The skeletal muscles in vertebrates develop from somites as the result of patterning and cell type specification events. These processes are mainly controlled by members of myogenic regulatory factor (MRF) family, including MyoD, Myf-5, Myogenin, and MRF4, which are all basic helix-loop-helix (bHLH) transcription factors (reviewed by Arnold and Winter, 1998; Arnold and Braun, 2000). Among these four genes, myf-5 and myoD are expressed earlier than myogenin and MRF4 (Cossu et al., 1996) and, therefore, are believed to play a central role in initiating myogenesis.
A comprehensive view of skeletal muscle development does not only depend on understanding the functions of MRF proteins but also on the knowledge of how MRF gene expression is regulated during embryogenesis. In Xenopus, myf-5 transcription begins at the early gastrula broadly in the dorsal half of the mesoderm and then quickly resolves into two dorsolateral domains directly adjacent to the Spemann organizer. After gastrulation, myf-5 transcripts disappear from the differentiating myocytes of anterior somites but remain detectable in newly formed somites and unsegmented paraxial mesoderm (Hopwood et al., 1991; Dosch et al., 1997; Jones and Smith, 1998; Takahashi et al., 1998). Understanding how this unique expression pattern is regulated would provide critical insights into both mesoderm patterning and myogenic development. We have demonstrated previously the existence of two negative regulatory elements in the Xmyf-5 promoter that control Xmyf-5 expression. In the Xmyf-5 promoter, a distal TCF-3 binding site restricts Xmyf-5 expression out of the midline mesoderm by means of Wnt/β-catenin signals (Yang et al., 2002) and an interferon regulatory factor (IRF) -like DNA binding element down-regulates Xmyf-5 expression in the differentiating myocytes (Mei et al., 2001). However, little is known about the mechanism of the myf-5 activation in Xenopus gastrula embryos. In an analysis of the Xenopus tropicalis myf-5 promoter, Polli and Amaya (2002) identified two TAAT motifs in a homeobox binding site (HBX) similar to the consensus binding site of Xvent-2, which is a transcriptional repressor. It was shown that these elements were required for both repression and activation of myf-5, but a candidate of transcription activators, Xcad-3, failed to activate myf-5 and resulted in the repression of myf-5 expression. Candidates of myf-5 activators remain unknown.
Members of T-box gene family have been demonstrated to control many and varied developmental processes in metazoans from hydra to humans (reviewed by Papaioannou, 2001). Several T-box transcription factors are involved in mesoderm formation (reviewed by Stennard et al., 1997). For instance, Xbra is both necessary and sufficient for mesoderm formation (reviewed by Smith, 2001). Maternal VegT, through TGFβ growth factors, is also essential for mesoderm induction (Zhang et al., 1998; Kofron et al., 1999). Mutation in Tbx6 in mice resulted in the formation of three neural tubes at the expense of the posterior somites (Chapman and Papaioannou, 1998), indicating the role of Tbx6 in the specification of the posterior paraxial mesoderm. All members of this gene family contain a conserved sequence coding for a DNA-binding motif known as the T-box, named after the first-discovered T-box gene, T, or Brachyury (Herrmann et al., 1990). Binding site selection experiments demonstrated that the DNA-binding motif of three T-box proteins, Xbra, VegT, and Eomesoderm, recognized the same core sequence of TCACACCT, with some differences in the flanking nucleotides (Conlon et al., 2001). Although several T-box proteins have been shown to function as transcriptional regulators, to date only a few of downstream target genes have been identified (Tada et al., 1998; Casey et al., 1998, 1999; Saka et al., 2000; Tada and Smith, 2000; Mitani et al., 2001).
In this study, we report the identification of a T-box binding site in the Xmyf-5 promoter by using restriction enzyme–mediated integration (REMI) transgenesis (Kroll and Amaya, 1999). We showed that this putative T-box binding site mediated the dorsal activation of Xmyf-5 transcription in Xenopus gastrula embryos and could be bound in vitro by either nuclear proteins of gastrula embryos or in vitro-synthesized Xbra protein. Moreover, this putative T-box binding site actively responded to T-box protein induction. Our findings, therefore, provide insights into how Xmyf-5 activation takes place and how T-box proteins are involved in the regulation of muscle development.
