Mechanical Strain Modulates Extracellular Matrix Degradation and Byproducts in an Isoform-Specific Manner

Abstract

Many studies have shown that mechanical forces can alter collagen degradation by proteases, and this mechanochemical effect may potentially serve an important role in determining extracellular matrix content and organization in load-bearing tissues. However, it is not yet known whether mechano-sensitive degradation depends on particular protease isoforms, nor is it yet known whether particular degradation byproducts can be altered by mechanical loading. In this study, we tested the hypothesis that different types of proteases exhibit different sensitivities to mechanical loading both in degradation rates and byproducts. Decellularized porcine pericardium samples were treated with human recombinant matrix metalloproteinases-1, -8, -9, cathepsin K, or a protease-free control while subjected to different levels of strain in a planar, biaxial mechanical tester. Tissue degradation was monitored by tracking the decay in mechanical stresses during displacement control tests, and byproducts were assessed by mass spectrometry analysis of the sample supernatant after degradation. Our key finding shows that cathepsin K-mediated degradation of collagenous tissue was enhanced with increasing strain, while MMP1-, MMP8-, and MMP9-mediated degradation were first decreased and then increased by strain. Degradation induced changes in tissue mechanical properties, and proteomic analysis revealed strain-sensitive degradome signatures with different ECM byproducts released at low vs. high strains. This evidence suggests a potentially new type of mechanobiology wherein mechanical forces alter the degradation products that can provide important signaling feedback functions during tissue remodeling.

Introduction

The extracellular matrix (ECM) forms connective tissue around cells and serves as a major regulator of tissue structure and function across development and disease [1–3]. The ECM is comprised of many molecules, most notably collagen fibrils and proteoglycans that are present in all connective tissues in various isoforms and assemblies [4]. Collagen is the most abundant protein in the human body and comprises approximately one third of the total protein [5]. There are more than 25 types of collagen, the most common being fibrillar types I and III, composed of three polypeptide subunits that exist in a triple helix form [5,6]. The form and structure of ECM depends mainly on collagen content and organization wherein collagen fibrils self-assemble into large-scale structures such as fibers and sheets [6,7]. The primary role of fibril forming collagens (I, II, III, V, and XI) is to bear and transmit mechanical loads along their main axis [8].

Collagen turnover is centrally involved in many diseases including cardiac fibrosis, pulmonary fibrosis, wound healing, cancer metastasis, and myocardial infarction (MI) where collagen accumulates in the infarct zone to form collagenous scar tissue [9,10]. The degradation of collagen is mediated by various proteases, primarily different isoforms of matrix metalloproteinases (MMPs) and cathepsins. The MMP family consists of many different members that can be categorized in different groups based on the substrate they prefer to degrade. The collagenases (MMP-1, MMP-8, and MMP-13) can cleave fibrillar type collagens and the gelatinases (MMP-2 and MMP-9) can degrade gelatins [1,11,12]. MMP-9 was first thought to be only gelatinase, but recent studies showed that it is also able to degrade full length interstitial collagens [9,11]. Cathepsins are a superfamily of cysteine proteases, and some members can proteolyze ECM. Cathepsin K is the most potent mammalian collagenase, and it can cleave type I collagen both in the native triple helix and in the telopeptide regions [13,14].

Previous studies have established that mechanical strain can modulate the enzymatic degradation of collagen fibers by altering the ‘mechanochemistry’ of collagen-protease binding [7,10,15]. This occurs presumably due to altered molecular conformations of the collagen molecule’s protease-binding sites (Table 1). However, while some groups report increased collagen turnover with increased loading, other groups report decreased collagen turnover with loading [8,10,12,15–27]. In the current study, we sought to test the hypothesis that different types of proteases might exhibit different sensitivities to mechanical loading. Specifically, we measured the degradation of porcine pericardium samples treated with MMP-1, MMP-8, MMP-9, or cathepsin K while subjected to different levels of mechanical loading. We also measured the proteomic signatures of degradation byproducts to assess how mechanical loading could alter the particular types of products released after degradation of porcine pericardium samples by each protease.

