1. Introduction
Fibrodysplasia ossificans progressiva (FOP) is a rare, one-in-two-million-occurring, autosomal dominant genetic disease characterized by progressive heterotopic bone formation (HO) where especially muscles, tendons, and ligaments are converted into bone [
1,
2,
3]. The clinical manifestation of the heterotopic ossification is rather diverse. HO can occur after a flare up, during inflammation, after injury, or even spontaneously. In some patients, HO is more progressive than in others, and episodes exist with complete absence of HO formation [
4]. Because of these differences in clinical manifestations, cell biological approaches that can shed light on the biochemical events that precede heterotopic ossification are of great importance as a prelude to therapy. Over the past decade, our understanding of the pathogenesis of the disease has improved considerably. The causative mutation in the gene encoding the Activin Receptor Type I receptor (ACVR1) was identified by the group of Shore et al. [
5]. The most frequent R206H mutation changes an arginine to a histidine in the glycine–serine rich cytoplasmic domain of this bone morphogenetic protein (BMP) type I receptor [
2,
5]. This amino acid change makes the receptor more sensitive to BMP signaling and simultaneously results in decreased binding of the receptor’s natural inhibitor, FKBP12, likely resulting in leaky signaling of the receptor through SMADs, the main signal transducers of the transforming growth factor-beta (TGF-β) family [
6,
7,
8,
9,
10]. More recently, the discovery by two independent groups that Activin-A can also function as an activator of the mutated ACVR1 [
11,
12] has given great impulse to the FOP research community, since now it is conceivable to specifically prevent heterotopic bone formation by inhibiting Activin-A. Under normal circumstances, Activin-A signals mainly via ACVR1B receptor complexes through SMAD 2/3 phosphorylation [
13,
14,
15]. Upon binding to ACVR1, it normally inhibits SMAD 1/5/9 signaling and subsequent osteogenic differentiation [
16], but, as mentioned before, when Activin-A binds to the mutated ACVR1, it activates SMAD1/5/9 phosphorylation and subsequent osteogenic differentiation [
11,
12]. This was demonstrated by Hatsell et al. in their FOP mouse model, but also in the currently conducted LUMINA-I trial investigating the Activin-A blocking antibody Garetosmab, which has revealed promising results in the prevention of newly formed HO [
17,
18]. Even though it has by now been clearly shown that capturing Activin-A protects against heterotopic ossification, this knowledge is mainly limited to mouse models for FOP. Moreover, the exact underlying mechanism at the gene expression level causing ACVR1 R206H-mediated ossification and progression of the disease is unknown. Thus, knowledge of the downstream transcriptomic changes after Activin-A-ACVR1 R206H binding using open-ended approaches such as RNA sequencing is warranted.
One of the challenges that FOP research faces is the lack of primary cells from patients, since surgical removal of the heterotopic bone, as a potential source of bone-forming osteoblasts, often results in the appearance of new heterotopic bone formation. Several alternative sources for bone-forming cells are currently being investigated as a tool for the unraveling of the FOP HO mechanism. Dermal fibroblasts have been demonstrated by Micha et al. to be a source of cells with osteoblastic properties. They showed that the FOP-derived fibroblasts have a high osteogenic potential [
19]. Another source of human primary cells are cells isolated from extracted teeth. Many FOP patients develop a locked jaw, making oral care extremely difficult. In some cases, tooth extraction is the only dental treatment possible. Additionally, teeth are sometimes extracted to provide extra space in the mouth. There have been no reports of HO in the sockets at the extraction sites. Some groups have made use of the cells from the discarded primary teeth (SHED cells) [
7,
20], showing higher osteogenic differentiation in the FOP cells compared to control cells. Our group has recently shown that periodontal ligament cells obtained from FOP patients after tooth extraction show both osteogenic- as well as osteoclastogenic-inducing capacities, as do PLF from healthy individuals [
21,
22,
23,
24]. Here, we used these cells to investigate the Activin-A-induced transcriptome differences between controls and FOP patients. This study explored for the first time RNA sequencing as an open-ended approach for non-biased identification of potential new differences in early gene expression between control and FOP fibroblasts, directly following Activin-A binding. Given the unique signaling by Activin-A via the mutated ACVR-1, as described in mouse models [
11] and iPS cell approaches [
12], we hypothesize that Activin-A will induce early transcriptomic differences specifically in primary cells from FOP patients.
4. Discussion
The discovery of the causative mutation in the
ACVR1 gene resulting in the altered responsiveness of this BMP receptor [
5] has paved the way to investigating the exact mechanism that underlies the heterotopic ossification seen in FOP. Especially during the last decade, great progress has been made, and the discovery of Activin-A as an activator of the mutated ACVR1 receptor [
11,
12] has stimulated the scientific FOP community to explore new therapeutic avenues. Here, we investigated the effect on early transcriptome differences induced by Activin-A in our recently established human primary FOP cell culturing system using periodontal ligament fibroblasts [
21].
