The negative regulation of Mesp2 by mouse Ripply2 is required to establish the rostro-caudal patterning within a somite

Somites are generated by sequential segregation of cell masses from the anterior part of the unsegmented presomitic mesoderm (PSM), in both a spatially and temporally coordinated manner every two hours(Iulianella et al., 2003; Pourquie, 2003; Saga and Takeda, 2001). The somites provide the basic axial structures that underlie the segmental architecture of not only the vertebra, ribs and muscles, which are all somite derivatives, but also of the vascular and nervous systems(Borycki and Emerson, 2000; Brand-Saberi and Christ, 2000; Monsoro-Burq and Le Douarin,2000). Periodicity is generated by Notch signal oscillations linked to the segmentation clock (Bessho et al., 2001; Huppert et al.,2005; Morimoto et al.,2005; Rida et al.,2004). The temporal information that results from this is translated into spatial patterns in the anterior PSM, which is defined by the so-called determination front (Dubrulle and Pourquie, 2004).

The Mesp2 transcription factor plays important roles during somitogenesis(Saga et al., 1997), and its expression is periodically activated by cyclic Notch signaling and Tbx6 at the anterior PSM in the determination front(Yasuhiko et al., 2006). Mesp2 demarcates the next segmental boundary and defines the rostro-caudal identity of somites (Takahashi et al.,2000). It has been shown that Mesp2-null embryos fail to segment and that the resulting non-segmented somites show caudalized properties (Saga et al.,1997). Previously, we have shown that Mesp2 suppresses Notch activity via the activation of Lfng, which might function as a negative regulator of Notch signaling(Morimoto et al., 2005). In addition, Mesp2 acts as the transcriptional activator of Epha4 in the anterior PSM (Nakajima et al.,2006). Mesp2 is also known to be a strong suppressor of genes such as Dll1 and Uncx4.1 that confer caudal properties upon the somitic cells via Notch signaling(Takahashi et al., 2000). However, the manner in which the caudal genes are suppressed is currently unknown. In our current study, which aimed to elucidate the molecular mechanisms underlying the regulation of somitogenesis by Mesp2, we have compared the gene expression patterns of Mesp2^+/- and Mesp2^-/- embryos, and found that several genes are affected by the Mesp2 knockout. Among the downregulated genes that we identified in the Mesp2-null embryo, we focused on a putative transcriptional repressor. This gene turned out to be Ripply2, which was recently reported as a mouse homolog of zebrafish ripply1(Kawamura et al., 2005). Morpholino-mediated knockdown analysis revealed that ripply1 is required for the proper transition from the PSM to somites. We generated a Ripply2-knockout mouse and now show that Ripply2 is activated by Mesp2, but also functions negatively toward Mesp2 to regulate the levels of Notch signaling in the anterior PSM. This negative regulation is required for the periodic generation of the rostro-caudal patterning within a somite.

Total RNA was purified from cells corresponding to the S-1 to S2 somites and PSM of wild-type, Mesp2-GFP knock-in heterozygous and homozygous embryos at E10.5 using the RNeasy Mini Kit (Qiagen) according to the manufacturer's instructions. First-strand cDNAs were synthesized by incubating 5 μg of total RNA with 200 U SuperScript II reverse transcriptase(Invitrogen) and 100 pmol T7-(dT)₂₄ primer[5′-GGCCAGTGAATTGTAATACGACTCACTATAGGGAGGCGG-(dT)₂₄-3′]. After second-strand synthesis, the double-stranded cDNAs were purified using a GeneChip Sample Cleanup Module (Affymetrix), according to the manufacturer's instructions. Our detailed methods for the labeling of the double-stranded cDNAs and hybridization to a GeneChip Mouse Genome 430 2.0 Array (Affymetrix),and the subsequent washing, staining and data analysis have been described previously (Kanno et al.,2006). All of these data are also now available online at the National Institute of Health Sciences(http://www.nihs.go.jp/tox/TtgSubmitted.htm).

