Activation of PPARγ redirects fibro-adipogenic progenitors to replace ectopic bone with fat in models of fibrodysplasia ossificans progressiva and trauma-induced heterotopic ossification

The pathologic, osteogenic differentiation of fibroadipogenic progenitor cells (FAPs) is the primary recognized contributor to ectopic bone formation in fibrodysplasia ossificans progressiva (FOP) and trauma -induced heterotopic ossification (HO). Both cond itions are characterized by up - regulated BMP signaling – the former by a gene mutation rendering the BMP receptor ACVR1 susceptible to activation by inflammatory ligands (Activin A), and the latter by up -regulated presence of BMP2 ligand in the setting of unmutated BMP receptor. We performed an unbiased assessment of FDA -approved therapies which would optimally target the transcriptional aberrations observed in developing FOP and HO lesions based on publicly -available datasets. This analysis uncovered rosiglitazone, a peroxisome proliferator -activated receptor gamma (PPARγ) agonist as the highest scoring therapeutic op tion across three data sets for both conditions. Rosiglitazone treatment eliminated ectopic bone lesions in a mouse model of FOP, and replaced these lesions with ectopic adipose tissue; similarly, systemic and local rosiglitazone treatment eliminated ectopic bone lesions in a mouse model of trauma-induced HO and replaced these lesions with ectopic adipose tissue. Our findings were corroborated by a single case report from 2010 showing positive results with rosiglitazone in a non-diabetic patient with FOP, with no subsequent studies. Overall, our findings suggest that a previously FDA-approved therapeutic is likely to be a successful therapeutic agent for both FOP and trauma-induced HO, both conditions for which current therapeutic options remain inadequate. One Sentence Summary: We show that a previously FDA-approved therapeutic known to induce adipogenesis reduces ectopic bone and induces ectopic fat formation in diseases of heterotopic ossification. .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2026. ; https://doi.org/10.64898/2026.02.26.708276doi: bioRxiv preprint

Heterotopic ossification (HO) is a pathologic condition in which extra -skeletal bone forms within soft tissues 1,2. A genetic form of HO known as fibrodysplasia ossificans progressiva (FOP), develops in patients with a hyperactivating mutation of the bone morphogenetic protein (BMP) receptor ACVR1 (ACVR1 R206H) 2. Patients with FOP develop painful , ectopic bony lesions at sites of inflammation such as may occur after even minor bruises or local injections. HO is also observed in patients with wild type ACVR1 who have experienced severe traumatic injury such as blast injuries, burns, and even surgical procedures 3-5. Like FOP, trauma-induced HO is mediated by excessive BMP signaling albeit due to elevated levels of BMP ligand rather than the presence of a hyperactivating mutation in the receptor itself 4,5. Previous studies have established fibro -adipogenic progenitor cells (FAPs) as precursor cells responsible for bone formation in FOP and trauma-induced HO 6,7. These FAPs are also capable of undergoing fibrogenic differentiation, as observed after muscle injury in individuals without hyperactivating BMP receptor mutations. A less understood differentiation pathway for FAPs is towards the adipogenic fate 8; however, this process is observed with ectopic fat infiltration in atrophying muscle 8 9. Much of the morbidity which ensues after HO formation is due to the rigidity and firmness of the bone, which leads to impaired mobility, open wounds due to pressure, and pain due to nerve impingement. Given the multi-potent capacity of FAPs, we hypothesized that these cells could be "nudged” towards adopting an adipogenic fate, rather than undergoing osteogenic differentiation. Peroxisome proliferator-activated receptor gamma (PPARγ) is a master transcriptional regulator of adipogenesis 10. Although in some contexts, elevated bone morphogenetic protein (BMP) signaling can up -regulate PPARγ expression 11, adipogenesis is not typically observed in FOP lesions or trauma-induced HO. In the clinical setting, the formation of soft, adipose tissue, would be preferable to rigid bone. Our objective in this study was to determine whether PPARγ could serve as a target for the developing lesions present in FOP and trauma-induced HO, and whether delivery of a PPARγ agonist could instead induce formation of ectopic fat.

