LvTbx2/3: a T-box family transcription factor involved in formation of the oral/aboral axis of the sea urchin embryo

Cell-fate specification in the sea urchin embryo is achieved through an initial polarization of the unfertilized egg along the animal/vegetal (A/V)axis and, after fertilization, through cell-cell interactions within and between blastomere layers during cleavage and blastula stages. Such polarizations, signaling events and cell-cell interactions activate region-specific genes and segregate the 60-cell stage embryo into five embryonic territories: the small micromeres; the skeletogenic mesenchyme; the vegetal plate endomesoderm; the oral ectoderm; and the aboral ectoderm(Davidson, 1998). Although there is much data about the development of the primary, A/V axis of the embryo (reviewed inAngerer and Angerer, 2000),less is known about the specification of the secondary, O/A axis in the sea urchin. In some species of sea urchins, the O/A axis can be predicted by the plane of first cleavage, whereas in others, the future axis cannot be related to the position of cleavage planes until the third cleavage or later(Cameron et al., 1989;Henry et al., 1992). Signaling events seem also to be involved(Wikramanayake et al., 1995;Wikramanayake and Klein,1997).

Several genes are expressed asymmetrically about the O/A axis in the endoderm, mesoderm and ectoderm, but no genes have yet been identified that reflect the oral/aboral polarity extending between all three. In the endoderm,asymmetrically localized gene products include apical LvNotch protein, which is enriched on the aboral side (Sherwood and McClay, 1997). In the nonskeletogenic mesoderm, CyII actin is distributed orally, and OrCT, CAPK and P1103 aborally(Miller et al., 1996;Rast et al., 2002). O/A patterning has been studied most in the ectoderm and, thus, many gene products with asymmetric distribution about the axis have been identified. Aboral genes include Spec1 and Spec2(Lynn et al., 1983),CyIIIa actin (Cox et al.,1986), arylsulfatase(Sasaki et al., 1988),Hox8/Hbox1 (Angerer et al.,1989), SpMTA (Nemer et al., 1995) and P3A2(Calzone et al., 1991); whereas oral genes include EctoV (Coffman and McClay, 1990) Otx (Li et al., 1997a; Yuh et al.,2002), SpCOUP-TF(Vlahou et al., 1996),BMP2/4 (Angerer et al.,2000), PlOtp (Di Bernardo et al., 1999), Brachyury(Gross and McClay, 2001) andgoosecoid (Angerer et al.,2001).

Based on the expression patterns and likely function of members of the family of T-box genes in regionalization of body plans in other organisms, we hypothesized that they might play a similar role in the sea urchin embryo. Members of the T-box family of transcription factors have been identified in all metazoan organisms in which they have been sought (reviewed byPapaioannou and Silver, 1998;Smith, 1999). T-box genes are characterized by homology to the DNA-binding domain of Brachyury, the founding member of the T-box gene family. The T-box encompasses ∼180 amino acids and can be located anywhere in the protein(Kispert and Hermann, 1993). T-box proteins share little homology outside this region and it is in the T-box that the specificity for target promoters resides(Conlon et al., 2001). The T-box family includes transcriptional activators such as Brachyury, Tbx5,VegT and Eomesodermin(Kispert et al., 1995;Horb and Thomsen 1997;Horb and Thomsen, 1999;Ryan et al., 1996) as well as transcriptional repressors such as Tbx2 and Tbx3 (Carreira et al., 1998; He et al.,1999). The importance of T-box genes in development is underscored by their involvement in a variety of human pathologies, including that ofTbx5 in Holt-Oram syndrome (Basson et al., 1997; Li et al.,1997b), Tbx1 in DiGeorge syndrome(Jerome and Papaioannou, 2001;Merscher et al., 2001),Tbx3 in ulnar-mammary syndrome(Bamshad et al., 1997) and,possibly, Tbx2 in breast cancer(Jacobs et al., 2000).

