Screening of FGF target genes in Xenopus by microarray: temporal dissection of the signalling pathway using a chemical inhibitor

Authors

  • Hyeyoung A. Chung,

    1. Department of Developmental Biology, National Institute for Basic Biology, Okazaki, 444-8585 Japan
    2. Department of Chemistry and Biotechnology, School of Engineering, The University of Tokyo, 1-3-7 Hongo, Bunkyo-ku, Tokyo, 113-8656 Japan
    Search for more papers by this author
  • Junko Hyodo-Miura,

    1. Department of Developmental Biology, National Institute for Basic Biology, Okazaki, 444-8585 Japan
    Search for more papers by this author
  • Atsushi Kitayama,

    1. Department of Developmental Biology, National Institute for Basic Biology, Okazaki, 444-8585 Japan
    Search for more papers by this author
  • Chie Terasaka,

    1. Department of Developmental Biology, National Institute for Basic Biology, Okazaki, 444-8585 Japan
    Search for more papers by this author
  • Teruyuki Nagamune,

    1. Department of Chemistry and Biotechnology, School of Engineering, The University of Tokyo, 1-3-7 Hongo, Bunkyo-ku, Tokyo, 113-8656 Japan
    Search for more papers by this author
  • Naoto Ueno

    Corresponding author
    1. Department of Chemistry and Biotechnology, School of Engineering, The University of Tokyo, 1-3-7 Hongo, Bunkyo-ku, Tokyo, 113-8656 Japan
      * Correspondence: E-mail: nueno@nibb.ac.jp
    Search for more papers by this author

  • Communicated by: Eisuke Nishida

* Correspondence: E-mail: nueno@nibb.ac.jp

Abstract

Microarray is a powerful tool for analysing gene expression patterns in genome-wide view and has greatly contributed to our understanding of spatiotemporal embryonic development at the molecular level. Members of FGF (fibroblast growth factor) family play important roles in embryogenesis, e.g. in organogenesis, proliferation, differentiation, cell migration, angiogenesis, and wound healing. To dissect spatiotemporally the versatile roles of FGF during embryogenesis, we profiled gene expression in Xenopus embryo explants treated with SU5402, a chemical inhibitor specific to FGF receptor 1 (FGFR1), by microarray. We identified 38 genes that were down-regulated and 5 that were up-regulated in response to SU5402 treatment from stage 10.5–11.5 and confirmed their FGF-dependent transcription with RT-PCR analysis and whole-mount in situ hybridization (WISH). Among the 43 genes, we identified 26 as encoding novel proteins and investigated their spatial expression pattern by WISH. Genes whose expression patterns were similar to FGFR1 were further analysed to test whether any of them represented functional FGF target molecules. Here, we report two interesting genes: one is a component of the canonical Ras-MAPK pathway, similar to mammalian mig6 (mitogen-inducible gene 6) acting in muscle differentiation; the other, similar to GPCR4 (G-protein coupled receptor 4), is a promising candidate for a gastrulation movement regulator. These results demonstrate that our approach is a promising strategy for scanning the genes that are essential for the regulation of a diverse array of developmental processes.

Introduction

Recent genome biology of several model organisms has been facilitating identification of functional genes based on their structure, spatiotemporal expression pattern, and the pattern of their transcriptional state. In particular, DNA-microarray technology has opened a new area of genes characterization based on a systematic and comprehensive study of gene expression. This has enabled us to examine biological processes with higher temporal and spatial resolutions. In addition to expression profiling, the annotation/identification of genes as a unigene, which will be obtained from the completion of the genome/EST projects, will enhance our understanding of the link between gene structure and function and eventually lead to our understanding the biological significance of each gene encoded by the genome (Altmann et al. 2001; Hoffman et al. 2002; Buttitta et al. 2003).

Xenopus is a model organism whose ESTs have been collected in a large-scale project and the genomic DNA is being sequenced. Combination of DNA information and its experimental advantages, such as ease of surgical operation and microinjection, provides a promising future for understanding developmental events remained to be solved.

FGFs are a family of polypeptides that regulate a number of biological phenomena including cell proliferation, differentiation and migration. In particular, FGFs are known to participate in early embryonic inductions, such as mesoderm induction and in cell-to-cell interactions during organogenesis known as epithelial-mesenchymal interactions (reviewed in Gotoh & Nishida 1996). In the central nervous system, FGF signals are not only implicated in neural induction, but also are believed to be essential for the patterning that includes the formation of the telencephalon and the midbrain-hindbrain barrier (MHB) (Walshe et al. 2002). In Xenopus, the roles of FGF signals have been studied extensively, and they have been shown to regulate mesoderm induction as well as the patterning of anterior-posterior axis. In mesoderm induction, embryonic FGF (eFGF) signal induces Xbra, which in turn activates the transcription of eFGF constituting a closed autoregulatory circuit (Musci et al. 1990; Amaya et al. 1991; Smith et al. 1991; Amaya et al. 1993; Gotoh et al. 1995). More recently, it has been proposed that FGF signal is also essential for the convergent extension of gastrulation cell movements and the signalling pathway regulating it differs from the one that regulates mesoderm induction (Nutt et al. 2001; Frazzetto et al. 2002).

It is well established that FGF family ligands signal thorough receptor tyrosine kinases, whose phosphorylation in turn triggers cytoplasmic events that transmit the external signal to the nucleus. For mesoderm induction, FGF signal is known to activate a canonical MAPK pathway in a ras-dependent fashion (reviewed in Gotoh & Nishida 1996). In contrast, the pathway regulating convergent extension is sensitive to an FGF signal inhibitor, Sprouty2 (Nutt et al. 2001). It is not clear how the FGF signal is bifurcated into these two distinct pathways. To achieve this, we took advantage of differential screening with DNA-microarray that was recently fabricated in our laboratory using 4600 Xenopus embryonic cDNAs and SU5402, which induces conformational changes in the nucleotide-binding loop of FGFR by producing hydrogen bond between the side chain in the hinge region of FGFR and the carboxyehyl group of it. SU5402 treatment of mammalian cells was previously shown to dramatically inhibit the activation of ERK (Mohammadi et al. 1997).

