eFGF is required for activation of XmyoD expression in the myogenic cell lineage of Xenopus laevis

An enormous breakthrough in our understanding of how the myogenic cell lineage is established came when the myogenic regulatory genes MyoD, Myf5, myogenin and Mrf4 were identified. These genes encode bHLH transcription factors, known as MRFs (myogenic regulatory factors), which act as dominant regulators of myogenesis: expression of any one of these genes in many non-muscle cell types will convert those cells to myoblasts (reviewed by Weintraub, 1993). Moreover, mouse knockout experiments have shown that MRFs are essential for myogenesis in vivo (reviewed by Arnold and Winter, 1998). MyoD (Myod1 – Mouse Genome Informatics) and Myf5 in particular are important for early events during myogenesis, as mice deprived of both Myod1 and myf5 do not form any skeletal muscle precursor cells (Rudnicki et al., 1993). Consistent with a role in determining the skeletal muscle cell lineage, MyoD and Myf5 are expressed in myogenic cells in all vertebrates but the precise regulation of gene expression varies among different vertebrates.

In mice, Myf5 is the earliest expressed MRF and is initially found in the epaxial domain of the epithelial somite at 8dpc, while MyoD expression is detected in somites significantly later at 10.5 dpc (Ott et al., 1991). In avians, high expression of both Myf5 and MyoD is closely associated with somitogenesis (Hirsinger et al., 2001; Pownall and Emerson, 1992). However, other recent studies have detected expression of Myf5 (Kiefer and Hauschka, 2001) and MyoD (Gerhart et al., 2000) in pre-segmented mesoderm and gastrula stage embryos. Transcription of Myf5 has also been detected in mouse pre-segmented mesoderm by RT-PCR (Lin-Jones and Hauschka, 1996) and in pre-segmented mesoderm of transgenic mice where lacZ has been knocked into the Myf5 locus (Cossu et al., 1996). The relevance of these low levels of MRF gene expression in early chick and mouse mesoderm is at present unknown, however, in frogs and fish MyoD and Myf5 are expressed at high levels in the mesoderm prior to somitogenesis (Chen et al., 2001; Hopwood et al., 1991; Hopwood et al., 1989; Weinberg et al., 1996). The exact nature of the signals that regulate MRF gene expression in the early mesoderm remains to be determined.

XmyoD is expressed dynamically during Xenopus development: there are maternal XmyoD transcripts present throughout the egg and when zygotic transcription begins, a low level of transient, ubiquitous XmyoD expression initiates (Harvey, 1990; Harvey, 1991; Rupp and Weintraub, 1991). Expression of XmyoD in the myogenic lineage is activated during early gastrula stages, where high levels of expression are detected in the cells of the marginal zone and XmyoD continues to be expressed in myogenic cells throughout somitogenesis (Hopwood et al., 1989). The skeletal muscle specific expression of XmyoD is activated in response to endogenous mesoderm inducing signals (Hopwood et al., 1989) and the growth factors bFGF and activin have also been shown to activate XmyoD expression (Harvey, 1991). In Xenopus, it is possible to determine whether the expression of a particular gene is a direct response to mesoderm inducing signals (Cascio and Gurdon, 1987; Rosa, 1989; Smith et al., 1991). It has been reported that the expression of XmyoD in response to factors such as activin, which mimic the endogenous mesoderm inducing signal, is a delayed response, requiring protein translation (Harvey, 1991; Steinbach et al., 1998), rather than a direct, immediate early response, as has been shown for Mix1 (Rosa, 1989) and Xbra (Smith et al., 1991).

FGF signalling has been shown to be required for the expression of many mesodermal genes (Amaya et al., 1991; Amaya et al., 1993) and notably to directly regulate the Xenopus homologue of the T-box gene brachyury (Xbra) (Isaacs et al., 1994; Schulte-Merker and Smith, 1995; Smith et al., 1991) as well as the caudal homologue Xcad3 (Isaacs et al., 1998; Pownall et al., 1996). Previous work also suggests a role for FGF signalling in regulating XmyoD: (1) XmyoD expression is activated in animal cap explants treated with bFGF (Harvey, 1990) and the induction of XmyoD in explants by activin requires a functional FGF signalling pathway (Cornell and Kimelman, 1994); (2) in whole embryos, blocking FGF signalling results in a dramatic downregulation of XmyoD expression (Isaacs et al., 1994) and loss of skeletal muscle (Amaya et al., 1991; Amaya et al., 1993); and (3) a community effect among muscle precursor cells is required for myogenic differentiation (Gurdon et al., 1993) and has recently been shown to be mediated by eFGF (Standley et al., 2001).

