characterization via atomic force microscopy of discrete...

Available online at www.sciencedirect.com

www.elsevier.com/locate/jmbbm

j o u r n a l o f t h e m e c h a n i c a l b e h a v i o r o f b i o m e d i c a l m a t e r i a l s 6 0 ( 2 0 1 6 ) 3 5 6 – 3 6 6

http://dx.doi.org/10.1751-6161/& 2016 El

Abbreviations: A

buffered saline; SEMnCorresponding a

Canada B3H 4R2. TeE-mail address:

Research Paper

Characterization via atomic force microscopyof discrete plasticity in collagen fibrilsfrom mechanically overloaded tendons: Nano-scalestructural changes mimic rope failure

Samuel J. Baldwina, Laurent Kreplaka,b, J. Michael Leeb,n

aDepartment of Physics and Atmospheric Science, Dalhousie University, Halifax, NS, CanadabSchool of Biomedical Engineering and Department of Applied Oral Sciences, Dalhousie University, 5981 UniversityAvenue, PO Box 15000, Halifax, NS, Canada B3H 4R2

a r t i c l e i n f o

Article history:

Received 2 November 2015

Received in revised form

22 January 2016

Accepted 3 February 2016

Available online 10 February 2016

Keywords:

Collagen fibrils

Discrete plasticity

Atomic force microscopy

Molecular organization

1016/j.jmbbm.2016.02.004sevier Ltd. All rights rese

FM, atomic force micro

, scanning electron miuthor at: School of Biomel.: þ1 902 494 6734; fax: þ[email protected] (J.M.

a b s t r a c t

Tendons exposed to tensile overload show a structural alteration at the fibril scale termed

discrete plasticity. Serial kinks appear along individual collagen fibrils that are susceptible

to enzymatic digestion and are thermally unstable. Using atomic force microscopy we

mapped the topography and mechanical properties in dehydrated and hydrated states of

25 control fibrils and 25 fibrils displaying periodic kinks, extracted from overloaded bovine

tail tendons. Using the measured modulus of the hydrated fibrils as a probe of molecular

density, we observed a non-linear negative correlation between molecular density and

kink density of individual fibrils. This is accompanied by an increase in water uptake with

kink density and a doubling of the coefficient of variation of the modulus between kinked,

and control fibrils. The mechanical property maps of kinked collagen fibrils show radial

heterogeneity that can be modeled as a high-density core surrounded by a low-density

shell. The core of the fibril contains the kink structures characteristic of discrete plasticity;

separated by inter-kink regions, which often retain the D-banding structure. We propose

that the shell and kink structures mimic characteristic damage motifs observed in laid rope

strands.

& 2016 Elsevier Ltd. All rights reserved.

rved.

scopy; PF-QNM, peak force quantitative nanomechanical mapping; PBS, phosphate

croscopydical Engineering, Dalhousie University, 5981 University Avenue, PO Box 15000, Halifax, NS,1 902 494 6621.Lee).

http://dx.doi.org/10.1016/j.jmbbm.2016.02.004




http://crossmark.crossref.org/dialog/?doi=10.1016/j.jmbbm.2016.02.004&domain=pdf

mailto:[email protected]


j o u r n a l o f t h e m e c h a n i c a l b e h a v i o r o f b i o m e d i c a l m a t e r i a l s 6 0 ( 2 0 1 6 ) 3 5 6 – 3 6 6 357

1. Introduction

Tendons are tension-bearing tissues that convert musclecontraction into skeletal movement used for every daylocomotion. The functional attributes of tendon are the resultof a highly organized hierarchical structure of extracellularmatrix proteins. The most mechanically structural element iscollagen, its most fundamental structural unit being thenano-scaled fibril (Benjamin et al., 2008; Kannus, 2001) Injurydue to mechanical overload of a tendon results in thedisruption of its ordered hierarchical structure, long-termalteration in its mechanical properties, and loss of function(Sharma and Maffulli, 2005; Spiesz et al., 2015). A structuralmotif for collagen damage due to tendon overload hasrecently been identified. It is characterized by a serial disrup-tion (kinking) of collagen fibril structure along their length(Veres and Lee, 2012). Termed as discrete plasticity, suchdamage is associated with increased enzyme susceptibility,decreased thermal stability and denaturation of collagen, andcellular recognition of locally kinked fibrils (Veres and Lee,2012; Veres et al., 2013, 2014, 2015).

The native structure of a collagen fibril (in bovine tendon)consists of tropocollagen molecules, self-assembled into fivemolecule cross-section microfibrils approximately 4 nm indiameter, these then assembled into sub-fibrils approxi-mately 10–40 nm in diameter (Wess, 2005). It has beendemonstrated that sub-fibrils and micro-fibrils are woundabout the axis of the fibril in a manner similar to that in a laidrope strand (Bozec et al., 2007; Hulmes et al., 1995; McKennaet al., 2004; Ottani et al., 2002; Silver et al., 1992; Zhao et al.,2011). Characteristic, periodic D-banding is observed alongthe length of native fibrils. This appearance is a result of thenon-integer staggering of tropocollagen molecules withinmicrofibrils, resulting in molecular density fluctuations alongthe fibril's length. A single period of the D-band is �67 nm intendon fibrils and consists of a gap and overlap componentwhere the gap has 4/5's of the molecular density of theoverlap region (Hodge and Schmitt, 1960; Orgel et al., 2001;Wess et al., 2005).

