The process of myogenesis requires the coordinated activation of many structural genes whose products are required for myofibril assembly, function, and regulation. Although numerous reports have documented the importance of the myogenic regulator myocyte enhancer factor 2 (MEF2) in muscle differentiation, the interaction of MEF2 with cofactors is critical to the realization of muscle fate. This study identified a genomic region required for full MEF2-mediated activation of actin gene expression in Drosophila, and the zinc finger transcriptional regulator Chorion factor 2 (CF2) was identified as a factor functioning alongside MEF2 via this region. Furthermore, although both MEF2 and CF2 can individually activate actin gene expression, these two factors collaborate in regulating the Actin57B target gene in vitro and in vivo. More globally, MEF2 and CF2 synergistically activate the enhancers of a number of muscle-specific genes, and loss of CF2 function in vivo results in reductions in the levels of several muscle structural gene transcripts. These findings validate a general importance of CF2 alongside MEF2 as a critical regulator of the myogenic program, identify a new regulator functioning with MEF2 to control cell fate, and provide insight into the network of regulatory events that shape the developing musculature (Tanaka, 2008).

A major function of the mesodermal germ layer during embryogenesis is the generation of myogenic tissue and the subdivision of cells into skeletal, cardiac, and visceral muscle precursors. This process has been broadly conserved during evolution. Specification of the three major muscle lineages is conserved from Drosophila to mammals, and conserved structural genes are activated in each lineage to provide power for locomotion, circulation, and digestion (Tanaka, 2008).

The transcription factors that mediate muscle cell differentiation are also broadly conserved. Foremost among these proteins are members of the myocyte enhancer factor 2 (MEF2) family of transcriptional regulators. MEF2 proteins dimerize and bind to DNA via conserved MADS and MEF2 protein domains, and MEF2 proteins recognize an AT-rich sequence, which is found in the promoters and enhancers of numerous muscle-specific genes. MEF2-encoding genes are found in all animal genomes, from single-gene copies in Caenorhabditis elegans (named CeMef2) and Drosophila melanogaster (named Mef2), to four genes in mammals (named mef2a to -d) (Tanaka, 2008).

Consistent with the pervasiveness of MEF2 sites in muscle-specific genes and the conservation of this gene family during evolution, mutational studies in several animal models have established important roles for MEF2 factors in muscle differentiation. In Drosophila, inactivation of the single Mef2 gene resulted in a complete failure of differentiation for all muscle lineages. In mice, inactivation of mef2c resulted in a failure of normal cardiac and visceral muscle development. Furthermore, inactivation of murine mef2a caused cardiac failure shortly after birth. Also, expression of dominant-negative isoforms of MEF2 predicted to inhibit all MEF2 function resulted in the inhibition of skeletal myogenesis in mammalian cells in vitro and in interference in cardiac development in vivo. Finally, a point mutation in the human gene encoding MEF2A is associated with susceptibility to cardiac disease. These studies present a strong argument for the importance of MEF2 proteins in muscle development and disease (Tanaka, 2008).

Despite the critical requirement of MEF2 for muscle differentiation, MEF2 proteins are not themselves capable of activating the myogenic program in naive cells in tissue culture. Similarly in Drosophila, ectopic expression of Mef2 in the ectoderm caused the activation of some muscle-specific markers, and yet failed to induce myogenesis at high levels. These findings suggested that MEF2 must act with specific cofactors in order to control myogenesis. Much progress has subsequently been made in defining both positive and negative cofactors for MEF2, in a variety of muscle tissues. MEF2 collaborates positively with skeletal muscle-specific members of the basic helix-loop-helix family of transcriptional regulators, including MyoD and myogenin, to control skeletal myogenesis. In addition, the SAP domain proteins Myocardin and MASTR also stimulate the transcriptional activity of MEF2 in cardiac and skeletal muscle tissue, respectively. MEF2 proteins also interact with ubiquitous factors, such as the p300 coactivator and members of the histone deacetylase family of transcriptional repressors (Tanaka, 2008).

