Hydrogel biomaterials that stiffen and soften on demand reveal that skeletal muscle stem cells harbor a mechanical memory

Significance

Muscle stem cells (MuSCs) are required for repairing muscle tissue throughout life, but their ability to regenerate damaged tissue declines with aging and in muscle wasting disorders. MuSCs are known to be sensitive to the stiffness of tissue culture substrates, and muscle tissue stiffness is also known to increase during aging and disease. We postulated that stem cells acquire a mechanical memory on stiff substrates and tested this hypothesis using dynamic hydrogels that soften or stiffen in response to light. We demonstrate that the mechanical memory MuSCs develop is rapidly acquired and not easily reversed by a change in stiffness. We identify potential signaling pathways that can be targeted to block this memory to enhance MuSC function.

Abstract

Muscle stem cells (MuSCs) are specialized cells that reside in adult skeletal muscle poised to repair muscle tissue. The ability of MuSCs to regenerate damaged tissues declines markedly with aging and in diseases such as Duchenne muscular dystrophy, but the underlying causes of MuSC dysfunction remain poorly understood. Both aging and disease result in dramatic increases in the stiffness of the muscle tissue microenvironment from fibrosis. MuSCs are known to lose their regenerative potential if cultured on stiff plastic substrates. We sought to determine whether MuSCs harbor a memory of their past microenvironment and if it can be overcome. We tested MuSCs in situ using dynamic hydrogel biomaterials that soften or stiffen on demand in response to light and found that freshly isolated MuSCs develop a persistent memory of substrate stiffness characterized by loss of proliferative progenitors within the first three days of culture on stiff substrates. MuSCs cultured on soft hydrogels had altered cytoskeletal organization and activity of Rho and Rac guanosine triphosphate hydrolase (GTPase) and Yes-associated protein mechanotransduction pathways compared to those on stiff hydrogels. Pharmacologic inhibition identified RhoA activation as responsible for the mechanical memory phenotype, and single-cell RNA sequencing revealed a molecular signature of the mechanical memory. These studies highlight that microenvironmental stiffness regulates MuSC fate and leads to MuSC dysfunction that is not readily reversed by changing stiffness. Our results suggest that stiffness can be circumvented by targeting downstream signaling pathways to overcome stem cell dysfunction in aged and disease states with aberrant fibrotic tissue mechanics.

Skeletal muscle stem cells (MuSCs) play a crucial role in the maintenance and repair of contractile skeletal muscle tissue throughout life (1–3). With aging and in heritable muscle wasting disorders such as Duchenne muscular dystrophy, MuSC function becomes impaired and muscle regenerative function is diminished (4, 5). MuSCs occupy a specialized niche within the skeletal muscle, sandwiched between mature muscle fibers and the basal lamina extracellular matrix (ECM) that surrounds the fibers (1, 2). Signals from the stem cell niche are potent regulators of cell fate, and in response to injury orchestrate the complex processes of MuSC activation, differentiation, and return to quiescence (2). In particular, MuSCs are exquisitely sensitive to cues from the ECM (6). Macromolecules in the ECM including laminin (7), collagen V (8), collagen VI (9), fibronectin (10), and hyaluronic acid (11), have been shown to potently regulate myogenesis.

Beyond biochemical signals provided by these ECM molecules, the ECM also provides biophysical cues, including the elasticity, or stiffness, of the cellular microenvironment, that can influence cell fate (12). MuSCs are well known to be mechanosensitive, altering their phenotype in response to the stiffness of the substrates on which they are cultured. We have previously demonstrated that freshly isolated MuSCs exhibit the greatest expansion and engraftment potential when cultured on hydrogel substrates with elastic moduli (E) mimicking the stiffness of healthy skeletal muscle (~12 kPa) (13–15). If MuSCs are cultured on substrates that are too stiff, they lose their regenerative potential. Changes in ECM composition and crosslinking that occur with aging and disease can substantially increase the stiffness of the ECM (16–19). As increased microenvironmental stiffness in aged tissues has been correlated with altered MuSC function (16), understanding how MuSCs respond to changes in matrix stiffness may be crucial to identify therapeutic approaches to enhance MuSC-mediated regeneration of aged and diseased tissues.

In vitro studies of myogenesis often employ cultured myogenic progenitor cells, or myoblasts, derived by expanding MuSCs on rigid collagen-coated tissue culture plastic. Whereas freshly isolated MuSCs expand significantly more on softer substrates whose stiffness is similar to healthy (E ~ 12 kPa) versus aged (E ~ 35 to 40 kPa) muscle (15), myoblasts show a trend toward enhanced proliferation on stiffer substrates (20, 21). In contrast to freshly isolated MuSCs, myoblasts have already activated and are in a proliferative state when these mechanotransduction experiments are performed. Freshly isolated MuSCs are initially quiescent, entering the cell cycle during the first 2 d in culture (22). We hypothesized that MuSCs may exhibit a “mechanical memory” of their biophysical microenvironment that is acquired during activation and dictates subsequent cellular fate decisions.

