Shisa2 promotes the maturation of somitic precursors and transition to the segmental fate in Xenopus embryos

A metameric vertebrate body pattern is established by the segmentation of presomitic mesoderm (PSM), by which somites are formed sequentially along the anteroposterior (AP) direction at a regular interval. As a consequence of the maturation of unsegmented somitic precursors in the caudal PSM, cells have committed to a definitive segmental fate and form somitomeres, prospective somites, in the rostral PSM. A caudorostrally decreasing morphogenetic gradient (high caudally and low rostrally) distributes in the PSM; this gradient controls the maturation of PSM and the position in which cells transit to the definitive segmental fate(Aulehla and Herrmann, 2004; Dubrulle and Pourquie, 2004a; Saga and Takeda, 2001). This transition point, termed a `wavefront', `differentiation wavefront' or`determination front' (Cooke and Zeeman,1976; Dubrulle and Pourquie,2004a), has been thought to be located at the border between the rostral and caudal PSM and marked by changes in the expression of two basic helix-loop-helix (bHLH) transcription factors. Mesogenin1(Mes1; Xenopus ortholog of Mespo) marks the caudal PSM (Joseph and Cassetta,1999; Yoon et al.,2000; Yoon and Wold,2000); Thylacine1 (Thy1; Xenopusortholog of Mesp) is expressed in two or three bilateral stripes that correspond to the anterior half of the somitomeres I-III(Fig. 3G) in the rostral PSM. The most posterior Thy1 stripe in S-III is thought to represent PSM cells that have just passed the wavefront(Buchberger et al., 1998; Saga et al., 1997; Sawada et al., 2000; Sparrow et al., 1998). As somitogenesis proceeds, the wavefront moves posteriorly; this movement is tightly coupled with axial elongation.

A well-known signal constituting the gradient is FGF. The increase in the FGF/MAPK signaling suppresses the maturation of PSM cells and delays the transition to the segmental fate, consequently generating smaller somites. By contrast, the inhibition of FGF/MAPK signaling promotes the maturation of PSM cells and the generation of larger somites(Delfini et al., 2005; Dubrulle et al., 2001; Dubrulle and Pourquie, 2004b; Sawada et al., 2001).

The Wnt signal emitted from the tailbud has been implicated in the mechanism of the segmentation clock by which periodicity of segmentation is generated. In the mutant mouse harboring the Wnt3a hypomorphic allele vestigial tail (vt), expressions of several segmental clock genes are severely reduced or disrupted, including Axin2, Lunatic fringe, Nkd1 and Snail1 (Snai1 - Mouse Genome Informatics) (Aulehla et al.,2003; Dale et al.,2006; Ishikawa et al.,2004). In the segmentation, increases and decreases in Wnt/β-catenin signaling generate smaller and larger somites,respectively, suggesting that Wnt signaling also functions as an inhibitor of the PSM maturation as FGF signaling does(Aulehla et al., 2003). It has thus been suggested that the wavefront is settled at the position where the level of FGF and Wnt signaling goes below a certain threshold(Aulehla and Herrmann, 2004; Dubrulle and Pourquie, 2004a). However, as the FGF8 expression is strongly reduced in vt/vtmice, Wnt might set up the wavefront indirectly by regulating the FGF signaling (Aulehla and Herrmann,2004). Thus it remains elusive whether individual regulation of both FGF and Wnt signals are required for the positioning of the wavefront.

The mechanisms establishing the PSM gradient have been explained in two ways. One is the axial elongation by which somitic precursors progressively move away from the tailbud, where cells actively transcribe and translate FGF8 and Wnt3a genes. The decay and diffusion of the ligand protein regulate the gradient activity. Furthermore, as a result of a slow decay of FGF8 transcripts, cells establish an FGF8 mRNA gradient in the PSM; this mRNA gradient has been suggested to generate a shallow FGF8 protein gradient and to regulate the maturation of PSM(Dubrulle and Pourquie,2004b). The other mechanism is the antagonistic relationship between retinoic acid (RA) and the FGF gradient. The RA signal constitutes a gradient that is rostrocaudally decreasing and has been suggested to promote the maturation of PSM. The increase and decrease in RA activity causes a posterior and anterior shift of the wavefront, respectively, through the indirect regulation of FGF signaling (Diez del Corral et al., 2003; Moreno and Kintner, 2004). By contrast to the gradient in extracellular ligands, however, it is largely unknown whether, and how, cells regulate their responsiveness to FGF and Wnt signaling for setting up the morphogenetic gradient in the PSM.