This genome-wide approach enabled us to identify 43 candidates as general FGF targets and 10 of them are promising candidates for components specifically acting in gastrulation cell movements. We performed functional analyses of two interesting genes. Here, we show that a homolog of mammalian mig6 regulates Xmyf5 expression and hence is essential for muscle differentiation and that a GPCR4 homolog is required for gastrulation cell movement during Xenopus embryogenesis.

Results

The effects of SU5402 on Xenopus embryos

The approximate concentration of SU5402 needed to block endogenous FGF signalling was determined according to previous studies of the ascidian, zebrafish, chick, and mouse (Hoffman et al. 2002; Mohammadi et al. 1997; Kim & Nishida 2001; Shinya et al. 2001; Montero et al. 2001; Malcolm et al. 2002). The effectiveness of SU5402 in shutting down the FGF signalling was confirmed by Western blotting with anti-phosphorylated-ERK antibody (Fig. S1A). We compared the amount of dp-ERK (diphosphorylated ERK) in eFGF-stimulated Keller explants either exposed to 50 µm SU5402 or over-expressing XFD or HAVØ, an inert form of XFD. Dose-dependent signal induction by eFGF and inhibition by XFD were confirmed. Compared with XFD, SU5402 treatment led to a more drastic inhibition of dp-ERK activation. This was consistent with the results of microarray analysis: data set of SU5402 included those of XFD (data not shown).

To determine the effective dose of SU5402, embryos and Keller explants were treated with 5 µm to 100 µm of SU5402 for the entire time from early gastrula to tailbud stage, and changes in morphology were examined. Compared with the DMSO-treated control, SU5402-exposed embryos exhibited incomplete closure of the blastopore, a short trunk and hardly distinguishable anterior-posterior structures; these changes increased in severity in a dose-dependent fashion. Fifty micromolar SU5402 was sufficient to inhibit the gastrulation of whole embryos and elongation of Keller explants, respectively.

Next, we investigated the effects of changing the time of exposure to SU5402 by morphological observation and by RT-PCR analysis. SU5402 was added to culture medium around the onset of gastrulation, i.e. stage 10.5, and was removed at stage 11.5. This relatively short exposure also caused the retardation of involution, leading to the incomplete closure of the blastopore and short trunk. Similarly, the exposed Keller explants displayed complete failure of elongation (Fig. S1B).

The expression pattern of marker genes in the Keller explants was investigated by RT-PCR analysis (data not shown). The results were consistent with previous reports (Amaya et al. 1993; Nutt et al. 2001; Malcolm et al. 2002; Monsoro-Burq et al. 2003): mesodermal markers including Xbra, myoD, myf5, and α-actin, as well as non-mesodermal markers, Xspry2 and Xmc, were down-regulated by SU5402. The neural markers we tested, the spinal chord marker HoxB9, presumptive notochord marker Xnot, and posterior marker Krox20 were also suppressed. More anterior markers, such as Otx-2 and En-2, exhibited unchanged or up-regulated expression, and the cement gland marker AG-1 was up-regulated. These results are consistent with the known function of FGF as a posteriorizing factor. Immunostaining with the MZ15 antibody showed that the notochord was partially present in the anterior but not in the posterior structures. Prolonged exposure to SU5402 exaggerated the above gene expression profile.

Microarray analysis of Keller explants

Changes in the gene expression pattern in the Keller explants cultured in the presence of SU5402 were analysed using NIBB non-redundant 4.6-K microarray. Explants from stage 10.5 embryos were treated with DMSO or 50 µm SU5402 until the sibling embryos reached stage 11.5. Through the microarray analysis shown as Fig. 1A and Table 1, we isolated 43 genes whose expression was affected by SU5402, including 38 down- regulated genes (ratio < 2) and 5 up-regulated ones (ratio > 2) (Fig. 1B), suggesting that FGF signalling functions mostly in transcriptional activation.

Figure 1.

We identified FGF target genes by microarray analysis and subclassified them based on their sequence. (A) The data set in the scatter plot, show the expression level of genes on a logarithmic scale as the median value of DMSO-exposed explants on the x-axis vs. the median value of SU5402-exposed explants on the y-axis. (B) Out of the 43 genes including 38 down-regulated genes and 5 up-regulated genes by combinatorial microarray. (C) Out of 43 genes, 26 (60%) had unknown function or were novel and 14 genes (33%) had known functions, of which 11 were previously identified FGF targets. Three (7%) were retrotransposon or reverse transcriptase related genes. (D) Nine transcription factors (21%), 4 transmembrane proteins (9%), 3 cytoskeleton interaction (7%), and 3 transport-related genes (7%), and 2 growth factors (5%); however, there were no clues to the functions of 17 genes (40%).