In silico drug discovery identifies rosiglitazone as a top candidate to reverse aberrant gene expression in FOP and trauma-induced HO To identify compounds capable of reversing pathological gene expression in fibrodysplasia ossificans progressiva (FOP) and trauma-induced heterotopic ossification (HO), we analyzed bulk RNA-sequencing (RNA -seq) datasets from mouse injury models (GSE220725, GSE218699, GSE126118) 12 13 14. Differential expression between FOP/HO -injured sites and controls was scored against drug -gene interactions from the Drug –Gene Interaction Database (DGIdb) to prioritize agents predicted to counteract these signatures. For FOP (GSE220725), injured gastrocnemius from mutant ACVR1 -expressing mice showed distinct profiles versus normal controls, with pairwise correlations, confirming clustering (Fig 1A). Volcan o plot highlighted dysregulated genes (|log ₂FC| ≥ 3, P < 0.01; Fig 1B). Drugs were scored by opposing effects (agonist/antagonist) to inverse differential patterns, filtering for non -immunotherapeutic, non - antineoplastic agents (Fig 1C). Rosiglitazone, a PPAR γ agonist, ranked amongst the highest. Rosiglitazone also led scores in HO datasets: early post -injury in dexamethasone -cardiotoxin model (GSE218699; Fig 1D) and chronic burn/tenotomy (GSE126118; Fig 1E). Thus, we advanced rosiglitazone therapy in mouse models of FOP and trauma-induced HO. Absence of adipocytes or adipogenic signaling in FOP lesions .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2026. ; https://doi.org/10.64898/2026.02.26.708276doi: bioRxiv preprint We examined FOP lesions in a validated genetic mouse model (Ub.CreERT/ACVR1 R206H) three weeks after cardiotoxin-induced muscle injury (Fig 2A). Based on our examination of histologic images, we found no evidence of adipocytes in the injury site using H&E, Picrosirius Red, and Safranin O/Fast green with no treatment (Fig 2B). Perilipin-1 immunostaining verified adipocytes inside the bone marrow as expected, but not within the surrounding soft tissue (Fig 2C). Treatment with a PPARγ agonist suppresses osteogenesis and induces adipogenesis in bone marrow-derived mesenchymal cells in vitro Next, we sought to determine the fate of mesenchymal cells derived from FOP mice, when cultured in media with a pro -adipogenic PPARγ agonist. FOP-derived cells cultured in Activin A and rosiglitazone exhibited reduced expression of osteogenic genes Col2a1, Aggrecan, and Sox9 relative to FOP-derived cells cultured in Activin A alone. Conversely, FOP-derived cells cultured in Activin A and rosiglitazone exhibited increased expression of adipogenic genes such as Lpl and Fabp4 relative to FOP-derived cells cultured in Activin A alone (Fig 3A) (* is p<0.05). To simulate trauma-induced HO, which is dependent on BMP signaling in the setting of wild type ACVR1, a similar series of experiments were performed with mesenchymal cells obtained from wild type mice, and treated with BMP2 and rosiglitazone or vehicle control. WT cells cultured in BMP2 and rosiglitazone exhibited reduced expression of osteoge nic genes Col2a1, Aggrecan, and Sox9 relative to WT cells cultured in BMP2 alone. Conversely, WT cells cultured in BMP2 and rosiglitazone exhibited increased expressio n of adipogenic genes such as Lpl and Fabp4 relative to WT cells cultured in BMP2 alone (Fig 3B) (* is p<0.05). Treatment with rosiglitazone significantly reduces formation of post-injury FOP lesions FOP mice which had received intramuscular cardiotoxin were treated with rosiglitazone (10 mg/kg) or vehicle control initiated on day of cardiotoxin injection and delivered twice weekly until euthanization after 3 weeks . Both MicroCT and x-ray verified a significant reduction in ectopic bone formation at the site of injury (Fig 4A, B) (* is p< 0.05). Instead, we observed new adipose tissue and decreased cartilage in the affected region using H&E, Picrosirius Red, and Safranin O/Fast green (Fig 4C). However, immunostaining of the affected site verified the presence of elevated BMP signaling (anti -pSMAD 1/5), coincident with increased expression of adipogenic markers (anti-perilipin-1 and anti-Pparg). Soft tissue resident adipocytes also displayed PDGFRα signaling (anti-PDGFRα) suggesting that FAPs served as the source of ectopic adipocytes (Fig 4D). Systemic treatment with rosiglitazone reduces formation of heterotopic bone in a trauma model Next, we sought to examine whether rosiglitazone could similarly reduce heterotopic bone in a model of trauma -induced H O. MicroCT verified ectopic bone formation in soft tissue and calcaneus in the absence of treatment (Fig 5 A,B). Mice which received tenotomy exhibited significantly reduced formation of ectopic cartilage production after systemic rosiglitazone treatment relative to control treatment (Safranin O, H&E) (Fig 5C). Immunostaining verified the presence of increased pe rilipin and PPAR γ, indicative of adipocytes , in mice treated with rosiglitazone (Fig 5D). Local treatment with rosiglitazone reduces formation of heterotopic bone in a trauma model Next, we sought to determine whether local rosiglitazone treatment would sufficiently reduce HO formation in a trauma model. Given the association between rosiglitazone and osteopenia/osteoporosis, such an approach would avoid systemic exposure in patients who may require normal skeletal healing to occur following trauma. Mice underwent tendon transection to .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2026. ; https://doi.org/10.64898/2026.02.26.708276doi: bioRxiv preprint induce local HO formation; additionally, they were treated with local injection of rosiglitazone (10 mg/kg; once a week for 3 weeks) (Fig 6A). As with systemic delivery, mice which received tenotomy exhibited significantly reduced formation of ectopic cart ilage production after local rosiglitazone treatment relative to local control treatment (Safranin O, H&E) (Fig 6B). There was no evidence of increased adipogenesis in the contralateral hindlimb. Immunostaining verified that perilipin and PPARγ expression was elevated in the affected area, indicative of adipocytes in mice locally treated with rosiglitazone (Fig 6C).

