencing grade (Roche)
(see Note 3).
7. Digestion termination: Formic acid (LC/MS grade).
8. Desalting media: C-18 cartridge (10–200 L NuTip; Glygen
Corp #NT2C18.96).
9. Solution A: H2O (HPLC grade) 0.1 % formic acid.
10. Solution B: 70 % Acetonitrile, 0.1 % formic acid.
184 Julian H.M. Kwan and Andrew Emili

3 Methods

1. Remove media from cell culture plates (see Note 4).

2. Lyse cells on dish in 1 ml cold lysis buffer per 150 mm dish,
and incubate for 5 min on ice or in cold room/fridge (see Note 5).
3. Scrape lysate into pre-chilled 1.5 ml microcentrifuge tubes
(see Note 6).
4. Spin down cell debris (20 min, 20,000 × g, 4 °C) (see Note 7).
5. Transfer supernatant to fresh pre-chilled tubes; the pellet of
cell debris may be discarded (see Note 8).
6. Add 1 μg of appropriate antibody to each sample (see Note 9).
7. Incubate samples with end-over-end rotation at 4 °C (cold
room) for 1–2 h.
8. Wash 40 μl Protein G (Dynabead) bead slurry per sample
with lysis buffer and resuspend in same volume of lysis buffer
(see Note 10).
9. Add 40 μl washed Dynabead Protein G slurry to each sample
and incubate with rotation at 4 °C for an additional 1 h
(see Note 11).
10. Collect beads and wash with 1.5 ml cold wash buffer (repeat)
(see Note 12).
11. Resuspend beads in 1 ml wash buffer and transfer to a new
pre-chilled microfuge tube (see Note 13).
12. Wash beads with 400 μl trypsin digestion buffer (see Note 14).
13. Resuspend beads in 20 μl trypsin digestion buffer and add
750 ng trypsin (see Note 15).
14. Digest with rotation (end over end) at 37 °C for 4 h to over-
night (see Note 16).
15. Magnetize beads, transfer supernatant to a fresh tube, and
then add another 750 ng trypsin. Incubate at 37 °C with agita-
tion for 4 h (see Note 17).
16. Add formic acid (2 % final concentration) to sample to termi-
nate digestion.
17. Condition desalting media using multiple rounds (10×) of aspi-
ration and discharge of 200 μl solution B (repeat) (see Note 18).
18. Condition desalting media by 10× aspiration and discharge of
200 μl solution A (repeat 2× with new aliquots of solution A)
(see Note 19).
19. Load sample onto desalting media by 20–50× aspiration and
discharge of digested sample (see Note 20).
20. Wash sample by 10× aspiration and discharge of 200 μl solution A
(repeat 2×) (see Note 21).
 Simple and Effective Affinity Purification Procedures for Mass Spectrometry-Based… 185

21. Elute sample by 20× aspiration and discharge of 20 μl solution B.

22. Lyophilize eluate using a vacuum concentrator. After drying,
samples may be stored at −20 °C until resuspension in 20 μl
of solution A immediately prior to analysis by liquid
chromatography-mass spectrometry (see Note 22).

4 Notes

1. Wash buffer omits detergent to reduce detergent contamina-

tion that can hinder mass spectrometry identification. Protease
and phosphatase inhibitors may also be omitted because the
sample is not expected to be exposed to the buffer for an
extended period of time.
2. Make with HPLC-grade water, pH should be about 8.0, and
do not adjust.
3. The lyophilized trypsin should be resuspended in trypsin
digestion buffer at 0.5 μg/μl (instead of acetic acid as recom-
mended by the manufacturer).
4. You may rinse the cells gently with cold PBS to remove remain-
ing media.
5. This lysis condition is suitable for solubilization of cytoplasmic
proteins, but is not well suited for solubilization of nuclear or
membrane proteins.
6. Since lysis occurs in a small volume, use a cell lifter or rubber
policeman to scrape cell lysate to the bottom of the dish; this
insures maximum sample recovery.
7. This spin can be done in a refrigerated tabletop centrifuge.
8. Input amount can be scaled up or down as needed; for 2×
150 mm plates per sample (suggested), the supernatant from
two microfuge tubes is combined in a 5 or 15 ml tube.
9. Use an antibody that has been successful for immunopre-
cipitation experiments. We typically use Life Technologies
anti-GFP (#G10362) or anti-FLAG (#F1804) as capture
reagents.
10. To wash magnetic beads, use a magnet bar to collect beads to
the tube wall for 1 min; while the beads are magnetized remove
the supernatant. The tube may be briefly centrifuged to collect
supernatant at the bottom of the tube, re-magnetize, and
remove the remaining supernatant. Next, remove the tube
from the magnet and resuspend the beads in fresh buffer by
inversion or agitation.
11. End-over-end rotation is important to keep the beads in solution
to maximize target recovery.
186 Julian H.M. Kwan and Andrew Emili

12. Use a magnetic bar designed for magnetic bead collection in

1.5 ml tubes. If the initial sample volume is greater than 1.5 ml,
magnetize the first 1.5 ml to collect the beads and discard the
supernatant. Repeat until all beads have been collected, and
then proceed with washes.
13. Changing tubes is an important step to reduce contamination
from nonspecifically adsorbed proteins and detergents that
adhere to the tube wall.
14. Using trypsin digestion buffer for the final wash reduces the
carryover of salt present in the lysis and wash buffers.
15. Ensure that the lyophilized trypsin was resuspended in trypsin
digestion buffer and not acetic acid; since the digestion volume
is small the addition of acid would inhibit digestion.
16. Briefly centrifuge tube to collect digestion buffer and beads to
the bottom of the tube prior to incubation with rotation;
surface tension will keep the sample at the bottom of the tube.
End-over-end rotation is required to keep beads in solution for
efficient digestion.
17. The supernatant now contains the digested sample. This addi-
tional incubation step allows more time for digestion to produce
peptides for mass spectrometry identification.
18. Desalting media attach to micropipette, aspirate, and discharge
a single aliquot of 200 μl solution B in a microfuge tube
containing excess solution B to avoid forming bubbles. Repeat
with a new aliquot of solution B.
19. Be careful not to contaminate solution A with carryover of
solution B that may adhere to outside of tip (use separate
aliquots as needed).
20. The digested peptides will bind to the C-18 tip, allowing salt
and other contaminants to be washed away. Try to avoid aspi-
rating air and generating bubbles. Longer/more aspiration
and discharge will allow more peptides to bind the C-18 media.
21. Do not confuse or contaminate solution A with solution
B. Solution B will elute the peptides.
22. 8 μl of resuspended sample is generally sufficient for mass
spectrometry identification.

References
1. Butland G, Peregrín-Alvarez JM, Li J et al tive map of physical and genetic interactions
(2005) Interaction network containing con- mediated by the hsp90 chaperone. Cell
served and essential protein complexes in 120:715–727. doi:10.1016/j.cell.2004.12.024
Escherichia coli. Nature 433:531–537. 3. Gavin A-C, Aloy P, Grandi P et al (2006)
doi:10.1038/nature03239 Proteome survey reveals modularity of the
2. Zhao R, Davey M, Hsu Y-C et al (2005) yeast cell machinery. Nature 440:631–636.
Navigating the chaperone network: an integra- doi:10.1038/nature04532
 Simple and Effective Affinity Purification Procedures for Mass Spectrometry-Based… 187

4. Krogan NJ, Cagney G, Yu H et al (2006) ized AP-MS. Nat Methods 10:307–314.

Global landscape of protein complexes in the doi:10.1038/nmeth.2400
yeast Saccharomyces cerevisiae. Nature 440: 9. Roncagalli R, Hauri S, Fiore F et al (2014)
637–643. doi:10.1038/nature04670 Quantitative proteomics analysis of signalo-
5. Sardiu ME, Cai Y, Jin J et al (2008) Probabilistic some dynamics in primary T cells identifies the
assembly of human protein interaction net- surface receptor CD6 as a Lat adaptor-
works from label-free quantitative proteomics. independent TCR signaling hub. Nat Immunol
Proc Natl Acad Sci U S A 105:1454–1459. 15:384–392. doi:10.1038/ni.2843
doi:10.1073/pnas.0706983105 10. Guruharsha KG, Rual J-F, Zhai B et al (2011)
6. Sowa ME, Bennett EJ, Gygi SP, Harper JW A protein complex network of Drosophila
(2009) Defining the human deubiquitinating melanogaster. Cell 147:690–703. doi:10.1016/
enzyme interaction landscape. Cell 138:389– j.cell.2011.08.047
403. doi:10.1016/j.cell.2009.04.042 11. Choi H, Larsen B, Lin Z-Y et al (2011) SAINT:
7. Behrends C, Sowa ME, Gygi SP, Harper JW probabilistic scoring of affinity purification-mass
(2010) Network organization of the human spectrometry data. Nat Methods 8:70–73.
autophagy system. Nature 466:68–76. doi:10.1038/nmeth.1541
doi:10.1038/nature09204 12. Teo G, Liu G, Zhang J et al (2014) SAINTexpress:
8. Varjosalo M, Sacco R, Stukalov A et al (2013) improvements and additional features in Significance
Interlaboratory reproducibility of large-scale Analysis of INTeractome software. J Proteomics
human protein-complex analysis by standard- 100:37–43. doi:10.1016/j.jprot.2013.10.023
 Chapter 14

A Systems Approach to Understand Antigen Presentation

and the Immune Response
Nadine L. Dudek, Nathan P. Croft, Ralf B. Schittenhelm,
Sri H. Ramarathinam, and Anthony W. Purcell

Abstract
The mammalian immune system has evolved to respond to pathogenic, environmental, and cellular
changes in order to maintain the health of the host. These responses include the comparatively primitive
innate immune response, which represents a rapid and relatively nonspecific reaction to challenge by
pathogens and the more complex cellular adaptive immune response. This adaptive response evolves with
the pathogenic challenge, involves the cross talk of several cell types, and is highly specific to the pathogen
due to the liberation of peptide antigens and their presentation on the surface of affected cells. Together
these two forms of immunity provide a surveillance mechanism for the system-wide scrutiny of cellular
function, environment, and health. As such the immune system is best understood at a systems biology
level, and studies that combine gene expression, protein expression, and liberation of peptides for antigen
presentation can be combined to provide a detailed understanding of immunity. This chapter details our
experience in identifying peptide antigens and combining this information with more traditional pro-
teomics approaches to understand the generation of immune responses on a holistic level.

Key words Major histocompatibility complex, Human leukocyte antigens, Peptide ligands, Mass
spectrometry, Antigen presentation

1 Introduction

The human major histocompatibility complex (MHC) is located

on the short arm of chromosome 6 and encompasses around
4 Mbp or 0.1 % of the genome. Around 220 genes have been iden-
tified in this region and at least 10 % of these genes have a direct
function in the immune response to pathogens or the regulation of
immunity. The human MHC can be divided into three regions
which encode the class I, class II, and class III human leukocyte
antigen (HLA) gene products. HLA molecules demonstrate tre-
mendous polymorphism, which reflects the natural evolution of
these genes in response to various microbial pathogens in different
populations. Moreover, HLA genes exhibit linkage disequilibrium,

Jörg Reinders (ed.), Proteomics in Systems Biology: Methods and Protocols, Methods in Molecular Biology, vol. 1394,
DOI 10.1007/978-1-4939-3341-9_14, © Springer Science+Business Media New York 2016

189
190 Nadine L. Dudek et al.

meaning that they are often inherited together in blocks of genes.

This dictates that studies take into account the role of HLA mole-
cules both in isolation and in the context of their naturally occur-
ring combinations or haplotypes.
HLA class I molecules, and the murine H-2 equivalent, are
expressed on all nucleated cells and bind short peptides (typically
8–11 amino acids in length) derived from both self and foreign
antigens. These peptide ligands are primarily generated in or
transported into the cytoplasm and subsequently translocated
into the endoplasmic reticulum (ER) where they assemble with
nascent MHC class I molecules. These mature, peptide-loaded
complexes are transported to the cell surface where they are scru-
tinized by CD8+ cytotoxic T lymphocytes (CTL). Should the
peptide ligand be derived from a pathogen and be recognized as
foreign in an immunocompetent host, the cell is killed via the
cytotoxic armoury of the CTL.
The expression of MHC class II molecules is confined to a
small subset of highly specialized cells called professional antigen-
presenting cells (APCs). MHC class II molecules associate with
longer peptides (9–25 amino acids in length) than class I molecules
and this association occurs in late endosomal compartments, a dis-
tinct and separate cellular compartment to the ER-Golgi route
inhabited by assembling MHC class I molecules. MHC class II
molecules are recognized by CD4+ T helper cells and functional
recognition of these complexes is intimately involved in both the
humoral and cellular immune response. MHC class I and class II
molecules form membrane-distal structures that comprise a cleft in
which the antigenic peptide ligands reside [1–3]. The T cell recep-
tor (TCR) on CD8+ or CD4+ T cells recognizes MHC molecules
in the context of both the polymorphic class I or class II molecules,
respectively, and the peptide antigen presented in the antigen-
binding groove of these cell surface molecules [4]. The class III
region-encoded molecules have quite different and diverse bio-
chemical properties and are involved in inflammation and other
immune activities. As such they include components of the com-
plement system, cytokines (such as tumor necrosis factor and lym-
photoxin), and heat-shock proteins.
Approaches that facilitate the direct isolation and identifica-
tion of peptide antigens associated with class I or class II mole-
cules have defined the ligand specificity of different MHC
molecules. Moreover they have allowed direct identification of
naturally processed and presented antigens derived from infec-
tious microorganisms as well as self-peptides associated with auto-
immune disorders and cancers. Several different approaches have
been used to isolate MHC-bound peptides from cells, including
analysis of acidified cell lysates [5–7], elution of peptides from the
cell surface [8, 9], and immunoaffinity purification of the MHC-
peptide complexes from detergent-solubilized cell lysates [10–14].
 A Systems Approach to Understand Antigen Presentation and the Immune Response 191

Table 1
Commonly used monoclonal antibodies for MHC-peptide immunoaffinity chromatography

Hybrid Specificity Isotype Ref ATCC Comments

Anti-human
L243 DRα IgG2a [23, 24] Y Lower peptide yield than LB3.1
LB3.1 DRα IgG2a [25] Y Use in preference to L243
SPV-L3 DQ IgG2a [23]
B721 DP (DP1-5) IgG3 [26]
BB7.2 A2, A69 IgG2b [27]
ME1 B7, Bw22, B27 IgG1 [28] Y Cross-reacts with HLA B14 and Bw46
W632 A, B, C IgG2a [29]
DT9 C, E IgG2b [30, 31] Marginally lower yield of peptides in
comparison to W6/32
Anti-mouse
Y-3 Kb IgG2b [32] Y Cross-reacts with H-2k, p, q, r and s
28-14-8 Db Lb [33] Higher yield of peptides for Db than
28.8.6s
SF1.1.1.10 Kd IgG2a [14] Y
28.8.6 s Kb, Db IgG2a [34] Y Lower peptide numbers for Db than
28-14-8s
MKD6 IAd IgG2a [35] Y
Y-3P IA IgG2a [36] Y Weak reactivity with I-Ak
10.2.16 IA g7,k,r,f,s IgG2b [37] Y

The use of immunoaffinity chromatography dramatically improves

the specificity of the peptide extraction process. A single MHC
allele can be isolated by the use of an appropriate monoclonal
antibody (see Table 1) and some antibodies can even select a sub-
population of MHC molecules with defined molecular or func-
tional properties [15, 16]. The use of immunoaffinity
chromatography to isolate specific MHC molecules provides the
most appropriate material for identifying individual peptide
ligands restricted by a defined MHC allele. Furthermore, the
complexity of the eluates/lysates can be decreased by using cell
lines that express limited numbers of HLA alleles, for example
homozygous lymphoblastoid or mutant cell lines such as C1R
which express very low levels of endogenous class I molecules but
support high-level expression of transfected class I molecules
[17]. This property makes these cells very attractive for examin-
ing endogenous peptides presented by individual class I alleles
under normal physiological conditions [18–20] or during infec-
tion [21, 22].
In order to understand immunity at a systems level several ele-
ments must be examined in parallel. One crucial step is defining
which peptides are selected by particular MHC molecules or com-
binations of MHC molecules for presentation on the surface of
cells; how this changes both qualitatively and quantitatively at
192 Nadine L. Dudek et al.

Fig. 1 A systems approach to antigen presentation using mass spectrometry. In this workflow, cells are infected
with virus in vitro. At various time points, cells are harvested and lysed. A sample of lysate is taken and subjected
to tryptic digestion. The remaining lysate is used to affinity purify MHC peptide complexes. Both tryptic peptides
and non-tryptic MHC peptides are subjected to RP-HPLC before mass spectrometric interrogation. Mass spectro-
metric analysis involves both a global discovery approach of LC-MS/MS and the targeted method of LC-MRM,
where a set of known peptides are detected and quantified. In this workflow it is possible to simultaneously
quantify the presentation of virus or host peptide-MHC complexes, and the levels of their source antigens at
multiple times during infection to develop a comprehensive picture of antigen presentation

different stages of development, during inflammation or infection,

will critically inform systems immunology studies. These parame-
ters may also be correlated with the host and pathogen proteome,
which can be analyzed and quantitated from the same sample. The
combination of peptidome and proteome data allows the relation-
ship between antigen expression kinetics, abundance, and epitope
liberation to be correlated with immune outcomes in animal mod-
els or humans [38]. This chapter explores methods for studying
antigen presentation at a systems level by identifying peptides iso-
lated from specific MHC class I or class II molecules from various
types of APCs. It focuses on the use of serial immunoaffinity chro-
matography to study peptide determinant selection by different
HLA haplotypes, the parallel determination of the proteome of the
APC, and the quantitation of specific MHC-peptide complexes on
the cell surface using targeted proteomics approaches (Fig. 1).

2 Materials

Prepare all solutions using ultrapure H2O (18 mΩ cm at 25 °C,

freshly drawn) and use MS-grade solvents, reagents, glassware, and
plasticware.
 A Systems Approach to Understand Antigen Presentation and the Immune Response 193

2.1 Generation All solutions for cross-linking with the exception of triethanolamine
of MHC Eluate and DMP should be filtered through a 0.2 μM filter.
2.1.1 Preparation 1. Phosphate-buffered saline (PBS): 137 mM NaCl, 2.7 mM
of Cross-Linked KCl, 10 mM Na2HPO4, and 2 mM KH2PO4, pH 7.4.
Immunoaffinity Column 2. Purified monoclonal antibody (mAb) at 1–10 mg/ml in PBS:
Ideally the mAb should only recognize the class I or class II
allele of interest, although affinity and specificity issues fre-
quently require a compromise (Table 1, see Note 1).
3. Suitable column (e.g., disposable plastic Econo-Column from
BioRad).
4. Protein A resin (e.g., Repligen CaptivA™ PriMAB).
5. Borate wash buffer: 0.05 M Borate buffer pH 8.0. For 100 ml
of buffer, add 3.97 ml of 0.1 N NaOH to 50 ml of 0.1 M boric
acid/0.1 M KCl stock solution and make up to 100 ml with
MS-grade H2O.
6. Protein A wash buffer: 0.2 M Triethanolamine, pH 8.2 at
RT. Prepare this solution fresh and pH just prior to use.
7. Dimethyl pimelimidate (DMP) cross-linker: 40 mM DMP-
2HCl in 0.2 M triethanolamine pH 8.3. Prepare DMP by dis-
solving 250 mg DMP-2HCl (Sigma) in 22 ml 0.2 M
triethanolamine pH 8.2. Adjust pH to 8.3 with NaOH, and
bring to 24.1 ml. Prepare this solution fresh and pH just prior
to use (see Note 2).
8. Termination buffer: Ice-cold 0.2 M Tris, pH 8.0.
9. Stripping buffer: 0.1 M Citrate, pH 3.0.

2.1.2 Generation MHC-bound peptides may be affinity purified from whole tissue,
of Cell Lysate isolated primary cells, or transformed cells grown in culture (see
Note 3). The amount of material required is dependent on the
downstream application (LC-MS/MS vs. LC-MRM) and will be
highly dependent on the expression levels of MHC on the cells
contained within the sample (see Note 4). Transformed cells grown
in culture are the simplest sample type for MHC/peptide isolation
with cell numbers ranging between 5 × 107 and 1 × 109 per isola-
tion. Cells may be expanded, washed in PBS, and the pellets snap
frozen in liquid nitrogen for storage at −80 °C for up to 6 months.
1. 2× Lysis buffer: 0.5 % NP-40 (IGEPAL 630 from Sigma is the
equivalent), 50 mM Tris, pH 8.0 (from 1 M stock solution),
150 mM NaCl, protease inhibitor cocktail (cOmplete protease
inhibitor from Roche or equivalent, should be made up fresh
each time) in MS-grade H2O. Prepare lysis buffer just prior to
use and keep on ice.
194 Nadine L. Dudek et al.

2.1.3 Immunoaffinity MHC class I and II molecules can be eluted from the same sample
Purification of MHC Class I/ by using tandem columns. For human cell lines, we routinely pass
Class II Molecules the cell lysate through a class I column, followed sequentially by a
column specific for HLA DR, then HLA DQ, and finally HLA DP.
1. mAb cross-linked protein A resin from Subheading 2.1.1.
2. Pre-column (non-cross-linked protein A sepharose in suitable
column): The bed volume should be half that of the cross-
linked column; that is, for a 1 ml protein A-mAb column, use
a 0.5 ml pre-column.
3. Pepstatin A: 1 mg/ml stock in isopropanol, aliquot, and store
at −20 °C.
4. PMSF: 0.1 M stock in absolute ethanol, aliquot, and store at
−20 °C.
5. Wash buffer 1: 0.005 % IGEPAL 630, 50 mM Tris, pH 8.0,
150 mM NaCl, 5 mM EDTA, 100 μM PMSF, 1 μg/ml pep-
statin A in MS-grade H2O.
6. Wash buffer 2: 50 mM Tris, pH 8.0, 150 mM NaCl, in MS-
grade H2O.
7. Wash buffer 3: 50 mM Tris, pH 8.0, 450 mM NaCl, in MS-
grade H2O.
8. Wash buffer 4: 50 mM Tris, pH 8.0, in MS-grade H2O.
9. Elution buffer: 10 % acetic acid in MS-grade water (use best
grade glacial acetic acid, e.g., Sigma ACS grade).

2.1.4 Separation of MHC 1. Peptide separation prior to loading on mass spectrometer is

Eluate by RP-HPLC performed by reversed-phase (RP) chromatography using a
C18 column in a high-pressure liquid chromatography
(HPLC) system (e.g., 4.6 internal diameter × 50 mm long
reversed-phase C18 endcapped HPLC column, Chromolith
Speed Rod, Merck). LC mobile phases: Buffer A is 0.1 % tri-
fluoroacetic acid (TFA) in MS-grade water and buffer B is
0.1 % TFA in MS-grade acetonitrile.
2. Low-protein-binding 1.5 ml tubes for fraction collection, e.g.,
Eppendorf LoBind tubes.

2.2 LC-MS/MS 1. Buffer A: 0.1 % formic acid (FA) in MS-grade H2O.

Analysis 2. Buffer B: 0.1 % FA in 80 % MS-grade acetonitrile.
of MHC Eluate
3. Autosampler vials for mass spectrometry.

2.3 Targeted Mass In cases where peptide epitopes are known, isotopically labeled
Spectrometric (AQUA) peptides can be used for absolute quantitation by tar-
Analysis geted LC-MRM (multiple reaction monitoring) [38, 39]. AQUA
of MHC Eluate peptides are composed of the same amino acid sequence as the
natural equivalent but bearing one or more heavy amino acids.
 A Systems Approach to Understand Antigen Presentation and the Immune Response 195

1. Buffer A: 0.1 % FA in MS-grade H2O.

2. Buffer B: 0.1 % FA in 98 % MS-grade acetonitrile.
3. Synthetic peptides at >98 % purity: Native peptide and isotopi-
cally labeled AQUA peptide (optimal monoisotopic shift
between the two species of 6–8 Da) solubilized in appropriate
buffer, e.g., DMSO (see Note 5).
4. Autosampler vials for mass spectrometry.

2.4 Analysis In order to correlate the cellular proteome with the immun-
of Cellular Proteome opeptidome, a small sample of the MHC lysate is subjected to
Using FASP Digestion tryptic digestion and mass spectrometric analysis [40]. Samples
of MHC Lysate for tryptic digestion may be taken immediately after lysis or from
the flow through of the MHC-affinity column. Retaining the
flow through from the affinity column and freezing at −80 °C is
advised so that if required, more tryptic digestions can be per-
formed. We utilize the Expedeon FASP Protein Digestion Kit as
a convenient way of generating tryptic peptides. This kit is com-
patible with a number of reducing agents; however we recom-
mend TCEP.
1. Expedeon FASP Protein Digestion Kit (reagents except tryp-
sin and reducing agent are provided in the FASP kit).
2. Ammonium bicarbonate (in kit): 50 mM.
3. NaCl (in kit): 0.5 M.
4. Urea sample solution (in kit): 1 ml of Tris solution to one tube
of urea, vortex until powder is dissolved. Prepare just before
use.
5. 10× Iodoacetamide solution (in kit): Add 100 μl of urea
sample solution to one tube of iodoacetamide. Mix and dis-
solve by pipetting up and down. Transfer into clean
microfuge tube, wrap in foil, and keep on ice. Prepare just
before use.
6. Digestion solution (in kit): Dissolve one 1 μg tube of trypsin
in 75 μl of 50 mM ammonium bicarbonate solution and
keep on ice. Aim for 1:100 trypsin-to-protein ratio. Scale
amount of trypsin according to how much lysate is used, i.e.,
if loading 400 μg of protein, use 4 μg trypsin. Prepare just
before use.
7. Bradford reagent.
8. Trypsin (single shot 1 μg, Sigma).
9. 0.1 % FA.
10. TCEP: 0.5 M in 50 mM ammonium bicarbonate (Sigma).
196 Nadine L. Dudek et al.

3 Methods

3.1 Generation 1. Remove required amount of protein A sepharose (supplied as

of MHC Eluate a 50 % slurry in 20 % ethanol) and add to column. Generally
cross-link 10 mg of antibody per 1 ml of resin. Allow to settle
3.1.1 Preparation
by gravity, check for air bubbles, and agitate the slurry if neces-
of Cross-Linked
sary to remove.
Immunoaffinity Column
2. Wash with 10 column volumes (c.v.) of MS-grade H 2O
followed by 10 c.v. of PBS.
3. Prepare mAb (ideally at 0.5–1 mg/ml in PBS) in 50 ml tube.
Remove washed resin from column and add to antibody in
tube. Rotate gently end-over-end at 4 °C for 1 h to allow bind-
ing (see Note 6).
4. Transfer resin and antibody back to column and allow anti-
body to flow through by gravity. Wash antibody-bound resin
with 20 c.v. of borate buffer followed by 15 c.v. of freshly pre-
pared 0.2 M triethanolamine, pH 8.2 at RT. Triethanolamine
is used to ensure that there are no residual primary amines
present that may interfere with the cross-linking reaction.
5. Flow 5 c.v. of freshly prepared DMP cross-linker through the
column at RT leaving a meniscus just over the protein A col-
umn bed. Seal the bottom of column and allow to sit at RT for
1 h.
6. Terminate the cross-linking reaction by adding 10 c.v. of ice-
cold termination buffer (0.2 M Tris, pH 8.0).
7. Remove unbound antibody by washing with 10 c.v of strip-
ping buffer (0.1 M citrate pH 3).
8. Flow 10 c.v. of PBS pH 7.4 until pH of flow through is >7 (it
may be convenient to stop here, wash, and store the column in
PBS pH 7.4 supplemented with 0.02 % NaN3). Generally col-
umns are best used within a month; however this will vary
depending on the specific antibody.

3.1.2 Generation 1. Cells (5 × 107 to 1 × 109) can be grown in spinner flasks, biore-
of Cell Lysate actors, or roller bottles (see Note 7) to appropriate numbers,
washed in PBS, and harvested by centrifugation (2000 × g,
10 min at 4 °C). Pellets should be snap frozen in liquid nitrogen
and may be harvested iteratively for storage at −80 °C for up to
6 months. If collecting tissues, they should be rinsed in cold
PBS containing protease inhibitors and immediately snap fro-
zen without liquid for later processing.
2. Prepare a 2× concentrated solution of lysis buffer. Lysis buffer
is added at 2× to allow for the volume of the cell pellet to be
taken into account prior to adjustment of the concentration of
the lysate to 1×. Cells are lysed at 5 × 107–1 × 108 cells per ml
 A Systems Approach to Understand Antigen Presentation and the Immune Response 197

of 1× lysis buffer. If the volume of the cell pellet is close or over

50 % of the final volume required you may need to lyse at a
lower cell density.
3. Add correct volume of 2× lysis buffer to the frozen cell pellets
and thaw the pellets quickly in a bath of tepid (i.e., RT) water.
The temperature of the material should remain cold to touch
so do not let the material equilibrate, thaw until small ice
clumps are left, and add ice-cold MS-grade water to a final
volume so as the lysis buffer is now at 1× strength.
4. Briefly homogenize the lysate (e.g., using a Polytron Disperser)
to disperse any left over ice pellets. For large cell pellets or tis-
sue samples we recommend grinding cells under liquid nitro-
gen (rather than homogenizing) using a cryogenic mill (e.g.,
Mixer Mill MM 400, Retsch). If the mill is used, pellets are
placed straight into the precooled mill pot and after grinding
the powder is scraped into lysis buffer.
5. Rotate lysate end-over-end at 4 °C for 1 h. Retain a sample
(100 μl to 1 ml depending on cell number and lysis volume)
for FASP digestion; retained lysate may be stored at −80 °C.
6. Centrifuge lysate for 10 min at 2000 × g (4 °C). This step
removes the nuclei.
7. Take supernatant from previous step and spin for 75 min in an
ultracentrifuge (100,000 × g) at 4 °C. Multiple spins may be
necessary to fully clarify the lysate.
8. Collect the supernatant. It should be clear. If there is an unclear
layer at top of the tubes carefully remove this lipid-containing
layer and filter through a 0.8 μm and a 0.45 μm filter.