Highly conserved Ripply2 upstream regions were identified using a cross-species DNA sequence comparison using the PipMaker website(http://pipmaker.bx.psu.edu/pipmaker/). We cloned a 5′ upstream genomic sequence of Ripply2 from a bacterial artificial clone (RP23) contained in a mouse genomic library. A 1.5 kb DNA fragment containing the 171 bp highly conserved region was isolated by EcoRI and BamHI digestion and then subcloned into the hsp-nlacZ reporter construct. Fertilized eggs from B6C3F1 female mice were collected for pronuclear injection and the injected eggs were then implanted into ICR female mice. Foster mothers were sacrificed at E10.5 and stained forβ-galactosidase (β-gal) activity with X-Gal. The genotypes of the embryos were then identified by PCR using DNA prepared from the yolk sac.

For luciferase reporter analysis under the control of the 1.5 kb Ripply2 anterior-PSM enhancer (EcoRI-BamHI)fragment (20 ng), reporter constructs were co-transfected with the expression vectors 3xFLAG-Mesp2 (0, 30, 100 ng) and 3xFLAG-E47 (0, 50 ng) into NIH3T3 cells (0.25×10⁵ cells per well in 24-multiwell plates) using Lipofectamine Plus (Invitrogen), following the manufacturer's instructions. The vector containing the Renilla luciferase gene under the control of the thymidine kinase promoter (1 ng) was co-transfected as an internal standard to normalize for transfection efficiency, and the amount of total plasmid was adjusted with pcDNA3.1. After 36 hours, luciferase activities were measured using a Dual Luciferase Assay Kit (Promega).

A 3xFLAG-Mesp2 protein was produced using the FreeStyle 293 Expression System (Invitrogen) and then collected via a nuclear extraction method. Double-stranded DNA oligonucleotide probes were end-labeled with DIG and protein-DNA complexes were detected using a DIG Gel Shift Kit (Roche). Binding reactions were carried out for 30 minutes on ice, and protein-DNA complexes were analyzed on 6% native polyacrylamide gels.

The mouse Ripply2 gene consists of four exons, the first of which harbors two putative in-frame translational initiation codons. We generated a targeting vector with a floxed neo cassette to remove a portion of exon 1, which would introduce a termination codon just after the second initiation codon and produce a null allele. The resulting linearized vector(25 μg) was then electroporated into TT2 ES cells(Yagi et al., 1993). G418-resistant cell clones were further selected by PCR. Correct homologous recombination was confirmed by Southern blotting, and targeted cell clones were aggregated with ICR 8-cells and then transferred to pseudopregnant female recipients. The resulting chimeric mice were bred with ICR females. Germline transmission of the targeted allele was confirmed by PCR. The floxed neomycin cassette was later removed by breeding with a CAG-Cre transgenic mouse(Sakai and Miyazaki,1997).

Methods for gene expression analysis by in situ hybridization of whole-mount samples and by skeletal staining have been described previously(Takahashi et al., 2000). The probes used in this study have been described previously(Takahashi et al., 2000; Takahashi et al., 2003; Nomura-Kitabayashi et al.,2002). For the Ripply2 RNA probe, we used a full-length cDNA clone containing intron 1. Section in situ hybridization and immunohistochemical detection of proteins were performed as previously described (Morimoto et al.,2005). For whole-mount detection of Mesp2-venus, embryos were fixed with 4% paraformaldehyde in PBS overnight at 4°C, incubated with rabbit anti-GFP (MBL; 1:1000), followed by Alexa-488-conjugated goat anti-rabbit IgG (Molecular Probes; 1:400) and observed using a fluorescent microscope (Olympus BX61).

