Elevated BMP and mechanical signaling through YAP1/RhoA Poises FOP Mesenchymal Progenitors for Osteogenesis

Abstract

Fibrodysplasia ossificans progressiva (FOP) is a rare genetic disease characterized by the formation of extra-skeletal bone, or heterotopic ossification (HO), in soft connective tissues such as skeletal muscle. All familial and sporadic cases with a classic clinical presentation of FOP carry a gain-of-function mutation (R206H; c.617G>A) in ACVR1, a cell surface receptor that mediates bone morphogenetic protein (BMP) signaling. The BMP signaling pathway is recognized for its chondro/osteogenic-induction potential, and HO in FOP patients forms ectopic but qualitatively normal endochondral bone tissue through misdirected cell fate decisions by tissue-resident mesenchymal stem cells. In addition to biochemical ligand-receptor signaling, mechanical cues from the physical environment are transduced to activate intracellular signaling, a process known as mechanotransduction, and can influence cell fates. Utilizing an established mesenchymal stem cell model of mouse embryonic fibroblasts (MEFs) from the Acvr1^R206H/+ mouse model that mimics the human disease, we demonstrated that activation of the mechanotransductive effectors Rho/ROCK and YAP1 are increased in Acvr1^R206H/+ cells. We show that on softer substrates, a condition associated with low mechanical signaling, the morphology of Acvr1^R206H/+ cells is similar to the morphology of control Acvr1^+/+ cells on stiffer substrates, a condition that activates mechanotransduction. We further determined that Acvr1^R206H/+ cells are poised for osteogenic differentiation, expressing increased levels of chondro/osteogenic markers as compared to Acvr1^+/+ cells. We also identified increased YAP1 nuclear localization in Acvr1^R206H/+ cells, which can be rescued by either BMP inhibition or Rho antagonism. Our results establish RhoA and YAP1 signaling as modulators of mechanotransduction in FOP and suggest that aberrant mechanical signals, combined with and as a result of the increased BMP pathway signaling through mutant ACVR1, lead to misinterpretation of the cellular microenvironment and a heightened sensitivity to mechanical stimuli that promotes commitment of Acvr1^R206H/+ progenitor cells to chondro/osteogenic lineages.

Introduction

Each of the many tissue types within the human body have highly specialized functions that are enabled by their specific physical properties ^{(1, 2)}. Bones support the transfer of large forces to enable human motion, and so are stiff and strong, relative to adipose tissue which does not play a loadbearing role ⁽²⁾. In order to achieve and maintain the properties required for their function, cells within these tissues sense and interpret their physical environment. Cells perceive the context of their surroundings through environmental cues, such as substrate stiffness and mechanical loading of the matrix, via mechanoreceptors on the membrane surface. These mechanical inputs can modulate the morphology and the fate decisions of cells through mechanotransductive signaling pathways that ultimately result in modification of chromatin organization and gene expression ^(3–5), implicating stiffness in impacting cell fate. Cell fates can be controlled by substrate stiffness comparably to induction through exogenous ligands or growth factors: softer substrates support adipogenesis and myogenesis, while stiffer substrates support chondrogenesis and osteogenesis ^{(2, 3, 6)}.

Evidence for increased mechanotransduction has been implicated in both non-genetic and genetic diseases ^(7–9). The formation of endochondral bone within soft connective tissues, called heterotopic ossification (HO), could be in part caused by increased mechanotransduction, stimulating a cell population to interpret its substrate as stiffer and modify its genetic profile in response ^{(8, 10)}. Fibrodysplasia ossificans progressiva (FOP; MIM #135100) is a rare human genetic disease characterized by the formation of HO within muscle and other soft connective tissues ^(11–14). 97% of patients with a classic clinical presentation of FOP have the same autosomal dominant mutation (R206H; c.617G>A) in the type l bone morphogenetic protein (BMP) receptor activin A receptor type 1 (ACVR1) ^{(15, 16)}. FOP patients present with HO within skeletal muscle and other soft connective tissues within the first five years of life ⁽¹³⁾. While this genetic form of HO is rare, HO arises more commonly from blast-initiated injuries and joint-replacement surgeries ⁽¹⁷⁾, making understanding the mechanism of HO formation highly relevant for the general population. Developing improved insight into the mechanical signals and pathways that regulate the functions of cells and bone formation may identify targets for therapeutic intervention in both genetic and non-genetic causes of HO. We previously established that HO progression driven by the FOP ACVR1^R206H mutation is influenced by disrupted mechanotransduction ⁽⁸⁾, but the specific mechanism as to how this contributes to the disease pathology of FOP is still not well understood.

Here we demonstrate that YAP-associated protein (YAP1) signaling is a main contributing factor in this process. The YAP signaling pathway ^{(18, 19)} is regulated by ECM stiffness and cell geometry, and is a key regulator of cell differentiation ^(20–23). YAP, and its paralogue TAZ, are key factors directing MSC lineage commitment ^{(24, 25)}. Phosphorylation of YAP promotes its cytoplasmic localization, preventing YAP-mediated transcriptional activation in the nucleus ⁽²⁰⁾. Cytoplasmic YAP is associated with a soft surrounding ECM, cell cycle arrest, and adipogenic conditions, while translocation into the nucleus occurs in response to a stiffer ECM, proliferation, and osteogenic condition ^{(20–23, 25)}.

Another intracellular mechanotransductive pathway, Rho GTPase, regulates downstream effectors such as Rho kinase ⁽²⁶⁾, necessary for cell migration, adhesion, and differentiation ⁽²⁷⁾. Rho signaling through ROCK stimulates actin polymerization, a vital part of cell contractility and cellular mechanotransduction ⁽²⁸⁾. One of the Rho GTPases, RhoA, regulates ROCK to influence actin filament stability through myosin light chain (MLC) and cofilin ^{(29, 30)}. Activation of RhoA in mesenchymal cells largely contributes to their chondro/osteogenic cellular identity ^{(27, 31)}. Osteogenic conditions increase cell spreading, ECM production, BMP signaling, RhoA activation, and nuclear localization of YAP1^(31–33). This suggests that elevated signaling by both YAP1 and BMP pathways could coordinately promote the enhanced chondro/osteogenic differentiation that occurs in FOP. YAP1 responds to cell-cell contact and contractility signals mediated by Rho ^{(34, 35)}, suggesting an intersection between RhoA, YAP1, and BMP pathway signaling ^(35–37). Interestingly, basal activation of BMP signaling pathways, even in the absence of ligand, also regulates cell contractility in mesenchymal stem cells ^(38–40), further supporting that the FOP ACVR1^R206H mutation could instigate aberrant mechano-signaling in FOP progenitor cells.

