Two axial structures, a neural tube and a notochord, are key structures in the chordate body plan and in understanding the origin of chordates. To expand our knowledge on mechanisms of development of the neural tube in lower chordates, we have undertaken isolation and characterization ofHrzicN, a new member of the Zic family gene of the ascidian,Halocynthia roretzi. HrzicN expression was detected by whole-mount in situ hybridization in all neural tube precursors, all notochord precursors,anterior mesenchyme precursors and a part of the primary muscle precursors. Expression of HrzicN in a- and b-line neural tube precursors was detected from early gastrula stage to the neural plate stage, while expression in other lineages was observed between the 32-cell and the 110-cell stages.HrzicN function was investigated by disturbing translation using a morpholino antisense oligonucleotide. Embryos injected with HrzicNmorpholino (`HrzicN knockdown embryos') exhibited failure of neurulation and tail elongation, and developed into larvae without a neural tube and notochord. Analysis of neural marker gene expression inHrzicN knockdown embryos revealed that HrzicN plays critical roles in distinct steps of neural tube formation in the a-line- and A-line precursors. In particular HrzicN is required for early specification of the neural tube fate in A-line precursors. Involvement of HrzicNin the neural tube development was also suggested by an overexpression experiment. However, analysis of mesodermal marker gene expression inHrzicN knockdown embryos revealed unexpected roles of this gene in the development of mesodermal tissues. HrzicN knockdown led to loss of HrBra (Halocynthia roretzi Brachyury) expression in all of the notochord precursors, which may be the cause for notochord deficiency.Hrsna (Halocynthia roretzi snail) expression was also lost from all the notochord and anterior mesenchyme precurosrs. By contrast,expression of Hrsna and the actin gene was unchanged in the primary muscle precursors. These results suggest that HrzicN is responsible for specification of the notochord and anterior mesenchyme. Finally,regulation of HrzicN expression by FGF-like signaling was investigated, which has been shown to be involved in induction of the a- and b-line neural tube, the notochord and the mesenchyme cells inHalocynthia embryos. Using an inhibitor of FGF-like signaling, we showed that HrzicN expression in the a- and b-line neural tube, but not in the A-line lineage and mesodermal lineage, depends on FGF-like signaling. Based on these data, we discussed roles of HrzicN as a key gene in the development of the neural tube and the notochord.

Two axial structures, a neural tube and a notochord, are characteristics of chordates and therefore, information about developmental mechanisms of the neural tube and the notochord in lower chordates is essential for an understanding of the origin of the chordate body plan(Satoh and Jeffery, 1995).

The neural tube of ascidian larvae is composed of about 340 cells, and is divided into three regions along the anteroposterior axis, which are, from anterior to posterior, the sensory vesicle, the visceral ganglion and the caudal neural tube (Nicol and Meinertzhagen, 1991). The sensory vesicle is composed solely of the a-line (anterior-animal) cells(Nishida, 1987). The visceral ganglion present at the junction between the trunk and tail consists of the A-line (anterior-vegetal) cells. The caudal neural tube running along the length of the tail consists of four (dorsal, ventral and two lateral) rows of ependymal cells: the lateral and ventral cells are of A-line origin and the dorsal cells are of b-line (posterior-animal) origin. Beneath the neural tube,a stack of exactly 40 notochord cells runs along the tail. The anterior 32 cells (primary notochord) and the posterior 8 cells (secondary notochord) are derived from A-line and B-line cells, respectively(Nishida, 1987).

Cellular interactions that specify the neural tube and notochord of ascidian embryos have been extensively demonstrated. Specification mechanisms of neural tube differ between the a- and b-line precursors and A-line precursors. The a-line neural tube precursors require an inductive influence from the vegetal hemisphere cells to form the sensory vesicle(Nishida and Satoh, 1989;Okado and Takahashi, 1990). Upon disturbance of the induction, they adopt epidermal fate, like most other animal hemisphere cells. Although timing of the induction has not been fully understood, it is likely that the induction starts at the 16-cell stage and becomes complete at the early gastrula stage, including multiple sequential steps (Darras and Nishida,2001b; Nishida and Satoh,1989; Okado and Takahashi,1990). The b-line precursors also require an inductive influence from the vegetal cells to differentiate into the caudal neural tube cells(Hudson and Lemaire, 2001). In the induction of a- and b-line neural tube cells, an FGF-like signaling pathway is likely involved, since human recombinant basic FGF mimics the inductive activity of the vegetal hemisphere, and block FGF signaling leads to inhibition of the sensory vesicle formation(Darras and Nishida, 2001b;Hudson and Lemaire, 2001;Inazawa et al., 1998;Kim and Nishida, 2001). This situation is very reminiscent of neural induction in vertebrates. By contrast,specification mechanisms of the A-line neural tube cells seems to be unique(Minokawa et al., 2001). At the 32-cell stage, anterior-most A-line vegetal cells (A6.2 and A6.4 blastomere pairs) have both neural tube and notochord fates, which separate into the daughter cells after the next cleavage. Anteriorly located daughters succeed to the neural tube fate while posterior ones that contact the endoderm precursors assume the notochord fate. A-line neural tube fate is specified autonomously without any cellular interaction(Minokawa et al., 2001).

A-line notochord fate, however, requires inductive influence from the endoderm precursors or the neighboring notochord precursors, which can be mimicked by basic FGF (Darras and Nishida,2001a; Kim and Nishida,2001; Nakatani and Nishida,1994; Nakatani et al.,1996; Shimauchi et al.,2001). Interestingly, all descendants of the isolated A6.2 or A6.4 blastomeres adopt the notochord fate when treated with basic FGF, and conversely, they all adopt the neural tube fate in the absence of the induction. Therefore, binary choice of the alternative fates is involved in specification of the A-line neural tube and the notochord(Minokawa et al., 2001). In contrast to our knowledge about cellular interactions involved in specification of the neural tube, little is known about transcription factors that participate in this process.

