Acta Biomaterialia 163 (2023) 35–49

Contents lists available at ScienceDirect

Acta Biomaterialia
journal homepage: www.elsevier.com/locate/actbio

Review article

Mechanics of isolated individual collagen fibrils✩

Orestis G. Andriotis, Mathis Nalbach, Philipp J. Thurner∗
Institute for Lightweight Design and Structural Biomechanics, TU Wien, Vienna, A-1060, Austria

a r t i c l e i n f o a b s t r a c t

Article history: Collagen fibrils are the fundamental structural elements in vertebrate animals and compose a structural
Received 30 March 2022 framework that provides mechanical support to load-bearing tissues. Understanding how these fibrils
Revised 15 November 2022
initially form and mechanically function has been the focus of a myriad of detailed investigations over
Accepted 5 December 2022
the last few decades. From these studies a great amount of knowledge has been acquired as well as a
Available online 10 December 2022
number of new questions to consider. In this review, we examine the current state of our knowledge
Keywords: of the mechanical properties of extant fibrils. We emphasize on the mechanical response and related
Collagen deformation of collagen fibrils upon tension, which is the predominant load imposed in most collagen-
Structure-function relationship rich tissues. We also illuminate the gaps in knowledge originating from the intriguing results that the
Mechanics field is still trying to interpret.

Statement of significance

Collagen is the result of millions of years of biological evolution and is a unique family of proteins, the
majority of which provide mechanical support to biological tissues. Cells produce collagen molecules that
self-assemble into larger structures, known as collagen fibrils. As simple as they appear under an optical
microscope, collagen fibrils display a complex ultrastructural architecture tuned to the external forces
that are imposed upon them. Even more complex is the way collagen fibrils deform under loading, and
the nature of the mechanisms that drive their formation in the first place. Here, we present a cogent
synthesis of the state-of-knowledge of collagen fibril mechanics. We focus on the information we have
from in vitro experiments on individual, isolated from tissues, collagen fibrils and the knowledge available
from in silico tests.
© 2022 The Authors. Published by Elsevier Ltd on behalf of Acta Materialia Inc.
This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/)

1. Introduction Collagen stands for a superfamily of closely related, but genet-

ically distinct molecules. To date, 28 different types of collagen
Ancient cultures (Greeks and Romans) used collagen as the have been identified in vertebrates, which are encoded by 44 genes
original source of glue. Glue was made through denaturation, i.e., [1]. The 28 members of the collagen superfamily have been fur-
boiling, of collagen-rich biological tissues such as bones, tendons ther classified, based on the supramolecular architecture collagen
and skin. Hence, the name collagen, which in Greek means, a glue- molecules form. An in-depth information on the collagen family
generating substance (from the latinized form of Greek kolla “glue” can be found in the review by Ricard-Blum [2] and further in-
+ -gen “giving birth to”). In a wider context, we may consider col- troduction to collagen structure and mechanics at various length
lagen to be the glue that holds cells, biological tissues, and ulti- scales can be found in the book edited by Fratzl [3]. In brief,
mately human bodies together. Importantly, collagen is much more fibril-forming collagens, as the name implies, self-assemble (in the
than that; it provides both macroscopic mechanical function and extra-cellular space) into heterotypic collagen fibrils from collagen
microscopic mechanical feedback in the human body. types I, II, III, V, XI, XXIV and XXVII. Collagen fibrils are reported
to possess a semi-crystalline ultrastructure [4]. Importantly, colla-
gen fibrils are heterotypic, i.e. they are built up by a mixture of
the different types of fibril-forming collagen molecules. Studies fo-
✩
Part of the Special Issue on the Mechanics of Cells and Fibers, guest-edited by cused on the mechanics of individual collagen fibrils have been
Professors Amrinder S. Nain, Derrick Dean, and Guy M. Genin.
∗
conducted on samples sourced from tendons. Such collagen fibrils,
Corresponding author.
E-mail address: [email protected] (P.J. Thurner).
are sometimes confusingly mentioned as collagen type I, but they
are in fact heterotypic made up of a mixture of collagen type I, III

https://doi.org/10.1016/j.actbio.2022.12.008
1742-7061/© 2022 The Authors. Published by Elsevier Ltd on behalf of Acta Materialia Inc. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/by/4.0/)
O.G. Andriotis, M. Nalbach and P.J. Thurner Acta Biomaterialia 163 (2023) 35–49

and V molecules [5,6]. The fibril-forming collagen type I is found fibril remodeling provides for a more dynamic interplay with cells.
in a variety of biological tissues across the body such as bones, In addition, the formation of collagen fibrils seems highly depen-
skin, tendons, ligaments, sclera, cornea and in blood vessels [21]. dent on the forces applied to them or their unassembled compo-
Accounting for about 90% of collagen, type I is the most ubiqui- nents. Collagen fibrils play a pivotal role for cell behavior [11,75,76]
tous of all collagens. As such, collagen type I molecules are found and the dynamic mechanobiological interplay between cells and
in the highest concentration compared to type III and V. But lungs collagen fibrils enriches our current understanding of their role
[7] and skin [8] contain relative amounts of type III to type I higher in tissue homeostasis and cell mechanotransduction in health and
compared to the relative amount found in tendons (or ligaments) disease. Efforts have been made for mechanistic models to describe
[9]. To date, studies on individual collagen fibril mechanics focused collagen fibril mechanics [40,44,77,78]. But there is yet a gap in
on samples primarily sourced from tendons, mainly because of the knowledge to be filled, rising from the fact that the current mod-
ease of extracting intact collagen fibrils and hence our review im- els either do not or use only a small amount of unbound water and
plicitly considers collagen types I, III and V. are limited to small spatial and temporal simulation times. Also
Other collagen subgroups, not considered in detail in this re- the molecular dynamics simulations aiming to mechanistically ex-
view, are the fibril-associated collagens with interrupted triple he- plain collagen fibril mechanics, are either employed on structural
lices (FACITs; type IX, XII, XIV, XVI, XIX, XX, XXI, XXII), the network models of a theoretical subunit of collagen fibrils [4] or simplify
forming collagens (type IV, VI, VII, VIII and X), the transmembrane the structure through coarse grain models [41], yet in both cases
collagens or also known as membrane-associated collagens with simulating the case of a dehydrated collagen fibril.
interrupted triple helices (MACITs; types XIII, XVII, XXIII, XXV), the In this review, we will discuss the current state-of-the-art of
multiplexins (types XV and XVIII) and two, currently, unclassified the mechanical properties of isolated and individual collagen fibrils
collagen types, the type XXVI and XXVIII. measured in vitro. Importantly, we make the distinction, for this
More than 25% of the total protein mass content of the human review, of focusing only on mechanics and related studies assessing
body is composed by collagen [10]. This makes collagen one of the individual and isolated collagen fibrils rather than a collection of
most abundant extracellular matrix (ECM) components in multi- fibrils or a fibril within a collection of other fibrils.
cellular organisms. In an evolutionary sense, collagen is strongly
conserved; vertebrates and echinoderms, share genetically almost
2. Origin of collagen fibril samples
identical collagen types. The primary function of collagen is to
provide structural support and mechanical function, particularly in
Included studies in this review tested either a) native collagen
collagen-rich tissues. Being the main structural proteins of ECM
fibrils isolated mechanically from tissues, such as tendons, the der-
and therefore of the cell microenvironment, collagens play a role
mis of the Cucumaria frondosa and human biopsies, b) or recon-
in cell behavior, and are involved in cell adhesion [11] and migra-
stituted collagen fibrils from soluble collagen in acetic acid solu-
tion [12–15]. In this context, collagen fibrils contribute to tissue
tion. Soluble collagen can either be prepared [79] or purchased,
remodeling during growth [16–18], differentiation [19] and wound
and native-type collagen fibrils are formed from that solution as a
healing [20].
result of the intrinsic property of collagen to self-assemble. In vitro
Collagen fibrils bear a unique architecture [4], which is simi-
reconstitution of collagen fibril from solution takes place, as pro-
lar across different biological tissues. Nevertheless, these tissues
posed by Holmes et al. [80] by diluting, neutralizing and warming
deliver a broad spectrum of functions characterized by different
of the collagen solution. A list of the publications and the sample
mechanical properties. In musculoskeletal tissues, collagen fibrils
origin is included in Table 1.
facilitate the force transfer from muscles to tendons and bones
to support locomotion. In the lungs, collagen fibrils and their mi-
croscale organization in both the airways and the parenchyma pro- 3. Quantifying the mechanics of individual collagen fibrils
vide the structural stability necessary during breathing [22,23].
Chronic lung pathologies such as asthma [24,25] and idiopathic The mechanics of individual collagen fibrils have experi-
pulmonary fibrosis [26] leading to impaired lung functionality from mentally been studied via atomic force microscopy (AFM) or
nanoscale alterations at the collagen fibril level, manifest the im- micro-electromechanical system (MEMS) devices and in different
portance of collagen fibril mechanics for lung mechanics and func- loading scenarios. Nanoindentation [49,69,81] and bending [71–
tion. In the cornea, the narrow diameter distribution of collagen 73] tests have been conducted with AFM. Moreover, nanoten-
fibrils and their unique highly aligned and highly packed organi- sile tests have been conducted with AFM [33,36,58,63–65,82],
zation into lamellae [27] forms a tissue transparent to the visi- MEMS [42,51,60,61] devices, while also dedicated instruments
ble spectrum of light, facilitating the ability of sight and vision. [34,83] were developed to overcome the low-throughput of both
Additionally, in the cornea, collagen fibrils and their organization AFM- and MEMS-based tests. To date, these mechanical tests
greatly contribute to the mechanical strength required for sustain- have been conducted in various environmental conditions, i.e.,
ing corneal curvature [28]. from air-dried, controlled-humidity to fully submerged in aque-
Molecular structure [29], post-translational modifications ous solutions and/or with varying salt concentration and pH
[30,31], intrafibrillar cross-linking [32–34] and collagen type [42,45,46,49,51,60,61,69–73,81].
[35] are some of the various parameters responsible for the Since the first tensile tests of collagen fibrils were reported [65],
specific mechanical function and properties of individual collagen 94 collagen fibrils were mechanically characterized under tension
fibrils. However, exactly how the mechanics of collagen fibrils are with AFM- and MEMs-based tests, 66 of which were tested while
specified and what the mechanical properties entail is a matter hydrated and the rest 15 while air-dried [33,36,42,51,60–65,71,82].
of ongoing research: how do collagen fibrils achieve the variety Svensson et al. [34] increased the specimen number tested in ten-
of mechanical functionalities required by the different tissues, sion by developing a device using a) a piezoelectric actuator to
which mechanisms are at work and how are these combined or apply and measure displacement, and b) a microscopic cantilever
interdependent on each other? Seeking answers to these questions with a capacitance sensor to measure the applied force. The device
by studying collagen fibril mechanics has been the focus of a developed by Svensson et al. [34] offers high throughput tensile
number of studies [29,31-34,36–74]. testing of collagen fibrils [34] up to fracture but suffers from noise
Collagen fibrils should not be considered merely passive ele- at low forces compared to an AFM (due to the long cantilever that
ments of the ECM. Instead, constant and tissue-dependent collagen has natural oscillating frequencies of tenths of Hz) and lacks the

