ABSTRACT
We have identified a novel frog gene, Pintallavis (the Catalan for lipstick), that is related to the fly fork head and rat HNF-3 genes. Pintallavis is expressed in the organizer region of gastrula embryos as a direct zygotic response to dorsal mesodermal induction. Subsequently, Pintallavis is expressed in axial midline cells of all three germ layers. In axial mesoderm expression is graded with highest levels posteriorly. Midline neural plate cells that give rise to the floor plate transiently express Pin tallavis, apparently in response to induction by the notochord. Overexpression of Pintallavis perturbs the development of the neural axis, suppressing the differentiation of anterior and dorsal neural cell types but causing an expansion of the posterior neural tube. Our results suggest that Pintallavis functions in the induction and patterning of the neural axis.
Introduction
The development of the basic body plan of vertebrate embryos depends on inductive interactions and morphogenetic movements that take place during gastrulation (Gerhart and Keller, 1986; Hamburger, 1988). The process of gastrulation results in the formation of two midline mesodermal structures, the prechordal plate and the notochord, and in the induction of the neural plate. Subsequently, the notochord induces the overlying neural ectoderm to differentiate into the floor plate, the ventral midline of the neural tube. The patterning properties of these midline cell groups are, in large part, responsible for the regional differentiation of the nervous system along the antero-posterior (A–P) axis, leading to the formation of the brain and spinal cord.
The first insight into the events that control the formation of the neural axis of the amphibian embryo derived from the transplantation assays of Spemann and Mangold (1924). In these experiments, transplantation of the dorsal lip of early gastrulae to the ventral region of a sibling embryo resulted in the formation of a twinned embryo with a second neural axis. The cells of the dorsal lip, termed the organizer, changed the fate of host ventral cells, resulting in the formation of axial mesoderm and induced neural tissue. In contrast, the midline structures of the induced second axis derived primarily from the descendents of the grafted dorsal lip cells (Spemann, 1938; Smith and Slack, 1983).
In normal Xenopus embryos, the dorsal lip contains mesodermal cells that involute at the blastopore during gas-trulation and extend along the A-P axis giving rise primarily to the prechordal plate and notochord. Ectodermal cells adjacent to the dorsal lip, a region known as the dorsal noninvoluting marginal zone (NIMZ, Keller and Danilchik, 1988) remain in the ectoderm but also extend along the midline, to form the notoplate, which is coextensive with the underlying notochord (Jacobson, 1981; Keller et al., 1985; Keller and Danilchik, 1988). In addition, the superficial cells of the dorsal lip give rise to the midline endodermal cells that line the archenteron roof. Thus, cells in and adjacent to the dorsal lip form midline structures in each of the three germ layers of the embryo after gastrulation.
The ability of the dorsal lip to induce neural tissue is maintained by its axial mesodermal derivatives. The notochord, in particular, is a potent inducer of neural tissue in competent ectoderm. These and other experiments have led to the idea that the involuted axial mesoderm is responsible for induction of the neural tube (Mangold, 1933; Holtfreter, 1933; Nieuwkoop et al., 1952; Sharpe and Gurdon, 1990). However, signals from the dorsal lip that spread through the plane of the ectoderm are also involved in neural induction and A-P patterning (Spemann, 1938; Savage and Phillips, 1989; Dixon and Kintner, 1989; Ruiz i Altaba, 1990, 1992). Moreover, recent studies suggest that the notoplate may be a source of planar signals involved in the formation of the neural tube and that these cooperate with vertical signals from the head mesoderm and notochord to induce the forebrain and ventral neuronal types in the hindbrain and spinal cord (Ruiz i Altaba, 1992).
The induction of ventral neurons by the notochord appears to involve the floor plate, a group of cells with notochord-like properties located along the ventral midline of the embryonic neural tube with the exception of the forebrain (Kingsbury, 1930; Placzek et al., 1990, 1991; Yamada et al., 1991; Hatta et al., 1991). The differentiation of the floor plate from cells at the midline of the neural plate is induced by contact with the underlying notochord (van Straaten et al., 1988; Placzek, et al., 1990; Clarke et al., 1991). Thus, in Xenopus, the floor plate is likely to be derived from the notoplate in response to signals from the notochord. The induction and patterning of the Xenopus nervous system may therefore require signals from midline mesodermal cell groups: the dorsal lip, the notochord and the prechordal mesoderm, and from midline neural cell groups: the notoplate and the floor plate. The different inductive properties of these midline cell groups could reflect distinct signals or the action of the same signal on cells with different response properties.