Here, we report the identification and characterization ofLvTbx2/3, a member of the Tbx2/3 subfamily of T-box genes,during development of the sea urchin embryo. LvTbx2/3 protein is concurrently expressed in the aboral territories of the endoderm, mesoderm and ectoderm. A series of perturbations to the molecular components that are thought to be involved in specifying the O/A axis revealed that the aboral distribution of LvTbx2/3 appears to be a common aspect of O/A specification in each of these tissues. Specifically, LvTbx2/3 expression is dependent on eitherβ-catenin or genes downstream of β-catenin, and is prevented by ventralization with NiCl₂, overexpression of LvBMP2/4 and disruption of the extracellular matrix (ECM). Thus, LvTbx2/3 is expressed downstream of, or relatively late in, the sequence of events that serve to specify this axis. That LvTbx2/3 expression can not be separated between the different tissues after perturbation indicates that O/A axis specification is linked in all three germ layers of the sea urchin embryo at the level ofLvTbx2/3 and may occur in parallel to the distinct specification events that give rise to the ectoderm, endoderm and mesoderm of the embryo. Ectopic expression of LvTbx2/3 supports this conclusion in that universal expression of LvTbx2/3 profoundly affects the morphogenesis of ectoderm, endoderm and mesoderm without altering specification events of embryonic territories. Combined with the loss of expression ofLvTbx2/3 after perturbation of O/A specification, these results indicate that LvTbx2/3 may be a downstream component of the O/A axis program that is involved specifically in morphogenesis of aboral territories in the embryo.

Sea urchins (L. variegatus) were obtained from either Susan Decker(Hollywood, FL) or Jennifer Keller (Duke University Marine Laboratory). Gametes were harvested and cultured at 23°C as described(Hardin et al., 1992).

Degenerate primers were designed that corresponded to the amino acids YIHPDSP (forward)/AVTAYQN (reverse) and used in a PCR reaction with cDNA template prepared from mid gastrula poly(A)⁺ mRNA. PCR conditions were 45 cycles of 96°C for 60 seconds, 40°C for 60 seconds, 72°C for 2 minutes 45 seconds. The amplified, 234 bp products were gel purified,cloned into the pGEMT vector (Promega) and sequenced bidirectionally (Duke Sequencing Core). Clones were identified as PCR products of LvTbx2/3by BLAST search.

Screens were performed essentially as described(Gross and McClay, 2001) with hybridizations performed at 55°C in 0.5 M NaHPO₄ pH 7.2, 1 mM EDTA, 7% SDS, after Church and Gilbert(1984). After rescreens, nine clones were excised, sequenced and identified as LvTbx2/3 fragments. A full-length open-reading frame was defined by overlapping individual fragments.

Northern blotting (RNA gel blot hybridization) for LvTbx2/3 was performed as described (Gross and McClay,2001). Blots were given two 5 minute washes with 6× SSPE,0.5% SDS at room temperature, one 45 minute wash with 1× SSPE, 0.1% SDS at 37°C, and one 45 minute wash with 1× SSPE, 0.1% SDS at 50°C. After washing, the blot was wrapped in plastic wrap and placed on film for 72 hours at -70°C with an intensifying screen. It was then stripped in 50%formamide, 6× SSPE for 30 minutes at 65°C and reprobed as above with an L. pictus ubiquitin fragment as a loading control.

LvTbx2/3 fusion protein was expressed following PCR amplification of aBamHI-XhoI fragment of LvTbx2/3 (encoding amino acids 11-339) and subcloning into the pGEX4T-1 glutathione S-transferase (GST)expression system (AmershamPharmacia Biotech). Expressed, affinity-purified protein (80 μg) was mixed 1:1 with Freund's complete adjuvant and injected into each of three guinea pigs (Charles River, Raleigh, NC). Animals were boosted with 80 μg protein mixed 1:1 with incomplete Freund's adjuvant after 21, 42 and 70 days. Bleeds were performed 31, 53 and 80 days after the last injection and serum isolated as described(Harlow and Lane, 1988).

