MEF2A regulates the Gtl2-Dio3 microRNA mega-cluster to modulate WNT signaling in skeletal muscle regeneration

Skeletal muscle has the ability to regenerate myofibers upon damage or disease. The muscle repair process involves activation of quiescent muscle stem cells, called satellite cells, which proliferate and subsequently enter the differentiation pathway leading to the formation of multi-nucleated myotubes (Shi and Garry, 2006; Kuang and Rudnicki, 2008; Tedesco et al., 2010). Genetic dissection of this process has revealed that developmental pathways required for embryonic myogenesis also regulate muscle regeneration (Parker et al., 2003; Chargé and Rudnicki, 2004; Tajbakhsh, 2009). For example, the WNT signaling transduction cascade provides positional cues that establish myogenic identity within the developing somite (Buckingham, 2006). This pathway is later deployed to regulate the differentiation of activated satellite cells in response to injury (Polesskaya et al., 2003; Brack et al., 2008). Likewise, the paired homeobox transcription factor PAX7 plays a crucial role in progenitor cells in embryonic myogenesis and early postnatal skeletal muscle regeneration (Kuang et al., 2006; Buckingham and Relaix, 2007; Lepper et al., 2009).

Myocyte enhancer factor 2 (MEF2), an evolutionarily conserved transcription factor, is required for the differentiation of all three muscle lineages in flies (Bour et al., 1995; Lilly et al., 1995). In mammals, which have four Mef2 genes, an unequivocal role for MEF2 in muscle differentiation in vivo has yet to be established. The lack of an overt muscle differentiation defect is likely to be due to functionally redundant MEF2-dependent gene regulatory mechanisms (Black and Olson, 1998; Potthoff and Olson, 2007). Nevertheless, MEF2 loss-of-function analysis in mice has revealed distinct roles for individual family members in skeletal muscle. Developing myoblasts in mice lacking MEF2C are able to differentiate, but the resulting skeletal myofibers display disorganized cytoarchitecture (Potthoff et al., 2007a). MEF2C and MEF2D are also required for the formation of specialized skeletal muscle fiber types (Potthoff et al., 2007b). Despite the finding that inactivating mutations in some MEF2 family members in mice result in specific perturbations in skeletal muscle, a detailed understanding of MEF2 function in this tissue is far from complete.

MicroRNAs (miRNAs) are small, non-coding RNAs which function post-transcriptionally by interacting directly with mRNAs to repress their expression (Bartel, 2004). All cell types express miRNAs but a subset exhibit restricted expression, such as the muscle-specific miRNAs miR-1 and miR-133 (Lagos-Quintana et al., 2002; Chen et al., 2006; Liu et al., 2007). These and other more broadly expressed miRNAs have been shown to play important modulatory roles in a variety of skeletal muscle processes (van Rooij et al., 2008; Chen et al., 2009). However, only a limited number of miRNAs have been linked to the regulation of skeletal muscle regeneration. miR-1 and miR-206 have been shown to control PAX7 levels in satellite cell proliferation and differentiation (Chen et al., 2010), whereas miR-27b regulates PAX3 expression in this process (Crist et al., 2009). As the balance between proliferation and differentiation of progenitor cells in muscle regeneration must be carefully orchestrated, it is likely that additional miRNAs exist that modulate this process.

In the present study we explored the function of MEF2A in skeletal muscle. We report that, contrary to skeletal muscle development, MEF2A is required for adult myogenesis in response to injury. Skeletal muscle injury in adult Mef2a knockout mice resulted in widespread necrosis and attenuated myofiber regeneration accompanied by a significant reduction in PAX7-positive (PAX7+) nuclei. This impaired injury response was associated with a downregulation of the largest known mammalian miRNA cluster, the Gtl2-Dio3 locus, harboring >40 miRNAs. A cohort of these miRNAs directly targets secreted Frizzled-related protein 2 (sFRP2), an inhibitor of the WNT signaling pathway. Accordingly, we detected an upregulation of sFRP2 in regenerating Mef2a knockout skeletal muscle along with reduced WNT activity. Finally, overexpression of miRNAs that directly repress sFRP2 or addition of recombinant WNTs was able to rescue myogenic differentiation in Mef2a-deficient myoblasts.

C2C12 and COS cells were cultured as described previously (McCalmon et al., 2010). For differentiation assays, C2C12 cells were switched to DMEM supplemented with 2% horse serum (differentiation media) 24 hours post-transfection/transduction. Primary skeletal myoblasts and single myofibers were isolated and cultured from uninjured or cardiotoxin-injured adult mouse hindlimb muscle as described by Springer et al. (Springer et al., 1997).