36
 Table 1

O.G. Andriotis, M. Nalbach and P.J. Thurner

Indentation modulus of individual collagen fibrils, native and reconstituted, (deposited on a stiff substrate) measured via indentation-type AFM.
Environment & Indentation
Indentation modulus, Eind Key finding Conditions Analysis method Contact (AFM tip/sample) ∼kAFM (N/m) depth 1 Sample origin Ref.
1–2 Gpa 2 Eind increases above 10% indentation depth Air-dry (<45% humidity) Hertz Sphere/cylinder 1–2 n/a Cucumaria frondosa - native [49]
5–11.5 GPa Employed Oliver-Pharr method Air-dry Oliver-Pharr Sphere/flat; 4.5 h/D 3 ≤ 0.1 Rat tail tendon - native [69]
Ac ( h ) = π ( 2Re f f h − h2 )
1.9 ± 0.5 GPa Eind decreased from GPa to MPa upon Air-dry Hertz Sphere/sphere 40 h/D ≤ 0.4 Bovine Achilles tendon - reconstituted [45]
hydration
1.2 ± 0.1 MPa Buffer (100 mM sodium Hertz Sphere/flat 0.3 h/D ≤ 0.4 Bovine Achilles tendon - reconstituted [45]
phosphate pH7)
2.1 ± 0.4 MPa - Eind can be tuned upon exposure to solutions Various solutions (salt, Hertz Sphere/sphere 0.3 h/D ≤ 0.4 Bovine Achilles tendon - reconstituted [46]
172.5 ± 59 MPa pH and ethanol)
1.2 – 2.2 GPa Eind higher in overlap compared to gap Air-dry Hertz; elliptical Sphere/cylinder 4.1 h/D ≤ 0.05 Bovine Achilles tendon - reconstituted [54]
regions contact area
7.0 ± 1.5 GPa Validation study: Air-dry Hertz Sphere/sphere 40 h/D ≤ 0.1 WT 4 mice tail tendon - native [81]
Reconstructed AFM tip shape. Compared
AFM-based nanoindentation vs. conventional
nanoindentation on polymers as well as
various data analysis for collagen fibril
characterization
6.9 ± 1.5 GPa Air-dry Hertz Sphere/cylinder 40 h/D ≤ 0.1 WT mice tail tendon - native [81]
9.3 ± 1.3 GPa Air-dry Oliver-Pharr Sphere/flat; 40 h/D ≤ 0.1 WT mice tail tendon - native [81]
Ac ( h ) = π ( 2Re f f h − h2 )
9.4 ± 1.7 GPa Air-dry Oliver-Pharr Empirical function 5 40 h/D ≤ 0.1 WT mice tail tendon - native [81]
3.2 ± 1.1 GPa Air-dry Oliver-Pharr Empirical function 40 h/D ≤ 0.1 Rat tail tendon - native [81]
6.6 ± 0.7 GPa Air-dry Oliver-Pharr Empirical function 40 Human bronchi - native [81]
h/D = 0.06 ± 0.01
∼2–50 Mpa 6 Eind Nanopure water Sneddon Conical tip 0.7 h ∼30 nm Rat tail tendon - native [39]
a. increased with indentation speed
b. decreased upon exposure to temperature
∼1–10 Mpa 7 h ∼30 nm
37

Nanopure water Sneddon Conical tip 0.25 Rat tail tendon - native [39]
7.9 ± 2.8 GPa Lower packing density in air-dried OIM fibrils, Air-dry Oliver-Pharr Empirical function 40 h/D ≤ 0.1 WT mice tails tendons - native [31]
resulting in lower Eind vs. WT ones
5.3 ± 2.2 GPa Air-dry Oliver-Pharr Empirical function 40 h/D ≤ 0.1 OIM 8 mice tails tendons - native [31]
3.3 ± 0.5 MPa Less hydration in OIM collagen fibrils resulted PBS 9 (pH7.4) Oliver-Pharr Conical with spherical apex 0.32 – 0.48 h/D ≤ 0.1 WT mice tails tendons - native [31]
in higher Eind vs. WT ones
16.2 ± 3.0 MPa PBS (pH7.4) Oliver-Pharr Conical with spherical apex 0.32 – 0.48 h/D ≤ 0.1 OIM mice tails tendons - native [31]
17.3 ± 3.9 MPa Eind reduced over damaged (“kinked”) regions PBS (pH7.4) / Control Sneddon Conical 0.7 Fitted portion Bovine tail tendons - native [38]
of the fibrils for h/D = 0.1
6.6 ± 2.1 MPa PBS (pH7.4) / Kinked Sneddon Conical 0.7 Fitted portion Bovine tail tendons - native [38]
for h/D = 0.1
2.1 ± 1.2 MPa PBS (pH7.4) / Very Sneddon Conical 0.7 Fitted portion Bovine tail tendons - native [38]
kinked for h/D = 0.1
0.1 – 1.4 MPa Eind increased with increasing concentration PBS Hertz Sphere/sphere 0.2 h/D ≤ 0.15 In vitro fibrillogenesis [35]
of collagen type I
∼15 Mpa Eind increased up to 24-fold upon partial PBS (pH7.4) Sneddon Conical 0.48 h/D ≤ 0.1 WT mice tails tendons - native [36]
dehydration
∼370 Mpa PBS (pH7.4) + 3.5M Sneddon Conical 0.48 h/D ≤ 0.1 WT mice tails tendons - native [36]
PEG10


Acta Biomaterialia 163 (2023) 35–49

Sphere (Rt )/cylinder (Rf ) Effective radius: Re f f = Rt2 R f / (Rt + R f )
Sphere (Rt )/sphere (Rf ) Effective radius: Re f f = Rt R f / (Rt + R f )
1 Indentation depth is noted here as percentage of collagen fibril height, D
2 Reduced modulus reported in Heim et al. (2007)
3 h/D: ratio of indentation depth to fibril diameter, D. For h/D≤0.1, Bueckle’s rule is satisfied [90].
4 WT = wild type
5 Empirical function of AFM tip determined after image reconstruction. AFM height image of TGT1 calibration grating
6 Data from three collagen fibrils. Baldwin et al. (2014) showed how the indentation modulus increases up to 10-fold with indentation speed.
7 Temperature exposure - indentation modulus decreased with increasing temperature exposure. Collagen fibrils were exposed to temperature, then cooled down to 25° and measured.
8 OIM = osteogenesis imperfecta mouse
9 PBS = phosphate buffer saline
10 PEG = poly-ethylene glycol
O.G. Andriotis, M. Nalbach and P.J. Thurner Acta Biomaterialia 163 (2023) 35–49

Fig. 1. (A) Three-dimensional view of an air-dried collagen fibril on a glass microscope slide. Collagen fibrils are identified by their characteristic D-periodicity. In a typical
AFM cantilever-based nanoindenation test, the cantilever is driven towards the sample and deflects (dmax ) while a small portion of the AFM tip indents (hmax ) the surface
of the sample (with permission from [81]). (B) Scanning electron microscopy and three-dimensional atomic force microscopy image of the reconstructed AFM tip apex,
accompanied by the cross-sectional area as a function of the tip height (with permission from [36]). (C) Force volume map of an isolated individual collagen fibril. The force
volume map is partitioned in a two-dimensional array of positions where the indents are conducted i.e., force-indentation data are recorded (with permission from [36]).
(D) Cross-section height profile of the collagen fibril from which the diameter is determined (with permission from [36]). (E) Experimental force-indentation data (with
permission from [81]).

control of force and displacement. The mechanics of collagen fib- force-indentation tests (for example in a force volume map format;
rils at high forces (>100 nN) up to fracture are important for in- Fig. 1C) over a region of interest. The pixel resolution of the region
forming tissue-level mechanics, whereas, studying collagen fibrils of interest i.e., force volume map should be such that at least one
at low forces (<100 nN, forces the cells exert on their environment pixel, i.e., a force-indentation test (Fig. 1D) will be conducted at
[84,85]) is important for mechanotransduction. A device recently the crest of the fibril. To analyze force-indentation data and pro-
developed by Nalbach et al. could bridge this gap [83] by perform- vide an estimate of the indentation modulus, several theories and
ing tests under various loading scenarios i.e., quasistatic, dynamic models based on contact mechanics can be used. What is impor-
loading, stress relaxation and creep tests (the latter currently not tant to know for data analysis is the shape of the projected con-
possible with most AFM instruments). tact which is defined by the geometry of the AFM tip apex and
the sample. The AFM tip apex, is either assumed to have a perfect
geometric shape (conical, spere, paraboloid) [38,39,45,46,49,69,86]
3.1. Nanoindentation tests
or is experimentally reconstructed via (AFM) imaging a calibration
grating [81], eventually providing an empirical function of the con-
Atomic force microscopy cantilever-based nanoindentation is
tact area (Fig. 1E). On the other hand, collagen fibrils have either
perhaps the most time-efficient test for mechanical characteriza-
been assumed as spheres or cylinders of radius R, and as flat.
tion of individual collagen fibrils [45,46,49,69,71–73,81] and gener-
Regardless of the limitations associated with nanoindentation
ally materials at the nanoscale.
tests on collagen fibrils, valuable knowledge has been gained over
AFM-based nanoindentation is easy to use, because of the mini-
the last two decades on the mechanical behavior of collagen fibrils.
mal sample preparation and yields high-throughput data compared
to bending or tensile tests at the nanoscale.
AFM also allows the identification of individual isolated colla- 3.2. Nanotensile tests
gen fibrils, via recognition of the characteristic D-banding or D-
periodicity through contact- or intermittent contact mode imaging The first nanotensile test methods were published in 2006 by
in air-dried state (∼67 nm; periodic corrugations of the collagen Eppel et al. [42], who designed a micro-electro-mechanical sys-
fibril surface as seen in Fig. 1A). The sharp AFM tip apex (Fig. 1B), tem (MEMS) (Fig. 2A) and by van der Rijt et al. [65], who em-
located at the end of a bendable cantilever, is used to perform ployed the atomic force microscope (AFM) in force spectroscopy

38
O.G. Andriotis, M. Nalbach and P.J. Thurner Acta Biomaterialia 163 (2023) 35–49

Fig. 2. Overview of the three different methods for tensile testing of individual, isolated collagen fibrils. (A) The MEMS platform (adapted from Liu et al. [52] with per-
mission). The collagen fibril is suspended between the fixed end and strain gage pad. The strain gage pad is connected to the moving pad via beams of known stiffness.
Displacement of the moving pad causes deflection of the beams that is proportional to the force exerted on the collagen fibril. The deflection is measured optically. (B)
With the AFM method, a segment of a collagen fibril is glued to the substrate and the AFM cantilever with epoxy resin. By an upward translation of the cantilever, the
collagen fibril is deformed and the force exerted is proportional to the measured cantilever deflection (adapted with permission from [36]). (C) Bowstring stretching with
the AFM. The collagen fibril is glued to the substrate on two sites and subsequently loaded in a bowstring geometry. The force is measured by torsional bending of the
cantilever. (D) NanoTens instrument (from Nalbach et al. [83], with permission under CC BY 4.0). A magnetic microsphere is glued to the collagen fibril. By application of an
external magnetic field, the magnetic microsphere is lifted and picked up with a fork-like gripper that is mounted on a fiber-optic force probe. The force, proportional to the
cantilever deflection, that arises due to an upward movement of the force probe is measured with an interferometer.

