postnatal mechanical loading drives adaptation of tissues … · 2020-04-23 · 1 1 post natal...

1

Postnatal mechanical loading drives adaptation of tissues primarily through modulation 1

of the non-collagenous matrix 2

3

D. E. Zamboulis1*, C. T. Thorpe2, Y. Ashraf Kharaz1, H. L. Birch3, H. R. C. Screen4†, P. D. 4

Clegg1† 5

* Corresponding author: Zamboulis, † Joint last authorship: Screen, Clegg 6

7

1Institute of Ageing and Chronic Disease, Faculty of Health and Life Sciences, University of 8

Liverpool, Liverpool, L7 8TX, United Kingdom 9

2Comparative Biomedical Sciences, The Royal Veterinary College, Royal College Street, 10

London, NW1 0TU, United Kingdom 11

3University College London, Department of Orthopaedics and Musculoskeletal Science, 12

Stanmore Campus, Royal National Orthopaedic Hospital, Stanmore, HA7 4LP, United 13

Kingdom 14

4Institute of Bioengineering, School of Engineering and Materials Science, Queen Mary 15

University of London, London, E1 4NS, United Kingdom 16

17

Abstract 18

Mature connective tissues demonstrate highly specialised properties, remarkably adapted to 19

meet their functional requirements. Tissue adaptation to environmental cues can occur 20

throughout life and poor adaptation commonly results in injury. However, the temporal 21

nature and drivers of functional adaptation remain undefined. Here, we explore functional 22

adaptation and specialisation of mechanically loaded tissues using tendon; a simple aligned 23

biological composite, in which the collagen (fibre phase) and surrounding predominantly 24

non-collagenous matrix (matrix phase) can be interrogated independently. Using an equine 25

.CC-BY 4.0 International licensemade available under a(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is

The copyright holder for this preprintthis version posted April 24, 2020. ; https://doi.org/10.1101/2020.04.23.058081doi: bioRxiv preprint

https://doi.org/10.1101/2020.04.23.058081

http://creativecommons.org/licenses/by/4.0/

2

model of late development, we report the first phase-specific analysis of biomechanical, 26

structural and compositional changes seen in functional adaptation, demonstrating adaptation 27

occurs postnatally, following mechanical loading, and is almost exclusively localised to the 28

non-collagenous matrix phase. These novel data redefine adaptation in connective tissue, 29

highlighting the fundamental importance of non-collagenous matrix and suggesting that 30

regenerative medicine strategies should change focus from the fibrous to the non-collagenous 31

matrix phase of tissue. 32

33

Introduction 34

Functional adaptation of load-bearing tissues such as tendon is crucial to ensure the tissue is 35

specialised appropriately to meet functional needs. Adaptation to mechanical requirements is 36

key in healthy development and homeostatic tissue maintenance, with poor tissue 37

optimisation during maturation likely a key contributor to increased injury risk later in life. 38

Dysregulated homeostasis and long-term under- or over-stimulation leads to maladaptation, 39

changes in tissue integrity, and reduced mechanical competence and is implicated in the 40

disease aetiology of load-bearing tissues(Freedman et al., 2015; Gardner et al., 2008). 41

Understanding the developmental drivers of structural specialisation and their association 42

with mechanobiology is thus of fundamental importance for healthy ageing and disease 43

prevention in musculoskeletal tissues(R. Choi et al., 2018; Chavaunne T. Thorpe et al., 2013). 44

Such knowledge will help identify future targets for therapeutic interventions, and thus 45

address the current lack of effective musculoskeletal disease treatments with new, evidence-46

based approaches to disease management. However, there is currently little knowledge of the 47

key extracellular matrix (ECM) components associated with structural specialisation, the 48

temporal nature of their adaptation, or the stimuli that drive adaptation. 49

As the principal structural component of connective tissues, collagen expression at the gene 50



https://doi.org/10.1101/2020.04.23.058081


3

and protein level has been the focus of the majority of studies in relation to loading, with 51

some studies reporting increases in collagen synthesis and others noting collagen degradation 52

in response to loading, depending on the tissue function or tissue structure in different 53

species(R. Choi et al., 2018; Magnusson & Kjaer, 2019). In materials science, this collagen 54

structural framework is commonly referred to as the “fibre phase” of a material, surrounded 55

by the primarily non-collagenous and glycoprotein-rich components of the ECM, termed the 56

“matrix phase”(Armiento et al., 2018; C. T. Thorpe & Screen, 2016). This distinction is 57

important, as it describes a fibre composite material, in which “fibre” and “matrix” phases 58

have different mechanical properties, and overall tissue mechanical properties and function 59

are governed by the interplay of these two phases. When looking to understand functional 60

adaption of a tissue, it is necessary to look at adaption of all ECM components. 61

Whilst fibre phase adaption has received some attention, adaptation of the matrix phase to 62

mechanical stimuli remains poorly defined. Indeed, it is notable that the numerous studies 63

investigating the mechanoresponsive nature of load-bearing tissues tend to restrict their focus 64

to specific fibre or matrix components with no spatial distinction, and also focus on a single 65

element of either structural or mechanical adaptation, such that limited information is 66

gained(Cherdchutham, Becker, et al., 2001; R. K. Choi et al., 2019; Mendias et al., 2012). In 67

order to provide the necessary complete profile of adaptive behaviour, it is crucial that phase-68

specific, temporospatial adaptation in the context of both structure and function is defined. 69

Identifying the drivers of adaptation requires use of a model system in which the 70

temporospatial nature of adaptation can be fully profiled. Tendon provides the ideal model 71

for such a study. It is well-established that mature tendons can present structural and 72

mechanical specialisms(Chavaunne T. Thorpe et al., 2015; Chavaunne T. Thorpe, 73

Karunaseelan, et al., 2016; Chavaunne T. Thorpe, Riley, et al., 2016) and be grouped into two 74

clear functional groups; stiff positional tendons simply connect muscle to bone to effectively 75



https://doi.org/10.1101/2020.04.23.058081


4

position limbs, whilst elastic energy-storing tendons increase locomotor efficiency by 76

stretching and storing energy which they return to the system on recoil(Batson et al., 2010; C. 77

T. Thorpe & Screen, 2016; C T Thorpe et al., 2012). Further, the simple aligned organisation 78

of tendon means that fibre and matrix phases are spatially distinct, enabling structural and 79

mechanical characterisation of each phase independently, by comparing fascicles (fibre 80

phase) and inter-fascicular matrix (IFM; matrix phase)(C. T. Thorpe & Screen, 2016; C T 81

Thorpe et al., 2012). Finally, use of equine tendon provides access to an exceptional model of 82

adaptation. The equine superficial digital flexor tendon (SDFT) has been shown to be highly 83

analogous to the human Achilles tendon in its capacity for energy storage, injury profile and 84

extent of specialization(Biewener, 1998; Patterson-Kane & Rich, 2014). Availability of 85

samples enabled us to explore the extensive adaptation processes associated with late stage 86

development, contrasting paired positional and energy-storing equine tendons through pre- 87

and post-natal development. 88

Using this model, we investigate the process and drivers of functional adaptation, when 89

tendons transition from an absence of loading (foetal: mid to end (6 to 9 months) gestation, 90

and 0 days: full-term foetuses, and foals that did not weight-bear); through to weight-bearing 91

(0-1 month) and then to an increase in body weight and physical activity (3-6 months; and 1-92

2 years). We hypothesise that early in development during gestation, the fibre and matrix 93

phases of functionally distinct tendons have identical compositional profiles and mechanical 94

properties, with tissue specialisation occurring as an adaptive response to the mechanical 95

stimulus of load-bearing, predominantly in the matrix phase of the elastic energy-storing 96

tendon. 97

98

Results 99



https://doi.org/10.1101/2020.04.23.058081


5

Mechanical adaptation is localised to the matrix phase 100

First, we determined how the mechanical properties of the fibre and matrix phase develop in 101

tendon, with a particular focus on the temporospatial nature of mechanical adaptation to 102

functional specialisation. Individual fascicles were dissected and tested as fibre samples, 103

while an isolated region of interfascicular matrix, matrix phase, was tested by shearing 104

fascicles apart. Samples were subjected to preconditioning followed by a pull to failure 105