MEF2A-FLAG and MEF2C-FLAG plasmids were generated by cloning full-length mouse cDNAs into pCMV-tag4 (Invitrogen). MEF2B-FLAG and MEF2D-FLAG (human) were kind gifts of T. Gulick (Sanford Burnham Medical Research Institute, Orlando, FL, USA). The mouse Gtl2 promoter (0.5 kb) containing the MEF2 binding site was cloned into pGL3-Basic (Promega). The mutant Gtl2 reporter was generated by mutating the –39 MEF2 site CTT to GGG, without altering the overlapping TATA box (–39MUT). For 3′UTR reporter assays, the 3′UTR (892 bp) of mouse sFRP2 was cloned into pMIR-REPORT (Ambion). The mutant 3′UTR-sFRP2 construct was generated by mutating the miR-410 seed sequence binding site TTATAT to GGGGGG. pMIR-REPORT-β-galactosidase (Ambion), pCMV-Renilla-luciferase and pTOPFLASH (Addgene) were also used for luciferase assays. MiRNA mimics were purchased from Ambion.

Adenoviruses carrying shRNAs specific for Mef2a or lacZ were generated as described previously (Ewen et al., 2011). Adenoviruses were used at a multiplicity of infection (MOI) of 25 for all assays.

Tibialis anterior (TA) muscle of wild-type (WT) and Mef2a knockout (KO) mice (Naya et al., 2002) were injected with 10 μM cardiotoxin (Naja nigricollis, EMD chemicals) and harvested several days post-injury. Transverse muscle sections were stained with Hematoxylin and Eosin for visualization of basic muscle morphology.

Phase contrast images of C2C12 DIFF 3 cells (post-transduction) were taken using an Olympus MX50 microscope. For immunofluorescence, primary antibodies included: anti-α-actinin (1:200; Sigma), anti-Pax7 [1:2; supernatant, Developmental Studies Hybridoma Bank (DSHB)], anti-MEF2 (1:200; Santa Cruz Biotechnology) and anti-MyoD (1:200 S-17, 1:100 M-318; Santa Cruz Biotechnology). Secondary antibodies were: anti-mouse Alexa Fluor 488, 568 IgG1 and 568 IgG (H+L) (1:300); anti-rabbit Alexa Fluor 488 and 568 (1:300); and anti-goat Alexa Fluor 647 (1:100) (Invitrogen). Terminal dUTP nick-end labeling (TUNEL) assays were performed using the Promega DeadEnd Fluorometric TUNEL System Kit. Vectashield mounting medium with DAPI was applied to all slides. Fluorescent images were taken using an Olympus DSU Spinning Disk Confocal microscope.

Transverse TA muscle cryosections (10 μm) were immunostained as described in Kanisicak et al. (Kanisicak et al., 2009). Primary antibodies were: anti-laminin (1:200; Sigma), anti-β-catenin (1:200, BD Transduction Labs), anti-Pax7 and anti-MEF2. Secondary antibodies were: anti-mouse Alexa-Fluor 488 IgG (H+L), anti-mouse Alexa-Fluor 568 IgG₁, and fluorescein anti-rabbit IgG (H+L) (1:500; Vector Labs). Cross-sectional area (CSA) of skeletal muscle was assessed by Metamorph (Molecular Devices, Downington, PA, USA) analysis of laminin-stained muscle sections.

Total RNA from WT (n=5) and Mef2a KO (n=5) TA muscle 7 days post-injury was prepared by TRIZOL isolation (Invitrogen), and were hybridized to the Mouse Gene 1.0 ST Array (Affymetrix) at the Boston University Microarray Facility. Microarray data are available in GEO with Accession Number GSE41871.

Pooled RNA from TA muscle or C2C12 MEF2A knockdown experiments (n=6) was used to synthesize cDNA using reverse transcriptase (M-MLV) with random hexamers according to the manufacturer’s instructions (Promega). cDNAs were synthesized using the TaqMan miRNA Reverse Transcriptase Kit (Applied Biosystems) for detection of mature miRNAs in vivo as described by Chen et al. (Chen et al., 2005). miRNA and 5S sequences were amplified using forward-specific primers and a universal reverse primer (Thompson et al., 2011). Quantitative RT-PCR was performed in triplicate wells using Power SYBR Green Master Mix (Applied Biosystems) with the 7900HT Sequence Detection System (Applied Biosystems). Primers used for all RT-PCR and qRT-PCR/stem-loop qRT-PCR analyses are listed in supplementary material Table S1.