In this study, we utilized mouse embryonic fibroblasts (MEFs) isolated from a knock-in Acvr1^R206H/+ mouse model ^{(41, 42)} that recapitulates the human disease progression to examine the YAP1 and Rho/ROCK mechano-signaling molecular pathways and investigate the ability of cells expressing the FOP mutation to properly sense and respond to the mechanical cues in their microenvironment. MEFs are used as an in vitro model system of mesenchymal stem cells (MSCs), including their ability to differentiate into adipogenic, chondrogenic, and osteogenic lineages ⁽⁴³⁾. We previously showed increased BMP pathway signaling in FOP patient-derived stem cells from human exfoliated deciduous teeth (SHED cells) ⁽⁴⁴⁾ and Acvr1^R206H/+ MEFs ⁽⁴³⁾ as measured by phosphorylated Smad1/5/8 (pSmad1/5/8) protein levels in the presence or absence of BMP ligand. Thus, BMP pathway signaling is increased downstream due to enhanced activity of ACVR1. Our data support that the ability of cells to sense their environment and properly signal through mechanical effectors is altered by the FOP ACVR1^R206H mutation, leading to increased chondro/osteogenic cell fate decisions.

Materials and Methods

Mice

Cells (mouse embryonic fibroblasts, MEFs) for in vitro experiments were obtained from the conditional-on knock-in mouse Acvr1^[R206H]FlEx was used to generate doxycycline-inducible global allele expression Acvr1^{[R206H]FlEx/+};Gt(ROSA)26Sor^{tm1(rtTA*M2)Jae};Tg(tetO-Cre)1Jaw mice (which we refer to as Acvr1^R206H/+), as described ^{(41, 42)}. Acvr1^+/+ controls were littermates that did not contain an Acvr1^R206H allele (as verified by PCR genotyping, primers in Supplementary Table 1). To induce recombination and global expression of the mutant allele, 4-week-old mice were provided a doxycycline diet chow (625mg/kg, Envigo RMS Inc., TD 01306) for at least three consecutive days. Cre- recombination was verified by PCR. Littermate Acvr1^+/+ controls were also treated with doxycycline chow. All procedures were reviewed and approved by the Institutional Animal Care and Use Committee at University of Pennsylvania.

Preparation of polyacrylamide (PA)-hydrogels

Stiffness of native tissues were mimicked using an adjustable polyacrylamide (PA) hydrogel system. PA hydrogels were prepared as described ^{(8, 45)}. Elastic moduli of the gels were verified through AFM force spectroscopy ⁽⁴⁶⁾. Fibronectin (20 mg/mL, Sigma Aldrich, catalog F1141) coating of gel surfaces was accomplished after treatment with 2 mg/mL sulfo-SANPAH (No. 22589; Pierce Protein Biology/Life Technologies, Rockford, IL) as previously described in ⁽⁴⁵⁾.

Generation of mouse embryonic fibroblasts (MEFs) and cell culture

Knock-in Acvr1^R206H/+ MEFs were isolated at embryonic day 13.5 (E13.5) as previously described ⁽⁴³⁾ from mice described in ⁽⁴¹⁾. Genotypes of cell lines (MEFs) and presence of the R206H mutation were confirmed by PCR (primers in Supplementary Table 1). Acvr1^+/+ controls were littermates that did not contain an Acvr1^R206H allele. Cells were cultured in Dulbecco’s modified Eagle’s medium, high glucose (DMEM, Gibco, catalog 11885–084) containing 10% fetal calf serum (FCS, Gibco, catalog 10438034) before being placed on PA gels and then in DMEM/1%FCS in basal media experiments on PA gels. For mixed media experiments, cells were cultured in bipotential adipogenic:osteogenic media (1:1) as described ⁽⁴⁷⁾ (without addition of BMP ligand) for 72 hours on PA gels and then harvested for qRT-PCR analysis.

For mechanosensing assays, subconfluent MEFs were seeded onto freshly prepared PA-hydrogels (described above) at 5×10³ cells/cm². In some experiments, cells were treated with 50nM LDN-193189 (Sigma, catalog SML0559) for 2 hours, 10uM Y272362 (Tocris, catalog 1254) for 1 hour, or 2μg/ml CNO₃ (Cytoskeleton Inc, Denver CO, catalog CNO3-A) for 3 hours (all in DMEM/1%FCS). After 18 hours, cells on hydrogels were fixed in 4% paraformaldehyde/PBS (Thermo Fisher Scientific, catalog J19943-K2), followed by Alexa Fluor 488 Phalloidin (Invitrogen, catalog A12379, 1:1000) to visualize the actin cytoskeleton and DAPI Fluoromount-G (SouthernBiotech, catalog 0100–20) for nucleus staining as previously described in ⁽⁸⁾. YAP1 (Abcam, catalog 52771, 1:200) protein was detected with Alexa Fluor 546 (Invitrogen, catalog A11035, 1:1000). Cell contractility was evaluated with primary paxillin antibody (Cell Signaling Technologies, catalog 3674, 1:500) followed by detection with Alexa Fluor 594 anti-rabbit antibody (Invitrogen, catalog A22107MP). Imaging used an Eclipse 90i microscope (Nikon) at consistent exposure times. Cell morphology parameters, YAP1 localization, and focal adhesion number were analyzed in ImageJ. Graphs of cell morphology measurements and YAP1 localization measurements represent the average of 3–4 individual experimental means, with each mean of a single experiment representing the mean of individual cell measurements. For YAP1 localization, nuclear images were utilized to delineate the nuclear area from the cytoplasmic region, and the average fluorescent intensity over each region was calculated using Image J.

Small interfering RNA (siRNA) transfection

MEFs (5×10³ cells/cm²) were cultured in DMEM/10%FCS for 24 hours then were transfected in DMEM/1%FCS for 24 hours with 2 mg/mL of siRNAs targeting YAP1 mRNA (siYAP1: 5’-GGCCAGAGAUAUUUCCUUATT-3’ and 5’-UAAGGAAAUAUCUCUGGCCTT-3’) (synthesized by Dharmacon, Inc) using Lipofectamine RNAimax (Invitrogen, catalog 1875238) following the manufacturers’ instructions. Knock-down efficiency was validated by RT-PCR for YAP1 mRNA and by Western Blot for YAP1 protein (Supplementary Fig. 1).

