Ascidians have been thought to be closely related to vertebrates, and the ascidian larva is often regarded as an organism close to the ancestral form of chordates (Satoh, 1994; Satoh and Jeffery, 1995). Therefore, the ascidian larva serves as an interesting and important model system for understanding a basic chordate body plan and its evolution. The body plan of the ascidian larva is simple but reminiscent of that of vertebrates (references therein, Satoh, 1994). The ascidian larva has a tadpole-shaped body, consisting of a trunk (corresponding to the head of a frog tadpole) and a tail. A hollow neural tube extends dorsally most of the length of the body. In the tail, the notochord runs through the central midline and muscle lies on both lateral sides of the notochord. In the trunk, endoderm that later develops into adult endodermal tissues, lies at the ventral side of the neural tube. The neural tube is divided into three domains (from anterior to posterior): the sensory vesicle, visceral ganglion, and caudal neural tube with morphologically distinct features for each domain (Nicol and Meinertzhagen, 1991; Satoh, 1994).
Recently, it has been found that homeobox genes Hroth (Wada et al., 1996), HrPax2/5/8 (Wada et al., 1998), HrHox-1 (Katsuyama et al., 1995), and Hrcad (Katsuyama et al., 1999), that are the ascidian homologues of vertebrate otx, Pax2/5/8, Hox-1, and cdx, respectively, are expressed in the central nervous system (CNS) with distinct expression domains along the anterior–posterior axis (Katsuyama et al., 1996; Wada et al., 1998). Hroth, HrHox-1, and Hrcad are expressed in the sensory vesicle, the visceral ganglion, and the caudal neural tube, respectively. HrPax2/5/8 is expressed at the border between the sensory vesicle and visceral ganglion. This situation is very similar to that in the developing CNS of vertebrates, in which otx, Pax, Hoxb-1, and cdx genes are expressed in the fore- and midbrain, the midbrain, the rhombomere 4, and the spinal cord, respectively (Nornes et al., 1990; Murphy and Hill, 1991; Simeone et al., 1992; Meyer and Gruss, 1993; Rowitch and McMahon, 1995; Joyner, 1996). These observations led to a suggestion that a mechanism for the patterning of the CNS along the anterior–posterior axis may be conserved between ascidians and vertebrates.
Among the genes described above, otx is of particular interest. Otx is a phylogenetically well-conserved homeobox gene. Otx genes have been isolated from wide range of animal species, including cnidaria, Drosophila, and vertebrates (Finkelstein and Perrimon, 1991; Simeone et al., 1992, 1993; Ang et al., 1994; Li et al., 1994; Bally-Cuif et al., 1995; Mercier et al., 1995; Pannese et al., 1995; Smith et al., 1999). They are expressed in the anterior part of the embryos and, therefore, are expected to have a role in formation and patterning of the anterior part. In vertebrates, two Otx genes, Otx1 and Otx2, have been identified (Simeone et al., 1992). Otx2 is expressed in the entire epiblast and visceral endoderm before gastrulation (Simeone et al., 1993). In later development, both Otx2 and Otx1 are expressed in the developing forebrain and midbrain, whereas the Otx1 expression domain is included within the Otx2 expression domain (Simeone et al., 1993). Mice mutated for Otx1 exhibited a subtle phenotype with the brain, sense organ, and pituitary (Acampora et al., 1996, 1998). By contrast, analysis of null mutant mice of Otx2 has shown that Otx2 functions as a head organizer component at the primitive streak stage and plays a critical role in the formation and patterning of anterior CNS at the neurula–pharyngula stage, (Acampora et al., 1995; Matsuo et al., 1995; Ang et al., 1996). However, transcription regulatory mechanisms for otx genes in the CNS are largely unknown, although cis-regulatory elements for the expression of Otx2 in the cephalic mesenchyme in mouse embryos have been reported (Kimura et al., 1997).
In the ascidian, Halocynthia roretzi, Hroth, the homologue of otx genes, is expressed in the precursors for the anterior CNS and epidermis, endoderm, and mesenchyme before and during gastrulation (Wada et al., 1996). After gastrulation, expression of Hroth is restricted to the anterior epidermis and sensory vesicle cells (Wada et al., 1996). Although complexity of the body plan of ascidian larva is much lower than that of vertebrates, overall expression pattern of Hroth is very similar to that of vertebrate otx genes, which leads to a possibility that Hroth and its vertebrate counterparts share a conserved transcription regulatory mechanism of the patterning in the neural tube formation.
