The Mef2c gene is a direct transcriptional target of myogenic bHLH and MEF2 proteins during skeletal muscle development

The formation of skeletal muscle during embryogenesis involves a multi-step developmental program in which mesodermal progenitors become committed to a skeletal muscle fate and then propagate and migrate to specific destinations before differentiating to form myofibers (Hauschka, 1994). Skeletal muscle development is controlled by the MyoD and myocyte enhancer factor 2 (MEF2) families of transcription factors, which interact to establish a unique transcriptional code for activation of skeletal muscle-specific genes (Molkentin and Olson, 1996). Members of the MyoD family – MyoD, myogenin, Myf5 and MRF4 – are expressed exclusively in the skeletal muscle lineage and can each activate the complete muscle differentiation program in transfected fibroblasts (Davis et al., 1987; Braun et al., 1989; Edmondson and Olson, 1989; Rhodes and Konieczny, 1989). During embryogenesis, MyoD and Myf5 are expressed in distinct early populations of skeletal muscle precursor cells in the somites (Ott et al., 1991; Smith et al., 1994). Mice that lack either MyoD or Myf5 are able to form skeletal muscle (Rudnicki et al., 1992; Braun et al., 1992), whereas mice lacking both genes fail to form any trace of the skeletal muscle lineage (Rudnicki et al., 1993), consistent with the notion that these genes play overlapping roles in myoblast specification. Myogenin is expressed as myoblasts enter the differentiation pathway and is required for muscle differentiation (Hasty et al., 1993; Nabeshima et al., 1993). Mrf4 is expressed transiently during early myogenesis and in differentiated muscle fibers (Hinterberger et al., 1991); its functions appear to overlap with those of MyoD in the muscle differentiation pathway (Rawls et al., 1998).

Members of the MyoD family share homology in a basic helix-loop-helix (bHLH) region that mediates dimerization and binding to the E-box consensus sequence (CANNTG), which is found in the control regions of many muscle-specific genes (Olson and Klein, 1994). The bHLH region also associates with MEF2 transcription factors, which lack myogenic activity alone, but potentiate the muscle-inducing activity of myogenic bHLH proteins and are required for the activation of muscle differentiation genes (Molkentin et al., 1995). The four vertebrate MEF2 factors – MEF2A, MEF2B, MEF2C and MEF2D – belong to the MADS (MCM1, agamous, deficiens, SRF) family of transcription factors (Black and Olson, 1998). The MADS domain mediates dimerization, association with myogenic bHLH proteins, and binding to an A/T-rich DNA sequence associated with muscle-specific genes (Gossett et al., 1989; Pollock and Treisman, 1991).

In contrast to the skeletal muscle specificity of myogenic bHLH factors, MEF2 factors are expressed in skeletal, cardiac and smooth muscle cells, as well as in neurons (Breitbart et al., 1993; Chambers et al., 1992; Yu et al., 1992; Leifer et al., 1993; Martin et al., 1993, Martin et al., 1994; Lyons et al., 1995; Edmondson et al., 1994; Ticho et al., 1996), and at lower levels in several other cell types. During embryogenesis, Mef2c is expressed at the onset of differentiation of the cardiac and skeletal muscle lineages and is followed by expression of the other MEF2 genes (Edmondson et al., 1994). Mice that lack Mef2c die at about E9.5 from cardiovascular defects, precluding analysis of the role of Mef2c in skeletal muscle development in vivo (Lin et al., 1997). Mice homozygous for mutations in Mef2a or Mef2b are viable, whereas mice lacking Mef2d die prior to gastrulation (our unpublished results). Loss-of-function mutations of the single Mef2 gene in Drosophila result in a block to differentiation of all muscle cell types (Lilly et al., 1995; Bour et al., 1995; Ranganayakulu et al., 1995), indicating that MEF2 genes are required for muscle differentiation.

