Xenopus VegT RNA is localized to the vegetal cortex during oogenesis and encodes a novel T-box transcription factor involved in mesodermal patterning

The basic spatial organization of the amphibian embryo is established early in development through the sequential formation of localized centers with inductive capacities. Evidence suggests that the first such center to be elaborated is the oocyte’s vegetal cortex, a region rich in cytoskeletal com-ponents and localized RNAs (reviewed by King, 1995; Elinson et al., 1993; Forristall et al., 1995). The initial event that intro-duces a dorsoventral axis into the radially symmetrical egg is the movement of cortex material relative to the deep ooplasm before first cleavage (reviewed by Gerhart et al., 1989). Cyto-plasmic transfer and UV rescue experiments (Gimlich and Gerhart, 1984; Gimlich, 1986) indicate that dorsal determinants are specifically localized in the vegetal cortex and that cortical rotation moves and/or modifies the determinant on the future dorsal side (Vincent et al., 1986; Holowacz and Elinson, 1993; Fujisue et al., 1993). Cortical rotation sets up a vegetal dorsalizing domain, the Nieuwkoop center, which consists of a small group of dorsovegetal cells (reviewed by Gerhart et al., 1989). These cells are competent to induce overlying marginal zone cells to form the third center, Spemann’s organizer, which is capable of organizing a complete second axis when grafted to the ventral side (Spemann, 1938). The organizer contains information for patterning both anteroposterior and dorsoventral mesoderm (reviewed by Dawid, 1994).

Results from dominant negative receptor experiments have strongly indicated that some members of the TGFβ family such as BVg1 or activin are involved in dorsal mesoderm induction (Hemmati-Brivanlou and Melton, 1992), while members of the FGF family and BMP4 are required for ventral mesoderm formation (Amaya et al., 1991; Graff et al., 1994). Other genes have been shown to modify the mesoderm, including Xwnt-8, noggin, goosecoid, chordin and nodal-related genes (reviewed by Kessler and Melton, 1994; reviewed by Harland, 1994; Jones et al., 1995). Although much information is available about mesoderm-inducing factors, little is known about how the initial signals are transduced to activate the early genes after MBT. It has been shown that activin can activate a maternal factor long before MBT (Huang et al., 1995). Recently identified Mad proteins are downstream effectors of mesoderm-inducing factors, but whether Mads directly activate immediate early genes has yet to be determined (Graff et al., 1996). Clearly, an important goal is to understand the relationship between maternal prepatterning and zygotic gene expression that results in cell type and tissue differentiation.

Although an inductive event is required for mesoderm formation (Gimlich, 1986), induction does not account for the observation that both goosecoid and Xwnt-8 can be activated in the correct dorsoventral region of the embryo in the complete absence of cell-cell interactions (Lemaire and Gurdon, 1994). Lemaire and Gurdon (1994) have suggested that formation and patterning of mesoderm requires the simul-taneous activation of two different pathways: an inductive pathway and a cell-autonomous pathway. The maternal factors that could participate in the later pathway are unknown.

Differential screening and expression cloning have been used to identify genes that are good candidates for involvement in the earliest steps of embryonic patterning. Our strategy was to select maternal transcripts localized in the vegetal cortex by differential screening and then to test for patterning activity. This approach has yielded VegT, a new member of the T-box family whose original member was the mouse Brachyury gene (reviewed by Herrmann and Kispert, 1994). Brachyury, in both mouse and zebrafish, functions as a DNA-sequence-specific transcription factor and is important in axial development (Kispert et al., 1995). The VegT expression pattern is dynamic along the animal/vegetal axis during oogenesis and, later, along the dorsoventral axis in the marginal zone. Unlike the Brachyury members of the T-box family, VegT is never expressed in the notochord. We show that VegT is a nuclear protein and that its C-terminal region can activate transcription in a yeast reporter system. We also show that ectopic expression of VegT can induce Xwnt-8 expression and converts prospec-tive ectoderm into ventral mesoderm. Mis-expression of localized VegT in dorsal animal blastomeres results in the sup-pression of head formation. The results presented here support a role for VegT as a localized transcription factor that patterns mesoderm along the dorsoventral and posterior axis in Xenopus.

An adult ovary was treated with collagenase to obtain oocytes free of follicle cells. Embryos were obtained by in vitro fertilization, dejellied and cultivated in 0.1× Modified Barth Saline with Hepes (MBSH) (Gurdon, 1977). Embryos were staged according to Nieuwkoop and Faber (1967). Animal cap explants were isolated at stage 8 and cultured in isolation in 0.5× MBSH until the desired stage.