An understanding of the molecular events involved in the induction and patterning of the nervous system requires the identification of genes that control the properties of these midline structures. In addition to the conserved signalling properties of the notochord and floor plate, both structures are unsegmented but are surrounded by segmented tissues. The notochord is adjacent to the somites and the floor plate is adjacent to the rhombomeres of the hindbrain (Vaage, 1969; Lumsden and Keynes, 1989). The division of tissues into segmented and unsegmented structures could reflect a fundamental and evolutionarily conserved decision. In flies, the fork head gene has been suggested to be involved in repressing thoracic segmentation in the head (Jürgens and Weigel, 1988). Fork head encodes a nuclear protein (Weigel et al., 1989) that contains a sequence very similar to that of the DNA-binding domain of the HNF-3 liver transcription factors (Weigel and Jäckle, 1990; Lai et al., 1990, 1991). The conservation in sequence between fork head and the HNF-3 genes enabled us to search for related genes expressed in midline cells of Xenopus embryos that could be involved in establishing their common properties, the lack of segmentation and their inductive abilities.
We describe here the isolation and characterization of a Xenopus gene, Pintallavis, that is related to the rat HNF-3 and Drosophila fork head genes. Pintallavis is expressed first in the dorsal lip as a direct response to dorsal mesodermal induction. After the onset of gastrulation, Pintallavis is transiently expressed at high levels in the notochord and at low levels in the prechordal mesoderm. Pintallavis is also expressed at the midline of the neural plate although it is absent from the notoplate in exogastrula embryos, suggesting that its expression is induced by the notochord. Injection of synthetic Pintallavis RNA into developing embryos results in axial defects. These include a disruption in the extension of the embryo and a suppression of anterior neural structures. In contrast, in posterior regions there is an overproduction of neural tissue and an absence of certain dorsal neurons. The restricted expression of Pintallavis in midline cells and the phenotype obtained after its overexpression suggest that this gene is involved in the development of the neural axis.
In recent studies, the cloning of XFKHI, a closely related Xenopus gene, has been reported (Dirksen and Jamrich, 1992). The distributions of Pintallavis and XFKHI are similar but not identical, suggesting that they may represent the two copies of the same gene in the tetraploid Xenopus genome.
Materials and methods
Embryological manipulations
Xenopus laevis embryos were obtained by inducing female pigmented or albino frogs to lay eggs with human chorionic gonadotropin which were fertilized in vitro with testis homogenates. Embryos were freed from their jelly coats by treatment with 3% cysteine pH 7.6 in tap water and reared in 0.1×MMR (Newport and Kirschner, 1982). Embryos were staged according to Nieuwkoop and Faber (1967).
Treatment of embryos before first cleavage with ultraviolet light was performed for 30 seconds, 1 minute and 1.5 minutes by irradiating the vegetal poles approximately 40 minutes after fertilization. Treatment of early blastula stage embryos (stage 6–7) with Li+ was performed by incubating embryos in 0.25 M LiCl in tap water for 8–10 minutes. A fraction of treated embryos was allowed to develop to the tadpole stage for the assessment of the degree of axial defects (Kao and Elinson, 1988).
Animal pole explants were cut from late blastula stage embryos (stage 7–8) and incubated in 0.5×MMR. Both activin and basic fibroblast growth factor were supplied as transformed COS cell supernatants and were a kind gift from C. Hume (Columbia University). Treatment of animal caps with all-trans retinoic acid (RA; Sigma) was performed by diluting the stock of 5×10−3M in dimethylsulfoxide with 0.1 ×MMR to 10−8M. RA-treated embryos were protected from light. Cycloheximide (5 μM; Sigma) was applied to whole embryos 1 hour before cutting the animal caps, which were also incubated in the drug.
Complete exogastrulae were obtained by incubation of embryos from which the vitelline membrane had been removed in 1.3×MMR. All explants were kept in agarose dishes in the presence of antibiotics.
Embryo injections
Injection of synthetic RNA was performed by injecting 1–10 nl of RNA solution in water with a pressure driven injector. Embryos were kept in 1×MMR, 3% Ficoll before, during and 1 hour following the injection. Rhodamine/lysine/dextran (Mr 10,000; Molecular Probes) at 50 mg/ml was mixed with the RNA solution at a ratio of ∼1:10 immediately before injection.
RNA isolation, cDNA synthesis and PCR
RNA was isolated from homogenized embryos by proteinase K digestion followed by precipitations with ethanol and LiCl. Dissection of embryo parts for RNA isolation was performed by fixing the embryos in acid alcohol (95% ethanol, 5% glacial acetic acid) for 5-20 minutes before dissection.