Our data show that indeed the mutated ACVR1 receptor has an altered responsiveness to Activin-A as compared to the non-mutated receptor. Initial gene expression analysis without false discovery rate correction showed that, when using a
p-value cutoff of
p < 0.01, 131 genes are differentially expressed in FOP cells as compared to 46 genes in the control cells. Pathway analysis of these genes using GeneMania shows an association with pathways involved in Activin, BMP, and TGFβ signaling only in the FOP cells. FOP is an autosomal dominant genetic disease, meaning that both the non-mutated as well as the mutated gene is being expressed, which has been demonstrated in FOP derived SHED cells and monocytes [
22,
51]. The association with the Activin and TGFβ signaling pathways indicates signaling via the non-mutated receptor, whereas the BMP signaling via SMAD1, 5, and 9 shows that the altered responsiveness of the mutated receptor to Activin-A as described by other groups [
20,
52] is also present in the periodontal ligament fibroblasts from FOP patients. Additionally, the gene enrichment analysis on these genes show a stronger reaction of the mutated receptor when exposed to Activin-A. There is an association with different GO terms, whereas no such association could be found in the control cells. HO in FOP often occurs after inflammation or injury [
3], a process that is possibly mediated by an increase in Activin-A [
3,
11]. The altered gene expression profile seen here in the FOP cells after short exposure to Activin-A shows an association with GO terms GO 0009611 (response to wounding), GO:0009719 (response to endogenous stimulus), and GO:0071495 (cellular response to endogenous stimulus). This probably reflects the altered reaction described in FOP after injury or infection [
53,
54,
55]. Under normal circumstances when Activin-A binds to the non-mutated ACVR1 receptor, it acts as an inhibitor of the canonical BMP-mediated signaling via ACVR1 and subsequent osteogenic differentiation. In FOP, however, the mutation results in activation of the mutated ACVR1 upon Activin-A binding, resulting in enhanced osteogenic differentiation. The association with the Activin-A-induced upregulated genes in the FOP cells with the collagen gene family observed in this study probably reflects the osteogenesis process taking place as a response to this altered signaling of the mutated ACVR1.
The inter-donor variability has a higher impact on the difference in overall expression compared to the experimental condition (e.g., without or with Activin-A), as described in
Figure 1, suggesting that differences on the single-gene-expression level induced by Activin-A treatment could be difficult to detect above the “noise” induced by the above-mentioned inter-donor variability. Surprisingly, despite this fact, supervised gene expression analysis using a false discovery rate of 10% showed that differential expression was seen, but only in the FOP cells. Some of the genes that showed a higher than two-fold upregulation in the FOP cells under the influence of Activin-A can be related to bone metabolism or heterotopic ossification and can thus be regarded as potential targets to inhibit HO in FOP. These genes will be discussed next in the context of the process of endochondral bone formation.
Heterotopic ossification, as seen in FOP, is believed to be formed via endochondral bone formation where chondrogenic differentiation of stem cells results in cartilage formation that is subsequently converted to bone. We find an increase in DOCK7 expression in the FOP cells, possibly resulting in increased chondrogenesis and osteogenesis, ultimately resulting in enhanced HO. The mutated ACVR1 has been shown to play an important role in enhanced chondrogenic differentiation in a gain-of-function mouse model [
56]. In mouse models of HO, one of the early steps in the formation of HO is the biogenesis of brown adipose tissue (BAT) [
57]. This BAT creates a hypoxic environment that is suggested to induce chondrogenic differentiation of mesenchymal stem cells [
58] and subsequent endochondral bone formation. Misty mice lack BAT and show decreased osteogenesis and increased osteoclast activity [
40]. This uncoupling of bone formation and bone resorption is the result of a loss of function mutation in the
DOCK7 gene [
39,
59]. Additionally, the 2.3-fold upregulation of PAPPS2 observed in the FOP cells here may relate to the increased early stage endochondral ossification. The cartilage matrix that is produced during the early stages of endochondral ossification contains high concentrations of sulfate groups bound to several extracellular matrix proteins. PAPPS2 is a phosphosulphate phosphatase that catalyzes the sulfation of these extracellular matrix proteins [
38]. Expression of PAPPS2 is partly regulated by TGFβ [
60]. Mutations in PAPPS2 result in different types of osteochondrodysplasias, examples of which are described by Wang et al. [
37]. The upregulation of TTC1 expression suggests that the RAS signaling pathway, which belongs to the osteogenic differentiation program, is activated. Osteogenic differentiation may take place via different pathways. One of the signaling pathways involved in osteogenic differentiation is the RAS signaling [
61,
62]. TTC1 is an adaptor protein that binds to Gα proteins, resulting in an interaction with small GTPase RAS that activates the RAS signaling pathway [
35,
36]. Similarly, the higher expression of SHOC2 seen here is an implication of increased RAS-MAPK signaling. SHOC2 is a leucine-rich repeat scaffolding protein that is involved in the RAS-MAPK signaling [
33,
34]. Mutations in this gene are linked to RASopathies like Mazzaniti syndrome and Noonan syndrome [
63,
64], both of which result in reduced postnatal growth. At a later stage of endochondral bone formation, the formation of collagen cross links is an important step that determines the integrity and strength of the bone. One of the enzymes involved in the formation of these crosslinks is Lysil Oxidase (LOX), a copper-dependent enzyme that catalyzes the deamination of lysine and hydroxylisine residues, thereby initiating the crosslinking of collagen molecules [
41,
42,
65,
66]. LOX-deficient mice show lower osteoblast differentiation and activity [
42], and lower numbers of crosslinks are related to bone fragility in, for instance, aging and osteoporosis [
65]. Interestingly, expression of LOX is partly regulated by the hypoxia-induced transcription factor HIF1 [
67,
68]. HIF1 is, together with Activin-A, upregulated during injury [
20,
53], providing potential links between hypoxia, FOP, and LOX expression.