Wnts bind to the Frizzled (Fz) family of the seven-pass transmembrane receptor and activates a downstream target, Dishevelled. In the Wnt/β-catenin (canonical) pathway, this suppresses the activity of a protein kinase GSK3 (a negative regulator of this signaling), stabilizesβ-catenin and activates transcriptional targets(Nusse, 2005). Wnt-Fz also initiates at least two other signaling cascades, planer cell polarity (PCP)and Wnt/Ca²⁺ pathway(Wallingford and Habas, 2005). In the early Xenopus embryos, integrations of this signaling govern numerous biological processes, including axis formation, convergent-extension cell movements, mesodermal differentiation and cell adhesion. FGF binds to the receptor tyrosine kinase FGF receptor family (FGFR1-4), and induces its dimerization and transphosphorylation. Subsequently, the small GTPase Ras transmits the FGFR signal and activates the protein kinase cascade Raf-MEK1/2-ERK1/2, which phosphorylates and activates various transcription factors (Goldfarb, 2001). FGF functions in the mesoderm and neural induction and their differentiation(De Robertis and Kuroda, 2004; Slack et al., 1996). We have recently reported isolation and functional characterization of Shisa1(previously called Shisa) (Yamamoto et al., 2005), a cell-autonomous inhibitor for both Wnt and FGF signaling, involved in Xenopus head formation. Shisa physically interacts with an immature form of Fz and FGFR in the endoplasmic reticulum(ER) and inhibits their protein maturation and cell surface transportation,thereby suppressing events being initiated by ligand-receptor interactions of Wnt and FGF signaling.

Here we have identified Shisa-related genes in Xenopus, Shisa2 and Shisa3, which inhibit both Wnt and FGF signals through the retention of their receptors in the ER as Shisa1 does. Knockdown study of Shisa2 suggests that it plays an essential role in the maturation of PSM cells by individual attenuation of both FGF and Wnt signaling.

Animal cap, tailbud and dorsal explants were prepared at the stage indicated in the figure legends for each experiment. The explants were dissected in low-calcium magnesium Ringer's solution (LCMR) and cultured at 22°C with LCMR containing 0.1% bovine serum albumin (BSA) alone or with bFGF (50 ng/ml, R&D Systems), recombinant Human WIF1 (20 ng/ml, R&D Systems), mFz8CRD-Fc (30 ng/ml), SU5402 (0.1 mg/ml, CALBIOCHEM) or RA (1μmol/l, CALBIOCHEM). mFz8CRDFc protein was prepared as described(Yamamoto et al., 2005). Morpholino antisense oligomers were obtained from Gene Tools: 5mis Shisa2MO1, 5′-GAGGCGTGCAACCACATCACTGGC-3′; Shisa2MO1, 5′-GAGCCCTCCAACCACATGACTGGG-3′; Shisa2MO2, 5′-ACTCCTCTCACGGGCAGCAAAAGTC-3′; 5mis Shisa2MO3, 5′-ACATGCCATTTATTAGCTCCTCTAG-3′; Shisa2MO3, 5′-AGATCCCATTTATTACCTGCTGTAG-3′. Shisa1MO was described previously(Yamamoto et al., 2005).

The expressed sequence tag (EST) clone (Accession number CF286494: IMAGE 5516153) encoded a partial Xenopus Shisa2 coding sequence (CDS),lacking 227 nt from the 3′ end of the CDS. To determine the full-length Shisa2 cDNA sequence, 3′ RACE of the tadpole stage total RNA was carried out with a SMART RACE cDNA Amplification Kit (CLONTECH). The two forward primers 5′-GGTGGCAATTTGCTGTTGCAGATGT-3′ and 5′-AGTGCGAGCTGCGCTACTGCTGTT-3′ were used in a nested way. By the sequences of five independent clones, a full-length sequence of Shisa2 cDNA was determined. The EST sequence of the BI449671 clone contained the full-length of Shisa3 CDS. The CDS of Shisa2and 3 were amplified by RT-PCR and subcloned into pCS2(Shisa2/pCS2 and Shisa3/pCS2). Shisa2-HA, Shisa2-Flag, Shisa3-HA and Shisa3-Flag were also generated by PCR. Other constructs were described previously (Yamamoto et al.,2005).