Luciferase assays were performed using Luciferase Assay Reagent (LAR, Promega), and results were normalized by β-galactosidase assay (Miller, 1972). miRNA mimics and expression plasmids were transfected using the RNAiMAX transfection reagent (Invitrogen). TOPFLASH assays were performed on C2C12 cells at DIFF 3 using LARII (Promega) and normalized by Renilla luciferase assay (Promega). All luciferase assays were performed in triplicate (n>3).

Western blots were performed as previously described (McCalmon et al., 2010). Antibodies included: anti-GAPDH (1:5000; Santa Cruz), anti-MEF2A (1:1000), anti-cleaved-caspase-3 (1:1000; Cell Signaling), anti-FLAG (1:10,000; Sigma), anti-sFRP2 (1:500; Millipore), anti-MF 20 (1:50; supernatant, DSHB) and anti-β-catenin (1:1000). Blots were incubated with horseradish peroxidase-conjugated secondary antibodies (1:10,000; Sigma) and reacted with Western Lightning Chemiluminescent Reagent (Perkin Elmer).

In vitro-translated mouse MEF2A (rabbit reticulocyte lysate; Promega) or nuclear extracts from C2C12 myotubes were used for gel shift assays. The mutant MEF2 site was generated in the same manner as the mutant Gtl2 reporter. Supershift assays were performed with anti-MEF2 antibodies (C-21, Santa Cruz). Competitions were performed with 100-fold molar excess of unlabeled probe. Gel shift reactions were fractionated on 5% non-denaturing polyacrylamide gels, dried, and exposed to a phosphor-imaging screen (Amersham Biosciences).

Chromatin was immunoprecipitated from C2C12 cells (DIFF 3) transfected with mouse MEF2A (pcDNA3-MEF2A-FLAG) using the anti-FLAG (Sigma) antibody, or negative control anti-HA antibody (Sigma). Chromatin was subjected to qRT-PCR for the detection of the Gtl2 promoter sequence containing the –39 MEF2 site. Primer sequences are listed in supplementary material Table S1.

For rescue experiments, C2C12 myoblasts were transduced with adenoviruses (shlacZ or shMef2a) at MOI 25 and transfected with 40 nM miRNA mimics (miR negative control #1, miR-410, miR-433, or an equimolar combination) 24 hours post-transduction. For miR knockdown experiments, C2C12 myoblasts were transfected with 100 pmol anti-miR-410 or anti-miR-433 (Ambion) using RNAi Max (Invitrogen). Cells were then differentiated in culture for three days prior to imaging or RNA/protein analysis.

C2C12 myoblasts were transduced with adenoviruses (shlacZ or shMef2a) at MOI 25 and recombinant WNTs were added 24 hours post-transduction. recombinant mouse WNT3A (rmWnt3a) was used at 100 ng/ml and recombinant human/mouse WNT5A (rh/mWnt5a) was used at 200 ng/ml (R&D Systems). For combinatorial rescue, WNT concentrations were halved. Cells were imaged on differentiation day three.

All numerical quantification is representative of the mean±s.e.m. of at least three independently performed experiments. Statistically significant differences between two populations of data were determined using Student’s t-test. P-values of ≤0.05 were considered to be statistically significant.

To dissect the role of MEF2A in skeletal muscle differentiation, C2C12 cells were transduced at various time points with adenovirus harboring a Mef2a-specific shRNA (shMef2a), which effectively knocked down MEF2A expression (Ewen et al., 2011) (supplementary material Fig. S1A). As shown in Fig. 1A, knockdown of MEF2A in proliferating C2C12 myoblasts (DIFF –1) resulted in severely compromised myotube formation by differentiation day three (DIFF 3). Knockdown of MEF2A in differentiated C2C12 cells (DIFF 3 and DIFF 7) also caused substantial loss of myotubes (Fig. 1A). Immunofluorescence analysis of Mef2a-deficient C2C12 cells at day three revealed a 46% reduction in differentiation (Fig. 1B) and a 49% decrease in myoblast fusion (Fig. 1B). The reduction in myotube number resulted, in part, from programmed cell death as shown by a 1.5-fold increase in TUNEL-positive cells (Fig. 1C) and increased expression of activated caspase-3 (Fig. 1D).

