Fig 1.
Skeletal muscles exhibit varied cis-regulatory repertoires.
(A) qPCR expression analysis of Myh isoforms measured in different muscle groups (EDL, Quad, Dia, Sol) (n = 5/group). Data are represented as means ± SEM. ***p < 0.001. (B) Number of H3K4me2 and H3K27ac peaks identified by ChIP-seq. (C) Heatmap distribution of H3K4me2 and H3K27ac tags within 6 kb of all H3K4me2 and/or H3K27ac peaks identified in EDL, Quad, Sol, and Dia; n = 2/group. (D) Similarity matrix for H3K4me2 (top) and H3K27ac (bottom) profiles across muscle groups. Spearman rank correlation coefficient (rs) is represented in color. (E,F) Representative UCSC browser tracks of H3K4me2 (E) and H3K27ac (F) ChIP-seq in EDL (orange), Quad (yellow), Sol (red), and Dia (purple) along Myh isoform genes. Myh6 is a cardiac isoform negative control gene, and Ngdn is a positive control gene invariant between different muscles. Numerical values for panels A, B, D, E, and F are available in S1 Data, and numerical values for panel C are available in S2 Data, S3 Data and S4 Data. ChIP-seq, chromatin immunoprecipitation sequencing; Dia, diaphragm; EDL, extensor digitorum longus; H3K4me2, histone 3 lysine 4 dimethylation; H3K27ac, histone 3 lysine 27 acetylation; Myh, myosin heavy chain; Ngdn, neuroguidin; Norm., normalized; qPCR, quantitative PCR; Sol, soleus; Quad, quadriceps femoris; UCSC, University of California, Santa Cruz.
Fig 2.
Patterns of open chromatin differ across muscle groups.
(A) Genomic distributions of ATAC-seq peaks across EDL, Quad, Sol, and Dia in promoter, intragenic, and intergenic regions. (B) Similarity matrix for ATAC-seq profiles across muscle groups. Spearman rank correlation coefficient (rs) is represented in color. (C) Representative UCSC browser tracks of ATAC-seq in EDL (orange), Quad (yellow), Sol (red), and Dia (purple) along Myh isoform genes. Myh6 is a cardiac isoform negative control gene, and Ngdn is a positive control gene invariant between different muscles. (D) Venn diagrams showing overlapping and unique regulatory regions (H3K4me2, H3K27ac, and ATAC-seq peaks) in EDL, Quad, Sol, and Dia. (E) Heatmap distribution of ATAC-seq, H3K4me2, and H3K27ac tags within 4 kb of all ATAC-seq peaks identified in EDL, Quad, Sol, and Dia; n = 1–2/group. Numerical values for panels A, B, C, and D are available in S1 Data, and numerical values for panel E are available in S5 Data, S6 Data, S7 Data, and S8 Data. ATAC-seq, assay for transposase-accessible chromatin sequencing; Dia, diaphragm; EDL, extensor digitorum longus; H3K4me2, histone 3 lysine 4 dimethylation; H3K27ac, histone 3 lysine 27 acetylation; Myh, myosin heavy chain; Ngdn, neuroguidin; Norm., normalized; Quad, quadriceps femoris; Sol, soleus; UCSC, University of California, Santa Cruz.
Fig 3.
SEs are distinct among muscle groups.
