Understanding how complex patterns of temporal and spatial expression are regulated is central to deciphering genetic programs that drive development. Gene expression is initiated through the action of transcription factors and their cofactors converging on enhancer elements leading to a defined activity. Specific constellations of combinatorial occupancy are therefore often conceptualized as rigid binding codes that give rise to a common output of spatio-temporal expression. Here, we assessed this assumption using the regulatory input of two essential transcription factors within the Drosophila myogenic network. Mutations in either Myocyte enhancing factor 2 (Mef2) or the zinc-finger transcription factor lame duck (lmd) lead to very similar defects in myoblast fusion, yet the underlying molecular mechanism for this shared phenotype is not understood. Using a combination of ChIP-on-chip analysis and expression profiling of loss-of-function mutants, we obtained a global view of the regulatory input of both factors during development. The majority of Lmd-bound enhancers are co-bound by Mef2, representing a subset of Mef2's transcriptional input during these stages of development. Systematic analyses of the regulatory contribution of both factors demonstrate diverse regulatory roles, despite their co-occupancy of shared enhancer elements. These results indicate that Lmd is a tissue-specific modulator of Mef2 activity, acting as both a transcriptional activator and repressor, which has important implications for myogenesis. More generally, this study demonstrates considerable flexibility in the regulatory output of two factors, leading to additive, cooperative, and repressive modes of co-regulation.
While genetic studies are essential to reveal the phenotypic relationships between genes, it is often very difficult to disentangle the molecular mechanism of two genes that phenocopy each other. In this study, we used global scale and single gene analysis to investigate the relationship between two transcription factors whose mutant embryos have a similar defect in myogenesis. In Drosophila, Mef2 mutant embryos display a block in myoblast fusion, which is very similar to what is observed in mutant embryos for lmd, a zinc-finger transcription factor. To understand the underlying nature of these defects we used ChIP-on-chip analysis to obtain a global view of their co-regulated enhancers, and we used expression profiling of mutant embryos to reveal their downstream transcriptional response. The results indicate that Lmd acts as a tissue specific modulator of Mef2 activity. Using in vivo and in vitro reporter assays, we show that co-binding to the same enhancer element can lead to diverse regulatory responses. The presence of Lmd has an additive, cooperative, or repressive effect on Mef2 activity, demonstrating that it acts as a molecular switch for gene expression during muscle differentiation. More broadly, our results highlight the difficulty in translating information on combinatorial binding data into a functional regulatory response.
Citation: Cunha PMF, Sandmann T, Gustafson EH, Ciglar L, Eichenlaub MP, Furlong EEM (2010) Combinatorial Binding Leads to Diverse Regulatory Responses: Lmd Is a Tissue-Specific Modulator of Mef2 Activity. PLoS Genet 6(7): e1001014. https://doi.org/10.1371/journal.pgen.1001014
Editor: Dirk Schübeler, Friedrich Miescher Institute for Biomedical Research, Switzerland
Received: October 19, 2009; Accepted: June 1, 2010; Published: July 1, 2010
Copyright: © 2010 Cunha et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was in part supported by funding from the EU FP6 grant MyoRes (EEMF). The funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Development is driven by precise patterns of spatio-temporal gene expression, which are regulated through the action of transcription factors and cell signaling cascades converging on cis-regulatory modules (CRMs). CRMs are typically bound by multiple transcription factors, whose concentrations and interactions change dynamically over time. It is this combinatorial and dynamic property of CRM occupancy which makes regulatory output difficult, if not impossible, to predict based on information from a single transcription factor (TF) . Understanding the regulation of complex developmental processes requires linking combinatorial binding at the molecular level to the regulation of these processes at the phenotypic level. We have assessed the contribution of two well-studied TFs, Mef2 (Myocyte Enhancing Factor 2) and Lmd (Lame duck) to the cellular process of myogenesis during Drosophila development. Although the phenotypic defects in myoblast fusion are almost identical in Mef2 or lmd loss-of-function mutant embryos –, the molecular relationship between these TFs activity is poorly understood.
Members of the Mef2 family of MADS-box proteins were first characterized in vertebrates as important regulators downstream of the MyoD family of transcription factors, and have since been identified as part of an evolutionarily ancient regulatory network driving myogenesis from flies to man . In vertebrates, Mef2 transcription factors act as central regulators of cell proliferation, survival, apoptosis and differentiation in a range of cell types, including skeletal, cardiac and smooth muscle, brain, neural crest, lymphocytes and bone (reviewed in ). This diversity in Mef2 function is achieved through regulation by extracellular signals and cooperative activity with specific co-regulators. In skeletal muscle, for example, Mef2 acts together with bHLH transcription factors to regulate the expression program that drives myogenic differentiation , . In neural crest cells, Mef2c acts cooperatively with the DLX5 and DLX6 homeodomain TFs to regulate craniofacial development , , while in smooth muscle cells Mef2 acts together with myocardin . Thus, Mef2 TFs have little inherent instructive potential by themselves but rather act together with tissue-specific TFs to drive specific gene expression programs. Given the diverse roles of the Mef2 gene family during development, many more co-regulators are likely to be required to generate the spectrum of transcriptional responses elicited by these factors.
In Drosophila embryos, the single Mef2 ortholog is expressed exclusively in mesoderm and its muscle derivatives. Even in this relatively simple model system, Mef2 regulates distinct batteries of target genes in precise spatial patterns (e.g. in the dorsal vessel, the somatic mesoderm and visceral mesoderm , ), and in a specific temporal order , . Global in vivo occupancy experiments revealed dynamic Mef2 enhancer binding; although Mef2 is expressed continuously, it binds to one group of enhancers only early in development and to another group only at late developmental stages . The temporal shift in the expression onset of Mef2 target genes ,  as well as their spatial diversity, indicates a requirement for co-regulators, similar to the mechanism of Mef2 action in vertebrates . holes-in-muscle (him) was recently identified as a potential repressor of Mef2-dependent transcriptional activation via the recruitment of the general co-repressor Groucho . Regulation by Him therefore provides one mechanism to alter the temporal output of Mef2 activity once it is bound to an enhancer. However, other co-regulators are clearly required to modulate Mef2's temporal enhancer occupancy and to restrict its spatial activity.
The Drosophila body wall muscle or somatic muscle is formed from two heterogeneous populations of cells- the founder cells (FCs), which represent 30 distinct cells in each hemisegment of the embryo, and the fusion competent myoblasts (FCMs) . Once specified, a single FC will fuse with a defined number of FCMs to give rise to a syncytial myotube of distinct identity, defined by its size, shape and attachments. Mef2 is required to initiate a program that regulates myoblast fusion and drives the differentiation program of the resulting myotube into a contractile myofiber . The zinc-finger transcription factor Lame duck maintains Mef2 expression in FCMs, and like Mef2, is also essential to regulate a program of muscle differentiation, the first step of which is myoblast fusion , , . lmd mutant embryos have defects in the specification or maintenance of FCMs , which results in an expansion of Zfh1-expressing pericardial cells and adult muscle precursor-like cells . In contrast to Mef2, the molecular function of Lmd is more poorly understood; its only known direct target gene being Mef2 itself .
Given the extensive co-expression of lmd and Mef2 and the similarity in the myoblast fusion phenotype observed in their loss-of-function mutants, we hypothesized that Lmd may act as an FCM-specific modulator of Mef2 activity. To assess this, we have systematically compared the in vivo enhancer occupancy of Lmd and Mef2 and identified a large number of combinatorially bound enhancers during myogenesis. Expression profiling of loss-of-function lmd and Mef2 mutants revealed that, although these TFs co-occupy the same enhancer region, they have different regulatory effects on the expression of the target genes. We used a combination of in vivo and in vitro approaches to demonstrate differential integration of regulatory input from Lmd and Mef2 at individual CRMs. Taken together, these data emphasize the diversity of transcriptional responses that can be generated by two transcription factors and identify Lmd as a new context-specific modulator of Mef2 activity.
