Fibrodysplasia ossificans progressiva (FOP) is a rare genetic disorder marked by the episodic deposition of heterotopic bone in the place of muscle and connective tissues throughout the life of a patient. All individuals with FOP have been found to carry a point mutation in one copy of the gene encoding the bone morphogenetic protein (BMP) type I receptor, ALK2/ACVR1 (Shore et al.,2006; Kaplan et al.,2009). FOP-associated mutations in ALK2/ACVR1 appear to result in a hyperactive receptor that produces inappropriate BMP signaling (Billings et al.,2008; Fukuda et al.,2009; Kaplan et al.,2009; van Dinther et al.,2010).
Transforming growth factor-β (TGF-β)/BMP type I receptors are highly conserved, transmembrane receptor serine/threonine kinases that are an integral part of the TGF-β/BMP signal transduction pathway acting in a diverse array of cellular and developmental processes. The TGF-β/BMP type I receptors are characterized by a cysteine-rich, extracellular ligand-binding domain; a single-pass transmembrane domain; and a well-conserved, intracellular kinase domain (Massagué,1998). The intracellular domain contains regulatory regions such as the L45 loop, which confers binding specificity for the intracellular transducer, the receptor-mediated Smad (R-Smad) (Feng and Derynck,1997; Persson et al.,1998), and a glycine-serine rich GS domain required for activation of the type I receptor kinase (Wrana et al.,1994; Franzén et al.,1995). Two type I receptors mediate extracellular TGF-β/BMP signals as part of a receptor complex with two related type II receptor. This heteromeric receptor complex has been shown to assemble by two different mechanisms: (1) secreted ligands can induce receptor complex formation by binding the extracellular ligand-binding domains of the type I and type II receptors (Groppe et al.,2008; Nickel et al.,2009), or (2) the type I and type II receptors can interact independently of ligand to generate preformed complexes that then bind ligand to initiate signal transduction (Nohe et al.,2002; Hassel et al.,2003; Ehrlich et al.,2011; Marom et al.,2011). Upon formation of the ligand-bound receptor complex, the type I receptor is activated by the constitutive kinase activity of the type II receptor through trans-phosphorylation. Once phosphorylated by the type II receptor, the GS domain forms a secondary binding site for R-Smads (Huse et al.,2001). The R-Smads are, in turn, phosphorylated by the activated type I receptors. Phosphorylated R-Smads complex with the co-Smad, accumulate in the nucleus, and interact with other proteins to regulate transcription of specific target genes (Massagué,1998; Wu and Hill,2009).
Based largely on sequence homology and ligand specificity, two main groups emerge from the family of BMP type I receptors represented by the mammalian ALKs and Drosophila Tkv and Sax receptors: ALK3/ALK6/Tkv and ALK1/ALK2(ACVR1)/Sax (Chen and Massagué,1999; Newfeld et al.,1999). These two groups exhibit higher affinities for secreted BMP ligands from the two subfamilies, BMP2/4/Dpp and BMP5/6/7/Gbb, respectively. Although some functional redundancy exists amongst ligand and receptor families, it is clear that the tissue context of various ligand/receptor combinations impacts signaling output. As such, mutations in type I receptors alone are associated with several unique diseases, including hereditary hemorrhagic telangiectasia type 2 (HHT2; ALK1/ACVRL1), juvenile polyposis syndrome (ALK3/BMPR1A), brachydactyly type A2 (ALK6/BMPR1B), and, the focus of this study, fibrodysplasia ossificans progressiva (FOP; ALK2/ACVR1) (Howe et al.,2001; Zhou et al.,2001; Kim et al.,2003; Lehmann et al.,2003,2006; Abdalla and Letarte,2006; Bayrak-Toydemir et al.,2006; Wehner et al.,2006; Olivieri et al.,2007).
FOP is characterized by progressive, heterotopic ossification that occurs through an endochondral process (Pignolo et al.,2005). Extraskeletal ossification is especially detrimental when it leads to immobilization of joints and restriction of organ function. Mortality associated with FOP often results from respiratory complications due to the fusion of rib bones that interfere with the function of muscles, connective tissue, and nerves in the intercostal space (Kaplan and Glaser,2005). Interestingly, for the most part, clinical features are not apparent at birth other than great toe malformations, a characteristic that is invariant in classic FOP (Shore et al.,2006). As a general rule, the onset of FOP is delayed until early childhood, suggesting that the disease is developmental in nature and may require other triggers. Ossification is episodic and tends to occur in association with trauma or inflammation, thus rendering surgery an ineffective treatment (Kaplan et al.,2008).
The vast majority of FOP patients harbor the classic mutation which is defined by a 617G>A mutation in one copy of their ALK2 gene, resulting in a histidine substitution at arginine 206 (R206H) (Shore et al.,2006). The R206H FOP mutation, which lies C-terminal to the GS domain in ALK2 (see Fig. 2A), leads to a high level of BMP signaling in a variety of systems (Billings et al.,2008; Shen et al.,2009; Song et al.,2010; van Dinther et al.,2010). Interestingly, this residue is located just N-terminal to a conserved Thr/Gln residue that, when mutated to Asp, confers constitutive activity to TGF-β and BMP type I receptors (Wieser et al.,1995; Attisano et al.,1996; Akiyama et al.,1997; Macías-Silva et al.,1998; Chen and Massagué,1999). Both the classic FOP mutation and the Thr/Gln-to-Asp mutation emphasize the importance of the region near the GS domain.
While the mutations responsible for FOP have been identified, the molecular details that result in the hyperactive behavior of mutated ALK2/ACVR1 type I receptors are not yet fully understood. The episodic nature of FOP and the long latency or quiescent period before heterotopic bone formation in patients, indicate that the hyperactivity of the mutant receptor must be unleashed. In order to advance our understanding of FOP, studies to elucidate the mechanism of ALK2/ACVR1 receptor activation and its function within an organismal context are critical. As such, identifying the molecular events that are responsible for FOP-induced hyperactive BMP signaling will reveal avenues for potential therapeutic approaches.