RT reaction was carried out with MLTV (Invitrogen) using 500 ng of total RNA, isolated with RNA-STAT-60 (TEL-TEST Inc.) from embryos, animal caps or tailbud explants. PCR amplification was carried out for 28 cycles with the following thermal cycle profile: denaturation at 94°C for 30 seconds,annealing at 55°C for 45 seconds and extension at 68°C for 45 seconds,followed by a final extension at 72°C for 5 minutes. The primers used were: Shisa2, forward 5′-ACGATTCGACCATCTGCTG-3′ and reverse 5′-CAGTTGGTTTGGGATCGAGT-3′; Mes1, forward 5′-GAGACAACGGAGCTCTCACC-3′ and reverse 5′-AATCCAGCCTGGTGTTTCAG-3′; FGF8, forward 5′-ACCTCCATCCTGGGCTATCT-3′ and reverse 5′-GCCCCTTCCATTAGTCTTCC-3′; Wnt3a, forward 5′-GCGATTTTTGGACCAGTGTT-3′ and reverse 5′-TTCTGCCTGCTTCATTGTTG-3′; Hes6, forward 5′-GGCTGCTGATCTTCTGAACC-3′ and reverse 5′-CCTTCTCCCCTTCAGATTCC-3′; Shisa2 F,5′-ACGATTCGACCATCTGCTG-3′; Shisa2 R1,5′-GAAATTCCATCATCCCAACC-3′; Shisa2 R2,5′-CAGTTGGTTTGGGATCGAGT-3′. Other primer sequences and conditions for PCR reaction were carried out as described previously(Yamamoto et al., 2005).

Whole-mount in situ hybridization was performed according to described procedure (Sive et al., 2000). Signals were developed with BM Purple (Roche) or BCIP (Roche). The probes used for in situ hybridization were transcribed from Mes1, XL322e02ex(NIBB); Thy1, XL220g19 (NIBB); ESR9, XL224g01ex (NIBB); Arp-A, XL146e16 (NIBB); Cyp26, XL322k18ex (NIBB); Rarg, XL275p11ex (NIBB); Raldh2, XL191i17 (NIBB); Shisa2, EXL1051-5991502 (Open Biosystems). Xbra, MyoD, Papc and Wnt3a were transcribed as described(Kim et al., 1998; Rupp and Weintraub, 1991; Smith et al., 1991; Wolda et al., 1993). FGF8 cDNA was isolated by RT-PCR and cloned into the pGEMT-E vector(Promega). Western blotting was performed as described previously(Yamamoto et al., 2005). Antibodies against phospho-p44/42 MAP kinase (dp-ERK)(Cell Signaling), p44/42 MAP kinase (Cell Signaling) and HA (Covance) were used at 1:1000 dilution. Whole-mount immunostaining was performed according to described procedure(Kuroda et al., 2005).

Immunofluorescent staining, luciferase assay and co-immunoprecipitation assay were carried out as described previously(Yamamoto et al., 2005). Cell transfection into COS cells was performed with Lipofectamine 2000(Invitrogen), according to the manufacturer's instructions.

Xenopus Shisa1 has two unique cysteine-rich domains (CRD1 and CRD2) in the amino-terminal half of the sequence(Fig. 1A)(Yamamoto et al., 2005). A search of a Xenopus EST database with the amino acid sequence of Shisa1 allowed us to identify two genes that encode Shisa family proteins harboring the two conserved CRDs (Fig. 1A,B); they are referred to as Shisa2 and Shisa3(Accession number CF286494 for Shisa2 and BI449671 for Shisa3). Among Shisa1, 2 and 3, the amino acid sequences are well conserved in the amino-terminal half, including the CRDs, but not in the carboxy-terminal half (Fig. 1C).

In the Xenopus animal cap assay, ectopic expression of Wnt8 or treatment with bFGF (FGF2) protein induces the expression of Xnr3 and Xbra, respectively (Brannon et al., 1997; McKendry et al.,1997; Pownall et al.,1996). Shisa2 and 3 inhibited this induction in a dose-dependent manner, as Shisa1 does (Fig. 1D,E) (Yamamoto et al.,2005). Shisa2 and 3 neither induced neuroectoderm in animal caps or a secondary axis on the ventral side, nor did they inhibit Nodal/Activin-dependent Mix2 expression (data not shown). These results indicate that Shisa2 and 3 uniquely inhibit Wnt and FGF signaling, but not Nodal/activin or BMP signaling.