The concept of a mechanical memory was first applied to mesenchymal stromal cells cultured on hydrogel substrates with controlled stiffness and exemplified as stiffness-driven cellular differentiation decisions that could not be overridden by later application of antagonistic soluble signaling factors (23). By either transferring cells between substrates of different stiffness or using photoresponsive materials to decrease substrate stiffness on demand, persistent changes in Yes-associated protein (YAP) activity (24), microRNA expression (25), and chromatin organization (26) were implicated as potential mechanisms underlying this mechanical memory. Despite the known mechanosensitivity of MuSCs, no such mechanical memory has been previously reported for these critical regulators of muscle repair. We therefore employed hydrogel platforms whose stiffness can be either decreased or increased in response to light to determine whether the duration of exposure to either stiff or soft microenvironments during MuSC activation could induce a mechanical memory. Notably, we used highly selective biorthogonal chemistries that do not cross-react with biological functional groups, including the first hydrogel-based application of phototriggered strain-promoted azide-alkyne cycloaddition (SPAAC), in contrast to prior approaches relying on free-radical based methods (27–29). While short duration (~24 h) culture on stiff substrates did not detrimentally impact expansion of myogenic progenitors, longer culture (~72 h) on stiff substrates decreased MuSC expansion and biased cells either toward premature commitment to differentiation or development of a dysfunctional quiescent-like state. Further, 3 d of culture on soft substrates, or inhibition of RhoA-mediated mechanotransduction for 3 d, was sufficient to preserve the expansion capacity of the myogenic progenitor pool even with subsequent culture on stiff substrates. Through single-cell RNA sequencing, our findings identify potential therapeutic targets to boost MuSC regenerative function in aged and diseased tissues. We anticipate that our dynamic hydrogel systems will be broadly useful to elucidate mechanisms of aberrant mechanosensing in other cell types from the diverse array of tissues susceptible to fibrosis.

Results

MuSCs Acquire a Mechanical Memory during Activation.

To confirm that the mechanosensitive proliferation of myogenic progenitors depends on whether the cells begin as quiescent stem cells or as activated progeny prior to exposure to different microenvironments, MuSCs were isolated from the hindlimb muscles of young (~2 mo) mice, following our previously published procedures (13, 15, 30). Quiescent MuSCs were plated directly on laminin-functionalized poly(ethylene glycol) (PEG) hydrogel substrates with relatively compliant stiffness similar to healthy muscle (E ~ 12 kPa, “soft”) or rigid fibrotic muscle (E ~ 40 kPa, “stiff”) and cultured for 7 d (Fig. 1A). This 2D culture platform was designed to mimic the microenvironment of regenerating muscle (SI Appendix, Supporting Note 1). Activated MuSC progeny, or myoblasts, were generated by plating the same freshly isolated MuSCs on collagen-coated plastic dishes and expanding the cells for one week prior to trypsinizing the cells and replating on the same hydrogel conditions as the quiescent MuSCs (Fig. 1B) (21). Consistent with our prior results (13–15), after 7 d of hydrogel culture, initially quiescent MuSCs exhibited ~twofold greater expansion on soft compared to stiff hydrogels (Fig. 1 C and D). Furthermore, consistent with other prior work (20, 21), activated myoblasts plated on stiff hydrogels showed moderately more expansion than those cultured on soft hydrogels (Fig. 1 E and F). Thus, quiescent MuSCs respond differently to microenvironmental stiffness than their activated progeny.

Fig. 1.

MuSCs exhibit a mechanical memory during activation. Schematics illustrating the culture protocols for (A) freshly isolated MuSCs and (B) expanded myoblasts on soft versus stiff hydrogel substrates. (C) Immunofluorescence analysis and (D) quantification of cell number per well after 7 d of culture revealed increased cell expansion on soft versus stiff substrates for freshly isolated MuSCs. In contrast, (E) immunofluorescence analysis and (F) quantification of cell number per well after 7 d of culture for myoblasts revealed a trend toward higher cell expansion on stiff compared to soft substrates. Schematics depicting (G) the culture protocol and (H) experimental and control conditions for culturing freshly isolated MuSCs on hydrogels with dynamic softening to determine the temporal window for mechanical memory formation. (I) Immunofluorescence analysis revealed that softening substrates during the process of MuSC activation prevents mechanical memory formation, with (J) increased proliferative fractions and (K) decreased committed fractions remaining on softened substrates compared to stiff controls. Error bars represent mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001. In J and K, statistical significance is determined relative to the static soft control condition. In K, the color of stars denotes which myogenic population is being compared.

This differential response to substrate stiffness between quiescent MuSCs and actively proliferating myoblasts suggested that MuSCs acquire a mechanical memory during activation that biases subsequent cell fate decisions. The most rigorous test to prove that a mechanical memory has formed is to culture freshly isolated cells on stiff substrates for defined periods of time prior to reducing the stiffness of the substrates and observing persistent changes in cell behavior (24). The duration of time in a stiff microenvironment required to form a mechanical memory can thus be determined without otherwise perturbing the system (24). Therefore, we employed a hydrogel system that was engineered to have an initially high stiffness that could be decreased on demand (15). Freshly isolated MuSCs were plated on hydrogels containing light-sensitive ortho-nitrobenzyl ester (oNB) crosslinks that could be selectively cleaved upon exposure to 365 nm light (15, 24). An advantage of our implementation of the oNB chemistry is that the hydrogel molecular architecture was engineered such that a defined fraction of the crosslinks was responsive to light, thus providing precise control over the extent of softening (15).

The hydrogels initially presented stiff (E ~ 40 kPa) mechanical cues. After 1, 2, or 3 d of culture post-MuSC isolation and seeding, the hydrogels were softened by exposure to light (E ~ 12 kPa), and the cells were maintained in culture for a total of 7 d (Fig. 1 G and H). The softened gel conditions were compared to cells cultured on persistently soft or stiff hydrogels as above, or persistently stiff gels composed of the oNB materials that were not exposed to 365 nm light. Additional controls to confirm that ultraviolet (UV) light exposure in the absence of mechanical changes did not alter cellular phenotype were also performed using the same light dose on materials that did not contain the oNB crosslinks (SI Appendix, Fig. S1). Further, atomic force microscopy (AFM) analysis of the hydrogels pre- and postlight exposure confirmed spatially homogeneous hydrogel stiffness and no substantial changes in surface morphology due to potential gel swelling after softening (SI Appendix, Fig. S2). Confocal microscopy analysis of fluorescently labeled laminin also revealed consistent presentation of laminin across the hydrogel conditions tested (SI Appendix, Fig. S3). Taken together, these results indicate that observed differences in MuSC behavior can be attributed to changes in hydrogel stiffness as opposed to off-target effects of hydrogel softening.