Mesp2 is known to function as a transcriptional activator of genes, such as Epha4 and Lfng, which are expressed in the rostral half of the presumptive somite (Morimoto et al.,2005; Nakajima et al.,2006). Moreover, expression of the Dll1 and Uncx4.1 genes, which are expressed in the caudal half of the somite(Bettenhausen et al., 1995; Leitges et al., 2000), is increased in the Mesp2-null mouse, indicating that Mesp2 is required for their suppression (Takahashi et al.,2000). However, the molecular mechanisms underlying this are unknown. To identify novel genes that operate downstream of Mesp2, we performed GeneChip analysis using RNAs prepared from both wild-type and Mesp2-null embryos. Among the genes that showed a reduction in expression in the Mesp2-null embryos (see Table S1 in the supplementary material), we selected a cDNA clone (corresponding to RIKEN cDNA C030002E08) that showed an identical expression pattern to that of Mesp2 by in situ screening of 11.5 dpc embryos. This cDNA was subsequently revealed to be the mouse Ripply2 gene recently reported by Kawamura et al. (Kawamura et al.,2005). The initial expression of Mesp2 was found to be restricted to the nascent mesoderm at E7.0, but Ripply2 expression appeared to be absent or very weak prior to somitogenesis(Fig. 1A). However, its expression became evident in the anterior PSM as a pair of bands by 8.0 dpc,similar to Mesp2 (Fig. 1A). The expression of Ripply2 then continued until 12.5 dpc, during the somite-forming period (Fig. 1A and data not shown). The expression domains of Mesp2and Ripply2 were next compared by double in situ hybridization of embryonic tail sections. Two typical patterns are shown in Fig. 1B. One shows single bands that are completely merged, whereas the other is of a single Mesp2band and two Ripply2 bands in which the caudal band is merged with a distinct Mesp2 band. This observation indicates that Mesp2expression precedes that of Ripply2, but that Ripply2persists for longer. In addition, Ripply2 expression was lost in the Mesp2-null embryo (Fig. 1C), as predicted from our GeneChip analysis. These data thus indicated that Ripply2 might be a target of Mesp2.

To examine this possibility, we searched for possible cis-regulatory sequences in the Ripply2 gene by comparing mouse, human, dog and chick genomic sequences using MultiPipMaker sequence alignment software(Fig. 2A). From these analyses,we identified a conserved region (-6917 to -6747, Fig. 2B). To investigate whether the 1.5 kb region containing this conserved 171 bp sequence(Fig. 2B) possessed enhancer activity, we performed transient transgenic analyses using a β-gal reporter (Fig. 2C). In five out of nine PCR-positive embryos, we detected specific β-gal expression in several segmented somites (Fig. 2D), which is a typical pattern for genes expressed in the anterior PSM, including Mesp2 and Epha4(Haraguchi et al., 2001; Nakajima et al., 2006). We next employed a luciferase reporter assay system to ascertain whether the enhancer activity was dependant upon Mesp2. Two reporter constructs were generated - one containing the 1.5 kb genomic fragment and the other harboring six repeats of the 171 bp consensus sequence. Both constructs were activated by the addition of Mesp2, but not in conjunction with E47 (also known as Tcfe2a - Mouse Genome Informatics) (Fig. 2E). This result was different from the findings of our previous study of the Epha4 enhancer(Nakajima et al., 2006), in which Mesp2 was observed to bind and transactivate the reporter activity only in the presence of E47, a possible heterodimeric partner. Since Mesp2 belongs to the bHLH-type transcription factor family, which is known to bind either to E-box or N-box motifs, we screened the 171 bp Ripply2 gene consensus sequence for E-boxes, or for an N-box which is capable of binding to Mesp2 with or without E47. We identified a DNA fragment containing a highly conserved E-box CATCTG sequence, and confirmed that this binds to Mesp2,whereas a mutated form did not (Fig. 2F,G). E47 was also found to bind to this E-box, but this might not be functional binding as no associated activity was detectable by luciferase reporter assay. Furthermore, the binding of Mesp2 was weakened by the addition of E47. These results are consistent with the idea that Mesp2 binds to this E-box in the enhancer of the Ripply2 gene, and that this enhancer does not require E47 for subsequent transactivation.

Because Mesp2 confers rostral properties to the somites and is involved in the formation of the somite boundary, we speculated whether Ripply2 might function in this Mesp2 pathway during somitogenesis. To elucidate this possibility, we generated Ripply2-knockout mice using ES cell-mediated gene targeting (Fig. 3A). Since the heterozygous mice were found to be normal, we performed timed intercross matings to analyze the phenotypes of the homozygotes. As expected from the expression patterns, the Ripply2^-/- embryos failed to proceed through normal somitogenesis and the embryos displayed no clear segmental borders(Fig. 4A-D). These homozygous mice also died soon after birth. The morphology of the 17.5-dpc fetus was found to be similar to that of the Mesp2-null embryo, with a short trunk and tail (Fig. 3B)(Saga et al., 1997). However,the vertebral phenotype of the Ripply2^-/- embryos, as revealed by skeletal staining, differed from that of Mesp2-null embryos as it features extensive fusion of the pedicles in the neural arches owing to the caudalized characteristics of the somitic mesoderm(Saga et al., 1997).

