Actomyosin-mediated cellular tension promotes Yap nuclear translocation and myocardial proliferation through α5 integrin signaling

The neonatal mammalian heart is capable of substantial regeneration following injury; however, this regenerative capacity largely disappears in the first week after birth (Porrello et al., 2011, 2013). As such, the adult human heart does not have an effective mechanism to regenerate the cardiomyocytes (CMs) lost following a heart attack. Instead, the lost muscle is replaced by the excessive deposition of extracellular matrix (ECM), which together reduces both systolic and diastolic left ventricular (LV) functions of the heart and often leads to heart failure. Thus, a better understanding of the molecular mechanisms controlling CM proliferation is required in order to identify novel targets that will stimulate cardiac regeneration following myocardial infarction.

The Rho-associated coiled-coiled containing protein serine/threonine kinases ROCK1 and ROCK2 are downstream effectors of the Rho subfamily of small GTPases (Hartmann et al., 2015; Noma et al., 2006; Shimokawa et al., 2016). As a key regulator of intracellular contractility, ROCK1/2 allows cells to respond to mechanical cues from the tissue microenvironment. Evidence from animal studies indicate an important role for ROCK in cardiac morphogenesis and in adult heart pathophysiology (Shi et al., 2011). Pharmacological inhibition of ROCK during early chick and mouse embryogenesis has been shown to cause defects in cardiac morphogenesis (Wei et al., 2001). Consistent with this finding, simultaneous deletion of both ROCK1 and ROCK2 in the heart using a cardiac-specific αMHC-Cre transgene causes embryonic lethality (Shi et al., 2019). Interestingly, cardiac-restricted knockout of ROCK2 using the αMHC-Cre transgene (ROCK2 CKO mouse model) did not result in a cardiac phenotype under basal conditions (Okamoto et al., 2013; Sunamura et al., 2018). However, following angiotensin II treatment or transverse aortic constriction ROCK2 CKO mice exhibited less cardiac hypertrophy, reduced end-diastolic wall thickness, and decreased LV mass compared with control littermates, supporting the idea that ROCK2 is required for CMs to respond to mechanical stimuli. Using a ROCK dominant-negative transgenic model (ROCK DN), endogenous ROCK activity was blocked in embryonic CMs, resulting in disruption of sarcomeres and reduced CM proliferation (Bailey et al., 2019), although no molecular mechanism was established to account for the decreased CM proliferation. These in vivo studies demonstrate the requirement for ROCK2 signaling in cardiac development and cardiac remodeling following mechanical stress, but its specific role in postnatal CM maturation remains poorly defined.

ROCK mediates the phosphorylation of myosin regulatory light chains leading to increased myosin ATPase activity, thus stimulating actomyosin contractility. Nonmuscle myosin IIB (NM IIB) is the primary NM isoform expressed in heart muscle. As CMs mature after birth, nonmuscle myosin activity decreases and it is reactivated in response to mechanical stress in the adult heart, e.g. after myocardial infarction (Ma and Adelstein, 2012; Pandey et al., 2018). Germline knockout of NM IIB in mice leads to multiple developmental abnormalities, including cardiac malformations and cardiac hypertrophy resulting in embryonic lethality (Tullio et al., 1997). Notably, defective cytokinesis causes an increased number of enlarged binucleated CMs in the NM IIB mutant hearts, supporting a role for NM IIB in myocardial proliferation (Takeda et al., 2003). Mechanistically, NM IIB actin cross-linking activity is required to generate tension that drives cytokinesis (Ma et al., 2012). Cardiac-specific NM IIB knockout (αMHC-Cre) mice also exhibit cytokinesis defects; however, the mutant mice survive and go on to develop progressive cardiomyopathy with age (Ma et al., 2009). Taken together, these data indicate the importance of NM IIB in regulating cardiac development and homeostasis, including CM cytokinesis.

The transition from hyperplastic to hypertrophic growth in the postnatal heart is accompanied by dynamic remodeling of cell–cell and cell–ECM adhesive junctions. Specifically, CM α5/β1 integrin-fibronectin (FN; FN1) adhesions decrease after birth and N-cadherin (cadherin 2, CDH2) junctions remodel and accumulate at myocyte termini leading to the formation of a specialized junction called the intercalated disc (ICD) (Vite and Radice, 2014). Establishment of functional cadherin and integrin adhesion complexes require the reorganization and assembly of the actin cytoskeletal network. Modulation of intracellular tension can have different effects on cell–cell and cell–ECM interactions depending on the cellular context (de Rooij et al., 2005; Dzamba et al., 2009; Liu et al., 2010; Martinez-Rico et al., 2010; Shewan et al., 2005). Here, we focus on the effects of cellular tension on the remodeling of myocardial adhesions in the postnatal heart prior to ICD assembly.

We previously reported that altering the linkage between N-cadherin and the actin cytoskeleton by depleting αE-/αT-catenins (CTNNA1/3) caused an increase in RhoA activity, mislocalization of actomyosin contractility, and enhanced CM proliferation in the postnatal mouse heart (Li et al., 2015; Vite et al., 2018). Whether enhanced intracellular tension is responsible for the CM hyperproliferation phenotype in the αE-/αT-catenin mutant mice is unknown. To address this possibility, here we studied the direct effect of activating ROCK2 signaling and consequent actomyosin contractility on CM maturation. We find that inducing actomyosin contractility leads to a rebalancing of myocardial cell–cell and cell–matrix adhesions, with the latter responsible for Yap (Yap1) nuclear translocation and CM proliferation. These results implicate the actomyosin cytoskeleton as a crucial mediator of crosstalk between cadherin and integrins, and show that activating nonmuscle myosin contractility leads to enhanced myocardial cell–FN interactions, which stimulate CM proliferation in the postnatal heart.

