Tissue-specific splicing of a ubiquitously expressed transcription factor is essential for muscle differentiation

The muscle-specific isoform of Mef2D drives the late stages of myogenesis

Advances in our understanding of the relative roles of Mef2D isoforms in myogenesis have been hampered by the lack of reagents that distinguish between the endogenous proteins. Consequently, we generated antibodies that discriminate between the muscle-specific Mef2Dα2 isoform and its ubiquitously expressed counterpart, Mef2Dα1 (Fig. 1A). Western blot of recombinant Mef2D proteins demonstrate the specificity of our antibodies (Supplemental Fig. 1). Furthermore, these antibodies recognize the endogenous Mef2D isoforms in differentiating myoblasts using Western blot, immunofluorescence, and immunoprecipitation (Supplemental Fig. 1; data not shown). Importantly, a similar efficiency of immunoprecipitation is observed with both isoform-specific antibodies.

Initially, we characterized the expression of Mef2D isoforms during murine myogenesis. Temporal analysis demonstrates a clearly distinct pattern of expression for the two Mef2D isoforms (Fig. 1B,C). In the case of the ubiquitously expressed Mef2Dα1, the protein is present in proliferating myoblasts and continues to accumulate throughout differentiation. In contrast, the muscle-specific Mef2Dα2 is absent from proliferating myoblasts and is strongly induced between 48 and 72 h of differentiation (Fig. 1B,C). This pattern of expression suggests a role for Mef2Dα2 in late myogenesis.

To further explore the role of Mef2D isoforms in myogenesis, we created isogenic C2 myoblast lines where the exogenous expression of either Mef2Dα1 or Mef2Dα2 can be induced upon doxycycline (Dox) treatment (Supplemental Fig. 2A). Interestingly, myoblasts induced to express Mef2Dα2 formed multinucleated, myosin heavy chain (MHC)-positive cells more quickly than the control cells (Fig. 2A; Supplemental Fig. 2B), suggesting an accelerated differentiation. In contrast, expression of Mef2Dα1 appeared to block myogenesis, as fewer MHC⁺ cells were observed in this population at late stages (120 h) of differentiation. Having observed opposing effects of Mef2Dα1 and Mef2Dα2 on differentiation in cultured myoblasts, we next examined whether these disparate roles of Mef2D isoforms are also observed in vivo using a model of mouse muscle regeneration. Indeed, injection of retrovirus expressing Mef2Dα1 into cardiotoxin-injured tibialis anterior (TA) muscle resulted in severely impaired regeneration where myofiber repair was delayed (Fig. 2B; Supplemental Fig. 3A), resulting in an accumulation of undifferentiated MyoD⁺ muscle cells in the interstitial space (Supplemental Fig. 4). In contrast, injection of retrovirus expressing Mef2Dα2 resulted in improved muscle regeneration with an increased muscle fiber caliber compared with the contralateral control TA muscle (Fig. 2B; Supplemental Fig. 3B). Furthermore, only the injection of Mef2Dα2 could mediate the previously described Mef2D-dependent fiber type switching in the TA muscle (Potthoff et al. 2007) from a predominantly fast type MHC-IIb myofiber toward a slower muscle isotype that contained a mixture of MHC-IIb and MHC-IIa myofibers (Supplemental Fig. 5). Thus, Mef2Dα1 and Mef2Dα2 have antagonistic effects on myogenesis and muscle regeneration.

Examination of Mef2Dα2-expressing myoblasts showed that MHC proteins were expressed at the earliest stages of differentiation (0 h) (Fig. 2A). This temporal shift in MHC expression led us to examine whether Mef2Dα2 expression would allow for the expression of muscle genes in proliferating myoblasts. To our surprise, RT-qPCR analysis demonstrates that expression of Mef2Dα2, but not Mef2Dα1, can activate Myh3 along with other late myogenic genes (Mylpf and Tnnt1) in proliferating myoblasts (Supplemental Fig. 6). To determine whether the expression of these late markers of myogenesis was due to a precocious differentiation, we examined the ability of Mef2D isoforms to activate expression of Myog and Ckm—two muscle-specific genes normally up-regulated at intermediate stages of myogenesis. In contrast to Myh3, which is strongly up-regulated by ectopic expression of Mef2Dα2 in proliferating myoblasts, we observed no change in transcription of Myog or Ckm upon expression of the muscle-specific isoform (Supplemental Fig. 6). Taken together, our findings suggest that alternate splicing generates the muscle-specific Mef2Dα2 isoform to temporally regulate activation of late muscle genes.