To examine the overexpression phenotype of Shisa2 and 3, synthesized RNA was injected into four-cell stage embryos(Fig. 1F-F″). Embryos receiving a high dose of Shisa2 or Shisa3 RNA (50 pg per blastomere) exhibited a shortened body axis with an open neural tube (arrow)and enlarged cement gland (arrowhead), demonstrating that Shisa2 and 3 affect both AP patterning and morphogenetic activities of the Xenopusembryo. A low-dose injection of Shisa2 (3 pg per blastomere)inhibited the elongation of the dorsolateral marginal zone explants of gastrulae without affecting the mesodermal differentiation(Fig. 1G-H′), showing that Shisa2 affects the convergent-extension cell movement of somitic precursors, possibly through regulation of the Wnt/PCP pathway. At the neurula stage, unilateral injection of a low dose of Shisa2 RNA disturbed convergence of the MyoD-positive PSM cells toward the dorsal midline(Fig. 1I,I′).

Shisa2 and Shisa3 do not have a known ER retention signal, as is also true of Shisa1. In Hek293T cells, however, HA-tagged Shisa2 and Shisa3 specifically localized in the ER (Fig. 2A-B″). We examined the mode of action of Shisa2 and 3 in Wnt signaling in Hek293T cells. Ligand cells expressing Wnt3a and receptor cells expressing Fz8 and Lrp6 together with TOPFLASH reporter(Korinek et al., 1997) were prepared independently and mixed for stimulation. The non-cell-autonomous action of Wnt3a elevated the reporter activity (threefold; Fig. 2C, lane 2). When either Shisa 2 or 3 was coexpressed with Wnt-ligand, reporter activity was not affected (Fig. 2C, lanes 5 and 6); however, expression of Shisa2 or 3 in the receptor cells suppressed the reporter activity below the basal level(Fig. 2C, lanes 3 and 4),demonstrating that Shisa2 and 3 cell-autonomously inhibit Wnt signaling in these cells.

To analyze whether Shisa2 and 3 physically interact with Fz and FGFR,receptors tagged with HA were coexpressed together with Shisa2- or Shisa3-tagged Flag, and immunoprecipitated with anti-Flag mAb. We found the low molecular weight form of Fz8 and FGFR1 could be immature glycosylated Fz and FGFR (Yamamoto et al.,2005) in the precipitates of the Shisa2 and 3, indicating that Shisa2 and 3 physically interact with immature forms of Fz and FGFR(Fig. 2D,E). Furthermore,Shisa2 and 3 retain the Fz8 and FGFR1 in the ER(Fig. 2F-I″ for Shisa2;data not shown for Shisa3). We also examined whether Shisa2 retains other Fz homologs in the ER. In Xenopus, Fz2 and 7 are expressed in the PSM:they share 78% identities in the amino acid level(Deardorff and Klein, 1999; Sumanas et al., 2000). We found that Shisa2 retained Fz7 in the ER (n=20/20; data not shown). Altogether these results indicate that Shisa2 and 3 inhibit both Wnt and FGF signaling through the regulation of protein maturation and cell surface transportation of their receptors within the ER as Shisa1 does.

To further test the functional similarity of Shisa family members in vivo,we examined whether Shisa2 rescues the Shisa1 knockdown phenotype. Knockdown of Shisa1 suppressed the expression of Otx2 at mid-gastrulation(Yamamoto et al., 2005). We found that injection of Shisa2 RNA rescued this phenotype, suggesting that Shisa1 and 2 are functionally exchangeable in vivo at a molecular level(Fig. 2J-J″).

RT-PCR analysis showed that the Shisa2 expression was weak throughout early embryogenesis and increased after the tailbud stage(Fig. 3A). Whole-mount in situ hybridization showed maternal and/or zygotic expression of Shisa2 in the entire animal hemisphere by blastula stage (data not shown). Shisa2 expression in the PSM was first detected at the beginning of neurulation (Fig. 3B). As somitogenesis proceeded, the Shisa2 expression moved posteriorly(Fig. 3C), covering all the somitomeres (S-I, -II, -III) that have committed to segmentation and are visualized by Thy1 expression(Fig. 3E,E′,G). The Thy1 expression in the S-I and -II extended into the lateral plate mesoderm; however, Shisa2 expression in this region was below the detectable level. In the caudal PSM, Shisa2 showed a graded expression: high anteriorly and low posteriorly, extended laterally and partly overlapping with Mes1 expression, a marker for the caudal PSM and tailbud (Fig. 3F,F′,G). At the tadpole stage, Shisa2 expression was detected in the middle of each somite, precursors of the ventral body wall muscle, the otic and optic vesicles, head mesenchyme and brachial arches(Fig. 3D). We also found unique Shisa3 expressions in ventral forebrain and ventral hindbrain at the tailbud stage (data not shown). Shisa3 might play a role in these tissues;however, this study focused on the role of Shisa2 in segmental patterning.