To better understand the mechanism, analysis of cis-regulatory elements for spatiotemporally regulated gene expression is absolutely necessary. In the present study, we analyzed a cis-regulatory region that drives Hroth expression in the sensory vesicle cells. First, we determined the genomic organization of Hroth and found the presence of three introns that are unknown with vertebrate counterparts. Comparison of nucleotide sequences of 5′-rapid amplification of cDNA ends (RACE) products with that of the genomic clone revealed that 5′-end of the Hroth mRNA was modified. Then, we carried out deletion analysis and identified a putative endogenous promoter region of Hroth. We found that a 4-kbp genomic DNA fragment, including the promoter region, was able to drive Hroth expression in the sensory vesicle lineage cells. By deletion analysis of the 4-kbp genomic fragment, we have narrowed down cis-regulatory regions responsible for Hroth expression in the sensory vesicle lineage cells. Among these regions, no apparent sequence conservation was observed. Lastly, based on these observations, a complex organization of regulatory regions for the ascidian otx is discussed.
Genomic Organization of Hroth
To determine the genomic organization and the transcription start site of Hroth, we first isolated a DNA fragment to be examined. We screened a genomic library of Halocynthia roretzi using probes prepared from the cDNA of Hroth (Wada et al., 1996). This strategy yielded a single clone, which contained a DNA fragment encompassing 11.4 kbp upstream from the translation start site and the first 1924 nucleotides of the open reading frame of the Hroth cDNA.
Then, we carried out 5′-RACE analysis, comparing nucleotide sequences of obtained 5′-RACE products with that of the genomic clone. We found a sequence approximately 20 bp in length at the 5′-end of 5′-RACE products, which does not correspond to the genomic sequence (Fig. 1A). Surprisingly, the sequence was also observed at the 5′-end of cDNAs of several Halocynthia genes, which had been isolated previously, including Hrsna (Wada and Saiga, 1999; Fig. 1A). Furthermore, the nucleotide sequence exhibited significant similarity to Ciona 5′ spliced leader RNA (Vandenberghe et al., 2001; Fig. 1A). These observations strongly suggest that trans-splicing occurs with Hroth transcripts. Because the detailed mechanism for this modification is unknown, it was not possible to identify the 5′-end of the primary transcript of Hroth and, in turn, we could not determine the transcription start site of Hroth.
Nevertheless, from the comparison between the 5′-RACE product sequence and the genomic sequence of Hroth, we found five introns as shown in Figure 1B,C. Distal three introns in the genomic map of Hroth were unknown among otx genes of vertebrates. Two proximal introns of Hroth were mapped immediately upstream of the homeodomain and at the position between the codons for the 46th glutamine and 47th valine residues within the homeodomain. These two intron positions appear to be conserved (Williams and Holland, 1998). In otd/otx genes of Drosophila, amphioxus, mouse, and human, two introns are present in common at the same positions (Fig. 1C; Williams and Holland, 1998).
Promoter of Hroth
To test the transcription driving activity of Hroth upstream region, first, we injected p5402-830, which contains the genomic region between 830 and 5402 base pairs upstream from the translation start site of Hroth into Halocynthia fertilized eggs, allowed the eggs to develop, and examined the embryos for lacZ activity by histochemical detection at the early tail bud stage. As shown in Figure 2, p5402-830 drove lacZ activity in most of the injected embryos, suggesting that transcription driving activity is included in this fragment. To characterize the transcription driving activity in the upstream region, various 5′ deletion constructs of p5402-830 were examined (Fig. 2). The constructs of p3139-830, p1752-830, and p1549-830 drove lacZ activity in more than 75% of the embryos injected with them. On the other hand, lacZ activity was scarcely detected in the embryos injected with p1512-830, p1364-830, p1251-830, and p1069-830. This finding suggests that a critical component for the transcription driving activity is present in the upstream region 1549-1513bp. In this region, a TATAA sequence was found to be present, which raised a possibility that this region may represent an endogenous promoter of Hroth.
To test this possibility, we examined the construct p5402-1473, which was prepared by removing a proximal 643 bp from p5402-830, for lacZ activity in the injected embryos at the early tail bud stage. We found p5402-1473 was capable of driving lacZ activity almost to the same degree as was observed when p5402-830 was injected (Fig. 2). The construct p1532-1473, which includes the TATAA sequence and its 3′-adjacent 55-bp region, was also capable of driving lacZ activity in the injected embryos at high frequency (Fig. 2). Furthermore, when the TATAA sequence in p1532-1473 was replaced by CGTAA, activity of lacZ was scarcely detected (p1532Δ-1473 in Fig. 2). These results strongly suggest that the upstream region 1549-1513bp includes an endogenous promoter of Hroth; therefore, this region was designated the putative endogenous promoter.