1500 late-gastrula embryos were homogenized in the presence of protease inhibitors, boiled and run on a 10% SDS-PAGE gel. Protein was blotted onto nitrocellulose, blocked overnight at 4°C in 2% milk, 1% bovine serum albumin (BSA) TBST and probed for 1.5 hours at room temperature with a 1:1000 dilution of either α-Tbx2/3 or preimmune serum in 2% milk, 1% BSA TBST. The blot was washed three times with TBS before applying goat α-guinea pig HRP-tagged secondary antibody (Jackson Immunoresearch Laboratories) at 1:5000 for 1 hour at room temperature. Labeled proteins were visualized by ECL(AmershamPharmacia Biotech).

Embryos were fixed in 2% paraformaldehyde, 60% artificial sea water (ASW)for 10-12 minutes at room temperature, before being permeablized for 60 seconds with ice cold, 100% methanol. They were then washed three times with PBS, blocked 10-20 minutes in PBS, 4% normal goat serum (NGS; GibcoBRL) and incubated overnight at 4°C in primary antibody, 4% NGS. After washing four times in PBS, they were blocked as above and incubated for 60 minutes at room temperature in secondary antibody, 4% NGS (either Cy3 or Cy5-conjugated;Jackson Immunoresearch Laboratories). Embryos were then washed four times in PBS and mounted in 70% glycerol. LvTbx2/3 and LvBrac sera were diluted 1:500 for all images. Undiluted supernatants of monoclonal antibodies (mABs) 5a7(EctoV), 5c7 (Endo1) and 295 were used with the above fixation and incubation conditions. All images were obtained using a 40× Plan-Neofluar oil-immersion objective (NA=1.3) on a Zeiss laser-scanning confocal microscope(Carl Zeiss, Thornwood, NY) mounted on a Zeiss Axiovert inverted microscope. Where necessary, 1 μm sections from single label images were rendered into 3D projections using Zeiss confocal software. Double labeled images were taken sequentially using appropriate filters and subsequently overlayed using Adobe Photoshop 5.0.

Treatment of embryos with either NiCl₂ orβ-aminopropionitrile (βAPN) were performed as described(Hardin et al., 1992;Wessel and McClay, 1987).

Full-length LvTbx2/3 was generated by subcloning fragments from individual excised cDNA clones obtained in library screening (details available on request). For ectopic overexpression studies, an SpOtx5′ UTR plus the first five amino acids of SpOtx was cloned in frame, 5′ to the LvTbx2/3 translation-start site. This leader sequence has been demonstrated to provide an excellent translation start site for mRNA constructs in the sea urchin(Sherwood and McClay, 1999). All clones were sequenced bidirectionally to verify fidelity.

ΔLvG-cadherin and LvBMP2/4 were linearized and injected as described (Logan et al.,1999; Angerer et al.,2000). LvTbx2/3 was linearized with XhoI and used as a template to generate in vitro-transcribed 5′ capped mRNA using the T3 mMessage mMachine kit (Ambion). Concentrations of mRNA were determined by spectrophotometry, and by comparison to known amounts of RNA using both gel electrophoresis and dotting onto a 0.6% agarose gel.

RNA was isolated using Trizol (Invitrogen). Reverse transcription reactions were performed using oligo dT priming and MMLV-reverse transcriptase (Gibco). Reactions were purified using a PCR-purification kit (Qiagen). QPCRs were performed using Roche LightCycler Fast Start Master SYBR as manufacturers instructions. Primers used were ubiquitin(Rast et al., 2000) and LvTbx2/3. A Tbx2/3 plasmid was used to generate a standard curve for quantification, and ubiquitin was used to normalize the cDNA samples. Each time point was determined from two independent batches, and each reaction was confirmed by gel electrophoresis.

LvTbx2/3 was PCR amplified from a mid-gastrula stage cDNA pool using degenerate oligonucleotides that correspond to evolutionarily conserved regions of the DNA-binding domain of other T-box proteins. Cloning and sequencing of the amplified fragment identified it as a L. variegatus Tbx2/3 ortholog (LvTbx2/3). A mid-gastrula cDNA library was then screened and nine LvTbx2/3 cDNA clones recovered. Alignment of the sequences of these clones defined the full coding region of the gene.LvTbx2/3 encodes a 637 amino acid protein, based on the predicted open reading frame from the primary sequence data(Fig. 1A; GenBank accession number AY120889). Supporting LvTbx2/3 as a member of this T-box subfamily, LvTbx2/3 aligns in a phylogenetic tree of Tbx2/3 subfamily proteins(Fig. 1B).

