A sea urchin homologue of the chordate Brachyury (T) gene is expressed in the secondary mesenchyme founder cells

The molecular mechanisms involved not only in ontogeny but also in phylogeny need to be elucidated. The discovery of a shared function of the HOM-C/Hox cluster genes in determining the anteroposterior axis of insects and vertebrates, permits investigation of how the widely divergent animal body plans have formed during evolution (e.g. Slack et al., 1993; Akam et al., 1994; Manak and Scott, 1994). Here, we investigate the molecular developmental mechanisms involved in the origin and evolution of chordates.

The evolutionary pathway from advanced invertebrates through primitive chordates to vertebrates has been a subject of extensive investigation and vigorous discussion for more than a century (e.g. Haeckel, 1868; Garstang, 1928; Berrill, 1955; Løvtrup, 1977; Jefferies, 1986; Willmer, 1990). Chordates (urochordates, cephalochordates and vertebrates) are categorized as deuterostomes together with two invertebrate groups, echinoderms and hemichordates (e.g. Brusca and Brusca, 1990). Deuterostomes share many features, such as radial cleavage, the fate of the blastopore that does not form a mouth and an enterocoelic coelom; in addition, chordates show several common embryological and anatomical features, including a notochord, a dorsal hollow nerve cord and pharyngeal gill slits (e.g. Brusca and Brusca, 1990; Willmer, 1990). Of these features, the notochord may be the most prominent characteristic, as the name ‘Chordata’ is derived from this organ. Molecular phylogenetic studies (Wada and Satoh, 1994; Turbeville et al., 1994) as well as a cladistic analysis (Schaeffer, 1987) suggest that echinoderms, hemi- chordates and chordates form a monophyletic group of deuterostomes. Therefore, the organization of the notochord during invertebrate evolution is of salient importance in attempts to understand chordate ontogeny and phylogeny (Satoh and Jeffery, 1995; Yasuo et al., 1995).

Recent cloning and characterization of the mouse Brachyury (T) gene by Herrmann et al. (1990) have advanced the under- standing of molecular mechanisms involved in the notochord formation in vertebrates (see the review by Herrmann and Kispert, 1994). The T gene is expressed both in the notochord and in the area from which the mesoderm is generated (Wilkinson et al., 1990). Loss of function of the T gene leads to a disturbance in the development of the primitive streak with the most affected region being the notochord (Chesley, 1935; Grüneberg, 1958; Yanagisawa et al., 1981). In addition, a single-copy transgene representing the wild-type T allele can rescue T-associated phenotypes expressed as a defective tail (Stott et al., 1993). Homologues of the mouse T gene have been isolated from Xenopus (Xbra; Smith et al., 1991) and zebrafish (Zf-T; Schulte-Merker et al., 1992). Zebrafish no tail is a mutation of this gene (Halpern et al., 1993; Schulte-Merker et al., 1994). In addition, an ascidian T gene (As-T) is expressed exclusively in blastomeres of notochord lineage and the timing of initiation of As-T transcription coincides with that of a restriction of developmental fate (Yasuo and Satoh, 1993, 1994). All of these studies demonstrated that the T gene is conserved in primitive chordates (ascidian) and vertebrates and that it is closely associated with notochord organization. Therefore, T may be a key gene with which to probe the origin and evolution of chordates from a molecular-embryological perspective.

We examined whether or not the T gene is expressed in echinoderms, a group of deuterostomes that do not have a notochord. A T-related gene (Trg) has been identified in Drosophila and two other insects; it is expressed in the hindgut primordium at the blastoderm stage, then in the differentiating hindgut (Kispert et al., 1994). This finding, together with the conservation of the T gene in chordates, however, does not prove the presence of the T gene in echinoderms and hemi- chordates. This study seeks to show that a T gene homologue is present and expressed in sea urchin embryos. Both cell lineage studies and cell-type-specific gene expression analyses have demonstrated that, in the sea urchin embryo, five territories for founder cells can be distinguished (e.g. Davidson, 1989; Cameron and Davidson, 1991). Therefore, the territory expressing the T gene can be determined. A comparison of the embryonic regions that express the T gene among an echinoderm, an ascidian and vertebrates may indicate how the T gene changed its function during the emergence of chordates.

