In beginning of muscle development, determination is induced in the mesoderm, and then differentiation occurs with accumulation of muscle structural proteins. Mesoderm cells differentiate to many type cells, but the direct signaling activator for muscle determination is still unknown. In this paper we report some of the conditions required for determination of muscle. Muscle determination during Xenopus development was found to be marked by Xmyf-5 and XmyoD expression, but not by expression of Xmyogenin or Xmrf4. Xmyf-5 and XmyoD expression was first detected in the early gastrula stage. Xmyf-5 expression was first detected on the dorsal side, whereas XmyoD was initially expressed on the ventral side. Subsequently, expression of both genes was strongly induced on the dorsal side. The expression of Xmyf-5 and XmyoD did not continue to increase on the ventral side when it was separated from the dorsal side, although muscle originates from the both sides. These findings suggest that a continuous increase in expression of both genes require the dorsalizing signal. The mesoderm inducers bFGF and Activin A induced both genes in animal caps, and the inductive activity of Activin A was stronger than that of bFGF. Overexpression of Xbra, a pan-mesoderm marker, alone induced both genes, but weakly. The inductive activity of Xbra was enhanced by co-injection with noggin. This suggests that inhibition of BMP4 by noggin in the mesoderm mediates dorsalizing signal, and may induce the direct dorsalizing activator genes of Xmyf-5 and XmyoD.
INTRODUCTION
The discovery of MyoD, a mouse gene that can convert cultured fibroblasts into myoblasts (Davis et al., 1987), has been followed by isolation of three more mammalian myo-genic factors related to MyoD: myogenin (Edmonson and Olson, 1989; Write et al., 1989), myf-5 (Braun et al., 1989), and MRF4/myf-6/herculin (Rhodes and Konieczny, 1989; Braun et al., 1990; Miner and Wold, 1990). They are all members of the basic helix-loop-helix (bHLH) family of DNA-bind-ing proteins (Murre et al., 1989) and can bind to muscle-specific promoters (Lassar et al., 1989; Brennan and Olson, 1990; Piette et al., 1990). The pattern of expression of the all four myogenic factors has been reported in normal mouse development. In axial skeletal muscle, myf-5 (day 8), myogenin (day 8.5), MRF4 (day 9) and MyoD (day 10.5) are expressed sequentially, but a different sequence of expression of these genes is observed in the developing limb bud: myf-5 was expressed transiently at day 10-12, myogenin and MyoD are expressed after day 10.5, and MRF4 was detected after day 16 (Sassoon et al., 1989; Bober et al., 1991; Hinterberger et al., 1991; Ott et al., 1991). In vitro and gene-targeting studies suggest that myf-5 and MyoD are involved in muscle cell determination and that myogenin and MRF4 are involved in differentiation and maturation (reviewed by Weintraub, 1993; Olson and Klein, 1994; Rudunicki and Jaenisch, 1995). In Xenopus, the complete cDNAs of XmyoD (Hopwood et al., 1989; Harvey, 1990; Scales et al., 1990), Xmyf-5 (Hopwood et al., 1991), and Xmrf4 (Jennings, 1992) have been cloned and described, and a partial genomic Xmyogenin clone with Xmrf4 was also described by Jennings (1992). Injection of both Xmyf-5 and XmyoD mRNAs at the 2-cell stage results in fairly normal embryos (Hopwood et al., 1991) with no large scale conversion of non-muscle cells into muscle (Gurdon et al., 1992). Late blastula stage animal caps from embryos injected with 1-9 ng of XmyoD (Hopwood and Gurdon, 1990) or Xmyf-5 mRNA (Hopwood et al., 1991) were found to express muscle-specific cardiac actin, but the differentiated muscle-specific antigen 12/101 was not expressed in these explants. Injection of XmyoD mRNA together with RNA encoding its dimerization partner Xenopus E12 (XE12) appeared to lead to limited muscle differentiation (12/101 antigen expressed), although morphological muscle was still not observed (Rashbass et al., 1992). A recent study, however, yielded different results, i.e., that injection of XmyoD or Xmyf-5 mRNA at the 2- to 32-stage activates precocious and ectopic expression of muscle-specific antigens and induces the formation of ectopic muscle. Phenotypically, the embryos displayed enlarged myotomes with increased numbers of myocytes that were shown to be derived at least in part by recruitment of cells of nonsomitic lineage (Ludolph et al., 1994). In either case, the results showed that XmyoD and Xmyf-5 have inducing activity for some muscle-specific genes, and indicate that XmyoD and Xmyf-5 play important roles in early muscle development in Xenopus.