We next sought to understand the mechanism underlying the differential ability of Mef2D isoforms to activate late muscle gene expression. Two potential models were envisaged: (1) Mef2Dα1 and Mef2Dα2 recruit similar coregulatory proteins to an alternate subset of muscle genes or (2) Mef2Dα1 and Mef2Dα2 bind to a similar set of genomic loci, leading to the recruitment of alternate coregulatory proteins. We therefore explored the contribution of these two models to the functional differences observed between Mef2D isoforms.

Genome-wide binding shows that Mef2D isoforms bind to a largely overlapping set of genes

We first explored the contribution of altered genomic targeting to the differential ability of Mef2D isoforms to activate transcription. For this, we performed chromatin immunoprecipitation (ChIP) sequencing (ChIP-seq) to examine the genomic distribution of Mef2Dα1 and Mef2Dα2 in myotubes. Fragmented chromatin was immunoprecipitated using our isoform-specific antibodies to isolate DNA associated with the endogenous Mef2Dα1 and Mef2Dα2 proteins (Supplemental Figs. 7, 8). Bioinformatic analysis of ChIP-seq data identified 7910 binding sites for Mef2α1, while the muscle-specific isoform Mef2Dα2 bound to 9103 sites that are not present in the control (Fig. 3A; Supplemental Table 1). Comparison of the Mef2D-binding sites using a highly stringent cutoff of P < 10⁻⁶ revealed that 66% of Mef2Dα1-bound peaks are also bound by Mef2Dα2 (Fig. 3A). However, when the stringency is modestly reduced to P < 10⁻⁴, the overlap in binding of Mef2Dα1 at Mef2Dα2-bound peaks increases to 85% (data not shown). Consistent with the fact that Mef2Dα1 and Mef2Dα2 bind to a largely overlapping set of genomic loci, position weight matrices for Mef2D-binding elements observed under these peaks show that both isoforms of the protein recognize a nearly identical consensus sequence of CTAAAAATAG (Fig. 3B). A de novo search for DNA sequences enriched under peaks identified Mef2 and E-box sequences (i.e., MyoD/Myog/Myf5/MRF4-binding sites) as the two most highly represented DNA elements at both Mef2Dα1 and Mef2Dα2 genomic targets (data not shown). Interestingly, examination of genomic loci bound preferentially by either Mef2Dα1 or Mef2Dα2 did not identify Mef2-binding sites as the most highly represented DNA sequences found under these peaks (Supplemental Fig. 9). This finding suggests that the differential binding observed between Mef2Dα1 and Mef2Dα2 at a small subset of genes is likely due to altered targeting of the transcription factor to DNA through an indirect tethering mechanism.

