Msx is a homeodomain-containing transcriptional factor that plays an essential role in pattern formation in vertebrata and invertebrata embryos. In Xenopus laevis, two msx genes have been identified (Xmsx-1 and Xmsx-2). In the present study, we attempted to elucidate the expression and function of Xmsx-2B (formerly designated as Xhox7.1′) in early embryogenesis. Whole mount in situ hybridization analyses showed that the expression pattern of Xmsx-2B at gastrula and neurula stages was very similar to that of Xmsx-1: the transcript of Xmsx-2B was observed in ventral and lateral sides of the embryo. At the tailbud stage, however, the expression pattern of Xmsx-2B in neural tissues was distinct from that of Xmsx-1. An RNA injection experiment revealed that, like BMP-4, Xmsx-2B has a strong ventralizing activity. In the Xmsx-2B -injected embryos, differentiation of axial structures such as the notochord, muscle, and neural tissue was completely suppressed, whereas α-globin mRNA, a blood cell marker, was highly expressed. Simultaneous injection of Xmsx-1 and Xmsx-2B RNAs showed that they function in an additive manner. It was also shown that coinjection of Xmsx-2B with a dominant-negative BMP-4 receptor (tBR), which can induce formation of secondary axis when injected alone in ventral blastomeres, suppressed secondary axis formation. Furthermore, Xmsx-2B also suppressed secondary axis formation, which was induced by a dominant-negative form of Xmsx-1 (VP16/msx-1). Therefore, like Xmsx-1, Xmsx-2B is a downstream nuclear factor of the BMP-4-derived ventralizing signal, and these two factors probably share the same target molecules. In conclusion, Xmsx-1 and Xmsx-2B function in dorso-ventral axis formation in early Xenopus laevis development.
A number of studies have indicated that BMP-4 (bone morphogenetic protein-4), a member of the TGF-β superfamily, is a key regulator in body planning, especially in dorsoventral patterning of mesodermal tissues and in neural tissue differentiation, in the Xenopus laevis embryo (Dale et al., 1992; Jones et al., 1992; Fainsod et al., 1994; Graff et al., 1994; Maéno et al., 1994; Suzuki et al., 1994; Schmidt et al., 1995). Recently, homeodomain-containing transcriptional factors that mediate the BMP-4 signal have been isolated and analyzed: Xmsx-1 (Maeda et al., 1997; Suzuki et al., 1997), Xvent-1 (PV. 1) (Gawantka et al., 1995; Ault et al., 1996), Xvent-2 (Xbr-1, Vox, Xom) (Onichtchouk et al., 1996; Papalopulu and Kintner 1996; Ladher et al., 1996) and Xvex-1 (Shapira et al., 1999). We and other investigators have revealed that Xmsx-1 is expressed in the prospective ventral ectoderm and mesoderm at mid to late gastrula stages and that it has a ventralizing activity if its RNA is injected into the prospective dorsal region (Maeda et al., 1997; Suzuki et al., 1997). In ectodermal cells, Xmsx-1 functions as an inhibitor of neurogenesis, and it has been shown that Xmsx-1 is a direct target gene of the BMP-4 signal (Suzuki et al., 1997). More recently, we have demonstrated that Xmsx-1 is an upstream factor of Xwnt-8 in the ventralizing signal cascade and a possible direct repressor of organizer genes such as goosecoid (Takeda et al., 2000).
In comparison to the numerous studies on Xmsx-1, quite a few studies have been carried out on the function of Xmsx-2 (Su et al., 1991; Gong and Kiba, 1999). In other species, msx-2 has been studied extensively in various developing tissues such as bone, tooth, and limb bud (Monachan et al., 1991; Coelho et al., 1991; Mackenzie et al., 1992; Jowett et al., 1993). Most likely, the function of msx-2 overlaps with that of msx-1 (Semenza et al., 1995), but in some cases msx-2 may have distinct expression patterns and functions. For instance, expression domains of msx-1 and -2 overlap in some regions but are discrete along with anterior-posterior identity in chicken and mouse limb buds (Davidson et al., 1991; Coelho et al., 1991; Muneoka and Sassoon, 1992; Nohno et al., 1992). The BMP-4/msx-2 signal is involved in elimination of cranial neural crest cells in rhombomeres 3 and 5 (Graham et al., 1993; Graham et al., 1994; Takahashi et al., 1998; Farlie et al., 1999) and the interdigit regions of developing limbs (Zou and Niswander 1996; Ganan et al., 1996; Ferrari et al., 1998) by mediating apoptosis. Furthermore, the mechanisms of expressional control in mouse mammary development are different between msx-1 and -2 (Friedmann et al., 1996; Phippard et al., 1996).