Hemicentrotus pulcherrimus was collected during the spawning season near the Ushimado Marine Laboratory, Okayama University, Okayama, Japan. Gametes were handled and fertilized eggs were cultured according to standard procedures. Embryos were raised in filtered seawater at about 15 °C. The embryos divided about every hour and became mesenchymal blastulae 24 hours after fertilization. Gastrulae, prism larvae and pluteus larvae were observed about 35, 45 and 60 hours after fertilization, respectively.

Eggs, embryos and larvae at appropriate stages were collected by low-speed centrifugation and frozen quickly in liquid nitrogen for northern blotting or they were fixed for in situ hybridization as whole- mount specimens.

Amino acid sequences of so-called T domain of the T gene products are highly conserved among mouse (Herrmann et al., 1990), Xenopus (Smith et al., 1991), zebrafish (Schulte-Merker et al., 1992) and ascidians (Yasuo and Satoh, 1994). The sense-strand oligonucleotide that corresponds to the amino acid sequence YIHPDSP and the antisense oligonucleotide that corresponds to the amino acid sequence NPFAKG(A)L(F) were synthesized using an automated DNA syn- thesizer (Applied Biosystmes Inc., Foster City, CA, USA). Using these oligonucleotides as primers, we amplified target fragments from an H. pulcherrimus gastrula cDNA library by means of PCR. The library was a gift from Drs H. Shimada and K. Akasaka in Hiroshima University. Probing with candidate cDNA fragments random-labeled with [³²P]-dCTP (Amersham), we screened the library at high strin- gency (hybridization: 6× SSPE, 0.1% SDS, 1× Denhardt’s solution, 50% formamide at 42 °C; washing: 2× SSC, 0.1% SDS at 65 °C). Among three cDNA clones that were isolated to search for the sea urchin T gene, the longest was subcloned into pBluescriptII SK(+). Both strands of the clones were sequenced by dideoxy chain termi- nation (Sanger et al., 1977) using Sequenase ver. 2.0 (USB; United States Biochemical Corp., Cleveland, OH, USA).

Total RNA was extracted from specimens at various stages using acid guanidinium thiocyanate-phenol-chloroform (AGPC; Chom- czynski and Sacchi, 1987). 20 μg of total RNAs were fractionated by electrophoresis on a 1% agarose gel containing 6% formaldehyde and RNA bands were transferred to a Hybond-N+ nylon membrane (Amersham). The membranes were prehybridized in 6× SSPE, 0.5% SDS, 5× Denhardt’s solution, 50% formamide, and 100 μg/ml salmon sperm DNA for 2 hours. Random-primed ³²P-labeled DNA probes were then incubated with the membranes overnight at 42°C. The length of the probe for both northern and Southern hybridizations was about 2.5 kb between positions 671 and 3189 of the cDNA clone shown in Fig. 1. The membranes were washed under high- stringency conditions (twice in 2× SSC, 0.1% SDS at 65 °C for 15 minutes).

High molecular weight H. pulcherrimus genomic DNA was provided from Dr H. Shimada. After exhaustive digestion with EcoRI, BamHI and HindIII and electrophoresis, the DNA fragments were blotted onto Hybond-N+ nylon membranes (Amersham). The blots were hybridized with [³²P]dCTP-labeled DNA probes at 42 °C for 18–24 hours and washed under high-stringency conditions.