Zic was originally identified as a gene encoding zinc finger protein that is expressed abundantly in the adult mouse cerebellum(Aruga et al., 1994). In vertebrates, multiple Zic family genes are known: for example, at least six distinct Zic family genes have been identified inXenopus (Nakata et al.,2000) and at least four in mouse(Aruga et al., 1996). Vertebrate Zic family genes so far identified are very similar to one another both in structure and expression pattern during development(Nagai et al., 1997;Nakata et al., 1998). At the gastrula stage, Zic family genes are expressed throughout the presumptive neural plate. Their expression becomes restricted to the lateral edges of the neural plate at the neurula stage and persists in the dorsal region of the forebrain and the midbrain, the roof plate of the spinal cord,the migratory neural crest and additionally in developing somites and limb buds. The expression pattern of vertebrate Zic family genes suggests their early roles in neural and neural crest development. In accordance with this, overexpression experiments show that Xenopus Zic genes promote differentiation of neural and neural crest derived tissues(Nakata et al., 1998). In mouse, knockout of Zic1 leads to aplasia of cerebellum and skeletal abnormalities (Aruga et al.,1998). Mutation in Zic2 and Zic3 in mouse and/or human cause holoprosencephaly and heterotaxis, respectively(Klootwijk et al., 2000;Nagai et al., 2000). These mutant phenotypes seem to be much weaker than expected from the results of overexpression experiments.

In ascidians, macho-1, a recently identified muscle determinant,encodes a zinc finger protein with the zinc finger domain most similar to that of Zic family genes (Nishida and Sawada, 2001). Transcripts of macho-1 are supplied to the eggs maternally, segregated into the primary (the B-line) muscle cells through rounds of cleavage, and have been shown to be necessary and sufficient for primary muscle formation. Thus expression and function of macho-1 are quite different from those of vertebrate Zic family genes. So far, no ascidian Zic family gene other than macho-1 has been reported.

In the present study, toward understanding of the neural development in ascidian embryos, we addressed whether there is another Zic family gene that may have a role in the neural development. We have clonedHrzicN, a new Zic family gene of the ascidian,Halocynthia roretzi and studied expression, function and regulation of HrzicN. We here report that HrzicN plays an essential role in neural tube development. Unexpectedly, we have also found that this gene is required for the formation of the notochord and anterior mesenchyme,representing a novel function of Zic genes. Thus, the present study has established that HrzicN is a key gene in the development of the two axial structures in the ascidian embryo.

Embryos

Adult ascidians, Halocynthia roretzi, were obtained from fishermen near Asamushi Marine Biological Station, Tohoku University, Aomori, Japan and Otsuchi Marine Research Center, Ocean Research Institute, University of Tokyo,Iwate, Japan. Naturally spawned eggs were fertilized with a suspension of sperm from other individuals. Fertilized eggs were raised at 11-13°C in Millipore-filtered seawater containing 100 μg/ml streptomycin and 100 units/ml penicillin. Nomenclature of the cell lineage is according to Nishida(Nishida, 1987).

Molecular cloning of HrzicN

PCR was carried out to amplify DNA fragments of a conserved region within the zinc finger domain of Zic family genes. The nucleotide sequences for forward and reverse primers were 5′-GCGAATTCTT(CT)AA(AG)GC(ACGT)AA(AG)TA(CT)AA-3′ and 5′-CGCTGCAGTG(ACGT)AC-(CT)TTCAT(AG)TG(CT)TT-3′, respectively. PCR reaction was carried out through 40 cycles of denaturing at 94°C for 1 minute, annealing at 48°C for 1 minute and elongation at 63°C for 2 minutes using Halocynthia roretzi genomic DNA as template. The PCR products were sequenced and two types of candidate clones for Zicfamily genes were identified. One was of macho-1 and the other represented a novel gene. Therefore, this was used as a probe for screening of a neurula cDNA library. A cDNA clone was isolated and sequenced. 5′- and 3′-most regions for the cDNA were isolated by 5′- and 3′-RACE procedures, respectively, using the RACE System (Gibco BRL).

Construction of expression plasmids for in vitro transcription of mRNA

To generate the expression plasmid for HrzicN, a full lengthHrzicN cDNA was cloned into pBluescriptRN3(Lemaire et al., 1995). To generate the expression plasmid for lacZ (pRN3/lacZ), full lengthlacZ of pSV-β-Gal was cloned into pBluescriptRN3. To generate the expression plasmid for HrzicN/lacZ, the 5′ UTR and the initiation codon of lacZ cDNA were substituted with the 5′ UTR and the first 183 bp of HrzicN coding region amplified by PCR. To generate the expression plasmid for lacZ/HrzicN, the first 52nd base pair of HrzicN cDNA was substituted with the first 123 bp oflacZ cDNA. In vitro transcription was carried out using mMessage mMachine (Ambion) as described previously(Wada and Saiga, 1999b).

Design of morpholino oligonucleotide and microinjection

Morpholino oligonucleotides were obtained from Gene Tools. Sequences of HrzicNMO and HrzicNMO2 are 5′-GCTGTTGCGTATGCCATTTTTGCTT-3′ (the underline indicates sequence complimentary to the putative initiation codon) and 5′-ATTCGCTCAATTAAATTACTGTTGT-3′, respectively. As a negative control, `standard control oligo' supplied by Gene Tools was used. Microinjection was carried out as described previously(Wada and Saiga, 1999b). Synthetic RNA and morpholino oligonucleotides to be injected were dissolved in distilled water and 0.1× TE, respectively. Each microinjection experiment was conducted twice or more.

To test the ability of HrzicNMO to inhibit translation, approximately 50 pg of HrzicN/lacZ mRNA was injected into fertilized eggs either with or without HrzicNMO (final concentration: 10 μM). Cleavage of the injected embryos was inhibited by treatment with cytochalasin B(Hirano et al., 1984). The injected embryos were tested for β-galactosidase activity at the middle tailbud equivalent stage as described previously(Hikosaka et al., 1994).

For the rescue experiment, a mixture of HrzicNMO (final concentration: 10μM) and approximately 50 pg of lacZ/HrzicN mRNA was injected into fertilized eggs and the injected embryos were examined for HrBraexpression at the 110-cell equivalent stage.

Treatment with MEK inhibitor

To inhibit the FGF-Ras-MAP kinase signaling pathway, the MEK (MAP kinase kinase) inhibitor (U0126, Promega) was used. Embryos were cultured in Millipore-filtered seawater containing 2 μM U0126 as described previously(Kim and Nishida, 2001).