In this study, we used an unbiased approach to identify the top -scoring FDA -approved drug predicted to intercede on transcriptional changes observed in developing FOP and HO lesions; this unbiased approach uncovered rosiglitazone as a top -scoring candidate and validated diminished PPARγ signaling as a modifiable aberration in FOP lesions. We found that delivery of rosiglitazone induces ectopic fat formation and suppresses ectopic bone formation in models of FOP and trauma-induced HO, thereby verifying that this drug has the desired effect in vivo. In mice with the genetic mutation responsible for FOP (ACVR1 R206H), rosiglitazone delivery led to the discernible presence of ectopic fat at the injury site. These fat cells expressed traditional adipocytes markers including perilipin and adiponectin, while also expressing PDGFRa, a marker for FAPs. Conversely, ectopic bone was significantly and substantially reduced with rosiglitazone treatment. Similarly, rosiglitazone, delivered either systemically or locally, induced ectopic fat formation and significantly reduced ectopic bone formation at the tendon transection site in a well- studied model of trauma-induced HO. Our findings of diminished ectopic bone and de novo ectopic adipogenesis with rosiglitazone treatment in FOP are consequential to the potential for this drug as a therapeutic option for FOP. Gatti et al have previously reported on an improvement observed in a patient with FOP treated with rosiglitazone 15. In that case report, the patient had substantial improvement relative to the year prior while on corticosteroid therapy during which time five episodes of flares were noted. In that patient, rosiglitazone was administered at a dose of 4-8 mg daily, which is consistent with the dosing used for patients with type 2 diabetes. However, beyond this case report, no further studies were pursued or published to further interrogate rosiglitazone in FOP. Currently, only one therapy has received FDA approval for FOP – palovarotene, a retinoic acid receptor agonist. However, adverse effects associated with this medication, such as premature joint plate fusion in pediatric patients have been reported 16. This is of particular interest given that ectopic bone formation in FOP occurs during childhood and accumulates throughout the patient’s lifespan; therefore, early intervention is paramount. Interestingly, while PPARg and RARg both bind to retinoid X receptor (RXR) to form separate heterodimer complexes which activate gene expression, rosiglitazone therapy has been recently shown to be safe in children without adverse effects 17. Therefore, repurposing rosiglitazone may represent an efficacious and safe therapeutic opportunity; it should be noted as well that palovarotene has similar roots as a drug developed initially developed for an alternative indication – emphysema 18. Previous studies have established FAPs as a progenitor cell population responsible for ectopic bone formation 6,7. FAPs are multipotent and have been shown to contribute to ectopic adipose tissue deposition in muscle atrophy, and to intramuscular fibrosis. The factors which modify the fate of FAPs in vivo are likely myriad. Our findings from this study suggest that FAPs responsible for ectopic bone formation may be directed towards an adipogenic fate through drug delivery. It should be noted that ectopic adipogenesis was not readily observed in the contralateral hindlimb which did not receive an injury, suggesting tha t the initial traumatic event is responsible for the influx of FAPs. .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2026. ; https://doi.org/10.64898/2026.02.26.708276doi: bioRxiv preprint We have also demonstrated that rosiglitazone reduces ectopic bone formation in trauma-induced HO. Both systemic and local treatment were highly effective in this role. In the setting of severe burns or high -impact trauma where the likely location of HO for mation may not be easily determined for local intervention, systemic delivery may present a viable option for HO prevention. For example, burn patients who are at elevated risk for HO formation (>30% total body surface area burns, arm burns, increased number of days on the ventilator or operations) 19 may benefit from systemic rosiglitazone therapy. In other patients who are elevated risk for HO, such as those with hip arthroplasty, systemic rosiglitazone may not be desirable due to the presence of comorbid osteopenia/osteoporosis in older patients. Local rosiglitazone delivery into the tissues surrou nding the surgical joint space which are susceptible to HO formation may represent a therapeutic option for these patients. Previously, poly(lactic-go-glycolic) acid (PLGA) particles have been used for rosiglitazone delivery to mitigate pulmonary arterial hyperplasia 20; separately, hydrogel loaded with rosiglitazone reduced joint contractures in a rabbit limb immobilization model 21. Other drugs have shown promise in prevention of heterotopic ossification. For example, rapamycin, an mTOR inhibitor has been shown to be efficacious in FOP 3 and has led to the development of a clinical trial in Japan 22. Whether a combinatorial treatment including both rapamycin and rosiglitazone may provide improved protection from FOP lesions with reduced dosing requirements may be of interest in the future. Separately, combinatorial treatment with palovarotene may also be an opportunity to reduce palovarotene dosing and adverse effects. A major challenge in the development of new drugs for FOP is the small patient population at risk for this condition. While it is certainly desirable for few patients to be struck by this condition, this is of little solace to those who do have FOP . Importantly, the small number of patients with FOP presents a challenge to clinical trial enrollment and drug development; this is compounded by reticence due to unknown risk profiles of newer medications. However the low risk profile of rosiglitazone may engender increased trial enrollment, relative to newer drugs without a previous track record of clinical use. In addition, consideration of studies with multiple treatment arms may be feasible when drugs are no longer on -patent. One strategy which may enable commercial interest in these drugs which are off -patent includes the FDA 505(b)(2) pathway which may provide market exclusivity up to 7 years in the setting of orphan drug designation. Regardless, our findings underscore the importance of iteratively and cont inuously examining new data and the literature for potential therapeutic avenues with a history of FDA approval.