Table 1.

Literature report of effect of loading on collagen degradation

Reference	Collagen structure	Collagenase type	Degradation effect
Camp (2011)	Single molecule	Bacterial collagenase	Decrease
Adhikari (2011)	Homotrimeric peptide	MMP-1	Increase
Adhikari (2012)	Single molecule	MMP-1	Increase
Bhole (2009)	Fibril Network	Bacterial collagenase	Decrease
Flynn (2010)	Fibril Network	MMP-8	Decrease
Flynn (2013)	Single fibril	Bacterial collagenase	Decrease
Huang (1977)	Reconstituted tape	Bacterial collagenase	Decrease
Nabeshima (1996)	Tendon-tibia units	Bacterial collagenase	Decrease
Ellsmere (1999)	Pericardial tissue	Bacterial collagenase	Increase
Ruberti (2005)	Corneal tissue	Bacterial collagenase	Decrease
Wyatt (2009)	Rat-tail tissue	Bacterial collagenase	Decrease
Zareian (2010)	Corneal tissue	Bacterial collagenase	Decrease
Yi (2016)	Lung tissue	Bacterial collagenase	Increase, then Decrease
Ghazanfari (2016)	Pericardial tissue	Bacterial collagenase	Decrease

Materials and Methods

Sample Preparation

In order to assess the effect of tensile loading on collagen degradation, we subjected porcine pericardium samples to different levels of equibiaxial tensile displacement with or without different proteases. Porcine pericardium samples were collected from fresh pig hearts at a local slaughterhouse. Pericardium samples were decellularized using a standard decellularization protocol that has previously been shown to have minimal effects on ECM content [28-29,32]. Shortly, upon receipt, tissue samples were placed in beakers of ddH₂O with ice. Tissue samples were then placed in tube containers in 4°C for 24 hours for cell lysis. Samples were then placed in decellularization solution containing 50 mM Tris, 0.15% (v/v) Triton x-100, 0.25% deoxycholic acid-sodium salt, 0.1% EDTA and 0.02% sodium azide and gently shaken at room temperature for three days. The decellularization solution was change with new decellularization solution at day three and samples were shaken for three more days. Samples were then rinsed two times with ddH₂O for 10 minutes at room temperature then two times with 70% ethanol for 10 minutes at room temperature and finally another two times with ddH₂O for 10 minutes at room temperature. Samples were then placed in 2X concentration of DNase/RNase solution containing DNase (3480 U/mg), RNase (97.1 U/mg), MgCl (5 mmol) and D-PBS at 37° C for 24 hours with shaking. Samples were then washed twice with ddH₂O and kept sterile in FBS at 4°C.

Pericardium samples were subjected to different mechanical strains (from ~5-40%) and different proteases in order to identify the effect of strain on collagen degradation by each protease. Recombinant human proMMP-1, proMMP-8, proMMP-9, and cathepsin K solutions were purchased commercially (Enzo Life Sciences, Farmingdale, New York) and proMMPs were activated based on provided activation protocols. ProMMP-1 was activated by 8μl of 0.5 mg/ml trypsin added to 100 μl of proMMP-1 and incubated for 30 minutes at 37°C. ProMMP-8 was activated by 2μl of 0.5 mg/ml trypsin added to 100 μl of proMMP-8 and incubated for 30 minutes at 37°C. ProMMP-9 was activated by 20μl of 0.5 mg/ml TCPK-trypsin added to 100 μl of proMMP-9 and incubated for 30 minutes at 37°C. After activation, 300 μl of ddH₂O was added to all protease solutions to reach the final volume of 400 μl and final concentration of 2.5 μg/ml.

