Anti-ACVR1 antibodies exacerbate heterotopic ossification in fibrodysplasia ossificans progressiva (FOP) by activating FOP-mutant ACVR1

Abstract

Fibrodysplasia ossificans progressiva (FOP) is a rare genetic disorder whose most debilitating pathology is progressive and cumulative heterotopic ossification (HO) of skeletal muscles, ligaments, tendons, and fascia. FOP is caused by mutations in the type I BMP receptor gene ACVR1, which enable ACVR1 to utilize its natural antagonist, activin A, as an agonistic ligand. The physiological relevance of this property is underscored by the fact that HO in FOP is exquisitely dependent on activation of FOP-mutant ACVR1 by activin A, an effect countered by inhibition of anti-activin A via monoclonal antibody treatment. Hence, we surmised that anti-ACVR1 antibodies that block activation of ACVR1 by ligands should also inhibit HO in FOP and provide an additional therapeutic option for this condition. Therefore, we generated anti-ACVR1 monoclonal antibodies that block ACVR1's activation by its ligands. Surprisingly, in vivo, these anti-ACVR1 antibodies stimulated HO and activated signaling of FOP-mutant ACVR1. This property was restricted to FOP-mutant ACVR1 and resulted from anti-ACVR1 antibody-mediated dimerization of ACVR1. Conversely, wild-type ACVR1 was inhibited by anti-ACVR1 antibodies. These results uncover an additional property of FOP-mutant ACVR1 and indicate that anti-ACVR1 antibodies should not be considered as therapeutics for FOP.

Keywords: Bone Biology; Bone disease; Genetic diseases; Signal transduction; Therapeutics.

Figure 1. Anti-ACVR1 antibodies block BMP7 and activin A signaling in HEK293.ACVR1[R206H] cells but increase heterotopic bone formation in FOP mice.

Activin A and BMP7 dose response was evaluated in stable pools of HEK293/BRE-luciferase reporter cells overexpressing ACVR1[R206H] (A). HEK293/BRE-luciferase reporter cells overexpressing ACVR1[R206H] were treated with a fixed concentration (2 nM) of BMP7 (B) or activin A (C). Anti-ACVR1 antibodies inhibited Smad1/5/8 phosphorylation induced by BMP7 or activin A (B and C). Data show the mean (n = 4) ± SEM. Three biological replicates were performed for the in vitro signaling assays. (D) Acvr1^{[R206H]FlEx/+}; GT(ROSA26)Sor^CreERT2/+ mice were injected with tamoxifen to initiate the model and concurrently injected with anti-ACVR1 antibodies or isotype control antibody at 10 mg/kg weekly (n = 7–8/group). Total heterotopic bone lesion volume was measured 4 weeks after initiation. Data show the mean ± SD. *P < 0.05, **P < 0.01 by 1-way ANOVA with Dunnett’s multiple-comparison test. (E) Representative μCT images of FOP mice [Acvr1^{[R206H]FlEx/+}; GT(ROSA26)Sor^CreERT2/+, after tamoxifen] treated with anti-ACVR1 antibody or isotype control antibody. Yellow arrows indicate the positions of heterotopic bone lesions.

Figure 2. Anti-ACVR1 antibody–induced changes in hepcidin and iron levels are consistent with inhibition of WT ACVR1 and activation of ACVR1[R206H] in vivo.

(A and C) In WT mice (n = 8/group), anti-ACVR1 mAb 1 decreased serum hepcidin (A) and increased serum iron (C). (B and D) In FOP mice [Acvr1^{[R206H]FlEx/+}; GT(ROSA26)Sor^CreERT2/+, after tamoxifen] (n = 5–6/group), anti-ACVR1 mAb 1 increased serum hepcidin (B) and decreased serum iron (D). ***P < 0.001 by Student’s t test.

Figure 3. Ligand-independent dimerization of ACVR1[R206H], but not WT ACVR1, induces Smad1/5/8 signaling.

HEK293 cells harboring p-Smad1/5/8–responsive luciferase reporter (BRE) were transfected with hACVR1-DmrB (A) or hACVR1[R206H]-DmrB (B). Homodimerization of C-terminally DmrB-tagged ACVR1 was induced with 20 nM B/B homodimerizer for 16 hours. Activin A activated Smad1/5/8 signaling only in hACVR1[R206H]-DmrB cells, but BMP6 activated Smad1/5/8 signaling both in hACVR1-DmrB and hACVR1[R206H]-DmrB cells (A and B). Intracellular homodimerization of hACVR1[R206H] activated Smad1/5/8 signaling in the absence of exogenous ligands (C) as well as in the presence of 300 nM ACVR2B-Fc ligand trap (D). Data show the mean (n = 4) ± SEM. Three biological replicates were performed for the in vitro signaling assays.

