A negative feedback loop between fibroadipogenic progenitors and muscle fibres involving endothelin promotes human muscle fibrosis

Abstract

Background: Fibrosis is defined as an excessive accumulation of extracellular matrix (ECM) components. Many organs are subjected to fibrosis including the lung, liver, heart, skin, kidney, and muscle. Muscle fibrosis occurs in response to trauma, aging, or dystrophies and impairs muscle function. Fibrosis represents a hurdle for the treatment of human muscular dystrophies. While data on the mechanisms of fibrosis have mostly been investigated in mice, dystrophic mouse models often do not recapitulate fibrosis as observed in human patients. Consequently, the cellular and molecular mechanisms that lead to fibrosis in human muscle still need to be identified.

Methods: Combining mass cytometry, transcriptome profiling, in vitro co-culture experiments, and in vivo transplantation in immunodeficient mice, we investigated the role and nature of nonmyogenic cells (fibroadipogenic progenitors, FAPs) from human fibrotic muscles of healthy individuals (FibM^CT ) and individuals with oculopharyngeal muscular dystrophy (OPMD; FibM^OP ), as compared with nonmyogenic cells from human nonfibrotic muscle (M^CT ).

Results: We found that the proliferation rate of FAPs from fibrotic muscle is 3-4 times higher than those of FAPs from nonfibrotic muscle (population doubling per day: M^CT 0.2 ± 0.1, FibM^CT 0.7 ± 0.1, and FibM^OP 0.8 ± 0.3). When cocultured with muscle cells, FAPs from fibrotic muscle impair the fusion index unlike M^CT FAPs (myoblasts alone 57.3 ± 11.1%, coculture with M^CT 43.1 ± 8.9%, with FibM^CT 31.7 ± 8.2%, and with FibM^OP 36.06 ± 10.29%). We also observed an increased proliferation of FAPs from fibrotic muscles in these co-cultures in differentiation conditions (FibM^CT +17.4%, P < 0.01 and FibM^OP +15.1%, P < 0.01). This effect is likely linked to the increased activation of the canonical TGFβ-SMAD pathway in FAPs from fibrotic muscles evidenced by pSMAD3 immunostaining (P < 0.05). In addition to the profibrogenic TGFβ pathway, we identified endothelin as a new actor implicated in the altered cross-talk between muscle cells and fibrotic FAPs, confirmed by an improvement of the fusion index in the presence of bosentan, an endothelin receptor antagonist (from 33.8 ± 10.9% to 52.9 ± 10.1%, P < 0.05).

Conclusions: Our data demonstrate the key role of FAPs and their cross-talk with muscle cells through a paracrine signalling pathway in fibrosis of human skeletal muscle and identify endothelin as a new druggable target to counteract human muscle fibrosis.

Keywords: ECM; Endothelin; FAPs; Fibrosis; Human; Pharyngeal muscle; Regeneration; Skeletal muscle; TGFβ.

Figure 1

Isolation of nonmyogenic cells from nonfibrotic muscles (M^CT) and muscles with physiological (FibM^CT) and pathological (FibM^OP) fibrosis. (A) Cross sections of human muscle biopsies stained with Sirius red to visualize and quantify fibrosis or haematoxylin and eosin (H&E). M^CT, control muscle; FibM^CT, fibrotic control muscle; FibM^OP, fibrotic OPMD muscle. (B) Quantification of fibrosis by Sirius red staining of M^CT, FibM^CT and FibM^OP human biopsies (n = 6–9 biopsies per condition). (C) Quantification of the number of interstitial cells (n = 6–7 muscle biopsies per condition). (D) RT‐qPCR quantification of TGFβ gene expression normalized to RPLP0 expression in M^CT, FibM^CT, and FibM^OP human biopsies (n = 5–6 biopsies per condition). (E) Experimental scheme used for MACS isolation of nonmyogenic cells (CD56− cells) and myogenic cells (CD56+ cells) from human muscle biopsies. The number of myogenic cells in each population was assessed by desmin staining (green). Data are presented as means ± SD, with P‐values obtained by ordinary one‐way ANOVA test followed by Tukey's multiple comparisons test (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001).

Figure 2

Human FAPs isolated from fibrotic muscles differ from those isolated from nonfibrotic muscles. (A) Uniform manifold approximation and projection (UMAP) map showing the distributions of cells from M^CT (yellow), FibM^CT (orange), and FibM^OP (red) muscle biopsies. For each condition, nonmyogenic cells were extracted, and data from three to four patients were concatenated. Each dot represents a single cell, and 300 000 cells were used to obtain the map. The 37 markers used for the analysis are listed in Table 2. (B) UMAP plots showing the expression patterns of PDGFRα, CD90, CD105, and CD73 in nonmyogenic cells from M^CT, FibM^CT, and FibM^OP muscle biopsies. Cells are coloured according to the intensity of the marker shown. (C) Expression dot plot of selected markers from the CyTOF analysis of nonmyogenic cells from fibrotic and nonfibrotic muscles. Dots are coloured according to the average intensity with which the marker was expressed, and the size of each dot shows the percentage of nonmyogenic cells expressing each marker in each condition: M^CT, FibM^CT, and FibM^OP. (D) Adipogenic and osteogenic differentiation of human FAPs isolated from fibrotic or nonfibrotic muscles. Left panel: Oil red O staining of FAPs from M^CT, FibM^CT, and FibM^OP muscle biopsies in adipogenic differentiation medium. Right panel: Alizarin red staining of FAP cells from M^CT, FibM^CT, and FibM^OP muscle biopsies in osteogenic differentiation medium.