In addition to activating subordinate muscle structural genes, myogenic bHLH and MEF2 factors auto- and crossregulate the expression of one another. Forced expression of myogenic bHLH genes in non-muscle cell types is sufficient to activate expression of the endogenous myogenic bHLH genes and to upregulate expression of MEF2 genes (Lassar et al., 1991; Cserjesi and Olson, 1991). MEF2 alone cannot induce expression of myogenic bHLH genes (Molkentin et al., 1995), but the promoters of the myogenin and Mrf4 genes contain binding sites for myogenic bHLH and MEF2 factors that control transcription during myoblast differentiation in vivo and in vitro (Cheng et al., 1993; Yee and Rigby, 1993; Black et al., 1995; Naidu et al., 1995). Binding of MEF2 to these sites provides a mechanism for amplifying the expression of these regulatory genes and stabilizing the muscle phenotype. Whether myogenic bHLH proteins act directly on MEF2 genes to upregulate their expression, or whether they induce MEF2 expression through an indirect mechanism is unknown because regulatory elements for vertebrate MEF2 genes have not been identified.

To further understand the mechanisms that regulate MEF2 expression during myogenesis, we sought to identify cis-regulatory elements responsible for transcription of the mouse Mef2c gene during skeletal muscle development. We describe a novel skeletal muscle-specific control region upstream of the Mef2c gene that is sufficient to direct the expression of a lacZ transgene in a spatiotemporal expression pattern that mimics that of the endogenous gene during mouse development. Mutational analysis of this control region shows that it is a direct target for myogenic bHLH and MEF2 factors in vivo, revealing a transcriptional circuit through which these transcription factors induce, amplify and maintain Mef2c expression during muscle development.

5′-RACE (rapid amplification of cDNA ends) cloning was performed as described previously (Wang et al., 1999). Briefly, total RNA was isolated from adult mouse skeletal muscle using Trizol reagent (LifeTech). 5 μg of RNA was used for first-strand cDNA synthesis with random hexamers. A PCR-based RACE procedure was carried out using the Marathon cDNA amplification kit (Clontech) according to the manufacturer’s instructions. Mef2c gene-specific primers were as follows: 2C-RACE1, 5′-GTGTTTCTTCTCTCTCTCGTCCCTG-3′; and 2C-RACE2, 5′-GCACAGCTCAGTTCCCAAATCCCTG-3′.

An aliquot of the first PCR products was also used for subsequent nested PCR. Amplified cDNAs were gel-purified and subcloned into the pGEM-T-Easy vector (Promega) and sequenced. Multiple overlapping clones were isolated through this approach.

A mouse genomic library (Stratagene) was screened using a cDNA fragment obtained from 5′-RACE as a probe. Three positive clones were isolated and sub-cloned into the pBlueScript vector (Stratagene). Restriction mapping and DNA sequencing were performed as described previously (Lin et al., 1998).

Different fragments of the Mef2c 5-flanking region were subcloned into the hsp68-lacZ vector (Kothary et al., 1989) to make reporter constructs (detailed in Fig. 2). Transgenes were prepared and injected into the male pronuclei of fertilized oocytes from B6C3F1 mice as described previously (Cheng et al., 1992). Injected oocytes were then transplanted into ICR pseudopregnant females and embryos were harvested at the desired embryonic or neonatal time points indicated in the text and figure legends.

lacZ transgene expression was detected as previously described (Cheng et al., 1992). Briefly, embryos were dissected out of sacrificed mothers. Yolk sacs were removed for lacZ PCR genotyping. The amnion was also removed and embryos were fixed at 4°C in 2% paraformaldehyde and 0.2% glutaraldehyde in phosphate-buffered saline (PBS) for variable amounts of time, depending on the size of the embryo. Fixed embryos were washed in PBS at 4°C for 30 minutes. Embryos were stained overnight at room temperature in 1 mg/ml X-gal, 2 mM MgCl₂, 5 mM K₃Fe(CN)₆ and 5 mM K₄Fe(CN)₆ in PBS.