The temporal expression of NM IIB, nonmuscle myosin light chain 2 (MLC2; MYL2), and active phosphorylated MLC2 (pMLC2) was examined in postnatal hearts of wild-type mice (C57BL/6J) by western blotting (Fig. 1A). NM IIB, which controls tension within the actin cytoskeleton, is activated by phosphorylation of its light chain at Ser19 (pMLC). Both NM IIB levels and MLC2 activity declined after birth (Fig. 1A), consistent with previous report of NM IIB expression in murine heart (Tullio et al., 1997), whereas ROCK2 levels remained constant over the same period. To investigate the effects of actomyosin contractility on CM differentiation and cell cycle arrest, we used a conditionally activated ROCK2:ER mouse model (Samuel et al., 2016), outlined in Fig. 1B. Fusion of the ROCK2 kinase domain with the estrogen receptor (ER) hormone-binding domain generates a ROCK2:ER chimeric protein that is inactive in the absence of ligand, but which can be conditionally activated in vivo by administering the estrogen analog 4-hydroxytamoxifen (Tam) (Rath et al., 2017; Samuel et al., 2016). ROCK2:ER mice were bred with αMHC/Cre mice to generate a cardiac-restricted ROCK2:ER mouse model. Western blot analysis confirmed expression of the ROCK2:ER fusion protein in ROCK2:ER; αMHC-Cre hearts (Fig. 1D). CM proliferation decreases after postnatal day (P) 4 in mice (Soonpaa et al., 1996). Therefore, to investigate the effects of cellular tension on the transition from hyperplastic to hypertrophic CM growth, mice were administered Tam via intraperitoneal injection for three consecutive days: P5, P6 and P7 (Fig. 1C). On P7, 4 h after the last Tam injection, hearts were harvested for analysis. The kinase ROCK2 is required for RhoA-initiated actomyosin contractility, acting through phosphorylation of substrates including MLC2 and MYPT1 (PPP1R12A). Western blot analysis confirmed increased expression of NM IIB, pMLC2 (Ser19) and pMYPT (Thr696) in ROCK2:ER; MHC-Cre hearts following Tam administration (Fig. 1E). Tam administration alone to wild-type C57BL/6J mice had no effect on actomyosin contractility, as determined by pMLC/MLC expression (Fig. S1). For all subsequent studies, pups were administered Tam as described above. Littermates lacking the Cre transgene (i.e. ROCK2:ER alone) served as controls, and ROCK2:ER; MHC-Cre mice are referred to as ‘ROCK’. Actomyosin contractility, as visualized by punctate pMLC immunofluorescence, was higher in P7 ROCK CMs compared with control (Fig. 1F). Consistent with increased nonmuscle myosin contractility, there was more NM IIB along the lateral membrane of ROCK CMs compared with control (Fig. 1G).

Histological analysis did not detect any cellular or morphological abnormalities in the P7 ROCK hearts (Fig. S2). To assess cardiac function, echocardiography was performed on P7 pups. There was no change in LV function as determined by ejection fraction and fraction shortening in the ROCK mice (Fig. S2). Moreover, LV anterior (LVAW) and posterior (LVPW) wall thickness was not changed in the ROCK mice. Thus, temporary (P5-P7) activation of ROCK2 did not perturb cardiac structure or function.

Mechanical equilibrium between cells and tissues is dependent on maintaining a proper balance between cell–cell and cell–ECM interactions (Han and de Rooij, 2016; Mui et al., 2016; Zuidema et al., 2020). To investigate the effects of actomyosin contractility on the crosstalk between cell–ECM and cell–cell interactions, we examined the spatial distribution and expression of α5/β1 integrin (the receptor for FN) and FN in P7 ROCK hearts (Fig. 2A). Increase in α5 integrin expression was accompanied by enhanced extracellular FN matrix along the lateral membrane, consistent with strengthening of myocardial cell–matrix interactions in the ROCK hearts. Integrin-FN binding leads to the formation of FN fibrils that are initially soluble in the detergent deoxycholate (DOC) but are gradually converted into a stable, DOC-insoluble form that comprises the mature matrix (Sechler et al., 1996). Both α5 and FN increased in the DOC-insoluble fraction, consistent with compensatory matrix assembly in response to altered intracellular tension in ROCK-activated CMs (Fig. 2B). At this developmental stage, N-cadherin and α5 integrin exhibited extensive colocalization at the membrane, with the notable exception of enhanced α5 integrin expression at the lateral membrane in the ROCK CMs (Fig. 2C, enlarged images).