Obtaining a systematic map of Lmd-bound enhancer regions in vivo
As a first step towards understanding the phenotype of lmd mutant embryos and its potential combinatorial regulation with Mef2, we set out to identify Lmd-bound enhancer regions and directly regulated target genes. To identify Lmd-bound enhancers within the developing embryo, we used chromatin immunoprecipitation followed by microarray analysis (ChIP-on-chip) during defined stages of muscle development. Lmd-associated DNA was precipitated from tightly staged embryos at two consecutive developmental time points, spanning most of the developmental stages when lmd is expressed (stages 10–13). To obtain data with high sensitivity and specificity we performed a total of eight independent chromatin immunoprecipitations per time-point using two different anti-Lmd antibodies for each time point (Materials and Methods). The enriched DNA sequences were analyzed on microarrays containing overlapping 3 kb fragments tiling across ∼50% of the Drosophila genome . Genomic regions were considered bound by Lmd if they were significantly enriched with both antibodies, thereby reducing potential false targets caused by non-specific antibody effects.
Lmd binding was detected at 154 unique genomic regions at one or both developmental time points (Table S1), including the only known Lmd-binding site upstream of the Mef2 locus (; Figure 1A). In addition, Lmd binds to a previously characterized enhancer of sns (; Figure 1B), a transmembrane protein that requires lmd for its expression in FCMs  indicating that Lmd directly regulates sns expression. The expression of the bHLH transcription factor twist persists longer in lmd loss-of-function mutants than in wild-type embryos . As the DNA-binding domain of Lmd is similar to the Gli-family of transcription factors, which can act both as transcriptional activators and repressors, it was proposed that Lmd may directly repress Twist . However, no significant Lmd-binding was detected in the twist locus (data not shown). Although we cannot exclude low-level Lmd occupancy below the detection limit of our assay, this result suggests an indirect regulatory connection between Lmd and twist.
(A) Schematic overview of a genomic region within an intron of the Mef2 locus: An exon (located on the antisense strand) is shown in orange. Genomic fragments on the tiling arrays are indicated as stacks of two horizontal bars in their corresponding genomic position. The top bar represents the results from the 6–8 hour (stages 10–11) ChIP-on-chip time point, the lower one corresponds to the 8–10 hour (stages 12–13) time point. Significantly enriched regions are indicated in blue for Lmd (top) and in green for Mef2 (bottom). The black bar indicates the location of the previously identified Lmd-binding site. Both Lmd and Mef2 are co-bound to genomic regions overlapping the known Lmd binding site at the 6–8 hr time point, as well as to other sites in this area. (B) Schematic overview of the genomic region upstream of the sns locus (exons show in orange): The known sns enhancer (black bar) partially overlaps with tiling array probes bound by Lmd (at both time points, blue bars) and Mef2 (at 8–10 hour time point, green bar). Additionally, Mef2 binds to other locations upstream and intronic of the sns locus. (C) Lmd and Mef2 co-occupy many genomic locations: Venn diagram displaying the number of non-overlapping regions significantly enriched in Lmd ChIPs (blue) and significantly enriched in Mef2 ChIPs at the same stages of development (green). Both factors co-occupy a large number of regions (overlap). (D) Co-regulation of common direct target genes by Lmd and Mef2: The majority of Lmd target genes (blue) are co-regulated by Mef2 (overlap), while Mef2 regulates a large number of additional genes (green). (E) Differential gene expression in lmd and Mef2 loss-of-function mutants [log2]: differences in expression between mutant and wt embryos were recorded in a timecourse for lmd (left) or Mef2 (right) mutant embryos. Shown are all direct target genes of Lmd that are significantly misregulated at one or more time points in either mutant background (fold change >1.6, q<1%). K-means clustering was used to highlight similar downregulation (top) in both mutants, similar upregulation (centre) or divergent expression changes (bottom). Color scale indicates fold changes [log2]. Genes studied in more detail are marked in grey.
The recovery of enhancers of both Mef2 and sns, genes known to be genetically downstream of lmd, underscores the accuracy of the ChIP-on-chip results. Moreover, a number of Lmd-bound regions overlap previously characterized muscle enhancers, including betaTub60D , Act57B , CG14687 and CG9416  (Figure S1 and Table S2) and are dependent on Lmd for their activity (see below). In addition, we have characterized the activity of four previously unknown Lmd-bound enhancers that are responsive to Lmd both in vivo (Figure 2) and in vitro (see below).
Left hand panels: Colorimetric in situ hybridization (black and white images). (A–I) In situ hybridization of wild-type embryos with probes specific for (A) βTub60D, (B) act57B, (C) CG5080, (D) blow, (E) sug, (F) sns, (G) CG14687, (H) CG9416 and (I) gol, detecting specific expression in the mesoderm. No specific staining was observed in the ectoderm (red brackets). An engrailed-Gal4 driver line was used to ectopically express (A′–I′) UAS-Lmd, (A″–I″) UAS-Mef2-HA or (A′″–I′″) both UAS-Mef2-HA and UAS-Lmd in ectodermal stripes. Lmd and Mef2 show differential ability to activate ectopic target gene expression (red arrows). Expression of 8 of 9 genes can be induced in the ectoderm when both factors are present (A′″–H′″), while gol expression was never observed in the ectoderm (I′″). Right hand panels: Double fluorescent in situ hybridization of the same genotypes with probes specific for the corresponding genes (green) and wingless (red). A high magnification of the ectodermal wingless staining reveals adjacent ectopic staining in the engrailed domain (white arrows), not detected in wild-type embryos. Full embryos are shown in Figure S2.
Combinatorial binding of Lmd and Mef2 to shared cis-regulatory modules
The activities of Lmd and Mef2 are required for the initiation of myoblast fusion, presumably due to the regulation of a battery of genes essential for this process or for the identity of the FCMs themselves. To investigate potential co-regulation of target genes, we compared the enhancer binding data from Lmd (presented here) to our previously reported ChIP-on-chip data for Mef2 performed at the same developmental time points . 106 out of 154 (68.8%) Lmd-bound regions are co-bound by Mef2 during muscle development (Figure 1C), suggesting that the majority of these regions are co-regulated by both TFs. This is likely to be a conservative estimate as regions bound by one or both transcription factors just below our thresholds are not considered. Nevertheless, the extensive level of enhancer co-occupancy (Figure 1C) indicates that combinatorial regulation by these two TFs is an important feature within the myogenic program. In many cases the temporal profile of Lmd and Mef2 binding to shared enhancers is identical, again indicating that these two TFs act together to co-regulate enhancer output. For example, both TFs only bind to the ttk (tramtrack) enhancer at stages 10–11, but not later, while the blow (blown fuse) and CG5080 enhancers are co-bound at stages 12–13, and not earlier (Figure 3).
(A–D) Schematic overviews of the tiling array probes covering the (A) blow, (B) CG5080, (C) gol and (D) ttk loci. Regions significantly enriched in Lmd ChIPs (blue) and Mef2 (green) were selected and intergenic sequences (black bars) assayed for regulatory activity in vivo. (A′–D″) Immuno-histochemistry using an anti-GFP antibody to detect reporter gene expression. All four enriched sequences were able to specifically activate GFP expression in transgenic embryos in the mesoderm as early as (A′–D′) stage 11, with persistent GFP-presence at (A″–D″) stage 13.
Differential requirements of Lmd and Mef2 activity for target gene expression
The combinatorial binding of Lmd and Mef2 to shared CRMs raises several interesting questions. How much regulatory input does each factor contribute to the activation of an enhancer? Are Lmd and Mef2 acting in a co-operative or additive manner to regulate target gene expression? Are both transcription factors required for enhancer activation, or do they act redundantly? We have used several approaches, both in vivo and in vitro, to address these questions. First, we used global expression profiling to determine which genes require Lmd and Mef2 activity for their correct expression in vivo (Figure 1E).