Drosophila has proven to be an outstanding model organism to study an increasing number of human diseases based on the high degree of molecular and functional conservation observed for genes known to be involved in both the signaling pathways and regulatory mechanisms governing development and homeostasis (Veraksa et al.,2000; Reiter and Bier,2002; O'Kane,2003; Bier,2005; Botas,2007; Chintapalli et al.,2007; Pandey and Nichols,2011). Drosophila signaling components are largely nonredundant, which circumvents the potential difficulty in interpreting pathway manipulations made in vertebrate systems where two or more closely related proteins may exhibit functional redundancy. As such, the initial identification of the core TGF-β/BMP signaling components benefited from the genetically tractable Drosophila system (Sekelsky et al.,1995; Newfeld et al.,1996; Zhang et al.,1996; Botas,2007). In addition, the high degree of functional conservation between the Drosophila and vertebrate BMP signaling pathway components is underscored by the interchangeability of their respective signaling components at each level of the pathway (Padgett et al.,1993; Sampath et al.,1993; Penton et al.,1994; Brummel et al.,1994; Newfeld et al.,1996; Fritsch et al.,2010).
Previous work from our lab has demonstrated that Sax, the Drosophila ALK2 orthologue, has a dual function that both promotes and antagonizes BMP signaling (Bangi and Wharton,2006b). The ability of type I receptors from other organisms to antagonize BMP signaling has not been investigated, although recent studies suggest that ALK2 is able to inhibit activin signaling in MA-10 cells and inhibit BMP6-induced signaling in COS cells (Renlund et al.,2007; van Dinther et al.,2010). These reports coupled with the evolutionary relatedness of ALK2 and Sax raises the possibility that ALK2, like Sax, may have the ability to inhibit BMP signaling.
Here, we report on a series of studies that investigated the use of Drosophila as a model to assess the consequences of hyperactive receptor kinase activity associated with FOP and the molecular factors required therein. Our findings reveal that the Drosophila components are able to mediate hyperactive signaling by the mutant human receptor ALK2R206H ligand-independently, both in vivo and in Drosophila cell culture. Importantly, our findings have contributed to the mechanistic understanding of how defective FOP receptors signal by revealing that the type II receptor is a critical, molecular determinant required for ALK2R206H mutant receptor signaling. Additionally, we investigated the functional similarities between Sax and ALK2 and found that wild-type ALK2 is also able to block BMP signaling but it achieves this inhibition in a manner different from that employed by Sax. These results provide an important advance in our understanding of both the molecular events required for hyperactive signaling by a FOP mutant receptor and the wild-type behavior of the ALK2/ACVR1 receptor. Moreover, these studies provide a new tool for future investigations of the mechanistic attributes and the triggers responsible for activating FOP mutant receptors.
FOP Mutant Receptor ALK2R206H Stimulates Increased BMP Signaling in Drosophila
In order to test the ability of the human ALK2R206H classical FOP mutant receptor to signal in Drosophila, we generated transgenic lines that allowed us to control the expression of ALK2R206H in a tissue-specific manner (Brand and Perrimon,1993) and assayed for the ability of ALK2R206H to induce BMP signaling in the developing wing. It is known that BMP signaling plays critical roles in growth, patterning, and differentiation in the wing imaginal disc during wing development (Rogulja and Irvine,2005; Bangi and Wharton,2006a; O;Connor et al., 2006; Affolter and Basler,2007; Blair,2007; Rogulja et al.,2008; Oh and Irvine,2011; Schwank et al.,2011; Wartlick et al.,2011a; Wartlick et al.,2011b). In the primordial cells of the wing proper, a gradient of BMP signaling activity is generated through the action of two Drosophila BMP ligands, Dpp and Gbb, two type I receptors, Sax and Tkv, and the type II receptor, Punt. The resulting phospho-Mad (pMad) gradient reflects the output of BMP signaling and is critical for regulating its transcriptional targets.
The expression of ALK2R206H was directed primarily to the dorsal compartment of the wing imaginal disc by crossing ap-GAL4 or A9-GAL4 lines to a UAS-ALK2R206H transgenic line. Adults were obtained, albeit unable to fully emerge from the pupal case. The wings from these individuals were misshapen and marked by ectopic veins (Fig. 1B; data not shown). In larvae of the same genotype we found a higher level of pMad throughout the dorsal compartment of the wing imaginal disc compared with endogenous levels of pMad observed in the ventral compartment (Fig. 1C′,D). The presence of ectopic pMad observed in these discs indicate that ALK2R206H is able to stimulate BMP signaling in Drosophila imaginal disc tissues, presumably through the direct phosphorylation of the Drosophila Smad1/5/8 orthologue Mad by its own kinase activity (Fig. 1D). As expected from the known role of BMP signaling in tissue growth (Capdevila and Guerrero,1994; Haerry et al.,1998; Rogulja and Irvine,2005; Affolter and Basler,2007), we observed an increase in the size of the dorsal compartment of ap-GAL4>ALK2R206H wing discs (Fig. 1D), which likely results in the downwardly curved wings evident in adults due to an enlargement of the dorsal surface.
As in our in vivo studies, we found that ALK2R206H can induce an increase in BMP signaling in a quantitative cell-based BMP signaling assay (Fig. 1F). This cell-based assay makes use of a lacZ reporter construct under the control of the brinker silencer element (brkS), which is known to quantitatively repress transcription in response to Mad-mediated signaling (Müller et al.,2003; Bangi and Wharton,2006b; Twombly et al.,2009). S2 cells transfected with a plasmid construct encoding the Drosophila ligand Gbb exhibit a reduction in β-galactatosidase (β-gal) activity, reflecting the repression of lacZ transcription as a result of an increase in BMP signaling (Fig. 1F). Cells transfected with a construct encoding the FOP mutant receptor ALK2R206H showed very high levels of BMP signaling as evidenced by the dramatic reduction in β-gal activity (Fig. 1F).