Both the cadherin- and integrin-based adhesion systems are linked to the cortical actin cytoskeleton and share common downstream actin regulatory pathways, including the cytoskeletal adaptor protein vinculin. Vinculin regulates cell adhesion by directly binding to actin, stimulating actin polymerization and recruiting actin remodeling proteins in response to intrinsic and extrinsic mechanical forces (Bays and DeMali, 2017). As a mechanosensitive protein, vinculin undergoes a conformational change in response to mechanical force that results in actin remodeling and adhesion strengthening at cell junctions, a prerequisite for cadherin junction maturation. The junction remodeling process requires phosphorylation of vinculin at tyrosine residue 822 by the nonreceptor tyrosine kinase c-Abl (Abl1) (Bays et al., 2014). Thus, increased levels of vinculin phosphorylation (Y822) correlate with less mature cadherin junctions. To investigate vinculin (VCL) activity following ROCK activation, we performed immunofluorescence and western blot analyses on P7 hearts using a phospho-specific antibody that recognizes pVCL-Y822 (Bays et al., 2014). Vinculin was primarily localized at the membrane as expected, whereas phosphorylated VCL (pVCL) exhibited a diffuse but heterogeneous distribution, which was enhanced in ROCK hearts compared with control (Fig. 3A,C). Western blot analysis confirmed the increased pVCL in ROCK hearts (Fig. 3B). A portion of the pVCL colocalized with N-cadherin at the membrane (Fig. 3C, arrowheads); however, most remained diffuse in the cytosol. Total VCL and N-cadherin protein levels did not change in ROCK hearts (Fig. 3B,D). These results are consistent with the previous observation that pharmacological inhibition of ROCK blocks force-induced VCL phosphorylation in cultured epithelial cells (Bays et al., 2014). We previously demonstrated that simultaneous depletion of both αE- and αT-catenin results in perturbation of nascent N-cadherin junctions, Rho activation and enhanced actomyosin contractility (Vite et al., 2018). Indeed, we found that pVCL expression was increased in the αE-/αT-catenin double knockout CMs (Fig. S3), comparable to observations in the ROCK model, providing independent support for pVCL as a marker of dynamic junction remodeling in the postnatal heart.

To investigate the N-cadherin/catenin complex, heart sections were co-stained with N-cadherin and either β-catenin, αE-catenin or αT-catenin. The expression and distribution of the N-cadherin/catenin adhesion complex appeared to be similar between ROCK and control P7 hearts (Fig. S4). Cadherin junctions mechanically couple to the actomyosin network via mechanosensing proteins, including VCL and catenins (Lecuit and Yap, 2015); therefore, we examined VCL with either β-catenin, αE-catenin or αT-catenin. The colocalization pattern of VCL and catenins was similar between ROCK and control P7 hearts (Fig. S5). Although expression and distribution of the N-cadherin/catenin complex appeared normal, the increase VCL Y822 phosphorylation suggests that cell–cell adhesions are more dynamic and less stable in ROCK-activated CMs.

The tissue microenvironment plays a key role in the fate and function of CMs. FN and ECM-associated growth factors are crucial regulators of CM proliferation and behavior (Ieda et al., 2009; Wu et al., 2020). Therefore, we examined incorporation of 5-ethynyl-2'-deoxyuridine (EdU; an indicator of S phase) and phosphorylation of histone H3 (pHH3; an indicator of M phase) in ROCK hearts at P7 (Fig. 4A,B). No increase in cell cycle activity was observed in control mice: ROCK alone, ROCK; MHC-Cre without Tam administration or ROCK alone plus Tam (Fig. S6). The percentage of EdU-positive CMs in ROCK hearts (10.66%±0.70%, n=12) was about 1.5-fold higher than in control (6.90%±0.51%, n=10, P<0.001) P7 hearts. Similarly, the percentage of pHH3-positive CMs in ROCK hearts (8.48%±0.61%, n=11) was about 1.9-fold higher than in control (4.50%±0.24%, n=11, P<0.0001) P7 hearts. Myocyte cross-sectional areas were reduced in the ROCK hearts (−12%; n=4, P<0.01; Fig. S7).

Mechanical cues, including rearrangement of cytoskeletal tension, control Yap activity independently of the Hippo kinase cascade (Dupont et al., 2011), and, importantly, Yap has been shown to be necessary and sufficient for CM proliferation (von Gise et al., 2012; Xin et al., 2013). Therefore, we examined the cellular distribution of Yap after activation of actomyosin contractility (Fig. 4C). Quantitative analysis demonstrated that the ROCK hearts had a 1.6-fold increase percentage of nuclear Yap-positive CMs (52.97%±2.93%, n=11) compared with control littermates (33.61%±2.23%, n=11, P<0.0001). There was a modest increase in total Yap in ROCK hearts, but no change in phosphorylated Yap (S112 or Y357) (Fig. 4D). Next, we examined expression of Yap target genes by quantitative reverse transcription polymerase chain reaction (qRT-PCR). The Yap target genes Ctgf (Ccn2) and Cyr61 (Ccn1) were upregulated twofold, consistent with increased Yap transcriptional activity in the ROCK hearts (Fig. 4E). Together, these data indicate that increasing actomyosin cytoskeletal tension is sufficient to drive Yap to the nucleus and increase cell cycle activity, supporting a role for nonmuscle myosin contractility in the regulation of CM proliferation in the postnatal heart.