Activation of myf-5 in Xenopus Embryos Requires a 372-bp Fragment in Xmyf-5 Upstream Region
Activation of myf-5 is the first key component of the myf-5 regulation network. Previously, we have demonstrated that the cis-elements required for the activation of myf-5 in Xenopus embryos resided in a 1.8-kb upstream region of the Xmyf-5 promoter (Yang et al., 2002). To further narrow down and identify the element(s) responsible for Xmyf-5 activation in vivo, we performed REMI transgenesis in Xenopus laevis embryos with serial deletion constructs of the 1.8-kb Xmyf-5 promoter (constructs diagrammed in Fig. 1A).
First, the integration efficiency of the REMI transgenesis was evaluated by using the p-1778GFP construct, whose expression was shown to be stable (Yang et al., 2002). An average integration efficiency of 66% was observed in the transgenesis, which was very similar to that reported previously (Kroll and Amaya, 1999). As shown in Figure 1C–E, and Table 1, the p-1778GFP reporter construct, which has been shown to contain the necessary cis-elements for the dorsal activation of Xmyf-5 (Yang et al., 2002), was able to drive reporter gene expression in the dorsal mesoderm of Xenopus gastrula embryos and the posterior myogenic mesoderm of embryos at later stages. A similar expression pattern of the reporter gene driven by a slightly shorter construct p-1662GFP was observed. By contrast, constructs p-1290GFP, p-1066GFP, and p-772GFP failed to give reporter gene expression in transgenic embryos. These results indicated that the 372-bp sequence between the nucleotides −1662 and −1290 of the Xmyf-5 promoter is essential for the Xmyf-5 activation in the dorsal half of the gastrula marginal zone and the maintenance of Xmyf-5 expression in the myogenic mesoderm of embryos at later stages.
Table 1. GFP Reporter Gene Expression of Various Constructs in REMI Embryos
Dorsal marginal zone expression (%)
No. expression or nonspecific expression
aREMI, restriction enzyme–mediated integration.
A Putative T-box Binding Site Is Responsible for the Activation of myf-5 in Xenopus Embryos
Consensus sequence analysis of the 372-bp DNA fragment of the Xmyf-5 promoter identified a putative T-box binding site (Tbx) located between nucleotides −1468 and −1459 (Fig. 1A). In Xenopus, Xbra has been shown to regulate eFGF expression through binding to a nonpalindromic element (Casey et al., 1998). Since then, several Xbra target genes were identified, including the Bix genes, Xwnt11, Xbtg1, and BIG3 (Tada et al., 1998, 2000; Saka et al., 2000). The sequence of the putative T-box binding site in the Xmyf-5 promoter exhibited high identity to the Xbra binding site in the promoter of these target genes (Fig. 1B, Casey et al., 1998, 1999; Tada et al., 1998). To test whether the putative T-box binding site we identified mediates the dorsal activation of Xmyf-5 in gastrula embryos, the pmTbxGFP reporter construct, in which Tbx was mutated by using site-directed mutagenesis, was subjected to REMI transgenesis in X. laevis embryos. The results showed that the expression of the reporter gene was significantly down-regulated in the pmTbxGFP REMI transgenic embryos (Fig. 1F–H; Table 1), indicating a requirement of the T-box binding site in the Xmyf-5 promoter for the dorsal activation of myf-5 in Xenopus gastrula embryos.
The Putative T-box Binding Site Can Specifically Interact With Nuclear Proteins of Gastrula Embryos and the T-Box Protein Xbra
Transcriptional regulation often requires physical interactions between the transcription regulator and its specific binding site in the target genes. To test whether the gastrula embryo contains proteins that can specifically interact with the putative T-box binding site, electrophoretic mobility shift assays (EMSAs) were carried out. Nuclear proteins extracted from stage 10.5–11 X. laevis embryos were incubated with a 42-bp probe containing the putative T-box binding site. A distinct shift complex was resolved by electrophoresis, whereas no DNA-protein complex was observed when the probe was incubated with bovine serum albumin (Fig. 2A). The shift complex was eliminated when 50-fold molar excess of unlabeled “cold” DNA probe was preincubated with nuclear extract. In contrast, the DNA fragment containing a mutated T-box binding site (as designated in pmTbxGFP/Luc) failed in the formation of this DNA-protein complex. As a parallel control, nuclear extract from mouse macrophage formed a distinguishable complex with this site (Fig. 2A).