Table 1.  The results of sequence based BLAST search and spatiotemporal expression patterns of identified genes during gastrulation stage are described. The remarkable spatiotemporal expression patterns that we described were detected by whole-mount in situ hybridization. No putative protein means no homologue by BLAST searches. Access our database website, XDB, for more information
Clone IDBLAST best hitExpression pattern
Early gastrulaMid-gastrula
XL001o05hox7.1 [xl]UbiquitousDorsal/ectoderm
XL003a10syrB gene [ps]Dorsal/ectodermUbiquitous
XL003f08NS5 [Hepatitis C virus]UbiquitousUbiquitous
XL005c15ARL5 [mm]DorsalDorsal
XL006h11gene 33/mig6 [mm]DorsalBlastopore; weak
XL007m13po [xl]UbiquitousBlastopore
XL010k19CDK5 activator 2 [hs]EctodermEctoderm
XL010o02ESR5 [xl]DorsalBlastopore
XL011a24no putative proteinDorsalDorsal
XL012d21PCR4 [hs]BlastoporeDorsal
XL012i17fat 1 cadherin [mm]UbiquitousUbiquitous
XL013k1838 k protein [xl]DorsalBlastopore
XL014n21fkh5 [xl]DorsalDorsal
XL015a12growth factor Livertine [xl]UbiquitousUbiquitous
XL015d13no putative proteinBlastoporeBlastopore/ectoderm
XL016e23derriere [xl]BlastoporeBlastopore
XL017b07LIG1 precursor [hs]DorsalDorsal
XL017o04btgl [xl]BlastoporeBlastopore
XLO18b03CL100 [xl]Dorsal/ectodermUbiquitous
XL019m15p75-like transmembrane protein [xl]BlastoporeBlastopore
XL020n17transposon-like element [xl]BlastoporeBlastopore
XL020o07no putative proteinBlastoporeBlastopore/ectoderm
XL021f10no putative proteinUbiquitousUbiquitous
XL022d02no putative proteinUbiquitousDorsal/ectoderm
XL022e09no putative proteinUbiquitousUbiquitous
XL022h10cad3 [xl]DorsalBlastopore
XL023m11reverse transcriptase [xl]UbiquitousUbiquitous
XL023o21ANKRD5 [hs]UbiquitousUbiquitous
XL025d04Golgi UDP-GicNAc transporter [hs]Dorsal/ectodermBlastopore
XL025f20no putative proteinUbiquitousBlastopore
XL025g0238 k protein [xl]UbiquitousBlastopore
XL025i04monooxygenase C-terminal interactor [hs]UbiquitousBlastopore/ectoderm
XL025n01no putative proteinDorsalBlastopore
XL027l07KLAA1022 protein [mm]Dorsal/ectodermEctodenn
XL032e19no putative proteinUbiquitousBlastopore/ectoderm
XL034k09no putative proteinUbiquitousUbiquitous
XL036h02NADH dehydrogenase subunit 5 [bs]UbiquitousBlastopore/ectoderm
XL039m17GATA4 [xl]Dorsal/ecdodermDorsal/ectoderm
XL042i04N-acetylglucosamine-specific receptor 1 precursor [hs]UbiquitousBlastopore
XL042p23lunatic fringe gene [xl]UbiquitousBlastopore/ectoderm
XL044i01NADH dehydrogenase [bt]UbiquitousUbiquitous
XL044k16CCCH Zn finger protein [xl]UbiquitousBlastopore/ectoderm
XL048p23reverse transcriptase [xl]UbiquitousUbiquitous

Using BLAST searches and XDB of the NIBB/NIG Xenopus EST database (http://xenopus.nibb.ac.jp), we classified the identified genes into several categories based on sequence information. Of the 43 genes identified by microarray analysis, 26 (60%) were novel genes. Fourteen were genes whose function is known, including 11 that had been previously identified as FGF target genes (Fig. 1C).

The details of the functional classification of these genes are given in Fig. 1D. The genes obtained by our screen included transcription factors, transport-related genes, transmembrane proteins, scaffolding proteins, growth factors, and cell-cycle regulating protein, among others.

Whole-mount in situ hybridization analysis of identified genes

The spatiotemporal expression patterns of the identified gene were investigated by careful comparison with FGFR1, because FGFR1-specific inhibitor was used in this screen. Since the candidate genes we chose were specifically expressed at higher levels in the blastopore region in the gastrula, we focused on expression during gastrula. The expression patterns of the novel genes are described briefly and shown in Fig. 2. The expression domains of all the identified genes are summarized in Table 1.

Figure 2.

Whole-mount in situ hybridization of several interesting genes. Among identified genes, we picked up six novel genes that are highly expressed around blastopore, which is showed similar expression pattern with FGFR1 expression in gastrula. From left to right, expression patterns of early gastrula, late gastrula, neurula, and tailbud stage are shown. All gastrula embryos are vegetal but late gastrula of ARL5. Late gastrula of ARL5 is lateral view and dorsal is right. All neurula embryos are dorsal view and anterior is left. For tailbud embryos, XL019m15 and XL027l07 are dorsal view but the others are lateral view. Anterior is left.

The expression of gene XL005c15, similar to ADP ribosylation factor-like protein 5 (ARL5) (Lin et al. 2002) was highly restricted to the dorsal blastopore region, and thus showed a steep gradient from the dorsal to ventral region. Gene XL011a24, of unknown function, was expressed around the blastopore, but its expression was remarkably confined to the dorsal blastopore as development proceeded. These two genes showed very similar expression to FGF8, which is expressed as a ring around the blastopore with increasing dorsal expression in the mid-to-late gastrula. XL006h11, like gene33/mig6 (Makkinje et al. 2000; Hackel et al. 2001), was expressed dorsally in the early gastrula and then in a ring-like pattern around the blastopore in the mid-gastrula. XL012d21, which probably encodes GPCR4 (Xu et al. 2000), was expressed weakly around the blastopore region and endoderm in a salt-and-pepper pattern. XL027l07, which is weakly homologous with mMGC, exhibited ubiquitous expression. Interestingly, XL019m15, encodes p75-like transmembrane protein similar to the previously identified gene fullback (GenBank accession number AF131890); this protein was named p75 neurotrophin receptor homologue (NRH) (Kanning et al. 2003) and had an expression pattern very similar to that of FGFR1. During gastrula, NRH is strongly expressed in a ring around the blastopore. These very similar expression patterns, known as ‘synexpression’ (Bottcher et al. 2004), may suggest that these genes carry out closely connected functions during gastrulation.