To assess the importance of overlapping Mef2Dα1 and Mef2Dα2 genomic binding to myogenesis, we performed gene ontology (GO) analysis of peaks bound by both isoforms. Genes targeted by both Mef2Dα1 and Mef2Dα2 are strongly associated with GO annotations defining a role in muscle development (Fig. 3C; Supplemental Table 1). Among these genes, we examined Mef2D binding at the HDAC9 gene, which is known to participate in a negative feedback loop with Mef2 proteins (Haberland et al. 2007). At this locus, we observed peaks within the HDAC9 promoter for both Mef2Dα1 and Mef2Dα2 that overlap with MyoD binding (Fig. 3D). Similarly, we observed that the promoters of Myh3, Titin, Myog, Tnnt1, Ckm, and Mylpf are bound equally well by both Mef2Dα1 and Mef2Dα2 in differentiating myotubes (Supplemental Fig. 8). RNA-seq analysis of transcripts that are up-regulated at 24 h of differentiation when expression of Mef2D isoforms is induced in C2 cells identified 227 genes activated by Mef2Dα2 (Supplemental Table 1). Combining the RNA-seq data set with Mef2D peaks within 50 kb of the transcription start site identified 107 genes whose expression may be directly modulated by Mef2Dα2 (Supplemental Table 1; Supplemental Fig. 10). Consistent with Mef2Dα2 modulating the late muscle gene expression program, GO analysis of putative Mef2Dα2 direct target genes showed a highly significant enrichment for genes involved in muscle development and muscle contraction (Supplemental Fig. 10A). Importantly, although Mef2Dα1 bound to most of the direct transcriptional targets of Mef2Dα2, exogenous expression of the ubiquitous Mef2D isoform did not lead to a significant increase in expression at any of these genes (Supplemental Table 1; Supplemental Fig. 10B). Similarly, ChIP-qPCR analysis of Mef2D isoform binding in proliferating myoblasts demonstrates that both Mef2Dα1 and Mef2Dα2 bind to the Myh3 and Mylpf promoters (Supplemental Fig. 6B), although only Mef2Dα2 binding gives rise to transcriptional activation. Taken together, these results demonstrate that altered DNA binding is not the major determinant of the differential activation of transcription by Mef2D isoforms.

The muscle-specific Mef2Dα2 isoform preferentially interacts with coactivator complexes

To assess the possibility that Mef2Dα1 and Mef2Dα2 associate with alternate coregulatory proteins to regulate muscle gene expression, we performed a proteomic analysis of proteins that interact with each Mef2D isoform. Initially, we examined proteins that interact with recombinant Mef2Dα1 and Mef2Dα2. In a pull-down assay, Flag-tagged Mef2Dα1 or Mef2Dα2 proteins were used as bait to isolate interacting proteins from nuclear extracts prepared from myotubes (Supplemental Fig. 11). Mass spectrometry analysis of Mef2Dα1- or Mef2Dα2-interacting proteins (Table 1) identified components of the coactivator Ash2L methyltransferase complexes, corepressor HDAC deacetylase complexes, and the SWI/SNF ATP-dependent chromatin remodeling complexes that have previously been reported to associate with Mef2D (Ohkawa et al. 2006, 2007; Rampalli et al. 2007; Du et al. 2008). Qualitative analysis of the identified peptides showed that recombinant Mef2Dα1 and Mef2Dα2 interact with the same sets of proteins (Table 1). This finding suggests that the exchange of mutually exclusive exons in Mef2D does not generate proteins with distinct protein–protein interaction domains. However, the experiment was performed with recombinant Mef2D isoforms. To determine whether Mef2Dα1 and Mef2Dα2 isoforms also interact with the same sets of proteins in their native cellular context, we immunoprecipitated endogenous Mef2Dα1 and Mef2Dα2 proteins from myotube extracts and tested the interaction with the previously identified cofactors by Western blot (Fig. 4A). Strikingly, in these conditions, we observed differential interactions for Mef2Dα1 versus Mef2Dα2. Precisely, the muscle-specific Mef2Dα2 interacts with the Ash2L coactivator complex but not the repressive HDAC4 and HDAC9 proteins. Conversely, the ubiquitously expressed Mef2Dα1 does not associate with Ash2L but interacts strongly with HDAC4 and HDAC9. Therefore, in contrast to their recombinant counterparts, endogenous Mef2D isoforms interact with distinct protein complexes in differentiating myotubes—Mef2Dα1 interacts with corepressors, while Mef2Dα2 interacts with coactivators.

