Skeletal myosin light chain kinase regulates skeletal myogenesis by phosphorylation of MEF2C

In order to identify proteins that regulate MEF2C, P19 stem cells expressing a TAP‐tagged MEF2C were generated and characterized, confirming that MEF2C‐TAP is functional and can enhance both cardiomyogenesis and skeletal myogenesis (Supplementary Figure S1A–C), as shown previously for untagged MEF2C (Skerjanc et al, 1998; Ridgeway et al, 2000). Purification and mass spectrometric analysis of proteins interacting with tagged MEF2C on day 5 of differentiation led to the identification of skMLCK as a novel MEF2C‐interacting protein (Supplementary Figure S1D). Western blot analysis with anti‐skMLCK antibodies validated the mass spectrometric findings (Supplementary Figure S1D) and demonstrated that skMLCK is present at low levels in C2C12 myoblasts and upregulated during differentiation (Supplementary Figure S1F). SkMLCK can be localized to the nucleus and the cytoplasm (Pujol et al, 1993) (Supplementary Figure S1E), indicating that it can be located in the same subcellular compartment as the transcription factor MEF2C.

Identification of the interaction of MEF2C and skMLCK led us to study the expression profile of skMLCK during P19 EC myogenesis. Quantitative real‐time PCR (Q–PCR) analysis showed that the expression of skMLCK transcripts increased during differentiation, almost 10‐fold by day 9, compared with day 0 (Figure 1A). The transcripts of MEF2C increased on days 2–5 and 9, representing cardiac and skeletal muscle induction, respectively. GATA‐4, found in cardiac muscle precursors (Grepin et al, 1997), showed increased levels starting at day 2 and peaking at days 5–6 (Figure 1A). Cardiomyocytes were observed by day 6 of differentiation (Figure 1A; Supplementary Figure S3). The expression of the skeletal premyogenic mesoderm gene, Pax3, peaked from days 4 to 6 and the myogenic regulatory factors (MRFs) increased starting on day 7, indicating commitment into the skeletal muscle lineage (Figure 1A). Skeletal myocytes were observed by day 9 of differentiation (Figure 1; Supplementary Figure S3). Expression of structural proteins such as myosin heavy chain 3 (MHC3), MHC6, and cardiac α‐actin showed increased levels from days 5–6 and 8–9, indicating the waves of cardiac and skeletal muscle formation, respectively (Figure 1A). The difficulty of discerning skeletal versus cardiac muscle‐specific structural protein markers is evident, although MHC3 appears to be predominantly skeletal muscle specific. The time line for the various stages of skeletal and cardiac muscle, along with the genes expressed, is outlined in Figure 1B.

The novel interaction between TAP‐tagged MEF2C and skMLCK was validated using co‐immunoprecipitation assays. Endogenous MEF2C was co‐immunoprecipitated with endogenous skMLCK in extracts from C2C12 myocytes using skMLCK‐specific antibodies (Figure 2A). Further, co‐transfection of HEK‐293 cells with Flag‐MEF2C with constructs containing wild‐type skMLCK, revealed that the full‐length skMLCK protein physically interacted with MEF2C (Figure 2B, lane 2). To determine which domain of skMLCK interacted with MEF2C, several skMCLK mutants were examined. MEF2C‐interacting mutants included His‐skMLCK‐ATP and His‐skMLCK‐Calm, defective in the putative ATP and calmodulin‐binding domains, respectively (Figure 2B, lanes 5 and 6). In contrast, a C‐terminally truncated skMLCK mutant (skMLCK420) was inefficiently co‐immunoprecipitated with MEF2C, indicating that the interaction required the C‐terminal domain of skMLCK (Figure 2B, lane 7).

To determine if skMLCK could phosphorylate MEF2C, an in vitro kinase assay was performed. Purified His‐tagged MEF2C protein was phosphorylated by a commercially available skMLCK protein preparation in the presence of Ca²⁺, Mg²⁺, and [³²P]‐γ‐ATP as a phosphate donor and the absence of calmodulin (Figure 2C, lane 3). Phosphorylation of MEF2C was lost in the absence of skMLCK and reduced in the presence of EGTA or the skMLCK inhibitor ML‐7 (Figure 2C, lanes 4 and 6). Phosphorylation of MEF2C also occurred in the presence of calmodulin, which is required for optimal myosin light chain phosphorylation, but the autophosphorylation of skMLCK was more intense (Gao et al, 1992) (Supplementary Figure S2). Mass spectrometric analysis of the skMLCK preparation did not reveal any contaminating kinases. Silver staining showed an equal loading of the blot (Supplementary Figure S2). Thus, skMLCK phosphorylates MEF2C in vitro.