Northern-blot analysis of LvTbx2/3 mRNA revealed that a 5.37 kb message appears first at the mesenchyme blastula stage and that it is present throughout the pluteus stage (Fig. 2). The highest concentrations of mRNA were observed during gastrula stages. Quantitative PCR analysis of the LvTbx2/3 RNA corresponds well with the Northern-blot data. No LvTbx2/3 mRNA is present in the egg but low concentrations start to be detected at the hatched blastula stage. In the mesenchyme blastula there are ∼200 copies ofLvTbx2/3 mRNA per aboral cell and this level is retained until the early prism stage when the number of copies per aboral cell drops. To characterize the temporal and spatial distribution of LvTbx2/3 protein, a polyclonal antiserum was generated in guinea pigs against recombinant LvTbx2/3(amino acids 11-339 fused to GST). This serum was tested for immunoreactivity by protein analysis on SDS-PAGE gels and whole-mount immunofluorescent staining of fixed embryos. Western blots of protein extracts from late gastrula were probed with LvTbx2/3 polyclonal and preimmune sera to ascertain specificity (Fig. 3). Two immunoreactive bands were observed when blots were probed with LvTbx2/3 serum,one of ∼70 kDa and one of 35 kDa. The 70 kDa band corresponds with the predicted size of LvTbx2/3 from primary sequence data (637 amino acids). The 35 kDa band was also recognized when blots were probed with preimmune serum,indicating that it is a nonspecific antigen. Whole-mount immunofluorescent analysis of fixed embryos stained with preimmune serum did not result in any distinct staining pattern at any stage examined(Fig. 4C). Additionally,embryos stained with LvTbx2/3 serum after preincubation with recombinant LvTbx2/3 protein did not stain positively at any stage examined(Fig. 4D).

Fig. 4A,B shows two different orientations of prism-stage embryos costained with anti-LvTbx2/3 antiserum (red) and 5a7 mAB (green). 5a7 recognizes EctoV, a protein that is localized to the foregut and oral ectoderm(Coffman and McClay, 1990). LvTbx2/3 was localized to the nucleus, as expected given its role as a transcription factor. A striking asymmetry of LvTbx2/3 distribution was observed in the ectoderm, endoderm and skeletogenic mesoderm at this stage of development. EctoV and LvTbx2/3 were present in complementary patterns,indicating that this asymmetry is about the O/A axis of the ectoderm and that LvTbx2/3 is restricted solely to the aboral territories of the embryo. The aboral distribution of LvTbx2/3 in the endoderm and mesoderm is apparent inFig. 4B where the protein is clearly localized to the aboral regions of the archenteron and skeletogenic mesoderm. As such, LvTbx2/3 is the first marker of O/A polarity expressed in the derivatives of all three germ layers of the sea urchin embryo. Additionally, the LvTbx2/3 characterization reported here is, to our knowledge, the first report of protein expression for a non-Brachyury T-box gene in any organism.

The ectoderm, endoderm and mesoderm are all specified prior to LvTbx2/3 expression. Because LvTbx2/3 was distributed in a subset of cells in each of these tissues, we next characterized the temporal details of LvTbx2/3 protein expression (Fig. 5). LvTbx2/3 was localized to the nucleus at all stages examined. At mesenchyme blastula stage, LvTbx2/3 protein was observed in cells of the presumptive endoderm and ectoderm but not the mesoderm, as neither the ingressed skeletogenic nor the presumptive nonskeletogenic mesoderm expressed LvTbx2/3 protein when these territories were defined by marker genes (data not shown). A view of the vegetal surface of an early-gastrula stage embryo is shown inFig. 5D. LvTbx2/3 was present at high concentrations in the presumptive endoderm and the ectoderm that surrounds the blastopore, whereas invaginated tissues contained much less protein. Asymmetric distribution in the endoderm and ectoderm continued through mid-gastrula stage (Fig. 5E,F). Between mid-gastrula and late-gastrula stages, LvTbx2/3 started to be expressed in cells of the skeletogenic mesenchyme lineage and the asymmetric localization in the invaginated endoderm became more apparent(Figs 5,6). LvTbx2/3 protein in early and late-plutei embryos is shown in Fig. 5I-K. From an animal view of an early pluteus embryo that has been optically sectioned and projected so that the animal-most ectoderm is removed,asymmetric distribution of LvTbx2/3 was observed in the ectoderm of the embryo and in the archenteron (Fig. 5I). A vegetal projection of a similarly staged pluteus embryo revealed that LvTbx2/3 is present in the ectoderm that surrounds the anus and was very strong in the distal-most portions of the extending embryonic arms(Fig. 5J). Although the concentration of LvTbx2/3 began to decline at the late pluteus stage, it was still observed asymmetrically in the ectoderm, endoderm and skeletogenic mesenchyme (Fig. 5K).