Sex as a biological variable Our study examined male and female animals, and similar findings are reported for both sexes. Ethics statement All experiments on animals were conducted in accordance with the protocol approved by Institutional Animal Care and Use Committee (IACUC) of Brigham and Women’s Hospital (#2020N000036). In silico drug discovery Bulk RNA -seq data was sourced from GEO accessions GSE220725 (FOP vs. normal injury; n=5/group), GSE218699 (cardiotoxin + dexamethasone HO vs. saline; n=4/group), and .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2026. ; https://doi.org/10.64898/2026.02.26.708276doi: bioRxiv preprint GSE126118 (burn/tenotomy HO vs. uninjured; n=2/group). Drug-gene interactions were obtained from DGIdb v5.0 (accessed August 2025; https://dgidb.org) 23. Differential expression involved group-wise mean calculation, log₂FC with pseudocount (0.01), and Welch's t test. Visualizations included pairwise Pearson correlations on log₂(TPM + 1) for top 3000 variable genes and volcano plots (|log₂FC| ≥ 3, P < 0.01), with top genes labeled by composite rank. Drug effects were signed (+1 agonist, -1 antagonist), score -scaled, and aggregated per pair; filtered to non - immunotherapeutic, non-antineoplastic agents. Animals C57BL/6 mice and Ub.CreERT/ACVR1 R206H mice were housed in the Brigham and Women’s Hospital (BWH) vivarium. Tamoxifen was dissolved in corn oil and intraperitoneally injected into Ub.CreERT/ACVR1R206H mice (100 mg/kg) for 4 consecutive days to induce mutation. 8 Ub.CreERT/ACVR1R206H mice were treated with rosiglitazone and 8 mice were used as control. 5 WT C57BL/6 mice were treated with rosiglitazone post tenotomy and 6 mice were used as control. Finally, 5 WT mice were treated with local rosiglitazone post tenotomy and 5 mice were used as control. Mouse model For FOP model, two weeks after the tamoxifen injection, cardiotoxin (0.1 mg/ml in PBS, 30 μl per hindlimb) was injected intramuscularly into the hindlimb of the Ub.CreERT/ACVR1R206H mice to induce muscle injury and promote heterotopic ossification. Hindlimb samples were harvested 3 weeks after the cardiotoxin injection. For tenotomy model, the achilles of C57BL/6 mice were transected. Hindlimb samples were harvested 3 weeks after the tenotomy. Sacrifice was performed via CO2 asphyxiation. Hindlimb samples were fixed in 10% formalin for 48h followed by 70% ethanol for further assessment. Rosiglitazone treatment For systemic treatment of rosiglitazone, it was dissolved in DMSO at the concentration of 25 mg/ml and was diluted in corn oil to the final concentration of 2.5 mg/ml before injection. Mice were intraperitoneally injected with rosiglitazone (10 mg/kg) twice a week started in parallel with cardiotoxin injection or surgery. Mice in the control group received coin oil injection at the same time points. For local treatment, 10 mg/kg rosiglitazone was injected locally around the site of tendon transection once a week for 3 weeks. Mice in the control group received PBS at the same time points. Micro-CT and analysis Hindlimb samples were scanned using high -resolution Micro -Computed Tomography ( μCT40, SCANCO Medical AG, Brüttisellen, Switzerland). Scan parameters were 30 μm3 isotropic voxel size, 55 kVp peak X -ray tube intensity, 145 µA X-ray tube current, and 300 ms i ntegration time. DICOM images were exported for analysis of ectopic bone volume. Histology Formalin-fixed samples were dehydrated, embedded in paraffin, and sectioned with 5 -micron- thickness. Sections were stained with Hematoxylin and Eosin (H&E), Picrosirius Red and Safranin-O/Fast green according to standard protocol. The images were taken by Olympus BX53 microscope. Immunofluorescent staining Tissue sections were de-paraffinized, rehydrated and blocked with 10% goat serum, 1% bovine serum albumin (BSA), and 0.3% Triton (MilliporeSigma) in PBS for 1 hour and incubated with primary antibodies at 4°C overnight. The sections were probed with antibodies for Perilipin-1 (Cell .