Biaxial Stretching

Thin, square pericardium samples (7 mm × 7 mm × 0.2 mm) were biaxially stretched using a commercially available planar biaxial testing system (BioTester; CellScale, Waterloo, Canada) for 30 seconds to different levels of static strain in PBS at 37°C (Fig. 1A). Biaxial tensile displacement control was used in order to load all fibers within the tissue regardless of their orientation. A custom rig was designed to hold the protease solution to allow us to achieve the desired high protease concentration within a temperature-controlled bath. Samples were placed in the rig and kept under protease solution during the whole process (Fig. 1B). After the loading step, samples were held at constant extension under displacement control for one hour to allow for viscoelastic stress relaxation, after which the PBS bath was replaced with either a protease solution or protease-free control, and stresses were monitored for two additional hours (Fig. 1C). Matrix degradation led to decaying stress levels, and the stress decay rate was quantified as the decay constant (b) of an exponential fit to the stress versus time curve for every strain level for each protease as done previously (Eq. 1) [15]. A total of 28 samples were tested from 3 different pigs (5-7 samples for each protease experimental group). An F-stat, non-linear, quadratic regression analysis of degradation decay rate on strain level was performed in order to test the accuracy of our predictions.

Figure 1: Experimental Setup.

A) Decellularized pericardium stretched biaxially with CellScale and the custom rig used for holding high concentration protease solution within a temperature-controlled water bath. B) Decellularized pericardium sample held under biaxial strain by rigid rake attachments. C) Example stress-time curve of pericardium under constant extension subjected to protease or protease-free control.

$$F=F_0{e}^{-bt}$$ (Equation 1)

Degradation Byproduct Analysis

Degradation fragments were potentially being released into the solution in the form of either large degradation products or small degradation products while the degradation process was happening. The solutions, containing either a protease solution or protease-free control and respective degradation products, were preserved after the 3-hour relaxion and degradation process. The solutions were then analyzed using silver stain analysis to confirm degradation, to study the difference between protease treated samples and controls, and to compare the level of degradation between proteases.

Samples (i.e. solutions containing the degradation products) were prepared for mass spectrometry (MS) analysis first by dissolving the protein lysate in 8M Urea/1M NH4HCO3 buffer followed by 1h reduction at 37°C with 120 mM Tris(2-carboxyethyl)phosphine (TCEP). Protein alkylation was then performed with 160 mM iodoacetamide (IAA) for 30 min at room temperature with shaking. The samples were then diluted 8-fold with water to reduce the urea content, pH was adjusted to 8.0 and trypsin added at a ratio of 1:25. Digestion occurred at 37°C overnight with shaking. Trypsin was deactivated by acidifying samples to pH <3.0 using formic acid. Samples were desalted and purified using 1cc C18 cartridge columns and peptides were recovered in 0.1% formic acid. Samples were subjected to nano-LC-MS/MS analysis using an UltiMate 3000 RSLCnano system (ThermoFisher) coupled to a Q Exactive Plus Hybrid Quadrupole-Orbitrap mass spectrometer (ThermoFisher) via a nanoelectrospray ionization source. For each injection, 4μL (~1 μg) of sample was first trapped on an Acclaim PepMap 100 20 mm × 0.075 mm trapping column (ThermoFisher Cat# 164,535; 5 μL/min at 98/2 v/v water/acetonitrile with 0.1% formic acid). Analytical separation was then performed over a 95 min gradient (flow rate of 250 nL/min) of 4–25% acetonitrile using a 2 μm EASY-Spray PepMap RSLC C18 75 μm × 250 mm column (ThermoFisher Cat# ES802A) with a column temperature of 45°C. MS1 was performed at 70,000 resolution, with an automatic gain control (AGC) target of 3 × 10⁶ ions and a maximum injection time (IT) of 100 ms. MS2 spectra were collected by data-dependent acquisition of the top 15 most abundant precursor ions with a charge greater than 1 per MS1 scan, with dynamic exclusion enabled for 20 s. Precursor ions isolation window was 1.5 m/z and normalized collision energy was 27. MS2 scans were performed at 17,500 resolution, maximum IT of 50 ms, and AGC target of 1 × 10⁵ ions. Proteome Discoverer 2.5 was used for raw data analysis, with default search parameters including oxidation (15.995 Da on M) as a variable modification and carbamidomethyl (57.021 Da on C) as a fixed modification. Data were searched against the NCBI Sus scrofa reference proteome (Taxonomy ID 9823). Peptide-spectrum matches were filtered to a 1% false discovery rate (FDR) and grouped to unique peptides while maintaining a 1% FDR at the peptide level. Peptides were grouped to proteins using the rules of strict parsimony and proteins were filtered to 1% FDR.