To track myogenic progression, the MuSCs used in these experiments were isolated from our transgenic Sun1-green fluorescent protein (GFP) reporter mouse strain in which myogenin activity labels the nuclear envelopes of the cells with GFP (15). To determine the fraction of proliferative cells remaining at day 7, the nontoxic thymidine analog F-ara-EdU (31) was added to the medium for 6 h on day 7 prior to fixing the cells. In addition to staining for EdU incorporation, the fixed cells were stained for the myogenic transcription factors Pax7 and MyoD and classified according to their expression of myogenic factors. Stem-like cells were those that were Pax7+ MyoD+ GFP-, progenitor cells were those that were Pax7- MyoD+ GFP-, and committed cells were those that were Pax7- MyoD+ GFP+.

Culturing freshly isolated MuSCs on soft gels for the full 7 d resulted in a larger population of proliferative cells at day 7 compared to persistent culture on both stiff control conditions, as expected from our initial experiments (Fig. 1 I and J). Strikingly, culture on an initially stiff substrate for only 1 d prior to softening resulted in an equivalent proliferative fraction on day 7 as persistent soft culture, whereas culture on an initially stiff substrate for 3 d prior to softening resulted in an equivalent proliferative fraction on day 7 as persistent stiff culture (Fig. 1 I and J). Thus, it appeared that a mechanical memory is acquired between days 1 and 3 during the activation process that dictates the proliferative capacity of MuSCs. We further quantified the myogenic state of the cells and observed a similar indication of a mechanical memory. Persistent culture on stiff substrates resulted in increased commitment to differentiation on day 7, as indicated by myogenin reporter activity, while persistent culture on soft substrates maintained a larger progenitor pool, indicated by high MyoD and low Pax7 staining (Fig. 1 I and K). Softening the substrates during the first few days of culture significantly reduced the fraction of the population that had committed to differentiation, instead favoring proregenerative stem and progenitor populations (Fig. 1 I and K). Exposure to equivalent doses of UV light did not alter the proliferative fraction or progenitor population observed on day 7 (SI Appendix, Fig. S1), indicating that the observed changes in cell phenotype are due to changes in substrate mechanical properties.

Substrate Stiffness Regulates MuSC Cytoskeletal Architecture and Mechanosensitive Signaling Pathways.

To determine the mechanistic origins of the mechanical memory, freshly isolated MuSCs from wild-type mice were seeded onto soft and stiff hydrogel substrates and cultured for either 1 d or 3 d to capture the early stages of activation before the mechanical memory is formed (day 1) and the end of activation, by which point the memory is likely present (day 3). We first considered whether differences in substrate stiffness change the time course of the activation process (Fig. 2 A and B). Staining for Pax7 and MyoD transcription factors revealed that nearly all cells on both substrates at both time points express Pax7 and MyoD (Fig. 2 A, C, and D and SI Appendix, Fig. S4 A and B). In addition to changes in myogenic transcription factor expression, activation is characterized by cell cycle entry. Cells on both substrates at day 1 were almost entirely negative for the cell cycle marker Ki67, while at day 3, the majority of cells on both substrates expressed Ki67 (Fig. 2 A and E and SI Appendix, Fig. S4C). Increased activity of p38 mitogen-activated protein kinase (MAPK) is another hallmark of activation (14, 32). Staining for phosphorylated p38 followed similar trends as Ki67, with relatively few cells staining positive at day 1 and most staining positive at day 3, with no substantial difference between substrate conditions (Fig. 2 B and F and SI Appendix, Fig. S4D). Taken together, these results suggest that substrate stiffness does not regulate the kinetics of MuSC activation.

Fig. 2.

Microenvironmental stiffness does not alter MuSC activation timing but does regulate MuSC morphology. (A and B) Immunofluorescence analysis of activation markers and morphological parameters for freshly isolated MuSCs cultured on soft versus stiff hydrogels for 1 or 3 d postisolation. The fraction of cells expressing the myogenic transcription factors (C) Pax7 and (D) MyoD and the activation markers (E) Ki67 and (F) phospho-p38 MAP kinase (P-p38) does not vary as a function of substrate stiffness. At both days 1 and 3 postisolation, cells cultured on soft versus stiff substrates exhibit (G) larger nuclear area, (H) more elongated nuclei, (I) larger cell spread area, and (J) more elongated cellular morphologies. Error bars represent median ± 95% CI. ****P < 0.0001.

While the timing of activation did not differ with substrate stiffness, the morphology of the cells and their nuclei exhibited marked differences as a function of stiffness. Cells cultured on soft substrates were significantly larger with larger nuclei that were much more elongated than cells cultured on stiff substrates at both day 1 and day 3 (Fig. 2 G–J). Cells on soft substrates exhibited distinctive bipolar morphologies, with two long processes extending from opposing ends of the cell (Fig. 2B), consistent with recent studies demonstrating the importance of cellular protrusions as indicators of MuSC state (33, 34). In contrast, cells on stiff substrates exhibited more rounded morphologies, with several short processes extending from various locations on the cells (Fig. 2B). Consistent with these differences in cell shape, cells cultured on stiff substrates exhibited altered organization of filamentous actin, with increased actin network density per cell area compared to the cells cultured on soft substrates (Fig. 3 A and B). These results highlight that cytoskeletal architecture is highly dependent on the underlying substrate stiffness.

