Myod1 and GR coordinate myofiber-specific transcriptional enhancers

Abstract

Skeletal muscle is a dynamic tissue the size of which can be remodeled through the concerted actions of various cues. Here, we investigated the skeletal muscle transcriptional program and identified key tissue-specific regulatory genetic elements. Our results show that Myod1 is bound to numerous skeletal muscle enhancers in collaboration with the glucocorticoid receptor (GR) to control gene expression. Remarkably, transcriptional activation controlled by these factors occurs through direct contacts with the promoter region of target genes, via the CpG-bound transcription factor Nrf1, and the formation of Ctcf-anchored chromatin loops, in a myofiber-specific manner. Moreover, we demonstrate that GR negatively controls muscle mass and strength in mice by down-regulating anabolic pathways. Taken together, our data establish Myod1, GR and Nrf1 as key players of muscle-specific enhancer-promoter communication that orchestrate myofiber size regulation.

Figure 1.

Characterization of mouse skeletal muscle enhancers. (A) Tag density map of mouse skeletal muscle ATAC, H3K27ac and H3K27me3 locations, ±5 kb from the ATAC-seq peak center. (B, C) Tag density map of mouse skeletal muscle H3K27ac, H3K4me1, H3K4me3, H3K27me3 and Polr2 associated locations, ±5 kb from the peak center of H3K27ac (B), and corresponding average tag density profiles of the two identified clusters (C). (D, E) Tag density map of prostate, skeletal muscle, adipose tissue and liver H3K27ac binding sites, ± 5 kb from the peak center of enhancers identified in these four tissues (D), and corresponding average tag density profiles (E). (F, G) HOMER known motif analysis of active promoters (F) and active enhancers (G) in skeletal muscle. BG refers to estimated background. (H) Normalized expression of indicated genes in mouse gastrocnemius muscles. (I, J) Tag density map of mouse skeletal muscle H3K27ac and H3K4me1, and C2C12 myotube Myod1 binding sites, ±5 kb from the peak center of skeletal muscle enhancers (I), and corresponding average tag density profile of Myod1-bound enhancers (J).

Figure 2.

GR binding profile in mouse skeletal muscles. (A) Normalized gene expression of indicated genes in mouse gastrocnemius muscles. (B, C) Tag density map of mouse skeletal muscle H3K27ac, H3K4me1 and GR binding sites, ±5 kb from the peak center of skeletal muscle-specific enhancers (B), and corresponding average tag density profile of GR-bound enhancers (C). (D, E) Tag density map of mouse skeletal muscle GR, H3K27ac, H3K4me1, H3K4me3 and Polr2 binding sites, ±5 kb from the peak center of GR (D), and corresponding average tag density profiles of the two identified clusters (E). (F) Localization of GR, H3K27ac, H3K4me1, H3K4me3 and Polr2 at the Eif4ebp2 locus. The two GR binding sites localized at the enhancer (GBSe1 and GBSe2) and the promoter (GBSp1), and the non-specific binding region (ns) are boxed in red. (G) HOMER de novo motif analysis of GR response elements (GREs) peaks located at enhancers and promoter regions. BG refers to estimated background. (H) Chromatin immunoprecipitation (ChIP) followed by qPCR analysis (ChIP-qPCR) performed with anti-GR antibodies in skeletal muscle of control and GR^{(i)skm−/−} mice at GBSe1 (GRE1), GBSe2 (GRE2) and GBSp1. The non-specific binding region (ns) depicted in (F) was used as a negative control. n = 3 mice. Mean + SEM. *P < 0.05; **P < 0.01. (I) Microscale thermophoresis analysis and corresponding binding affinities of GR DBD to indicated DNA probes.

Figure 3.

Phenotypic characterization of GR^{(i)skm−/−} mice. (A) Volcano plot depicting in red genes differentially expressed in gastrocnemius muscles of GR^{(i)skm−/−} mice one week after GR ablation. (B) Overlap between genes with GR peaks and those down-regulated one week upon GR loss. (C) Pathway analysis of up- and down-regulated genes in gastrocnemius muscle of 9-week-old GR^{(i)skm−/−} mice. (D) Relative transcript levels of representative differentially expressed genes in gastrocnemius muscle of 16-week-old control and GR^{(i)skm−/−} mice. (E) Mass of gastrocnemius muscle from control and GR^{(i)skm−/−} mice at indicated ages. (F) qNMR analysis of total fat, lean and free body fluid (FBF) content of 16-week-old control and GR^{(i)skm−/−} mice. (G, H) Mean cross section area (CSA) (G) and fiber CSA distribution (H) of gastrocnemius muscle from control and GR^{(i)skm−/−} mice at 16 weeks. (I) Grip strength of 8- to 20-week-old control and GR^{(i)skm−/−} mice. (J) In vivo absolute maximal isometric tetanic force of tibialis anterior muscle from control and GR^{(i)skm−/−} mice at 16 weeks. (D−F and I) n = 10 mice, (G and H) n = 4 mice, (J) n = 5. Mean + SEM. *P < 0.05; **P < 0.01; ***P < 0.001.

Figure 4.