3.1.3 Immunoaffinity 1. Using gravity flow or a peristaltic pump in a cold room, load
Purification of MHC Class I/ cell lysate onto a protein A sepharose pre-column that has been
Class II pre-equilibrated in 10 c.v. wash buffer 1. Multiple pre-columns
may be required depending on the size and type of sample and
should be replaced upon clogging.
2. Collect pre-cleared lysate and slowly load onto the cross-linked
mAb column(s). If gravity flow is too quick (<1 h for lysate to
pass), use a peristaltic pump. For small samples (1–4 ml lysate),
it is generally better to add lysate to several of 2 ml LoBind
Eppendorf tubes containing resin and rotate slowly end over
end at 4 °C for 1 h.
3. For maximal yield the lysate should be run through the col-
umn twice. Retain flow through and freeze at −80 °C for sub-
sequent FASP analysis. If multiple columns are being run (i.e.,
class I and II elutions from the same lysate), we routinely run
the lysate once only over each column. The columns can be set
up on a retort stand in tandem so that the flow through from
198 Nadine L. Dudek et al.

one column runs directly to the next. It may be necessary to

temporarily stop the flow of the different columns throughout
the process, to ensure that they do not dry out (if the lysate is
passing at slightly different flow rates through each). If binding
has been performed in 2 ml LoBind Eppendorf tubes, after the
1-h incubation, spin resin gently and transfer supernatant to
Eppendorf containing next mAb resin. Repeat incubation
while washing and eluting from the first batch of mAb resin.
4. Wash the column(s) with 20 c.v. of each wash buffer in the fol-
lowing order: Wash buffer 1, wash buffer 2 (to remove deter-
gent), wash buffer 3 (to remove nonspecifically bound material),
wash buffer 4 (removes salt to prevent crystal formation).
5. Elute MHC molecules in 5 c.v. of elution buffer. Add AQUA
peptides here if used, i.e., post-elution from column and pre-
separation by RP-HPLC.
6. Empty column(s) can be discarded or soaked overnight in ace-
tic acid, washed in MS-grade H2O, and reused.
7. Progress to RP-HPLC fractionation of flow through.
Alternatively the eluate can be frozen at −80 °C; however this
will result in some sample loss.

3.1.4 Separation of MHC A single RP-HPLC step may be used to isolate peptides if an
Eluate by RP-HPLC immunological readout is used to assay peptide fractions.
However biochemical analysis by mass spectrometry requires a
minimum of two dimensions of RP-HPLC to achieve sufficient
separation (see Note 8).
1. Peptides are separated from MHC heavy chain, β2m (for class
I molecules), leached antibody, and contaminating detergent
using a C18 reverse-phase column running on a mobile-phase
buffer A of 0.1 % TFA and buffer B of 80 % acetonitrile/0.1 %
TFA (see Note 9). We routinely use a 4.6 mm internal diame-
ter × 50 mm (or 100 mm for greater separation see Note 10)
long reversed-phase C18 endcapped HPLC column
(Chromolith Speed Rod, Merck) on an ÄKTAmicro™ HPLC
system (GE Healthcare).
2. Separate based on a rapid gradient of buffer A to B, which
results in 10–30 peptide-containing fractions (e.g., 2–40 % B
for 4 min, 40–45 % for 4 min, and a rapid 2 min increase to
100 % B; Fig. 2). Using this approach a small number of early
fractions contain greater than 95 % of the peptides, whilst the
later fractions contain IGEPAL 630 polymers which hamper
MS analysis severely and β2-microglobulin (see Note 11).
3. Collect fractions (500 μl) into LoBind Eppendorf tubes.
At this point fractions may be frozen at −80 °C.
 A Systems Approach to Understand Antigen Presentation and the Immune Response 199

Fig. 2 Monolithic separation of affinity-purified MHC-peptide complexes and fractionation of bound peptides.
Representative UV trace showing fractionation of the eluted mixture of peptides and heavy chains from both
MHC class I and class II eluates. The early fractions contain MHC peptides and the later fractions contain heavy
chains, detergent, and leached antibody. The β2-microglobulin peak is highlighted for the MHC class I eluate

4. Vacuum concentrate peptide-containing fractions (generally

peptide-containing fractions contain up to 45 % buffer B) to
reduce the concentration of acetonitrile. Typically dry to 10 μl
and dilute to 15–25 μl in 0.1 % formic acid. Do not dry to
completeness as this may result in sample loss due to adsorp-
tion to the plasticware.

3.2 LC-MS/MS Two forms of mass spectrometry can be used to analyze fractions
Analysis containing HLA-bound peptides (Fig. 3). This may consist of the
of MHC Eluate traditional global LC-MS/MS analysis or more targeted method-
ologies such as multiple reaction monitoring. Global LC-MS/MS
is recommended for samples where the MHC peptide repertoire
composition is unknown. Peptide species are separated by an LC
gradient and paired MS and MS/MS spectra are acquired by the
mass spectrometer. Downstream analysis and identification of
acquired spectra are facilitated by either manual sequencing or,
preferably, the use of protein identification software algorithms
(e.g., Mascot (MatrixScience), ProteinPilot™ (SCIEX)).
Numerous factors can affect the resulting number of peptide
identifications from a global LC-MS/MS analysis, including initial
starting cell number, expression level of the MHC molecules at the
cell surface, efficiency of cell lysis, quantity of antibody used for
immunoaffinity capture, appropriate and sufficient upstream
HPLC fractionation, online LC gradient, and specific MS param-
eter settings. It is recommended to optimize these variables.
1. Place samples in a sonicating water bath for 5 min to detach
peptides bound to plastic.
200 Nadine L. Dudek et al.

Fig. 3 A comparison of information-dependent acquisition (IDA) or LC-MS/MS and LC-MRM analysis. In global
LC-MS/MS analysis a defined number of precursor ions are selected for fragmentation during each duty cycle
of the mass spectrometer. The selection of precursors is a stochastic process typically based on abundance or
ion intensity and limited to the top 10–50 most intense ions entering the instrument in that particular cycle. In
contrast, LC-MRM involves the selection of specific, predefined, precursors that are targeted for analysis and
detected based on a defined set of fragment ions. This allows relatively low-intensity ions to be selected in
preference to more abundant co-eluting species. Area under the curve quantitation of the MRM transitions can
then afford very accurate and specific quantitation

2. Centrifuge at 13,000 × g for 10 min to pellet any particulates

and transfer supernatant into an autosampler vial.
3. Add retention time standard peptides at appropriate concen-
trations, e.g., iRT peptide mix. This step is optional but highly
recommended (see Note 12).
4. Load sample onto mass spectrometer and run optimized gradi-
ent. We routinely run MHC peptides on a 5600+ TripleTOF
(SCIEX) equipped with a Nanospray III ion source, where
samples are loaded onto a microfluidic trap column packed
with ChromXP C18-CL 3 μm particles (300 Å nominal pore
size; equilibrated in 0.1 % FA/2 % acetonitrile) at 5 μl/min
using a NanoUltra cHiPLC system. An analytical
(75 μm × 15 cm ChromXP C18-CL 3 μm, 120 Å, Eksigent)
microfluidic column is switched in line and peptides separated
using linear gradient elution of 0–80 % acetonitrile over 90 min
(300 nl/min). MS/MS switch criteria includes ions of
m/z > 200 amu, charge state +2 to +5, intensity >40 cps, and
the top 20 ions meeting this criteria are selected for MS/MS
per cycle (see Note 13). Some examples of MS/MS spectra of
class I and II peptides are shown in Fig. 4.
 A Systems Approach to Understand Antigen Presentation and the Immune Response 201

Fig. 4 Examples of biochemical analysis of peptides eluted from MHC class I and class II molecules. (a) A typical
total ion chromatogram (TIC) of LC-separated MHC-bound peptides analyzed on a 5600+ TripleTOF mass
spectrometer (SCIEX). (b–d) Annotated MS/MS spectra of various MHC-bound peptides. ARFDSDVEVY (panel b)
and phosphorylated RSLSPMS*GLFGSIW (panel d) were eluted from the MHC class I molecules HLA-B*27:05
and HLA-B*57:01, respectively. AGQLVFLATEGDHL (panel c) was eluted from human MHC class II molecules.
The insets of each panel show the corresponding precursor MS1 regions

3.3 Targeted Mass Targeted methodologies, e.g., LC-MRM, are of use when specific
Spectrometric peptide epitopes are known and a qualitative and/or quantitative
Analysis readout is desired. Here, MS instrument parameters are set to tar-
of MHC Eluate get only the peptides of interest, ignoring the rest of the sample
allowing low-abundance peptides to be detected in complex sam-
ples (see Note 14). LC-MRM also negates the need for high cell
numbers as such targeted methods are, by their nature, highly spe-
cific and more sensitive than global LC-MS/MS approaches. For
LC-MRM, synthetic peptides corresponding to those of interest
can be used to design and optimize MRM parameters [41] or
alternatively MS/MS data from discovery-based experiments can
be used to identify optimal transition parameters. The detection of
spiked AQUA peptides by LC-MRM allows integration of the area
202 Nadine L. Dudek et al.

under the curve of the light and heavy peptides, providing quanti-
tation of the light peptides. These values can be related back to the
starting number of cells, to give the most accurate assessment pos-
sible of epitope copy number per cell [38, 39].
1. Prepare AQUA peptide stock, we generally reconstitute pep-
tides at 5 mM in 100 % DMSO (choice of buffer will be user
dependent, see Note 5). Run approximately 50 fmol of peptide
diluted in 0.1 % FA on mass spectrometer to determine the
dominant precursor ion. Peptides are then run targeting the
dominant precursor ion and fragmented across a range of col-
lision energies (CE) in order to determine the optimal energy
to generate the most intense fragment ions [41]. Note that
each MRM transition can use its own CE value, so different
fragment ions will sometimes require different CE values.
Typically the top four most intense fragment ions are used to
build the MRM for a given peptide.
2. For each AQUA peptide perform a standard curve and choose
a concentration in the linear part of the curve that will be used
to spike MHC samples. Typically this will be between 1 and
100 fmol. Determine which fraction AQUA peptides elute
during RP-HPLC for initial separation on the C18 column
(prior to loading onto mass spec). By doing this it will be pos-
sible to run only fractions that contain the peptide of interest
which will substantially reduce instrument time depending on
the number of AQUA peptides being analyzed. Moreover, co-
elution of the AQUA peptide provides further confidence in
the detection of the target peptide.
3. Add isotopically labeled (AQUA) peptides of known concen-
tration into the MHC eluate immediately following elution
from the antibody affinity column. This will allow for the co-
elution of the light (i.e., endogenous) and heavy (i.e., AQUA)
peptide during RP-HPLC.
4. We routinely run LC-MRM experiments on a 5500 QTRAP
(SCIEX) with similar LC conditions as above. MRM transi-
tions are used with a dwell time between 5 and 40 ms, opti-
mized to result in a cycle time that will lead to at least eight
data points across a detected peptide. MS parameters are unit
resolution for Q1 and Q3, with the MRM experiment coupled
to an information-dependent acquisition (IDA) criterion set to
trigger an EPI scan (10,000 Da/s; rolling CE; unit resolution)
following any MRM transition exceeding 500 counts.
5. To quantitate the amount of the peptide of interest present in
each HPLC fraction, the area under each MRM transition
peak is calculated (using for example MultiQuant 2.0, SCIEX).
After peak integration, the area value for all MRM transitions
for the peptide is combined and compared against the area
 A Systems Approach to Understand Antigen Presentation and the Immune Response 203

value for the combined transitions of the AQUA peptide. The

area ratio between the light and heavy peptides is used to
determine the molar amount of the peptide of interest.
Multiplying the molar amount of peptide by Avogadro’s num-
ber and dividing by the cell number will give the number of
peptide copies per cell.

3.4 Analysis Tryptic digestion can be performed on samples taken before or

of Cellular Proteome after loading onto affinity column. If quantitation of the lysate can-
Using FASP Digestion not be performed, we would recommend digesting 100 μl (from a
of MHC Lysate lysate made of 1 × 108 cells) with 1 μg of trypsin as a starting point.
1. If lysate has been frozen, thaw sample on ice.
2. Centrifuge at 16,000 × g for 5 min, and collect supernatant.
3. Quantitate lysate by using Bradford reagent.
4. Treat up to 400 μg protein lysate with TCEP at 5 mM final
concentration for 30 min at 60 °C to reduce sample.
5. Add 200 μl of urea sample solution, transfer to spin filter, and
centrifuge 14,000 × g for 15 min (maximum volume of the
spin filter is 300 μl; if more volume of lysate is required to
reach capacity, multiple loads through the spin filter can be
performed).
6. Pass flow-through through the column again (spin at 14,000 × g
for 15 min).
7. Add 200 μl of urea sample solution to spin filter (see Note 15),
centrifuge at 14,000 × g for 15 min, and discard flow-through
from collection tube.
8. Add 10 μl of 10× iodoacetamide solution and 90 μl of urea
sample solution to spin filter and vortex for 1 min.
9. Incubate without mixing for 20 min in the dark.
10. Centrifuge for 10 min 14,000 × g to remove iodoacetamide.
11. Add 100 μl of urea sample solution to the spin filter and cen-
trifuge at 14,000 × g for 10 min; repeat this step twice and
discard flow-through.
12. Add 100 μl of 50 mM ammonium bicarbonate solution to
spin filter and centrifuge at 14,000 × g for 15 min; repeat this
step twice.
13. Change collection tube and add 75 μl digestion solution
(1:100 enzyme:protein) and vortex for 1 min.
14. Wrap tubes with parafilm to minimize evaporation, and incu-
bate spin filters at 37 °C overnight.
15. Add 40 μl of 50 mM ammonium bicarbonate solution.
16. Centrifuge spin filter at 14,000 × g for 10 min; repeat this step
once.
204 Nadine L. Dudek et al.

17. Add 50 μl 0.5 M NaCl and centrifuge spin filter at 14,000 × g

for 10 min.
18. Transfer to LoBind tube, and spin down at 16,000 × g for
10 min to remove debris.
19. Transfer to new LoBind tube and store peptides at −20 °C or
−80 °C until ready for reversed-phase fractionation as per
MHC eluate (Subheading 3.1.4, see Note 16).

4 Notes

1. The choice of monoclonal antibody is closely allied to the

choice of cell line and the specificity and efficacy of the mono-
clonal antibody/antibodies used in the immunoaffinity isola-
tion of MHC molecules. Monoclonal antibodies with specificity
towards classes of MHC molecules, families of MHC mole-
cules, individual alleles of MHC molecules, and even subsets of
molecules of an individual allotype have been generated over
the years and many hybridomas are readily accessible commer-
cially through bodies such as the ATCC (www.atcc.org). Those
we have successfully used for MHC/peptide elution experi-
ments are shown in Table 1. This highlights that some anti-
bodies are better suited for MHC elution than others. For
example, although both L243 and LB3.1 antibodies affinity
purify HLA-DR, LB3.1 yields at least double the number of
MHC peptides. It is also important to determine whether the
DMP cross-linker interferes with the ability of the antibody to
immunoprecipitate (in which case other cross-linking methods
may be required).
2. DMP can be bought in larger amounts; however the 250 mg
vials are relatively cheap and work well for single use. This
eliminates the possibility of decreased efficiency of cross-linking
due to long-term storage of larger quantities of DMP at
−20 °C. Generally one 250 mg vial is used to cross-link up to
20 mg of antibody to 2 ml of resin.
3. Choice of cell type: In order to maximize the yield of MHC
class I or class II molecules the cell line used must be given
serious consideration. Epstein–Barr virus-transformed B cell
lines that express high levels of HLA A, B, or C class I mole-
cules or HLA DR, DQ, or DP molecules are easily sourced
from depositories such as ATCC. These cells can be grown to
high density in cell culture and used to great effect in bio-
chemical studies of bound ligands. Homozygous cell lines for
most common class I or class II alleles are well documented
and often express haplotypes of interest. The use of B-LCLs
dictates the use of a discriminating antibody should a single
A Systems Approach to Understand Antigen Presentation and the Immune Response 205

allele be required to be purified. In the absence of such an

antibody, the ideal cell type for these experiments would
express a single MHC molecule and have intact antigen pro-
cessing and presentation pathways. Several mutant cell lines
have been generated that approximate such a cell type. The
B-lymphoblastoid cell line Hmy2.C1R was generated by
gamma irradiation of LICR.LON.Hmy2 [42] and selected
with antibodies against HLA A and HLA B alleles and comple-
ment. This resulted in a cell line with no detectable HLA A or
B gene products, yet with intact antigen processing and pre-
sentation pathways [43]. Thus, these cells are able to support
high-level expression of individually transfected HLA A, B, or
C gene products [43]. Similar cell lines exist for class II elution
studies. For example, the murine cell line M12.C3 lacks endog-
enous Ia and functional I-Ak expression can be restored by
introduction of I-AK α and β chains via transfection [44]. It
should be noted however that not all cells express class II mol-
ecules endogenously, thus restricting the array of APCs ame-
nable for the creation of appropriate cell lines. The same
considerations apply to tissue samples, where MHC levels may
vary and where there may be a mix of different cell types.
4. The number of cells required will vary depending on the appli-
cation and the MHC levels expressed by the cells. Global
LC-MS/MS analysis requires the largest number of cells. For a
basic repertoire analysis of a cell line expressing high levels of
MHC we would generally use 1 × 109 cells and expect to
sequence between 1000 and 3000 peptides depending on the
allele. During LC-MS/MS analysis the mass spectrometer is
limited to the number of ions per second that can be frag-
mented to obtain MS/MS data. This dictates that it is the most
abundant ions that are sequenced such that increasing the cell
number does not necessarily increase the number of peptide
identifications. Increasing the number of peptides requires a
combination of adequate cell number, and further fraction-
ation of the eluate prior to mass spectrometric analysis to
decrease the complexity of the peptide mix. LC-MRM analysis
requires fewer cells but can only be utilized when the sequence
of the peptides to be targeted is known. We generally start with
pellets of 1 × 108 for MRM analysis. Cell numbers can be lower
than 1 × 108; however detection and quantitation of peptides
from smaller cell numbers are dependent on how well the tar-
get peptide ionizes and how much of the given peptide/MHC
complex is expressed in the surface of the cell.
5. AQUA peptides should be synthesized to the highest purity
(>98 %) and solubilised in an appropriate buffer (e.g., buffer A,
DMSO). The amino acid composition of the peptide will infer
optimal buffer to use; a number of guides are available such as
206 Nadine L. Dudek et al.

http://www.genscript.com/peptide_solubility_and_stablity.
html. Peptides should be quantified to an exact concentration
(amino acid analysis; DirectDetect (Merck Millipore)). AQUA
peptides are spiked into the sample immediately post-acid elu-
tion from the affinity column. The amount of each AQUA
peptide to add requires optimization on a per-peptide basis,
but typically ranges from 1 to 100 fmol per sample.
6. For each new antibody it is advisable to check the efficiency of
the cross-linking reaction by SDS-PAGE. Take a sample of the
original antibody (1) and a sample of the flow through (2)
following incubation with the resin. Take aliquots of resin
(25 μl) before the addition of DMP (3) and after incubation
with DMP (4). It is also advisable to concentrate the flow
through from the citric acid wash (5) using a 15–30 kDa cut-
off concentrator to ~500 μl (this is done to see how much
antibody has been left unbound), and take 25 μl to add to 2×
SDS-PAGE loading dye. Samples should be run on a reducing
12 % SDS-PAGE gel and Commassie stained. Heavy and light
chains of the antibody should be seen in samples 1 and 3.
There should be little antibody in samples 2, 4, and 5. If the
antibody has not been used for DMP cross-linking before,
perform a small-scale purification to test that the antibody still
retains its binding affinity.
7. Cells can be expanded by standard tissue culture techniques;
however particular care should be taken with adherent cells.
Removing cells with trypsin can strip coatings from treated tis-
sue culture plastics, which may interfere with subsequent mass
spectral analysis. We routinely passage cells using trypsin; how-
ever when harvesting for snap freezing we remove cells using
5–10 mM EDTA in PBS.
8. Column and mobile phase choice for multidimensional RP-
HPLC are dictated by sample composition but consider-
ations should include altered ion pair agent, altered mobile
phase pH, altered stationary-phase ligand or mode, and col-
umn dimensions.
9. A common practice to separate peptides from heavy chain and
β2m in the case of class I molecules is to use a low-protein-
binding spin filter. In our experience, this results in significant
loss of peptides, irrespective of filter brand or pre-blocking
with BSA. We find that separation using RP-HPLC increases
the yield of peptides at least tenfold.
10. We routinely separate β2m from heavy chains using a 4.6 mm
internal diameter × 50 or 100 mm long reversed-phase C18
endcapped HPLC column (Chromolith Speed Rod, Merck).
The choice between 50 and 100 mm is dependent on the cell
number and whether the samples are to be used for LC-MS/MS
A Systems Approach to Understand Antigen Presentation and the Immune Response 207

or LC-MRM. For LC-MS/MS where larger cell numbers are

used and the goal is to sequence as many peptides as possible,
we find that the 100 mm column gives better separation of the
MHC peptides and greater peptide numbers (two- to three-
fold over the 50 mm column). Peptide numbers may also be
increased by using orthogonal separation methods such as
SCX or HILIC. For MRM analysis, which is normally per-
formed on smaller samples, the 50 mm rod is sufficient; how-
ever the impact of sample complexity on any given peptide of
interest should be determined to optimize the amount of frac-
tionation required.
11. The number of fractions collected needs to be a compromise
between reducing the complexity of the sample for mass spec-
trometric analysis and keeping the number of samples to a
practical number. We routinely collect 500 μl fractions, and
then pool these into a single concatenated sample consisting of
four to six fractions spread well across the gradient separation.
In this way, the number of samples for LC-MS/MS is reduced,
and the full gradient on the column attached to the mass spec-
trometer is utilized. Caution should be exercised with fractions
coming off in high acetonitrile; although these may contain
peptides of interest, the highly hydrophobic nature of the pep-
tides in these samples may interfere with detection of peptides
eluting earlier in the gradient and may be better run as indi-
vidual fractions.
12. In addition to AQUA peptides, it is often desired for both LC-
MS/MS and LC-MRM experiments to spike in a set of well-
characterized peptides, which can be used (1) to normalize
retention times across different experiments independent of
the chromatographic system and/or (2) to precisely predict
retention times of known target peptides. The indexed reten-
tion time (iRT) peptide mix, a set of 11 peptides derived from
Leptospira interrogans, was specifically designed for this pur-
pose [45].
13. Although a time-of-flight (TOF) mass spectrometer such as
the 5600+ TripleTOF (SCIEX) is calibrated every three to five
LC runs using a standard such as glu-fibrinopeptide, we also
recommend preparing a standard sample of MHC peptides
(i.e., non-tryptic) to run as a quality control for instrument
performance. We often see a drop in the number of MHC
peptides identified using this standard before a decrease in the
intensity of the glu-fibrinopeptide calibration peak is detected.
14. Although the sample complexity is not particularly a problem for
MRM analysis, it is our experience that some peptides are harder
to detect when present in an increasingly complex sample or a
sample containing highly hydrophobic sequences. For this rea-
son fractionation is still recommended for LC-MRM analysis.
208 Nadine L. Dudek et al.

15. The FASP kit comes with 30 kDa cutoff spin filters; we how-
ever prefer to use 10 kDa cutoff spin filters from Pall.
16. Mass spectrometric analysis can be performed on the FASP
digest without fractionation; however we find that this dra-
matically reduces the number of protein identifications. We
recommend an off-line separation by RP-HPLC; alternatively
other separation methods such as SCX may be used.

References
1. Madden DR (1995) The three-dimensional 12. Scull KE, Dudek NL, Corbett AJ et al (2012)
structure of peptide-MHC complexes. Annu Secreted HLA recapitulates the immunopepti-
Rev Immunol 15:587–622 dome and allows in-depth coverage of HLA
2. Stern LJ, Brown JH, Jardetzky TS et al (1994) A*02:01 ligands. Mol Immunol 51(2):136–142
Crystal structure of the human class II MHC 13. Illing PT, Vivian JP, Dudek NL et al (2012)
protein HLA-DR1 complexed with an influ- Immune self-reactivity triggered by drug-
enza virus peptide. Nature 368(6468): modified HLA-peptide repertoire. Nature
215–221 486(7404):554–558
3. Fremont DH, Hendrickson WA, Marrack P 14. Dudek NL, Tan CT, Gorasia DG et al (2012)
et al (1996) Structures of an MHC class II Constitutive and inflammatory immunopepti-
molecule with covalently bound single pep- dome of pancreatic beta-cells. Diabetes
tides. Science 272(5264):1001–1004 61(11):3018–3025
4. Garcia KC, Teyton L, Wilson IA (1999) 15. Urban RG, Chicz RM, Lane WS et al (1994)
Structural basis of T cell recognition. Annu A subset of HLA-B27 molecules contains pep-
Rev Immunol 17:369–397 tides much longer than nonamers. Proc Natl
5. Falk K, Rötzschke O, Deres K et al (1991) Acad Sci U S A 91(4):1534–1538
Identification of naturally processed viral non- 16. Purcell AW, Kelly AJ, Peh CA et al (2000)
apeptides allows their quantification in infected Endogenous and exogenous factors contribut-
cells and suggests an allele-specific T cell epit- ing to the surface expression of HLA B27 on
ope forecast. J Exp Med 174:425–434 mutant APC. Hum Immunol 61(2):120–130
6. Sijts AJ, Neisig A, Neefjes J et al (1996) Two 17. Storkus WJ, Howell DN, Salter RD et al
Listeria monocytogenes CTL epitopes are pro- (1987) NK susceptibility varies inversely with
cessed from the same antigen with different target cell class I HLA antigen expression.
efficiencies. J Immunol 156(2):683–692 J Immunol 138(6):1657–1659
7. Rotzschke O, Falk K, Deres K et al (1990) 18. Macdonald WA, Purcell AW, Mifsud NA et al
Isolation and analysis of naturally processed (2003) A naturally selected dimorphism within
viral peptides as recognized by cytotoxic the HLA-B44 supertype alters class I struc-
T cells. Nature 348(6298):252–254 ture, peptide repertoire, and T cell recogni-
8. Storkus WJ, Zeh HJ, Maeurer MJ et al (1993) tion. J Exp Med 198(5):679–691
Identification of human melanoma peptides 19. Zernich D, Purcell AW, Macdonald WA et al
recognized by class I restricted tumor infiltrat- (2004) Natural HLA class I polymorphism
ing T lymphocytes. J Immunol 151(7): controls the pathway of antigen presentation
3719–3727 and susceptibility to viral evasion. J Exp Med
9. Storkus WJ, Zeh HJ, Salter RD et al (1993) 200(1):13–24
Identification of T-cell epitopes: rapid isolation 20. Scally SW, Petersen J, Law SC et al (2013) A
of class I-presented peptides from viable cells molecular basis for the association of the HLA-
by mild acid elution. J Immunother 14(2): DRB1 locus, citrullination, and rheumatoid
94–103 arthritis. J Exp Med 210(12):2569–2582
10. Falk K, Rötzschke O, Stevanovic S et al (1991) 21. Ringrose JH, Yard BA, Muijsers A et al (1996)
Allele-specific motifs revealed by sequencing of Comparison of peptides eluted from the
self-peptides eluted from MHC molecules. groove of HLA-B27 from Salmonella infected
Nature 351:290–296 and non-infected cells. Clin Rheumatol 15
11. Rammensee HG, Falk K, Rotzschke O (1993) Suppl 1:74–78
Peptides naturally presented by MHC class I 22. van Els CA, Herberts CA, van der Heeft E
molecules. Annu Rev Immunol 11:213–244 et al (2000) A single naturally processed
 A Systems Approach to Understand Antigen Presentation and the Immune Response 209

measles virus peptide fully dominates the 34. Ozato K, Sachs DH (1981) Monoclonal anti-
HLA-A*0201-associated peptide display bodies to mouse MHC antigens. III. Hybridoma
and is mutated at its anchor position in per- antibodies reacting to antigens of the H-2b
sistent viral strains. Eur J Immunol haplotype reveal genetic control of isotype
30(4):1172–1181 expression. J Immunol 126(1):317–321
23. Nepom BS, Nepom GT, Coleman M et al 35. Kappler JW, Skidmore B, White J et al (1981)
(1996) Critical contribution of beta chain resi- Antigen-inducible, H-2-restricted, interleukin-
due 57 in peptide binding ability of both 2-producing T cell hybridomas. Lack of inde-
HLA-DR and -DQ molecules. Proc Natl Acad pendent antigen and H-2 recognition. J Exp
Sci U S A 93(14):7202–7206 Med 153(5):1198–1214
24. Lampson LA, Levy R (1980) Two populations 36. Janeway CA Jr, Conrad PJ, Lerner EA et al
of Ia-like molecules on a human B cell line. (1984) Monoclonal antibodies specific for Ia
J Immunol 125(1):293–299 glycoproteins raised by immunization with
25. Gorga JC, Knudsen PJ, Foran JA et al (1986) activated T cells: possible role of T cellbound
Immunochemically purified DR antigens in Ia antigens as targets of immunoregulatory T
liposomes stimulate xenogeneic cytolytic T cells. J Immunol 132(2):662–667
cells in secondary in vitro cultures. Cell 37. Oi VT, Jones PP, Goding JW et al (1978)
Immunol 103(1):160–173 Properties of monoclonal antibodies to mouse
26. Watson AJ, DeMars R, Trowbridge IS et al Ig allotypes, H-2, and Ia antigens. Curr Top
(1983) Detection of a novel human class II Microbiol Immunol 81:115–120
HLA antigen. Nature 304(5924):358–361 38. Croft NP, Smith SA, Wong YC et al (2013)
27. Parham P, Brodsky FM (1981) Partial purifica- Kinetics of antigen expression and epitope pre-
tion and some properties of BB7.2. A cyto- sentation during virus infection. PLoS Pathog
toxic monoclonal antibody with specificity for 9(1), e1003129
HLA-A2 and a variant of HLA-A28. Hum 39. Tan CT, Croft NP, Dudek NL et al (2011)
Immunol 3(4):277–299 Direct quantitation of MHC-bound peptide
28. Ellis SA, Taylor C, McMichael A (1982) epitopes by selected reaction monitoring.
Recognition of HLA-B27 and related antigen Proteomics 11(11):2336–2340
by a monoclonal antibody. Hum Immunol 40. Wisniewski JR, Zougman A, Nagaraj N et al
5(1):49–59 (2009) Universal sample preparation method
29. Schittenhelm RB, Dudek NL, Croft NP et al for proteome analysis. Nat Methods 6(5):359–
(2014) A comprehensive analysis of constitu- 362. doi:10.1038/nmeth.1322
tive naturally processed and presented HLA- 41. Lange V, Picotti P, Domon B et al (2008)
C*04:01 (Cw4)-specific peptides. Tissue Selected reaction monitoring for quantitative
Antigens 83(3):174–179 proteomics: a tutorial. Mol Syst Biol 4:222
30. Braud VM, Allan DS, Wilson D et al (1998) 42. Edwards PA, Smith CM, Neville AM et al
TAP- and tapasin-dependent HLA-E sur- (1982) A human-hybridoma system based on
face expression correlates with the binding a fast-growing mutant of the ARH-77 plasma
of an MHC class I leader peptide. Curr Biol cell leukemia-derived line. Eur J Immunol
8(1):1–10 12(8):641–648
31. Thomas R, Apps R, Qi Y et al (2009) HLA-C 43. Alexander J, Payne JA, Murray R et al (1989)
cell surface expression and control of HIV/ Differential transport requirements of HLA
AIDS correlate with a variant upstream of and H-2 class I glycoproteins. Immunogenetics
HLA-C. Nat Genet 41(12):1290–1294 29(6):380–388
32. Hammerling GJ, Rusch E, Tada N et al (1982) 44. Nelson CA, Roof RW, McCourt DW et al
Localization of allodeterminants on H-2Kb (1992) Identification of the naturally pro-
antigens determined with monoclonal anti- cessed form of hen egg white lysozyme bound
bodies and H-2 mutant mice. Proc Natl Acad to the murine major histocompatibility com-
Sci U S A 79(15):4737–4741 plex class II molecule I-Ak. Proc Natl Acad Sci
33. Straus DS, Stroynowski I, Schiffer SG et al U S A 89(16):7380–7383
(1985) Expression of hybrid class I genes of 45. Escher C, Reiter L, MacLean B et al (2012)
the major histocompatibility complex in mouse Using iRT, a normalized retention time for
L cells. Proc Natl Acad Sci U S A 82(18): more targeted measurement of peptides.
6245–6249 Proteomics 12(8):1111–1121
 Chapter 15

Profiling of Small Molecules by Chemical Proteomics

Kilian V.M. Huber and Giulio Superti-Furga

Abstract
Chemical proteomics provides a powerful means to gain systems-level insight into the mode of action of
small molecules and/or natural products. In contrast to high-throughput screening efforts which only
interrogate selected subproteomes such as kinases and often only consider individual domains, the meth-
odology presented herein allows for the determination of the molecular targets of small molecules or drugs
in a more relevant physiological setting. As such, the compound of interest is exposed to the entire variety
of cellular proteins considering all naturally occurring posttranslational modifications and activation states.
Samples prepared according to the procedures described in this protocol are compatible with lysates from
cultured cell lines, primary cells, or samples from biopsies. In combination with state-of-the-art mass spec-
trometry techniques this approach grants access to a comprehensive view of small molecule-target protein
interactions.