Whole-mount in situ hybridization

Gene expression was visualized by whole-mount in situ hybridization as described previously (Wada et al.,1995).

Structure of HrzicN

The HrzicN cDNA (accession no. AB092643) encodes a protein of 468 amino acids with five zinc finger motifs(Fig. 1A). Over the zinc finger domain, HrzicN protein shows significant sequence similarity to the known members of the Zic family proteins. The sequence outside the zinc finger domain shows no similarity to the Zic or other known proteins, lacking a motif conserved between vertebrate Zic and Drosophila odd-paired proteins. To see whether HrzicN belongs to the Zic family, a phylogenic tree was constructed with amino acid sequences of zinc finger motifs of HrzicN and other Zic family proteins, using sequences of vertebrate Gli proteins (Ruiz i Altaba,1999), which comprise a related family of zinc finger proteins as an outgroup (Fig. 1B). The tree supported close relationship between HrzicN and a group consisting of vertebrate Zic 1, 2 and 3. Therefore, it is highly likely that HrzicN represents a new member of Zic family.

Expression pattern of HrzicN during embryogenesis

The spatial and temporal expression pattern of HrzicN during larval development was examined by whole-mount in situ hybridization(Fig. 2). As described below,HrzicN expression was detected in the cells of neural and mesodermal lineages from the early 32-cell stage to the neural plate stage.

Expression in the neural lineage

The larval ascidian neural tube originates from the a4.2, b4.2 and A4.1 cell pairs at the 8-cell stage. HrzicN expression was evident in all three lineages of neural tube precursor cells. HrzicN expression was first detected in A6.2 and A6.4 blastomere pairs, each of which contains both neural tube and notochord fates, at the early 32-cell stage(Fig. 2A,F,R). At the 44- and 64-cell stages, HrzicN expression was found in their daughter cells,the A-line neural tube precursors (A7.4 and A7.8 pairs) and the A-line notochord precursors (A7.3 and A7.7 pairs)(Fig. 2B,C,G,H,R). The expression continued in these cell lineages until the 110-cell stage(Fig. 2D,E,I,J,R) but became undetectable by the early gastrula stage(Fig. 2K,P). At this stage,however, HrzicN expression started in all the a- and b- line neural tube precursors (a8.17, a8.19, a8.25, b8.17 and b8.19 pairs;Fig. 2K,P). The expression continued during gastrulation, disappeared by the early neurula stage(Fig. 2L-N,Q) and no longer detected afterwards (Fig. 2O).

Expression in the mesodermal lineage

As mentioned above, HrzicN was expressed in the primary notochord lineage cells at the early 32-cell stage through the 110-cell stage. In addition, HrzicN was expressed until the 110-cell stage in the B6.2 pair at the late 32-cell stage and their descendants(Fig. 2R), which develop into the secondary notochord, mesenchyme and primary muscle(Fig. 2B-E,G-J,R). Like the expression in the A-line cells, the expression in the B-line cells became undetectable by the early gastrula stage(Fig. 2K,P). In summary,HrzicN expression was detected in all notochord precursors, one of the two pairs of mesenchyme precursors and two out of the five pairs of primary muscle precursors of the 110-cell stage embryo.

Phenotype induced by morpholino antisense oligonucleotide-based translational inhibition of HrzicN

Recently, morpholino antisense oligonucleotide-based translational inhibition was shown to be an effective tool for loss-of-function experiments in the ascidian embryos (Satou et al.,2001). We applied this technique to deduce functions ofHrzicN during embryogenesis. A morpholino oligonucleotide we prepared(HrzicNMO) targets the initiation codon and its flanking regions. The ability of HrzicNMO to inhibit translation was assessed by examining the effect of HrzicNMO on translation of HrzicN/lacZ mRNA (a chimeric mRNA in which the 5′ UTR and the initiation codon of lacZ mRNA were substituted with the 5′ UTR and the first 183 nucleotides ofHrzicN mRNA coding region) in cleavage-arrested embryos. As summarized in Table 1, HrzicNMO was capable of disturbing translation of the mRNA with its target site.

To investigate the effect of translational inhibition of HrzicNmRNA on ascidian development, we injected HrzicNMO into fertilized eggs to achieve a final concentration of 1, 5 and 10 μM and reared them up to the swimming larva equivalent stage. As a negative control, we injected the`standard control oligo' supplied by Gene Tools and found that embryos injected at a final concentration of 10 μM or lower developed normally(Fig. 3C,D). Eggs injected with HrzicNMO at 1 or 5 μM developed into normal larvae (data not shown). However, almost all eggs injected with 10 μM developed into larvae with severe defects such as shortening of the tail, no differentiated notochord cells, failure of neural tube formation and lack of sensory pigment cells(Fig. 3B). However,protrusions, which are likely the adhesive organ, formed at the tip of the trunk (arrowhead in Fig. 3B). In the course of development, HrzicNMO-injected embryos seemed to be normal until early gastrula stage but delay in involution became evident in the later half of the gastrula stage (data not shown). The most notable abnormality was that they exhibited no sign of neurulation. Elongation of the tail was also inhibited, although a tail tip-like structure formed(Fig. 3A,B).

The phenotype generated by HrzicNMO injection seemed to be unique to translational inhibition of HrzicN. First, an essentially identical phenotype was observed upon injection of HrzicNMO2, another morpholino nucleotide against HrzicN with a different and non-overlapping target site from that of HrzicNMO (see inset inFig. 3B). This phenotype was completely different from those observed upon injection of a morpholino oligonucleotide against Hroth (S. W. and H. S., unpublished) orβ-catenin mRNA (S. W., K. W. Makabe and H. S., unpublished). Second, to verify specificity of effects of HrzicNMO, a rescue experiment was carried out using lacZ/HrzicN mRNA, in which a translational initiation site was provided by insertion of a lacZ fragment into the HrZicNplasmid DNA so as to shift the translation initiation site 81 bp upstream to the original initiation site of HrzicN. Thus, it is expected that mRNA from this construct is free of translational inhibition by HrzicNMO,since it has been shown that morpholinos that target more than a few bases 3′ to the initiation codon exhibits a quite low efficiencySummerton, 1999). Co-injection of lacZ/HrzicN mRNA and HrzicNMO recovered HrBra expression otherwise lost by injection of HrzicNMO(Fig. 6B,G; for details about marker gene expression in HrzicNMO-injected embryos, see below).