Fig. 3.

Microenvironmental stiffness regulates cytoskeletal network architecture and activation of mechanosensitive signaling pathways. Immunofluorescence analysis of (A, B) filamentous actin, (C, D) YAP localization, (E, F) RhoA activity, and (G, H) Rac1 activity revealed increased actin network density, increased nuclear YAP localization, increased RhoA activity, and increased Rac1 activity, respectively, in cells cultured on stiff compared to soft substrates at both days 1 and 3 postisolation. Error bars represent median ± 95% CI. ****P < 0.0001. Data in B, F, and H are presented as mean intensity (integrated intensity normalized to cell area) to account for differences in cell spreading.

To probe which mechanosensitive pathways are implicated in the observed differences in cell morphology and mechanical memory acquisition, we assayed for the activity of three pathways known to play a role in myogenic progression: Rho and Rac GTPases and YAP (21, 33, 35, 36). Both at early (day 1) and later (day 3) time points during activation, all three pathways showed elevated activity on stiff compared to soft substrates. The nuclear localization of YAP and the level of active (GTP-bound) RhoA were significantly increased on stiff versus soft substrates (Fig. 3 C–F), consistent with a more contractile phenotype in cells cultured on stiff gels. Rac1 activity, which is correlated with protrusive cellular forces, was also greater on stiffer substrates, but to a more modest degree than RhoA (Fig. 3 G and H). Thus, YAP, Rho, and Rac signaling are sensitive to the underlying substrate stiffness and may contribute to the observed mechanical memory phenotype.

Acquisition of a Mechanical Memory Is Time Dependent.

To determine whether the formation of the mechanical memory can be prevented, we engineered hydrogels that can stiffen on demand in response to specific wavelengths of light. To achieve this, we developed a technology for generating phototriggered changes in hydrogel stiffness that was compatible with MuSC culture. To date, the most reliable method for in situ stiffening of hydrogel substrates is phototriggered radical polymerization (27, 28). However, this chemistry requires the generation of free radicals that can detrimentally impact MuSCs. Other prior light-responsive chemistries rely on changes in molecular conformation and do not generate potentially toxic side products (37–39). However, these chemistries result in only modest changes in hydrogel stiffness, whereas aging and disease-associated fibrosis can result in multifold changes in stiffness (16–19).

To address this technological gap, we developed hydrogel substrates with on-demand stiffening. The gels were prepared by functionalizing 8-arm PEG macromers both with bicyclo[6.1.0] non-4-yne (BCN) groups that will spontaneously react with azides on a second PEG macromer to form a hydrogel network and with photocaged oxa-dibenzocyclooctyne (pODIBO) groups that can be activated by exposure to 365 nm light. The photocaged ODIBO molecules contain cyclopropenones that decompose under irradiation to yield strained cyclooctynes that can subsequently react with excess azides in the hydrogel network to stiffen the gels (SI Appendix, Fig. S5 A and B). Our approach leverages a phototriggered bioorthogonal SPAAC reaction originally developed by Popik and colleagues (40, 41). This reaction enabled facile integration into our PEG gels containing SPAAC-based crosslinks, while still responding to the same 365 nm light source used for our softening materials based on oNB crosslinks. An unexpected benefit of the photo-SPAAC chemistry is that the pODIBO groups are very stable under cell culture conditions, which is not true of a second set of hydrogels we developed based on photooxidation-triggered tetrazine ligation (42, 43) (SI Appendix, Fig. S6 A and B) and was essential for these experiments that culture cells in high serum media for three days prior to stiffening. Based on spectroscopic analysis, PEGylated pODIBO in solution did not appreciably degrade after 3 wk in cell culture medium with 10% fetal bovine serum (FBS) at 37 °C (SI Appendix, Fig. S5 C and D), while retaining rapid photoactivation kinetics after the 3 wk period (SI Appendix, Fig. S5 E and F). This contrasts with on-demand stiffening gels that use photooxidation-triggered tetrazine ligation. While we successfully developed cytocompatible strategies to trigger tetrazine ligation and corresponding gel stiffening using far red (680 nm) light (SI Appendix, Fig. S6 C–G), the poor serum stability of the tetrazine precursor molecules precluded the use of this system in MuSC mechanical memory experiments (SI Appendix, Fig. S6H).

To validate that hydrogels containing pODIBO groups could still stiffen after being maintained under cell culture conditions, we incubated pODIBO hydrogels under identical conditions used in MuSC culture with 15% FBS for 1 wk and compared the stiffness of the gels pre- and postlight exposure. The hydrogels reliably stiffened to a similar extent, even after 1 wk in cell culture conditions (SI Appendix, Fig. S5G). The pre- and postlight exposure hydrogels exhibited spatially homogeneous stiffness as measured by AFM (SI Appendix, Fig. S7 A–D), consistent with the secondary crosslinking reaction proceeding by the intended step-growth mechanism, as opposed to chain-growth mechanisms employed by free-radical-mediated approaches that result in more heterogeneous gels (44–46). While a moderate decrease in gel swelling was observed poststiffening, AFM analysis showed no substantial surface deformation of the gels (SI Appendix, Fig. S7 E and F), consistent with limited changes in gel shape expected for constrained hydrogels similar to those used in this study (47). Confocal microscopy analysis of fluorescently labeled laminin confirmed uniform presentation of laminin pre- and poststiffening (SI Appendix, Fig. S7 G and H). Thus, on-demand stiffening hydrogels based on photo-SPAAC chemistry could be used to determine whether subsequent stiffening of the hydrogels alters MuSC phenotype similar to initial culture on stiff substrates.