During the course of deletion experiments, we also found that the upstream region 1069-830bp, including two copies of the TATAC/T sequence, was capable of directing lacZ activity when linked to an upstream region possessing transcription-driving activity (see below, Figs. 4, 5), although it did not exhibit such activity by itself (Fig. 2). Thus, as described below, we used this region as a pseudopromoter for analysis of a region with transcription regulatory activity.
Upstream Region 5402-1473bp Is Capable of Driving Hroth Expression in the Sensory Vesicle Cells at the Tail Bud Stage
Because p5402-1473 was capable of driving lacZ activity in the early tail bud stage embryo almost to the same degree as was found upon injection of p5402-830, we decided to further analyze the region at 5402-1473bp. We first examined in which tissue p5402-1473 drove lacZ expression. Hroth is expressed in the sensory vesicle and anterior epidermis of the trunk (Wada et al., 1996) in the normal tail bud stage embryo (Fig. 3A). The construct of p5402-1473bp drove lacZ activity in various tissues of injected embryos, including the sensory vesicle, trunk epidermis, endoderm, muscle, and mesenchyme (Fig. 3B,C). We also examined tissue distribution of lacZ transcripts by whole-mount in situ hybridization (Fig. 3D), which showed a more or less similar expression pattern to that of lacZ activity. Activity of lacZ was scarcely detected in the posterior neural tube, tail epidermis, and notochord. It is noteworthy that activity of lacZ was detected in the sensory vesicle cells with significantly high frequency (Fig. 3E). This finding suggests that the upstream region 5402-1473bp includes regulatory regions responsible for expression of Hroth in the sensory vesicle cells at the tail bud stage.
Upstream Region Close to the Putative Endogenous Promoter Supports Expression of Hroth in the Sensory Vesicle Cells at the Tail Bud Stage
To narrow down the regulatory regions in 5402-1473bp that drove Hroth expression in the sensory vesicle cells, we carried out the deletion analysis. In this analysis, we found that p1752-1473 was capable of driving lacZ activity in the sensory vesicle lineage cells at the early tail bud stage in almost all injected embryos (Fig. 4A). In embryos injected with p1752-1473 or p5402-1473, a large number of lacZ-positive cells were observed in the sensory vesicle as shown in Figure 4B. The construct of p1549-1473, a more deleted construct (Fig. 4A), also drove lacZ activity in the sensory vesicle cells with high frequency, although the number of lacZ-positive cells in the sensory vesicle was relatively small. This finding suggests that the region 1549-1473bp contains a transcription regulatory region driving the expression in the sensory vesicle cells. In the previous section, we have shown that p1532-1473 is capable of driving expression of lacZ at the tail bud stage (Fig. 2), in which the number of lacZ-positive cells in the sensory vesicle was comparable to the case of p1549-1473 injection (data not shown). This finding indicates that the regulatory region is present close to the putative endogenous promoter.
Upstream Region 1752-1550bp Contains Two Regulatory Regions
Because the number of lacZ-positive cells in the embryo injected with p1549-1473 was smaller than that in the embryo injected with p5402-1473 or p1752-1473 (Fig. 4), a possibility was raised that another regulatory region may be present in 1752-1550bp that drives Hroth expression in the sensory vesicle cells. To address this question, we first examined a construct p1752-1550pp, in which the region 1752-1550bp was combined with the pseudopromoter region. As shown in Figure 4, in 67% of the embryos injected with this construct, a large number of sensory vesicle cells exhibited lacZ activity. This result confirms that the region 1752-1550bp contains a regulatory region that drives expression of Hroth in the sensory vesicle cells at the tail bud stage. Then, we carried out deletion analysis of p1752-1550pp to locate the regulatory region. As shown in Figure 5, 3′-deletion analysis indicated that deletion of the regions of 1628-1613bp and 1659-1650bp resulted in significant decrease in frequency of the embryo positive for lacZ activity in the sensory vesicle cells. In 5′-deletion analysis, when the upstream region 1628-1613bp was deleted, the frequency of lacZ-positive embryos decreased (Fig. 5), whereas the deletion of upstream region 1659-1650bp exhibited a smaller decrease. We further examined p1642-1613pp, in which the upstream region 1642-1613bp was combined with the pseudopromoter region (Fig. 5). The construct was capable of driving lacZ activity in the sensory vesicle cells in 70% of the injected embryos. This further supports a view that this region functions as the regulatory region that drives Hroth expression in the sensory vesicle cells at the tail bud stage.