Injection of HrzicNMO into eggs at higher than 10 μM resulted in a phenotype similar to that obtained by injection at 10 μM (data not shown),so that injection at 10 μM seemed to be sufficient for inducing a representative phenotype by HrzicNMO. Therefore, we refer to embryos injected with HrzicNMO at 10 μM at the 1-cell stage as “HrzicNknockdown embryos”.

Neural tube differentiation is disturbed in HrzicN knockdown embryos

Since neurulation was blocked in HrzicN knockdown embryos, we investigated the neural tube development in these embryos by analyzing expression of neural markers. Initially, we assessed neural tube differentiation in HrzicN knockdown embryos at the early tailbud equivalent stage. First, we examined expression of a pan-neural marker,HrETR-1 (Yagi and Makabe,2001), which is expressed in the whole neural tube except for the dorsal and ventral walls of the caudal neural tube, and the peripheral neurons in the normal early tailbud stage embryo(Fig. 4D). In HrzicNknockdown embryos, weak HrETR-1 expression was found in superficial cells of the trunk that seemed to be presumptive a-line neural tube cells, but not in presumptive A-line neural tube precursors(Fig. 4A). Next, we examined expression of another neural marker, HrTBB2 (Halocynthia roretziβ-tubulin gene) (Miya and Satoh,1997), which is expressed in the neurons of the adhesive organ,the neural tube in the trunk and peripheral epidermal neurons in the tail(Fig. 4E). In HrzicNknockdown embryos, expression of HrTBB2 was evident in the adhesive organ-forming region and the epidermal neurons, but expression in the neural tube-forming region was completely lost(Fig. 4B). We examined expression of HrTRP encoding the tyrosinase related protein(Sato et al., 1999) andHroth (Halocynthia roretzi otx gene)(Wada et al., 1996).HrTRP is expressed in dorsal and lateral parts of the sensory vesicle in the normal early tailbud stage embryo(Fig. 4F), but it was not expressed in HrzicN knockdown embryos(Fig. 4C). Hrothexpression was also abnormal in HrzicN knockdown embryos. In the normal early tailbud stage embryo (Fig. 4J) this gene is expressed in the sensory vesicle and the anterior epidermis. In HrzicN knockdown embryos, Hroth expression was lost from the sensory vesicle precursors, while it was detected only in the anterior epidermis (Fig. 4G). Together, these results indicate that differentiation of the neural tube is severely affected in HrzicN knockdown embryos.

Neural fate specification occurs in a- but not A-line precursors

In the a- and b-line precursors, HrzicN expression starts at the early gastrula stage. Several lines of evidence suggest that fate choice between epidermis and neural tube fates occurs earlier than this stage(Darras and Nishida, 2001b;Ishida et al., 1996;Yagi and Makabe, 2001). However, in the A-line neural tube cells, HrzicN expression starts at the early 32-cell stage. Although it is unclear when the neural tube fate is established in these A-line cells, the onset of HrzicN expression is well before the appearance of neural properties such as HrETR-1expression (this starts at the 110-cell stage). To see when abnormality in the neural tube development arises in HrzicN knockdown embryos and whether there is a difference in this process between a-line cells and A-line cells, we next examined expression of these neural markers at earlier stages of development. At the 110-cell stage, in control embryos, HrETR-1 is expressed in the A-line neural tube precursors and only weakly in the a-line neural tube precursors (Fig. 5F,G). At this stage, Hroth is also expressed in the a-line neural tube precursors (Fig. 5H-J). In HrzicN knockdown embryos, HrETR-1expression was evident in the a-line precursors but not in the A-line neural tube precursors (Fig. 5A,B). The a-line precursors of these embryos were also positive for Hrothexpression (Fig. 5C-E). These results suggest that initial specification of the neural tube precursors occurs normally in a-line but not A-line precursors in HrzicNknockdown embryos.

We then examined expression of the markers at the neural plate equivalent stage. At this stage, in control embryos expression of HrETR-1 andHroth continues in the a- and A- line neural tube precursors and in the a-line precursors, respectively (Fig. 5Q,R). In the a-line precursors, HrTRP is also expressed(Fig. 5S). In HrzicNknockdown embryos, HrETR-1 expression was again absent from the A-line neural tube precursors (Fig. 5L) but evident in the a-line neural tube precursors, although the level of the expression was lower than that in control embryos(Fig. 5L,Q). Similarly, the expression of Hroth in the a-line neural tube precursors was reduced in HrzicN knockdown embryos (Fig. 5M,R). Furthermore, HrTRP expression was lost in them(Fig. 5N). These results suggest that the neural fate is once specified but not maintained at later stages of development in the a-line neural tube precursors.

Since early specification of the neural fate likely occurs in the a-line neural tube precursors of HrzicN knockdown embryos, it is expected that the epidermal fate is excluded from these cells. To test this possibility, we first examined HrzicN knockdown embryos for expression of HrEpiG (Ishida et al., 1996), which occurs only in the epidermis precursors after the 76-cell stage in normal development. HrEpiG expression was normal, being excluded from both of the a- and b-line neural tube precursors in HrzicN knockdown embryos at the early gastrula equivalent stage(Fig. 5K,P). Next, we examinedHrzicN knockdown embryos at the early tailbud equivalent stage for expression of HrEpiD (Ishida et al., 1996), which is another epidermis-specific gene expressed in the whole epidermis except for a small area around the neuropore in the normal embryo at this stage (Fig. 4K).HrEpiD expression was excluded from a group of cells at the dorsal side of HrzicN knockdown embryos. These cells were thought to be descendants of the original neural tube precursors that had rejected the epidermis fate (Fig. 4H). These results indicate that the fate choice between neural tube and epidermis is made successfully in HrzicN knockdown embryos.