It is now clear that overload damage in a tendon involvesalteration to molecular conformation or packing order at thenano-scale within individual collagen fibrils (Veres and Lee,2012; Veres et al., 2013, 2014, 2015). In particular, twosignificant structural alterations to fibril morphology havebeen reported in association with discrete plasticity underscanning electron microscopy (SEM): (i) the appearance of ashell layer which obscures or replaces D-banding, and (ii)serial kink sites along the fibril's length. These changesappear to relate to radial and longitudinal alterations respec-tively within the collagen fibril (Veres and Lee, 2012). To thispoint in time, discrete plasticity has been observed onlyunder SEM, and that after dehydration and sputter coatingto achieve nanometer resolution. In the present study, wehave sought to circumvent these alterations to collagen fibrilsbefore observation by application of atomic forcemicroscopy (AFM).

We have previously demonstrated acquisition of high-resolution mechanical maps from hydrated collagen fibrilsvia AFM, using the peak force, quantitative nanomechanical

mapping (PF-QNM) mode (Baldwin et al., 2014). From theacquired data we are able to calculate the maximum pene-tration depth of the tip into the collagen fibril, and the fibril'smodulus, all with a spatial resolution of around 10 nm. Thesemechanical properties were shown to be a measure of thelocal molecular density of the fibril. By this means, thepenetration depth and modulus images visualize the interiorof the fibril at distinct depths. In the present study, we haveused this technique to analyze 25 control and 25 overload-kinked collagen fibrils, seeking to further our understandingof the molecular alterations in collagen fibrils associated withdiscrete plasticity damage. Based on the resulting data, andmaking use of previous description of the collagen fibril as ananoscale rope and/or as a tube-like structure, we suggest anew model of kink formation which resembles damagemotifs found in laid rope strands (Bozec et al., 2007;Gutsmann et al., 2003; McKenna et al., 2004).

2. Methods

2.1. Sample acquisition

Tissue harvesting and handling were approved by the HealthSciences Research Ethics Board of Dalhousie University.Bovine tails were acquired from 18 to 24 month old steersimmediately after slaughter for meat at a local abattoir. Thetails were stored for no more than 2 h at 4 1C, after which thetendons were dissected from the dorsal region of each. In thisstudy tail tendons from 5 animals were used, with a singletendon sourced from each tail.

2.2. Tendon preparation

A segment 1 cm in length was cut from each tendon to serveas a control and stored in phosphate-buffered saline solution(PBS) with a pH of 7.4 at 4 1C. The remainder of the tendon wasthen clamped into an MTS Series 458 servo-hydraulic testingmachine for mechanical loading. Each tendon underwent apre-loading cycle at 1% strain per second to a maximum strainof 10%. This was followed by 5 overload cycles into the plasticregion of the load–deformation curve under computer controlas previously described (Veres et al., 2013). This repeatedoverload regime has been demonstrated to produce the mostcollagen fibrils displaying discrete plasticity in the tendoncollagen while retaining a wide range of kink density amongthe individual fibrils. During loading, the tendon samples werekept hydrated by continuous application of PBS with a pipette.At completion of the loading procedure, the overloaded tendonwas removed from the apparatus, and the clamped ends werecut away. Both the control tendon sample and the trimmedoverload samples were then subjected to a decellularizationprocess. (See Fig. 1.)

2.3. Decellularization

Decellularization was undertaken to allow clear visualization ofindividual collagen fibrils and has previously been demonstratedto preserve discrete plasticity damage (Veres et al., 2015). Theprocess has been fully described previously (Ariganello et al.,

Fig. 1


2009). It began with treatment in a hypotonic cell-lysing solutionlasting 36 h at 4 1C, followed by exposure to a Triton X-100solution at 4 1C for 24 h, a sterile 3 h wash in a DNase/RNasesolution at 37 1C, and finally a sterile 24 h wash in Triton X-100solution at room temperature. All solutions were replaced withfresh solution every 12 h where required. Following the finalwash in the Triton X-100 solution, samples were rinsed in sterilePBS, in antifungal/antibiotic solution, and twice in a sterile PBS/antibiotic solution, each wash occurring at room temperatureand lasting 30min. A final wash in sterile PBS/antibiotic solutionat room temperature lasted 48 h and occurred before storage inPBS/antibiotic solution at 4 1C. The maximum storage time forthe now-sterile, decellularized tendon samples never exceeded3 weeks.

2.4. AFM sample preparation

AFM sample preparation followed the method described byBaldwin et al. (2014). Decellularized tendon samples weretaken from storage and placed on a plastic dissection dish in1 ml PBS at pH 7.4. Collagen fibrils were dissected from thetendons using tweezers and a thin glass needle. After suffi-cient dissection, the PBS/fibril solution was poured onto aglass-bottomed dish and allowed to settle for a minimum of45 min to facilitate fibril adsorption to the bottom of the dish.These AFM samples were then rinsed 3 times with deionizedwater and dried under a flow of nitrogen for 10 min. At thispoint the samples were ready for AFM imaging.

