Activin-dependent signaling in fibro/adipogenic progenitors causes fibrodysplasia ossificans progressiva

Abstract

Fibrodysplasia ossificans progressiva (FOP) is a rare autosomal-dominant disorder characterized by progressive and profoundly disabling heterotopic ossification (HO). Here we show that fibro/adipogenic progenitors (FAPs) are a major cell-of-origin of HO in an accurate genetic mouse model of FOP (Acvr1 ^tnR206H ). Targeted expression of the disease-causing type I bone morphogenetic protein (BMP) receptor, ACVR1(R206H), to FAPs recapitulates the full spectrum of HO observed in FOP patients. ACVR1(R206H)-expressing FAPs, but not wild-type FAPs, activate osteogenic signaling in response to activin ligands. Conditional loss of the wild-type Acvr1 allele dramatically exacerbates FAP-directed HO, suggesting that mutant and wild-type ACVR1 receptor complexes compete for activin ligands or type II BMP receptor binding partners. Finally, systemic inhibition of activin A completely blocks HO and restores wild-type-like behavior to transplanted Acvr1 ^R206H/+ FAPs. Understanding the cells that drive HO may facilitate the development of cell-specific therapeutic approaches to inhibit catastrophic bone formation in FOP.

Fig. 1

Targeted recombination of Acvr1^tnR206H in FAPs causes HO. a Organization of the conditional Acvr1^tnR206H knockin allele. b–e Validation of the fluorescence marking system using the MyoD^iCre driver. In Acvr1^tnR206H/+;R26^NG/+;MyoD^iCre/+ mice, developing skeletal muscle expressed the GFP lineage tracer (R26^NG), but not tdTomato. All other tissues express tdTomato but not GFP. c–e Cryosection through a developing hindlimb muscle bed. The approximate section plane is shown in b. f,g Representative µCT images of the distal hindlimb 14 days following pinch injury of the gastrocnemius muscle in wild-type (f; n = 25 mice) and Acvr1^tnR206H/+;R26^NG/+;Tie2-Cre (g; n = 42 mice) mice. h Representative µCT image of the distal hindlimb of a SCID host 21 days following intramuscular transplantation of Acvr1^R206H/+ FAPs (n = 7 mice). HO in g and h is pseudocolored green and heterotopic bone volume is given (mm³)

Fig. 2

Histological and lineage analysis of FAP-driven HO. a–i IHC (brown staining) for ACVR1 and SOX9 in paraffin sections of pinch-injured gastrocnemius muscle in wild-type (n = 5 mice per timepoint) and FOP (Acvr1^tnR206H/+;R26^NG/+;Tie2-Cre; n = 5 mice per timepoint) mice. a–c At Day 3 post-injury, areas of weak ACVR1 staining (arrows) are evident in both wild-type (a) and FOP (b) muscle. Occasional SOX9+ cells were observed in FOP (c, arrowheads) and wild-type (see Supplementary Fig. 3) muscle. A few nascent muscle fibers (identified by their central nucleation; arrowheads) are present in wild-type muscle (a). d–f At Day 6, regenerating muscle fibers in wild-type muscle are (d, examples at arrowheads) surrounded by ACVR1+ interstitium (arrows). In FOP muscle, early cartilage and associated mesenchyme (asterisks) is intensely stained for ACVR1 (e) and SOX9 (f). Uninjured muscle adjacent to lesions exhibit thickened bands of interstitial cells that stain strongly for ACVR1 and SOX9 (arrowheads in e, f). g–i At Day 14, regenerated muscle fibers in wild-type muscle are surrounded by an ACVR1+ endomysial layer (g, arrows). In FOP muscle, cartilage (C) continued to strongly express ACVR1 (h, asterisk) and SOX9 (i, arrowheads). ACVR1 staining in some heterotopic bone cells (arrowheads) and encapsulating fibroblastic cells (arrow) is evident in h. M, uninjured muscle. Sections were counterstained with hematoxylin. j–m Fluorescence images of a cryosection through a heterotopic lesion (asterisks) 12 days after cardiotoxin injection (n = 7 mice). j Unrecombined, tdTomato+ cells were absent from skeletal lesional tissue, whereas muscle fibers (M) exhibited intense red fluorescence. k At this stage, most lesional tissue stained positively for the bone marker, osterix (purple). l Virtually all lesional cells are GFP+. m Merge of panels (j–l). Scale bars in a–i and j–m = 100 µm