Finally, we examined expression of HrzicN itself inHrzicN knockdown embryos. At the 110-cell stage, HrzicNexpression in HrzicN knockdown embryos seemed to be identical to that in control embryos, indicating that HrzicN is not required for the maintenance of its early expression (data not shown). As mentioned previously,HrzicN expression disappears from the vegetal cells by the early gastrula stage in the normal development. By contrast, HrzicNexpression was detected not only in the a- and b-line but also in the A-line neural tube precursors in HrzicN knockdown embryos at the middle gastrula equivalent stage (Fig. 5O,T). Thus, HrzicN expression in the A-line neural tube precursors failed to be suppressed in HrzicN knockdown embryos. Furthermore, HrzicN expression continued in dorsal superficial cells until early tailbud equivalent stage (Fig. 4I,L). From their position, these cells were thought to be descendants of the neural tube precursors that failed to form a neural tube. Thus, HrzicN expression in HrzicN knockdown embryos continued until a much later stage of development. This suggests that the activity of HrzicN is required for proper suppression of its own transcription.

Development of mesodermal tissues in HrzicN knockdown embryos

As mentioned above, HrzicN was expressed in precursors for mesodermal tissues, and HrzicN knockdown larvae exhibited a shortened tail phenotype without differentiated notochord cells as judged by morphological criteria. Therefore, we investigated development of the notochord and other mesodermal tissues in HrzicN knockdown embryos by examining marker gene expression. First we examined expression ofHrBra (Halocynthia roretzi Brachyury)(Yasuo and Satoh, 1993) to test whether the notochord fate was specified in HrzicN knockdown embryos. Normally, HrBra is expressed in the A- and B-line notochord precursors after the 64- and 110-cell stages, respectively(Fig. 6F). We foundHrBra was not expressed in HrzicN knockdown embryos at the 110-cell stage (Fig. 6A) or at the neural plate equivalent stage (data not shown). We also noticed another abnormality in the gene expression profile of the notochord precursors inHrzicN knockdown embryos. Expression of Hroth, which is excluded from the notochord precursors in normal development(Fig. 5H,I), was observed ectopically in some notochord precursors in HrzicN knockdown embryos(Fig. 5C,D). These suggest that the notochord fate is not successfully specified in HrzicN knockdown embryos.

HrzicN is also expressed in the anterior pair (B8.5 pair) of the two mesenchyme precursor pairs and two (B8.7 and B8.8 pairs) of the five muscle precursor pairs at the 110-cell stage(Fig. 2R). Therefore,expression of the muscle-specific actin gene was examined in HrzicNknockdown embryos at the 110-cell stage. In control embryos, all five pairs of the primary muscle precursors expressed the actin gene(Fig. 6J). This expression pattern was also observed in HrzicN knockdown embryos(Fig. 6E), suggesting that the muscle fate is properly specified in B8.7 and B8.8 pairs withoutHrzicN function.

To test this possibility further and to investigate development of the mesenchyme precursors, we examined expression of Hrsna(Halocynthia roretzi snail) (Wada and Saiga, 1999a) at the 110-cell stage. Normally, Hrsnais expressed in all notochord precursors, two A-line neural tube precursor pairs (A8.15 and A8.16 pairs), both mesenchyme precursor pairs and all primary muscle precursors (Fig. 6H,I). In HrzicN knockdown embryos, Hrsna expression was lost from all the notochord precursors, the two A-line neural tube precursor pairs(A8.15 and A8.16 pairs) and the anterior mesenchyme precursor pair (B8.5 pair;Fig. 6C,D). By contrast,expression in the primary muscle precursors and the posterior mesenchyme precursor pair (B7.7 pair) was unaffected(Fig. 6C,D). This observation strengthens the idea that the muscle fate is properly specified in B8.7 and B8.8 pairs in HrzicN knockdown embryos. Also this result points to a possibility that specification of the anterior mesenchyme precursor pair (B8.5 pair) requires HrzicN function.

In summary, these results suggest that among HrzicN-expressing cells, specification of all notochord precursors and anterior mesenchyme precursors was disturbed, while specification of muscle precursors was unaffected in HrzicN knockdown embryos.

HrzicN overexpression promotes neural development but not notochord development

As described above, HrzicN seems to be essential for development of the neural tube and notochord. To verify this idea, we next carried out overexpression of HrzicN as a complementary experiment. Eggs were injected with approximately 50 pg of HrzicN mRNA, cultured up to the middle gastrula equivalent stage and examined for expression of marker genes. We found that HrETR-1 was expressed in a half of the body ofHrzicN-overexpressing embryos. Judging from the size of cells, the expression domain seemed to correspond to the animal hemisphere and its level of expression was higher than that in control embryos(Fig. 7A,B,D). Thus,HrETR-1 expression was upregulated in HrzicN-overexpressing embryos. This suggests that HrzicN promotes neural development by activating downstream neural genes, directly or indirectly. However,expression of HrBra was not detected in embryos injected with 50 pg of HrzicN mRNA (Fig. 7C,E). Upon injection with approximately 5 or 15 pg ofHrzicN mRNA, HrBra expression was reduced as compared with that in the control embryos (data not shown). Thus, HrzicNoverexpression affects development of the notochord and neural tube differently. HrzicN alone may be insufficient to promote the notochord fate. Alternatively, it is possible that the level ofHrzicN expression must be controlled precisely and/or that temporal down-regulation of HrzicN expression may be important for this gene to promote the notochord fate.

Regulation of HrzicN expression by FGF-like signaling pathway

In ascidians, FGF-like signaling has been thought to be involved in inductive interactions that are responsible for formation of the neural tube,the notochord and the mesenchyme (Darras and Nishida, 2001a; Darras and Nishida, 2001b; Hudson and Lemaire, 2001; Inazawa et al.,1998; Kim and Nishida,2001; Kim et al.,2000; Minokawa et al.,2001; Nakatani and Nishida,1997; Nakatani et al.,1996; Shimauchi et al.,2001). For example, treatment of embryos with a MEK inhibitor,U0126, blocks formation of these tissues(Darras and Nishida, 2001a;Kim and Nishida, 2001). Since our analyses so far showed significant correlation between HrzicNexpression and the neural tube, notochord and mesenchyme fates, we addressed whether HrzicN expression is dependent on FGF-like signaling. Embryos were treated with U0126 from the 1-cell stage onward and fixed at the 76-cell stage or the middle gastrula equivalent stage to examine their HrzicNexpression as well as HrBra expression, which has been shown to depend on FGF-like signaling as control(Nakatani et al., 1996).HrzicN expression in the vegetal hemisphere at the 76-cell stage was normal in U0126-treated embryos (Fig. 8A,E). By contrast, expression of HrzicN in the a- and b-line neural tube precursors at the middle gastrula stage was inhibited in U0126-treated embryos (Fig. 8B,F). As expected, HrBra expression was not detected in U0126-treated embryos at both stages (Fig. 8C,D,G,H). These data indicate that HrzicN expression in the vegetal cells is independent of FGF-like signaling, while HrzicNexpression in a- and b-line neural tube precursors depends on FGF-like signaling.