To address whether this T-box binding site could be bound by T-box proteins, we performed EMSA with in vitro synthesized myc-tagged XbraDBD protein, which contains the entire DNA-binding domain of Xbra and had been shown to be able to bind to both palindromic and single T-box binding sites (Casey et al., 1998). This feature of XbraDBD protein was also displayed in the case of the T-box binding site in the Xenopus myf-5 promoter (Fig. 2B). Furthermore, when the XbraDBD protein was incubated together with anti–Myc-tag antibody, a supershift complex was yielded, whereas the shift complex remained the same when a control immunoglobulin (Ig) G antibody was used. Probe containing the mutant T-box binding site showed no binding with XbraDBD protein (data not shown). Thus, the EMSA analyses demonstrated the presence of nuclear proteins in Xenopus gastrula embryos that specifically interact with the putative T-box binding site in the Xmyf-5 promoter and the capability of the T-box proteins in binding to this T-box binding site.
Xmyf-5 Can be Induced by T-Box Genes
The REMI transgenesis results suggested that T-box proteins might regulate Xmyf-5 expression. To test this possibility, we first injected in vitro synthesized mRNA of Xbra, which has been shown necessary and sufficient to induce mesoderm formation (Smith, 2001), into the animal pole of 2-cell stage embryos. The induction of Xmyf-5 in animal caps harvested at stage 11 was assayed. As shown in Figure 3A, Xmyf-5 expression was induced by ectopic expression of Xbra. We next examined whether Xtbx6, a Xenopus Tbx6 homolog that has been implicated in the specification of myogenic mesoderm (Uchiyama et al., 2001), was also able to induce Xmyf-5 in animal cap assays. The result indicated that Xtbx6, when ectopically expressed in the animal cap, indeed induced Xmyf-5 expression (Fig. 3A). Moreover, induction of Xmyf-5 by Xbra seemed to be independent of protein synthesis. Although the induction of BIG3, or 1A11, by Xbra was inhibited by cycloheximide (Saka et al., 2000), induction of Xmyf-5 by Xbra was not affected (Fig. 3B). Induction of Xmyf-5 by Xtbx6 was not subjected to cycloheximide treatment, because the transcripts of Xtbx6 were barely detected before the onset of Xmyf-5 expression (Uchiyama et al., 2001).
The T-box Binding Site Is Responsive to T-box Proteins
The EMSA analyses suggested that the putative T-box site we identified in the Xmyf-5 promoter could be bound by proteins present in Xenopus gastrula embryos and the in vitro synthesized T-box protein Xbra. Members of the T-box family have been shown to be capable of regulating transcription by both transcriptional activation and repression. For instance, Brachyury has a complex structure of both activation and repression domains. To address whether the putative T-box binding site we isolated could respond to T-box proteins through transcriptional activation, we carried out Luciferase assays. First, the reporter constructs of p-1778Luc (which contains the intact T-box binding site) and pmTbxLuc (in which the T-box binding site was mutated) were injected into the dorsolateral marginal zone of 4-cell stage embryos. Reporter gene activity was measured at stage 12.5, when Xmyf-5 activity peaks. As shown in Figure 4A, reporter gene activity was decreased when the T-box binding site was mutated, indicating that the putative T-box binding site can respond to the inputs present in the dorsolateral marginal zone, where Xmyf-5 is expressed.
We next tested whether the putative binding site can respond to T-box proteins. Response of reporter constructs to the Xbra or Xtbx6 induction was measured in animal caps injected with reporter constructs and Xbra/Xtbx6 mRNA. Induction levels were determined by reporter gene activities as described (Yang et al., 2002). The reporter construct activity of p-1778Luc showed approximately a threefold induction to Xbra, and the response to Xbra induction was decreased when the T-box binding site was mutated in pmTbxLuc construct (Fig. 4B). However, when the 4×Tbx-TK and 4×mTbx-TK reporter constructs (in which four wild-type or mutant T-box binding sites, each flanked with two nucleotides at both ends, were combined and inserted into the upstream of TK promoter) were used, significant difference could be observed. The 4×Tbx-TK, upon Xbra/Xtbx6 induction, had a much higher fold induction and an average of eight times over that of 4×mTbx-TK was observed (Fig. 4C), suggesting that the putative T-box binding site, by itself, can respond to the input mimicked by the T-box protein Xbra or Xtbx6. However, this phenomenon was not observed in cell culture cotransfection assays (data not shown), probably because of the absence of necessary cofactors in the cell line used.