Collectively, the results of the C2C12 knockdown experiments revealed that MEF2A is essential for myogenic differentiation. This differentiation defect is in stark contrast to Mef2a knockout (KO) mice which do not display developmental skeletal muscle abnormalities (Potthoff et al., 2007b). The phenotypic differences may be explained, in part, by the radically different Mef2 temporal expression in C2C12 cells compared with that observed during muscle development (Edmondson et al., 1994). Contrary to embryonic myogenesis, in which Mef2c is expressed first, in C2C12 cells Mef2a is expressed earlier than Mef2c and in proliferating myoblasts (Mbs) (Fig. 1E). Expression of Mef2a is subsequently upregulated upon induction of myogenic differentiation and this higher expression is maintained throughout myotube differentiation and maturation. In C2C12 cells, Mef2d and Mef2c are expressed one and two days after Mef2a, respectively. Mef2b is also expressed in proliferating myoblasts but its levels remain constant throughout differentiation. As C2C12 myoblasts were originally isolated from injured muscle (Yaffe and Saxel, 1977; Blau et al., 1983), these results suggest that the C2C12 differentiation process might reflect muscle regeneration. To test this notion, we isolated primary myoblasts from injured muscle and analyzed Mef2 temporal expression. Indeed, the pattern of Mef2 expression in proliferating and differentiated primary myoblasts was similar to that observed in C2C12 cells (Fig. 1E; supplementary material Fig. S1D and Fig. S2). Thus, we postulate that Mef2 regulation is reprogrammed during regeneration and that the pronounced defect in C2C12 differentiation reflects a specialized function for MEF2A in skeletal muscle regeneration.

To determine whether MEF2A plays a role in skeletal muscle regeneration, we induced muscle injury in adult Mef2a KO mice (129Sv/C57Bl6 mixed genetic background) by injecting the tibialis anterior (TA) muscle with cardiotoxin. TA muscle from Mef2a KO mice displayed widespread necrosis at various time points after injury compared with wild-type (WT) animals (Fig. 2A). Transverse muscle sections revealed a similar extent of damage in both WT and KO tissue at day 3 (Fig. 2B). However, by day 7, unlike WT muscle, Mef2a KO muscle displayed an intense basophilic appearance, consistent with the presence of necrotic tissue (Fig. 2B). By day 14, necrosis was diminished in Mef2a KO muscle and these areas were replaced by regenerating myofibers (Fig. 2B). By day 21, abundant regenerating myofibers were clearly evident and necrosis was largely absent (supplementary material Fig. S1E), suggesting a delay, but not a block, in regeneration.

To characterize this impaired regenerative response in greater detail, cross-sectional area (CSA) of non-necrotic myofibers possessing centrally located nuclei, a hallmark of regenerating myofibers, was measured in TA muscle. Metamorph analysis at day 7 of TA muscle sections immunostained with anti-laminin revealed significantly reduced CSA of regenerating Mef2a KO tissue (Fig. 2C). By day 14, Mef2a KO muscle displayed a slight increase in CSA relative to earlier time points, but not to the extent observed in WT muscle (Fig. 2C), supporting the notion that muscle regeneration is delayed but not blocked in these mice.

As global Mef2a KO mice were used for this study, we wanted to determine next whether impaired muscle regeneration could be attributable to a specific defect in myoblasts. For this characterization, we isolated primary myoblasts from injured WT and KO muscle and induced them to differentiate in culture. As shown in Fig. 3A, Mef2a KO myoblasts were unable to differentiate effectively, as demonstrated by a 78% reduction in α-actinin staining. Moreover, there was no significant difference in cell proliferation and viability (data not shown) or in MyoD-positive myoblasts (Fig. 3A), demonstrating that Mef2a KO myoblasts are appropriately specified and have committed to the myogenic pathway.

The defect in myogenic differentiation in vitro prompted us to analyze injured Mef2a KO muscle in greater detail using the muscle satellite cell marker PAX7 as a perturbation in the PAX7+ progenitor population and/or its expression is one reliable indicator of myogenic impairment. At day 7 post-injury, there was a 50% reduction in the number of PAX7+ nuclei colocalized with DAPI in regenerating Mef2a KO muscle compared with regenerating WT muscle (Fig. 3B). However, uninjured Mef2a KO muscle displayed a similar number of quiescent PAX7+/DAPI nuclei compared with uninjured WT muscle (data not shown) indicating that the reduction in the PAX7+ progenitor population is only observed in a muscle injury setting. PAX7 transcripts were also significantly downregulated in both regenerating Mef2a KO muscle and in primary myoblasts isolated from these injured mice (Fig. 3C). These results indicate that impaired regeneration in Mef2a KO muscle is, in part, due to defective muscle differentiation and suggest that loss of MEF2A affects the ability of injured muscle to generate (or maintain) sufficient activated satellite cells for subsequent myotube formation.

To dissect the molecular mechanisms of this defective muscle regeneration, we isolated RNA from WT and Mef2a KO injured TA muscle and performed microarray analysis. The most striking result of this analysis was the downregulation of >40 microRNAs (miRNAs) (Fig. 4A; supplementary material Table S2). Further characterization revealed that these miRNAs all localize to a single locus on mouse chromosome 12 (Fig. 4B). This locus is known as the Gtl2-Dio3 domain and is the largest known miRNA cluster in the mammalian genome (Seitz et al., 2004; Kircher et al., 2008).