When LvTbx2/3 first appeared in the skeletogenic mesoderm, it was present in all of the PMCs (Fig. 5G,Hand data not shown). However, as development proceeded to the prism and pluteus stages, LvTbx2/3 became restricted to the aboral PMCs (Figs4,5). The spatial and temporal aspects of LvTbx2/3 expression in the skeletogenic mesenchyme were further characterized to determine precisely when this restriction occurs(Fig. 6).Fig. 6A shows the initial,panskeletogenic mesoderm distribution of LvTbx2/3. Two different levels of confocal projections from the same prismstage embryo are shown(Fig. 6B,C). The asymmetric distribution of LvTbx2/3 protein is clearly confined to the aboral PMCs and not present in the ventrolateral clusters. An oblique view of another late-prism stage embryo stained for LvTbx2/3 (red) and 5a7 (green) reinforced this observation, because LvTbx2/3 persists in the aboral ectoderm and endoderm of the embryo whereas no expression is observed in the ventrolateral clusters of PMCs (Fig. 6D). In early-pluteus stage embryos, asymmetric distribution of LvTbx2/3 persists in the PMCs, endoderm and ectoderm (Fig. 6E,F). Thus, in the skeletogenic mesoderm LvTbx2/3 is restricted to the aboral PMCs between late gastrula and early prism stages at the time when skeletal patterning begins to shape the spicule skeleton.

The striking asymmetry of the distribution of LvTbx2/3 about the O/A axis in the endoderm, ectoderm and mesoderm raises the possibility that O/A polarity might be either established or maintained by the same molecular component(s) in all three germ layers of the embryo. To place LvTbx2/3 in the framework of specification pathways and patterning events that impinge on the formation of the O/A axis, and to gain further insights into the mechanisms of O/A axis specification, the distribution of LvTbx2/3, 5a7(Fig. 7A) and LvBrac(Fig. 7B) were examined under a variety of perturbations to this axis.

The influence of β-catenin and β-catenin-dependent signaling on LvTbx2/3 expression was assayed first. Injection of a construct that encodes the cytoplasmic tail of the sea urchin ortholog of E-cadherin,LvG-cadherin (ΔLvG-cadherin)(Logan et al., 1999), serves as a `sink' for cytoplasmic β-catenin by binding to it and preventing nuclear translocation and gene activation. Such embryos develop without endoderm or mesoderm, and β-catenin depletion eliminates the O/A axis in the ectoderm (Wikramanayake et al.,1998; Logan et al.,1999). Expression of this construct resulted in uniform expression of EctoV (Fig. 7C) and prevented the expression of LvTbx2/3 (Fig. 7D). LvBrac and SpGsc (Goosecoid), two gene products that normally localize to the stomodaeum, are also not expressed when the nuclear translocation of β-catenin is prevented(Gross and McClay, 2001;Angerer et al., 2001). This indicates that the blockage of β-catenin nuclear localization prevents both aboral and oral (stomodael) specification.

