Author response:
The following is the authors’ response to the original reviews.
Our revised version of the manuscript addresses all the comments and suggestions raised, as clarified in our point-by-point answer to the reviewers. We have performed additional experiments regarding the effects on proliferation and differentiation of additional cell types in the muscle, such as myogenic and mesenchymal progenitors as well as chondrogenesis in parental hMSCs that did not express exogenous ACVR1. Moreover, as suggested by reviewer #2, we performed all the chondrogenic experiments with addition of TGFβ in the differentiation media and analyzed chondrogenesis by both Alcian blue staining and qPCR analysis of gene markers (Sox9, Acan, Col2a1 and Mmp3). We also extended our RNA-seq analysis and included new data using both hMSCs expression wild type or R206H ACVR1 receptor, with or without different ACVR1 ligands (BMP6 and Activin A) and treated or not with the inhibitor BYL719. The new data suggests that BYL719 is able to inhibit the expression of genes involved in ossification and osteoblast differentiation irrespective of the presence of the mutation. We also discuss the effect of BYL719 in mTOR signaling and addressed all the minor comments suggested by both reviewers.
We addressed the specific comments of the reviewers as follows:
Reviewer # 1:
Specific points:
Point #1 and #2. The authors showed that BYL719 inhibited HO in FOP model mice. Did they have HO not only in the muscle but also in the bone marrow? The progenitor cells of chondrocytes and osteoblasts may differ between the muscle and bone marrow. The authors should examine the effects of BYL719 on some other types of cells in the muscle, such as myoblasts and fibro-adipogenic cells, in addition to the bone marrow-derived MSCs. Furthermore, it was unclear whether they were human or murine MSCs in the text.
The inhibitory effect of BYL719 on HO in FOP mice was clear, but the molecular mechanisms or target cells were still unclear because BYL719 affected multiple types of cells and molecules. The authors are encouraged to show clearer mechanisms and target cells' critical inhibition of HO. Again, this reviewer believes that in vivo and in vitro experiments using muscle and bone marrow and cells prepared from them should provide additional critical information.
As detailed in the introduction, it is known that Heterotopic Ossification develops in the skeletal muscle and connective tissues. Consistent with the current knowledge of the field, none of the mice showed HO in the bone marrow. Additionally, since activation of the mutant allele is achieved by injection of CRE-expressing adenovirus and cardiotoxin in the muscle hindlimb, it is unlikely that mesenchymal progenitors in the bone marrow would be strongly affected. Interestingly, single-cell RNA sequencing from multiple mouse tissues identified a very strong transcriptional similarity between FAPs and non-muscle mesenchymal progenitors (PMID: 37599828). As suggested, we examined the effects of BYL719 in proliferation and differentiation in additional cell types such as muscle progenitors. In this new version of the manuscript, we show that BYL719 reduces the proliferation of muscle and mesenchymal progenitors while it blocks myoblast differentiation in vitro (Figure 7, Figure Supplement 1). MSCs were murine on those experiments shown in Figure 3; whereas assays shown in Figures 5 and 6 were of human origin. We have further clarified this in the respective Figure legends.
All the data generated strongly suggests that there is not a single mechanism supporting all the effects of BYL719 in HO. Instead, BYL719 affects multiple cell types involved in efficient HO (e.g. reduction in proliferation and osteochondrogenic specification of mesenchymal precursors (MPs), reduction on proliferation, migration, and inflammatory gene expression on monocytes, etc.). Interestingly, our data suggests that BYL719 is able to inhibit these effects on MPs and monocytes irrespective of the presence of the ACVR1-R206H mutation (Figures 5, 6 and 7). Additionally, there are several signaling mechanisms affected. BYL719 reduces SMAD1/5, p38, AKT and mTOR signaling in parental MPs or with mutations in ACVR1 (Figure 3 and our previous publication PMID: 31373426), being all these pathways required for efficient osteochondrogenic specification of MPs. We consider that the different detailed mechanisms by which BYL719 inhibits osteochondrogenic specification enhances the robustness of the findings in this study.
Point #3. In FOP model mice, ACVR1 was mutated as Q207D. However, R206H was used in in vitro experiments. Do they have the same characteristics? This reviewer would like to recommend examining the effect of BYL719 on wild-type ACVR1, R206H, and Q207D simultaneously in each experiment.