In Drosophila, there is relatively little direct data concerning the identification of cofactors that might collaborate with MEF2 to control myogenesis. Nevertheless, a number of enhancers for muscle-specific genes have been described, many of which contain MEF2 binding sites that are critical for full gene activation. Among these, the promoter has been described for the Act57B actin gene in Drosophila (Kelly, 2002). Act57B is the predominant embryonic muscle actin, being expressed at high levels in all of the major muscle lineages and also being one of the earliest markers of muscle differentiation in the embryo. Activation of Act57B expression in all embryonic muscle lineages is controlled by a proximal 595-bp promoter element. Within this region, full transcriptional activation required the integrity of a single MEF2 binding site at position -209 relative to the transcriptional start site, validating the importance of MEF2 as a direct regulator of myogenesis in this system. This finding was also supported by the study of Sandmann (2006), who showed, using microarray analysis, that Act57B expression was strongly downregulated in Mef2 mutants. Furthermore, the Sandmann study utilized chromatin immunoprecipitation-microarray analyses to identify the MEF2 site at position -209 as an in vivo target of MEF2 (Tanaka, 2008).

An ~300-bp region distal to the MEF2 site is critical for full Act57B activation in embryonic muscle tissue, suggesting the existence of a MEF2 cofactor that interacted with this distal sequence. This analysis permitted a directed approach to be carried out to identify and functional characterize a MEF2 collaborating factor in Drosophila. This study delineates sequences within the 300-bp distal region that are required for full gene activation in vivo; a nuclear factor binds to this region with the characteristics identical to that of the C2H2 zinc finger transcriptional regulator chorion factor 2 (CF2) (Hsu, 1992). It is further shown that MEF2 and CF2 can each activate Act57B expression in vitro and in vivo and that these factors function synergistically to maintain high levels of actin expression in the Drosophila embryo. More globally, MEF2 and CF2 synergistically activate other muscle structural genes, and loss of CF2 function results in reductions in the expression levels of several muscle structural genes, including Act57B. These findings describe the first collaborating factor for MEF2 in the Drosophila system and further delineate the network of transcriptional events required for normal muscle development (Tanaka, 2008).

A region of the Act57B promoter from positions -593 to +2 can provide robust levels of muscle-specific expression of a lacZ reporter gene in transgenic animals. Within this fragment, at position -209, is a conserved MEF2 binding site that was critical for full promoter activity. Furthermore, 5' deletion of 315 nucleotides to generate a -278/+2 Act57B-lacZ reduced reporter gene expression. These findings established the importance of the MEF2 binding site for full activation of the Act57B gene (Kelly, 2002). However, since the -278/+2 construct showed reduced reporter activity despite retaining the MEF2 binding site at position -209, the data also indicated that a genomic region separate from the MEF2 site was critical for normal levels of gene expression, which might therefore interact with a cofactor for MEF2 (Tanaka, 2008).

To more precisely delineate regions critical to Act57B expression, a further 5' deletion analysis of the -593/+2 fragment was performed. At least three transgenic lines were assessed for each construct, and representative images are presented. Expression of the full-length construct was first detected at stage 11 of embryogenesis in visceral muscle precursors, and by stage 12 in skeletal muscle and visceral muscle precursors; high levels of reporter gene expression were still observed at stage 16 in all muscle lineages. 5' deletion of ~70 bp to position -521 did not affect the initiation of Act57B expression at stage 11, although reporter gene expression levels were consistently reduced at stage 16. This was apparent by the lesser ease with which individual muscle fibers could be distinguished in stained embryos at higher magnifications. A further 5' deletion of 130 bp to position -390/+2 showed no additional strong reduction in reporter gene expression. In contrast, the smallest -278/+2 Act57B-lacZ showed reduced reporter activity at stage 12, and reporter levels also appeared to be further reduced at stage 16 compared to larger genomic fragments that were tested (Tanaka, 2008).

These studies demonstrated that more than one region of the Act57B upstream promoter is required for normal reporter gene expression in vivo. Although it is difficult to assign relative levels of expression in these immunohistochemical stains, two regions seem most critical to normal expression: the region between positions -593 and -521 is a genomic element required for full gene activation at later stages of development, and a more proximal region from positions -390 to -278 is important for the initiation of Act57B expression at stages 11 and 12, as well as for later sustained Act57B expression. Subsequent studies therefore focused initially upon the identification of factors interacting with this more proximal region (Tanaka, 2008).

In order to determine whether factors existed that were capable of interacting with the Act57B promoter region, electrophoretic mobility shift assays were carried out using as a probe a -390/-245 region of the Act57B gene. When this fragment was mixed with nuclear extracts prepared from 12- to 24-h-old Drosophila embryos, the formation was observed of a slowly migrating complex, which is hypothesized to corresponded to a transcription factor capable of regulating Act57B expression (Tanaka, 2008).