To investigate CM proliferation further, we took advantage of the integration of the loxP-STOP-loxP-EGFP:ROCK2:ER transgene at the hypoxanthine phosphoribosyltransferase (Hprt) locus on the X chromosome (Samuel et al., 2016). In female mammals, dosage compensation of X-linked genes between males and females occurs by genetic inactivation of one of the two X chromosomes (Goto and Monk, 1998). The choice of X chromosome to be inactivated is random in female somatic cells. We hypothesized that ROCK-activated CMs will undergo increased cell division resulting in large GFP-positive clones containing a high percentage of EdU-positive, GFP-positive CMs. To generate mosaic X^RX, MHC-Cre females, we bred MHC-Cre male mice with ROCK2:ER (R) homozygous (X^RX^R) females (Fig. 5A). The experiment is outlined in Fig. 5B. The fusion of GFP to the ROCK kinase domain enabled the identification of GFP-positive CMs using an anti-GFP antibody. In the absence of MHC-Cre, the loxP-STOP-loxP-EGFP:ROCK2:ER transgene is not expressed. Hence, CMs exhibited no GFP signal (Fig. 5C), whereas the presence of MHC-Cre in male ROCK (X^RY) mice resulted in GFP expression in all CMs (Fig. 5C). Next, we examined GFP-positive clones in mosaic (X^RX) female hearts with and without Tam treatment (Fig. 5D). In ROCK-activated (+Tam) female (X^RX) hearts, CM clones consisted of more GFP-positive CMs compared with control (−Tam) GFP-positive clones. Importantly, the percentage of EdU-positive, GFP-positive CMs/clone was greater in ROCK hearts compared with control female X^RX hearts. Wheat germ agglutinin (WGA) was used to stain the cell borders. The close proximity of mononucleated EdU-positive CMs is consistent with cell division and the formation of two daughter cells in ROCK-activated hearts (Fig. 5E). Taken together, these results indicate that activation of ROCK and consequent actomyosin contractility is sufficient to stimulate CM proliferation in the postnatal heart.

ECM and ECM-associated proteins play a crucial role in regulating CM proliferation (Ieda et al., 2009; Wu et al., 2020). To determine whether CM–FN interactions are required for the hyperproliferation phenotype in the ROCK hearts, we introduced an α5 integrin (Itga5) floxed allele into the ROCK2:ER model to generate ROCK2:ER; Itga5 f/+; MHC-Cre (ROCK; α5^+/−) and ROCK2:ER; Itga5 f/+ (control) mice (Fig. 6A). Deletion of one copy of Itga5 resulted in reduced α5 integrin membrane expression in ROCK; α5^+/− CMs (Fig. 6B). Consistent with knockdown of α5 integrin, FN matrix surrounding the ROCK; α5^+/− CMs was reduced (Fig. 6B). Western blot analysis confirmed reduction of α5 integrin in ROCK; α5^+/− heart lysates (Fig. 6C).

Next, we compared CM proliferation rates of ROCK2:ER; Itga5 f/+ (control), ROCK2:ER; MHC-Cre (ROCK) and ROCK2:ER; Itga5 f/+; MHC-Cre (ROCK; α5^+/−) mice (Fig. 7A,B). The percentage of EdU-positive CMs in ROCK hearts (9.25%±0.57, n=7) was significantly higher compared with control (4.82%±0.76%, n=4, P<0.001) and ROCK; α5^+/− hearts (6.78%±0.32%, n=7, P<0.01) at P7. In addition to using sarcomeric α-actinin to identify CMs, EdU analysis was performed with the nuclear CM marker Nkx2-5 (Fig. S8). Comparable EdU results were obtained with either CM marker. Similarly, the percentage of pHH3-positive CMs in ROCK hearts (7.56%±0.09%, n=6) was increased significantly compared with control (5.31%±0.12%, n=4, P<0.001) and ROCK; α5^+/− hearts (6.01%±0.35%, n=7, P<0.01) at P7, whereas EdU and pHH3 were not significantly different between ROCK; α5^+/− and control hearts. Thus, deletion of one copy of Itga5 was sufficient to rescue the excessive mitotic CMs in the ROCK hearts. Next, we assessed Yap cellular distribution in the rescued hearts. Indeed, nuclear translocation of Yap was reduced in ROCK hearts upon deletion of one Itga5 allele (Fig. 7C). The percentage of nuclear Yap-positive CMs in ROCK hearts (56.62%±2.73%, n=6) was increased significantly compared with control (31.23%±2.27%, n=4, P<0.001) and ROCK; α5^+/− CMs (43.49%±3.62%, n=7, P<0.05) at P7, whereas the percentage of nuclear Yap positive CMs was not significantly different between ROCK; α5^+/− and control hearts. Additionally, CM cytokinesis, as determined by Aurora-B kinase (AURKB) localization to the midbody, was increased in the ROCK (6.53 AURKB+CMs/field, n=7, P<0.001) compared with control (3.93 AURKB+CMs/field, n=4) and ROCK; α5^+/− (3.88 AURKB+CMs/field, n=7) hearts (Fig. 7D). Collectively, our results demonstrate that high nonmuscle myosin contractility shifts the balance from cell–cell to cell–matrix adhesion, leading to Yap nuclear translocation and CM proliferation (Fig. 8). Furthermore, the effect of genetic knockdown of Itga5 supports the proposition that CM–ECM interactions (i.e. α5/β1–FN) regulate Yap activity and CM proliferation in the postnatal ROCK heart.