We already performed these experiments, assaying in parallel ACVR1-WT, ACVR1-Q207D and ACVR1-R206H, in the transcriptional responses of MPs in our previous work (PMID: 31373426). Both mutations had similar responses, being ACVR1-Q207D stronger than ACVR1-R206H, as it has been shown in vivo in mouse models of HO (PMID: 34633114). In any case, BYL719 inhibits these transcriptional responses induced by both mutant alleles.
Point #4. Figure 5: What was the effect of BYL719 on the differentiation of parental cells that did not express exogenous ACVR1?
We performed new assays of chondrogenic differentiation of hMSCs that are shown in the new Figure 5. BYL719 inhibits chondrogenic differentiation of parental hMSCs and also inhibits chondrogenic specification irrespective of the expression of either wild type or mutant ACVR1.
Point #5. Figure 6: In this experiment, gene expression was examined in pretreated MSCs-ALK2 (ACVR1?) R206H with and without BYL719. It was clear whether suppression of gene expression by BYL719 was specifically caused in cells expressing R206H. What were the effects of BYL719 on parental cells that did not express exogenous ACVR1?
To be consistent, we relabeled ALK2 to ACVR1 in the figure. We expanded the conditions analyzed in the RNA-sequencing. We included conditions where we activate ACVR1 (either WT or R206H) with their known physiological ligand BMP6. In both, human MSCs expressing ACVR1-R206H and human MSCs expressing Wild Type ACVR1, we observed a downregulation of differentially expressed genes upon addition of BYL719, irrespective of ligand (BMP6 or Activin A) or receptor (RH or WT) (added new Figure 6: B and C).
Point #6. Figure 7: BYL719 suppressed cell proliferation of all cells examined partially at 2 uM and almost completely at 10 uM, respectively. There is a possibility that BYL719 inhibits HO by inhibiting osteochondroprogenitor proliferation. The authors are encouraged to show data on the effect of BYL719 on the proliferation of other types of cells, such as myoblasts, fibro-adipogenic cells, or bone marrow cells.
We examined the effects of BYL719 in proliferation in additional cell types such as muscle and mesenchymal progenitors. BYL719 slightly reduced the proliferation of myoblasts and mesenchymal cells in vitro (Figure 7, Figure Supplement 1). However, the reduction in the proliferation in myoblasts or MPs did not reach the extent to that observed in monocytes or macrophages (Figure 7).
Point #7. Figure 8: How was the effect of BYL719 on muscle regeneration in wild-type? It was reported that mTOR signaling is important in HO in FOP. The authors are encouraged to show the effect of BYL719 on mTOR signaling.
Muscle regeneration in wild-type mice has also been shown in our previous results PMID: 31373426. In addition, we included images of the muscle regeneration after 23 days of treatment with BYL719 in mice ACVR1Q207D with or without PI3Kα deletion after induction of HO in the new Figure 2, Figure Supplement 2. These mice showed full muscle regeneration or small calcifications surrounded by muscle at most. The effects of PI3Kα inhibitors, either BYL719 or A66, on mTOR signaling had been previously shown by our group (PMID: 31373426). Both inhibitors strongly reduced signaling of mTOR, visualized by activation of p70 S6-kinase, a surrogate marker of mTOR activity.
Minor points:
(9) SMAD 1/5 should be SMAD1/5.
(10) The source of human MSCs should be indicated in the text.
(11) ALK2 should be ACVR1 in Figure 6A.
(12) The protein levels of each receptor should be examined in Fig. 4.
We introduced the suggested changes in the manuscript and Figure 6 and indicated the source of human MSCs in Materials and Methods. We also examined the levels of each receptor that are shown in the new Figure 4, Figure Supplement 1.
Reviewer # 2:
Specific points:
Point #1. Because the involvement of PI3K in HO of FOP, was already reported by authors' group and also others (Hino et al, Clin Invest, 2017), the main purpose of this study was to disclose the mechanism of how PI3K was activated in FOP cells. In the published study (Hino et al, Clin Invest, 2017), PI3K was activated by the ENPP2-LPA-LPR cascade. Unfortunately, there were no new data for this important issue.
