Transcriptional Regulation of Mesoderm Genes by MEF2D during Early Xenopus Development

In Xenopus, specification of the three germ layers is one of the earliest developmental decisions occurring prior to gastrulation. The maternally-expressed vegetally-localized transcription factor VegT has a central role in cell autonomous specification of endoderm and in the generation of mesoderm-inducing signals. Yet, marginally-expressed transcription factors that cooperate with mesoderm-inducing signals are less investigated. Here we report that the transcription factors MEF2A and MEF2D are expressed in the animal hemisphere before mid-blastula transition. At the initiation of zygotic transcription, expression of MEF2D expands into the marginal region that gives rise to mesoderm. Knockdown of MEF2D delayed gastrulation movements, prevented embryo elongation at the subsequent tailbud stage and caused severe defects in axial tissues. At the molecular level, MEF2D knockdown reduced the expression of genes involved in mesoderm formation and patterning. We also report that MEF2D functions with FGF signaling in a positive feedback loop; each augments the expression of the other in the marginal region and both are necessary for mesodermal gene expression. One target of MEF2D is the Nodal-related 1 gene (Xnr1) that mediates some of MEF2D mesodermal activities. Chromatin immunoprecipitation analysis revealed that MEF2D associates with transcriptional regulatory sequences of the Xnr1 gene. Several MEF2 binding sites within the proximal promoter region of Xnr1 were identified by their in vitro association with MEF2D protein. The same promoter region was necessary but not sufficient to mediate MEF2D activity in a reporter gene assay. In sum, our results indicate that the MEF2D protein is a key transcription factor in the marginal zone acting in a positive feedback loop with FGF signaling that promotes mesoderm specification at late blastula stages.


Introduction
Specification of the three primary germ layers in Xenopus embryos is one of the earliest developmental events occurring before the end of the blastula stage [1]. In this process, vegetal, marginal and animal cells are specified to form endoderm, mesoderm and ectoderm, respectively. Since zygotic transcription is initiated only at the 12 th cell cycle (mid blastula transition; MBT), maternal mRNA inherited from the oocyte plays a significant role in early specification and patterning [2]. Maternal transcripts encode transcription factors and signaling molecules and are often localized to specific areas of the early embryo such as the vegetal pole or the animal pole [3]. For example, the T box transcription factor, VegT, is localized in the vegetal hemisphere and plays a central role in endoderm and mesoderm specification [4]. VegT and the dorsally-expressed b-catenin induce the transcription of nodal related genes, Xnr5 and Xnr6, at the 256 cell stage [5]. At MBT, these nodal-related genes induce the nonautonomous expression of Xnrs 1, 2 and 4 that is essential for mesoderm induction [6,7]. As a result, Xnr genes are expressed at higher levels on the dorsal side of the embryo (Nieuwkoop center), creating a dorsal-ventral gradient in late blastula stages. Nodal signaling induces mesoderm identity and patterning in the overlying marginal zone [8,9]. Fibroblast growth factor (FGF) signaling that is initiated at late blastula in the marginal zone, induces the expression of the brachyury gene (XBra) that encodes a key transcription factor in mesoderm specification [10,11,12,13,14,15]. Yet, the transcription programs responsible for Nodal and FGF signaling in the specification and patterning of mesoderm are still largely unknown.
The MEF2 (Myocyte Enhancer Factor 2) family of transcription factors is involved in many aspects of cell differentiation and organogenesis. There are four members of MEF2 in vertebrates: MEF2A, B, C and D. Their conserved DNA binding and dimerization motifs mediate binding of homo-or hetero-dimers to a consensus A/T rich DNA sequence [16].
Three isoforms of MEF2 have been characterized in Xenopus: MEF2A, MEF2D and MEF2C. Early studies demonstrated the involvement of MEF2A and MEF2D in somitic skeletal muscle development and cardiac gene expression [17,18]. An additional role of MEF2A in the formation of anterior neural structure downstream to the Nemo like kinase (NLK) was reported recently [19]. The effects of MEF2A were attributed to its expression during neural development. Unlike MEF2A and MEF2D, MEF2C does not participate in muscle development, but in the development of tendons where it cooperates with Xscleraxis [20].
Interestingly, the expression of MEF2A and MEF2D is detected before MBT, unlike the expression of MEF2C that is initiated at zygotic stage 13 [17,20]. Early expression of MEF2A and MEF2D proteins may indicate their early function in germ layer specification and patterning.
In the present study we investigated early expression patterns of MEF2A and MEF2D and their possible involvement in germ layer specification. Our results indicate that the two genes are maternally expressed and localized to animal cells. Following MBT, zygotic expression of MEF2D expands into the marginal region. The role of MEF2D at the blastula stage was studied by its antisense morpholino-mediated knockdown. It was found that absence of MEF2D prevented normal development of axial organs of mesoderm and ectoderm origin. We also show that the expression of MEF2D in the marginal zone was necessary for the response of this tissue to vegetal signals and the induction of mesoderm gene expression. We report that MEF2D is involved in the expression of many mesodermal genes including that of signaling molecules of the BMP, Nodal and FGF pathways. MEF2D and FGF signaling maintained each other's expression, constituting a positive feedback loop. Lastly, we show that MEF2D associates directly with regulatory sequences of one target gene of the Nodal family, Xnr1. Thus, our study uncovers the involvement of MEF2D in the establishment of early mesoderm.