Whole-mount specimens were hybridized in situ basically as described by Ransick et al. (1993). Embryos were fixed in 4% paraformaldehyde in 0.5 M NaCl, 0.1 M MOPS pH 7.5 on ice for overnight prior to storage in 80% ethanol at −20 °C. The fixed specimens were thoroughly washed with PBT (phosphate-buffered saline containing 0.1% Tween 20), and partially digested with 2 μg/ml proteinase K (Sigma) in PBT for several minutes at 37 °C. After washing with PBT, the specimens were postfixed with 4% paraformaldehyde in PBT for 1 hour at room temperature, then washed again with PBT. Specimens were then treated with PBT:hybridization buffer (1:1, v/v) for 10 minutes at room tempera- ture and with hybridization buffer alone for 10 minutes at room temperature. After a 1 hour prehybridization at 42 °C, the specimens were hybridized with digoxigenin(DIG)-labeled antisense probes for about 16 hours at 42°C. The hybridization buffer contained 50% formamide, 10% PEG#6000, 2× Denhardt’s solution, 500 μg/ml yeast RNA, 0.6 M NaCl, 5 mM EDTA, 20 mM Tris-HCl pH 7.5 and 0.1% Tween 20. The probe was synthesized following the instructions supplied with the kit (Boehringer Mannheim DIG RNA Labelling kit), and the hybridized fragments were reduced to about 200-500 nucleotides by alkaline hydrolysis. The probe was used at 0.1 μg/ml in the hybridization buffer.

After hybridization, the specimens were washed in 1× SSC for 15 minutes at 65 °C, three times. This solution was replaced with PBT at room temperature. The DIG label was detected as follows. The specimens were incubated with 5% goat serum in PBT for 30 minutes at room temperature, then with 1:1000 Boehringer Mannheim alkaline-phosphatase-conjugated anti-DIG in PBT. Washes proceeded in PBT (10 minutes, four times), then in alkaline phosphatase buffer (100 mM NaCl, 50 mM MgCl₂, 0.1% Tween 20, 100 mM Tris-HCl pH 9.5) before signal detection using NBT and BCIP following the supplier’s instructions. The reactions were stopped in PBT after 30 minutes to 3 hours, then the embryos were dehydrated in a graded series of ethanol solution and cleared by placing them in a 1:2 mixture (v/v) of benzyl alcohol:benzyl benzoate.

As a technical control, Strongylocentrotus purpuratus prism larvae were hybridized with CyIIa probe. The samples and probe are a gift from Dr Eric Davidson’s laboratory in California Institute of Tech- nology.

The chordate T gene products consist of two distinguishable domains (Herrmann et al., 1990; Smith et al., 1991; Schulte- Merker et al., 1992; Yasuo and Satoh, 1994). The N-terminal half is highly conserved among the T proteins, while the degree of conservation is not high in the C-terminal region. In particular, a stretch of about 200 amino acids, known as the T domain, in the N terminus show extensive similarity. This T domain confers specific DNA binding (Kispert and Herrmann, 1994). Using oligonucleotide primers corresponding to the shared sequences, we amplified target fragments from H. pul- cherrimus gastrula cDNA by the PCR reaction. Sequencing the amplified 300 bp-long fragments after subcloning them into pBluescriptII SK(+), revealed that the sea urchin library contains at least two independent cDNA clones, with sequences similar to that of the chordate T gene (data not shown). Genomic Southern analysis suggested that these sequences corresponded to two independent genes in the sea urchin genome (data not shown; cf. Fig. 3). We therefore designated the genes for the clones, HpTa and HpTb, respectively. Northern analysis indicated that HpTa and HpTb transcripts were about 3.7 kb (cf. Fig. 4) and 5 kb (data not shown) in length respectively. In this study, we further analysed HpTa, because the nucleotide sequence of the HpTa gene resembled that of the mouse T gene more than HpTb.

Screening of 2.7×10⁵ pfu of H. pulcherrimus gastrula cDNA library with the random-labeled HpTa fragment, yielded three positive cDNA clones. Sequencing the longest showed that it consisted of about 3,700 nucleotides, suggesting that the clone was nearly the entire length of the transcript. As shown in Fig. 1, the cloned fragment contained a single long, open reading frame of 1,302 nucleotides that encode a polypeptide of 434 amino acids. The open reading frame extended from nucleotides 349 to 1,651 of the fragment. The calculated relative molecular mass (M_r) of the protein was 46×10³. A putative polyadenylation signal was found at position 3,659- 3,664.