The initial trigger of myogenesis in early development remains unknown. In amphibian development, mesoderm is formed in the equatorial region of the blastula by induction of the nearby animal pole by growth factors released by vegetal pole cells (Nieuwkoop, 1969; Nakamura et al., 1971; Asashima, 1975). Recently, members of the TGFβ superfamily and basic fibroblast growth factor (bFGF) have been reported to induce mesoderm (reviewed in Asashima, 1994). Activin A has the strongest mesoderm-inducing activity of these factors (Asashima et al., 1989, 1990; Smith et al., 1990; van den Eijnden-Van Raaij et al., 1990). Activin A induces mesoder-mal gene-expression and tissues in a concentration-dependent manner (Green and Smith, 1990; Ariizumi et al., 1991; Green et al., 1992). Xbra has been reported as an early response gene (Smith et al., 1991). Overexpression of Xbra induces ectopic muscle in the animal cap (Cunliffe and Smith, 1992), and acts synergistically with noggin (Cunliffe and Smith, 1994) and pintallavis (O'Reilly et al., 1995). These observations may provide an important clue to the identity of the initial muscle determination gene. Very recently, a number of Xenopus genes encoding a T-box, a motif also found in Xbra, have been reported, including Eomesodermin (Ryan et al., 1996), Antipodean (Stennard et al., 1996), Xombi (Lustig et al., 1996), VegT (Zhang and King, 1996), and Brat (Horb and Thomsen, 1997). These genes are expressed at an early stage of embryogenesis, suggesting that they play a role in mesoderm determination.
MATERIALS AND METHODS
Eggs and embryos
Xenopus laevis eggs were obtained by injecting of female animals with 600 IU of human chorionic gonadotropin (Gestron; Denka Seiyaku Co., Kanagawa, Japan). Fertilized eggs were chemically dejellied by treatment containing 3% cystine hydrochloride in Steinberg's solution (pH 7.8) with kanamycin sulfate (100 mg/l; Banyu Pharmaceutical Co., Tokyo, Japan), then washed with sterile Steinberg's solution (pH 7.4). Embryos were transferred to Steinberg's solution and allowed to develop until stage 9 (Nieuwkoop and Faber, 1956).
RNA extraction, RT-PCR and Southern blotting
RT-PCR analysis of RNA samples was performed as described by Sambrook et al. (1989). Total RNAs were isolated by the acid guanidinium thiocyanate-phenol-chloroform (AGPC) method with several modifications (Chomczynski and Sacchi, 1987). Oligo(dT)-primed first strand cDNA was prepared from the total RNA of Xenopus whole embryos and explants, and PCR reactions were carried out in a Thermal Cycler (Perkin-Elmer Cetus). Internal negative controls to which no reverse transcriptase was added were prepared in parallel. After amplification, RT-PCR products were subcloned for Southern blotting, and the sequences were confirmed with an automatic DNA sequencing analyzer (ABI). 32P-labeled probes were used to perform Southern blotting. The PCR products were transferred to a nylon membrane, and signals were detected with X-ray film. The sequences of the primers used in this study were as follows: in the 5′ to the 3′ orientation, Xmyf-5 at 27 cycles, upstream CAACTCCACTGAGCA-TCTTTCTAAG, downstream CGTCTTCATCCGATTCTTCAAGGTC; XmyoD at 27 cycles, upstream TGCCAAGAGTCCAGATTTCCTACAA, downstream TTATGGTGGGGTTCCTCTGGTTTCA; Xmyogenin at 27 cycles, upstream AGGTGTGCAAGAGGAAGACG, downstream GCCAATAGTGTCTGCAAGCG; Xmrf4 at 27 cycles, upstream CACAGTTTGGATCAGCAGGACAAGC, downstream GGATAGTA-GAGCAGTTGATCCTGTA; alpha skeletal muscle actin (muscle specific actin; ms-actin), (Stutz and Spohr, 1986) at 27 cycles, upstream AACAGCAGCTTCTTCCTCAT, downstream TACACAGAGCGAC-TTGAACA; ef1-α (Krieg et al., 1989) at 28 cycles, upstream TTGCCACACTGCTCACATTGCTTGC, downstream ATCCTGCTG-CCTTCTTTTCCACTGC; ornithine decarboxylase (odc), (Bassez et al., 1990; Osborne et al., 1991) at 27 cycles, upstream GTCAAT-GATGGAGTGTATGGATC, downstream TCCATTCCGCTCTC-CTGAGCAC.