To prepare cDNA probes for differential screening, ten vegetal or animal cortices were isolated from stage VI oocytes and total cortical RNA was prepared as described previously (Elinson et al., 1993). The synthesis and PCR amplification of cortical cDNA were performed according to Brady et al. (1990) with the following modifications. RNA equivalent to three cortices was denatured at 70°C for 10 minutes and then added to reverse transcription buffer (50 mM Tris HCl [pH 8.3], 75 mM KCl, 3 mM MgCl_2, 0.25% NP-40, 10 mM DTT, containing 250 ng/ml primer-adaptor XC1: CTC GAG ACG CTG TCT AGA (T)₁₅, 2 U/μl RNasin [Promega], and 2.5 mM each of dATP, dGTP, dCTP and dTTP). Reverse transcription was initiated by adding 200 U of SuperScript™ II RNase H⁻ reverse transcriptase (Gibco BRL) followed by incubation at 37°C for 20 minutes. Reaction mixtures were inactivated at 70°C for 10 minutes. One third of the mixture was used in the poly(dA)-tailing reaction carried out according to the manufacturer’s instructions (Boehringer). One third of the poly(dA)-tailed cDNA was brought to 100 μl in PCR buffer: 10 mM Tris-HCl [pH 9.0], 50 mM KCl, 2.5 mM MgCl₂, 100 mg/ml bovine serum albumin, 0.1% Triton X-100 and containing 0.5 mM of dATP, dGTP, dCTP, dTTP, 10 U of Taq polymerase and 5 ng of the primer-adaptor XC-1. PCR amplification was performed as follows: 94°C (6 minutes), 42°C (3 minutes) and 72°C (6 minutes) followed by an additional 45 cycles: 94°C (1 minute), 53°C (2 minutes) and 72°C (6 minutes plus 10 seconds extension for each cycle). The PCR products were analyzed by Southern blot hybridization. The blot was probed with α-³²P-labelled Xcat-3, a vegetally localized RNA, to ensure cortical RNAs were amplified in a representative manner (Elinson et al., 1993). Amplified cDNAs were labeled with [α-³²P]dCTP by random priming (Boehringer). Vg1, Xcat-2 and Xcat-3 cDNA probes were synthesized, mixed together and used to screen the oocyte cDNA library.

An amplified oocyte cDNA library (7,000 pfu) 50-fold enriched for vegetally localized mRNAs (see Mosquera et al., 1993) was plated and lifted onto triplicate filters (NEN). The filters were hybridized with the animal, vegetal and the above mixed probes, respectively. Those plaques were selected as positive clones that gave a strong signal with the vegetal cortical probe and a weak to no signal with the animal cortical and mixed probes. 35 clones were positive after the second round of screening. These clones were classified into five groups by cross-hybridization using dot blot analysis and partial DNA sequencing. Three clones of VegT were isolated, the longest one was 1.7 kb (clone B12). Northern blot analysis indicated that the endogenous transcript was approximately 3.0 kb in length. Additional screening of the same library failed to recover a longer clone.

To obtain a full-length cDNA for VegT, 5′ RACE was used to clone the 5′ end (Frohman et al., 1988). The primer sequences were: gene-specific primer 1 (GSP1) 5′-GGA ACC ACC CAC TGA TCT GCT CCA T-3′; gene-specific primer 2 (GSP2) 5′-AAC ATG TGC AGT TCT GGA GC-3′; the anchor primer was XC1 as shown above. 5′ RACE was performed according to the manufacturer’s instructions (Gibco BRL). Briefly, total RNA from stage I oocytes was reverse transcribed into cDNA using GSP1 and poly(dA)-tailed with terminal transferase. 40-cycle PCR was performed at 94°C (1 minute), 60°C (1 minute) and 72°C (2 minutes). Four independent PCRs were used to generate the correct sequence for VegT. DNA sequence was obtained by double-strand sequencing using the Sequenase version 2 kit (USB Biomedical). Sequence analysis was carried out using the GCG package (University of Wisconsin).

pGBD-VegT, encoding a GBD-VegT fusion plasmid, was constructed as a chimeric DNA-binding protein. An AocI fragment (encoding aa from 261 to 437) from VegT was cloned into the BamHI site of pGBT9 (Fields and Song, 1989) in such an orientation that the VegT C-terminal open reading frame (ORF) was expressed as a translation fusion to GBD. Yeast strain Y526 (Legrain and Rosbash, 1989) was transformed with pGBD-VegT. Transformants were selected on SC medium lacking tryptophan. Filter and quantitative assays were performed as previously described (Preker et al., 1995).

Total RNA was isolated from different stage oocytes, embryos and animal explants (Forristall et al., 1995). Northern blot analysis was carried out as previously detailed (Mosquera et al., 1993). Fibroblast growth factor receptor (FGFR) was used as an RNA loading control and was a gift from Jan Christian.