To identify the sites in MEF2C that are phosphorylated by skMLCK, in vitro kinase assays were performed followed by LC–MS/MS analysis (Abu‐Farha et al, 2008). Phosphorylation of MEF2C was observed on a tryptic peptide corresponding to amino acids 80–89 of MEF2C, identifying a single phosphorylation in this peptide at Threonine‐80 (T⁸⁰). Notably, the peptide was mapped to the MEF2 domain, which is essential for homodimerization and heterodimerization of MEF2 family members, binding to DNA, and interaction with the inhibitor HDAC4 (Lu et al, 2000). Alignment of the MEF2 domain shows that T⁸⁰ is conserved in MEF2C proteins from different species (Figure 2E).

To investigate whether skMLCK phosphorylates MEF2C in cells, in vivo phosphorylation assays were conducted. After incubation of transfected HEK‐293 cells with [³²P]‐orthophosphate, Flag‐MEF2C was purified on an anti‐Flag‐agarose resin and examined by western blot analysis to determine Flag‐MEF2C expression and by autoradiography to identify the level of phosphorylation. Quantification revealed a statistically significant 2.5‐fold increase in the intensity of Flag‐MEF2C phosphorylation in the presence of skMLCK (Figure 2D). Mutation of T⁸⁰ to Alanine, creating MEFT80A, resulted in a lack of enhancement of phosphorylation in the presence of skMLCK, suggesting that T⁸⁰ is the major site of MEF2C phosphorylation by skMLCK (Figure 2D). HEK‐293 cells did not contain endogenous MLCK activity, since treatment with ML‐7 did not change the baseline MEF2C phosphorylation levels (Figure 2D) and skMLCK protein was not detected by western blotting (Supplementary Figure S1F). Immunoblotting with anti‐flag antibodies indicated an equal loading of the purified Flag‐MEF2C protein (Figure 2D). Therefore, skMLCK binds and phosphorylates MEF2C on T⁸⁰.

To explore the functional relevance of MEF2C phosphorylation at T⁸⁰ on endogenous promoters during myogenesis, stable P19 cells expressing wild‐type Flag‐MEF2C, Flag‐MEFT80A, or Flag‐MEFT80D mutants were isolated and aggregated to induce differentiation. While MEFT80A is phosphorylation deficient, MEFT80D should be a phosphomimetic mutant. On day 9 of differentiation, immunofluorescence using antibodies against the muscle‐specific marker MHC showed that both P19[MEF2C] and P19[MEFT80D] cultures, but not P19[MEFT80A], enhanced skeletal muscle generation three‐fold compared with P19[Control] cultures (Figure 3A and B). Notably, all three cell lines induced comparable numbers of cardiomyocytes expressing MHC (MHC^+ve), indicated by their rounded morphology (data not shown), and contained comparable levels of Flag‐tagged wild‐type or mutant MEF2C protein (Figure 3C). Taken together, these data indicate that phosphorylation of T⁸⁰ in MEF2C is required for MEF2C‐enhanced skeletal myogenesis.

Q–PCR analyses supported the results obtained by immunofluorescence. While P19[MEF2C] and P19[MEFT80D] cultures showed a statistically significant enhancement of the skeletal myoblast/myocyte‐specific markers MyoD, myogenin and Myf5, compared with P19[Control] cells, P19[MEFT80A] cultures did not (Figure 3D). A similar pattern was observed for mRNA transcripts of the muscle structural proteins cardiac α‐actin and MHC3 on day 9, indicating a lack of upregulation of skeletal myogenesis in P19[MEFT80A] cultures. Note that under the conditions used, MEF2C enhanced skeletal myogenesis more efficiently than cardiomyogenesis (Figure 3D). Interestingly, the cardiac muscle markers GATA‐4 and Nkx2‐5 had similar increases in transcript levels in the wild‐type MEF2C and its mutant cell cultures (Figure 3D). Furthermore, cardiac α‐actin and MHC6 transcript levels on day 6 of differentiation, representative of cardiac muscle, were similar in all cell lines examined (Figure 3D), indicating a similar extent of cardiomyogenesis. The reduced ability to generate skeletal muscle by the MEFT80A mutant was not due to lower levels of MEF2C or skMLCK expression, since the transcript levels of wild‐type and mutant MEF2C and skMLCK on day 9 were similar in all three cell lines (Figure 3D). Taken together, our results show that the phosphorylation of T⁸⁰ in the MADS/MEF2 domain has a critical effect on the ability of MEF2C to enhance skeletal but not cardiac myogenesis.