To confirm the microarray data, we examined the expression of coding and non-coding transcripts in the Gtl2-Dio3 locus. Expression of the upstream Dlk1 gene, which has been shown to play a role in skeletal muscle development and regeneration (Waddell et al., 2010), was unaffected in regenerating Mef2a KO muscle (Fig. 4C). By contrast, analysis of selected miRNA transcripts in the Gtl2-Dio3 locus, which were retrospectively chosen based on their common target mRNAs (see Fig. 5A), displayed significant downregulation in this tissue (Fig. 4C), in primary KO myoblasts (supplementary material Fig. S3E-I) and in Mef2a-deficient C2C12 cells (Fig. 4D). We also determined that these miRNAs are upregulated in WT primary myoblast and C2C12 differentiation (supplementary material Fig. S3D-I), paralleling the expression profile of MEF2A (supplementary material Fig. S3B,C). Furthermore, our tissue expression profiling revealed that, in addition to the brain (Seitz et al., 2004; Schratt et al., 2006), Gtl2-Dio3 miRNAs are enriched in skeletal muscle and the heart (supplementary material Fig. S3A). These data strongly suggest that MEF2A regulates the expression of Gtl2-Dio3 miRNAs.

Previous studies have shown that the miRNAs in the Gtl2-Dio3 cluster are coordinately misregulated in mouse models harboring mutations that affect expression of this locus (Lin et al., 2003; Zhou et al., 2010). Because all miRNAs in the Gtl2-Dio3 locus are downregulated in regenerating Mef2a KO muscle, we hypothesized that transcription of this cluster may be controlled by MEF2A through an upstream cis-acting sequence.

We identified a conserved MEF2 site located at position –39 upstream of the Gtl2 transcription start site, which overlaps the only defined TATA box in the entire locus (Zhou et al., 2010). In vitro gel shift assays showed that MEF2A binds to this MEF2 site, which is effectively competed by the unlabeled oligonucleotide, but is unable to bind to a mutant probe harboring a specific mutation in the MEF2 sequence (Fig. 4E). Nuclear extracts from C2C12 myotubes incubated with the –39 MEF2 site showed robust DNA binding, and this complex was supershifted when incubated with a MEF2A antibody (Fig. 4E). To demonstrate that MEF2A associates with the Gtl2 proximal promoter region in vivo, we subjected chromatin from C2C12 myotubes transfected with MEF2A-FLAG to chromatin immunoprecipitation assays. Incubation of these extracts with anti-FLAG showed highly enriched MEF2A binding to this region (Fig. 4F).

To determine whether the Gtl2 promoter is active in muscle, we compared the activity of the proximal promoter in non-muscle COS cells and C2C12 myoblasts. Transfection of the Gtl2 proximal promoter revealed a significantly higher activity in C2C12 myoblasts that was dependent on an intact MEF2 site (Fig. 4G). We next asked whether MEF2A could activate the Gtl2 proximal promoter harboring the MEF2 site. COS cells co-transfected with MEF2A and the Gtl2 promoter reporter resulted in a threefold higher level of activation compared with reporter alone (Fig. 4H). Mutation of the MEF2 site, without affecting the TATA element, resulted in a significant decrease in MEF2A-dependent transcriptional activation (Fig. 4H). The results above suggest that MEF2A directly regulates the Gtl2-Dio3 cluster from the Gtl2 proximal promoter region.

To understand the mechanism by which MEF2A regulates skeletal muscle regeneration through this miRNA mega-cluster we subjected all miRNA sequences within the cluster to miRNA prediction algorithms (Target Scan and MirANDA). The results of these analyses revealed that many of the miRNAs in the Gtl2-Dio3 mega-cluster are predicted to target secreted Frizzled-related proteins (sFRPs) 1, 2 and 4 (Fig. 5A), members of a well-known family of inhibitors in the WNT signaling pathway (Kawano and Kypta, 2003; MacDonald et al., 2009).