mode (Fig. 2B) for tensile testing of individual isolated collagen fib- accompanied by an up to 30% increase of fibril height [39]. In the
rils. Both methods were further adapted to enable dynamic tensile same study, Baldwin et al. [39] showed that increasing the inden-
tests of fully hydrated collagen fibrils up to failure [33,52,60,71]. In tation speeds above 100 μm/s, resulted in an increase of indenta-
2016, a further AFM method was presented by Quigley et al. [58], tion modulus. In the above cases (with an exception to the effect
who loaded collagen fibrils in a bowstring geometry (Fig. 2C), in of indentation speed), increase in indentation modulus was accom-
contrast to the force spectroscopy geometry. Svensson et al. [34,87] panied by a decrease in fibril height and vice versa.
built a custom instrument to perform tensile studies by pulling the Upon hydration, the axial separation between collagen
collagen fibril in the focal plane of a light microscope and optically molecules increases [88] and therefore the molecular packing
measuring the local elongation of the collagen fibril. Recently, Nal- density is reduced in the radial direction of the fibril. One may
bach et al. [83] developed a tensile testing device, the (NanoTens then conclude as result of the decrease in molecular packing
Fig. 2D), which allows for quick coupling and uncoupling of sam- density (also manifesting upon exposure to higher temperatures),
ples via a non-permanent connection between collagen fibrils and collagen molecules have more space to deform and rearrange upon
the force sensor. an axial compressive load. Therefore, the molecules would be more
compliant, which results in a reduced indentation modulus at the
4. Collagen fibril mechanics collagen fibril level.
The influence of indentation speed on indentation modulus
4.1. Indentation modulus influenced by environmental/experimental highlights the time-dependent response to axial loading. As inden-
factors tation speeds increase, certain relaxation mechanisms of the colla-
gen molecules do not take place resulting in an apparent increase
Hydration of collagen fibrils in an aqueous solution of molecular packing density and increase in indentation modulus.
[31,36,45,46], reduced their indentation modulus by three-orders
of magnitude (down to several MPa from GPa) and was accompa-
nied by an increase in collagen fibril height (diameter) compared 4.2. Collagen fibril ultrastructure and indentation modulus
to the air-dried state. Conversely, dehydration of collagen fibrils in
ethanol [31,46] or by exposing them to various concentrations of The D-periodicity of 67 nm in collagen fibrils is characterized
poly-ethylene glycol [36] increased their indentation modulus and by the overlap and gap regions because of a periodic offset of col-
decreased their height (or diameter). lagen molecules along the long axis of collagen fibrils [89]. Over-
Additionally, exposing collagen fibrils to temperatures above lap regions have slightly higher packing density compared to the
50 °C and up to 62 °C decreased their indentation modulus, also gap regions [89] and Minary-Jolandan et al. [54] and Grant et al.

39
O.G. Andriotis, M. Nalbach and P.J. Thurner Acta Biomaterialia 163 (2023) 35–49

[47] reported a slightly higher indentation modulus in the overlap permanent damage behavior [33,56]. However, the non-linearity
compared to the gap regions. and viscous behavior do not necessarily arise from the same defor-
Kink formations localized at damaged regions along collagen mation mechanisms. Phenomenologically, the unique and distinct
fibrils and sharp bents of the fibrils are also domains of the fib- response of collagen fibrils under tension is described by a three-
rils that have been associated with reduced indentation modulus phase behavior (phase I, II and II; Fig. 3) [33].
[38,39]. The D-periodicity, damaged areas and bents of the fibrils
show variations in the molecular packing density that has influ- 4.3.1. Three-phase stress-strain evolution: deformation mechanisms
enced the indentation modulus measured in these areas. Collagen molecules, within the collagen fibril structure, show
An intriguing question is, what makes collagen fibrils to kink local variations along their structure characterized by kink forma-
periodically upon overload at the fascicle level or close to the frac- tions and micro-unfolding of the triple helix [30,44]. The kinks
ture site of collagen fibrils? As speculated by Baldwin et al. [38], are prevalent at the gap regions [4,44]. Molecular dynamics sim-
kinks could be the result of compressive load normal to the long ulations predicted the different possible deformation mechanisms
fibril axis, once the normal tensile load is removed. Yet unknown is during tension of a collagen fibril to be straightening, uncoiling, in-
the origin of this periodic compressive load and why collagen fib- termolecular sliding, intramolecular sliding, and backbone stretch-
ril kink formations vary among tendon locations [68] and between ing of collagen molecules [41,44,92,93].
collagen fibrils from positional and energy-storing tendons [56]. To Upon tension at lower forces molecular kinks straighten
this end, tensile tests on individual collagen fibrils by means of [44] and small regions of the triple helix uncoil [41], while at
quasistatic, creep or stress relaxation, dynamic mechanical analysis higher loads the backbone of the triple helix is stretched [44].
performed in combination with structural assessment and molecu- As predicted by Zhou et al. uncoiling and straightening take place
lar dynamics simulations may provide further insights. at lower forces leading to lower stiffness, compared to backbone
Beyond changes in ultrastructure, the composition of collagen stretching of the collagen molecule, which leads to a higher stiff-
fibrils has been reported to influence the indentation modulus too. ness [92]. Depalle et al. suggested these deformation mechanisms
For example, in the osteogenesis imperfecta mouse (OIM) model, take place at increasing stretch levels upon fibril tension [41].
genetic mutations [31] resulting in primary structure alterations of However, experiments yield varying results which suggest that the
collagen molecules, were associated with impaired hydration at the different deformation mechanisms may occur simultaneously with
collagen fibril level and increased indentation modulus. In addi- some being prevalent depending on load and sample conditions.
tion, a relative increase in the amount of collagen moleculestype In phase I, straightening, and aligning of collagen molecules
III compared to type I constituting heterotypic collagen fibrils was gradually takes place resulting in a convex “toe” region of the
associated with a reduction in the indentation modulus [35]. Al- stress-strain evolution. From about 2% strain, this region transitions
though, the latter effect lacks a mechanistic explanation as to why to an almost linear increase characterized by straightening and un-
this decrease in mechanics is influenced by the presence of colla- coiling of the collagen triple helix [41]. With increasing strain in
gen type III, in the case of OIM, the increase in indentation mod- phase I, intermolecular sliding starts to take place [41] until it,
ulus of hydrated collagen fibrils (compared to WT ones) was sug- eventually, becomes the dominant deformation mechanism.
gested to result from changes in the molecular packing density due Phase II initiates at about 10% of fibril strain. Onset of phase
to less hydrated OIM compared to WT collagen fibrils [31]. II is characterised by a decline of the stress-strain slope, hence the
tangent tensile modulus decreases (softening). Collagen fibrils with
4.3. Tensile behavior of collagen fibrils low intermolecular cross-link density, or containing only immature
cross-links, deform to failure in phase II. In contrast, collagen fib-
Based on estimations of their length being 14–50 mm [91], col- rils with a sufficiently large number of mature or sugar-based in-
lagen fibrils have an aspect ratio in the range of 14 × 104 - 50 × termolecular cross-links enter phase III at 15% to 20% strain [33].
104 . Because of this, collagen fibrils in mammals predominantly Phase III is indicated by a second increase in the tangent tensile
withstand tensile loads in vivo. Under tension, collagen fibrils show modulus. This increase is thought to be due to hindrance of fur-
a non-linear elastic [36,65], viscous (time-dependent) [64,70], and ther intermolecular sliding by intermolecular cross-links, resulting

Fig. 3. (A) Experimental tensile stress-strain curve up to fracture of a collagen fibril from human patellar tendon measured with AFM. The stress-strain curve shows three
distinct phases. The presence of the three-phase stress-strain curves is assumed to be associated with different deformation mechanisms taking place in each phase. In phase
I, collagen molecules are straightened, and uncoiled/unwound. In phase II, the collagen fibril begins to deform by intermolecular sliding, resulting in a decrease of the slope
of the stress-strain diagram. In phase III, intermolecular cross-links restrict further intermolecular sliding, and the molecular backbone is stretched. This leads to an elevated
resistance against deformation as shown by the increased slope in the stress-strain diagram. The slope of the stress-strain diagram is plotted in B and corresponds to the
tangent modulus. One can observe two local maxima of the tangent modulus, one in phase I and one in phase III. In case of reduced intermolecular cross-link density, the
collagen fibril fails without distinct stiffening, i.e., the phase III behavior is absent. Image from Svensson et al. [33] with permission.

40
O.G. Andriotis, M. Nalbach and P.J. Thurner Acta Biomaterialia 163 (2023) 35–49

Table 2
Nanotensile properties of individual collagen fibrils, namely phase I and III tangent tensile modulus, strength and failure strain. Data is represented mean ± std or geometric
mean (geometric SE).

Phase I Phase III

Tangent modulus Tangent Strength Failure strain
(GPa) modulus (GPa) (MPa) (%) Sample origin Method Environment & Conditions Reference

0.4 n/a n/a n/a Rat tail tendon AFM Hydrated, [82]
10 weeks old phosphate-buffered saline
(PBS)
1.1 n/a n/a n/a Cross-linked with AFM Hydrated, PBS [82]
glutaraldehyde
0.5 ± 0.4 n/a 230 ± 160 80 ± 44 Sea cucumber (dermis) MEMS Hydrated, PBS [60]
0.08 – 0.25 n/a n/a Sea cucumber (dermis) MEMS Hydrated, PBS [61]
0.6 ± 0.2 n/a 60 ± 10 13 ± 2 Bovine Achilles tendon - AFM Hydrated, PBS [71]
reconstituted
0.7 ± 0.1 n/a 170 ± 10 30 ± 2 Cross-linked with EDC/NHS AFM Hydrated, PBS [71]
1.8 ± 0.3 n/a 290 ± 10 22 ± 2 Cross-linked with AFM Hydrated, PBS [71]
glutaraldehyde
3.5 ± 0.4 4.3 ± 1.4 540 ± 140 20 ± 1 Human patellar tendon AFM Hydrated, PBS [33]
2.2 ± 0.9 n/a 200 ± 110 16 ± 4 Rat tail tendon, 12 weeks old AFM Hydrated, PBS [33]
0.3 ± 0.1 n/a 71 ± 23 63 ± 21 Rat patellar tendon MEMS Not specified (presumably [52]
hydrated in PBS)
1.8 – 4.8 n/a 638 – 1134 n/a Reconstituted collagen fibril MEMS Ambient, 60% relative [51]
from calf skin humidity
1.6 (1.3 – 2.0) n/a 80 (70-100) 20 (16-25) Rat tail tendon, 16 weeks Custom Hydrated, PBS [34]
2.8 (2.4 – 3.1) 6.0 (3.5–10.3) 420 (310–570) 23 (20-26) Cross-linked with MGO Custom Hydrated, PBS [34]
2.0 (1.6 – 2.4) 2.6 150 [120–190] 14 (11-18) Rat Achilles tendon, 16 Custom Hydrated, PBS [34]
weeks
2.4 (2.3–2.6) 4.7 (3.3–6.8) 410 [330–500] 37 (31-45) Cross-linked with MGO Custom Hydrated, PBS [34]
0.4 – 0.8 n/a n/a n/a Mice tail tendon, AFM Hydrated, PBS [36]
6-month-old
3.9 ± 0.7 9.1 ± 4.0 1020 ± 330 26.6 ± 3.1 Human patellar tendon, Custom Hydrated, PBS [32]
male, 26 ± 6 years old
4.6 ± 1.3 10.8 ± 3.3 1210 ± 280 26.9 ± 3.6 Human patellar tendon, Custom Hydrated, PBS [32]
male, 66 ± 1 years old
0.1 – 1.0 n/a n/a n/a Reconstituted collagen fibril MEMS Hydrated, PBS [70]
from calf skin