We present evidence that eFGF is necessary for the initial activation of XmyoD expression during Xenopus myogenesis. We show that during early gastrula stages, XmyoD and embryonic FGF (eFGF), which is the Xenopus homologue of Fgf4, are co-expressed in the nascent mesoderm and that XmyoD is activated as an immediate early response to eFGF. Moreover, we use morpholino antisense oligonucleotides directed against eFGF to show that the specific inhibition of eFGF eliminates early expression of XmyoD. We further show that the expression of eFGF itself is a direct response to mesoderm induction by activin-like signalling molecules and propose a molecular pathway for XmyoD activation in the early mesoderm. We also present evidence that the expression of XmyoD may be regulated by an unstable transcriptional repressor.

Xenopus laevis embryos were obtained by artificial fertilisation and cultured in 0.1×normal amphibian medium (NAM) at 23°C. Embryos were de-jellied in a 2.5% cysteine solution and staged according to (Nieuwkoop and Faber, 1967). Animal cap explants were taken at NF stage 7.5-8.5. Manipulations were performed in 0.5×NAM + 0.5% bovine serum albumen (BSA). Caps were cultured in 0.5×NAM + 0.5% BSA, with and without growth factor, (eFGF 20 units/ml or Activin 13 ng/ml). Whole embryos and caps were frozen at the appropriate control stage for later RNAase protection analysis.

For experiments involving cycloheximide (CHX) treatment animal cap explants were taken at stage 8-8.5 and cultured for 30 minutes in 0.5×NAM + 0.5% BSA, with and without 7.5 μg/ml CHX. Control explants were left in culture while explants for growth factor treatment were cultured for a further 90 minutes with 20 μ/ml eFGF or 13.5 ng/ml activin in 0.5×NAM + 0.5% BSA, with and without 7.5 μg/ml CHX. After growth factor and CHX treatment, caps were washed twice in 0.5×NAM + 0.5% BSA and cultured in this medium until the appropriate control stage.

Animal cap culture and CHX treatment was as described above except 15 μCi/ml ³⁵S-methionine was included in the culture medium following the initial 30 minute CHX exposure. Control and CHX treated caps were collected at the appropriate control stage. Caps were homogenised in 10 μl of 0.1 M NaCl, 50 mM Tris (pH 8), 5 mM EDTA and 0.5% SDS (NETS) and 5 μl of homogenate was spotted onto filter paper. Filters were then washed in 10% trichloroacetic acid (TCA) for 30 minutes, twice in water for 10 minutes, once in 95% ethanol for 10 minutes and once in acetone for 10 minutes. Filters were then air dried and activity measured by scintillation counter. Inhibition of radio-methionine incorporation was typically 80-90% over the time course of the experiment.

The CM control morpholino oligonucleotide is a standard control provided by GeneTools. The eFM oligonucleotide was designed by GeneTools and has the sequence ATGGAACAGTCATCCC-GATCAAC, which is complementary to the nucleotides –10 to +13, which span the initiating codon AUG of eFGF(i). Injections were carried out on four-cell stage embryos cultured in 0.3×NAM + 5% Ficoll 400. Bilaterally injected embryos, for use in RNAase protection analysis, were injected with up to 20 ng into the pigment boundary of each cell. Unilaterally injected embryos, for analysis by in situ hybridisation, were injected with 20 ng into each cell on the left of the embryo. Oligonucleotide-injected embryos for use in animal cap experiments were injected with 40 ng into the animal region of both cells at the two-cell stage. Animal caps for protein treatment taken at stage 8 and were cultured with and without bFGF protein (20 units) until the appropriate control stage. eFGF mRNA co-injection experiments used synthetic capped eFGF mRNA made from the pSP64T eFGF (un) plasmid which contains the 5′UTR and coding region of eFGF(i). 10 pg eFGF(i) mRNA was injected with and without oligonucleotide into both cells at the two-cell stage. Animal caps were excised at stage 8 and cultured until the appropriate control stage. ssbFGF mRNA co-injection experiments used synthetic capped ssbFGF mRNA as described in (Thompson and Slack, 1992). ssbFGF (100 pg) was injected with and without oligonucleotide into both cells at the two-cell stage embryo. Embryos were cultured until stage 13.