Ethics Statement
All Xenopus laevis experiments in this study have been conducted under protocol # IL-149-12-2010 that is valid until December 2014 and approved by the Technion Committee for Care and Use of Laboratory Animals. The Technion holds a valid assurance (#A5026-01) of the US Department of Health and Human Services for humane care and use of laboratory animals.

Ovulation and Fertilization
Female Xenopus laevis frogs were injected with human Chorionic Gonadotropin (hCG, 1000 units per frog) 24 h prior to fertilization. Following fertilization in vitro, eggs were maintained in Modified Ringers (1xMR) salt solution. Fertilized eggs were dejellied with 2% L-Cysteine solution, washed in 1/3xMR, and injected in 3% Ficoll with different concentrations of capped sense in vitro transcribed RNA during the single cell stage. Embryos were staged according to [21]. Animal Cap (AC) and Vegetal Pole (VP) explants were removed at stage 9 using watchmakers' forceps. Dorsal Marginal Zone (DMZ) and Ventral Marginal Zone (VMZ) explants were removed at stage 10.5 with an eyebrow knife. Explants were cultured in 1XLCMR solution with gentamycin (50 mg/ml) until sibling embryos reached the desired stage.

Recombinant Explants
AC explants excised at stage 9 from naïve and XMEF2D AMOmicroinjected embryos were conjugated with naïve vegetal pole explants (excised at the same time) in 1xLCMR solution with gentamycine, at 13uc for 3 hrs. AC and vegetal explants were then separated; AC explants were cultured until sibling embryos reached stage 12 and subjected to RT-PCR analysis.

RT-PCR
Analysis was performed as described [23], except that random hexamers (100 ng per reaction) were used for reverse transcription. Primers are described in http://www.xenbase.org.

Real Time RT-PCR
Total RNA extraction and cDNA synthesis was performed as described above. Quantitative real-time PCR was carried out using Stratagene Mx3000P device and SYBR Premix Ex Taq II reaction mix (TaKaRa). The PCR conditions were as follows: an initial 2-min induction step at 95uC, followed by 40 cycles of amplification: 20-sec at 95uC, 30-sec at 55uC, and 30-sec at 72uC. The run is finished by a melting curve from 95uC to 55uC to ensure PCR product purity. Two negative controls were used to verify the authenticity of the results: reverse transcriptase negative control from the cDNA synthesis step and water-only control. In all cases those negative controls failed to produce specific products. Ornithine decarboxylase (ODC) was used as a loading control and all values were normalized to its levels. Each experiment was repeated a minimum of three times in independent experiments to verify reproducibility of results. Primer sequences used to amplify cDNA are presented in Table 1.

Histology
Slides were deparaffinized and hydrated with decreasing alcohol concentrations. The hematoxylin-eosin method was used to stain the tissue. Slides were analyzed by optical microscopy.
In situ Hybridization (ISH) and Immunohistochemistry ISH was carried out with digoxigenin-labeled probes [24,25,26]. Embryos were cultured to different stages and subsequently fixed for in situ hybridization. Mef2a, Mef2d, goosecoid, chordin, myod, Xnr1 and brachyury antisense probes were used. For immunohistochemistry on sections, slides were deparaffinized and hydrated with a decreasing alcohol gradient. Then sections were blocked with goat serum for 1 hour followed by 14 h incubation at 40uC with anti-MEF2 antibody (Santa Cruz C21, 1:100). This was followed by incubation with an appropriate biotinylated secondary antibody, streptavidin-peroxidase conjugate, and S-(2-aminoethyl)-Lcysteine (AEC) as a substrate (Histostain-SP kit; Zymed Laboratory, San Francisco, CA, USA); counterstaining was done with hematoxylin.