in an increase of shear resistance [33,41]. Hence, increased tensile expected to build around collagen molecules (Fig. 4B) [88,94,95].
load is exerted directly onto the bonded chains of the molecules, Weak intermolecular forces decrease with hydration resulting in
and the molecular backbones are stretched. Table 2 summarizes a facilitated molecular deformation mechanisms.
selection of the tensile mechanical properties of individual collagen Andriotis et al. [36] systematically explored and quantified the
fibrils measured in aqueous solution only. impact of partial dehydration by exerting osmotic pressure on col-
lagen fibrils within buffer solution. They achieved this by adding
4.3.2. Non-linear stress-strain evolution in phase I poly-ethylene glycol (PEG) of different concentrations to phosphate
A modulus has often been calculated from the first deriva- buffered saline (PBS) in between tensile tests on a single colla-
tive of stress vs. strain and presented as a function of strain [32– gen fibril from a mouse tail tendon. The tensile modulus increased
34,62,63,65], which we refer to as tangent modulus. The majority, by up to 6-fold in partially hydrated collagen fibril in 2.6 M PEG
if not all, of current literature reports tangent moduli data from compared to the exact same fibril fully hydrated in PBS. By AFM
the loading curve only, because to date tests conducted a single cy- nanoindentation, collagen fibrils were found to shrink in presence
cle until fracture [32–34,62,63]. We believe that subsequent stud- of higher osmotic pressure by about 15% applying the same de-
ies relied on the tangent modulus of the loading curve to cross- hydration protocol. The effect of partial dehydration was gradually
compare values with preceding literature. However, this may be increasing with increasing PEG concentration leading to the con-
less intuitive because only the unloading part of the curve, before clusion that collagen fibril mechanical properties can be tuned by
permanent damage settles in, would reflect the elastic response of hydration and osmotic pressure. This is in agreement with results
the collagen fibril. from studies that employed MEMS devices where experiments are
The tangent modulus in phase I has been reported to range conducted in a controlled humidity environment, but not in a fully
from 0.4 GPa, for collagen fibrils from rat tail tendon of an ani- hydrated aqueous solution [51,60,70].
mal at 10 weeks age [65], up to (4.6 ± 1.3) GPa for human patellar
tendon collagen fibrils of patients at (66±1) years age [87]. 4.3.4. Effect of intermolecular cross-links in phase I
Intermolecular cross-linking, i.e., covalent bonds between adja-
4.3.3. Effect of hydration in phase I cent collagen molecules, is the second proposed mechanism to in-
Hydration and intermolecular cross-linking have also shown fluence the collagen fibril tensile properties at <10% strain. Phys-
to affect the tangent modulus which was decreased upon hydra- iologically, two main types of cross-links have been identified in
tion from 5 GPa to 0.5 GPa (collagen fibrils from rat tail tendon) collagen-rich tissues a) enzymatic (subdivided into immature and
[65] and from 0.9 GPa to 0.5 GPa (collagen fibrils from sea cucum- mature) and b) non-enzymatic (sugar-induced) cross-links [96,97].
ber) [60]. Corroborated by AFM cantilever-based nanoindentation For studying the effect of enzymatic (i.e. physiological) occur-
tests, hydration of collagen fibrils results in increased intermolecu- ring cross-links on collagen fibril mechanics, collagen fibrils from
lar separation is increased due to the presence of higher amount of energy storing tendons, e.g., patellar, Achilles’ or superficial digi-
water molecules found as bound and unbound (bulk) water within tal flexor tendons (SDFT) have been compared to ones from po-
the fibril (Fig. 4A). At native hydration, a monolayer of water is sitional tendons, e.g., tail or common digital extensor tendons

41
O.G. Andriotis, M. Nalbach and P.J. Thurner Acta Biomaterialia 163 (2023) 35–49

Fig. 4. (A) High osmotic pressure results in lower intermolecular distance. Weak intermolecular interactions change proportionally resulting in the stiffness being a function
of the water concentration. At lower water concentration (purple-brown) or intermolecular distance the collagen fibril is stiffer compared to collagen fibrils with high water
concentration (yellow). The figure is adapted with permission from Andriotis et al. [36]. (B) The tiple helix of a 30-residue long part of a collagen molecule (PDB: 1CGD) is
shown on top without and below with the water monolayer as predicted by [95], with permission under a CC By 4.0 license from the RCSB PDB (10.2210/pdb1CGD/pdb).

(CDET) [33,34,98]. This is because these tendons show differences whether human patellar tendon collagen fibrils reach a maximum
in the cross-link profile and density [98]. value for tensile modulus due to a finite number of possible cross-
The effect of nonenzymatic cross-linking on fibril mechanics, linking sites along a collagen molecule. A small number of human
has been investigated by either a) in vitro cross-linking collagen patellar tendon samples, which Svensson et al. cross-linked with
fibrils with a methylglyoxal (MGO; a fast reacting agent mimicking MGO only resulted in a modest increase in tensile modulus op-
the formation of cross-links between lysine and arginine residues posed to the strong increase previously seen in rat tail and Achilles’
[97]) or b) by testing collagen fibrils harvested by young and old tendon samples [34].
individuals [87]. Both enzymatic and non-enzymatic cross-links Experiments show that cross-linking increases the tangent
form at specific locations on the collagen molecules. There were modulus in phase I. The large divergence of phase I tangent mod-
however studies in which artificial unspecific cross-linking was uli in different species and between specific cross-link densities is
investigated via 1-ethyl-3-(3-dimethyl-aminopropyl) carbodiimide however elusive [33,34]. This raises the question, if a positive cor-
hydrochloride (EDC) in the presence of N-hydroxysuccinimide relation between cross-link density and intermolecular spacing ex-
(NHS) [71] and glutaraldehyde [71,82]. Below we narrow our dis- ists, which may originate in a superposition of hydration and cross-
cussion to the enzymatic and non-enzymatic cross-links. linking mechanisms [37].
Enzymatic cross-links: Svensson et al. attributed the higher In light of the results obtained from multiscale modeling, the
tangent moduli in collagen fibrils from human patellar tendon influence of cross-links on the phase I tangent moduli is puzzling
compared to rat tail tendon samples (3.5 GPa vs. 2.2 GPa) to [40,41]. Both Buehler [40] and Depalle et al. [41] did not find an
the higher concentration of mature enzymatic HP cross-links in effect of cross-link density on the phase I tensile modulus of colla-
the human patellar tendons (890 mmol/mol), compared to rat tail gen fibrils by molecular dynamics and multiscale models. But cur-
one (8.7 mmol/mol) [33]. Additionally, Svensson et al. reported an rently, there is no mechanistic explanation of how intermolecular
increase of the mature HP cross-link with age in both Achilles’ cross-links contribute to the mechanical behavior at low strains.
(90 vs. 278 mmol/mol) and tail tendon (0 vs. 2 mmol/mol) [34]. Generally, the thought seems to be that the predominant de-
Non-enzymatic cross-links: Svensson et al. did not find signif- formation mechanisms in phase I are straightening and uncoiling.
icant differences in the tangent modulus of fibrils due to age and From this perspective it is not intuitive that cross-linking should
tissue type, i.e. collagen fibrils from tail and Achilles’ tendons [34]. influence phase I. Nevertheless, studies [32–34] report stiffer be-
However, tangent modulus in collagen fibrils from Achille’s ten- havior in phase I in samples that exhibit higher concentration
dons increased from 1.9 GPa in to 2.6 GPa after MGO treatment of some form of cross-links. One possible explanation is that in-
[34]. termolecular sliding is present already in phase I, but perhaps
Because of the inconsistency with previous results comparing not dominant. If this sliding is inhibited by the cross-linking, the
human patellar and rat tail tendons, this raised the question of stiffness would increase in phase I. Performing creep experiments

42
O.G. Andriotis, M. Nalbach and P.J. Thurner Acta Biomaterialia 163 (2023) 35–49

rat tail tendons [33]. Tangent modulus of the human patellar ten-
don fibrils reached a mean of 4.3 GPa in phase III compared to
3.5 GPa in phase I. Quigley et al. [56] tested collagen fibrils sourced
from two functionally distinct tendons, namely the positional CDET
(common digital extensor tendon) and the energy storing SDFT
(superficial digital flexor tendon) of 24–36 month old steer, up to
fracture. Collagen fibrils from SDFT had a 0.8 GPa average tangent
modulus in phase III, which was about 15% higher than the tan-
gent modulus they reported for phase II. In contrast, collagen fib-
rils from CDET tendons did not show such an increase in tangent
modulus at strains above 10%. In total, 19 out of 21 SDFT (energy
storing) collagen fibrils while only 5 out of 17 CDET (positional)
ones showed three-phase-behavior. As Quigley et al. [56] did not
conduct assays to determine cross-link density they could not cor-
relate chemical modifications with mechanical data. Nevertheless,
Fig. 5. Stress-strain diagram of human patellar tendon, native and MGO cross-
based on previous data the observations by Quigley et al. were
linked rat tail tendon collagen fibrils. All collagen fibrils were tested up to failure. explained as in vivo tuning of the mechanical properties of indi-
The human patellar tendon fibril and the MGO cross-linked rat tail tendon fibril vidual collagen fibrils via mature cross-links to meet physiologi-
show a three-phase behavior with a distinct phase of stiffening from strains of cal requirements at the macro scale [56]. Svensson et al. used a
about 15%. In contrast, the native rat tail tendon collagen fibril only shows a two-
similar approach, testing collagen fibrils from tendons with differ-
phase behavior. This figure is adapted from Svensson et al. [32], with permission
under CC BY 4.0 license. ent physiological requirements [34]. In contrast to Quigley et al.
[56], Svensson et al. did not observe three-phase-behavior for ei-
ther, energy storing Achilles’ or positional tail tendon collagen fib-
in phase I while intermolecular sliding is assumed to occur, one rils, sourced from 16-week-old Wistar rats despite observed dif-
would expect a) not to reach a plateau in strain increase and b) ference in cross-link density of mature enzymatic HP cross-links
failure should occur already in phase I. But recent studies [51,70] (278 mmol/mol vs. 2 mmol/mol). The absence of phase III in
have not confirmed this which could suggest that intermolecular Achilles’ tendon collagen fibrils is at first glance contradictory to
sliding does not take place but rather cross-linking inhibits uncoil- the findings in energy storing SDFT collagen fibrils by Quigley et al.
ing and straightening. This hypothesis seems to favor rather AGEs, [34,57,58], leading Svensson et al. to raise the question whether
to enzymatic cross-links, as the theoretical attainable concentra- mature enzymatic cross-links are truly the sole cause of three-
tion of AGEs per molecule is higher and not confined at the end phase-behavior [34].
regions of the collagen molecules. However, it is important to note that the cross-link density in
the fibrils from Wistar rats Achilles tendon (278 mmol/mol) was
4.3.5. Transitions phase I-II and phase II-III much lower compared to fibrils from human patellar tendon pre-
Commencing with a strain of between 5% to 10% a gradual de- viously investigated by Svensson et al. (890 mmol/mol). It seems
crease of the tangent modulus is observed in stress strain curves that there is a critical concentration required to achieve three-
(Fig. 3) this marks the transition from phase I to phase II. The cur- phase behavior. This idea is also supported by coarse grain mod-
rent mechanistic interpretation for the decrease in stiffness and els of collagen fibrils [41]. Interestingly, Depalle et al. could only
tangent modulus is molecular sliding, i.e., between adjacent col- achieve three-phase-behavior in their models when doubling the
lagen molecules [41]. Intermolecular sliding is thought to be the strength of mature cross-links and above 60% cross-link densities,
dominant deformation mechanism, analogous to stick-slip friction, which corresponds to 1200 mmol/mol [41]. In contrast, the model
and happens after most collagen molecules are straightened and by Depalle et al. did not show three-phase-behavior at a density
(partly) uncoiled [40,41,99–101]. There is no strict definition or cri- of 20% (or concentration of 400 mmol/mol) of double-strong ma-
terion of a force or strain threshold for the onset of phase II, so no ture cross-links [41]. Nevertheless, Svensson et al. observed distinct
quantitative value for the onset have been reported. Qualitatively, three-phase-behavior after in vitro cross-linking the Wistar rat tail
the decrease of tensile modulus and hence molecular sliding is ob- and Achilles’ tendon collagen fibrils with MGO [33].
served for all types of collagen fibrils, independent from hydration This suggests that in sufficient concentration, both mature en-
[51] or cross-link density [33,34,56]. zymatic cross-links and AGE cross-links could produce phase III be-
In phase II, cross-links play a well investigated role in the havior. In this context, it is important to note, that three-phase be-
deformation mechanism. With increasing force and strain, cross- havior was never observed in collagen fibrils from rats or mice at
links restrict molecular sliding [40]. However, as experimentally re- tens of weeks of age, in contrast to collagen fibrils from energy-
ported by Svensson et al. [33] not all collagen fibrils show the abil- storing tendons of animals at tens of years of age. Even though
ity to inhibit molecular sliding during tension. A subset of collagen there is experimental evidence that both, enzymatic and non-
fibrils deform until failure and shown only a phase II behavior [33]. enzymatic cross-links, contribute to the overall mechanical behav-
In contrast, other collagen fibrils display a gradual increase of the ior, it is not clear to which extent each type of the natural cross-
tangent modulus at strain values of 15% to 20% [33,56]. This marks links contribute individually. Molecular dynamics simulations by
the onset to phase III behavior. Buehler et al. hint the necessity of two cross-links per collagen
Phase III is governed by molecular backbone stretching [41]. molecule for substantial hardening to occur, while this effect is en-
Human patellar tendon samples did show three-phase behavior hanced by the presence of mature, trivalent cross-links [40]. To fur-
in contrast to rat tail tendon samples [33]. Fig. 5 shows the ten- ther investigate this question, mechanical data of a wider age span,
sile three-phase stress-strain diagram of individual collagen fibrils from weeks to tens of years of age, of one species in combination
originated from human patellar tendon, MGO cross-linked rat tail with comprehensive quantification of a broader range of enzymatic
tendon, and the tensile two-phase stress-strain of a collagen fibril and non-enzymatic cross-links is required.
from native rat tail tendon. From the studies conducted to date, the consensus is that spe-
Collagen fibrils from human patellar tendon bear a larger con- cific deformation mechanisms are dominant in phases I, II and
centration of mature enzymatic HP cross-links than the ones from III (Table 3), yet it is unclear if and how much deformation