Shortly after birth, nonmuscle myosin contractility declines as CMs mature in response to increased cardiac demand and growth during the postnatal period. During the same period, myocardial cell–ECM and cell–cell interactions reorganize, resulting in the assembly of strong cell–cell contacts comprising N-cadherin/catenin adhesion complexes. The postnatal heart thus offers a unique system to investigate how actomyosin-generated contractile force regulates cell adhesion, CM maturation and cell cycle arrest. We previously reported that modification of N-cadherin junctions as a result of α-catenin depletion was associated with increased actomyosin activity, enhanced Yap nuclear translocation and increased CM proliferation in the postnatal heart (Vite et al., 2018). Based on these findings, it was hypothesized that downregulation of nonmuscle myosin contractility might be required for proper remodeling and maturation of cadherin-based and integrin-based adhesions during postnatal heart development. Here, we utilize a cardiac-restricted ROCK2 conditionally activated gain-of-function murine model (ROCK2:ER) to evaluate directly the consequences of actomyosin contractility on postnatal CM cell adhesions and cell cycle arrest. The Tam-activated ROCK2:ER model has been used to study the effects of cellular tension on tissue homeostasis in epidermis, mammary gland, intestine and pancreas (Boyle et al., 2020; Rath et al., 2017; Samuel et al., 2011, 2016). To our knowledge, this is the first time the ROCK2:ER model has been used to manipulate actomyosin contractility in a cell type expressing both muscle myosins and nonmuscle myosins.

We report here that activation of ROCK enhances CM–ECM interactions (i.e. α5/β1–FN), thus promoting nuclear translocation of Yap, which results in increased CM proliferation in the postnatal heart. These results are consistent with our previous studies manipulating cytoskeletal tension in cultured CMs (Vite et al., 2018). We had previously shown that activation of RhoA (CN03) is sufficient to drive Yap to the nucleus in neonatal CMs grown on polyacrylamide hydrogels corresponding to physiological stiffness (Vite et al., 2018). Conversely, relieving cytoskeletal tension with pharmacological inhibitors of RhoA (CT04), ROCK (Y27632) or myosin II (blebbistatin) caused Yap to exit the nucleus (Vite et al., 2018). Moreover, RhoA activity is required for cyclic stretch-induced Yap nuclear translocation in cultured neonatal CMs (Byun et al., 2019). These in vitro studies support the idea that increased actomyosin contractility is responsible for Yap accumulation in the nucleus and consequent proliferation of ROCK-activated CMs.

We took advantage of X-chromosome inactivation (XCI) in mice (Lyon, 1961) to perform clonal analysis in mosaic ROCK female hearts. Transgenes targeted to the Hprt locus have proved useful to study XCI and its effects on regional phenotypic variability in heterozygous females (Ciavatta et al., 2006; Prosser et al., 2008; Wu et al., 2014). Using a dual fluorescent reporter system combined with a cell type-specific Cre transgene, XCI can be visualized at single-cell resolution (Wu et al., 2014). Here, we use MHC-Cre to delete the STOP signal that allows CAG-EGFP:ROCK2:ER transgene expression and CM-specific GFP signal. In ROCK (X^RY) males, all CMs express GFP, whereas heterozygous ROCK (X^RX) females exhibit a mosaic GFP staining pattern as a result of XCI. Using EdU incorporation to mark cells in S phase, we found that ROCK activation (+Tam) leads to larger GFP⁺ clones and, importantly, the clones consist of a higher percentage of EdU⁺, GFP⁺ CMs compared with control (−Tam) GFP⁺ clones. These data further support the idea that cytoskeletal tension drives Yap to the nucleus, leading to increased CM cell cycle activity and cell division.

It is intriguing that nonmuscle myosin activity is re-activated in the adult myocardium following myocardial infarction, with pMLC found colocalized with vinculin at costameres (Pandey et al., 2018). Whether activation of actomyosin activity in the setting of ischemic injury is compensatory and beneficial remains to be determined. In support of a beneficial effect, expression of modest levels of constitutively active RhoA protected mice from ischemia/reperfusion injury, as determined by reduced infarct size and improved contractile function (Xiang et al., 2011). In the same study, RhoA loss of function resulted in worse cardiac outcome following ischemia/reperfusion injury. Thus, we can speculate that activation of nonmuscle myosin following ischemic injury may serve a dual purpose: reinforcement of cell–matrix adhesions at costameres, thus strengthening cardiac contractility, and activation of CM cell cycle activity. Future experiments will be necessary to address these and other alternative possibilities.