Mice carrying lacZ transgenes were identified by PCR analysis. Genomic DNA isolated from tail biopsies or embryonic yolk sacs was digested in lysis buffer (10 mM Tris (pH 8.0), 25 mM EDTA, 100 mM NaCl, 1% SDS, 0.2 mg/ml Proteinase K) at 55°C overnight, followed by removal of protein by phenol/chloroform extraction and ethanol precipitation. A typical PCR reaction contained genomic DNA template, 50 mM KCl, 10 mM Tris-HCl (pH 9.0), 1.5 mM MgCl₂, 0.1% Triton X-100, 0.16 mM of each dNTP, 1 μM of each primer and 2.5U Taq polymerase (Promega) in a 25 μl total volume. A typical temperature profile included 32 cycles of DNA strand melting at 95°C for 30 seconds, primer annealing at 55°C for 30 seconds and polymerization at 72°C for 30 seconds. lacZ genotype analysis was performed by PCR on genomic DNA using the following primers: 5′-CAAACTGGCAGATGCACGGTTAC-3′ and 5′-CAGTACAGCGCGGCTGAAATC-3′, producing a 450 bp lacZ-specific product.

The preparation, sectioning and staining of embryos was performed using standard procedures. Briefly, β-galactosidase-stained embryos were post-fixed in 4% paraformaldehyde in PBS overnight at 4°C and embedded in paraffin after a stepwise dehydration with progressively higher concentrations of ethanol and two changes of xylene. Embryos were sectioned in increments of 5-7 μm. Sections were counterstained with nuclear Fast Red and cover-slipped with cytoseal or permount.

Proteins for electrophoretic mobility shift assays (EMSA) were produced using the TNT T3/T7 Coupled Reticulocyte Lysate System (Promega). The DNA templates used for in vitro transcription of mouse MyoD, E12 and Myc-tagged MEF2C were: EMSV-MyoD, pCITE-E12 and pcDNA3.1-MEF2C-Myc, respectively. pEMSV-MyoD consists of a full length MyoD cDNA subcloned into the expression vector pEMSVscribe (Harland and Weintraub, 1985). pCITE-E12 contains the E12 cDNA inserted into the pCITE expression vector (Novagen). pcDNA3.1-MEF2C-Myc contains the full-length mouse MEF2C cDNA with a C-terminal Myc epitope tag (McKinsey et al., 2000a; McKinsey et al., 2000b) subcloned into the pcDNA3.1 expression vector (Invitrogen). To ensure that proteins were appropriately translated, parallel transcription-translation reactions were performed in the presence of [³⁵S]methionine, separated by 10% SDS-PAGE and visualized using autoradiography.

The sequences of the sense strands of the oligonucleotides used as probes in the gel mobility shift assays were as follows: MEF2 site, 5′-ACCTTTACAGCTAAATTTACTCCAGAGTG-3′; and E-box, 5′-GAGTGACATGAACAGGTGCACCCTGGCCT-3′.

Gel mobility shift assays were also performed with oligonucleotides corresponding to the high-affinity right E-box (Chakraborty et al., 1991) and MEF2 sites (Cserjesi et al., 1994) from the MCK enhancer as a positive control. The oligonucleotides were generated with four extra nucleotides, GAGG, at their 5′ ends. Oligonucleotides were annealed at a concentration of 100 ng/μl and were end-labeled with [α-³²P]dCTP using the Klenow fragment of DNA polymerase I. All binding reactions were performed using 2 μg poly(dI:dC) and 100,000 cpm of probe in a total volume of 20 μl of binding buffer (400 mM KCl, 150 mM Hepes (pH 7.9), 10 mM EDTA, 5 mM DTT, 50% glycerol). Where indicated, unlabelled annealed oligonucleotide was used as competitor. Reactions were incubated at room temperature for 15-20 minutes. Where indicated, mouse monoclonal anti-Myc (Santa Cruz) and anti-MyoD (Pharmingen) antibodies were added to the reactions at concentrations of 45 and 33 ng/μl, respectively, and incubated for an additional 10-15 minutes. Binding reactions were analyzed by electrophoresis on 4% (MyoD binding) and 5% (MEF2C binding) non-denaturing polyacrylamide gels in 0.5× TBE. Gels were dried and binding was visualized by autoradiography.