The importance of the T⁸⁰ phosphorylation and the interaction of MEF2C with skMLCK prompted us to study the effect of skMLCK stable overexpression during P19 cell myogenesis. Cells were examined by immunofluorescence with MF20 on day 9 of differentiation and the number of MHC^+ve skeletal myocytes was counted and found to be increased four‐fold in P19[skMLCK] cultures compared with P19[Control] cells (Figure 4A and B). Q–PCR analysis of MRF expression revealed a 10–15‐fold induction over control of MyoD, Myogenin, and Myf5 transcripts, indicating enhanced commitment into skeletal muscle (Figure 4C). Further, a five‐fold increase was observed in mRNA levels of the structural genes cardiac α‐actin and MHC in P19[skMLCK] compared with P19[Control] cultures (Figure 4C). In contrast, the cardiac muscle marker, GATA‐4, and the skeletal muscle progenitor markers Pax3 and Meox1, were not significantly upregulated on day 6. These results suggest that skMLCK exerts its effect at the commitment stage of P19 skeletal myogenesis, and not cardiac myogenesis, by amplifying the expression of the MRFs and not the premyogenic mesoderm markers.

To investigate the effect of skMLCK inhibition on skeletal myogenesis in mES and P19 cells, ML‐7, a cell‐permeable, potent, and selective competitive inhibitor for the ATP‐binding site of MLCK was used (Kuhlmann et al, 2007). mES cells were differentiated in the presence of ML‐7 (0, 300, 600 nM) and examined by immunofluorescence for MHC on day 15. ML‐7 treatment resulted in a decrease of skeletal, but not cardiac, muscle formation (Figure 4D). The inhibition of skeletal myogenesis was due to a significant reduction (35–55%) in MyoD, myogenin, Myf5, and MHC3 mRNA transcript levels (Figure 4D). In contrast, GATA‐4 transcripts were unaffected (Figure 4D), indicating no loss of cardiomyogenesis. Furthermore, the skeletal premyogenic mesoderm markers, Meox1 and Pax3 showed no significant changes in their mRNA levels (Figure 4D). Inhibition of skeletal, but not cardiac, myogenesis at the stage of muscle commitment was also observed in P19 cells treated with ML‐7 (Supplementary Figure S3). Therefore, gain‐and‐loss‐of‐function studies identify skMLCK as a key regulatory kinase during the commitment stage of skeletal, but not cardiac, muscle formation in differentiating embryonic and P19 stem cells.

Since previous studies have shown that MEF2C is highly expressed in quiescent satellite cells (Pallafacchina et al, 2010), we isolated adult mouse satellite cells and examined their activation and differentiation status when freshly isolated (day 0) and after 3 days culture, with or without ML‐7. Q–PCR analysis of satellite cells cultured for 3 days showed upregulated levels of Pax7, MyoD, and Myf5, indicative of satellite cell activation, and upregulated levels of myogenin and MHC3, indicative of early myoblast differentiation (Figure 5). In the presence of ML‐7, MyoD, Myf5, myogenin, and MHC3 were significantly downregulated but Pax7 levels remained relatively unchanged (Figure 5). MEF2C was expressed at high levels in the freshly isolated satellite cells, in agreement with previous findings (Pallafacchina et al, 2010), and was still present but at reduced levels after 3 days of culture. Thus, MEF2C appears to be expressed at all stages of satellite cell myogenesis. Notably, treatment with ML‐7 reduced MEF2C expression significantly by day 3. Taken together, our data suggest that skMLCK is important for the activation of satellite cells and their subsequent differentiation.