The entire VegT cDNA was subcloned into pNB40 (a gift of R. Lehmann) and designated pNB-VegT. A C-terminal truncated construct of VegT was made by deletion of the AocI fragment (aa. 261 to 437) from pNB40-VegT and was designated as pNB-VegTΔAD. For myc-tagged VegT mRNA, VegT cDNA was subcloned into pCS2+MT (a gift of D. Turner) and designated as pCS2+MT-VegT. pCS2+nβgal is also a gift from D. Turner. The VegT, VegTΔAD, MT-VegT and nβgal mRNAs were synthesized in vitro in the presence of cap analog and GTP (ratio 4:1) using the Megascript kit (Ambion) from the plasmids linearized with NotI, respectively. The mRNAs were dissolved in RNAse-free water at concentrations indicated in the figure legends and injected at 5-10 nl in volume.

Whole-mount in situ hybridization was done as described previously (Harland, 1991). The VegT in situ hybridization probe was synthe-sized using SP6 RNA polymerase from the plasmid B12 linearized with NotI. Probes for En2 (Hemmati-Brivanlou et al., 1991) and Xlim-1 (Taira et al., 1994) were prepared as described previously.

Embryos and animal cap explants were fixed in MEMFA for 2 hours, embedded in paraffin wax, and 7 μm sections stained with a combination of Feulgen, light green and orange G as described in Green et al. (1990), unless stated otherwise.

For immunostaining, myc-tagged VegT RNA-injected embryos were frozen and dehydrated in 100% ethanol at −80°C for 3 days. The embryos were then embedded in paraffin wax and 7 μm sections were cut with a microtome. The slides were dewaxed and rehydrated through a series of ethanols and treated with 3 M urea for 3 minutes (Hausen and Dreyer, 1982). Positive signal was detected with c-myc monoclonal antibody 9E10 (Oncogene Science) at 10 mg/ml and goat anti-mouse secondary antibody conjugated with horseradish peroxi-dase (Gibco BRL). Notochord staining was performed with MZ-15 antibody (Smith and Watt, 1985). β-galactosidase staining was carried out as described (Vize and Melton, 1991).

A rare set of localized RNAs have been shown to be selectively retained in the vegetal cortex of the stage VI Xenopus oocyte (Weeks and Melton, 1987; Mosquera et al., 1993; Ku and Melton, 1993; Kloc et al., 1993; Elinson et al., 1993). This observation has led to a view of the vegetal cortex as a unique cytoskeletal domain, which functions to anchor specific RNAs and thus to polarize genetic information within the oocyte (Elinson et al., 1993; reviewed by King, 1995). Some of these RNAs are likely to encode proteins involved in dorsoventral patterning (Elinson et al., 1993; Holowacz and Elinson, 1993; Fujisue et al., 1993). To isolate such determinants, we con-structed a cDNA library from oocyte RNA enriched for other known localized RNAs such as Vg1 and Xcat-2 (Mosquera et al., 1993). To screen this library, we developed a PCR-based method to generate cDNA probes from hand-isolated vegetal or animal cortices (see Materials and Methods). The library was also screened with a mixture of Xcat-2, Xcat-3 and Vg1 to eliminate them from consideration. Five independent groups of cDNAs were isolated, each enriched in vegetal cortices. Interestingly, no clone was selected that represented RNAs enriched in animal cortices, lending further support to the premise that the vegetal cortex is a specialized region for anchoring RNAs.

Whole-mount in situ hybridization analysis (Harland, 1991) confirmed that all of the clones were localized to the vegetal pole (data not shown). Their localization patterns during oogenesis are virtually identical to that of Vg1 or Xcat-2 and further support our finding of two major RNA localization pathways during oogenesis (Forristall et al., 1995; J. Z. and M. L. K., unpublished data). One group of three clones (B12) rep-resented an RNA transiently expressed on the dorsal side and was selected for further characterization.

The longest B12 clone isolated from our library was 1.7 kb and an additional 1.0 kb of sequence was obtained by 5′ RACE. The amplified DNAs were sequenced and the correct RACE fragments were used to generate the 2.7 kb cDNA sequence. The first methionine at amino acid position 57 in the longest ORF was assigned as the translational start point. The cDNA encodes a 437 aa polypeptide with a predicted molecular mass of 49.6 kDa. The cDNA includes a 1.2 kb 3′ untranslated region (3′UTR) (data not shown, accession number U59483). The presence of an AAUAAA consensus polyadenylation signal suggests that the cDNA clone has a complete 3′ end.

A full-length RNA transcript synthesized from the cDNA was translated in vitro and yielded a 50 kDa protein product, in agreement with the predicted size (data not shown). Com-parison of the deduced amino acid sequence with a protein database revealed that it represents a previously unknown member of the T-box family. We have named the new gene VegT. Mouse (Herrmann et al., 1990), zebrafish (Schulte-Merker et al., 1994) and Xenopus Brachyury (Xbra) (Smith et al., 1991) members of this emerging family are essential in pat-terning posterior mesoderm.