Mutagenesis of the MEF2 site and E-box was achieved using the overlap extension method, previously described (Horton and Pease, 1991). The DNA template used for mutagenesis and cloned into pBlueScript (Stratagene) included the nucleotide region from –512 bp to +41 bp. Mutant PCR products were then sub-cloned into the pGEM-T-Easy Vector (Promega). Mutated fragments were then excised using endogenous flanking restriction enzyme sites and cloned into the appropriate transgenic expression vector. All mutations were confirmed by DNA sequencing.

The 5′-untranslated sequences of the Mef2c gene have not been previously defined. Therefore, as a first step toward identifying regulatory regions responsible for Mef2c transcription in skeletal muscle, we performed 5′-RACE on mouse skeletal muscle mRNA using primers from the cDNA sequence immediately 5′ of the translation initiation codon. Several overlapping cDNA clones containing extended 5′-untranslated sequence were obtained. One of the 5′-RACE clones was then used to screen a mouse genomic library, resulting in three overlapping genomic clones.

The structure of the mouse Mef2c gene was characterized by genomic DNA sequencing and restriction mapping, as well as comparison of human and mouse genomic DNA sequences from several databases. The deduced structure of the mouse Mef2c gene is shown in Fig. 1A. The protein-coding region of the gene comprises 11 exons (exons 4-14) distributed over approximately 200 kb of genomic DNA. As reported previously, several exons are contained in all Mef2c transcripts, while others are used alternatively (Martin et al., 1993; McDermott et al., 1993). Based on RT-PCR analysis of RNA from different tissue sources and on the presence of exon sequences in expressed sequence tags, exons 6 and 7 appear to be mutually exclusive, with exon 6 being used in transcripts from heart and brain, and exon 7 being specific to skeletal muscle. Exons 9 and 14b are also detected in heart and brain transcripts, but not in skeletal muscle transcripts; and exon 11 is specific to brain. The AUG codon is contained in a 193 bp exon (exon 4) that, in skeletal muscle transcripts, is spliced to an approximately 250 bp exon (exon 1), located approximately 80 kb upstream. The sequence of the 5′-UTR from skeletal muscle Mef2c transcripts are shown in Fig. 1B.

We also performed 5′-RACE with RNA from mouse heart and brain and identified two additional 5′ exons that appear to be preferentially used in those tissues (Fig. 1A). We did not identify additional 5′ exon sequence in RACE products from these tissues, suggesting that different promoters are used in skeletal muscle, brain and heart to generate the distinct 5′-untranslated regions of the transcripts from these tissues.

To search for the regulatory region responsible for skeletal muscle expression of Mef2c, we fused a series of 5′ genomic fragments to the hsp68 basal promoter upstream of a lacZ reporter gene and tested for expression in F₀ transgenic mouse embryos (Fig. 2). We initially included the hsp68 basal promoter in the transgene constructs because this promoter is extremely sensitive and permissive to heterologous enhancers (Lien et al., 1999; McFadden et al., 2000) and it therefore enabled us to search for potential Mef2c regulatory elements without knowing the precise location of the promoter. As shown in Fig. 3, the 6.6 kb region immediately 5′ of the skeletal muscle-specific exon 1 was sufficient to direct strong skeletal muscle-specific expression in vivo (construct 1). As a positive control for lacZ transgene expression, a well-defined neural crest enhancer (NCE) of the mouse dHAND (Hand2 – Mouse Genome Informatics) gene, which has been shown to be active specifically in the branchial arches of developing embryos (Charité et al., 2001), was fused upstream of this 6.6 kb genomic sequence. The NCE gave rise to an expression pattern in the branchial arches and their derivatives (Fig. 3), as predicted. Because this enhancer is active in cell types in which Mef2c is not expressed, we were able to use it initially to rapidly survey numerous genomic fragments for the Mef2c skeletal muscle enhancer by identifying embryos that contained lacZ transgenes integrated into regions of chromatin permissive to gene expression. Once we had identified the region of genomic DNA with skeletal muscle regulatory activity, we no longer included the NCE in subsequent transgenes. Expression from construct 1 was localized to the somite myotomes at E9.5-11.5 (Fig. 3A-D). At E14.5, this construct was highly active in differentiated skeletal muscle fibers throughout the body (Fig. 3E). No expression in other cell types, including cardiac and smooth muscle, was detected at any developmental stage examined.