FN has emerged as a crucial regulator of cardiac development and myocardial growth control (Ieda et al., 2009; Mittal et al., 2013; Trinh and Stainier, 2004; Wu et al., 2020). FN matrix assembly is mediated by integrin receptors and regulated by intracellular signals, cytoskeletal organization, and availability of FN (Singh et al., 2010). Cardiac integrin α5/β1 is the primary receptor for FN. In co-culture experiments, secretion of FN by embryonic cardiac fibroblasts induces embryonic CM proliferation, which is dependent on β1 integrin signaling (Ieda et al., 2009). However, Yap activity was not investigated in this earlier study. During the postnatal period, cardiac ECM undergoes significant remodeling resulting in increased tissue stiffness (Hortells et al., 2019). FN protein levels decline 1 week after birth, as collagen levels increase and remain high in the adult myocardium. Following ischemic injury in humans, FN is re-expressed in the infarct and border zone (Willems et al., 1996). Here, we show that CM-specific activation of actomyosin contractility is sufficient to increase interstitial FN matrix. Importantly, cardiac-specific knockdown of α5 integrin (i.e. Itga5 f/+) reduces FN matrix and restores CM nuclear Yap and CM cell cycle activity to control levels, demonstrating that CM–FN interactions are required for enhanced CM proliferation in ROCK hearts. In cultured mesenchymal stem cells, FN-induced Yap nuclear translocation requires cytoskeletal tension (Stanton et al., 2019), consistent with our findings in ROCK-activated CMs. In postnatal CMs, Yap was shown to directly target genes involved in cytoskeletal remodeling at integrin adhesions (Morikawa et al., 2015). Moreover, Yap loss of function in nonmuscle cells leads to downregulation of α5 integrin and reduced assembly of focal adhesions (Nardone et al., 2017). In endothelium, α5 integrin/FN/Yap mechanotransduction plays a key role in inflammation and vascular remodeling in atherosclerosis (Wang et al., 2016; Yun et al., 2019). Together, these data suggest that ROCK/integrin/Yap mechanotransduction may function as a positive-feedback loop to maintain cytoskeletal tension via cell–ECM adhesions, thus promoting Yap nuclear translocation during development and disease.

Cardiac fibroblasts (cFbs) are the major cell type contributing to the synthesis of the ECM, which surrounds the CMs and acts as a scaffold, maintaining the myocardial tissue architecture and contractile force of the heart (Hortells et al., 2019; Tallquist, 2020). Periostin (Postn) is a marker of activated cFbs, which have important roles in tissue regeneration and wound healing (Kanisicak et al., 2016). Recent lineage-tracing studies identified a heterogeneous population of cFbs in the postnatal heart consisting of Tcf21⁺ cFbs and Postn⁺ cFbs (Hortells et al., 2020). Although less abundant than Tcf21⁺ cFbs, Postn⁺ cFbs are highly proliferative and, importantly, their ablation during the first week leads to decreased CM mitotic activity and a reduction in binucleation. The correlation between high nonmuscle myosin activity and the appearance of Postn⁺ cFbs in both the postnatal heart and in the ischemic adult heart suggest that enhanced CM intracellular tension may provide a mechanical cue to activate cFbs. Further studies are necessary to determine whether activation of actomyosin contractility is sufficient to induce Postn⁺ cFbs in the heart.

Mechanosensitive proteins such as α-catenin and vinculin undergo force-induced conformational changes resulting in distinct biochemical functions, allowing cells to sense and respond to mechanical stimuli (Charras and Yap, 2018; Pannekoek et al., 2019). Although cytoskeletal tension is important for the formation of both cell–cell and cell–matrix adhesions, excessive tension might induce changes in protein conformation that lead to junctional instability (Weber et al., 2011). In the case of vinculin, site-specific tyrosine phosphorylation distinguishes its mechanical role in cadherin adhesions and integrin adhesions (Bays and DeMali, 2017). Vinculin conformational activation at cadherin-containing adherens junctions requires tension and phosphorylation at tyrosine residue Y822, which is regulated by the nonreceptor tyrosine kinase c-Abl (Bays et al., 2014; Bertocchi et al., 2017). Here, we show that ROCK activation leads to increased VCL Y822 phosphorylation, consistent with its role in force-mediated junction remodeling in epithelial cells (Bays et al., 2014). Conversely, inhibition of ROCK blocks VCL phosphorylation in epithelial cells exposed to force (Bays et al., 2014). A key avenue for future investigation will be to explore the consequences of blocking vinculin phosphorylation during N-cadherin junction remodeling in the postnatal heart.

In summary, our findings show that nonmuscle myosin contractile forces regulate the balance between myocardial cell–cell and cell–ECM adhesions, which, in turn, control Yap nuclear translocation and CM proliferation in the postnatal heart. The ability to induce as well as reverse ROCK activity specifically in CMs will prove invaluable in understanding the role of cellular tension in cardiac development and regeneration, and thus provide useful insights into the potential use of actomyosin agonists for stimulating myocardial proliferation following ischemic injury.

Lox-Stop-Lox ROCK2:ER mice (Samuel et al., 2016), designated ROCK2:ER mice, were bred with α-Myosin heavy chain (MHC)-Cre mice (Agah et al., 1997) to generate a cardiac-restricted, conditionally-active ROCK model. α5 integrin (Itga5) floxed mice (van der Flier et al., 2010) were bred with MHC-Cre mice, and resulting Itga5 f/f; MHC-Cre mice were bred with ROCK2:ER mice to generate ROCK2:ER; Itga5 f/+; MHC-Cre (ROCK; α5^+/−) mice. The mice were maintained on C57BL/6J genetic background. To conditionally activate ROCK2 kinase in the postnatal heart, P5, P6 and P7 mice were injected intraperitoneally once a day with Tam (Sigma-Aldrich, T5648) at 80 mg/kg body weight. To ensure inclusion of both sexes for these studies, the sex of the neonatal pups was confirmed by the presence or absence of the Y chromosome by PCR. All animal studies were performed in accordance with the guidelines of the Institutional Animal Care and Use Committee of Rhode Island Hospital.