Key words Drug discovery, Target deconvolution, Chemical proteomics, Mode of action

1 Introduction

Phenotypicscreening represents an interesting strategy for identifying

new potential therapeutics [1]. The fact that the compounds dis-
covered by this approach exhibit a desired phenotype in living cells
makes them attractive candidates for further development.
However, understanding the molecular target and thus the mode
of action can pose a significant challenge [2]. There are also a num-
ber of approved drugs which have proven efficacious in the treat-
ment of human disease over decades for which it is not clear by
which mechanism they work [1]. Knowledge of the relevant cel-
lular target(s) would allow for the development of more potent
and selective drugs with probably less side effects. Even for drugs
whose mode of action is known a comprehensive target profile can
assist further patient stratification both in terms of applicability and
managing side effects [3]. Moreover, it can also reveal new poten-
tial indications due to previously unknown off-targets. Common
“high-throughput” in vitro approaches to determine the targets of

Jörg Reinders (ed.), Proteomics in Systems Biology: Methods and Protocols, Methods in Molecular Biology, vol. 1394,
DOI 10.1007/978-1-4939-3341-9_15, © Springer Science+Business Media New York 2016

211
212 Kilian V.M. Huber and Giulio Superti-Furga

small molecules or natural products are often both time and cost
intensive and usually only cover a selected range of protein classes,
e.g., kinases or histone deacetylases. Furthermore, these assays do
not provide any information if the targets identified in the screen
are “real-and-relevant” interactors as such issues as differential tis-
sue expression or even posttranslational modifications are not
considered.
Several target deconvolution approaches have been developed
covering a wide range of biochemical and genetic techniques as
well as in silico methods [2, 4]. Among all those affinity-based
approaches have contributed to the discovery of a number of
important drug classes such as immunosuppressants [5, 6] and his-
tone deacetylase (HDAC) inhibitors [7]. The combination of this
methodology with protein mass spectrometry has been termed
“chemical proteomics” and has lately been applied to diverse areas
of research including the elucidation of the molecular mechanism
of thalidomide teratogenicity [8] as well as the identification of
novel regulators of necroptosis signaling [9] and potential antican-
cer targets [10]. In a typical chemical proteomic experiment, the
small molecule or compound of interest is incubated with a rele-
vant cell lysate which can be prepared from tissue culture cell lines
or even primary cells and biopsy samples. Using the small-molecule
compound as bait by means of a drug matrix, cellular interactors
are captured and purified by affinity enrichment. After washing,
the eluted proteins are digested to peptides and can subsequently
be identified by mass spectrometry followed by bioinformatic anal-
ysis (Fig. 1). This procedure requires the compound of interest
itself or a corresponding analogue to be amenable to chemical
derivatization in order to be immobilized on the solid phase.
The design of a suitable analogue is facilitated by prior knowledge
of structure-activity relationships (SAR) or co-crystal structures of
annotated targets. Alternatively, the so-called east–west approach
may be applied to compounds devoid of those data [11]. The concept
of this strategy is to prepare two coupleable analogues of the
compound of interest of which each is modified at a different,
preferably most distant site.
An alternative chemical proteomic approach is activity-based
probe profiling (ABPP) which does not require chemical modifica-
tion of the query compound by taking advantage of a reactive
probe which binds covalently to a given class of proteins, e.g.,
serine hydrolases [12–14]. In this case, comparison of treated versus
untreated sample yields the putative interactors. Recently, another
proteomic target identification strategy based on ligand-induced
thermal stabilization of proteins has been established [15]. This
methodology termed thermal stability profiling does also not
depend on chemical derivatization and instead uses the unmodified
original compound of interest allowing for the detection of protein-
ligand interactions in intact living cells. For a general overview of
Fig. 1 Schematic outline of a chemical proteomic experiment. A drug matrix consisting of the immobilized small molecule on a solid phase is incubated with lysates
from either cell lines, primary cells, or biopsy samples. After incubation, contaminants are removed by extensive washing and finally target proteins are eluted and
analyzed by mass spectrometry followed by bioinformatic processing
Profiling of Small Molecules by Chemical Proteomics
213
214 Kilian V.M. Huber and Giulio Superti-Furga

recent proteomic technologies relevant to the field of drug and

target discovery the reader is directed to the literature [14].
The procedures presented herein assume that the investigative
compounds bear a nucleophilic handle suitable to react with
N-hydroxy-succinimide (NHS) esters, e.g., a functional group
such as a primary or secondary aliphatic amine. However, this
protocol can easily be adapted to using biotinylated drugs by
changing the solid phase accordingly.

2 Materials

2.1 Reagents 1. Nonidet P-40 (NP-40).

2.1.1 Lysis Buffer 2. Tris–HCl pH 7.5.
for Preparation of Whole- 3. Glycerol.
Cell Lysates 4. MgCl2.
5. NaCl.
6. NaF.
7. Na3VO4.
8. Phenylmethylsulfonyl fluoride (PMSF).
9. Dithiothreitol (DTT).
10. N-p-Tosyl-L-phenylalanine chloromethyl ketone (TPCK).
11. Inhibitor cocktail containing leupeptin, aprotinin, soybean
trypsin inhibitor.
12. Bradford reagent.
Lysis buffer (LB): 0.20 % NP-40, 50 mM Tris–HCl pH 7.5,
5 % glycerol, 1.5 mM MgCl2, 100 mM NaCl, 25 mM NaF, 1 mM
Na3VO4, 1 mM PMSF, 1 mM DTT, 10 μg/mL TPCK, 1 μg/mL
leupeptin, 1 μg/mL aprotinin, 10 μg/mL soybean trypsin inhibitor;
prepare freshly, keep on ice.

2.1.2 Compound 1. Coupleable compound (drug) (see Note 1).

Coupling Protocol 2. NHS-activated sepharose beads (50 % slurry in isopropanol).
3. Dimethyl sulfoxide (DMSO) (abs.).
4. Triethylamine (TEA).
5. Ethanolamine.
6. Isopropanol.

2.1.3 Drug Pull-Down 1. Drug matrix (compound immobilized on beads).

Protocol 2. DMSO (abs.).
3. Lysis buffer (LB).
4. HEPES.
 Profiling of Small Molecules by Chemical Proteomics 215

5. EDTA.
6. NaCl.
7. NaOH.
8. Purified water.
9. Formic acid.
HEPES-NaOH-EDTA buffer pH 7.5: 50 mM HEPES, 0.5 μM
EDTA pH 8.0, 100 mM NaCl; adjust pH to 7.5 using NaOH,
prepare freshly, keep on ice.

2.2 Equipment 1. Needles and syringes.

2. Pipettes.
3. Filter tips.
4. Polycarbonate ultracentrifuge tubes.
5. Ultracentrifuge.
6. Photometer.
7. Tabletop centrifuge.
8. Rotoshaker.
9. HPLC with MS and/or UV detector.
10. Chromatography columns.
11. Mass spectrometry glass vials.

3 Methods

3.1 Preparation 1. (a) For cell pellets, thaw pellets on ice and resuspend in lysis
of Whole-Cell Lysates buffer (depending on pellet size, rule-of-thumb 1:1 ratio),
(Timing 2 h) (See transfer into homogenizer/Dounce apparatus (e.g., 0.9 mm
Notes 2 and 3) syringe) and homogenize sample ten times (optional).
(b) For tissues, transfer the sample into a tissue homoge-
nizer/Dounce apparatus, wash with lysis buffer, and adjust to
desired volume; homogenize sample ten times.
2. Transfer the homogenate to a Falcon tube and incubate on ice
for 30 min.
3. Transfer homogenate to polycarbonate ultracentrifuge tubes,
balance tubes, and centrifuge lysate for 10 min at 4 °C at
20,000 × g.
4. Transfer supernatant to fresh polycarbonate ultracentrifuge
tubes, balance tubes, and centrifuge for 1 h at 4 °C at 90,000 × g.
5. Transfer supernatant (remove most of lipid layer, if possible)
to a fresh Falcon centrifuge tube, and keep on ice.
6. Determine protein concentration (e.g., Bradford).
7. Prepare lysate aliquots (e.g., 5–10 mg total protein) or use
directly for pull-downs (see Note 4).
216 Kilian V.M. Huber and Giulio Superti-Furga

Pausing point: After shock-freezing in liquid nitrogen the

lysate aliquots may be stored at −80 °C until use.

3.2 Coupling 1. Pipet 100 μL of slurry (≈50 μL settled bed volume) in a 1.5
of Compounds mL Eppendorf tube. Caution: Always use filter tips and cut
to Sepharose Beads pipet tips for pipetting beads!
(Timing 3 Days) 2. Centrifuge beads for 3 min at room temperature at 75 × g, and
remove supernatant.
3. Add 50 μL of DMSO (abs) to beads, suspend gently by invert-
ing several times, centrifuge (as before), and discard
supernatant.
4. Add 500 μL DMSO (abs), resuspend beads gently by inverting
several times, transfer beads in a 2 mL Eppendorf tube, centri-
fuge (as before), and discard supernatant; repeat wash step
another two times.
5. Resuspend beads in 50 μL DMSO (abs).
6. Add 0.025 μmol of coupleable compound to 50 % bead slurry;
add 0.75 μL triethylamine, and mix carefully.
7. Incubate on roto-shaker for 16–24 h at RT with 10 rpm.
8. To check coupling efficiency, centrifuge beads (as before) and
remove 10 μL (≈5 nmol) from supernatant.
9. Check for remaining unreacted compound by HPLC; if there
are still significant amounts detected in the supernatant go
back to step 7 and extend coupling reaction time. Repeat steps
8 and 9.
10. Add 2.5 μL ethanolamine to drug-bead mixture in order to
block unreacted NHS-ester groups.
11. Incubate on roto-shaker for at least 8 h at room temperature
with 10 rpm.
12. Centrifuge beads (see above), remove supernatant, and wash
twice with 500 μL of DMSO (abs).
13. Proceed directly with pull-down and wash with lysis buffer.
Pausing point: Alternatively, the drug-bead matrix can be
stored for up to 2 weeks using the following procedure: Remove
supernatant and add 50 μL of isopropanol, resuspend beads gently
by inverting several times, centrifuge (as before), and discard
supernatant. Add 500 μL isopropanol, resuspend beads gently by
inverting several times, centrifuge (as before), and discard
supernatant; repeat wash step once. Add 50 μL isopropanol to
beads and resuspend gently; store coupled beads at 4 °C (away
from light) until further use.

3.3 Pull-Down 1. Centrifuge beads for 3 min at 75 × g, and remove supernatant.

Procedure (Timing 6 h) 2. Add 1 mL lysis buffer and wash beads gently by resuspending
(See Note 5) and inverting several times.
 Profiling of Small Molecules by Chemical Proteomics 217

3. Centrifuge beads (see above), remove supernatant, and repeat

wash step three times.
4. Remove supernatant.
5. Dilute cell lysate with lysis buffer to a final protein concentra-
tion of 15 mg/mL.
6. Transfer to ultracentrifuge tube, balance tubes, and centrifuge
for 20 min at 4 °C at 90,000 × g.
7. Remove 200 μg protein of whole-cell lysate from supernatant
(as input control for Western blot).
8. Decant remaining supernatant directly onto compound-beads.
9. Gently resuspend washed compound-beads in cell lysate.
10. Incubate on roto-shaker for 2 h at 4 °C with 10 rpm.
11. After the incubation centrifuge beads for 3 min at 75 × g.
12. Wash disposable chromatography column twice with 1 mL
lysis buffer.
13. Caution: Perform the following steps at 4 °C! Resuspend beads
gently by pipetting up and down and transfer to plugged col-
umns. Let beads settle by gravity, unplug column, and then
drain remaining buffer by gravity flow.
14. Add 5 mL lysis buffer, and let buffer drain by gravity flow.
15. Add 2.5 mL 50 mM HEPES-NaOH buffer, and let buffer
drain by gravity flow.
16. Wash column tip twice with 0.5 mL HEPES buffer, place in
centrifuge tube, and spin down for 1 min with 100 × g.
17. For MS analysis place a suitable glass vial under the column to
collect the sample. Add 250 μL formic acid, let the sample
drain by gravity, and remove remaining liquid from matrix by
plugging/unplugging the column lid (3×). Submit sample to
MS processing.

3.4 Concluding The described procedures provide an effective means to reveal the
Remarks interactors of small molecules and natural compounds. However,
due to the high complexity of the samples and varying protein abun-
dance it is recommendable to include control pull-downs which can
be either the unreacted bead matrix itself or an unrelated compound
matrix to estimate random and unspecific binding. Alternatively,
lysates can be preincubated with the original, unmodified compound
of interest to determine competitive binding (Fig. 1). If coupled
with quantitative MS technologies such as iTRAQor TMT this
approach also allows for the determination of Kds.
218 Kilian V.M. Huber and Giulio Superti-Furga

4 Notes

1. To maximize pull-down efficacy, coupleable compound

analogues should be evaluated in vitro prior to the pull-down
experiment.
2. This procedure is in general applicable for most protein targets
of small molecules; however, some nuclear and transmembrane
proteins may require optimized lysis conditions.
3. Depending on the stability of certain proteins it may be
advisable to perform cell lysis and compound pull-down on
the same day to avoid detrimental freeze-thawing cycles.
4. The required amount of total protein per pull-down depends
largely on the sensitivity of the subsequent analytical method.
As a rule of thumb, 5–10 mg should suffice for both applica-
tions described herein.
5. For the preparation of mass spectrometry samples it is crucial to
avoid any potential contamination with keratin. Also, materials
which are not sensitive to the chemicals used in this procedure
should be considered preferably in order to avoid contamina-
tion with degradation products and/or polymers.

References

1. Swinney DC, Anthony J (2011) How were mide teratogenicity. Science 327(5971):
new medicines discovered? Nat Rev Drug 1345–1350
Discov 10(7):507–519 9. Sun L, Wang H, Wang Z et al (2012) Mixed
2. Schenone M, Dancik V, Wagner BK et al lineage kinase domain-like protein mediates
(2013) Target identification and mechanism of necrosis signaling downstream of RIP3 kinase.
action in chemical biology and drug discovery. Cell 148(1–2):213–227
Nat Chem Biol 9(4):232–240 10. Huber KVM, Salah E, Radic B et al (2014)
3. Sillaber C, Herrmann H, Bennett K et al (2009) Stereospecific targeting of MTH1 by (S)-
Immunosuppression and atypical infections in crizotinib as an anticancer strategy. Nature
CML patients treated with dasatinib at 140 mg 508(7495):222–227
daily. Eur J Clin Invest 39(12):1098–1109 11. Rix U, Superti-Furga G (2009) Target profil-
4. Terstappen GC, Schlupen C, Raggiaschi R et al ing of small molecules by chemical proteomics.
(2007) Target deconvolution strategies in drug Nat Chem Biol 5(9):616–624
discovery. Nat Rev Drug Discov 6(11):891–903 12. Nomura DK, Dix MM, Cravatt BF (2010)
5. Brown EJ, Albers MW, Bum Shin T et al Activity-based protein profiling for biochemi-
(1994) A mammalian protein targeted by cal pathway discovery in cancer. Nat Rev
G1-arresting rapamycin-receptor complex. Cancer 10(9):630–638
Nature 369(6483):756–758 13. Savitski MM, Reinhard FBM, Franken H et al
6. Harding MW, Galat A, Uehling DE et al (2014) Tracking cancer drugs in living cells by
(1989) A receptor for the immuno-suppressant thermal profiling of the proteome. Science
FK506 is a cis-trans peptidyl-prolyl isomerase. 346(6205):1255784
Nature 341(6244):758–760 14. Huber KVM, Olek KM, Müller AC et al (2015)
7. Taunton J, Hassig CA, Schreiber SL (1996) A Proteome-wide drug and metabolite interaction
mammalian histone deacetylase related to the mapping by thermal-stability profiling. Nat
yeast transcriptional regulator Rpd3p. Science Meth 12(11):1055–1057
272(5260):408–411 15. Schirle M, Bantscheff M, Kuster B (2012) Mass
8. Ito T, Ando H, Suzuki T et al (2010) spectrometry-based proteomics in preclinical
Identification of a primary target of thalido- drug discovery. Chem Biol 19(1):72–84
 Chapter 16

Generating Sample-Specific Databases for Mass

Spectrometry-Based Proteomic Analysis by Using RNA
Sequencing
Toni Luge and Sascha Sauer

Abstract
Mass spectrometry-based methods allow for the direct, comprehensive analysis of expressed proteins and
their quantification among different conditions. However, in general identification of proteins by assigning
experimental mass spectra to peptide sequences of proteins relies on matching mass spectra to theoretical
spectra derived from genomic databases of organisms. This conventional approach limits the applicability
of proteomic methodologies to species for which a genome reference sequence is available. Recently,
RNA-sequencing (RNA-Seq) became a valuable tool to overcome this limitation by de novo construction
of databases for organisms for which no DNA sequence is available, or by refining existing genomic data-
bases with transcriptomic data. Here we present a generic pipeline to make use of transcriptomic data for
proteomics experiments. We show in particular how to efficiently fuel proteomic analysis workflows with
sample-specific RNA-sequencing databases. This approach is useful for the proteomic analysis of so far
unsequenced organisms, complex microbial metatranscriptomes/metaproteomes (for example in the
human body), and for refining current proteomics data analysis that solely relies on the genomic sequence
and predicted gene expression but not on validated gene products. Finally, the approach used in the here
presented protocol can help to improve the data quality of conventional proteomics experiments that can
be influenced by genetic variation or splicing events.

Key words Mass spectrometry, Metaproteomics, Proteogenomics, Proteomics informed by transcrip-

tomics, Gene expression

1 Introduction

Improvements in high-throughput liquid chromatography-coupled

tandem mass spectrometry (LC-MS/MS) and the broad availabil-
ity of quantitative techniques, such as stable isotope labeling of
amino acids in cell culture (SILAC) [1], dimethyl-labeling [2], iso-
baric tags for relative and absolute quantitation (iTRAQ) [3],
absolute quantification (AQUA) peptides [4], and label-free
methods like intensity-based absolute quantification (iBAQ) [5]
amongst others, paved the way for comprehensive relative protein

Jörg Reinders (ed.), Proteomics in Systems Biology: Methods and Protocols, Methods in Molecular Biology, vol. 1394,
DOI 10.1007/978-1-4939-3341-9_16, © Springer Science+Business Media New York 2016

219
220 Toni Luge and Sascha Sauer

expression profiling in complex samples over different conditions

[6, 7]. Typically, in a shotgun discovery proteomics workflow, pro-
teins are digested to peptides by sequence-specific proteases, for
instance with trypsin or Lys-C. For in-depth analysis, samples are
often pre-fractionated prior to (e.g., by SDS-PAGE [8]) or after
the digestion step (e.g., by isoelectric focusing or strong anion-
exchange chromatography [9]) and subsequently submitted to
LC-MS/MS.
The characteristic mass-to-charge ratios of the proteolytic pep-
tides and their fragment ions, such as y and b ions that are typically
recorded in data-dependent acquisition (DDA) mode by LC-MS/
MS instruments [10], form distinct mass spectra that contain
sequence information. To determine peptide sequences from frag-
ment mass spectra and to infer proteins, sequence database search—
by, e.g., Mascot [11], Sequest [12], X!Tandem [13], MyriMatch
[14], Paragon [15], or Andromeda [16]—is the most commonly
applied approach for peptide and protein identification [17].
Thereby experimental mass spectra are matched with in silico-
generated mass spectra that are derived from available sequence
databases. Thus, the chosen sequence database can directly affect
search results because only peptides of which the amino acid
sequences are deposited can be assigned. Retrieving protein
sequence databases from public resources like Uniprot and
ENSEMBL is a straightforward strategy for well-annotated model
organisms with fully sequenced genomes, high-quality definition
of transcriptional units, and predicted proteomes. But this approach
represents the primary bottleneck to the widespread use of quanti-
tative shotgun proteomics for the analysis of orphan organisms or
metaproteomes. Furthermore, public databases are collections of
all known and predicted proteins in a species but incomplete with
respect to single-nucleotide variants (SNVs) and RNA-splice and
-editing variants. Thus, they do not represent the real protein com-
position in a specific cell or tissue type [18].
With the increasing affordability of next-generation sequenc-
ing (NGS) technologies [19] many proteomic studies incorporated
large-scale RNA expression analysis to correlate and verify protein
by transcript expression data. A growing number of researchers
start using this valuable resource to construct or refine their search
space for improved protein identification and quantification [18,
20–30].
For instance, Wang et al. [18] investigated SNVs on protein
level in human colorectal cancer cell lines SW480 and
RKO. Therefore they performed RNA-sequencing (RNA-Seq),
aligned short reads to the human genome, and added nonsynony-
mous protein-coding SNVs to the regular protein sequence data-
base. Assuming that proteins with low-abundance transcripts are
likely to be undetectable by shotgun proteomics, an RPKM (reads
per kilobase of exon per million mapped reads) threshold for tran-
 Sample Specific Databases for Proteomics 221

script abundance was applied to reduce the size of reference pro-

tein sequence database. This strategy improved sensitivity of
peptide identification and reduced ambiguity in protein assembly.
The same researchers incorporated their approach in the R package
customProDB [31].
Recently, procedures have evolved that even do not need refer-
ence sequences for alignment. These emerging procedures are
based on applying de novo assemblies of transcriptomes. Thus,
these methods are in particular powerful when studying non-
sequenced organisms or even communities of different species, as
is often the case in the emerging field of metaproteomics.
The great potential of this technique has been benchmarked
by Evans et al. [20], who used RNA-Seq to build a reference pro-
tein database by six-frame translation of the transcriptome of
adenovirus-infected HeLa cells. This proteomics informed by tran-
scriptomics (PIT) workflow identified more than 99 % of the pro-
teins when using traditional protein sequence databases with
annotated human and adenovirus proteins. Furthermore, we
recently adopted the PIT approach to study the effect of
“Candidatus Phytoplasma mali” strain AT infection on protein
expression in Nicotiana occidentalis (tobacco) plants [32] and
additionally applied Blast2GO [33] analysis for de novo annota-
tion of identified proteins. Thereby, we could show that these
resources can also be fruitfully applied for the construction of tar-
geted proteomics assays. Also custom protein sequences derived
from RNA-Seq might be incorporated in the analysis of mass spec-
trometry data acquired in data-independent (DIA) modes such as
SWATH [34] and MSX [35].
In general, the PIT approach might be useful for various pro-
teomics applications such as investigating metatranscriptomes/
metaproteomes, and refining proteomics data analysis to cope with
just predicted but non-validated gene/protein expression as well as
incomplete annotation of genetic variation and splicing events.
The general workflow for mass spectrometric analysis of protein
samples by using RNA-Seq data is outlined in Fig. 1.

2 Materials

1. Users need to install Trinity software [36] in order to de novo

assemble the transcriptome from short read RNA-seq data,
available for unix-type operating systems from http://trinityr-
naseq.github.io. It is best run in a high-performance comput-
ing environment with ~1 GB of RAM per 1 million paired-end
reads. Alternatively, when lacking these resources, Trinity is
also accessible on the Data Intensive Academic Grid (DIAG,
http://diagcomputing.org/), a shared computational cloud
for academic and nonprofit institutions.
222 Toni Luge and Sascha Sauer

Fig. 1 Scheme of the bioinformatics workflow for the integrative analysis of shot-
gun mass spectrometry and RNA-Seq data. Proteins are subjected to LC-MS/MS
analysis whereas RNA, ideally isolated from the same samples used for pro-
teomics, is sequenced on next-generation sequencing (NGS) instruments [19].
Short read sequencing data is used to reconstruct protein sequences used as
sequence database in the peptide search engine for identification of peptides
and proteins from mass spectrometric raw data. Finally, identified proteins can
be de novo annotated based on sequence homology search to known proteins

2. The “getorf” tool of the EMBOSS suite [37] is needed and

available on the open-source, Web-based platform Galaxy
(http://www.usegalaxy.org/).
3. Additional analysis steps take place in the statistical program-
ming environment R [38], downloadable from http://cran.r--
project.org/. Besides the base installation the Biostrings R
package, available from http://www.bioconductor.org/, is
needed.
 Sample Specific Databases for Proteomics 223

4. Database searching of proteomic data is performed in the

MaxQuant computational proteomics platform [16, 39]
(http://www.maxquant.org/), which requires Windows oper-
ating system and the MS File Reader for accessing Thermo
Scientific instrument files (Thermo Scientific, https://thermo.
flexnetoperations.com/control/thmo/login).
5. For functional annotation of sequences and analysis of annota-
tion data install Blast2GO [33] (http://www.blast2go.com/
b2ghome).

3 Methods

The presented bioinformatics pipeline is a guide to the analysis of

proteomic data generated by mass spectrometry with the help of a
sample-specific protein sequence database for the identification and
quantification of proteins. The workflow is composed of four major
steps: RNA-Seq, protein sequence database construction, database
searching of MS/MS data, and annotation of identified proteins.

3.1 RNA-Seq Consult responsible NGS bioinformatics expert to define an appro-

priate experimental design matching your needs (see Note 1). We
highly recommend isolating total RNA from the same samples
used for protein extraction and proteomic analysis. In typical RNA-
Seq workflows mRNA is enriched by Poly A+ selection prior to
sequencing to increase the informative fraction in samples.
However, when dealing with prokaryotes whole-transcriptome
sequencing after depleting ribosomal RNAs with, e.g., RiboMinus-
Kits (Invitrogen) might be the better choice. Alternatively, publicly
available RNA-Seq data sets can be used as well (see Note 2).

3.2 Construction The fundament of the protein sequence database construction is

of Protein Sequence formed by the de novo transcriptome assembly from short read
Database RNA-Seq data using Trinity software [36] (see Note 3). This
assembly pipeline consists of the three consecutive modules
Inchworm, Chrysalis, and Butterfly (Fig. 2). Briefly, the first mod-
ule Inchworm generates transcript contigs from the RNA-Seq
reads which are clustered in the second step by Chrysalis into
regions that have probably originated from alternatively spliced
transcripts or closely related gene families. Chrysalis encodes this
structural complexity by building the Bruijn graphs for each clus-
ter. Finally, Butterfly traces the RNA-Seq reads through the graphs
and traverses supported graph paths to reconstruct full-length
transcripts for alternatively spliced isoforms while teasing apart
transcripts that correspond to paralogous genes. Trinity accepts
pre-processed single- or paired-end short read data in either
FASTQ or FASTA formats. Pre-processing involves removing
barcodes used for multiplexing on the sequencing instrument
224 Toni Luge and Sascha Sauer

Fig 2 Overview of the reconstruction of protein sequences from short read RNA-
sequencing (RNA-Seq) data. The Trinity pipeline [36] is used to build full-length
transcripts. This software first assembles contigs (sets of overlapping sequences)
in the Inchworm module. The Chrysalis module clusters contigs and processes
each cluster to a de Bruijn graph component. Butterfly, the third module, extracts
all probable sequences from the graph components and reports reconstructed
sequences and their isoforms. Protein sequences are inferred by the EMBOSS
tool “getorf” [37] and subjected to the Andromeda search engine [16] for peptide
and protein identification from mass spectrometry raw data

and reads that probably contain sequencing errors (see Note 4).
The reconstructed transcripts are translated to protein sequences
using the EMBOSS tool “getorf.” This file is subjected to the
Andromeda search engine for peptide identification from mass
spectrometry data (see Subheading 3.3).
1. Log in from your local working station to your Unix server
where Trinity is installed. For instance, use the free SSH and
telnet client PuTTY (or appropriate alternatives) when work-
ing on Windows operating system. Open terminal (“$” denotes
the shell prompt) and copy raw or if necessary pre-processed
sequencing data in FASTQ format to the working folder:
$ mkdir workingfolder
$ cp rawdata/Illumina_singleread_*.fq /
workingfolder
$ cd workingfolder
 Sample Specific Databases for Proteomics 225

2. When multiple sequencing runs are conducted for a single

experiment, concatenate RNA-Seq data into a single file to cre-
ate a single reference Trinity assembly. When using paired-end
data, combine all left and right reads into separate files:
$ cat Illumina_singleread_*.fq > combined_sin-
gleread.fq
3. Assemble reads into transcripts using Trinity:
$ trinity_directory/Trinity.pl --seqType fq --JM 50G
--single combined_singleread.fq --CPU 10
The “--seqType” option specifies input data format (“fq”
for FASTQ, “fa” for FASTA), the “--JM” option controls
amount of RAM used by processes of the Inchworm module,
the “--single” option defines whether short reads are single
or paired-end, and the “--CPU” option facilitates paralleliza-
tion of processes. For more options see the Trinity documenta-
tion and choose appropriate parameters to fit your
computational resources as well as RNA-Seq design like paired-
end reads and strand specification of reads. Trinity per default
outputs the de novo-assembled transcripts in FASTA format to
“trinity_out_dir/Trinity.fasta” (see Note 5).
4. Copy “Trinity.fasta” to your local working station and
upload the file to Galaxy (http://www.usegalaxy.org/) using the
“Get Data” “Upload File” function of the homepage.
5. Translate “Trinity.fasta” into proteins using the “EMBOSS”
“getorf” [37] function in Galaxy. Use transcripts with a min-
imum nucleotide length of 200 bp and output translated
regions between start and stop codons. Save the resulting mul-
tiple FASTA file of protein sequences “Galaxy4-[getorf_
on_data_1].fasta” to your local working station.

