Dynamics of cellular states of fibro-adipogenic progenitors during myogenesis and muscular dystrophy

Abstract

Fibro-adipogenic progenitors (FAPs) are currently defined by their anatomical position, expression of non-specific membrane-associated proteins, and ability to adopt multiple lineages in vitro. Gene expression analysis at single-cell level reveals that FAPs undergo dynamic transitions through a spectrum of cell states that can be identified by differential expression levels of Tie2 and Vcam1. Different patterns of Vcam1-negative Tie2^high or Tie2^low and Tie2^low/Vcam1-expressing FAPs are detected during neonatal myogenesis, response to acute injury and Duchenne Muscular Dystrophy (DMD). RNA sequencing analysis identified cell state-specific transcriptional profiles that predict functional interactions with satellite and inflammatory cells. In particular, Vcam1-expressing FAPs, which exhibit a pro-fibrotic expression profile, are transiently activated by acute injury in concomitance with the inflammatory response. Aberrant persistence of Vcam1-expressing FAPs is detected in DMD muscles or upon macrophage depletion, and is associated with muscle fibrosis, thereby revealing how disruption of inflammation-regulated FAPs dynamics leads to a pathogenic outcome.

Fig. 1

Heterogeneous FAPs population consists of distinct subpopulations of cells. a Experimental workflow for single cell gene expression analysis. Hindlimb muscles of C57Bl/10 mice were isolated, minced, and enzymatically digested. FAPs were isolated by FACS and loaded on the C1 System (Fluidigm) to extract RNA, reverse transcribe RNA to cDNA and pre-amplify cDNA from each single cell. Real-time qPCR analysis of single cell-derived cDNA was performed on the Biomark platform (Fluidigm) for 87 genes. b Principal component analysis (PCA) of single cell gene expression values of FAPs isolated from WT, WT notexin-injured day 3 (WT-inj d3) and dystrophic MDX mice. c Self organizing maps (SOM) representation of gene expression in clusters of single FAP cells. Each circle is a cluster of single cells and the fill color represents the level of expression for each gene shown. The expression scale is shown on the left for each gene individually. Expression is measured as Log₂Ex (Log₂Ex = Ct_(LOD)-Ct_(gene)) with LOD = 24 (limit of detection) and Ct = cycle threshold. d Correlation matrix of single cell gene expression across all cells. Orange color marks high positive correlation, green color marks high negative correlation. Groups of genes outlined in blue are positively correlated, while the genes outlined in green are negatively correlated. e Expression scatterplot of Tek and Vcam1 gene expression. Cutoff is set at 7 Log₂Ex for both genes based on the SOM graph (c). Tek and Vcam1 expression levels define the predicted subpopulations, marked as Tek(Tie2)+/Vcam− (Tek+) in dark blue, Vcam1+/Tek(Tie2)− (Vcam1+) in brown, double positive (DP) in light blue and double negative (DN) in gold. f The same PCA as in Fig. 1a but with cells color coded based on the FAPs subpopulations predicted in Fig. 1e. g Distribution of subpopulations in each experimental condition (n = 2)

Fig. 2

Vcam1+ and Tie2-expressing cells are dynamic subpopulations of FAPs. a Representative FACS plots of FAPs isolated from hindlimb muscles and analyzed based on Tie2 and Vcam1 expression in wild type (WT), WT notexin-injured day 3 (WT-inj d3) and dystrophic mdx C57Bl/10 mice (MDX). b Distribution of subpopulations in each experimental condition by FACS analysis (mean + s.e.m., n = 4, *p-value P < 0.05; **P < 0.01, ***P < 0.001, ****P < 0.0001). Statistical significance was determined by one-way ANOVA with Bonferroni post hoc test, and comparisons to the WT control group are reported. c Number of cells in each FAPs subpopulation for each experimental condition by FACS analysis (n = 4, mean + s.e.m., one-way ANOVA, *P < 0.05, **P < 0.01, ***P < 0.001). d Representative FACS plots of FAPs based on the expression of cell surface markers Tie2 and Vcam1 following notexin (NTX) injury in WT mice at indicated time points. FAPs were isolated from tibialis anterior (TA) muscles of C57Bl/6J mice. e Quantification of cell numbers for the subpopulations of FAPs in TA muscles during the time course in d (n = 3 representing independent experiments, mean + s.e.m.)

Fig. 3

SubFAPs have unique transcriptional profiles that are modulated by injury. a Principal component analysis (PCA) of gene expression data from RNA-seq analysis of FAPs subpopulations isolated from wild type (WT), WT NTX-injured at day 1 and day 3 (WT-inj d1 and d3) and dystrophic mice C57Bl/10 (MDX). b Expression heatmap of genes differentially expressed in FAPs subpopulations (Tie2^high, Tie2^low and Vcam1+) compared to bulk FAPs (adjusted p-value <0.001). Gene expression is represented as z-score calculated across the rows. c, d Biological functions predicted to be differentially activated or inhibited in each FAP subpopulation compared to bulk FAPs by IPA comparison analysis. Selected altered functions are shown in d. e, f Biological functions predicted to be differentially activated or inhibited in FAP subpopulations in each treatment condition (WT-inj d1, WT-inj d3 and MDX) compared to bulk WT FAPs by IPA comparison analysis. Selected altered functions are shown in f. g Selected altered canonical pathways in FAP subpopulations in each treatment condition (WT-inj d1, WT-inj d3 and MDX) compared to bulk WT FAPs by IPA comparison analysis. d, f and g Gray bars represent z-score (predicted level of activation/repression of the gene ontology category) with scale on the left. Red lines represent the significance of the prediction in log10(p-value), with scale on the right

Fig. 4

SubFAPs are distinctly associated with neonatal and injury-induced myogenesis. a Representative FACS plots of FAPs analyzed based on Tie2 and Vcam1 expression in adult wild type (WT), adult WT notexin-injured day 3 (WT-inj d3) and neonatal WT C57Bl/6J mice (postnatal day 8–10). FAPs were isolated from hindlimb muscles. b Distribution of subpopulations in each experimental condition by FACS analysis (mean + s.e.m., n = 6 (three independent experiments, each including biological duplicates), *P < 0.05, ****P < 0.0001). Statistical significance was determined by one-way ANOVA with Bonferroni post hoc test, and comparisons to the WT control group are reported. c Experimental design of the macrophage depletion study. ITGAM-DTR mice were used for this study. Acute muscle injury (Inj) was induced with either NTX (10 µl of 10 µg/ml notexin) or CTX (10 μl of 10 μM cardiotoxin). Macrophage depletion was achieved with diphtheria toxin (DT) injection (12 ng/g) at the time points indicated. Muscles were collected at day 0, 3 or 7 after injury. d, e Hematoxylin/Eosin (d) and Sirius Red (e-left panel) stainings of gastrocnemius muscle sections 7 days after an acute injury (ITGAM-inj d7) and 7 days after an acute injury in the context of macrophage depletion (ITGAM-inj d7 + 2xDT). Quantification of fibrosis in muscle sections stained with Sirius Red (e-right panel; n = 4, mean + s.d., t-test, **P < 0.01). Scale bar 200 µm. f Representative FACS plots profiles of Vcam1+ cells within FAPs population. g Percentage of Vcam1+ FAPs in the conditions analyzed in f. Statistical significance was determined by one-way ANOVA with Bonferroni post hoc test, and only comparisons to the control ITGAM (*) and to ITGAM-Inj. 7d (#) groups are reported (mean + s.d., n = 3 independent experiments, ANOVA, ***P < 0.001, ##P < 0.01)