To more precisely locate the region of the DNA probe with which the nuclear factor was interacting, binding reactions were compated with sets of double-stranded oligonucleotides spanning the entire probe sequence. It was found that, whereas most of the oligonucleotide competitors failed to reduce the intensity of the bound complex, double-stranded DNA probes of length approximately 30 bp, termed here oligonucleotides 11 and 12, effectively competed for the formation of the complex (Tanaka, 2008).

These findings were important for two reasons. First, the observation that some oligonucleotides were capable of competition, whereas others were not, indicated that the interaction of the nuclear factor with the Act57B probe was sequence specific. Second, competition assays identified more precisely the region of the Act57B gene that was interacting with the nuclear factor (Tanaka, 2008).

Observation of the DNA sequence corresponding to the locations of oligonucleotides 11 and 12 revealed several repeats of a 5'-TATA-3' motif, which lies at the core of the recognition sequence of the transcription factor CF2 (Gorgos, 1992). Since CF2 is a C2H2 zinc finger protein that regulates gene expression in the chorion cells of the maturing egg in Drosophila (Gogos, 1992; Gogos, 1996), and since embryonic expression of CF2 is exclusive to the three muscle lineages of the mesoderm (Bagni, 2002), it was reasoned that CF2 might be interacting specifically with the Act57B enhancer (Tanaka, 2008).

To determine whether CF2 could interact with the Act57B gene enhancer, an electrophoretic mobility shift assay was carried out using as a probe the -390/-245 region and using the CF2 protein generated in vitro for binding. A robust interaction was found between CF2 and the -390/-245 region, and this interaction was competed for with excess nonradioactive -390/-245 sequence. More importantly, when the interaction of CF2 and -390/-245 probe was competed using the oligonucleotides spanning the region, CF2 binding was competed for only by oligonucleotides 11 and 12 (Tanaka, 2008).

These findings firstly demonstrated that CF2 is capable of interacting with the Act57B promoter. This is likely to be a relevant interaction given the broad expression of CF2 in the embryonic mesoderm. Second, these results showing identical responses to binding competition by a nuclear factor and by CF2 protein strongly suggested that CF2 is the interacting factor identified in nuclear extracts. It is noted that when CF2 protein is synthesized in vitro the electrophoretic mobility shift assays show two shifted bands of very similar mobilities. This probably arises from the synthesis of two CF2 isoforms in the in vitro expression system. Whether this arises from posttranslational modification of the CF2 protein or from internal translation initiation is not clear. Nevertheless, there is a strong and specific interaction between CF2 protein and the Act57B promoter region (Tanaka, 2008).

Clearly, the C2H2 zinc finger transcription factor CF2 is able to bind potently to regions of the Act57B promoter that were critical for full levels of gene expression, suggesting that it might be a positive regulator of Act57B transcription. To test the hypothesis that CF2 is an Act57B activator, a series of cell culture experiments was initiated aimed at determining whether MEF2 or CF2 can activate Act57B promoter-lacZ constructs. First, Drosophila S2 cells were cotransfected with the full-length -593/+2 promoter-lacZ, along with a MEF2 expression plasmid. This resulted in strong and reproducible activation of the reporter. In contrast, a point mutation of the putative MEF2-binding site within the -593/+2 promoter-lacZ construct dramatically impaired MEF2-dependent reporter responsiveness. These results confirmed the importance of the sole MEF2 site in this construct for Act57B gene activation (Tanaka, 2008). When the ability of CF2 to activate the wild-type Act57B-lacZ construct was tested, significant activation of the reporter occured, suggesting that CF2 was indeed an activator of this gene. To determine whether CF2 activates Act57B via the sites identified in vitro, the ability of CF2 to activate the 5'-deleted reporter constructs was tested. Indeed, deletion of 5' nucleotides to -390/+2, predicted to remove the most distal CF2 binding sites, reduced reporter activation by CF2 by ca. 50% compared to the full-length promoter fragment. The shortest version of the Act57B promoter, -278/+2, lacking both of the predicted CF2-binding regions, was completely unable to respond to CF2. Similarly, tests were performed to see whether the CF2-binding regions alone are sufficient for reporter activation, in the proximally truncated construct -593/-245. Here, there was strong CF2-dependent activation, although it did not reach full strength compared to the -593/+2 promoter construct. Apparently, some unknown regulatory elements in the proximal part of the Act57B promoter, perhaps its basal promoter, may additionally boost transcriptional activation initially conferred by the CF2-binding regions. Also, comparative analysis of this construct might be complicated by the observation that its basic activity was more than two times higher than that of the other tested constructs (Tanaka, 2008).