Confirmation of FGF target genes by semiquantitative RT-PCR analysis and WISH

To confirm the reliability and reproducibility of the microarray analysis in the context of molecular embryology, semiquantitative RT-PCR was performed for the identified genes (Fig. 3). FGF target genes Xbtg1 (Saka et al. 2000) and derriere (White et al. 2002), which were found to be down-regulated by microarray analysis, showed FGF-dependent up-regulation and XFD/SU5402-dependent down-regulation. Up-regulated genes in the microarray analysis Xhox7.1 and XGATA4 showed FGF-dependent down-regulation and XFD/SU5402-dependent up-regulation, thus confirming the effectiveness of the microarray approach. Other familiar FGF targets that were excluded from the microarray Xbra, Xwnt11, eFGF, Xspry2, and Xmc displayed FGF- and SU5402-dependent responses (data not shown), demonstrating that the experiments were designed properly.

Figure 3.

Semiquantitative RT-PCR confirmed FGF-dependent regulation of the identified genes. The amplification of cDNAs obtained from eFGF (0.01 ng), HAVØ, or XFD (0.5 ng)-injected Keller explants, and SU5402, or DMSO-exposed Keller explants was performed. Each ratio represents the relative intensity of the respective explants divided by the intensity of the uninjected explants. These results are representative of three experiments.

We also carried out RT-PCR analysis of the novel genes. The up-regulation of GPCR4, gene33/mig6, ARL5, and NRH in response to eFGF and their down-regulation in response to XFD and SU5402 suggest that FGF is necessary and sufficient for the activation of these genes.

We further confirmed that the genes we identified were FGF targets via WISH (Fig. S2A). The expression of FGF target genes fkh5, ESR5, and derriere diminished in XFD-injected embryos. ARL5, mig6, and NRH also displayed XFD-dependent loss of expression. GPCR4, XL011a24, and other identified genes exhibited similar regulation profiles.

Through the RT-PCR and WISH analyses, we confirmed FGF-dependent regulation of identified genes. That is, the genes that were down-regulated in the microarray analysis showed up-regulation in response to eFGF and down-regulation in response to XFD or SU5402. Moreover, Xhox7.1 and XGATA4, which exhibited up-regulation in the microarray analysis, were down-regulated in response to eFGF and up-regulated is response to XFD or SU5402. These two unequivocal results demonstrate the reliability of the microarray analysis.

How are the novel genes relevant to FGF signalling pathways?

Results obtained in Xenopus have shown that FGFs are involved in early mesodermal inductive signals and in the maintenance of Xbra expression. More recently, Xspry2 was shown to prevent convergent extension movements during gastrulation but not to prevent mesodermal induction and patterning (Nutt et al. 2001). To understand how our candidates are related to known FGF components, we carried out RT-PCR and WISH of embryos that received injection of RNAs of the novel genes as well as of embryos that were injected with the RNAs of known FGF components (Fig. 4).

Figure 4.

How novel genes are related to other FGF components. (A) RT-PCR analysis of animal cap explants was performed. cDNAs from 10 animal caps that had received injections of Xbra, Xspry2 (1 ng) co-injected with eFGF (0.01 ng), and Xmc (1 ng) were amplified by primers to ARL5, GPCR4, mig6, and NRH. (B) Using a MEK inhibitor, U0126, we further examined and confirmed that mig6 is regulated in a Ras-MAPK dependent manner, whereas the other genes are not. RT-PCR analysis of Keller explants was carried out in the presence of U0126 or DMSO.

Through RT-PCR analysis, we confirmed that ARL5, mig6, GPCR4, and NRH were up-regulated by ectopic expression of eFGF. The expression level of ARL5, GPCR4, and NRH remained steady in response to Xbra mRNA and no changes were observed. Interestingly, mig6 was slightly increased by Xbra injection, which led us to investigate it in more detail as a canonical FGF target gene. Xspry2 and Xmc, which are believed to be non-canonical FGF targets, did not cause any significant changes when they were microinjected. Although up-regulated transcriptional levels of ARL5, GPCR4, and mig6 by eFGF appear to be suppressed in the presence of Xspry2 somewhat, the inhibitory effect of Xspry2 was not concentration-dependant. For this reason, we concluded that newly identified genes are not necessarily related to the Xspry2-dependent pathway (Fig. 4A). Possibly, this may be repressed directly or indirectly by Xspry2 that could remarkably inhibit Ca2+ mobilization induced by the activation of FGFR although its molecular mechanism still remains elusive (Nutt et al. 2001). Marker gene expression in ARL5, GPCR4, mig6 and NRH-injected embryos was compared with the expression in control embryos. We found that the spatiotemporal patterns of Xbra and Xwnt11 did not change. Likewise, Xgsc, Xmc and Xspry2 showed no change in expression (Fig. S2B).

We also used a specific MEK inhibitor, U0126 (Kim & Nishida 2001) to investigate whether the regulation of the novel FGF target genes depended on Ras-MAPK signalling. Although other genes were stable when Xbra expression was completely suppressed, mig6 was thoroughly down-regulated, suggesting that mig6 is specifically regulated by Ras-MAPK signalling and by Xbra (Fig. 4B).

Xenopus mig6 (Xmig6) is regulated brachyury-dependent manner

The profile of Xmig6 transcriptional regulation described above, namely, activation by FGFs or Xbra and inhibition by XFD, SU5402, or U0126, suggested that Xmig6 might be involved in mesodermal induction during early embryogenesis. To determine this we again used U0126 and MAPK phosphatase (XCL100 or MKP-1) to inhibit Ras-MAPK signalling pathway (Keyse & Emslie 1992; Gotoh & Nishida 1996; Favata et al. 1998; LaBonne et al. 1995). The ectopic induction of Xmig6 by eFGF was completely down-regulated by XCL100 and severely suppressed by Xbra-EnR (Conlon et al. 1996) (Fig. 5A). In the presence of U0126, ectopically induced Xmig6 by eFGF was completely suppressed but this induction was unaffected by Xbra co-injection. The up-regulation of Xmig6 by Xbra was not affected by U0126 (Fig. 5B) or XCL100 (data not shown). These results suggest that Xmig6 is regulated by FGF-MAPK-bra pathway.

Figure 5.