Extracellular Ligand Binding Domain Is Not Required for Hyperactivity of ALK2R206H
The wild-type ALK2 receptor has been shown to promote Müllerian-inhibiting substance (MIS)-dependent signaling in mammalian systems (Clarke et al.,2001; Visser et al.,2001) and to bind the vertebrate ligands Activin and BMP7 (Attisano et al.,1993; ten Dijke et al.,1994). In zebrafish embryos and mammalian cells, the mutant ALK2R206H receptor has been reported to signal independently of BMP ligands (Billings et al.,2008; Fukuda et al.,2009; Shen et al.,2009). However, the possibility that the hyperactive nature of ALK2R206H depends on interactions with other ligands was not fully excluded in these studies. In Drosophila, orthologues are evident for both BMP ligand subfamilies (BMP2/4 = Dpp and BMP5/6/7 = Gbb) and for the TGF-β/Activin subfamily (Daw, Actβ, Myo, Mav; Kutty et al.,1998; Lo and Frasch,1999; Nguyen et al.,2000; Parker et al.,2004,2006; Moustakas and Heldin,2009). Given ALK2's promiscuity in binding ligands from different TGF-β families, we generated an ALK2R206H receptor that lacked the cysteine-rich (C38-C99), ligand-binding domain (LBD) to definitively test for the ability of ALK2R206H to signal independently of ligand in the Drosophila system. Cells transfected with the ALK2ΔLBD R206H construct were able to induce BMP signaling at comparable levels to that achieved by the full-length ALK2R206H receptor (Fig. 1F), indicating that the signaling activity of the FOP mutant receptor is not only ligand-independent but that it can promote BMP signaling despite lacking a large portion of its extracellular domain (ECD; Fig. 1E).
Hyperactivity of ALK2R206H Requires Type II Receptor Function
To transduce BMP signals, type I receptors must be activated through phosphorylation of its GS domain by type II receptors. Given the proximity of the R206H FOP mutation to the GS domain, we questioned whether the hyperactive nature of the ALK2R206H receptor depended on a phosphorylated GS domain. The importance of type II receptor kinase-mediated trans-phosphorylation of the ALK2R206H GS domain was tested by mutating GS domain Ser/Thr residues to Ala and assaying the mutated constructs for signaling activity (Fig. 2A,B). Both ALK2GS1-R206H (three Ser mutated to Ala) and ALK2GS2-R206H (three Ser and one Thr mutated to Ala) resulted in the abrogation of signaling as indicated by the failure of brkS-lacZ expression to be repressed (Fig. 2B). These results indicate that the Ser or Thr residues are critical for signaling, suggesting that their phosphorylation is required for the hyperactive signaling of the ALK2R206H receptor.
To specifically test the importance of type II receptor function for ALK2R206H hyperactivity, we made use of a UAS-put RNAi construct to knock down endogenous expression of the Drosophila type II receptor, Punt, in vivo. Directed expression of put RNAi to the dorsal compartment of the wing imaginal disc using ap-GAL4 resulted in a dramatic loss of pMad (Fig. 2C), consistent with the requirement for put in BMP signaling (Letsou et al.,1995; Ruberte et al.,1995). The elevated levels of pMad induced by expression of ALK2R206H in the dorsal wing compartment (Fig. 2D) were largely suppressed when put RNAi was coexpressed (Fig. 2E), demonstrating that the activity of ALK2R206H is dependent on the presence of type II receptor function in vivo.
We next tested if the activity of ALK2R206H depends on a specific type II receptor. Given that knocking down endogenous Punt completely suppressed the elevated pMad levels in wing discs induced by ALK2R206H, we tested for the ability of the other Drosophila type II receptor, Wit, to restore hyperactive signaling in this experimental context. Indeed, we found that expression of wit-HA with ALK2R206H and put RNAi led to elevated pMad, indicating that the ability of ALK2R206H to signal is not limited to a specific type II receptor (Supp. Fig. S1, which is available online). The ability of Wit to restore signaling by ALK2R206H in the absence of Punt is consistent with our finding that ALK2R206H signaling requires the presence of a type II receptor.
The Activating QD Mutation in BMP Type I Receptors Is Dependent on Type II Receptor Function
Previous reports have shown that mutation of the conserved Thr/Gln residue neighboring R206 (in ALK2) to Asp results in constitutive signaling in all members of the TGF-β/BMP family of type I receptors (Wieser et al.,1995; Attisano et al.,1996; Akiyama et al.,1997; Macías-Silva et al.,1998; Chen and Massagué,1999). In particular, the constitutive activity of TβR1T204D has been described as being independent of TGF-β type II receptor (TβRII) activity (Wieser et al.,1995). However, as shown above, we found that the presence of a type II receptor was absolutely required for the signaling hyperactivity associated with ALK2R206H. These conflicting observations led us to question whether activating mutations in other type I receptors have a requirement for type II receptors. Indeed, we found that unlike TβR1T204D, constitutive signaling produced by the Drosophila BMP type I receptor, Tkv, carrying the equivalent mutation (Gln-to-Asp; TkvQD) is type II receptor-dependent. The high levels of pMad induced by TkvQD are suppressed by knocking down Punt with put RNAi (Fig. 2F,G). These results suggest a fundamental difference in how TβRI and the BMP type I receptors, Tkv and Alk2, respond to an activating mutation. However, it remains a possibility that in previous studies BMP or Activin type II receptors could have substituted for TβRII to enable constitutive signaling by TβR1T204D (Wieser et al.,1995; Chen et al.,1997).