Taken together, the upregulated genes described above play a role in either the early, chondrogenic, stages of endochondral bone formation (
DOCK7 and
PAPPS2) or the later, more osteogenic, stages of this process (
TTC1,
SHOC2. and
LOX). This indicates that the altered responsiveness of the mutated ACVR1 gene to Activin-A could induce heterotopic bone formation. Further studies to elucidate the exact role of athe bove-described genes in HO in FOP cells need to be performed. In these studies, blocking experiments with, for instance, follistatin will be included. We and other groups have previously shown that follistatin indeed inhibits the effect of Activin-A on other FOP derived primary cells [
20,
22]. We did not include such blocking experiments in this RNA sequencing experiment, because we first wanted to investigate which genes were influenced in the first place. Additionally, the subsequent data and statistical analysis where we would compare four different experimental conditions would become rather complex.
As discussed above, the up- and downregulated genes were identified only in FOP patient-derived cells. Nevertheless, and strikingly, the fold-change was relatively small, between 1.3- and 2.5-fold changes in the upregulated genes and 1.3- and 3.9-fold changes in the downregulated group. This suggests that the advanced heterotopic ossification in FOP could be the result of additive gene expression effects, rather than being caused by one dominantly over- or under expressed gene. QPCR experiments did not confirm the statistical differences, as seen in the RNAsequencing data analysis. This could be due to the fact that, as stated in
Figure 1, the inter-donor variability is even higher than the differences induced by the experimental condition (e.g., addition of Activin-A), but probably more importantly, the fact that the average expression differences induced by Activin-A were below two-fold. In our experience, it is hard to reach statistical differences in QPCR experiments on human material with a low
n as used here. In this paper, we use the periodontal ligament fibroblast as a cell biological tool to study early effects of Activin-A on osteogenic differentiation. Different groups have previously shown the osteoblast-like properties of these cells, making them a relatively accessible source of cells to study osteogenic differentiation in, for instance, FOP research [
69,
70,
71,
72]. However, as previously stated, following tooth extraction, no HO formation in the socket has been reported. This may imply that the periodontal ligament fibroblasts exhibit HO formation inhibiting properties, which may be the underlying reason for the relatively small fold changes witnessed, though an alternative possibility is the short incubation time with Activin-A. Longer exposure to the molecule might reveal more and possibly stronger effects on the transcriptome, resulting, when using osteogenic medium, in Activin-A-enhanced osteogenesis, as recently shown with tooth-associated cells by Wang et al. [
20]. Gene enrichment and pathway analysis did not identify many or any new pathways associated to Activin-A signaling through the mutated receptor. This may be due to the relatively small number of differentially expressed genes, which does not allow for extensive pathway analysis. Nevertheless, when applying supervised gene expression analysis with a false discovery rate of 10%, our data show that Activin-A only induces differential gene expression in the FOP cells bearing the R206H mutation. The highest upregulated genes have all been linked to bone metabolism, and some even to heterotopic ossification in general, but none of these genes have, to our knowledge, previously been described to be linked to HO in FOP. The differential expression of genes somehow involved in bone formation by Activin-A exclusively in cells carrying the FOP mutation is in strong support of the therapeutic treatment rationale of inhibiting Activin-A. This study adds evidence to the notion that when disarming Activin-A’s malevolent effect, heterotopic bone formation can be tempered in the lives of people who suffer the daily anxiety of progressive ectopic bone formation with all its disabling consequences.