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2026. ; https://doi.org/10.64898/2026.02.26.708276doi: bioRxiv preprint Signaling Technology, cat# 3467S, 1:100), PPAR γ ( Proteintech, cat 16643 -1-AP, 1:100), pSMAD1/5 (Cell Signaling Technology, cat# 9516S, 1:100). After washing, the sections were incubated for one hour in the dark using the following secondary antibodies: Alexa Fluor Plus 488 goat anti-rabbit IgG (1:1000) and Alexa Fluor 594 goat anti-rabbit IgG (1:1000). ProLong Diamond Antifade Mountant with DAPI (P36971, Invitrogen) was used to stain the nuclei and mount the samples. Fluorescent images were taken by Olympus Fluoview Confocal microscope. Cell culture Bone marrow-derived mesenchymal cells were isolated from both C57BL/6 mice and tamoxifen treated Ub.CreERT/ACVR1R206H mice. Briefly, femur and tibia were obtained after the euthanation of the mice. The proximal and distal end of the femurs and tibia were cut using sharp scissors. Ice-cold PBS was injected into the femur and tibia to flush out the bone marrow from the proximal end onto a 70 µm nylon cell strainer. The cells were centrifuged at 1500 rpm for 5 minutes at 4 °C. Cells were cultured in DMEM medium with 10% FBS and 1% penicillin/streptomycin. All cells were cultured in a humidified atmosphere with 5% CO2 at 37 °C. Cell treatment Bone marrow-derived mesenchymal cells from C57BL/6 mice were treated with BMP2 (25 ng/ml). Bone marrow-derived mesenchymal cells from tamoxifen treated Ub.CreERT/ACVR1R206H mice were treated with Activin A (25 ng/ml). Rosiglitazone (10 µM) was administered at the same time. All cells were harvested on day 7. RT-qPCR Total mRNA was isolated from cells using Direct -zol RNA Miniprep Kits (Zymo Research). Reverse transcription was performed using qMax cDNA Synthesis Kit (Accuris). Real-Time PCR was performed using the following primers for Lpl, Fabp4, Sox9, Col2a1, Aggrecan, and Actin respectively: 5’-GCGTAGCAGGAAGTCTGACCAA-3’, 5’ -AGCGTCATCAGGAGAAAGGCGA-3’; 5’-TGAAATCAC-CGCAGACGACAGG-3’, 5’ -GCTTGTCACCATCTCGTTTTCTC-3’; 5’ - CACACGTCAAGCGACCCA-TGAA-3’, 5’ -TCTTCTCGCTCTCGTTCAGCAG-3’; 5’ GCTGGTGAAGAAGGCAAACGAG-3’, 5’ -CCATCTTGACCTGGGAATCCAC-3’; 5’ - CAGGCTATGAGCAGTGTGATGC-3’, 5’ -GCTGCTGTCTT-TGTCACCCACA-3’; 5’- CATTGCTGACAGGATGCAGAAGG-3’, 5’-TGCTGGAAGGTGGACAGTG-AGG-3’. The PCR condition was: 95 °C for 10 minutes; 40 cycles 95°C for 15 s, 60 °C for 1 minute, 72 °C for 30 s; 72 °C for 10 min The PCR was performed using the machine QuantStudio 7 Flex Real-Time PCR Systemand analyzed using QuantStudio™ Real-Time PCR Software. Statistics All data are presented as mean ± S.D. Statistical analysis was performed using GraphPad Prism 9. Comparisons between two groups were performed by using 2 -tailed Student’s t -tests. Comparisons among multiple groups were performed by using a one-way ANOVA with a Tukey’s multiple-comparisons test. Significant difference was established when P value was smaller than 0.05.

1. Kaplan, F.S., Tabas, J.A., Gannon, F.H., Finkel, G., Hahn, G.V., and Zasloff, M.A. (1993). The histopathology of fibrodysplasia ossificans progressiva. An endochondral process. J Bone Joint Surg Am 75, 220-230. 10.2106/00004623-199302000-00009. 2. Shore, E.M., Xu, M., Feldman, G.J., Fenstermacher, D.A., Cho, T.J., Choi, I.H., Connor, J.M., Delai, P., Glaser, D.L., LeMerrer, M., Morhart, R., et al. (2006). A recurrent mutation .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2026. ; https://doi.org/10.64898/2026.02.26.708276doi: bioRxiv preprint in the BMP type I receptor ACVR1 causes inherited and sporadic fibrodysplasia ossificans progressiva. Nat Genet 38, 525-527. 10.1038/ng1783. 3. Agarwal, S., Loder, S., Brownley, C., Cholok, D., Mangiavini, L., Li, J., Breuler, C., Sung, H.H., Li, S., Ranganathan, K., Peterson, J., et al. (2016). Inhibition of Hif1alpha prevents both trauma-induced and genetic heterotopic ossification. Proc Natl Acad Sci U S A 113, E338-347. 10.1073/pnas.1515397113. 4. Agarwal, S., Loder, S.J., Breuler, C., Li, J., Cholok, D., Brownley, C., Peterson, J., Hsieh, H.H., Drake, J., Ranganathan, K., Niknafs, Y.S., et al. (2017). Strategic Targeting of Multiple BMP Receptors Prevents Trauma-Induced Heterotopic Ossification. Mol Ther 25, 1974-1987. 10.1016/j.ymthe.2017.01.008. 5. Peterson, J.R., De La Rosa, S., Eboda, O., Cilwa, K.E., Agarwal, S., Buchman, S.R., Cederna, P .S., Xi, C., Morris, M.D., Herndon, D.N., Xiao, W., et al. (2014). Treatment of heterotopic ossification through remote ATP hydrolysis. Sci Transl Med 6, 255ra132. 10.1126/scitranslmed.3008810. 6. Lees-Shepard, J.B., Stoessel, S.J., Chandler, J.T., Bouchard, K., Bento, P., Apuzzo, L.N., Devarakonda, P .M., Hunter, J.W., and Goldhamer, D.J. (2022). An anti-ACVR1 antibody exacerbates heterotopic ossification by fibro-adipogenic progenitors in fibrodysplasia ossificans progressiva mice. J Clin Invest 132. 10.1172/JCI153795. 7. Lees-Shepard, J.B., Yamamoto, M., Biswas, A.A., Stoessel, S.J., Nicholas, S.E., Cogswell, C.A., Devarakonda, P.M., Schneider, M.J., Jr., Cummins, S.M., Legendre, N.P ., Yamamoto, S., et al. (2018). Activin-dependent signaling in fibro/adipogenic progenitors causes fibrodysplasia ossificans progressiva. Nat Commun 9, 471. 10.1038/s41467-018-02872-2. 8. Tan, J., Li, Y ., Zhang, J., Qi, B., Chen, J., and Sun, Y . (2025). Role of aberrant activated fibro/adipogenic progenitors and suppressed ferroptosis in disused skeletal muscle atrophy and fatty infiltration. J Mol Med (Berl) 103, 713-724. 