mRNA was extracted in NETS solution followed by phenol/chloroform extraction and ethanol precipitation. RNAase protection analyses were performed using the Ambion RPA III^TM kit with a hybridisation temperature of 45°C. Ornithine Decarboxylase (ODC) mRNA was used as a loading control for all RNAase protections. Templates prepared were Actin (Gurdon et al., 1985), eFGF (Isaacs et al., 1992), XmyoDb (Harvey, 1991), Xmyf5 linearised with EcoR1 and transcribed using T7 polymerase, Xbra (Smith et al., 1991), ODC (Isaacs et al., 1992) and Xsna (Sargent and Bennett, 1990).

Embryos were cultured to appropriate stages and then fixed in MEMFA (0.1 M MOPS, 2 mM EDTA, l mM MgSO₄, 3.7% formaldehyde) for l hour at room temperature and stored in l00% ethanol at –20° C until further processing. Embryos were rehydrated through a graded series of ethanols and then rinsed in PBS with 0.1% Tween. Proteinase K treatment was carried out for 10 minutes at room temperature with l0 μg/ml of Proteinase K. Hybridisation was carried out overnight at 60°C in 50% formamide, 5×SSC, 1 mg/ml total RNA, 100 μg/ml heparin, 1×Denhardts, 0.1% Tween 0.1% CHAPS, 10 mM EDTA. Extensive washes in 2×SSC and 0.2×SSC at 60°C were followed by washes at room temperature with maleic acid buffer, MAB (0.1 M maleic acid, 0.15 M NaCl, 0.1% Tween, pH 7.8), and blocking in 2% Roche Blocking Reagent and 20% heat-treated lamb serum for 2 hours at room temperature. Embryos were then incubated with anti-DIG antibody at a dilution of 1/2000 in blocking solution at 4°C overnight. The antibody is detected after extensive washes at room temperature in MAB by a colour reaction using BM purple precipitating alkaline phosphatase detection system (Roche). Probes for in situ hybridisation were transcribed using l×DIG RNA labelling mix (Roche) from linearised plasmids: XmyoDb as for RNAase protection (Harvey, 1991); eFGF as for RNAase protection (Isaacs et al., 1992); Xsna as for RNAase protection (Sargent and Bennett, 1990).

As discussed above, there is evidence to suggest a role for FGF signalling during myogenesis. We show that the expression domains of XmyoD and eFGF overlap in the early mesoderm. Both genes are initially co-expressed within a region that encompasses much of the mesoderm (Fig. 1).

Using an animal cap assay, we show that eFGF is sufficient to activate the expression of XmyoD. Ectodermal explants (or ‘animal caps’), which when cultured alone give rise to only ectodermal derivatives, were treated with eFGF protein and assayed for the expression of XmyoD and Xbra by RNAase protection analysis. Untreated animal caps do not express mesodermal genes, while the expression of the pan-mesodermal gene Xbra is a well documented response to induction by FGF signalling (Isaacs et al., 1994; Schulte-Merker and Smith, 1995; Smith et al., 1991). Treatment with 20 units of eFGF protein activates the expression of both XmyoD and Xbra in ectodermal explants (Fig. 2). This activation is rapid and occurs by early gastrula stage 10, becoming stronger at later gastrula stage 11.5. Another study found that injection of synthetic bFGF mRNA (coding for a secreted form of bFGF) was unable to induce XmyoD expression in animal caps without the co-injection of Xwnt8 mRNA (Steinbach et al., 1998). However, we show here that eFGF treatment alone is sufficient to rapidly activate the expression of XmyoD.

As previously reported, the TGFβ signalling molecule, activin, which is known to have strong mesoderm inducing activity, also induces the expression of XmyoD and Xbra in this assay (Harvey, 1991; Smith et al., 1991). However, the expression of these genes over the timescale shown here is much weaker in response to activin than to eFGF. Control animal cap explants show a low level of XmyoD expression during early gastrula stages, due to the transient, ubiquitous pulse of XmyoD expression that occurs in Xenopus embryos at the mid-blastula transition when the zygotic genome becomes active (Harvey, 1990; Rupp and Weintraub, 1991). This low level expression in control caps is no longer apparent by mid to late gastrula stages, so in most of the following experiments we assay for gene expression at late gastrula stage 13.