Two antisense morpholino oligonucleotides (MO) were generated toward the translation initiation site and 5′ untranslated region (UTR) of Shisa2, MO1 and MO2, respectively(Fig. 4A). MO1 and MO2 inhibited the translation of Shisa2, but not that of α-tubulin(Fig. 4B). In these studies,the MOs were unilaterally injected, the uninjected side serving as an internal control. During gastrulation, these MOs had no effect on the expression of Xbra (a panmesodermal marker) or MyoD (a myogenic marker),suggesting that Shisa2 has no role in mesoderm induction, or in the dorsoventral and anteroposterior patterning(Fig. 4C-D″). At the early neurula stage, the paraxial mesodermal region stained by MyoDexpression was symmetrical in the MO-injected and uninjected side, suggesting that the early allocation of somitic precursors was not affected by the MO injection (Fig. 4E-E″).

In the early segmentation period (stage 18; in Xenopus, the first somite buds off at stage 16-17) (Hamilton,1969), unilateral injection of MO1 and MO2 elicited the anterior shift of the expression of Thy1 and Papc(Kim et al., 1998) for a distance of one to two segments; the level and the mediolateral width of their expression was not affected (Fig. 4F-G″). The anterior borders of Mes1, Xbra, FGF8and Wnt3a expressions in the caudal PSM and the tailbud were also expanded anteriorly in the Shisa2MO-injected side(Fig. 4H-K″). These results suggest that knockdown of Shisa2 causes delay of the maturation of PSM cells.

Although we tried to rescue the Shisa2 morphant phenotypes by co-injection of carefully titrated Shisa2 RNAs, the morphogenetic defects of gastrulae caused by ectopic Shisa2 expression severely disturbed early allocation of somitic precursors (as shown in Fig. 1F-I′): this made it difficult for us to come to any conclusion on this issue. To further confirm the specificity of the Shisa2 morphant phenotype, we generated a third Shisa2MO (MO3; Fig. 4A′), which inhibits splicing of endogenous Shisa2mRNA (Fig. 4B′). We found that unilateral injection of MO3 also resulted in the anterior shift of the expression of Thy1 and Mes1(Fig. 4L-M′). These results further support the specificity of the Shisa2 morphant phenotypes.

Next we examined whether knockdown of Shisa2 caused anterior extension of Wnt and FGF signaling activities. In mice, Axin2 has been thought to be a direct downstream target of Wnt3a in the PSM(Aulehla et al., 2003). In Xenopus, expression of the Axin-related gene (Arp-A)(Itoh et al., 2000) was also under the control of Wnt signaling but not FGF (see Fig. 8C-C″). In the Shisa2 morphant, the expression of Arp-A and phospho-ERK (dp-ERK)staining were extended anteriorly (Fig. 4N-O′), indicating that knockdown of Shisa2 caused anterior extension of both Wnt and FGF signaling activities. We also found the anterior extension of retinoic acid hydroxylase Cyp26(Hollemann et al., 1998)expression and RA receptor gamma Rarg(Pfeffer and De Robertis,1994), but not that of a dehydrogenase of RA Raldh2(Chen et al., 2001) in the Shisa2 morphant (Fig. 4P,Q;data not shown for Raldh2).

Next we examined whether the depletion of Shisa2 affects the cyclic expression of the segmental clock gene. In Xenopus, three distinct phases of cyclic expression are observed for ESR9(Hairy/Enhancer-of-split related 9), a possible component in the Notch signaling cascade (Fig. 5A-C) (Li et al.,2003). By the Shisa2 depletion, the anterior border of the ESR9 expression was expanded anteriorly; however, the phase of the expression was not affected (Fig. 5D-F for MO1; data not shown for MO2).