Chromatin Immunoprecipitation
Chromatin immunoprecipitation assays were performed as described [27]. Embryos were injected at the one cell stage with mRNA encoding MEF2D-Flag (1.5 ng per embryo). A second group of control embryos were not injected. Embryos were collected at stage 10 (50 embryos/sample) and processed according to the protocol. Antibodies used to immunoprecipitate chromatin were polyclonal anti-Flag (''specific'') and Rabbit preimmune serum (''non-specific''). Sequences of primers used to amplify the immunoprecipitated fragmented DNA are presented in Table 2. Data was analyzed by the DDC(t) (or Livak) method as described [27].

Luciferase Assay
A 696 bp fragment including proximal promoter sequences of Xnr1 was PCR-amplified from genomic Xenopus DNA and was inserted into the pGL2-basic vector (Promega). The correct sequence was verified by sequencing of the plasmid. This reporter gene or x3 Mef2-Luc were either injected to embryos (20 pg DNA) or transfected to 293T HEK cells (2 mg DNA). Several MEF2expression plasmids were co-transfected with each of the reporter genes. Extract preparation and luciferase assay were performed using the Luciferase Assay System (Promega). Proteins were extracted from transfected cells (6 cm plate in 500 ml buffer) or embryos injected with luciferase reporter gene using the Reporter Lysis Buffer (Promega, 10 embryos in 200 ml buffer). Luciferase activity was measured by integrating total light emission over a period of 30 sec using a Berthold luminometer. Luciferase activity was normalized to total protein concentration.

Statistical Analysis
The graphs in the figures are presented as means 6 S.E. Statistical significance was analyzed using the Student's t-test. In the figures, * indicates p,0.05, ** p,0.01, *** p,0.001.

Maternal MEF2 is Localized at the Animal Hemisphere of the Embryo
The expression levels of the mef2a and mef2d gene transcripts from fertilization until gastrula stages were analyzed by quantitative PCR (qPCR). In agreement with previous studies [17,20], pre-MBT maternal transcripts of both isoforms were detected ( Figure 1A). These transcripts were localized animally at the 4 cell stage ( Figure 1B). While mef2a expression levels declined after MBT, zygotic mef2d transcripts progressively accumulated due to the onset of zygotic transcription ( Figure 1A). Mef2c was not detected by qPCR at these stages (data not shown) [20]. Localization of the mef2d mRNA and protein was analyzed in embryos at stages 8 and 9. Reverse transcription-PCR and Western analyses indicated that MEF2D was confined to the animal hemisphere at stage 8 ( Figure 1C, left panels). Likewise, sections that were immunostained with anti MEF2 antibody also indicated animal expression of MEF2D ( Figure 1C, right panel). This antibody detected a single band of the MEF2D expected size that was diminished in embryos that were injected with specific antisense morpholino to mef2d (see Figure 2A). In addition, the same anti MEF2 antibody specifically and robustly stained cell nuclei of early neurula stage paraxial mesoderm as was expected for MEF2D ( Figure S1). At stage 9, both in situ hybridization (ISH) and immunostaining analyses indicated that the expression of zygotic MEF2D was expanded into the marginal region ( Figure 1D). At stage 10.5, mef2d transcript was localized to the dorsal marginal zone (DMZ) ( Figure 1E, left panel). At that stage, the expression of injected Mef2-reporter gene (x3 Mef2 Luc) indicated that MEF2 activity was enriched in the DMZ relative to the ventral marginal zone (VMZ) ( Figure 1E, right panel). In situ hybridization at later stages revealed zygotic expression of mef2d in the paraxial mesoderm, while expression of zygotic mef2a was detected first at stage 15 ( Figure S2). In conclusion, expression of MEF2D before MBT is confined to the animal half of embryos whereas at late blastula it expands to the marginal zone. At early gastrula stage MEF2D expression is reduced to the DMZ and in the latter neurula stages it takes place in the paraxial mesoderm.