We performed a developmental time-course of gene expression, comparing the transcriptional state of wild-type embryos to that of lmd-mutant embryos at six consecutive one-hour windows of development, providing a high-resolution map of lmd-dependent changes in gene expression. These experiments identified 640 genes that are genetically downstream of lmd during the stages of myoblast fusion and the initiation of terminal muscle differentiation (Table S3). By integrating this differential gene expression data with information on Lmd enhancer occupancy (ChIP-chip data) and muscle-specific gene expression patterns (from BDGP in situ hybridizations, ) we defined a high-confidence set of 74 target genes  that are likely to be directly regulated by Lmd (Table S1 and Table S4). Among these are a number of genes known to be involved in myoblast fusion, including Mef2, sns and blow, as well as genes with characterized roles in other aspects of muscle development, suggesting that Lmd may have a broader role in myogenesis than previously anticipated.
In a previous study, we used ChIP-on-chip experiments and expression profiling to identify a stringent set of Mef2 direct target genes at multiple stages of development . Comparing these data to that of Lmd revealed that a large (79.7%) and highly significant (p<2.2×10−16, Fisher's exact test) proportion of Lmd direct target genes are also directly regulated by Mef2 (Figure 1D). Thus the majority of the regulatory input provided by Lmd is mediated in conjunction with Mef2, which is not the case in the other direction. Mef2 regulates many target genes independently of Lmd, reflecting its broader expression and role in muscle development (Table S4).
To assess the regulation of Lmd direct target genes we first examined their transcriptional response to loss of regulatory input from either Lmd or Mef2. 57 of the 74 Lmd direct target genes are differentially expressed in lmd and/or Mef2 mutant embryos compared to stage-matched wild-type controls (visualized by K-means clustering in Figure 1E). This analysis revealed an unanticipated diversity in transcriptional responses, despite the fact that the majority of genes have an enhancer bound by both transcription factors. One group of genes (cluster I, Figure 1E) is downregulated in both lmd and Mef2 mutants compared to the stage matched wild-type embryos. This group contains many genes coding for structural muscle proteins, including Act57B, Act87E and betaTub60D (Figure 1E). As Mef2 expression is also strongly reduced in lmd mutant embryos , these target genes either depend on input from Mef2 alone or on a combination of Mef2- and Lmd-mediated activation. Other genes are affected differently in the two mutants. Several are upregulated in lmd mutants (e.g. CG9416 and CG30035), but are either unchanged or have slightly decreased or increased levels in Mef2 mutants (cluster II, Figure 1E). In contrast, a third cluster of genes, including blow, goliath (gol) and tramtrack (ttk), have decreased expression at the late time points in lmd mutants and increased expression in Mef2 mutants, suggesting activation by Lmd and repression by Mef2 (Figure 1E). We note that, although we have used several methods to assess the role of Mef2 in regulating the expression of these genes (see below), we have not found any evidence that Mef2 may act as a transcriptional repressor. Therefore the apparent de-repression of these genes is most likely due to a secondary effect within the Mef2 mutant embryos. Despite this, the vast majority of genes known to be genetically downstream of Mef2 had significantly reduced expression, indicating that the expression profiling data accurately recapitulates what is expected from genetic studies .
Different combinations of Lmd and Mef2 trigger divergent gene expression
As a complementary approach to assess the regulatory inputs of Lmd and Mef2, we asked if these transcription factors are sufficient, either alone or in combination, to induce target gene expression in vivo. The transcription factors were ectopically expressed in parasegmental stripes under the control of the engrailed-Gal4 driver  (Figure 2). Lmd has been reported to activate Mef2 expression in the CNS but not in the remainder of the ectoderm under these conditions , allowing us to assess the contribution of the two transcription factors independently. As the transcription factors are acting outside of their normal cellular context, this is a stringent assay to investigate regulatory connections.
The transcriptional response of shared target genes to ectopic TF expression was examined using colorimetric in situ hybridization (ISH) (Figure 2), and confirmed by double fluorescent ISH (Figure 2, Figure S2). This analysis revealed a range of regulatory responses. We examined three genes that showed reduced expression in both lmd and Mef2 mutant embryos (Figure 1E, cluster I). betaTub60D and Act57B are ectopically induced by Mef2 alone, but not by Lmd alone (Figure 2A–2B″). As expected, co-expression of both transcription factors also led to ectopic expression (Figure 2A′″, 2B′″). A third gene, CG5080, was neither ectopically activated by Lmd nor Mef2 alone (Figure 2C′–2C″). However, when both transcription factors were co-expressed, their combined activity was sufficient to drive ectopic expression, revealing a synergistic regulation of this target gene (Figure 2C′″). Ubiquitous over-expression of Mef2 using a daughterless-Gal4 driver was previously reported to ectopically activate CG5080 in the head mesoderm . The fact that Mef2 is sufficient to regulate CG5080 expression in this context, but not in ectodermal stripes, strongly suggests that Mef2 requires additional tissue-specific co-activators also in other tissues of the embryo.
The expression levels of blow and sug were also strongly reduced in lmd mutants, and weakly reduced in Mef2 mutants (Figure 1E, cluster III). Similar to CG5080, neither expression of lmd nor Mef2 alone was sufficient to activate expression of blow, sug or sns, yet ectopic activation was detected upon co-expression of both transcription factors (Figure 2D–2F′″). Although Mef2 is not required for sns expression , our data demonstrates that Mef2, in combination with Lmd, is sufficient to activate the expression of sns in ectodermal cells. CG14687 showed the opposite response to bTub60D and act57B, in that it could be activated by Lmd alone, but not by Mef2 (Figure 2G′–2G″). These data correlate with the expression profiling data, showing a strong requirement of lmd activity for CG14687 expression (Figure 1E, cluster IV). Although in situ hybridization is not quantitative, the fluorescent ISH suggests a higher level of expression when both Lmd and Mef2 are co-expressed (Figure 2G, fluorescent panels, Figure S2).
The gene CG9416 revealed yet another mode of regulatory integration: Mef2 activated ectopic expression of CG9416 in the absence of Lmd, but this effect appears to be attenuated when both transcription factors were co-expressed (Figure 2H–2H′″, fluorescent panel, Figure S2), indicating opposing regulatory inputs from Mef2 (activation) and Lmd (inhibition). The repressive effect of Lmd is consistent with the dramatic increase in CG9416 expression in lmd mutant embryos (Figure 1E). Finally, the gol gene represents the only example tested where even the combination of both Mef2 and Lmd was not sufficient to induce ectopic expression (Figure 2I′″).
In summary, although all genes investigated are directly co-regulated by Lmd and Mef2, ectopically supplying one or both factors revealed considerable flexibility in how information is integrated at each individual locus. In higher eukaryotes, many genes have multiple regulatory elements, which collectively contribute to the complete expression pattern of a gene. To investigate whether the different transcriptional responses to Lmd and Mef2 activity are reflected by the integration of inputs at single enhancers or by the combined activity of multiple cis-regulatory elements, we next studied regulatory integration at the CRM level.
Delimiting enhancer regions co-bound by Lmd and Mef2
Individual enhancer regions in Drosophila commonly range from 0.5 to 1 kb in size. The Lmd-bound DNA fragments immunoprecipitated in our ChIP experiments were in a similar size range, however the genomic tiling arrays used in this study limited our resolution to overlapping 3 kb sequences. To achieve higher resolution, we used quantitative real-time PCR to assay the enrichment of shorter sequences within individual 3 kb-bound regions using both the Lmd and Mef2 chromatin immunoprecipitates. In all eight cases examined, the highest enrichment of Mef2 and Lmd binding coincided within a common 0.1 to 1 kb region (data not shown), suggesting that the transcription factors co-occupy a single enhancer element. In addition, each refined sequence was found to contain at least one Mef2 consensus binding site conserved in Drosophila pseudoobscura (data not shown).