Figure 3

FAPs from fibrotic muscles have a higher proliferative capacity and a negative effect on fusion compared to those from nonfibrotic muscle. (A) Histogram showing the proliferation rate as the mean population doubling (PDL)/day for nonmyogenic cells isolated from M^CT, FibM^CT, and FibM^OP muscle biopsies over time (n = 6–8 biological replicates). (B) Violin plot representing the areas of nonmyogenic cells from M^CT, FibM^CT and FibM^OP (n = 26–112 cells per condition). (C) Representative trajectory of cells over 24 h, (D) distance covered over 12 h (n = 25–39 cells per biological replicate; n = 3 per condition), and (E) speed of motion over 12 h evaluated by manual tracking using the SkyPad algorithm on live‐imaging videos of nonmyogenic cells from M^CT, FibM^CT, and FibM^OP (n = 25–39 cells per biological replicate; n = 3 per condition). (F) Percentage of basal pSMAD3+ cells assessed by immunostaining of serum‐starved FAPs from M^CT, FibM^CT, and FibM^OP muscle biopsies (n = 3 biological replicates). (G) Experimental scheme used to inject FAPs isolated from M^CT, FibM^CT, and FibM^OP muscle biopsies into the regenerating TA muscle of immunodeficient mice. A total of 1.4 × 10⁵ cells were injected at D0 after cryodamage and at D4 and D8 (left). Immunofluorescence analysis of cryosections was carried out using a human‐specific Lamin A/C antibody (hlaminA/C, green), a human‐specific fibronectin (hFN,red) antibody and a pan‐laminin antibody (blue) (middle). Quantification of ECM secretion as a ratio of hFN signal divided by the number of human cells (n = 4 per condition) (right). (H) Experimental scheme of the coculture experiments. FAPs from nonfibrotic and fibrotic muscles were cocultured with myoblasts at a 30%/70% ratio for 5 days in differentiation medium. Proliferation and the fusion index were assessed to evaluate the cross‐talk between nonmyogenic and myogenic cells. (I) Representative coimmunostaining of desmin (green) and EdU (red) after 5 days of coculture of FAPs and myogenic cells. Nuclei were counterstained with Hoechst (blue). (J) Quantification of EdU incorporation in FAPs from M^CT, FibM^CT, and FibM^OP alone or in coculture with myotubes (hatched bars) after 5 days in differentiation medium (n = 3–4 biological replicates; ns = nonsignificant). (K) the fusion index after 5 days of differentiation was assessed by Desmin staining in myogenic cells alone and in coculture with 30% M^CT, FibM^CT, or FibM^OP FAPs (n = 3–4 biological replicates). Data are presented as means ± SD, with P‐values obtained by ordinary or RM one‐way ANOVA test followed by Tukeys multiple comparisons test (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001).

Figure 4

Targeting ET‐1‐mediated secretion of fibrotic FAPs improves myoblast fusion. (A) Principal component analysis (PCA) plot prepared following transcriptome analysis of FAPs from M^CT (yellow), FibM^CT (orange), and FibM^OP (red) muscle biopsies. Each dot represents cells from one patient. (B) Quantification of ET‐1 protein levels in unconcentrated conditioned medium from differentiated human myoblasts after 24, 48, and 72 h of differentiation, reanalyzed from a previous study. (C) RT‐qPCR quantification of EDNRb gene expression normalized to RPLP0 expression in FAPs from M^CT, FibM^CT, and FibM^OP muscle biopsies (n = 5 biological replicates). (D) RT‐qPCR quantification of Col7a1 gene expression normalized to RPLP0 expression in FAPs from M^CT, FibM^CT, and FibM^OP muscle biopsies (n = 3 biological replicates). (E) FAPs from M^CT, FibM^CT, and FibM^OP muscle biopsies were cultured in the presence of ET‐1 (40 nM) or ET‐1 and bosentan (10 μM) for 3 days in proliferation medium containing 1% FBS. Left: Immunofluorescence analysis of Phalloidin (red), Hoechst (blue) and collagen 7a1 (green) was performed. Right: Quantification of the percentage of COL7A1‐positive cells (n = 3–4 biological replicates). (F) Quantification of the percentage of EdU incorporation after treatment of FAPs from M^CT, FibM^CT, and FibM^OP muscle biopsies with ET‐1 (40 nM) or ET‐1 and bosentan (10 μM) for 3 days in proliferation medium containing 1% FBS (n = 3–4 biological replicates). (G) Myoblasts were cocultured alone or with FAPs from fibrotic muscles at a 70/30% ratio for 5 days in differentiation medium. Bosentan (10 μM) was added on days 0 and 3 of differentiation. Left: The fusion index was assessed by Desmin staining (n = 3 biological replicates). Right: Desmin immunostaining and Hoechst staining (scale bar = 100 μm). Data are presented as means ± SD, with P‐values obtained by ordinary or RM one‐way ANOVA test followed by Tukey's multiple comparisons test (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001).