We tested whether formation of the detrimental mechanical memory of a stiff microenvironment could be prevented by initially culturing the cells on soft substrates. Freshly isolated MuSCs were seeded onto pODIBO-containing gels and control soft and stiff gels to determine whether time on a soft hydrogel would create a favorable mechanical memory that persisted even once hydrogel stiffness was increased. Myogenin Sun1-GFP reporter MuSCs were cultured on the gels for 3 d of activation, after which stiffness was increased (Fig. 4 A and B). As an additional control, another set of pODIBO gels were maintained without light exposure to present persistently soft conditions in the same material system. After 7 d of culture, all of the conditions that started out soft exhibited a significantly higher proliferative (EdU+) fraction of cells compared to persistent culture on stiff (Fig. 4 C and D). Furthermore, early exposure to soft microenvironments maintained the stem and progenitor pools, with only persistent culture on stiff hydrogels resulting in increased commitment to differentiation (Fig. 4 C and E). Cells cultured on soft substrates that were stiffened on day 3 retained a proliferative progenitor phenotype, similar to cells that were persistently cultured on soft substrates. Thus, the favorable mechanical memory formed during the first three days of activation on soft hydrogels is resilient to subsequent stiffening of the microenvironment during myogenic progenitor expansion.

Fig. 4.

The mechanical memory is established within the first 3 d of activation on soft gels and can be blocked by interfering with RhoA-mediated mechanotransduction. Schematics depicting (A) the culture protocol and (B) experimental and control conditions for culturing freshly isolated MuSCs on hydrogels with dynamic stiffening. (C) Immunofluorescence analysis revealed that stiffening hydrogels after 3 d of culture under soft conditions maintained similar (D) proliferative fractions and (E) committed fractions of cells present at day 7 compared to persistent culture on soft substrates. (F) Schematic depicting the experimental and control conditions for culturing freshly isolated MuSCs on stiff hydrogels and inhibiting mechanotransduction. (G) Transient Rac1 and YAP inhibition suppress cell expansion, while transient RhoA inhibition does not. (H) Immunofluorescence analysis revealed that only inhibition of RhoA (I) increased the proliferative fraction and (J) decreased the committed fraction of cells present at day 7 compared to vehicle controls on stiff substrates. Error bars represent mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001. In D and E, statistical significance is determined relative to the static stiff control condition. In I and J, statistical significance is determined relative to the vehicle-treated control condition. In E and J, the color of stars denotes which myogenic population is being compared.

Mechanical Memory Can Be Blocked by Targeting Downstream Signaling.

In a second experiment, we sought to block formation of the mechanical memory by treating MuSCs on stiff substrates with small molecule inhibitors of the mechanosensitive pathways we identified above. Freshly isolated myogenin Sun1-GFP reporter MuSCs were seeded onto stiff hydrogel substrates and were treated with inhibitors of RhoA (rhosin), Rac1 (NSC23766), and YAP (verteporfin) for 3 d, after which the inhibitors were washed out for the remaining 4 d of culture (Fig. 4F). Interfering with Rac1 and YAP signaling markedly reduced overall cell expansion (Fig. 4G), suggesting that Rac1 and YAP were not driving formation of a mechanical memory on stiff substrates that also limits MuSC expansion. Rather, Rac1 and YAP appear to be required for MuSC expansion. In contrast, RhoA inhibition did not impair MuSC expansion (Fig. 4G). Of the cells present at day 7, only initial RhoA inhibition resulted in an increase in the proliferative (EdU+) population of cells compared to vehicle-treated stiff controls (Fig. 4 H and I). Furthermore, RhoA inhibition decreased the fraction of cells that had committed to differentiation compared to vehicle-treated stiff controls, while Rac1 inhibition increased commitment and dramatically decreased the progenitor pool, and YAP inhibition did not alter commitment (Fig. 4 H and J). Taken together, these results suggest that RhoA-mediated signaling is responsible for induction of the mechanical memory. Further, inhibiting RhoA during MuSC activation blocks the formation of a detrimental mechanical memory.

The Mechanical Memory Manifests as a Decrease in Myogenic Progenitor Cells by Day 7.

To better understand the consequences of the mechanical memory on myogenic progression, we characterized the populations of cells present after 7 d of culture on soft versus stiff substrates. By immunofluorescence staining of myogenin Sun1-GFP reporter cells, we observed an approximately twofold increase in the fraction of proliferative (EdU+) cells at day 7 on soft compared to stiff substrates (Fig. 5 A and B). Furthermore, the myogenic state of the cells was significantly altered on stiff substrates. Compared to soft substrates, greater fractions of cells cultured on stiff substrates were classified as either committed cells or nonproliferative stem-like cells, with a dramatic reduction in the progenitor (Pax7^low, MyoD^high) population (Fig. 5 A and C). These results suggest that the formation of a mechanical memory for stiff substrates results in a loss of progenitor cells.

Fig. 5.

The mechanical memory manifests as a loss of myogenic progenitors at day 7. (A) Immunofluorescence analysis revealed (B) decreased proliferative fraction and (C) increased progenitor fraction and decreased stem and committed fractions of cells after 7 d of culture on soft versus stiff substrates. (D) Single-cell RNA sequencing analysis shows reduced presence of cells in the progenitor cluster when cultured for 7 d on stiff compared to soft substrates. (E) Heatmap of myogenic genes and corresponding myogenic state assignments for a representative subset of cells analyzed in D similarly reveals that the progenitor population consists mostly of cells cultured on soft substrates. (F) Single-cell protein immunofluorescence analysis confirms reduced presence of cells in the progenitor cluster when cultured for 7 d on stiff compared to soft substrates. (G) Heatmap of immunofluorescence markers and corresponding myogenic state assignments for a representative subset of cells analyzed in F similarly confirms that the progenitor population consists mostly of cells cultured on soft substrates. (H) Comparing a relevant subset of differentially expressed genes between the stem (enriched on stiff substrates) and the progenitor (enriched on soft substrates) populations reveals differences in expression of genes associated with expanding myogenic cells, quiescent MuSCs, and Rho GTPase signaling. Error bars represent mean ± SD. *P < 0.05, **P < 0.01, ****P < 0.0001. In C, the color of stars denotes which myogenic population is being compared.