ALK2 Can Inhibit BMP Signaling Like its Drosophila Ortholog, Sax
Clearly, the FOP mutant receptor, ALK2R206H, exhibits high levels of signaling in Drosophila when expressed in vivo as well as in cell culture (Figs. 1, 2). We have previously shown that the Drosophila ALK2 orthologue, Sax, exhibits a dual function in its transduction of BMP signals (Bangi and Wharton,2006b). Given that wing phenotypes associated with ectopic expression of Gbb or Dpp can be suppressed by overexpression of Sax and enhanced by the loss of endogenous Sax, we proposed that Sax antagonizes BMP signaling by binding ligand into signaling incompetent receptor complexes. Consistent with this hypothesis, we found that Sax was able to block Gbb-induced signaling in a quantitative manner (Bangi and Wharton,2006b). This ability to inhibit BMP signaling is specific to Sax, as overexpression of Tkv enhances rather than inhibits signaling induced by either Gbb or Dpp. Despite detection of this inhibitory behavior, the complete loss of endogenous Sax results in a reduction in BMP signaling, indicating that Sax must also play a role in promoting signaling output. However, Sax must not be able to promote signaling on its own, given that all BMP signaling is eliminated in cells lacking the other type I receptor, Tkv. Given this unusual behavior of Sax and its evolutionary relatedness to ALK2, we considered the possibility that ALK2 mutations associated with FOP may in fact mask a normally occurring dual function of ALK2, such that the inhibitory function is lost and the FOP mutation manifests itself as one that hyperactivates ALK2.
To test for the ability of wild-type ALK2 to inhibit BMP signaling we first compared the effect of overexpressing wild-type ALK2 to that of Sax under conditions known to reveal the inhibitory function of Sax in vivo (Fig. 3A,B). In both cases, we observed a loss or thinning of longitudinal vein 5 (L5), a phenotype associated with a loss of gbb function (Wharton et al.,1999), as well as a reduction in the overall size of the wing. In general, the ectopic expression of ALK2 produced phenotypes indicative of a more severe reduction in BMP signaling than those achieved by overexpression of sax, including a greater reduction in wing size and the additional loss of L4 (Fig. 3A, B). Consistent with this, overexpression of ALK2 results in a greater reduction in pMad than that observed with overexpression of sax (Fig. 3C-D′). Thus, like overexpression of Sax, overexpression of ALK2 leads to an effective reduction in BMP signaling.
We and others have observed that Sax binds Gbb more effectively than Dpp, and as such we found that Sax can suppress the wing phenotypes produced by overexpression of Gbb better than those produced by the overexpression of Dpp (Haerry et al.,1998; Bangi and Wharton,2006b; Haerry,2010). ALK2 has been shown to bind BMP7 but not BMP4 (ten Dijke et al.,1994; Macías-Silva et al.,1998; Greenwald et al.,2003) and given the evolutionary relatedness of Gbb and Dpp to BMP7 and BMP4, respectively (Sampath et al.,1993; Fritsch et al.,2010), we hypothesized that ALK2 would be able to effectively inhibit Gbb-induced BMP signaling. As observed previously, A9-GAL4>UAS-gbb resulted in an array of wing phenotypes marked by ectopic vein material, indicative of an increase in BMP signaling. The distribution of wing phenotypes is shifted toward less severe phenotypic classes when sax is coexpressed with gbb (Bangi and Wharton,2006b; Fig. 4A). In a second set of experiments we made use of this phenotypic assay to test for the ability of ALK2 to antagonize signaling. We found that not only did coexpression of ALK2 with gbb suppress wing defects associated with ectopic Gbb signaling but that all A9-GAL4>UAS-ALK2; UAS-gbb wings exhibited phenotypes consistent with a decrease in endogenous BMP signaling, such as a reduction in wing size and a loss of longitudinal vein material (class 6; Fig. 4A). An examination of pMad distribution in the wing disc confirmed this conclusion as not only was ectopic pMad induced by gbb overexpression eliminated, but pMad associated with endogenous BMP signaling was also dramatically reduced (Fig. 4B).
The ability of ALK2 to inhibit BMP signaling was also tested in the quantitative, cell-based BMP signaling assay. Cotransfection of either sax or ALK2 with gbb resulted in a suppression of Gbb-induced signaling, indicating that both receptors were capable of inhibiting signaling (Fig. 4C). Interestingly, we found that signaling induced by transfections with either dpp or human BMP4 was also inhibited by ALK2 (Fig. 4D) whereas signaling induced by transfection of mouse BMP7 (see Supp. Fig. S2) was enhanced by ALK2, indicating that the ability of ALK2 to inhibit or promote BMP signaling is ligand-specific (Fig. 4D,E).
Ligand Binding Is not Required for ALK2-mediated Inhibition
The fact that ALK2 inhibits signaling induced by BMP4, a ligand it does not bind, raised the possibility that ALK2 may only antagonize signaling induced by ligands that do not interact with ALK2. We investigated this possibility by testing the ability of ALK2 to bind Gbb and Dpp by co-immunoprecipitation. While the expected association between Gbb and Sax (Fig. 5A, lane 2) was apparent, as was a strong interaction between Dpp and Tkv (Fig. 5B, lane 11), we were not able to detect an interaction between ALK2 and either Gbb or Dpp (Fig. 5A, lane 4 & Fig. 5B, lane 13, respectively). It is possible that the affinity of ALK2 for the Drosophila BMP ligands was below the detectable limit of co-immunoprecipitations. Therefore, to more rigorously test for the importance of ALK2-ligand interactions, we deleted the cysteine-rich (C38-C99) ligand-binding domain of ALK2 (ALK2ΔLBD; Fig. 5C) and tested for the ability of this mutated receptor to block signaling. Interestingly, we found that ALK2ΔLBD was able to effectively block Gbb-induced signaling in S2 cells (Fig. 5D), indicating that the ability of ALK2 to block BMP signaling is independent of a direct interaction with ligand and, for that matter, independent of a large portion of its extracellular domain. Thus, ALK2 must inhibit BMP signaling by a mechanism other than sequestration of ligands into signaling incompetent receptor complexes.