A comparison of the predicted 191 amino acid N-terminal T-box domain of VegT with T-domains in other members of this family reveals a 55% identity and 72% similarity to mouse or human Tbx2 (Fig. 1A). The next closest known relatives are Drosophila optomotor-blind (omb), which plays a role in the development of specific neurons of the optic lobe (Pflugfelder et al., 1992), followed by F21H11.3 (function unknown; accession number U11279) in C. elegans. VegT is just slightly more related to these genes in other species than it is to Xbra, showing a 51% identity and 72% similarity to the Xbra T-box domain. It is unlikely that Tbx2 or omb are homologs of VegT. The similarity between T domains among VegT and other T-box proteins suggests that the T-domain in VegT may also bind to specific DNA targets.

As is typical for T proteins, no homologous sequences in the databases were found outside the T-box domain. However, the C-terminal part of VegT is rich in serine and threonine residues. Computer analysis of the VegT sequence shows several putative protein kinase C and casein kinase II sites. It has been suggested that phosphorylation of serine and threonine residues would increase the net negative charge of the domain, thereby mimicking an acidic activation domain (Hunter and Karin, 1992). The potential net negative charge of this region provides an interface for interactions with other proteins, including transcriptional co-activators or compo-nents of the basal transcription complex (Roberts et al., 1993).

Northern blot analysis confirmed that VegT transcripts are concentrated within the vegetal half of fully grown oocytes (Fig. 1B). A single transcript about 3.0 kb was observed. During development, maternal RNA levels remain constant from ovulated egg to late blastula. Zygotic transcripts appear to accumulate from late blastula, reaching maximum levels during gastrulation and greatly declining at the time of blasto-pore closure at stage 13 (Fig. 1C) (staging according to Nieuwkoop and Faber, 1967). Expression beyond neurula was not detected on this blot.

Fig. 2 shows the dynamic spatial and temporal pattern of expression of VegT as revealed by whole-mount in situ hybrid-ization (Harland, 1991). VegT mRNA is uniformly distributed in stage I/II oocytes, and localizes to the vegetal cortex during late stage III and early stage IV in a pattern similar to that of Vg1 (Melton, 1987; Forristall et al., 1995) (Fig. 2A,B). In the ovulated egg, VegT RNA remains concentrated in the vegetal hemisphere (Fig. 2C).

Zygotic VegT expression is first detected at stage 9.5, where it appears in a highly restricted pattern on the dorsal side of the marginal zone at the site of the future dorsal lip (Fig. 2D,E). A vegetal view of a cleared embryo at this stage shows faint expression along the lateral and ventral sides (Fig. 2E). The strong dorsal pattern of VegT expression continues into the early gastrula embryo (stage 10) but with increased staining along the lateral and ventral sides (Fig. 2F). The staining pattern and its relation to the dorsal lip was confirmed using pigmented embryos where the dorsal lip is quite easily observed (data not shown). Transverse sections of the stained embryo in Fig. 2F showed that the transcripts were in the presumptive mesoderm on the ventral lateral side. On the dorsal side, both epithelial and deeper layers of the organizer express VegT, although staining was less intense in the epithelial cells (data not shown).

An hour later, at stage 10.5, expression is now uniformly distributed within the entire marginal zone (Fig. 2G). At stage 12.5, intense staining is now observed around the closing blastopore at the posterior end of the embryo. VegT expression is specifically absent from a narrow region in the dorsalmost area, which corresponds to where the notochord will form (Fig. 2H). Histological sections of this stage show that the staining is restricted to the posterior paraxial and lateral mesodermal layers and is not detected anteriorly nor in the epithelial layer (data not shown). At the mid-neural fold stage (stage 16-17), VegT transcripts are expressed exclusively within the deep mesoderm in the posterior region of the embryo, similar to Xbra (Smith et al., 1991). Importantly, the pattern of VegT expression differs from Xbra in that VegT transcripts are never observed in the notochord (Fig. 2I). By early tail bud (stage 23), the posterior expression of VegT has greatly diminished and is barely detected. Instead, new expression is observed in approximately 15 to 20 cells of the posterior neural tube (Fig. 2J). These cells appear to be a subset of Rohon-Beard cells, a type of primary sensory neuron, populating the embryo at this stage. The number of cells stained remained constant at-stage 31 tadpole (Fig. 2K,L) and through late tadpole stages (not shown). Sections of stage 31 show that these cells are in the dorsal-lateral region of the neural tube (Fig. 2K′). Embryos hybridized with the control sense probe were negative except for non-specific staining within cavities in the anterior region (also observed in Fig. 2K).

In summary, zygotic expression patterns of VegT shift from an intense dorsal expression in late blastula and early gastrula to a ventral/lateral expression at the posterior end of later stage embryos. VegT appears excluded from the notochord. Finally, in tailbud and tadpole stages, VegT is expressed exclusively in a subset of posterior Rohon-Beard neurons.