Mef2Dα2 protein evades inhibitory protein kinase A (PKA) signaling

Based on the difference in protein-binding specificity observed between recombinant and endogenous proteins, we reasoned that the altered interactions could be modulated by post-translational modifications (PTMs). Indeed, several PTMs have been shown to regulate Mef2D function in vivo (Potthoff and Olson 2007). Important to transcriptional activation, phosphorylation of Mef2D by p38α MAPK (p38α) at positions Thr308 (T308) and T315 (see Fig. 4B) is required for recruitment of the Ash2L coactivator complexes to muscle genes (Rampalli et al. 2007). We therefore examined the propensity of Mef2Dα1 and Mef2Dα2 to be phosphorylated by p38α. Using recombinant proteins, in vitro kinase assays demonstrate that both Mef2Dα1 and Mef2Dα2 are efficiently phosphorylated by p38α (Fig. 4C). Having observed that the activating phosphorylation of Mef2D by p38α occurs in both Mef2Dα1 and Mef2Dα2, we next explored the possibility that the isoforms may be differentially affected by inhibitory PTMs. Transactivation by Mef2D is known to be inhibited through phosphorylation of Ser119 (S119) and S190 by PKA, which enables an association with class II HDACs (Du et al. 2008). Importantly, alternative exon usage in Mef2Dα2 removes the S119 residue targeted by PKA in the Mef2Dα1 isoform (see Fig. 4B). Thus, we examined whether the Mef2Dα1 and Mef2Dα2 isoforms are differentially phosphorylated by PKA. Consistent with previous observations (Du et al. 2008), Mef2Dα1 is strongly phosphorylated by PKA (Fig. 4C). In contrast, Mef2Dα2 is resistant to phosphorylation by PKA even though the phosphorylation target site S189(190) remains intact (Fig. 4C). These findings confirm the hypothesis that Mef2D isoforms are differentially modified by PTMs. Thus, overall, our results show that alternative exon usage regulates the ability of Mef2D protein to be phosphorylated by PKA.

Mutation of PKA phosphorylation sites allows Mef2Dα1 to activate myogenic transcription

To better understand the importance of the differential phosphorylation of Mef2D isoforms by PKA, we examined how PKA activity changes during differentiation. Previous studies have shown that PKA signaling is required at various stages of myogenesis to mediate CREB-dependent gene expression (Berdeaux and Stewart 2012). Using antibodies that recognize CREB phosphorylated at S133 to measure PKA activity, we observed that the PKA phosphorylates CREB in proliferating myoblasts, and the kinase remains active as myoblasts transition to myotubes (Fig. 4D). Thus, consistent with the requirement for activated CREB in myogenesis (Berdeaux and Stewart 2012), PKA activity is maintained throughout myotube formation.

The fact that PKA activity is maintained throughout differentiation suggests that the differential propensity to be phosphorylated by PKA could be responsible for the differential interaction of Mef2D isoforms with corepressor versus coactivators. To test this possibility, we generated the mutant Mef2Dα1(S119A/S190A), which is resistant to phosphorylated by PKA (Du et al. 2008). Similar to results observed with the endogenous protein, overexpressed Mef2Dα1 did not interact with Ash2L but showed a strong interaction with HDAC9 (Fig. 5A). However, inhibition of phosphorylation by PKA using Mef2Dα1(S119/S190A) led to a marked reduction in the interaction between the mutant protein and the repressive HDAC9 protein. Surprisingly, the loss in interaction with HDAC9 is concurrent with the appearance of an association with the coactivator Ash2L that normally interacts with Mef2Dα2. Thus, phosphorylation of Mef2D proteins by PKA is a primary determinant of the interaction with transcriptional coactivators or corepressors.