10.1007/s00109-025-02548- 7. 9. Joe, A.W., Yi, L., Natarajan, A., Le Grand, F., So, L., Wang, J., Rudnicki, M.A., and Rossi, F.M. (2010). Muscle injury activates resident fibro/adipogenic progenitors that facilitate myogenesis. Nat Cell Biol 12, 153-163. 10.1038/ncb2015. 10. Rosen, E.D., Sarraf, P., Troy, A.E., Bradwin, G., Moore, K., Milstone, D.S., Spiegelman, B.M., and Mortensen, R.M. (1999). PPAR gamma is required for the differentiation of adipose tissue in vivo and in vitro. Mol Cell 4, 611-617. 10.1016/s1097-2765(00)80211-7. 11. Takada, I., Yogiashi, Y ., and Kato, S. (2012). Signaling Crosstalk between PPARgamma and BMP2 in Mesenchymal Stem Cells. PPAR Res 2012, 607141. 10.1155/2012/607141. 12. Sun, L., Jin, Y., Nishio, M., Watanabe, M., Kamakura, T., Nagata, S., Fukuda, M., Maekawa, H., Kawai, S., Yamamoto, T., and Toguchida, J. (2024). Oxidative phosphorylation is a pivotal therapeutic target of fibrodysplasia ossificans progressiva. Life Sci Alliance 7. 10.26508/lsa.202302219. 13. Alexander, K.A., Tseng, H.W., Lao, H.W., Girard, D., Barbier, V., Ungerer, J.P .J., McWhinney, B.C., Samuel, S.G., Fleming, W., Winkler, I.G., Salga, M., et al. (2024). A glucocorticoid spike derails muscle repair to heterotopic ossification after spinal cord injury. Cell Rep Med 5, 101849. 10.1016/j.xcrm.2024.101849. 14. Sorkin, M., Huber, A.K., Hwang, C., Carson, W.F.t., Menon, R., Li, J., Vasquez, K., Pagani, C., Patel, N., Li, S., Visser, N.D., et al. (2020). Regulation of heterotopic ossification by monocytes in a mouse model of aberrant wound healing. Nat Commun 11, 722. 10.1038/s41467-019-14172-4. 15. Gatti, D., Viapiana, O., Rossini, M., and Silvano, A. (2010). Rosiglitazone therapy is associated with major clinical improvements in a patient with fibrodysplasia ossificans progressiva. J Bone Miner Res 25, 1460-1462. 10.1359/jbmr.091113. .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2026. ; https://doi.org/10.64898/2026.02.26.708276doi: bioRxiv preprint 16. Lees-Shepard, J.B., Nicholas, S.E., Stoessel, S.J., Devarakonda, P .M., Schneider, M.J., Yamamoto, M., and Goldhamer, D.J. (2018). Palovarotene reduces heterotopic ossification in juvenile FOP mice but exhibits pronounced skeletal toxicity. Elife 7. 10.7554/eLife.40814. 17. Varo, R., Crowley, V.M., Mucasse, H., Sitoe, A., Bramugy, J., Serghides, L., Weckman, A.M., Erice, C., Bila, R., Vitorino, P., Mucasse, C., et al. (2024). Adjunctive rosiglitazone treatment for severe pediatric malaria: A randomized placebo-controlled trial in Mozambican children. Int J Infect Dis 139, 34-40. 10.1016/j.ijid.2023.11.031. 18. Stolk, J., Cooper, B.G., Stoel, B., Rames, A., Rutman, O., Soliman, S., and Stockley, R. (2010). Retinoid treatment of Emphysema in Patients on the Alpha-1 International Registry. The REPAIR study: study design, methodology and quality control of study assessments. Ther Adv Respir Dis 4, 319-332. 10.1177/1753465810379617. 19. Levi, B., Jayakumar, P., Giladi, A., Jupiter, J.B., Ring, D.C., Kowalske, K., Gibran, N.S., Herndon, D., Schneider, J.C., and Ryan, C.M. (2015). Risk factors for the development of heterotopic ossification in seriously burned adults: A National Institute on Disability, Independent Living and Rehabilitation Research burn model system database analysis. J Trauma Acute Care Surg 79, 870-876. 10.1097/TA.0000000000000838. 20. Rashid, J., Alobaida, A., Al-Hilal, T.A., Hammouda, S., McMurtry, I.F., Nozik-Grayck, E., Stenmark, K.R., and Ahsan, F. (2018). Repurposing rosiglitazone, a PPAR-gamma agonist and oral antidiabetic, as an inhaled formulation, for the treatment of PAH. J Control Release 280, 113-123. 10.1016/j.jconrel.2018.04.049. 21. Arsoy, D., Salib, C.G., Trousdale, W.H., Tibbo, M.E., Limberg, A.K., Viste, A., Lewallen, E.A., Reina, N., Yaszemski, M.J., Berry, D.J., van Wijnen, A.J., et al. (2018). Joint contracture is reduced by intra-articular implantation of rosiglitazone-loaded hydrogels in a rabbit model of arthrofibrosis. J Orthop Res 36, 2949-2955. 10.1002/jor.24068. 22. Hino, K., Zhao, C., Horigome, K., Nishio, M., Okanishi, Y., Nagata, S., Komura, S., Yamada, Y., Toguchida, J., Ohta, A., and Ikeya, M. (2018). An mTOR Signaling Modulator Suppressed Heterotopic Ossification of Fibrodysplasia Ossificans Progressiva. Stem Cell Reports 11, 1106-1119. 10.1016/j.stemcr.2018.10.007. 23. Cannon, M., Stevenson, J., Stahl, K., Basu, R., Coffman, A., Kiwala, S., McMichael, J.F., Kuzma, K., Morrissey, D., Cotto, K., Mardis, E.R., et al. (2024). DGIdb 5.0: rebuilding the drug-gene interaction database for precision medicine and drug discovery platforms. Nucleic Acids Res 52, D1227-D1235. 10.1093/nar/gkad1040. .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2026. ; https://doi.org/10.64898/2026.02.26.708276doi: bioRxiv preprint