2.5. AFM imaging

AFM imaging was performed with a Bioscope Catalyst atomicforce microscope (Bruker, USA) mounted on an IX71 invertedmicroscope (Olympus, Japan) equipped with differentialinterference contrast (DIC) and a 100� , 1.3 NA, oil immersion

objective, which allowed optical targeting of single collagenfibrils. ScanAsyst liquidþ cantilevers (Bruker, USA) with aspring constant of �0.7 N/m, resonance frequency of �150

KHz, tip radius of �2 nm, and cone half-angle of 17.51, wereused in this work. All cantilevers were calibrated usingabsolute calibration. The deflection sensitivity measured onfused silica and spring constant was determined using thethermal tune component of the Nanoscope 8.15 software(Bruker) which makes use of the thermal noise method ofcantilever calibration (Sader et al., 1995). All AFM images wereacquired in peak force quantitative nanomechanical mappingmode (PF-QNM) with a tip velocity of 1.2 mm/s and a peakforce set point of 10 nN, resulting in a 256�256 grid of forcedisplacement curves allowing analysis of the sample areasmechanical properties (Baldwin et al., 2014).

2.6. AFM data acquisition

In this study, ten fibrils from each of the 5 tendons wereimaged along 10 mm of their length. We selected five fibrilsfrom each control sample and five fibrils from each loadedsample which displayed sites of discrete plasticity that wewill subsequently refer to as kinks. In total, we imaged 50fibrils, 25 control and 25 kinked for a total axial length of

500 mm. Three essential steps were followed in this order forall data acquisition: (i) optical targeting of dehydrated col-lagen fibrils, (ii) AFM imaging of each dehydrated fibril usingPF-QNM with a pixel size of 20 nm, and (iii) AFM imaging ofrehydrated fibrils using PF-QNM imaging with a pixel size of8 nm. Imaging of hydrated collagen fibrils was carried out inPBS with at pH 7.4 and at room temperature. All collagenfibrils were exposed to a single dehydration step duringsample preparation and dehydrated imaging, followed by asingle hydration step prior to hydrated imaging. Sampleswere allowed 30 min to rehydrate prior to hydrated imaging.

2.7. Data analysis

All data analysis was performed using SPIP 6.3.3 software(Image Metrology). From the acquired data, 6 properties of thecollagen fibrils were of interest: the height and deformationof the fibrils in both hydrated and dehydrated states, themodulus of fibrils in the hydrated state, and the averageserial kink density (kinks/mm) along each fibril over a 10 mmlength. The deformation of a fibril at any point was defined

from the force/distance data as the distance between theextension curve and maximum indentation at 15% of thepeak force (max force). (Fig. 2). All height and deformationdata were extracted from a 20 nm wide strip along the apex ofthe collagen fibril. The deformations in the hydrated anddehydrated states were summed with the correspondingheight measure to provide a value termed the zero-forceheight. This measure represented the corrected height of thefibril in the absence of applied force due to the AFM tip(Baldwin et al., 2014).

Using the averages of the hydrated and dehydrated zero-force heights, a swelling ratio was calculated for each fibril asshown in Eq.(1):


Swelling Ratio¼ Hydrated Zero Force HeightDehydrated Zero Force Height

ð1Þ

This ratio provided insight into the fibril's interaction with

the surrounding PBS medium. The modulus along the radial

direction of the hydrated collagen fibrils was calculated by

fitting the Sneddon model as described in Eq.(2):

F¼ 2Eπð1�υ2Þ ð tan δ2Þ ð2Þ

Fig. 2

Fig.

where F is the applied force, ν is the Poisson ratio, α is the half

angle of the cone, δ is the depth of indentation into the

sample and E is the elastic modulus of the sample (Maugis

and Barquins, 1978; Sneddon, 1965; Stolz et al., 2004). The

Poisson ratio of the fibril was taken to be 0.5, characteristic of

incompressibility (Baldwin et al., 2014; Grant et al., 2009). The

modulus obtained by fitting Eq. 2 serves as a probe for the

molecular density of the collagen fibril (Baldwin et al., 2014).

The Sneddon model was selected for modulus analysis due to

the AFM tip geometry and a minimum indentation depth of

�30 nm upon the apex of the control collagen fibrils. The fit

range of each force curve was determined using Bueckles'

rule, corresponding to a maximum fit range equivalent to 10%

of the fibril's zero-force height (Bueckle, 1973; Persch et al.,

1994). This process produced maps of the radial modulus of

the collagen fibril from which values were extracted along a

20 nm wide strip along the apex of the fibril. Outliers in the

modulus of a single fibril were determined using JMP software

(version 11.0.0, SAS Institute Inc.) and removed from the data

set. The average modulus and the coefficient of variation of

the modulus were calculated for each fibril. The average

modulus was used to quantify the molecular density of the

fibrils as justified in Baldwin et al., 2014, while the coefficient

of variation of the modulus provided a measure of the

heterogeneity of the molecular density for the collagen fibril.

3


The average, serial kink density along each fibril wasdetermined by counting the number of kinks along a 10 mmlength. This was performed on 10 mmx10 mm dehydratedheight images of kinked fibrils where a kink was classifiedas a transversal fault line on the fibril (Fig. 3). Kinked fibrilswere split into two separate groups: kinked (average kinkdensity less than 1.5 kinks/mm), and very kinked (averagekink density greater than 1.5 kinks/mm).

2.8. Image analysis

To visualize the data, maps of height, deformation, andmodulus were extracted using SPIP 6.3.3 and stitchedtogether in Adobe Photoshop. All cross-sectional data pre-sented were extracted along a 20 nm-wide strip along theapex of the corresponding collagen fibril.