To further explore the heterogeneity of cells present after 7 d of culture for freshly isolated wild type MuSCs, we employed single-cell RNA sequencing via a SPLiTseq barcoding approach to mitigate potential confounding batch effects (48). The sequenced cells exhibited the diversity of myogenic cells expected 7 d postisolation, including activated progenitor cells, committed progenitors, differentiating cells, and stem-like cells exhibiting signatures of a return to quiescence (Fig. 5 D and E and SI Appendix, Fig. S8A). When overlaying the stiffness of the cell culture substrates onto our clusters of MuSC progeny, both soft and stiff cultured cells appear in clusters corresponding to cycling progenitor cells and committed and differentiated cells. However, a striking separation was observed for the stem and progenitor cell clusters. Cells cultured on stiff hydrogels predominantly comprised the stem cluster, whereas cells cultured on soft hydrogels predominantly comprised the progenitor cluster (Fig. 5 D and E). This is consistent with our population-level analysis based on immunofluorescence of Sun1-GFP reporter cells (Fig. 5C). To validate our transcriptomic data, we also performed quantitative single-cell immunofluorescence analysis using a subset of markers that would allow us to cluster cell fate into similar categories: Pax7, MyoD, Ki67, myogenin, and nuclear size (SI Appendix, Fig. S8B). A nearly identical trend was observed compared to the transcriptomic data. While both soft and stiff cultured cells appeared in the cycling progenitor and committed and differentiated cell clusters, the stem cluster was predominantly composed of cells cultured on stiff substrates, and the progenitor cluster was predominantly composed of cells cultured on soft substrates (Fig. 5 F and G). Notably, however, the stem population is likely dysfunctional, as described below.

Having validated that our single-cell transcriptomic and immunofluorescence data gave similar classifications of cell populations, we performed differential expression analysis on our transcriptomic data to provide molecular insight into the differences between the stem and progenitor populations that were particularly sensitive to the stiffness of the culture substrate. As expected, cells in the stem cluster expressed higher levels of Pax7 and very low MyoD, and cells in the progenitor cluster expressed lower levels of Pax7 and higher levels of MyoD (Fig. 5H). Cells in the progenitor cluster also expressed higher levels of genes associated with MuSC expansion, such as cyclins D1 and D2, cyclin-dependent kinase 6 (49, 50), and the prostaglandin E2 receptor EP4 (51). In contrast, cells in the stem cluster expressed higher levels of genes associated with quiescence, including components of the Notch, epidermal growth factor, fibroblast growth factor, and Wnt pathways, type V collagens, Sprouty family members, and phosphatase and tensin homolog (PTEN) (2, 8, 52–55). However, these quiescent-like cells also express higher levels of genes linked to loss of MuSC function with aging, including CD47 and growth-arrest specific 1 (GAS1) (56, 57). Thus, while exhibiting many hallmarks of quiescent MuSCs, this stem-like population is likely a dysfunctional, rather than proregenerative, state. Consistent with the mechanoregulation of these cell populations, genes associated with Rho GTPase signaling were also differentially expressed. The stem population expressed higher levels of Rho kinases (ROCK1/2), whereas the progenitor population expressed higher levels of PAK1, downstream of Rac1 signaling (Fig. 5H). Taken together, these results indicate that the mechanical memory is characterized by premature commitment to differentiation or induction of a dysfunctional quiescent-like state in myogenic progeny cultured on stiff substrates versus an expansion of a progenitor population on soft substrates.

Characterizing the Mechanical Memory Molecular Signature.

We sought to characterize persistent transcriptomic changes that could serve as a molecular signature for the mechanical memory. Freshly isolated MuSCs from wild type young (~2 mo) and aged (~24 mo) mice were seeded onto soft or stiff hydrogels and cultured for 1, 3, or 7 d prior to collection for single-cell RNA sequencing. Cells from aged animals were included to determine whether cells in a dysfunctional tissue state characterized by increased microenvironmental stiffness (16–19) were still capable of mechanosensing and forming a mechanical memory. Because the mechanical memory could be blocked with 3 d of RhoA inhibition, an additional condition of treating young cells on stiff gels with RhoA inhibitor was also included. Cells at each time point were enzymatically removed from the gels, fixed, and processed together for RNA sequencing.

The sequenced cells clustered into the expected phenotypes, spanning stem cells at day 1 to differentiated cells at day 7, as classified based on expression of myogenic genes (Fig. 6A). During activation, cells on soft and stiff substrates largely clustered together, suggesting that substantial changes in cell fate were not apparent at these early time points. However, by day 7, there is a bifurcation of cells cultured on soft versus stiff gels in the late progenitor and cycling progenitor clusters. The Rho inhibitor-treated cells at day 3 clustered between the soft day 3 and soft day 7 progenitor cells, consistent with the ability of Rho inhibition to block mechanical memory formation on stiff substrates (Fig. 6A). No substantial differences in the clustering of young versus aged cells were observed, suggesting that aged MuSCs are still capable of mechanosensing, like young MuSCs (Fig. 6A and SI Appendix, Fig. S9). Regardless of age or time point, cells cultured on soft versus stiff substrates maintained a larger fraction of cells in clusters associated with earlier stages of myogenesis. However, aged cells were prone to more rapid commitment and differentiation (SI Appendix, Fig. S10), consistent with our prior work (14). Cell fate trajectory inference was performed using RNA velocity, which considers the relative proportions of spliced and unspliced messenger RNAs to predict the temporal progression of cell state (58). The RNA velocity analysis revealed two primary stable states for the cells after activation, consistent with the formation of separate differentiated and stem/progenitor populations after 7 d in culture (Fig. 6A). Trajectory analysis also indicated two different routes toward differentiation: direct commitment starting at day 1 and expansion of a progenitor pool prior to commitment to differentiation.