(Supplementary Figure 1). The yield point of samples was identified, denoting the point at 106

which the sample became irreversibly damaged and was unable to recover from the applied 107

load, and the sample failure properties also recorded, highlighting the maximum stress and 108

strain the sample could withstand. 109

A significant increase in fibre phase yield and failure properties was evident when comparing 110

embryonic fibre samples to those acquired immediately at birth. However, data indicate 111

minimal distinction in fibre phase mechanics between functionally distinct tendons (Figure 1) 112

and, significantly, no specialisation for energy storage in response to loading during postnatal 113

development. 114

Contrasting with fibre phase mechanics, the failure properties of the matrix phase continued 115

to alter throughout development with failure properties increasing markedly from 6 months 116

onwards. We also identify the emergence of an extended region of low stiffness at the start of 117

the loading curve (ie an extended toe region) specific to the SDFT matrix phase pull to failure 118

curve. This indicates less resistance to extension, and together with the concomitant increase 119

in matrix yield force and extension at yield, demonstrates development of an overall greater 120

capacity for extension in the SDFT matrix phase behaviour. A summary of these findings is 121

achieved by plotting the amount of matrix phase extension at different percentages of failure 122

force (Figure 2b,h-k), highlighting how the matrix phase of the energy-storing SDFT became 123



https://doi.org/10.1101/2020.04.23.058081


6

significantly less stiff than that of the positional CDET during the initial toe region of the 124

loading curve as the tendon adapts. 125

The viscoelastic properties of the developing matrix phase also showed significant 126

interactions between tendon type and development, with matrix viscoelasticity significantly 127

decreasing with development specifically in the energy-storing SDFT (Figure 2a,c-d), 128

resulting in specialisation towards a more energy efficient structure. 129



https://doi.org/10.1101/2020.04.23.058081


7

130

131

Figure 1. Fibre phase response to mechanical testing shows increase in strength with 132 development but few significant differences between tendon types, indicating that the fibre 133 phase shows minimal structural specialisation in response to loading. (a) Representative curves 134 for 10 preconditioning cycles for the SDFT and CDET fibre phase in the foetus and 1-2 years age 135 group. (b) Representative force-extension curves to failure for the SDFT and CDET fibre phase in the 136 same age groups. (c-i) Mean SDFT and CDET fibre phase biomechanical properties are presented 137 across development, with data grouped into age groups: foetus, 0 days (did not weight-bear), 0-1 138 month, 3-6 months, 1-2 years. ‡ significant interaction between tendon type and development, * 139 significant difference between tendons, a-b significant difference between age groups. Error bars 140 depict standard deviation. 141



https://doi.org/10.1101/2020.04.23.058081


8

142

Figure 2. Mechanical testing of the matrix phase shows an equivalent increase in failure 143 properties between the SDFT and CDET with development, but development of an extended 144 low stiffness toe region and more elastic behaviour in the SDFT. (a) Representative curves for 10 145 preconditioning cycles for the SDFT and CDET matrix phase in the foetus and 1-2 years age group. 146 (b) Representative force-extension curves to failure for the SDFT and CDET matrix phase in the same 147 age groups. (c-i) Mean SDFT and CDET matrix phase biomechanical properties are presented across 148 development, with data grouped into age groups: foetus, 0 days (did not weight-bear), 0-1 month, 3-6 149 months, 1-2 years. (j-k) To visualise the extended low stiffness toe region in the SDFT matrix phase, 150 the amount of matrix phase extension at increasing percentages of failure force is presented, 151 comparing the SDFT and CDET in the foetus and 1-2 years age group. ‡ significant interaction 152 between tendon type and development, * significant difference between tendons, a-g significant 153 difference between age groups. Error bars depict standard deviation. 154



https://doi.org/10.1101/2020.04.23.058081


9

155

Structural adaptation is localised to the matrix phase 156

Having described mechanical adaptation of the matrix phase to meet functional demand, we 157

next performed a histological and immunohistochemical comparison of developing energy-158

storing and positional tendons to determine how temporospatial structural adaptation may 159

dictate this evolving mechanical behaviour. 160

The energy-storing SDFT and positional CDET appeared histologically similar in the foetus, 161

in both instances showing surprisingly poor demarcation of the matrix phase, which only 162

became structurally distinct after birth and the initiation of loading (Figure 3a). Fibre 163

development was generally consistent in both tendon types with cellularity and crimp 164

showing a reduction with development, cells displaying more elongated nuclei, and collagen 165

showing a more linear organisation. In contrast, the matrix phase demonstrated divergence 166

between tendon types with only the SDFT matrix phase showing an increase in cellularity 167

following tendon loading and a retention of matrix phase width throughout development 168

(Figure 3, Supplementary Figure 2). 169

The abundance of major ECM proteins was also generally consistent across fibre and matrix 170

regions in the foetus, with divergence of protein composition between phases only evident 171

with further development. Notably adaptation was driven by changes in non-collagenous 172

ECM components specifically, levels of which reduced in the fibre phase and increased 173

dramatically in the matrix phase through postnatal development (Figure 3c, Supplementary 174

Figure 3). Of particular note, we demonstrate the distribution of PRG4 (commonly known as 175

lubricin) and TNC predominantly in the matrix phase of tendons. We also demonstrate that 176

elastin is preferentially localised to the matrix phase with its abundance decreasing only in 177

the CDET with development. Furthermore, we show how structural changes do not manifest 178

until after birth and the initiation of loading, but that structural adaptation to loading then 179



https://doi.org/10.1101/2020.04.23.058081


10

occurs over a period of months, involving both upregulation and downregulation of different 180

ECM constituents. 181

182



https://doi.org/10.1101/2020.04.23.058081


11

183

184

Figure 3. The SDFT and CDET are histologically similar at birth and differentiate with 185 development especially in the matrix phase. (a) Representative images of H&E sections of the 186 SDFT and CDET demonstrate structural development: foetus, 0 days (did not weight-bear), 0-1 year, 187 and 1-2 years whilst (b) Radar plots enable the mean histology scores of the fibre and matrix phase 188 for the SDFT and CDET to be compared between the foetus and 1-2 years age group (all data shown 189

in Supplementary Figure 2). A decrease in cell numbers, crimp, and matrix phase width is visible 190 with progression of age, and the aspect ratio of cells in the fibre phase increases. (c) Use of 191 immunohistochemical assays shows divergence of PGR4 (lubricin) and elastin with maturation 192 between functionally distinct tendons. Matrix and fibre phase staining scores are shown for decorin 193 (DCN), fibromodulin (FMOD), lubricin (PRG4), and tenascin-C (TNC) in the SDFT and CDET, 194 alongside representative images of immunohistochemical staining in the postnatal SDFT. A 195 quantitative measure of elastin (ELN) is provided as percentage of wet weight, alongside a 196 representative image of immunohistochemical staining in the postnatal SDFT. Staining scores for 197

elastin are provided in Supplementary Figure 3. ‡ significant change in tendon phase with 198 development, ‡‡ significant interaction between tendon phase and development, * significant 199 difference between tendons, a-d significant difference between age groups. Scale bar 100 µm. Error 200 bars depict standard deviation. 201



https://doi.org/10.1101/2020.04.23.058081


12

202

Adaptation relies on evolution of matrix phase composition only 203

To explore these concepts in further detail and to scrutinise the capacity for ECM adaptation, 204

proteomic methodologies were adopted. With the mechanical and histological data 205

identifying that functional adaptation is particular to the energy-storing SDFT, mass 206

spectrometry analysis focused on a more detailed comparison of the matrix and fibre phase 207

development and adaptation in this tendon specifically. 208

Our results demonstrated that the proteomic profile of the matrix phase was more complex 209

(more identified proteins) and a higher percentage of matrix phase proteins were cellular, 210

supporting the histological findings of a more cellular matrix phase (Supplementary Figure 211

4). Notably, despite the two phases being structurally distinct, they had 14 collagens and 11 212

proteoglycans in common (Supplementary Table 4). Overall, proteomic heatmap analysis 213

correlated very strongly with immunohistochemical findings, showing that alterations in the 214

fibre phase proteome reduced through development, with minimal changes following the 215

initiation of loading (Table 1 and 2, Figure 4a), whilst numerous matrisome and matrisome-216

related proteins progressively increase in abundance through development in the matrix phase 217

(Table 1 and 2). Detailed consideration of protein changes also highlights that post loading 218

changes in the matrix phase appear more specific to proteoglycans and glycoproteins. 219

Proteomic data also enable insight into the turnover of proteins in the different tendon phases, 220

through a comparison of the neopeptides produced by protein breakdown(Chavaunne T. 221

Thorpe, Peffers, et al., 2016). In the current study, we are able to profile the temporal nature 222

of fibre and matrix turnover, demonstrating that both phases display turnover during 223

development but that fibre phase turnover slows down towards the end of maturation, whilst 224



https://doi.org/10.1101/2020.04.23.058081


13

matrix phase turnover rates are maintained, suggesting structural and/or compositional 225

plasticity (Figure 4b). 226

Having identified the matrix phase as the location of functional adaptation, we next 227

investigated the regulation of this process, to detect targets for modulation in regeneration 228

strategies addressing functional impairment. Pathway analysis of the differentially abundant 229

proteins across age groups supported an ECM-integrin-cytoskeleton to nucleus signalling 230

pathway for the mediation of the observed mitogenesis and matrigenesis in response to 231

tendon loading. In addition, TGF-β1 was highlighted as a regulator of ECM organisation and 232

functional adaptation, upregulated in the energy-storing tendon following loading (Figure 4). 233