Results

We stretched collagenous samples to different levels of static strain and treated them with MMP-1, MMP-8, MMP-9, Cathepsin K, or a protease-free buffer control. Across all strain levels tested, pericardium samples treated with protease-free buffer control produced very few degradation products evident by silver staining (Fig. 2A, 2C, 2D) and maintained steady stresses during the 2-hour treatment period, leading to near-zero stress decay constants (Fig. 3E). In contrast, samples treated with any of the four proteases exhibited increases in degradation, which was evident in both silver staining (Fig. 2) and stress decay rates (Fig. 3). All four protease types produced significantly more of the large degradation products quantified by silver staining (Fig. 2A and 2B), while MMP1 and MMP9 both produced significantly more small degradation products compared to all other groups (Fig. 2A and 2D, p<0.05 by ANOVA).

Figure 2. Silver stain analysis of degradation byproducts.

A and B) show silver stain blots of fluid samples collected after treating decellularized pericardial tissues with proteases or protease-free control. Quantification of silver stain bands indicated both large degradation products (C) and small degradation products (D) in the protease samples but very little degradation in the control sample. * indicates significant difference from all other groups by ANOVA (p<0.05).

Figure 3. Stress decay rates of loaded pericardium samples during protease degradation.

A) Degradation by cathepsin K was increased with increasing strains. B, C, and D) Degradation by MMP-1, MMP-8 and MMP-9 first decreased and then increased with the strain magnitude forming a V-shaped curve. E) No degradation detected in protease-free control samples. E) Degradation decay rates for all proteases. P-values obtained from quadratic regression tests.

Repeating the degradation tests across multiple strain levels demonstrated that the particular rate of degradation was affected by the sample strain (as previous studies have shown), but the particular influence of strain depended upon the type of protease. Specifically, degradation by cathepsin K was increased with increasing strains, and degradation by MMP-1, MMP-8 and MMP-9 first decreased and then increased with the strain magnitude forming a V-shaped curve (Fig. 3). The stress degradation decay rate reaches a minimum at strain level of approximately ε_min = 20% for MMP-1 and MMP-8 and approximately ε_min = 25% for MMP-9. Across nearly all strain levels, MMP-1 demonstrated the highest degree of stress decay while the other three proteases swapped degradation rate rankings based on the particular level of strain.

Mass spectrometry identified 936 different proteins whose byproducts were released into the substrate solution during the tests (Fig. 4A, full proteomic data available upon request). After removing proteins that were also detected in protease-free control samples, each protease isoform produced a unique degradome signature (Table 2) with many degradation products observed in only 1 protease type (Fig. 4B). MMP1 and MMP8 shared the most overlapping products with all four proteases, while 93% of cathepsin K products (13/14) and 71% of MMP9 products (54/76) were unique to those proteases. Most interestingly, the degraded protein signatures released into the solution also depended on strain magnitude (Fig. 4C, Table 3). While there was some overlap in degradation products spanning both low strain and high strain conditions, many proteins were only detected under low strain or high strain and not under both conditions.

Figure 4. Proteomic analysis of degradation byproducts.

A) Distinct degradome signatures for each protease type. B) Although the protease isoforms produced some overlapping degradation byproducts, most proteins detected in the degraded solution were unique to a single protease. C) Degradome profile depends on strain magnitude with different degradation byproducts detected under low vs. high strain environments.