Bisection of the 6.6 kb region into fragments from –6.6 to –4.6 and from –4.4 to the first exon (constructs 2 and 3, respectively) showed that all skeletal muscle activity was localized to the 3′ DNA fragment (Fig. 4 and data not shown). Further dissection of this region localized the skeletal muscle control region to a fragment extending from –1.1 kb to exon 1 (construct 7, Fig. 4).

Because there is a precedent in other muscle genes for modularity of regulatory elements in which individual control regions direct only part of the muscle expression pattern and the complete pattern requires combinations of independent regulatory regions (Firulli and Olson, 1997), we carefully analyzed the expression pattern of construct 7 at various stages of development from E9.0 to E16.5 (Fig. 4). This construct was active in all embryonic skeletal myocytes that express Mef2c. The construct was also expressed at extremely high levels in all post-natal skeletal muscle fibers (Fig. 5). These results suggest that this upstream DNA region contains the cis-regulatory elements sufficient to direct the skeletal muscle expression pattern of Mef2c in vivo.

The proximal position of the Mef2c control region relative to exon 1 suggested that this region might function as a skeletal muscle-specific promoter, although no consensus TATA-binding sites are found in this 1.1 kb region. To test this, we fused the region from –1058 bp to +27 bp directly to a promoter-less lacZ transgene (construct 14). This transgene showed an expression pattern at E11.5 (Fig. 6A) that was the same as that of construct 7, although its level of expression was weaker, which we presume reflects the stronger potential activity of the hsp68 basal promoter, which was included in construct 7. This indicated that the 1.1 kb proximal regulatory region could also function as a skeletal muscle-specific promoter for Mef2c.

When the proximal 1.1 kb fragment (construct 7) was cut approximately in half, each part directed a distinct pattern of lacZ expression within the myotome. At E11.5, the distal portion (–1058/–507, construct 8) was expressed only in the dorsomedial lip of the myotome and in ventrolateral myoblasts in the limb (Fig. 6B-D). By contrast, the proximal portion (–512/+41, construct 9) was expressed in the entire myotome (Fig. 6E,F). The expression pattern of construct 9 was indistinguishable from the expression pattern of construct 7, from which it was derived. Thus, constructs 8 and 9 identify distinct, but overlapping myogenic precursor populations.

Because construct 9 was expressed throughout the entire myotome, and was as strong as construct 7, we attempted to further localize the cis-regulatory elements within this fragment. Bisection of construct 7 into two fragments (–512/–174, construct 10, and –158/+4, construct 11) demonstrated that all of its transcriptional activity was contained in the 3′ fragment (Fig. 6G,H and data not shown).

Together, the above results demonstrate the existence of two independent Mef2c regulatory regions capable of directing transcription in the skeletal muscle lineage. The region from –158/+4, which acts as a promoter, appears to be primarily responsible for the skeletal muscle-specific expression of Mef2c. This region can also direct muscle-specific expression in the opposite orientation when combined with the hsp68 promoter, which suggests that it can also act as an enhancer.

The sequence of the minimal skeletal muscle regulatory region is shown in Fig. 7A. Comparison of the sequence from the mouse and human genes showed extensive cross-species homology (data not shown). Within the minimal regulatory region, we identified an E-box and a potential MEF2 site (Fig. 7A,B). To determine whether myogenic bHLH or MEF2 proteins could bind these sequences, we performed gel mobility shift assays using oligonucleotide probes and in vitro translated proteins. As shown in Fig. 7C, Myc-tagged MEF2C bound avidly to the MEF2 site-containing oligonucleotide and binding was competed by the cognate sequence, but not by a nonspecific sequence (data not shown). The identity of the MEF2C-containing complex was confirmed by its supershift with anti-Myc antibody. Similarly, the E-box-containing sequence was bound by MyoD/E12 heterodimers, which were supershifted with anti-MyoD antibody (Fig. 7D). Homodimers of MyoD or E12 showed only weak binding to this site.