Finally, we examined whether the switch in Mef2D isoforms could represent a requirement for the transcription factor to evade inhibitory PKA signaling to mediate activation of late gene transcription. Using RT-qPCR, we observed that, unlike wild-type Mef2Dα1, the PKA phosphorylation mutant Mef2Dα1(S119/S190A) gives rise to a strong activation of both Myh3 and Mylpf expression (Fig. 5B). Consistent with the differential ability to interact with coregulatory proteins, ChIP analysis showed that expression of Mef2Dα2 or Mef2Dα1(S119/S190A) leads to an enrichment of Ash2L at both the Mylpf and Myh3 gene promoters, while expression of Mef2Dα1 results in an increased enrichment of HDAC9 at these same loci (Fig. 5C). These findings demonstrate that Mef2Dα1(S119/S190A) and Mef2Dα2 both activate expression of late muscle genes through the recruitment of coactivator proteins led us to examine whether the inability of the ubiquitously expressed Mef2Dα1 protein to stimulate myotube formation may be due to PKA phosphorylation. As observed above, immunofluorescence shows that myoblasts expressing Mef2Dα1 are resistant to the formation of MHC⁺ myotubes at 48 h of differentiation (Fig. 5D). In contrast, expression of the PKA mutant Mef2Dα1(S119/S190) gave rise to MHC⁺ myotubes with a level of differentiation similar to that observed with Mef2Dα2 (Fig. 5D; Supplemental Fig. 12). This demonstrates that the differential ability of Mef2Dα1 and Mef2Dα2 to drive myogenesis depends on their propensity for PKA-mediated phosphorylation. Thus, our results reveal a new role for alternative use of mutually exclusive exons in generating distinct transcription factors that have antagonizing roles in regulating differentiation based on their susceptibility to cell signaling-induced PTMs.

Antibodies

Rabbit polyclonal antibodies that specifically recognize the alternate isoforms of Mef2D were generated by Genscript using keyhole limpet hemocyanin-coupled peptides—CELDGLFRRYGST (for Mef2Dα1) or CPQTEEKYKKIDEEF (for Mef2Dα2). The pan-Mef2D mouse monoclonal antibody was purchased from BD Biosciences (catalog no. 610774). Ash2L antibody has been previously described (Demers et al. 2007). The DPY-30 antibody was generated in a rabbit using a His-tagged full-length mouse DPY-30 protein as antigen. The following antibodies recognizing different forms of MHC were obtained from the Developmental Studies Hybridoma Bank: embryonic MHC (MF20), MHC-I (A4.840), MHC-IIa (SC-71), and MHC-IIb (BF-F3). CREB (Santa Cruz Biotechnology, catalog no. SC-186), phospho-CREB (Ser133) (Cell Signaling, catalog no. 9191), HDAC4 (Cell Signaling, catalog no. 2072), HDAC9 (Abcam, catalog no. 109446), tubulin (Developmental Studies Hybridoma Bank, catalog no. 6G7), and myogenin (Santa Cruz Biotechnology, catalog no. sc12732) antibodies are commercially available.

Cell culture

C2 myoblasts (Yaffe and Saxel 1977) were maintained in Dulbecco's modified medium (4.5 g/L glucose) supplemented with 10% fetal calf serum, 4 mM L-glutamine, and 1.5 g/L sodium bicarbonate while not allowing cell density to achieve >80% confluent. To differentiate myoblasts, cells were allowed to achieve 90% confluency prior to a shift in culture medium to Dulbecco's modified medium containing 2% horse serum with insulin (10 μg/mL) and transferrin (10 μg/mL). Differentiation medium was changed every 24 h for up to 5 d.

Generation of isogenic C2 lines expressing Mef2D isoforms

A C2 line containing a single insertion of an FRT recombination element was generated using the Flp-In system (Invitrogen). The C2iFRT cell line was then generated to constitutively express the Tet repressor by selecting for cells transfected with the pPyCAGIP-TetR plasmid. Clones were then screened by immunofluorescence to ensure that they retained the potential to differentiate (express Myog and Myh3 while forming multinucleated myotubes) upon serum withdrawal. The cDNAs encoding Mef2Dα1, Mef2Dα2, Mef2Dα2(S189A), Mef2Dα1(S190A), or Mef2Dα1(S119/190A) were cloned into the multiple cloning site of a modified pCDNA5/FRT plasmid, where the CMV promoter was replaced by a CMV containing a Tet operator. Plasmids were transfected separately into C2iFRT cells and selected for hygromycin expression as outlined in the Flp-In system protocol. Batch cultures were then screened for their ability to differentiate (as described above) and Mef2D expression. Batch cell lines were termed C2iFRT-Mef2Dα1, C2iFRT-Mef2Dα2, C2iFRT-Mef2Dα2(S189A), C2iFRT-Mef2Dα1(S190A), or C2iFRT-Mef2Dα1(S119/190A) based on the form of Mef2D that was exogenously expressed when induced with Dox at a concentration of 0.5 μM.