Table 2.

List of secreted proteins released by proteolytic cleavage. Columns group secreted proteins released by individual enzyme; names in black font were uniquely identified in that enzyme secretome, while names in italicized blue represent proteins secreted by 2 or more enzymes. Proteins listed were detected only in the enzyme treated samples, i.e., proteins also detected in the respective control samples are not included in the list.

CatK	MMP1		MMP8		MMP9
BASEMENT MEMBRANE PROTEINS
LAMA4	LAMB1		LAMA1		LAMA1	LAMA2	LAMC3
ENZYMES & INHIBITORS
C1R	ADAM23	QSOX1	ADAMTSL4	SERPINB13	A2M	EPHB4	LIPH	RELN
ITIH2	F10	REN	ALDOC		ADAM28	F9	LYG1	ROCK1
ITIH3	FGFR4	SERPINB13	GALNT1		ARSG	FAP	NIT2	SERPINB13
	GDI2	SERPINB6	LYG1		CAT	FGFR4	PLA2G3	SERPINC1
	LYG1		NIT2		CES4A	GALNT1	PON3	SPINT1
	NIT2		PIP		CFI	GARS1	PZP
	PLG		PLG		CLCA3P	GDI2	QSOX1
FIBRILLAR COLLAGENS & FIBRIL ORGANIZATION-ASSOCIATED PROTEINS
COL1A2	CHADL		CHADL		CHADL
COL5A3	COL1A2		COL1A2		LOX
FMOD	GPC1
HORMONES & GROWTH FACTORS
ERFE					RETN
GDF9					SST
NETWORK-FORMING COLLAGENS
COL6A1	COL8A1				COL4A3
COL6A3
NON-STRUCTURAL EXTRACELLULAR PROTEINS
ICOS	ADNP	PDCD6IP	AHNAK	SEMA3F	A1BG	DEFA1	HUWE1	SDCBP
SPATA20	APOLD1	RBP3	APOB	SFTPD	ACTN4	DNAJC3	IGHG4	SEMA3F
WNT6	C1orf54	SEMA6D	APOLD1	SLC12A1	ACTR10	DOCK2	IL17B	SEMA6B
	CD274	SLC12A1	CD274	WNT10A	AFM	DST	IL4R	SLC12A1
	CYRIB	TUBB4B	CYRIB		ADM2	EMILIN2	KERA	SORL1
	DOCK2	UNC13D	HSP90AB1		CALML5	FERMT3	LRIG3	TSPEAR
	DST	WNT10A	IL17B		CD200R1	FIBIN	LRRTM2	UNC13D
	HSP90AB1		NCAM1		CD274	FRAS1	MAPT	VWF
	HSPA8		PDCD6IP		CFAP43	GSDMD	MVP	WNT10A
	IL17B		ODAD4		COL28A1	HSP90AA1	ODAD4
	LTBP1		SELP		CYRIB	HSP90AB1	PSMD7
OTHER STRUCTURAL PROTEINS
	ACAN	FBN2	ACAN	FBN2

Table 3.

List of secreted proteins released by proteolytic cleavage dependent on tissue strain level. Columns group secreted proteins released by individual enzyme from samples in the lower half of mechanical strains, samples in the upper half of mechanical strains, or samples in both. Proteins listed were detected only in the enzyme treated samples, i.e., proteins also detected in the respective control samples are not included in the list.