2.9. Statistical analysis

Statistical analysis was performed using JMP software (ver-sion 11.0.0, SAS Institute Inc.). A 2-way ANOVA was per-formed for swelling ratio, modulus, and coefficient ofvariance data with variables of five animal tails and threekink density groups (control, kinked, very kinked). Whereappropriate after the 2-way ANOVA, Bonferroni-correctedpost-hoc testing was carried out using Fisher's Least

Fig.

Significant Difference (LSD) test. All parametric data arepresented as the mean7the standard deviation.

3. Results

Topographic mapping of dehydrated collagen fibrils extractedfrom control tendon samples was free of curvature andshowed the periodic D-banding pattern indicative of proper,native molecular packing within the fibrils (Fig. 3A). Bycontrast, fibrils extracted from overloaded tendon displaydiscrete plasticity damage, following a twisted path withthe D-band interrupted at sites of sudden directional changesin the fibril axis (Fig. 3B). These sites were consistent with thekink structures previously observed under SEM (Veres andLee, 2012; Veres et al., 2013, 2014, 2015). The persistence of theD-banding between kinks suggests retention of orderedmolecular packing in these regions (Fig. 3B, arrows).

Upon hydration in PBS, both control and damaged fibrilsswelled significantly (Fig. 3). In the hydrated state, D-bandingwas barely observable for the control fibrils (5–10 nm fluctua-tion) and not observable for the kinked collagen fibrils (Fig. 3Cand D). This observation suggests that the gap and overlapregions of the D-band have distinct swelling ratios in PBS,resulting in a nearly constant cross-sectional area along thelength of control fibrils. The heterogeneity in swelling does,

4


however, permit observation of D-banding in the modulus

maps of hydrated collagen fibrils (Baldwin et al., 2014;

Spitzner et al. 2015). Indeed, in the present study, D-

banding of control fibrils was observed in both deformation

and modulus images (Fig. 4A and C). Kinked fibrils, by

contrast, often showed no D-banding in either deformation

or modulus images (Fig. 4B and D). Instead, we observed a

highly penetrable shell layer enveloping the collagen fibril

(Fig. 4B, arrow) and a new periodic modulus fluctuation

occurring at the kink sites along the collagen fibril (Fig. 4D).

An inverse relationship is expected between the deformation

and modulus values of a collagen fibril. A highly deformed

region should have a lower modulus and vice versa. This

relationship broke down in the kinked regions of damaged

collagen fibrils (Fig. 4B and D, rectangles) where a decrease in

modulus was observed in the kinked region —without an

increase in deformation. This observation is an artifact, due

to the method in which the deformation is measured on force

curves taken from an area where high penetration occurs.Comparison of modulus images from control, kinked, and

very kinked collagen fibrils demonstrated the progression of

damage as serial kink density increased along collagen fibrils

(Fig. 5). No correlation between tendon source and a region of

the kink density range was observed, demonstrating no

dependence on kink density and the macroscale loading of

the tendons. Again, not all kinked collagen fibrils displayed a

complete lack of D-banding when visualized via modulus

(Fig. 5B and E). Plotting the modulus, coefficient of variation of

the modulus, and the swelling ratio of the collagen fibrils

Fig.

against serial kink density revealed an increase in swellingratio and decrease in modulus with increasing kink density(Fig. 6A, C and E). This is consistent with the notion that adecrease in modulus is due to a decrease in moleculardensity. Comparison of the mean properties from the threefibril groups showed significant, systematic differencesbetween the control, kinked, and very kinked collagen fibrils(Fig. 6B, D and F; Table 1). The modulus was distinct betweeneach of the three groups, while the swelling ratio increasedsignificantly between the control and very kinked fibrils(Fig. 6B and F). It was striking that the heterogeneity inmodulus changed so greatly with damage: the coefficient ofvariation for modulus in control fibrils was more thandoubled in both kinked and very kinked groups, largelyindependent of kink density (Fig. 6C; Table 1).

4. Discussion

Our previous work has addressed two basic questions. First,what does it mean for collagen to be mechanically damagedin an overload event? Second are there identifiable structuralmotifs, at some characteristic scale, which are associatedwith damage? These are questions which are important forour understanding of evolutionary “design” in collagen andfor biomedical identification and treatment of soft tissueinjuries. Examining overload-damaged bovine tail tendonswith SEM, we have identified a characteristic, nano-scaledmotif for damage in the fundamental collagen fibril: localkinks which increased in serial density under increasing

5

Fig. 6

Table 1 – Fibril properties. Mean Sneddon modulus,coefficient of variation of the modulus, and the swellingratio for the control, kinked, and very kinked collagenfibril groups. Mean7standard deviation.

Kinkdensity

Sneddonmodulus (MPa)

Coefficient ofvariation formodulus

Swellingratio

Control 17.373.9 0.2070.02 1.970.2Kinked 6.672.1* 0.4370.04* 2.270.4Verykinked

2.171.2*† 0.4670.03* 2.470.4*

* Difference with value for control fibrils is significant with po0.05,Bonferroni-corrected.