234



https://doi.org/10.1101/2020.04.23.058081


14

235 236 Figure 4. The fibre phase proteome remains the same during postnatal development and tendon 237 loading whereas the matrix phase starts changing following tendon loading in postnatal 238 development. (a) Heatmap of differentially abundant proteins in foetus, 0 days (did not weight-bear), 239 0-1 month, 3-6 months, and 1-2 years SDFT matrix and fibre phases (p<0.05, fold change ≥2). 240 Heatmap colour scale ranges from blue to white to red with blue representing lower abundance and 241 red higher abundance. (b) Proteins with identified neopeptides and proteins showing differential total 242 neopeptide abundance across age groups. Graph of proteins showing differential total neopeptide 243 abundance in the SDFT fibre phase across development (p<0.05, fold change≥2, FDR 5%). No 244 proteins showed differential total neopeptide abundance in the matrix phase. (c) IPA networks for 245 TGFB1 in the foetus and at 3-6 months in the SDFT matrix phase. TGFB1 regulation in the matrix 246 phase is predicted to be inhibited in the foetus and activated at 3-6 months in the SDFT. Red nodes, 247 upregulated proteins, green nodes, downregulated proteins, intensity of colour is related to higher 248 fold-change, orange nodes, predicted upregulated proteins in the dataset, blue nodes, predicted 249 downregulated proteins. (d) Whole tendon relative mRNA expression for TGFB1 in the SDFT and 250 CDET during postnatal development. * significant difference between tendons, a significant 251 difference between age groups. Error bars depict standard deviation. 252



https://doi.org/10.1101/2020.04.23.058081


15

253 Table 1. Matrix phase differentially abundant matrisome and matrisome-associated proteins 254 through development organised by highest mean condition (p<0.05, fold change≥2). Proteins are 255 arranged into colour coded divisions and categories. Bar graphs profile the relative abundance of each 256 protein at each development stage, with the development age reporting the highest mean protein level 257 also specified. 258

Protein Division Category Highest mean cond.

SERPINH1 Matrisome-associated ECM Regulators Foetus

COL14A1 Core matrisome Collagens 0-1 month

ASPN Core matrisome Proteoglycans 0-1 month

FMOD Core matrisome Proteoglycans 0-1 month

KERA Core matrisome Proteoglycans 0-1 month

FBLN5 Core matrisome ECM Glycoproteins 0-1 month

FGB Core matrisome ECM Glycoproteins 0-1 month

FGG Core matrisome ECM Glycoproteins 0-1 month

COL1A2 Core matrisome Collagens 3-6 months





BGN Core matrisome Proteoglycans 3-6 months

HSPG2 Core matrisome Proteoglycans 3-6 months

ADIPOQ Core matrisome ECM Glycoproteins 3-6 months

FBN1 Core matrisome ECM Glycoproteins 3-6 months

FN1 Core matrisome ECM Glycoproteins 3-6 months

LAMB2 Core matrisome ECM Glycoproteins 3-6 months

LAMC1 Core matrisome ECM Glycoproteins 3-6 months

NID1 Core matrisome ECM Glycoproteins 3-6 months

ANXA4 Matrisome-associated ECM-affiliated 3-6 months

S100A4 Matrisome-associated Secreted Factors 3-6 months

COL21A1 Core matrisome Collagens 1-2 years






DCN Core matrisome Proteoglycans 1-2 years

LUM Core matrisome Proteoglycans 1-2 years

OGN Core matrisome Proteoglycans 1-2 years

PRELP Core matrisome Proteoglycans 1-2 years

COMP Core matrisome ECM Glycoproteins 1-2 years

DPT Core matrisome ECM Glycoproteins 1-2 years

TGFBI Core matrisome ECM Glycoproteins 1-2 years



https://doi.org/10.1101/2020.04.23.058081


16

259

Table 2. Fibre phase differentially abundant matrisome and matrisome-associated proteins 260 through development organised by highest mean condition (p<0.05, fold change≥2). Proteins are 261 arranged into colour coded divisions and categories. Bar graphs on the right profile the relative 262 abundance of each protein at each development stage, with the development age reporting the highest 263 mean protein level also specified. 264

265

266

Discussion 267

In this study, we describe the process and drivers of functional adaptation in tendon 268

development integrating mechanical, structural, and compositional analysis in tendons 269

transitioning from an absence of loading through to weight-bearing and then to an increase in 270

body weight and physical activity. Whilst the limited previously available data on the 271

development of tendon gross mechanical properties show an increase in mechanical 272

properties with development(Ansorge et al., 2011; Cherdchutham, Meershoek, et al., 2001), 273

no such phase-specific analysis of the development of tissue mechanics has been carried out 274

previously. Similarly, available research into tendon morphogenesis and maturation has 275

Protein Division Category Highest mean cond.

COL11A1 Core matrisome Collagens Foetus

DCN Core matrisome Proteoglycans Foetus

FMOD Core matrisome Proteoglycans Foetus

KERA Core matrisome Proteoglycans Foetus

PCOLCE Core matrisome ECM Glycoproteins Foetus

SERPINF1 Matrisome-associated ECM Regulators Foetus

ANXA1 Matrisome-associated ECM-affiliated Proteins Foetus



LGALS1 Matrisome-associated ECM-affiliated Proteins Foetus

COL12A1 Core matrisome Collagens 0 days


PRELP Core matrisome Proteoglycans 1-2 years

COMP Core matrisome ECM Glycoproteins 1-2 years

FN1 Core matrisome ECM Glycoproteins 1-2 years



https://doi.org/10.1101/2020.04.23.058081


17

previously focused on the development of the collagenous network that comprises the fibrous 276

phase of the tendon and is often focussed on early foetal development(Kalson et al., 2011; 277

Marturano et al., 2013; Pan et al., 2018). 278

Here, examining the fibre and matrix phase mechanical properties independently, we show 279

minimal distinction in fibre phase mechanics between functionally distinct tendons and, 280

significantly, no specialisation for energy storage in response to loading during postnatal 281

development. In contrast, the matrix phase mechanical properties display continuous 282

alterations through development with properties in the foetus being comparable between 283

functionally distinct tendons, and a low stiffness region emerging in the initial non-linear 284

region (toe region) of the pull to failure curve of the SDFT matrix phase only following 285

tendon loading postnatally. This is coupled with a concomitant increase in matrix force and 286

extension at yield, point at which the sample became irreversibly damaged, highlighting that 287

the energy-storing SDFT matrix phase becomes significantly less stiff than that of the 288

positional CDET during the initial toe region of the loading curve. We have previously 289

indicated that this low stiffness behaviour allows sliding between the fascicles of the fibre 290

phase, enabling non-uniform loading of tissue and is fundamental for effective extension and 291

recoil in energy-storing tendons(Chavaunne T. Thorpe et al., 2015). Furthermore, with ageing 292

the low stiffness behaviour of the matrix phase of energy-storing tendons is lost, possibly 293

contributing to disease development(Chavaunne T. Thorpe et al., 2013). The only other 294

studies considering the mechanical properties of developing tendons have focused simply on 295

changes in whole tissue mechanics, and have thus not been able to identify the drivers of 296

change within the tissue(Ansorge et al., 2011; Cherdchutham, Meershoek, et al., 2001). Here, 297

we identify that the matrix phase is the key region in which mechanical adaptation to meet 298

function occurs, and that this occurs after the initiation of loading, primarily 1-2 years 299

postnatally. 300



https://doi.org/10.1101/2020.04.23.058081


18

We subsequently examine how temporospatial structural adaptation may dictate this evolving 301

mechanical behaviour and uncover a divergence in structural characteristics between tendon 302

types in the matrix phase only, with a retention of matrix phase width throughout 303

development and an increase in cellularity in the SDFT matrix phase only. It is well 304

recognised that foetal and early postnatal tendons are highly cellular, and cellularity is 305

generally considered to decrease postnatally(Russo et al., 2015; Stanley et al., 2008), but here 306

we show that the described reduction in cellularity only occurs in the fibre phase, and 307

cellularity in fact increases in the matrix phase with development. By following the 308

alterations in cellularity across fibre and matrix, here we can determine that the marked 309

difference in regional cellularity is likely driven by a maintenance of cell numbers following 310

tendon loading in the thinning matrix region, while cell numbers in the fibre region appear to 311

reduce as a result of the fibre phase ECM increase. Greater cellularity is commonly 312

associated with a requirement for rapid adaptive organisation of ECM components(Russo et 313

al., 2015), suggesting the matrix phase, particularly in the energy-storing tendons, may adapt 314

to be more mechanoresponsive, a necessary aspect of healthy homeostatic maintenance of a 315

tissue. Immunohistochemical analysis reveals the distribution of major ECM proteins is 316

consistent across tendons in the foetus and becomes distinct across phases through 317

development. Of note, PRG4 (lubricin), a large proteoglycan which is important in ECM 318

lubrication, is found mainly distributed in the matrix phase of tendons. Using a lubricin-319

knockout mouse, this proteoglycan has been demonstrated to facilitate interfascicular 320

sliding(Kohrs et al., 2011), indicating that this structural adaptation may be key in achieving 321

the previously identified mechanical adaptation in the energy-storing tendon matrix phase. In 322

addition, Kostrominova and Brooks (2013)(Kostrominova & Brooks, 2013) report PRG4 323