ChIPs

C2 myoblasts (7 × 10⁷ cells) were differentiated for 5 d. Proteins were cross-linked to DNA for 30 min using 1% formaldehyde in Dulbecco's modified Eagle medium (DMEM) at 22°C. After 30 min, the formaldehyde-containing medium was removed, and cells were washed with 125 mM glycine in PBS for 10 min at 22°C. Recovered cells were resuspended in a hypotonic lysis buffer (25 mM Hepes at pH 7.9, 1.5 mM MgCl₂, 10 mM KCl, 0.1% NP-40, 0.5 mM DTT, 0.3 mM PMSF). After 45 min on ice, myotubes were lysed using a 2-mL dounce homogenizer (B-pestle). Pellets recovered from homogenate were resuspended in nuclease buffer (50 mM Tris HCl at pH 7.5, 140 mM NaCl, 5mM MgCl₂, 1 mM CaCl₂). MNase (Sigma) was added to the chromatin at a concentration of 0.12 U per 100 μg of ribonucleic acids (measured as in Brand et al. 2008) and incubated for 5 min at 37°C. Fragmentation of the DNA was stopped by the addition of one-tenth volume of MNase stop buffer (11 mM EDTA, 11 mM EGTA). The chromatin was then pelleted at 4000 rpm for 10 min at 4°C. The pellet was resuspended in shearing buffer (50mM Tris HCl at pH 7.5, 140 mM NaCl, 1 mM EDTA, 0.5% SDS, 1% Triton X-100) and sonicated using a Bioruptor for 80 cycles of 30 sec on and 60 sec off to further fragment the chromatin to obtain fragments of 150–300 base pairs (bp). After sonication, the chromatin was then centrifuged to pellet debris, and the supernatant was transferred to fresh Eppendorf tubes for immunoprecipitation with antibodies of interest as previously described (Seenundun et al. 2010). Libraries were generated from immunoprecipitated DNA as outlined in the Illumina DNA Library Construction kit protocol, except that 21 rounds of PCR were performed for amplification, and DNA fragments ranging from 200 to 350 bp were isolated during size selection on agarose gel. DNA libraries were sequenced on an Illumina Hi-Seq 2000 at the McGill University and Génome Québec Innovation Center. The DNA sequences of primer sets used for ChIP-qPCR analysis are available on request.

High-throughput DNA sequencing analysis

Sequences were extracted using GAPipeline. Reads were aligned by Burrows-Wheeler alignment (BWA) to the mouse genome (mm9). Duplicate sequences were discarded to minimize the effects of PCR amplification. Each read was extended in the sequencing orientation to a total of 200 bases to infer the coverage at each genomic position. Peak calling was performed by an in-house-developed R package as previously described (Palii et al. 2011; Fong et al. 2012). Motif analyses were performed using in-house-developed Bioconductor package motifRG (Palii et al. 2011; Fong et al. 2012). To detect differential peaks between two samples, we first took the square root of the peak heights in both samples and fit a local regression curve on the difference of the two versus the mean. The residual is defined as the difference subtracted by the fitted value, which adjusts for mean-dependent bias. We fit another local regression model on squared residual versus mean to estimate mean-dependent variance. The Z-value is defined by the residual divided by standard deviation (square root of the estimated variance). This approach draws some similarity to the mean-dependent dispersion estimate by DESEQ (Anders and Huber 2010), except that we used a nonparametric approach instead of modeling by a negative binomial distribution. ChIP-seq data were deposited in Gene Expression Omnibus (GEO; accession no. GSE43223).

RNA-seq

C2iFRT-Mef2Dα1, C2iFRT-Mef2Dα2, or control (C2iFRT) cells were grown in proliferative conditions for 24 h in the presence of Dox. Near confluence, cells were then differentiated for a further 24 h in the presence of Dox. Total RNA was isolated from cells using the RNeasy minikit (Qiagen) using standard conditions. Library preparation and 100-bp paired-end RNA-seq were performed using standard Illumina procedures for the HiSeq 2000 platform.