In the present study, we have isolated a new member of Zic family gene from Halocynthia roretzi. This gene is distinct frommacho-1, the muscle determinant, and plays multiple roles in neural tube formation and notochord development as well as mesenchyme specification as discussed below.

HrzicN is required for maintenance, but not for initial specification, of neural tube fate in the a-line precursors

It has been shown that an inductive signal from the vegetal hemisphere cells is required for formation of the a- and b-line neural tube(Darras and Nishida, 2001b;Nishida and Satoh, 1989;Okado and Takahashi, 1990). The induction likely occurs between the 16-cell and the early gastrula stage and is mediated by FGF-like molecules(Darras and Nishida, 2001b;Hudson and Lemaire, 2001;Inazawa et al., 1998;Kim and Nishida, 2001;Nishida and Satoh, 1989). In response to the induction, specification of the neural tube fate takes place with activation of early neural genes such as HrETR-1 andHroth and suppression of the epidermis fate. HrzicNexpression starts in the a- and b-line precursors later than the onset of expression of these genes and therefore, HrzicN may be unnecessary for specification of the neural tube fate in these lineages. We further tested this possibility by examining gene expression in HrzicN knockdown embryos and found this is the case with the a-line precursors(Fig. 9A). Furthermore, our data suggest that HrzicN in a-line neural cells may be involved in maintenance of the neural tube fate by keeping expression of early neural genes active and by activating late neural genes such as HrTRP(Fig. 9A). This view is supported by the results of the overexpression experiment, showing thatHrzicN enhances HrETR-1 expression. It should be noted,however, that some aspect of the genetic program for sensory vesicle formation may be kept active in the knockdown embryo, because expression ofHrETR-1 did not vanish completely in the putative sensory vesicle region of the knockdown embryo. Since the neural marker genes we used here are not expressed in the b-line neural tube precursors, it was not determined whether this is also the case with the b-line precursors.

In Xenopus, overexpression of any of the Zic family genes caused transformation of the epidermis cells into neural and/or neural crest-derived tissues (Mizuseki et al.,1998; Nakata et al.,1998). Therefore, it has been thought that they are involved in fate choice between epidermal and neural/neural crest fates. This seems to be different from the function of the ascidian Zic suggested above. The reason for this discrepancy is unknown but may be simply becauseHrzicN and Xenopus Zic family genes play different roles during neural fate specification, or it may be due to the difference in methodologies or experimental systems. Further analysis of function of vertebrate Zic genes and HrzicN may resolve this problem.

HrzicN is necessary for specification of the neural tube fate in the A-line precursors

The A-line neural tube cells are derived from A6.2 and A6.4 blastomeres of the 32-cell stage embryo (Fig. 9B). In this lineage, unlike the a-line neural tube precursors,HrzicN seems to be involved in the initial specification of the neural fate, since HrETR-1 was not expressed in A-line precursors ofHrzicN knockdown embryos. Previous experiments showed that A6.2 and A6.4 descendants adopt the neural tube fate autonomously without any cellular interactions (Minokawa et al.,2001). We have shown that vegetal expression of HrzicN is independent of FGF-like signaling, a crucial and multifunctional regulator,occurring around the onset of HrzicN expression. Therefore, a possible model is that HrzicN expression is activated in the A6.2 and A6.4 blastomeres autonomously and this in turn promotes initial specification of the neural tube fate through activation of early neural genes likeHrETR-1 (Fig. 9B).

It has been shown that removal of the A-line neural tube precursors at the 64-cell stage leads to failure in pigment cell differentiation but not in early neural fate specification in the a-line neural tube precursors(Darras and Nishida, 2001b). Therefore, it is possible that disturbance of the specification of the A-line neural tube precursors may be the cause of defects in the development of the a-line neural tube precursors in HrzicN knockdown embryos. However,this cannot fully explain the a-line defects, because our preliminary experiments showed that injection of HrzicNMO into a4.2 blastomeres of the 8-cell stage embryos leads to a failure in the sensory vesicle differentiation similar to that shown in HrzicN knockdown embryos. Defects in development of the a-line neural tube precursors found in HrzicNknockdown embryos may occur as combined consequences of loss ofHrzicN function in both a- and A-line precursors.

HrzicN plays a novel role in notochord formation

We have shown that HrzicN is required for both of the primary and secondary notochord cell formation. It has been shown that HrBraexpression in the notochord precursors and the notochord formation depend on a cellular interaction with the endoderm precursors, which can be mimicked by FGF (Darras and Nishida, 2001a;Kim and Nishida, 2001;Nakatani and Nishida, 1994;Nakatani et al., 1996;Shimauchi et al., 2001).HrBMPb (Halocynthia roretzi BMP2/4) is also involved in this process (Darras and Nishida,2001a). Together with the result that HrzicN expression is independent of FGF-like signaling, we suggest that HrzicN is required for A6.2, A6.4 and B6.2 blastomeres to respond to FGF-like molecules(and/or HrBMPb) emanating from the endoderm precursors, and to activateHrBra, which in turn promotes notochord development after the next cleavage (Fig. 9B,C).

Previous reports showed that all descendants of the A6.2 and A6.4 blastomeres assume the neural fate when they are in isolation and, conversely,they adopt the notochord fate after treatment with human recombinant basic FGF(Minokawa et al., 2001;Nakatani and Nishida, 1994;Nakatani et al., 1996). SinceHrzicN is required for the A6.2 and A6.4 blastomeres to develop into both the neural tube and the notochord, it is possible that HrzicNprompts these blastomeres to pursue the neural fate in the absence of FGF-like signaling while it allows them to follow the notochord fate in the presence of the FGF-like signaling.