Western Blotting

MEFs were recovered from PA gels using Accumax (Sigma, catalog A7089) for 10 minutes at room temperature. Cells were lysed in buffer (RIPA, Sigma-Aldrich, catalog R0278) supplemented with Halt Protease and Halt Phosphatase Inhibitor Cocktails (Thermo Fisher Scientific, catalog 87785), cleared by centrifugation and quantified using BCA Protein Kit (Thermo Fisher Scientific, catalog 23227). Proteins were electrophoresed through 4–12% Bis-Tris protein gels (Thermo Fisher Scientific, catalog 12313623) and transferred to polyvinylidene difluoride (PVDF) membranes (Thermo Fisher Scientific, catalog 88585). Membranes were blocked in Odyssey Blocking Buffer (LI-COR, catalog 927–40100) and incubated overnight at 4°C with primary antibody for YAP1 or p-YAP1 or RhoA, and Gapdh (Abcam, catalog 56701, 1:1000; Cell Signaling Technologies, 4911S, 1:1000; Novus Biologicals, 1:500, NB100–91273; Cell Signaling Technologies, 5174S, 1:2500, respectively) then with IR Dye 800CW goat anti-rabbit IgG (LI-COR Biosciences, catalog 926–32211, 1:10,000) in Odyssey blocking buffer for 1 hour at room temperature. Following washing with tris-buffered saline with Tween20 (TBS-T) 2 × 5minutes, membranes were imaged with Odyssey Infrared Imaging System (LI-COR Biosciences, model 9120) and protein levels quantified with Image Studio Lite 5.2 (LI-COR Biosciences). Phospho-YAP and total YAP values were normalized to Gapdh to correct for variation in protein loading.

RNA isolation and real-time RT-PCR analysis

RNA was isolated from MEF monolayers on PA gels using Accumax (Sigma, catalog A7089) for 10 min at room temperature, followed by TRIzol extraction (Thermo Fisher, catalog 15596026). Genomic DNA was removed from RNA samples by digestion with RNase-free DNase (Promega, catalog M6101). RNA concentration was determined by spectrophotometer (NanoDrop) and equivalent amounts for each sample were used for cDNA synthesis using High Capacity RNA-to-cDNA reagents (Applied Biosystems, catalog 4385612). Quantitative RT-PCR analysis was performed to detect mRNA expression of PPARγ, LP1, CEBPβ, Sox9, Runx2, Osx, and Gapdh. Real-time quantitative PCR reactions contained forward/reverse primers (0.37mM). cDNA (1:5 dilution), and Fast SYBR Green PCR Master Mix (Applied Biosystems; 4385612); each sample was analyzed in triplicate. Target gene mRNAs were quantified from standard curves and normalized to Gapdh followed by normalization to respective gene expression by Acvr1^+/+controls. Forward and reverse primer sequences are in Supporting Table 1.

Traction Force Microscopy (TFM)

Traction force microscopy was performed as described previously ^{(23, 38)} Briefly, prior to polymerization, polyacrylamide (PA) hydrogels were UV-cured to a modulus of 10kPa (as verified by AFM). Hydrogels were subsequently washed three times with PBS and incubated with fibronectin solution (20 μg/ml, F1141, Sigma Aldrich) for 1 hour. MEFs were seeded on PA hydrogels at 1000 cells/cm² and cultured for 18 hours in DMEM/10% FCS before carrying out traction force microscopy (TFM). Cells (phase) and embedded beads (0.2-μ-diameter fluorescent microspheres, F8810, Invitrogen, Carlsbad, CA) were imaged using ZEISS Axio AXIO Observer fluorescence microscope at 40X magnification. Image sequences for each cell were taken before and after cell lysis with lysis buffer (10% Sodium dodecyl sulfate and Triton X-100 at 1:100).

TFM data analysis was performed using a Fourier transform traction cytometry plugin in Image J and a custom MATLAB (The MathWorks) as in ^{(23, 38)}. In some studies, cells were treated with 10μM Fasudil (Sigma Aldrich, catalog CDS021620), a potent RhoA inhibitor ⁽⁴⁸⁾, for 30 min before measurement of traction forces.

Statistical analysis

Data were analyzed statistically using GraphPad (La Jolla, CA, USA) Prism 7 software. Results are presented as the mean ± SD or SEM. Paired or unpaired data sets were analyzed using two-tailed, unpaired, equal variance Student’s t test or one-way ANOVA (Tukey’s multiple comparison post-hoc test) to determine significance. Differences were considered statistically significant at p<0.05. Significance and sample size are indicated for each data set in the figure legends.

Results

Acvr1^R206H/+ cells misinterpret substrate rigidity through increased BMP pathway signaling.

To address how the stiffness of the cell microenvironment affects Acvr1^R206H/+ cells, Acvr1^+/+ and Acvr1^R206H/+ primary MEFs were cultured on polyacrylamide (PA) substrates mimicking adipogenic (5kPa), myogenic (10kPa), chondro/osteogenic (55kPa) stiffness. Thin layers of PA-hydrogels (h=100μm) were polymerized onto glass slides and then coated with fibronectin to aid cell attachment. Hydrogel stiffness was controlled by acrylamide content, and the stiffness verified by atomic force microscopy (AFM) ⁽⁴⁶⁾. Cell-cell contacts can override cell-ECM-induced mechanosignaling ⁽⁴⁹⁾, so MEFs were seeded at a low density. Cellular response to substrate stiffness, or mechanosensing, was quantified by measuring cell area. Acvr1^+/+ cells responded to increasing levels of stiffness appropriately, with cell spread area increasing with more apparent stress fibers, as substrate stiffness increased (Fig. 1A, upper half of panel). However, on softer substrates (5 and 10kPa), Acvr1^R206H/+ cells showed a similar morphology to control cells on stiffer substrates (55kPa) (Fig. 1A, lower half of panel). Differences were most pronounced at 10kPa (Fig. 1C), a stiffness comparable to normal skeletal muscle tissue ^{(6, 46)}, a site where HO commonly forms in FOP patients ⁽¹³⁾.

Figure 1. Acvr1^R206H/+ cells misinterpret substrate rigidity through increased BMP pathway signaling.

A) Response of Acvr1^+/+ and Acvr1^R206H/+ mouse embryonic fibroblasts (MEFs) to substrate stiffness corresponding to adipogenic, myogenic, and osteogenic tissues was tested using polyacrylamide (PA) hydrogels of 5, 10, and 55 kPa. Cells were seeded at low density and stained with phalloidin (green) and DAPI (blue). On soft (5 and 10kPa) substrates, Acvr1^R206H/+ cells showed a similar morphology to Acvr1^+/+ cells on stiffer (55kPa) substrates. B) Cells were treated with 50nM of the Acvr1 inhibitor LDN-193189 (LDN) for 2 hours, which notably decreased the spread morphology of Acvr1^R206H/+ cells on softer substrates. Scale bar= 100μm. C) Cell area, analyzed as a function of matrix elasticity and cell spreading, was quantified. Without LDN, cell area scaled with substrate rigidity in Acvr1^+/+ cells but not Acvr1^R206H/+ cells. This effect is ameliorated with the addition of LDN. Graph represents mean ± SEM from 4 independent experiments (>350 cells per experiment). Statistical significance determined by 1-way ANOVA; #p<0.0001.