Transthoracic echocardiography was performed without anesthesia and with mice lightly restrained in a supine position, at ambient temperatures >25°C, using VisualSonics Vevo 2100 ultrasound system (FUJIFILM VisualSonics Inc.) (Fatkin et al., 1999). Two-dimensional images were obtained using a MS700 transducer (50 MHz). Heart rate, ejection fraction, fractional shortening, left ventricular internal dimensions, left ventricular volume, LVPW and LVAW, were measured from M-mode images at the plane bisecting the papillary muscles. The researcher was blinded to the genotype of the pups.

Mouse hearts were isolated and washed in PBS. The hearts were then fixed in fresh 4% paraformaldehyde at 4°C overnight, dehydrated, and embedded in paraffin. Longitudinal 5-μm deparaffinized sections were stained with Hematoxylin and Eosin, mounted and imaged with Olympus BX43 microscope system. Images were elaborated with cellSens™ imaging software.

Heart tissues were harvested 4 h after the third and final Tam injection and embedded in OCT (Sakura Finetek). OCT sections (6 µm) were fixed in 4% paraformaldehyde for 10 min and then washed in PBS three times. The heart sections were then permeabilized in 0.1% Triton X-100 (in PBS) for 10 min at room temperature. Sections were blocked by T-BSA (5% BSA in 0.01% Triton X-100 in PBS) for 1 h at room temperature and incubated overnight at 4°C with primary antibodies: pHH3 (1:200, 06-570, Millipore), YAP (1:200, 4912, Cell Signaling Technology), α-actinin (1:400, A7732, Sigma-Aldrich), N-cadherin (1:400, 610921, BD Biosciences), pMLC2 (1:100, 3674, Cell Signaling Technology), myosin IIb (1:200, M7939, Sigma-Aldrich), α5 integrin (1:200, 553319, BD Biosciences), FN (1:100, PA1-23693, Thermo Fisher Scientific), phospho-VCL-Y822 (pVCL) (1:200, ab61071, Abcam), VCL (1:100, V4505, Sigma-Aldrich), Aurora B (1:400, 611082, BD Biosciences), cardiac actin (1:400, A2103, Sigma-Aldrich), Nkx2-5 (1:200, ab35842, Abcam), αE-catenin (1:50, 711200, Invitrogen), αT-catenin (1:100, 13974-1-AP, Proteintech), β-catenin (1:200, C2206, Sigma-Aldrich). For detection of GFP, OCT sections were fixed in methanol and incubated overnight at 37°C with primary antibody (1:400, 632593, Takara Bio). After washing in PBS, sections were incubated with secondary antibodies (Alexa Fluor 488 goat anti-rabbit, A11008; Alexa Fluor 488 goat anti-mouse IgG, A11001; Alexa Fluor 555 donkey anti-mouse IgG, A31570; or Alexa Fluor 586 donkey anti-rat IgG, A175475; Life Technologies) for 1 h at room temperature. Finally, the tissue sections were washed in PBS and mounted with ProLong Gold Antifade Reagent containing DAPI (P36935, Life Technologies). Images were acquired using a Nikon A1R confocal microscope system. z-stack images were acquired in a range of 5 µm with a step of 0.36 µm. Images were elaborated and analyzed using Nikon NIS-Elements software and ImageJ 1.53 software.

To assess CM proliferation in heart tissue, mice were injected intraperitoneally with EdU (20 mg/kg body weight) at P7. Four hours after the injection, the hearts were removed, washed in PBS, and embedded in OCT. DNA synthesis was revealed using the Click-iT EdU Alexa Fluor 488 Imaging Kit (C10337, Life Technologies) following the manufacturer's instructions and then co-stained with the cardiac marker α-actinin (1:400, A7732, Sigma-Aldrich). A minimum of ten fields at 40× with a total number of >4000 CMs per animal were analyzed.

The harvested mice heart tissues were homogenized in a modified RIPA buffer (50 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM EDTA pH 8.0, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS) or SDS-Urea buffer (for the detection of MLC2 and phospho-MLC2; 1% SDS, 8 mM urea, 10 mM Tris pH 7.5, 140 mM NaCl, 5 mM EDTA, 2 mM EGTA), containing protease inhibitor (11697498001, Roche Diagnostics) and phosphatase inhibitor cocktails II and III (P5726 and P0044, Sigma-Aldrich). After rotating at 4°C for 2 h, the mixtures were centrifuged at 13,000 g for 15 min at 4°C. Primary antibodies used were: phospho-MLC2 (1:1000, 3674, Cell Signaling Technology), MLC2 (1:1000, 3672, Cell Signaling Technology), myosin IIb (1:1000, M7939, Sigma-Aldrich), ROCK2 (1:1000, HPA007459, Sigma-Aldrich Atlas Antibodies), GFP phospho-MYPT1 (1:1000, ABS45, Millipore), MYPT1 (1:1000, 612164, BD Biosciences), α5 integrin (1:200, sc-166681, Santa Cruz Biotechnology), pVCL (1:1000, ab200825, Abcam), VCL (1:1000, V4505, Sigma-Aldrich), N-cadherin (1:1000, 610921, BD Biosciences), GAPDH (1:3000, 6c5, RDI), Yap (1:1000, 4912, Cell Signaling Technology), phospho-Yap (Y357) (1:1000, ab62751, Abcam), phospho-Yap (S127) (1:1000, 4911, Cell Signaling Technology), FN (1:3000, PA1-23693, Thermo Fisher Scientific), desmin (1:1000, 32362, Abcam). For normalization of signals, blots were also analyzed with anti-GAPDH antibody followed by IRDye 680- or IRDye 800CW-conjugated secondary antibody (1:15,000, LI-COR). Anti-mouse-HRP (1:3000, 1706516, Bio-Rad) or anti-rabbit-HRP secondary antibodies (1:3000, 1706515, Bio-Rad) were also used and developed by SuperSignal West Pico PLUS Chemiluminescent Substrate (Thermo Scientific). Membranes were imaged with Odyssey Infrared Imaging System (LI-COR) or ChemiDoc Imaging System (Bio-Rad). Quantification was performed with Fiji ImageJ 1.53 software.