Figure 4. Dimeric anti-ACVR1 antibodies activate, whereas monomeric anti-ACVR1 Fabs block, ACVR1[R206H].

(A) Acvr1^{[R206H]FlEx/+}; GT(ROSA26)Sor^CreERT2/+ mice (n = 7–9/group) received plasmids expressing anti-ACVR1 Fabs or a plasmid encoding a control mAb by hydrodynamic delivery (HDD) 5 days after initiation of the model with tamoxifen. HO was triggered in the hind limb by muscle pinch 7 days after HDD and total heterotopic bone volume was measured 6 weeks after injury. FOP mice [Acvr1^{[R206H]FlEx/+}; GT(ROSA26)Sor^CreERT2/+, after tamoxifen] expressing anti-ACVR1 Fab showed reduced HO compared with control mice. Data show the mean ± SD. *P < 0.05 by 1-way ANOVA with Dunnett’s multiple-comparison test. (B) Representative μCT images of FOP mice expressing either anti-ACVR1 Fab or an isotype control antibody. (C) Acvr1^[R206H]/+; GT(ROSA26)Sor^CreERT2/+ ([R206H]/+) mES cells (mESC) were treated with activin A, anti-ACVR1 mAb 2, anti-ACVR1 Fab 2, or anti-activin A mAb (REGN2476) in various combinations for 1 hour. Activin A and anti-ACVR1 mAb 2 but not anti-ACVR1 Fab 2 induced Smad1/5/8 phosphorylation. Anti-ACVR1 Fab 2 significantly reduced activin A–induced Smad1/5/8 phosphorylation, whereas anti-ACVR1 mAb 2 only slightly reduced activin A–induced Smad1/5/8 phosphorylation. (D) Anti-ACVR1 antibody activation of ACVR1[R206H] is independent of activin A. Acvr1^{[R206H]FlEx/+}; GT(ROSA26)Sor^CreERT2/+ mice (n = 6–8/group) were injected with tamoxifen to initiate the model and concurrently injected with antibodies at 10 mg/kg weekly. Total heterotopic bone volume was measured 3 weeks after initiation. Data show the mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001 by 1-way ANOVA with Dunnett’s multiple-comparison test.

Figure 5. Anti-ACVR1 antibody activation of ACVR1[R206H] is type II receptor dependent.

(A) Acvr1^[R206H]/+; GT(ROSA26)Sor^CreERT2/+ ([R206H]/+) mES cells lacking Acvr2a plus Acvr2b, or Bmpr2 or all 3 of these type II receptor genes were treated with 10 nM activin A, BMP7, BMP2, BMP10, or anti-ACVR1 mAb 1 for 1 hour. Activin A, BMP7, BMP2, BMP10, and anti-ACVR1 mAb 1 induced Smad1/5/8 phosphorylation in cells that lack Bmpr2 but retain Acvr2a and Acvr2b, but not in cells where Acvr2a and Acvr2b or all 3 type II receptors have been knocked out. (B) ACVR2B coimmunoprecipitates with both ACVR1 and ACVR1[R206H] from W20 cells expressing Myc-tagged ACVR1 and/or HA-tagged ACVR2B. Myc-ACVR1 was immunoprecipitated using an anti-Myc antibody. ACVR1 and ACVR2B were detected using an anti-ACVR1 or anti-HA antibody, respectively.

Figure 6. ACVR1[R206H;S330P] is activated by anti-ACVR1 antibodies but to a lesser degree than ACVR1[R206H].

(A and B) Acvr1^{[R206H]FlEx/+}; GT(ROSA26)Sor^CreERT2/+ mice or Acvr1^{huecto;[R206H]FlEx;[S330P]/+}; GT(ROSA26)Sor^CreERT2/+ (FOP^[S330P]) mice were injected with tamoxifen to initiate the model and concurrently injected with anti-ACVR1 mAb 1 or isotype control antibody at 10 mg/kg weekly (n = 8/group). Total heterotopic bone volume was measured 3 weeks after initiation of the model. ACVR1 mAb 1 increased HO compared with isotype control in both mouse models, though to a lesser degree in FOP^[S330P] mice. Data show the mean ± SD. *P < 0.05, ***P < 0.001 by 1-way ANOVA with Bonferroni’s multiple-comparison test. (C) Acvr1^{huecto;[R206H;S330P]/+}; GT(ROSA26)Sor^CreERT2/+ ([R206H;S330P]/+) mES cells were treated with 10 nM activin A, anti-ACVR1 mAb 1, or anti-hACVR1 antibody and assessed for phosphorylated Smad1/5/8. Anti-hACVR1 mAb (that only binds ACVR1[huecto;R206H;S330P] and not WT mouse ACVR1) induced Smad1/5/8 phosphorylation, whereas mAb 1 (which recognizes both human and mouse ACVR1) did not drive an appreciable level of Smad1/5/8 phosphorylation.