The potential roles of the above transcription factor binding sites in Mef2c transcriptional regulation were determined by mutagenesis of each site individually within the context of the –512/+41 bp region (constructs 12 and 13 of Fig. 2). As shown in Fig. 8D-F, the E-box mutant (construct 13) was inactive at both E9.5 and 11.5. Nine F₀ transgenic embryos harboring this construct were analyzed. None showed expression in the skeletal muscle lineage, but several showed ectopic expression in other cell types, reflecting random integration sites of the transgene. This data indicates that the E-box is necessary for initiation, as well as maintenance, of Mef2c expression in vivo.

In contrast to the complete inactivity of the E-box mutant, the MEF2 site mutant (construct 12) was weakly active in the myotome at E9.5 (Fig. 8A,B). However, at E11.5 this transgene was inactive (Fig. 8C). These results suggest that the MEF2 site is not required for the initiation of Mef2c transcription, but is essential for the amplification and maintenance of Mef2c expression at later stages in vivo.

MEF2 transcription factors play a central role in the control of skeletal muscle development by enhancing the muscle-inducing activity of myogenic bHLH proteins. While much has been learned about the mechanisms whereby MEF2 proteins activate muscle-specific transcription, little is known of the mechanisms that regulate expression of MEF2 genes and no cis-regulatory elements that control transcription of vertebrate MEF2 genes have been previously identified. This has been a difficult problem because the 5′-UTRs of MEF2 transcripts are encoded by multiple alternative exons. For Mef2c, these exons are distributed over approximately 80 kb of genomic DNA. To locate the control region responsible for the expression of Mef2c in skeletal muscle, we used tissue-specific 5′-RACE to identify 5′-exons encoding skeletal muscle transcripts. We also took advantage of the completed human genome sequence by searching the human genome database using the tissue-specific 5′-untranslated sequences that we identified. Our results demonstrate the presence of tissue-specific alternatively spliced exons in the 5′-UTR of the Mef2c gene and suggest that distinct Mef2c transcripts are directed by different regulatory elements.

Our results are consistent with the possibility that the Mef2c gene is a direct target for transcriptional activation by myogenic bHLH and MEF2 proteins in the developing skeletal muscle lineage in vivo. These results confirm and extend previous findings that MEF2 DNA binding activity is upregulated by myogenic bHLH proteins (Lassar et al., 1991; Cserjesi and Olson, 1991) and reveal a direct positive feedback loop between these two classes of transcription factors. This is the first demonstration of a direct role of myogenic bHLH factors in the activation of MEF2 gene transcription in any organism.

Skeletal muscle development is initiated in the rostral somites of the mouse embryo at E8.0 (Hauschka, 1994). Signals from the notochord induce cells from the ventral region of the somite to undergo an epithelial-to-mesenchymal transformation and migrate away from the somite, forming the sclerotome (Christ et al., 1978), which gives rise to the axial skeleton. The remaining epithelial cells of the dorsal somite give rise to the dermomyotome, which serves as the source of myogenic precursors of the trunk and limb musculature, and the myotome, from which the muscles of the back are derived (Denetclaw et al., 1997; Ordahl et al., 2001). Myf5 is the first marker of the skeletal muscle lineage and is expressed in the dorsomedial lip of the dermomyotome at E8.0 in the mouse followed soon after by the expression of myogenin, MRF4 and MyoD at E8.5, E9.5 and E10.5, (Sassoon et al., 1989; Ott et al., 1991; Hinterberger et al., 1991), respectively.