These observations suggest that msx-2 could have a distinct expression pattern and function in the early development of Xenopus laevis. In spite of the importance of the issue, expressional and functional natures of msx-2 in Xenopus embryogenesis are poorly understood. Xenopus msx-2 (Xhox7.1′) was first isolated by Su et al. (1991). In their report, however, major attention was paid to the expression of Xmsx-1. Afterward, no report has described the expression and function of Xmsx-2 except for a very recent one in which a related but distinct gene from Xhox7.1′ was described (Gong and Kiba, 1999). In this study, we attempted to elucidate the expression and the function of Xhox7.1′. We propose designating Xhox7.1′ as Xmsx-2B in order to distinguish it from Xmsx-2, which was isolated by Gong and Kiba (1999). Our results showed that there is a strong correlation between the expression patterns of Xmsx-1 and Xmsx-2B at gastrula to neurula stages and that those two genes play a role in the establishment of the dorso-ventral axis in developing Xenopus embryos.
MATERIALS AND METHODS
Cloning and cDNA construction
Xmsx-2B cDNA fragments of two different sizes were amplified by polymerase chain reaction (PCR) from Xenopus stage10 cDNA. A short fragment including 402 N-terminal base pairs was used as a probe for in situ hybridization, and another fragment including the whole protein-coding region (815 bp) was used for an RNA injection assay. The primer sequences for the PCR were as follows: 5′-TC T-GCG-CCC-CTT-TAC-TCA-CCG-CA-3′ and 5′-TCA-AAG-TGC-AAG-AGG-ATG-GGC-TCA-3′ for the 402-bp fragment; and 5′-TCT-GCG-CCC-CTT-TAC-TCA-CCG-CA-3′ and 5′-CTA-TGT-CAT-GTC-ACC-CTC-CTC-AGA-TA-3′ for the 815-bp fragment (Su et al., 1991). Each product was ligated into pCRTM II (In vitro gene) for subcloning, and the DNA sequence of the inserts was determined using an Auto Read Sequencing kit (ALF DNA Sequencer, Pharmacia Biotech). The obtained sequence was completely identical to that of Xhox-7.1′, which was previously reported (Su et al., 1991). For injection of sense RNA of Xmsx-2, an 815-bp fragment including an open reading frame was excised with Not I, blunt ended, and ligated into a blunt-ended pCS2+ expression vector.
Embryos, mRNA micro-injection, and DAI judgement
Xenopus laevis embryos were obtained by artificial insemination after induction of females with 250 i.u. of human chorionic gonadotropin. Fertilized embryos were dejellied with 2.5% thioglycolic acid (pH8.3) and washed several times in 5% MMR. After the embryos were transferred into 3% Ficoll/Steinberg's solution, equatorial regions of either two ventral or dorsal blastomeres at the 4-cell stage were injected with mRNAs using a Nanoject microinjector (3.7-6 nl/embryo). pCS2+ plasmids containing Xmsx-1 whole cDNA (1.8 kb) or the 815-bp Xmsx-2B fragment were linearized with Xho I, and capped mRNA was synthesized using SP6 polymerase (Ambion). VP16/msx-1, a dominant-negative form of Xmsx-1, has been described previously (Takeda et al., 2000). Injected embryos were incubated in 50% Steinberg's solution at 23°C until the sibling embryos reached stage 33/34 for determining DAI (dorso-anterior index) as described by Kao and Elinson (1988). Developmental stages were designated according to Nieuwkoop and Faber (1967).