To investigate whether BMP pathway signaling through mutant ACVR1 contributes to the spread morphology of Acvr1^R206H/+ cells on soft substrates, we treated Acvr1^+/+ and Acvr1^R206H/+ cells with the inhibitor LDN-193189 (LDN) ^{(50, 51)} in basal media on 5,10, and 55kPa substrates for 2 hours. LDN-treated Acvr1^R206H/+ cells were less spread than without inhibitor, with cell areas more similar to Acvr1^+/+, particularly cells on softer substrates (Fig. 1B and C). These data indicate that increased activity of ACVR1 signaling modulates cell spread area, and that increased BMP signaling through the ACVR1^R206H mutation contributes to the spread morphology of Acvr1^R206H/+ cells on softer substrates. During osteogenic differentiation, cells spread and increase their adhesions to their substrate ⁽³⁾. This is further increased when BMP is overexpressed ⁽⁵²⁾, indicating that BMP pathway signaling can modulate cell morphology. Our data support that increased BMP pathway signaling via ACVR1 influences Acvr1^R206H/+ cell interpretation of its substrate stiffness, with Acvr1^R206H/+ cells sensing and interpreting their environment as stiffer than it actually is, causing them to adopt a more spread cell morphology on softer substrates.

Acvr1^R206H/+ cells are poised for osteogenesis.

Mechanical (physical) cues to cells from their surrounding environments contribute to cell fates ^{(2, 6, 53)}. Our findings that elevated BMP pathway signaling through ACVR1 in Acvr1^R206H/+ cells causes them to respond to soft substrates as though they were stiffer (Fig. 1) suggested that the FOP mutation leads to promiscuous diversion from normal cell fate though misregulation of mechanosignaling pathways. We detected the greatest differences between Acvr1^+/+ and Acvr1^R206H/+ cell morphology on 10kPa substrates (Fig. 1C), which represents connective tissues like skeletal muscle where HO commonly forms in FOP patients ⁽¹³⁾. We therefore investigated gene expression of adipogenic and chondro/osteogenic markers by Acvr1^+/+ and Acvr1^R206H/+ cells on 10kPa (soft) PA gels.

Acvr1^+/+ and Acvr1^R206H/+ cells on 10kPa substrates cultured in basal media expressed similar mRNA levels of the adipogenic regulator PPARγ; however, Acvr1^+/+ cells expressed significantly higher levels of the adipogenic markers CEBPβ and LP1 compared to Acvr1^R206H/+ cells (Fig. 2A). Lipoprotein lipase 1 (LP1), peroxisome proliferator-activated receptor gamma (PPARγ), and CCAAT/enhancer-binding protein beta (CEBPβ) are all expressed in early adipogenesis ⁽⁵⁴⁾, with CEBPβ transactivating expression of PPARγ ⁽⁵⁵⁾. On a stiffer (55kPa) substrate, there was a slight increase in adipogenic gene expression in both Acvr1^+/+ and Acvr1^R206H/+ cells, but not at significant levels. We further investigated the expression of chondro/osteogenic genes (Sox9, Runx2, Osx). Sox9 marks the early stages of chondrogenesis ⁽⁵⁶⁾, while Runt-related transcription factor 2 (Runx2; alias CBFA1) marks the initial commitment of mesenchymal cells to the process of osteogenesis ^{(57, 58)}. Nuclear localization of Runx2 is essential for its ability to act as a transcription factor to induce osteogenic gene expression ⁽⁵⁸⁾, and we previously identified increased nuclear localization of Runx2 protein in Acvr1^R206H/+ cells compared to Acvr1^{+/+ (8)}. Osterix (Osx; alias Sp7) is a master-regulator required for formation of bone by osteoblasts ⁽⁵⁹⁾. We found that expression of these chondro/osteogenic genes were significantly increased in Acvr1^R206H/+ cells on soft substrates cultured in basal media as compared to Acvr1^+/+ cells, which further increased on 55kPa (Fig. 2A). These data support that Acvr1^R206H/+ cells may be poised towards an osteogenic lineage.

Figure 2. Acvr1 ^R206H/+ cells are poised for osteogenesis.

A) Cultures of Acvr1^+/+ and Acvr1^R206H/+ cells were maintained on 10kPa or 55kPa PA gels in basal media for 24 hours (as shown in the schematic above) and expression of adipogenic and chondro/osteogenic genes was quantified by RT-PCR. Adipogenic genes were expressed similarly in Acvr1^+/+ and Acvr1^R206H/+ cells, but expression of chondro/osteogenic genes were increased significantly in Acvr1^R206H/+ cells on both 10kPa and 55kPa (n=3 biologic samples run in triplicate). B) Cells were cultured on 10kPa or 55kPa PA gels in bipotential adipogenic:osteogenic media (1:1) for 72 hours (as shown in the lower schematic) and expression of adipogenic and chondro/osteogenic genes of Acvr1^+/+ and Acvr1^R206H/+ cells was quantified by RT-PCR. Expression levels of chondro/osteogenic genes were significantly higher in Acvr1^R206H/+ cells on both 10kPa and 55kPa (n=3 biologic samples run in triplicate). Data are relative to Acvr1^+/+. Graphs represent mean ± SEM. Significance determined by two-tailed Student’s t-test, *p<0.05, **p<0.01, #p<0.0001.

The modulus of substrate can influence stem cell spreading, traction generation, and fate, including in the absence of soluble differentiation factors ^{(2, 6)}. As described above, we investigated markers of adipogenic and osteogenic lineage to examine whether cells were predisposed toward one lineage or another in the absence of osteogenic or adipogenic inducers (Fig. 2A). To investigate whether control and mutant cells would show a bias toward one lineage over another, we cultured the cells in bipotential adipogenic:osteogenic (1:1) media that contained equivalent inducers of osteogenesis and adipogenesis ^{(47, 60)}. We previously showed that Acvr1^+/+ and Acvr1^R206H/+ cells differentiated to similar extents under adipogenic conditions, and that Acvr1^R206H/+ cells expressed higher levels of osteogenic markers than Acvr1^+/+ cells under osteogenic conditions ⁽⁴³⁾. We found that while exposure to adipogenic and osteogenic inducers increased markers for both lineages in both Acvr1^+/+ and Acvr1^R206H/+ cells, Acvr1^R206H/+ cells expressed chondro/osteogenic markers more robustly compared to Acvr1^+/+ cells (and compared to adipogenic markers) on both soft (10kPa) and stiff (55kPa) substrates (Fig. 2B). Acvr1^+/+ cells also showed significantly increased expression of the adipogenic genes PPARγ and LP1 under the mixed media conditions compared to Acvr1^R206H/+ cells (Fig. 2B).