expression as wells as elastin expression decreased with ageing in rat tendon suggesting an 324

association with an increased risk of disease with ageing. We also demonstrate that elastin is 325



https://doi.org/10.1101/2020.04.23.058081


19

preferentially localised to the matrix phase, potentially having a role in the capacity for 326

matrix recoil after loading which is necessary for the healthy function of energy-storing 327

tendons(Godinho et al., 2017; Ritty et al., 2002). Further supporting a role for elastin in the 328

energy-storing function, elastin appears to be redundant in positional tendons with its 329

abundance decreasing in the CDET with development. Together, these findings show that 330

structural adaptation of tendon post loading is primarily focused to the matrix phase, and 331

observed predominantly in the energy-storing SDFT. 332

Compositional analysis of the energy-storing SDFT phases using mass spectrometry 333

corroborates the immunohistochemical analysis and shows a more complex proteomic profile 334

for the matrix phase. It additionally shows that abundance of the majority of proteins in the 335

fibre phase is higher in the foetus and reduces through development with minimal changes 336

following the initiation of loading, whereas in the matrix phase numerous ECM and ECM-337

related proteins progressively increase in abundance following tendon loading and through 338

development. Neopeptide analysis demonstrated fibre phase ECM protein turnover slows 339

down towards the end of maturation, whilst matrix phase ECM protein turnover rates are 340

maintained. Once a tendon is mature, little collagen turnover occurs(Birch, 2007; KM et al., 341

2013) and we have previously shown that the minimal turnover in mature tendon is focused 342

to the matrix phase(C. T. Thorpe et al., 2015; Chavaunne T. Thorpe et al., 2010; Chavaunne 343

T. Thorpe, Peffers, et al., 2016). The maintenance of turnover rates observed in the matrix 344

phase here suggests structural and/or compositional plasticity of the matrix phase. Integrated, 345

our data convincingly show a continual temporal change in the matrix phase proteome 346

specifically, which would enable adaptation and specialisation to the load environment, and 347

highlight the compositional plasticity of the matrix phase in responding to dynamic altered 348

conditions such as those occurring during development and regeneration. It is critical that the 349

difference in capacity for functional adaptation across fibre and matrix phases identified here 350



https://doi.org/10.1101/2020.04.23.058081


20

is considered if regenerative medicine and tissue engineering approaches are to be successful. 351

Here, we demonstrate the temporal pattern of structure-function adaptation, with 352

compositional changes occurring in the first months after loading, and leading to the 353

mechanical specialisation we have previously observed in adult energy-storing tendon(Birch, 354

2007; C T Thorpe et al., 2012; Chavaunne T. Thorpe et al., 2015). With the fibrous phase 355

primarily responsible for the mechanical strength of a tissue, biomaterial and regenerative 356

medicine studies have unsurprisingly placed considerable emphasis on this region to 357

date(Sensini et al., 2019; Watts et al., 2017). Here, we not only highlight the importance of 358

the matrix phase in modulating mechanical behaviour, but also demonstrate how the matrix 359

phase must be targeted to support adaptation and optimum tissue quality. 360

Finally, pathway analysis of our proteomic data highlighted TGF-β1 as a regulator of ECM 361

organisation and functional adaptation, predicted to be upregulated following loading in the 362

energy-storing tendon. TGF-β is known to have a role in cellular mechanobiology and 363

connective tissue homeostasis, regulating ECM synthesis and remodelling in a force-364

dependent way following mechanical stimulation, to specify the quality of the ECM and help 365

coordinate cytoskeletal tension(Maeda et al., 2011). A previous study of developing chick 366

tendon detected TGF-β1 staining in the matrix phase only during development, highlighting 367

its localised distribution in development(Kuo et al., 2008). In the current study, we are also 368

able to associate TGF- expression with functional adaptation of the matrix phase of tendon. 369

In addition to tissue development and homeostasis, TGF-β1 is involved in connective tissue 370

injury and repair with abnormal expression levels reported in both processes suggesting a 371

pleiotropic mode of action(Gao et al., 2019). The above may suggest a role for TGF-β1 in 372

tissue development and homeostasis and that its dysregulation is associated with tissue injury 373

and repair expression levels. 374

375



https://doi.org/10.1101/2020.04.23.058081


21

Outlook 376

We demonstrate for the first time that functional adaptation in tendon is predominantly reliant 377

on adaptation of the metabolically active matrix phase, which responds to the mechanical 378

environment through TGF- signalling, resulting in modulations in ECM turnover and 379

composition to fine-tune mechanical properties. Traditionally, the matrix phase of connective 380

tissues has received considerably less attention than the fibre phase, with regenerative 381

medicine, biomimetics and biomechanics studies all largely focused on investigating and 382

recapitulating the organisation and mechanical properties of the collagenous fibrous network. 383

Following tendon injury, normal tissue architecture is not recovered, and in particular, the 384

cellular matrix phase is not regenerated. There is great potential gain from understanding the 385

convergence of biology underpinning adaptation, function and pathology and here, we 386

propose a paradigm shift to consider the metabolically active matrix phase as a key target for 387

regenerative medicine strategies aimed at addressing functional impairment of tendons and 388

other connective tissues following disease. Regeneration of the matrix phase following 389

tendon injury could be key for tendon health and low re-injury risk. 390

391

Materials and Methods 392

Experimental design 393

Using an equine tendon model, we investigate the process and drivers of functional 394

adaptation in the SDFT and CDET, two functionally distinct tendons, when tendons transition 395

from an absence of loading (foetal: mid to end (6 to 9 months) gestation, and 0 days: full-396

term foetuses, and foals that did not weight-bear); through to weight-bearing (0-1 month) and 397

then to an increase in body weight and physical activity (3-6 months; and 1-2 years). We use 398

a phase-specific approach to characterise each tendon phase independently, by comparing 399



https://doi.org/10.1101/2020.04.23.058081


22

fascicles (fibre phase) and inter-fascicular matrix (IFM; matrix phase) mechanical properties, 400

structure and composition. 401

For this purpose, we used mechanical testing, histological and immunohistochemical analysis 402

and mass spectrometry analysis following laser capture microdissection. Sample size was 403

selected based on previous experiments and restricted by sample availability and the cost of 404

mass spectrometry analysis. 405

Sample collection 406

Both forelimbs were collected from foetuses and foals aged 0-2 years (n=19) euthanised for 407

reasons unrelated to this project at a commercial abattoir or equine practices following owner 408

consent under ethical approval for use of the cadaveric material granted by the Veterinary 409

Research Ethics Committee, School of Veterinary Science, University of Liverpool 410

(VREC352). Collected tendons were split in the following age groups: Foetus (between 6 and 411

9 months of gestation; n=4); 0 days (full-term foetuses (average gestation 11-12 months) and 412

foals that did not weight-bear; n=4): 0-1 month (n=3); 3-6 months (n=4); 1-2 years (n=4). 413

The SDFT and CDET from one forelimb were dissected and wrapped in phosphate-buffered 414

saline dampened tissue paper and foil and stored at -80 oC for biomechanical testing. Two 1-2 415

cm segments from the mid-metacarpal area of the SDFT and CDET of the other forelimb 416

were dissected, and one fixed in 4% paraformaldehyde for histology and 417

immunohistochemistry, and the other snap frozen in isopentane and stored at -80 oC for laser 418

capture microdissection. 419

Biomechanical testing of the fibre phase 420

On the day of testing, samples were defrosted within their tissue paper wrap, then 421

immediately prepared for testing. Fascicles (fibre phase samples) were dissected from the 422

mid-metacarpal region of the SDFT and CDET and subjected to a quasi-static test to failure 423



https://doi.org/10.1101/2020.04.23.058081


23

according to Thorpe et al. (2015)(Chavaunne T. Thorpe et al., 2015). Briefly, prior to testing, 424

the diameter of each fascicle was measured along a 1 cm length in the middle of the fascicle 425

with a non-contact laser micrometre (LSM-501, Mitotuyo, Japan, resolution = 0.5 um) and 426

the smallest diameter recorded and used to calculate cross-sectional area (CSA), assuming a 427

circular shape. 428

Fibre phase samples were loaded in an electrodynamic testing machine (Instron ElectroPuls 429