CatK	MMP1		MMP8			MMP9
DETECTED ONLY UNDER LOW TISSUE STRAIN
ICOS	ACAN		ACE	COL4A2	KLKB1	A2M	LIPH
LAMA4	ADNP		ADAMTSL4	DEFB106	LAMA1	ACTN4	MNDA
	F5		AFP	EFEMP1	LAMB1	ARHGAP9	MSR1
	FBN2		ALDOC	ENTPD5	LILRA3	CD47	MUC5B
	GPC1		AOC1	F2	LPA	CHI3L1	PSMD7
	PDCD6IP		APOC3	FAN1	LRRC7	CLCA3P	RELN
	PIGR		APOE	FBN2	NCAM1	CPAMD8	RETN
	RBP3		ARSA	FGFR4	PDCD6IP	CPB2	SEMA6B
	REN		C1QA	GPC1	PIP	FGF2	SEMA7A
	VWF		CAP1	HAUS3	PRTN3	FNDC1
	WNT10A		CD274	HSP90AB1	QSOX1	FRAS1
			CFHR2	HYOU1	SEMA	FREM2
			CHADL	IGKC	SLC12A1	HSPA8
			COL1A2	IGKV2-40	TNXA	IGHG4
			COL3A1	IL17B	WNT10A	LAMA1
DETECTED ONLY UNDER HIGH TISSUE STRAIN
C1R	ADAM23	HDLBP	GALNT1			ACTR10	LYG1	QSOX2
COL6A1	AHNAK	HSP90AB1	SFTPD			BRPF3	MERTK	ROCK1
COL6A3	C1orf54	HYOU1	SST			CCT8	NPHP3	SERPINC1
DHRS7C	COL16A1	LAMA1				CKB	OLFML2A	SETX
GLG1	COL19A1	LAMA2				COL1A2	PLA2G3	TGFB1
	DNAH5	LTBP3				COL5A1	PLG	TSPEAR
	DOCK2	RAB11A				CORIN	QSOX1
	F10	SEMA6D				DNAJC3	GARS1
	FGFR4	SERPINB6				DOCK2	HSP90AA1
	FLNA	SERPINC1				DST	IL4R
	FRAS1	THBS4				F9	JHY
	GDI2	UNC13D				FBN3	KERA
	HAUS3					FNDC5	LAMC3
DETECTED UNDER BOTH LOW AND HIGH TISSUE STRAIN
COL1A2	APOLD1	SLC12A1	ACAN			A1BG	DEFA1	MAPT
FMOD	CD274	TUBB4B	AHNAK			ADAM28	EMILIN2	MVP
GDF9	CHADL		APOB			ADM2	EPHB4	NIT2
WNT6	COL1A2		APOLD1			AFM	FAP	ODAD4
	COL8A1		CYRIB			ARSG	FERMT3	PON3
	CYRIB		LYG1			CALML5	FGFR4	PZP
	DST		NIT2			CAT	FIBIN	SDCBP
	HSPA8		ODAD4			CD200R1	GALNT1	SEMA3F
	IL17B		PLG			CD274	GDI2	SERPINB13
	LAMB1		SERPINB13			CES4A	GSDMD	SLC12A1
	LTBP1					CFAP43	HSP90AB1	SORL1
	LYG1					CFI	HUWE1	SPINT1
	NIT2					CHADL	IL17B	SST
	PLG					COL28A1	LAMA2	UNC13D
	QSOX1					COL4A3	LOX	VWF
	SERPINB13					CYRIB	LRIG3	WNT10A

Discussion and Conclusions

In the current study, we aimed to study the effect of strain on collagenous tissue degradation by different proteases. Treating decellularized pericardium samples with proteases under different levels of biaxial deformation, our results demonstrate that the degree of mechanical strain can significantly alter protease-mediated stress decay rates of tissue, but the particular effect depends on the particular type of protease. We showed that cathepsin K-mediated degradation was enhanced with increasing strain, while MMP-1, MMP-8, and MMP-9-mediated degradation rates were first decreased and then increased by strain, forming a V-shaped curve with a minimum at strain level of approximately ε_min = 20 – 25%. These findings are consistent with past studies by Huang (1977) and Ghazanfari (2016), who found that collagen degradation by bacterial collagenase was influenced by strain and discovered bacterial collagenase-mediated degradation rate also formed a V-shaped curve [15,26]. Huang found that the lowest degradation rate in a reconstituted collagen tape occurred at 4% strain. However, Ghazanfari reported the minimum degradation value was around 20% strain in pericardial tissue. The collagen in Huang’s reconstituted construct was comprised of relatively straight fibers compared to the pericardium, and the shift in the curve might be a result of the larger strain needed to uncrimp the fibers prior to stretching their individual fiber lengths. Given that cathepsin K is capable of cleaving collagens both intrahelically and at the telopeptide regions [33], a potential explanation for why cathepsin K shows a monotonically increasing response to mechanical strain compared to the MMPs may be that increased strain continues to enhance the intrahelical exposure of CatK (but not MMP) binding. Overall, our results highlight that the relative contributions of different proteases to tissue turnover depend not only on protease concentrations but also on the local level of tissue’s mechanical conditions. More pointedly, high protease isoform expression does not necessarily mean that a particular protease dominates degradation – it also depends on localized tissue mechanics, which are often dynamically and spatially modulated during complex remodeling contexts like wound healing or morphogenesis.