43
O.G. Andriotis, M. Nalbach and P.J. Thurner Acta Biomaterialia 163 (2023) 35–49

Table 3
Description and dominance of the different deformation mechanisms during collagen fibril tensile stretching.

Deformation mechanism Description Phase Reference

Straightening Alignment of a collagen molecule by straightening of kinks and twisted segments I [44]
Uncoiling Partial unfolding of helical domains, breaking of H-bonds I [40,44]
Intermolecular sliding Shearing of one molecule with respect to one or more neighboring molecules II [43,104]
Backbone stretching Alignment and stretching of peptide bonds with load direction III [40]
Disentanglement of the alpha chains within one collagen molecule III [93]

Fig. 6. (A) AFM images of rupture sites of collagen fibrils that were tested up to failure. Two types of failure modes can be distinguished (with permission from Svensson
et al. [33]). One type, that is largely intact over the entire length as determined by the presence of D-banding in the AFM images and the other type with a highly disrupted
topography. (B) AFM image over the entire length of a collagen fibril of the latter type. (C) Confocal microscope image of the same collagen fibril after staining with CHP.
Kinking and delamination as observed with the AFM correlates to molecular damage indicated by the fluorescence signal in confocal microscopy. One site, indicated with a
white arrow in panels B and C, is intact and D-banding is present in the AFM image. (B) and (C) are from Quigley et al. [56] with permission under the CC BY 4.0 license.

mechanisms, dominant in one phase, also contribute to the other ing. Given the nature of CHP binding only those locations, damage
phases. here is defined by one alpha chain being pulled out of an initially
intact collagen molecule resulting in a disruption of the triple he-
4.3.6. Damage of collagen fibril under tensile loading lix. Only ruptured collagen fibrils from positional tendon showed
Closely related to the diverging mechanical behavior at strain an elevated fluorescent signal indicating collagen molecule damage
above 10% is the observation of diverging failure modes that seem (Fig. 6).
to be associated with the cross-link density and the related defor-
mation mechanism contributing to damage of individual collagen 4.3.7. Cyclic loading
fibrils [40,41]. Svensson et al. [33] observed largely intact human Molecular damaged within the collagen fibril occurs when an
patellar tendon collagen fibrils after loading to failure in contrast alpha chain is pulled out of the collagen molecule resulting in a
to heavily disrupted rat tail tendon collagen fibrils throughout the disrupted and unfolded collagen molecule. To further asses molec-
entire length. Quigley et al. further investigated the failure modes ular damage in the context of physiological repeated mechanical
through AFM imaging, second harmonic generation (SHG) imag- deformation, Iqbal et al. applied AFM imaging and a CHP assay
ing microscopy and a collagen hybridizing peptide assay (CHP) of on collagen fibrils from positional tendon after a single cycle test
ruptured collagen fibrils of a positional and an energy storing ten- [50]. They found a sigmoidal relationship between mean fluores-
don [57]. The surface of ruptured collagen fibrils from positional cence intensity and collagen fibril strain with a steep increase of
tendons is highly disorganized and shows delamination as well as the fluorescence intensity being between 7% and 13% strain. A sim-
kink formation in contrast to the highly organized D-banding in ilar observation was made by Zitnay et al. [93], who measured the
undisrupted collagen fibrils (Fig. 6) as determined by AFM imag- fluorescence intensity of CHP stained collagen fascicles after a sin-
ing. Interestingly, delamination and kink formation are interrupted gle stretch cycle. In comparison to results from molecular dynam-
by small sections of intact D-banding. SHG imaging resulted in 30% ics simulations, it is suggested that in a shear-dominated defor-
reduction in maximum forward scattered intensity projection im- mation, collagen alpha- chains from one collagen molecule, cova-
ages of positional tendon collagen fibrils that indicate molecular lently bound via cross-links to alpha-chains of neighboring colla-
level disruption. In energy storing tendon a reduction of only 2% gen molecules are pulled-out of the triple helical structure [99].
was measured. Although SHG imaging allows for quantification of Damage in phase I was also inferred from observing the residual
overall molecular damage, it does not discriminate between molec- strain and hysteresis in other studies, in which collagen fibrils were
ular alignment and molecular denaturation. To measure molecular cyclically loaded and unloaded. Van der Rijt et al. [65] cyclically
denaturation solely, both types of ruptured collagen fibrils were loaded collagen fibrils, and observed residual strain and hysteresis,
stained with fluorescently labeled CHP. CHP binds to unfolded re- upon stretching a collagen fibril above 90 MPa in a dry or above
gions of the alpha chains of collagen molecules, i.e. by forming a 20 MPa in a hydrated state. The occurrence of residual strain af-
triple helix with two of the unfolded alpha chains, but not native ter loading a collagen fibril is confirmed by the findings in sev-
collagen [93], resulting in an increased fluorescence signal of dena- eral other studies [59,64,71]. Svensson et al. [64], however, did not
tured collagen fibrils, when performing confocal microscope imag- measure residual strains after stretching collagen fibrils about to

44
O.G. Andriotis, M. Nalbach and P.J. Thurner Acta Biomaterialia 163 (2023) 35–49

Fig. 7. Stress-strain diagram of a cyclically loaded collagen fibril at 60% humidity. After cyclical loading, the collagen fibril is tested up to failure. Hysteresis is observed in the
stress-strain diagram that is quantified in panel E. After the first cycle, a drop of the hysteresis is observed. The inelastic strain, e.g. the strain at zero stress after unloading,
is increasing over the first three cycles to reach a constant value as shown in panel C. From Liu et al. [51] with permission.

4% strain (phase I). Hence, there may be a threshold for the onset hydrogen bonds [102], which partially reform to drive recovery of
of residual strains, perhaps associated with the onset of molecular mechanical properties as suggested by the authors. Collagen fibrils,
sliding and possibly the onset of the damage mechanism proposed that are stretched to phase II, are thought to exhibit intermolecular
by Zitnay et al. [93]. Shen et al. [59] cyclically loaded dry collagen sliding, resulting in gradual increase of residual strain during cyclic
fibrils, which resulted in a gradual decrease of the point at which loading [51]. In addition, intermolecular sliding is hypothesized to
the slope of the stress-strain curves begins to decrease (phase I to be the dominant mechanism behind strengthening and toughen-
phase II transition), which may be explained with the accumula- ing in the subsequent fracture test. While Liu et al. [51] suggested
tion of damage. that there is no nanoscale damage accumulation in phase I, this
To explore this further, Liu et al. [51] systematically investi- is contradicting the interpretation and opinions expressed of dam-
gated the hysteresis and progression of residual strain of cyclically age accumulation in phase I [33,50,70]. However, direct compari-
stretched collagen fibrils to phase I, II, and post-phase II strains. son of these hypotheses is intricate as any observations involving
In their study, the authors used partially hydrated collagen fib- molecular sliding heavily involve hydration of collagen fibrils due
rils (60% humidity in air). After cyclic loading, all collagen fibrils to molecular separation and a potential role of water molecules as
were tested up to fracture. Liu et al. observed a residual strain a lubricant [88,103].
that gradually increased until a plateau after about three cycles In summary, although phase I is generally thought to be dom-
is reached, which is higher for fibrils pulled to higher strains, i.e., inated by deformation mechanisms of uncoiling and straightening,
corresponding to phase I, II, or III (Fig. 7). The hysteresis generally existing evidence suggest that damage accumulation may also oc-
dropped after the first cycle to reach a steady state. After 60 min cur, i.e., disruption of the tight triple helix as an alpha chain is
of relaxation, residual strains as well as the first cycle hysteresis pulled out of the molecule. The evidence in question is permanent
partially recovered. Partial recovery of the residual strain was in strains observed after stretching in phase I as well as higher phase
agreement with reported results by Svensson et al. and Shen et al. I stiffness observed after artificial crosslinking of fibrils with low
[59,64]. Subsequent tensile tests to fracture reveal toughening and native crosslink density (cf. Fig. 6). Additionally, creep experiments
strengthening of collagen fibrils that were cyclically stretched to in phase I also suggest damage in phase I as a possible deforma-
strains in phase II and post-phase II (phase III). The drop of the tion mechanism. However, a systematic investigation of damage
hysteresis can be explained by molecular straightening and uncoil- onset lacks to date. Overall, the various deformation mechanisms
ing, thought to be dominant deformation mechanisms in phase I. are likely dominant in a specific phase but may also contribute to
Uncoiling of collagen molecules can be associated with breaking of other phases.

45
O.G. Andriotis, M. Nalbach and P.J. Thurner Acta Biomaterialia 163 (2023) 35–49

Fig. 8. (A) Stress-strain diagram of a collagen fibril loaded at strain-rates from 0%/s to 157%/s. The derivative modulus at maximum strain increases with strain-rate, an
indication of time-dependent deformation effects (From Svensson et al. [64] with permission). (B) Stress-relaxation of individual collagen fibril by Yang et al. [71]. The
position of a piezo actuator, that performs the pulling motion, is held constant at a target collagen fibril strain while the force is recorded. A two-term Prony-series is fitted
to the stress over time data to acquire the time constants τ 1 and τ 2 . Panel B from Yang et al. [71] with permission.