1000) equipped with a 250 N load cell and pneumatic grips (4 bar) coated with rubber and 430

sandpaper to prevent sample slippage(C T Thorpe et al., 2012). The distance between the 431

grips was set to 20 mm and fibre phases preloaded to 0.1 N (approx. 2% fibre phase failure 432

load) to remove any slack in the sample. Following preload, the distance between the grips 433

was recorded as the gauge length, then fibre phase samples preconditioned with 10 loading 434

cycles between 0 and 3% strain (approximately 25% failure strain) using a sine wave at 1 Hz 435

frequency. Immediately after preconditioning, fibre samples were pulled to failure at a strain 436

rate of 5%/s. Force and extension data were continuously recorded at 100 Hz during both 437

preconditioning and the quasi-static test to failure. Acquired data were smoothed to reduce 438

noise before calculations with a 3rd order Savitzky-Golay low pass filter, with a frame of 15 439

for the preconditioning data and 51 for the pull to failure data. 440

Using the preconditioning data, the total percentage hysteresis and stress relaxation were 441

calculated, between the first and last preconditioning cycle. Failure force, extension, stress, 442

and strain were calculated from the test to failure, and a continuous modulus calculated 443

across every 10 data points of each stress strain curve, from which the maximum modulus 444

value was determined. The point of maximum modulus was defined as the yield point from 445

which yield stress and yield strain were determined. 446

Biomechanical testing of the matrix phase 447



https://doi.org/10.1101/2020.04.23.058081


24

On the day of testing, tendons were defrosted within their tissue paper wrap, and matrix 448

phase samples immediately dissected and prepared for biomechanical testing as described 449

previously by Thorpe et al. (2012)(C T Thorpe et al., 2012). Briefly, a group of two adjacent 450

intact fascicles (bound by the matrix phase) were dissected, after which the opposing end of 451

each fascicle was cut transversely 10 mm apart, to leave a consistent 10 mm length of intact 452

matrix phase that could be tested in shear (Supplementary Figure 1). 453

Utilising the same electrodynamic testing machine and pneumatic grips as described for the 454

fibre phase, the intact end of each fascicle was gripped with a grip to grip distance of 20 mm, 455

and a pre-load of 0.02 N (approx. 1% matrix phase failure load) applied. Matrix phase 456

samples were preconditioned with 10 cycles between 0 and 0.5 mm extension (approx. 25% 457

failure extension) using a sine wave at 1 Hz frequency, then pulled to failure at a speed of 1 458

mm/s. Force and extension data were continuously recorded at 100 Hz during both 459

preconditioning and the quasi-static test to failure. Acquired data was smoothed to reduce 460

noise before calculations with a 3rd order Savitzky-Golay low pass filter, with a frame of 15 461

for the preconditioning data and 51 for the pull to failure data. 462

Total percentage hysteresis and stress relaxation were again calculated between the first and 463

last preconditioning cycle. Failure force and extension were determined from the quasi-static 464

pull to failure curve, and a continual stiffness curve was calculated across every 10 data 465

points of the curve, from which maximum stiffness was determined, and yield force and yield 466

extension at maximum stiffness reported. Based on previous data demonstrating notable 467

differences in the toe region of the matrix phase curve of functionally distinct tendons, the 468

shape of failure curves was also compared between samples by calculating the amount of 469

matrix phase extension at different percentages of matrix phase failure load(Chavaunne T. 470

Thorpe et al., 2015). 471

Histology scoring 472



https://doi.org/10.1101/2020.04.23.058081


25

Paraformaldehyde-fixed paraffin-embedded longitudinal SDFT and CDET segments were 473

sectioned at 6 µm thickness and stained with H&E for histologic examination and scoring 474

(n=13; 3 from each age group, 4 from 1-2 years age group). The examined histologic 475

variables are reported in Supplementary Table 1. 476

For parameters scored by investigators, the sections were blinded and histologic variables 477

assigned a grade from 0 to 3 by two independent investigators. Weighted Kappa showed 478

moderate to good agreement in all instances, hence the average of the two scores was used. 479

Other histologic variables were measured using image analysis (TissueFaxs, Tissuegnostics 480

and Adobe Photoshop CS3) and then assigned a grade from 0 to 3. Cumulative scores for the 481

fibre and matrix phase for each horse were obtained by summing the scores of the fibre and 482

matrix phase variables, respectively, excluding matrix phase percentage to ensure matrix 483

phase dimensions were not over-weighted in final reporting. 484

Immunohistochemistry 485

Immunohistochemical analysis for DCN, FMOD, PRG4, TNC and ELN was carried out on 486

paraformaldehyde-fixed paraffin-embedded longitudinal SDFT and CDET sections (6 µm 487

thickness) (n=12; 3 from each age group) as previously described by Zamboulis et al. 488

(2013)(Zamboulis et al., 2013). Antigen retrieval was carried out with 0.2 U/mL 489

Chondroitinase ABC (C2905, Sigma, Merck, Darmstadt, Germany) at 37 oC for two hours for 490

DCN, FMOD, PRG4, and TNC or with 4800 U/mL hyaluronidase (H3506, Sigma, Merck, 491

Darmstadt, Germany) at 37 oC for two hours for ELN. Primary antibodies were used at a 492

concentration of 1:1500 for DCN (mouse IgG), 1:400 for FMOD (rabbit IgG), 1:200 for 493

PRG4 (mouse IgG), 1:250 for TNC (mouse IgG, sc-59884, Santa Cruz Biotechnology, 494

Dallas, Texas), and 1:100 for ELN (mouse IgG, ab9519, Abcam, Cambridge, UK). 495

Antibodies for DCN and PRG4 were a kind gift from Prof. Caterson, Cardiff University, UK, 496

and the FMOD antibody was kindly provided by Prof. Roughley, McGill University, Canada. 497



https://doi.org/10.1101/2020.04.23.058081


26

The secondary antibody incubation was performed with the Zytochem Plus HRP Polymer 498

anti-rabbit for FMOD and anti-mouse for DCN, PRG4, TNC, and ELN (ZUC032-006 and 499

ZUC050-006, Zytomed Systems, Berlin, Germany). Immunohistochemical staining was 500

graded from 0 to 3 (low to high) on blinded sections, assessing stained area and staining 501

intensity for DCN, FMOD, TNC, and ELN. For PRG4, where staining was confined to the 502

pericellular area, staining intensity was measured using TissueFaxs (Tissuegnostics). 503

Quantification of tendon elastin 504

The elastin content of SDFT and CDET samples from each age group (n=12; 3 from each 505

age group) was quantified using the FASTINTM Elastin Assay (Biocolor, Carrickfergus, 506

UK)(Godinho et al., 2017). Briefly, SDFT and CDET tissue was powdered (~15 mg wet 507

weight) and incubated with 750 µl of 0.25 M oxalic acid at 100 °C for 2 one hour cycles to 508

extract all soluble a-elastin from the tissue. Preliminary tests showed two extractions were 509

sufficient to solubilise all a-elastin from developing SDFT and CDET. Following 510

extraction, samples and standards were processed in duplicate according to the 511

manufacturer’s instructions and their absorbance determined spectrophotometrically at 512

513 nm (Spectrostar Nano microplate reader, BMG Labtech, Aylesbury, UK). A standard 513

curve was used to calculate the samples’ elastin concentration and elastin was expressed as 514

a percentage of tendon wet weight. 515

Laser-capture microdissection 516

Laser-capture microdissection was used to collect samples from the fibre and matrix phase of 517

SDFT samples from all age groups (n=4 for each age group with the exception of the 0-1 518

month group where n=3). For this purpose, 12 µm transverse cryosections were cut from the 519

SDFT samples and mounted on steel frame membrane slides (1.4 µm PET membrane, Leica 520

Microsystems, Wetzlar, Germany). Frozen sections were dehydrated in 70% and 100% ice-521



https://doi.org/10.1101/2020.04.23.058081


27

cold ethanol, allowed to briefly dry, and regions of fibre and matrix phase laser-captured on 522

an LMD7000 laser microdissection microscope (Leica Microsystems, Wetzlar, Germany) and 523

collected in LC/MS grade water (FisherScientific, Hampton, New Hampshire). Collected 524

samples were immediately snap frozen and stored at -80oC for mass spectrometry analysis. 525

Mass spectrometry analysis 526

Mass spectrometry analysis of laser-captured SDFT matrix and fibre phase samples was 527

carried out as previously described by Thorpe et al. (2016)(Chavaunne T. Thorpe, Peffers, et 528

al., 2016). Samples were digested for mass spectrometry analysis with incubation in 0.1% 529

(w/v) Rapigest (Waters, Herts, UK) for 30 min at room temperature followed by 60 min at 60 530

°C and subsequent trypsin digestion. LC MS/MS was carried out at the University of 531

Liverpool Centre for Proteome Research using an Ultimate 3000 Nano system 532

(Dionex/Thermo Fisher Scientific, Waltham, Massachusetts) for peptide separation coupled 533

online to a Q-Exactive Quadrupole-Orbitrap mass spectrometer (Thermo Scientific, 534

Waltham, Massachusetts) for MS/MS acquisition. Initial ranging runs on short gradients were 535

carried out to determine the sample volume to be loaded on the column and subsequently 536

between 1-9 µL of sample was loaded onto the column on a one hour gradient with an inter-537

sample 30 min blank. 538

Protein identification and label-free quantification 539

Fibre and matrix phase proteins were identified using Peaks® 8.5 PTM software 540

(Bioinformatics Solutions, Waterloo, Canada), searching against the UniHorse database 541

(http://www.uniprot.org/proteomes/). Search parameters used were: peptide mass tolerance 542

10 ppm, fragment mass tolerance 0.01 Da, fixed modification carbamidomethylation, variable 543

modifications methionine oxidation and hydroxylation. Search results for peptide 544

identification were filtered with a false discovery rate (FDR) of 1%, and for protein 545



https://doi.org/10.1101/2020.04.23.058081


28

identification with a minimum of 2 unique peptides per protein, and a confidence score >20 (-546