Fig. 2 shows a comparison of the amino acid sequences of the T domain of HpTa with those of the putative proteins encoded by the mouse T (Herrmann et al., 1990), Xenopus Xbra (Smith et al., 1991), zebrafish Zf-T (no tail; Schulte- Merker et al., 1992), ascidian As-T (Yasuo and Satoh, 1994) and Drosophila Trg (Kispert et al., 1994) genes. Although the overall degree of amino acid identity was not very high (52%, sea urchin/mouse; 41%, sea urchin/ascidian), in the T domain shown in this figure, the full extent of amino acid identity was 73% (sea urchin/mouse), 73% (sea urchin/frog), 75% (sea urchin/zebrafish) and 68% (sea urchin/ascidian). The high degree of identity in the T domain between the sea urchin protein and the mouse T, Xbra, Zf-T and As-T proteins demonstrated that this cDNA clone corresponds to the sea urchin homologue of the chordate T gene.

We determined the number of different sequences that corresponded to HpTa in the sea urchin genome by means of genomic Southern hybridization. As shown in Fig. 3, one major band and a few minor bands were detected in each lane of DNAs digested with EcoRI, BamHI and HindIII, suggest- ing that HpTa is present as a single copy gene and that the sea urchin genome contains T domain genes other than HpTa. In addition, HpTa and HpTb had completely different hybridization patterns (data not shown).

Northern blotting of the HpTa gene during embryogenesis of H. pulcherrimus revealed that the HpTa transcripts are transient.

Hybridization signals were undetectable in unfertilized eggs and early embryos at the cleavage and early blastula stages (Fig. 4). This result was confirmed by longer exposure. A weak but distinct hybridization signal was first detected in hatching blastulae. A distinct band was found at the mesenchymal blastula and gastrula stages, at which time the intensity of the band was maximal. The intensity of the band decreased after the gastrula stage. Although a band was detected in prism larvae, it was barely detectable at the pluteus-larva stage (Fig. 4). This pattern of transient HpTa expression coincides with that of the chordate T genes.

Cell lineage analyses of sea urchin embryos have demonstrated that unique sets of founder cells generate five territories of gene expression by means of an invarient pattern of complete cleavage (e.g. Davidson, 1989; Cameron and Davidson, 1991). The five territories and cell types are aboral ectoderm, oral ectoderm, ectoderm-derived neurons, primary skeletogenic mesenchyme cells and the definitive vegetal plate. The vegetal plate cells after gastrulation, contribute to the endoderm and four types of secondary mesenchyme, namely basal cells, coelomic pouches, pigment cells and circumesophageal muscle. We studied which territories or cell types express HpTa by means of in situ hybridization of whole-mount specimens.

Control embryos hybridized with a sense probe did not show signals above the level of background (Fig. 5B,D,G,I). Sea urchin embryos processed for in situ hybridization were fragile, particularly those up to the swimming blastula stage. Most embryos broke down during the later half of hybridization process. The first stage at which we obtained a clear result was that of the mesenchyme blastula, when the distribution of hybridization signals was not even but restricted to cells of the vegetal plate (Fig. 5A). At the early gastrula stage, the signals were found in cells of the invaginating tip of the vegetal plate or newly formed archenteron (Fig. 5C). Neither aboral and oral ectoderm nor primary mesenchyme cells that have delaminated from the vegetal wall into the blastocoel showed signals above the background (Fig. 5A,C).

At the mid-gastrula stage, the archenteron has extended to nearly a half of the blastocoel and two islands of skeletalgenetic sites are formed by the primary mesenchyme cells (Fig. 5E,F). Hybridization signals were found in cells of extending tip of the archenteron and secondary mesenchyme cells migrating from there (Fig. 5E,F). In contrast, signals were not so evident in cells of the middle and posterior parts of the archenteron (Fig. 5E,F).