Developmental fate of the A6.2 and A6.4 blastomeres in HrzicNknockdown embryos is unclear. It is unlikely that they adopt the epidermis fate because expression of HrEpiG was excluded from them. It is also unlikely that they assume the endoderm fate because endoderm-specific alkaline phosphatase activity was restricted to the original endoderm cells inHrzicN knockdown embryos (data not shown). One possibility is that they remain undifferentiated, although expression of other markers must be examined to verify this possibility.

HrzicN is required for specification of the anterior mesenchyme but not for the primary muscle

It has been shown that the B6.2 blastomere requires FGF-like signaling from the endoderm precursors to form not only the notochord but also the mesenchyme(Kim and Nishida, 1999;Kim and Nishida, 2001;Kim et al., 2000). In the absence of the signaling, all of B6.2 descendants pursue muscle fate possibly because of the action of muscle determinants they inherit(Kim and Nishida, 1999;Kim and Nishida, 2001). We found that HrzicN knockdown led to failure in specification of B8.5 blastomeres into mesenchyme. In this case, the blastomeres did not exhibit a muscle character either. Together with a role of HrzicN in the notochord specification in B8.6 blastomeres discussed above, we suggest thatHrzicN is required for the B6.2 blastomeres to respond to FGF-like molecules to facilitate the notochord and the mesenchyme fates. Suppression of the muscle fate probably occurs independently of HrzicN(Fig. 9C).

Expression of the actin gene and Hrsna was normal in two primary muscle precursor pairs, the B8.8 and B8.7 pairs, irrespective ofHrzicN expression there. This indicates that HrzicN is unnecessary for these cells to adopt muscle fate. It is well known that the primary muscle of ascidian embryos develops autonomously owing to the action of maternally provided cytoplasmic determinants. Recently, macho-1, a muscle determinant has been isolated from Halocynthia roretzi(Nishida and Sawada, 2001). Depletion of the transcripts results in loss of all of the primary muscle cells and overexpression of macho-1 caused ectopic muscle formation in non-muscle-lineage such as endoderm and epidermis. Therefore, it is highly likely that HrzicN is not required for primary muscle development: it is dependent on macho-1(Fig. 9C). However, since bothmacho-1 and HrzicN belong to the Zic family, they may interact to provide the B8.8 and B8.7 pairs with unknown specific characters. Such a possibility should be tested in future study.

The authors thank the staff in Asamushi Marine Biological Station, Tohoku University and the staff in Otuchi Marine Research Center, University of Tokyo for providing us with research facilities. Thanks are also due to Dr Nori Satoh for HrEpiD, HrEpiG, HrTBB2 and HrBra cDNAs. The authors thank Dr Kazuhiro W. Makabe for HrETR-1 cDNA; Dr Hiroaki Yamamoto for HrTRP cDNA; Dr Patrick Lemaire for pBluescriptRN3. This work was supported by a JSPS Postdoctoral Fellowship for Japanese Junior Scientists to S. W. and by Grants-in-Aid from JSPS (12480222) and from the Ministry of Education, Science, Sports and Culture, Japan (13045038) to H. S.

