Matrix metalloproteinase-9 deficiency confers resilience in fibrodysplasia ossificans progressiva in a man and mice

Abstract

Single case studies of extraordinary disease resilience may provide therapeutic insight into conditions for which no definitive treatments exist. An otherwise healthy 35-year-old man (patient-R) with the canonical pathogenic ACVR1R206H variant and the classic congenital great toe malformation of fibrodysplasia ossificans progressiva (FOP) had extreme paucity of post-natal heterotopic ossification (HO) and nearly normal mobility. We hypothesized that patient-R lacked a sufficient post-natal inflammatory trigger for HO. A plasma biomarker survey revealed a reduction in total matrix metalloproteinase-9 (MMP-9) compared to healthy controls and individuals with quiescent FOP. Whole exome sequencing identified compound heterozygous variants in MMP-9 (c.59C > T, p.A20V and c.493G > A, p.D165N). Structural analysis of the D165N variant predicted both decreased MMP-9 secretion and activity that were confirmed by enzyme-linked immunosorbent assay and gelatin zymography. Further, human proinflammatory M1-like macrophages expressing either MMP-9 variant produced significantly less Activin A, an obligate ligand for HO in FOP, compared to wildtype controls. Importantly, MMP-9 inhibition by genetic, biologic, or pharmacologic means in multiple FOP mouse models abrogated trauma-induced HO, sequestered Activin A in the extracellular matrix (ECM), and induced regeneration of injured skeletal muscle. Our data suggest that MMP-9 is a druggable node linking inflammation to HO, orchestrates an existential role in the pathogenesis of FOP, and illustrates that a single patient's clinical phenotype can reveal critical molecular mechanisms of disease that unveil novel treatment strategies.

Keywords: ACVR1; Activin A; MMP-9; disease resilience; fibrodysplasia ossificans progressiva (FOP); heterotopic ossification.

Figure 1

FOP in patient-R. (A) Photograph of malformations of the great toes (arrowheads). (B) Radiographs of feet. Malformed great toes (arrows) with absence of the interphalangeal joints (arrowheads). Circled areas show additional digital malformations. (C) Electropherogram showing the classic ACVR1 mutation (617G > A). (D) Scout image of a whole-body low dose CT scan showing a paucity of HO. Arrowhead indicates a small area of HO associated with the lumbar spine.

Figure 2

MMP-9 in patient-R. (A) Box and whisker plot of total plasma MMP-9 in FOP patients (n = 38) and in patient-R (n = 2, taken 2 years apart) by Myriad RBM Multiplex Luminex analysis. (B) Box and whisker plot of total plasma MMP-9 (Myriad RBM Multiplex Luminex analysis) in patient R (n = 2) relative to those from individuals with quiescent FOP flare-up activity (n = 11). (C) Gelatin zymography assay reveals less active MMP-9 in cell culture supernatant of PBMCs (cultured for 72 h with TNF-α) of patient-R versus controls and FOP patients (n = 3). (D) Quantification of gelatin zymography in (C). (E) Schematic of MMP-9 protein denoting the locations of the A20V and D165N variants in patient-R resulting in amino acid residue changes. (F) PCR amplicons confirm the presence of A20V and D165N variants in patient-R. These variants were not seen in 31 randomly selected patients with classic FOP. A: Mann–Whitney test. B: Unpaired t-test. D: One-way ANOVA with Tukey’s multiple comparisons test. ^**P < 0.01, ^*P < 0.05.

Figure 3

MMP-9 is expressed in early inflammatory lesional tissues of Acvr1^{[R206H]FlEx/+};CreERT2^-/+ mice at 1 day (A-B), 3 days (C-D) but not 5 days (E-F) after cardiotoxin injury. Sections A, C, and E were stained with rabbit antibodies against MMP-9 followed by anti-rabbit antibodies conjugated with Alexa 594 (red). Nuclei (blue) were stained with DAPI, and immunostaining and DAPI images were merged. B, D, and F are representative hematoxylin and eosin stained sections at days 1, 3, and 5, respectively. Scale bar = 50 μm. (G) Quantification of MMP-9 protein in lysates from lesional tissue demonstrates that levels peak on day 1 and plateau by day 5 after injury. These complementary measurements confirm that MMP-9 is produced and secreted at early time points after injury that correspond to the inflammatory phase of lesion formation. One-way ANOVA with Tukey’s multiple comparisons test; ^****P < 0.0001, ^***P < 0.001, ^**P < 0.01.