Fig. 3

FAPs drive spontaneous HO via endochondral ossification. a–c Spontaneous HO extended along the spine of a 22-week-old Acvr1^tnR206H/+;R26^NG/+;Tie2-Cre mouse. a µCT detects calcified heterotopic bone (bracket) but not the distal cartilaginous region of the lesion (asterisk). b Whole mount ABAR staining of the same mouse detects both lesional cartilage (blue; arrow) and bone (red; bracket). c Magnified image of the boxed region in b. d–g Spontaneous HO (bracket) along the spine of a 6-week-old Acvr1^tnR206H/+;R26^NG/+;Pdgfrα-Cre mouse. The site of the most posterior chondrogenic area is shown at the asterisk in d and e. e ABAR staining shows the presence of two chondrogenic areas at the periphery of the predominantly boney lesion. f, g Higher magnification images of areas denoted in e. Chondrogenic areas are shown at the arrow or bracket. h,i ABAR staining of the distal hindlimb of two 6-week-old Acvr1^tnR206H/+;R26^NG/+;Pdgfrα-Cre mice. Multiple chondrogenic areas (examples at arrows) are evident at the periphery of intramuscular boney lesions (h). In i, peripheral cartilage (arrow) is associated with HO of the Achilles tendon (arrowhead). B, heterotopic bone associated with the distal tibia/fibula. j–m Paraffin section of spontaneous HO from an adult Acvr1^tnR206H/+;R26^NG/+;Pdgfrα-Cre mouse stained by IHC for ACVR1 (brown) and counterstained with hematoxylin (n = 7 mice). j Low magnification image showing multiple areas of heterotopic cartilage and bone. Areas of mature (asterisk) and immature cartilage are evident. Staining for ACVR1 was most intense in immature cartilage, peripheral regions of mature cartilage, and in fibroproliferative and chondrogenic regions at the junction between HO and muscle (brackets). k–m Higher magnification images of the labeled boxes in j. k ACVR1 expression was not detected in apparently mature heterotopic bone, characterized by abundant marrow elements. l Growth plate-like cartilaginous regions strongly stain for ACVR1 at the chondro-osseous junction (bracket). m Peripheral fibroproliferative and chondrogenic regions (asterisk) strongly expressed ACVR1. Scale bars = 500 µm (j) and 100 µm (k–m)

Fig. 4

Activin A is an osteogenic ligand for Acvr1^R206H/+ FAPs. a Osteogenic differentiation of Acvr1^R206H/+ FAPs (R206H/+; n = 6 mice) and wild-type FAPs (n = 6 mice) cultured in base media (5% FBS/DMEM), with or without 25 ng/mL (~1 nM) activin A or BMP2, was assessed by ALP staining (purple). ActA-mAb was used at 1 μg/mL (7-fold molar excess over ligands). b Western blot of SMAD 1/5/8 phosphorylation (p-SMAD 1/5/8) in response to 50 ng/mL BMP2 or 50 ng/mL activin A. β-actin was used as a loading control (n = 3 mice per genotype). c, d µCT of the distal hindlimb of Acvr1^tnR206H/+;R26^NG/+;Tie2-Cre mice 14 days post-injection of 1% methylcellulose carrier alone (c; n = 8 mice) or 5 µg activin A in methylcellulose (d; n = 3 mice). e, f µCT of the distal hindlimb at day 21 post-transplantation of Acvr1^R206H/+ FAPs (R206H/+; n = 8 mice) into the gastrocnemius of SCID hosts without e or with f a single dose of ActA-mAb administered to the SCID host at the time of muscle injury (1 day prior to transplantation). HO in d and e is pseudocolored green and heterotopic bone volume is given (mm³)

Fig. 5

Activin blockade inhibits FAP-mediated spontaneous HO. a, b Representative µCT images of 6-week-old Acvr1^tnR206H/+;R26^NG/+;Pdgfrα-Cre mice treated with 10 mg/kg IgG2a isotype control antibody (a) or ActA-mAb (b) twice weekly from 14–42 days-of-age. Arrows denote sites of HO. c Survival curve of isotype control (n = 14) and ActA-mAb-treated (n = 9) mice. Percent survival reflects animal deaths as well as mice euthanized due to weight loss of >20%. d, e Representative µCT images of 112-day-old Acvr1^tnR206H/+;R26^NG/+;Pdgfrα-Cre mice either untreated (d; n = 1) and heavily burdened with HO of the musculature, tendons/ligaments and jaw (arrows), or 70 days post-cessation of ActA-mAb treatment (e; n = 6) and presenting no overt HO. f–i GFP-labeled Acvr1^R206H/+ FAPs were isolated and expanded from 42-day-old Acvr1^tnR206H/+;R26^NG/+;Pdgfrα-Cre mice (n = 3) that had been treated with ActA-mAb from 14–42 days-of-age, and transplanted into the distal hindlimb of SCID hosts that were either untreated (f, g) (n = 6) or administered ActA-mAb on the day of injury (h, i) (n = 6). f, h Whole mount GFP images showing FAP engraftment (examples at arrowheads). g, i µCT images at 21 days post-transplantation. HO was only observed in untreated hosts. TA tibialis anterior, GA gastrocnemius. HO in g is pseudocolored yellow. Heterotopic bone volume is given (mm³)

Fig. 6

HO is greatly exacerbated by loss of wild-type Acvr1. a, b Representative µCT images of HO (pseudocolored green) 14 days after pinch injury of the gastrocnemius muscle of Acvr1^tnR206H/+;R26^NG/+;Tie2-Cre (a; n = 7) or Acvr1^tnR206H/flox;R26^NG/+;Tie2-Cre (b; n = 8) mice. c, d ActA-mAb administered at 10 mg/kg in both mouse genetic models, either twice weekly beginning at the time of injury (Day 0–14; tnR206H/+, n = 12; tnR206H/flox, n = 7), as a single dose at the time of injury (Day 0; tnR206H/+, n = 8; tnR206H/flox, n = 4), or as a single dose 3 days post-injury (Day 3; tnR206H/+, n = 12; tnR206H/flox, n = 4), was highly effective at inhibiting HO. Residual HO in these treated mice is shown at arrows. HO volume (mm³) in a–d is given. e Quantification of HO volume at day 14 post-injury from untreated and ActA-mAb-treated mice. Bars represent ± SEM. ***p < 0.001, ****p < 0.0001