In this study, we elucidated the genomic organization of the 5′-upstream region of Hroth and identified a region likely to be the endogenous promoter for this gene. Also, we identified the cis-regulatory regions capable of driving expression of Hroth in the sensory vesicle cells at the tail bud stage.
Intron/Exon Organization and Promoter of Hroth
Because of the unexpected modification of the 5′-end of Hroth transcript, we could not completely determine the extreme end of the transcription unit of Hroth. However, we found the presence of five introns in the 5′-end region of Hroth. Two of the five introns are found at the positions phylogenetically well conserved: one is in the homeobox, generally present close to the 5′-end of the translated region in otx/otd genes and the other in the immediate upstream of the homeobox. Because the positions of the two introns are highly conserved between vertebrate otx and Drosophila orthodenticle (Williams and Holland, 1998), the origin of these introns must be very old. By contrast, the remaining three introns appear to be unique to ascidians, because a similar intron/exon organization has been observed in other ascidian species (B. Degnan, P. Lemaire, personal communication). This finding suggests that the 5′-most region of ascidian otx might have been altered, leaving the positions of the two introns unchanged, after diversification of ascidians from the lineage to vertebrate.
Modification of the 5′-end of Hroth transcript is likely exerted by trans-splicing, originally known in trypanosome and recently reported in the ascidian Ciona intestinalis (Vandenberghe et al., 2001). The mechanism and biological meaning of this modification are still in enigma. Halocynthia and Ciona belong to two different orders, which constitute the whole class of Ascidiacea. Therefore, it seems likely that this modification has spread widely among ascidians.
Despite this novel modification of the 5′-end of Hroth transcript, the putative endogenous promoter region of Hroth was mapped through the deletion analysis. The basis for this mapping was as follows. In our deletion analysis, when deleted from the 5′-upstream side, the lacZ reporter constructs without this region exhibited no lacZ activity. On the other hand, the 5′-upstream DNA fragment of 5402-1473, which contains the putative promoter region but not its 3′-juxtaposed region, exhibited lacZ activity. This region includes the TATAA sequence, which is expected to be a putative TATA box. When this sequence was mutated, lacZ activity was lost. These observations suggest that the region is likely an endogenous promoter of Hroth. Additionally, this region supports transcription quite efficiently when introduced into the embryo. Under our lacZ staining conditions, constructs driven by the putative endogenous promoter required a shorter time for staining than those driven by the pseudopromoter.
Property of Transcription Regulatory Activity of the Upstream Region 5402-1473bp
In this study, we have found that the upstream region, 5402-1473bp is capable of driving lacZ expression in the sensory vesicle cells, in which Hroth is expressed in the normal tail bud stage embryo. However, this region led to lacZ activity not only Hroth-expressing cells but also in other tissues such as epidermis, endoderm, mesenchyme, and others, which were thought to be ectopic at the tail bud stage (Fig. 3C,E). With regard to such ectopic expression, there may be two possibilities. One is that the upstream region 1473-5402bp alone cannot restrict expression of Hroth to the endogenous expression sites. We did not observe any significant change in the frequency and/or degree of ectopic expression in our deletion experiments, suggesting that a region that has activity to repress such ectopic expression should be, if present, somewhere surrounding the Hroth gene region not examined in the present study. Another possibility, which appears more likely, is that the ectopic expression may be due to carryover of lacZ protein and/or lacZ transcript known to be highly stable. The tissues in which lacZ expression was detected were mostly descendants of the cells expressing Hroth up to the gastrula stage (Wada et al., 1996; Fig. 3E). In agreement with the latter possibility, it was found that, in 64-cell stage embryos, the upstream region 5402-1473bp was capable of driving transcription of lacZ in the cells, in which endogenous Hroth expression is detected (data not shown). Furthermore, activity of lacZ was never detected in the notochord, in which Hroth is not expressed in normal development. In the posterior neural tube and tail epidermis, which are derived from the A- and b-lineage precursor cells, respectively, activity of lacZ was scarcely detected. In these precursor cells, Hroth is not expressed or expressed up to the 64-cell stage (Wada et al., 1996). To further confirm this possibility, detailed analysis of the early lacZ expression driven by the region 5402-1473bp is required in future study.