Like the myogenic bHLH transcription factors, MEF2 factors are expressed early in myogenic precursors and in developing skeletal muscle. Mef2c is the first member of the MEF2 family to be expressed in skeletal muscle in vivo; its expression is initially detected at E9.0 in the rostral myotomes at the onset of myocyte differentiation (Edmondson et al., 1994) and expression is maintained throughout skeletal muscle development. In the current study, we were able to detect expression of the Mef2c-lacZ transgene in the myotome as early as E9.0, throughout skeletal muscle development, and into adult muscle, recapitulating that of the endogenous Mef2c gene. The lag between the expression of Myf5/myogenin and Mef2c in developing somites suggests that the myogenic bHLH factors, but not MEF2C, are required for the initiation of skeletal muscle differentiation, whereas MEF2 proteins may be involved in the maintenance and/or amplification of the skeletal muscle differentiation program.

Forced expression of myogenic bHLH proteins in non-muscle cells is sufficient to upregulate MEF2 expression (Lassar et al., 1991; Cserjesi and Olson, 1991). Our results suggest that this reflects the direct binding of myogenic bHLH proteins to an E-box in the proximal promoter of the Mef2c gene. Our data also indicate that in the context of the Mef2c gene (–512 to +41) region, this E-box is essential for the initial activation of Mef2c transcription in the skeletal muscle lineage. An E-box mutation in the Mef2c promoter completely abolished lacZ reporter expression at E9.5. Given that the E-box mutation also abolished expression at later stages of development, binding of myogenic bHLH proteins to this site also appears to be required for the maintenance of Mef2c expression. As Myf5 and myogenin are expressed prior to Mef2c, these factors are potential initiators of Mef2c transcription. In mature skeletal muscle fibers, it is likely that MRF4 maintains Mef2c expression, as this factor is the most highly expressed in postnatal muscle.

In contrast to the requirement of the E-box for initiation of Mef2c transcription, a mutation of the MEF2 site did not affect the timing for initial activation of the reporter gene such that weak lacZ expression was detected in the somites at E9.5. However, at later stages, this MEF2 site mutant was unable to maintain lacZ expression and was completely inactive at E11.5. Together, these results suggest that MEF2C acts on its own promoter to amplify and maintain its expression in differentiating myoblasts and differentiated muscle. Other members of the MEF2 family also become expressed at high levels in the somite myotome after E9.5 (Edmondson et al., 1994), which could further reinforce the expression of Mef2c.

Our results suggest that myogenic bHLH transcription factors are required for the initial activation of Mef2c expression in vivo. Although MEF2 proteins are not sufficient to activate Mef2c transcription, they cooperate with the myogenic bHLH proteins, which provide the muscle specificity necessary to direct the expression of Mef2c in cells of the myogenic lineage throughout development and into adulthood. A model of how myogenic bHLH and MEF2 proteins might cooperatively regulate the expression of Mef2c during skeletal myogenesis is shown in Fig. 9A.

Recently, we have shown that MEF2 proteins associate with histone deacetylase5 (HDAC5) in myoblasts, resulting in repression of muscle genes controlled by MEF2 sites (McKinsey et al., 2000a). When myoblasts are triggered to differentiate, HDAC5 is exported to the cytoplasm, allowing for enhanced activity of MEF2 factors and upregulation of MEF2 target genes. As the Mef2c gene is itself a target for the MEF2 transcription factor, this type of derepression mechanism could also account for the upregulation of MEF2 expression during myogenesis.

The structure of the Mef2c skeletal muscle promoter/enhancer is remarkably similar to that of the myogenin and MRF4 promoters, both of which contain MyoD- and MEF2-binding sites (Edmondson et al., 1992; Cheng et al., 1993; Yee and Rigby, 1993; Black et al., 1995; Naidu et al., 1995). Like Mef2c, myogenin and MRF4 are upregulated during myocyte differentiation, presumably through direct transactivation by bHLH and MEF2 factors. This type of crosstalk provides a powerful mechanism for amplification of both types of regulatory genes, thereby reinforcing and stabilizing the transcriptional program for myogenesis.