Together, our data demonstrate that Acvr1^R206H/+ cells are poised for chondro/osteogenic differentiation in the absence of established biochemical or biomechanical cues, with a preference for osteogenic commitment when differentiation signals are provided, even on soft substrates that normally directs cells towards an adipogenic/myogenic lineage.

Acvr1^R206H/+ cells show increased nuclear YAP localization on soft substrates.

YAP signaling is influenced by substrate stiffness ^(20–23) and nuclear localization of YAP1 and activation of signaling is associated with a stiff extracellular environment, cell motility, and chondro/osteogenic differentiation ^(20–23). To investigate YAP signaling, we examined YAP1 protein localization via immunofluorescence and found increased nuclear YAP1 in Acvr1^R206H/+ cells on softer substrates compared to Acvr1^+/+ cells (Fig. 3A and C). In addition to increased cell area (Fig. 3D), two other measurements, circularity and aspect ratio, indicated that Acvr1^R206H/+ cells were more spread and less circular with more protrusions on 10 and 55kPa (Fig. 3E and F). Aspect ratio (AR) is the proportion of the width of an object to its height and circularity represents how close an object is to a perfect circle (AR=1). These data suggest that YAP1 contributes to the irregular spread cell morphology ⁽⁶¹⁾ of Acvr1^R206H/+ MEFs on softer substrates as identified in Fig. 1.

Figure 3. Acvr1^R206H/+ cells show increased nuclear YAP localization.

A) Acvr1^R206H/+ cells have more nuclear YAP on softer substrates (10kPa) compared to Acvr1^+/+ cells. Nuclear YAP1 is shown alone in monochromatic (cyan) columns, and DAPI (blue), phallodin (green), and YAP1 (red) are shown together in multicolor columns. B) Cells treated with siYAP1 have decreased nuclear YAP1 localization. Scale bar= 50μm. C) Quantification of YAP1 nuclear/cytoplasmic ratio with or without siYAP1. Quantification of cell morphology parameters with or without siYAP1 D) cell area, E) circularity, and F) aspect ratio. Circularity indicates how close an object is to a perfect circle. Aspect ratio is the proportion of the width of an object to its height. Graphs represent mean ± SEM from 4 biologic experiments (>50 cells per experiment). Statistical significance determined by 1-way ANOVA, *p<0.05, ***p<0.001, #p<0.0001.

To demonstrate the influence of YAP1 in cell interpretation of mechanical environment, we utilized a YAP1 siRNA (siYAP1) to knockdown YAP1 mRNA. After confirming YAP1 targeting with our siRNA (Supplementary Fig. 1), we examined changes in YAP1 localization and cell morphology. Nuclear YAP1 levels decreased in Acvr1^R206H/+ cells in the presence of siYAP1 (Fig. 3B and C), similarly to Acvr1^+/+ cells, and cellular morphology parameters (cell area, aspect ratio, and circularity) of Acvr1^R206H/+ cells were rescued and more similar to Acvr1^+/+ cells (Fig. 3D, E, F). Thus, decreasing AR and increasing circularity indicate that YAP1 inhibition mitigates increased mechanotransduction in Acvr1^R206H/+ cells.

Interaction between BMP pathway signaling and YAP1 signaling in Acvr1^R206H/+ cells.

The BMP signaling pathway has been previously established to communicate with the YAP signaling pathway ^{(37, 62)}. Given the increased nuclear localization of YAP1 in Acvr1^R206H/+ cells on soft substrates (Fig. 3A), we further investigated the relationship between BMP and YAP signaling pathways in Acvr1^+/+ and Acvr1^R206H/+ cells by examining YAP1 protein expression and localization in response to LDN-193189 (LDN). Inhibition by LDN decreased YAP1 nuclear localization in Acvr1^R206H/+ cells on softer (10 kPa) substrates to levels similar to Acvr1^+/+ cells (with or without LDN treatment) (Fig. 4A and B). These data are consistent with the effects on cell morphology by LDN-treated Acvr1^R206H/+ cells, which appear more similar to Acvr1^+/+ when BMP pathway signaling is inhibited (Fig. 1C). These data indicate that increased BMP pathway signaling conferred by enhanced activation of ACVR1 alters cell morphology and influences the localization of YAP1.

Figure 4. Interaction between BMP signaling and YAP1 signaling pathways in Acvr1^R206H/+ cells.

A) Inhibition of BMP signaling with Acvr1 inhibitor LDN-193189 (LDN) in basal media decreased nuclear YAP1. Yellow dotted line outlines cell nucleus. Scale bar= 100μm. B) Quantification of nuclear/cytoplasmic YAP1 localization in the presence or absence of LDN. C) pYAP (cytoplasmic YAP1) protein was quantified in Acvr1^+/+ and Acvr1^R206H/+ cells on 10 and 55kPa substrates. Graph represents the mean from three biologic replicates ± SEM from three biologic replicates done in triplicate. D) Expression of YAP/TAZ mRNA and the YAP target gene Cyr61 in Acvr1^+/+ and Acvr1^R206H/+ cells on 10 and 55kPa substrates for 24 hours in basal media were quantified by RT-PCR. Expression of YAP1 and Cyr61 was significantly expressed in Acvr1^R206H/+ cells on 10 and 55kPa substrates. Graphs represent the mean ± SEM from three biologic replicates done in triplicate. Data are relative to Acvr1^+/+. E) Quantification of total YAP1 protein in Acvr1^+/+ and Acvr1^R206H/+ cells on 10 and 55kPa substrates. Graph represents the mean from three biologic replicates ± SEM. Statistical significance determined by 1-way ANOVA *p<0.01, **p<0.01, #p<0.0001.

Progressively decreasing cytoplasmic pYAP as substrate stiffness increases is indicative of increased nuclear YAP1 localization and increased cellular spreading ^{(20, 21, 34)}. To determine the effects of Acvr1^R206H/+ on levels of YAP1 targeted for degradation, we analyzed phosphorylated-YAP (pYAP) protein levels in cells on 10 and 55kPa by Western blot. Both Acvr1^+/+ and Acvr1^R206H/+ cells showed decreased expression of pYAP on 55kPa as compared to 10kPa substrates (Fig. 4C). Acvr1^R206H/+ cells showed slightly more, although not statistically significant, decreased pYAP protein expression on both 10 and 55kPa compared to Acvr1^+/+ (Fig. 4C), consistent with the increased nuclear YAP1 in Acvr1^R206H/+ cells (Fig. 3C). These data were further correlated with increased mRNA expression of YAP1 and its target gene Cysteine-rich angiogenic inducer 61 (Cyr61) ^{(24, 63, 64)} (Fig. 4D), and increased expression of total YAP1 protein in Acvr1^R206H/+ cells (Fig. 4E). While total YAP1 was not significantly different between Acvr1^+/+ and Acvr1^R206H/+ cells, localization of YAP1 is more indicative of YAP signaling activity than overall protein abundance ^{(20, 65)}; nuclear YAP (active YAP) is increased in Acvr1^R206H/+ cells (Fig 4B). Taken together, these data suggest that increased cellular spreading of Acvr1^R206H/+ cells on soft substrates is due to increased nuclear localization of YAP1, causing Acvr1^R206H/+ misinterpretation of substrate stiffness.