Recent work has shown that Pax7 expression in satellite cells is regulated by the p38α kinase‐dependent repression by PRC2 (Palacios et al, 2010). We compared inhibition of skMLCK, by ML‐7, with inhibition of p38 kinase by SB203580 (SB). In agreement with the previous study (Palacios et al, 2010), SB markedly enhanced Pax7 expression and efficiently inhibited expression of MEF2C, myogenin, and MHC (Figure 5). Notably, both MyoD and Myf5 were upregulated by SB, although MyoD was upregulated only at the lowest concentration. Thus, in accordance with what has been previously reported (Palacios et al, 2010), SB expanded the population of activated satellite cells and inhibited their subsequent differentiation.

In comparing the two pathways, SB was a better inhibitor of differentiation than ML‐7 and the latter was a superior inhibitor of satellite cell activation than SB. To determine how the two pathways would interact, we treated satellite cells with both SB and ML‐7. The observed outcome was more similar to that of SB inhibition, with high Pax7 and Myf5 expression and greatly reduced MEF2C, myogenin, and MHC3 (Figure 5). Thus, the two pathways did not synergize but the SB treatment overrode the inhibition of MyoD and Myf5 by ML‐7. In summary, skMLCK regulates the expression of MyoD and Myf5 during satellite cell activation.

In light of the ability of skMLCK to phosphorylate MEF2C and enhance skeletal muscle commitment in P19 cells, the effect of mutant MEF2C and skMLCK was examined on the myogenic conversion of C3H10T1/2 embryonic fibroblasts by MyoD. Myogenic conversion was evaluated by Q–PCR analysis of cardiac α‐actin and myogenin transcripts, in cells transiently transfected with MyoD and/or skMLCK, MEF2C or its T⁸⁰ mutants. Co‐transfection of MEF2C or MEFT80D with MyoD significantly increased the transcript levels of cardiac α‐actin and myogenin (Figure 6A). In contrast, MEFT80A did not increase myogenin or cardiac α‐actin transcript levels over MyoD alone (Figure 6A). Co‐transfection with skMLCK significantly enhanced MyoD‐directed myogenic conversion and this enhancement was reduced by the skMLCK inhibitor, ML‐7. Finally, ML‐7 reduced the observed synergy between MyoD and MEF2C (Figure 6A). Similar levels of wild‐type and mutant MEF2C proteins were detected in these experiments, as shown by a western blot analysis with an anti‐Flag antibody (Figure 6B). SkMLCK transcripts were present at low levels in 10T1/2 fibroblasts and were upregulated by MyoD and MEF2C expression (data not shown), similar to the upregulation of skMLCK in myogenesis of C2C12 (Supplementary Figure S1F) and P19 cells (Figure 1). These results indicate that the synergy between MyoD and MEF2C activity on endogenous promoters is regulated, at least in part, by the phosphorylation of MEF2C‐T⁸⁰ by skMLCK.

Promoter analysis was performed to determine if the phosphorylation‐deficient (MEFT80A) or phosphomimetic (MEFT80D) mutants would modulate the transcriptional activity of MEF2C. Transient expression of wild‐type or mutant MEF2C resulted in similar levels of enhancement of luciferase activity, driven by a MyoD‐ and MEF2‐responsive promoter, in the presence or absence of MyoD (Figure 6C). Thus, in contrast to the above findings analysing endogenous gene expression, the phosphorylation of T⁸⁰ is not essential for the transcriptional activity of MEF2C, or its synergistic activation with MyoD, on non‐chromatinized, exogenous promoters.

Since MEFT80A could activate exogenous (Figure 6C) but not endogenous (Figures 3 and 6A) promoters, we examined the ability of MEFT80A to bind to endogenous MEF2 DNA‐binding sites, using chromatin immunoprecipitation (ChIP) followed by Q–PCR. Both MEF2C and MEFT80A bound to the myogenin and GATA‐4 promoters (Figure 7A), specifically in the region of a conserved MEF2 DNA‐binding site. Thus, the MEFT80A mutant appeared to bind MEF2 sites in chromatin as efficiently as wild‐type MEF2C protein.

