Pax7 remodels the chromatin landscape in skeletal muscle stem cells

Abstract

Pluripotent stem cells (PSC) hold great promise for the treatment of human skeletal muscle diseases. However, it remains challenging to convert PSC to skeletal muscle cells, and the mechanisms by which the master regulatory transcription factor, Pax7, promotes muscle stem (satellite) cell identity are not yet understood. We have taken advantage of PSC-derived skeletal muscle precursor cells (iPax7), wherein the induced expression of Pax7 robustly initiates the muscle program and enables the in vitro generation of precursors that seed the satellite cell compartment upon transplantation. Remarkably, we found that chromatin accessibility in myogenic precursors pre-figures subsequent activation of myogenic differentiation genes. We also found that Pax7 binding is generally restricted to euchromatic regions and excluded from H3K27 tri-methylated regions in muscle cells, suggesting that recruitment of this factor is circumscribed by chromatin state. Further, we show that Pax7 binding induces dramatic, localized remodeling of chromatin characterized by the acquisition of histone marks associated with enhancer activity and induction of chromatin accessibility in both muscle precursors and lineage-committed myoblasts. Conversely, removal of Pax7 leads to rapid reversal of these features on a subset of enhancers. Interestingly, another cluster of Pax7 binding sites is associated with a durably accessible and remodeled chromatin state after removal of Pax7, and persistent enhancer accessibility is associated with subsequent, proximal binding by the muscle regulatory factors, MyoD1 and myogenin. Our studies provide new insights into the epigenetic landscape of skeletal muscle stem cells and precursors and the role of Pax7 in satellite cell specification.

Fig 1. Identification and characterization of Pax7 bound regions in iPax7 cells.

(A) Schematic showing iPax7 cell isolation after inducing Pax7 expression with doxycycline (Dox) during embryoid body (EB) formation and sorting the PDGFαR+Flk-1- fraction (Darabi et al., 2011). (B) Pax7 ChIP-seq was performed using iPax7 cells grown in the presence of Dox or after removal of Dox for 3d. Regions that were more highly enriched in the presence of Dox were identified as Pax7 binding sites. (C) The majority of 2455 Pax7 binding sites are > 5 kb from the nearest transcription start site (TSS), whereas 300 binding sites were found at promoter regions within 1 kb of the nearest TSS. (D) Representative read density profiles for Pax7, H3K4me1, and H3K27Ac enrichment, and input in +Dox (blue) and -Dox (grey) conditions. Enlarged region shows enhancer-associated histone modifications (H3K4me1 and H3K27Ac) are enriched at distal Pax7 binding sites in +Dox versus -Dox conditions. The y-axis corresponds to normalized read densities. (E) Pie chart depicting percentages of distal (>1 kb from TSS) Pax7 enriched regions that overlap with H3K4me1, H3K27Ac, both marks, or neither histone modification.

Fig 2. Pax7 binding demarcates and maintains potentially active enhancers.

(A) Heatmap generated using k-means clustering analysis indicates that Pax7 binding sites are preferentially and globally enriched for histone modifications associated with enhancer marks (H3K27Ac and H3K4me1) in iPax7 cells treated with Dox. Enhancer marks were reduced upon removal of Pax7, indicating that maintenance of a significant number of enhancers could depend on Pax7 binding. Only a fraction of Pax7 binding regions enriched for the enhancer signature in Dox-treated cells were also enriched in C2C12 myoblasts and myotubes. ChIP-seq data were normalized for each antibody and plotted for regions 1.5 kb upstream and downstream of the center of Pax7 binding sites. MB and MT, myoblasts and myotubes, respectively. (B) Metagene analysis of average signal for ChIP-seq data displayed in panel A, plotted for regions 1.5 kb upstream and downstream of the center of Pax7 binding sites. Levels of indicated enhancer marks were either reduced or unchanged after removal of Pax7.

Fig 3. Pax7 is essential to maintain open chromatin and enhancer signatures at a subset of target genes.