4.3.8. Collagen fibril viscoelastic and inelastic behavior (a nonlinear spring and a nonlinear dashpot in series), exhibiting
The hysteresis observed in the study by Liu et al., (Fig. 7) in- thixotropic behavior, similar to the interpretation of Svensson et al.
dicate that phenomenologically collagen fibrils behave viscoelastic [64].
and/or inelastic [51]. In this subsection the current phenomeno- Finally, also stress relaxation experiments have been performed
logical findings are discussed and related to possible mechanisms and interpreted in a small number of studies. Specifically, Shen
identified mostly from molecular dynamics studies. Liu et al. quan- et al. [61] and Yang et al. [71] fitted a Maxwell-Wiechert model
tified hysteresis by calculating the loss coefficient, defined as the (a Prony series with n = 2) to stress relaxation data from sea cu-
ratio of dissipated to the supplied elastic energy per cycle and cumber and bovine Achilles’ tendon collagen fibrils (Fig. 8B), re-
found a moderate loss of about 5–10%. They also found inelastic spectively.
strains of up to 13% (depending on the cyclic regime imposed). It Shen et al. [61] and Yang et al. [71] performed relaxation stress
is important to note that the collagen fibrils the authors investi- tests in phase II (14% and 30% strain) and phase I (5%−7% strain),
gated were in a humid environment (air-dried with 60% humidity) respectively, and both reported short and long relaxation times
rather than fully hydrated in aqueous solution, which may have in- suggesting that there are at least two time-dependent mechanisms.
fluenced the results. The relaxation time was associated to the molecular rearrangement
Viscoelasticity does not only manifest itself in energy dissipa- of water and collagen molecules by Shen et al. [61]. In addition,
tion, but also in deformation-rate-dependent mechanical behav- Yang et al. [71] investigated, how cross-linking affected the vis-
ior. Deformation-rate-dependent tensile moduli was observed by a coelasticity of collagen fibrils, by performing similar relaxation ex-
number of studies [36,61,63,64,70,71] and is qualitatively shown in periments (in phase I) on native and artificially cross-linked colla-
the stress-strain diagram in Fig. 8A. Svensson et al. [63] reported a gen fibrils (with glutaraldehyde). Importantly, Yang et al. [71] sug-
7.6% increase of the tangent modulus on average across all samples gested (by fitting a two-term Prony series model [106] to re-
by increasing the pulling velocity from 9.8 μm/s to 314.0 μm/s. laxation data) that glutaraldehyde-induced cross-linking signifi-
Svensson et al. [64] separated a viscous and an elastic compo- cantly increased the short and long-time constants [71]. Yang et al.
nent of the stress-strain curve by deriving a stress-strain curve that [71] suggested intermolecular sliding to be the major mechanism
only represents the quasistatic elastic component. They did this behind both time constants. Intermolecular sliding is hampered by
by performing a stepwise tensile experiment with step sizes be- intermolecular cross-linking of both types thus both long time con-
tween 0.3 μm and 0.5 μm and allowing the collagen fibril to relax stants increase.
for 5 min between each step. This elastic component is subtracted In vitro experiments proposed that the viscoelastic behavior
from dynamic stress-strain curves of deformation-rates from 5% of collagen fibrils under tension could be explained by water hy-
strain/s to 157% strain/s (Fig. 8A). The resulting viscous compo- droplaning [103] as interstitial water and water rearrangement
nents showed linear stress-strain relationships and their slopes in- [61] promotes sub-fibrillar elements to slip against each other. In-
creased with increasing strain rates. The slope increase was best termolecular sliding was another mechanism proposed to influ-
fitted to strain-rate with a power-law function and a power lower ence the viscous behavior of collagen fibrils under tension [71] and
than one. Svensson et al. concluded that the viscosity is a de- water rearrangement . However, molecular dynamics simulations
creasing function of strain rate, which associated with interstitial suggested that rupture and reformation of H-bonds is the most
water in the collagen fibril. This causes hydroplaning of collagen highly cause of the time-dependent mechanical properties in col-
molecules at high strain rates and therefore reduces the friction lagen fibrils [107–109]. Creep tests on collagen molecules revealed
between them [64,103]. that while the collagen triple helix unwinds and straightens, in-
Viscoelasticity as a function of hydration and strain rate was tramolecular H-bonds rupture and collagen-water H-bonds are
also explored by Andriotis et al. [36], which is in agreement with forming [107–109]. This mechanism goes well in agreement with
the ideas and analysis proposed by Svensson et al. [64]. Andriotis earlier work on statistical mechanics (Bell’s model) [110]) that sug-
et al. also reported an increase in stress at a given strain with in- gested the velocity at which H-bond rupture depends on the ap-
creasing strain rate at all hydration levels investigated. In a further plied force, which in turn shows the time-dependency of the me-
analysis, Handelshauser et al. [105] reported that the tensile load chanical response of H-bonds. It is very well possible that not only
of collagen fibrils can be fitted with a non-linear Maxwell model one mechanism is at play, i.e., H-bond rupture and reformation,

46
O.G. Andriotis, M. Nalbach and P.J. Thurner Acta Biomaterialia 163 (2023) 35–49

but it is not clear which effect is dominating the viscous response direly needed to deepen our understanding. Nevertheless, experi-
of collagen fibrils. In contrast, it seems that intermolecular sliding mental advances and molecular dynamics simulations have already
is the dominant mechanism responsible for the inelastic behavior. provided us with important mechanistic insights.

5. Insights and outlook on collagen fibril nanomechanics Declaration of Competing Interest

Collagen fibrils are complex structures, made up of differ- The authors declare that they have no known competing finan-
ent types of collagen molecules, and undergoing chemical mod- cial interests or personal relationships that could have appeared to
ifications throughout their lifetime. Mechanically, collagen fibrils influence the work reported in this paper.
serve a multitude of functions such as providing toughness to
hard tissues, extensibility to lungs and skin, and transmission Acknowledgement
of forces from muscles to bones in tendons facilitating, in this
way, locomotion. The mechanical properties of individual colla- O.G.A and P.J.T. gratefully acknowledge funding from the Vi-
gen fibrils are influenced by their composition and chemical (post- enna Science and Technology Fund (WWTF), Grant No. LS19-035.
translational) modifications, i.e., the formation and density of enzy- The authors acknowledge TU Wien Bibliothek for financial support
matic and non-enzymatic cross-links. AFM-based nanoindentation through its Open Access Funding Programme.
results have shown that hydration, temperature, and collagen type
References
mixture influence indentation modulus, mostly via influencing the
molecular packing density. However, collagen fibrils transmit and [1] J. Myllyharju, K.I. Kivirikko, Collagens and collagen-related diseases, Ann.
withstand predominantly tensile forces. In tensile loading experi- Med. 33 (1) (2001) 7–21.
ments, isolated collagen fibrils show either a two- or a three-phase [2] S. Ricard-Blum, The collagen family, Cold Spring Harb. Perspect. Biol. 3 (1)
(2011) a004978.
behavior, which is linked to cross-link types and densities present. [3] P. Fratzl, Collagen: structure and mechanics, an introduction, in: Collagen,
To date, several deformation mechanisms have been suggested and, Springer, 2008, pp. 1–13.
in this context, phenomenological studies are complemented with [4] J.P. Orgel, T.C. Irving, A. Miller, T.J. Wess, Microfibrillar structure of type I col-
lagen in situ, Proc. Natl. Acad. Sci. 103 (24) (2006) 9001–9005.
molecular dynamics simulations. Existing results suggest that sev- [5] G. Cameron, I. Alberts, J. Laing, T.J. Wess, Structure of type I and type III het-
eral deformation mechanisms are present in different phases with erotypic collagen fibrils: an X-ray diffraction study, J. Struct. Biol. 137 (1–2)
one or two dominating certain phases. Studies have shown that (2002) 15–22.
[6] D.R. Keene, L.Y. Sakai, H.P. Bächinger, R.E. Burgeson, Type III collagen can be
hydration and cross-linking largely influence nanotensile mechan-
present on banded collagen fibrils regardless of fibril diameter, J. Cell Biol.
ics in terms of stiffness and strength. In addition, collagen fib- 105 (5) (1987) 2393–2402.
rils show both viscoelastic and inelastic behavior. Here, H-bond [7] J.M. Kirk, B.E. Heard, I. Kerr, M. Turner-Warwick, G.J. Laurent, Quantitation of
breakage and reformation as well as molecular sliding could be types I and III collagen in biopsy lung samples from patients with cryptogenic
fibrosing alveolitis, Coll. Relat. Res. 4 (3) (1984) 169–182.
the key mechanisms explaining rising stiffness with strain-rate and [8] C. Lovell, K. Smolenski, V. Duance, N. Light, S. Young, M. Dyson, Type I and III
the thixotropic behavior. Overall, the situation is likely more com- collagen content and fibre distribution in normal human skin during ageing,
plex. Collagen fibrils are composed of more than one collagen type, Br. J. Dermatol. 117 (4) (1987) 419–428.
[9] I. Williams, K. McCullagh, I. Silver, The distribution of types I and III collagen
which may influence enzymatic and non-enzymatic cross-linking and fibronectin in the healing equine tendon, Connect. Tissue Res. 12 (3–4)
and the extent of hydration. At the same time, chemical modifica- (1984) 211–227.
tions may also influence hydration, resulting in interrelated factors [10] G.B. Smejkal, C. Fitzgerald, Revised estimate of total collagen in the human
body, J. Biomed. Res. Environ. Sci. 3 (8) (2017) 0 01–0 02.
influencing the nanotensile mechanics. This is schematically shown [11] J. Jokinen, E. Dadu, P. Nykvist, J. Käpylä, D.J. White, J. Ivaska, P. Vehviläinen,
in Fig. 9. Given the already sparse data available, further studies H. Reunanen, H. Larjava, L. Häkkinen, Integrin-mediated cell adhesion to type
ideally coupled with chemical analysis and computer models are I collagen fibrils, J. Biol. Chem. 279 (30) (2004) 31956–31963.
[12] E. Hadjipanayi, V. Mudera, R.A. Brown, Guiding cell migration in 3D: a colla-
gen matrix with graded directional stiffness, Cell Motil. Cytoskeleton 66 (3)
(2009) 121–128.
[13] K. Poole, K. Khairy, J. Friedrichs, C. Franz, D.A. Cisneros, J. Howard, D. Mueller,
Molecular-scale topographic cues induce the orientation and directional
movement of fibroblasts on two-dimensional collagen surfaces, J. Mol. Biol.
349 (2) (2005) 380–386.
[14] J. Friedrichs, A. Taubenberger, C.M. Franz, D.J. Muller, Cellular remodelling of
individual collagen fibrils visualized by time-lapse AFM, J. Mol. Biol. 372 (3)
(2007) 594–607.
[15] C. Xue, J. Wyckoff, F. Liang, M. Sidani, S. Violini, K.-.L. Tsai, Z.-.Y. Zhang, E. Sa-
hai, J. Condeelis, J.E. Segall, Epidermal growth factor receptor overexpression
results in increased tumor cell motility in vivo coordinately with enhanced
intravasation and metastasis, Cancer Res. 66 (1) (2006) 192–197.
[16] J.E. Marturano, J.D. Arena, Z.A. Schiller, I. Georgakoudi, C.K. Kuo, Character-
ization of mechanical and biochemical properties of developing embryonic
tendon, Proc. Natl. Acad. Sci. 110 (16) (2013) 6370–6375.
[17] D.E. Birk, J.F. Southern, E.I. Zycband, J.T. Fallon, R.L. Trelstad, Collagen fibril
bundles: a branching assembly unit in tendon morphogenesis, Development
107 (3) (1989) 437–443.
[18] H.K. Graham, D.F. Holmes, R.B. Watson, K.E. Kadler, Identification of collagen
fibril fusion during vertebrate tendon morphogenesis. The process relies on
unipolar fibrils and is regulated by collagen-proteoglycan interaction, J. Mol.
Biol. 295 (4) (20 0 0) 891–902.
[19] X. Liu, X. Long, W. Liu, Y. Zhao, T. Hayashi, M. Yamato, K. Mizuno, H. Fujisaki,
S. Hattori, S.-i. Tashiro, Type I collagen induces mesenchymal cell differenti-
ation into myofibroblasts through YAP-induced TGF-β 1 activation, Biochimie
150 (2018) 110–130.
Fig. 9. Collagen fibril mechanics is influenced by a multitude of factors. The rela-
[20] E. Brauer, E. Lippens, O. Klein, G. Nebrich, S. Schreivogel, G. Korus, G.N. Duda,
tive amount and diversity of collagen molecules composing the fibril may influence
A. Petersen, Collagen fibrils mechanically contribute to tissue contraction in
what exactly chemical modifications take place, the interactions of water molecules an in vitro wound healing scenario, Advanced Science 6 (9) (2019) 1801780.
and hence hydration, which may be influenced by the chemical modifications as [21] M.J. Mienaltowski, D.E. Birk, Structure, physiology, and biochemistry of col-
well (Artist’s impression of collagen molecules and collagen fibril; courtesy of Lydia lagens, in: Progress in Heritable Soft Connective Tissue Diseases, 2014,
Andrioti). pp. 5–29.