Histological analysis of the longitudinal horizontal section demonstrated that Shisa2 depletion elicited anterior displacement of the most newly formed somite for a distance of one to two somites, accompanying the anterior expansion of PSM. The serial sections showed that the position of the first somite was symmetrically at the same level in the MO-injected and uninjected sides, and the segment number was reduced by the knockdown of Shisa2(Fig. 5G-L). Our data suggest that the altered morphogenetic gradient by the Shisa2 deficiency affected the generation of somites for a few segments and consequently reduced their number.

The maturation of PSM and transition to the segmental fate are known to be regulated by a morphogenetic gradient established by the interaction among RA,FGF and Wnt signaling. To address the role of Shisa2 in the setting up of the PSM gradient, we examined the relationship between Shisa2 function and the RA/FGF gradient in the maturation of the PSM cells. Embryos unilaterally receiving Shisa2MO1 were treated with RA or SU5402 (a chemical inhibitor of FGFR function) at the early neurula stage for 1.5 hours, which approximately corresponded to the period in which two segments are generated(Hamilton, 1969), and then stained for Mes1 and Thy1 expressions. These treatments affect the maturation of the caudal PSM and the most newly formed segment in the S-III, but not older segments (e.g. S-I). In the uninjected side, both RA and SU5402 treatments abolished Mes1 expression(Fig. 6A-C) and induced ectopic Thy1 expression in the caudal PSM(Fig. 6D-F; ectopic Thy1 expression is indicated by red arrowheads)(Moreno and Kintner, 2004). The treatments did not affect the endogenous Shisa2 expression(Fig. 6J-J″). In the Shisa2-depleted side, both the reduction of Mes1 expression and the ectopic induction of Thy1 in the caudal PSM were significantly inhibited (Fig. 6A-F,G-I).

Although the knockdown of Shisa2 inhibited the effects of RA on the positioning of the S-III/Thy1 stripe, it did not inhibit the RA-mediated expansion of Thy1 expression(Fig. 6E; the expanded Thy1 stripes are marked by white brackets), which is a direct target of RA signaling (Moreno and Kintner,2004). RA treatment in wild-type embryos directly activated ubiquitous expression of Cyp26(Loudig et al., 2000)(Fig. 6K′). The knockdown of Shisa2 had no effect on the RA-induced Cyp26 expression(Fig. 6K″). These results suggest that the inhibition of the effects of RA on the wavefront by the Shisa2 knockdown would be indirect.

In the tailbud explants, the depletion of Shisa2 also inhibited the reduction of Mes1 expression by SU5402 treatment, the expression of FGF8 and Wnt3a being unchanged(Fig. 6L). This Shisa2 knockdown effect was not caused by the reactivation of FGF signaling, as the SU5402 treatment strongly suppressed the activation of ERK/MAPK in both control and Shisa2-depleted embryos (Fig. 6M). These results strongly suggest that Shisa2 functions in the maturation of the PSM cells by regulating signals other than RA and FGF signaling, or by both FGF and Wnt signaling together.

As Shisa2 inhibits Wnt and FGF signaling, it is possible that inhibition of both these types of signaling canceled the anterior shift of the S-III/Thy1 stripe in the Shisa2 morphants. To inhibit Wnt signaling,we injected a small amount of Gsk3 RNA, an inhibitor of the canonical Wnt pathway but not of Wnt/PCP: this had no or little effect on the early allocation of PSM precursors at gastrulation (data not shown). Embryos radially receiving Gsk3 RNA were subsequently injected with Shisa2MO1 unilaterally and were treated with SU5402 at the early neurula stage for 1.5 hours. This manipulation generated a larger posterior shift of the S-III/Thy1 stripe in the Shisa2-depleted side and positioned the stripes in a symmetrical manner(Fig. 7A-A‴).

Next we further examined this issue using dorsal explants that were generated at the early neurula stage (stage 15)(Fig. 7B). The explants were treated for 1.5 hours with SU5402 and/or WIF1 protein, and subsequently analyzed for Mes1 and Thy1 expression. The knockdown of Shisa2 expanded Mes1 expression anteriorly, as seen in the whole embryo (Fig. 7C). The inhibition of either Wnt or FGF signaling alone reduced Mes1expression in the uninjected side, whereas it was insufficient to abolish this expression in the Shisa2-depleted side(Fig. 7C′,C″). Inhibition of both types of signaling together, however, efficiently abolished Mes1 expression in the caudal PSM of the Shisa2 morphants(Fig. 7C‴,F).