3.3 Searching MS/ Discovery proteomic raw data acquired on high-resolution MS

MS Data instruments operating in data-dependent mode is analyzed in the
MaxQuant computational proteomics platform [39] (Fig. 3). The
integrated Andromeda peptide search engine [16] is used to search
MS/MS spectra against the RNA-Seq-derived protein sequence
database by taking the target decoy approach to control the false
discovery rate (see Note 6).
1. Execute the “AndromedaConfig.exe” and switch to the
Sequences tab (Fig. 4).
2. Select “Add new row” to create a new protein sequence data-
base. Load the “Galaxy4-[getorf_on_data_1].
fasta” file in the “Database” field.
3. Select an appropriate regular expression rule to correctly parse
FASTA headers in the multiple FASTA file. The rule “>([^
]*)” will output the whole header up to the first space in the
MaxQuant search result files.
226 Toni Luge and Sascha Sauer

Fig 3 Main steps performed by the MaxQuant software [39] for the analysis of shotgun proteomic data acquired
on high-resolution liquid chromatography-coupled tandem mass spectrometry (LC-MS/MS) instruments. The
software suite contains algorithms to efficiently extract information from raw mass spectrometry data includ-
ing peptide and protein identities as well as their high-accuracy quantification. First features are detected on
MS1 level as three-dimensional objects in m/z, retention time, and intensity space. Isotope label partners are
identified, features quantified and normalized, and their masses calculated precisely. To achieve mass accura-
cies in the p.p.b. range, mass errors (Δm/z ) are estimated and mass offsets corrected by nonlinear mass
recalibration. Furthermore MS/MS spectra are preprocessed and filtered prior to the main MS/MS ion search
by the Andromeda search engine [16]. Peptide spectrum matches are scored and the false discovery rate
(FDR) controlled on peptide and protein level by decoy database searching [17]. Additionally MaxQuant
accounts for co-fragmented peptides occasionally observed due to the complexity of protein samples.
Optionally, isotope-labeled peptide pairs are re-quantified if one label partner is missing and peptide identifica-
tions matched between different LC-MS/MS runs after their retention time alignment. In the consolidation
phase proteins are assembled from peptide identifications, protein ratios calculated from peptide ratios by
robust median averaging, and results compiled in multiple cross referenced tab-delimited .txt files that are
used for further analysis steps in Perseus (www.maxquant.org) or R [38]

4. Save the rule in the “Rule” field. Afterwards click on “File”

“Save” “databases”. The RNA-Seq-derived protein
sequence database has now been created.
5. Launch the “MaxQuant.exe” to analyze mass spectrometric
raw data. Load the data and specify the experimental design of
the experiment by assigning experiment and fraction to each
file. Adjust parameters like isotope labels and their multiplicity,
enzyme used for digestion of proteins to peptides, allowed
missed cleavages, variable and fixed modifications, protein
identification, and quantification criteria to fit your sample
preparation methods and experimental design. In the field
“Sequences” add the FASTA file you configured for
Andromeda search engine using the “AndromedaConfig.
exe”. Check the box “Include contaminants”, which
will include a list of typically observed contaminants in samples
like keratins. Select the number of threads you reserve for mul-
tithreading and speeding up the analysis. Start MaxQuant. The
software exports result files in tab-delimited .txt format to the
 Sample Specific Databases for Proteomics 227

Fig. 4 Configuration of the RNA-Seq-derived protein sequences as search space in the Andromeda peptide
search engine of the MaxQuant computational proteomics platform [16, 39]. After installation of the MaxQuant
software package launch the “AndromedaConfig.exe” from the MaxQuant directory. The screenshot shows the
graphical user interface of version 1.3.0.5 with the EMBOSS “getorf”- processed Trinity de novo assembly
“Galaxy4-[getorf_on_data_1].fasta” selected as a new database

new folder “txt”, a subfolder in the “combined” directory

that was created when starting MaxQuant. For instance, the
“proteinGroups.txt” contains information on the identi-
fied proteins in the processed raw files. Each single row con-
tains the group of proteins that could be reconstructed from a
set of peptides alongside with, e.g., quantitative information
like normalized ratios between isotope label partners.
6. Inspect search results and perform secondary data analysis
steps including testing for differential protein expression in
Perseus (http://www.maxquant.org/) or R [38]. Apply
appropriate statistics to identify proteins whose expression
level changes between conditions.

3.4 Annotation In the last step the novel sequences identified to be present in the
of Sequence Data sample on RNA and protein level are to be annotated based on
homology search to known sequence data. Annotating the whole
de novo-assembled transcriptome is time consuming. However,
taking the fraction that could be positively identified in the pro-
teomics experiment will reduce data amount and speed up the
analysis. Therefore the FASTA header of the protein sequences
reported in the MaxQuant output files is used to parse the custom
protein sequence database via regular expressions and extract the
228 Toni Luge and Sascha Sauer

relevant protein sequences. This reduced multiple FASTA file of

protein sequences is subjected to the Blast2GO software [33], a
tool for the annotation of sequences and the analysis of annotation
data (see Note 7).
1. Extract the amino acid sequences of the proteins that were
identified in the proteomics experiment from the protein
sequence database. For instance, this can be performed in R
using the following commands (“>” denotes the R prompt).
After starting R, download the Biostrings R Package or
load the library if already installed:
>source("http://bioconductor.org/biocLite.R")
>biocLite("Biostrings")
>library(Biostrings)
Load the protein sequence database and the MaxQuant
search result file “proteinGroups.txt”:
>proteinsequences=read.AAStringSet("C:/working-
folder/MaxQuant_analysis/Galaxy4-[getorf_on_
data_1].fasta")
>proteingroups= read.table("C:/workingfolder/
MaxQuant_analysis/combined/txt/proteinGroups.
txt", header=TRUE, sep="\t", na.strings=c("NaN"))

Create a vector of the FASTA headers of identified proteins:

>major_protein_id=as.vector(proteingroups [,"Majority.
protein.IDs"])
Parse the protein sequence database via FASTA headers of
identified proteins and save the new multiple FASTA file:
>name=names(proteinsequences)
>id=unique(grep(paste(major_protein_
id,collapse="|"), name, value=FALSE))
>identified_proteins_sequences=proteinsequences[id]
>writeXStringSet(identified_proteins_sequences,
file="C:/workingfolder/Blast2GO_analysis/pro-
tein_sequences.fasta")
2. Load the file “protein_sequences.fasta” into the Blast2GO
software by using the “File” “Load Sequences (e.g.:
.fasta)” option in the menu bar of the graphical user
interface.
3. Start the Blast step by selecting “Blast” → “Run BLAST Step”.
This will use NCBI’s BLAST service. You may choose the blast
program and database to match your needs. For instance, use
blastp to search protein sequences against the nr protein
sequence database of NCBI.
4. The mapping step will link Blast hits to information stored in
the Gene Ontology database. Select “Mapping” → “Run GO-
Mapping Step”.
 Sample Specific Databases for Proteomics 229

5. Perform the annotation step. The mapped GO terms will now

be assigned to the query sequences. Choose “Annotation” in
the menu bar and “Run Annotation Step”. Additionally run
the “InterProScan” and “Merge InterProScan GOs to
Annotation” which uses the functionality of InterPro annota-
tions to retrieve motif and domain information. The corre-
sponding GO terms are merged with already existing GO
terms from the mapping step.

4 Notes

1. Take into account that sequencing depth and read length

directly influence quality of de novo transcriptome assembly. A
coverage of the protein-coding transcriptome as low as 11×
might be sufficient to identify most of the present peptides in
proteomics experiment [20], but underpowers other analyses
like differential gene expression and SNV calling. Although
short read sequencing on Illumina’s Genome Analyzer for only
35 cycles has been used successfully in protein sequence data-
base construction [25], keep in mind that longer reads reduce
read ambiguity during transcriptome assembly, especially when
dealing with metatranscriptomes. For instance, Adamidi et al.
[23] combined long and relatively few reads obtained by 454
GS FLX (Roche) sequencing with shorter but many more
reads from Illumina GAIIX (Illumina) sequencing to assemble
a high-quality transcriptome of Schmidtea mediterranea.
2. Publicly available RNA-Seq data sets can be retrieved from reposi-
tories like ArrayExpress (https://www.ebi.ac.uk/arrayexpress/),
Gene Expression Omnibus (GEO, http://www.ncbi.nlm.nih.
gov/geo/), NCBI’s Sequence Read Archive (SRA, http://www.
ncbi.nlm.nih.gov/Traces/sra), European Nucleotide Archive
(ENA, http://www.ebi.ac.uk/ena/), and the RNA-Seq Atlas
(http://www.medicalgenomics.org/rna_seq_atlas).
3. Amongst Trinity several other software tools have been devel-
oped to assemble the transcriptome from RNA-Seq data with-
out the need for a reference sequence. Alternatives are, e.g.,
DNAStar (http://www.dnastar.com/), Trans-ABySS (http://
www.bcgsc.ca/platform/bioinfo/software/trans-abyss ),
SOAPdenovo-Trans (http://soap.genomics.org.cn/
SOAPdenovo-Trans.html), Newbler (http://my454.com/
products/analysis-software/index.asp), iAssembler (http://
bioinfo.bti.cornell.edu/tool/iAssembler/), and Oases
(https://www.ebi.ac.uk/~zerbino/oases/).
4. When the read length of the sequencing machine is longer
than the molecule that is sequenced removing adapter
sequences from reads becomes necessary. For data from
230 Toni Luge and Sascha Sauer

Illumina platforms these pre-processing steps can be performed

with, e.g., Trimmomatic (http://www.usadellab.org/cms/
index.php?page=trimmomatic) that is implemented as addi-
tional option in Trinity. A platform-independent solution is
cutadapt (https://cutadapt.readthedocs.org/en/stable/).
5. Statistic parameters of the Trinity assembly can be examined
using the “TrinityStats.pl” script to be found in the
“utilities” folder in the Trinity installation directory. The
number of transcripts, components, and the transcript contig
N50 value will be reported. The contig N50 is a weighted
median statistic such that 50 % of the entire assembly is con-
tained in contigs or scaffolds equal to or larger than this value.
Thus this value should be expected to be near to the average
transcript length and can be used to confirm success of assem-
bly whenever information about transcript length from refer-
ence sequences or assemblies are available.
6. MaxQuant currently enables the analysis of high-resolution
MS data from, e.g., Thermo Scientific Orbitrap, and from
Bruker’s maXis qTOF and FT-ICR instruments only. Note
that similar alternative data analysis tools, covering detection
and quantification of peaks, isotope clusters and isotope-
labeled peptide pairs, as well as MS/MS ion search, validation,
and scoring of peptide identifications and their assembly to
protein identifications including determination of protein
ratios, can in principle be applied as well. For instance when
using the Mascot search engine as implemented in the
Proteome Discoverer software (Thermo Scientific) make sure
that you have a local Mascot server installed to configure your
RNA-Seq-based protein sequence database as search space.
7. Sequence data may be annotated with other tools as well.
Mercator together with MapMan (http://mapman.gabipd.
org/web/guest), GOanna as part of the AgBase (http://
agbase.msstate.edu/cgi-bin/tools/GOanna.cgi), the KEGG
Automatic Annotation Server (KAAS, http://www.genome.
jp/kegg/kaas/), the KEGG Orthology Based Annotation
System (KOBAS, http://kobas.cbi.pku.edu.cn/home.do), and
the Trinotate pipeline for transcriptome functional annotation
and analysis (http://trinotate.github.io) represent valuable
replacement options for Blast2GO.

Acknowledgement

Our work was supported by the German Ministry for Education

and Research (BMBF, grant number 0315082, 01EA1303 to
S.S.), the European Union (FP7/2007-2013), under grant agree-
ment n° 262055 (ESGI), and the Max Planck Society. This work is
part of the Ph.D. thesis of T.L.
 Sample Specific Databases for Proteomics 231

References
1. Ong S-E, Blagoev B, Kratchmarova I et al variate hypergeometric analysis. J Proteome
(2002) Stable isotope labeling by amino acids Res 6:654–661
in cell culture, SILAC, as a simple and accurate 15. Shilov IV, Seymour SL, Patel AA et al (2007)
approach to expression proteomics. Mol Cell The Paragon Algorithm, a next generation
Proteomics 1:376–386 search engine that uses sequence temperature
2. Hsu J, Huang S, Chow N et al (2003) Stable- values and feature probabilities to identify pep-
isotope dimethyl labeling for quantitative pro- tides from tandem mass spectra. Mol Cell
teomics. Anal Chem 75:6843–6852 Proteomics 6:1638–1655
3. Wiese S, Reidegeld KA, Meyer HE et al (2007) 16. Cox J, Neuhauser N, Michalski A et al (2011)
Protein labeling by iTRAQ: a new tool for Andromeda: a peptide search engine integrated
quantitative mass spectrometry in proteome into the MaxQuant environment. J Proteome
research. Proteomics 7:340–350 Res 10:1794–1805
4. Gerber SA, Rush J, Stemman O et al (2003) 17. Nesvizhskii AI (2010) A survey of computa-
Absolute quantification of proteins and phos- tional methods and error rate estimation proce-
phoproteins from cell lysates by tandem dures for peptide and protein identification in
MS. Proc Natl Acad Sci U S A 100: shotgun proteomics. J Proteomics
6940–6945 73:2092–2123
5. Schwanhäusser B, Busse D, Li N et al (2011) 18. Wang X, Slebos RJC, Wang D et al (2012)
Global quantification of mammalian gene Protein identification using customized protein
expression control. Nature 473:337–342 sequence databases derived from RNA-Seq
6. Mann M, Kulak NA, Nagaraj N et al (2013) data. J Proteome Res 11:1009–1017
The coming age of complete, accurate, and 19. Metzker ML (2010) Sequencing technolo-
ubiquitous proteomes. Mol Cell 49:583–590 gies—the next generation. Nat Rev Genet
7. Gstaiger M, Aebersold R (2009) Applying 11:31–46
mass spectrometry-based proteomics to genet- 20. Evans VC, Barker G, Heesom KJ et al (2012)
ics, genomics and network biology. Nat Rev De novo derivation of proteomes from tran-
Genet 10:617–627 scriptomes for transcript and protein identifica-
8. Freiwald A, Weidner C, Witzke A et al (2013) tion. Nat Methods 9:1207–1211
Comprehensive proteomic data sets for study- 21. Lopez-Casado G, Covey PA, Bedinger PA et al
ing adipocyte-macrophage cell-cell communi- (2012) Enabling proteomic studies with RNA-
cation. Proteomics 13:3424–3428 Seq: the proteome of tomato pollen as a test
9. Meierhofer D, Weidner C, Hartmann L et al case. Proteomics 12:761–774
(2013) Protein sets define disease states and 22. Sheynkman GM, Shortreed MR, Frey BL et al
predict in vivo effects of drug treatment. Mol (2013) Discovery and mass spectrometric
Cell Proteomics 12:1965–1979 analysis of novel splice-junction peptides using
10. Mann M, Hendrickson RC, Pandey A (2001) RNA-Seq. Mol Cell Proteomics 12:2341–2353
Analysis of proteins and proteomes by mass 23. Adamidi C, Wang Y, Gruen D et al (2011) De
spectrometry. Annu Rev Biochem 70: novo assembly and validation of planaria tran-
437–473 scriptome by massive parallel sequencing and
11. Perkins DN, Pappin DJC, Creasy DM et al shotgun proteomics. Genome Res
(1999) Probability-based protein identification 21:1193–1200
by searching sequence databases using mass 24. He R, Kim M-J, Nelson W et al (2012) Next-
spectrometry data. Electrophoresis generation sequencing-based transcriptomic
20:3551–3567 and proteomic analysis of the common reed,
12. Eng JK, McCormack AL, Yates JR (1994) An Phragmites australis (Poaceae), reveals genes
approach to correlate tandem mass spectral involved in invasiveness and rhizome specificity.
data of peptides with amino acid sequences in a Am J Bot 99:232–247
protein database. J Am Soc Mass Spectrom 25. Song J, Sun R, Li D et al (2012) An improve-
5:976–989 ment of shotgun proteomics analysis by adding
13. Craig R, Beavis RC (2004) TANDEM: match- next-generation sequencing transcriptome data
ing proteins with tandem mass spectra. in orange. PLoS One 7, e39494
Bioinformatics 20:1466–1467 26. Romero-Rodríguez MC, Pascual J, Valledor L
14. Tabb DL, Fernando CG, Chambers MC et al (2014) Improving the quality of protein
(2007) MyriMatch: highly accurate tandem identification in non-model species.
mass spectral peptide identification by multi- Characterization of Quercus ilex seed and
232 Toni Luge and Sascha Sauer

Pinus radiata needle proteomes by using tion, visualization and analysis in functional
SEQUEST and custom databases. J Proteomics genomics research. Bioinformatics 21:
105:85–91 3674–3676
27. Wu X, Xu L, Gu W et al (2014) Iterative 34. Gillet LC, Navarro P, Tate S et al (2012)
genome correction largely improves proteomic Targeted data extraction of the MS/MS spectra
analysis of nonmodel organisms. J Proteome generated by data-independent acquisition: a
Res 13:2724–2734 new concept for consistent and accurate pro-
28. Woo S, Cha SW, Merrihew G et al (2014) teome analysis. Mol Cell Proteomics
Proteogenomic database construction driven 11:O111.016717
from large scale RNA-seq data. J Proteome Res 35. Egertson JD, Kuehn A, Merrihew GE et al
13:21–28 (2013) Multiplexed MS/MS for improved
29. Armengaud J, Trapp J, Pible O et al (2014) data-independent acquisition. Nat Methods
Non-model organisms, a species endangered 10:744–746
by proteogenomics. J Proteomics 105:5–18 36. Haas BJ, Papanicolaou A, Yassour M et al
30. Wang X, Zhang B (2014) Integrating genomic, (2013) De novo transcript sequence recon-
transcriptomic, and interactome data to improve struction from RNA-seq using the Trinity plat-
peptide and protein identification in shotgun form for reference generation and analysis. Nat
proteomics. J Proteome Res 13:2715–2723 Protoc 8:1494–1512
31. Wang X, Zhang B (2013) customProDB: an R 37. Rice P (2000) The European Molecular
package to generate customized protein data- Biology Open Software Suite EMBOSS: the
bases from RNA-Seq data for proteomics European Molecular Biology Open Software
search. Bioinformatics 29:3235–3237 Suite. Trends Genet 16:2–3
32. Luge T, Kube M, Freiwald A et al (2014) 38. R Development Core Team R (2011) R: a lan-
Transcriptomics assisted proteomic analysis of guage and environment for statistical comput-
Nicotiana occidentalis infected by “Candidatus ing. R Found Stat Comput 1:409
Phytoplasma mali” strain AT. Proteomics 39. Cox J, Matic I, Hilger M et al (2009) A practi-
14:1882–1889 cal guide to the MaxQuant computational plat-
33. Conesa A, Götz S, García-Gómez JM et al form for SILAC-based quantitative proteomics.
(2005) Blast2GO: a universal tool for annota- Nat Protoc 4:698–705
 Chapter 17

A Proteomic Workflow Using High-Throughput De Novo

Sequencing Towards Complementation of Genome
Information for Improved Comparative Crop Science
Reinhard Turetschek, David Lyon, Getinet Desalegn, Hans-Peter Kaul,
and Stefanie Wienkoop

Abstract
The proteomic study of non-model organisms, such as many crop plants, is challenging due to the lack
of comprehensive genome information. Changing environmental conditions require the study and selec-
tion of adapted cultivars. Mutations, inherent to cultivars, hamper protein identification and thus consid-
erably complicate the qualitative and quantitative comparison in large-scale systems biology approaches.
With this workflow, cultivar-specific mutations are detected from high-throughput comparative MS anal-
yses, by extracting sequence polymorphisms with de novo sequencing. Stringent criteria are suggested to
filter for confidential mutations. Subsequently, these polymorphisms complement the initially used data-
base, which is ready to use with any preferred database search algorithm. In our example, we thereby
identified 26 specific mutations in two cultivars of Pisum sativum and achieved an increased number
(17 %) of peptide spectrum matches.

Key words Proteomics, De novo sequencing, Polymorphism, Crop science, Cultivars, Mass spec-
trometry, Pisum sativum

1 Introduction

In recent decades entire genome sequences of many organisms

were acquired and the amount of sequence information is continu-
ously expanding at an increasing rate. Advanced functional annota-
tion of genomic data in model organisms facilitates interpretation
of newly generated data. Despite the fact that specific sequence
information is unavailable for non-model organisms, a growing
number and a broad range of phylogenetic diverse species, reach-
ing from snake venoms [1–6] to whole microbial communities
[7–11], are being subjected to proteomic studies. A great evolu-
tionary distance to well-characterized species considerably
complicates the compilation of comprehensive databases (DBs),

Jörg Reinders (ed.), Proteomics in Systems Biology: Methods and Protocols, Methods in Molecular Biology, vol. 1394,
DOI 10.1007/978-1-4939-3341-9_17, © Springer Science+Business Media New York 2016

233
234 Reinhard Turetschek et al.

which is crucial for every bottom-up proteomic approach.

Prevalently this hurdle can be overcome by combining a relatively
large unspecific database (e.g., viridiplantae, NCBI) with a cus-
tom-built specific database consisting of translated RNA-sequences
(e.g., 6-frame translation of nucleotide to amino acid sequences,
BLAST homology searches for functional annotation) [12]. In
most of the cases, such a composite database is sufficient to gain
new insights into the protein level. However, conclusions are often
hampered when it comes to comparison of cultivars within the
same species. Among cultivar-specific sequence polymorphisms
match to a greater or lesser extent with the aforementioned data-
base and consequently result in distinct identifications, not related
to functional differences. Thus, differentiation of cultivars in the
proteomic domain may be composed of both sequence variation
and molecular processes. However, the identification of molecular
adaptations of cultivars upon environmental constraints is a major
focus of crop science. With the use of shotgun proteomics, com-
parative crop science not only aims to identify homologues but
moreover quantifies differences among cultivars, thus supporting
the development of breeding strategies [13]. Yet, most common
database search algorithms (e.g., SEQUEST) require a good match
with in silico-generated spectra (PMF and fragment ion series) and
fail to identify polymorphisms derived from cultivar-specific
sequences. Hence, the amino acid sequence is required for detailed
DB comparison.
The sequence may be acquired de novo, by deriving the amino
acid composition from fragment ions of peptides.
The idea of determining peptide primary structure via mass
spectrometry without prior knowledge of the sequence was already
developed in the 1970s by studying penicillinase [14]. In the
1980s, first tandem MS scans were manually sequenced with 2200
mass resolution [15]. Today, various automated de novo sequenc-
ing algorithms are available enabling high-throughput processing
of MS/MS data [16–19]. Still, the reliability scoring, inherent to
all de novo algorithms, remains an ongoing issue which highly
influences accuracy and computation time [20]. Once confident
sequence tags are obtained, these can be matched to a database
with the help of various search engines [21–24] to retrieve homo-
logue proteins.
After assigning homologues, comparing de novo tags with an
adequate database is just one more step to extract sequence differ-
ences in order to determine mutations. This additional step, how-
ever, requires special care as de novo sequencing is prone to specific
errors (e.g., the inability to distinguish between K and E in low
mass accuracy measurement). Such errors are taken into account
by a few programs, such as SPIDER [21], TagRecon [22], and
OpenSea [23], that correct de novo tags and additionally allow
inexact matches to DB sequences. By matching de novo tags
 A Proteomic Workflow Using High-Throughput De Novo Sequencing Towards… 235

3.5 PTM &

3.3 DB compilation 3.4 De novo sequencing
Homology search

3.7 Application of 3.6 DB extension 3.5 Mutation

extended DB with mutations assessment

Fig. 1 Workflow with main steps from initial DB compilation and de novo
sequence analysis to the application of the new organism aligned DB, explained
in detail as follows (cf. Subheadings 3.2–3.6)

inexactly to the DB these algorithms show the capacity to more or

less accurately—depending on the spectra quality and the search
algorithm—identify posttranslational modifications, homologies,
and mutations. With the use of such an automated search for
homologies and mutations, our workflow (Fig. 1) aims to amend
the initial DB with newly identified sequences (see Subheading 3.4).
In case of cultivars, mutated sequences are not replacing original
entries in the database, but are added with a header corresponding
to the cultivar. Identified homologies to DB entries from different
organisms are as well amending the new DB, but additionally the
name of the organism must remain in the header not to confuse a
homology with a mutation. Using this (extended) DB with any
conventional algorithm (e.g., SEQUEST, Mascot) facilitates high-
throughput MS/MS data analysis (compared to de novo sequenc-
ing) and additionally increases the confidence- and probability-based
peptide identification (e.g., Xcorr) as well as protein sequence cov-
erage, which results in more accurate quantification of cultivar-
specific proteins (see Subheading 3.6). However, the determination
of mutations via de novo sequencing is delicate and requires a few
criteria. Therefore, particular attention has to be paid to reliable
identification of mutations by critically taking mismatches with
PTMs into account (see Subheading 3.2) and setting further crite-
ria for stringent consideration of mutations (see Subheading 3.3).

2 Materials

2.1 Plant Material 1. Seeds from P. sativum ssp. cultivar Messire were provided by
the Institute for Sustainable Agriculture CISC (Department of
Plant Breeding, Cordoba, Spain). Cultivar Protecta was
obtained from Probstdorfer Saatzucht GmbH & Co KG
(Probstdorf, Austria).
236 Reinhard Turetschek et al.

2.2 Protein 1. Lyophilized plant material.

Extraction: Materials 2. TRIzol Reagent® RNA Isolation Reagent.
3. Precipitation solution: 0.5 % β-Mercaptoethanol in acetone.
4. Protein digestion: Endoproteinase LysC Sequencing grade,
Poroszyme® Immobilized Trypsin Bulk Media.

2.3 LC-MS/MS 1. One-dimensional nano-flow LC (Dionex UltiMate 3000;

Instrumentation Thermo Scientific, USA).
2. EASY-Spray column, 15 cm × 75 μm ID, PepMap C18, 3 μm
(Thermo Scientific, USA).
3. LTQ Orbitrap Elite (Thermo Scientific, USA).

2.4 LC-MS/MS 1. Mobile-phase solvent A: 0.1 % Formic acid; solvent B: 90 %

Analysis: Materials acetonitrile, 0.1 % formic acid.

2.5 Software 1. mEMBOSS 6.5 (European Molecular Biology Open Software

Suite).
2. PEAKS 7.0 (Bioinformatics Solutions Incorporation, Canada).
3. SEQUEST (Proteome Discoverer 1.3; Thermo Scientific, USA).

3 Methods

3.1 Protein Leaves of 4-week-old plants were sampled, immediately quenched,

Extraction and ground in liquid N2. Protein from lyophilized material was
extracted in TRIzol® according to Lee et al. [25] with a few modifi-
cations: 3 ml of β-mercaptoethanol in acetone was used for precipi-
tation overnight at −20 °C. The protein pellet was washed and
digested with LysC and trypsin according to Staudinger et al. [26].

3.2 LC-MS/MS Peptide digests (1 μg) were applied to a one-dimensional nano-

Analysis flow LC. The peptides were separated using a 95-min nonlinear
gradient from 98 % of solvent A to 45 % of solvent B at a flow rate
of 300 nl/min. The nLC-ESI-MS/MS analysis was optimized for
standard high-throughput analysis at a resolution of 120,000
(FTMS) with 20 MS/MS scans in the LTQ at the following set-
tings: rapid scan mode, minimum signal threshold counts 1000,
prediction of ion injection time, repeat count 1, repeat duration
30 s, exclusion list size 500, exclusion duration 60 s, exclusion
mass width 5 ppm relative to reference mass, early expiration ena-
bled (count 1, S/N threshold 2), monoisotopic precursor selection
enabled, rejected charge state: 1, normalized collision energy: 35,
and activation time 30 ms.
 A Proteomic Workflow Using High-Throughput De Novo Sequencing Towards… 237

3.3 Custom A composite protein-fasta file was created by merging the following
Database Design six databases:
Uniprot UniRef100 (all identical sequences and subfragments
with 11 or more residues are placed into a single record—http://
www.uniprot.org/help/uniref) sourced at 15-05-2013 from the
following Taxa:
1. Pisum sativum
2. Rhizobium leguminosarum
3. Glomus
4. Mycosphaerella
5. Legume-specific protein database (LegProt) [27] including
information from the following organisms: Pisum sativum,
Lotus japonicus, Medicago sativa, Glycine max, Lupinus albus,
Phaseolus vulgaris.
6. Processed dbEST NCBI sourced from http://www.coolsea-
sonfoodlegume.org/
Pisum Sativum Unigene v1, P. Sativum Unigene wa1, Pisum
Sativum Unigene v2;
Nucleotide sequences were six-frame-translated using mEM-
BOSS. For each accession number the longest continuous amino
acid sequence (longest ORF) within a frame was chosen. If multi-
ple sequences (of different frames) were of the same maximum
length, all of them were kept (each with a different accession num-
ber, including the frame number).
The 6 fasta files described above were combined, producing a
new fasta containing 135,754 entries. Protein sequences 100 %
identical in sequence and length were combined by subsequently
adding one header after the other, separating them by the follow-
ing characters “ __***__ ” (no matter if the redundancies origi-
nated from one or multiple fasta files). All other entries were simply
added to the end of the new file. The first accession number of the
header was repeatedly written at the very beginning of the header
line, separated by a “ | ” in order to consistently view and parse the
accession numbers.

3.4 De Novo Several automated software solutions for de novo sequencing are
Sequencing available to date. The outcome very much depends on the quality
and Homology Search of processed spectra and the selected algorithm [28]. Higher spec-
tra quality can be achieved by adaptation of the fragmentation (see
Note 1). Here, the de novo search was performed with PEAKS
[18] employing settings according to the resolution and mass
accuracy of the mass spectrometer used (see Subheading 2.3). By
calculating the narrowest possible mass error tolerance most occur-
ring PTMs cannot be mistaken for a mutation. Hence, a mass error
238 Reinhard Turetschek et al.

of 5 ppm from the monoisotopic precursor was allowed with fragment

ion mass error of 0.5 Da (see Note 2). De novo tags with a mini-
mum average local confidence (ALC) of ≥15 were subjected to
PTM identification to determine the most frequent modifications
in order to recalculate an adequate mass error tolerance (see Note 2).
Accordingly, de novo tags were matched against the designed data-
base (see Subheading 3.1) and searched for mutations with the
implemented SPIDER tool. Maximum three of the previously
searched PTMs were allowed. The peptide spectrum matching
score (−10 lgP) was set to 20. A maximum of two missed cleavages
per peptide and nonspecific cleavage at one end of the peptide
were allowed—this apparently loose restriction facilitates the iden-
tification of mutated K or R residues.