10lgp>20). Label-free quantification was also carried out using Peaks® 8.5 PTM software for 547

the SDFT fibre and matrix phase separately. Protein abundances were normalised for 548

collected laser-capture area and volume loaded onto the mass spectrometry column and 549

differentially abundant proteins between the age groups in the SDFT fibre and matrix phase 550

were identified using a fold change ≥2 and p<0.05 (PEAKS adjusted p values). The mass 551

spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via 552

the PRIDE partner repository, with the dataset identifier PXD012169 and 553

10.6019/PXD012169. 554

Gene ontology and network analysis 555

The dataset of identified proteins in the SDFT fibre and matrix phase were classified for cell 556

compartment association with the Ingenuity Pathway Analysis software (IPA, Qiagen, 557

Hilden, Germany) and for matrisome categories with The Matrisome Project database(Hynes 558

& Naba, 2012). Protein pathway analysis for the differentially abundant proteins between age 559

groups in the SDFT fibre and matrix phase was carried out in IPA. Protein interactions maps 560

were created in IPA allowing for experimental evidence and highly predicted functional 561

links. 562

Neopeptide identification 563

For neopeptide identification, mass spectrometry data was analysed using Mascot server 564

(Matrix Science) with the search parameters: enzyme semiTrypsin, peptide mass tolerance 10 565

ppm, fragment mass tolerance 0.01 Da, charge 2+ and 3+ ions, and missed cleavages 1. The 566

included modifications were: fixed carbamidomethyl cysteine, variables oxidation of 567

methionine, proline, and lysine, and the instrument type selected was electrospray ionization-568

TRAP (ESI-TRAP). The Mascot-derived ion score was used to determine true matches 569



https://doi.org/10.1101/2020.04.23.058081


29

(p<0.05), where p was the probability that an observed match was a random event. The 570

peptide list was exported and processed with the Neopeptide Analyser, a software tool for the 571

discovery of neopeptides in proteomic data(Peffers et al., 2017). Obtained neopeptide 572

abundances for each sample were normalised for total peptide abundance for that protein and 573

sample, and normalised neopeptide abundances were subsequently summed for each protein 574

and the total neopeptide abundance analysed for differential abundance across the age groups 575

in the SDFT fibre and matrix phase using p<0.05 and FDR 5% (ANOVA and Benjamini-576

Hochberg FDR). 577

Relative mRNA expression 578

RNA extraction from whole SDFT and CDET was carried out followed by reverse 579

transcription. Quantitative real-time PCR (qRT-PCR) was performed on an ABI7300 system 580

(Thermo Fisher Scientific Waltham, Massachusetts) using the Takyon ROX SYBR 2X 581

MasterMix (Eurogentec, Liege, Belgium). qRT-PCR was undertaken using gene-specific 582

primers for DCN, FMOD, BGN, COMP, COL1A1, COL1A2, COL3A1, TGFB1, and 583

GAPDH as a reference gene previously validated (Supplementary Table 2). Relative 584

expression levels were normalised to GAPDH expression and calculated with the formula 585

E−ΔCt following primer efficiency calculation. 586

Statistical analysis 587

Statistical analysis was carried out in SigmaPlot (Systat Software Inc, San Jose, California) 588

unless otherwise stated. Details of the n numbers for each experiment and the statistical test 589

used for the analysis of the data are listed in Supplementary Table 3. Heatmaps were 590

designed in GProX(Rigbolt et al., 2011). The Central Limit Theorem (CLT) was used to 591

assume normality where n>30 and where n<30 normality was tested using the Shapiro-Wilks 592



https://doi.org/10.1101/2020.04.23.058081


30

test. If data were found not to be normally distributed their log10 transformation or ANOVA 593

on Ranks was used for statistical analysis but the original data was presented in graphs. 594

595

Acknowledgements 596

This work was funded by the Horserace Betting Levy Board, PRJ/776. We would like to 597

acknowledge the equine practices that provided samples for this study. 598

599

Author Contributions 600

DEZ designed and performed experiments, analysed the data, and wrote the manuscript. 601

CTT assisted with study design, data analysis, and edited the manuscript. YAK assisted 602

with data collection and analysis. HLB, HRCS, PDC conceived the study, assisted with 603

data analysis, and edited the manuscript. 604

605

Competing Interests statement 606

The authors declare that they have no competing interests. 607

608

Data and materials availability 609

The mass spectrometry proteomics data have been deposited to the ProteomeXchange 610

Consortium via the PRIDE partner repository, with the dataset identifier PXD012169 and 611

10.6019/PXD012169. 612

613

References 614

Ansorge, H. L., Adams, S., Birk, D. E., & Soslowsky, L. J. (2011). Mechanical, 615



https://doi.org/10.1101/2020.04.23.058081


31

compositional, and structural properties of the post-natal mouse Achilles tendon. Annals 616

of Biomedical Engineering, 39(7), 1904–1913. https://doi.org/10.1007/s10439-011-617

0299-0 618

Armiento, A. R., Stoddart, M. J., Alini, M., & Eglin, D. (2018). Biomaterials for articular 619

cartilage tissue engineering: Learning from biology. Acta Biomaterialia, 65, 1–20. 620

https://doi.org/10.1016/j.actbio.2017.11.021 621

Batson, E. L., Paramour, R. J., Smith, T. J., Birch, H. L., Patterson-Kane, J. C., & Goodship, 622

A. E. (2010). Are the material properties and matrix composition of equine flexor and 623

extensor tendons determined by their functions? Equine Veterinary Journal, 35(3), 314–624

318. https://doi.org/10.2746/042516403776148327 625

Biewener, A. A. (1998). Muscle-tendon stresses and elastic energy storage during locomotion 626

in the horse. Comparative Biochemistry and Physiology - B Biochemistry and Molecular 627

Biology, 120(1), 73–87. https://doi.org/10.1016/S0305-0491(98)00024-8 628

Birch, H. L. (2007). Tendon matrix composition and turnover in relation to functional 629

requirements. International Journal of Experimental Pathology, 88(4), 241–248. 630

https://doi.org/10.1111/j.1365-2613.2007.00552.x 631

Cherdchutham, W., Becker, C. K., Spek, E. R., Voorhout, W. E., & Van Weeren, P. R. 632

(2001). Effects of exercise on the diameter of collagen fibrils in the central core and 633

periphery of the superficial digital flexor tendon in foals. American Journal of 634

Veterinary Research, 62(10), 1563–1570. https://doi.org/10.2460/ajvr.2001.62.1563 635

Cherdchutham, W., Meershoek, L. S., Weeren, P. R. Van, & Barneveld, A. (2001). Effects of 636

exercise on biomechanical properties of the superficial digital flexor tendon in foals. 637

American Journal of Veterinary Research, 62(12), 1859–1864. https://avmajournals-638

avma-org.proxy1.cl.msu.edu/doi/pdf/10.2460/ajvr.2001.62.1859 639

Choi, R. K., Smith, M. M., Smith, S., Little, C. B., & Clarke, E. C. (2019). Functionally 640



https://doi.org/10.1101/2020.04.23.058081


32

distinct tendons have different biomechanical, biochemical and histological responses to 641

in vitro unloading. Journal of Biomechanics, 95, 109321. 642

https://doi.org/10.1016/j.jbiomech.2019.109321 643

Choi, R., Smith, M., Clarke, E., & Little, C. (2018). Cellular, matrix, and mechano-biological 644

differences in load-bearing versus positional tendons throughout development and 645

aging: a narrative review. Connective Tissue Research, 59(5), 483–494. 646

https://doi.org/10.1080/03008207.2018.1504929 647

Freedman, B. R., Bade, N. D., Riggin, C. N., Zhang, S., Haines, P. G., Ong, K. L., & Janmey, 648

P. A. (2015). The (dys)functional extracellular matrix. Biochimica et Biophysica Acta - 649

Molecular Cell Research, 1853(11), 3153–3164. 650

https://doi.org/10.1016/j.bbamcr.2015.04.015 651

Gao, F., Li, C., Zhou, J., Shen, X., Hu, B., Lu, M., Liu, Y., Zhang, B., Fang, Y., & Liu, Y. 652

(2019). Progress on relationship between transforming growth factor-beta1 and 653

tendinopathy. China Journal of Orthopaedics and Traumatology, 32(4), 377–382. 654

Gardner, K., Arnoczky, S. P., Caballero, O., & Lavagnino, M. (2008). The effect of stress-655

deprivation and cyclic loading on the TIMP/MMP ratio in tendon cells: An in vitro 656

experimental study. Disability and Rehabilitation, 30(20–22), 1523–1529. 657

https://doi.org/10.1080/09638280701785395 658

Godinho, M. S. C., Thorpe, C. T., Greenwald, S. E., & Screen, H. R. C. (2017). Elastin is 659

Localised to the Interfascicular Matrix of Energy Storing Tendons and Becomes 660