We then examined whether the inhibition of Wnt signaling by the WIF1 or conditioned media containing FzCRD (a soluble form of Fz8), together with SU5402 treatment, abolishes the anterior shifted S-III/Thy1 stripe in the Shisa2 morphants. As in the case of Mes1 expression, the inhibition of either signaling alone was insufficient to cancel the anterior shift of the stripe (Fig. 7D′,D″,E,G); however, inhibition of both types of signaling together positioned the stripe symmetrically(Fig. 7E′,E″,G). Altogether, the present results strongly suggest that Shisa2 promotes maturation of the PSM cells by the individual inhibition of both Wnt and FGF signaling.

The dorsal explants from wild-type embryos were treated with WIF1 or SU5402 for 1.5 or 3 hours. In the 3 hour treatment, but not in the 1.5 hour treatment, WIF1 and SU5402 suppressed MAPK phosphorylation and Arp-Aexpression, respectively (Fig. 8A-C″). These results show that the inhibition of each signaling for a longer period induces mutual regulation of their signaling activities.

To analyze the individual role in the positioning of the wavefront, we examined the average distance between S-II/Thy1 and S-III stripes of the 1.5 hour explants. Compared with the SII-III distance of the control explants (n=36), that of SU5402-treated and WIF1-treated explants were 1.10-fold (n=51) and 1.15-fold (n=32), respectively(Fig. 8D-D″,E),suggesting that these two types of signaling individually reposition the S-III/Thy1 stripes. We further asked whether the inhibition of these two types of signaling together synergistically reposition the wavefront posteriorly. The SII-III distance of the explants treated with WIF1 and SU5402 was 1.14-fold (n=34) (Fig. 8E). Thus we did not observe further posterior shift of the S-III stripe with the inhibition of both Wnt and FGF signaling together.

In a previous study we reported the characterization of Shisa1, which promotes Xenopus head formation(Yamamoto et al., 2005). Shisa1 uniquely inhibits Wnt and FGF signaling by suppressing the protein maturation of their receptors in the ER. It remains uncertain, however,whether this regulatory mechanism functions in other Wnt- and FGF-related events during vertebrate embryogenesis. To address this issue, we isolated and characterized Xenopus Shisarelated genes, Shisa2 and 3. Our data indicate that Shisa-related molecules constitute a functionally conserved new gene family. Furthermore, we show that Shisa2-mediated regulation of FGF and Wnt signaling plays an essential role in segmental patterning during somitogenesis.

In segmentation, positional information provided by morphogenetic gradients controls the maturation of the somitic precursors and transition to the segmental fate. This crucial process takes place in the caudal PSM, where cells express the bHLH transcription factor Mes1, an essential factor for maintenance of the immature state of PSM as well as activation of the segmental clock (Yoon et al.,2000; Yoon and Wold,2000). It seems likely that the termination of Mes1expression is a prerequisite to activate the expression of another bHLH transcription factor, Thy1, an essential factor for specifying a position of the segment boundary and the anteroposterior polarity of somites(Morimoto et al., 2005; Nomura-Kitabayashi et al.,2002; Saga et al.,1997; Sawada et al.,2000; Sparrow et al.,1998). It has recently been reported that Mesp2 (mouse ortholog of Thy1) arrests the oscillation of Notch activity and initiates the segmentation program in the rostral PSM(Morimoto et al., 2005; Takahashi et al., 2000). Thus,the transition of the expression of these two transcription factors seems to be coincident with the wavefront. We found that both Wnt and FGF signaling are required in the initiation and/or maintenance of Mes1 expression, and that the Shisa2-mediated inhibition of these two types of signaling is required for the proper termination of Mes1 expression(Fig. 7). FGFR1, Fz7 and Fz2 are expressed in the PSM (Deardorff and Klein, 1999; Golub et al.,2000; Sumanas et al.,2000); Shisa2 would play an essential role in the regulation of the morphogenetic gradient by controlling protein maturation of these receptors. These present results provide an additional context, segmentation in the PSM, in which Shisa-mediated signaling regulation controls cell-autonomous competence to respond to Wnt and FGF signaling for the establishment of vertebrate body patterning.