Treatment of embryos with NiCl₂ at any point between the hatched blastula and early gastrula stages disrupts O/A patterning events(Hardin et al., 1992). Embryos perturbed in this manner are oralized, displaying defects in ectodermal patterning manifested by the formation of a circumferential stomodaeum around the animal pole, rather than at a localized site, and the formation of ectopic spicule clusters. These animals express EctoV and LvBrac around their entire circumference except the vegetal plate(Hardin et al., 1992;Gross and McClay, 2001). Treatment of embryos with 1 mM NiCl₂ resulted in expression of LvBrac throughout the entire ectoderm (Fig. 7E) and elimination of LvTbx2/3 expression in all tissues(Fig. 7F).

Recent evidence indicates that an animally derived BMP2/4 ortholog affects O/A specification (Angerer et al.,2000). In situ analysis localizes BMP2/4 mRNA to presumptive oral ectoderm at the hatching blastula stage. Ubiquitous overexpression of BMP2/4 mRNA animalizes the embryo, causing it to form a ball of squamous epithelium whereas lower concentrations radialize the ectoderm of the embryo, as indicated by the formation of multiple clusters of spicules. At concentrations of BMP2/4 mRNA that radialize the spicules, oral expression of LvBrac was prevented but vegetal LvBrac expression was normal (Fig. 7G). Aboral expression of LvTbx2/3 was not observed in the ectoderm, endoderm or skeletogenic mesoderm under such conditions(Fig. 7H). This indicates that ectopic expression of BMP2/4 prevented the normal expression ofLvTbx2/3 in all three germ layers, and that O/A polarity in the ectoderm, mesoderm and endoderm is linked by some common genetic or molecular mechanism that is likely to be sensitive to changes in BMP2/4 levels. The failure to observe stomodael LvBrac protein after ectopic expression ofBMP2/4 indicates that BMP2/4 signals prevent the expression of a subset of genes in the oral ectoderm and do not uniformly oralize the embryo. Expression of LvBrac is normal in the vegetal blastopore region,indicating that, unlike in the stomodaeum, LvBrac regulation in this region is refractory to ectopic BMP2/4 injected at this level.

Disruption of the ECM with βAPN, a drug that prevents collagen crosslinking implicates the ECM in O/A specification or maintenance. Embryos treated with βAPN do not gastrulate and do not express the aboral-ectoderm-specific Spec1 gene(Wessel et al., 1989). The effects of ECM disruption on LvBrac and LvTbx2/3 expression was assayed(Fig. 7I,J). Neither stomodael LvBrac (Fig. 7I) nor aboral LvTbx2/3 (Fig. 7J) were expressed following treatment with βAPN. This indicates that an intact ECM is necessary for specification and/or maintenance of gene expression in both the oral and aboral territories of the ectoderm, not solely in the aboral territory as previously thought. Normal expression of LvBrac in the vegetal blastopore region indicates that this perturbation did not affect LvBrac regulation in this region.

The results of the perturbation studies detailed above place aboralLvTbx2/3 expression downstream of several signals and specification events that are known to be involved in the formation of the O/A axis. To determine the role of LvTbx2/3 in the formation of this axis, ectopic mRNA expression studies were performed. Ectopic LvTbx2/3 expression produced drastic morphological defects in derivatives of all germ layers,suggesting that LvTbx2/3 functions in each germ layer(Fig. 8). Between 60-75% of embryos that ubiquitously express LvTbx2/3 mRNA displayed severe morphological abnormalities 24-48 hours post-fertilization (three- to fivefold overexpression obtained following injection of 0.75-1 pg/pl of mRNA amounting to 600-1000 copies of LvTbx2/3 mRNA per cell).

By 24 hours, control embryos injected with glycerol had formed a tripartite gut, characteristic skeletal structures and the embryonic shape appropriate for these stages of development (Fig. 8A,B). Embryos that expressed LvTbx2/3 ectopically were often delayed in gastrulation but did invaginate endoderm and gastrulate normally several hours after controls (data not shown). At 24 hours post-injection, embryos injected with LvTbx2/3 mRNA lacked normal skeletal rods and had a grossly mispatterned skeleton with several spicule clusters forming around the circumference of the embryo(Fig. 8C,D). Consistent with the observation that LvTbx2/3 is downstream of both germ-layer specification and O/A patterning events, embryos that ectopically expressedLvTbx2/3 had endoderm, ectoderm, skeletogenic mesoderm, pigment and blastocoelar cells. This indicates that germ-layer specification was not perturbed noticeably, rather, it is likely that aspects of patterning and later morphogenesis were affected. Because ectopic expression of T-box-family members might affect the function of other T-box proteins, these results must be considered cautiously. However, the interpretation that the disruption is at the level of patterning and not at the level of specification appears to be conservative.