The T protein encoded by Brachyury has been shown to be a tissue-specific transcription factor (Kispert et al., 1995). Analysis of the VegT sequence indicated a putative nuclear localization sequence from residue 250 to 256 (KRQKRKK). To determine if VegT could enter the nuclei of cells, we injected myc-tagged VegT (MT-VegT) into one blastomere of a 2-cell embryo. The other blastomere served as the negative control. Immunostaining with anti-myc antibody of sections taken from a blastula embryo (stage 7 or 9) revealed myc-tagged VegT protein only in the nuclei of the progeny of the injected blastomere (Fig. 3A). We conclude that VegT protein can enter and accumulate within nuclei consistent with the hypothesis that it functions as a transcription factor.

Next we asked whether the C-terminal segment of VegT could function as a regulatory domain in a yeast reporter assay (Fields and Song, 1989). The C-terminal 176 aa (from 261 aa to 437 aa) of VegT fused to a GAL-4 DNA-binding domain significantly transactivated the β-gal reporter gene in yeast (Fig. 3B). The activation activity of the chimeric VegT was at least 70-fold higher than that of the negative control. From these results, we conclude that VegT protein likely functions as a transcription factor and that the C-terminal region of the protein could regulate its transcriptional activity.

The initial expression of zygotic VegT in the presumptive dorsal lip region suggested that VegT might be involved in dorsal patterning. To test whether VegT has dorsalizing activity, 2.5 ng of the full-length transcript was injected into both ventrovegetal blastomeres of the 8-cell embryo. The first overt indication that VegT had altered the fate of ventral cells was the appearance of a second invagination on the ventral side shortly after gastrulation had started (Fig. 4B). Secondary axes formed in 46% of the injected embryos (n=301). Secondary axis formation seems to be dose-dependent (0.7 ng resulted in 17%, n=41; 2.5 ng in 46%, n=301; 5 ng in 56%, n=32), although a fine dose analysis has not yet been done. Embryos injected with control RNAs encoding either a truncated VegT (VegTΔAD, missing the presumptive regulatory domain) or a nuclear form of β-galactosidase (nβgal), were unaffected (Fig. 4A,C,E).

Typical duplicated axes that formed are shown in Fig. 4D. Secondary axes lacked anterior structures such as head, cement gland and eyes. The formation of a separate tail in some of the secondary axes is reminiscent of the tail-organizing activity observed in chordin-injected embryos (Sasai et al., 1994). His-tological analysis of the secondary axis revealed massive amounts of muscle and, at lower frequencies, neural tubes (Fig. 4F). Immunostaining with an anti-notochord-specific antibody showed that 55% of the embryos (n=24) have positively stained cells in the secondary axis (Fig. 4G). In some cases, organized notochords can be seen (Fig. 4G). Also evident were otic vesicles in 30% (n=24) of the cases with or without an organized notochord, suggesting anterior development extends at least as far as the hindbrain. Injection of nβgal and VegTΔAD did not result in any specific phenotype and were comparable to non-injected embryos. From this analysis, we conclude that VegT can trigger a partial duplicated axis, which includes hindbrain and the dorsalmost mesodermal derivative, notochord.

To determine if VegT were acting within an inductive pathway or whether it only altered the fate of injected cells, the lineage tracer nβ-gal was co-injected on the ventral side with VegT. In each case, X-gal nuclear staining was found within embryonic endoderm directly underlying the unstained dorsal tissues of the secondary axes (Fig. 5A). Therefore, VegT-injected ventrovegetal blastomeres were able to induce a secondary axis. Histological examination of representative embryos shows that the X-gal-stained cells constitute a small domain forming a clear boundary between the descendants of the injected blastomeres and the induced tissues, suggesting a paracrine mechanism may be involved in the process (Fig. 5B). The progeny of VegT-injected blastomeres appear to be converted into a cell group resembling Nieuwkoop’s center (Nieuwkoop, 1973), which functions to induce other blastomeres to form and organize a dorsal axis.

Although VegT can induce mesodermal tissues in a partial secondary dorsal axis, the question remained whether VegT is capable of directly inducing mesoderm. To address this question, embryos were injected at the 2-cell stage with VegT mRNA (0.7 ng, 2 ng, 5 ng). Animal caps were then explanted at stage 8 and cultured in isolation. All three doses of VegT caused the formation of mesoderm in the explants (100%, n=126). The convergence and extension observed at stage 11 was moderate compared to that of dorsal marginal zone explants (data not shown). Histological examination of the cultured animal caps revealed extensive mesenchyme, blood and muscle, indicating that ventral and lateral mesoderm had been formed (Fig. 6D,E). We did not observe a clear dose-dependent induction of ventral versus lateral mesoderm which has been shown for Xbra (O’Reilly et al., 1995). Animal caps injected with control RNAs (VegTΔAD or nβ-gal) formed only atypical epidermis (Fig. 6A,C). This ventral mesodermal phenotype is similar to that triggered by ventralizing factors, such as Xwnt-8 (Christian and Moon, 1993). To understand how VegT may function in this pathway, we asked whether VegT can activate the expression of Xwnt-8 in VegT-injected animal caps. The results indicate that VegT is capable of causing the de novo expression of Xwnt-8 (Fig. 7). Interest-ingly, VegT can also positively regulate itself (Fig. 7). Thus, VegT can trigger the formation of ventral lateral mesoderm and may do this through a Wnt signalling pathway although other factors like eFGF could also be involved.