FGF8 and Wnt3a are expressed in the posteriormost mesoderm and generate signaling gradients: low rostorally and high caudally. Coupled with axial elongation during segmentation, diffusion of FGF8 and Wnt3a proteins, and also possibly decay of their transcripts, generates the morphogenetic gradient(Dubrulle and Pourquie,2004b). We show, however, that in the absence of Shisa2 function,the regulation of FGF and Wnt ligands is not sufficient to generate the morphogenetic gradient for proper segmental patterning. Shisa2-mediated suppression of FGF and Wnt signaling in the PSM is required for setting up the gradient. Shisa2 is expressed strongly in the region covering the wavefront, but its expression is low in the posteriormost PSM(Fig. 3). As somitogenesis proceeds, Shisa2 expression moves posteriorly, coupled with axial elongation. Shisa2 inhibits surface expression of Fz and FGFR by retaining them in the ER. Thus, it is likely that Fz and FGFR are expressed at a high level in the posteriormost PSM and at a lower level in the anterior PSM. In this context, cells in the posterior PSM can respond to FGFs and Wnts better than those in the anterior PSM. In cooperating with the gradients of the ligand proteins, the gradients in the receptor expression may play a crucial role in setting up the morphogenetic gradient. As the knockdown of Shisa2 delays the maturation of PSM cells through the anterior extension of FGF and Wnt signaling activities, this mechanism would contribute to the formation of the morphogenetic gradient in the PSM.

In Xenopus, RA signaling is reported to inhibit FGF signaling by upregulating the MAPK phosphatase MKP3. A negative regulation of FGF signaling by RA is also reported in chick embryos(Diez del Corral et al., 2003; Moreno and Kintner, 2004). These reports suggest that RA signaling modifies the morphogenetic gradient at least in part by suppressing FGF signaling. Although the RA treatment or inhibition of FGFR caused a posterior shift of the wavefront(Moreno and Kintner, 2004)(Fig. 6), neither of them could suppress the phenotypes of the Shisa2 knockdown. The data indicate that inhibition of FGF by Shisa2 alone cannot explain the loss-of-function phenotype. The inhibition of both FGF and Wnt signals, however, strongly suppressed the phenotype and shifted the wavefront posteriorly(Fig. 7). Furthermore, in the normal condition we found that the inhibition of these two types of signaling independently repositioned the third Thy1 stripes posteriorly(Fig. 8). Altogether these results strongly suggest that Shisa2-mediated individual inhibition of Wnt and FGF signaling is required for the proper positioning of the wavefront.

The components of the Notch signaling cascade are involved in the segmental clock (Bessho and Kageyama,2003; Giudicelli and Lewis,2004; Pourquie,2003; Rida et al.,2004). In Xenopus, the expression of ESR9 and the closely related ESR10 are oscillated in the caudal PSM(Li et al., 2003). We have found that the loss of function of Shisa2 expanded the anterior border of ESR9 expression but did not affect its oscillation(Fig. 5). It has been suggested that FGF and Wnt signaling not only controls the maturation of the PSM and the position of the wavefront but also the segmental clock system. In vt/vt mutants, the expression of the cyclic genes is affected(Aulehla et al., 2003; Dale et al., 2006; Ishikawa et al., 2004). LEF/TCF factors, a component of the Wnt signaling cascade, directly control the expression of Delta-like1, thereby controlling the segmental clock (Galceran et al., 2004; Hofmann et al., 2004). In zebrafish, Her13.2 (Hes6-related hairy/Enhancer split-related), which functions downstream of FGF signaling, controls the expression of cyclic genes such as Her1 and Her7(Kawamura et al., 2005). By contrast to these previous reports, Shisa2-mediated inhibition of FGF and Wnt signals did not affect the phase of expression of the ESR9. It is tempting to speculate that the high level of FGF and Wnt signals in the tailbud is the source of generation of the cyclic expression of the genes. Shisa2 controls the FGF and Wnt signals in the region more anterior to the tailbud; thus it controls the wavefront but not the segmental clock.

In summary, Shisa2 plays an essential role in the maturation of PSM and establishment of proper segmental patterning by the individual inhibition of Wnt and FGF signaling.

We thank the National Institute for Basic Biology of Japan for gifts of plasmids. We gratefully acknowledge Drs H. Takeda, Y. Saga, M. Hibi and M. Royle for helpful discussion and critical comments on this manuscript. This work was supported by Grants-in-Aid for Scientific Research on Priority Areas from the Ministry of Education, Culture, Sports, Science and Technology of Japan.