To determine if histone methylation/acetylation levels were changed, ChIP was performed with antibodies against trimethylated‐K4 on Histone 3 (H3K4^me3) and against acetylated H3K14 (H3K14^ac), indicative of actively transcribed chromatin. The levels of H3K4^me3 on the myogenin promoter were comparable in P19[MEF2C] and P19[MEFT80A] cell lines, indicating similar levels of H3K4 trimethylation in these cell lines (Figure 7A). However, ChIP performed with antibodies against acetylated H3K14 indicated that the levels of H3K14^ac were significantly lower at the myogenin promoter, but not the GATA‐4 promoter, in P19[MEFT80A] cells as compared with P19[MEF2C] cells (Figure 7A).

To determine if the loss of acetylation was due to deficient recruitment of a histone acetyltransferase (HAT), we performed ChIP with antibodies against p300 and PCAF, which have been shown to bind MEF2C (Sartorelli et al, 1997). While wild‐type MEF2C could recruit both p300 and PCAF to the myogenin promoter, the MEFT80A mutant could not (Figure 7A). In contrast, p300 was recruited efficiently by both MEF2C and MEFT80A to the GATA‐4 promoter.

To determine if the decrease of p300 recruitment was due to inefficient binding to MEFT80A, co‐immunoprecipitation studies were performed in HEK‐293 cells transfected with wild‐type MEF2C, MEFT80A, or MEFT80D. A decrease in the interaction of p300 with MEFT80A was observed, compared with the interaction with MEF2C and MEFT80D (Figure 7B). Therefore, the MEFT80A mutant appears defective in enhancing H3K14 acetylation on skeletal, but not cardiac, promoters, likely due to a loss of recruitment of p300 and/or PCAF.

P19 EC cells were cultured as described (Wilton and Skerjanc, 1999; Karamboulas et al, 2006a) and stable cell lines were generated following the published procedures (Ridgeway and Skerjanc, 2001; Karamboulas et al, 2006b). Briefly, P19 cells were transfected using the FuGENE‐6 transfection system (Roche Diagnostics, USA) with 0.7 μg of PGK‐Puro, 1 μg of PGK‐LacZ, and 2.5 μg of B17 in the presence of 2.04 μg MEF2C‐TAP, Flag‐MEF2C, Flag‐MEFT80A, Flag‐MEFT80D, or His‐skMLCK constructs. P19 control cells were transfected with the appropriate vector control. Stable clonal populations were selected in a media containing 2 μg/ml puromycin. Clones with relatively similar expression of the exogenous genes were selected for further studies.

The differentiation protocol was performed as described previously (Karamboulas et al, 2006a, 2006b) with minor modifications. Cells were aggregated at a density of 5 × 10⁵ in Petri dishes in serum‐supplemented medium containing 1% DMSO. Four days later, the aggregates were plated in tissue culture plates. At days 6 or 9 following differentiation, cultures were fixed or RNA/protein were harvested, as described (Skerjanc et al, 1998; Skerjanc and Wilton, 2000). Suboptimal serum conditions were used to minimize the extent of skeletal muscle development in the control cultures (Skerjanc, 1999; Wilton and Skerjanc, 1999).

D3 mouse ES cells were grown and differentiated as described previously (Kennedy et al, 2009). Briefly, differentiation was induced by aggregating the cells at a density of 4 × 10⁴ cells/ml, for 2 days in hanging drops and 5 days in NTC plates. On day 7, the embryoid bodies were transferred to TC plates or 0.1% gelatin‐coated glass coverslips. On day 10, the cells were cultured in low serum medium containing N2 supplement and allowed to grow for a further 5 days, at which time skeletal myocytes were observed. The cells were treated with ML‐7 (0, 300, or 600 nM) from days 3 to 15 of differentiation. These concentrations are far below the concentrations at which ML‐7 inhibits other kinases. According to the manufacturer (http://www.emdchemicals.com/life-science-research/ml-7-hydrochloride/EMD_BIO-475880/p_uuid?ProductID=PwGb.s1OzZcAAAEiLTxCeVC_), the primary target for ML‐7 is MLCK with K_i=300 nM, the secondary targets are PKA (K_i=21 μM) and PKC (K_i=42 μM).