Fig. 6.

Single-cell RNA sequencing enables molecular characterization of the mechanical memory. (A) Uniform manifold approximation and projection (UMAP) of single-cell RNA sequencing results with myogenic states assigned to clusters and overlaid with cell fate trajectories inferred from RNA velocity analysis revealed differential clustering for cells in the progenitor pool at day 7 for culture on soft versus stiff substrates and at day 3 for culture on stiff substrates with RhoA inhibition, with no substantial differences in clustering observed as a function of age. (B) Differential gene expression analysis revealed a subset of genes that were consistently up- or down-regulated on soft versus stiff substrates at days 3 and 7. (C) Differential gene expression analysis also revealed a separate subset of genes that were consistently up- or down-regulated on soft and RhoA-inhibited stiff substrates versus control stiff substrates at day 3. (D) By overlapping both sets of genes found in B and C, a set of genes corresponding to a mechanical memory signature was obtained.

To identify a molecular signature of the mechanical memory, we performed differential gene expression analysis in the stem and progenitor populations to determine persistent transcriptomic changes on soft versus stiff substrates. Very few differentially regulated genes were identified at day 1, consistent with formation of the mechanical memory after day 1. By day 3, more genes were differentially expressed, many of which remained either up- or down-regulated at day 7 (Fig. 6B). We performed a similar differential gene expression analysis on day 3 samples comparing soft control and stiff Rho-inhibited samples to stiff control samples. Fifty-five genes were differentially regulated on both soft and Rho-inhibited samples, suggesting these genes may play a role in the mechanical memory formation (Fig. 6C). By comparing genes that were persistently differentially expressed on both days 3 and 7 with genes that were differentially expressed in both soft and Rho-inhibited conditions, a subset of 14 genes consistently appeared, forming a molecular signature for the mechanical memory (Fig. 6D). Some of these genes are known to regulate MuSC fate, while others remain to be explored further. Of those with known effects on MuSCs, GAS1 has been reported to impair MuSC expansion and self-renewal in aged tissues, and glial-derived neurotrophic factor (GDNF) can counteract the effects of GAS1 (57). The fact that GAS1 is down-regulated and GDNF is up-regulated in soft versus stiff conditions is consistent with our observation that soft substrates promote progenitor expansion, while stiff substrates favor dysfunctional quiescence or differentiation. NOTCH3 is a key regulator for maintaining MuSC quiescence (54), and it is also more highly expressed in cells cultured on stiff substrates that favor a premature return to quiescence.

Discussion

Fibrosis, an excessive deposition of ECM that increases the stiffness of affected tissues, contributes to tissue dysfunction and failure in aging and disease (59, 60). We reasoned that altered microenvironmental stiffness in fibrotic tissues could directly impair cell function. While many prior studies have focused on fibrotic biochemistry, considerably less attention has been paid to how the changing biophysical properties of the ECM contribute to pathological cell behavior and tissue dysfunction. Fibrosis is inherently dynamic. Therefore, to mimic fibrotic pathogenesis in vitro, cell culture platforms enabling highly selective, real-time manipulation of matrix mechanics are required. This is particularly true for studies designed to probe the existence of a cellular mechanical memory. Cells have been shown to persistently alter their phenotype in response to transient mechanical cues (23–26, 61). It is therefore imperative to understand how this mechanical memory occurs to design new therapeutic approaches to block or overcome it and restore proper function in fibrotic tissues.

Our studies identifying a mechanical memory in MuSCs were made possible by the development of a synthetic hydrogel system that stiffened in response to light through a bioorthogonal SPAAC reaction (Fig. 4 A–E and SI Appendix, Fig. S5). Using this system, we showed that initial culture on soft substrate conditions was sufficient to prevent formation of a detrimental mechanical memory. Our hydrogel materials constitute an important addition to the biomaterials toolkit for studying and leveraging mechanotransduction in tissue regeneration. Prior hydrogel materials with on-demand stiffening typically used free-radical initiated polymerization (27, 28), which can lead to cytotoxicity in sensitive cell types such as MuSCs and confound biochemical mechanisms that interact with reactive oxygen species. The photo-SPAAC reaction presented here is activated by mild exposure to UV light that is generally well-tolerated by cells (62, 63) and forms additional crosslinks by a highly selective reaction that does not interfere with biological function (64). The ability to add crosslinks into the hydrogel networks enables substantially greater increases in stiffness than can be obtained by altering just the molecular conformation of existing crosslinks (37–39). This chemistry enables the multifold changes in tissue stiffness observed in disease and aging to be recapitulated in vitro (16–19, 65). We therefore anticipate that our on-demand stiffening materials will be broadly useful to uncover mechanisms of mechanically induced dysfunction in other tissues that are susceptible to fibrosis.