Cell contractility is increased in in Acvr1^R206H/+ cells.

The mechanotransductive effector RhoA, which is associated with cell contractility and motility through actin cytoskeleton dynamics ^{(28, 30)}, is more highly expressed in Acvr1^R206H/+ cells (Supplementary Fig. 2), consistent with our previous observations in ⁽⁸⁾. RhoA facilitates phosphorylation of Smad1/5/8, connecting RhoA activation and BMP signaling as two critical pathways that impact cell fate decisions ^{(31, 66–68)}. Activation of Rho is required for the formation of focal adhesions and their associated stress fibers ^{(28, 30, 60, 69)}. Focal adhesions are located at the edges of the cellular membrane and attach cells to the ECM ^{(61, 70, 71)}. Focal adhesions are also regulated downstream from RhoA by YAP signaling ^{(22, 34, 35, 61, 72)}. To examine the effect of the increased RhoA expression by Acvr1^R206H/+ cells, we detected paxillin, part of the focal adhesion complex ^{(73, 74)} by immunofluorescence, and found an increased number of focal adhesions at the edges of Acvr1^R206H/+ cells on softer substrates compared to Acvr1^+/+ cells (Fig. 5A). Focal adhesions also appeared more clustered in Acvr1^R206H/+ cells compared to the more sparsely distributed focal adhesions in Acvr1^+/+cells (Fig. 5A). Elevated paxillin staining in Acvr1^R206H/+ indicates increased cellular adhesion as a consequence of increased RhoA signaling activity, implicating RhoA signaling through focal adhesion formation as a key player in improper mechano-sensation in Acvr1^R206H/+ cells.

Figure 5. Cell contractility increases in Acvr1^R206H/+ cells.

A) Acvr1^+/+ and Acvr1^R206H/+ cells were cultured in basal media on 10 and 55kPa substrates and detected for paxillin (part of the focal adhesion complex). There were an increased number of focal adhesions in Acvr1^R206H/+ cells on 10 and 55kPa substrates. Cells were also treated with the B) ROCK antagonist Y272362 (Y27) or C) ROCK agonist CNO₃. Cell area as a function of substrate stiffness is decreased with Y27 addition (Bii) and increased with CNO₃ addition (Cii). Total number of focal adhesions decreased when Y27 was added (Biii) and increased with CNO₃ (Ciii). D) Cells treated with Y27 have decreased nuclear YAP1 localization. Scale bar= 100μm. Graphs represent the mean ± SEM (SD for focal adhesion number) of >50 cells analyzed from three biologic replicates. Statistical significance determined by 1-way ANOVA, **p<0.01, ***p<0.001, #p<0.0001.

In order to investigate the impact of Rho signaling on focal adhesion complex formation and mechanotransduction in Acvr1^+/+ and Acvr1^R206H/+ cells, we treated cells with 10μM of the ROCK inhibitor Y72362 (Y27) ⁽⁷⁵⁾. All cells responded by significantly decreasing paxillin staining and cell area (Fig. 5B ii and iii) relative to untreated cells (Fig. 5A ii and iii). We also found that YAP1 protein was more localized to the cytoplasm on both 10 and 55kPa in Acvr1^+/+ and Acvr1^R206H/+ cells with Y27 (Fig. 5D). These data support interactions between YAP and RhoA signaling pathways to regulate cell shape and spreading.

To probe this aspect further, we augmented cell contractility with the RhoA-agonist CNO₃ ⁽⁷⁶⁾ and determined that both Acvr1^+/+ and Acvr1^R206H/+ cells spread more with increased paxillin staining and more clustered focal adhesions at cell edges on all stiffnesses (Fig. 5C). Relative to control cells, Acvr1^R206H/+ cells treated with CNO₃ were more spread, with more paxillin staining, on softer 10kPa substrates (Fig. 5C) and were often detached from the substrate on stiffer 55kPa substrates. On softer substrates, RhoA and YAP1 signaling pathways are normally less active and cells are smaller and rounder, however we found that in Acvr1^R206H/+ cells on soft substrates, these two pathways were over-active, suggesting increased cell contractility.

To further directly assess cell contractility in Acvr1^+/+ and Acvr1^R206H/+ cells, we used traction force microscopy (TFM) ⁽⁷⁷⁾ and determined that Acvr1^R206H/+ cells exert a higher traction stress on their substrates, and generate a higher total force per cell (Fig. 6A). To examine the cellular response and traction force in response of a potent ROCK inhibitor Fasudil ⁽⁴⁸⁾, we cultured the cells with 10μM of Fasudil, for 30 min. Our TFM measurements demonstrated that Fasudil reduced the relative traction stress of Acvr1^R206H/+ cells on 10kPa to levels by Acvr1^+/+ cells (Fig. 6B). Taken together, these data indicate that cellular contractility is increased in Acvr1^R206H/+ cells, which can be rescued with ROCK inhibition via Fasudil, presenting a possible therapeutic target in the treatment of FOP.

Figure 6. Acvr1^R206H/+ cells generate higher levels of contractile forces.

A) Representative traction stress vector maps for Acvr1^+/+ and Acvr1^R206H/+ cells on 10kPa substrates in basal media. Quantification shows average traction stress and average traction force per cell. Significance determined by two-tailed Student’s t-test, *p<0.05,***p<0.001. B) Representative traction stress vector maps for Acvr1^+/+ and Acvr1^R206H/+ cells on 10kPa substrates with the ROCK inhibitor Fasudil. Quantification shows relative average traction stress and average traction force per cell normalized to Acvr1^+/+ without Fasudil. Contractile forces were decreased in cells treated with Fasudil. Scale bar= 20μm. Statistical significance determined by 1-way ANOVA, **p<0.01, ***p<0.001, #p<0.0001 (n>15 cells/group, mean ± SD).