The main purpose of this study is to demonstrate that the pharmacological and genetic inhibition of PI3Kα in HO progenitors at injury sites reduces HO in vivo, to extend the insights into the molecular and cellular mechanisms responsible for the therapeutic effect of PI3K inhibition, and to optimize the timing of the administration of BYL719. Class I PI3Ks are heterodimers of a p110 catalytic subunit in complex with a regulatory subunit. They engage in signaling downstream of tyrosine kinases, G protein-coupled receptors and monomeric small GTPases. Therefore, a plethora of growth factors, cytokines, inflammatory agents, hormones and additional external and internal stimuli are able to activate PI3Kα (PMID: 31110302). In fact, TGF-β family members, including activin A, are able to activate PI3K and mediate some of their non-canonical responses (PMID: 19114990). Multiple factors with known increased expression in the ossifying niche in HO and FOP (e.g. activin A, TGF-β, inflammatory agents such as TNFα, IL6, IL3, etc.) are known activators of PI3K (PMID: 30429363). Interestingly, in our RNA-seq analysis in hMSCs we did not observe increased expression levels of Enpp2 when comparing wild type and R206H mutated cells treated with activin A.
Point #2. The HO formation of ACVR1/Q207D model mice in this study is extremely unstable (Figure 1B, DMSO). Even the bone volume of some red symbols, which indicate the presence of HO, is located on the base (0.00) line. I would examine carefully the credibility of the data. Also, it is well known that the molecular behavior of mice Acvr1/Q207D and human ACVR1/R206H was different.
We agree with the reviewer that induction of HO is variable between mice showing variations in penetrance and intensity of the ossifying lesions. This variability is a known common trend that appears in all the models of HO published so far (e.g. PMID: 28758906, PMID: 26333933). Accordingly, we did not exclude any animal that has been injected with CRE-expressing adenovirus plus cardiotoxin in the μCT analysis. Regarding the behavior of mice Acvr1/Q207D and human ACVR1/R206H, it is well known that Q207D produces more robust and stronger responses in terms of signaling and formation of heterotopic ossification (PMID: 34633114). Therefore, reduction of HO by BYL719 would be more stringent in the Acvr1/Q207D model.
Point #3. The experimental design of Figure 5 experiments is confusing. Although the authors mentioned that the data in Figure 5A were taken seven days after chondrogenic induction, I am skeptical whether the chondrogenic induction was successful. Based on the description of Material and Methods, the authors did not include TGFβ in their "Differentiation Medium", which is an essential growth factor to induce chondrogenic differentiation of human MSC. Why did the ALP activity increase after chondrogenic induction? The authors should demonstrate the evidence of successful chondrogenic induction by showing the expression of key chondrogenic genes such as SOX9, ACAN, or COL2A1. The data in Figure 5B-E are also confusing. The addition of Activin A showed no difference between ACVR1/WT and ACVR1/R206H cells, suggesting that these cells did not reproduce the situation of FOP.
We performed new assays of chondrogenic differentiation of hMSCs that are shown in the new Figure 5. We included TGFβ1 in the differentiation medium and also included the parental cell line in the analysis. In addition of being a marker of osteoblast differentiation, alkaline phosphatase (ALPL) has also been shown to be induced during chondroblast differentiation in vitro (PMID: 19855136; PMID: 9457080; PMID: 18377198; PMID: 23388029). Moreover, expression data of SOX9, COL2A1, ACAN and MMP13 of cells after chondrogenic differentiation is included in the new Figure 5. Expression of some markers (e.g. ACAN) are increased by the expression of ACVR1R206H, however, we did not observe significant differences in chondroblast differentiation gene expression between ACVR1wt and ACVR1R206H expressing cells. In any case, BYL719 could inhibit chondrogenic differentiation of parental hMSCs and also the chondrogenic specification irrespective of the expression of either wild type or mutant ACVR1.