The authors acknowledge 1K08AR082031 (NIH/NIAMS), Plastic Surgery Foundation National Endowment for Plastic Surgery Award, Million Dollar Bike Ride Foundation, International FOP Association, Stepping Strong Foundation and Beal Fellowship to SA, Stepping Strong Innovator Award to PK, National Nature Science Foundation of China (82404113), Basic and Applied Basic Research Foundation of Guangdong Province (2023A1515111068) and Shenzhen Science and Technology Program (JCYJ20230807095121041) to ZC, Hale fellowship, Coller Award and PSF fellowship to AS. This work in part was supported by NIH T35 fellowship to ZM, NIH T35 HL110843 fellowship to HS. Author contributions: Conceptualization: PK, ZC, SNH, SA; Methodology: PK, ZC, SNH, AS, CL, ZM, HS, DM; Investigation: PK, ZC, SNH, SA; Visualization: PK, ZC, SNH, HS; Funding acquisition: ZC, SA, Project administration: PK, ZC, SNH, SA, Supervision: YM, VR, SA, Writing – original draft: PK, ZC, HS, SA, Writing – review & editing: PK, ZC, YM, VR, SA. Competing interests: Authors declare that they have no competing interests. Data and materials availability: All data are available in the main text. .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2026. ; https://doi.org/10.64898/2026.02.26.708276doi: bioRxiv preprint FIGURES Fig. 1. Drug repurposing for fibrodysplasia ossificans progressiva (FOP) and heterotopic ossification (HO). (A) Pairwise correlation heatmap of RNA-seq profiles. Pearson correlations for log₂(TPM + 1)–normalized expression of top 3000 variable genes in injured gastrocnemius from normal (n = 5) and FOP (n = 5) mice (GSE220725). (B) Volcano plot of FOP vs. normal injury differential expression. Points show genes with log₂ fold change (x axis) and –log₁₀(P value) from Welch's t test (y axis). Gray: all genes; red: significant (|log₂FC| ≥ 3, P < 0.01). Dashed blue lines: thresholds. Top 10 genes per direction labeled by composite rank of fold change and significance. (C) Drug scoring for FOP gene expression reversal. Scatter plot of non-immunotherapeutic, non-antineoplastic drugs. horizontal axis: total score (sum of [drug effect × –log₂FC]); vertical axis: SD of contributions. Rosiglitazone highlighted in red/bold. (D) Drug scoring for HO reversal (cardiotoxin model). Scoring as in (C), using differential expression from cardiotoxin + dexamethasone (HO; n = 4) vs. cardiotoxin + saline (control; n = 4) at day 4 (GEO: GSE218699). (E) Drug scoring for HO reversal (burn/tenotomy model). Scoring as in (C), using differential expression from burn/tenotomy sites (HO; n = 2) vs. uninjured contralateral limbs (n = 2) at 3 weeks (GEO: GSE126118). .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2026. ; https://doi.org/10.64898/2026.02.26.708276doi: bioRxiv preprint Fig. 2. FOP lesions do not contain adipocytes or adipogenic signals. (A) Schematic representation of the procedure used to develop the FOP mouse model. Mice were treated with tamoxifen to induce the R206H mutation in ACVR1, and cardiotoxin was injected to induce heterotopic ossification (HO). Tissue was harvested after 3 weeks of cardiotoxin injection. (B) Picrosirius Red, Safranin-O/Fast Green, and H&E staining of hindlimbs from FOP mice treated with cardiotoxin show bone lesions but no adipocytes. (C) Immunostaining of FOP mice treated with cardiotoxin shows Perilipin-1–positive cells in the bone marrow, but not in the soft tissue surrounding the bone. .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2026. ; https://doi.org/10.64898/2026.02.26.708276doi: bioRxiv preprint Fig. 3. Treatment with a PPARγ agonist suppresses osteogenesis and induces adipogenesis in bone marrow-derived mesenchymal cells in vitro. (A) Bone marrow– derived mesenchymal cells from FOP mice were treated with 10µM rosiglitazone, a PPARγ agonist, in the presence or absence of 25ng/ml Activin A. Activin A induces osteogenic gene expression in FOP cell lines. Expression levels of adipogenic genes such as LPL and FABP4 increased with rosiglitazone treatment, whereas expression of osteogenic genes such as Sox9, Col2A1, and Aggrecan decreased. (n = 3) Statistical significance between two groups was determined using Student's t-test (p < 0.05). B-actin was used for normalization. (B) Bone marrow–derived mesenchymal cells from WT mice were treated with 10µM rosiglitazone in the presence or absence of 25ng/ml BMP2, which induces osteogenic gene expression. Like FOP cells, expression of adipogenic genes (LPL and FABP4) increased, while expression of osteogenic genes (Sox9, Col2A1, and Aggrecan) decreased. (n = 3) Statistical significance between two groups was determined using Student's t-test (p < 0.05). B-actin was used for normalization. .