Discussion

Cellular behavior (proliferation, differentiation, motility, protein production) is regulated in large part by the surrounding environment (including ECM, exogenous ligands, and small molecules) ^{(2, 3, 6, 78)}. Many signals from the cell environment, including the physical stiffness of tissues, are transduced through mechanical signaling effectors to influence gene expression and protein secretion ^{(3, 4)}. The cellular interpretation of such cues, termed mechanotransduction, is critical for cell response to its environment ^{(2, 6)}. The rare genetic disease fibrodysplasia ossificans progressiva (FOP) is caused by mutations in the BMP receptor ACVR1 that lead to increased BMP pathway signaling activity and the formation of endochondral bone tissue within soft connective tissues, including skeletal muscles, as a result of misdirection of cell fate decisions ^(11–14). Utilizing cells from a knock-in Acvr1^R206H/+ mouse model ^{(41, 42)}, we show here that the ACVR1^R206H mutation poises cells for chondro/osteogenic differentiation in the absence of inductive factors. In this basal, unstimulated state, we additionally found that mechanotransduction through the YAP1 and RhoA signaling pathways is enhanced, leading to increased cellular contractility, and indicating that cellular perception of environment is altered in Acvr1^R206H/+ cells. The combined effects of enhanced BMP/Rho/YAP1 pathway signaling combine to create a permissive cell state for chondro/osteogenic differentiation.

Of note, although HO forms in skeletal muscles and other soft connective tissues, it has not been observed in some skeletal muscles such as the tongue or diaphragm muscles in FOP ⁽¹²⁾. It is reasonable to speculate that these tissues are spared from the effects of the FOP mutation or are less susceptible due to a requirement for a permissive environment that can support progenitor cells with osteogenic potential for HO formation; specific types of muscle (fast vs slow twitch), movement (isometric vs isotonic) or other environment forces/factors may be needed to curate an environment conducive to HO formation. In this study, we focused on skeletal muscle stiffness due to the high incidence of HO in this tissue ⁽¹⁵⁾. Future studies will address the implications of muscle type and function on the incidence of HO formation in FOP, as well as addressing factors that contribute to HO in additional sites such as tendon and ligaments.

Increased cellular spreading is observed in cells exposed to osteogenic conditions and/or cells cultured on stiff matrices ⁽⁷⁸⁾. We observed that Acvr1^R206H/+ cells have an atypical spread morphology on softer, skeletal muscle tissue-like (10kPa), substrates. Further, these mutant cells show increased expression of chondro/osteogenic genes (Sox9, Runx2, Osx) in both basal and mixed adipogenic/osteogenic media, cellular responses that indicate improper interpretation of mechanical environment. We previously demonstrated that after injury in the FOP Acvr1^R206H/+ mouse model, the skeletal muscle tissue environment becomes stiffer than injured control muscle ⁽⁸⁾. Elevated tissue stiffness leads to increased activation of mechanotransductive signaling pathways, and promotes chondro/osteogenic differentiation over adipogenic/myogenic cell fates ^{(78, 79)}. Formation of HO in FOP patients is not continuous but episodic ^{(13, 15)}, indicating that although Acvr1^R206H/+ cells are in a state poised for chondro/osteogenic differentiation, a secondary condition appears necessary to reach a threshold to send them down this lineage pathway ⁽¹¹⁾. Interestingly, when we examined markers of adipogenesis (PPARγ, CEBPβ, LP1), we found a significantly increased expression of CEBPβ in Acvr1^+/+ cells on soft (10kPa) substrates relative to Acvr1^R206H/+ cells. CCAAT/enhancer-binding protein beta (CEBPβ) is crucial in macrophage-mediated skeletal muscle repair ⁽⁸⁰⁾, and we previously demonstrated that muscle repairs more slowly and less efficiently in FOP skeletal muscle after injury ⁽⁸¹⁾. This provides further support that the ACVR1^R206H mutation poises cells toward a chondro/osteogenic lineage, and away from an adipogenic/myogenic one.

The YAP/TAZ complex, part of the Hippo signaling pathway ^{(18, 19)}, is regulated by ECM elasticity and cell geometry, and is a key intracellular regulator of cell differentiation ^(20–23). YAP/TAZ proteins shuttle between the cytoplasm and nucleus in response to substrate stiffness and other environmental cues. Cytoplasmic localization of YAP/TAZ is associated with a soft surrounding ECM and adipogenic conditions, while YAP translocation to the nucleus occurs with a stiffer ECM, proliferation, and osteogenic conditions ^(20–23). We found that nuclear YAP1, an indicator of activated pathway signaling, is increased in Acvr1^R206H/+ cells on softer substrates relative to Acvr1^+/+ cells. Interestingly, this activation could be blocked by the inhibitor LDN-193189 (LDN), supporting that BMP pathway signaling through increased activity of ACVR1 and YAP signaling crosstalk in Acvr1^R206H/+ cells. LDN is commonly used to inhibit BMP-pSmad1/5/8 signaling ^{(50, 51, 82)}, however, side effects on pSmad2/3 signaling have also been reported in some cell types ^(83–85), we cannot exclude an influence from this pathway as well. This increased nuclear YAP correlated with less cytoplasmic (inactive) phosphorylated-YAP (pYAP) in Acvr1^R206H/+ cells on all substrates. Expression of the pro-osteogenic YAP1 target gene Cyr61, an ECM-associated signaling protein that is a key regulator of cell adhesion, migration, proliferation, and osteogenic differentiation ^{(24, 63, 64)}, was also increased, suggesting that Cyr61 may be a downstream mediator of YAP1 regulation of cell differentiation and stimulation of chondro/osteogenic gene expression. BMP and YAP1 pathway signaling have both been shown to promote osteogenesis under normal conditions ^{(25, 31, 86)}; our data support that increased signaling in both pathways may promote abnormal cell fate decisions (i.e., bone within skeletal muscle tissue).

The Rho/ROCK pathway provides a link to the extracellular matrix to affect cellular morphology, migration, contractility, and adhesion to surface substrates ^{(4, 26, 71, 87)}. RhoA is a small GTPase that regulates its kinase effector ROCK to influence actin filament polymerization through myosin light chain and Rho/ROCK tension-sensing of the ECM through formation of focal adhesions at the termini of actin filaments ^{(87, 88)}. RhoA facilitates phosphorylation of Smad1/5/8, downstream effectors of activated BMP receptors ^{(31, 33, 89, 90)}; YAP1 also responds to cell-cell contact and contractility signals mediated by Rho ^{(34, 35)}, suggesting an intersection between RhoA, YAP1, and BMP pathway signaling ⁽³⁶⁾. We previously demonstrated that the Rho/ROCK signaling pathway was over-activated in Acvr1^R206H/+ cells ⁽⁸⁾, and here we investigated its interactions with YAP1 and the impact on cell contractility. YAP1 signaling has been shown to control focal adhesion formation downstream of the RhoA pathway, thereby impacting cell morphological parameters ^{(20, 61)}. We quantified focal adhesions to assess Rho/ROCK mechanotransduction in Acvr1^R206H/+ cells, and identified an increase in focal adhesion formation and clustering in Acvr1^R206H/+ cells on soft substrates. A larger number of focal adhesions correlates with cellular spreading and increased adhesion strength ^(91–93); thus, the increased number and length of focal adhesions in Acvr1^R206H/+ cells supports increased cellular contractility and increased total force generated by these cells. ROCK inhibition by the ROCK inhibitor Y272362 (Y27) also decreased nuclear YAP1 localization, confirming that interaction of these pathways contributes to the improper mechanotransduction by Acvr1^R206H/+ cells on soft substrates.