We tested the ability of the refined Lmd-Mef2-bound regions to regulate expression in vivo by generating transgenic reporter lines. All tested enhancer regions specifically activated GFP-reporter expression in the developing muscle (Figure 3 and Figure S1). At stage 11, when both lmd and Mef2 are co-expressed in fusion-competent myoblasts, the enhancers of blow (Figure 3A′) and gol (Figure 3C′) activated GFP-expression broadly in the visceral and somatic mesoderm. At the same stages, the CG5080 (Figure 3B′) and tramtrack enhancers (ttk, Figure 3D′) induced GFP-expression in a subset of myoblasts. At stage 13, when myoblast fusion is in progress, all four enhancers showed almost identical expression patterns throughout the somatic muscle (Figure 3A″–3D″). We also re-examined the spatio-temporal activity of the previously characterized Act57B , betaTub60D  and CG14687  enhancers (Table S2 and Figure S1) and included them in the set of combinatorially-bound enhancers investigated in the remainder of this study.
Lmd and Mef2 are differentially required for enhancer activity in vivo
We used the in vivo enhancer-reporter lines to study the integration of Lmd and Mef2 regulatory input by comparing CRM activity in wild-type and mutant embryos. Six transgenic reporter lines (Figure 3, Table S2) were placed in the genetic background of lmd1 and Mef222.21, two characterized loss-of-function alleles for these transcription factors ,  (Figure 4, Figure S3).
In situ hybridization of GFP-reporter mRNA in (A–F) wt embryos, (A′–F′) homozygous lmd1 mutant embryos and (A″–F″) homozygous Mef2P544 mutant embryos. The intronic βTub60D enhancer requires Lmd enhancer activity (A′), but remains active in Mef2 mutants (A″). Similarly, activity of the act57B enhancer is strongly dependent on Lmd (B′), but only mildly reduced in Mef2 mutants (B″). CG5080 reporter expression is clearly reduced in both background (C′,C″), while activity of the blow enhancer is not detectable in the absence of either Lmd or Mef2 (D′,D″). The CG14687 enhancer show different requirements in different muscle types: while expression in somatic muscles requires Lmd (E′), expression in the visceral muscle is not affected in neither Lmd nor Mef2 mutants (E″). The gol enhancer requires Lmd activity (F′) but is active normally in Mef2 mutants (F″).
The expression of the betaTub60D gene is controlled by several independent cis-regulatory modules , , . An upstream enhancer, 5′ to the betaTub60D gene, requires Mef2 activity for its full activation . In contrast, the intronic betaTub60D enhancer under study here, although co-occupied by Mef2 and Lmd, appeared unaffected in Mef2 mutant embryos while having strongly reduced expression in lmd mutants (Figure 4A′ and 4A″). The strong reduction in the expression of the betaTub60D gene in Mef2 and lmd mutant embryos detected by expression profiling (Figure 1E) therefore reflects the combined activity of at least two enhancers: one strongly responsive to Mef2 levels and a second one depending on Lmd (but not Mef2) for activation.
The Act57B enhancer drives GFP-expression in somatic and visceral muscles in wild-type embryos at stage 13 (Figure 4B). This expression was completely lost in lmd mutant embryos (Figure 4B′), while Mef2 mutant embryos showed reduced, but detectable reporter expression, as observed previously  (Figure 4B″). In contrast, expression driven by the CG5080 enhancer was reduced in lmd (Figure 4C and 4C′) and to a lesser extent in Mef2 mutant embryos (Figure 4C and 4C″). Similarly, reporter expression in the somatic muscle driven by the blow enhancer was lost in both lmd and Mef2 mutant embryos (Figure 4D–4D″). Enhancer expression in the hindgut visceral muscle persisted in Mef2 mutant embryos (Figure 4D″), indicating additional tissue-specific input at this enhancer.
The CG14687 enhancer is activated in somatic and visceral muscle in wild-type embryos (Figure 4E). Expression in somatic muscle required lmd expression (Figure 4E′), but is unaffected in Mef2 mutant embryos (Figure 4E″). Interestingly, expression in the visceral muscle was independent of both lmd and Mef2 expression (Figure 4E′ and 4E″), implicating additional tissue-specific factors in the activation of this enhancer. Both the homeodomain transcription factor bagpipe (bap) and the fork head domain transcription factor biniou are recruited to this enhancer in vivo  and most likely activate gene expression in this tissue. Finally, the gol enhancer required lmd activity (Figure 4F and 4F′), but robustly activated gene expression in the absence of Mef2 (Figure 4F″).
In summary, all six muscle enhancers examined showed reduced activity in one or both mutant conditions, demonstrating that the in vivo occupancy of these modules by Mef2 and Lmd has regulatory function. lmd mutants generally displayed a stronger reduction in enhancer activity compared to Mef2 mutant embryos. As Lmd is required to maintain Mef2 expression, lmd mutant embryos are effectively double mutants for both transcription factors. This is reflected by the stronger reduction in enhancer activity in this genetic background and underscores the combinatorial regulation of these enhancers by both transcription factors.
Cooperative, additive, and repressive effects of Lmd and Mef2 on CRM activity
We next assessed if the combined regulatory inputs of Lmd and Mef2 on these enhancers are integrated in an additive, cooperative or repressive manner. Drosophila S2 cells, which express neither endogenous lmd nor Mef2 , were used to study the regulatory logic of the different enhancers in vitro. Eight regulatory regions that are co-bound by Lmd and Mef2 in vivo were placed upstream of a minimal Hsp70 promoter driving a firefly luciferase reporter and co-expressed with increasing amounts of Lmd and/or Mef2 expression vectors.
Co-transfection of either Lmd or Mef2 alone was sufficient to activate the CG14687, CG5080 and gol enhancers (Figure 5A–5C), while co-expression of both regulators led to an approximately additive level of reporter activity. For example, transfection of 10 ng of the Lmd expression vector led to a 3.6 fold increase in the luciferase activity driven by the CG14687 enhancer, while 1 ng of the Mef2 expression vector led to a 2.4 fold increase in expression. Co-expression of both factors resulted in a 5 fold increase in enhancer activity (Figure 5A).
Minimal regions required for Firefly luciferase reporter activity in vitro were identified in the (A) CG14687, (B) CG5080, (C) gol, (D) ttk, (E) blow, (F) βTub60D, (G) CG9416 and (H) CG30035 enhancers which were cloned and assayed for activity in vitro. Expression plasmids encoding for Lmd or Mef2 proteins were co-transfected with the reporter plasmids and a Renilla normalization control at different concentrations (1 ng/10 ng) alone or in combination (x-axis). Dual-luciferase readout was normalized to reporter-only controls and fold changes are indicated as mean +/− 1 standard error (at least three biological replicates, each done in triplicate). Informative combinations of transcription factor transfections are indicated (brackets). (A–C) Both Lmd and Mef2 can activate the CG14687, CG5080 and gol reporters. Co-transfection of both transcription factors leads to roughly additive fold changes (brackets). (D–F) Presence of both Lmd and Mef2 yields higher activity from the ttk, blow and Tub60D enhancers than expected by summing the individual fold changes (brackets), indicating cooperative regulation. (G, H) The CG9416 and CG30035 reporters can readily be activated by Mef2, but show reduced activity upon co-transfection with Lmd, revealing its inhibitory influence in this context. The regulatory interactions are highly significant (unpaired, two-tailed student′s t-test, (*) p<0.05, (**) p<0.01, (***) p<0.001).
In contrast, Lmd and Mef2 acted cooperatively to regulate the ttk, blow and betaTub60D enhancers (Figure 5D–5F). For example, expression of either Lmd or Mef2 alone yielded only low levels of reporter gene activity via the blow enhancer (Figure 5E). However, co-expression of both transcription factors resulted in much higher levels of activity, indicating a cooperative interaction between Lmd and Mef2 in the context of this enhancer.