Possible Transcription Factor Binding Sites in the Regulatory Regions
In the present study, we identified three regions as a regulatory region responsible for the expression of Hroth in the sensory vesicle cells at the tail bud stage. Among the regulatory regions narrowed down in the present study, no sequence conservation was recognized. We tried to search for a putative binding site for transcription factors in the three upstream regions by using a computer program, TFSEARCH. As shown in Figure 6, a putative binding site for MZF-1 is present around the upstream region 1650-1659bp. MZF-1 is a zinc finger gene, which has been reported to be preferentially expressed in hematopoietic cells in vertebrates (Hromas et al., 1991). In ascidians, however, MZF-1 homologue has not been identified yet. In the upstream region 1613-1628bp, putative binding sites for CDXA, HFH-2, and HNF3-β, are present. Hrcad has been identified as the single Halocynthia roretzi cdx/caudal homologue, which has been reported to be expressed in the posterior part of the developing neural tube (Katsuyama et al., 1999). Therefore, Hrcad is unlikely to be a good candidate for a regulatory factor of Hroth expressed in the sensory vesicle lineage cells. Another homeobox gene may bind to the TAAT sequence, which is a conserved core element for homeodomain protein binding sites (Gehring et al., 1994). With regard to the homologous genes of HNF3-β and HFH-2, Cs-FoxA5, and Cs-FoxD, respectively, have been identified in Ciona savignyi, another ascidian species. However, their temporal expression patterns during embryogenesis (Shimauchi et al., 2001; Imai et al., 2002) make it unlikely that they are directly responsible for regulating expression of Hroth in the sensory vesicle.
The upstream region 1549-1473bp includes binding sites for Sox-5, GATA, CdxA, and HSF2. However, except for CdxA, the ascidian homologues have not yet been isolated. Furthermore, there has been no evidence accumulated to suggest their participation in patterning of the CNS. With regard to the expression in the sensory vesicle, upstream genes of otx have not yet been identified. In three regulatory regions, there is no otx binding consensus sequence (Fig. 6).
Complex Regulatory Regions for Hroth
In the present study, we have focused on the regulatory region responsible for the expression in the anterior CNS. Because the upstream region, 5402-1473bp seems to drive lacZ transcription in the sensory vesicle cells or sensory vesicle lineage cells, the regulatory regions we have identified in the present study likely drive transcription of Hroth in the sensory vesicle lineage cells. However, we cannot tell all of the three or which of the three are responsible for the transcription of Hroth at the early tail bud stage. All of the three regulatory regions exhibited weaker activity in terms of the number of lacZ-positive sensory vesicle cells and intensity of lacZ staining (data not shown) as the regulatory region became narrowed down by deletion analysis, although some quantitative difference in activity to drive transcription was observed. Deletion analysis showed that a 3′-deletion of the upstream region 1650-1659bp resulted in a significant decrease in the number of embryos positive for lacZ activity in the sensory vesicle, whereas a 5′-deletion had no significant influence. Explanation for this finding may be that expression driving activity of the regulatory region 1613-1628bp is stronger than that of the regulatory region 1650-1659bp. Thus, the loss of the upstream region 1650-1659bp did not largely affect the transcription driving activity of the reporter constructs with the regulatory region 1613-1628bp.
It is likely that expression of Hroth in the sensory vesicle cells is under control of the multiple regulatory regions, although the possibility of the presence of a silencing region to repress ectopic expression at the tail bud stage cannot be excluded. It should be noted that Hroth exhibits a complex expression pattern, dynamically switching expression sites during development. Expression sites include precursors for endoderm, mesenchyme, some muscle cells and anterior epidermis, as well as sensory vesicle during earlier developmental stage (Wada et al., 1996). Other than the regulatory regions identified in the present study, it is possible that a regulatory region specific for early but not later expression of Hroth may also be present. As described, the genomic organization of the ascidian otx appears to be distinct among other chordate counterparts. On the other hand, the overall expression pattern of otx is rather well conserved. Whether or not there is a conserved regulatory mechanism for expression of otx in chordates is an interesting issue to be studied further.
Adult ascidians Halocynthia roretzi were purchased from fishermen near Asamushi Marine Biological Station, Tohoku University, Aomori, or Otsuchi Marine Research Center of the Ocean Research Institute, University of Tokyo, Iwate, Japan.