We demonstrate here that the expression of XmyoD is a direct response to FGF signalling by repeating our induction assay in the presence of the translational inhibitor cycloheximide. This type of analysis has been carried out previously to determine which genes are activated by mesoderm inducing factors in an immediate-early, or cycloheximide-insensitive, manner (Cascio and Gurdon, 1987; Rosa, 1989; Smith et al., 1991). It has been well documented that the expression of XmyoD is not an immediate-early response to mesoderm induction by activin-like molecules (Harvey, 1991; Steinbach et al., 1998). However, we demonstrate here that the expression of XmyoD in response to FGF signalling does not require protein synthesis, as the activation of XmyoD by eFGF protein is insensitive to cycloheximide (Fig. 3A). This is also true for the expression of Xbra, which was shown previously to be an immediate-early response to FGF signalling (Smith et al., 1991). Transcription of actin does require protein synthesis and is sensitive to cycloheximide (Fig. 3B).

Interestingly we find that expression of XmyoD is super-induced in animal caps treated with eFGF and cycloheximide. We also see that animal caps treated with cycloheximide alone express XmyoD (Fig. 3A). Our favoured interpretation of these results is that the regulation of XmyoD involves an unstable transcriptional repressor and in the presence of cycloheximide the rapid turnover of this factor is not replenished so that XmyoD is no longer repressed. Other studies investigating XmyoD expression in animal caps treated with cycloheximide have not described this phenomenon (Harvey, 1991). The timing of our analysis is likely the basis of this difference, as animal caps treated with cycloheximide alone express XmyoD during early and late gastrula stages, but no longer express XmyoD at neurula stages (Fig. 3B), which is when other studies assayed gene expression (Harvey, 1991).

While we have shown that eFGF is sufficient to activate the expression of XmyoD, there are many FGF family members that could be playing this role in vivo. eFGF, Fgf3 and Fgf8 are expressed in the early mesoderm of Xenopus laevis (Christen and Slack, 1997; Isaacs et al., 1995; Lombardo et al., 1998), and all could have some role in regulating XmyoD expression. In order to determine if it is eFGF that is required for XmyoD expression during normal development, we have adopted an antisense inhibition approach using morpholino oligonucleotides directed against eFGF. Injection of antisense morpholino oligos into Xenopus embryos has previously been shown to be an effective way of inhibiting translation from specific target mRNAs during early development (Heasman et al., 2000). We designed a 23-mer morpholino oligonucleotide, designated eFM, which is directed against a region spanning the initiating AUG of the eFGF(i) mRNA.

Although our experiments seek to analyse the role of eFGF after the period of mesoderm induction, the well characterised mesoderm-inducing activity of eFGF provides an ideal assay to test out the efficacy and specificity of the eFM morpholino oligonucleotide. In order to demonstrate that injection of eFM can inhibit the activity of eFGF mRNA, we took animal cap explants that were injected with synthetic mRNA coding for eFGF alone or co-injected with eFGF mRNA plus eFM or a control morpholino oligo (CM). Uninjected control animal caps remain as a rounded ball of ectodermal tissue (Fig. 4A, part i). Animal cap explants that are injected with eFGF mRNA are induced, and undergo morphogenetic movements and elongation (Fig. 4A, part ii) that are characteristic of the cellular behaviour of mesoderm during gastrula stages. Caps co-injected with eFGF mRNA and CM (Fig. 4A, part iii) also undergo elongation movements. However, explants that are co-injected with eFGF mRNA and the eFM antisense morpholino oligo are uninduced (Fig. 4A, part iv) and look much the same as the uninjected control explants.

Animal cap elongation is an accurate indicator of mesoderm formation, but we further analysed the effect of eFM by looking at the ability of eFM to block mesodermal gene expression induced by injection of eFGF mRNA. Fig. 4B shows that co-injection of eFM, but not CM, with eFGF mRNA blocks the activation of expression of both XmyoD and Xbra in gastrula stage animal caps. We conclude that the eFM oligo, but not the CM control oligo, inhibits the ability of the injected eFGF mRNA to induce mesoderm.