(A) Venn diagrams showing overlapping and unique SEs identified using three different methods (H3K27ac, Med1, and ATAC-seq) in EDL, Quad, Dia and Sol. (B) Overlap between the high-confidence SEs across the four muscle groups. High-confidence SEs were defined as those identified by at least two out of the three methods. (C) Heatmap of top ontologies for genes annotated to the high-confidence SEs in EDL, Quad, Sol, and Dia, with p-values represented in color. (D) Representative UCSC browser tracks of H3K27ac ChIP-seq (top), ATAC-seq (middle), and MED1 ChIP-seq (bottom) in EDL (orange), Quad (yellow), Sol (red), and Dia (purple) at a SE along Acta1 common to all muscles (left) and a glycolytic muscle-specific SE along Pvalb (right). Numerical values for all panels are available in S1 Data. Acta1, actin alpha-1; ATAC, assay for transposase-accessible chromatin sequencing; ATAC-seq, ATAC sequencing; AUF1, AU rich element RNA binding protein 1; ChIP-seq, chromatin immunoprecipitation sequencing; Dia, diaphragm; EDL, extensor digitorum longus; H3K27ac, histone 3 lysine 27 acetylation; MAPK, mitogen-activated protein kinase; Med1, mediator complex subunit 1; MHC, myosin heavy chain; Pvalb, parvalbumin; Quad, quadriceps femoris; ROS, reactive oxygen species; SE, super-enhancer; Sol, soleus; UCSC, University of California, Santa Cruz.
Fig 4.
Enhancer repertoires are uncorrelated between MTs in vitro and muscle tissue in vivo.
(A,B) Venn diagram showing overlapping and unique (A) H3K27ac and (B) ATAC peaks in C2C12 MT and Quad (left). Genomic distributions of the peaks common to MT and Quad in promoter, intergenic, and intragenic regions are shown (right). (C) Scatterplots of normalized H3K27ac (left) and ATAC (right) tag counts at peaks identified in MT, Quad, or both, respectively. The correlation coefficient (rs) is listed. (D) Histogram of H3K27ac (left) and ATAC (right) tags in MT and Quad centered on all peaks identified in Quad. (E) H3K27ac tags within 50 kb around the TSS for genes differentially expressed in Quad > MT (left) or MT > Quad (right); n = 2/group, ***p < 0.001. Violin plots display median (white square), interquartile range (black rectangle), 95% confidence interval (black line), and frequency (colored density plot). (F) Top ontologies for H3K27ac regions that annotate near genes expressed in Quad > MT (top) and MT > Quad (bottom). (G) Heatmap showing percentage of H3K27ac and ATAC peaks containing the top transcription factor motifs in MT or Quad. Numerical values for all panels are available in S1 Data. ATAC, assay for transposase-accessible chromatin; ETC, electron transport chain; ECM, extracellular matrix; ETS, E-twenty-six family of transcription factors; H3K27ac, histone 3 lysine 27 acetylation; Jun, jun proto-oncogene; MSigDB, Molecular Signatures database; MEF2, myocyte enhancer factor 2; MYF5, myogenic factor 5; NFY, nuclear transcription factor Y; MT, myotube; norm tag ct, normalized tag count; PAX, paired box; Quad, quadriceps femoris; RUNX, runt-related transcription factor; SP1, specificity protein 1; TCA, citric acid cycle; TEAD, TEA domain transcription factor 1; TSS, transcription start site.
Fig 5.
Enhancer regions reflect muscle group–specific function.
(A) Heatmap of pairwise comparisons showing the number of differential H3K4me2 (left) and H3K27ac (right) peaks in EDL, Quad, Dia, and Sol. Differential peak number is represented by color. (B) Heatmap distribution of H3K4me2 and H3K27ac tags within 6 kb of Quad-specific (top) and Sol-specific (bottom) H3K27ac peaks in Quad and Sol; n = 2/group. Color bars on the right correspond to promoter regions, promoter-distal regions with H3K4me2 peaks, or promoter-distal regions without H3K4me2 peaks. (C) Top three motifs enriched within H3K27ac peaks in Quad compared with Sol (top) and Sol compared with Quad (bottom). (D, E) Top five categories of genes annotated to H3K27ac peaks specific to Quad (D) or Sol (E), respectively. Numerical values for panels A, C, D, and E are available in S1 Data, and numerical values for panel B are available in S10 Data. AKT, protein kinase B; Dia, diaphragm; EDL, extensor digitorum longus; FA, fatty acid; FGFR, fibroblast growth factor receptor; GO, gene ontology; H3K4me2, histone 3 lysine 4 dimethylation; H3K27ac, histone 3 lysine 27 acetylation; KB, ketone body; MSigDB, Molecular Signatures database; MEF, myocyte enhancer factor; NFAT, nuclear factor of activated T cells; Norm., normalized; NUR77, nuclear hormone receptor 77; PDGFR, platelet-derived growth factor receptor; PPAR, peroxisome proliferator–activated receptor; Quad, quadriceps femoris; SIX, sine oculis homeobox factor; Sol, soleus; TG, triacylglycerol.