Morpholino-mediated Knockdown of MEF2D Interferes with Axial Development
To study the involvement of MEF2D protein in early specification and patterning, antisense morpholino (AMO) to MEF2D was injected to single-cell embryos. Injected mismatched MEF2D AMO served as controls. Injection of MEF2D AMO significantly reduced the expression of endogenous MEF2D protein at different developmental stages before and following MBT (Figure 2A). In addition, MEF2D AMO prevented the ectopic expression of Xenopus MEF2D-Flag protein (Figure 2A). Injection of MEF2D AMO delayed blastopore closure relative to uninjected or mismatch AMO-injected embryos ( Figure 2B). Mismatch MEF2D AMO mildly-reduced MEF2D protein levels ( Figure 2B, right panel). This result might explain the subtle effects of injected mismatch AMO relative to uninjected embryos that were observed in the rest of the study. Increasing amounts of injected MEF2D AMO induced significant shortening of the body axis whereas the mismatch-AMO displayed a minimal effect on embryo elongation at stage 26 ( Figure 2C). Histological sections of stage 26 embryos revealed that axial structures were significantly affected in MEF2D-depleted embryos ( Figure 2D). The spinal cord was absent while the structures of the notochord and somites were greatly reduced. In conclusion, these results indicated that MEF2D is involved in gastrulation movements and the formation of axial organs of mesoderm and ectoderm origin.

MEF2D is an Inducer of Mesoderm and the Spemann Organizer
Expression of MEF2D in the marginal zone starts at the late blastula stage, and the loss of axial mesodermal structures including the somites and notochord in MEF2-depleted embryos suggested that mesoderm might be affected by MEF2D. To investigate this possibility, MEF2D AMO and control mismatch AMO were injected to one cell embryos and expression patterns of several mesodermal genes were analyzed by qPCR ( Figure 3A). At late blastula stage, the expression of early mesodermal genes was significantly reduced whereas the expression of the vegetal gene, vegt was not affected in MEF2D AMO-injected embryos     Figure 3A). At gastrulae, expression of Spemann organizer genes, goosecoid, chordin and noggin was diminished in MEF2D-depleted embryos while that of the vegetal gene, mixer was not affected ( Figure 3B). At this stage, the normal ring-like expression of brachyury (Xbra) around the blastopore was barely detected and organizer expression of goosecoid and chordin was significantly standardized accordingly. Data are presented as means 6 SE of two independent experiments with duplicates. (D) fgf8 mRNA was injected vegetally to one of four cell embryos. At stage 9 embryos were analyzed by ISH with a probe to mef2d. Upper panel: whole embryos. Lower panel: hemisected embryos. (E) One cell embryos were injected with a plasmid containing x3 Mef-Luc reporter gene without or with fgf8 mRNA or FGF-DNR mRNA. At stage 10.5 proteins were extracted and luciferase activity was measured and normalized according to the total amount of proteins. Data are presented as means 6 SE of three independent experiments with duplicates. (F) qPCR analysis of control embryos and embryos injected with MEF2D AMO. Five embryos on each group were injected and RNA was extracted at stage 10.5. Expression levels of control embryos were arbitrarily set to a value of 1, and values of injected embryos were standardized accordingly. Data are presented as means 6 SE of two independent experiments with duplicates. doi:10.1371/journal.pone.0069693.g005 reduced in the MEF2D AMO-injected embryos ( Figure 3C). At the neurula stage, the paraxial expression of myod and muscle actin was substantially decreased in MEF2D AMO-injected embryos relative to control embryos ( Figure 3D). At the tailbud stage, the expression of myosin heavy chain was completely absent in MEF2D AMO-injected embryos, yet the somite and notochord structures are observable but smaller than in control embryos ( Figure 3E). It therefore appeared that MEF2D was required for the early expression of mesoderm inducers such as Xnr1 and Xbra, as well as for the subsequent expression of organizer and paraxial mesoderm genes such as chordin, noggin, myod and muscle actin. The specificity of MEF2D knockdown was investigated by injection of mef2d mRNA containing three mismatches along the AMO-targeted sequence. Co-injection of this transcript rescued the expression of mesoderm genes in MEF2D-knockdown embryos ( Figure 3F). In a second approach to block MEF2 activity, a chimera protein composed of MEF2D and the Engrailed repressor domain of drosophila (MEF2D-En R ) was expressed in embryos. Injection of mef2d-en R mRNA repressed the promoter activity of a MEF2-dependent reporter gene (x3 Mef2-Luc) suggesting that MEF2D-En R chimera functioned as a dominant negative of MEF2 activity ( Figure S3). It also decreased transcript levels of the mesodermal genes, Xbra, goosecoid, noggin and chordin in a concentration-dependent manner ( Figure S3).