Whole-mount in situ hybridization
Whole-mount in situ hybridization was performed according to the method described by Harland (1991). The subcloned RT-PCR products were used for synthesis of the digoxigenin-labeled RNA probe. Embryos obtained from albino females were used. Anti-digoxigenin antibodies were purchased from Boehringer Mannheim GmbH. (Mannheim, Germany).
mRNA synthesis and embryo manipulations
pSP64T vector cDNA was provided by Dr. D. A. Melton. Full-length Xbra of pXT1, provided by Dr. J. C. Smith, was ligated into pSP64T. The noggin template was Δ5′-noggin provided by Drs. W. C. Smith and R. M. Harland (Smith and Harland, 1992). Capped mRNA was synthesized in vitro as described previously (Krieg and Melton, 1984). The mRNAs dissolved in Gurdon's buffer (88 mM NaCl, 1 mM KCl, 15 mM Tris-HCl, pH 7.5) were injected into both blastomeres at the 2-cell stage in 5% Ficoll-Steinberg's solution. Animal caps were dissected from stage 9 embryos, then cultured in Steinberg's solution (pH 7.4) containing 0.1 % BSA and 0.1 g/l kanamycin sulfate at 20°C, in the presence and absence of human recombinant Activin A or bFGF. Human recombinant Activin A was kindly provided by Dr. Yuzuru Eto of the Central Research Laboratory, Ajinomoto Co., Kawasaki, Japan (Eto et al., 1987; Murata et al., 1988). Human recombinant bFGF was obtained from Mallinckropt Co. (Paris, France).
RESULTS
Xmyf-5 and XmyoD are expressed during determination of muscle
In Xenopus, the complete cDNAs of three myogenic factors, Xmyf-5, XmyoD and Xmrf4, and part of the sequence of genomic Xmyogenin DNA have been cloned. Analysis by Northern blotting detected expression of all three cDNAs during normal development, but did not detect Xmyogenin expression at any time. RT-PCR was used to examine the patterns of expression of these factors with greater sensitivity (Fig. 1). Xmyf-5and XmyoD were expressed at stage 10, with the level of transcripts increasing during gastrulation, as reported previously. Very weak Xmyogenin expression was detected at stage 15 but not during the early gastrula stage, similar to Xmrf4 and muscle-specific actin. The highest expression of Xmyogenin was transient, at stage 35–40. Thus only Xmyf-5 and XmyoD were expressed at the muscle-determination stage.
Comparison of Xmyf-5 and XmyoD expression
Xmyf-5 transcripts increased up to the neurula stage and then decreased more rapidly than XmyoD (Fig. 1). This unique pattern of expression of Xmyf-5 suggested that it may operate in a different activation pathway. We therefore closely compared the pattern of expression of Xmyf-5 and XmyoD during muscle determination. Whole-mount in situ hybridization was used to compare the expression of Xmyf-5 and XmyoD in early stage embryos (Fig. 2). No signals were detected in the late blastula (stage 9; Fig. 2a, f). In early gastrula (stage 10; Fig. 2b, g), very weak expression of both XmyoD and Xmyf-5 were detected by RT-PCR (Fig. 1), but the region of expression was ill-defined. In the mid-gastrula (stage 11; Fig. 2c, h), both genes were strongly and specifically expressed in developing somitic mesoderm, but not in the presumptive notochord. XmyoD expression was detected in all somitic mesoderm, but Xmyf-5 expression was restricted to the posterior region. In the late gastrula (stage 12; Fig. 2d, i) and neurula (stage 15; Fig. 2e, j), this difference became more distinct. Xmyf-5 was transiently expressed, and expression then gradually decreased as the cells extended. Xmyf-5 was expressed in a dorsal-to-ventral gradient (Fig. 2e, j). The region of highest Xmyf-5 expression was immediately adjacent to the notochord, which did not express Xmyf-5 at all.