Silver stain analysis confirmed that protein degradation in the samples accompanied the mechanical stress decay observed. It also showed that MMP-1 and especially MMP-9 are able to degrade the matrix into smaller fragments, which is expected since MMP-9 is a well-known gelatinase [9]. Proteomic analysis revealed distinctly different degradome signatures in the protein degradation profiles generated by each protease isoform as well as the profiles generated between low strain vs. high strain contexts. This result should provide caution when simplifying proteases into categories like “collagenase” vs. “gelatinase” vs. “elastase”, etc., since our findings suggest that a single protease could have different ECM targets depending upon the mechanical context. It is also important to mention that changes in ECM degradation products can be critical to regulating cellular response during tissue remodeling. Degradation of ECM proteins generates various byproducts that can interact with cell surface receptors and alter inflammatory, fibrogenic, angiogenic, and reparative cascades [30-31]. Thus, our results suggest a potentially new form of mechanobiology wherein tissue strain can modulate the degradome to alter local cell signaling events via interactions with degradation fragment.

Our approach comes with a few notable limitations. First, we have only tested one tissue type herein: porcine pericardium. This tissue was selected as a readily available tissue primarily composed of fibrillar collagens, therefore providing ample material to test hypotheses related to mechano-sensitive degradation. These samples are thin which eases the decellularization and loading processes, and they are large enough to divide each sample across multiple experimental conditions. Further, they are relatively isotropic compared to other collagenous tissues like skin and tendon, whose highly aligned structures can confound mechanical loading distributions.

As a second limitation, we recognize that the pericardium is not only comprised of fibrillar collagen, but other ECM proteins as well, and we recognize that many important proteases exist apart from cathepsin K and MMP-1, 8, and 9. Expanding this study across more proteases while also testing protease inhibitors and protease cocktails can expand our knowledge of the effect of tensile loading on collagenous degradation. Indeed, previous work by Parks et al [34] has demonstrated complex interactions of multiple proteases in combinations when treating sequentially vs. simultaneously.

A third limitation of this study is the use of bottom-up mass spectrometry, which results in many trypsin-generated peptides and makes it impossible to accurately assess whether the detected ECM products represent decayed protein fragments or intact proteins liberated during ECM degradation, which is likely to affect downstream cellular feedback signaling that may arise from the degradation productions.

In spite of these limitations, our collective results highlight that mechanical context is an important regulator of protease function. Therapeutic strategies for modulating tissue remodeling might benefit from tailoring such strategies to the particular mechanical context in vivo, which can affect not only degradation rates, but also the mechanobiological signaling feedback from degradation byproducts.

Highlights.

• Mechanical strain can both increase and decrease protease-mediate degradation of collagenous tissues, depending on the type of protease and on the strain regime.

• Cathepsin K-mediated degradation of collagenous tissue was enhanced with increasing strain.

• MMP1-, MMP8-, and MMP9-mediated degradation were first decreased and then increased by strain.

• Mechanical forces alter the degradation products that can provide important signaling feedback functions during tissue remodeling.