Xmig6 is regulated in an eFGF, MAPK, and Xbra-dependent fashion. (A) To look for the relationship among Xmig6, eFGF (0.01 ng), and Xbra (1 ng), we performed RT-PCR analysis MPK-1 (0.5 ng) and Xbra-EnR (Xbra repressor, 1 ng) with animal cap assay. Animal cells that received injection were proceeded into RT-PCR (B) In the same condition as (A), we further confirmed that the relationship between Xmig6 and FGF-MAPK pathway. In the presence of U0126 (50 µm) or DMSO as a control, animal cells that received injection were proceed to RT-PCR analysis.

Xmig6 is required for muscle differentiation

Xmig6 encodes a polypeptide of 404 amino acids that shares a ~56% sequence identity with mouse and human mig6, which consist of 461 and 462 amino acids, respectively (Fig. S3). Database analysis revealed that mig6 contains a potential CRIB domain, SH3 domain, 14-3-3 domain, and ACK-related non-receptor tyrosine kinase domain (AH domain). These proteins share well-conserved serine/proline-rich carboxy-terminal domains (Makkinje et al. 2000; Hackel et al. 2001).

Because of the way Xmig6 was regulated, we scrutinized the relationship between Xmig6 and the downstream signalling of FGF in mesodermal induction. To confirm that the Xmig6 MO (morpholino antisense-oligonucleolide)-specific transcription inhibition was specific, we used Western blotting (Fig. 6A). Xmig6 MO was specifically inhibiting transcription of 5′ UTR including the Xmig6 venus-tagged (Nagai et al. 2002) construct since the Xmig6 MO recognizes the 5′ UTR sequence. But this inhibition was rescued using a full length Xmig6 construct. We examined the role of mig6 in regulating other mesodermal markers using Xmig6 MO to suppress mig6 function. The result of WISH revealed that the knockdown of Xmig6 by the MO completely abolished Xmyf5 expression (Fig. 6B). The expression of Xbra and Xnot was intact, as revealed by WISH (Fig. 6B) and RT-PCR analyses (Fig. 6C). The down-regulation of Xmyf5 by the Xmig6 MO was also efficiently reversed by co-injecting mig6 mRNA construct (Fig. 6B). However, another muscle marker XmyoD was only partially down-regulated. Together, these data prove the specificity of the Xmig6 MO. About 30% (6 of 20) embryos that were mig6 depleted lacked well-differentiated muscle, as assessed by immunostaining with the muscle-specific antibody, 12/101. No change was found in the notochord (Fig. 6D). Although Xmig6 MO showed Xmyf5-specific down-regulation, Xmig6 itself did not induce Xmyf5 in the whole embryos when over-expressed (data not shown). In higher doses of Xmig6 (> 500 pg), the embryos showed abnormal cell division (data not shown). The results suggest that Xmig6 is a potential downstream gene of eFGF and Xbra, and has specific roles in muscle differentiation that are mediated in an Xmyf5-dependent fashion.

Figure 6.

Xmig6 depleted embryos showed muscle differentiation defect in a Xmyf5-dependent manner. (A) Transcription inhibition by Xmig6 MO was confirmed by Western blotting. Protein expression was detected by a rabbit polyclonal anti-GFP antibody and anti-rabbit IgG-HRP antibody. (B) X-Gal-stained embryos that received injection of Xmig6 MO (16.8 ng) or Xmig6 RNA (0.15–0.25 ng) to the one dorsal cell of 4-cell-stage were used for WISH with a DIG-labelled Xmyf5 anti-sense RNA probe. Vegetal view. Xmig6 MO introduced embryos caused the completely depleted Xmyf5 expression, which was rescued by mig6 RNA co-injection. (C) In eFGF induced animal caps, Xmig6 MO (8.4–16.8 ng) specifically suppressed the expression of Xmyf5 and slightly suppressed XmyoD. This suppression was rescued by Xmig6 RNA (0.2–0.25 ng). No changes were observed in Xbra and Xnot expression. (D) In later stages, embryos that received Xmig6 MO injection were stained by 12/101 antibody and showed muscle differentiation defect whereas control MO injected embryos showed well-developed somites. We counted the embryos that proceeded 12/101 antibody staining and six of 20embryos were failed to be stained by the antibody. Xmig6MO introduced embryos were no effect on notochord detected by MZ15 antibody.

Xenopus G-protein coupled receptor 4 (XGPCR4) might involve in gastrulation cell movements

We examined the role in early embryogenesis of novel Xenopus GPCR4 (XGPCR4), which most likely to encode a G-protein coupled receptor. As little is known about the molecular nature of GPCR4, we first searched the database for homologous sequences. GPCR4 shares homology with several other GPCRs, including ORG1, G2A, T-cell death associated gene 8, and 12 A. These have ∼51% sequence identity to each other and approximately 30% sequence identity with the platelet-activating factor (PAF) receptor. XGPCR4, whose cDNA sequence contained an open reading frame of 364 amino acids, shared 36% and 55% sequence identity with human and porcine GPCR4 and 35% and 55% identity with murine and rat GPCR4, respectively (Zhu et al. 2001; Xu 2002). XGPCR4 contains 7 passed transmembrane protein from 33 to 284.

To expand our understanding of GPCR4 in early embryogenesis we carried out gain-of-function and loss-of-function analyses. XGPCR4 mRNA-injected embryos showed dose-dependently incomplete closure of the blastopore and spina bifida (Fig. 7A). We examined the XGPCR4 MO-mediated transcription inhibition (Fig. 7B) and phenotypes using mut-XGPCR4 (rescue construct) mRNA and one including the 5′ UTR. To obtain molecular clues regarding the role of GPCR4, WISH and RT-PCR analyses were performed. Neither XGPCR mRNA-injected embryos (Fig. S2B) nor XGPCR4 MO-injected embryos showed perturbed expression of Xbra, Xwnt11, Xgsc (Fig. 7C). XGPCR4 MO-injected embryos displayed gastrulation defects and a dorsal open phenotype or short trunk (Fig. 7D). Using a rescue construct we tried to rescue the spina bifida phenotype caused by the MO-induced XGPCR4 knockdown. We achieved a partial rescue: the occurrence of short trunk and spina bifida decreased from 53% to 24%, and the percentage of normal embryos doubled (Fig. 7D).