Whole-mount in situ hybridization
Whole-mount in situ hybridization was performed as described previously (Shain et al., 1996). Digoxygenin-labeled antisense mRNA probes were synthesized from the following plasmids: Xmsx-1, an N-terminal fragment (EcoRI/HindIII, 440 bp) in pSP72; and Xmsx-2B, an N-terminal fragment (402 bp) in pCRTMII. The former plasmid was linearized with EcoRI, and RNA was synthesized by SP6 polymerase. The latter plasmid was linearized with BamHI, and RNA was synthesized by T7 polymerase. The positive signals were visualized using BM purple (Moors et al., 1995). For neural marker probes, fragments of slug and krox-20 were amplified by the PCR method from the cDNAs of stage18 whole embryo and of stage 27 head tissue, respectively. The primer sequences for the PCR were as follows: 5′-GAA-TCT-GGT-TGC-TGT-GTA-G-3′ and 5′-CGT-TGC-TGG-ATT-GTC-TAG-GC-3′ for slug; and 5′-CCA-ACC-GCC-CCA-GTA-AGA-CC-3′ and 5′-GTG-TCA-GCC-TGT-CCT-GTT-AG-3′ for krox-20. Both fragments were subcloned into a pCR2.1 vector (In vitro gene), and their DNA sequences were determined. For making an antisense RNA probe, the plasmids were cut with BamHI, and RNAs were synthesized by T7 polymerase.
Expression pattern of Xmsx-2B in the early embryo
To investigate the expression pattern of Xmsx-2B during embryogenesis and to compare it with that of Xmsx-1, wholemount in situ hybridization analysis was performed using various stages of albino embryos. As shown in Fig. 1, during gastrula and early neurula stages, the Xmsx-2B transcript was observed in the ventral and lateral marginal zones and animal pole area (Fig. 1 B, D and F), as was also found in the case of Xmsx-1 (Maeda et al., 1997; Suzuki et al., 1997; Fig. 1A, C and E). At the neurula stage (stages 16–17), two parallel lines of transcript were observed at the dorsal side along the border between the prospective neural tissue and epidermis (arrowheads in Fig. 1G and H), and also strong spot signals were detected on the medial and lateral sides of the prospective neural crest (arrows in Fig. 1 G and H), with reference to the expression of slug, a neural crest marker (Mayor et al., 1995; Fig. 1 J). We also compared the site of expression with that of krox-20, a marker of rhombomeres 3 and 5 (Bradley et al., 1992; Fig. 1 I and L), and we found that strong spot signals were located at the level of the hind brain (rhombomeres 3–5) (Fig. 1 H and K). At the early tailbud stage (stage 23), expression patterns became different between Xmsx-1 and Xmsx-2B in the neural tissue (Fig. 1 M, N, P and Q). Xmsx-1 was highly expressed in the broad areas including fore-, mid-and hind-brains, according to the expression of krox-20 (Fig. 1 O and R), and spinal cord, while Xmsx-2B was expressed in the medial region of neural tissues. These observations suggest that Xmsx-1 and Xmsx-2B have the same or similar functions in gastrula stages but have different functions after the neurula stage.
Xmsx-2B has a ventralizing activity
It has been shown in previous studies that Xmsx-1 has ventralizing and anti-neuralizing activities in early embryogenesis (Maeda et al., 1997; Suzuki et al., 1997; Ishimura et al., 2000; Takeda et al., 2000). To examine whether Xmsx-2B has the same or similar activity, we performed mRNA injection experiments. Two dorsal blastomeres of 4-cell stage embryos were injected with various concentrations of mRNAs encoding for the full-length Xmsx-2B protein, and the injected embryos were incubated for 2 days until the sibling embryos reached the tailbud stage. The embryos injected with Xmsx-2B mRNA showed ventralized phenotypes and the average DAI was reduced in dose-dependent manner (Fig. 2 D, E, G, J and M). This activity of Xmsx-2B seemed very strong among the known ventralizing agents, and as little as 10 pg of mRNA was enough to exert a ventralizing effect (average DAI=4.3, n=154; data not shown). Injection of 1ng RNA led to a hyperventralized phenotype completely lacking the dorso-ventral axis (average DAI=0.3, n=98; Fig. 2 J, M). The phenotype induced by Xmsx-2B overexpression was very similar to that obtained by the injection of BMP-4 mRNA (average DAI=0.1, n=92; Fig. 2 I, L). In these embryos, ectodermal, mesodermal and endodermal layers were clearly distinguished, but differentiation of dorsal tissues such as the notochord, neural tube or muscle was not observed (Fig. 2 O and P). Furthermore, in situ hybridization analysis indicated that in these hyperventralized embryos, α-globin mRNA was strongly expressed in the anterior region (Fig. 2 I, J) but α-actin was not (Fig. 2 L, M).