In the context of skeletal muscle, our results demonstrate the existence of a mechanical memory in cultured MuSCs. After 3 d on a soft or stiff hydrogel, a memory is established that alters subsequent cell fate decisions and is not reversed by a switch in hydrogel elasticity. Previously, we demonstrated that freshly isolated MuSCs were capable of engrafting and maintaining their stem cell function upon transplantation into skeletal muscle tissue, whereas myoblasts that were first cultured on rigid tissue culture plastic did not successfully engraft (30). However, culturing freshly isolated MuSCs on hydrogel substrates with muscle-like stiffness prior to transplantation preserved the ability of the MuSC progeny to engraft and retain their regenerative stem cell properties (13). These findings suggested that culturing MuSCs on rigid plastic prior to transplantation could be partially responsible for the diminished engraftment potential of these cells. Furthermore, culturing MuSCs on hydrogels softer than plastic, but with increased stiffness similar to fibrotic muscle tissue, reduced cell expansion, and engraftment (13, 15), suggesting that tissue stiffness could impair MuSC function. Our prior results stood in contrast to the mechanosensitive behavior of myoblasts that were first expanded on tissue culture plastic. In two different materials systems, myoblasts showed increased proliferation as substrate stiffness is increased. However, myoblasts only exhibited moderate increases in proliferation on substrates spanning the stiffness of healthy versus fibrotic muscle tissue (20, 21). The present study reconciles these two sets of observations by identifying a mechanosensitive window during MuSC activation when a mechanical memory is formed, dictating whether MuSCs expand as functional progenitors (memory for soft) or prematurely differentiate or enter a dysfunctional quiescent-like stem cell state (memory for stiff). These results could account for the limited engraftment and contribution of myoblasts to muscle tissues when used as a cell therapy in early clinical trials for the treatment of Duchenne muscular dystrophy (66–68).

While the present results are limited to in vitro analysis of cell fate, the fold-change in substrate stiffness sufficient to induce the mechanical memory phenotype is in accordance with that observed in aged and diseased tissues (16–19, 65). We demonstrate that not only young MuSCs but also aged MuSCs have the ability to mechanosense and develop the molecular signature of the mechanical memory (Fig. 6 and SI Appendix, Fig. S9). These findings suggest that the stiffened microenvironment of fibrotic muscle may play a role in impaired MuSC-mediated regeneration in aged muscles (4). Indeed, we have previously demonstrated that restoration of engraftment potential in aged MuSCs required both inhibition of overactive p38 MAPK and culture on substrates with stiffness similar to healthy young muscle (14). Given that we have now demonstrated that p38 activity during activation is not dependent on substrate stiffness (Fig. 4 B and F), our data underscore that treatments to restore the function of aged MuSCs will need to overcome both cell-intrinsic defects, like overactive p38, and extrinsic microenvironmental defects (4, 14). Two genes that are up-regulated in the dysfunctional stem-like cells on stiff substrates, CD47 and GAS1, have already been confirmed by us and others to impair the regenerative potential of aged MuSCs (56, 57), suggesting a potential mechanical link to the increased expression of these genes in MuSCs residing in a stiffened fibrotic niche. The identified signaling pathways constitute potential avenues for future research to block or surmount formation of the mechanical memory during MuSC activation in fibrotic tissue. One potential approach is to interfere with signaling downstream of RhoA, such as ROCK-mediated activation of PTEN (69), to promote expansion of myogenic progenitors. ROCK inhibition has recently been shown by others to promote a more elongated morphology in MuSCs (33, 70), similar to our observations of MuSCs cultured on soft substrates. Another approach may be to activate GDNF signaling to overcome the effects of up-regulated GAS1 expression (57). In aged or diseased muscle, mechanical memory blocking approaches will likely need to be combined with other treatments that enhance MuSC expansion, such as inhibition of p38 MAPK or activation of prostaglandin signaling (14, 51, 71), to overcome both biophysical and biochemical dysregulation to promote effective regeneration. Our studies underscore that an in-depth mechanistic understanding of the signaling cascades induced by mechanosensing of cells on hydrogels that mimic a fibrotic microenvironment can inform strategies to increase the efficacy of stem cell-focused therapies.

Materials and Methods

Synthesis and Characterization of Hydrogels.

Detailed synthetic protocols for hydrogel components are provided in SI Appendix and in our previously published work (15). Hydrogels were formed by reacting azide- and BCN-containing PEG precursors at the desired polymer concentration at 37 °C for 30 min. The resulting hydrogels were allowed to swell for at least 1 h before subsequent characterization or use. For dynamically softening or stiffening hydrogels, following equilibration, the hydrogels were exposed to 365 nm light to trigger changes in stiffness. Detailed characterization protocols are provided in SI Appendix.

MuSC Isolation and Culture.

All animal protocols were approved by the Stanford University Administrative Panel on Laboratory Animal Care and experiments were performed in compliance with the institutional guidelines of Stanford University. Myogenin Sun1-GFP reporter mice (15), young (~2 mo) C57BL/6 mice (The Jackson Laboratory), and aged (~22 mo) C57BL/6 mice (National Institute on Aging) were used. MuSCs were isolated from the hindlimb muscles of mice following our published procedures (13, 15, 30). Freshly isolated MuSCs were immediately seeded onto hydrogels and cultured for 1, 3, or 7 d before fixation and subsequent analysis by immunofluorescence, as in our published procedures (15). Additional details are provided in SI Appendix.

Single-Cell RNA Sequencing and Bioinformatics Analysis.

Cells cultured on hydrogels were washed with phosphate buffered saline and detached from the substrates by treatment with TrypLE Express and manual pipetting. The cells were collected by centrifugation and processed using the Evercode Mini WT and Evercode WT kits from Parse Biosciences, following the manufacturer's protocols. Sequencing was performed by Novogene on an Illumina NovaSeq 6000 system. Cells were sequenced to an average read depth of 180 k reads/cell. Downstream analysis of transcriptomic data was performed as described in SI Appendix.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

Single cell RNA sequencing data have been deposited in GEO (GSE263619) (72). All other data are included in the article and/or SI Appendix.