Suppression of heterotopic ossification in fibrodysplasia ossificans progressiva using AAV gene delivery

Heterotopic ossification is the most disabling feature of fibrodysplasia ossificans progressiva, an ultra-rare genetic disorder for which there is currently no prevention or treatment. Most patients with this disease harbor a heterozygous activating mutation (c.617 G > A;p.R206H) in ACVR1. Here, we identify recombinant AAV9 as the most effective serotype for transduction of the major cells-of-origin of heterotopic ossification. We use AAV9 delivery for gene replacement by expression of codon-optimized human ACVR1, ACVR1R206H allele-specific silencing by AAV-compatible artificial miRNA and a combination of gene replacement and silencing. In mouse skeletal cells harboring a conditional knock-in allele of human mutant ACVR1 and in patient-derived induced pluripotent stem cells, AAV gene therapy ablated aberrant Activin A signaling and chondrogenic and osteogenic differentiation. In Acvr1(R206H) knock-in mice treated locally in early adulthood or systemically at birth, trauma-induced endochondral bone formation was markedly reduced, while inflammation and fibroproliferative responses remained largely intact in the injured muscle. Remarkably, spontaneous heterotopic ossification also substantially decreased in in Acvr1(R206H) knock-in mice treated systemically at birth or in early adulthood. Collectively, we develop promising gene therapeutics that can prevent disabling heterotopic ossification in mice, supporting clinical translation to patients with fibrodysplasia ossificans progressiva.

There are a number of questions that require clarification or explanation, which are summarized below. 1. Page 8 line 151: Have the authors also considered using transdifferentiation of dermal fibroblast cultures instead of FOP IPSCs, thus preventing the loss of valuable characteristics? 2. Page 13 line 281: chondrocyte lineage cells in growth plates did not appear to be rAAV9transducible in experiments with Acvr1(R206H)KI mice, but did the authors show the same for knee cartilage? 3. Page 15 line 330: Here is stated ''administration of rAAV9.amiR-RH6.ACVR1opt at birth prevented trauma-induced development of heterotopic endochondral ossification in Acvr1(R206H)KI FOP mice but showed minimal effects on early post-traumatic injury responses'' Does this mean that it has to be given more often or that cells other than FAP play a role during the already ongoing HO process? Could the authors perhaps bring their opinion about this topic into the discussion? 4. Page 18 line 422: Can the authors explain why HO formation of the cervical spine appeared to specifically occur after systemic administration 5. Page 19 line 450: It is intriguing why no effect of the administration on the immune system is seen. Can the authors indicate whether it has to do with the type of vector? 6. An article has recently been published in which the possibility of a study such as the one you have carried out is discussed. You may consider including this in the references.

Reviewer #2 (Remarks to the Author):
The manuscript presented by Yang and colleagues clearly shows the possibility to use a combination of gene replacement and inhibition to address heterotopic ossification in human and mouse cells and in vivo in rodents. The manuscript is well written and the data presented are robust. The authors clearly showed proof of concept. The authors have to address the following comments: 1. The mouse models used have to be described in detail in the method section. For example, the different CRE used and whether they are inducible or not as well as the ACVR1 mutant that has the mutation in the mouse sequence, and this information was not easily found. 2. Please comment on the fact that you selected your artificial miRNA on a human sequence and that sequence was effective also on the mouse sequence. How similar are the target sequences in the two species? 3. One potential limitation of this work is that the use of inducible models not recapitulating the human disease and AAV treatment at different ages failed to demonstrate efficacy in a clinically relevant context. Could you please discuss this point in light of the potential target patient population? 4. The authors clearly state that rAAV has high transduction efficacy in skeletal muscle and skeleton. Please comment on the adverse events reported when doses efficiently targeting those tissues were applied in patients. This is a clear limitation of the clinical translation. Are there alternatives to target those tissues with lower vector doses?

Reviewer #1: The authors conducted an extremely interesting in-depth study to identify the best strategy to find a genetic therapy for Fibrodysplasia ossificans progressive (FOP). They identified recombinant adenoassociated virus 9 (rAAV9) as the most effective serotype for transduction of the major cells-of-origin of HO in soft tissues and found that the combination of gene replacement and silencing (ACVR1OPT;amiR-ACVR1R206H) ablated aberrant Activin A signaling and chondrogenic and
osteogenic differentiation of mouse skeletal cells harboring a conditional knock-in allele of human ACVR1R206H (Acvr1(R206H)KI) and human FOP patient-derived induced pluripotent stem cells (iPSCs). In Acvr1(R206H) mice, trauma-induced HO was also markedly reduced when treated locally in early adulthood or systemically at birth. This is the first study focusing on a genetic therapy of FOP, for which there is currently no therapy worldwide. This is an extremely complicated disease and the authors have achieved impressive results that give hope for the future. This manuscript is very innovative, original, the methods are well described, the results are of great importance to the field of FOP, I would certainly consider it for publication. The videos are also very interesting.
-We thank the reviewer for summarizing and highlighting the significance of our manuscript. We agree that these findings will pave the way for treating HO in FOP patients. We believe that addressing these comments has significantly improved the revised manuscript.