† Difference with value for kinked group is significant with po0.05,Bonferroni-corrected.


cycles of plastic overloading (Veres and Lee, 2012; Veres et al.,2013). As well, treatment with the acetyl trypsin, a serineprotease, demonstrated that molecular collagen was dena-tured during this process, gradually obscuring D-banding inSEM images and occurring most intensely at the kink sites(Veres and Lee, 2012; Veres et al., 2013). Mechanical evidencesuggests that these kinks may be part of a structuralmechanism which absorbs overload energy via sacrificialdestruction of certain subfibrils, toughening collagen at avery basic level. Due to the very local nature of the kinkformation, we have termed this damage mechanism “discreteplasticity”. Culture of decellularized native and damagedtendons with immortal macrophage-like U937 cells has giventhe first evidence that inflammatory cells may be able torecognize and respond to discrete plasticity damage (Vereset al., 2015).

We recognized that the kink morphology observed by SEMonly provides surface characterization of the collagen fibrilsfollowing significant alteration due to dehydration and metalsputtering. In the present study, we have sought to removethis barrier and to look at native and damaged fibrils in anunfixed, hydrated state. By comparing features in dehydratedand hydrated specimens, we directly examine the impor-tance of water solvation in our imaging. Further, by carryingout nano-scaled mechanical examination of fibrils, we areable to probe the interior of damaged fibrils. In short, AFMoffered the opportunity to determine whether the kinkmorphology of discrete plasticity damage was artifactual toany significant degree, and to greatly expand our structural

understanding of accumulating mechanical damage at thefibril level in collagen.

4.1. Hydration reveals structural features associated withdiscrete plasticity

Comparing the dehydrated height, hydrated height, modulus,and deformation of kinked and control collagen fibrilsrevealed alteration of molecular packing due to tendon over-load. Newly-validated structures include a low-density shellwhich encompassed a higher-density core in the damagedcollagen fibril. SEM images under acetyl trypsin digestionsuggest that this layer is composed of denatured collagen(Veres et al., 2013,, 2014). Along the core of the fibril, theperiodic kink structures of discrete plasticity have beenconfirmed, as have the relatively undamaged inter-kinkregions spanning consecutive kinks (Fig. 4). The kink struc-tures are mechanically characterized by a sharp decrease inmodulus compared to the inter-kink regions and an increasein deformation by the tip (Fig. 4). The alteration in mechan-ical response of the shell and kink sites is most likely due touptake of water associated with the loss of native molecularpacking and local denaturation of triple-helical tropocollagen.This is supported by the preferential enzymatic digestion ofmaterial from the shell and kink structures by trypsin, as wellas the decrease in thermal stability of the overloaded tendon(Veres and Lee, 2012; Veres et al., 2014).

The dramatic increase in the coefficient of variation of themodulus between the control fibrils and discrete plasticity-kinked fibrils (Fig. 6D) suggests that the kink/inter-kinkmodulus difference is conserved, independent of kink den-sity—that is, a kink is a local structure which has limitedinfluence on the mechanical behavior of neighboring regionsin the fibril. This finding contrasts with the decrease in themean modulus of damaged fibrils as kink density increases(Fig. 6A). Fitting the mean modulus versus kink density datawith a linear (r2¼0.042) and an exponential fit (r2¼0.62)demonstrates the mean modulus is not solely due to anaccumulation of discrete kink structures. It follows that,while the kink/inter-kink heterogeneity was captured by thecoefficient of variation, the fall in mean modulus withincreasing kink density is associated with both the formationof the shell structure and accumulation of kink sites (Fig. 7).

The D-banding seen in the modulus maps of hydratedcontrol bovine tail collagen fibrils is similar to that seen in rattail collagen fibrils (Baldwin et al., 2014). Increasing levels of

Fig. 7

Fig. 8

Fig. 9


damage in various tendon overload fibrils were associated

with an increase in kink density, decreasing mean modulus,

and disappearance of the D-banding in modulus images

(Fig. 8). The loss of D-banding is tied to increasing thickness

of the fibril shell which parallels increasing serial kink

density. The swelling ratio in our experiments also increased

with greater kink density (Fig. 6E). When the sample was

dehydrated, this shell layer collapsed and the D-band could

be seen in the topography of the inter-kinked regions of the

kinked collagen fibrils (Fig. 3B and D). This suggests that the

shell structure consists of a gel-like layer of denatured

collagen molecules, highly solvated in water. This denatured

collagen is likely not solubilized due to the presence of

crosslinking between neighboring alpha chains (Eyre and

Wu, 2005; Hormann and Schlebusch, 1971). The similarity in

the mean dehydrated heights of control and kink collagen

fibrils suggests that no significant molecular content is lost

during kink formation, suggesting retention of the entire

shell layer (Fig. S1).

The higher-density core inside discrete plasticity-damaged fibrils was primarily observed in the deformationmaps rather than in the modulus maps. This is due to thedifference in depth sensitivity of the two measurements.Modulus maps were calculated by fitting the force curves toa maximum depth of 10% of the fibrils zero force height(�40 nm indentation), while the deformation maps are cal-culated from the maximum indentation depth of the AFM tipinto the collagen fibril (�100 nm). Therefore, the deformationmaps have increased sensitivity to features buried deepwithin the core of the fibrils: e.g. the appearance of D-banding in some inter-kink regions of very kinked collagenfibrils (Fig. 9). The inconsistency of D-band observation in theinter-kink regions may be due to regional variation in thefibril shell thickness.