FGF8 Induces the Expression of MEF2D in the Marginal Zone
Since MEF2D-Flag activates only minimally the expression of mesodermal genes relative to MEF2-VP16, we hypothesized that MEF2 may cooperate with one of the signaling events that induces mesoderm. It was demonstrated before that marginal FGF signaling at the late blastula stage initiates mesodermal gene expression [10,11,12,13,14,15]. Therefore, we investigated how FGF signaling and MEF2D interact to elicit mesoderm gene expression. First, we inquired whether MEF2D protein was required for FGF8-induced mesoderm gene expression ( Figure 5A). As expected, injection of fgf8 mRNA induced the expression of mesodermal genes, Xbra and myf5 in AC explants. Co-injection of fgf8 mRNA and MEF2D AMO significantly reduced the expression of mesodermal genes, indicating that MEF2D is necessary for the induction by FGF8 of mesoderm gene expression. Next, we investigated whether FGF signaling affected mef2d gene expression levels ( Figure 5B). Dominant negative FGF receptor (XFD) was ectopically expressed in embryos and the expression of mef2d was followed by ISH ( Figure 5B) and qPCR ( Figure 5C). The marginal expression of mef2d was reduced in xfdinjected embryos while the animal expression appeared to remain unaffected in stage 9 embryos ( Figure 5B). qPCR analysis of stage 10.5 embryos revealed that injected xfd mRNA significantly reduced the expression of mef2d as well as the expression of Xbra without affecting the expression of vegt ( Figure 5C). In a gain-offunction experiment, injection of fgf8 mRNA into vegetal blastomeres induced robust ectopic and localized expression of mef2d ( Figure 5D). Next, we inquired whether the activity of MEF2D was affected by FGF signaling. MEF2 activity was monitored by measuring luciferase activity of injected x3 MEF2-Luc reporter gene. Co-injection of xfd mRNA together with the reporter gene reduced whereas of fgf8 mRNA increased the reporter activity ( Figure 5E). We conclude therefore that FGF signaling sufficed to induce mef2 expression and activity. Further insight into the effect of MEF2D on FGF expression was gained by observing that injected MEF2D AMO significantly reduced fgf4 and fgf8 expression at late blastula ( Figure 5F). Overall, our results indicated the existence of a positive feedback loop between FGF and MEF2D in the marginal zone that participates in the early expression of mesoderm genes.

MEF2 Cooperates with Vegetal Signals to Induce Mesoderm Gene Transcription
Next, we asked whether MEF2D functions as a mediator of vegetal signals in the initiation of mesoderm gene expression. Injection of vegt mRNA induced the expression of pro-mesodermal genes in control AC explants [28], yet expression of these genes by vegt mRNA was prevented in MEF2D-knockdown embryos ( Figure 6A). Injected MEF2D AMO prevented the expression of Xnr1 and Xbra at a late blastula stage and myf5 and muscle actin at a late gastrula stage. We therefore conclude that MEF2D is necessary for VegT-induced expression of mesoderm genes.
Since MEF2 is expressed in the marginal zone at the late blastula stage, it may function in situ to induce the expression of  [47]. Arrows show the two transcription start site and ''M'' the translation initiation codon. Lower panel: EMSA of each of the MEF2 binding elements coupled with protein extracts of stage 9 control embryos as well as embryos injected with mef2d-flag mRNA. Anti-flag antibody (1 ml, 0.1 mg/ml) was included in some reaction mixtures while unlabeled homologous double stranded oligonucleotides in 100 fold excess over the probe was included in others, as indicated. Unbound probes are not shown. Arrow indicates the MEF2D-DNA complex. Arrowhead indicates the Anti MEF2-MEF2D-DNA complex. (C) 293T HEK cells were transfected as indicated. Thirty six hours later, proteins were extracted and luciferase activity was measured and was normalized to total protein levels. Activity of the reported gene with an empty vector was set to a value of 1 and values of other treatments were standardized accordingly. Means of two independent experiments are presented. doi:10.1371/journal.pone.0069693.g008 mesodermal genes. To test this possibility, we used a recombinant explant experiment [29] in which vegetal explants from control embryos were combined with animal cap (AC) explants from control or MEF2D AMO-injected late blastula stage embryos. After a 3 hour co-culture period, animal caps were separated from vegetal masses, cultured until sibling embryos reached late gastrula stage and analyzed for the expression of mesodermal genes ( Figure 6B, left panel). As expected, the expression of mesodermal genes Xbra, myod, muscle actin and noggin was induced in AC explants from wild type embryos that were co-cultured with vegetal explants ( Figure 6B, right panel). In addition to mesodermal genes, the animal-specific gene, Epidermal keratin, was expressed in AC explants. However, the same mesodermal genes were only mildly induced in AC explants that originated from MEF2D-depleted embryos. Interestingly, the expression of Epidermal keratin was diminished in MEF2D-depleted AC explants. This result indicated that in situ expression of MEF2D protein in the animal cells was necessary for their response to the vegetal mesoderm-inducing signals and to the expression of animal ectoderm marker, Epidermal keratin.