(A) Heatmap of ATAC-seq and ChIP-seq data highlight two distinct types of Pax7 binding sites, as indicated. Type 2 sites are accessible in satellite cells, iPax7 cells, and throughout myogenesis, and these sites are proximal to regions bound by MyoD in myoblasts (MB). Type 1 sites exhibit accessibility and enhancer signatures that strictly depend on Pax7 binding. (B) Pax7 and MyoD binding sites are proximal to one another. The center-to-center distances between MyoD- and Pax7 ChIP-seq peaks were analyzed at promoters and enhancers. (C) Illustrative read density profiles for ATAC-seq, ChIP-seq, and RNA-seq data corresponding to a Type 2 Pax7 binding site. Highlighted region indicates the proximity of Pax7 and MyoD binding sites at a potential enhancer that is accessible in satellite cells, iPax7 cells, and throughout myogenesis. Promoter regions bound initially by Pax7 are accessible throughout myogenesis in the absence of Pax7, as indicated by times after Dox withdrawal. Pax7-dependent Type 1 enhancers (encompassing right-most Pax7 binding site) are accessible only in the presence of Pax7, and these regions lose both enhancer marks (H3K27Ac and H3K4me1) upon removal of Pax7. MB and MT, myoblasts and myotubes, respectively. MB cntrl and MB Pax7 indicate C2C12 cell lines without or with ectopic expression of Flag-Pax7, respectively.

Fig 4. Pax7 functions to remodel chromatin.

(A) Pax7 bound regions are accessible in the presence of Dox, and there is a significant and rapid loss of accessibility near Pax7 binding sites upon removal of Dox at 12h, 24h, and 3d (*, p<0.001). A significant number of sites bound by Pax7 in iPax7 cells also become accessible after ectopic expression of Pax7 in C2C12 cells. (B) Type 2, but not Type 1, sites (both denoted within dashed rectangles) remain accessible throughout myogenesis in satellite cells, iPax7 cells, myoblasts ectopically expressing Pax7, and myotubes and do not exhibit dramatic alterations in levels of H3K27Ac and H3K4me1 at these Pax7-bound regions. Type 1 sites further display de novo formation of open chromatin and enhancer marks in C2C12 myoblasts expressing Pax7. (C) Pie chart showing that the majority of Type 1 enhancers are lost upon removal of doxycyline whereas in (D) Type 2 enhancers do not display similar dramatic loss of enhancer modifications. MB and MT, myoblasts and myotubes, respectively. (E) Representative position-weight matrices (PWMs; adjusted p-value <0.05) found in CentriMo using MEME for a 250 bp window on both sides of the Pax7 peak center for Type 1 and Type 2 sites.

Fig 5. Chromatin accessibility in progenitors precedes differentiation-dependent gene expression.

(A) The majority of Pax7 bound regions accessible at 3d post-Dox removal are already accessible in satellite cells, Dox-treated iPax7 cells, and earlier time points (Dox removal for 12h and 24h). (A’) Few sites (5.5%) exhibit de novo accessibility 3d after Dox removal. (B) Genes important for myogenesis and induced upon loss of Pax7 remain accessible throughout myogenesis, despite their relatively low expression. Normalized read densities for ATAC-seq and RNA-seq in iPax7 cells and satellite cells are indicated on the y-axis. Scatter plots of RNA-seq normalized read counts for replicates from satellite cells (n = 3), iPax7 +Dox (n = 3), and iPax7 -Dox after 3d (n = 2). (C) Accessibility at Pax7 sites throughout differentiation allows for gene expression plasticity. Genes that are up-regulated and down-regulated after loss of Pax7 retain accessibility at promoters throughout myogenesis. Read density for ATAC-seq and RNA-seq in iPax7 cells and satellite cells. Scatter plots showing normalized RNA-seq read counts for replicates from satellite cells, Dox-treated iPax7 cells, and iPax7 cells 3d after Dox removal. MB and MT, myoblasts and myotubes, respectively.

Fig 6. Induced accessibility of Pax7 binding sites associated with the formation of novel Pax7-specific enhancers.

(A) mESC do not exhibit features of enhancers seen in myogenic precursors, and sites bound by MyoD in myoblasts are Pax7-specific as they lack chromatin accessibility in mESC. Histone modifications (H3K27Ac, H3K4me1, H3K4me3) and chromatin accessibility in myotubes (MT) differentiated from either iPax7 cells or C2C12 cells share similar patterns. MB from C2C12 with and without Flag control (Cntrl) also exhibit similar histone modification patterns. (B) Venn diagram depicting the number of shared and unique peaks with the indicated histone modifications at Pax7 binding sites (as defined in Dox-treated iPax7 cells) in control (MB Cntrl) and Pax7 expressing (MB Pax7) C2C12 cells. (C) Accessibility at Pax7 sites changes upon Pax7 expression in C2C12. Bar graph showing the quantification of accessible Pax7 binding sites with associated histone modifications at these sites in C2C12 control versus cells ectopically expressing Pax7. (D) Schematic showing the percentage of sites marked with individual histone modifications (as indicated) at known Pax7 binding sites in control C2C12 cells that undergo a novel conversion to an enhancer signature after Pax7 expression in C2C12 myoblasts.