The RNA-seq data were analyzed using the Bowtie, TopHat, CuffLinks, and CummeRbund software suite (Trapnell et al. 2012). Briefly, paired-end 100-bp reads were mapped to the mm9 genome (University of California at Santa Cruz) and known mm9 transcripts (RefSeq mm9 build37.2) using Bowtie version 2.0.2 (Langmead and Salzberg 2012) with default parameters. Splice junctions were determined from the mapped reads using TopHat version 2.0.6 (Trapnell et al. 2009) with default values with the exception of −r = 0. Transcript assembly and quantification was performed with CuffLinks version 2.0.2 (Trapnell et al. 2010) using default values. Identification of differential gene expression was performed with the CummeRbund bioconductor package (http://compbio.mit.edu/cummeRbund). Genes with a Q-value ≤0.05 were considered to be significant. RNA-seq data were deposited in GEO (accession no. GSE46363).

Regeneration in mice

Mouse studies were performed in accordance with the University of Ottawa regulations for animal care and handling. The left and right TA muscles of wild-type CD-1 mice (obtained from Charles River) were injured by injection of cardiotoxin as previously described (Liu et al. 2012). After 48 h of recovery, retrovirus expressing either Mef2Dα1 or Mef2Dα2 (multiplicity of infection [MOI] 1 × 10⁶) was injected into one of the regenerating TA muscles, while the contralateral regenerating muscle was injected with a control retrovirus. After 3, 7, or 15 d of recovery, TA muscles were isolated and sectioned (12 μm thickness). For TA samples taken at 15 d of regeneration, the muscles were prepared for paraffin inclusion as described in Liu et al. (2012). For TA samples prepared at 3 and 7 d of regeneration, unfixed muscles were embedded in OCT compound, frozen in isopentane cooled with liquid nitrogen, and then sectioned (12 μm thickness). To assess regeneration, cross-sections were prepared for histological examination on a ScanScope CS microscope (Aperio) after staining with hematoxylin and eosin. Changes in fiber size distribution between muscles injected with Mef2D-expressing virus and the control were determined using a nonparametric (Wilcoxon) test. Fiber type switching was assessed by immunofluorescence using antibodies recognizing different isoforms of MHC. After immunostaining, coverslips were mounted in mounting medium with DAPI (Vector Laboratories, Inc., H-1500) and imaged using a Zeiss LSM 510 Meta confocal microscope.

Protein–protein interaction analysis

Nuclear extracts were prepared from differentiated C2 myotubes (5 d) as previously described (Weil et al. 1979) with the following modifications. Nuclear pellets were resuspended in one pellet volume of modified buffer C (20 mM Hepes at pH 7.6, 1.5 mM MgCl₂, 650 mM KCl, benzonase [2.5 U per 10⁷ cells], 0.2 mM PMSF, 0.5 mM DTT, 5 mM β-glycerolphosphate, 1 mM sodium orthovanadate) and rotated for 30 min at 4°C. Nuclear homogenates were then diluted with one pellet volume of buffer E (20 mM Hepes at pH 7.6, 1.5 mM MgCl₂, 0.2 mM EDTA, 1 mM PMSF, 0.5 mM DTT, 5 mM β-glycerolphosphate, 1 mM sodium orthovanadate). Extracted proteins were recovered from the nuclei by centrifugation at 15,000g for 30 min at 4°C. Nuclear extracts were then quantitated for protein content prior to interaction studies.

For immunoprecipitation studies of endogenous proteins, 10 μg of specific antibody or rabbit IgG were prebound to 20 μL of protein A Dynabeads. Flag immunoprecipitations were performed using M2-agarose (Sigma). Mef2D-associated proteins were immunoprecipitated as previously described (Chaturvedi et al. 2012) and eluted by incubating with 5% acetic acid for 5 min at 22°C. The acetic acid was removed from the recovered proteins by evaporation using a SpeedVac (Eppendorf) before proceeding with Western blot analysis.