4.2. Laid rope damage motif model of collagen fibrils

It is clear that discrete plasticity damage results from tendonoverload, and that the intensity of the associated structuralchanges increases with repeated loading into the plasticregion of the tendon stress–strain curve. Nonetheless, thekinks which we have now observed in dehydrated andhydrated specimens, under both SEM and AFM, are reboundstructures which result after unloading of damaged fibrils.We know little about the mechanism via which the localdamage occurs and how the kinks form.

A collagen fibril consists of helically wound sub-fibrils andmicro-fibrils composed primarily of crosslinked, triple-helicalmolecules (Hulmes et al., 1995; Ottani et al., 2002; Silver et al.,1992; Wess, 2005). The crosslinking of the tropocollagenmolecules and the twisted nature of the collagen fibrils sub-components suggest that an apt macroscopic, structuralanalog of the collagen fibril might be that of a polymerizedlaid rope strand (McKenna et al., 2004). While not in thecontext of understanding damage, models of this sort havebeen previously proposed. Some investigators have suggestedthat collagen fibrils are nanoscale ropes, while others havesuggested a tube-like organization or a liquid crystallinemodel for the fibrils structure (Bozec et al., 2007; Brownet al., 2014; Gutsmann et al. 2003). If we consider the collagenfibril as a laid rope strand within a larger rope, it is intriguingto consider whether the shell and kink structures observed indiscrete plasticity samples might resemble damage motifs


found in laid ropes, associated with internal abrasion andaxial compression fatigue (McKenna et al., 2004).

The radial variation of the shell–core structures of kinked,damaged fibrils, coupled with the increases in both swellingcoefficient and shell thickness with kink density, offers thepossibility that the shell layer may be the result of forcesacting directly on the surface of the collagen fibril. There issupport in the literature to suggest that shear stress plays astrong role in the behavior of tendons under tension(Bruehlmann et al., 2004; Screen et al., 2004; Szczesny andElliott, 2014). Indeed, interfibril shear stress has recently beencalculated as being 32 kPa (Szczesny et al., 2015). In overload,if shear stress were applied to a particular collagen fibril, wewould expect that any shear-induced damage might propa-gate inwards from the affected surface. Molecular alterationdue to interfibrillar shear, would then explain the formationof the denatured shell layer of the damaged collagen fibril.One possible mechanism for such denaturation in the shellmight be inter-strand delamination; a phenomenon whereinα-chains participating in a crosslink are pulled out of theirrespective tropocollagen molecules, thereby leading to apermanent, stable, denaturation (Marino, 2015; Svenssonet al., 2013; Uzel and Beuhler, 2011; Veres et al., 2014). Inshear loading of the fibril, such delamination would propa-gate radially inwards from the surface toward the core of thefibril. This mechanism is similar to internal abrasion within arope strand, characterized by a low density shell and loss ofsubstructure at the ropes surface, features seen in the shell ofdiscrete plasticity damaged collagen fibrils (McKenna et al.,2004). In considering such a possibility, we must also consideran alternative hypothesis: that there is a pre-existing radialheterogeneity in the fibril structure which preferentially pre-disposes the outermost molecular layers of the fibril todamage: perhaps sacrificial damage which absorbs energyprior to fibril fracture, toughening the fibril. Such a hetero-geneity has been proposed in the tube-like model of acollagen fibril (Gutsmann et al., 2003). It is not possible atpresent to differentiate which of these mechanisms may beat play, if not both.

The kink structures observed in association with discreteplasticity are thought to be caused by structural damage whichoccurred during the overloading of the tendon, but actuallyformed upon removal of load. Observation of the kink struc-tures, now with both AFM and SEM, confirms their similarityto kink structures seen in laid ropes when their lowerhierarchical components experience compression (McKennaet al., 2004; Veres and Lee, 2012; Veres et al., 2013, 2014, 2015).One experimental difficulty which emerges when intact ten-dons are mechanically mounted and overloaded is that theresulting damage is heterogeneous: that is, only sub-populations of the collagen fibrils within the tendon displaydiscrete plasticity damage. Some regions appear to have beenspared overload and remain entirely intact (Veres and Lee,2012). Such heterogeneous damage could lead to regionswhich experience some degree of compression upon removalof external load, perhaps resulting in kink formation. Shouldthis be the case, and morphological similarity between engi-neered rope and collagen fibril structure certainly does notconfirm this mechanism, we might ask whether compressivedamage would occur during physiological overloading of a

muscle–tendon–bone complex, or whether it is peculiar to atesting machine sample mounted in grips.

One damage motif found in laid ropes, termed axialcompression fatigue, is characterized by periodic Z-shapedkink structures along the longitudinal axis of lower hierarch-ical structures. These structures in some ways resemble thekink structures observed in discrete plasticity. Formation ofrope kinks in axial compression fatigue has been shown to bethe result of twisting of the hierarchal structures undertension, and a mismatch in component lengths and mechan-ical properties. Their locations are thought to be determinedby regions of low torsional resistance or the non-uniformloading of rope components (McKenna et al., 2004). We havepreviously shown micron-scale variation in a collagen fibrilsdensity along its length. This observation suggests weakpoints may exist along the fibril (Baldwin et al., 2014).Furthermore, there is heterogeneity in fibril diameter intendon, which may imply accompanying heterogeneity infibril mechanical properties correlated with diameter (Moelleret al., 1995). It may be the case that some fibrils or fibrilregions are particularly susceptible to the effects of localstress concentration, whether via torsion or other modes.Whether the morphological similarities between laid ropefailure motifs and discrete plasticity in tendon collagen fibrilsare more than intriguing coincidences remains the subject forfurther study.