The above experiments do not rule out the possibility that the eFM oligo inhibits all mesoderm induction in a nonspecific fashion. To test the specificity of eFM we show that eFM does not block the ability of exogenously supplied FGF protein to induce mesoderm. We find that control caps or caps injected with CM or eFM, treated with exogenous FGF protein, show normal induction of mesodermal gene expression (Fig. 4C). We conclude that the eFM oligo does not block mesoderm induction in a nonspecific fashion.

We next showed that eFGF is required for XmyoD expression in vivo. Using RNAase protection analysis, we found that although 80 ng of CM has no effect on gene expression, the same amount of eFM completely eliminates XmyoD expression during gastrula stages (Fig. 5A, part i). This effect is not a general inhibition of mesodermal gene expression as another pan-mesodermal gene, Xsna, continues to be expressed in embryos injected with eFM. At later stages, the expression of XmyoD is no longer inhibited by the eFM oligo. In Fig. 5A, part ii we show, as a control for specificity, the rescue of XmyoD expression with the co-injection of an mRNA coding for a secreted form of basic FGF (ss-bFGF) (Thompson and Slack, 1992), a protein with similar biological activity to eFGF but with different sequence.

The loss of inhibition at later stages is also seen when the eFM oligo is injected unilaterally into two cells at the four-cell stage and gene expression is assayed at gastrula and neurula stages by whole mount in situ hybridisation (Fig. 5B). Expression of XmyoD is found to be completely abolished on the eFM injected side of gastrula embryos (Fig. 5B, part i). The expression of Xsna is unaffected (Fig. 5B, part iii). The injected side of sibling embryos allowed to develop to neurula stages shows the expression of XmyoD is recovering on the injected side (Fig. 5B, part ii). These data support the notion that while the activation and early expression of XmyoD in the nascent mesoderm requires eFGF, the later maintenance of XmyoD expression in the paraxial mesoderm is independent of eFGF and is regulated by other mechanisms.

We have shown that eFGF is required in vivo for the activation and early expression of XmyoD, and that XmyoD expression is a direct response to induction by eFGF. To further characterise the molecular pathway leading to XmyoD expression we examine if eFGF expression is an immediate early response to mesoderm induction by activin-like molecules. Fig. 6 shows the results of an animal cap experiment where we test the ability of activin protein (which is used to mimic the endogenous mesoderm inducing factor) to activate the expression of eFGF. We find that activin does activate eFGF expression even in the presence of cycloheximide.

We have addressed the molecular mechanism for the transcriptional activation of XmyoD in cells of the myogenic lineage of the frog Xenopus laevis. In Xenopus, the earliest expression of XmyoD in the skeletal muscle lineage occurs in the nascent mesoderm at the start of gastrulation (Hopwood et al., 1989). This expression is notably earlier than in other vertebrates: in avians MyoD expression is associated with somitogenesis (Hirsinger et al., 2001; Pownall and Emerson, 1992) and in mice MyoD is expressed in somites sometime after they have formed (Ott et al., 1991). While there is some evidence for low-level MyoD transcription in avians prior to somitogenesis (Gerhart et al., 2000), it is clear that both frogs and fish express MyoD at high levels in the newly formed mesoderm well before the onset of somitogenesis. (Hopwood et al., 1989; Weinberg et al., 1996).

Although XmyoD is expressed in the early mesoderm, it has been reported that the expression of XmyoD is not a direct response to mesoderm induction (Harvey, 1991; Steinbach et al., 1998). Current thinking is that the endogenous mesoderm inducing signals are members of the TGFβ family of growth factors and may include the Xenopus nodal related factors (Jones et al., 1995; Kofron et al., 1999). Historically the FGFs were also considered to play a part in this role; however, as the mesoderm inducing activity of FGF was first described (Slack et al., 1987), the role of FGF signalling during the establishment of the mesoderm has been re-evaluated (Cornell and Kimelman, 1994; Cornell et al., 1995; Isaacs, 1997; Isaacs et al., 1994). Despite its activity in mesoderm induction assays, the bulk of evidence points to a later role for FGFs in regulating mesodermal gene expression during gastrula stages, after the initial phase of mesoderm induction. In this study we examine a possible role for FGF signalling in regulating XmyoD expression within the early mesoderm. It has previously been shown that blocking FGF signalling inhibits myogenesis (Amaya et al., 1991; Amaya et al., 1993) and dramatically downregulates XmyoD expression (Isaacs et al., 1994) in developing embryos. It has been also been demonstrated that eFGF is able to maintain the expression of XmyoD in disaggregated mesoderm cells of gastrula stage embryos (Standley et al., 2001).