In transfected cells, MyoD and MEF2 can cooperatively activate transcription through a binding site for either factor alone (Molkentin et al., 1995). By contrast, mutation of either the MEF2 or MyoD site in the Mef2c promoter severely diminished transcriptional activity in vivo. These findings suggest that the levels of expression of the endogenous factors are insufficient to activate transcription solely through protein-protein interactions and without binding sites for both factors. This further demonstrates the existent crossregulation between the myogenic bHLH transcription factors and the MEF2 proteins in skeletal muscle development. A model to account for the regulation of Mef2c transcription during myogenesis is shown in Fig. 9B.

The Drosophila Mef2 gene is also expressed at high levels in differentiated skeletal muscle fibers (Lilly et al., 1995; Bour et al., 1995). Previously, we and others showed that transcription of Mef2 is controlled by an array of independent enhancers that are activated in specific subsets of myocytes at distinct developmental stages (Cripps et al., 1998; Cripps et al., 1999; Nguyen and Xu, 1998). Recently, we identified an enhancer that directs Mef2 transcription specifically in differentiated myofibers of the Drosophila embryo (R. Cripps and E. N. O., unpublished). This enhancer contains both a high-affinity MEF2-binding site that is essential for transcriptional activity and an E-box that is necessary for full activity. Thus, the type of positive auto- and crossregulatory loops identified in the present study seems likely to reflect an evolutionarily conserved mechanism for the control of MEF2 gene expression in the skeletal muscle lineage.

A common theme that has emerged through the analysis of muscle gene transcription is the modularity of cis-regulatory elements, in which multiple independent regulatory regions are required to generate the complete spatiotemporal expression pattern of a gene throughout development (Firulli and Olson, 1997). Our results demonstrate the existence of two independent regulatory regions that direct Mef2c transcription in discrete, but overlapping, sets of embryonic skeletal muscle cells.

A distal upstream region (–1058/–507) activates transcription specifically in cells from the dorsomedial myotome and ventrolateral dermomyotome, which give rise to extreme epaxial, as well as hypaxial and limb muscles, respectively. To our knowledge, these two populations of muscle cells related and have not been previously recognized as being molecularly distinct from other mnyogenic populations. Thus, this Mef2c transgene has revealed unique compartments of the myogenic lineage. The specific transcription factors that activate this distal regulatory region remain to be identified.

The proximal ∼500bp regulatory region of Mef2c acts independently of this distal enhancer and is sufficient to direct expression of lacZ in a pattern that appears to recapitulate the expression pattern of the endogenous Mef2c gene during pre- and postnatal skeletal muscle development. Although our studies do not allow us to conclude that these two regions are solely responsible for Mef2c transcription in the skeletal muscle lineage, we have found no other skeletal muscle regulatory elements within 24 kb of genomic DNA analyzed.

In addition to its expression in the skeletal muscle lineage, Mef2c is expressed in the developing heart, in specific sets of neurons in the brain, and in the spleen (Lyons et al., 1995; Martin et al., 1993; Edmondson et al., 1994). Unlike many other muscle-specific enhancers, the Mef2c enhancer described here is absolutely specific for the skeletal muscle lineage with no expression in the heart or other organs, suggesting the presence of completely separate regulatory mechanisms for expression in tissues other than skeletal muscle. We have not yet identified the regulatory elements that control Mef2c transcription in the latter cell types. However, our studies suggest that cardiac and neural transcripts for Mef2c contain unique 5′ exons that probably reflect alternate promoters in these tissues. The mechanisms that regulate MEF2 expression in these cell types and whether MEF2 positively autoregulates its expression in cardiac myocytes and neurons, as in skeletal myocytes, is currently under investigation.

We thank J. Page for editorial assistance and A. Tizenor for graphics. Supported by grants from the National Institutes of Health, and the Donald W. Reynolds Cardiovascular Clinical Research Center, Dallas, Texas to E. N. O. D. W. was supported by a grant from the Muscular Dystrophy Association and M. R. V. was supported by a Training Grant from the National Institutes of Health.