Aruga, J., Minowa, O., Yaginuma, H., Kuno, J., Nagai, T., Noda,T. and Mikoshiba, K. (
1998
). Mouse Zic1 is involved in cerebellar development.
J. Neurosci.
18
,
284
-293.
Aruga, J., Yokota, N., Hashimoto, M., Furuichi, T., Fukuda, M. and Mikoshiba, K. (
1994
). A novel zinc finger protein,zic, is involved in neurogenesis, especially in the cell lineage of cerebellar granule cells.
J. Neurochem.
63
,
1880
-1890.
Aruga, J., Yozu, A., Hayashizaki, Y., Okazaki, Y., Chapman, V. M. and Mikoshiba, K. (
1996
). Identification and characterization of Zic4, a new member of the mouse Zic gene family.
Gene
172
,
291
-294.
Darras, S. and Nishida, H. (
2001a
). The BMP signaling pathway is required together with the FGF pathway for notochord induction in the ascidian embryo.
Development
128
,
2629
-2638.
Darras, S. and Nishida, H. (
2001b
). The BMP/CHORDIN antagonism controls sensory pigment cell specification and differentiation in the ascidian embryo.
Dev. Biol.
236
,
271
-288.
Hikosaka, A., Kusakabe, T. and Satoh, N.(
1994
). Short upstream sequences associated with the muscle-specific expression of an actin gene in ascidian embryos.
Dev. Biol.
1
66,
763
-769.
Hirano, T., Takahashi, K. and Yamashita, N.(
1984
). Determination of excitability types in blastomeres of the cleavage-arrested but differentiated embryos of an ascidian.
J. Physiol.
347
,
301
-325.
Hudson, C. and Lemaire, P. (
2001
). Induction of anterior neural fates in the ascidian Ciona intestinalis.
Mech. Dev.
100
,
189
-203.
Inazawa, T., Okamura, Y. and Takahashi, K.(
1998
). Basic fibroblast growth factor induction of neuronal ion channel expression in ascidian ectodermal blastomeres.
J. Physiol.
511
,
347
-359.
Ishida, K., Ueki, T. and Satoh, N. (
1996
). Spatio-temporal expression patterns of eight epidernis-specific genes in the ascidian embryos.
Zool. Sci.
13
,
699
-709.
Kim, G. J. and Nishida, H. (
1999
). Suppression of muscle fate by cellular interaction is required for mesenchyme formation during ascidian embryogenesis.
Dev. Biol.
214
,
9
-22.
Kim, G. J. and Nishida, H. (
2001
). Role of the FGF and MEK signaling pathway in the ascidian embryo.
Dev. Growth Differ.
43
,
521
-533.
Kim, G. J., Yamada, A. and Nishida, H. (
2000
). An FGF signal from endoderm and localized factors in the posterior-vegetal egg cytoplasm pattern the mesodermal tissues in the ascidian embryo.
Development
127
,
2853
-2862.
Klootwijk, R., Franke, B., van der Zee, C. E., de Boer, R. T.,Wilms, W., Hol, F. A. and Mariman, E. C. (
2000
). A deletion encompassing Zic3 in bent tail, a mouse model for X-linked neural tube defects.
Hum. Mol. Genet.
9
,
1615
-1622.
Lemaire, P., Garrett, N. and Gurdon, J. B.(
1995
). Expression cloning of Siamois, aXenopus homeobox gene expressed in dorsal-vegetal cells of blastulae and able to induce a complete secondary axis.
Cell
81
,
85
-94.
Minokawa, T., Yagi, K., Makabe, K. W. and Nishida, H.(
2001
). Binary specification of nerve cord and notochord cell fates in ascidian embryos.
Development
128
,
2007
-2017.
Miya, T. and Satoh, N. (
1997
). Isolation and characterization of cDNA clones for beta-tubulin genes as a molecular marker for neural cell differentiation in the ascidian embryo.
Int. J. Dev. Biol.
41
,
551
-557.
Mizuseki, K., Kishi, M., Matsui, M., Nakanishi, S. and Sasai,Y. (
1998
). Xenopus Zic-related-1 and Sox-2, two factors induced by chordin, have distinct activities in the initiation of neural induction.
Development
125
,
579
-587.
Nagai, T., Aruga, J., Minowa, O., Sugimoto, T., Ohno, Y., Noda,T. and Mikoshiba, K. (
2000
). Zic2 regulates the kinetics of neurulation.
Proc. Natl. Acad. Sci. USA
97
,
1618
-1623.
Nagai, T., Aruga, J., Takada, S., Gunther, T., Sporle, R.,Schughart, K. and Mikoshiba, K. (
1997
). The expression of the mouse Zic1, Zic2, and Zic3 gene suggests an essential role for Zic genes in body pattern formation.
Dev. Biol.
182
,
299
-313.
Nakata, K., Koyabu, Y., Aruga, J. and Mikoshiba, K.(
2000
). A novel member of the Xenopus Zic family,Zic5, mediates neural crest development.
Mech. Dev.
99
,
83
-91.
Nakata, K., Nagai, T., Aruga, J. and Mikoshiba, K.(
1998
). Xenopus Zic family and its role in neural and neural crest development.
Mech. Dev.
75
,
43
-51.
Nakatani, Y. and Nishida, H. (
1994
). Induction of notochord during ascidian embryogenesis.
Dev. Biol.
166
,
289
-299.
Nakatani, Y. and Nishida, H. (
1997
). Ras is an essential component for notochord formation during ascidian embryogenesis.
Mech. Dev.
68
,
81
-89.
Nakatani, Y., Yasuo, H., Satoh, N. and Nishida, H.(
1996
). Basic fibroblast growth factor induces notochord formation and the expression of As-T, a Brachyury homolog,during ascidian embryogenesis.
Development
122
,
2023
-2031.
Nicol, D. and Meinertzhagen, I. A. (
1991
). Cell counts and maps in the larval central nervous system of the ascidian Ciona intestinalis (L.).
J. Comp. Neurol.
309
,
415
-429.
Nishida, H. (
1987
). Cell lineage analysis in ascidian embryos by intracellular injection of a tracer enzyme. III. Up to the tissue restricted stage.
Dev. Biol.
121
,
526
-541.
Nishida, H. and Satoh, N. (
1989
). Determination and regulation in the pigment cell lineage of the ascidian embryo.
Dev Biol
132
,
355
-367.
Nishida, H. and Sawada, K. (
2001
).macho-1 encodes a localized mRNA in ascidian eggs that specifies muscle fate during embryogenesis.
Nature
409
,
724
-729.
Okado, H. and Takahashi, K. (
1990
). Induced neural-type differentiation in the cleavage-arrested blastomere isolated from early ascidian embryos.
J. Physiol.
427
,
603
-623.
Ruiz i Altaba, A. (
1999
). Gli proteins and Hedgehog signaling: development and cancer.
Trends Genet.
15
,
418
-425.
Sato, S., Toyoda, R., Katsuyama, Y., Saiga, H., Numakunai, T.,Ikeo, K., Gojobori, T., Yajima, I. and Yamamoto, H. (
1999
). Structure and developmental expression of the ascidian TRP gene: insights into the evolution of pigment cell-specific gene expression.
Dev. Dyn.
215
,
225
-237.
Satoh, N. and Jeffery, W. R. (
1995
). Chasing tails in ascidians: developmental insights into the origin and evolution of chordates.
Trends Genet.
11
,
354
-359.
Satou, Y., Imai, K. S. and Satoh, N. (
2001
). Action of morpholinos in Ciona embryos.
Genesis
30
,
103
-106.
Shimauchi, Y., Murakami, S. D. and Satoh, N.(
2001
). FGF signals are involved in the differentiation of notochord cells and mesenchyme cells of the ascidian Halocynthia roretzi.
Development
128
,
2711
-2721.
Summerton, J. (
1999
). Morpholino antisense oligomers: the case for an RNase H-independent structural type.
Biochim. Biophys. Acta
1489
,
141
-158.
Wada, S., Katsuyama, Y., Sato, Y., Itoh, C. and Saiga, H.(
1996
). Hroth an orthodenticle-related homeobox gene of the ascidian, Halocynthia roretzi: its expression and putative roles in the axis formation during embryogenesis.
Mech. Dev.
60
,
59
-71.
Wada, S., Katsuyama, Y., Yasugi, S. and Saiga, H.(
1995
). Spatially and temporally regulated expression of the LIM class homeobox gene Hrlim suggests multiple distinct functions in development of the ascidian, Halocynthia roretzi.
Mech. Dev.
51
,
115
-126.
Wada, S. and Saiga, H. (
1999a
). Cloning and embryonic expression of Hrsna, a snail family gene of the ascidian Halocynthia roretzi: implication in the origins of mechanisms for mesoderm specification and body axis formation in chordates.
Dev. Growth Differ.
41
,
9
-18.
Wada, S. and Saiga, H. (
1999b
). Vegetal cell fate specification and anterior neuroectoderm formation by Hroth, the ascidian homologue of orthodenticle/otx.
Mech. Dev.
82
,
67
-77.
Yagi, K. and Makabe, K. W. (
2001
). Isolation of an early neural maker gene abundantly expressed in the nervous system of the ascidian, Halocynthia roretzi.
Dev. Genes Evol.
211
,
49
-53.
Yasuo, H. and Satoh, N. (
1993
). Function of vertebrate T gene.
Nature
364
,
582
-583.