MEF2D Binds Directly to Xnr1 Transcription Regulatory Sequences
One of the targets of MEF2D that were identified in this work was the Xnr1 gene whose expression was initiated at MBT (stage 8). We could expect therefore similar expression patterns of mef2d and Xnr1. Indeed, ISH analysis indicated that at late blastula these two genes are expressed marginally and to a lesser extent vegetally ( Figure 7A) Xenopus Nodal-Related (Xnr) genes are required for mesendoderm induction and gastrulation movements [30]. Next, we asked whether MEF2 activity was sufficient to induce ectopic Xnr1 expression. Marginal injection of mef2-vp16 mRNA induced robust ectopic animal expression of Xnr1 ( Figure 7B) and low Xnr1 expression in the vegetal region (data not shown). Thus, MEF2 activity was sufficient to induce Xnr1 expression in the animal but not in the vegetal region. We next examined whether Xnr1 was able to substitute MEF2D activity in the induction of mesoderm genes. Injection of Xnr1 mRNA induced significant expression of Xbra both marginally and animally while injection of MEF2D-AMO completely abolished physiological marginal Xbra expression ( Figure 7C). Injection of both MEF2D-AMO and Xnr1 mRNA induced intermediate levels of Xbra expression. Similarly, injected Xnr1 rescued the expression of Xbra and Myf5 in MEF2Ddepleted embryos as was analyzed by qPCR ( Figure 7D). Moreover, injection of Xnr1 mRNA also induced ectopic mesodermal gene expression in AC explants even in the presence of co-injected MEF2D AMO ( Figure S4). These results indicated that Xnr1, one of MEF2D target genes, mediated at least some of MEF2D activities in mesoderm specification. Next, we asked whether MEF2D associates directly with Xnr1 regulatory sequences at the chromatin level. To explore this question we employed chromatin immunoprecipitation (ChIP) assay in embryos that were injected with mRNA encoding for MEF2D-Flag protein ( Figure 8A). Fragmented chromatin was immunoprecipitated (IPed) with either anti-Flag (specific) or pre immune (nonspecific) serum and qPCR analysis of fragments scattered along the Xnr1 gene was performed. IP with anti-Flag antibodies resulted in a significant and reproducible enrichment of fragments found around 1900 (distal) and 900 (proximal) base pairs upstream to Xnr1 translation initiation site and within the first intron of Xnr1 as compared to IP with pre-immune serum ( Figure 8A). Intron sequences of the tubulin gene and other upstream regions of the Xnr1 gene were not enriched in anti-Flag complexes relative to pre-immune complexes indicating the specificity of the assay. We conclude, therefore, that MEF2D associates with several regulatory regions of the Xnr1 gene. Sequence analysis revealed several putative Mef2 binding sites in the different regions that were enriched in the ChIP analysis. Using electrophoretic mobility shift assay (EMSA), we further analyzed the in vitro MEF2 binding to 4 elements located within 600bp from the translation initiation site (proximal region) ( Figure 8B) [31,32]. For that purpose, protein extracts were prepared from embryos injected with mRNA encoding the MEF2D-Flag protein and incubated with four different DNA probes, each containing one of the identified elements. In all four cases, a shifted band representing a Mef2-DNA complex was detected. The most proximal element (PE) displayed the lowest binding whereas the most distal element (DE) exhibited the highest binding efficiency. Binding of MEF2D-Flag to any of these elements appeared weaker than its binding to the optimal element of the MCK enhancer ( Figure 8B). The in vitro binding to the proximal region (between 2540 and 2230 relative to translation initiation site) possibly corresponds to the enrichment of the 2900 fragment in the ChIP assay, in which the average sonicated fragment size was around 500-1000 bps (data not shown). A reporter gene carrying these elements (696bp upstream regulatory region) was not activated by MEF2D-Flag both in embryos (data not shown) and in transfected cells ( Figure 8C). The same reporter gene was only mildly activated by exogenous MEF2-VP16 and was significantly repressed by the Mef2 repressor protein, MEF2D-En R . In contrast, the activity of a control reporter containing three optimal Mef2 binding sites (x3 Mef2-Luc) was potently activated by MEF2D-Flag or by MEF2-VP16 and was repressed by MEF2D-En R . We conclude, therefore, that although the proximal promoter contains Mef2 binding sites, this region is not sufficient for transcriptional activation of the Xnr1 gene by MEF2D.