3.5 Evaluation Various automated programs [21, 23, 24] crucially simplify the
of Homology identification of sequence variance. However, these identifications
and Mutation must be critically filtered to obtain only confident sequence amend-
ments to the original database. First, exclusively proteotypic pep-
tides are added to the original database, because a variation in any
other peptide cannot be specifically attributed to one protein.
However, as a result of using a merged DB containing sequence
information of several RNA data (see Subheading 3.1), variations
are sometimes assigned to multiple protein entries with the same
function but slightly different sequences. In such case, the protein
entry with the most assigned peptides and highest coverage is cho-
sen for further processing. If the number of assigned peptides and
coverage is similar for several DB entries, all of them are chosen for
further processing. Second, the mutated peptide must not have a
non-mutated counterpart: if a mutated and a non-mutated peptide
are attributed to the same sequence in the database, the identified
mutation is likely to be a false positive. Third, the sequence of the
mutation must be confirmed by at least two MS/MS spectra.
Thereby, a de novo error—identifying a mutation—caused by a
low-quality spectrum is largely avoided and mutations gain confi-
dence. A typical quantitative shotgun proteomics experiment
requires the measurement of several replicates, which usually
acquire enough MS/MS to confirm mutations.

3.6 Database Confidently identified mutated amino acid residues are added to
Extension the database by copying the original fasta entries with the mutated
with Mutations sequence. Original entries must be kept, because the located muta-
tions may be characteristic for just one cultivar. The mutated
sequences are found in new entries with modified accession and
header (see example below).
Accession: ACU20233.1_m1
Header: unknown [glycine max] [Me_IV25]
 A Proteomic Workflow Using High-Throughput De Novo Sequencing Towards… 239

The accession shows a suffix (“_m1”) indicating a sequence

alteration. If cultivars show different mutations at the same protein
entry, the number of the suffix is ascending. The header is supple-
mented with squared brackets including the information about the
cultivar and the polymorphisms in the sequence (in the current
example in cv. Messire isoleucine substitutes valine at position 25).
Here, an entry of the LegProt database (NCBI) [27] is shown
which is not yet annotated. After inserting the mutation, the
sequence can again be blasted to achieve improved annotation,
albeit a change in only one amino acid will result in a similar BLAST
result. Additionally, when working with non-model organisms, it is
worth considering re-annotation of the genome by use of pro-
teomic data in a proteogenomics approach [29].

3.7 Application The database, amended with mutated sequences, potentially

of Extended Database enhances the identification of any preferred DB search algorithm
and enables high-throughput processing of MS/MS data. By
increasing peptide scores (e.g., Xcorr) and protein sequence cov-
erage, proteins are more confidently identified. Moreover, inclu-
sion of exclusively proteotypic peptides (see Subheading 3.5)
expands the list of candidates for other proteomic approaches
(e.g., SRM, MRM).

3.8 Iterated Search The new extended DB was used for a standard DB search using the
with DB Search SEQUEST algorithm with the following settings: 5 ppm precursor
Algorithm mass tolerance, 0.5 Da fragment mass tolerance, acetylation of the
N-terminus, and oxidation of methionine as dynamic modifica-
tions. Minimum peptide confidence was set to medium, and mini-
mum Xcorr to 2. A minimum of two peptides per protein were
required for identification.
In the present study of P. sativum with the cultivars Messire
and Protecta we identified 48 variations to original DB entries, of
which 26 are mutations showing high cultivar specificity. Both cul-
tivars have five mutations in common. Messire showed 12 and
Protecta 9 characteristic mutations. Furthermore, 22 homologues
were identified from entries of different species (e.g., G. max from
the LegProt DB). The ratio of replaced and substituting amino
acids (Fig. 2) shows that most frequently valine and alanine are
both replacing and substituting other amino acids in our
experiment.
The number of peptide spectral matches (PSMs) shows how
many of fragmented ions match to the applied DB. Thus, a rather
complete DB will result in a higher number of PSMs compared to
an imperfect DB. Here Fig. 3 shows that the number of PSMs
increased significantly (17 %) for the two studied cultivars after
amending the initial DB with sequence variations. Besides improv-
ing protein identification, the elevated number of PSMs crucially
contributes to more accurate and confident quantification.
240 Reinhard Turetschek et al.

Fig. 2 Ratio of replaced (from the initial DB) and substituting amino acids

PSMs after Database Extension

5000 *
*
4000

3000
# PSMs

2000

1000

0
Messire Protecta
original DB extended DB

Fig. 3 The total number of PSMs is significantly increased (student’s t-test, p < 0.05) in both cultivars when the
extended DB is applied; n = 24, error bars indicate standard error at 95 % confidence intervals

4 Notes

1. Optimizing prerequisites for de novo sequencing

De novo sequencing essentially depends on a complete series
of fragment ions which are generated according to the applied
fragmentation technique. Thus, the quality of de novo
sequencing can be enhanced crucially by applying different
fragmentations (CID/HCD/ETD) to the same precursor
and subsequently merging the spectra in order to achieve a
great number of fragment ions [30]. The choice of instru-
mentation often points to high speed and sensitivity (e.g.,
LTQ using CID) with the drawbacks of reduced mass resolu-
tion and accuracy, resulting in an impossibility to resolve
A Proteomic Workflow Using High-Throughput De Novo Sequencing Towards… 241

fragment ions’ charge state. This deficit causes considerable

difficulties to de novo sequencing because of additional ambi-
guity. Through performance of MS/MS scans at high resolu-
tion (i.e., FT) the precision of de novo sequencing benefits
especially for precursors of higher charge states (≥3+).
However, high resolution requires higher AGC values, which
increases acquisition time remarkably. The trade-off between
spectra-quality (FT) and speed (LTQ) can be accounted for
by a data-dependent decision tree [31], where the site of the
fragment scan depends on the charge state and the m/z of the
peptide. Consequently MS/MS scans from ≥3+ charged pre-
cursors crucially improve in quality.
2. Definition of mass error tolerance for accurate de novo
sequencing
The calculation of the mass error tolerance comprises the
comparison of an AA mass with a PTM to another AA. This
gap is the mass difference which needs to be resolved to dis-
tinguish a PTM from a mutation. Since a database search
with more than 200 possible PTMs [32] requires infeasible
computing capacity, the calculation refers to the seven most
observed PTMs in this experiment. The most frequently
occurring PTMs were determined by peaks with 5 ppm pre-
cursor mass error tolerance and 0.6 Da fragment mass error
tolerance: oxidation of methionine (+15.99), sodium adduct
(+21.98, on D-, E-, and C-term), carbamylation (+43.01 on
N-term), methyl ester (+14.02 on D-, E-, and C-term),
deamidation (+0.98 on N and Q), acetylation (+42.01 on
N-term), and replacement of two protons by calcium (+37.95
on D-, E-, and C-term). The mass error tolerance can be cal-
culated as follows: The mass of a possible modification is
added to the AA’s mass. When modifications affect the C- or
N-term, the PTM’s mass is added to each AA’s mass. These
values are subtracted from the masses of each amino acid.
Thus, a mass error window to differentiate between an AA
and a modified AA is calculated. For MS/MS scans in an ion
trap a mass accuracy of 100–200 ppm must be additionally
subtracted from this mass difference. Accordingly, a bulk of
false positives (modified peptides identified as mutations) can
already be excluded by setting the mass error tolerance to
0.5 Da in an MS/MS scan, that is, e.g., the identification of
valine as asparagine which mass differs in 0.94 Da from a pep-
tide with methyl ester at the C-terminus. Considering a mass
accuracy of 200 ppm in an ion trap (0.4 Da at 2000 m/z) the
mass difference narrows to 0.54 Da. For an FT MS/MS scan
the mass error tolerance is preferably set a lot lower (e.g.,
0.05 Da). Additionally, setting the precursor mass tolerance
to ≤5 ppm critically minimizes false positives.
242 Reinhard Turetschek et al.

References
1. Fox JW, Ma L, Nelson K et al (2006) 12. Romero-Rodríguez MC, Pascual J, Valledor L
Comparison of indirect and direct approaches et al (2014) Improving the quality of protein
using ion-trap and Fourier transform ion identification in non-model species.
cyclotron resonance mass spectrometry for Characterization of Quercus ilex seed and
exploring viperid venom proteomes. Toxicon Pinus radiata needle proteomes by using
47(6):700–714 SEQUEST and custom databases. J Proteomics
2. Mackessy SP (2002) Biochemistry and phar- 105:85–91
macology of colubrid snake venoms. J Toxicol 13. Vanderschuren H, Lentz E, Zainuddin I et al
Toxin Rev 21(1–2):43–83 (2013) Proteomics of model and crop plant
3. Nawarak J, Sinchaikul S, Wu CY et al (2003) species: status, current limitations and strategic
Proteomics of snake venoms from Elapidae advances for crop improvement. J Proteomics
and Viperidae families by multidimensional 93:5–19
chromatographic methods. Electrophoresis 14. Morris HR, Williams DH, Ambler RP (1971)
24(16):2838–2854 Determination of the sequences of protein-
4. OmPraba G, Chapeaurouge A, Doley R et al derived peptides and peptide mixtures by mass
(2010) Identification of a novel family of snake spectrometry. Biochem J 125(1):189–201
venom proteins Veficolins from Cerberus ryn- 15. Johnson RS, Biemann K (1987) The primary
chops using a venom gland transcriptomics structure of thioredoxin from Chromatium
and proteomics approach. J Proteome Res vinosum determined by high-performance
9(4):1882–1893 tandem mass spectrometry. Biochemistry
5. Serrano SMT, Shannon JD, Wang D et al 26(5):1209–1214
(2005) A multifaceted analysis of viperid snake 16. Frank A, Pevzner P (2005) PepNovo: de novo
venoms by two-dimensional gel electrophore- peptide sequencing via probabilistic network
sis: an approach to understanding venom pro- modeling. Anal Chem 77(4):964–973
teomics. Proteomics 5(2):501–510 17. Fischer B, Roth V, Roos F et al (2005)
6. Wong ESW, Morgenstern D, Mofiz E et al NovoHMM: a hidden Markov model for de
(2012) Proteomics and deep sequencing com- novo peptide sequencing. Anal Chem
parison of seasonally active venom glands in 77(22):7265–7273
the platypus reveals novel venom peptides and 18. Ma B, Zhang K, Hendrie C et al (2003)
distinct expression profiles. Mol Cell PEAKS: powerful software for peptide de novo
Proteomics 11(11):1354–1364 sequencing by tandem mass spectrometry.
7. Abraham PE, Giannone RJ, Xiong W et al Rapid Commun Mass Spectrom 17(20):
(2014) Metaproteomics: extracting and min- 2337–2342
ing proteome information to characterize 19. Taylor JA, Johnson RS (1997) Sequence data-
metabolic activities in microbial communities. base searches via de novo peptide sequencing
Curr Protoc Bioinformatics 46:13.26. by tandem mass spectrometry. Rapid Commun
1–13.26.14 Mass Spectrom 11(9):1067–1075
8. Becher D, Bernhardt J, Fuchs S et al (2013) 20. Ma B, Johnson R (2012) De novo sequencing
Metaproteomics to unravel major microbial and homology searching. Mol Cell Proteomics
players in leaf litter and soil environments: 11(2):O111.014902
challenges and perspectives. Proteomics 21. Han Y, Ma B, Zhang K (2005) SPIDER: soft-
13(18–19):2895–2909 ware for protein identification from sequence
9. Hao C, Liu Q, Yang J et al (2008) tags with de novo sequencing error.
Metaproteomics: exploration of the functions J Bioinform Comput Biol 3(3):697–716
of microbial ecosystems. Chin J Appl Environ 22. Dasari S, Chambers MC, Slebos RJ et al
Biol 14(2):270–275 (2010) TagRecon: high-throughput mutation
10. Siggins A, Gunnigle E, Abram F (2012) identification through sequence tagging.
Exploring mixed microbial community func- J Proteome Res 9(4):1716–1726
tioning: recent advances in metaproteomics. 23. Searle BC, Dasari S, Turner M et al (2004)
FEMS Microbiol Ecol 80(2):265–280 High-throughput identification of proteins
11. Wang H-B, Zhang ZX, Li H et al (2011) and unanticipated sequence modifications
Characterization of metaproteomics in crop using a mass-based alignment algorithm for
rhizospheric soil. J Proteome Res 10(3): MS/MS de novo sequencing results. Anal
932–940 Chem 76(8):2220–2230
 A Proteomic Workflow Using High-Throughput De Novo Sequencing Towards… 243

24. Tabb DL, Saraf A, Yates JR 3rd (2003) sequencing algorithms. J Proteome Res 5(11):
GutenTag: high-throughput sequence tagging 3018–3028
via an empirically derived fragmentation 29. Armengaud J, Trapp J, Pible O et al (2014)
model. Anal Chem 75(23):6415–6421 Non-model organisms, a species endangered
25. Lee FW, Lo SC (2008) The use of Trizol by proteogenomics. J Proteomics 105:5–18
reagent (phenol/guanidine isothiocyanate) for 30. Guthals A, Clauser KR, Frank AM et al
producing high quality two-dimensional gel (2013) Sequencing-grade de novo analysis of
electrophoretograms (2-DE) of dinoflagel- MS/MS triplets (CID/HCD/ETD) from
lates. J Microbiol Methods 73(1):26–32 overlapping peptides. J Proteome Res
26. Staudinger C, Mehmeti V, Turetschek R et al 12(6):2846–2857
(2012) Possible role of nutritional priming for 31. Frese CK, Altelaar AF, Hennrich ML et al
early salt and drought stress responses in (2011) Improved peptide identification by tar-
Medicago truncatula. Front Plant Sci 3:285 geted fragmentation using CID, HCD and
27. Lei Z, Dai X, Watson BS et al (2011) A legume ETD on an LTQ-Orbitrap Velos. J Proteome
specific protein database (LegProt) improves Res 10(5):2377–2388
the number of identified peptides, confidence 32. Gooley AA, Packer NH (1997) The impor-
scores and overall protein identification success tance of protein co- and post-translational
rates for legume proteomics. Phytochemistry modifications in proteome projects. In: Wilkins
72(10):1020–1027 MR et al (eds) Proteome research: new fron-
28. Pevtsov S, Fedulova I, Mirzaei H et al (2006) tiers in functional genomics. Springer, Berlin,
Performance evaluation of existing de novo pp 65–91
 Chapter 18

From Phosphoproteome to Modeling of Plant Signaling

Pathways
Maksim Zakhartsev, Heidi Pertl-Obermeyer, and Waltraud X. Schulze

Abstract
Quantitative proteomic experiments in recent years became almost routine in many aspects of biology.
Particularly the quantification of peptides and corresponding phosphorylated counterparts from a single experi-
ment is highly important for understanding of dynamics of signaling pathways. We developed an analytical
method to quantify phosphopeptides (pP) in relation to the quantity of the corresponding non-­phosphorylated
parent peptides (P). We used mixed-mode solid-phase extraction to purify total peptides from tryptic digest and
separated them from most of the phosphorous-containing compounds (e.g., phospholipids, nucleotides) which
enhances pP enrichment on TiO2 beads. Phosphoproteomic data derived with this designed method allows
quantifying pP/P stoichiometry, and qualifying experimental data for mathematical modeling.

Key words Phosphopeptide enrichment, Mixed-mode solid-phase extraction, Metal oxide affinity
chromatography, Mathematical modeling

1 Introduction

Mathematical modelingand dynamic simulation of signal transduc-

tion pathways is an important topic in systems biology [1, 2]. One
of the purposes of the dynamic modeling in plant physiology is to
evaluate the degree of involvement of different signaling pathways
in plant responses to external perturbations [3, 4] or to explain
phenotypic appearances of plant mutants. Protein phosphorylation
is one of the fastest posttranslational modifications (PTM) that is
an intrinsic mechanism of the signal transduction in some signaling
pathways (e.g., MAPK cascades). Phosphorylation of signaling
proteins traced in time allows revealing involvement of corre-
sponding pathways into adaptive responses [5, 6]. Signaling path-
ways are organized in cascades of counteracting (e.g., cyclic)
irreversible reactions [7], which generate and amplify the cellular
signal (Fig. 1). Phosphorylation (by kinases)/de-phosphorylation
(by phosphatases) of substrate-proteins is an elemental event in
many signal transduction pathways [1, 3, 7]. Normally, proteins in

Jörg Reinders (ed.), Proteomics in Systems Biology: Methods and Protocols, Methods in Molecular Biology, vol. 1394,
DOI 10.1007/978-1-4939-3341-9_18, © Springer Science+Business Media New York 2016

245
246 Maksim Zakhartsev et al.

(S)ignal
ATP ADP

receptor R

R′ level 1 R P

PP1

ATP ADP

Conserved moiety
Kinase K1
Kipool = Ki + K i′
K′2 level 2 K2 P
Ki K i′
+ =1
PP2 Ki pool Kipool

P
...

Kn−1
ATP ADP

target T

T′ level n T P functional readout

PPn

Fig. 1 A simplified hypothetical scheme of a signal (S) transfer from receptor (R) to
target (T) through a linear signal transduction pathway consisting of different
kinases (K). At each cascade level, only two interconvertable forms are shown:
active and inactive (marked by ′). The mechanism of activation is through phos-
phorylation by means of group transfer from phosphate-donor (e.g., ATP, GTP). The
active form of a species from a preceding level activates species at a subsequent
level. PPi—ith phosphatase. The figure was compiled from [1, 7]

cyclic reactions function under assumption of a conserved moiety

(Fig. 1): a total pool of a protein remains unchanged in short time,
but its degree of phosphorylation shows a fast response to the envi-
ronmental stimuli. Therefore, it is crucial to measure the total pool
of the signaling proteins as well as the degree of their phosphoryla-
tion, which, when measured in a time course, can provide one with
sufficient information for parameter estimation of mathematical
models based on kinetic expressions of individual reactions.

1.1 MS-Based Shotgun mass spectrometry (MS)-based proteomics offers a unique

Proteomics opportunity for identification and quantification of thousands of
peptides (P) in a single analysis, and in combination with
 From Phosphoproteome to Modeling of Plant Signaling Pathways 247

computational methods provides information on expression of

complete proteome of a research object at a given state [8, 9].
The shotgun MS-based proteomics also allows identifying most
PTMs of proteins, such as phosphorylation, acetylation, methyla-
tion, ­glycosylation, and ubiquitination [10, 11], which drive sig-
naling pathways, and their dynamics can be used for decoding
signaling networks. The common workflow of sample preparation
for MS-based proteomics relies on different analytical techniques:
enzymatic protein digest, sample fractionation based on various
chromatographic techniques, sample enrichment or desalting by
solid-phase extraction (SPE), etc. [12]. Since the fraction of phos-
phopeptides is relatively low (<1 %) regarding to total peptides in
trypsinized protein digests, usually an enrichment step is required
for their confident identification and quantification [12–14].

1.2 Total Peptide The classical approach to purify total peptides from complex digest
Purification mixes exploits hydrophobic interactions between side chains of
hydrophobic amino acids and C18-groups of reverse-phase SPE sor-
bent [15]. However, this approach results in co-purification of a
variety of nonpolar components, such as lipids, pigments, or sterols.
This often is particularly problematic when working with strongly
pigmented plant tissues. At low pH (e.g., <3.0) the zwitterion of a
peptide becomes fully protonated (i.e., a weak cation or weak base)
which allows the use of strong cation exchanger (SCX) for their
purification, but it also results in co-purification of charged impuri-
ties such as nucleotides. Thus, a combination of the reverse-­phase
and SCX modes provides enhanced peptide purification and addi-
tionally a better removal of nonpolar and charged impurities. Oasis®
MCX (mixed-mode cation-exchange and reversed-phase sorbent)
provides such dual modes of retention by (1) strong sulfonic-­acid-
cation exchanger and (2) reversed-phase interactions in combination
with hydrophilic interactions on a single organic co-polymer. The
sorbent is highly selective and sensitive for extraction of basic com-
pounds from acidified biological matrices (such as plasma, urine,
bile, and pigmented plant tissue) and demonstrates very good capac-
ity for the peptide purification from tryptic protein digests. Sample
preparation based on use of mixed-mode solid phases provides supe-
rior removal of nearly all phospholipids and weaker acids, achieving
double-positive effects by (1) eliminating phosphorous-containing
compounds (phospholipids, nucleic acids, etc.) that strongly inter-
fere with phosphopeptide enrichment and (2) eliminating the major
sources of matrix effects (e.g., neutrals), a known case of ion
suppression, loss of sensitivity, and inaccurate quantification by liquid
chromatography with a tandem mass analyzer (LC-MS/MS).

1.3 Phosphopeptide The typical pP enrichment protocols rely on several unit operations,
(pP) Enrichment such as total peptide purification/desalting after the digest, pP
enrichment, and further desalting, in order to provide salt-free
samples for the LC-MS/MS analysis. Usually, all desalting steps are
248 Maksim Zakhartsev et al.

carried out on C18 Stop-and-go-extraction tips (StageTips) [15].

Most common pP enrichment protocols are based on metal oxide
affinity chromatography (MOAC) on titanium, aluminum, and
zirconium oxides [5, 13, 16]. The MOAC of pP is based on the
affinity of phosphates to metals, but this interaction is interfered by
non-phosphorylated acidic peptides which also show affinity for
the metals [14]. Therefore, enhancers of the phosphopeptide
selectivity on MOAC are widely used (e.g., lactic or
β-hydroxypropanoic acids [16], 2,5-dihydroxybenzoic or phthalic
acids [17]). The enhancers reduce unspecific binding of non-phos-
phorylated acidic peptides, in this sense giving preferences for pP
to bind to the metal oxides. Obviously, the second desalting step
after the use of enhancers is extremely important, but it leads to an
additional loss of phosphopeptides.
In some research, the phosphoprotein enrichment [18, 19] is
used prior to the digest followed by the phosphopeptide enrichment
procedure to increase yield of phosphopeptides. However, in the
context of our approach it is meaningless since non-­phosphorylated
cognates will be removed.
Other phosphorous-containing compounds (phospholipids,
nucleotides, etc.) also compete with pP for the metal centers to bind.
Therefore, we have chosen a strategy of eliminating the phospho-
rous-containing compounds on the level of total peptide purifica-
tion/desalting (mixed-mode SPE) to give the advantages for the pP
at the enrichment step on conventional TiO2 without enhancers.
Our method describes pP enrichment from a microsomal fraction
[20] of roots of wild-type Arabidopsis thaliana (Fig. 2).

1.4 Peptide (P) There are different methods of peptide quantification in quantita-
and Phosphopeptide tive proteomics, which implement either absolute or relative pep-
(pP) Quantification tide quantifications. The quantification techniques are based on
the use of stable isotope labeling (e.g., SILAC), isobaric labeling
(e.g., TMT, iTRAQ), isotope-coded affinity tag (ICAT), internal
peptide standards [21], and label-free quantification (e.g., iBAQ).
The workflow we developed is based on label-free relative peptide
quantification (iBAQ by MaxQuant [22]) before and after pP
enrichment, using the amount (literally a mass in μg) of total pep-
tides and its loss during pP enrichment for the iBAQ value normal-
ization (Fig. 2). This approach allows us to estimate the degree of
phosphorylation of the signaling protein pool (see Note 1).
However, peptides and their phosphorylated counterparts have
significant differences in ionization/detection efficiency (so-­called
flyability) [23]. Determination of flyability ratio for a particular
pP/P pair (based on the technical replicates and conserved moiety
assumption) allows to correct the signal intensities of the corre-
sponding species and to quantify the absolute phosphorylation stoi-
chiometry in each obtained pP/P pair [23]. This aspect also has to
be taken into account for the final quantification.
 From Phosphoproteome to Modeling of Plant Signaling Pathways 249

Sample
(homogenate, CF, MF)

digest

mixed-mode SPE
on Oasis MCX

[tP]1
aliquot 1 aliquot 2

TiO2 -enrichment

[tP]2

LC-MS/MS LC-MS/MS

iBAQ1-j iBAQ2-j

peptide : phosphopeptide matching

Fig. 2 Workflow of label-free phosphopeptide/peptide ratio quantification based

on TiO2 enrichment of phosphopeptides. Refer to Note 1 for the calculation algo-
rithm and notations. CF cytosolic fraction, MF microsomal fraction, SPE solid-­
phase extraction, tP total peptide concentration, iBAQ intensity-based absolute
quantification is used for label-free quantitation (for further details please refer
to MaxQuant manual instructions)

2 Materials

Prepare all solutions using double-deionized ultrapure water

(0.055 μS/cm; see Note 2) and analytical grade reagents. Use
pipette tips and microcentrifuge tubes made from low-binding-­
capacity plastics to minimize peptide loss by adsorption to the plas-
tic. Take maximum care to avoid keratin contamination. Prepare
and store all reagents at room temperature (unless indicated
otherwise).
250 Maksim Zakhartsev et al.

2.1 Protein 1. Protein quantification reagents according to manual instruc-

Quantification tion for use of NanoOrange® protein quantification kit
(Molecular Probe).
2. Bovine serum albumin (BSA) standard solutions in water, 1
mL of 0, 0.1, 0.3, 0.6, 1, 3, 6, 10, 30, 60, 100, 300, 600,
1000, 1500, 2000 μg/mL each.
3. BSA standard solutions in UTU buffer, 1 mL of 0, 0.1, 0.3,
0.6, 1, 3, 6, 10, 30, 60, 100, 300, 600, 1000, 1500, 2000 μg/
mL each.
4. BSA standard solutions in loading buffer for LC-MS/MS, 1
mL of 0, 0.1, 0.3, 0.6, 1, 3, 6, 10, 30, 60, 100, 300, 600,
1000, 1500, 2000 μg/mL each.

2.2 Total Protein 1. Hammer and aluminum foil.

Extraction 2. Liquid nitrogen.
3. Miracloth® filter paper (Merk Millipore).
4. Potter® homogenizer (10 mL; VWR).
5. Homogenization buffer: 330 mM sucrose, 100 mM KCl, 1
mM EDTA, 50 mM Tris–HCl adjusted with MES to pH 7.5,
6.5 mM dithiothreitol (DTT; add freshly). Protease inhibitor
cocktail (Sigma), phosphatase inhibitor cocktail 2 (Sigma), and
phosphatase inhibitor cocktail 3 (Sigma) were added from
stock solutions in 50 μL each/10 mL of the ice-cold homog-
enization buffer just before use.
6. UTU-buffer: 6 M Urea, 2 M thiourea, 50 mM Tris–HCl, pH
8.0.
7. Ultrasound bath (e.g., Ultrasonic Cleaner USC 300TH,
VWR).
8. High-speed refrigerated benchtop centrifuge for max. speed
65,000 × g (e.g., Sigma 3-30KS).

2.3 In-Solution 1. Reduction buffer: 6.5 mM (or 1 μg/μL; w/v) DTT in water.
Digest 2. Alkylation buffer: 27 mM (or 5 μg/μL; w/v) iodoacetamide in
water.
3. 10 mM Tris–HCl, pH 8.0.
4. Lysyl endopeptidase (LysC) stock solution: 0.5 μg/μL
(Promega).
5. Trypsin stock solution: 0.5 μg/μL (Promega).
6. 2 % Trifluoroacetic acid (TFA) in water (v/v).

2.4 Peptide 1. Oasis® MCX 1 cc Vac Cartridge, 30 mg per Cartridge, 30 μm

Purification particle size (Waters).
2. 100 % Methanol.
 From Phosphoproteome to Modeling of Plant Signaling Pathways 251

3. Water.
4. 2 % Formic acid in water (v/v).
5. 5 % Ammonium hydroxide in 80 % methanol (pH 11.0) (v/v).
6. Thermo-Strips and Caps, 8 × 0.2 mL (ThermoScientific).

2.5 C8-TiO2-StageTip The 200 μL C8-StageTips are commercially available (e.g.,

Column Preparation ThermoScientific; or elsewhere) or can be in-house manufactured
according to [15, 24].
1. C8-StageTip: 200 μL Pipet tip packed with two C8 disks.
2. Titanium dioxide (TiO2) beads: Titansphere® beads (5–10 μm
particle size, 100 Å pore size, spherical particle shape, TiO2
crystals; GL-Sciences). Store under dry conditions to keep
specificity against phosphopeptides (see Note 3).
3. 100 % Methanol.

2.6 Phosphopeptide 1. 5 % Acetonitrile, 0.2 % TFA (pH 2.0) in water (v/v).

Enrichment 2. 80 % Acetonitrile, 0.2 % TFA (pH 2.0) in water (v/v).
on C8-TiO2-­StageTip
3. 5 % Ammonium hydroxide (pH 11.0) in water (v/v).
4. 5 % Piperidin in water (v/v).
5. 20 % Phosphoric acid in water (v/v).

2.7 C18-StageTip The 200 μL C18-StageTips are commercially available (e.g.,

Column Preparation ThermoScientific; or elsewhere) or can be in-house manufactured
according to [15, 24].
1. C18-StageTip: 200 μL Pipet tip packed with two C18 disks.
2. 80 % Acetonitrile, 0.2 % TFA.
3. 0.5 % Acetic acid in water (v/v).
4. 80 % Acetonitrile, 0.5 % acetic acid (v/v).
5. 5 % Acetonitrile, 0.2 % TFA (v/v).

2.8 LC-MS/MS 1. Chromatographic system: Easy-nLC 1000 (ThermoScientific).

2.8.1 Liquid 2. Column: EASY-Spray column, PepMap® RSLC, C18
Chromatography (ThermoScientific); particle size 2 μm; pore size 100 Å; col-
umn dimensions 75 μm × 50 cm (I.D. × L).
3. Loading buffer: 5 % Acetonitrile, 0.2 % TFA, pH 2.0.
4. Buffer A: 0.5 % Acetic acid, pH 2.0.
5. Buffer B: 0.5 % Acetic acid, 80 % acetonitrile, pH 2.0.

2.8.2 Mass Spectrometry 1. Mass analyzer: Q Exactive Plus (ThermoScientific).

Equipment
252 Maksim Zakhartsev et al.

3 Methods

3.1 Protein 1. Use 3 μL of proteins/ peptides contenting solution for quan-

Quantification tification, which is performed according to the manual instruc-
tion for use of NanoOrange ® protein quantification kit
(see Note 4). Choose the BSA standard (in water, in UTU
buffer, or in loading buffer for LC-MS/MS) in accordance
with the matrix of the protein/peptide-contenting solution
(see Note 5).