Given the results described above, we examined the expression levels of Sfrp1, Sfrp2 and Sfrp4 in regenerating WT and Mef2a KO muscle. We found that expression of Sfrp1, Sfrp2 and Sfrp4 was significantly upregulated (Fig. 5B) and that sFRP2 protein is upregulated in regenerating Mef2a KO muscle (Fig. 5C). We extended these results by examining the temporal expression of Sfrp2 and selected Gtl2-Dio3 miRNAs predicted to target Sfrp2 in regeneration. As shown in Fig. 5D, between days one and three post-injury, when muscle is degenerating, the Gtl2-Dio3 miRNAs were downregulated whereas Sfrp2 was upregulated (Fig. 5D). At these early time points, we did not find any significant difference in expression of either the miRNAs or Sfrp2 between WT and Mef2a KO muscle. Between days three and seven post-injury, when activated myoblasts begin forming nascent myofibers, miRNA expression increased in WT muscle approaching pre-injury levels but failed to be upregulated in injured Mef2a KO muscle (Fig. 5D). In a reciprocal fashion, the expression of Sfrp2 decreased to near pre-injury levels in WT muscle at these later time points whereas its expression was significantly upregulated in regenerating Mef2a KO muscle (Fig. 5D). These results reveal an inverse correlation between Gtl2-Dio3 miRNA levels and Sfrp2 expression in regenerating WT and KO muscle, which is coordinately dysregulated in Mef2a KO mice.

Because many of the miRNAs from the Gtl2-Dio3 mega-cluster are predicted to target Sfrp2, we focused on this gene to determine whether these miRNAs could directly repress its expression. We cloned the 3′UTR of murine Sfrp2 into the pMIR-REPORT vector (Ambion) and used this construct in transient transfection assays to examine the ability of selected miRNAs to repress this reporter. miRNA mimics (Ambion) were generated for miR-410 (located within Mirg) and miR-433 (located within anti-Rtl1) because the seed sequences of these miRNAs and their target sequences in Sfrp2 are absolutely conserved between mouse and human (Fig. 5E). Co-transfection of the Sfrp2 3′UTR with either miR-410 or miR-433 mimics resulted in a significant inhibition of the reporter; however, these miRNAs in combination did not inhibit the reporter further (Fig. 5F). Mutation of the miR-410 binding site reduced the ability of miR-410 mimic to repress the reporter (Fig. 5F). These results demonstrate that two miRNAs within the Gtl2-Dio3 mega-cluster are capable of directly inhibiting Sfrp2.

Based on the knowledge that sFRPs are inhibitors in the WNT signaling pathway, we would expect that upregulation of sFRP2 in regenerating Mef2a KO muscle inhibits WNT activity. To test this possibility, we initially examined Sfrp expression and WNT activity in C2C12 cells transduced with shMef2a. At differentiation day three, we observed a significant upregulation of Sfrp1, Sfrp2 and Sfrp4, consistent with our observations in regenerating Mef2a KO muscle (Fig. 6A). We then analyzed WNT activity by transfecting Mef2a-deficient C2C12 myoblasts with the WNT-sensitive reporter TOPFLASH (Addgene). Luciferase assays revealed an 80% decrease in TOPFLASH activity in shMef2a transduced cells (Fig. 6B). Supporting these observations, qRT-PCR analysis of endogenous Axin2, a WNT-responsive gene (Clevers, 2006), showed a significant 50% decrease in Mef2a-deficient C2C12 cells (Fig. 6C).

We next sought to assess WNT activity in regenerating Mef2a KO muscle. Translocation of β-catenin from the cytoplasm to the nucleus is one of the major indicators that the WNT signaling pathway has been activated (Clevers, 2006; MacDonald et al., 2009). Therefore, we performed immunostaining analysis using a pan β-catenin antibody and found a 75% decrease in β-catenin positive nuclei in Mef2a KO regenerating muscle (Fig. 6D) along with an apparent increase in cytoplasmic β-catenin immunoreactivity, suggesting targeting of this protein for proteasomal degradation. Supporting these observations, western blot analysis revealed a reduction in total β-catenin levels in regenerating Mef2a KO muscle (Fig. 6D). Moreover, in primary Mef2a KO myoblasts isolated from injured muscle Axin2 expression was reduced (Fig. 6E) and Sfrp2 was upregulated (Fig. 6F), further demonstrating inhibited WNT activity. Collectively, these results reveal that WNT activity is attenuated in Mef2a KO muscle in response to injury, suggesting that this perturbation contributes to impaired myogenic differentiation in this process.

To demonstrate that miRNAs within this mega-cluster are relevant downstream targets of MEF2A in myogenic differentiation, we transfected miRNA mimics into shMef2a-transduced C2C12 myoblasts. As shown in Fig. 7A, transfection of either miR-410, miR-433, or the two miRs combined was sufficient to rescue the myogenic differentiation defect as demonstrated by increased myotube formation and myosin heavy chain (MHC) expression. We then assessed rescue of WNT activity by way of Axin2 expression, and observed that Axin2 was restored to equal to or greater than control levels in the presence of the miRNA mimics individually or in combination (Fig. 7B). Furthermore, transfection of C2C12 myoblasts with inhibitors (anti-miRs) of miR-410 and miR-433 resulted in impaired myogenesis (Fig. 7C) supporting the notion that their downregulation plays a role in the differentiation defect in Mef2a-deficient myoblasts. Finally, if reduced WNT activity is also contributing to impaired myogenesis in Mef2a-deficient myoblasts then activating this pathway would be expected to properly restore differentiation. Treatment of Mef2a-deficient C2C12 myoblasts with recombinant WNT3A or WNT5A effectively rescued differentiation (Fig. 7D). These data demonstrate that the Gtl2-Dio3 miRNAs function downstream of MEF2A to modulate WNT signaling in myogenic differentiation.