Both Xmyf-5 and XmyoD were expressed in the early gastrula stage (Fig. 1), but the specific sites of expression could not be clarified by whole-mount in situ hybridization. Therefore RT-PCR with divided embryos was used to resolve Xmyf-5 and XmyoD expression (Fig. 3). Embryos were divided into dorsal and ventral explants at stage 10 and cultured until the stage at which they were sampled. Xmyf-5 was expressed in stage 10 whole embryos, but expression began on the dorsal side and was not detected on the ventral side. In contrast, XmyoD was expressed on both sides, but more strongly on the ventral side than the dorsal side at stage 10. Xmyf-5 expression increased greatly on the dorsal side as well as in the whole embryo, with some low-level expression becoming evident on the ventral side. XmyoD expression was also detected and increased on the dorsal side of advanced stage explants, with increased transcript levels compared to the ventral side. The ratio of ventral/whole embryo XmyoD expression was higher than that of Xmyf-5 expression. These results suggest that Xmyf-5 induction was affected by dorsalizing and that this effect on Xmyf-5 was larger than on XmyoD.
Xmyf-5 was induced by growth factors
bFGF and Activins are mesoderm-inducing factors, with bFGF generally inducing ventral mesoderm in animal caps, and Activins inducing both ventral and dorsal mesoderm, depending on the dose. Figure 4 shows Xmyf-5 and XmyoD induction by these factors in animal caps. High-dose bFGF induced Xmyf-5in the animal caps, but a lower concentration of bFGF (1 ng/ml) did not induce Xmyf-5. The greatest induction of Xmyf-5 by Activin A was at a dose of 10 ng/ml. XmyoD was induced at all concentrations of both growth factors, including 1 ng/ml bFGF. These results indicate that Xmyf-5 and XmyoD can be induced by mesoderm-inducing factors, and that Activin A, which include a dorsalizing signal, is more effective than bFGF.
Xbra and noggin induce Xmyf-5 expression
Other studies have shown that ectopic expression of Xbra in animal caps can induce muscle differentiation, and that Xbra is an activator of XmyoD and muscle-specific actin expression (Cunliffe and Smith, 1992; Horb and Thomsen, 1997). Xbra is also an early response gene for both bFGF andx Activin A. We showed that bFGF and Activin A can induce Xmyf-5 in animal caps, similar to XmyoD. We therefore then examined whether Xbra could also induce Xmyf-5 expression (Fig. 5a). When a lower dose of Xbra (0.5-2 ng/embryo) was injected into both blastomeres at the 2-cell stage, Xmyf-5 and XmyoD were not induced in the animal caps, but high-dose Xbra (4 ng/embryo) induced expression of both genes. Only very weak Xmyf-5expression was induced, however, and required a long exposure time for detection (compare with whole embryos; WE, at the right of Fig. 5a and b). We therefore co-injected noggin and Xbra, since this has been described as leading to high induction of muscle-specific actin at a low dose of Xbra (Cunliffe and Smith, 1994). Neither noggin (200 pg/embryo) nor Xbra (1 ng/embryo) alone induced Xmyf-5 or XmyoD (Fig. 5b), but when noggin (200 pg/embryo) and Xbra (1 ng/embryo) were injected together at the same doses they induced both Xmyf-5 and XmyoD.
DISCUSSION
Four myogenic factors that contain a bHLH domain are expressed sequentially and play an important role in determination and differentiation of muscle. We have identified the initial stage of Xmyogenin expression in Xenopus for the first time. Jennings (1992) isolated a genomic DNA fragment of Xmyogenin, but was unable to detect any transcripts and could not isolate any cDNAs. Highly sensitive RT-PCR Southern blotting analysis detected very low levels of transcripts at the same stage as Xmrf4 transcripts were detected before. A transient peak of expression was observed at stage 35-40. At this point in the development of the muscle cell lineage, myofibers accumulate before the start of multinucleation (Boujelida and Muntz, 1987). Transient expression of Xmyogenin transcripts in forming myotubes has been reported during regeneration of adult muscle following cardiotoxin injury (Nicolas et al., 1996). These observations indicate that Xmyogenin may play a role in muscle differentiation and may not function in muscle determination. We therefore concluded that only two of the four factors, Xmyf-5 and XmyoD, function in the muscle determination step.