It is generally accepted that vertebrate myogenesis is coordinated by actions of several signaling inputs, including FGFs (Grass et al., 1996), BMPs, Wnts, and Shh (Marcelle et al., 1997). For instance, growing lines of evidence have suggested that the canonical Wnt/β-catenin pathway is involved in the regulation of myf-5 expression (reviewed in Cossu and Borello, 1999; Shi et al., 2002; Yang et al., 2002). Recently, Friday and Pavlath (2001) showed that a calcineurin/NFAT pathway could also regulate myf-5 gene expression in the skeletal muscle reserve cells. Sonic Hedgehog, which is emanated from the notochord, is also believed to participate in the regulation of myf-5 expression (Borycki et al., 1999; Coutelle et al., 2001), and probably functions as a direct upstream regulator (Gustafsson et al., 2002).
Accumulating evidence suggests that T-box transcription factors are involved in the mesoderm patterning and myogenesis, in both vertebrate and invertebrate animals. Xbra has been shown necessary and sufficient for normal mesoderm development in Xenopus embryos (reviewed in Smith, 2001). In mouse, Tbx6 functioned in the paraxial mesoderm formation, and mutation in Tbx6 resulted in the formation of three neural tubes at the expense of the posterior somites (Chapman and Papaioannou, 1998). The Tbx6-related gene As-T2 in ascidian is expressed in muscle precursor cells and is associated with embryonic muscle cell differentiation (Mitani et al., 2001). ske-T, a member of T-brain subfamily recently cloned in the sea urchin, is expressed in the skeletal mesenchyme and is also implicated in the process of muscle development (Croce et al., 2001). Our findings presented in this study also demonstrate a role of T-box proteins in myogenesis by regulating MRF factors, like myf-5. We demonstrated that Xbra and Xtbx6 could induce Xmyf-5 expression in animal caps, and the induction of Xmyf-5 by Xbra did not depend on protein synthesis (Fig. 3). Through a transgenic approach, we also identified a putative T-box binding site in the Xmyf-5 promoter and showed that this T-box binding site was required for the activation of myf-5 gene expression in Xenopus embryos. Furthermore, the putative T-box binding site could specifically bind to nuclear proteins prepared from gastrula stage embryos, and in vitro synthesized Xbra protein, as shown by the EMSAs. Finally, the T-box binding site was able to actively respond to T-box proteins induction (Fig. 4).
Compared with the mouse myf-5 gene, whose regulatory region appears to spread out over more than 90 kb (see Carvajal et al., 2001, and reference therein), the smaller size of the Xenopus myf-5 regulatory contig (Yang et al., 2002) is very advantageous for further delineation of the transacting factors and signaling inputs involved in the transcriptional control of myf-5 during primary myogenesis. Based on our previous work and the findings we report here, a simplified model of Xmyf-5 expression regulation is proposed (Fig. 5). We propose that Xmyf-5 expression is first activated by T-box factor(s), such as Xbra, and is further orchestrated to achieve its unique expression pattern. A distal TCF binding site may mediate the repression of Xmyf-5 in midline mesoderm, probably by the antagonism of Frezb to Wnt8 (Yang et al., 2002). The absence of Xmyf-5 expression in the ventral gastrula mesoderm may result from a repression by Xvent (Polli and Amaya, 2002). In neurula and later stages, down-regulation of Xmyf-5 in differentiating myocytes is mediated, at least partially, by an IRF-like DNA binding element (Mei et al., 2001).
However, it is most likely that other factors are also involved in the regulation of myf-5 expression during myogenesis. For example, the expression pattern of the p-1778GFP reporter construct (Fig. 1C) in the transgenic embryos was restricted within the dorsal marginal zone, whereas endogenous myf-5 expression pattern extends into the ventral half. This pattern of Xmyf-5 expression is particularly obvious before the mid-gastrula stage when the myf-5 expression reaches its ventral-most extension. The factor(s) involved in this extension is still unknown. Activin/Nodal signaling has been implicated in the activation of Xmyf-5 in the ventral mesoderm of gastrula embryos (Yang et al., 2002), but the mechanism remains unclear.
Constructs and Mutagenesis
The serial deletion constructs were generated by inserting restriction fragment of FL-SK into the HindIII and BglII (blunted by Mung Bean nuclease) restriction sites of the pGL3-basic reporter plasmid, as shown in Figure 1A. For the transgenic reporter assays, the luciferase gene was removed with XbaI and HindIII and replaced by the GFP cDNA (Zernicka-Goetz et al., 1996).