47
O.G. Andriotis, M. Nalbach and P.J. Thurner Acta Biomaterialia 163 (2023) 35–49

[22] B. Suki, J.H. Bates, Extracellular matrix mechanics in lung parenchymal dis- [53] A. Masic, L. Bertinetti, R. Schuetz, S.-.W. Chang, T.H. Metzger, M.J. Buehler,
eases, Respir. Physiol. Neurobiol. 163 (1–3) (2008) 33–43. P. Fratzl, Osmotic pressure induced tensile forces in tendon collagen, Nat.
[23] E.R. Weibel, J. Gil, Structure-function relationships at the alveolar level, Bio- Commun. 6 (1) (2015) 1–8.
eng. Aspects Lung 3 (1977) 1–81. [54] M. Minary-Jolandan, M.-.F. Yu, Nanomechanical heterogeneity in the gap and
[24] S.T. Holgate, H.S. Arshad, G.C. Roberts, P.H. Howarth, P. Thurner, D.E. Davies, overlap regions of type I collagen fibrils with implications for bone hetero-
A new look at the pathogenesis of asthma, Clin. Sci. 118 (7) (2010) 439–450. geneity, Biomacromolecules 10 (9) (2009) 2565–2570.
[25] W. Manuyakorn, P.H. Howarth, S.T. Holgate, Airway remodelling in asthma [55] C.J. Peacock, L. Kreplak, Nanomechanical mapping of single collagen fibrils
and novel therapy, Asian Pac. J. Allergy Immunol. 31 (1) (2013) 3. under tension, Nanoscale 11 (30) (2019) 14417–14425.
[26] M.G. Jones, O.G. Andriotis, J.J. Roberts, K. Lunn, V.J. Tear, L. Cao, K. Ask, [56] A.S. Quigley, S. Bancelin, D. Deska-Gauthier, F. Légaré, L. Kreplak, S.P. Veres,
D.E. Smart, A. Bonfanti, P. Johnson, Nanoscale dysregulation of collagen struc- In tendons, differing physiological requirements lead to functionally distinct
ture-function disrupts mechano-homeostasis and mediates pulmonary fibro- nanostructures, Sci. Rep. 8 (1) (2018) 1–14.
sis, Elife 7 (2018) e36354. [57] A.S. Quigley, S. Bancelin, D. Deska-Gauthier, F. Légaré, S.P. Veres, L. Kreplak,
[27] D.M. Maurice, The structure and transparency of the cornea, J. Physiol. 136 Combining tensile testing and structural analysis at the single collagen fibril
(2) (1957) 263–286. level, Sci. Data 5 (1) (2018) 1–8.
[28] C. Boote, S. Dennis, R.H. Newton, H. Puri, K.M. Meek, Collagen fibrils appear [58] A.S. Quigley, S.P. Veres, L. Kreplak, Bowstring stretching and quantitative
more closely packed in the prepupillary cornea: optical and biomechanical imaging of single collagen fibrils via atomic force microscopy, PLoS One 11
implications, Invest. Ophthalmol. Vis. Sci. 44 (7) (2003) 2941–2948. (9) (2016) e0161951.
[29] M.J. Buehler, Nature designs tough collagen: explaining the nanostructure of [59] Z.L. Shen, M.R. Dodge, H. Kahn, R. Ballarini, S.J. Eppell, Stress-strain ex-
collagen fibrils, Proc. Natl. Acad. Sci. 103 (33) (2006) 12285–12290. periments on individual collagen fibrils, Biophys. J. 95 (8) (2008) 3956–
[30] S.-.W. Chang, S.J. Shefelbine, M.J. Buehler, Structural and mechanical differ- 3963.
ences between collagen homo-and heterotrimers: relevance for the molecular [60] Z.L. Shen, M.R. Dodge, H. Kahn, R. Ballarini, S.J. Eppell, In vitro fracture test-
origin of brittle bone disease, Biophys. J. 102 (3) (2012) 640–648. ing of submicron diameter collagen fibril specimens, Biophys. J. 99 (6) (2010)
[31] O. Andriotis, S. Chang, M. Vanleene, P. Howarth, D. Davies, S. Shefelbine, 1986–1995.
M. Buehler, P. Thurner, Structure–mechanics relationships of collagen fibrils [61] Z.L. Shen, H. Kahn, R. Ballarini, S.J. Eppell, Viscoelastic properties of isolated
in the osteogenesis imperfecta mouse model, J. R. Soc. Interface 12 (111) collagen fibrils, Biophys. J. 100 (12) (2011) 3008–3015.
(2015) 20150701. [62] R.B. Svensson, P. Hansen, T. Hassenkam, B.T. Haraldsson, P. Aagaard, V. Kova-
[32] R.B. Svensson, C.S. Eriksen, P.H. Tran, M. Kjaer, S.P. Magnusson, Mechanical nen, M. Krogsgaard, M. Kjaer, S.P. Magnusson, Mechanical properties of hu-
properties of human patellar tendon collagen fibrils. An exploratory study of man patellar tendon at the hierarchical levels of tendon and fibril, J. Appl.
aging and sex, J. Mech. Behav. Biomed. Mater. 124 (2021) 104864. Physiol. 112 (3) (2012) 419–426.
[33] R.B. Svensson, H. Mulder, V. Kovanen, S.P. Magnusson, Fracture mechanics of [63] R.B. Svensson, T. Hassenkam, C.A. Grant, S.P. Magnusson, Tensile Properties of
collagen fibrils: influence of natural cross-links, Biophys. J. 104 (11) (2013) Human Collagen Fibrils and Fascicles Are Insensitive to Environmental Salts,
2476–2484. Biophys. J. 99 (12) (2010) 4020–4027.
[34] R.B. Svensson, S.T. Smith, P.J. Moyer, S.P. Magnusson, Effects of maturation and [64] R.B. Svensson, T. Hassenkam, P. Hansen, S.P. Magnusson, Viscoelastic behavior
advanced glycation on tensile mechanics of collagen fibrils from rat tail and of discrete human collagen fibrils, J. Mech. Behav. Biomed. Mater. 3 (1) (2010)
Achilles tendons, Acta Biomater. 70 (2018) 270–280. 112–115.
[35] M. Asgari, N. Latifi, H.K. Heris, H. Vali, L. Mongeau, In vitro fibrillogenesis of [65] J.A. Van Der Rijt, K.O. Van Der Werf, M.L. Bennink, P.J. Dijkstra, J. Feijen, Mi-
tropocollagen type III in collagen type I affects its relative fibrillar topology cromechanical testing of individual collagen fibrils, Macromol. Biosci. 6 (9)
and mechanics, Sci. Rep. 7 (1) (2017) 1–10. (2006) 697–702.
[36] O.G. Andriotis, S. Desissaire, P.J. Thurner, Collagen fibrils: nature’s highly tun- [66] S.P. Veres, J.M. Harrison, J.M. Lee, Repeated subrupture overload causes pro-
able nonlinear springs, ACS Nano 12 (4) (2018) 3671–3680. gression of nanoscaled discrete plasticity damage in tendon collagen fibrils, J.
[37] O.G. Andriotis, K. Elsayad, D.E. Smart, M. Nalbach, D.E. Davies, P.J. Thurner, Orthop. Res. 31 (5) (2013) 731–737.
Hydration and nanomechanical changes in collagen fibrils bearing advanced [67] S.P. Veres, J.M. Harrison, J.M. Lee, Mechanically overloading collagen fibrils
glycation end-products, Biomed. Opt. Express. 10 (4) (2019) 1841–1855. uncoils collagen molecules, placing them in a stable, denatured state, Matrix
[38] S.J. Baldwin, L. Kreplak, J.M. Lee, Characterization via atomic force microscopy Biol. 33 (2014) 54–59.
of discrete plasticity in collagen fibrils from mechanically overloaded ten- [68] S.P. Veres, J.M. Lee, Designed to fail: a novel mode of collagen fibril disrup-
dons: nano-scale structural changes mimic rope failure, J. Mech. Behav. tion and its relevance to tissue toughness, Biophys. J. 102 (12) (2012) 2876–
Biomed. Mater. 60 (2016) 356–366. 2884.
[39] S.J. Baldwin, A.S. Quigley, C. Clegg, L. Kreplak, Nanomechanical mapping [69] M.P. Wenger, L. Bozec, M.A. Horton, P. Mesquida, Mechanical properties of
of hydrated rat tail tendon collagen I fibrils, Biophys. J. 107 (8) (2014) collagen fibrils, Biophys. J. 93 (4) (2007) 1255–1263.
1794–1801. [70] F. Yang, D. Das, K. Karunakaran, G.M. Genin, S. Thomopoulos, I. Chasiotis,
[40] M.J. Buehler, Nanomechanics of collagen fibrils under varying cross-link den- Nonlinear time-dependent mechanical behavior of Mammalian collagen fib-
sities: atomistic and continuum studies, J. Mech. Behav. Biomed. Mater. 1 (1) rils, Acta Biomater. (2022).
(2008) 59–67. [71] L. Yang, K.O. van der Werf, P.J. Dijkstra, J. Feijen, M.L. Bennink, Microme-
[41] B. Depalle, Z. Qin, S.J. Shefelbine, M.J. Buehler, Influence of cross-link struc- chanical analysis of native and cross-linked collagen type I fibrils supports
ture, density and mechanical properties in the mesoscale deformation mech- the existence of microfibrils, J. Mech. Behav. Biomed. Mater. 6 (2012) 148–
anisms of collagen fibrils, J. Mech. Behav. Biomed. Mater. 52 (2015) 1–13. 158.
[42] S. Eppell, B. Smith, H. Kahn, R. Ballarini, Nano measurements with micro-de- [72] L. Yang, K.O. van der Werf, C.F. Fitie, M.L. Bennink, P.J. Dijkstra, J. Feijen, Me-
vices: mechanical properties of hydrated collagen fibrils, J. R. Soc. Interface 3 chanical properties of native and cross-linked type I collagen fibrils, Biophys.
(6) (2006) 117–121. J. 94 (6) (2008) 2204–2211.
[43] P. Fratzl, K. Misof, I. Zizak, G. Rapp, H. Amenitsch, S. Bernstorff, Fibrillar struc- [73] L. Yang, K.O. van der Werf, B.F. Koopman, V. Subramaniam, M.L. Bennink,
ture and mechanical properties of collagen, J. Struct. Biol. 122 (1–2) (1998) P.J. Dijkstra, J. Feijen, Micromechanical bending of single collagen fibrils us-
119–122. ing atomic force microscopy, J. Biomed. Mater. Res. A 82 (1) (2007) 160–168.
[44] A. Gautieri, S. Vesentini, A. Redaelli, M.J. Buehler, Hierarchical structure and [74] S.R. Inamdar, D.P. Knight, N.J. Terrill, A. Karunaratne, F. Cacho-Nerin,
nanomechanics of collagen microfibrils from the atomistic scale up, Nano M.M. Knight, H.S. Gupta, The secret life of collagen: temporal changes in
Lett. 11 (2) (2011) 757–766. nanoscale fibrillar pre-strain and molecular organization during physiological
[45] C.A. Grant, D.J. Brockwell, S.E. Radford, N.H. Thomson, Effects of hydration loading of cartilage, ACS Nano 11 (10) (2017) 9728–9737.
on the mechanical response of individual collagen fibrils, Appl. Phys. Lett. 92 [75] J. Heino, The collagen family members as cell adhesion proteins, Bioessays 29
(2008) 233902. (10) (2007) 1001–1010.
[46] C.A. Grant, D.J. Brockwell, S.E. Radford, N.H. Thomson, Tuning the elastic mod- [76] D.P. McDaniel, G.A. Shaw, J.T. Elliott, K. Bhadriraju, C. Meuse, K.-.H. Chung,
ulus of hydrated collagen fibrils, Biophys. J. 97 (11) (2009) 2985–2992. A.L. Plant, The stiffness of collagen fibrils influences vascular smooth muscle
[47] C.A. Grant, M.A. Phillips, N.H. Thomson, Dynamic mechanical analysis of col- cell phenotype, Biophys. J. 92 (5) (2007) 1759–1769.
lagen fibrils at the nanoscale, J. Mech. Behav. Biomed. Mater. 5 (1) (2012) [77] M. Milazzo, A. David, G.S. Jung, S. Danti, M.J. Buehler, Molecular origin
165–170. of viscoelasticity in mineralized collagen fibrils, Biomater. Sci. 9 (9) (2021)
[48] A.J. Heim, T.J. Koob, W.G. Matthews, Low strain nanomechanics of collagen 3390–3400.
fibrils, Biomacromolecules 8 (11) (2007) 3298–3301. [78] M. Milazzo, G.S. Jung, S. Danti, M.J. Buehler, Mechanics of mineralized colla-
[49] A.J. Heim, W.G. Matthews, T.J. Koob, Determination of the elastic modulus of gen fibrils upon transient loads, ACS Nano 14 (7) (2020) 8307–8316.
native collagen fibrils via radial indentation, Appl. Phys. Lett. 89 (18) (2006) [79] N. Rajan, J. Habermehl, M.-.F. Coté, C.J. Doillon, D. Mantovani, Preparation of
181902. ready-to-use, storable and reconstituted type I collagen from rat tail tendon
[50] S.A. Iqbal, D. Deska-Gauthier, L. Kreplak, Assessing collagen fibrils molecu- for tissue engineering applications, Nat. Protoc. 1 (6) (2006) 2753–2758.
lar damage after a single stretch–release cycle, Soft Matter 15 (30) (2019) [80] D. Holmes, M. Capaldi, J. Chapman, Reconstitution of collagen fibrils in vitro;
6237–6246. the assembly process depends on the initiating procedure, Int. J. Biol. Macro-
[51] J. Liu, D. Das, F. Yang, A.G. Schwartz, G.M. Genin, S. Thomopoulos, I. Chasi- mol. 8 (3) (1986) 161–166.
otis, Energy dissipation in mammalian collagen fibrils: cyclic strain-induced [81] O.G. Andriotis, W. Manuyakorn, J. Zekonyte, O.L. Katsamenis, S. Fabri,
damping, toughening, and strengthening, Acta Biomater. 80 (2018) 217–227. P.H. Howarth, D.E. Davies, P.J. Thurner, Nanomechanical assessment of hu-
[52] Y. Liu, R. Ballarini, S.J. Eppell, Tension tests on mammalian collagen fibrils, man and murine collagen fibrils via atomic force microscopy cantilever-based
Interface Focus 6 (1) (2016) 20150080. nanoindentation, J. Mech. Behav. Biomed. Mater. 39 (2014) 9–26.