The hind limb muscles were isolated from 3‐week‐old C57BL/6 mice (five animals treated individually). The muscles were digested with 10 mg/ml collagenase in DMEM for 1 h at 37°C. Satellite cells were isolated by three cycles (1 h each) of incubation in tissue culture plates to remove fibroblasts. The satellite cells were placed on Matrigel‐coated (Invitrogen) plates in growth media containing 10% fetal‐calf serum, 100 U/ml penicillin/streptomycin, and 292 ng/ml l‐glutamine in DMEM for 3 days. Under these conditions, both the activation of satellite cells, as shown by MyoD and Myf5 upregulation, and their early differentiation, as shown by myogenin and MHC3 upregulation, were observed. However, the cultures appeared to be just beginning to differentiate and multinucleated myotubes were not yet observed. On a daily basis, cells were treated with ML‐7 or SB203580 (Calbiochem) at final concentrations of 0.6, 1.2, or 5 μM, or both drugs at 1.2 μM.

Human MEF2C cDNA was subcloned into the appropriate restriction sites of the eukaryotic expression plasmids pCMV‐Taq2 and pCTAP (Stratagene, USA). MEF2C mutants, MEFT80A and MEFT80D, were generated by a Site‐Directed Mutagenesis Kit (Stratagene, USA) using the primers in Supplementary Table S1 and following the protocol recommended by the manufacturer. Human skMLCK was cloned from cDNA generated from human skeletal RNA (Clontech, USA) by PCR using Advanced‐GC cDNA PCR kit (Clontech, USA) and skMLCK‐specific primers (Supplementary Table S1). The skMLCK mutants resulting in the loss of the ATP‐binding site, or the calmodulin‐binding site were generated using a Site‐Directed Mutagenesis Kit (Stratagene, USA; Supplementary Table S1). The skMLCK C‐terminal deletion mutant was generated by an insertion of a stop codon after amino‐acid 420, resulting in a deletion of the C‐terminal 300 aa containing the putative kinase active site and the calmodulin domains.

Total cellular proteins were prepared by lysing cells in modified RIPA buffer and western blots were performed as described (Al‐Madhoun et al, 2004; Abu‐Farha et al, 2008). Western blots were probed with antibodies against the calmodulin‐binding peptide, MEF2C, skMLCK (Santa Cruz Biotechnology, USA), anti‐His tag (Invitrogen, USA), monoclonal anti‐FLAG M2‐peroxidase (HRP) or β‐actin (Sigma‐Aldrich, USA). Co‐immunoprecipitation and silver staining was performed as described previously (Al‐Madhoun et al, 2007; Abu‐Farha et al, 2008).

For immunofluorescent labelling, cells were fixed in cold methanol and incubated first with mouse anti‐MHC monoclonal antibody supernatant MF20 (Bader et al, 1982), followed by goat anti‐mouse IgG(H+L) Cy3‐linked antibody (Jackson ImmunoResearch Laboratories), as described previously (Gianakopoulos and Skerjanc, 2005). Luciferase activity was measured using the Dual luciferase assay system (Promega, USA) as previously described (Al‐Madhoun et al, 2004).

Total RNA was extracted from cells using the RNeasy Kit (Qiagen Inc., Canada) following the protocol described by the manufacturers. First‐strand cDNA was synthesized from 1 μg RNA by reverse transcription using QuantiTect Reverse Transcription Kit (Qiagen Inc., Canada). Q–PCR reactions were performed as described (Savage et al, 2009). Primer pairs (Supplementary Table S2) were selected from the PrimerBank (Wang and Seed, 2003) at website (http://pga.mgh.harvard.edu/primerbank). The reactions and data analysis were performed on the ABI 7300 system (Applied Biosystems, Canada) using SDS software. Relative gene expression was calculated using the comparative Ct method as previously described (Savage et al, 2009). Unless indicated otherwise, results were normalized to β‐actin, and averages±s.e.m. are shown expressed as fold‐change relative to P19[Control] cells.

MEF2C cDNA was constructed into PET30a bacterial expression vector and was expressed in Escherichia coli, BL 21 (DH3) pLys host cells, which were induced by 0.5 mM IPTG (Al‐Madhoun et al, 2002). The protein was purified on Ni‐NTA His‐binding resins column (Invitrogen, USA) (Al‐Madhoun et al, 2002). Purified human skMLCK was obtained from a commercially available source (ProQinase Tools and Tests, USA).