Conversely, the CG9416 enhancer is readily activated by Mef2, but cannot be induced by Lmd (Figure 5G). Instead, co-expression of both transcription factors revealed that Lmd counteracts the positive input of Mef2 to this module, essentially blocking activation by Mef2 in a dose dependent manner. This repressive activity, in combination with the in vivo occupancy of Lmd on this enhancer (Figure S1), the increase in CG9416 gene expression in lmd mutant embryos (Figure 1E) and the ability of Lmd to attenuate the ectopic activation of CG9416 by Mef2 (Figure 2H, fluorescent panels), provides strong evidence that Lmd can provide direct inhibitory input to enhancer activity. Similar to CG9416, the expression of endogenous CG30035 was de-repressed in lmd mutant embryos (Figure 1E). The Lmd-Mef2 bound enhancer region close to the CG30035 locus displayed a similar dose-dependent inhibitory effect of Lmd on Mef2-mediated transcriptional activation (Figure 5H).
Collectively, our results demonstrate that Lmd and Mef2 can induce different regulatory responses depending on the context of the enhancer. This may reflect differences in the relative positioning of Mef2 and Lmd binding to each other or the recruitment of additional unknown factors. As there is no consensus binding site known for Lmd, we used de novo motif discovery  to identify possible Lmd binding motifs. Since we observed that Lmd and Mef2 are commonly bound within close proximity to each other, we reduced the search space to a 400 bp window around each predicted Mef2 site within the group of 57 co-bound regions. This analysis did not reveal any candidate motifs matching the only known site occupied by Lmd , precluding further analysis of individual Lmd binding sites.
Metazoan cells must activate and inactivate the expression of large cohorts of genes in a precise spatio-temporal manner to progress through development. To achieve a molecular understanding of the regulatory networks controlling cellular decision-making, it is essential to understand how inputs from different regulators are being integrated to give rise to defined patterns of gene expression. In this study, we approached this challenge from a genomic perspective by examining the combinatorial input of two key myogenic regulators, Mef2 and Lmd. ChIP-on-chip experiments and expression profiling of loss-of-function mutants were used to systematically identify the direct target genes of the zinc-finger protein Lmd, an important regulator of myogenesis, for which only a single target gene had previously been identified. Integrating these data with data previously obtained for Mef2 revealed that Lmd regulates the majority of its targets in a combinatorial manner together with Mef2. In a few cases these two transcription factors target the same locus through different regulatory regions (e.g. ladybird-early, PAK-kinase or short stop), however in the majority of cases Lmd- and Mef2-binding could be mapped to the same genomic location (Table S1, Table S4). Examining the contribution of both Lmd and Mef2 to regulatory activity, at both the enhancer and gene level, revealed a number of important insights into the contribution of both transcription factors to the myogenic developmental program.
Combinatorial binding to enhancers leads to diverse regulatory responses
Genes that are co-regulated by the same two (or more) transcription factors are generally expected to have very similar spatio-temporal expression profiles. In fact, this assumption has been used by many studies to computationally predict the location of enhancer elements by searching for common TF binding motifs in the vicinity of clusters of co-expressed genes (or synexpression groups) –. It was therefore surprising when our comparison of experimentally-identified enhancer regions bound by the same two transcription factors uncovered a diverse range of regulatory responses. The 59 genes with enhancer elements co-bound by Lmd and Mef2 at the same stages of development are regulated either in a cooperative, additive or repressive manner depending on the individual enhancers. These data suggest that enhancer regions integrate regulatory inputs more flexibly than previously anticipated. By focusing on individual enhancer elements, we evaluated how Lmd and Mef2 influence regulatory activity in different contexts both in vivo and in vitro. Combining a number of complementary approaches allowed us to identify three different modes of TF integration at developmental enhancers leading to additive, cooperative or repressive regulation.
Lmd and Mef2 operate under additive, cooperative, and repressive regulatory logic
Mef2 and Lmd provide an additive positive input to the regulation of the Act57B locus. Ectopic Mef2 expression in the ectoderm is sufficient to induce Act57B expression, while providing Lmd alone is not (Figure 2B–2B″). Conversely, enhancer-reporter gene expression is completely lost in lmd mutant embryos and only slightly reduced in Mef2 loss-of-function mutant embryos (Figure 4B–4B″). Together, these data reveal a role for both transcription factors at this enhancer. Previous studies demonstrated that the initiation of Act57B expression at stage 11 requires Mef2 for its activation. Yet, artificially increasing Mef2 levels at this stage does not lead to premature activation of this locus . Our findings offer an explanation for this result: at this stage of development, combined input from Lmd and Mef2 is required to drive gene expression, while the presence of Mef2 alone is not sufficient to activate transcription. At later stages, when lmd expression is lost, Mef2 concentration has increased sufficiently to maintain Act57B expression. Conversely, the CG14687 locus can be activated by ectopic Lmd in the ectoderm, but not by Mef2 alone (Figure 2G′–2G′″) and requires lmd, but not Mef2, for its expression in the somatic muscle (Figure 4E′–4E″). Combined ectopic expression of the two TFs, on the other hand, leads to a marked increase of reporter signal, again indicating combinatorial positive regulation by both TFs (Figure 2G, fluorescent panels). These findings are supported by the ability of both Lmd and Mef2 to separately activate reporter gene expression in vitro and to yield additive reporter activity in combination (Figure 5A).
The blow enhancer shows a different mode of regulation and is synergistically activated by both factors. While neither Mef2 nor Lmd alone are sufficient to activate ectopic gene expression in vivo, supplying both factors simultaneously leads to robust target gene expression (Figure 2D–2D′″). Assaying for reporter gene activation in the two mutant backgrounds yields a complementary result; Mef2 and Lmd activity is required to activate transcription in the somatic mesoderm via the blow enhancer (Figure 4D–4D″). Moreover, the in vitro reporter assay reveals a positive interaction between the two proteins (Figure 5E), indicating that the blow enhancer functions as a cooperative switch.
Analysis of the CG9416 enhancer revealed an antagonistic interaction between Lmd and Mef2. While ectopic expression of Mef2 leads to enhancer activation (Figure 2H–2H″), simultaneous expression of Lmd markedly attenuates the transcriptional output from this locus (Figure 2H, fluorescent panels, Figure S2H). This effect can be reproduced in vitro: while providing Mef2 alone leads to robust activation of the CG9416 enhancer, Lmd is not able to activate gene expression (Figure 5G). Instead, Lmd antagonizes the activating input of Mef2 in a concentration-dependent manner. To our knowledge, this is the first example of direct negative regulation by Lmd. To identify additional examples of a repressive role for Lmd, we re-examined the expression profiles of lmd and Mef2 mutant embryos (Figure 1E). CG9416 is markedly upregulated in lmd mutants, but shows reduced expression in embryos lacking Mef2 (Figure 1E). We selected another direct target gene with similar expression changes in these genetic backgrounds, CG30035 (Figure 1E) and after determining the limits of the ChIP-enriched region we assayed its ability to drive reporter gene expression in vitro. Similar to the CG9416 enhancer, the CG30035 enhancer is robustly activated by Mef2, and this activation is inhibited by Lmd in a dose-dependent manner (Figure 5H). This provides a second, independent example for Lmd-mediated repression of gene expression.
In summary, starting from a genomic perspective, we have identified a large cohort of genes co-regulated by a pair of tissue-specific transcription factors. Lmd modulates the activity of Mef2 at different enhancers in a context-dependent fashion, allowing for additive, cooperative or antagonistic interactions in the same cells. In this way, the timing and expression levels of Mef2 target genes can be further refined, as exemplified by the Act57B locus, which may owe its early activation during embryonic development to the combined activity of both proteins. Lmd shows homology with the Gli superfamily of transcription factors , which can act both as transcriptional activators and repressors, depending on proteolytic cleavage regulated by the hedgehog signaling pathway. To date, there is no evidence for proteolytic cleavage of Lmd and an irreversible conversion of Lmd from a transcriptional activator to an inhibitor is difficult to reconcile with our observation that Lmd can perform both roles at different loci at the same time, in the same tissue. For the same reason, we also consider it unlikely that Lmd interferes with transcriptional activation simply by binding to Mef2 and sequestering the protein in the cytoplasm. Instead, we propose that Lmd exerts a dominant inhibitory influence over a transcriptional activator, either by locally quenching Mef2's activity or through direct repression of the locus, similar to transcriptional repressors described in other developmental networks , . Our results provide a molecular understanding for the genetic observation that restoring Mef2 activity in lmd mutant embryos is not sufficient to rescue muscle differentiation . Both transcription factors are required to provide different regulatory inputs to a large number of co-regulated target genes during myogenesis. Their associated enhancers have revealed considerable flexibility in integrating regulatory inputs from these two TFs at individual cis-regulatory regions.