48
O.G. Andriotis, M. Nalbach and P.J. Thurner Acta Biomaterialia 163 (2023) 35–49

[82] P. Hansen, T. Hassenkam, R.B. Svensson, P. Aagaard, T. Trappe, B.T. Haraldsson, [96] M. Saito, K. Marumo, Collagen cross-links as a determinant of bone quality:
M. Kjaer, P. Magnusson, Glutaraldehyde cross-linking of tendon—Mechanical a possible explanation for bone fragility in aging, osteoporosis, and diabetes
effects at the level of the tendon fascicle and fibril, Connect. Tissue Res. 50 mellitus, Osteop. Int. 21 (2) (2010) 195–214.
(4) (2009) 211–222. [97] V.M. Monnier, G.T. Mustata, K.L. Biemel, O. Reihl, M.O. Lederer, D. Zhenyu,
[83] M. Nalbach, F. Chalupa-Gantner, F. Spoerl, V. de Bar, B. Baumgartner, O.G. An- D.R. Sell, Cross-linking of the extracellular matrix by the maillard reaction
driotis, S. Ito, A. Ovsianikov, G. Schitter, P.J. Thurner, Instrument for tensile in aging and diabetes: an update on “a puzzle nearing resolution, Ann. N. Y.
testing of individual collagen fibrils with facile sample coupling and uncou- Acad. Sci. 1043 (1) (2005) 533–544.
pling, Rev. Sci. Instrum. 93 (5) (2022) 054103. [98] H. Birch, T. Smith, E. Draper, A. Bailey, N. Avery, A. Goodship, Collagen
[84] J.L. Tan, J. Tien, D.M. Pirone, D.S. Gray, K. Bhadriraju, C.S. Chen, Cells lying on crosslink profile relates to tendon material properties, Matrix Biol. (25)
a bed of microneedles: an approach to isolate mechanical force, Proc. Natl. (2006) S74.
Acad. Sci. 100 (4) (2003) 1484–1489. [99] M.J. Buehler, Atomistic and continuum modeling of mechanical properties of
[85] O. Du Roure, A. Saez, A. Buguin, R.H. Austin, P. Chavrier, P. Siberzan, B. Ladoux, collagen: elasticity, fracture, and self-assembly, J. Mater. Res. 21 (8) (2006)
Force mapping in epithelial cell migration, Proc. Natl. Acad. Sci. 102 (7) 1947–1961.
(2005) 2390–2395. [100] S.G. Uzel, M.J. Buehler, Molecular structure, mechanical behavior and failure
[86] S.J. Baldwin, L. Kreplak, J.M. Lee, MMP-9 selectively cleaves non-D-banded mechanism of the C-terminal cross-link domain in type I collagen, J. Mech.
material on collagen fibrils with discrete plasticity damage in mechanical- Behav. Biomed. Mater. 4 (2) (2011) 153–161.
ly-overloaded tendon, J. Mech. Behav. Biomed. Mater. 95 (2019) 67–75. [101] R. Usha, V. Subramanian, T. Ramasami, Role of secondary structure on the
[87] C. Couppé, R.B. Svensson, S.V. Skovlund, J.K. Jensen, C.S. Eriksen, N.M. Malm- stress relaxation processes in rat tail tendon (RTT) collagen fibre, Macromol.
gaard-Clausen, J.D. Nybing, M. Kjaer, S.P. Magnussson, Habitual side-specific Biosci. 1 (3) (2001) 100–107.
loading leads to structural, mechanical, and compositional changes in the [102] M.E. Launey, M.J. Buehler, R.O. Ritchie, On the mechanistic origins of tough-
patellar tendon of young and senior lifelong male athletes, J. Appl. Physiol. ness in bone, Annu. Rev. Mater. Res. 40 (2010) 25–53.
131 (4) (2021) 1187–1199. [103] F.H. Silver, A. Ebrahimi, P.B. Snowhill, Viscoelastic properties of self-assem-
[88] G.D. Fullerton, A. Rahal, Collagen structure: the molecular source of the ten- bled type I collagen fibers: molecular basis of elastic and viscous behaviors,
don magic angle effect, J. Magn. Reson. Imaging 25 (2) (2007) 345–361. Connect. Tissue Res. 43 (4) (2002) 569–580.
[89] R. Fraser, T. MacRae, A. Miller, E. Suzuki, Molecular conformation and packing [104] N. Sasaki, S. Odajima, Elongation mechanism of collagen fibrils and force-s-
in collagen fibrils, J. Mol. Biol. 167 (2) (1983) 497–521. train relations of tendon at each level of structural hierarchy, J. Biomech. 29
[90] H. Bueckle, J. Westbrook, H. Conrad, The Science of Hardness Testing and (9) (1996) 1131–1136.
Its Research Applications, American Society for Metals, Materials Park, Ohio, [105] M. Handelshauser, M. Marchetti-Deschmann, P.J. Thurner, O.G. Andriotis, Col-
1973. lagen Fibril Tensile Response Described by a Nonlinear Maxwell Fluid Model,
[91] R.B. Svensson, A. Herchenhan, T. Starborg, M. Larsen, K.E. Kadler, K. Qvortrup, Available at SSRN 3940868.
S.P. Magnusson, Evidence of structurally continuous collagen fibrils in ten- [106] A.S. Wineman, K.R. Rajagopal, Mechanical Response of Polymers: An Intro-
dons, Acta Biomater. 50 (2017) 293–301. duction, Cambridge University Press, 20 0 0.
[92] Z. Zhou, M. Minary-Jolandan, D. Qian, A simulation study on the signifi- [107] A. Gautieri, M.J. Buehler, A. Redaelli, Deformation rate controls elasticity and
cant nanomechanical heterogeneous properties of collagen, Biomech. Model. unfolding pathway of single tropocollagen molecules, J. Mech. Behav. Biomed.
Mechanobiol. 14 (3) (2015) 445–457. Mater. 2 (2) (2009) 130–137.
[93] J.L. Zitnay, Y. Li, Z. Qin, B.H. San, B. Depalle, S.P. Reese, M.J. Buehler, S.M. Yu, [108] A. Gautieri, S. Vesentini, A. Redaelli, M.J. Buehler, Viscoelastic properties
J.A. Weiss, Molecular level detection and localization of mechanical damage in of model segments of collagen molecules, Matrix Biol. 31 (2) (2012) 141–
collagen enabled by collagen hybridizing peptides, Nat. Commun. 8 (1) (2017) 149.
1–12. [109] H. Ghodsi, K. Darvish, Investigation of mechanisms of viscoelastic behavior of
[94] G.D. Fullerton, M.R. Amurao, Evidence that collagen and tendon have mono- collagen molecule, J. Mech. Behav. Biomed. Mater. 51 (2015) 194–204.
layer water coverage in the native state, Cell Biol. Int. 30 (1) (2006) 56–65. [110] G.I. Bell, Models for the specific adhesion of cells to cells: a theoretical
[95] J. Bella, B. Brodsky, H.M. Berman, Hydration structure of a collagen peptide, framework for adhesion mediated by reversible bonds between cell surface
Structure 3 (9) (1995) 893–906. molecules, Science 200 (4342) (1978) 618–627.

49