The Ripply2^-/- mouse fetus showed fewer pedicles of neural arches (Fig. 4E,F), and the phenotype resembled that of the presenilin 1 (Psen1)-null mouse(Fig. 4G), which lacks Notch signaling (Koizumi et al.,2001). The findings of our gene expression studies using both rostral and caudal molecular markers are consistent with these skeletal defects. In Ripply2^+/- embryos, the expression of the caudal markers Uncx4.1 and Dll1 in the segmented somites was restricted to the caudal compartments of the somites(Fig. 5A,C). The expression of these genes is increased and more expansive in Mesp2-null embryos(Takahashi et al., 2000), but was greatly reduced in the Ripply2-null embryos,(Fig. 5B,D). In addition, no Dll1 stripe could be observed within the somitic region or in the anterior PSM, although the expression in the posterior PSM was intact in the Ripply2^-/- embryo (Fig. 5D). By contrast, the rostral markers were found to be present in the Ripply2^-/- embryo(Fig. 5E-H). Tbx18,which is known to be involved in the maintenance of the rostral properties of the somites (Bussen et al.,2004; Kraus et al.,2001), was expressed in the rostral compartment of the segmented somites (Fig. 5E). In Ripply2^-/- embryos, this expression was expanded throughout the entire somite region and no clear segmental pattern was evident(Fig. 5F). In addition, Epha4 was expressed in the rostral compartment of S0 and S1 somites in the Ripply2^+/- embryo(Fig. 5G), and this expression in the Ripply2^-/- embryo was increased and the expression domain expanded as compared with the wild type(Fig. 5H). We thus conclude that the Ripply2-null mouse displays a rostralized phenotype.

As we have previously reported, the rostro-caudal polarity of the somites is generated by the interaction between Mesp2 and the Notch signaling pathway in the anterior PSM (Morimoto et al.,2005). To identify the underlying cause of the rostralized phenotype in the Ripply2-null embryo, the Mesp2 gene expression profile was examined. During somitogenesis in the anterior PSM,both wild-type and Ripply2^+/- embryos generally showed either a single Mesp2 expression band of variable width or no band,depending on the cyclic expression stage(Fig. 5I and data not shown). However, we observed that Mesp2 is expressed in the Ripply2^-/- embryo as either one or two bands(Fig. 5J). In other words, an additional band was frequently observed in the more-rostral region (four out of six examined). In addition, we did not observe any Ripply2^-/- embryos without Mesp2 expression,suggesting either that Mesp2 expression is prolonged or that the Mesp2 transcripts are stabilized in a Ripply2^-/-background. To distinguish these possibilities, we performed in situ hybridization using an intron probe. Although the signal obtained was low, we frequently detected two bands in the Ripply2^-/- embryos(Fig. 5L; in all three examined), but only one band in the Ripply2^+/- embryos(Fig. 5K). Hence, the transcription of Mesp2 appears to be prolonged in the absence of Ripply2, although the possibility that differences exist in their mRNA stability cannot yet be excluded.

We next examined how the expression of the Mesp2 protein is influenced in the Ripply2^-/- background. As we have shown previously,Mesp2-venus can be used to visualize functional Mesp2 proteins in vivo because the homozygous knock-in mouse is viable and shows normal somitogenesis(Morimoto et al., 2005). In a typical case, a single Mesp2-venus band was detectable in the area just caudal to the next presumptive segmental border in the Ripply2^+/-background (Fig. 5M). However,in the Ripply2^-/- embryo, two broader and interconnected bands could be discerned (Fig. 5N). These data suggest that Mesp2 is negatively regulated by Ripply2, and that these factors form a negative-feedback loop to restrict the levels of Mesp2.