Dynamic Expression of MEF2 Proteins at Early Developmental Stages
The present study aimed at understanding expression patterns and roles of MEF2 proteins during blastula to gastrula stages. We find that MEF2 expression pattern is very dynamic: before MBT, maternal MEF2A and MEF2D proteins are localized in the animal hemisphere. Following MBT, the expression of zygotic MEF2D expands primarily to the marginal region whereas MEF2A expression declines and is initiated at later neurula stages. Our study focused, therefore, on the role of post-MBT MEF2D expression in the marginal region. As is expected from the dynamic early expression of MEF2D, the phenotype of MEF2Ddepleted embryos is complex; gastrulation is delayed, embryos do not properly elongate and axial structures of mesoderm and ectoderm origins are absent. The present investigation focused on the mode of MEF2D involvement in mesoderm gene expression, leaving for future studies the issue of ectoderm gene expression. Results presented in this study indicate that MEF2D plays a major role in mesoderm specification.

MEF2D Affects Mesoderm Specification and Patterning
Mesoderm induction occurs after MBT when zygotic transcription begins under the influence of vegetally-localized maternal molecules [33,34]. The immediate marginal expression of MEF2D following MBT suggested its involvement in mesoderm specification. Knockdown of MEF2D reduced the expression of a wide array of mesodermal genes encoding for signaling molecules and transcription factors that affect early mesoderm. The expression of marginal zygotic genes such as fgf8, Xbra, vent1, Xnr1 and goosecoid diminished whereas the expression levels of vegetal genes; Xnr5, mixer and vegt was not affected. These results indicate that MEF2D is an early zygotic pan-mesodermal master gene which is involved directly or indirectly in the expression of majority of marginal genes. A possible explanation for this wide effect of MEF2D is a likely reflection of its regulatory role in the expression of critical signaling molecules of the FGF, Nodal, BMP and Wnt signaling pathways that are essential for mesoderm specification and patterning.
The maternal VegT transcription factor is directly involved in the vegetal expression of Nodal ligands that induce mesoderm in the overlaying marginal cells [7,33,34]. We show that MEF2D expression is required in an animal explant for its response to vegetal signals in the initiation of mesoderm gene transcription. MEF2D is, therefore, a marginally-localized transcription factor that transforms vegetal signals to mesoderm gene expression. MEF2D may cooperate with vegetal signals of Xnr5 and Xnr6 in a mechanism involving interaction with the downstream Nodal effector, Smad2 at the regulatory regions of their target genes [7,35]. Alternatively, MEF2D could be phosphorylated by a kinase of non-canonical Nodal signaling [19,36]. Thus, MEF2D may be viewed as a hub that integrates vegetal-originated signals into marginal transcription program of the mesoderm.

MEF2D and FGF Function in a Positive Feedback Loop to Induce Mesoderm Specification
FGF signaling plays a crucial role in mesoderm specification [10,11,12,13,14,15]. Similarly to MEF2D, several FGF ligands such as FGF2, FGF4 and FGF8 are expressed in the marginal zone of late blastula stage embryos and in the DMZ of early gastrula stage [37]. Little is known about transcription factors that act downstream of FGF signaling in Xenopus during mesoderm formation. One downstream effector of FGF is the SRFinteracting Ets family member Elk-1, which is phosphorylated in response to FGF signaling and is required for Brachyury expression [38]. Our results indicate that MEF2D is a newly identified downstream target of FGF signaling. We show that FGF signaling is required for the expression of MEF2D in the marginal zone. Loss of FGF activity prevents marginal expression of MEF2D, whereas exogenous FGF activity induces ectopic MEF2D expression in the vegetal region. Moreover, in the absence of MEF2D expression, FGF failed to induce mesodermal gene expression in animal cap assay. Presently, we do not know if MEF2D is phosphorylated in response to FGF signaling in Xenopus. This possibility is likely since in Xenopus, FGF signals through MAPK to induce mesoderm [39,40,41] and MEF2 proteins are known to be phosphorylated by mitogen activated protein kinase (MAPK) pathways [42,43,44,45,46]. Notably, it is not just that FGF affects MEF2D expression but also reciprocally MEF2D is required for the expression of FGF ligands. Therefore, FGF and MEF2D maintain each other overlapping expression in the forming mesoderm at blastula to gastrula stages.