Figure 7.

XGPCR4 is potentially a non-canonical FGF target gene. We investigated the role of XGPCR4 in gastrulation by injecting its mRNA or XGPCR4 MO into the two dorsal cells of 4-cell-stage embryos. (A) The gain-of-function effect showed dose-dependent A-P axis truncation or spina bifida but embryos given mRNA injection (n = 60) did not show any changes in the differentiation of the notochord and somites. (B) The specificity of XGPCR4 MO was confirmed by Western blotting, using venus-tagged constructs in animal caps. (C) In XGPCR4-MO injected embryos, neither the expression of mesodermal markers nor the formation of the organizer was affected compared with control MO-injected embryos. (D) Embryos given an injection of XGPCR4 MO (16.8 ng) (n = 45) showed incomplete closure of the dorsal blastopore and the spina bifida phenotype. However, no changes in the formation of the notochord and somites were observed. Using mut-GPCR4 (500 pg) construct, we partially rescued the spina bifida phenotype caused by the GPCR4 MO (n = 48). Mut-GPCR4 RNA itself (n = 45) exerted the same effects on embryos as intact GPCR4 RNA.

The results from immunostaining of the notochord and somites showed no changes in tissue differentiation (Fig. 7A,D). Supporting this, we did not identify any transcriptional changes as a result of changing the GPCR4 expression by microinjection, which suggests that GPCR4 regulates gastrulation cell movements rather than mesodermal induction and patterning.

Although the physiological roles of GPCR4 remain unclear, our data may suggest that GPCR4 is a promising candidate for the component that acts exclusively in the FGF pathway to regulate gastrulation. In other species, several studies of GPCR4 homologues have reported a receptor-ligand relationship and receptor-mediated signal transduction (reviewed in Xu 2002). To gain more insight into the role of GPCR4, we are currently investigating how it is involved in gastrulation cell movement.

Discussions

Traditional approaches to molecular embryology have generally been to performed one experiment on one gene and therefore have provided only an extremely limited understanding of the overall picture of embryonic development. Microarray technology is well suited for studying differential gene expression patterns between two or more cell populations or developmental stages, and can be performed on much larger scales (Altmann et al. 2001; Hoffman et al. 2002; Buttitta et al. 2003). This technique allows us to do functional genomic research as well as spatiotemporal analyses of embryogenesis. The power of the microarray-based approach is the capability to identify both up-regulated and down-regulated genes in the same run. Despite the potential physiological importance of understanding down-regulation, the majority of studies focused on up-regulation.

Much effort has been spent trying to elucidate the role of FGF signalling in embryonic development (Hoffman et al. 2002; Pownall et al. 2003; Tateossian et al. 2004), and a number of genes that regulate or modify FGF signalling have been identified (Gotoh & Nishida 1996; Walshe et al. 2002; Musci et al. 1990; Amaya et al. 1991, 1993; Smith et al. 1991; Nutt et al. 2001; Frazzetto et al. 2002; Malcolm et al. 2002; Monsoro-Burq et al. 2003; Saka et al. 2000; White et al. 2002). Although all these results have contributed to our understanding of the molecular mechanisms of FGF signalling, many experiments, particularly those using Xenopus, have been performed on whole embryos using gain-of-function assay or induction. These approaches, however, can be difficult to interpret, because it is hard to distinguish direct action on the target tissues from indirect actions on other tissues. In addition, a critical technical limitation is the incomplete diffusion of the injected RNA. This may introduce mosaic expression of the injected RNA and dispersion in the experiment. Drug treatment has some advantages over conventional approaches. For instance, a larger number of developing embryos can be treated. It is easy to control the exposure time to the drug, permitting transient or semipermanent exposure, which is difficult to do with microinjection. By treating embryos at various developmental stages, the researcher can determine the critical period for FGF signalling in embryos by restricting the hours of treatment and anatomical classification. The use of SU5402 allowed us to study the role of FGF signalling at a desired point, e.g. the gastrula in this report, and to focus on signal molecules that are specifically expressed in the developing dorsal marginal zone.

In this study we successfully introduced a combinatorial screening strategy using Keller explants, SU5402, and subtraction by microarray analysis, and we achieved our primary goal—to identify FGF target genes that are expressed during gastrulation. The genes we identified included a number of well-known FGF target genes (Fig. 1 and Table 1). The finding of known FGF targets implies high-fidelity of the microarray analysis. Through the analysis of two interesting genes, we expanded our understanding of when and how two different FGF pathways bifurcate during embryogenesis.

The inhibitory effects of SU5402 on gene expression and morphology were very similar to those of XFD (Amaya et al. 1993) but showed more diverse phenotypes. In stage-subdividing experiments, treatment starting at the late gastrula caused severe neural tube closure defect, but the effect was minimal in embryos treated from the mid-neurula. By assaying the elongation of Keller explants, we demonstrated that the inhibitory effect of SU5402 was most pronounced when applied to stage 10.5–11.5 embryos. Our data imply that there must exist a critical and/or sensitive period during which FGF signalling exerts its influence on convergent extension movements. In addition, the divergent phenotypes may suggest a multiple function or recursive role of FGF in embryogenesis. We found this discovery intriguing, and we further accumulated convincing clues through stage-subdividing microarray analysis (unpublished data) and the functional analysis of two interesting genes, mig6 and GPCR4, in developing embryos. Interestingly, microarray analyses from stage 10.5–11.5 and 10.5–15 identified many overlapping genes but they did not necessarily exclude genes related to mesodermal induction (unpublished data); mig6 was one such gene.