At the late gastrula stage, the archenteron has almost reached the wall of the most anterior part of the oral ectoderm (Fig. 5H) and some secondary mesenchyme cells have separated from the archenteron wall into blastocoel. Intense hybridization signals were detected in the secondary mesenchyme cells and cells of the archenteron tip, whereas the cells of the archenteron showed barely detectable signals (Fig. 5H).

At the prism larval stage, the secondary mesenchyme cells are scattered over the inner wall of blastocoel. At this stage, hybridization signals were not detected above the background level (data not shown). Fig. 5J is a control in which the CyIIa gene in Strongylocentrotus purpuratus embryos is expressed in the primary mesenchyme cells. As development proceeded, we could not detect the distinct signal above the background level at the late prism and the pluteus larval stages (data not shown).

Thus, we concluded that the HpTa gene is expressed in embryonic cells that give rise to the secondary mesenchyme cells. As mentioned before, the vegetal plate cells give rise to the archenteron (larval gut) and the secondary mesenchyme cells. HpTa was expressed in the secondary mesenchyme founder cells, but not in the gut cells. In addition, the secondary mesenchyme cells give rise to four types of cells, circumesophageal muscles, pigment cells, basal cells and coelomic pouches. However, further differentiation of the secondary mesenchyme cells into these four types occurs after the prism stage. At this stage, the hybridization signals were undetectable. Therefore, we were unable to identify which components of the secondary mesenchyme are the positive founder cells of all four types in this study.

This study revealed that the sea urchin Hemicentrotus pul- cherrimus contains homologues (HpTa and HpTb) of the chordate T gene. HpTa is expressed transiently at the blastula and gastrula stages and the distribution of HpTa transcripts is restricted to the secondary mesenchyme founder cells.

Molecular phylogenetic studies based upon a comparison of 18S rDNA sequences (Wada and Satoh, 1994; Turbeville et al., 1994) and a cladistic analysis based on the comparison of embryological and morphological features (Schaeffer, 1987) suggest that echinoderms, hemichordates and chordates are monophyletic and, thus, have evolved from a common ancestor. As shown in this study, T gene expression is conserved in the sea urchin embryo. Recent studies in our own laboratory (Terazawa and Satoh, 1995) and from Herrmann’s group (Dr B. G. Herrmann, personal communication) have identified an amphioxus homologue of the T gene, which is expressed in the mesoderm of the gastrula. Although a hemi- chordate homologue of the chordate T gene should be identified, it is highly likely that the T gene is conserved and expressed in embryos of all deuterostome groups. In other words, the T gene was organized in the common ancestor of deuterostomes.

Therefore, the T gene is unlikely to have originated along with notochord organization during the emergence and evolution of chordates. Instead, the gene had some functions in embryogenesis, for example in the formation of mesoderm, and the gene or a duplicate was used for notochord formation during chordate evolution (Yasuo et al., 1995). In addition, the T-related gene (Trg) has been identified in Drosophila, Tribolium and Locusta (Kispert et al., 1994). This suggests that the root of this gene extends to more primitive animals. Because the T domain is involved in DNA binding and in the regulation of trascriptional activities of downstream genes, the function of this gene might be more fundamental than previ- ously assumed (Kispert et al., 1994).

With respect to the wide distribution of the T gene in various animal groups, the T domain genes may constitute a family as pointed out by Kispert et al. (1994). In Drosophila, the T domain was identified in the gene optomoter blind (omb) (Pflugfelder et al., 1992). Here, we isolated another T domain gene, HpTb, from the sea urchin gastrula cDNA library. We are now examining HpTb gene expression to obtain more exact understanding of the sea urchin T genes. In addition, an ascidian genome contains two T domain genes other than the As-T gene (Yasuo and Satoh, unpublished). A preliminary molecular phylogenetic tree constructed by a comparison of sequences of about 50 amino acids in the highly conserved T domain among chordates and insects, suggests that in the T- gene family, a group of T genes (chordate T and insect Trg) branched from that of omb (ascidian omb-like gene and omb) (Yasuo et al., 1995). The HpTa gene was included in the former group while the HpTb was diverged one. Because some T genes such as HpTb have diverged from these two groups, there should be other groups within the T-gene family.