The distribution of maternal VegT is highly restricted to the vegetal pole of eggs (Fig. 2C). To assess the effect of mis-expression, 2 ng of VegT mRNA was injected into the animal pole of 2-cell embryos and the embryos were left to develop through gastrulation. The results from injecting MT-VegT RNA at the animal pole (Fig. 3A) had indicated that the myc-tagged protein diffused well into the vegetal hemisphere (data not shown). As a control, identical amounts of VegTΔAD were also injected (Fig. 6F). The VegTΔAD-injected embryos developed through blastula and started gastrulation normally compared to non-injected embryos. However, the VegT-injected embryos did not close their blastopores, a phenotype similar to that described for Xbra-injected embryos (Cunliffe and Smith, 1992) (Fig. 6G,H). Apparently invagination was not affected, but involution ceased and the embryos failed to close their blastopores.

In VegT-injected embryos, the blastocoel collapsed and formed a cavity sinking into the vegetal mass (Fig. 6G). Pro-trusions of the animal cap epithelium (Fig. 6G,H) were common and expected since mesoderm is induced in the ectoderm by contact with the endoderm (Nieuwkoop, 1973). Histological examination of these injected embryos revealed that the blastocoel was affected. In some cases, the blastocoel fluid had escaped into the vegetal mass of these embryos indicating the integrity of the enclosing epithelium had been lost (Fig. 6H). In contrast, VegTΔAD or nβ-gal RNA-injected embryos were indistinguishable from non-injected embryos (Fig. 6F). Therefore, the effects observed depended on a functional VegT protein and were not simply due to toxic effects. These results show that ectopic expression of VegT at the animal pole interferes, most likely indirectly, with normal cell-cell contact and the morphogenetic movements of gastrulation.

To more accurately evaluate VegT mis-expression at the animal pole, 2 ng VegT mRNA was injected into the dorsal animal blastomeres of the 8-cell embryo whose progeny are fated to contribute largely to head mesoderm (Moody, 1987a,b; Dale and Slack, 1987). VegT, but not the control RNAs, sup-pressed head formation similar to FGF (Isaacs et al., 1994) and Xwnt-8 (Christian and Moon, 1993). Significant numbers of tadpoles developed without cement glands and eyes (83%, n=61; Fig. 8). Interestingly, the VegT-induced secondary axis also lacked anterior head structures (Fig. 4D, 4G). It appears that VegT expression is involved in trunk and tail formation.

To determine more precisely the boundary of the anterior depletion, we examined En-2 expression in such embryos by whole-mount in situ hybridization. En-2 is mainly expressed at the midbrain-hindbrain junction (Hemmati-Brivanlou et al., 1991). In some VegT-injected tadpoles (Fig. 8D), En-2 expression was missing, indicating that both forebrain and midbrain were suppressed. In other examples where the En-2 transcripts (arrowhead) could be detected, the tadpoles lacked cement gland (open arrow) and eyes (Fig. 8D), suggesting that only the most anterior structures, such as forebrain were missing.

Since Lim1 is required for head formation in mouse (Shawlot and Behringer, 1995), we tested whether injection of VegT into dorsal animal blastomeres had any effect on Xlim1 (Lim1 homolog in Xenopus) expression. Whole-mount in situ hybridization revealed no differences in Xlim1 expression between VegT-injected embryos and control embryos at gastrula (data not shown), suggesting that inhibition of anterior structures by VegT may not act through a Xlim1 pathway.

In this report, we describe a novel approach for identifying maternally expressed genes involved in embryonic patterning. A differential screening procedure was designed to identify RNAs uniquely localized to the vegetal cortex. We targeted this region because of its demonstrated importance in dorsal axis formation (Nieuwkoop, 1973; Wylie et al., 1987; Fujisue et al., 1993) and potential significance in mesoderm induction (Thomsen and Melton, 1993; Kessler and Melton, 1995). The cDNA library screened was constructed from oocyte RNA 50-fold enriched for localized RNAs (Mosquera et al., 1993). VegT cDNA was one of five isolated by this approach and was selected for further study because of its dynamic and unique pattern of expression along the dorsal/ventral axis in the marginal zone.