Xmyf-5 and XmyoD expression was previously reported to begin in the early gastrula (stage 10) (Hopwood et al., 1989, 1991; Harvey, 1990). Our present study showed that Xmyf-5 expression begins at the early stage in the dorsal hemisphere, and that in contrast to XmyoD, it is expressed in the ventrolat-eral mesoderm (Frank and Harland, 1991). Xmyf-5 was subsequently detected in the divided ventral side, but was only weakly and unstably expressed (Fig. 3). XmyoD did not show continuous expression on the ventral side too. Moreover, the ventral explants expressed very little muscle-specific actin at stage 28 (data not shown). It is well known that the presumptive fate of muscle in Xenopus blastula lies in the mesoderm region of both the dorsal and ventral hemispheres (Keller, 1975, 1976; Dale and Slack, 1987; Moody, 1987a,b; Moody and Kline, 1990). Previous reports and our current whole-mount in situ hybridization study have shown that Xmyf-5 and XmyoD are also expressed at the neurula stage in the posterior region derived from the ventral hemisphere of early gastrula embryos. However, Xmyf-5 transcripts did not accumulate in ventral explants from which the dorsal side had been cut off. Therefore, these phenomena suggest that the dorsalizing effect that was released from the organizer and led to muscle formation actually persists during gastrulation, and that this continuing dorsalization occurs under convergent-extending movement (Vogt, 1929; Gerhart and Keller, 1986). The cells may continuously express the muscle determination genes Xmyf-5 and XmyoD in response to a dorsalizing signal from the extended presumptive notochord when the presumptive muscle cells derived from the ventral hemisphere of early gastrula embryos move to the dorsal side.
Activins and bFGF are potent mesoderm-inducing factors (reviewed by Slack, 1994; Asashima, 1994), and XmyoD is induced by bFGF and XTC-MIF (Harvey, 1990). We examined the ability of bFGF and Activin A to induce Xmyf-5. Both mesoderm-inducing factors induced Xmyf-5, similarto XmyoD, but Activin A was the stronger inducer of both genes. Thus, both factors may have basal inducing activity, and Activin A may also be a dorsalization signal. Activin A induces gene expression and differentiation of dorsal mesoderm depending on the dose (Green et al., 1990; Ariizumi et al., 1991). Activin A induced the strongest expression of the both myogenic factors at 10 ng/ml, the concentration at which explant elongation and muscle tissue induction occur. Treatment of Activin A at 100 ng/ml showed weaker inductive activity than at 10 ng/ml. The reason for this is suspected of being that high-dose Activin A mainly induces the notochord. If the dorsalization signal is excessive, it may cause deactivation of organizer genes and notochord formation, and the myogenic factors Xmyf-5 and XmyoD may be suppressed. Thus, myogenesis may be both up- and down-regulated by dorsalization.
Both bFGF and Activin A have been reported to induce the T-box gene Xbra, a pan-mesoderm marker, and ectopic expression of Xbra induces mesoderm, including muscle (Cunliffe and Smith, 1992). Our experiments suggest that induction of muscle by Xbra is mediated by Xmyf-5 and XmyoD expression. Injection of high doses of Xbra was required to induce of these genes, especially Xmyf-5, and lower doses acted synergistically action at lower doses with noggin, a dorsalization molecule. Therefore, muscle determination may be activated by two different signals, basal mesoderm induction and dorsalization, and Xbra and noggin may be the mediating molecules in vivo. Xbra encodes a DNA-binding nuclear protein containing a T-box and may direct activation of Xmyf-5 and XmyoD. Other recently cloned T-box genes may have similar functions. The secreted proteins noggin and chordin have been reported to act as BMP4 suppressors by direct binding (Piccolo et al., 1996; Zimmerman et al., 1996). This suggests that inhibition of BMP4 by noggin and chordin in the mesoderm mediates dorsalization signal and may induce the direct dorsalizing activator genes of Xmyf-5 and XmyoD. A candidate direct dorsalizing activator is pintallavis, which encodes a nuclear protein, and has been reported to act synergistically with Xbra (O'Reilly et al., 1995).
Acknowledgments
This work was supported by Grants-in-Aid for Scientific Research from the Ministry of Education, Science, Sports and Culture of Japan, and by CREST (Core Research for Evolutional Science and Technology) of the Japan Science and Technology Corporation.