Increasingly Disorganised With Ageing. Scientific Reports, 7(1), 1–11. 661

https://doi.org/10.1038/s41598-017-09995-4 662

Hynes, R. O., & Naba, A. (2012). Overview of the matrisome-An inventory of extracellular 663

matrix constituents and functions. Cold Spring Harbor Perspectives in Biology, 4(1), 1–664

16. https://doi.org/10.1101/cshperspect.a004903 665



https://doi.org/10.1101/2020.04.23.058081


33

Kalson, N. S., Holmes, D. F., Herchenhan, A., Lu, Y., Starborg, T., & Kadler, K. E. (2011). 666

Slow stretching that mimics embryonic growth rate stimulates structural and mechanical 667

development of tendon-like tissue in vitro. Developmental Dynamics, 240(11), 2520–668

2528. https://doi.org/10.1002/dvdy.22760 669

KM, H., P, S., J, H., SP, M., & M, K. (2013). Lack of tissue renewal in human adult Achilles 670

tendon is revealed by nuclear bomb 14C (pp. 2074–2079). 671

Kohrs, R. T., Zhao, C., Sun, Y. L., Jay, G. D., Zhang, L., Warman, M. L., An, K. N., & 672

Amadio, P. C. (2011). Tendon fascicle gliding in wild type, heterozygous, and lubricin 673

knockout mice. Journal of Orthopaedic Research, 29(3), 384–389. 674

https://doi.org/10.1002/jor.21247 675

Kostrominova, T. Y., & Brooks, S. V. (2013). Age-related changes in structure and 676

extracellular matrix protein expression levels in rat tendons. Age, 35(6), 2203–2214. 677

https://doi.org/10.1007/s11357-013-9514-2 678

Kuo, C., Petersen, B., & Tuan, R. (2008). Spatiotemporal protein distribution of TGF-ß’s, 679

their receptors, and extracellular matrix molecules during embryonic tendon 680

development. Developmental Dynamics, 237(5), 1477–1489. 681

https://doi.org/10.1002/dvdy.21547.Spatiotemporal 682

Maeda, T., Sakabe, T., Sunaga, A., Sakai, K., Rivera, A. L., Keene, D. R., Sasaki, T., 683

Stavnezer, E., Iannotti, J., Schweitzer, R., Ilic, D., Baskaran, H., & Sakai, T. (2011). 684

Conversion of mechanical force into TGF-β-mediated biochemical signals. Current 685

Biology, 21(11), 933–941. https://doi.org/10.1016/j.cortex.2009.08.003.Predictive 686

Magnusson, S. P., & Kjaer, M. (2019). The impact of loading, unloading, ageing and injury 687

on the human tendon. Journal of Physiology, 597(5), 1283–1298. 688

https://doi.org/10.1113/JP275450 689

Marturano, J. E., Arena, J. D., Schiller, Z. A., Georgakoudi, I., & Kuo, C. K. (2013). 690



https://doi.org/10.1101/2020.04.23.058081


34

Characterization of mechanical and biochemical properties of developing embryonic 691

tendon. Proceedings of the National Academy of Sciences of the United States of 692

America, 110(16), 6370–6375. https://doi.org/10.1073/pnas.1300135110 693

Mendias, C. L., Gumucio, J. P., & Lynch, E. B. (2012). Mechanical loading and TGF-β 694

change the expression of multiple miRNAs in tendon fibroblasts. Journal of Applied 695

Physiology, 113(1), 56–62. https://doi.org/10.1152/japplphysiol.00301.2012 696

Pan, X. S., Li, J., Brown, E. B., & Kuo, C. K. (2018). Embryo movements regulate tendon 697

mechanical property development. Philosophical Transactions of the Royal Society B: 698

Biological Sciences, 373(1759). https://doi.org/10.1098/rstb.2017.0325 699

Patterson-Kane, J. C., & Rich, T. (2014). Achilles tendon injuries in elite athletes: Lessons in 700

pathophysiology from their equine Counterparts. ILAR Journal, 55(1), 86–99. 701

https://doi.org/10.1093/ilar/ilu004 702

Peffers, M. J., Jones, A. R., McCabe, A., & Anderson, J. (2017). Neopeptide Analyser: A 703

software tool for neopeptide discovery in proteomics data. Wellcome Open Research, 704

2(May), 1–9. https://doi.org/10.12688/wellcomeopenres.11275.1 705

Rigbolt, K. T. G., Vanselow, J. T., & Blagoev, B. (2011). GProX, a user-friendly platform for 706

bioinformatics analysis and visualization of quantitative proteomics data. Molecular and 707

Cellular Proteomics, 10(8), 1–10. https://doi.org/10.1074/mcp.O110.007450 708

Ritty, T. M., Ditsios, K., & Starcher, B. C. (2002). Distribution of the elastic fiber and 709

associated proteins in flexor tendon reflects function. Anatomical Record, 268(4), 430–710

440. https://doi.org/10.1002/ar.10175 711

Russo, V., Mauro, A., Martelli, A., Di Giacinto, O., Di Marcantonio, L., Nardinocchi, D., 712

Berardinelli, P., & Barboni, B. (2015). Cellular and molecular maturation in fetal and 713

adult ovine calcaneal tendons. Journal of Anatomy, 226(2), 126–142. 714

https://doi.org/10.1111/joa.12269 715



https://doi.org/10.1101/2020.04.23.058081


35

Sensini, A., Gualandi, C., Focarete, M., Belcari, J., Zucchelli, A., Boyle, L., Reilly, G., Kao, 716

A., Tozzi, G., & Cristofolini, L. (2019). Multiscale hierarchical bioresorbable scaffolds 717

for the regeneration of tendons and ligaments. Biofabrication, 11(3), 035026. 718

Spiesz, E. M., Thorpe, C. T., Chaudhry, S., Riley, G. P., Birch, H. L., Clegg, P. D., & Screen, 719

H. R. C. (2015). Tendon extracellular matrix damage, degradation and inflammation in 720

response to in vitro overload exercise. Journal of Orthopaedic Research, 33(6), 889–721

897. https://doi.org/10.1002/jor.22879 722

Stanley, R. L., Edwards, L. J., Goodship, A. E., Firth, E. C., & Patterson-Kane, J. C. (2008). 723

Effects of exercise on tenocyte cellularity and tenocyte nuclear morphology in immature 724

and mature equine digital tendons. Equine Veterinary Journal, 40(2), 141–146. 725

https://doi.org/10.2746/042516408X266097 726

Thorpe, C. T., Chaudhry, S., Lei, I. I., Varone, A., Riley, G. P., Birch, H. L., Clegg, P. D., & 727

Screen, H. R. C. (2015). Tendon overload results in alterations in cell shape and 728

increased markers of inflammation and matrix degradation. Scandinavian Journal of 729

Medicine and Science in Sports, 25(4), e381–e391. https://doi.org/10.1111/sms.12333 730

Thorpe, C. T., & Screen, H. R. C. (2016). Tendon structure and composition. In Advances in 731

Experimental Medicine and Biology (Vol. 920). https://doi.org/10.1007/978-3-319-732

33943-6 733

Thorpe, C T, Udeze, C. P., Birch, H. L., Clegg, P. D., & Screen, H. R. C. (2012). 734

Specialization of Tendon Mechanical Properties Results From Interfascicular 735

Differences. Journal of The Royal Society Interface, 9(76), 3108–3117. 736

Thorpe, Chavaunne T., Godinho, M. S. C., Riley, G. P., Birch, H. L., Clegg, P. D., & Screen, 737

H. R. C. (2015). The interfascicular matrix enables fascicle sliding and recovery in 738

tendon, and behaves more elastically in energy storing tendons. Journal of the 739

Mechanical Behavior of Biomedical Materials, 52, 85–94. 740



https://doi.org/10.1101/2020.04.23.058081


36

https://doi.org/10.1016/j.jmbbm.2015.04.009 741

Thorpe, Chavaunne T., Karunaseelan, K. J., Ng Chieng Hin, J., Riley, G. P., Birch, H. L., 742

Clegg, P. D., & Screen, H. R. C. (2016). Distribution of proteins within different 743

compartments of tendon varies according to tendon type. Journal of Anatomy, 229(3), 744

450–458. https://doi.org/10.1111/joa.12485 745

Thorpe, Chavaunne T., Peffers, M. J., Simpson, D., Halliwell, E., Screen, H. R. C., & Clegg, 746

P. D. (2016). Anatomical heterogeneity of tendon: Fascicular and interfascicular tendon 747

compartments have distinct proteomic composition. Scientific Reports, 6(February), 1–748

12. https://doi.org/10.1038/srep20455 749

Thorpe, Chavaunne T., Riley, G. P., Birch, H. L., Clegg, P. D., & Screen, H. R. C. (2016). 750

Fascicles and the interfascicular matrix show adaptation for fatigue resistance in energy 751

storing tendons. Acta Biomaterialia, 42, 308–315. 752

https://doi.org/10.1016/j.actbio.2016.06.012 753

Thorpe, Chavaunne T., Streeter, I., Pinchbeck, G. L., Goodship, A. E., Clegg, P. D., & Birch, 754