Transcriptomic and protein analyses of gastrocnemius muscles of control and GR^{(i)skm−/−} mice. (A, B) Pathway analysis of down- (A) and up- regulated genes (B) in gastrocnemius muscle of 16-week-old GR^{(i)skm−/−} mice. (C) Heatmap depicting the mean centered normalized expression of indicated genes selected by pathway analysis from the differentially expressed genes obtained by RNA-seq analysis performed in gastrocnemius muscle of 16-week-old control and GR^{(i)skm−/−} mice. (D, E) Representative Western blot analysis (D) and relative levels of the indicated proteins (E) in gastrocnemius muscle of 16-week-old control and GR^{(i)skm−/−} mice. βTubulin was used as a loading control. D, E: n = 10 mice. Mean + SEM. *P < 0.05, **P < 0.01.

Figure 5.

Characterization of Myod1 and GR binding at the Eif4ebp2 and Pik3r1 loci. (A) HOMER known motif analysis on Myod1 peaks located at GR-bound enhancer regions. BG refers to estimated background. (B, C) Tag density map of H3K27ac, H3K4me1, H3K4me3 and GR in skeletal muscles, and Myod1 in C2C12 myotubes, ± 5 kb from the GR peak center (B) and corresponding average tag density profiles of cluster 2 (C). (D) Overlap between genes bound by GR in skeletal muscle and by Myod1 in C2C12 myotubes obtained from 3 merged datasets, and genes that are down regulated in GR^{(i)skm−/−} mice. (E) Western blot analysis of skeletal muscle nuclear extracts immunoprecipitated with anti-GR or anti-Myod1 antibodies. Membranes were decorated with anti-GR and anti-Myod1 antibodies. rIgG served as a control for immunoprecipitation. Non-immunoprecipitated extracts (10% input) were also analyzed. (F) ChIP-qPCR analysis performed at indicated locations with anti-Myod1 antibodies or rIgG in skeletal muscle of wild-type mice. (G, H) ChIP-qPCR analysis performed at indicated locations with anti-GR (G) or anti-Myod1 (H) antibodies in C2C12 myotubes transfected with siRNA directed against Myod1 (siMyod1) (G), GR (siGR) (H) or scrambled siRNA (siCtrl). (I) Relative Eif4ebp2 and Pik3r1 transcript levels determined in C2C12 myotubes transfected with siCtrl, siGR or siMyod1. (F−I): n = 3 independent experiments in triplicate. Mean + SEM. *P < 0.05; **P < 0.01; ***P < 0.001.

Figure 6.

Characterization of Pik3r1 enhancer-promoter communication. (A) Quantitative local 4C-seq signal for the Pik3r1 viewpoint in skeletal muscle. GR binding sites (GR) are depicted with blue lines, GREs with blue stars, Ctcf binding sites identified by ChIP-seq with cyan arrowheads, indicating convergent motif orientation. H3K4me1 and Myod1 ChIP-seq profiles are presented in green and red, respectively. Enhancer-promoter contact regions are highlighted in orange on the 4C-seq chart. Ctcf-bound promoter and enhancer regions are boxed in cyan. (B, C) Quantitative local 4C-seq signal for the Pik3r1 viewpoint in skeletal muscle of control (upper panels in B and C) mice compared with skeletal muscle GR^{(i)skm−/−} mice (B, lower panel), or with double positive (DP) thymocytes (C, lower panel). Enhancer-promoter contact regions lost upon GR ablation and absent in DP thymocytes are highlighted in red. Contact regions present in both control and GR^{(i)skm−/−} mice are highlighted in light blue. Contact regions present in both control and DP thymocytes are highlighted in light green. Contact regions that are specific for GR^{(i)skm−/−} mice and DP thymocytes are highlighted in dark blue and dark green, respectively. GR binding sites (GBS) are depicted with black lines and GREs with black stars.

Figure 7.

Analysis of enhancer-promoter interactions in skeletal muscle. (A) Tag density map of H3K27ac, H3K4me1 and Ctcf, ± 5 kb from center of skeletal muscle-specific enhancers. (B, C) Tag density map of H3K27ac, H3K4me1, H3K4me3, GR and Ctcf in skeletal muscles, and of Myod1 in C2C12 myotubes, ± 5 kb from GR peak center (B) and corresponding average tag density profiles of indicated clusters (C). (D) HOMER de novo motif analysis at GR-bound promoter (−1000 bp to 100 bp) regions. BG refers to estimated background. (E) ChIP-qPCR analysis at GR-bound promoter regions of Eif4ebp2 (GBSp1) and Pik3r1 (GBSp2), performed with anti-GR, anti-Nrf1 or mouse immunoglobulin G (mIgG) in skeletal muscle of wild-type mice. The non-specific regions (ns) depicted in Supplementary Figure S5D and E were used as a negative control. (F) ChIP-qPCR analysis performed at indicated locations with anti-GR or anti-Nrf1 antibodies in C2C12 myotubes transfected with siRNA directed against Nrf1 (siNrf1) or scrambled siRNA (siCtrl). (G) Western blot analysis of skeletal muscle nuclear extracts immunoprecipitated (IP) with anti-GR antibodies. Membranes were decorated with anti-GR and anti-Nrf1 antibodies. Rabbit IgG (rIgG) served as a negative control for immunoprecipitation. Non-immunoprecipitated extracts (10% input) were also analyzed. (E, F): n = 3 independent experiments in triplicate. Mean + SEM. *P < 0.05; **P < 0.01; ***P < 0.001.