MEF2D Associates with Regulatory Sequences of xnr1 Gene
Regulation of Xnr1 transcription was thoroughly investigated. Xnr1 is expressed from blastula stages in the vegetal marginal border, and at the gastrulation stage it is predominantly expressed in the dorsal marginal zone. Xnr1 is regulated by maternal factors VegT and b catenin, and zygotic Nodal signaling (Xnr5, and 6). Xnr1 upstream regulatory sequences were demonstrated to be directly regulated by VegT and effectors of Wnt signaling (TCF/ LEF) [47]. The first intron contains activin response element (ARE) that was demonstrated to bind a complex of FoxH1 and Smad2 [35]. Our results indicate that MEF2D may be added to the growing list of transcription factors regulating Xnr1 expression. Evidence supports the notion that Xnr1 is a target gene of MEF2D. First, knockdown of MEF2D diminishes Xnr1 expression while excessive Mef2 activity induces ectopic Xnr1 expression in the animal region. Second, ectopic expression of Xnr1 partly rescues mesoderm gene expression in the absence of MEF2D. Third, chromatin IP analysis indicates that MEF2D associates with several regions along the Xnr1 gene, both upstream to the transcription start sites and within its first intron. These findings indicate a complex regulation of Xnr1 transcription by MEF2D since several regions along the gene associate with MEF2D, and probably the combinatorial activity of distinct regions is necessary to activate transcription. In the present study we focused on the proximal promoter region that contains several putative MEF2 binding sites (YTA(A/T) 4 TAR). Our results demonstrate the in vitro binding of MEF2D to some of these sites. Some of the putative MEF2 sites are adjacent to previously identified sites of VegT and FoxH1/Smad2, indicating a possible cooperation between the different transcription factors in the regulation of Xnr1 transcription. Transcription from the same proximal promoter region was not induced by expression of MEF2D, but was repressed by the MEF2 repressor protein, MEF2D-En R . Therefore, we conclude that MEF2 binds to sites at the proximal promoter region and is necessary but not sufficient to drive transcription from this element. Future studies will clarify whether binding of the MEF2D protein to additional regions of the Xnr1 gene that were identified by the ChIP assay play a role in Xnr1 transcription.

Complex Activities of MEF2 Proteins during Early Stages of Development
While previous studies concentrated on late developmental activities of MEF2 proteins, in this study we analyzed expression and function of MEF2 proteins from fertilization to early gastrula stages. Expression patterns of MEF2 proteins indicate their participation in ectoderm and mesoderm specification. This study of the role of MEF2D in early mesoderm introduces new components to the already complex network of molecules that participate in mesoderm formation ( Figure 9). We suggest that following MBT MEF2D is involved in the intricate array of signaling pathways affecting mesoderm. FGF signaling regulates marginal expression of MEF2D which is necessary to mediate FGF activity in mesoderm gene expression. Also, MEF2D is required for the expression of FGF ligands (FGF4, 8). Therefore, FGF and MEF2D function in a positive autoregulatory loop to maintain marginal mesoderm gene expression. The marginal expression of MEF2D is necessary to mediate mesoderm gene expression by vegetal Nodal signaling. MEF2D induces the transcription of Xnr1 and some other mesodermal genes, and the Xnr1 protein partly mediates the mesodermal regulatory function of MEF2D. Overall, MEF2D responds to and activates signals necessary for mesoderm specific transcription. Global analysis of MEF2 target genes is expected to reveal a much broader role for this family of transcription factors as was demonstrated in Drosophila [48]. Several interesting questions that arise from this work remain to be explored; which additional developmental programs are affected by MEF2 at these early stages and how? Which genes are direct targets of MEF2 proteins? Is the early role of MEF2 in germ layer specification conserved between species? Our ongoing investigations focus on some of these questions.