DOC soluble and insoluble fractions of harvested heart tissues were prepared by a modified method as previously reported (Wierzbicka-Patynowski et al., 2004). Briefly, fresh heart tissues were homogenized in DOC lysis buffer (Tris-HCl pH 8.8, 20 mM, EDTA 2 mM, PMSF 2 mM, iodoacetic acid 2 mM, N-ethylmaleimide 2 mM, DOC 2%) with protease inhibitors (11697498001, Roche Diagnostics), and the lysates were centrifuged at 13,000 g 4°C for 15 min. The supernatants were saved as the DOC soluble fraction. The pellets were re-suspended in SDS solubilization buffer (Tris-HCl pH 8.8, 20 mM, EDTA 2 mM, PMSF 2 mM, iodoacetic acid 2 mM, N-ethylmaleimide 2 mM, SDS 1%) with protease inhibitors (11697498001, Roche Diagnostics). The protein concentration of DOC soluble fractions was measured by BCA assay (23225, Thermo Fisher Scientific). DOC insoluble sample volume for SDS-PAGE is based on protein concentration in the corresponding DOC soluble fraction.

For quantification of EdU/pHH3/nuclear Yap-positive CMs in immunofluorescence images, a minimum of ten fields (40×) per heart were analyzed using programmed analysis functions in Nikon NIS-Element AR software (Version 5.21.03). α-Actinin served as a CM marker and DAPI was used to stain nuclei. CMs in which Yap staining overlapped with DAPI were counted as nuclear Yap-positive CMs. CMs in which pHH3 or EdU staining overlapped with DAPI were counted as pHH3- or EdU-positive CMs. Total number of DAPI within α-actinin-positive cells was regarded as the total number of CMs. Ratios of nuclear Yap-, pHH3- and EdU- positive CMs to total CMs were calculated and plotted in GraphPad Prism 9.

For clonal analysis of mosaic female ROCK hearts, ten random clones per heart from each group (n=5 hearts/group) were imaged at 40× and analyzed using Nikon NIS-Element AR software (version 5.21.03). A contiguous GFP-positive area was traced to represent a clone. The total number of GFP-positive, DAPI-positive cells was regarded as the total number of CMs per clone. Within the GFP-positive clonal area, EdU-positive, GFP-positive, DAPI-positive CMs were counted and the percentage of EdU-positive CMs per clone was calculated and plotted in GraphPad Prism 9.

Total mRNA was isolated from hearts with TRIzol (Life Technologies) and precipitated with isopropyl alcohol. The concentration of isolated mRNA was determined by NanoDrop followed by treatment with the TURBO DNA-free™ kit (Life Technologies) to remove contaminating DNA. cDNA was synthesized from 1 μg of total mRNA with the LunaScript^® RT SuperMix Kit (New England Biolabs) following the manufacturer's protocol. qRT-PCR was performed using the QuantStudio™ 5 System (Thermo Fisher Scientific) and iTaq Universal SYBR^® Green Supermix (Bio-Rad) according to the manufacturers’ instructions. Specific primers designed for RT-PCR were as follows: GAPDH forward 5′-CATCTTCCAGGAGCGAGACC-3′, GAPDH reverse 5′-CTCGTGGTTCACACCCATCA-3′, CTGF forward 5′-GAACAAATGCTGTGCAGGTGA-3′, CTGF reverse 5′-TCCTGGTAGGAATCGGACCTT-3′, CYR61 forward 5′-CTGCGCTAAACAACTCAACGA-3′, CYR61 reverse 5′-GCAGATCCCTTTCAGAGCGG-3′. For each qRT-PCR, a standard curve for each gene assay was used to account for the difference in PCR efficiencies between each assay reaction. For each target gene, raw data were normalized to Gapdh mRNA level expressed as 2^−deltaCt.

Paraffin-embedded samples were sectioned at 5 μm thickness. The sections were deparaffinized and stained with WGA (1:400, W11261 and W11262, Invitrogen) following the manufacturer's protocol. Fluorescent images were obtained using a Nikon A1R confocal microscope system. Images were elaborated and analyzed using Nikon NIS-Elements software.

Statistical differences were assessed by unpaired, two-tailed Student's t-test or one-way ANOVA followed by Tukey's multiple comparison of individual means. A P<0.05 was considered statistically significant.

We thank Richard Hynes for the floxed Itga5 mice provided by Sophie Astrof. We thank the Genomics and Pathology core facilities at Brown University, supported in part by the National Institutes of Health (NIGMS P30GM103410 and NIEHS P42ES013660).