The BMP signaling pathway exhibits a high degree of conservation throughout metazoans (Newfeld et al.,1999). Consistent with this, we found that when the mutant form of the human ALK2 type I receptor (ALK2R206H), which is associated with the vast majority of fibrodysplasia ossificans progressiva (FOP) cases, (Shore et al.,2006; Billings et al.,2008) is expressed in Drosophila, it mimics the misregulation of BMP signaling previously described in vertebrate systems (Billings et al.,2008; Fukuda et al.,2009; Shen et al.,2009; Song et al.,2010; van Dinther et al.,2010). Our results provide clear evidence that the Drosophila BMP signaling components are compatible with the human ALK2 type I receptor, such that the classic FOP mutation (R206H) manifests as hyperactive BMP signaling in Drosophila tissues as well. This finding bodes well for the use of the Drosophila system as a future tool to elucidate the molecular details responsible for misregulated BMP signaling associated with FOP despite the obvious differences in the ultimate consequence of this hyperactive signaling in Drosophila compared to heterotopic bone formation in mammals.
Hyperactive BMP Signaling Requires Type II Receptor Function
Consistent with experiments conducted in mammalian cells, we found that, in Drosophila, ALK2R206H is able to induce high levels of phosphorylated Mad in the absence of ligand binding. Importantly, we found that the signaling hyperactivity of ALK2R206H requires the function of a type II receptor kinase, which is responsible for activating type I receptors at their GS domain (Fig. 6A). The ability of ALK2R206H to induce high levels of pMad is abrogated when the activation domain (GS domain) of the ALK2 receptor is mutated. While it remains possible, in the various Drosophila assay systems used in this study, that the endogenous Drosophila type I receptors Sax or Tkv are actually responsible for Mad phosphorylation in response to ALK2R206H expression, it is unlikely considering that the level of these receptors must be far lower than that of overexpressed ALK2R206H. Furthermore, our demonstration that the Ser/Thr residues within the GS domain of the ALK2R206H receptor must be retained for hyperactive BMP signaling further supports the conclusion that ALK2R206H itself is responsible for elevated pMad in our studies. The fact that ALK2R206H can signal independently of ligand, but requires type II receptor function, indicates that ALK2R206H must be able to interact with type II receptors independently of ligand. While interactions between type I and type II receptors have generally been thought to be induced by complex formation with secreted ligands (Groppe et al.,2008; Nickel et al.,2009), ligand-independent interactions have also been reported (Haerry,2010). Furthermore, BMP type I and type II receptors can interact independently of ligand to generate preformed complexes that then bind ligand to initiate signal transduction (Nohe et al.,2002; Hassel et al.,2003; Ehrlich et al.,2011; Marom et al.,2011). Taken together, we envision a model in which the classic FOP mutation exposes the serine/threonine residues in the GS domain to phosphorylation by the type II receptor, thus circumventing the requirement for ligand to activate signaling (Fig. 6A).
Interestingly, our results show that the constitutively active BMP type I receptor TkvQD also shows a dependency for type II receptor function (Fig. 2G). This result is in contrast to that previously shown for the constitutively active TβR1T204D which signals in the absence of the TGF-β type II receptor, TβR-II (Wieser et al.,1995; Chen et al.,1997). While it has not yet been definitively shown that TβR1T204D signals independently of BMP or Activin type II receptors, these apparently conflicting data could reflect a fundamental difference in either the requirement for type II receptors or for the interaction of type I and type II receptors in BMP versus TGF-β signaling. Other key distinctions between TGF-β and BMP receptor signaling have been previously noted. For example, structural studies have shown that in BMP receptor complexes, the extracellular domains of the BMP type I and II receptors do not contact one another, whereas in TGF-β receptor complexes, an N-terminal extension in the extracellular domain of TGF-β type II receptors directly interacts with the TGF-β type I receptor (Kirsch et al.,2000; Allendorph et al.,2006; Groppe et al.,2008). In addition, the minimal receptor complex required for BMP versus TGF-β signaling appears to differ such that a heterotrimeric (type I:type II:type II) BMP receptor complex is required to transduce BMP signals (Isaacs et al.,2010), whereas autonomously functioning TβRI:TβRII (type I:type II) heterodimers have been shown to be sufficient for transducing TGF-β signals (Huang et al.,2011).
In addition to divergent type II receptor requirements, mutations that confer hyperactivity or constitutive activity to TGF-β/BMP type I receptors differ in their respective effect on binding of the intracellular inhibitor FKBP12 to the type I receptor. FKBP12 has been proposed to prevent “leaky” ligand-independent signaling by masking the GS domain in absence of ligand binding (Chen et al.,1997; Huse et al.,1999,2001; Wang and Donahoe,2004). In a number of experiments, it has been shown that the R206H mutation reduces binding of FKBP12, making this an attractive molecular explanation for the hyperactivity displayed by ALK2R206H (Groppe et al.,2007,2011; Shen et al.,2009; Song et al.,2010). In contrast, the constitutively active Q207D mutation in ALK2 does not disrupt binding of FKBP12, whereas the equivalent constitutively active mutation in TβR1(T204D) does (Chen et al.,1997). In the case of the Drosophila FKBP12 orthologue FKBP2, our preliminary studies indicate that the loss of FKBP2 function in vivo did not produce phenotypes consistent with a substantial increase in BMP signaling (V. Le, S. Ballard, data not shown). Taken together, there does not appear to be a clear correlation between a loss or disruption of FKBP12 binding and the hyperactivity of mutant type I receptors. While we do not yet understand the mechanisms underlying the differential association of FKBP12 with ALK2R026H versus ALK2Q207D, such differences raise the possibility that in vivo, the constitutively active ALK2Q207D receptor behaves differently from the ALK2R206H FOP mutant receptor.