Point #4. The experimental design and data analyses of RNA-seq were inappropriate and insufficient, which is disappointing for the reviewer because this will be a key experiment in this study. Because the most important point is to identify the signal for PI3Kα induced by Activin A via ACVR1/R206H, they should also use hMSC-ACVR1/WT for this experiment. Because the authors clearly demonstrated that TGFBR were not targets of BYL719, they should compare the expression profiles between MSC-ACVR1/WT and MSC-ACVR1/WT with BYL719 to identify the targets of BYL719 unrelated to Activin A signal. Then the expression profiles of ACVR1/R206H cells treated with Activin A and Activin A plus BYL719 were compared. Among down-regulated signals by BYL719, those found also in MSC-ACVR1/WT should be discarded. It is important to investigate whether the GO term of ossification or osteoblast differentiation is found also in MSC-ACVR1/WT. If it is so, the effect of BYL719 is not specific for FOP cells.
We extended our RNA sequencing analysis with additional experimental conditions and comparisons. In new Figure 6, we now compare hMSCs expressing wild type or R206H receptors, with or without BYL719 inhibition, and with or without different ligand activations (BMP6 or Activin A) (New Figure 6A). New Figure 6B shows the Gene ontology analysis of the differentially expressed genes between cells expressing WT and RH receptors under control conditions. We can observe that ossification (GO:0001503) and osteoblast differentiation (GO:0001649) were detected within the top 10 significantly differentially regulated biological processes between these conditions. Therefore, we analyzed these relevant identified GO terms in 5 different comparisons upon GO enrichment analysis (Figure 6C). In addition to the comparison between cells expressing WT and RH receptors under control conditions explained above, we also compared cells expressing WT or RH receptor, with different ACVR1 ligands (BMP6 and Activin A), and with or without BYL719 inhibitor. The addition of BYL719 resulted in a downregulation of the GO terms “ossification” and “osteoblast differentiation” (new Figure 6C). These results confirm the inhibitory effect of BYL719 on ossification and osteoblast differentiation biological processes, and inform that this inhibitory effect remains consistent upon BMP6 or Activin A ligand activation, and with ACVR1 WT and RH expression.
Point #5. The data in Figure 7 were not related to the aim of this study because cell lines used in these experiments did not have ACVR1/R206H mutations. It is not appropriate to extrapolate these data in the FOP situation.
We utilized immune cell lines where we could activate ACVR1 with their known physiological ligand BMP6. Mutated ACVR1 gains response to activin A in addition to maintaining the physiological response to BMP6 as the wild type form. Therefore, in these assays we interrogated in vitro, with addition of BMP6, the effects of BYL719 in the growth, migration and inflammatory gene expression upon conditions of activated ACVR1 receptor downstream signaling. We consider that understanding the effects of PI3Kα inhibition in the regulation of proliferation, migration and inflammatory cytokine expression in monocytes, macrophages and mast cells is essential to better define the potential outcome of BYL719 treatment for heterotopic ossifications.
Minor comments:
(1) The legends for Figure 1C were those for Figure 1D, and there were no descriptions for Figure 1C in the legends and methods section. The reviewer was unable to understand the meaning of BV/TV. What is TV?
(2) “However, in PI3Kα deficient mice ACVR1Q207D expression only led to minor ectopic calcifications that were already surrounded by fully regenerated muscle tissue on the 23rd day after injury (Figure 2D, Figure 2-Figure Supplement 1B)": There were no histological data either Figure 2D, Figure 2-Figure Supplement 1B), which showed muscle tissues.
(3) "The overexpression of Acvr1R206H increased basal and activin dependent expression of canonical (Id1 and Sp7) and non-canonical (Ptgs2) BMP target genes (Figure 3C),": There was no increase of Ptgs2 gene in basal level.
(4) Materials and Methods. Production of human fetal mesenchymal stem cells expressing ACVR1.: Is it derived from a fetus?
(5) Figure 6C: There was no description of the meaning of each column. What does AA mean and what is the number?
We introduced the missing information in the manuscript, Figure legends and material and methods section for points #1, 4 and 5. AA was Activin A, the number was the number of replicates. This has been detailed in the figure legend. We included images of the muscle regeneration after 23 days of treatment with BYL719 in mice after induction of HO in the new Figure 2, Figure Supplement 2 (point #2). We corrected the mistake in the manuscript refraining for suggesting increase of Ptgs2 gene expression by ACVR1-R206 at the basal level (Point #3).