This work demonstrates that eFGF is necessary and sufficient to activate XmyoD expression in the early mesoderm. Treatment of animal cap ectodermal explants with eFGF rapidly activates the expression of XmyoD and can do so even in the presence of the translational inhibitor cycloheximide, indicating that this induction is direct. Interestingly, we have some preliminary data that show that Xmyf5 is not an immediate early response to eFGF (data not shown) and is therefore regulated by a distinct mechanism.

An unexpected finding from some of our experiments is that cycloheximide treatment alone activates expression of XmyoD during gastrula stages. One interpretation of this finding is that a labile transcriptional repressor inhibits XmyoD expression prior to its activation by eFGF. Such unstable inhibitors have been postulated to be involved in the regulation of other genes (Novak et al., 1991). Alternatively, cycloheximide has been shown to activate other signal transduction pathways and could be activating XmyoD in this way (Edwards and Mahadevan, 1992). Clearly, however, cycloheximide is not a general activator of all FGF target genes, as expression of Xcad3 (Isaacs et al., 1998) and Xbra (Smith et al., 1991) is not upregulated by cycloheximide treatment alone.

We favour the idea of an unstable repressor of XmyoD for several reasons. During blastula stages, there is a transient low-level burst of ubiquitous XmyoD expression that is quickly silenced (Harvey, 1990; Rupp and Weintraub, 1991). High levels of mesoderm induction-dependent XmyoD expression in the cells of the myogenic lineage could be activated through the inactivation of the transcriptional repressor by modification via FGF signalling. Rupp and colleagues have demonstrated a dominant mechanism restricting XmyoD expression in response to known inducers until early gastrula stages, and, although auto-regulation by the XmyoD protein was shown to be direct, its ability to activate XmyoD transcription is also restricted until these later stages (Steinbach et al., 1998). The human MYOD (MYOD1 – Human Gene Nomenclature Database) enhancer drives very specific skeletal muscle expression in transgenic mice. However, in tissue culture cells expression driven by the same enhancer is remarkably promiscuous, possibly reflecting the loss of a repressor needed to silence MyoD in these derived, non-muscle cell lines (Goldhamer et al., 1992).

Consistent with a role for regulating XmyoD in the early mesoderm, several members of the FGF family of signalling molecules (including eFGF, FGF3 and FGF8) are co-expressed with XmyoD as a ring around the blastopore in the newly formed mesoderm (Christen and Slack, 1997; Isaacs et al., 1995; Lombardo et al., 1998). To determine if eFGF specifically is necessary for XmyoD expression in the early mesoderm, we used a morpholino-based antisense oligonucleotide directed against eFGF(i). Our data show that inhibiting eFGF activity results in no XmyoD expression in gastrula mesoderm; however, at later stages of development, XmyoD expression recovers. This may indicate that later mechanisms for maintaining XmyoD expression in the paraxial mesoderm are distinct. Another interpretation is that the antisense oligo is no longer functioning at these later stages; however, work from other laboratories has shown that morpholino oligos are stable and functioning until at least tailbud stages (NF St20+) (Heasman et al., 2000). We favour the interpretation that while XmyoD expression during gastrula stages requires eFGF, later maintenance of XmyoD in the paraxial mesoderm and somites is independent of eFGF and regulated by another molecular mechanism.