5. Conclusion

In this work, we have used AFM in a hydrated environmentto quantify the effects of tendon overload on the structure ofindividual collagen fibrils. We have confirmed the morphol-ogy observed under SEM, but have now demonstrated thedependence of water uptake and mean fibril modulus onserial kink density. Further, we have shown that kink andinter-kink regions maintain a constant modulus ratio, inde-pendent of kink density, suggesting a truly local mechanismfor kink formation. Our data supports the conclusion thatthe shell and kink structures are associated with moleculardenaturation. We have also described a set of interestingresemblances between failures in engineered laid ropes anddiscrete plasticity kinking seen in tendon collagen. It mayfollow that discrete plasticity kink formation is an effect offibril sub-structure and longitudinal, structural heterogene-ities which play out under the sort of uneven loading whichoccurs during in vitro testing of tendon samples. It remainsto be demonstrated if this is also the case during physiolo-gical overloads in high-load bearing tendons as in, forinstance, strain injuries. Discrete plasticity in tendon col-lagen is an important and complex phenomenon, andatomic force microscopy is a particularly valuable tool forits examination.

Acknowledgments

This work was supported by grants to LK and JML from theNational Sciences and Engineering Research Council ofCanada (NSERC) (Grant #: RGPIN 355291-2013). SJB received


salary support from the ASPIRE CREATE program of NSERC.The authors would like to express their thanks to JasminAstle for valuable assistance with tissue decellularization andto Dr. Sam Veres for assistance with the mechanical overloadprotocol. Gratitude is also expressed to Oulton's Meats,Windsor, Nova Scotia for assistance in harvesting tendonsamples.

Appendix A. Supplementary material

Supplementary data associated with this article can be foundin the online version at http://dx.doi.org/10.1016/j.jmbbm.2016.02.004.

r e f e r e n c e s

Ariganello, M.B., Labow, R.S., Lee, J.M., 2009. Response ofmacrophage-like U937 cells to decellularized tissue heartvalve materials. J. Heart Valve Dis. 18, 187–197.

Baldwin, S.J., Quigley, A.S., Clegg, C., Kreplak, L., 2014. Nanome-chanical mapping of hydrated rat tail tendon collagen I fibrils.Biophys. J. 107 (8), 1794–1801.

Benjamin, M., Kaiser, E., Milz, S., 2008. Structure–function rela-tionships in tendons: a review. J. Anat. 212, 211–228.

Bozec, L., Van der Heijden, G., Horton, M., 2007. Collagen fibrils:nanoscale ropes. Biophys. J. 92, 70–75.

Brown, A.I., Kreplak, L., Rutenberg, A.D., 2014. An equilibriumdouble-twist model for the radial structure of collagen fibrils.Soft Matter 10, 8500–8511.

Bruehlmann, S.B., Matyas, J.R., Duncan, N.A., 2004. ISSLS prizewinner: collagen fibril sliding governs cell mechanics in theannulus fibrosus: an in situ confocal microscopy study ofbovine discs. Spine 29, 2612–2620.

Bueckle, H., 1973. The Science of Hardness Testing and itsResearch Applications. American Society for Metals, Ohio.

Eyre, D.R., Wu, J.J., 2005. Collagen cross-links. Top. Curr. Chem.247, 207–229.

Grant, C.A., Brockwell, D.J., Radford, S.E., Thomson, N.H., 2009.Tuning the elastic modulus of hydrated collagen fibrils.Biophys. J. 97, 2985–2992.

Gutsmann, T., Fantner, G.E., Venturoni, M., Ekani-Nkodo, A.,Thompson, J.B., Kindt, J.H., Morse, D.K., Fygenson, D.K.,Hansma, P.K., 2003. Evidence that collagen fibrils in tendonsare inhomogeneously structured in a tube-like manner. Bio-phys. J. 84, 2593–2598.

Hodge, A.J., Schmitt, F.O., 1960. The charge profile of the tropo-collagen macromolecule and the packing arrangement innative-type collagen fibrils. Proc. Natl. Acad. Sci. USA 46,186–197.

Hormann, H., Schlebusch, H., 1971. Reversible andirreversible denaturation of collagen fibers. Biochemtry 10 (6),932–937.

Hulmes, D.J.S., Wess, T.J., Prockop, D.J., Fratzl, P., 1995. Radialpacking order, and disorder in collagen fibrils. Biophys. J. 68,1661–1670.

Kannus, P., 2001. Structure of the tendon connective tissue.Scand. J. Med. Sci. Sports 10, 312–320.

Marino, M., 2015. Molecular and intermolecular effects in collagenfibril mechanics: a multiscale analytical model compared withatomistic and experimental studies. Biomech. Model. Mechan-obiol. http://dxdoi.org/10.1007/s10237-015-0707-8.

Maugis, D., Barquins, M., 1978. Fracture mechanics and adherenceof viscoelastic bodies. J. Phys. D Appl. Phys. 11, 1989–2023.