The results of this study reveal an essential role for MEF2A in skeletal muscle regeneration. Our data demonstrate that regenerating myoblasts initiate a gene program that is distinct from that occurring during development and is characterized by the early expression of MEF2A. Accordingly, skeletal muscle lacking MEF2A is unable to regenerate properly in response to injury. We show that MEF2A functions in regenerative myogenesis by regulating transcription of the Gtl2-Dio3 miRNA mega-cluster and that several of these miRNAs target sFRP2, an inhibitor of WNT signaling. Consistent with these observations, injured Mef2a KO muscle exhibits increased sFRP expression and attenuated WNT activity. These results firmly establish a molecular link between two important developmental regulators, MEF2 and WNT, in the process of adult regenerative myogenesis. To our knowledge, this is the first evidence that MEF2 participates in the regulation of WNT signaling.

Previous studies on Mef2a KO mice concluded that this factor is largely dispensable in skeletal muscle development (Potthoff et al., 2007b). By contrast, the present study shows that MEF2A is essential for skeletal muscle regeneration. These observations suggest that MEF2A activity is regulated in a context-dependent manner, i.e. embryonic versus adult muscle. Alternatively, the skeletal muscle regeneration defect might reflect differences in the temporal expression of Mef2a. In embryonic muscle development, Mef2a is expressed after Mef2b and Mef2c (Edmondson et al., 1994). But in activated, primary and C2C12 myoblasts Mef2a is one of the earliest Mef2 genes expressed with transcripts detectable prior to Mef2c induction, which is consistent with this factor playing an essential role in myoblast differentiation (this study) (Seok et al., 2011). Moreover, upon injury, expression of Mef2a is upregulated (supplementary material Fig. S1C and Fig. S3B), whereas Mef2c is downregulated in regenerating muscle (Zhao and Hoffman, 2004). Although it remains to be determined whether other MEF2 family members function in skeletal muscle regeneration, our data suggest that the regulation and/or activity of MEF2A has been reprogrammed such that it plays a more prominent role in adult myogenesis. This notion is supported by a recent report demonstrating co-expression of MEF2 and MyoD in activated satellite cells in regenerating muscle (Mokalled et al., 2012).

Muscle injury in adult mice triggers the activation of quiescent PAX7+ satellite cells, which proliferate as muscle progenitors (myoblasts) and subsequently differentiate into myotubes (Zammit et al., 2004; Olguin and Olwin, 2004; Chargé and Rudnicki, 2004; Zammit, 2008). We noted a significant reduction in PAX7+ nuclei and expression in injured Mef2a KO muscle, suggesting that mutant muscle is unable to maintain or generate the appropriate number of activated progenitors resulting in impaired myogenesis with smaller, immature myofibers. These results suggest that MEF2A functions upstream of or in parallel to PAX7 in muscle progenitors, though a definitive answer will require detailed characterization of sorted myoblasts from mutant tissue. Alternatively, the depletion in PAX7+ nuclei could be linked to a downstream differentiation function of MEF2A in the muscle regeneration process. That is, regenerating nascent or mature myotubes lacking MEF2A are unable to properly communicate with the injured microenvironment thereby adversely affecting the muscle progenitor population. Based on the established role of MEF2 proteins in muscle differentiation this latter notion appears to be the most plausible interpretation of the PAX7 deficiency. Our findings that PAX7 expression is not significantly reduced in Mef2a KO muscle until day 3 post-injury or in mutant primary myoblasts until the onset of differentiation (day 1) support this hypothesis.

The delay in regeneration suggests that other factors or pathways are ultimately able to compensate for the loss of MEF2A. One obvious explanation is that the remaining MEF2 proteins are partially functionally redundant with MEF2A and are less effective regulators of MEF2A-dependent gene programs in the muscle regeneration process. Because we used a global Mef2a KO it is also possible that the lack of MEF2A in non-muscle cells, such as inflammatory cells, contributes to the regeneration defect and delay in myofiber formation. In particular, the observation that injured Mef2a KO muscle eventually recovers suggests that inefficient clearance of necrotic myofibers, which is associated with the immune response, is an integral part of the phenotype.