Statistical significance was estimated with a one‐tailed Student's t‐test assuming equal variance and error is s.e.m. (^* indicates P<0.05). Sequences were extracted from PubMed website and alignments were performed using Clustal W program (http://www.ebi.ac.uk/Tools/clustalw2). The accession numbers for MEF2C were H. sapiens NP_002388; M. musculus NP_079558.1; C. lupus XP_536298.1; B. taurus NP_001039578.1; G. gallus XP_001231662.1; D. rerio AAC05226.1; C. elegans NM_060040.4; Drosophila NP_477018.1.

The skMLCK phosphorylation assay was carried out as described (Kennelly et al, 1990) with minor modifications. Purified His‐MEF2C and skMLCK were incubated at 30°C for 30 min in a buffer system containing 20 mM Tris–HCl, pH 7.5, 10 mM MgCl₂, 0.5 mM NaCl, 5 mM DTT, 1 μM ATP, 0.3 μM [γ‐³²P]‐ATP (NEN Life Science Products, USA), with and without 0.5 mM CaCl₂, 25 mM EGTA, or 5 μM ML‐7. The enzyme activity was terminated by heat inactivation in the presence of 1 × SDS‐loading buffer.

The in vivo kinase assay was performed as described previously (Huang and Chen, 2005) with some modifications. HEK‐293 were co‐transfected with Flag‐MEF2C or Flag‐MEFT80A constructs with/without His‐skMLCK in the presence or absence of ML‐7. Eighteen hours later, the cells were incubated in phosphate‐free Dulbecco's modified Eagle's medium for 5 h. Fresh phosphate‐free DMEM media containing 0.05 mCi/ml [³²P]‐orthophosphate (NEN Life Science Products, USA) was then added to the cells for another 4 h. The labelled cells were washed with Tris‐buffered saline and lysed in 1 ml of lysis buffer containing 50 mM Tris–HCl, pH 7.5, 300 mM NaCl, and 1 × Mini Complete protease‐inhibitor cocktail and 1 × phosphatase‐inhibitor cocktail, PhosSTOP (Roche, Laval, Canada). Labelled Flag‐MEF2C was purified from cell extracts on anti‐Flag‐agarose resin as described in the manufacturer's protocol (Sigma‐Aldrich, USA), and examined by western blot using monoclonal anti‐FLAG M2‐peroxidase (HRP; Sigma‐Aldrich, USA) antibodies. Labelled proteins were visualized using a phosphorimager (Typhoon TRIO, variable mode imager, GE Healthcare Bio‐science).

10T1/2 fibroblasts were cultured in 10% 1:1 cosmic calf‐fetal bovine serum in α‐minimum Eagle's medium. One day before transfection, cells were seeded in a six‐well plate at a density of 1 × 10⁵ cells/well. The cells were then transfected using the FuGENE‐6 transfection reagent following the procedure recommended by the manufacturer (Roche Diagnostics, USA). The myogenic conversion was driven by 1 μg Flag‐MyoD in the absence or presence of 1 μg His‐skMLCK, Flag‐MEF2C, Flag‐MEFT80A, or Flag‐MEF2T80D constructs and 0.2 μg pEGFP‐N1 vector (Clontech Laboratories, Inc., Palo Alto, CA). After 48 h, cells were transferred to differentiation medium, containing 0.5% horse serum, for 4 days with or without daily administration of 600 nM ML‐7 (Calbiochem, USA). Transfection efficiency was scored by analysing GFP transcripts and myogenic conversion was quantified by real‐time Q–PCR analysis.

ChIP assays were performed as described (Savage et al, 2009) with minor modifications. P19, P19[MEF2C], or P19[MEFT80A] clones were aggregated for 4 days in the presence of DMSO, and cells were harvested for ChIP analysis on day 7. Relative enrichment of binding sites compared with the IgG negative control immunoprecipitation was analysed from 50 μg of chromatin using Q–PCR, as described above. Primers sequences for myogenin and GATA‐4 are listed in (Supplementary Table S2).

Supplementary data are available at The EMBO Journal Online (http://www.embojournal.org).