Since the phenotypes of the embryos injected with Xmsx-2B and BMP-4 mRNA were similar, we examined the blood cell-inducing ability of Xmsx-2B in the dorsal marginal zone (DMZ) explants. After injection of several concentrations of Xmsx-2B mRNAs (200–800 pg/embryo) into the marginal zone of dorsal blastomeres, DMZs were excised at stage10.5 and cultured until stage 39. Northern blot analysis showed that α-globin mRNA was not induced in the explants even though the highest amount of mRNA was injected (data not shown). Thus, we concluded that among the target genes of the BMP-4 signal, Xmsx-1 and Xmsx-2B are not sufficient for blood cell differentiation.
To investigate whether Xmsx-1 and Xmsx-2B act cooperatively, these RNAs were injected alone or simultaneously into two dorsal blastomeres (Fig. 2 B-G). The average DAI obtained by 25 pg of Xmsx-2B mRNA was 3.1±1.4 (n=122), whereas that of embryos that received 400 pg of Xmsx-1 mRNA was 3.5±1.1 (n=185). When these mRNAs were coinjected, the DAI fell to 2.3±1.0 (n=140). Since similar values of DAI were obtained in the embryos injected with 50 pg Xmsx-2B (DAI=2.3±1.4; n=152) or 800 pg Xmsx-1 (DAI=2.2±1.2; n=77), we concluded that Xmsx-1 and Xmsx-2B function in ventralization in an additive manner.
Xmsx-2B mediates BMP-4 signaling
It has been shown that the expression of Xmsx-1 is regulated by the BMP-4 signal (Maeda et al., 1997; Suzuki et al., 1997). Taken together with these expression patterns of Xmsx-2B in the early embryo (Fig. 1), it seems that Xmsx-2B, like Xmsx-1, is regulated by the BMP-4 signal. To confirm this hypothesis, we examined whether Xmsx-2B could disturb the formation of the secondary axis, which was induced by tBR mRNA injection. Two ventral blastomeres of 4-cell stage embryos were injected with 500 pg of tBR mRNA alone or coinjected with 1ng of Xmsx-1 or 200 pg of Xmsx-2B. As shown in Table 1 and Fig. 3 B, 56 % (n=127) of tBR-injected embryos formed the secondary axis. In contrast, the formation of the secondary axis was completely suppressed if either Xmsx-1 (n=112) or Xmsx-2B (n=117) was coinjected with tBR mRNA.
Incidence of secondary axis formation in injected embryos
We also examined whether the ventralizing activity of Xmsx-1 can be replaced by that of Xmsx-2B. For this purpose, we used a fusion construct of Xmsx-1 and VP16, a dominant-negative form of Xmsx-1, which was shown to induce formation of the secondary axis (Takeda et al., 2000). The embryos injected ventrally with VP16/msx-1 mRNA formed an incomplete secondary axis without a head structure (65%; n=95, Fig. 3 F). Simultaneous injection either of Xmsx-1 (Fig. 3 G; n=81) or Xmsx-2B (Fig. 3 H; n=118) completely suppressed the formation of the secondary axis. These results provide us with evidence that the activity of Xmsx-1 can be replaced by that of Xmsx-2B and, therefore, these two transcriptional factors share the same target molecules in the ventralizing signal.
The expression of Xmsx-2B overlaps that of Xmsx-1 in gastrula and neurula stages
Since Su et al. (1991) first reported the isolation of an Xhox7.1′ (Xmsx-2B) cDNA clone in Xenopus laevis, there have been no reports describing the expression and function of Xenopus msx-2, despite the general view of importance in various vertebrate embryogenesis. Very recently, however, Gong and Kiba (1999) isolated a msx-2 cDNA clone and emphasized its function in antero-posterior patterning of mesoderm formation. From the viewpoints of sequence similarity, expression pattern, and biological activity, Xmsx-2B and the cDNA isolated by Gong and Kiba (1999) are subtypes of Xe-nopus msx-2.