Materials and Methods
Chromatin immunoprecipitation and DNA amplification
Embryo collections and chromatin immunoprecipitations were performed as described previously , . Two antisera were raised against the amino terminus of Lmd and purified from E. coli by poly-His tag affinity purification. Four independent staged wild-type embryo populations were collected at 6–8 and 8–10 hrs after egg-laying and fixed with formaldehyde. For each time point, chromatin from all four populations was precipitated with both antisera as well as the respective preimmunesera, leading to a total of 16 reactions (8 mock, 8 anti-Lmd) per time point. DNA amplification, labeling and hybridizations were performed as described previously ,  and dye swaps were included to account for possible dye biases.
Expression profiling of lmd loss-of-function mutants
The assayed lmd1  line was outcrossed to wild-type flies (Canton S) twice to remove any spurious mutants. Six one-hour embryo collections were assayed in an expression profiling timecourse (between 5 and 11 hours after egg-laying). At each time point, 4 independent populations of lmd mutant and stage-matched Canton S embryos were collected and aged. Homozygous mutants were selected with an automated embryo sorter , . The staging of all collections was verified by formaldehyde fixation of a small sample to ensure that wild-type and mutant embryos were tightly stage matched. Total RNA was extracted using Trizol (Invitrogen, Carlsbad, US), amplified, reverse-transcribed and labeled as described previously .
Microarray data analysis
For expression profiling analysis, mutant and stage-matched control cDNA was hybridized directly against each other. Raw data was normalized using print-tip LOESS. Differentially expressed genes were identified using Significance analysis of microarrays (SAM) . Genes with a q<1% and a fold change >1.6 (log2>0.7 or <−0.7) were considered to be differentially regulated (Table S5). Immunoprecipitated DNA from Lmd-specific or mock precipitations was hybridized against a total genomic reference DNA sample. Sequences significantly enriched by the anti-Lmd-antibodies were identified by comparing rank products  and the false-discovery rate was estimated. Only fragments with an FDR <2% and a fold enrichment >1.5 (log2 >0.58 or <−0.58) were considered to be significantly enriched (Table S1).
Automatic assignment of ChIP-enriched fragments to target genes was performed as described previously . The majority of regions co-occupied by Mef2 and Lmd was independently assigned to the same target genes using either Mef2-mutant or lmd-mutant expression profiling data. For a small number of regions, data from this study indicated a more likely target gene than had been assigned previously with Mef2 data alone ; in these cases, we chose the updated target prediction for further analysis. A complete list of ChIP-enriched regions, expression profiling results and target assignments are available in Tables S1, S3, S4, S5. All raw microarray data is available from ArrayExpress (Lmd ChIP (E-TABM-895) and lmd expression profiling (E-TABM-894). Lmd- and/or Mef2-bound regions and mutant expression data can be visualized at http://furlonglab.embl.de/data/.
Generation of transgenic reporter strains
Fragments within the following coordinates (based on BDGP genome release 5) were cloned into the pH-stinger (AF242365) vector for germline transformation : chr2R:16831306-16831372 (actin57B), chr2R:20197035-20197429 (betaTub60D), chr3R:6619371-6620063 (CG14687), chr3R:27529661-27530409 (ttk), chr2R:3472616-3473387 (blow), chr3R:27538572-27539618 (ttk), chr2R:8813219-8814579 (sug), chr2R:20966587-20969610 (gol). For all constructs at least two independent transgenic lines were obtained and assayed. The UAS-Mef2 line used in this study has been described previously . The UAS-lmd line was previously referred to as UAS-gfl .
Double fluorescent in situ hybridizations were done as described previously . To minimize experimental differences, the embryo fixations and the in situ hybridizations were done in parallel and the confocal imaging was performed with identical laser and gain settings for each gene in the four genetic backgrounds. The following ESTs were used to generate digoxigenin or fluorescein-labeled probes: RE53159 (betaTub60D), LD04994 (act57B), LD34147 (CG5080), LP02193 (blow), LD36528 (sug), RE74890 (CG14687), RE28322 (CG9416), GH20973 (gol), AT15089 (twi) and RE02607 (wg). The full-length sns cDNA was a kind gift from S. Abmayr. GFP expression in transgenic animals was detected by immunohistochemistry with rabbit α-GFP antibody (Torrey Pines Biolabs) at a concentration of 1:500. Biotinylated secondary antibodies were used in combination with the Vector Elite ABC kit (Vector Laboratories).
Luciferase reporter assays
Drosophila S2 cells were transiently transfected using Cellfectin (Invitrogen). Lmd and Mef2 were expressed from full-length ESTs (LD47926 and GH24154, respectively) in pAc5.1 vector (Invitrogen). The enhancers (coordinates given above) were assayed in a pGL3 luciferase reporter vector (Promega) with an Hsp70 minimal promoter and the luciferase activity was normalized to Renilla standard (Promega). The total amount of transfected DNA was kept constant by supplementing empty pAc5.1 vector. The measurements were performed according to the supplier's recommendations (Dual-Luciferase Reporter Assay, Promega) with a PerkinElmer 1420 Luminescence Counter.
Previously identified enhancer regions are co-occupied by Lmd and Mef2 and reproduce target gene expression in vivo. (A–D) Schematic overviews of the tiling array probes covering the (A) act57B, (B) Tub60D, (C) CG9416, and (D) CG14687 loci. These previously described regulatory regions (black bars) show significant binding of Lmd (blue) and Mef2 (green). (A″–D″) Immuno-histochemistry using an anti-GFP antibody to detect reporter gene expression. All four enriched sequences are able to specifically activate GFP-reporter expression in transgenic embryos in the mesoderm as early as (A′–D′) stage 11, with persistent GFP-presence at (A″–D″) stage 13.
(2.79 MB TIF)
Ectopic expression of Lmd and Mef2 reveals differential regulatory input on target gene expression. Double fluorescent in situ hybridisation (FISH) of the gene of interest (green) and endogenous wg gene (red). Ectopic expression (using the engrailed GAL4 driver) should be visible juxtaposed to the wg stripe. The wg expression was therefore used to ensure that the confocal imaging was performed with identical laser and gain settings for each gene within the four genetic backgrounds. (A–I) FISH of wild-type embryos with probes specific for (A) βTub60D, (B) act57B, (C) CG5080, (D) blow, (E) sug, (F) sns, (G) CG14687, (H) CG9416, and (I) gol, detecting specific expression in the mesoderm. No specific staining was observed in the ectoderm (white brackets). An engrailed-Gal4 driver line was used to ectopically express (A′–I′) UAS-Lmd, (A″–I″) UAS-Mef2-HA or (A″–I″) both UAS-Mef2-HA and UAS-Lmd in ectodermal stripes. Lmd and Mef2 show differential ability to activate ectopic target gene expression (white brackets). Note: The area indicated by the white brackets highlights the ectopic expression.
(9.37 MB TIF)
Enhancer activity in lmd and Mef2 loss-of-function mutant embryos. In situ hybridisations described in Figure 4 were performed by double-staining of GFP (green, first and third column, indicating specific reporter activity) and either endogenous twist mRNA (red, in lmd mutants) or lacZ mRNA expressed from a balancer (red, in Mef2 mutants) to identify homozygous mutant embryos. Twist expression persists longer in lmd mutants (C′. G′, K′, O′, S′, W′) than in wt embryos (A′, E′, I′, M′, Q′, U′). LacZ expression is associated only with heterozygous, balancer containing embryos (B′, F′, J′, N′, R′, T′, X′) (or embryos carrying two balancers).
(4.22 MB TIF)
ChIP-on-chip enriched fragments.