Mig6 may act as an adaptor molecule that interacts with other proteins through CRIB, 14-3-3, and/or SH3 domains (Makkinje et al. 2000; Hackel et al. 2001). Mig6 is reported to be a binding partner of the EGFR and to antagonize EGF signalling but not FGF (Hackel et al. 2001). Here, we demonstrated that Xmig6 expression was induced by FGF and Xbra and depleted by FGF inhibitor, MAKK inhibitor, and Xbra repressor. Xmig6 is regulated by FGF-MAPK-Xbra pathway and regulates muscle differentiation Xmyf5-depentently. Thus, we speculate that the interactions mediated by the domains named above modulate FGF signalling and regulate tissue specification by relaying FGF, Xbra, and Xmyf5 signal.

GPCR4 is most highly expressed in the placenta, lung, liver, skeletal muscle, ovary, lymph node and heart. It shares a significant identity with the ORG1 subfamily, which are specific receptors for sphingosylphosphorylcholine (SPC) and lysophosphatidylcholine (LPC). Like ORG1, mGPCR4 responds to both SPC and LPC, although less to the latter (Xu et al. 2000; Xu 2002). SPC is a bioactive lipid that regulates diverse cellular functions, including proliferation, growth inhibition, muscle contraction, and wound healing. SPC stimulates signalling pathways that induce protein tyrosine phosphorylation, activate MAP kinase and/or protein kinase C (PKC), modify ion channel activity, and increase intracellular calcium concentrations. GPCR4 mediates not only an SPC-induced PMA (phorbol myristate acetate)-sensitive intracellular calcium increase, but also PTX (pertussis toxin)-sensitive ERK activation, suggesting that PKC activity and the Gi/o type of G protein are involved. Ligands of GPCR4 induce cell shape changes, suggesting that SPC and LPC might affect the cellular cytoskeleton. In fact, Rho-dependent activities induced by SPC through its receptors include actin rearrangement and cell migration (Zhu et al. 2001; Xu 2002). Further investigation will be required to understand the role and signalling mechanism of GPCR4 in embryogenesis.

FGF signalling controls a diverse array of developmental events, but the number of signal bifurcations and how these branches collaborate is still unclear. MAPK-dependent pathway is a well-characterized pathway and is essential for mesodermal induction and patterning. Among the FGF targets identified in this study, only mig6 showed MAPK-dependent transcription regulation. GPCR4 was insensitive to any known FGF components, but showed perturbation of cell movement when it was depleted. It might regulate PKC or a calcium-dependent pathway or some other novel pathway.

Although the present study has demonstrated that microarray analysis combined with specific inhibitors is very efficient for uncovering novel FGF target genes and has implied potential for dissecting other signalling pathways, the method could be improved in a number of ways. One is to apply cyclohexamide to growth factor-stimulated animal caps to identify the earliest responsive target genes. We believe that such modifications of the analysis should lead to the identification of more novel genes and help us solve some important questions.

Experimental procedures

Embryo manipulations and microinjection

Xenopus eggs were collected as described (Yamamoto et al. 2001) and the embryos were staged according to Nieuwkoop and Faber (Nieuwkoop & Faber 1967). Keller explants were excised at stage 10.5 and placed in 1× Steinberg's solution supplemented with 0.1% bovine serum albumin (BSA) and then were incubated with 50 µm SU5402 (Calbiochem, CA, USA) or DMSO until the stages of interest were reached. Plasmids to be used for microinjections, were linearized with NotI. Capped mRNAs were synthesized with an mMESSAGE mMACHINE kit (Ambion, Austin, TX, USA) and purified on a NICK column (Pharmacia, Uppsala, Sweden). All plasmids were transcribed with SP6 polymerase.

Western blotting

To investigate the effects of SU5402 on Xenopus embryos, dominant-negative form of the FGFR (XFD) RNAs and eFGF RNAs were injected into the two dorsal cells of 4-cell-stage embryos. Ten Keller explants, embryonic explants consisting of presumptive dorsal mesoderm and neuroectoderm, were dissected at stage 10.5 and exposed to SU5402. These explants were incubated until stage 11.5. Explants that either received injection or were exposed to SU5402 were homogenized in 100 µL lysis buffer (50 mm Tris-Cl [pH 7.4], 150 mm NaCl, 50 mm NaF, 5 mm EDTA, 0.5% NP-40, 1 mm Na3VO4), with the exception that 1 mm p-APMSF, 10 µg/mL aprotinin, 20 µg/mL leupeptin, 10 µg/mL pepstatin, and 1 mm DTT were added to lysis buffer as indicated. Uninjected explants were used as an internal loading control. Boiled lysates (10 µL) were separated by 10% SDS-PAGE, transferred to PDVF membranes (BIO-RAD, CA, USA), and then immunoblotted as described (Nutt et al. 2001).

Microarray

NIBB cDNA microarray ver.1 was prepared using the tentatively selected non-redundant 4600 cDNA set from the NIBB Mochii normalized gastrula and tailbud cDNA libraries for the NIBB/NIG Xenopus laevis EST project (A. Kitayama, unpublished). The list of the cDNA genes that were spotted is available on our web site, XDB (http://xenopus.nibb.ac.jp).

Total RNAs were extracted by TRIzol™ reagent (Life Technologies, Carlsbad, CA, USA) from 10 Keller explants that were or were not treated with SU5402. Using these total RNAs (∼1 µg), T7 RNA polymerase-based mRNA amplification was carried out with a RiboAmp RNA Amplification Kit (ARCTURUS). Five micrograms of anti-sense RNA from each sample was labelled with Cy3- or Cy5-dCTP (Amersham, Stockholm, Sweden) using random primers and SuperScriptII reverse transcriptase (Life Technologies).

Hybridization and washing were carried out according to the slide manufacture's protocol (Amersham). Scanning was performed on a GenePix 4000B (Axon, Union City, CA) to generate two 16-bit TIFF images corresponding to the Cy3 and Cy5 channels. After image analysis using GenePix Pro 3.0 or 4.1 (Axon), the data were analysed with an Excel-based analysis tool MarC-V (Schageman et al. 2002).

