Regulation of proline-directed kinases and the trans-histone code H3K9me3/H4K20me3 during human myogenesis

Abstract

We present a system-level analysis of proteome, phosphoproteome, and chromatin state of precursors of muscle cells (myoblasts) differentiating into specialized myotubes. Using stable isotope labeling of amino acids in cell culture and nano-liqud chromatography-mass spectrometry/mass spectrometry, we found that phosphorylation motifs targeted by the kinases protein kinase C, cyclin-dependent kinase, and mitogen-activated protein kinase showed increased phosphorylation during myodifferentiation of LHCN-M2 human skeletal myoblast cell line. Drugs known to inhibit these kinases either promoted (PD0325901 and GW8510) or stalled (CHIR99021 and roscovitine) differentiation, resulting in myotube and myoblast phenotypes, respectively. The proteomes, especially the myogenic and chromatin-related proteins including histone methyltransferases, correlated with their phenotypes, leading us to quantify histone post-translational modifications and identify two gene-silencing marks, H3K9me3 and H4K20me3, with relative abundances changing in correlation with these phenotypes. ChIP-quantitative PCR demonstrated that H3K9me3 is erased from the gene loci of myogenic regulatory factors namely MYOD1, MYOG, and MYF5 in differentiating myotubes. Together, our work integrating histone post-translational modification, phosphoproteomics, and full proteome analysis gives a comprehensive understanding of the close connection between signaling pathways and epigenetics during myodifferentiation in vitro.

Keywords: Kinase; acetylation; acetyltransferase; chromatin; histone; mass spectrometry; myogenesis; myotube.

Figure 1.

Workflow for proteome-wide phosphorylation analysis. The “forward” experiment was performed in three biological replicates to predict kinases actively involved in myogenesis. Proliferating mononuclear myoblasts were SILAC-labeled. In the “reverse” experiment (drug treatment analysis), SILAC labeling was applied to differentiating myotubes, whereas unlabeled cell cultures were proliferating myoblasts (control) or cells treated with drugs. The kinase inhibitors were 1 μm PD325901, 1 μm GW8510, 10 nm roscovitine, and 3 μm CHIR99021. Light and heavy amino acid–labeled cells were lysed and mixed in equal amounts prior sample processing. Peptides were purified and fractionated using SCX. The fractions collected were enriched for phosphopeptides using TiO₂, and flow-through was analyzed for total proteome. The data shown are averages of at least three independent experiments. nLC, nano-LC.

Figure 2.

Prediction of kinase families playing active roles in myodifferentiation. A, heat map displaying the 193 phosphopeptides (phospho) and respective protein abundance from triplicate experiments represented as high confidence log2 ratios of myoblast/myotube. “Normalized” indicates phosphopeptide ratio normalized to protein ratio. FT, flowthrough. B, serine-centered motif analysis performed with motif-x (http://motif-x.med.harvard.edu) (65)⁴ searching phosphosites up- and down-regulated in myoblasts and myotubes. The background set was the human IPI database. C, statistically overrepresented signaling pathways for 3641 phosphosites using PathwayLinker. Proteins in the current study were compared with all proteins for the pathway. The cells are colored by increased p value enrichment.

Figure 3.

Staining via fluorescent microscopy of MHC in cell culture to assess myotube differentiation. A, phase (top left panels), Alexa 488 (top right panels), Hoechst 33341 (bottom left panels), and merged images representing the characterization of proliferating myoblast (panel i) and differentiating myotubes (panel ii). Scale bars for all the images indicate 0.05 mm. Alexa 488 (green) stains for MHC, which is the marker of myotube phenotype. MHC also shows the fusion of cells into a multinucleate myotube. B, Alexa 488 staining (left panels) and phase images (right panels) show cell differentiation for control (day 2, panel i), MAPK inhibitor (panel ii), CDK inhibitor (panel iii), control (day 6; panel iv), CMGC/GSK inhibitor (panel v), and CMGC/CDK/CDC2 inhibitor (panel vi). PD, PD325901; GW, GW8510; CHIR, CHIR99021.

Figure 4.

k means clustering analysis of protein trends across analyzed conditions. A, k means analysis grouped data into 18 clusters with different trends estimated using z score normalization of protein abundance across conditions. The first represented cluster (left panel) shows proteins more abundant in GW8510 (GW) and PD0325901 (PD) treatment, because the y axis represents the normalized z score for ratio of normal differentiation/drug treatment. A negative value indicates that the function is promoted by the drug. On the right, the biological process was enriched using the Gene Ontology database. p values are illustrated on the right. Connector lines indicate similar biological functions. Ball size represents relative number of genes that belong to that function. B, left panel, cluster including proteins down-regulated in GW8510, PD0325901, and CHIR99021 (CHIR) treatment. Right panel, most of these proteins are unrelated to muscle development.

Figure 5.

Label-free analysis of post-translational modifications on histone H3 and H4. A, heat map representing regulation of histone PTMs in the six conditions (three biological replicates each) analyzed. Color coding represents relative abundance across conditions (z score normalized). H3K9me3 and H4K20me3 were selected for further analysis, because they show an overall lower abundance in myotubes, PD0325901 (PD) and GW8510 (GW). B, z score summed values for all modification forms. Although acetylation is enriched in myotube-like states, methylations are higher among all myoblast-like cell states. C, methylations on specific histone H3 and H4 sites. D, log2 fold change of different methyl transferases known to be catalyzing the given histone PTMs (listed below). The bar plot represents their abundance in differentiation (Diff) relative to the rest of the five conditions. Prolif, proliferation; CHIR, CHIR99021; Rosco, roscovitine.

Figure 6.

Analysis of myotube biomarkers and genome mapping of the repressive trans-histone code H3K9me3/H4K20me3. A, global relative abundance of H3K9me3 and H4K20me3. B, qPCR of the myogenic genes MYOD, MYOG, and MYF5 in different conditions (n = 3 cultures each). C, occupancy of the histone marks H3K9me3 and H4K20me3 at promoters (prom.), TSS, and 3′-end of gene body of MYOD, MYOG, and MYF5 using ChIP–qPCR. The values are normalized to input and to background IgG. Blue indicates that the modification disappears in drug treatment, or normal differentiation (Diff), because these values are represented as log2 ratios to proliferation (Prolif). As expected, in normal differentiation and PD0325901 (PD), these marks are depleted and allow expression of these genes, promoting muscle formation. GW8510 (GW) treatment did not lead to removal of the repressive marks, as would have been expected in myotube samples. CHIR, CHIR99021; Rosco, roscovitine.