The GeneEditor site-directed mutagenesis system (Promega, catalog no.Q9280) was used to generate the mutant of the putative T-box binding site, pmTbxGFP/Luc. The primer shown below and its complement were used (nucleotides altered shown in bold): 5′-gccgagtgtgtaatgaacGGGCCCgaaaaggtcccacaaatatcgcag-3′. The alteration of nucleotides to GGGCCC introduced an ApaI restriction endonuclease site for facilitating positive clone selection.
To generate 4×Tbx-TK construct, oligos 5′-cgcgtacatgtgtgaaaagacatgtgtgaaaagacatgtgtgaaaagacatgtgtgaaaa-3′ and 5′-gatcttttcacacatgtcttttcacacatgtcttttcacacatgtcttttcacacatgta-3′ were annealed and inserted into the MluI and BglII restriction sites before the TK promotor in TK-basicLuc (Mei et al., 2001). For 4 × mTbx-TK, oligos used were: 5′-cgcgtacGGGCCCgaaaagacGGGCCCgaaaagacGGGCCCgaaaagacGGGCCCgaaaa-3′ and 5′-gatcttttcGGGCCCgtcttttcGGGCCCgtcttttcGGGCCCgtcttttcGGGCCCgta-3′.
Restriction Enzyme–Mediated Integration Transgenesis and In Situ Hybridization
REMI transgenic Xenopus laevis embryos were generated as described previously (Kroll and Amaya, 1999). Plasmids used for transgenesis were linearized by NotI digestion. In situ hybridizations were performed as described (Steinbach et al., 1998).
Xenopus Embryo Manipulation, Quantitation of Reporter Gene Activity, and Reverse Transcriptase-Polymerase Chain Reaction
Eggs were obtained from Xenopus females, cultured, and staged as described previously (Ding et al., 1998; Nieuwkoop and Faber, 1967). Reporter constructs were injected into the dorsolateral marginal zone of 4-cell stage embryos or the animal pole of 2-cell stage embryos with or without Xbra/Xtbx6 plasmids. Animal caps were prepared at stage 8.5 and harvested at mid-gastrula stage for reverse transcriptase-polymerase chain reaction (RT-PCR) or Luciferase reporter assays as described (Yang et al., 2002). Primers for BIG3 were 5′-ccccagtctttcagcaca-3′ (sense) and 5′-cttccacccactcccttc-3′ (antisense).
Protein Synthesis Inhibition Expriment
Embryos injected with Xbra mRNA were cultured in 0.1×MBS until stage 7 and then subdivided into two groups. One group of embryos was treated with 5 μg/ml cycloheximide (Sigma, C-7698) in 0.1×MBS until stage 8, when animal caps were dissected. The animal caps were cultured in 0.5×MBS containing 5 μg/ml cycloheximide and harvested for RT-PCR when sibling embryos reached stage 11. The embryos of the other group were treated similarly but without cycloheximide incubation. Uninjected animal caps were prepared as control.
Electrophoretic Mobility Shift Assay
Electrophoretic mobility shift assays were performed essentially as described previously (Yang et al., 2002). Xenopus laevis embryo nuclear extracts were prepared from stage 10.5–11 embryos as described (Barth et al., 1998). XbraDBD protein was in vitro synthesized by using Promega TNT coupled reticulocyte lysate system (L4600). Used were 5 μg of nuclear extract or 3 μl of the lysate equilibrium in each reaction and 0.5×TBE. For supershift assay, 2 μg anti-Myc antibody (a gift from J. Yang, University of Pennsylvania) or control IgG were used. The single-strand sequence of the Tbx DNA probe was 5′-tgtgtaatgaacatgtgtgaaaaggtcccacaaatatcgcag-3′, and the sequence of the mutant mTbx DNA was 5′-tgtgtaatgaacGGGCCCgaaaaggtcccacaaatatcgcag-3′. The palindromic Brachyury sequence was 5′-attagtcacacctaggtgtgaagagcc-3′.
The authors thank Dr. Jim Smith (the Wellcome/CRC Institute) for constructs. We also thank Dr. Wenyan Mei and Dr. Jing Yang (Howard Hughes Medical Institute, University of Pennsylvania) for discussion of the project, and Dr. YiPing Chen (Tulane University) and Dr. PanFeng Fang for critical reading of the manuscript. X.D. received funding from the National Natural Science Foundation of China and the Bureau of Life Science and Biotechnology, Chinese Academy of Sciences.