As expected, the expression pattern of Xmsx-2B was very similar to that of Xmsx-1 (Maeda et al., 1997; Suzuki et al., 1997) during gastrula and neurula stages. Concomitant expressions of msx-1 and -2 have been reported in various tissues in mammalian species (Mackenzie et al., 1992; Jowett et al., 1993; Wang and Sassoon, 1995; Friedmann et al., 1996; Phippard et al., 1996) and in birds (Coelho et al., 1991; Ganan et al., 1996; Ganan et al., 1998). At the gastrulation stage, the transcript was detected in the lateral and ventral marginal zone (Fig. 1), suggesting that Xmsx-2B and Xmsx-1 have the same or similar functions at this stage. At the tailbud stage, however, the expression pattern in neural tissues was clearly distinct. We suggest that the mechanism of expressional control in the two msx genes is conserved at least in early embryogenesis.
Xmsx-2B acts as a ventralizing factor
As suggested by the expression pattern of Xmsx-2B in gastrulating embryos, the forced expression of Xmsx-2B mRNA in dorsal blastomeres induced a strong ventralized phenotype (Fig. 2 J, M), and α-globin mRNA was strongly expressed in the embryos (Fig. 2 J). By analyzing the ventralizing activity of Xmsx-1, we noticed that, unlike BMP-4, Xmsx-1 alone lacks activity to upregulate the expression of ventral genes (Maeda et al., 1997). On the other hand, Xmsx-2B -injected embryos completely lost the body axis, while the same dose of Xmsx-1 had a weaker ventralizing effect. We predicted, therefore, that Xmsx-2B may be a cofactor of Xmsx-1 in blood cell-inducing activity. However, Northern blot analysis revealed that Xmsx-2B could not induce α-globin expression in the dorsal marginal zone explants, while BMP-4 could. Although there was no difference in the translational activities of Xmsx-1 and Xmsx-2B RNAs in the reticulocyte lysate (data not shown), the stabilities of proteins in the cells were not determined. As the Xmsx-2B cDNA used in this study lacks the 3′ UTR, there is a possibility that this 3′ UTR might regulate the stability of Xmsx-2B mRNA. From these observations, we concluded that even though Xmsx-2B has a strong activity, Xmsx-1 and Xmsx-2B ventralize the embryo by suppressing the expression of organizer genes (Takeda et al., 2000), and in the prospective ventral cells in which the BMP-4 signal is activated, another unknown factor is required to trigger the blood cell differentiation.
Xmsx-1 and Xmsx-2B function in the dorso-ventral patterning of mesoderm
In the present study, we found that Xmsx-1 and Xmsx-2B function in the ventralizing cascade in an additive manner (Fig. 2 A-G). In addition, the ability of Xmsx-2B to suppress the secondary axis formation induced by VP16/msx-1 RNA (a dominant negative form of msx-1 protein) supports our conclusion that the ventralizing activity of msx-1 can be replaced by that of Xmsx-2B and that their target gene(s) might be common. So far, the molecular nature of how these msx proteins interact with each other is unclear, but previous reports have suggested that msx and dlx proteins can form homo-and heterodimers in various combinations (Zhang et al., 1997). Since msx proteins are transcriptional repressors (Catron et al., 1995) and dlx proteins are transcriptional activators (Zhang et al., 1997), the biological activities in transcriptional regulation are mutually exclusive (Zhang et al., 1997). Other candidates of msx's partner are Xvent-1 (PV. 1) (Gawantka et al., 1995; Ault et al., 1996) and Xvent-2 (Xbr-1, Vox, Xom) (Onichtchouk et al., 1996; Papalopulu and Kintner, 1996; Ladher et al., 1996). These factors are also homeodomaincontaining proteins and possessing ventralizing activity. The similarity of expression patterns and functions between msx and vent suggests that these proteins might interact with each other, directly or indirectly, in prospective ventral cells. To elucidate the molecular mechanisms of msx's function in the ventralizing activity and the following specification of tissue distribution, isolation and analysis of the interacting molecules will be required in the future.
This work was partly supported by grants from The Saneyoshi Foundation, The Sumitomo Foundation and The Ministry of Education, Science and Culture of Japan (No. 09680721).