VegT can be divided into two domains which, either by sequence or function, identify VegT as a member of the growing T-box family (reviewed by Herrmann and Kispert, 1994). The N-terminal domain (46aa-242aa) shows significant homology to the T domain of other T-box genes, including Brachyury and Xbra (Fig. 1A). Evolutionarily conserved T domains in Drosophila (Pflugfelder et al., 1992), Xenopus, mouse and zebrafish have all been shown to specifically bind a 20 bp DNA sequence (T site) in vitro (reviewed by Herrmann and Kispert, 1994). Mouse Brachyury has further been shown to act as a transcription factor in transformed HeLa cells (Kispert et al., 1995). The transcriptional regulatory domain lies within the C-terminal domain, which bears no homology to other proteins (Kispert et al., 1995). Immediately down-stream to the T domain, VegT was also found to have a putative nuclear localization signal. Expression of a myc-tagged VegT at 2-cell stage confirmed the competency of VegT to enter the nuclei of the developing embryo. The C-terminal domain of VegT is capable of activating β-galactosidase through a Gal4 DNA-binding domain in a yeast reporter assay. The activation level is not as high as in the positive control and two explanations may account for this. First, the chimeric protein may not fold into the optimal conformation to execute its activation activity. Alternatively, the C-terminal domain may contain both activation and repression motif(s) that result in a dampening of activation in yeast. In fact, Brachyury does appear to have activation and repression domains in its regulatory domain (Kispert et al., 1995). Taken together, the data support a role for VegT as a transcriptional regulator related to Brachyury.

The dynamic and unique expression pattern of VegT suggests that it has multiple functions during embryogenesis. As a putative transcription factor, VegT can only regulate down-stream mediators of patterning, after the MBT, when zygotic transcription is initiated. The strict localization of VegT to one pole of the amphibian egg is intriguing and suggestive. Such localization could provide a mechanism for generating a gradient in VegT protein along the A/V axis that would result in different blastomeres inheriting different concentrations of the transcription factor. In addition, strict localization prevents VegT from being present in the animal hemisphere where it might activate or repress inappropriate genetic pathways. Lastly, VegT’s cortical location might allow it to act syner-gistically with other maternal components localized there. Interestingly, VegT is localized at the same time and within the same cortical domain as Vg1 (Weeks and Melton, 1987; For-ristall et al., 1993). Processed Vg1 protein can induce dorsal mesoderm (Thomson and Melton, 1993; Dale et al., 1993; Kessler and Melton, 1995) and it could be co-expressed with VegT. This maternal pattern is very different from that of Xbra mRNA whose low expression level is detected only by RNase protection and which appears to be uniformly distributed (Smith et al., 1991).

At MBT, VegT is expressed in the marginal zone along the entire dorsal/ventral axis underlying the blastocoel. Two important features of this expression pattern should be noted. Zygotic VegT is initially expressed in a gradient and the highest concentration is at the dorsal end. Shortly thereafter the gradient is gone and expression is uniform throughout the marginal zone, similar to Xbra. Later at stage 12.5, VegT expression is missing in the presumptive notochordal area where Xbra is expressed. At the end of gastrulation, VegT is expressed around the closing blastopore ventrally and laterally, where it persists through neurula stages. The different expression patterns of VegT and Xbra in posterior mesoderm, especially in notochord, suggests distinct roles for these two genes in mesodermal patterning.

Injection of VegT RNA into the animal pole region caused ventral mesodermal differentiation in animal caps. Consistent with this finding, we have shown that VegT can activate Xwnt-8, a ventral marker in animal caps. Xwnt-8 is zygotically expressed in the marginal zone, but is excluded from the organizer during gastruation (Smith and Harland, 1991). Goosecoid, an immediate early response gene to mesoderm induction expressed in the organizer, has been shown to repress Xwnt-8 expression there (Christian and Moon, 1993).

Our data can be explained in the context of VegT activating Xwnt-8. First, VegT misexpressed in the dorsal animal pole blastomere results in the suppression of head structures, most predominantly forebrain as indicated by the midbrain/hindbrain boundary marker En-2. Xwnt-8 misexpression also results in the loss of anterior structures. Second, both VegT and Xwnt-8 have ventralizing activity. Third, overexpression of VegT prior to the MBT in ventral vegetal blastomeres can induce secondary dorsal axes, just like ectopic Xwnt-8 expression in these blas-tomeres, with one important difference (Christian and Moon, 1993). Xwnt-8 can induce formation of a complete secondary axis (Smith and Harland, 1991; Sokol et al., 1991), while VegT can only cause a partial one to form (Fig. 4). Dorsal formation is unlikely to be a simple event elicited by a single molecule. The combination of induction and cell autonomous action and possible cross-talk between these two pathways suggests that a number of factors are required to orchestrate this complex process. Different molecules may have different regional and temporal roles. Molecules such as VegT may only specify the posterior dorsal axis, while noggin and unidentified maternal Wnts may act on anterior dorsal structures (Smith and Harland, 1992; Steinbeisser et al., 1993). Novel vegetally localized RNAs isolated in our laboratory may implicate additional com-ponents that may also be required to specify the dorsal axis and mesoderm and bear further investigation.