Taken together, these provide strong support for the identification of CF2 as an activator of Act57B: CF2 is capable of binding to two regions of the promoter, which were shown to be important for full reporter gene expression in vivo; furthermore, CF2 is identified as a strong transcriptional activator of the Act57B promoter via these same sequences in tissue culture (Tanaka, 2008).

Analyses performed on the promoter of Act57B provided direct evidence that CF2 can independently initiate and synergistically collaborate with MEF2 to maximize expression of this muscle-specific gene. However, it was of particular interest to investigate whether CF2 is more broadly involved in the regulation of the transcriptional activities of other structural muscle genes. In order to address this question, promoter reporters were analyzed of two genes for which the muscle enhancers had been identified: Troponin I (TnI) and Myosin heavy chain (Mhc) (Tanaka, 2008).

Transfection studies in cell culture indicated that CF2, as well as MEF2, could individually be effective transcriptional activators for both TnI and Mhc reporters. This is the first direct evidence in support of the roles of MEF2 and CF2 upon these structural genes. More importantly, it was found that CF2 and MEF2 also synergistically act upon the promoter-lacZ constructs for both TnI and Mhc. These findings supported a general role for the collaboration of CF2 and MEF2 in the control of embryonic muscle development in Drosophila. It was interesting that the titration analyses of the TnI and Mhc reporters showed different degrees of synergism between MEF2 and CF2: whereas dual activation of the TnI reporter with MEF2 and CF2 just slightly exceeded the predicted additive limit, the Mhc reporter showed a remarkable boost of activation conferred by the combination of these two factors (Tanaka, 2008).

Given the demonstrations that CF2 could activate Act57B gene expression in vivo and in tissue culture, it was shown that tests of loss of CF2 function resulted in a reduction or loss of expression of muscle structural genes. A Cf2 mutant named Df(2L)gamma, described by Bagni (2002), comprises a deletion of the Cf2 transcribed region and is homozygous larval lethal. In order to determine whether the loss of Cf2 function affects muscle structural gene expression, homozygous mutant embryos (at stage 16-17) or first-instar larvae were collected and assayed for the accumulation of the muscle-specific Act57B, TnI, and Mhc transcripts relative to the expression of the housekeeping Act5C cytoplasmic actin gene by using quantitative RT-PCR. As controls, heterozygous siblings were collected at embryonic or larval stages; these were assayed for expression of the same genes (Tanaka, 2008).

In embryos, there was an ~10% reduction in the levels of expression of all three muscle structural genes assayed relative to the Act5C control expression levels. While this change in expression is relatively small, it must be borne in mind that removal of the CF2 sites from the Act57B-lacZ reporter construct results in similarly moderate effects upon reporter gene expression, even though these two separate experiments cannot be directly compared. More strikingly, at the larval stage the reductions in the levels of muscle gene expression were much more severe, suggesting that a major function of CF2 might be to maintain high levels of expression in fully differentiated tissues. This reduction in the expression of the three genes tested is highly reproducible (Tanaka, 2008).

Overall, these studies identify CF2 as a potent activator of muscle gene expression both in vitro and in vivo and show that CF2 functions alongside MEF2 in controlling high levels of expression of muscle-specific genes (Tanaka, 2008).

This study has identified the first known collaborative factor for MEF2 function in Drosophila. CF2, a transcriptional regulator first identified as an activator of chorion protein genes in the female ovary, is expressed in all three muscle lineages in the embryonic mesoderm. The data indicate that CF2 interacts with the Act57B promoter both in the context of embryonic nuclear extracts and when expressed in vitro, and CF2 binding sites correspond precisely to the genomic regions required for full Act57B-lacZ activation. Furthermore, overexpression studies in vitro and in vivo establish a synergistic relationship between MEF2 and CF2 in the activation of Act57B. These studies define the first known embryonic function for this zinc finger transcriptional regulator. Nevertheless, it is also noted that even in the absence of CF2 sites, in the -270/+2 Act57B-lacZ construct there is significant muscle-specific expression. This indicates that CF2 contributes to high levels of structural gene expression but that it is not essential for the activation of Act57B (Tanaka, 2008).