In the pull-down studies, baculovirus-expressed Flag-Mef2Dα1 or Flag-Mef2Dα2 proteins were purified from Sf9 insect cells using M2-agarose (Sigma) as previously described (Rampalli et al. 2007). Briefly, M2-agarose-bound proteins were washed in stringent conditions (500 mM KCl) and then equilibrated in immunoprecipitation buffer (20 mM Hepes at pH 7.6, 150 mM KCl, 1.5 mM MgCl₂, 0.2 mM EDTA, 0.2 mM PMSF, 0.5 mM DTT). Flag-Mef2D-bound M2 agarose beads were then incubated with nuclear extracts for 4 h at 4°C. After washing, proteins were eluted with 3X-Flag peptide (Sigma) and separated by SDS-PAGE. After visualization with Coomassie blue, protein bands were excised for mass spectrometry analysis.

Mass spectrometry

Protein identification was performed at the Ottawa Hospital Research Institute Proteomics Core Facility (Ottawa, Canada). Gel bands were in-gel-digested according to the method of Shevchenko et al. (2006). Peptide extracts were concentrated by SpeedVac (Eppendorf) and resuspended in 0.1% trifluoroacetic acid. Peptides were loaded onto a peptide trap (Zorbax SB300-C18, Agilent) for 5 min at 15 μL/min using a Dionex UltiMate 3000 RSLC nano high-pressure liquid chromatography (HPLC). Peptides were eluted over a 60-min gradient of 3%–45% acetonitrile with 0.1% formic acid at 0.3 μL/min onto a 10-cm analytical column (New Objective Picofrit self-packed with Zorbax C18) and sprayed directly into a LTQ Orbitrap XL hybrid mass spectrometer using a nanospray source (Thermo Scientific). Mass spectra were acquired in a data-dependent fashion, with MS scans acquired in the Fourier Transform cell, while MS² scans were acquired in the ion trap module.

Tandem mass spectrometry (MS/MS) spectra were matched against a custom database comprised of mouse sequences in SwissProt (2011_07 version of uniprot_sprot.fasta.gz from ftp://ftp.uniprot.org) concatenated with a database of common contaminants or against the full SwissProt database using MASCOT 2.3.01 software (Matrix Science) with mass spectrometry tolerance of ±5 parts per million (ppm) and MS/MS tolerance of 0.6 Da. For standard Mascot search, oxidation of methionine, carbamidomethylation of cysteine, deamidation, protein N-terminal acetylation, and conversion of peptide N-terminal Glu or Gln to Pyro-Glu were allowed as potential protein modifications.

Immunofluorescence

Cells were fixed using 4% paraformaldehyde, permeabilized with 0.5% Triton X-100/PBS, and then blocked with 5% BSA in PBS. Primary antibodies were diluted with PBS/5% BSA solution and incubated overnight at 4°C. After washing, cells were incubated with secondary antibodies conjugated with Alexa Fluor 488 (Invitrogen, A11001) and/or Alexa Fluor 546 (Invitrogen, A11010). Prior to visualization, coverslips were mounted in mounting medium with DAPI (Vector Laboratories, Inc., H-1500). Images were taken using an Olympus FluoView FV1000 confocal microscope.

Kinase assay

Baculovirus-expressed Flag-Mef2Dα1 or Flag-Mef2Dα2 proteins were purified from Sf9 insect cells using M2-agarose (Sigma). His-p38α was purified from bacteria in its active (phosphorylated) form as previously described (Khokhlatchev et al. 1997). Active PKA was purchased from New England Biolabs. Recombinant Mef2D proteins (100 ng) were incubated with either His-p38α (80, 250, or 750 fmol) or PKA (50, 100, or 200 U) in the presence of 1 pmol (4 μCi) of [³²P]γ−ATP in a buffer containing 10 mM Hepes (pH 7.4), 50 mM KCl, 1.5 mM MgCl2, 10% glycerol, 0.5 mM EGTA, and 10 mM β-glycerophosphate for 30 min at 37°C. Reactions were then separated on 10% SDS-PAGE gels, dried, and exposed to autoradiographic film. Parallel reactions lacking [³²P]γ−ATP were run on 10% SDS-PAGE gels and stained for protein content with Coomassie blue.