VegT generated different cell types when expressed in different embryonic domains. Different competences of the responding cells may contribute to these observed results. Animal cap cells normally do not form mesoderm, and their ability to respond to inducing factors and cell autonomous factors, could be quite different from those cells in the marginal zone. Furthermore, VegT most likely co-operates with other mesoderm-inducing and/or pat-terning factors to lay down the final pattern of mesoderm. It has been previously demonstrated that Xbra causes dorsal mesodermal differentiation when co-expressed with the secreted proteins noggin and Xwnt-8 (Cunliffe and Smith, 1994). None of these three molecules cause dorsal mesoderm formation when expressed alone (Cunliffe and Smith, 1994). Synergistic interaction was also observed between Xbra and another mesoderm-specific transcription factor, Pintallavis (O’Reilly et al., 1995). Taken together, these data suggest that phenotypes generated by ectopic expression of single molecules may not reflect what occurs in vivo.

It is not surprising that misexpression of VegT at the animal pole caused the arrest of gastrulation given that VegT is capable of converting ectodermal cells to mesoderm. However, the molecular basis underlying this arrest is unknown. In VegT RNA-injected embryos, the complete loss of the blastocoel, in some cases, indicates the loss of cell-cell adhesion (Fig. 6H). Recently, it has been suggested that Brachyury may act during gastrula-tion to alter cell surface properties as cells pass through the primitive streak (Wilson et al., 1995). Ectopically expressed VegT may share a similar activity with Xbra.

Drosophila early development is regulated by four systems all using the same basic strategy to define the anteropos-terior or the dorsoventral axis (Nusslein-Volhard, 1991). In each system, an asymmetrically localized transcription factor results in a concentration gradient that defines the spatial limits of expression of one or more zygotic genes (Nusslein-Volhard, 1991). For example, the anterior morphogen bicoid is essential for hunchback expression and normal head/thorax development (Driever and Nusslein-Volhard, 1988). Like the Drosophila syncytial blastoderm, molecular gradients may form in Xenopus eggs and positional information likewise be inherited by individual blastomeres. Our working hypothesis is that maternal VegT may form such a concentration gradient along the A/V axis. VegT activity along such a gradient could determine regional or cellular identity by regulating the expression of zygotic target genes in a dose-dependent way. Different levels of VegT could elicit either the activation or the repression of the same gene as has been shown for Drosophila hunchback. Hunchback activates Kruppel at high concentrations and represses it at low concentrations (Schulz and Tautz, 1994). Brachyury has both activation and repression regions in its regulatory domain (Kispert et al., 1995) and this may also be important for VegT regulatory activity as well. The suggestion that VegT may act as a morphogen may appear to contradict our observation that VegT is likely a dose-independent agent in mesoderm formation. However, given its mutiple expression patterns during early embryogenesis, VegT may possess other yet unidentified activities, which may also be important in patterning.

After MBT, with the induction of organizer-specific genes like goosecoid, Xwnt-8 is downregulated (Christian and Moon, 1993) and VegT is subsequently downregulated in this region. The evidence suggests that Xwnt-8 is affected rapidly and VegT secondarily so. Xwnt-8 may be required for the maintenance of VegT in a similar fashion as has been described for the Xbra and FGF regulatory loop (Isaacs et al., 1994; Schulte-Merker and Smith, 1995). In related experiments, we have also seen that VegT can autoregulate and turns itself on when expressed in animal caps (Fig. 7). Misexpression of Xwnt-8 in the presumptive region of the notochord results in embryos missing forebrain and notochord, indicating that it is important to downregulate Xwnt-8 in the organizer (Christian and Moon, 1993). If VegT turns Xwnt-8 on in vivo as it does in animal caps, then it follows that VegT must also be eliminated from the presumptive notochordal region. Such a prediction is consistent with our observations as noted. It will be important to determine the immediate downstream factors of VegT and whether mesoderm-inducing factors can modulate VegT’s function at a post-translational level.

We are grateful to Dr F. Watts for MZ-15 antibody, Drs R. Harland, E. M. De Robertis, I. Dawid, A. Hemmati-Brivanlou, J. L. Christian, J. B. Gurdon, P. Lemaire, D. Turner, and R. Lehmann for kindly providing the various gene markers used in this study. We thank H. He for the yeast reporter system, Dr S. A. Moody for identification of Rohon-Beard neurons, and Dr M. Kirschner and K. Lustig for sharing information with us prior to publication. We thank Drs G. Grotendorst, R. Voellmy, M. Flajnik, P. Salas and D. Houston for critical reading of the manuscript. J. Z. thanks Hui Dong for being supportive and helpful during the completion of this work. The accession number for the sequence reported in this paper is U59483. This work was supported by National Institute of Health grant GM33932.

An allele to VegT, Xombi, has been isolated and characterized by Kevin Lustig and Marc Kirschner (this issue).