RNA isolation and RT-PCR assays

Total RNA was isolated from 10 explants using Trizol™ reagent, according to the manufacturer's instructions. Isolated RNAs from the explants were used for cDNA synthesis in the presence of 2 mm dNTP, 0.1 m DTT, 5X first strand buffer, RNase inhibitor, and Reverse Transcriptase (M-MLV, Life Technologies) in a 20 µL reaction volume for 1 h at 37 °C. A-twentieth volume of cDNA was used for PCR amplification with Ampli Taq polymerase (Applied Biosystems, Branchburg, NJ). The primer pairs and number of cycles used in this study were as follows: Xbtg1 (20) 5′GTTTCATCACGAAGTTCCTC3′ (F) and 5′-CTGGCTGCTGAGTCCAATAC3′ (R), XGATA4 (20) 5′AGCAGAGAGGTCTCCTACAGT3′ (F) and 5′TCTATGTTTGGGTGCCTTGA3′ (R), ARL5 (20) 5′TCATCAGTGTTGGACACTTA3′ (F) and 5′AGCGATCATGAGCAACTCTGT3′ (R), GPCR4 (18) 5′TCTGATGTGCTGTGAAGTCT3′ (F) and 5′ACTATAAAGACACTGCTGTA3′ (R), NRH (18) 5′ACGTGGTGAGAACAGCACTA3′ (F) and 5′TGTATCAAGGAAGACGCCACT3′ (R), Xspry2 (18) 5′TCTGCTCCACGACAGACTACCCA3′ (F) and 5′TCAGGTTTAGGCTGTACTCGAAT3′ (R), Xmc (20) 5′ATTCAGGTGAGGTACAGCTC3′ (F) and 5′AAACATCCCTTGTCCTTCGCTG3′ (R). PCR products were separated on 6% non-denaturing polyacrylamide gels and visualized by CYBR GreenI staining. The gel images were scanned and quantified by FLA2000 bioimager and built-in software. (Fuji Film, Japan)

Whole-mount in situ hybridization and immunostaining

Embryos to be used for in situ hybridization were incubated until the proper stages, fixed in MEMFA (0.1 m Mops [pH 7.4], 2 mm EGTA, 1 mm MgSO4, 3.7% formaldehyde) solution and stored in methanol at −20°C until use. Embryos that received an injection of mRNA and β-galactosidase mRNA were fixed primarily in MEMFA solution, rinsed with PBS, and stained with 6-chloro-3-indolyl-β-D-galactoside, as a lineage tracer. Following the staining procedure, the embryos were fixed again in MEMFA. Synthesis of DIG-labelled anti-sense RNA probes and whole-mount in situ hybridization (WISH) were performed as previously described. (Harland 1991) Probes were purified with a Sephadex G 50 spin column (Amersham Pharmacia). Hybridized RNAs were detected with alkaline-phosphatase-conjugated anti-DIG-antibody (Roche, Mannheim, Germany) and developed using BM purple (Roche). Stained embryos were bleached in 10% hydrogen peroxide supplemented methanol.

Immunostaining was perforsmed with minor modifications as described (Klymkowsky & Hanken 1991), using monoclonal antibodies MZ15 and 12/101 (Hybridoma bank) to stain the notochord and somites, respectively. Embryos and explants were fixed in 4% MEMFA at stage of 28, then incubated first in diluted primary antibody, and then in secondary antibody. Labeling was visualized with H2O2. For notochord observation, the tissue was cleared using benzyl benzoate solution.

Plasmid construction

Using PCR we constructed plasmids for microinjection and subcellular localization studies. Full-length cDNAs were subcloned in-frame into ClaI and XhoI sites for Xmig6 (GenBank accession no. AY553188) and EcoRI and XhoI sites for XGPCR4 (GenBank accession no. AY553188) of pCS2+ and pCS2-venus (Nagai et al. 2002), respectively. The primer pairs used in the PCR were: 5′-CCATCGATATGACAACTGCTGGGATTGCA-3′, 5′-CCGCTCGAGTCAGATCTGCCCCACTAAGCA-3′ for Xmig6; and 5′-TCGGAATTCATGTGTAACCAGAGCGTGTCG-3′, 5′-CCGCTCGAGCTACAACCTAGACTGTATTT3′ for XGPCR4, respectively. The primer used for 5 nucleotides displacement in XGPCR4 that becomes XGPCR4 MO rescue construct is indicated italic and bold font (mut-XGPCR4) was 5′-CGGAATTCATGTGCAATCAATCGTGTCGT-3′. The restriction sites are underlined and a stop codon was deleted in the downstream primers for the pCS2-venus construct.

MO sequence and detection of transcription inhibition

Two kinds morpholino oligonucleotides (MOs) were used: Xmig6 MO, 5′GCGGCTGTCATTCCTTCAACTATGG3′; and XGPCR4 MO, 5′CACGACACGCTCTGGTTACACATTC3′ (GENE TOOLS, LLC, OR). Transcription inhibition by the MOs was confirmed through Western blotting. Venus-tagged constructs were used and detected with a rabbit anti-GFP antibody (MBL) followed by and HRP-conjugated anti-rabbit anti-IgG (Amersham).

Acknowledgements

We thank Drs Fiona Watt for the MZ15 antibody, Takeharu Nagai for pCS2-venus plasmid, Elizabeth A. Emslie and Yukiko Gotoh for XCL100 (MKP-1), and Masazumi Tada for Xbra-EnR constructs. This work was supported by a grant from the Japan Society for the Promotion of Science (Research for the Future) and Ministry of Education, Culture, Sports, Science and Technology (to N.U.). Hyeyoung A. Chung was supported by the Ministry of Education, Culture, Sports, Science and Technology, Japan.

Ancillary