H. L. (2010). Aspartic acid racemization and collagen degradation markers reveal an 755

accumulation of damage in tendon collagen that is enhanced with aging. Journal of 756

Biological Chemistry, 285(21), 15674–15681. https://doi.org/10.1074/jbc.M109.077503 757

Thorpe, Chavaunne T., Udeze, C. P., Birch, H. L., Clegg, P. D., & Screen, H. R. C. (2013). 758

Capacity for sliding between tendon fascicles decreases with ageing in injury prone 759

equine tendons: A possible mechanism for age-related tendinopathy? European Cells 760

and Materials, 25(0), 48–60. 761

Watts, A. E., Millar, N. L., Platt, J., Kitson, S. M., Akbar, M., Rech, R., Griffin, J., Pool, R., 762

Hughes, T., McInnes, I. B., & Gilchrist, D. S. (2017). MicroRNA29a Treatment 763

Improves Early Tendon Injury. Molecular Therapy, 25(10), 2415–2426. 764

https://doi.org/10.1016/j.ymthe.2017.07.015 765



https://doi.org/10.1101/2020.04.23.058081


37

Zamboulis, D. E., Senior, J. M., Clegg, P. D., Gallagher, J. A., Carter, S. D., & Milner, P. I. 766

(2013). Distribution of purinergic P2X receptors in the equine digit, cervical spinal cord 767

and dorsal root ganglia. Purinergic Signalling, 9(3), 383–393. 768

https://doi.org/10.1007/s11302-013-9356-5 769

770

771



https://doi.org/10.1101/2020.04.23.058081


38

Supplementary data 772

Supplementary Figure 1. Tendon structure and graphical abstract for biomechanical testing. 773

Supplementary Figure 2. Scoring of histologic variables for the matrix and fibre phases in the 774

SDFT and CDET through postnatal development. 775

Supplementary Figure 3. Scoring of ELN staining for the matrix and fibre phases in the 776


Supplementary Figure 4. Classification of SDFT matrix and fibre phase identified proteins 778

and differentially abundant proteins according to their associated location. 779

Supplementary Figure 5. Relative mRNA expression of major ECM genes in whole tissue 780


Supplementary Table 1. Histologic variables used in the H&E scoring of the SDFT and 782

CDET sections and the analysis method and reporting criteria adopted. 783

Supplementary Table 2. Gene primer sequences used in relative mRNA expression analysis. 784

Supplementary Table 3. Samples used for analysis along with statistical test used for analysis. 785

Supplementary Table 4. Collagens and proteoglycans identified in SDFT matrix and fibre 786

phases. 787

788

789

790

791

792

793

794

795



https://doi.org/10.1101/2020.04.23.058081


39

796 Supplementary Figure 1. Tendon structure and graphical abstract for biomechanical 797

testing. (a) Tendon structure (adapted from Spiesz et al. 2015)(Spiesz et al., 2015). (b) H&E 798

section of fibre and matrix phase and schematic of fibre and matrix phase dissection and 799

biomechanical testing. 800

801

802



https://doi.org/10.1101/2020.04.23.058081


40

803 Supplementary Figure 2. Scoring of histologic variables for the matrix and fibre phases 804

in the SDFT and CDET through postnatal development. * significant difference between 805

tendons, a-f significant difference between age groups. Error bars depict standard deviation. 806



https://doi.org/10.1101/2020.04.23.058081


41

807

808

809 Supplementary Figure 3. Scoring of ELN staining for the matrix and fibre phases in the 810

SDFT and CDET through postnatal development. Error bars depict standard deviation. 811

812

813

814

815

Supplementary Figure 4. Classification of SDFT matrix and fibre phase identified 816

proteins and differentially abundant proteins (p<0.05, fold change≥2) according to their 817

associated location. 818

819

820



https://doi.org/10.1101/2020.04.23.058081


42

821

Supplementary Figure 5. Relative mRNA expression of major ECM genes in whole 822

tissue SDFT and CDET through postnatal development. * significant difference between 823

tendons, a-e significant difference between age groups. Error bars depict standard deviation. 824



https://doi.org/10.1101/2020.04.23.058081


43

Supplementary Table 1. Histologic variables used in the H&E scoring of the SDFT and 825

CDET sections and the analysis method and reporting criteria adopted. When software 826

was adopted to acquire a parameter, the software is additionally reported. “Scored” is used to 827

denote parameters analysed by blinded investigators. 828

Histologic variable Method Criteria

Fibre phase

Cellularity measured (TissueFaxs) 0 – 3, fewer - more

Shape of tendon cells scored 0 – 3, elongated to rounded

Organisation of collagen fibres scored 0 – 3, linear to non-linear

Crimp angle measured (PhotoShop) 0 – 3, smaller to larger

Matrix phase

Percentage of matrix phase measured (TissueFaxs) 0 – 3, lower to higher

Matrix phase width measured (TissueFaxs) 0 – 3, smaller to larger

Cellularity measured (TissueFaxs) 0 – 3, fewer - more

829

830

831

832

833

834

835

836

837

838

839



https://doi.org/10.1101/2020.04.23.058081


44

Supplementary Table 2. Gene primer sequences used in relative mRNA expression 840

analysis. 841

Gene Primer sequence

DCN F: CATCCAGGTTGTCTACCTTCATAACA

R: CCAGGTGGGCAGAAGTCATT

FMOD F: CTTGGCTCCAGACCCTGAAA

R: TGCCCCTCGCGTCAGA

BGN F: TCACCTTCCAGCCCCTAGAGT

R: AGAAGCAGCCCCTCCTCAA

COMP F: GGTGCGGCTGCTATGGAA

R: CCAGCTCAGGGCCCTCAT

COL1A1 F: GACTGGCAACCTCAAGAAGG

R: CAATATCCAAGGGAGCCACA

COL1A2 F: GCACATGCCGTGACTTGAGA

R: CATCCATAGTGCATCCTTGATTAGG

COL3A1 F: ACGCAAGGCCGTGAGACTA

R: TGATCAGGACCACCAACATCA

TGFB1 F: CCCTGCCCCTACATTTGGA

R: CGGGTTGTGCTGGTTGTACA

GAPDH F: GCATCGTGGAGGGACTCA

R: GCCACATCTTCCCAGAGG

842



https://doi.org/10.1101/2020.04.23.058081


45

Supplementary Table 3. Samples used for analysis along with statistical test used for analysis. n numbers represent biological replicates. 843 Analysis Tendon Foetus 0 days 0-1 month 3-6 months 1-2 years Statistical analysis

Biomechanics SDFT fibre n=4 n=4 n=3 n=4 n=4 Two-way ANOVA. Outliers < or >

mean 2xSD were removed. The

average of 4-7 technical replicates was

used (depending on sample availability).

SDFT matrix n=4 n=4 n=3 n=4 n=4

CDET fibre n=4 n=4 n=3 n=4 n=4


Histology and Immunohistochemistry SDFT n=3 n=3 n=3 (0-1 year) n=4 (hist.), 3 (IHC) Kruskal-Wallis, Rank Sum Test

CDET n=3 n=3 n=3 (0-1 year) n=4 (hist.), 3 (IHC)

Elastin quantification SDFT n=3 n=3 n=3 (0-1 year) n=3 Two-way ANOVA. The average of 2

technical replicates was used. CDET n=3 n=3 n=3 (0-1 year) n=3

Proteomics SDFT fibre n=3 n=4 n=3 n=4 n=4 Differentially abundant proteins –

PEAKS Q


Differential total neopeptide abundance

– One-way ANOVA, Benjamini-

Hochberg correction

Relative mRNA expression SDFT - n=4 n=3 n=4 n=4 Two-way ANOVA

CDET - n=4 n=3 n=4 n=4

844



https://doi.org/10.1101/2020.04.23.058081


46

Supplementary Table 4. Collagens and proteoglycans identified in SDFT matrix and 845

fibre phases (1% FDR, protein -10lgP >20, and ≥2 unique peptides). 846

Collagens Matrix Fibre Proteoglycans Matrix Fibre

COL1A1 + + ACAN + +

COL1A2 + + ASPN + +

COL2A1 + + BGN + +

COL3A1 + + CHAD - +

COL4A1 + - DCN + +

COL4A2 + - FMOD + +

COL4A3 - + HAPLN1 + +

COL4A6 + - HSPG2 + +

COL5A1 + + KERA + +

COL5A2 + + LUM + +

COL5A3 + - OGN + +

COL6A1 + + PRELP + +

COL6A2 + + VCAN + -

COL6A3 + +

COL8A1 + -

COL11A1 + +

COL12A1 + +

COL14A1 + +

COL15A1 + +

COL17A1 - +

COL18A1 +

COL21A1 + +

COL28A1 +

847

848



https://doi.org/10.1101/2020.04.23.058081




https://doi.org/10.1101/2020.04.23.058081


postnatal mechanical loading drives adaptation of tissues … · 2020-04-23 · 1 1 post natal...

Documents