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2026. ; https://doi.org/10.64898/2026.02.26.708276doi: bioRxiv preprint Fig. 4. Treatment with rosiglitazone significantly reduces formation of post-injury FOP lesions. (A) X-ray and Micro-CT imaging revealed a greater volume of ectopic bone in the hindlimbs of FOP mice treated with the corn oil vehicle control compared to those treated with rosiglitazone. (B) Quantification of ectopic bone showed an approximately 10-fold decrease in .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2026. ; https://doi.org/10.64898/2026.02.26.708276doi: bioRxiv preprint ectopic bone volume following rosiglitazone treatment. (n = 8) Statistical significance between two groups was determined using Student’s t-test (p < 0.05). (C) H&E, Picrosirius Red, and Safranin-O/Fast Green staining of FOP mouse hindlimbs showed a reduction in FOP lesions and an increase in soft tissue–resident adipocytes after rosiglitazone treatment. Chondrocytes are represented by the red staining of Saffranin-O whereas the Fast Green stain represents the bone. (D) Immunofluorescence imaging of rosiglitazone-treated FOP mice showed an increase in soft tissue–resident adipocytes, indicated by PPARγ and Perilipin-1–positive cells. In contrast, hindlimbs of control mice exhibited adipogenic signals only within the bone marrow, consistent with bone marrow–resident adipocytes. Soft tissue–resident adipocytes were absent in control mice. pSMAD1/5 staining revealed abundant positive cells in the bone marrow of control mice and in the soft tissue of rosiglitazone-treated mice. Soft tissue resident adipocytes were also positive for PDGFRα, suggesting that the adipocytes were derived from FAPs. .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2026. ; https://doi.org/10.64898/2026.02.26.708276doi: bioRxiv preprint Fig. 5. Systemic treatment with rosiglitazone reduces heterotopic bone formation in a trauma-induced HO model. (A) Schematic representation of the procedure used to induce heterotopic ossification (HO) in wild-type (WT) mice via Achilles tendon tenotomy. Tissue was .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2026. ; https://doi.org/10.64898/2026.02.26.708276doi: bioRxiv preprint harvested 3 weeks post tenotomy. (B) Micro-CT imaging showed ectopic bone in the soft tissue and surrounding the calcaneus of the right hindlimb where tenotomy was performed. (C) H&E and Safranin-O/Fast Green staining of tenotomized WT mice showed a reduction in HO lesions and an increase in soft tissue–resident adipocytes following rosiglitazone treatment. (D) Immunofluorescence imaging of rosiglitazone-treated tenotomy mice revealed an increase in soft tissue–resident adipocytes, indicated by PPARγ and Perilipin-1–positive cells. In contrast, control mice displayed adipogenic signals only within the bone marrow, consistent with bone marrow–resident adipocytes. Soft tissue–resident adipocytes were absent in control mice. pSMAD1/5 staining showed abundant positive cells in the bone marrow of control mice and in the soft tissue of rosiglitazone-treated mice. .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2026. ; https://doi.org/10.64898/2026.02.26.708276doi: bioRxiv preprint Fig. 6. Local treatment with rosiglitazone reduces heterotopic bone formation in a trauma-induced HO model. (A) Schematic representation of the procedure used to induce heterotopic ossification (HO) in wild-type (WT) mice via Achilles tendon tenotomy, followed by local rosiglitazone injection. Tissue was harvested 3 weeks post tenotomy. (B) H&E and Safranin-O/Fast Green staining of tenotomized WT mice showed a reduction in HO lesions and an increase in soft tissue–resident adipocytes after rosiglitazone treatment. (C) Immunofluorescence imaging of rosiglitazone-treated tenotomy mice revealed an increase in soft tissue–resident adipocytes, as indicated by PPARγ and Perilipin-1–positive cells. In .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2026. ; https://doi.org/10.64898/2026.02.26.708276doi: bioRxiv preprint contrast, control mice showed adipogenic signals only within the bone marrow, consistent with bone marrow–resident adipocytes, while soft tissue–resident adipocytes were absent. pSMAD1/5 staining showed abundant positive cells in the bone marrow of control mice and in the soft tissue of rosiglitazone-treated mice. .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2026. ; https://doi.org/10.64898/2026.02.26.708276doi: bioRxiv preprint