Fig 6.
Exercise and PGC1α modify the epigenome at different regions.
(A) Experimental scheme of voluntary wheel running (Ex) and skeletal muscle–specific Pgc1a overexpression (mTg). Ex mice were placed in cages with wheels for 4 weeks. (B) Scatterplots of normalized H3K27ac tag counts at regions marked by H3K27ac in Sed versus Ex Quad (left) and Wt versus mTg Quad (right). Correlation coefficient (rs) was calculated for each scatterplot; n = 2/group. (C) Similarity matrix with Spearman correlations for H3K27ac profiles in Quad and Sol, Ex and mTg Quad with correlation coefficient (rs) in color. (D) Pie charts with locations and H3K27ac signal changes at EMRs and mTg-modified regions (PMRs). Inner charts show peak distributions in promoter, intragenic, and intergenic regions. Outer charts indicate the direction of the acetylation change, with red for hyperacetylation and blue for hypoacetylation. (E) Heatmap distribution of H3K4me2 and H3K27ac tags within 6 kb of EMR (top) and PMR (bottom) peaks in Sed versus Ex Quads and Wt versus mTg Quads, respectively; n = 2/group. Color bars on the right correspond to promoter regions, promoter-distal regions with H3K4me2 peaks, or promoter-distal regions without H3K4me2 peaks. (F) Venn diagram with overlap between EMRs and PMRs. (G) Histograms showing distribution of H3K27ac tags within 6 kb of EMR peak centers in Ex and mTg mice and controls; n = 2/group. (H) Quantification of H3K4me2 (left) and H3K27ac (right) tag densities at EMR H3K27ac peaks in Sed, Ex, Wt, and mTg Quads. n = 2/group, ***p < 0.001. Numerical values for panels B, C, D, F, G, and H are available in S1 Data, and numerical values for panel E are available in S11 Data. EMR, exercise-modified region; Ex, exercised; H3K4me2, histone 3 lysine 4 dimethylation; H3K27ac, histone 3 lysine 27 acetylation; mTg, muscle-specific Pgc1a transgenic; norm tag ct, normalized tag count; Pgc1a, peroxisome proliferator–activated receptor gamma, coactivator 1 alpha; PMR, PGC1α-modified region; Quad, quadriceps femoris; Sed, sedentary; Sol, soleus; Wt, wild-type.
Fig 7.
Motif predictions and validations at EMRs.
(A) Top ontologies for genes annotated near hyperacetylated H3K27ac regions in Ex Quad. (B) Motifs enriched at hyperacetylated EMRs. (C) ChIP qPCR validation of c-JUN binding in Sed and Ex Quad. Enrichment is plotted as percent of input using technical duplicate control IgG and c-JUN ChIPs. (D) Top ontologies for genes annotated near hypoacetylated H3K27ac regions in Ex Quad. (E) Motifs enriched at hypoacetylated EMRs. (F) ChIP qPCR validation of SIX1 binding in Sed and Ex Quad. Enrichment is plotted as percent of input using technical duplicate control IgG and SIX1 ChIPs. Numerical values for all panels are available in S1 Data. CTF, CCAAT box-binding transcription factor; ChIP, chromatin immunoprecipitation; chr, chromosome; EMR, exercise-modified region; Ex, exercised; ECM, extracellular matrix; FA, fatty acid; H3K27ac, histone 3 lysine 27 acetylation; IgG, immunoglobulin G; Jun, jun proto-oncogene; KB, ketone body; MAFB, MAF bZIP transcription factor B; NKX2, NK2 homeobox; NUR77, nuclear hormone receptor 77; NR, nuclear receptor; PBX1, pre-B-cell leukemia homeobox 1; PDGF, platelet-derived growth factor; qPCR, quantitative PCR; Quad, quadriceps femoris; Sed, sedentary; SIX, sine oculis homeobox factor; SF1 steroidogenic factor 1; TCF3, transcription factor 3; TG, triacylglycerol.