Removing the genomic regions containing CF2 binding sites had clear effects upon Act57B-lacZ expression at stage 16. In contrast, removal of CF2 function from the embryos in the mutant analysis resulted in a more modest reduction in Act57B transcription. Although these two experiments utilize different approaches and different readouts, what might be the cause of this apparent discrepancy? The most reasonable explanation is that there are probably additional regulators of Act57B that bind to the genomic regions affected in the deletion analyses; thus, the deletions remove the influence not only of CF2 but also of other muscle-specific activators. In contrast, deletion of the CF2 gene might remove only the influence of CF2 from actin gene expression. Future deletion and point mutant analyses, carried out in the manner applied in this study, would test this hypothesis and could ultimately shed light upon the identity of such additional factors (Tanaka, 2008).

More broadly, MEF2 and CF2 can activate other muscle structural genes in Drosophila. Both of the enhancers that tested have predicted binding sites for MEF2 and CF2 proteins (TnI and Mhc); however, the current study privides the first direct functional tests of whether MEF2 and CF2 can transcriptionally activate these enhancers. The findings suggest strongly that the collaboration of MEF2 and CF2 during fly embryonic muscle development is a general and important phenomenon (Tanaka, 2008).

A simple tissue culture assay was used to demonstrate that MEF2 and CF2 collaborate to activate endogenous genes in the context of the intact genome, in addition to transiently transfected reporter constructs. The assay closely supports the findings from other experiments presented in this study and is somewhat more quantifiable than ectopic expression assays in embryos. This is because embryos already express significant levels of the targets genes in the mesoderm, whereas in vitro the baseline from nontransfected cells is essentially zero. Similarly, the lacZ reporter vectors that were use frequently show leaky promoter activity, which might ultimately result in an underestimation of the true activation of target promoter-lacZ constructs. In this sense, the activation of endogenous genes in S2 cells is highly informative (Tanaka, 2008).

Presumably, a more complete set of MEF2/CF2 targets could be generated by cotransfecting S2 cells with expression plasmids for CF2 and MEF2 and then performing an array analysis of transcripts upregulated after this treatment. Such a data set would significantly overlap with the analysis of mesoderm differentiation described by Sandmann (2006), although it might have the advantage of indicating which genes are direct targets of combinatorial activation by MEF2/CF2 (Tanaka, 2008).

Along these lines, the current studies also indicate that there are subtle, yet potentially important, differences between the responsiveness of different target genes to activation by MEF2 and CF2 alone. In tissue culture, Act57B is readily activated by MEF2 and CF2 individually, and yet TnI and Mhc induction are detected far less readily. This might be a simple result of different ease of detection for different targets. Alternatively, there might be a more mechanistic basis for these differences. One explanation might be related to the proximity of binding sites of the respective factors to the target promoter. For Act57B, both MEF2 and CF2 sites that were mapped are within 600 bp of the transcription start site. In contrast, for Mhc most of the candidate sites are located in the first intron, at least 800 bp away from the transcription start site. In contrast the other hand, putative binding sites for MEF2 and CF2 are in relatively close proximity to the TnI transcription start site. Thus, it is not clear whether binding site proximity plays a role in the activation of genes in vitro, although it should be noted that for TnI and Mhc many of the sites have yet to be validated by DNA-binding assays (Tanaka, 2008).

The data of Bagni (2002) indicate that Cf2 expression is dependent upon the function of the Mef2 gene in Drosophila embryos; these authors also show that MEF2 levels are unaffected in Cf2 mutants, at least at the embryonic stage. It is therefore most likely that the reduction in muscle structural gene transcripts in Cf2 mutants arises directly from a loss of CF2 protein rather than from indirect effects upon MEF2 levels. Clearly, one of the means by which MEF2 activates myogenesis at high levels is via the activation of its own collaborative factors. There is some precedent for this mechanism, both in muscle and other systems. In Drosophila cardiac development, the homeodomain transcription factor Tinman is a direct activator of the GATA factor pannier, after which Tin and Pnr can collaborate to activate structural genes such as Sulfonylurea receptor. The activation of, and then collaboration with, a cofactor appears to be a commonly used genomic regulatory mechanism (Tanaka, 2008).