3.2 Total Protein This protocol is adapted from [20]. All protein extraction steps
Extraction must be done on ice.
1. Weigh root samples (g of fresh weight; gFW), wrap them indi-
vidually in aluminum foil, and freeze them immediately in liq-
uid nitrogen (see Note 6).
2. Break the frozen samples into small pieces with the hammer
while keeping them wrapped in aluminum foil. Make sure that
the samples are constantly frozen after harvesting to avoid
rapid dephosphorylation.
3. Transfer the cell material into ice-cold homogenization buffer
in a ratio of 5 mL homogenization buffer per 1 gFW.
4. Resuspend the cell material thoroughly by gentle stirring or
shaking to get rid of clots.
5. Grind the sample manually in a Potter® homogenizer on ice,
50 smooth strokes per 7 mL sample.
6. Filter the homogenate through four layers of Miracloth® to get
rid of cell wall material and tissue debris.
7. Centrifuge the homogenate at 7.5 × 103 × g for 15 min at 4 °C
to get rid of unbroken cells and organelles.
8. Collect the supernatant.
9. Centrifuge the supernatant at 48 × 103 × g for 80 min at 4 °C to
precipitate microsomal vesicles.
10. Collect the supernatant which represents the cytosolic fraction
(i.e., water-soluble proteins, can be used for other
experiments).
11. Resuspend the pellet, which represents the microsomal frac-
tion (i.e., endomembranes and membrane-associated pro-
teins), in 500 μL of ice-cold UTU-buffer.
12. Rigorously vortex the suspension and ultra-sonicate it for
approximately 30 s.
13. Quantify total protein content in the microsomal fraction
using NanoOrange ® protein quantification kit and BSA stan-
dards in UTU buffer.
 From Phosphoproteome to Modeling of Plant Signaling Pathways 253

3.3 In-Solution 1. Reduction: Add 1 μL of the reduction buffer per each 50 μg of

Protein Digest the total protein and incubate for 30 min at 25 °C and 260 rpm.
2. Alkylation: Add 1 μL of alkylation buffer per each 50 μg of the
total protein and incubate for 20 min at 25 °C in the dark and
260 rpm.
3. Digest 1: Add 0.5 μL of LysC stock solution per each 50 μg of
the total protein and incubate for 3 h at 37 °C and 260 rpm.
4. Dilution: Dilute the sample by fivefold with 10 mM Tris–HCl,
pH 8.0 (see Note 7).
5. Digest 2: Add 1 μL of trypsin stock solution per each 50 μg of
the total protein and incubate overnight at 37 °C and 260 rpm.
6. Stop the digest: Acidify the digest to 0.2 % TFA final concen-
tration (add 1/10 volume of 2 % TFA to reach pH 2.0).
7. Spin the sample on centrifuge at 20 × 103 × g, 3 min, at room
temperature to pellet any insoluble materials.

3.4 Peptide 1. Conditioning: Condition the MCX cartridge with 1 mL of

Purification 100 % methanol (see Note 8).
2. Equilibrate the MCX cartridge with 1 mL of water.
3. Sample loading: Load the digest to the conditioned MCX car-
tridge (see Note 9). Collect the flow-through and re-load it
two more times.
4. Wash 1: Wash the cartridge with 1 mL of 2 % formic acid (pH
2.0) to lock ionized compounds on MCX (see Note 10).
5. Wash 2: Wash the cartridge with 1 mL of 100 % methanol to
remove interfering unionized weaker acids and neutrals.
Completely expel the methanol from the cartridge.
6. Elution: Elute peptides from the cartridge with 5 × 200 μL of
5 % NH4OH in 80 % methanol (pH 11) (see Note 11). The
final total volume is 1000 μL.
7. Split the eluate into 15 % (“aliquot 1”) and 85 % (“aliquot 2”)
of the sample volume (see Fig. 2).
8. Dry both aliquots of the eluate to complete dryness in a vac-
uum centrifuge (10 mbar, 235 × g, rotor temperature 37 °C,
e.g., Christ® RVC 2–25 CD plus).
9. Redissolve the dried “aliquot 1” in 50 μL of LC-MS/MS load-
ing buffer (e.g., 5 % acetonitrile, 0.2 % TFA).
10. Rigorously vortex the sample, sonicate it for 30 s, and centri-
fuge at 2 × 103 × g for 5 min.
11. Transfer the supernatant of “aliquot 1” into fresh micro-tubes
(0.2 mL).
12. Quantify total peptide content in “aliquot 1” using
NanoOrange®protein quantification kit and BSA standards in
254 Maksim Zakhartsev et al.

loading buffer for LC-MS/MS. This is [Peptides]1 according

to notations at Fig. 2.
13. Reserve the “aliquot 1” for the further LC-MS/MS analysis.

3.5 C8-TiO2-StageTip 1. Activation of TiO2 beads: Activate TiO2 beads at 130 °C for
Column Preparation 30 min prior to use (see Note 3).
2. Preparing TiO2 bead stock suspension: Weigh 25 mg of TiO2
beads, resuspend it in 500 μL of 100 % methanol, and vortex
the suspension well.
3. Loading of TiO2 bead stock suspension: Load 20 μL of the
stock suspension (overall 1 mg; see Note 12) on top of the C8
disk in the 200 μL C8-StageTip (see Note 13).
4. Let the suspension settle down under the gravity force in order
to distribute the beads evenly.
5. Spin (see Note 14) the C8-TiO2-StageTip to force the solution
through.

3.6 Phosphopeptide 1. Redissolve the dried “aliquot 2” in 50 μL of 80 % acetonitrile

Enrichment and 0.2 % TFA.
on C8-TiO2-­StageTip 2. Rigorously vortex the sample, sonicate it for 30 s, and centri-
fuge at 2 × 103 × g for 5 min.
3. Insert the 200 μL C8-TiO2-StageTip into a spin adapter and
place it in a fresh microcentrifuge tube.
4. C8-TiO2-StageTip conditioning: Load 100 μL of 80 % acetoni-
trile and 0.2 % TFA to the C8-TiO2-StageTip and spin it to
force the solution through.
5. Replace the waste microcentrifuge tube with a fresh tube.
6. Sample loading: Load the sample (step 1) onto conditioned C8-
TiO2-StageTip (step 4) and spin it to force the sample through.
7. Sample reloading: Collect the flow-through, reload the sample
again, and then spin it to force the sample through. Repeat this
step twice.
8. Wash: Load 100 μL of 5 % acetonitrile and 0.2 % TFA to the
C8-TiO2-StageTip and then spin it to force the sample through
into waste microcentrifuge tube.
9. Add 50 μL of 20 % phosphoric acid into a fresh microcentri-
fuge tube where the phosphopeptides will be eluted in.
10. Elution 1: Elute the phosphopeptides with 50 μL of 5 %
NH4OH (pH 11.0) from the 200 μL C8-TiO2-StageTip into a
microcentrifuge tube with 20 % phosphoric acid (step 10)
(see Note 11).
11. Elution 2: Elute the remaining phosphopeptides with 50 μL of
5 % piperidine from the C8-TiO2-StageTip into the same tube.
The final volume of the eluate is 150 μL.
 From Phosphoproteome to Modeling of Plant Signaling Pathways 255

3.7 Desalting 1. Conditioning of C18-StageTips: Load 100 μL of 80 % acetoni-

on C18-StageTip trile and 0.2 % TFA to the C18-StageTip and spin it to force the
solution through into a waste microcentrifuge tube.
2. Load 2 × 100 μL of 0.5 % acetic acid to the C18-StageTip and
spin it to force the solution through into a waste microcentri-
fuge tube.
3. Sample loading: Load the sample (Subheading 3.6, step 11)
onto pre-conditioned C18-StageTip and spin it to force the
solution through into a waste microcentrifuge tube.
4. Washing: Load 2 × 100 μL of 0.5 % acetic acid to the C18-­
StageTip and spin it to force the solution through into a waste
microcentrifuge tube.
5. Elution: Elute the phosphopeptide enriched fraction by 2 × 20
μL of 80 % acetonitrile and 0.2 % TFA into fresh microcentri-
fuge tube.
6. Spin down the eluate to dryness (10 mbar, 235 × g, rotor tem-
perature 37 °C, e.g., Christ® RVC 2–25 CD plus).
7. Redissolve the phosphopeptides in 50 μL of loading buffer for
LC-MS/MS (i.e., 5 % acetonitrile, 0.2 % TFA).
8. Rigorously vortex the sample, sonicate it for 30 s, and centrifuge
at 2 × 103 × g for 5 min.
9. Transfer the supernatant of “aliquot 2” into a fresh micro-­tube
(0.2 mL).
10. Quantify the total peptide content in the sample using
NanoOrange®protein quantification kit and BSA standards in
loading buffer for LC-MS/MS. This is [Peptides]2 according
to notations at Fig. 2.

3.8 LC-MS/MS 1. Injection volume: 1–5 μL to achieve at least 2 μg of overall

column load with the total peptides.
3.8.1 Liquid
Chromatography 2. Flow rate: 250 nL/min.
3. Gradient %B: 0 min 5 %, 200 min 35 %, 240 min 60 %, 242 min
90 %, 257 min 90 %, 258 min 5 %, 263 min 5 %.
4. Operation column temperature: 50 °C.
5. Operation pressure: 500 bar.

3.8.2 Mass Spectrometry 1. Polarity: Positive.

2. Full MS: Resolution 70,000 (at m/z = 200 Th); AGC target
1e6; maximum IT 20 ms; scan range 300–1600 m/z.
3. dd-MS2: Resolution 17,500 (at m/z = 200 Th); AGC target
1e5; maximum IT 120 ms; TopN 5; isolation window 2.2
m/z; scan range 200–2000 m/z; NCE 25.
4. dd-Settings: Underfill ratio 0.1 %; dynamic exclusion 40 s.
256 Maksim Zakhartsev et al.

3.9 Data This method is mainly designated for quantification of stoichiom-

Employment etry in pairs of phosphorylated peptide and its corresponding
unmodified cognate, i.e., to measure phosphorylation stoichiom-
etry. However, this method can also be applied for search/screen
of gross phosphorylation sites or qualitative assessment of phos-
phorylation of proteins from certain signaling pathways, without
detecting the unmodified cognates.
Stimulus response (i.e., dynamic perturbation) experimental
approaches are widely used in systems biology to provoke dynamic
responses of the studied system. The phosphorylation stoichiom-
etry of signaling proteins is a state variable in mathematical models
of the conserved moieties or cascade reactions of the signaling
pathways (Fig. 1). Steady-state phosphorylation stoichiometry and
its time-dependent dynamics in response to a perturbation event
allow parameter estimation of the kinetic equations that describe
the corresponding cascades in signaling pathways [1]. Measurements
of steady-state phosphorylation stoichiometry under different sig-
nal strength allow quantification of local and global response coef-
ficients of the signaling pathway, if the kinetic properties of the
reaction cascades are known [7]. This type of modeling can be
performed either in package programs like MATLAB (The
MathWorks) and MATHEMATICA (Wolfram Research) or in spe-
cialized software like Simmune [25].

4 Notes

1. Please refer to Fig. 2 for the notations associated with corre-

sponding workflow steps. The loss of peptides’ amounts (l)
during phosphopeptide enrichment can be estimated as
[tP]1
l= (1)
[tP]2

where [tP]i—concentration of total peptides before (1) and

after (2) enrichment (μg/mL). The peptide quantification
must be accurate; therefore please refer to Subheading 3.1.
The amount (i.e., mass in μg) of the injected peptides (mi)
for LC-MS/MS analysis can be calculated as
mi = xi ´[tP]i (2)

where xi—injection volume used for the LC-MS/MS analysis

(i = 1,2) (μL).
The mass-specific content of individual peptide (Pj) and its
phosphorylated species (pPj) can be estimated from its label-­
freequantification (iBAQi,j) and normalized per correspond-
ing mi (j = N):
From Phosphoproteome to Modeling of Plant Signaling Pathways 257

ì iBAQ1. j
ï Pj =
ï m1
í (3)
ï pP = iBAQ 2. j ´ l
ï j m2
î

where the correction factor l is calculated in Eq. 1. The peptide

content after phosphopeptide enrichment must be corrected
with the loss of the total peptides mass (Eq. 1). The total pool
pool
of the particular peptide (Pj ) consists of non-phosphorylated
and phosphorylated species:
Pjpool = Pj + pPj (4)

and correspondingly a part of each species is

Pj pPj
pool
+ =1 (5)
Pj Pjpool

2. Hereinafter designated just as water.

3. The specificity of TiO2 beads against phosphopeptides is
reduced by water absorption when it is kept without desicca-
tion. The specificity can be recovered by heating the beads in a
drying oven at 130 °C for 30 min [16].
4. Accurate and highly specific quantification of protein/peptide is
essential for this approach. We have selected NanoOrange®protein
quantification kit (Molecular Probe) for this purpose, because it
is very sensitive and specific to proteins/peptides, has a wide
quantification range (0.1–2000 μg/mL), and is compatible with
nucleic acids, reducing agents, and detergents.
5. UTU buffer or 5 % acetonitrile and 0.2 % TFA significantly
quench the fluorescence and therefore they must be included
into the solution matrix for compensation.
6. The frozen samples can be stored at −80 °C for further analysis.
7. The dilution step is required in order to get final 1.2 M urea,
0.4 M thiourea, and 10 mM Tris–HCl (pH 8.0), which is
favorable for trypsin operation.
8. Do not let the cartridge dry; always expel one mobile phase
with another.
9. The 1 cc cartridge from Waters has a load volume of a matrix
with an analyst up to 50 mL. At this step, the sample is in 1.2 M
urea, 0.4 thiourea, 10 mM Tris–HCl, and 0.2 % TFA, pH 2.0.
10. This step also removes undigested proteins and salts.
11. Phosphopeptides are not stable in alkaline conditions; there-
fore, in order to minimize the exposure time, it is advised to
258 Maksim Zakhartsev et al.

dry the eluate immediately at vacuum centrifuge, as it is

exemplified in Subheading 3.4, steps 7 and 8, or immediately
neutralize the alkaline solution with strong acid as it is exem-
plified in Subheading 3.6, steps 10 and 11.
12. 1 mg TiO2 beads per a single C8-StageTip column have a binding
capacity of ~100 μg of total peptides from Arabidopsis [16].
13. The choice of the C8 material is based on the idea that the
membrane disk is only used to retain the TiO2 beads inside the
tip, but the C8-disk itself does not participate in the phospho-
peptide enrichment. The C8-StageTips can be stored at room
temperature.
14. “Spin” hereinafter refers to centrifugation of a StageTip in a
bench microcentrifuge (e.g., Mini Star Silverline, Galaxy Mini
Centrifuge, VWR) at 2 × 103 g (or 6 × 103 rpm) for 1 min at
room temperature.

References
1. Klipp E, Liebermeister W (2006) Mathematical 8. Hein MY, Sharma K, Cox J et al (2013)
modeling of intracellular signaling pathways. Chapter 1—proteomic analysis of cellular
BMC Neurosci 7(Suppl 1):S10. doi:10.1186/ systems. In: Walhout AJM, Vidal M, Dekker
1471-2202-7-S1-S10 J (eds) Handbook of systems biology.
2. Mariottini C, Iyengar R (2013) Chapter 16—sys- Academic, San Diego, pp 3–25
tem biology of cell signaling. In: Walhout AJM, 9. Cox J, Mann M (2011) Quantitative, high-­
Vidal M, Dekker J (eds) Handbook of systems resolution proteomics for data-driven systems biol-
biology. Academic, San Diego, pp 311–327 ogy. Annu Rev Biochem 80(1):273–299.
3. Duan G, Walther D, Schulze W (2013) doi:10.1146/annurev-biochem-061308-093216
Reconstruction and analysis of nutrient-­ 10. Choudhary C, Mann M (2010) Decoding sig-
induced phosphorylation networks in nalling networks by mass spectrometry-based
Arabidopsis thaliana. Front Plant Sci 4:540. proteomics. Nat Rev Mol Cell Biol 11(6):427–
doi:10.3389/fpls.2013.00540 439. doi:10.1038/nrm2900
4. Niittylä T, Fuglsang AT, Palmgren MG et al 11. Altelaar AFM, Munoz J, Heck AJR (2013)
(2007) Temporal analysis of sucrose-induced Next-generation proteomics: towards an inte-
phosphorylation changes in plasma membrane grative view of proteome dynamics. Nat Rev
proteins of arabidopsis. Mol Cell Proteomics Genet 14(1):35–48. doi:10.1038/nrg3356
6(10):1711–1726. doi:10.1074/mcp. 12. Olsen J, Macek B (2009) High accuracy mass
M700164-MCP200 spectrometry in large-scale analysis of protein
5. Schulze WX (2010) Proteomics approaches to phosphorylation. In: Lipton M, Paša-Tolic
understand protein phosphorylation in p ­ athway L (eds) Mass spectrometry of proteins and
modulation. Curr Opin Plant Biol 13(3): peptides. Humana Press, Totowa, NJ,
279–286. doi:10.1016/j.pbi.2009.12.008 pp 131–142
6. Blagoev B, Ong S-E, Kratchmarova I et al 13. Schmelzle K, White FM (2006) Phosphoproteomic
(2004) Temporal analysis of phosphotyrosine-­ approaches to elucidate cellular signaling net-
dependent signaling networks by quantitative works. Curr Opin Biotechnol 17(4):406–414.
proteomics. Nat Biotechnol 22(9):1139–1145. doi:10.1016/j.copbio.2006.06.004
doi:10.1038/nbt1005 14. Larsen MR, Thingholm TE, Jensen ON et al
7. Kholodenko BN, Hoek JB, Westerhoff HV (2005) Highly selective enrichment of phos-
et al (1997) Quantification of information phorylated peptides from peptide mixtures
transfer via cellular signal transduction path- using titanium dioxide microcolumns. Mol
ways. FEBS Lett 414(2):430–434. Cell Proteomics 4(7):873–886. doi:10.1074/
doi:10.1016/S0014-5793(97)01018-1 mcp.T500007-MCP200
 From Phosphoproteome to Modeling of Plant Signaling Pathways 259

15. Wisniewski JR, Zougman A, Nagaraj N et al H+ ATPase during pollen grain germination
(2009) Universal sample preparation method and tube growth. Planta 213(1):132–141.
for proteome analysis. Nat Methods 6(5): doi:10.1007/s004250000483
359–362. doi:10.1038/nmeth.1322 21. Pratt JM, Simpson DM, Doherty MK et al
16. Nakagami H (2014) StageTip-based (2006) Multiplexed absolute quantification for
HAMMOC, an efficient and inexpensive phos- proteomics using concatenated signature pep-
phopeptide enrichment method for plant shot- tides encoded by QconCAT genes. Nat Protoc
gun phosphoproteomics. In: Jorrin-Novo JV 1(2):1029–1043
et al (eds) Plant proteomics. Humana Press, 22. Cox J, Mann M (2008) MaxQuant enables
New York, pp 595–607 high peptide identification rates, individual-
17. Thingholm TE, Jorgensen TJD, Jensen ON ized p.p.b.-range mass accuracies and
et al (2006) Highly selective enrichment of proteome-­ wide protein quantification. Nat
phosphorylated peptides using titanium dioxide. Biotechnol 26(12):1367–1372. doi:10.1038/
Nat Protoc 1(4):1929–1935. doi:10.1038/ nbt.1511
nprot.2006.185 23. Steen H, Jebanathirajah JA, Springer M et al
18. Beckers GM, Hoehenwarter W, Röhrig H et al (2005) Stable isotope-free relative and abso-
(2014) Tandem metal-oxide affinity chroma- lute quantitation of protein phosphorylation
tography for enhanced depth of phosphopro- stoichiometr y by MS. Proc Natl Acad Sci
teome analysis. In: Jorrin-Novo JV et al (eds) U S A 102(11):3948–3953. doi:10.1073/
Plant proteomics. Humana Press, New York, pnas.0409536102
pp 621–632 24. Rappsilber J, Mann M, Ishihama Y (2007)
19. Colby T, Röhrig H, Harzen A et al (2011) Protocol for micro-purification, enrichment,
Modified metal-oxide affinity enrichment com- pre-fractionation and storage of peptides for
bined with 2D-PAGE and analysis of phospho- proteomics using StageTips. Nat Protoc 2(8):
proteomes. In: Dissmeyer N, Schnittger A 1896–1906
(eds) Plant kinases. Humana Press, New York, 25. Meier-Schellersheim M, Xu X, Angermann B
pp 273–286 et al (2006) Key role of local regulation in
20. Pertl H, Himly M, Gehwolf R et al (2001) chemosensing revealed by a new molecular
Molecular and physiological characterisation interaction-­
based modeling method. PLoS
of a 14-3-3 protein from lily pollen grains reg- Comput Biol 2(7), e82. ­doi:10.1371/journal.
ulating the activity of the plasma membrane pcbi.0020082
 Chapter 19

Interpretation of Quantitative Shotgun Proteomic Data

Elise Aasebø, Frode S. Berven, Frode Selheim, Harald Barsnes,
and Marc Vaudel

Abstract
In quantitative proteomics, large lists of identified and quantified proteins are used to answer biological
questions in a systemic approach. However, working with such extensive datasets can be challenging, espe-
cially when complex experimental designs are involved. Here, we demonstrate how to post-process large
quantitative datasets, detect proteins of interest, and annotate the data with biological knowledge. The
protocol presented can be achieved without advanced computational knowledge thanks to the user-friendly
Perseus interface (available from the MaxQuant website, www.maxquant.org). Various visualization tech-
niques facilitating the interpretation of quantitative results in complex biological systems are also
highlighted.

Key words Quantification, Data interpretation, Perseus, Data post-processing

1 Introduction

Quantitative shotgun proteomics has become the method of choice

for the description of large scale biological systems [1]. The
approach relies on the proteome-wide quantification of proteins,
often including screening for posttranslational modifications [2, 3].
Different quantification methods are available to the researcher,
ultimately providing a list of relatively and/or absolutely quantified
proteins [4–6]. Before inferring any biological sense from the
quantitative results, the data must undergo several post-processing
steps, typically including normalization, statistical evaluation, and
functional enrichment—which can be challenging due to the
amount of data, and its specificity and complexity [7].
In this chapter, we present a simple workflow for the post-
processing of proteome-wide quantification results which can be
applied without advanced knowledge in computer science thanks
to the user friendly Perseus interface. The workflow is here applied
to a freely available dataset used for illustrative purposes, consisting

Jörg Reinders (ed.), Proteomics in Systems Biology: Methods and Protocols, Methods in Molecular Biology, vol. 1394,
DOI 10.1007/978-1-4939-3341-9_19, © Springer Science+Business Media New York 2016

261
262 Elise Aasebø et al.

of five cell lines derived from acute myeloid leukemia (AML)

patients measured with a spiked-in internal standard (IS) obtained
from the combination of the same five AML cell lines metabolically
labeled with heavy isotopes [8] and analyzed using MaxQuant
version 1.4.1.2 [9]. Subsequently, we exemplify the use of several
visualization techniques allowing the critical interpretation of the
data: volcano plots, principal component analysis (PCA), and
hierarchical clustering.

2 Material

1. The dataset here used for illustrative purposes is freely available

through the ProteomeXchange [10] consortium via the
PRIDE [11] partner repository under the accession number
PXD000441 (Results_MQ_5cell-line-mix.zip).
2. Perseus is a software tool freely available upon registration on
the MaxQuant website (http://www.maxquant.org). It can be
downloaded directly from http://www.perseus-framework.
org. After downloading the software and extracting the zipped
folder, open the Perseus program located in the Perseus folder.
For Perseus v1.5.0.0 and newer, download the plugin
PluginProteomicRuler.dll from http://perseus-framework.
org/plugins and put it in the main Perseus folder. See Note 1.

3 Methods

3.1 Loading Data 1. Open the Perseus program (annotated with the Perseus logo)
in Perseus located in the main Perseus folder.
2. Download annotations: Click the blue tab indicated in Fig. 1a
and then Annotation download (Fig. 1b) in the menu thatw-
wshows up. Go to the provided Dropbox folder and download
the appropriate mainAnnot.txt.gz file in the OrganismSpecific
folder (in this example use mainAnnot.homo_sapiens.txt.gz),
put the downloaded annotation file into Perseus\conf\annotation.
Extract the zipped file (see Note 2).
3. Import your output file: Press the green upwards-pointing
arrow to load your data (Fig. 1c). When you hover over the
arrow you will get the message: Generic matrix upload.
4. After clicking the green arrow, a matrix opens where you can
upload your .txt file (Fig. 2), in this case use the
“proteinGroups_5cell-line-mix.txt” file (see Note 3). Transfer
quantitative data (ratios) into the Expression field (see Note 4).
You can also choose other parameters of interest, such as
“Number of proteins”, “Unique peptides”, etc. Put numerical
 Interpretation of Quantitative Shotgun Proteomic Data 263

Fig. 1 When using Perseus for the first time, it is recommended to download
annotations (a, b). To import a data set (Generic matrix upload), press the green
arrow (c)

columns into the Numerical field, and columns containing

text into the Text field. Some columns are preselected if you
are working with MaxQuant output. Click OK and you will see
the selected columns in the Perseus matrix.

3.2 Filtering 1. Start by filtering on the categorical columns. Go to: Filter

and Rearranging rows → Filter rows based on categorical column. In the appearing
Columns window, select the categorical column you wish to remove. In
this example, remove contaminants, reverse hits and only identi-
fied by site.
2. Remove the empty columns: Rearrange → Remove empty columns.
3. Rename the columns: Rearrange → Rename columns. In this
dataset “1” is Molm-13, “2” is MV4-11, “3” is NB4, “4” is
OCI-AML3 and “5” is THP-1.
4. Remove proteins without an expression value: Filter
rows → Filter rows based on valid values. Use default parameters
for this dataset (see Note 5).
5. Have a look to the right in your Perseus window. All the steps
you have performed this far appear as individual matrices. Here
you can easily navigate between the matrices, and inspect the
steps you have performed. If you want to delete a matrix,
264 Elise Aasebø et al.

Fig. 2 Select the output file (.txt format) to upload (usually the ProteinGroups.txt file for MaxQuant).
The unique column headers in the output are listed in the left column. Select the columns to analyze in
Perseus and press the arrows to load them into the columns to the right. Be sure to load protein expression
data into the Expression fields and numerical information—such as the number of peptides or sequence
coverage—into the Numerical field

simply select the matrix and click the red cross. If you are using
Perseus version 1.5.0.15 or newer, you can (re-)name the
matrices by double clicking them (Fig. 3).
6. The MaxQuant output display the ratios as Heavy/Light, so in
order to compare all light samples against the heavy sample
(i.e., the internal standard) we have to invert the comparison
to get Light/Heavy. Go to: Basic → Transform, and write “1/
(x)” in the Transformation field.
7. Transform the expression values into log values (see Note 6).
Go to: Basic → Transform → choose “log2(x)”.

3.3 Normalization 1. The example dataset is now in log2 values; thus the normal
distribution should be centered on zero. Visualize the distri-
bution of the dataset by using a histogram. Go to:
Visualization → Histogram. Accept the default settings by
 Interpretation of Quantitative Shotgun Proteomic Data 265

Fig. 3 Matrix overview. Each new task that changes the Data matrix will appear as a new matrix. It is possible
to navigate between different matrices, delete paths and (re-) name matrices, thus allowing inspecting inter-
mediate results

a b
Counts

Counts

0 5 0 5
MV4-11 MV4-11

Fig. 4 Histogram displaying the protein ratio distribution before (a) and after (b) normalization, exemplified by
two of the cell lines. The Histogram tab can be found next to the Data tab in the same matrix

clicking OK. The histogram will appear in the same matrix,

you find it as a new tab next to the Data tab. You may notice
that you ought to normalize the data (Fig. 4a).
2. To normalize, go to: Normalization → Subtract. Choose
Columns in the Matrix access field and keep the default for
subtracting the median. Click OK. Make a new histogram and
check that the center of distribution has changed (Fig. 4b).

3.4 Gene Annotation 1. Add categorical annotations. Go to: Annot. Columns → Add
annotation. Select the mainAnnot.homo_sapiens.txt file as the
Source (Fig. 5). Choose from “GOBP name” down to
“Keyword” in the Annotations to be added field and transfer
the selected annotations over to the empty field using the
arrow. Click OK.
266 Elise Aasebø et al.

Fig. 5 Add annotation—a prerequisite is that annotations are already downloaded (described in Subheading
3.1.2 and Fig. 1)

Fig. 6 Identify regulated proteins in each sample by using the Significance A test with the indicated parame-
ters. Regulated proteins are marked with the symbol “+” in the Data matrix

3.5 Statistical 1. Go to: Basic → Significance A. Select all cell lines and transfer
Evaluation them to the empty field using the arrow. In the Use for trunca-
tion field choose P value and use the default (0.05) as Threshold
value (Fig. 6). Click OK. The protein ratios that are significant
outliers relative to the sample population are now annotated
with “+” in the matrix.
2. Compare the cell lines derived from patients at time of diagno-
sis (MV4-11 and OCI-AML3) to the other cells lines derived
from patients during relapse (Molm-13, NB4 and THP-1).
Start by making two groups: Annot. rows → Categorical anno-
tations rows and specify the groups as “Diagnosis” or “Relapse”
as shown in Fig. 7. Click OK.
3. Do a two-samples t-test to compare the groups: Tests → Two
samples t-test. Select P value in the Use for truncation field and
keep 0.05 as Threshold value, and default settings for other
parameters (Fig. 8). Click OK.
 Interpretation of Quantitative Shotgun Proteomic Data 267

Fig. 7 Categorical annotation rows is used to create groups. In this case we create “Group1” (default name)
and mark the samples with either “Diagnosis” or “Relapse”

Fig. 8 Find significantly expressed proteins between the groups “Diagnosis” or “Relapse” by applying a Two-
samples tests. Use the indicated parameters. The significantly expressed proteins are marked with the symbol
“+” in the Data matrix

3.6 Volcano Plot 1. Compare the two groups. Go to: Visualization → Scatter plot
and keep Columns in the Matrix access field. Click OK.
2. Find the Scatter plot tab next to the Data tab in the same
matrix. Select t-test difference and –Log t-test p value in the tabs
at right of the scatter plot (Fig. 9a).
3. To look for specific pathways or annotations go to the
Categories tab and select terms related to apoptosis (Fig. 9b).
4. Alternatively, go to: Misc. → Volcano Plot. Keep the default
parameters and click OK.

3.7 Principal 1. To perform principal component analysis, valid expression val-

Component Analysis ues for each protein are required. To filter, go to: Filter
(PCA) rows → Filter rows based on valid values and write “5” in the
Minimum number of values field. Click OK.
2. Create a PCA plot: Clustering/PCA → Principal component
analysis. Tick in the tab for Categoryenrichment in components
and keep the default settings (see Note 7). Click OK.
268 Elise Aasebø et al.

Fig. 9 Volcano plot. Start by creating a scatter plot and compare the t-test difference to the -Log t-test p value
(a) to visualize the volcano plot. Inspect enriched terms, such as “apoptosis,” by going to the Categories tab
(b). The proteins annotated by the term appear in red in the volcano plot

3. Find the PCA plot next to the Data tab and change from No
labels to All labels (Fig. 10a).
4. Color the groups “Diagnosis” and “Relapse” with two differ-
ent colors by clicking on the color symbol (Fig. 10b).
5. Select the proteins responsible for the separation in the PCA
plot by changing the mode from zoom (magnifying glass) to
select (square) (Fig. 10c). The selected proteins in this example
are marked in red in the PCA plot (Fig. 10d) and the protein
name is marked in blue in the box indicated in Fig. 10e.