Recently, good evidence has been reported for eFGF-mediated regulation of XmyoD during gastrula stages. Standley et al. (Standley et al. 2001) tested members of several different signal transduction pathways for the ability to maintain XmyoD expression in disaggregated mesoderm, and found that only eFGF was able to mimic the endogenous community effect (Gurdon et al., 1993; Standley et al., 2001). Their work shows that continued presence of eFGF protein in disaggregated mesoderm cultures is necessary during gastrula stages to maintain high levels of XmyoD expression and to substitute for the community effect. Consistent with this, we find that although a pulse of eFGF is able to activate XmyoD expression in ectodermal explants (see Fig. 3A), prolonged exposure to eFGF results in higher levels of XmyoD expression (see Fig. 2). Our findings further support the work of Standley et al. (Standley et al., 2001) in that we show that eFGF-mediated regulation of XmyoD expression is distinct and downstream from mesoderm induction by an activin-like signal. However we extend this consensus by showing that eFGF is required in vivo for XmyoD expression and that the expression of XmyoD in the myogenic lineage is an immediate-early response to induction by eFGF.

Induction and patterning of the early mesoderm has been well studied in frogs and combinations of signalling factors have been shown to act during gastrula stages to pattern the dorsoventral axis. BMP4 and Xwnt8 activity in the ventrolateral mesoderm is modulated by factors such as noggin, chordin and FrzB secreted from the dorsal most mesoderm to promote more dorsolateral fates, such as skeletal muscle (reviewed by Heasman, 1997; Slack, 1994; Smith, 1995). Xwnt8 has been shown to be necessary for XmyoD expression during gastrula stages (Hoppler, 1996). However, Xwnt8 signalling is not sufficient for normal myogenesis in vivo. We have previously shown that embryos in which FGF signalling has been blocked do not express XmyoD, despite expressing Xwnt8 at near normal levels (Isaacs et al., 1994).

Some of these same signals important for myogenesis in frogs have also been shown to play a role in regulating MRF gene expression in somites of higher vertebrates (Munsterberg et al., 1995; Reshef et al., 1998). An intriguing question is whether the molecular mechanisms involving FGF regulation of XmyoD in frogs are conserved in higher vertebrates? It is possible that as the early mesodermal expression of MyoD has been largely lost, perhaps this mode of MyoD activation is also no longer important. However there are some data that support the notion that FGF signalling is involved in regulating myogenesis in chick and mouse. FGFs have long been known to be mitogenic for cultured myoblasts (Linkhart et al., 1981) while inhibiting terminal differentiation. This is also true for muscle-forming cells in the chick limb (Seed and Hauschka, 1988). FGF receptors have been found to be expressed in the segmental plate mesoderm and newly formed somites in the mouse (Yamaguchi et al., 1992). Furthermore, an antibody that can neutralise bFGF has been shown to inhibit dorsal neural tube-derived myogenic signals in a chick explant study (Stern et al., 1997). These data support the notion of some role for FGF signalling during the regulation of myogenesis in chicks and perhaps mice, but the level at which this regulation occurs remains to be determined.

Much work has been undertaken to investigate the maternal factors involved in setting off a cascade that specifies the mesoderm in Xenopus laevis (Heasman, 1997). Depletion of VegT, a maternally deposited T-box transcription factor, results in the extinction of expression of many mesodermal genes, including XmyoD and eFGF. However, when the TGFβ signalling molecule Xnr2 is introduced into VegT depleted frog embryos, the expression of XmyoD and eFGF is rescued (Kofron et al., 1999). Consistent with this, a mutation in the zebrafish homologue of VegT, spadetail, results in fish which lack early expression of MyoD (Weinberg et al., 1996).

In this study we use activin to mimic the nodal-related endogenous mesoderm inducing factor, and show that eFGF expression is a direct response to mesoderm inducing signals. We also show that eFGF can activate XmyoD expression even in the presence of protein synthesis inhibitors and that furthermore, eFGF is required for the activation of XmyoD in vivo.

We are now in a position to propose a molecular pathway from maternally deposited factors to myogenic specification (Fig. 7). Vegetally localised maternal factors such as VegT regulate the expression of TGFβ signalling molecules such as the nodal related factor Xnr2. This factor(s) induces mesoderm during blastula stages and activates expression from a subset of genes, including eFGF, in the nascent mesoderm. eFGF induces the expression of XmyoD in the mesoderm and specifies the cells of the myogenic lineage, possibly via the inhibition of a labile transcriptional repressor of XmyoD.

We thank John Gurdon for helpful discussions and for sharing data prior to publication. This work was funded by a grant from the Wellcome Trust to M. E. P. M. E. F. is supported by a BBSCR studentship.