Figure 4

Down-regulation of MMP-9 in heterozygotes and homozygotes decreases cardiotoxin-induced HO in Acvr1^{[R206H]FlEx/+};CreERT2^-/+ mice. (A) Micro CT images of MMP-9^+/+, MMP-9^-/+ and MMP-9^-/- on genetic background in Acvr1^{[R206H]FlEx/+};CreERT2^-/+ mice at 14 days after cardiotoxin injury. (B) Heterotopic bone volume (arrows in A) was quantified (n = 6). (C) MMP-9 was measured by ELISA in plasma of MMP-9^+/+, MMP-9^-/+, and MMP-9^-/- on genetic background in Acvr1^{[R206H]FlEx/+};CreERT2^-/+ mice. One-way Anova with Tukey’s multiple comparisons test; ^****P < 0.0001, ^***P < 0.001. (D) MMP-9 inhibition leads to regeneration of injured skeletal muscle in FOP mice rather than formation of heterotopic endochondral bone. Tissue sections from Acvr1^+/+;MMP-9^+/+, Acvr1^+/+;MMP-9^-/-, Acvr1^{[R206H]FlEx/+};CreERT2^-/+;MMP-9^+/+, and Acvr1^{[R206H]FlEx/+};CreERT2^-/+;MMP-9^-/- mice after injury with cardiotoxin at 14 days were stained with hematoxylin and eosin. Arrows indicate regeneration in skeletal muscle (central nuclei). Scale bar = 50 μm.

Figure 5

Treatment with minocycline or with a MAb against MMP-9 decreases HO in FOP mouse models at 14 days after injury with cardiotoxin. (A) Micro CT images of Acvr1^{[R206H]FlEx/+};CreERT2^-/+ mice not treated (left panel) and treated with minocycline (right panel). Heterotopic bone (arrows) was quantified (n = 6). Mann–Whitney test; ^**P < 0.01. (B) Micro CT images of Acvr1^Q207D/+ mice not treated (left panel) and treated with minocycline (right panel). Heterotopic bone (arrows) was quantified (n = 6). Unpaired t-test; ^***P < 0.001. (C) Micro CT images of Acvr1^ARC-R206H/+;CreERT2^-/+ mice not treated (n = 18) (left panel) and treated with a MAb against MMP-9 (n = 16) (right panel). Heterotopic bone (arrows) was quantified. (D) Micro CT images of Acvr1^Q207D/+ mice not treated (left panel) and treated with MAbs against MMP-9 (right panel). Heterotopic bone (arrows) was quantified (n = 8). Unpaired t-test; ^****P < 0.0001, ^**P < 0.01.

Figure 6

MMP-9 and Activin A in human macrophages and mouse FOP lesional tissue. (A-D) Analysis of secreted MMP-9 and Activin A from THP-1 derived M1-like and M2-like macrophages (wild type, A20V, and D165N). Cell culture supernatants were from M1-like and M2-like macrophages at 48 hours after polarization of wild type, A20V, and D165N M0 macrophages. Activity and amount of MMP-9 (B and D, respectively) were measured from the same cultures. Data for B, C, and D were shown as mean ± SEM; two-way ANOVA; ^****P < 0.0001, ^**P < 0.01, ^*P < 0.05. (E and F) Inhibition of MMP-9 promotes sequestration of Activin A in lesional tissue of FOP mice. (A) Gelatin zymography of active MMP-9. (B) Quantification of gelatin zymography in (A). (C) ELISA for Activin A in cell culture supernatants from M1-like and M2-like macrophages. The level of Activin A in the supernatant is dramatically reduced by MMP-9^A20V and MMP-9^D165N compared to MMP-9^WT. Activin A in cell culture supernatants was normalized to protein in cell lysates. (D) ELISA for MMP-9 in cell culture supernatants as in (C). MMP-9 in cell culture supernatants was normalized to protein in cell lysates. (E) Tissue levels of Activin A and MMP-9 at days 1, 3, 5, 7, 11, and 14 after cardiotoxin muscle injury, in both FOP mice with and without genetic depletion of MMP-9. In FOP mice with intact MMP-9 (MMP-9^+/+), tissue levels of MMP-9 are elevated on day 1 after muscle injury and subsequently decline until about day 5, after which levels plateau through day 14. In the same animal model, tissue sequestration of Activin A becomes elevated by day 7 after injury and then declines slightly through day 14. In contrast, FOP mice with undetectable tissue levels of MMP-9 (MMP-9^-/-), in which HO does not form (Figure 6A and B), demonstrate about 2-fold higher levels of Activin A sequestration by day 11 after injury, and higher levels of Activin A through day 14 compared to FOP mice with intact MMP-9. (F) Comparison of tissue sequestration of Activin A in FOP mice with and without MMP-9 expression, where Activin A expression is normalized to MMP-9 level, the ratio of Activin A/MMP-9 confirms that tissue sequestration of Activin A is substantially higher in the latter across all time points.

Figure 7

Hypothetical schema of MMP-9 activity in HO in FOP. Chevrons = major categories; arrows = activation; arrowhead = activation complex; blunt-end red line = inhibition; MMP-9 (green) = target; FAPs = fibro-adipogenic progenitor cells; HSPG = heparan sulfate proteoglycan; HO^FOP = heterotopic ossification in FOP. Original photograph of FOP skeleton courtesy of F. Kaplan from Shafritz et al., N Engl J Med, Vol. 335, 1996.