At 48 hours post-injection, the skeletons of embryos injected withLvTbx2/3 lacked a consistent pattern, with each embryo elaborating a different, abnormal skeletal phenotype. Two such embryos are presented inFig. 8, and it is clear that,when compared to a normal pluteus-stage embryo(Fig. 8A,B), patterning of the skeletogenic mesoderm was grossly perturbed(Fig. 8E-H). Embryos that expressed LvTbx2/3 ectopically also had severe endodermal defects. In a few cases, exogastrulae were observed following ectopic LvTbx2/3expression (data not shown) but, most often, defects were manifest in an archenteron that had multiple `chambers' rather than a typical tripartite structure. Despite their abnormal morphology embryos stained positively for the Endo1 antigen (5c7), which is normally expressed in the midgut and hindgut(Fig. 8I). Vegetal (blastopore)LvBrac expression in embryos that ectopically expressLvTbx2/3 was also normal, indicating that the endodermal defect was independent of LvBrac in the vegetal plate. In other words, it occurred after gastrulation (Fig. 8J).

It is well established that, in the sea urchin embryo, the skeletogenic mesoderm uses spatial and temporal patterning cues that are localized to the ectoderm to form appropriate skeletal structures (reviewed byMcClay, 1999). The morphological skeletal abnormalities observed in embryos that expressLvTbx2/3 ectopically could result from inappropriate expression of either oral-specific or aboral-specific genes in the ectoderm that are induced by ectopic LvTbx2/3 expression. Thus, downstream patterning cues would also be misexpressed or absent. Embryos were stained either 24 hours(data not shown) or 48 hours after ectopic expression of LvTbx2/3using antibodies against the two markers of oral ectoderm, EctoV and LvBrac(Fig. 8K,L). EctoV expression was confined to one region of the embryo, which indicates that the ectoderm contained an oral territory (Fig. 8K). LvBrac was expressed in a stomodael domain, indicating that substructures in the oral ectoderm were also specified(Fig. 8L). mAb 295 is an antibody that recognizes the ciliary band, a neurogenic region composed of both oral and aboral cells (Cameron et al.,1989). In embryos injected with LvTbx2/3, mAb 295 stained an amorphous region around the embryo, indicating that although oral and aboral territories have been specified and subdivided in the ectoderm, the boundary is not tightly localized (Fig. 8M).

Here, we report the identification and characterization of a novel sea urchin T-box gene, an ortholog belonging to the Tbx2/3 subfamily. The spatial restriction of LvTbx2/3 protein to the aboral regions of each germ layer demonstrates that there is polarized gene expression about the O/A axis in the ectoderm, endoderm and mesoderm of the sea urchin embryo. Because all three tissues share this molecular component, O/A specification in each does not involve totally unique sets of proteins. Perturbations of either molecules or pathways involved in O/A axis formation indicate that LvTbx2/3 acts downstream of these events and may be proximal to the morphogenetic events that shape aboral structures. To our knowledge, no member of the T-box gene family, or of any transcription factor family, has been described that is distributed in such a strikingly polarized manner in the three germ layers of any organism.

That LvTbx2/3 is expressed in the aboral territories of the endoderm, ectoderm and mesoderm provides a point of entry to examine the regulation of gene expression along the O/A axis in each of these tissues. To this end, we perturbed several events that are thought to be involved in patterning this axis and examined the effects on the expression of LvTbx2/3 and LvBrac, markers of aboral and oral gene expression, respectively. Results of these experiments suggest that either β-catenin or the expression of genes downstream of β-catenin is necessary for the expression of both proteins and, thus, gene expression along both the A/V and the O/A axes of the sea urchin embryo. Pharmacologically blocking formation of the aboral axis with NiCl₂ prevented LvTbx2/3 expression in all tissues, suggesting that gene expression along this axis is uniformly sensitive to this perturbation. In addition, it is possible that gene expression of axial information is controlled by mechanisms common to each germ layer rather than through different pathways in each.