3.8 Hierarchical 1. Normalize on the protein level: Before clustering you need to
Clustering normalize the protein values at the row level. Go to:
Normalization → Z-score and select Rows in the Matrix Access
field (see Note 8). Use default settings for the other parame-
ters. Click OK.
 Interpretation of Quantitative Shotgun Proteomic Data 269

Fig. 10 Principal Component Analysis (PCA). Change from No labels to All labels (a) to reveal the sample names
of the dots in the PCA plot. Color the samples belonging to each group in different colors (b) to see if the cell
lines of one group cluster together. To look at the proteins that cause the separation of the samples, see the
box at the lower left. Change from the zoom to select option (c) and select the proteins to the left (d). Note that
the selected proteins appear with a blue line in the table at lower right (e)

2. We want to make a cluster of the proteins that are significantly

regulated between the two groups “Diagnosis” and “Relapse”.
Filter to keep the significant proteins: Filter rows → Filter rows
based on categorical column, select t-test Significant in the Column
field and Keep matching rows in the Mode field. Click OK.
3. To perform hierarchical clustering, go to: Clustering/
PCA → Hierarchical clustering. Use the default settings and
click OK.
4. Define clusters at the protein (row) and group (column) level as
indicated in Fig. 11a. In this dataset Number of clusters was set to
“2” at both the protein and group level. Accept with Apply and
note that two clusters appear in the figure and click OK.
 270
Elise Aasebø et al.

Fig. 11 Hierarchical clustering. The clustering is performed both at protein (row) and sample (column) level. Specify the number of clusters (in this example two) at
both levels by clicking the symbols indicated in (a). Configure the row and column names by clicking the symbols indicated in (b). At the upper right there is a table
of Row clusters (c), you can select a cluster by clicking the line. The proteins of the selected cluster are displayed in the profile plot (d). By selecting Protein names
in (e), you can see the table of protein names in the Members tab
 Interpretation of Quantitative Shotgun Proteomic Data 271

5. Configure the row and column names by clicking the symbols

indicated in Fig. 11b. In the Row names field choose Protein
names and in the Column names field select Group1 under
Addtl. Column names. Click OK.
6. Inspect the row clusters by selecting one of the clusters as indi-
cated in Fig. 11c. The expression profiles of the proteins cor-
responding to the selected cluster appear in the plot below
(Fig. 11d).
7. In order to explore which proteins that belong to a selected
cluster, change the tab indicated in Fig. 11e from < None > to
Protein names and look under the Members tab. If any
annotation (i.e., KEGG pathway, GO term) is enriched, it
will appear if you click the bar symbol next to the Protein
names field.

4 Notes

1. For more help and discussion with other users, visit the
Perseus Google Group at https://groups.google.com/
forum/#!forum/perseus-list.
2. You need a program to extract zipped files. Here we used
7-Zip File Manager, downloaded freely from the internet.
After installing 7-Zip, right click on the zipped folder and
“extract” the folder.
3. You can upload tab separated files, such as the .txt files from a
MaxQuant search, but also output files from other software
and types of experiments (genomics, transcriptomics, metabo-
lomics, etc.), as long as the file contains a unique column
header and is in the .txt format. In this example we look at the
protein expression data, but you can also explore and analyze
peptide and modification data with this software.
4. The expression values will differ between ratios (in case of
SILAC, TMT, iTRAQ, etc.) and intensities (in case of label
free), depending on your experiment.
5. In general you should have expression values in minimum 50
% of your samples, this allows for better statistics and cluster-
ing. It is also possible to specify the minimum number of valid
values in each group or in at least one group in the Mode field.
This can be relevant if a protein is expressed in one group and
absent in another group.
6. If you have intensity values you might use a base 10 logarith-
mic transformation, while base 2 logarithm is recommended
for ratios and is used in this example.
272 Elise Aasebø et al.

7. By ticking off Categoryenrichment in components you can later

explore the annotations the proteins are related to. Go to the
Categories tab above the protein list and click one of the anno-
tations. Sometimes you will see that the proteins that contrib-
ute most to the PCA plot are related to the same annotations,
and this might have biological implications. Look for instance
at the proteins related to “GTPaseactivation”.
8. With Z-score normalization the mean of the protein row is
subtracted from each protein and divided by the standard
deviation of protein row. This should be performed at pro-
tein level before hierarchical clustering, as it allows visualiz-
ing the protein clusters differentially expressed between the
samples analyzed, independently of average protein abun-
dance. It is then important to note that, even though pro-
teins in one cluster have the same expression profile, for
instance lower in OCI-AML3 and MV4-11 than the three
other cell lines, these proteins will not necessarily have the
same expression intensity.
9. Even though Perseus was initially created for the analysis of
MaxQuant results, its versatile input format also makes it com-
patible with most proteomic software, including OpenMS
[12, 13], the TransProteomic Pipeline (TPP) [14], or
PeptideShaker [15].
10. Although the methods presented here are generic and can be
applied to most proteomics studies, it is important to critically
tailor the post-processing workflow to the specificities of each
experimental design.
11. Perseus is not limited to proteomic analyses, and can be oper-
ated on other types of omics datasets. It can also be used for
multi-omic studies [16].

Acknowledgements

F.S.B. and F.S. acknowledge the support by the Norwegian Cancer

Society. H.B. is supported by the Research Council of Norway.

References

1. Aebersold R, Mann M (2003) Mass spectrometry- 3. Altelaar AF, Munoz J, Heck AJ (2013) Next-
based proteomics. Nature 422:198–207 generation proteomics: towards an integrative view
2. Olsen JV, Mann M (2013) Status of large-scale of proteome dynamics. Nat Rev Genet 14:35–48
analysis of post-translational modifications by 4. Vaudel M, Sickmann A, Martens L (2010)
mass spectrometry. Mol Cell Proteomics Peptide and protein quantification: a map of
12:3444–3452 the minefield. Proteomics 10:650–670
 Interpretation of Quantitative Shotgun Proteomic Data 273

5. Bantscheff M, Lemeer S, Savitski MM et al nated proteomics data submission and dissemi-

(2012) Quantitative mass spectrometry in pro- nation. Nat Biotechnol 32:223–226
teomics: critical review update from 2007 to 11. Martens L, Hermjakob H, Jones P et al (2005)
the present. Anal Bioanal Chem 404:939–965 PRIDE: the proteomics identifications data-
6. Bantscheff M, Schirle M, Sweetman G et al base. Proteomics 5:3537–3545
(2007) Quantitative mass spectrometry in pro- 12. Kohlbacher O, Reinert K, Gropl C et al (2007)
teomics: a critical review. Anal Bioanal Chem TOPP—the OpenMS proteomics pipeline.
389:1017–1031 Bioinformatics 23:e191–e197
7. Vaudel M, Sickmann A, Martens L (2014) 13. Bertsch A, Gropl C, Reinert K et al (2011)
Introduction to opportunities and pitfalls in OpenMS and TOPP: open source software for
functional mass spectrometry based pro- LC-MS data analysis. Methods Mol Biol
teomics. Biochim Biophys Acta 1844:12–20 696:353–367
8. Aasebo E, Vaudel M, Mjaavatten O et al (2014) 14. Deutsch EW, Mendoza L, Shteynberg D et al
Performance of super-SILAC based quantita- (2010) A guided tour of the Trans-Proteomic
tive proteomics for comparison of different Pipeline. Proteomics 10:1150–1159
acute myeloid leukemia (AML) cell lines. 15. Vaudel M, Burkhart JM, Zahedi RP et al
Proteomics 14:1971–1976 (2015) PeptideShaker enables reanalysis of
9. Cox J, Mann M (2008) MaxQuant enables MS-derived proteomics data sets. Nature bio-
high peptide identification rates, individualized technology 33:22–24
p.p.b.-range mass accuracies and proteome- 16. Cox J, Mann M (2012) 1D and 2D annotation
wide protein quantification. Nat Biotechnol enrichment: a statistical method integrating
26:1367–1372 quantitative proteomics with complementary
10. Vizcaino JA, Deutsch EW, Wang R et al (2014) high-throughput data. BMC Bioinformatics
ProteomeXchange provides globally coordi- 13(Suppl 16):S12
 Chapter 20

A Simple Workflow for Large Scale Shotgun

Glycoproteomics
Astrid Guldbrandsen, Harald Barsnes, Ann Cathrine Kroksveen,
Frode S. Berven, and Marc Vaudel

Abstract
Targeting subproteomes is a good strategy to decrease the complexity of a sample, for example in body
fluid biomarker studies. Glycoproteins are proteins with carbohydrates of varying size and structure
attached to the polypeptide chain, and it has been shown that glycosylation plays essential roles in several
vital cellular processes, making glycosylation a particularly interesting field of study. Here, we describe a
method for the enrichment of glycosylated peptides from trypsin digested proteins in human cerebrospinal
fluid. We also describe how to perform the data analysis on the mass spectrometry data for such samples,
focusing on site-specific identification of glycosylation sites, using user friendly open source software.

Key words Glycoproteomics, Enrichment, Data interpretation

1 Introduction

Enrichment for subproteomes can help circumvent the challenge

of a few high abundant proteins masking proteins of lower abun-
dance in biological samples and body fluids [1]. An example of
such a subproteome are the glycoproteins, proteins carrying one or
more carbohydrates (glycans) of varying size and structure at par-
ticular amino acid residues in the protein sequence [2, 3]. When a
glycan is attached to a protein it is referred to as a glycosylation—
one of the most common post-translational modifications.
Glycoproteins are most often secreted or membrane-attached [2],
and are known to be involved in protein folding and protection
from degradation [3–6]. They also play important roles in cell
communication, signaling, aging and cell adhesion [7–9]. Several
known clinical biomarkers, as well as therapeutic targets, are glyco-
proteins [10–17].
In this chapter, we present a simple protocol for glycopeptide
enrichment, and subsequent protein identification after shotgun

Jörg Reinders (ed.), Proteomics in Systems Biology: Methods and Protocols, Methods in Molecular Biology, vol. 1394,
DOI 10.1007/978-1-4939-3341-9_20, © Springer Science+Business Media New York 2016

275
276 Astrid Guldbrandsen et al.

proteomic analysis. Note that the liquid-chromatography coupled

to mass spectrometry (LC-MS) acquisition is not detailed as it does
not differ from standard shotgun proteomics [18]. The glycopep-
tide enrichment described here is based on solid-phase extraction
of N-linked glycopeptides, as described by Tian et al. [19] and
Berven et al. [20], with minor modifications as described in detail
in [18]. In the last steps of the protocol, the glycans are released
from the glycopeptides by the enzyme PNGase F, rendering the
identification of the glycans’ chemical composition and structure
not possible. The glycan release by PNGase F leads to a deamida-
tion of the asparagine residue where the glycan was attached, con-
verting it to an aspartic acid residue, resulting in a 1 Da mass
increase when an amide group is exchanged for a hydroxyl group
[21]. This mass shift allows for glycosylation site identification in
the mass spectrometry data. However, in order to distinguish gly-
cosylation sites from natural deamidation, it is necessary to (manu-
ally or automatically) validate the detected glycosylation sites, e.g.,
specify that the deamidated asparagine must satisfy the following
glycosylation pattern [N X^P S/T] (where X^P is any amino acid
except proline), see [18] Supplementary File 1C. For the experi-
ment described here, the peptides were manually inspected and
filtered as part of the post-processing. It should be noted that in
this approach for identifying glycosylation sites, a deamidation of
an asparagine residue in the glycosylation pattern is only an indica-
tor of a likely former glycosylation site, i.e., there is no direct detec-
tion of the glycosylation.
The data interpretation protocol is here demonstrated using a
dataset of human cerebrospinal fluid (CSF) obtained from the
Cerebrospinal Fluid Proteome Resource (CSF-PR) [18] (www.
probe.uib.no/csf-pr), and is conducted using SearchGUI [22]
(http://searchgui.googlecode.com) and PeptideShaker [23]
(http://peptide-shaker.googlecode.com), both freely available
from their respective Web pages.

2 Materials

If not otherwise stated, all chemicals and products are purchased

from Sigma-Aldrich (St. Louis, MO, USA). All solutions should be
prepared with deionized water.
1. 0.1 % N-octyl-β-D-glucopyranoside (NOG).
2. 3 kDa ultracentrifugation filters (Amicon Ultra-4, Merck
Millipore, Billerica, MA, USA).
3. Denaturation buffer: 8 M urea/0.4 M ammonium bicarbon-
ate/0.1 % SDS (Bio-Rad Laboratories, Hercules, CA, USA).
4. 120 mM Tris(2-carboxyethyl)phosphine (TCEP).
 A Simple Workflow for Large Scale Shotgun Glycoproteomics 277

5. 160 mM iodoacetamide (IAA), light sensitive.

6. 100 mM ammonium bicarbonate (Ambic).
7. Trypsin porcine (Promega).
8. Oasis™ HLB 10 mg (30 μm) plates (Waters, Milford, MA,
USA).
9. Oasis™ HLB μElution plates (Waters, Milford, MA, USA).
10. 100 % formic acid (FA).
11. 0.1 % FA.
12. 0.1 % trifluoroacetic acid (TFA).
13. 50 % acetonitrile (ACN)/0.1 % TFA.
14. 80 % ACN–0.1 % TFA.
15. 80 % ACN–0.1 % FA.
16. 0.1 M sodium periodate, light sensitive.
17. BcMag® hydrazide-modified magnetic beads, 30 mg/mL
(BioClone Inc. San Diego, CA, USA).
18. Dynal® magnetic bead separation rack (Life Technologies).
19. 100 % N,N-dimethylformamide (DMF, toxic).
20. PNGase F enzyme for proteomics 1 unit/μL.
21. 5 M hydrochloric acid (HCl).
22. Dataset: The dataset used here for illustrative purposes is freely
available from the ProteomeXchange Consortium [24] via the
PRIDE partner repository [25], with the identifiers
PXD000651-PXD000657. The dataset can also be inspected
(and proteins and peptides exported) via CSF-PR at http://
probe.uib.no/csf-pr.
23. Software: SearchGUI [22] is an open source user friendly
interface for simple use of multiple search engines (http://
searchgui.googlecode.com).
24. Software: PeptideShaker [23] is an open source user friendly
interface for the interpretation of results from multiple search
engines (http://peptide-shaker.googlecode.com).

3 Methods

3.1 Glycopeptide Protocol modified from [19] and [20].

Enrichment
1. Purify and concentrate the CSF sample using 3 kDa ultracen-
trifugation filters, pre-cleansed with 1 mL NOG. Add CSF
sample + 1 mL deionized water (MQ) and spin at 3000 × g for
45 min at 4 °C. Add 1 mL MQ and spin for approximately 1
h, or until there is between 50 and 100 μL left in the filter.
Transfer to Eppendorf tube and concentrate to ≈15 μL.
278 Astrid Guldbrandsen et al.

2. Add 135 μL denaturation buffer (for 100–1000 μg protein)

and vortex.
3. Add TCEP to a final concentration of 10 mM. Incubate with
shaking for 1 h at 37 °C.
4. Add IAA to a final concentration of 12 mM. Incubate with
shaking for 30 min in the dark at 20 °C.
5. Add 1 mL 100 mM Ambic to get the urea concentration
below 1 M.
6. Add trypsin in a 1:50 trypsin to protein ratio. Incubate with
gentle shaking for 12–16 h at 37 °C.
7. Acidify sample by adding approximately 12 μL 100 % FA, drop
by drop. Keep the lid open to avoid pressure building up inside
the tube.
8. Perform cleanup at 4 °C using Oasis HLB 10 mg (30 μm)
plates. Condition column with 1 mL 50 % ACN–0.1 % TFA,
wash × 2 with 1 mL 0.1 % TFA, all at 200 × g for 1 min, before
addition of sample and spinning at 150 × g for 3 min. Wash
again using 1 mL 0.1 % TFA × 3 and elute with 200 μL 50 %
ACN–0.1 % TFA × 2, all at 200 × g for 1 min.
9. Transfer the sample to a new Eppendorf tube and concentrate
to dryness.
10. Reconstitute the sample in 400 μL 0.1 % TFA and vortex.
11. Add sodium periodate to a final concentration of 10
mM. Incubate with shaking for 1 h at 20 °C in the dark.
12. Repeat step 8, but use 80 % ACN–0.1 % TFA for conditioning
and elution. Leave the sample in the Oasis collection plate.
13. Vortex the hydrazide modified magnetic beads and pipet 133
μL (4 mg) to a new tube. Wash with 1 mL 80 % ACN–0.1 %
TFA for 5 min with extensive shaking (1200 rpm should be
used for all bead incubations, see Notes 1 and 2).
14. Add the peptides from the Oasis well to the washed beads and
incubate overnight at 20 °C.
15. Spin the sample and save supernatant (contains the unbound
peptides) for future analysis.
16. Wash the beads for 5 min with 1 mL of the following:
(a) 80 % ACN–0.1 % TFA × 3
(b) Denaturation buffer × 3
(c) 100 % DMF × 3 (see Note 3)
(d) 100 mM Ambic × 3
17. Add 100 μL 100 mM Ambic to the beads.
18. Add 1.5 μL PNGase F enzyme. Incubate overnight at 37 °C.
19. Collect supernatant containing the released deglycosylated
peptides into new Eppendorf tubes.
 A Simple Workflow for Large Scale Shotgun Glycoproteomics 279

20. Add 200 μL 100 mM Ambic to the beads and wash for 5 min
at 20 °C.
21. Collect the supernatant and pool with the peptides collected
in step 19.
22. Acidify the sample by adding 7 μL 5 M HCl, drop by drop.
Keep lid open to avoid pressure building up inside the tube.
23. Add 200 μL 0.1 % FA and vortex.
24. Perform clean-up at 4 °C as is described in steps 8 and 12, but
this time use Oasis μElution plates (see Note 4) and 500 μL 80
% ACN–0.1 % FA for conditioning and elution and 500 μL 0.1
% FA for washing, and place the sample tube in the magnetic
rack before transfer to the Oasis plate to remove any remain-
ing beads. Elute with 150 μL × 2.
25. Transfer sample (300 μL) to a new Eppendorf tube, concen-
trate to dryness and freeze at −80 °C until LC-MS analysis.
26. Dissolve in appropriate solvent and volume for LC-MS analy-
sis (see Note 5).

3.2 Data The data interpretation consists of two main parts: (1) match the
Interpretation spectra to a database, and (2) interpret the matches to infer pro-
teins and glycosylation sites. For the first part, so-called search
engines are used, retrieving a list of Peptide Spectrum Matches
(PSMs). In the second, the PSMs are assembled into inferred pep-
tides and proteins, the quality of the identification results is evalu-
ated in order to limit the prevalence of false positive hits, and
post-translational modification (PTM) sites are inferred. More
details on the identification process can be found in the following
reviews [26–28].
In this chapter, the above task is demonstrated using user
friendly open source software, SearchGUI (version 1.20.8) and
PeptideShaker (version 0.33.6). These tools notably present the
advantage to support multiple search engines for PTM studies in a
user friendly environment. For more details on how to operate
these tools, please refer to the respective free tutorials [29] (http://
compomics.com/bioinformatics-for-proteomics). Note that the
concepts introduced here can be transposed to most proteomics
applications, like OpenMS [30, 31], the TransProteomic Pipeline
(TPP) [32], or MaxQuant [33].

3.3 Database Search 1. After starting SearchGUI, Click Add and select the spectrum
files to search in the Input & Output panel at the top of the
dialog shown in Fig. 1. Here, spectrum files consist of peak
lists of the original MS2 spectra in the mgf format (http://
www.matrixscience.com/help/data_file_help.html#GEN).
To convert raw data to the mgf format it is recommended to
use ProteoWizard [34]. In this experiment, fractionation has
been performed, so there is a total of 20 files to be searched.
280 Astrid Guldbrandsen et al.

Fig. 1 SearchGUI main dialog. The main dialog appearing when the tool is started. Spectrum files, search set-
tings, and output folder have to be loaded and specified. The search engines shown are all automatically
selected, but can be unchecked. The search can be automatically processed in PeptideShaker if this option is
selected

2. Edit the Search Settings or Load an already saved search settings

file. As displayed in Fig. 2, the following input is required: (a)
choose a Database (FASTA) file; here, the human comple-
ment of the UniProtKB/Swiss-Prot database [35] available
from the UniProt website (www.uniprot.org). SearchGUI
then proposes to add decoy sequences, press yes. (b) Add fixed
and variable PTMs by selecting the relevant modifications and
pressing ‘≪’ to add to the appropriate list; here we use carb-
amidomethyl c as fixed, and oxidation of m and deamidation of
n as variable modifications. (c) Set the Protease & Fragmentation
settings; Enzyme: Trypsin, Precursor Mass Tolerance: 10 ppm,
Fragment Mass Tolerance: 0.7 Da and Max Missed Cleavages
(by trypsin): 2, and for the rest keep the defaults. For more
information on how to set the search parameters, please see
[36]. Save the settings file for future reuse in other searches.
3. Select the Output Folder where the search output will be
stored by pressing Browse and navigating to the desired folder.
4. (a) It is possible to directly open the project in PeptideShaker
after the search. To do this, check the box under Post Processing.
A window will appear allowing the setting of the PeptideShaker
parameters. Under Project Details give a Project Name, a
 A Simple Workflow for Large Scale Shotgun Glycoproteomics 281

Fig. 2 SearchGUI search settings dialog. This dialog appears when editing the search settings, the parameters
used in this experiment are selected. The database (FASTA file) is the human complement of the UniProtKB/
Swiss-Prot database available from the UniProt website (www.uniprot.org)

Sample Name, and select the Ensembl [37] species by pressing

Edit; in this case Vertebrates as species type and Human (Homo
Sapiens) [Ensembl 76] as species, where 76 is the version of
Ensembl. Then select the folder to save the output of the
search by pressing Browse and browsing to the desired folder.
or
(b) It is also possible to first do the search and then later
load the results in PeptideShaker. Then, leave the box under
Post Processing unchecked.
5. Press Start the Search!.
6. While the search is running, a dialog shows updates on the
progress of the search. When finished, the search output is
written to the output folder in the form of a zipped file con-
taining the result file of every search engine, and which can be
loaded in PeptideShaker (see Note 6).
3.4 Glycosylation Steps 1–4 should be skipped if automatic post-processing in
Site Identification PeptideShaker was selected in the previous section.
1. If the data was not directly processed in PeptideShaker, start
PeptideShaker, press New Project and give a name for Project
Reference, Sample Name, and edit species (see previous
section).
282 Astrid Guldbrandsen et al.

2. Browse to find and select the correct identification file(s), i.e.,

SearchGUI output file(s). If not automatically selected when
identification files are loaded, also select spectrum file(s) (mgf)
and database file (FASTA).
3. Edit Search Settings and Import Filters if not automatically
loaded and leave Preferences to default.
4. Press Load Data! and a dialog appears showing updates on the
progress of the project creation.
5. Upon completion, the main display of PeptideShaker allows
the browsing of the proteomics dataset, as displayed in Fig. 3.
Note that glycosylation sites are highlighted in the peptide
and protein sequences using the PTM color coding set when
editing the search parameters. An enlarged Spectrum &
Fragment Ions window for a selected glycopeptide in the data-
set is displayed in Fig. 4.

3.5 Data Export Project features and result reports containing the possible glycosyl-
ation sites (at the protein, peptide, and PSMs level) can be exported
by pressing Export → Identification Features. In the Export Features
window displayed in Fig. 5 the type of report to export can be
chosen and custom reports created (see Notes 7 and 8).

4 Notes

1. All following washes/incubations with beads must be done

with extensive shaking (1200 rpm) to avoid beads depositing
at the bottom of the tube. After incubation, spin down the
tube briefly to collect beads that might have been stuck in the
lid during shaking. Then use the magnetic rack for bead–
supernatant separation. Wait for all beads to gather and get
attached to the back of the tube where the magnet is before
pipetting the supernatant gently without inducing stress on
the beads.
2. If using non-magnetic Macroporous beads, pipet 50 μL, wash
with deionized water and spin at 13,000 × g.
3. DMF is toxic and teratogenic (dangerous for the developing
embryo/fetus), so it should be handled with care and suitable
protection, and not by pregnant women.
4. Oasis plates with lower binding capacity (μElution) are used
because the amount of peptides is substantially lower after gly-
copeptide enrichment.
5. It is suggested to reconstitute in 10 μL 3 % ACN–5 % FA, and
to inject 5 μL for 250 μg starting material and 2 μL for 1 mg
starting material. However, injection volume depends on the
sensitivity of the instrument.
Fig. 3 PeptideShaker’s main display. The overview panel of PeptideShaker displays the result in a top down fashion upon data import and processing. The top table
lists all identified proteins and their details. Below, the upper left table lists all peptide matches identified for the selected protein, and under it a table lists the
peptide-spectrum matches (PSMs) for the selected peptide. The spectrum annotated with fragment ions deduced from the selected PSM is displayed to the bottom
A Simple Workflow for Large Scale Shotgun Glycoproteomics

right. At the bottom of the display, the protein sequence is displayed with the identified peptides color coded. Note that peptides can be navigated by clicking the
sequence. Deamidated asparagines (glycosylation sites) are highlighted in the peptide and protein sequences using the color coding from the search parameters.
More details on the project are available via the other tabs in the upper right corner
283
284 Astrid Guldbrandsen et al.

Fig. 4 PeptideShaker’s Spectrum & Fragment Ions display. The Spectrum & Fragment Ions panel displays
information on the annotation of the spectrum based on the selected peptide. At the top left, the intensity of
the fragment ions annotated on the spectrum is illustrated with bars at the possible fragmentation site on the
peptide sequence, where the modification is color coded. The intensities of b and y ions are in blue and red,
respectively. Here, the deamidation of the asparagine on the ENAT motif is displayed in brown and clearly
flanked with fragment ions. At the top center, a histogram shows the respective shares of annotated and not
annotated peaks, in green and grey, respectively, using intensity bins. At the top right, the m/z deviation of
every fragment ion is plotted against the fragment ion m/z. Below, the spectrum is displayed with the anno-
tated peaks in red. It is possible to customize the annotation of the spectrum using the menu at the bottom

6. If following Subheading 3.3, step 4b, go to the folder where

the zipped SearchGUI output file is stored and select the iden-
tification files in PeptideShaker.
7. Further (manual or automatic) validation is required to con-
firm if the detected sites are real glycosylation sites, see [18]
Supplementary File 1C.
8. It is crucial to remember that this approach identifies deami-
dated asparagine residues in the glycosylation patterns. While
these are highly likely to be former glycosylation sites, there is
no direct measurement of the glycosylation event, and no
information on the structure of the glycan.
 A Simple Workflow for Large Scale Shotgun Glycoproteomics 285

Fig. 5 PeptideShaker export features dialog. This dialog is available from the Export → Identification Features.
Under the panel Reports, several default reports can be selected and exported, while custom made reports can
be created and edited by right-clicking the table

Acknowledgements

A.C.K. is supported by the Kristian Gerhard Jebsen Foundation.

H.B. is supported by the Research Council of Norway.