Another contributing factor to osteogenic cell fate decisions is increased cell contractility ⁽⁹⁴⁾. Increased substrate stiffness leads to the formation of more focal adhesions to better anchor the cell to its substrate. Tension in these contractile actomyosin fibers increases activity of downstream effectors like YAP, resulting in osteogenic gene expression ^{(25, 95)}. Traction force microscopy (TFM) evaluates the extent to which a cell deforms its underlying ECM-coated substrate. From this, one can calculate the total force generated by a single cell and the average traction stress (force per unit area) it applies to the substrate ⁽⁷⁷⁾. A higher average traction stress indicates that a cell is in a more contractile state ^{(94, 95)}. TFM showed that Acvr1^R206H/+ cells generated higher contractile forces compared to Acvr1^+/+ cells on 10kPa (soft) substrate, providing further evidence that Acvr1^R206H/+ cells interpret their substrate as stiffer than Acvr1^+/+ cells do. This response by Acvr1^R206H/+ cells was ameliorated by the ROCK inhibitor Fasudil. Fasudil has been demonstrated as a potential therapeutic in several diseases ^(96–101) and is currently in clinical trials for treatment of cardiac disease ⁽¹⁰²⁾ and subarachnoid hemorrhage⁽¹⁰³⁾. This raises the potential for repurposing this treatment as a possible therapeutic for FOP and other forms of HO, in combination with other treatment approaches. Further studies investigating the impact of small molecule inhibitors of mechanotransduction, such as Fasudil, on the formation of heterotopic ossification in Acvr1^R206H/+ mouse models will help determine if this is a plausible new therapeutic target for future treatment of FOP and potentially in other HO patients.

While our studies provide important insight into the molecular pathophysiology of the mutant ACVR1 receptor on mechanotransductive signaling pathways in FOP, interpretations of these data are limited in part by experimental considerations. In our studies, we used MEFs as an experimentally amenable MSC model system. MEFs can be readily isolated in sufficient numbers for experiments, respond consistently in vitro, and we and others have demonstrated that MEFs are able to adopt multiple mesenchymal cell fates, including osteogenic ^{(43, 104)}. Within skeletal muscle tissue, fibro/adipogenic progenitor cells (FAPs) have been identified as an endogenous mesenchymal cell population with osteogenic potential that can contribute to the formation of HO ⁽¹⁰⁵⁾. Future studies are needed to address the mechanical signaling in Acvr1^R206H/+ FAPs, and the impact on chondro/osteogenic potential. Additionally, over twenty BMP family ligands ⁽¹⁰⁶⁾ have been identified and different ligands may activate various combinations of type I and type II receptors to induce different downstream interactions with mechanotransductive pathways. We previously demonstrated that the ACVR1^R206H mutation acts downstream of BMP ligands as well as being capable of stimulating increased SMAD1/5/8 phosphorylation independently of BMP ligand ⁽¹⁰⁷⁾. Here we demonstrate a mechanism by which increased BMP pathway signaling through ACVR1 mis-regulates mechanotransduction to poise cells for chondro/osteogenic differentiation independently of ligand. Future studies aiming to address ACVR1 responsiveness to specific ligands will provide further insights.

Taken together, our data elucidate the mechanism by which elevated BMP pathway and mechanical signaling combine to increase cell contractility and direct cells toward a chondro/osteogenic lineage pathway (depicted in the model in Fig. 7), contributing to the progression of disease in FOP. We showed the efficiency of several inhibitors in mitigating the aberrant mechanotransduction in Acvr1^R206H/+ cells (Fig. 4), and the downstream effect on cell contractility (Fig. 5 and 6). We show that altered mechanosensing by Acvr1^R206H/+ cells is downstream of increased BMP pathway signaling, and that Acvr1^R206H/+ cells are poised towards an osteogenic lineage in the absence of osteogenic biochemical factors and the absence of pro-osteogenic (stiff) mechanical signals. Inhibition of the mechanical signaling YAP pathway mitigates increased mechanotransduction by Acvr1^R206H/+ cells, and inhibition of Rho/ROCK signaling reduces YAP pathway activity supporting interactions between these two mechanosignaling pathways in mediating the misregulation of mechanosignaling in Acvr1^R206H/+ cells.

Figure 7. Schematic for mechanotransduction and BMP signaling interactions in Acvr1^R206H/+ cells.

In Acvr1^+/+ progenitor cells on soft substrates, mechanotransduction activity is low, with lower expression of BMP target genes; on stiffer substrates, mechanotransduction activity is activated, leading to reinforcement of BMP and osteogenic gene signaling through RhoA and YAP/TAZ positive feedback. With the ACVR1^R206H mutation, cells on soft substrates have activated mechanotransduction (as indicated by red arrows), demonstrating cell mis-interpretation of substrate stiffness. Increased RhoA, YAP1, and BMP signaling converge to activate cellular contractility and chondro/osteogenic gene expression. Our data suggest that FOP Acvr1^R206H/+ cells are poised for chondro/osteogenesis, leading to endochondral ossification within skeletal muscle following injury. Asterisk (*) indicates location of the FOP ACVR1^R206H mutation.

This increased understanding of the mechanism by which cells that are poised for chondro/osteogenesis are directed to that fate by a secondary cue of improper mechanotransduction through Rho/ROCK and YAP1 signaling also may have relevance in identifying new potential treatment approaches for FOP such as a strategy that combines inhibitors of mechano-signaling and BMP signaling pathways. Similar mechanisms and approaches could possibly apply to other genetic diseases of aberrant cell fate such as progressive osseous heteroplasia ^{(108, 109)} and non-hereditary forms of heterotopic ossification ^{(10, 14, 110)}. Our data emphasize the importance of mechanotransduction signaling in cell fate determination, and provide evidence that increased BMP pathway through ACVR1 predisposes Acvr1^R206H/+ cells towards chondro/osteogenic cell fates and acts synergistically with increased activity of mechanical-signaling pathways to promote HO formation.