We previously reported that Lfng expression is activated by Mesp2 in the anterior PSM and is subsequently involved in the suppression of Notch signaling. Moreover, Lfng expression shows a cyclic wave-like pattern in the posterior PSM, but its expression in the anterior PSM is similar to that of Mesp2 in Ripply2^+/- embryos. The width of this Lfng band becomes thinner before disappearing from the rostral end of the expression domain in the Ripply2^+/- embryo(Fig. 5O). However, in the Ripply2^-/- embryos, the anterior-most Lfng band was found to be wider and to persist for much longer as compared with the Ripply2^+/- embryos(Fig. 5P). This persistent expression of Lfng was also evident from 20-minute explant culture experiments with a half-PSM (Fig. 5Q,R). These results suggest that Notch signaling might be suppressed, even in the presumptive caudal compartment of the somites, by prolonged Mesp2 and/or Lfng expression in the Ripply2^-/-embryo.

In somite-stage embryos, Notch activity oscillates in the posterior PSM and stabilizes as a clear stripe in the anterior PSM with elevated activity(Huppert et al., 2005; Morimoto et al., 2005). To further understand the molecular events operating in the anterior PSM of Ripply2^-/- embryos, we first examined the expression of Notch1 mRNA (Fig. 6A,B) and Notch1 protein (Fig. 6C,D) in these embryos. Interestingly, these expression patterns were found to be expanded in the anterior PSM in the Ripply2^-/- embryo (Fig. 6B,D), but the Notch activity appeared to be lost as judged from the fact that the expression of Hes5, a Notch target gene(Ohtsuka et al., 1999), was absent (Fig. 6E,F). To further confirm this reduced Notch1 activity and its relationship to Mesp2 expression,we conducted double immunostaining analysis using anti-active Notch1 and anti-Mesp2 antibodies in both wild-type and Ripply2^-/-embryos. In the wild-type embryos, the Notch activity in the anterior PSM exhibited a sharp boundary with Mesp2 that determines the next segmental boundary (Fig. 6G). In addition, the contrast between Notch activities leads to the generation of future rostral and caudal compartments of the somites, whereby the Notch active site becomes the future caudal compartment. In the Ripply2^-/- embryo, the Notch1 signals oscillated normally in the posterior PSM (Fig. 6H,I). However, the elevation of Notch activity in the anterior PSM appeared to be repressed in these null embryos, whereas the Mesp2 expression banding was found to upregulated, as shown previously(Fig. 6H,I).

Since the expression of Lfng is under the control of Mesp2, we speculated that the suppression of Notch signaling might be the result of the prolonged activation of Lfng in the Ripply2^-/-embryo. To test this possibility, we generated a Ripply2/Lfng double-knockout embryo from which we prepared skeletal specimens, and then examined the somite properties by analyzing the expression of the caudal molecular marker Uncx4.1(Fig. 7). Intriguingly, the vertebral morphology of the Ripply2/Lfng double-knockout mouse was not recovered, and was more rostralized as compared with either the Ripply2^-/- (compare Fig. 4F with Fig. 7C) or Lfng-null fetus (Fig. 7B). The expression of Uncx4.1 was also not recovered by the additional loss of Lfng(compare Fig. 5B with Fig. 7C), and was found to be completely diminished in the double-knockout embryos.

To determine whether the suppression of Notch signaling is mainly due to the function of Mesp2, we also generated Mesp2/Ripply2double-null mice and analyzed the resulting skeletal phenotypes. As expected,the vertebral morphology of these fetuses was found to be very similar to the Mesp2 single-null fetus, and exhibited a caudalized phenotype(Fig. 7D,E). The expression of Uncx4.1 was also upregulated to similar levels as in the Mesp2-null embryo (Fig. 7E). These results clearly showed that Mesp2 suppresses the expression of this gene independent of Ripply2, and that the defect observed in the Ripply2^-/- mouse can be attributed to the function of Mesp2.