In vertebrates, T gene expression begins in both prospective posterolateral mesoderm and notochord cells (Herrmann and Kispert, 1994). In contrast, the spatial expression of the ascidian As-T is restricted to the primordial notochord cells and As-T was not expressed in other mesodermal precursor cells that give rise to mesenchyme and muscle (Yasuo and Satoh, 1993, 1994). What this difference means is uncertain at present. Regardless, it is highly likely that HpTa is orthologus with the chordate T genes. Because HpTa is expressed in the secondary mesenchyme founder cells and the chordate T is expressed in the notochord, we question whether there is any phylogenetic relationship between the secondary mesenchyme and notochord. However, secondary mesenchyme cells have no known features of molecular differentiation in common with notochord cells, and no developmental function or fate in common with notochord. It will require analysis of genes that are upstream and downstream with respect to T gene in these organisms. In addition, we should analyze sea urchin homologues of genes that are expressed in the vertebrate notochord, for example, the HNF3-related genes (e.g. Sasaki and Hogan, 1993; Ang and Rossant, 1994), to support further argumentation.

HpTa gene expression begins around the time of hatching of the embryo. At this stage, the secondary mesenchyme founder cells are buried in the vegetal plate. The earliest feature of the differentiation of secondary mesenchyme cells reported to date is the separation of the cells from the anterior tip of the invaginating archenteron. Specific antigens are expressed in various derivatives of the vegetal plate territory. For example, an endoderm antigen is expressed prior to archenteron invagination through- out the vegetal plate and later in the cells of intestine and stomach (Wessel and McClay, 1985). Endo16 is expressed in cells of the vegetal plate of S. purpuratus blastulae and in cells of the archenteron of gastrulae (Ransick et al., 1993). Therefore, Endo16 expression is an early vegetal plate marker. However, the archenteron tip cells did not express Endo16 (Ransick et al, 1993), where HpTa was expressed. Because HpTa expression is restricted to secondary mesenchyme founder cells and it begins early, HpTa transcripts might be used as an early marker of differentiation of these cells.

The mechanisms of specification of sea urchin secondary mesenchyme cells are not fully understood (Davidson, 1989), although the vegetal plate cells are specified by cell-cell inter- action (e.g. Ransick and Davidson, 1993). The secondary mesenchyme cells take the place of the primary mesenchyme cells when sea urchin embryos are manipulated to deprive the primary mesenchyme cells from the blastocoel (Ettensohn, 1992). However, the detailed mechanism of secondary mesenchyme specification remains to be determined. As demon- strated with the vertebrate T gene (Herrmann and Kispert, 1994), the product of the HpTa gene may function as a gene regulatory factor involved in the specification of the secondary mesenchyme cells. The ectopic expression of the HpTa gene or the injection of antisense RNA into the vegetal plate cells may help answer this question.

We thank Drs Hiraku Shimada and Koji Akasaka for supplying Hemicentrotus pulcherrimus cDNA library and genomic DNA. We also thank Dr Eric H. Davidson for S. purpuratus specimens and CyIIa probe. This research was supported by a Grant-in-Aid for Specially Promoted Research from the Ministry of Education, Science and Culture, Japan and from the Sumitomo Foundation to N. S.; H. Y. was supported by a Predoctoral Fellowship from the Japan Society for the Promotion of Science for Japanese Junior Scientists with Research Grant 3082.

The nucleotide sequence data reported in this paper will appear in the DDBJ, EMBL and GenBank Nucleotide Sequence Databases with the accession number (D50332).