The results on protein distribution after ectopic expression ofBMP2/4 are particularly interesting because previous experiments inStrongelocentrotus purpuratus embryos demonstrated that oral expression of EctoV was blocked by ectopic expression of BMP2/4 but the aboral domain of Spec1 expression increased(Angerer et al., 2000). In our study, ectopic expression of BMP2/4 in L. variegatusprevented the expression of both aboral LvTbx2/3 and oral LvBrac(Fig. 7G,H). Thus, two aboral genes, Spec1 and LvTbx2/3 differ in their response to ectopic expression of BMP2/4. The ectoderm of these embryos may be a locked in some sort of pre-aboral ectoderm state in which some aboral proteins are expressed but others are not because the signals necessary for their expression are inhibited by increased concentrations of BMP2/4. The most obvious explanation is that BMP2/4 might be a component of aboral specification, but, given the different Spec1 and LvTbx2/3responses to BMP2/4 perturbation, other, aboral-specification mechanisms must also exist. Evidence for a veg1-derived signal to overlying animal tissues has been observed recently (D.R.M. and J.M.G., unpublished observations), and this signal is sensitive to ectopic BMP2/4 expression. Thus, the defects in O/A gene expression described here might result from a perturbation to this signal. Further characterization of the BMP2/4 pathway,and the identification of more markers for O/A-axis formation in the ectoderm will likely clarify this issue. It is also possible that species-specific differences in specification of the oral and aboral axes might explain this discrepancy. S. purpuratus embryos at least partially differentiate aboral ectoderm autonomously, whereas L. pictus embryos require vegetal signaling to do so (Wikramanayake et al., 1995). Therefore, the loss of LvTbx2/3 expression in embryos of L. variegatus following overexpression ofBMP2/4 could reflect a slightly different role of this pathway in specifying structures along the O/A axis in Lytechinus species of urchins than that in S. purpuratus.

Based on perturbation studies, LvTbx2/3 expression is downstream of the specification of endoderm, mesoderm and ectoderm, including initial O/A specification events in these tissues. When ectopically expressed,LvTbx2/3 consistently produces abnormal morphological phenotypes and patterning deficiencies in derivatives of each tissue. Nevertheless, markers for specific germ layers and axial regions are expressed(Fig. 8). Thus, what is the role of LvTbx2/3 in the aboral territories? Genes downstream of LvTbx2/3 may be involved directly in patterning and morphogenesis, as suggested by the skeletal and endodermal phenotypes that result from the ectopic LvTbx2/3 expression studies presented here. Several other T-box genes have also been noted to have distinct functions during morphogenesis. These include Brachyury in gastrulation movements(Kimmel et al., 1989;Conlon et al., 1996;Wilson and Beddington 1997;Gross and McClay, 2001),Eomesodermin in the formation of bottle cells and initiation of gastrulation (Ryan et al.,1996; Russ et al.,2000), spadetail in paraxial mesoderm migration(Griffin et al., 1998;Yamamoto et al., 1998), andTbx24 in somite segmentation(Nikaido et al., 2002). It will be of great interest to refine the position of LvTbx2/3 in a network of O/A specification when more genes are identified in this gene-regulatory network. Also, using the differential screening technologies that are available currently, it should be possible to identify downstream targets of LvTbx2/3 and determine their roles in patterning and morphogenesis along this axis.

We are indebted to Drs. Dave Sherwood (CalTech) and Cyndi Bradham for discussion of these studies and invaluable comments and criticisms on the text of this manuscript. Specifically, we are grateful to Cyndi Bradham for advice on QPCR experiments. We would also like to thank Robert E. Keen for technical support throughout the injection phase of this work. This work was supported by NIH grants HD 14483 and GM64164 to D.R.M. and Department of Defense Breast Cancer Research Program grant BC990657 to J.M.G.