Our current study establishes the hypothesis that the negative-feedback regulation of Mesp2 by Ripply2 constitutes a core component of the regulatory network involved in establishing rostrocaudal patterning. The periodicity of somitogenesis is established by mechanisms based on the negative regulation of several genes in the mouse posterior PSM(Rida et al., 2004), in which the clock genes Hes7 and Lfng are negatively regulated by several mechanisms, including transcriptional suppression, protein degradation and destabilization of mRNA (Bessho et al.,2003; Chen et al.,2005; Cole et al.,2002; Hirata et al.,2004; Morales et al.,2002). In the anterior PSM, the levels of Mesp2 are strictly regulated to achieve the periodic suppression of Notch signaling, and also to establish the correct rostro-caudal polarity. During this activation step, the cooperation between Tbx6 and cyclic activated Notch-signaling is crucial for the periodic induction of Mesp2(Yasuhiko et al., 2006)(Fig. 8A). However, these processes must be regulated by both activation and inhibition. Previously, we reported that Mesp2 is regulated negatively by the proteasome pathway(Morimoto et al., 2006). In addition, our current study has identified Ripply2 as a potent negative regulator of Mesp2 transcription, and as a factor that is required for the correct establishment of rostro-caudal patterning. In the absence of Ripply2, Mesp2 expression is maintained over a longer period and leads to the suppression of caudal properties (Fig. 8B). It is noteworthy in this regard that Ripply2 might function exclusively to negatively regulate Mesp2, because the phenotype of the Ripply2-knockout mouse is almost completely reversed by the additional loss of Mesp2.

The Ripply2-null mutant exhibits not only an expansion of rostral marker genes but also a reduction in the expression of caudal markers. Immunohistochemical analysis further revealed a decrease in the activated form of Notch1 in the anterior PSM in these Ripply2-null embryos. Previously, we have shown that Mesp2 suppresses Notch signaling to establish segmental boundaries via the activation of Lfng. However, Lfng appears not to be crucial for the suppression of Notch signaling in the Ripply2-null embryo as this suppression was not rescued by the additional loss of Lfng,and, in fact, this results in a further reduction in Notch signaling activity. We speculate that this is caused by the function of Lfng during Mesp2 distribution, based upon our observations of the Mesp2-venus knock-in mouse. In the wild-type embryo, the Mesp2-venus expression pattern shows a clear gradient, being higher in the presumptive rostral compartment. However, in the absence of Lfng, such a biased gradient is not generated, and the Mesp2-venus pattern shows a diffuse distribution in this background (our unpublished data). The phenotype of the Ripply2/Lfng double-knockout mouse appears also to reflect this distribution defect. In this double-null mouse, the expression of Mesp2 is prolonged owing to the lack of Ripply2, and is distributed across a much wider area along the anterior-posterior axis because of the lack of Lfng. This in turn enhances the function of Mesp2 that suppresses Notch signaling in the anterior PSM, and results in the somites becoming completely rostralized in these double mutants.

The mechanisms underlying the suppression of Mesp2 by Ripply2 are currently unknown. Ripply2 appears to be required for the termination of Mesp2 expression at an appropriate time. Moreover, because Ripply2 has no apparent DNA-binding domain, it is plausible to assume that it suppresses Mesp2 by recruiting the Groucho homolog Tle1 and/or Tle3 via the WRPW motif, as revealed previously by in vitro assays in both zebrafish and mouse(Kawamura et al., 2005) (data not shown). Tle1 and Tle3 are known to be expressed in the PSM, but their expression patterns are not segmental(Dehni et al., 1995) (our unpublished data), and no loss-of-function studies have yet been reported. In the zebrafish, ripply1 morphants also display upregulation of mespb in their somitic regions, and this is accompanied by the upregulation of tbx24, deltaC and deltaD. This might also account for the upregulation of mespb(Kawamura et al., 2005). We have previously identified a 300 bp upstream region of the Mesp2 gene as a promoter-enhancer sequence required for the faithful expression of Mesp2 in the anterior PSM where T-box factor binding in combination with Notch signaling is involved in the gene activation(Yasuhiko et al., 2006). However, Tbx6 expression is unchanged (data not shown) and the Dll1expression profile is somewhat decreased in the Ripply2-null embryos. Hence, although the impact of the loss of Ripply proteins upon Mesp gene expression appears to be similar between mouse and zebrafish, the underlying mechanisms might well be different.

We are particularly thankful to Yuki Takahashi and Aya Satoh for their valuable technical support and for maintaining the mice used in this study. We thank Randy Johnson for generously providing the Lfng-knockout mouse and Masaru Tamura for permitting us to use the FreeStyle 293 expression system. This work was supported by Grants-in-Aid for Science Research on Priority Areas (B), the Organized Research Combination System and National BioResource Project of the Ministry of Education, Culture, Sports, Science and Technology, Japan.