Isolation of Genomic Clones
A Halocynthia roretzi genomic library constructed by using λDASH (Stratagene) was used to isolate 5′-upstream of Hroth in the present study. Two DNA fragments, 173bp of 5′-end and 133bp of 3′-end, of the Hroth cDNA (Wada et al., 1996) were used as probes for screening of the genomic DNA library. The nucleotide sequence data of the 5′-upstream region of Hroth has been deposited in the DDBJ database under accession no. AB104851.
Preparation of Fusion Gene Constructs
All reporter constructs were prepared by inserting genomic DNA fragments from 5′-upstream region of Hroth into the multicloning site of pPD46.21 vector, which is a variant of pPD1.27 (Fire et al., 1990). This vector encodes the lacZ gene (bacterial β-galactosidase) and a nuclear localization signal immediate upstream of its N-terminus.
Constructs of p1069-830, p1752-1473, p1549-1473, p1532-1473, and p1752-830 were prepared by inserting PCR products into the SmaI/BamHI, BamHI/SalI, BamHI/SalI, BamHI/SalI, and SmaI/SalI sites within the pPD46.21 vector, respectively. The numbers in the name of each construct indicate positions from the first A of the translation start codon. For making p1532Δ-1473, mutations at two nucleotide residues were introduced into p1532-1473 by PCR using mutagenic primers. The construct p5402-1473 was made by inserting the 3.6-kbp Hroth genomic fragment into the SalI/HindIII sites within the p1752-1473. Constructs of p5402-830 and p3139-830 were prepared by inserting the 3.6 kbp and 1.4 kbp Hroth genomic fragments into the SalI/HindIII and SalI/PstI sites within the p1752-830, respectively. Constructs of p1549-830, p1512-830, p1364-830, and p1251-830 were prepared by digesting p1752-830 with exonuclease III. For preparation of the deletion constructs of p1752-1550pp (“pp” stands for the pseudopromoter, see text), various primer sets were used to obtain a set of deleted DNA fragments of 1752-1550 bp, which were inserted into the BamHI/SalI sites of the p1069-830 plasmid DNA. All PCR-amplified fragments were determined for the nucleotide sequence to confirm identity to the original sequence.
Microinjection of Reporter Constructs, Histochemical Detection of lacZ Activity, and Detection of lacZ Transcripts by Whole-Mount In Situ Hybridization
Microinjection of fusion gene constructs into fertilized eggs and histochemical detection of lacZ activity were carried out according to Kusakabe et al. (1995).
Fifteen minutes after insemination, fertilized eggs were treated with 4 ml of sea water containing 1% sodium thioglycolate, 0.05% Actinase E, and 9 ml of 1 N NaOH for 10 min to soften the chorion. Construct plasmid DNAs of a circular form were dissolved in 1 mM Tris-HCl, 0.1 mM EDTA, pH 8.0, and injected into the eggs at the concentration of 3 or 13 μg/ml. Injected embryos were allowed to develop and fixed with Millipore-filtered seawater containing 1% glutaraldehyde for 30 min at room temperature. Fixed embryos were washed in phosphate-buffered saline (PBS) for 5 min followed by washing in PBS containing 0.1% Tween 20, 1 mM MgCl2, 3 mM K4[Fe(CN)6], and 3 mM K3[Fe(CN)6] for 5 min. Then, they were incubated in PBS containing 250 mM 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside (X-gal), 0.1% Tween 20, 1 mM MgCl2, 3 mM K4[Fe(CN)6], and 3 mM K3[Fe(CN)6] at 37°C for 12 hr. The embryos were washed in PBS to stop the staining reaction and examined for the staining under a stereomicroscope (Olympus SZH10). For an experiment for each construct, 20–70 embryos were injected, cultured, and examined for lacZ expression. Also at least two independent experiments were carried out for each construct by using different batches of embryos. Whole-mount in situ hybridization was carried out according to Oda and Saiga (2001).
The authors thank the staff of Asamushi Marine Biological Station of Tohoku University and of Otsuchi Marine Research Center of the University Tokyo for supplying living animals and facilitating our experiments there. Thanks also to Patrick Lemaire and Vincent Bertrand (Marseille, France) for their helpful comments and discussions and Bernie Degnan (Qeensland, Australia) for communicating unpublished sequence data. H.S. received support from JSPS and the Ministry of Education, Science, Sports, and Culture, of Japan, and I.O received a Sasakawa Scientific Research Grant from the Japan Science Society Foundation.