References
1. Gevaert K, Van Damme P, Ghesquiere B et al ical insights and biomedical implications. Cell
(2007) A la carte proteomics with an emphasis on Mol Life Sci 64:2133–2152
gel-free techniques. Proteomics 7:2698–2718 6. Varki A (1993) Biological roles of oligosaccha-
2. Zhang H, Li XJ, Martin DB et al (2003) rides: all of the theories are correct.
Identification and quantification of N-linked Glycobiology 3:97–130
glycoproteins using hydrazide chemistry, stable 7. Roth J (2002) Protein N-glycosylation along
isotope labeling and mass spectrometry. Nat the secretory pathway: relationship to organ-
Biotechnol 21:660–666 elle topography and function, protein quality
3. Gamblin DP, Scanlan EM, Davis BG (2009) control, and cell interactions. Chem Rev
Glycoprotein synthesis: an update. Chem Rev 102:285–303
109:131–163 8. Sato Y, Endo T (2010) Alteration of brain gly-
4. Shental-Bechor D, Levy Y (2008) Effect of gly- coproteins during aging. Geriatr Gerontol Int
cosylation on protein folding: a close look at 10 Suppl 1:S32–S40
thermodynamic stabilization. Proc Natl Acad 9. Ruggeri ZM, Mendolicchio GL (2007)
Sci U S A 105:8256–8261 Adhesion mechanisms in platelet function. Circ
5. Sola RJ, Rodriguez-Martinez JA, Griebenow K Res 100:1673–1685
(2007) Modulation of protein biophysical 10. Berger MS, Locher GW, Saurer S et al (1988)
properties by chemical glycosylation: biochem- Correlation of c-erbB-2 gene amplification and
286 Astrid Guldbrandsen et al.

protein expression in human breast carcinoma interface for simultaneous OMSSA and
with nodal status and nuclear grading. Cancer X!Tandem searches. Proteomics 11:996–999
Res 48:1238–1243 23. Vaudel M, Burkhart JM, Zahedi RP et al
11. Hudziak RM, Schlessinger J, Ullrich A (1987) (2015) PeptideShaker enables reanalysis of
Increased expression of the putative growth MS-derived proteomics data sets. Nature bio-
factor receptor p185HER2 causes transforma- technology 33:22–24
tion and tumorigenesis of NIH 3T3 cells. Proc 24. Vizcaino JA, Deutsch EW, Wang R et al (2014)
Natl Acad Sci U S A 84:7159–7163 ProteomeXchange provides globally coordi-
12. Vogelzang NJ, Lange PH, Goldman A et al nated proteomics data submission and dissemi-
(1982) Acute changes of alpha-fetoprotein and nation. Nat Biotechnol 32:223–226
human chorionic gonadotropin during induc- 25. Martens L, Hermjakob H, Jones P et al (2005)
tion chemotherapy of germ cell tumors. Cancer PRIDE: the proteomics identifications data-
Res 42:4855–4861 base. Proteomics 5:3537–3545
13. Bosl GJ, Lange PH, Fraley EE et al (1981) 26. Vaudel M, Sickmann A, Martens L (2012)
Human chorionic gonadotropin and alphafeto- Current methods for global proteome iden-
protein in the staging of nonseminomatous tes- tification. Expert Rev Proteomics 9:
ticular cancer. Cancer 47:328–332 519–532
14. Thompson DK, Haddow JE (1979) Serial 27. Nesvizhskii AI (2010) A survey of computa-
monitoring of serum alpha-fetoprotein and tional methods and error rate estimation pro-
chorionic gonadotropin in males with germ cedures for peptide and protein identification
cell tumors. Cancer 43:1820–1829 in shotgun proteomics. J Proteomics
15. Catalona WJ, Richie JP, Ahmann FR et al 73:2092–2123
(1994) Comparison of digital rectal examina- 28. Chalkley RJ, Clauser KR (2012) Modification
tion and serum prostate specific antigen in the site localization scoring: strategies and perfor-
early detection of prostate cancer: results of a mance. Mol Cell Proteomics 11:3–14
multicenter clinical trial of 6,630 men. J Urol 29. Vaudel M, Venne AS, Berven FS et al (2014)
151:1283–1290 Shedding light on black boxes in protein iden-
16. Canney PA, Moore M, Wilkinson PM et al tification. Proteomics 14:1001–1005
(1984) Ovarian cancer antigen CA125: a pro- 30. Kohlbacher O, Reinert K, Gropl C et al (2007)
spective clinical assessment of its role as a TOPP—the OpenMS proteomics pipeline.
tumour marker. Br J Cancer 50:765–769 Bioinformatics 23:e191–e197
17. Topol EJ, Byzova TV, Plow EF (1999) Platelet 31. Bertsch A, Gropl C, Reinert K et al (2011)
GPIIb-IIIa blockers. Lancet 353:227–231 OpenMS and TOPP: open source software for
18. Guldbrandsen A, Vethe H, Farag Y et al (2014) LC-MS data analysis. Methods Mol Biol 696:
In-depth characterization of the cerebrospinal 353–367
fluid proteome displayed through the CSF 32. Deutsch EW, Mendoza L, Shteynberg D et al
Proteome Resource (CSF-PR). Mol Cell (2010) A guided tour of the Trans-Proteomic
Proteomics 13(11):3152–63 Pipeline. Proteomics 10:1150–1159
19. Tian Y, Zhou Y, Elliott S et al (2007) Solid- 33. Cox J, Mann M (2008) MaxQuant enables
phase extraction of N-linked glycopeptides. high peptide identification rates, individualized
Nat Protoc 2:334–339 p.p.b.-range mass accuracies and proteome-
20. Berven FS, Ahmad R, Clauser KR et al (2010) wide protein quantification. Nat Biotechnol
Optimizing performance of glycopeptide cap- 26:1367–1372
ture for plasma proteomics. J Proteome Res 34. Kessner D, Chambers M, Burke R et al (2008)
9:1706–1715 ProteoWizard: open source software for rapid
21. Gonzalez J, Takao T, Hori H et al (1992) A proteomics tools development. Bioinformatics
method for determination of N-glycosylation 24:2534–2536
sites in glycoproteins by collision-induced 35. Apweiler R, Bairoch A, Wu CH et al (2004)
dissociation analysis in fast atom bombard- UniProt: the Universal Protein knowledge-
ment mass spectrometry: identification of the base. Nucleic Acids Res 32:D115–D119
positions of carbohydrate-linked asparagine 36. Vaudel M, Burkhart JM, Sickmann A et al
in recombinant alpha-amylase by treatment (2011) Peptide identification quality control.
with peptide-N-glycosidase F in 18O-labeled Proteomics 11:2105–2114
water. Anal Biochem 205:151–158 37. Flicek P, Amode MR, Barrell D et al (2014)
22. Vaudel M, Barsnes H, Berven FS et al (2011) Ensembl 2014. Nucleic Acids Res 42:
SearchGUI: an open-source graphical user D749–D755
 Chapter 21

Systemic Analysis of Regulated Functional Networks

Luis Francisco Hernández Sánchez, Elise Aasebø, Frode Selheim,
Frode S. Berven, Helge Ræder, Harald Barsnes, and Marc Vaudel

Abstract
In biological and medical sciences, high throughput analytical methods are now commonly used to investigate
samples of different conditions, e.g., patients versus controls. Systemic functional analyses emerged as a
reference method to go beyond a list of regulated compounds, and identify activated or inactivated bio-
logical functions. This approach holds the promise for a better understanding of biological systems, of the
mechanisms involved in disease progression, and thus improved diagnosis, prognosis, and treatment. In
this chapter, we present a simple workflow to conduct pathway analyses on biological data using the freely
available Reactome platform (http://www.reactome.org).

Key words Pathway analysis, Data interpretation, Functional proteomics

1 Introduction

In systems biology, thousands of compounds—metabolites, genes,

transcripts, proteins, etc.—are studied in a global approach to
investigate the biological processes differentially triggered between
samples [1, 2]. Identified and quantified compounds are mapped
against databases of known interactions and functions, and differ-
entially expressed pathways extracted, providing the user with
insight on the significantly regulated biological processes [3, 4].
Two factors are crucial for the success of this procedure [5, 6]: (1)
the quality of the functional knowledgebase used as reference for
the study, and (2) the accuracy of the matching of experimental
data with the knowledge bases. Notably, functional knowledge is
not as strongly established as other resources for omics studies, like
gene or protein databases, and is hence more subject to changes
and updates [5].
Several knowledge bases exist, along with tools to query them,
allowing the functional interpretation of large scale biological
studies [7]. For example, the commercial resource QIAGEN’s
Ingenuity® Pathway Analysis (IPA®, QIAGEN Redwood City,

Jörg Reinders (ed.), Proteomics in Systems Biology: Methods and Protocols, Methods in Molecular Biology, vol. 1394,
DOI 10.1007/978-1-4939-3341-9_21, © Springer Science+Business Media New York 2016

287
288 Luis Francisco Hernández Sánchez et al.

www.qiagen.com/ingenuity) allows the query of large omics

datasets and pathway analyses in a user friendly environment.
However, the dynamic nature of the functional knowledge
resources, always subject to evolution, makes it crucial to docu-
ment the changes which can affect the analysis. Since this task is
impossible in the case of private databases, open resources appear
as an alternative of choice [5].
Among freely available academic resources, the most
encountered are the Kyoto Encyclopedia of Genes and Genomes
[8, 9] (KEGG, http://www.genome.jp/kegg), WikiPathways
[10] (http://www.wikipathways.org), and Reactome [11]
(http://www.reactome.org). These resources allow the brows-
ing of pathways and analysis of data from their respective web-
sites. KEGG can also be queried via the metabolic and
physiological potential evaluator (MAPLE) [12] and Reactome
using a plugin in Cytoscape [13]. Here, we present the use of
Reactome for functional analyses. First, browsing pathways on
the website are illustrated taking the JAK/STAT pathway [14]
as an example. Second, the analysis of quantitative data is dem-
onstrated with the processing of a quantitative proteomic data-
set of different acute myeloid leukemia (AML) derived cell
lines [15].

2 Material

1. The dataset here used for illustrative purposes is freely

available through the ProteomeXchange [16] consortium via
the PRIDE [17] partner repository under the accession num-
ber PXD000441. Here, the relative quantification of proteins
from the cell line Molm-13 to an internal standard of five
AML cell lines is used. Details can be found in Supplementary
Table S2 of [15].
2. Reactome can be used directly from the Reactome website
(http://www.reactome.org) or via the dedicated Cytoscape [13]
plugin. Here, the use of the online version is demonstrated.

3 Methods

This section describes how to browse for a specific pathway and

then navigate it, using the JAK/STAT pathway as an example.
1. Go to the homepage of Reactome (http://www.reactome.org)
(Fig. 1).
2. In the search field type in the keywords related to the pathway
you are looking for, here: “JAK STAT”. The results of the
search are then displayed (Fig. 2).
 Systemic Analysis of Regulated Functional Networks 289

Fig. 1 Reactome homepage

3. In the results screen select the option corresponding to the

pathway of interest for more details. Select: Phosphorylation of
STAT1 by JAK kinases (Homo sapiens) (Fig. 3).
4. Click the “+” symbol to show the hierarchy of pathways cor-
responding to this function (Fig. 4). Note that you can
select the pathway or reaction of interest in the Pathway
Browser.
5. Go back to the Reactome homepage. Then go to the Pathway
Browser by clicking the Browse Pathways button (Fig. 5). As an
alternative, you can select the Pathway Browser option in the
Tools menu (Fig. 6). The Pathway Browser will appear as dis-
played (Fig. 7).
6. The Hierarchy Panel on the left shows the available Reactome
pathway topics sorted alphabetically. Find the topic associated
with this pathway, i.e., Immune System (Homo sapiens), and
click it to show the pathway diagram.
7. You can go deeper in the events hierarchy to find the sub-
pathways related to JAK/STAT by clicking the “+” symbol
Fig. 2 Reactome search results of JAK STAT

Fig. 3 Details of pathway Phosphorylation of STAT1 by JAK kinases (Homo sapiens)

Fig. 4 Events hierarchy for the pathway Phosphorylation of STAT1 by JAK kinases (Homo sapiens)

Fig. 5 Button on the homepage to go to the Pathway Browser of Reactome

292 Luis Francisco Hernández Sánchez et al.

Fig. 6 The option on the Tools menu at the homepage to go to the Pathway Browser of Reactome

to the left of each pathway or sub-pathway. Here select

Cytokine Signaling in Immune system to show the sub-pathway
diagram (Fig. 8).
8. It is also possible to explore a sub-pathway by double clicking
the appropriate diagram box, double click the Interferon
Signaling box and then Interferon gamma signaling. The
pathway diagram is now updated to show the selected sub-
pathway and the hierarchy panel highlights the name of the
pathway with its currently selected sub-pathways.
9. Select the desired pathway, Phosphorylation of STAT1 by JAK
kinases, the pathway diagram focuses on that reaction high-
lighting in blue the components associated with that func-
tion (Fig. 9). Also note that the details panel at the bottom
of the screen shows information about the selected pathway
or reaction.
10. Click the diagram objects to display more information about
specific elements of the reaction in the details panel.
11. Zoom in or out using the “+” and “–” symbols in the upper
left corner of the diagram panel.
12. At any time, you can click and drag the pathway diagram to
move to other areas, or use the small arrow buttons at the top
left corner.
Systemic Analysis of Regulated Functional Networks 293

Fig. 7 The Pathway Browser of Reactome

 294
Luis Francisco Hernández Sánchez et al.

Fig. 8 The Pathway diagram for Cytokine Signaling in Immune system

Fig. 9 The reaction diagram for Phosphorylation of STAT1 by JAK kinases
Systemic Analysis of Regulated Functional Networks
295
296 Luis Francisco Hernández Sánchez et al.

13. In the upper right corner, click the down arrow button to
show the Diagram Key, describing the shapes corresponding to
every type of entity in the diagram. To hide the key, click the
arrow button again. Note that proteins are shown as rectangles
with rounded corners.
14. In the Hierarchy Panel to the left, select the reaction called
Transphosphorylation of JAK1 inside the same sub-pathway
Interferon gamma signaling.
15. Right click on the protein PTPN6, and select the option Other
pathways to show the diagram of other pathways related to that
protein. For the protein uniquely involved in this pathway, the
label No other pathways is shown.
16. On the Hierarchy Panel on the left, select Binding of STAT1 to
p-IFNGR1. Right click on the protein STAT1-1 and choose
Display interactors to show the compounds interacting with
this protein in this pathway. Here 10 out of 52 interactors
associated to this protein are highlighted with new entity boxes
with blue thick borders. Note the small white square at the top
right corner of the entity box of the protein indicating the
number of interactors (Fig. 10).
17. You can switch to other pathway diagrams at any time by
selecting another pathway, sub-pathway or reaction name in
the hierarchy panel. The details panel at the bottom will be
updated according to what is currently selected in the hierar-
chy panel or the pathway diagram.
This section describes how to analyze quantitative data using
Reactome with the dataset material indicated in Subheading 2,
item 1. First, the analysis of the list of protein accessions will be
used to demonstrate how to find pathways of interest, a use case
relevant to both qualitative and quantitative datasets. Second, the
quantitative information will be provided to Reactome along with
the protein identifiers.
1. From the Reactome homepage (Fig. 1), click Analyze Data to
show the analysis tools (Fig. 11).
2. Prepare a data file containing the list of proteins according to
the required format:
(a) Create a new text file and name it protein_list.txt.
(b) In Supplementary Table S2 of [15], go to the worksheet
called Results_Merged_Five_and_four_IS. There you will
find the table entitled Supplementary Table 2: Merged data
from the five and four cell lines experiments.
(c) Select the protein accessions, here the column called
Protein IDs, along with the header, copy the data into the
text file and save the file.
 Systemic Analysis of Regulated Functional Networks 297

Fig. 10 The interactors of protein STAT1-1 in the Phosphorylation of STAT1 by JAK kinases reaction

(d) Change the first row of the file from Protein IDs to #GBM
Uniprot (see Note 1 and 2).
(e) Delete any empty lines after the protein list.
3. Upload the data file with the set of proteins to analyze:
(a) First click the Browse button to select your protein_list.txt
file.
(b) Make sure that the checkbox Project to human is selected,
as the identifiers to analyze are related to human
pathways.
(c) Click Analyze to upload the file and start the analysis of the
inserted data.
4. As an alternative, you can simply paste the information in the
textbox within the Analysis Tools section.
(a) In the Analyze Data screen, click Click here to paste you
data or try example data sets… to display the input field.
At the right you can also select example datasets.
 298
Luis Francisco Hernández Sánchez et al.

Fig. 11 The Data Analysis screen with the Analysis tools of Reactome
 Systemic Analysis of Regulated Functional Networks 299

(b) Paste the set of proteins into the text field.

(c) Click on the Analyze button in the lower right corner.
5. Review the results in the details panel at the bottom of the
screen, showing the pathways containing at least one protein
from the dataset. There are columns with different types of
information, for example you can see pathway names and how
many entities were found in those pathways, as well as the con-
fidence in the pathway identification (Fig. 12).
6. Choose the pathway Translation. In the diagram, the box enti-
ties are colored depending on the number of entities matched.
Encapsulated pathways are represented by black border rect-
angles (Fig. 13).
7. Double click on Eukaryotic Translation Elongation to go
deeper in the diagram and show the sub-pathway of interest
(Fig. 14).
8. Sets of proteins are indicated by double border rectangle with
rounded corners. Right click on EEF1A1-like proteins and
choose the option Display Participating Molecules. This will
show a small table with the components of the set colored in
yellow (Fig. 15).
In the following, the procedure above will be repeated, but
now including the quantitative information (Figs. 16, 17, 18, 19).
1. Prepare the data file to be uploaded in the Analysis screen of
Reactome.
(a) In Supplementary Table S2 of [15], go to the worksheet
called Results_Merged_Five_and_four_IS. There you will
find the table entitled Supplementary Table 2: Merged data
from the five and four cell lines experiments.
(b) Copy the columns Protein Ids and Molm-13 under Five cell
lines as IS, and paste them in a new worksheet. Rename the
first column from Protein IDs to #Probeset (see Note 3 and 4).
(c) Create a new text file called expression_data.txt and copy
paste the two columns into the new text file.
2. Reopen the Analysis tools panel by clicking on the button rep-
resenting a loupe over a pathway at the top of the screen
(Fig. 20).
3. Click the Browse button and select the expression_data.txt file.
Next, click the Analyze button. A table presenting the results of
the analysis appears in the Details Panel upon completion. Note
that the first eight columns are the same as in the previous exam-
ple, but the table now also includes the quantitative results.
4. You can explore any pathway using the Hierarchy Panel or the
Details Panel. Entities in the diagram are colored according to
 300
Luis Francisco Hernández Sánchez et al.

Fig. 12 The Analysis results of the protein sample data set

 Systemic Analysis of Regulated Functional Networks

Fig. 13 The diagram for Translation pathway partially colored in yellow to indicate the percentage of structures in the sample data set found in this pathway
301
 302
Luis Francisco Hernández Sánchez et al.

Fig. 14 The diagram for Eukaryotic Translation Elongation sub-pathway with partially colored entity boxes
 Systemic Analysis of Regulated Functional Networks 303

Fig. 15 The table of participating components of the protein set EEF1A1-like proteins highlighting in yellow the
table cells of the components found in the sample protein data set

the quantitative results, ranging from blue to yellow for the

lower and upper ratios, respectively. Only the entities with
numerical data associated will be colored.
5. In the Hierarchy Panel select the Metabolism pathway. Note
that next to its name, there is a number saying how many mol-
ecules were identified in the submitted data, in this case 343
out of 1588. Then select the sub-pathway called Metabolism of
lipids and lipoproteins, then select Lipid digestion, mobilization,
and transport and finally Lipoprotein metabolism (Fig. 21).
6. Note that some complex entities are partially colored. This
means that only some participating molecules of that complex
where present in the dataset.
7. Right click on the complex called ApoB-48:TG:PL. Then select
Display Participating Molecules. Note, that only one out of
three molecules where present in the dataset, the rectangle
representing the complex therefore has one third of its area
colored in gray. The gray is due to the numerical value of the
protein of accession P04114 (Fig. 22).
 304
Luis Francisco Hernández Sánchez et al.

Fig. 16 Analysis Tools screen with expression data

 Systemic Analysis of Regulated Functional Networks

Fig. 17 Diagram panel with analysis results coloring entities according to the submitted numerical data
305
306 Luis Francisco Hernández Sánchez et al.

Fig. 18 Expression data shown by hovering over a protein

Fig. 19 Colored table of components of a complex entity

4 Notes

1. Proteomics dataset result files compatible with Reactome can

be easily exported from most proteomic software like
MaxQuant [18], OpenMS [19], the Trans-Proteomic Pipeline
(TPP) [20], or PeptideShaker [21]. It is however recommended
 Systemic Analysis of Regulated Functional Networks 307

Fig. 20 The button to show the Analysis Tools

to post-process the results, for example conducting normaliza-

tion and imputation of missing values. This can be easily con-
ducted using the Perseus interface, available from the
MaxQuant website (www.maxquant.org).
2. Reactome can be operated on a wide variety of omics datasets:
metabolomics, genomics, transcriptomics, and proteomics.
Data can be loaded separately or together in a unified dataset.
3. Make sure that column headers are recognized by Reactome.
When the dada set is composed of only one column of data,
Reactome can recognize different types of identifiers such as
UniProt accessions for proteins and ChEBI IDs for small mol-
ecules. Quantitative information will be ignored if not indi-
cated by the correct header (see Subheading 3).
4. Reactome also recognizes HGNC gene symbols, ENSEMBL
IDs for DNA/RNA molecules, HUGO gene symbols,
GenBank/EMBL/DDBJ, RefPep, RefSeq, EntrezGene, MIM,
InterPro, EnsEMBL protein, EnsEMBL gene, EnsEMBL tran-
script, and some Affymetrix and Agilent probe IDs.
 308
Luis Francisco Hernández Sánchez et al.

Fig. 21 Diagram of the Lipoprotein metabolism pathway after the analysis of the submitted data
 Systemic Analysis of Regulated Functional Networks 309

Fig. 22 Component molecules of the ApoB-48:TG:PL complex

Acknowledgements

F.S.B. and F.S. acknowledge support from the Norwegian Cancer

Society. H.B. is supported by the Research Council of Norway.

References
1. Mouse Genome Sequencing Consortium, outstanding challenges. PLoS Comput Biol 8,
Waterston RH, Lindblad-Toh K et al (2002) e1002375
Initial sequencing and comparative analysis of 7. Vizcaino JA, Mueller M, Hermjakob H et al
the mouse genome. Nature 420:520–562 (2009) Charting online OMICS resources: a
2. Rosen R (1968) A means toward a new holism. navigational chart for clinical researchers.
Science 161:34–35 Proteomics Clin Appl 3:18–29
3. Barsnes H, Martens L (2013) Crowdsourcing 8. Kanehisa M (1997) A database for post-
in proteomics: public resources lead to better genome analysis. Trends Genet 13:375–376
experiments. Amino Acids 44:1129–1137 9. Ogata H, Goto S, Sato K et al (1999) KEGG:
4. Vaudel M, Sickmann A, Martens L (2014) Kyoto Encyclopedia of Genes and Genomes.
Introduction to opportunities and pitfalls in Nucleic Acids Res 27:29–34
functional mass spectrometry based pro- 10. Pico AR, Kelder T, van Iersel MP et al (2008)
teomics. Biochim Biophys Acta 1844:12–20 WikiPathways: pathway editing for the people.
5. Muller T, Schrotter A, Loosse C et al (2011) PLoS Biol 6, e184
Sense and nonsense of pathway analysis software 11. Croft D, O’Kelly G, Wu G et al (2011)
in proteomics. J Proteome Res 10:5398–5408 Reactome: a database of reactions, pathways
6. Khatri P, Sirota M, Butte AJ (2012) Ten years and biological processes. Nucleic Acids Res
of pathway analysis: current approaches and 39:D691–D697
310 Luis Francisco Hernández Sánchez et al.

12. Takami H, Taniguchi T, Moriya Y et al (2012) nated proteomics data submission and dissemi-
Evaluation method for the potential function- nation. Nat Biotechnol 32:223–226
ome harbored in the genome and metage- 17. Martens L, Hermjakob H, Jones P et al (2005)
nome. BMC Genomics 13:699 PRIDE: the proteomics identifications data-
13. Shannon P, Markiel A, Ozier O et al (2003) base. Proteomics 5:3537–3545
Cytoscape: a software environment for inte- 18. Cox J, Mann M (2008) MaxQuant enables high
grated models of biomolecular interaction net- peptide identification rates, individualized p.p.b.-
works. Genome Res 13:2498–2504 range mass accuracies and proteome-wide protein
14. Kisseleva T, Bhattacharya S, Braunstein J et al quantification. Nat Biotechnol 26:1367–1372
(2002) Signaling through the JAK/STAT 19. Bertsch A, Gropl C, Reinert K et al (2011)
pathway, recent advances and future chal- OpenMS and TOPP: open source software for
lenges. Gene 285:1–24 LC-MS data analysis. Methods Mol Biol
15. Aasebo E, Vaudel M, Mjaavatten O et al 696:353–367
(2014) Performance of super-SILAC based 20. Deutsch EW, Mendoza L, Shteynberg D et al
quantitative proteomics for comparison of dif- (2010) A guided tour of the Trans-Proteomic
ferent acute myeloid leukemia (AML) cell Pipeline. Proteomics 10:1150–1159
lines. Proteomics 14(17-18):1971–6 21. Vaudel M, Burkhart JM, Zahedi RP et al (2015)
16. Vizcaino JA, Deutsch EW, Wang R et al (2014) PeptideShaker enables reanalysis of MS-derived
ProteomeXchange provides globally coordi- proteomics data sets. Nat Biotechnol 33:22–24
 INDEX

A Glycolysis .....................................................................57–73
Glycopeptides .................................................. 163–178, 275,
Adipocytes ....................................................................57, 58 277–279, 282
Affinity purification-mass spectrometry (APMS), ..181–183 Glycoproteomics.......................................................275–284
Antigen presentation ................................................189–208
Assay design .................................................................47–49 H
Assay validation ...................................................... 50–51, 54
High-performance liquid chromatography
B (HPLC) ........................12, 16–18, 20–23, 27, 28,
30, 31, 34, 35, 39, 45, 63, 78, 81, 89, 90, 93, 94,
Biomarkers ..............................................44, 87, 88, 151, 275 96, 97, 104–106, 183, 185, 194, 198, 199, 202,
206, 215, 216
C
Human leukocyte antigen (HLA) ................... 189–192, 194,
Cell signaling pathways ............................................181–186 199, 201, 204, 205
Chemical cross-linking ......................109, 112, 114–115, 124 Hydrophilic interaction liquid chromatography
Chemical proteomics ................................................211–218 (HILIC) ...........................26, 28, 33–35, 40, 152,
Collision-induced dissociation (CID) ............... 8, 60, 97, 98, 155–156, 207
120, 122, 171, 172, 176, 178, 240
Cotton ...................................................................... 152, 160 I
Crop science .............................................................233–241 ImagePrep .........................................130, 131, 137–142, 145
Cryoconserved tissue ........................................ 131, 132, 135 Immobilized metal ion affinity chromatography
Cultivar..................................................... 234, 235, 238–240 (IMAC) ................................................ 88, 90, 94
Immune response .....................................................189–208
D
In-gel digestion ........................113, 116–119, 165, 168–169,
Data interpretation ................................................... 276, 279 171–172, 174–175
Data post-processing ........................................................261 In-solution digestions .......................113–114, 116, 119, 165,
De Novo Sequencing................................................233–241 171, 174, 175
Drug discovery .................................................................212 Isobaric labels ...............................................................3, 248
Isoelectric focusing ......................................... 16–18, 22, 220
E iTRAQ ................................... 3, 15–40, 87, 97, 98, 101, 217,
Enrichment .............................. 2, 3, 30, 32, 33, 39, 152, 158, 219, 248, 270
164–165, 167–169, 172, 177, 183, 212, 247, 261,
K
267, 272, 276–279
Ethyl esterification (EE) ..........................................151–161 Kinases ....................................... 59, 212, 245, 246, 289–292,
295, 297
F
L
Formalin fixed paraffin embedded (FFPE) ............. 130–132,
135–137 Label-free proteomics.......................................................102
Functional Networks ................................................287–307 Linkage-specific ....................................................... 152, 157
Functional proteomics ......................................................287 Liquid chromatography-mass spectrometry
(LC-MS) ...................... 25–28, 30, 35, 48, 50, 55,
G 79–80, 89–91, 94, 95, 97, 98, 114, 119–120, 185,
Gene expression................................................................229 192–194, 199–201, 205–207, 220, 222, 226, 236,
Gene/protein expression ..................................................221 251, 255, 276, 279

Jörg Reinders (ed.), Proteomics in Systems Biology: Methods and Protocols, Methods in Molecular Biology, vol. 1394,
DOI 10.1007/978-1-4939-3341-9, © Springer Science+Business Media New York 2016

311
 PROTEOMICS IN SYSTEMS BIOLOGY: METHODS AND PROTOCOLS
312 Index

M Protein interaction ............................................ 103, 111, 183

Protein quantitation.............................................. 15, 67, 205
Major histocompatibility complex (MHC) ............. 189–192, Proteogenomics ................................................................239
194–205, 207 Proteomics ........................ 2, 10, 15–25, 44, 48, 88, 220–223,
MALDI imaging......................................................129–149 227–229, 246, 272, 277, 279, 282, 306, 307
Mass spectrometry (MS) ................ 2–4, 8, 15, 18, 22–23, 37, Proteomics informed by transcriptomics (PIT) ................221
43, 44, 58, 78, 87, 102, 104, 110, 129–131, 143,
152, 164, 165, 183, 185, 186, 198, 212, 213, 215, Q
218, 221, 223, 224, 226, 234, 255, 276
Quadrupole mass spectrometer .............................. 59, 63, 76
Mass spectroscopy imaging (MSI) ...................................129
Quantification .........................2, 3, 11, 16, 44, 47, 50, 53, 55,
Mathematical models ....................................... 245, 246, 256
58, 60, 61, 87, 88, 101–103, 106, 107, 117, 220,
Matrix application .................................... 130, 134, 137–142
223, 226, 230, 235, 239, 246–250, 252, 253,
Matrix assisted laser desorption/ionization (MALDI), .........
255–257, 261, 288
151–161
Quantitation ....................12, 16, 52, 76, 83, 98, 99, 192, 194,
Melanoma cell ..................................................................106
200, 202, 203, 249
Metabolic control ...............................................................57
Quantitative phosphoproteomics..................................26, 35
Metal oxide affinity chromatography (MOAC) ......... 27, 248
Quantitative proteomics ................................... 1–12, 75, 248
Metaproteomics................................................................221
Mixed-mode solid phase extraction, ................................247 R
Mode of action .................................................................211
Multiple reaction monitoring (MRM) ........................ 75–85, RNA sequencing ......................................................219–230
87–99, 200, 202, 207, 239
Multiplexing ......................................2, 16, 59, 102, 106, 223
S
Selected reaction monitoring (SRM)
N assay design.............................................................47–49
NanoESI MS ....................................164, 165, 171, 172, 176 assay validation ................................................. 50–51, 54
N-glycan release .......................................................152–155 Sepharose .......................... 152–156, 160, 194, 197, 214, 216
N-Glycosylation .......................................................163–178 Shotgun proteomics.......................... 43, 47, 48, 54, 220, 234,
238, 261–272, 276
P Sialic acid (N-acetylneuraminic acid) ............... 152, 157–159
SIS peptides......................................................58, 60, 61, 66,
Pathway Analysis ..............................................................287
67, 72, 73
Pathway modelling ...................................................245–258
Small molecule profiling ...........................................211–218
Peptide ligands ......................................................... 190, 191
Solid phase extraction (SPE) ....................................152–156
Peptide-N-glycosidase F (PNGase F) ..................... 152–155,
Solid-phase extraction (SPE) ...........................11, 27, 34, 39,
157, 159, 164, 276–278
93, 94, 96, 164, 165, 247–249, 276
Peptides ........................ 2, 3, 7–11, 16–18, 20–22, 24, 30–33,
Selected/multiple reaction monitoring
35–40, 43–55, 58–61, 63, 66–73, 75–79, 81–82,
(SRM/MRM).....................43–55, 57–73, 87–99,
88, 90, 106, 107, 110, 111, 116–122, 129–149,
102, 239
161, 165, 167, 168, 172, 175, 177, 178, 186,
Stabilization......................................................................152
190–195, 198–207, 212, 219–222, 224–227, 229,
Stable isotope labeling in cell culture
230, 234–236, 238, 239, 241, 246–252, 256–258,
(SILAC) ...........................2, 87, 97, 98, 101–107,
262, 264, 270, 276–279, 282–284
219, 248, 270
Perseus ............................... 226, 227, 261–264, 270, 272, 307
Subcellular fractionation ...............................................16–17
Phosphatases ............................................................ 245, 246
SunCollect .........................................130, 131, 137–141, 148
Phosphopeptide-enrichment ................26, 27, 32–34, 38–40,
SWATH-MS ........................................................... 102, 103
88, 93, 247–249, 251, 254, 256–258
Systemic analysis ......................................................287–307
Phosphoproteomics ............................................................88
Photo-affinity labeling (PAL) ..........................................111 T
Photo-cross-linking .......................................... 111, 112, 125
Pisum sativum ........................................................... 237, 239 Tandem mass tags (TMT) ...............................3, 7–8, 10, 12,
Polymorphisms ................................................. 189, 234, 239 87, 217, 248, 270
Proteases ............................ 10, 16, 45, 46, 164, 171, 177, 220 Target deconvolution ........................................................212
Protein 3D-structure ................................................109–127 Targeted proteomics ........................................... 47, 192, 221
 PROTEOMICS IN SYSTEMS BIOLOGY: METHODS AND PROTOCOLS
Index
313
Tempo-spatial proteomics ............................................75–85 183–186, 195, 203, 206, 214, 220, 236, 250, 253,
Trypsin ..........................3–6, 9, 10, 16, 18, 20, 26, 30, 37–39, 257, 277, 278, 280
43, 45, 46, 49, 59, 62–64, 66, 81, 82, 84, 88, 92,
95, 96, 98, 104, 105, 113, 114, 118, 119, 121, 125, U
130, 132, 138, 140, 144, 147, 164, 171, 177, Unnatural amino acids ......................................................109