Fig 8.
Exercise and Pgc1a overexpression convergently control gene expression.
(A) Volcano plots showing log2 FC in gene expression in Ex (left) or mTg (right) compared with Con quadriceps. Red represents absolute FC > 1.5 with an adjusted p-value < 0.05, blue represents p-value < 0.05, and black represents nonsignificant genes (n = 5/group). (B) Principal component analysis of all significantly altered genes (p < 0.05) in Ex or mTg compared with control. (C) Venn diagram showing transcriptional overlap in Ex and mTg. (D) Quadrant plot of log2 FC in gene expression in Ex/Con versus log2 FC in mTg/Con. (E) Heatmap of K-means clustering of transcriptomic profiles for differentially expressed genes. The number of genes in each cluster is in parenthesis. Each cluster, indicated by a vertical color bar, corresponds to quadrant colors from the previous panel. Two clusters were unique to Ex-regulated genes (bottom). Pathways enriched in each cluster and corresponding −log10 p-values are shown on the right. Numerical values for all panels are available in S1 Data. Con, control; ECM, extracellular matrix; ETC, electron transport chain; Ex, exercised; FC, fold change; Fpkm, Fragments Per Kilobase of transcript per Million mapped reads; mTg, muscle-specific Pgc1a transgenic; mTOR, mammalian target of rapamycin; PC1, principal component 1; PC2, principal component 2; Pgc1a, peroxisome proliferator–activated receptor gamma, coactivator 1 alpha; ROS, reactive oxygen species; Sed, sedentary; TCA, citric acid cycle; Wt, wild-type.
Fig 9.
Integrative cis-regulatory analysis predicts causal factors for exercise-induced reprogramming.
(A) Flowchart depicting the two approaches (BETA and IMAGE) used to perform integrative transcriptomic and epigenomic analysis. (B) Gene ontology analysis of differentially marked H3K27ac peaks (EMRs) associated with differentially expressed genes in Ex Quad versus Con identified by BETA. Directly up-regulated (red) and down-regulated (blue) gene numbers and ontologies are shown. (C) Representative UCSC browser tracks of H3K27ac (top) and RNA-seq (bottom) in Sed Quad (light purple) and Ex Quad (dark purple). (D) Motifs associated with the directly up-regulated (left) and down-regulated (right) genes identified by BETA. (E) Log p-value versus Pearson correlation plotted for causal transcription factors identified by IMAGE. Red = high-confidence factors, blue = medium-confidence factors, and gray = low-confidence factors. (F) Venn diagram showing overlap of EMRs with ERRγ peaks (left) and overlap of genes annotated near ERRγ binding sites with exercise-regulated genes (right). Numerical values for all panels are available in S1 Data. Acadl, Acyl-CoA dehydrogenase long chain; BETA, Binding and Expression Target Analysis; bZIP, basic leucine zipper; Con, control; EMR, exercise-modified region; ERR, estrogen-related receptor; Ex, exercised; H3K27ac, histone 3 lysine 27 acetylation; IMAGE, Integrated analysis of Motif Activity and Gene Expression; Myh, myosin heavy chain; NR, nuclear receptor; Quad, quadriceps femoris; RNA-seq, RNA sequencing; Sed, sedentary; TSS, transcription start site; UCSC, University of California, Santa Cruz.