The finding that ALK2R206H hyperactive signaling depends on type II receptor function has not previously been appreciated in studies of FOP and provides a new angle in the search for FOP therapeutics. Current strategies for drug development have focused on identifying small molecule inhibitors of type I receptor kinase activity (Yu et al.,2008a,b; Hao et al.,2010). One such inhibitor, dorsomorphin, has been shown to effectively inhibit ALK2R206H kinase activity (Yu et al.,2008a; Fukuda et al.,2009; Shen et al.,2009; van Dinther et al.,2010), but unfortunately dorsomorphin nonspecifically inhibits the kinase activity of other BMP type I receptors and appears to exhibit “off-target” effects on VEGF signaling (Yu et al.,2008b; Hao et al.,2010). In addition to future efforts to improve the selectivity of dorsomorphin analogs (Hao et al.,2010), alternative approaches that disrupt FOP-induced signaling are needed. An exciting new prospect for drug development could exploit our recently identified type II receptor requirement for ALK2R206H hyperactivity by focusing on the identification of molecules or factors that specifically block the interaction between ALK2R206H and type II receptors in FOP cells.
Wild-Type ALK2 Receptor Can Inhibit BMP Signaling
In addition to our studies of ALK2R206H in the Drosophila system, we analyzed the ability of wild-type ALK2 to mediate signaling. Since FOP is a dominant autosomal disease and all mutations isolated thus far are protein-coding point mutations, the FOP mutant receptors must always be expressed in the presence of wild-type ALK2 receptor. In order to understand the mechanistic underpinnings of FOP it is therefore critical that we have a full understanding of wild-type ALK2 receptor function in addition to elucidating the consequences of the R206H mutation. Thus, we investigated the possibility that ALK2 can both promote and antagonize signaling, a behavior exhibited by the Drosophila ALK2 orthologue, Sax (Bangi and Wharton,2006b). Our results revealed that wild-type ALK2 receptor is indeed able to inhibit BMP signaling in vivo as well as in Drosophila cell culture. Interestingly, we found that the mechanism by which ALK2 accomplishes signaling inhibition likely differs from that employed by Sax. Whereas Sax likely inhibits signaling by means of incorporating its high-affinity ligand, Gbb, into inactive complexes, ALK2 appears to inhibit signaling induced by ligands that ALK2 itself does not actually bind (i.e. Gbb, Dpp, and BMP4). Furthermore, ALK2 has been observed to inhibit signaling induced by BMP6 (van Dinther et al.,2010), a ligand that ALK2 has not been reported to bind. Based on these observations, we propose that ALK2 inhibits signaling by interacting with a type II receptor, such as Punt, and prevents binding of ligands, such as Dpp, Gbb, BMP4, or BMP6, to the ALK2/Punt complex (Fig 6b, left). A similar mechanism has been proposed to explain the negative effect of ALK2/ACVR1 on signaling induced by Activin, which acts through a different set of core signaling components (Renlund et al.,2007). Conversely, in the case of BMP7, a ligand that has been shown to bind ALK2 (ten Dijke et al., 1994; Greenwald et al.,2003), the presence of ALK2 in complex with the type II receptor would facilitate BMP7 binding and enhance BMP7-induced signaling (Fig. 6A). In our model, we propose that ALK2 acts as a modifier of receptor complex activity by dictating which BMP ligand can or cannot bind the complex of ALK2 and type II receptor. Through this function of imparting ligand specificity, ALK2 ultimately determines whether the receptor complex in which a type II receptor participates is active or inactive, depending on which BMP ligand is present (Fig. 6B, left). Moreover, in the absence of BMP7 and under conditions where ALK2 becomes enriched over other type I receptors at the cell membrane, BMP signaling will be suppressed as ALK2 titrates type II receptors (Fig. 6B, right). Therefore, the ability of ALK2 to regulate signaling based on BMP ligand type may have a profound impact on ligand-specific responses and warrants further investigation to determine if this dual behavior of ALK2 is observed in tissues that normally express ALK2.
On a separate note, the inability of ALK2 to bind Gbb was unexpected based on the demonstrated ability of ALK2 to bind BMP7 and the evolutionary relatedness of Gbb to the BMP5/6/7 subgroup. While the conserved domains of BMP5, BMP6, and BMP7 can reportedly rescue gbb mutant phenotypes (Fritsch et al.,2010), our results suggest that it is unlikely that Gbb can fully substitute for BMP7 function in vertebrates, specifically for BMP7-induced signaling mediated by ALK2.
Impact of Drosophila Models for the Study of FOP
Perhaps the least well-understood aspect of FOP, and most difficult for patients, is the sporadic and progressive nature of the disease. One of the primary difficulties still facing the FOP field is reconciling the molecular events of hyperactive signaling induced by the FOP receptor in animal models with the clinical features that manifest in patients. The sporadic nature of the disease contrasts with the hyperactivity that the mutant receptor displays in experimental assays suggesting that under endogenous conditions the activity of the mutant receptor must be regulated or muted until an event triggers a flare-up.