Our studies also reveal that the WNT signal transduction cascade is modulated by MEF2A in skeletal muscle regeneration. WNT signaling has been established as an important regulator in adult regenerative myogenesis yet the mechanisms by which this pathway is modulated remain incompletely understood. WNT signaling in regeneration is complex and has been reported to be required for satellite cell proliferation, self-renewal and differentiation (Polesskaya et al., 2003; Brack et al., 2008; Otto et al., 2008). Given the diverse roles of WNT, the activity of this pathway would need to be tightly controlled throughout adult myogenesis in order to maintain a balance between proliferating, activated progenitors and myoblasts committed to differentiation. We show that defective regenerative myogenesis in Mef2a KO muscle is associated with aberrant WNT signaling. The WNT reporter and target gene expression assays performed with Mef2a-deficient primary myoblasts and C2C12 cells reveal reduced WNT activity in differentiated myotubes and suggest that it impacts this myogenic process, resulting in impaired myofiber formation. This notion is reinforced by the observation that treatment of Mef2a-deficient C2C12 cells with recombinant WNTs rescued myotube differentiation. However, as stated above, the reduction in PAX7+ nuclei and PAX7 expression in vivo implies that the depletion of activated progenitors also contributes to diminished myofiber formation. Although it is presently unclear whether inadequate proliferation or reduced viability of these myogenic progenitors is the reason for the depletion, the reduction in this progenitor population can also be attributable to attenuated WNT activity. Conceivably, WNT signaling might be acting on both the maturation of myotubes and undifferentiated, proliferating myoblasts. Because WNT signaling in regenerative myogenesis has pleiotropic effects it is likely that various cellular decisions are affected in this mutant.

MEF2A functions in regeneration by controlling transcription of the Gtl2-Dio3 miRNA locus. The Gtl2-Dio3 domain is the largest known non-coding RNA cluster in mammals and contains >40 miRNAs (Seitz et al., 2004; Kircher et al., 2008). Previous studies have suggested that transcription of miRNA sub-clusters within the Gtl2-Dio3 locus is regulated by different promoters (Song and Wang, 2008; Fiore et al., 2009). Although the aforementioned studies suggest the presence of independent cis-acting sequences within the Gtl2-Dio3 locus, our data support the notion that the Gtl2-Dio3 mega-cluster is coordinately regulated by an upstream MEF2A-dependent promoter. This rationale is based on two observations. First, a mouse knockout of the Gtl2 proximal promoter, which harbors the MEF2 binding site, results in downregulation of miRNAs in the Gtl2-Dio3 cluster with associated developmental skeletal muscle defects (Zhou et al., 2010). Second, we show that the tissue expression pattern of multiple miRNAs distributed throughout the cluster is similar, and that they are all downregulated in regenerating Mef2a knockout muscle. Although it is possible that sub-clusters of miRNAs within the Gtl2-Dio3 domain are differentially regulated by independent cis-acting sequences, the most logical explanation for the widespread, coordinated downregulation of this cluster in our and other models is through a single upstream control element. It remains to be determined whether the Gtl2-Dio3 mega-cluster is transcribed as a single, large polycistronic primary RNA transcript.

Given the connection of the MEF2A/Gtl2-Dio3 miRNA pathway to regeneration, it is tempting to speculate that this genetic pathway also plays an important role in muscular dystrophies. It is known that many muscular dystrophies are characterized by continuous cycles of degeneration and regeneration, ultimately resulting in failure to form new myofibers because muscle progenitors have been exhausted in this process (Wallace and McNally, 2009). Recently, it was reported that expression and splicing of MEF2A is dysregulated in various muscular dystrophies (Bachinski et al., 2010). Perhaps these and other muscle dystrophies might be associated with perturbed Gtl2-Dio3 miRNA expression resulting from altered MEF2A regulation. Indeed, numerous miRNAs are dysregulated in several muscular dystrophies (Eisenberg et al., 2007) and we have found that some of these map to the Gtl2-Dio3 locus. Our findings point to the importance of interrogating further MEF2A function in muscle diseases involving impaired muscle regeneration, and represent an attractive area of investigation for degenerative muscle disorders.

We thank Ning Liu and Eric Olson (UT Southwestern) and Dazhi Wang (Children’s Hospital, Boston) for sharing unpublished miRNA data. We also thank David Goldhamer (University of Connecticut) for providing us with a Pax7 immunostaining protocol and Tod Gulick (Sanford Burnham Institute) for the MEF2-FLAG expression constructs. We are grateful to Kim McCall, Tom Gilmore and Isabel Dominguez at Boston University for providing us with additional reagents and advice.