(0.08 MB XLS)
Overlap of ChIP data with previously characterised enhancer regions.
(0.04 MB PDF)
Lmd expression profiling timecourse.
(0.15 MB XLS)
Direct targets genes of Lmd and Mef2.
(0.04 MB XLS)
We are very grateful to all members of the Furlong lab for critical reading of the manuscript.
Conceived and designed the experiments: PMFC TS EHG LC EEMF. Performed the experiments: PMFC TS EHG LC MPE. Analyzed the data: PMFC TS EHG LC MPE EEMF. Wrote the paper: TS EEMF.
- 1. Bonn S, Furlong EE (2008) cis-Regulatory networks during development: a view of Drosophila. Curr Opin Genet Dev 18: 513–520.S. BonnEE Furlong2008cis-Regulatory networks during development: a view of Drosophila.Curr Opin Genet Dev18513520
- 2. Bour BA, O'Brien MA, Lockwood WL, Goldstein ES, Bodmer R, et al. (1995) Drosophila MEF2, a transcription factor that is essential for myogenesis. Genes Dev 9: 730–741.BA BourMA O'BrienWL LockwoodES GoldsteinR. Bodmer1995Drosophila MEF2, a transcription factor that is essential for myogenesis.Genes Dev9730741
- 3. Duan H, Skeath JB, Nguyen HT (2001) Drosophila Lame duck, a novel member of the Gli superfamily, acts as a key regulator of myogenesis by controlling fusion-competent myoblast development. Development 128: 4489–4500.H. DuanJB SkeathHT Nguyen2001Drosophila Lame duck, a novel member of the Gli superfamily, acts as a key regulator of myogenesis by controlling fusion-competent myoblast development.Development12844894500
- 4. Ruiz-Gomez M, Coutts N, Suster ML, Landgraf M, Bate M (2002) myoblasts incompetent encodes a zinc finger transcription factor required to specify fusion-competent myoblasts in Drosophila. Development 129: 133–141.M. Ruiz-GomezN. CouttsML SusterM. LandgrafM. Bate2002myoblasts incompetent encodes a zinc finger transcription factor required to specify fusion-competent myoblasts in Drosophila.Development129133141
- 5. Potthoff MJ, Olson EN (2007) MEF2: a central regulator of diverse developmental programs. Development 134: 4131–4140.MJ PotthoffEN Olson2007MEF2: a central regulator of diverse developmental programs.Development13441314140
- 6. Molkentin JD, Black BL, Martin JF, Olson EN (1995) Cooperative activation of muscle gene expression by MEF2 and myogenic bHLH proteins. Cell 83: 1125–1136.JD MolkentinBL BlackJF MartinEN Olson1995Cooperative activation of muscle gene expression by MEF2 and myogenic bHLH proteins.Cell8311251136
- 7. Penn BH, Bergstrom DA, Dilworth FJ, Bengal E, Tapscott SJ (2004) A MyoD-generated feed-forward circuit temporally patterns gene expression during skeletal muscle differentiation. Genes Dev 18: 2348–2353.BH PennDA BergstromFJ DilworthE. BengalSJ Tapscott2004A MyoD-generated feed-forward circuit temporally patterns gene expression during skeletal muscle differentiation.Genes Dev1823482353
- 8. Verzi MP, Agarwal P, Brown C, McCulley DJ, Schwarz JJ, et al. (2007) The transcription factor MEF2C is required for craniofacial development. Dev Cell 12: 645–652.MP VerziP. AgarwalC. BrownDJ McCulleyJJ Schwarz2007The transcription factor MEF2C is required for craniofacial development.Dev Cell12645652
- 9. Miller CT, Swartz ME, Khuu PA, Walker MB, Eberhart JK, et al. (2007) mef2ca is required in cranial neural crest to effect Endothelin1 signaling in zebrafish. Dev Biol 308: 144–157.CT MillerME SwartzPA KhuuMB WalkerJK Eberhart2007mef2ca is required in cranial neural crest to effect Endothelin1 signaling in zebrafish.Dev Biol308144157
- 10. Wang Z, Wang DZ, Pipes GC, Olson EN (2003) Myocardin is a master regulator of smooth muscle gene expression. Proc Natl Acad Sci U S A 100: 7129–7134.Z. WangDZ WangGC PipesEN Olson2003Myocardin is a master regulator of smooth muscle gene expression.Proc Natl Acad Sci U S A10071297134
- 11. Sandmann T, Jensen LJ, Jakobsen JS, Karzynski MM, Eichenlaub MP, et al. (2006) A temporal map of transcription factor activity: mef2 directly regulates target genes at all stages of muscle development. Dev Cell 10: 797–807.T. SandmannLJ JensenJS JakobsenMM KarzynskiMP Eichenlaub2006A temporal map of transcription factor activity: mef2 directly regulates target genes at all stages of muscle development.Dev Cell10797807
- 12. Junion G, Jagla T, Duplant S, Tapin R, Da Ponte JP, et al. (2005) Mapping Dmef2-binding regulatory modules by using a ChIP-enriched in silico targets approach. Proc Natl Acad Sci U S A 102: 18479–18484.G. JunionT. JaglaS. DuplantR. TapinJP Da Ponte2005Mapping Dmef2-binding regulatory modules by using a ChIP-enriched in silico targets approach.Proc Natl Acad Sci U S A1021847918484
- 13. Elgar SJ, Han J, Taylor MV (2008) mef2 activity levels differentially affect gene expression during Drosophila muscle development. Proc Natl Acad Sci U S A 105: 918–923.SJ ElgarJ. HanMV Taylor2008mef2 activity levels differentially affect gene expression during Drosophila muscle development.Proc Natl Acad Sci U S A105918923
- 14. Liotta D, Han J, Elgar S, Garvey C, Han Z, et al. (2007) The Him gene reveals a balance of inputs controlling muscle differentiation in Drosophila. Curr Biol 17: 1409–1413.D. LiottaJ. HanS. ElgarC. GarveyZ. Han2007The Him gene reveals a balance of inputs controlling muscle differentiation in Drosophila.Curr Biol1714091413
- 15. Baylies MK, Michelson AM (2001) Invertebrate myogenesis: looking back to the future of muscle development. Curr Opin Genet Dev 11: 431–439.MK BayliesAM Michelson2001Invertebrate myogenesis: looking back to the future of muscle development.Curr Opin Genet Dev11431439
- 16. Furlong EE, Andersen EC, Null B, White KP, Scott MP (2001) Patterns of gene expression during Drosophila mesoderm development. Science 293: 1629–1633.EE FurlongEC AndersenB. NullKP WhiteMP Scott2001Patterns of gene expression during Drosophila mesoderm development.Science29316291633
- 17. Sellin J, Drechsler M, Nguyen HT, Paululat A (2009) Antagonistic function of Lmd and Zfh1 fine tunes cell fate decisions in the Twi and Tin positive mesoderm of Drosophila melanogaster. Dev Biol 326: 444–455.J. SellinM. DrechslerHT NguyenA. Paululat2009Antagonistic function of Lmd and Zfh1 fine tunes cell fate decisions in the Twi and Tin positive mesoderm of Drosophila melanogaster.Dev Biol326444455
- 18. Stute C (2004) The determination of the founder and fusion competent myoblasts of the visceral mesoderm of Drosophila melanogaster depends on Notch as well as on Jeb/Alk mediated RTK-signalling. Marburg: Universität Marburg. C. Stute2004The determination of the founder and fusion competent myoblasts of the visceral mesoderm of Drosophila melanogaster depends on Notch as well as on Jeb/Alk mediated RTK-signalling.MarburgUniversität Marburg
- 19. Hinz U, Wolk A, Renkawitz-Pohl R (1992) Ultrabithorax is a regulator of beta 3 tubulin expression in the Drosophila visceral mesoderm. Development 116: 543–554.U. HinzA. WolkR. Renkawitz-Pohl1992Ultrabithorax is a regulator of beta 3 tubulin expression in the Drosophila visceral mesoderm.Development116543554