McKenna, H.A., Hearle, J.W.S., O’Hear, N., 2004. Handbook of Fibre

Rope Technology. Woodhead Publishing LTD, Cambridge (UK).Moeller, H.D., Bosch, U., Decker, B., 1995. Collagen fibril diameter

distribution in patellar tendon autografts after posterior

cruciate ligament reconstruction in sheep: changes over time.

J. Anat. 187 (1), 161–167.Orgel, J.P., Miller, A., Wess, T.J., 2001. The in situ supermolecular

structure of type I collagen. Structure 9, 1061–1069.Ottani, V., Martini, D., Franchi, M., Ruggeri, A., Raspanti, M., 2002.

Hierarchical structures in fibrillary collagens. Micron 33,

587–596.Persch, G., Born, C., Utesch, B., 1994. Nano-hardness investiga-

tions of thin-films by an atomic-force microscope. Microelec-

tron. Eng. 24, 113–121.Sader, J.E., Larson, I., Mulvaney, P., White, L.R., 1995. Method for

the Calibration of atomic-force microscope cantilevers. Rev.

Sci. Instrum. 66, 3789–3798.Screen, H.R.C., Lee, D.A., Bader, D.L., Shelton, J.C., 2004. An

investigation into the effects of the hierarchical structure of

tendon fascicles on micromechanical properties. Proc. Inst.

Mech. Eng. H 218, 109–119.Sharma, P., Maffulli, N., 2005. Tendon injury and tendinopathy:

healing and repair. J. Bone Jt. Surg. Am. 87 (1), 187–202.Silver, D., Miller, J., Harrison, R., Prockop, D.J., 1992. Helical model

of nucleation and propagation to account for the growth of

type I collagen fibrils from the symmetrical pointed tips: a

special example of self-assembly of rod-like monomers. Proc.

Natl. Acad. Sci. USA 89, 9860–9864.Sneddon, I., 1965. The relation between load and penetration in

the axisymmetric Boussinesq problem for a punch of arbitrary

profile. Int. J. Eng. Sci. 3, 47–57.Spiesz, E.M., Thorpe, C.T., Chaudhry, S., Riley, G.P., Birch, H.L.,

Clegg, P.D., Screen, H.R.C., 2015. Tendon extracellular matrix

damage, degradation and inflammation in response to in vitro

overload exercise. J. Orthop. Res. 33 (6), 889–897.Spitzner, E.C., Roper, S., Zerson, M., Bernstein, A., Magerle, R.,

2015. Nanoscale swelling heterogeneities in type I collagen

fibrils. ACS Nano 9 (6), 5683–5694.Stolz, M., Raiteri, R., Daniels, A.U., VanLandingham, M.R.,

Baschong, W., Aebi, U., 2004. Dynamic elastic modulus of

porcine articular cartilage determined at two different levels

of tissue organization by indentation-type atomic force

microscopy. Biophys. J. 86, 3269–3283.Svensson, R.B., Mulder, H., Kovanen, V., Magnusson, S.P., 2013.

Fracture mechanics of collagen fibrils: influence of natural

cross-links. Biophys. J. 104 (11), 2476–2484.Szczesny, S.E., Elliott, D.M., 2014. Interfibrillar shear stress is the

loading mechanism of collagen fibrils in tendon. Acta Bio-

mater. 10, 2582–2590.Szczesny, S.E., Caplan, J.L., Pedersen, P., Elliott, D.M., 2015.

Quantification of interfibrillar shear stress in aligned soft

collagenous tissues via notch tension testing. Sci. Rep. 5,

14649.Uzel, S.G.M., Beuhler, M.J., 2011. Molecular structure, mechanical

behavior and failure mechanism of the C-terminal cross-link

domain in type I collagen. J. Mech. Behav. Biomed. Mater. 4,

153–161.Veres, S.P., Lee, J.M., 2012. Designed to fail: a novel mode of

collagen fibril disruption and its relevance to tissue tough-

ness. Biophys. J. 102 (12), 2876–2884.Veres, S.P., Harrison, J.M., Lee, J.M., 2013. Repeated subrupture

overload causes progression of nanoscaled discrete plasticity

damage in tendon collagen fibrils. J. Orthop. Res. 31 (5),

731–737.Veres, S.P., Harrison, J.M., Lee, J.M., 2014. Mechanically overload-

ing collagen fibrils uncoils collagen molecules placing them in

a stable, denatured state. Matrix Biol. 33, 54–59.



http://refhub.elsevier.com/S1751-6161(16)00042-4/sbref1









































http://dx.doi.org/10.1007/s10237-015-0707-8

http://dx.doi.org/10.1007/s10237-015-0707-8

http://dx.doi.org/10.1007/s10237-015-0707-8







































































Veres, S.P., Brennan-Pierce, E.P., Lee, J.M., 2015. Macrophage-likeU937 cells recognize collagen fibrils with strain-induced dis-crete plasticity damage. J. Biomed. Mater. Res. A 103 (1),397–408.

Wess, T.J., 2005. Collagen fibril form and function. Adv. ProteinChem. 70, 341–375.

Zhao, T., Weinhold, P.S., Dahners, L.E., 2011. Some observationson the subfibrillar structure of collagen fibrils as noted duringtreatment with NKISK and cathepsin G with mechanicalagitation. J. Electron Microsc. (Tokyo) 60, 177–182.











characterization via atomic force microscopy of discrete...

Documents