First-strand cDNA was prepared by priming the RNA with random primers in the presence of Superscript reverse transcriptase (BRL). The polymerase chain reaction (PCR) was performed with degenerate primers on cDNA samples with a 40°C annealing step in the presence of AmpliTaq DNA polymerase (Perkin Elmer-Cetus). Oligonucleotides flanked by EcoRI or BamHI restriction sites with inosines in completely degenerate positions were directed against the conserved sequences ITMAIQ and MFENGC of the HNF-3 (Lai et al., 1991) and fork head (Weigel et al., 1989) genes. PCR reactions were also performed with primers directed to the internal sequence QNQQRWQ. The presence of an amplified band was dependent on the addition of reverse transcriptase to the cDNA reaction ruling out contamination by genomic DNA. The PCR products were digested with restriction enzymes, purified in low melting point agarose gels and ligated to a phage M13 vector for sequencing with sequenase (USB) in the presence of 35S-labelled dATP. The sequence of 20 independent clones showed only a few third-base changes and one conservative amino acid substitution. These variations are likely to reflect the polymorphisms of single genes in the teraploid Xeno pus genome.
Library screens and subclones
Screening of a neurula stage cDNA library in lambda gt10 (Kintner and Melton, 1987) was performed at high stringency with 32Plabelled random-primed M13 PCR inserts on phage lifts transferred to Hybond (DuPont) filters. Phage DNA was cross-linked to filters by UV light.
The insert of the longest cDNA isolated in the library screens did not have internal EcoRI sites. The ∼1.8 kb cDNA insert was then subcloned into the EcoRI site of pBSKS (Stratagene) to yield pF5.
RNase protections and in vitro RNA synthesis
RNase protection assays and the EF-I alpha probe were as described (Ruiz i Altaba and Melton, 1989a). For the Pintallavis probe, a PstI deletion subclone of pF5, pF5Pstdelta, was digested with XmnI and transcribed with phage T7 RNA poymerase. Ten and one embryo or animal cap equivalents were used for the Pin tallavis and the EF-I alpha probes, respectively. RNase digestion was carried out at room temperature for 1 hour. The undigested products were analysed in denaturing sequencing gels.
Synthetic, capped RNA was made by transcription of the fulllength Pintallavis cDNA insert at 30°C, cloned by blunt-end ligation into the filled BglII site of pSP64T, with SP6 RNA polymerase (Krieg and Melton, 1984). pSP64T (Melton, 1985) provides 5 ′ and 3 ′ untranslated sequences and a poly(A)-poly(C) tail that enhance the translation and the stability of injected messages. In vitro translation of Pintallavis RNA in reticulocyte lysates (Promega) was performed with 0.5 μg of RNA per reaction in the presence of 35S-labelled methionine using bromo mosaic virus RNA as a control. The products were analysed by electrophoresis in 10% polyacrylamide gels.
For control studies, a frame-shift mutation was introduced into the Pintallavis sequence by digestion with BglII (nucleotide 507), filled-in with Klenow polymerase and religated, yielding pF5(B). The translation of this RNA is predicted to terminate after an additional 29 amino acids that are not present in Pintallavis, thus truncating the putative DNA-binding domain.
Whole-mount in situ hybridization
Whole-mount in situ hybridization was performed with digoxigenin-labelled probes according to Harland (1991; see also Hemmati-Brivanlou et al., 1990). The full-length cDNA insert was used as template to generate an antisense RNA probe from pF5 after digestion with HindIII and transcription with T7 RNA polymerase in the presence of dig-11 UTP. Hybridization of embryos with the sense probe gave no signal. Hybridization with an anti-sense probe derived from the 3′ untranslated region gave the same results as the full-length probe. In situ hybridization with Xenopus Brachyury was performed by transcribing the full insert of pXbra6 with T7 RNA polymerase, kindly provided by J. Smith (Smith et al., 1991). Transcription reactions were traced by adding a small amount of 32P-labelled UTP. The RNA probes were not hydrolysed and used at a concentration of 1 μg/ml. The hybridization signal was detected by a secondary antibody coupled to alkaline phosphatase, used at 1:2000, followed by the phosphatase reaction with BCIP and X-phosphate (Boehringer Mannheim). Reaction times varied from 1 to 8 hours.
Whole-mount antibody labelling and histology
Whole-mount antibody labelling was performed essentially as described by Dent et al. (1989), using MEMFA as fixative (Patel et al., 1989; Hemmati-Brivanlou and Harland, 1989). Monoclonal antibodies (mAbs) Xen1 (Ruiz i Altaba, 1992) and HNK-1 (Abo and Balch, 1981) were used as culture supernatants diluted 1:1. mAb 12/101 (Kintner and Brockes, 1984) was used as ascites fluid diluted 1:500. Secondary antibodies (Fab fragments) coupled to horseradish peroxidase (Boehringer Mannheim) were used at 1:100. Antibody binding was detected by the peroxidase reaction in the presence of H2O2 and diaminobenzidine. