Genome-wide binding of the basic helix-loop-helix myogenic inhibitor musculin has substantial overlap with MyoD: implications for buffering activity

Abstract

Background: Musculin (MSC) is a basic helix-loop-helix transcription factor that inhibits myogenesis during normal development and contributes to the differentiation defect in rhabdomyosarcoma. As one of many transcription factors that impede myogenesis, its binding on a genome-wide scale relative to the widespread binding of the myogenic factor MyoD is unknown.

Methods: Chromatin immunoprecipitation coupled to high-throughput sequencing was performed for endogenous MSC in rhabdomyosarcoma cells and its binding was compared to that of MyoD in the same type of cells.

Results: MSC binds throughout the genome, in a pattern very similar to MyoD. Its binding overlaps strongly with regions enriched for acetylated histone H4, as well as regions that score high for DNase hypersensitivity in human myoblasts. In contrast to MyoD, MSC has a more relaxed binding sequence preference in the nucleotides that flank the core E-box motif.

Conclusions: The myogenic inhibitor MSC binds throughout the genome of rhabdomyosarcoma cells, in a pattern highly similar to that of MyoD, suggesting a broad role in buffering the activity of MyoD in development and rhabdomyosarcomas.

Figure 1

MSC has similar, but non-identical, DNA binding characteristics to MyoD and binds at many of the same genomic locations. (A) E-box motif enrichment of MSC and MyoD bound sites in RDs identifies a similar preference for central dinucleotide identity (GC and GG), but differing preferences in the E-box flanking nucleotides. (B) Comparison of the top 30,000 MyoD and MSC peaks in RDs demonstrates substantial overlap in the sites bound by each factor. Peaks were ranked by P value, and grouped into bins that increase by 3,000 peaks each time (that is, first the 3,000 most significant peaks are considered, than the 6,000 most significant, and so on). The fraction of the overlap is indicated by color, as depicted in the legend. (C)De novo motif analysis of peaks specific to MSC identifies an 8 bp motif (row 2) enriched at MSC-specific binding sites. The motif analysis compared MSC-specific binding sites to those sites that bound both MyoD and MSC. bp, base pair; fg.frac, bg.frac: fraction of foreground/background sequences that contain at least one motif occurrence; MSC, musculin; ratio, enriched/depleted ratio of motifs.

Figure 2

MyoD and MSC bind at unique identical and overlapping but non-identical sites in the genome. MSC and MyoD have both unique and overlapping binding patterns at various sites in the genome. Screenshots are shown from the UCSC Genome browser for MyoD and MSC ChIP-seq results at four distinct genomic locations (indicated below each panel, representing positions in hg19). The identity of the bHLH factor is indicated along the left, and E-boxes are represented as black marks along the bottom of each panel. Note that the number of MyoD reads in the ‘MSC only’ panel is five, in contrast to 298 reads for MSC, and are not centered on an E-box, and thus do not likely represent true MyoD binding. bHLH, basic helix-loop-helix; ChIP-seq, chromatin immunoprecipitation coupled to high-throughput sequencing; MSC, musculin.

Figure 3

MSC binding is associated with open chromatin. (A) Sites bound by MyoD and MSC are associated with acetylated histones. ChIP-seq for acetylated histone H4 (AcH4) was performed in RD cells and density plots constructed to compare the square root of the AcH4 value at all sites bound by MSC, MyoD or both factors. (B) MSC-specific and MyoD/MSC shared peaks are associated with higher levels of AcH4 near the transcription start site (TSS) of genes compared to MyoD-specific peaks. Density plots were constructed as in (A) for categories of peaks first split by peak identity (MyoD, MSC, shared), then subcategorized on distance from the nearest TSS. (C) Sites bound by MSC in RD cells overlap with DNase hypersensitive (HSS) sites in normal human myoblasts. Publicly available DNase HSS data from human myotubes were compared to the sites bound by MyoD and MSC in RD cells. Data for each factor category (for example, MSC-specific) are plotted as the fraction of peaks that overlap with locations that have a signal in the HSS data (that is, the graphed fraction = 1 – fraction of peaks at HSS score of ‘0’). AcH4, acetylated histone H4; bHLH, basic helix-loop-helix; ChIP, chromatin immunoprecipitation; ChIP-seq, chromatin immunoprecipitation coupled to high-throughput sequencing; HSS, hypersensitive; K, thousands of bp; MSC, musculin; TSS, transcription start site.

Figure 4

MSC dimers have relaxed requirements for flanking sequence compared to MyoD dimers. (A) MyoD homodimers and MyoD:E-protein heterodimers do not bind well to MSC-specific sequences, but bind after a small number of sequence changes. Electrophoretic mobility shift assays (EMSAs) were performed using in vitro translated proteins and probes as indicated. The asterisks indicate the location of what are, judging by their relative mobility, small amounts of E-protein homodimers. (B) MSC heterodimers can be competed off a preferred binding site equally well by competitors with variations in their flanking sequence, while MyoD heterodimers cannot. MyoD:E and MSC:E heterodimers were subjected to competition by excesses of cold probes as indicated. 25× and 50× refer to the excess mass of cold probe relative to hot probe. Variations in competitor sequences are indicated, and ‘CG Ebox’ refers to a probe with an inverted central dinucleotide sequence that abolishes all binding of MyoD and MSC. (C) Single nucleotide changes in flanking sequence can completely abrogate MyoD dimer binding, but still be permissive of MSC dimer binding. Shift assays were performed using proteins and probes as indicated. Each type of dimer combination was run in two lanes, with one lane having a probe with ‘A’ in the −2 position relative to the E-box, and the other lane having a probe with ‘T’ in that position, as indicated in red. All shifts were performed using a sufficient excess of probe so that visible free probe was present for all lanes (not shown in 4A and 4B). All probe counts were quantitated before addition to ensure there were roughly equivalent amounts in all compared lanes. Negative control lanes indicate lanes where probes were tested with an in vitro translated empty CS2 vector to identify any non-specific binding. EMSA, electrophoretic mobility shift assay; MSC, musculin.