Page 8 line 151: Have the authors also considered using transdifferentiation of dermal fibroblast cultures instead of FOP IPSCs, thus preventing the loss of valuable characteristics?
-We thank the reviewer for suggesting this approach. The main purpose of the FOP iPSCs experiments was to demonstrate that the AAV targeting of ACVR1 R206H could result in normalization of gene expression levels and reduction of osteogenic markers using a multipotent cell type as a starting point. FOP iPSCs were previously shown to have increased mineralization and osteogenic gene expression (PMID 24321451); thus, the changes described in Figure 2 and Extended Data Figure 4 demonstrate these findings. -Osteogenic transdifferentiation of dermal fibroblasts from patients with FOP have been previously reported, using retroviral transduction of osteogenic transcription factors and reprogramming factors, human platelet lysate, or ascorbic acid, beta glycerol phosphate, and dexamethasone (PMID 26769004, 28705683). We believe that the use of our established, highly-characterized, episomal, iPSC model, which has now been used successfully in multiple studies (PMID: 24321451, 35442931, 34755602, 34311122, 27530160), provides several key advantages for our study: The iPSCs used in this study have no retroviral reprogramming factors and do not induce cellular differentiation by overexpression of master transcription factors, both of which could over-ride cellular differentiation signals modified by ACVR1 R206H . In addition, the use of human platelet lysate is an undirected differentiation process, making the final cell linages difficult to interpret. The ascorbic acid/beta glycerol phosphate/dexamethasone method used for dermal fibroblasts is similar to what we used with the iPSCs, but the use of iPSCs provides a key advantage of starting from a cell type capable of differentiation into multiple skeletal lineages (which could be detected using NGS), rather than starting from a stable cell type that may have limited differentiation capacity.
-Currently, we are also testing the ability of rAAV9.amiR-RH6.ACVR1 opt to suppress osteogenic differentiation of periodontal ligament fibroblasts derived from human FOP patients.

Page 13 line 281: chondrocyte lineage cells in growth plates did not appear to be rAAV9-transducible in experiments with Acvr1(R206H)KI mice, but did the authors show the same for knee cartilage?
-We thank the reviewer for pointing this out. A new fluorescence microscopy data showing that chondrocyte lineage cells in articular cartilage are not rAAV9-transducible was added to Extended Data Figure 8c (top panel).

Page 15 line 330:
Here is stated ''administration of rAAV9.amiR-RH6.ACVR1opt at birth prevented trauma-induced development of heterotopic endochondral ossification in Acvr1(R206H)KI FOP mice but showed minimal effects on early post-traumatic injury responses'' Does this mean that it has to be given more often or that cells other than FAP play a role during the already ongoing HO process? Could the authors perhaps bring their opinion about this topic into the discussion? -We thank the reviewer for raising these questions. Tissue-specific tropism of rAAV9 serotype might be responsible for minimal effects of rAAV9.amiR-RH6.ACVR1 opt on early post-traumatic injury responses. Our present and previous studies (Figure 3 and Extended Data Figure 3, PMID: 32405514) and others (PMID: 35871478) demonstrated that rAAV9 serotype is effective for transduction of PDGFRa + or Tie2 + FAP-lineage cells, osteoblast-lineage cells (osteoprogenitors, mature osteoblasts, osteocytes), and myoblast-lineage cells (myogenic progenitors, myoblasts, myocytes), but not fibroblasts and immune cells (monocytes, macrophages, dendritic cells, neutrophils, and T and B lymphocytes). These results suggest that systemically delivered amiR-RH6.ACVR1 opt mainly impacts on rAAV9-transducible cells, including FAP-lineage cells, osteoblast-lineage cells, and myoblast-lineage cells in injured sites, resulting in suppression of chondrogenesis and osteogenesis while facilitating muscle regeneration. However, amiR-RH6.ACVR1 opt is likely to be dispensable for trauma-induced inflammation, fibroproliferation, and muscle damage (early post-traumatic injury responses) due to low transduction efficiency to fibroblasts and immune cells. This was added to the result section of the revised manuscript.
4. Page 18 line 422: Can the authors explain why HO formation of the cervical spine appeared to specifically occur after systemic administration? -We thank the reviewer for pointing this out. As described in a previous study (PMID: 26333933), 6week-old Acvr1 (R206H)Fl ;Cre-ER T2 mice develop spontaneous HO at multiple anatomical locations, including the cervical spine, hips and knees, 10-12 weeks post-injection of tamoxifen. These HO phenotypes might not be affected by IV injection of rAAV vectors.

Page 19 line 450:
It is intriguing why no effect of the administration on the immune system is seen. Can the authors indicate whether it has to do with the type of vector? -We thank the reviewer for raising this interesting question. From humoral responses point of view, since mouse is not a natural host of primate AAVs, there is little to no pre-existing neutralizing antibodies against any AAV serotypes and no negative impact on the first administration of mice. However, it will trigger the production of AAV neutralization antibodies, preventing redosing of AAVs. In terms of cellular responses, among the commonly used viral vectors, AAV is the only viral vector that elicits little to no T cell responses to AAV transduction in mice with a few exceptions for some highly immunogenic transgenes such as ovalbumin.
6. An article has recently been published in which the possibility of a study such as the one you have carried out is discussed. You may consider including this in the references.