- 20. Kelly KK, Meadows SM, Cripps RM (2002) Drosophila MEF2 is a direct regulator of Actin57B transcription in cardiac, skeletal, and visceral muscle lineages. Mech Dev 110: 39–50.KK KellySM MeadowsRM Cripps2002Drosophila MEF2 is a direct regulator of Actin57B transcription in cardiac, skeletal, and visceral muscle lineages.Mech Dev1103950
- 21. Tomancak P, Beaton A, Weiszmann R, Kwan E, Shu S, et al. (2002) Systematic determination of patterns of gene expression during Drosophila embryogenesis. Genome Biol 3: RESEARCH0088.P. TomancakA. BeatonR. WeiszmannE. KwanS. Shu2002Systematic determination of patterns of gene expression during Drosophila embryogenesis.Genome Biol3RESEARCH0088
- 22. Brand AH, Perrimon N (1993) Targeted gene expression as a means of altering cell fates and generating dominant phenotypes. Development 118: 401–415.AH BrandN. Perrimon1993Targeted gene expression as a means of altering cell fates and generating dominant phenotypes.Development118401415
- 23. Bour BA, Chakravarti M, West JM, Abmayr SM (2000) Drosophila SNS, a member of the immunoglobulin superfamily that is essential for myoblast fusion. Genes Dev 14: 1498–1511.BA BourM. ChakravartiJM WestSM Abmayr2000Drosophila SNS, a member of the immunoglobulin superfamily that is essential for myoblast fusion.Genes Dev1414981511
- 24. Damm C, Wolk A, Buttgereit D, Loher K, Wagner E, et al. (1998) Independent regulatory elements in the upstream region of the Drosophila beta 3 tubulin gene (beta Tub60D) guide expression in the dorsal vessel and the somatic muscles. Dev Biol 199: 138–149.C. DammA. WolkD. ButtgereitK. LoherE. Wagner1998Independent regulatory elements in the upstream region of the Drosophila beta 3 tubulin gene (beta Tub60D) guide expression in the dorsal vessel and the somatic muscles.Dev Biol199138149
- 25. Kremser T, Gajewski K, Schulz RA, Renkawitz-Pohl R (1999) Tinman regulates the transcription of the beta3 tubulin gene (betaTub60D) in the dorsal vessel of Drosophila. Dev Biol 216: 327–339.T. KremserK. GajewskiRA SchulzR. Renkawitz-Pohl1999Tinman regulates the transcription of the beta3 tubulin gene (betaTub60D) in the dorsal vessel of Drosophila.Dev Biol216327339
- 26. Jakobsen JS, Braun M, Astorga J, Gustafson EH, Sandmann T, et al. (2007) Temporal ChIP-on-chip reveals Biniou as a universal regulator of the visceral muscle transcriptional network. Genes Dev 21: 2448–2460.JS JakobsenM. BraunJ. AstorgaEH GustafsonT. Sandmann2007Temporal ChIP-on-chip reveals Biniou as a universal regulator of the visceral muscle transcriptional network.Genes Dev2124482460
- 27. Sims D, Bursteinas B, Gao Q, Zvelebil M, Baum B (2006) FLIGHT: database and tools for the integration and cross-correlation of large-scale RNAi phenotypic datasets. Nucleic Acids Res 34: D479–483.D. SimsB. BursteinasQ. GaoM. ZvelebilB. Baum2006FLIGHT: database and tools for the integration and cross-correlation of large-scale RNAi phenotypic datasets.Nucleic Acids Res34D479483
- 28. Pavesi G, Mereghetti P, Mauri G, Pesole G (2004) Weeder Web: discovery of transcription factor binding sites in a set of sequences from co-regulated genes. Nucleic Acids Res 32: W199–203.G. PavesiP. MereghettiG. MauriG. Pesole2004Weeder Web: discovery of transcription factor binding sites in a set of sequences from co-regulated genes.Nucleic Acids Res32W199203
- 29. Ho Sui SJ, Fulton DL, Arenillas DJ, Kwon AT, Wasserman WW (2007) oPOSSUM: integrated tools for analysis of regulatory motif over-representation. Nucleic Acids Res 35: W245–252.SJ Ho SuiDL FultonDJ ArenillasAT KwonWW Wasserman2007oPOSSUM: integrated tools for analysis of regulatory motif over-representation.Nucleic Acids Res35W245252
- 30. Cliften P, Sudarsanam P, Desikan A, Fulton L, Fulton B, et al. (2003) Finding functional features in Saccharomyces genomes by phylogenetic footprinting. Science 301: 71–76.P. CliftenP. SudarsanamA. DesikanL. FultonB. Fulton2003Finding functional features in Saccharomyces genomes by phylogenetic footprinting.Science3017176
- 31. Aerts S, Thijs G, Coessens B, Staes M, Moreau Y, et al. (2003) Toucan: deciphering the cis-regulatory logic of coregulated genes. Nucleic Acids Res 31: 1753–1764.S. AertsG. ThijsB. CoessensM. StaesY. Moreau2003Toucan: deciphering the cis-regulatory logic of coregulated genes.Nucleic Acids Res3117531764
- 32. Gelfand MS, Novichkov PS, Novichkova ES, Mironov AA (2000) Comparative analysis of regulatory patterns in bacterial genomes. Brief Bioinform 1: 357–371.MS GelfandPS NovichkovES NovichkovaAA Mironov2000Comparative analysis of regulatory patterns in bacterial genomes.Brief Bioinform1357371
- 33. Gray S, Cai H, Barolo S, Levine M (1995) Transcriptional repression in the Drosophila embryo. Philos Trans R Soc Lond B Biol Sci 349: 257–262.S. GrayH. CaiS. BaroloM. Levine1995Transcriptional repression in the Drosophila embryo.Philos Trans R Soc Lond B Biol Sci349257262
- 34. Gray S, Levine M (1996) Short-range transcriptional repressors mediate both quenching and direct repression within complex loci in Drosophila. Genes Dev 10: 700–710.S. GrayM. Levine1996Short-range transcriptional repressors mediate both quenching and direct repression within complex loci in Drosophila.Genes Dev10700710
- 35. Sandmann T, Jakobsen JS, Furlong EE (2006) ChIP-on-chip protocol for genome-wide analysis of transcription factor binding in Drosophila melanogaster embryos. Nat Protoc 1: 2839–2855.T. SandmannJS JakobsenEE Furlong2006ChIP-on-chip protocol for genome-wide analysis of transcription factor binding in Drosophila melanogaster embryos.Nat Protoc128392855
- 36. Furlong EE, Profitt D, Scott MP (2001) Automated sorting of live transgenic embryos. Nat Biotechnol 19: 153–156.EE FurlongD. ProfittMP Scott2001Automated sorting of live transgenic embryos.Nat Biotechnol19153156
- 37. Tusher VG, Tibshirani R, Chu G (2001) Significance analysis of microarrays applied to the ionizing radiation response. Proc Natl Acad Sci U S A 98: 5116–5121.VG TusherR. TibshiraniG. Chu2001Significance analysis of microarrays applied to the ionizing radiation response.Proc Natl Acad Sci U S A9851165121
- 38. Breitling R, Armengaud P, Amtmann A, Herzyk P (2004) Rank products: a simple, yet powerful, new method to detect differentially regulated genes in replicated microarray experiments. FEBS Lett 573: 83–92.R. BreitlingP. ArmengaudA. AmtmannP. Herzyk2004Rank products: a simple, yet powerful, new method to detect differentially regulated genes in replicated microarray experiments.FEBS Lett5738392
- 39. Barolo S, Carver LA, Posakony JW (2000) GFP and beta-galactosidase transformation vectors for promoter/enhancer analysis in Drosophila. Biotechniques 29: 726, 728, 730, 732.S. BaroloLA CarverJW Posakony2000GFP and beta-galactosidase transformation vectors for promoter/enhancer analysis in Drosophila.Biotechniques29726, 728, 730, 732