To date, all FOP patients are heterozygous for mutations in ALK2 regardless of whether they harbor the classic R206H or an atypical mutation. It is possible that one copy of ALK2WT can compete with ALK2FOP receptors for type II availability, thereby keeping final output of BMP signaling below a threshold required for bone formation. Therefore, in an endogenous context, the relative ratio between FOP type I receptors, wild-type type I receptors, and type II receptors could be the determining factor in whether or not activation of the pathway reaches a threshold necessary for bone formation. As such, it is possible that at physiological levels the FOP mutant receptor activity could be inhibited by a factor in trans and only when the mutant receptor is overexpressed under experimental conditions does it escape this negative regulation. Thus, in the future, it will be important to study the behavior of the FOP mutant receptors at physiologically relevant levels achievable through homologous recombination. Making use of the Drosophila model system to express both mutant and wild-type receptors at endogenous levels will enable in vivo mutagenic screens to identify factors that suppress or enhance the effects of the ALK2R206H activity and in turn provide us with new targets for therapy and treatment. Furthermore, given the correlation between ossification and trauma, it has been suggested that inflammation associated with injury may in some way trigger heterotopic ossification. Such a triggering event could act to increase the accessibility of the GS domain to the kinase activity of the type II receptor by disrupting the binding of a putative inhibitor or by influencing the ability of the ALK2R206H receptor to interact with available type II receptors (Fig. 6C). Although the precise mechanism(s) by which such putative modulators may influence the behavior of ALK2R206H remains unknown, the Drosophila system is a particularly attractive model organism in which to undertake such studies given the high conservation of pathways governing cellular physiology.
In closing, our work has demonstrated the value of using a Drosophila genetic system to study the molecular foundation of altered BMP signaling characteristic of individuals with FOP. Our experiments reveal a requirement for type II receptor function in the hyperactivity displayed by the ALK2R206H mutant receptor, a fact previously unappreciated. Although the majority of FOP patients are characterized by the classic R206H mutation, a small but growing list of variant mutations in other domains of the receptor are being identified (Billings et al., 2008; Furuya et al.,2008; Bocciardi et al.,2009; Petrie et al.,2009; Ohte et al.,2011). Whether the activity of these variant mutant receptors also requires type II receptor function will be the subject of future studies. Lastly, we have also observed the ability of the wild-type ALK2 receptor to inhibit BMP signaling in a ligand-specific manner. How these findings contribute to the sporadic nature of FOP as well as impact our broad understanding of other diseases associated with misregulated type I receptor activity warrants further investigation. We intend to exploit the comprehensive genetic tools in Drosophila system to screen for potential modifiers of FOP mutant receptor activity as a means to bridge this gap.
Gateway cloning (Invitrogen) was used to clone all cDNAs into the following Drosophila Gateway Vectors: pTWF for GAL4-UAS driven expression in transgenic animals and the Actin5C vector pAWF (C-terminal 3xFLAG) for constitutive expression in cell culture. ALK2 and ALK2R206H cDNAs were a generous gift from Eileen Shore. pAW gbb, pAW dppHA, and pAW hBMP4 were constructed by Takuya Akiyama.
Ligand-binding domain deletions.
Ligand-binding domain deletion mutants were generated by Quikchange Site-directed Mutagenesis (Stratagene). For ALK2ΔLBD, sequences corresponding to Cys35 to Cys 99 were removed using the following primers: fwd 5′-CAA CCC CAA ACT CTA CAT GAA CAG GAA CAT CAC GGC C-3′ and rev 5′-GGC CGT GAT GTT CCT GTT CAT GTA GAG TTT GGG GTT G-3′. For Sax, sequences corresponding to Cys67 to Cys148 were removed using the following primers: fwd 5′-CGC ATC CCA GAT ACA AAA ATG AGG GAG ACT TTC C-3′ and rev 5′-GGA AAG TCT CCC TCA TTT TTG TAT CTG GGA TGC G-3′.
GS domain mutants.
Two sets of GS domain mutations were generated in ALK2R206H (Quikchange Site-Directed Mutagenesis). ALK2GS1-R206H: all three serines were mutated to alanine (TSGSGSG > TAGAGAG) using the following primers: (ALK2 S190,192,194A fwd) 5′-CAG ATT TAT TGG ATC ATT CGT GTA CAG CAG GAG CTG GCG CTG GTC TTC CTT TTC TGG TAC-3′ and (ALK2 S190,192,194A rev) 5′-GTA CCA GAA AAG GAA GAC CAG CGC CAG CTC CTG CTG TAC ACG AAT GAT CCA ATA AAT CTG-3′. ALK2GS2-R206H: a threonine and all three serines were mutated to alanine (TSGSGSG > AAGAGAG) using the following primers: (ALK2 T189A S190,192,194A fwd) 5′-CAG ATT TAT TGG ATC ATT CGT GTG CAG CAG GAG CTG GCG CTG GTC TTC CTT TTC TGG TAC-3′ and (ALK2 T189A S190,192,194A rev) 5′-GTA CCA GAA AAG GAA GAC CAG CGC CAG CTC CTG CTG CAC ACG AAT GAT CCA ATA AAT CTG-3′.
Drosophila melanogaster Strains and Crosses
All fly strains were cultured using standard sucrose, yeast extract agar food at 25°C. All fly strains are described in Flybase and obtained from Bloomington Stock Center except where noted: UAS-gbb9.1 (Khalsa et al.,1998), A9-GAL4, UAS- tkvQD (Haerry et al.,1998), UAS-wit-HA31 (Michael O'Connor). UAS-putRNAi (from NIG-FLY, NIG 7904 R-2D). UAS-sax-3xFLAG(1-1M-A), UAS-ALK2-3xFLAG(8-1-9M-1a), and UAS-ALK2R206H-3xFlag(3-4F1-a) were germline transformants derived from constructs described above.
Receptor and gbb Overexpression
Receptors and gbb were overexpressed using the GAL4-UAS system (Brand and Perrimon,1993). A9-GAL4 and ap-GAL4 drivers express primarily in the dorsal compartment of the wing imaginal disc.
In Vivo Gbb Signaling Assay
A previously described in vivo assay (Bangi and Wharton,2006b) was used to test for the ability of BMP type I receptors to affect Gbb signaling. Adult wings from the following genotypes were mounted (DPX, EM Sciences) and scored: w A9-GAL4/yw; +/+; UAS-gbb9.1/+ were compared with w A9-GAL4/yw, UAS-sax(1-1M-A)/+; UAS-gbb9.1/+ and w A9-GAL4/yw; UAS-ALK2(8-1-9M-1a)/+; UAS-gbb9.1.
Everted late third instar larvae were dissected and fixed in 4% paraformaldehyde/phosphate buffered saline (PBS; v/v) for 20 min at room temperature followed by 5 washes in PBST (0.3% Triton X-100). Fixed tissues were then incubated overnight in blocking solution (10% normal goat serum in PBST) at 4°C. After blocking, the cuticles were incubated overnight in primary antibody diluted in blocking solution at the following dilutions: 1:1,000 anti-FLAG M2 (Sigma, F3165), 1:1,000 anti-HA 3F10 (Roche) and 1:1,000 anti-PS3 (Epitomics, 1880-1). Tissues were then washed 5 times with PBST and incubated overnight in secondary antibody in blocking solution at the following dilutions: 1:1,000 GAM Alexa Fluor 633, 1:1,000 GARt Alexa Fluor405 (in WitHA experiments), 1:1,000 GARb Alexa Fluor568. Following 5 washes in PBST, wing discs were removed and mounted in 80% glycerol/0.5% N-propyl gallate. Confocal images were collected using a Zeiss LSM510 Meta confocal laser scanning microscope.
Drosophila Schneider 2 (S2) Cell Maintenance and Transfections
S2 cells were cultured in Shields and Sang M3 Insect Medium (Sigma S8398) pH 6.5 containing 10% Insect Medium Supplement (Sigma I7267) and 2% fetal bovine serum (F3018). Transient transfections were carried out using Effectene Transfection Reagent (Qiagen 301427).
Quantitative Cell-Based BMP Signaling Assay
An adapted protocol based on a previously described assay was used to measure BMP signaling activity (Bangi and Wharton,2006b; Müller et al.,2003; Twombly et al.,2009). This assay makes use of a reporter construct expressing lacZ under the control of a Su(H) transcriptional activation response element as well as a brk transcriptional silencer element (Su(H)/brkS-lacZ). Cotransfection of the reporter construct with plasmids encoding Su(H) and an activated form of Notch (N*) lead to lacZ transcription while the activation of BMP signaling leads to a repression of lacZ expression by virtue of the BMP-responsive brk silencer element. BMP signaling can thus be measured as a loss of β-galactosidase activity. 2.8 × 106 cells were cotransfected with Su(H), N*, Su(H)/brkS-lacZ, and luciferase plasmids, all under the control of the Actin 5C promoter. Constructs and their concentrations used in this assay are indicated in the figure legends. Cells were harvested and lysed 3 days post-transfection and β-galactosidase activity of cleared lysate was measured using the dual luciferase assay system (Dual-Light, Applied Biosystems) and normalized to luciferase activity, which served as a transfection control for each sample. The normalized value obtained from the cleared lysate of cells cotransfected with only Su(H), N*, Su(H)/brkS-lacZ and luciferase was set to 100%. Statistical significance was determined using two-tailed t-test with significance value of P < 0.05. Epitope tagged and untagged versions of the type I receptors investigated in this study were compared for signaling activity and showed no significant difference (data not shown).
8 × 106 S2 cells were transiently transfected with 1 μg total DNA at the following ratios: Receptor–Ligand interaction - 300 ng pAWF type I receptor constructs and 700 ng of either pAW dppHA or pAW gbb; cells were incubated at 25°C for 4 days for protein production. Cells were solubilized in 1% Triton X-100 at 4°C for 1 hr. Cleared lysate was incubated with 1ug anti-FLAG M2 (Sigma F3165) bound to 20 μl of Dynabeads Protein G Dynabead (Invitrogen 100-04D) per sample at 4°C for 1 hr. An aliquot of cleared lysate was saved as soluble input for Western blot analysis. The beads were then washed once with one volume of Wash Buffer 1 (20 mM Tris-HCl pH 7.4, 150 mM NaCl, 0.2% Triton X-100), twice with one volume of Wash Buffer 2 (20 mM Tris-HCl pH 7.4, 150 mM NaCl), and boiled for 5 min in 50 μl of 2 × sodium dodecyl sulfate (SDS) buffer. IP and soluble input fractions were run on 12% SDS-polyacrylamide gels and analyzed by western blot using standard protocols. Anti-HA 3F10 (Roche) was used at 0.1 ng/μl. Anti-Flag M2 (Sigma) was used at 4 ng/μl, mouse anti-Gbb (gift from Guillermo Marquez) was used at a 1:1,000 dilution. Secondary antibodies GAM IgG-horseradish peroxidase (HRP) light-chain specific (Jackson) and GARat HRP (Jackson, preabsorbed) were used at a 1:10,000 dilution.
pMad intensity profiles.
Intensity profiles of pMad distribution were measured in confocal Z stacks of wing discs using the Fiji Image Processing Package (http://fiji.sc/wiki/index.php/Fiji). The profiles shown are the average intensity plots measured in the dorsal and ventral compartments of five wing discs and aligned by the posterior and anterior peaks of pMad distribution in the ventral compartment.
We thank Eileen Shore for ALK2 and ALK2R206H clones, Petra Seeman for the BMP7 clone and Michael O'Connor for the punt clone. We enjoyed fruitful discussions with Drs. E. Shore, F. Kaplan, and J. Groppe. We also thank members of the Wharton lab for their input during this project. We are indebted to the many FOP patients and Dr. Fred Kaplan who inspired us to pursue this work. This work was supported in part by a generous Cali Family Developmental Grant from The Center for Research in FOP and Related Disorders in The Department of Orthopedic Surgery at The Perelman School of Medicine of The University of Pennsylvania and by a grant from the NIH to K.A.W. and a NIH Training Grant to V.Q.L.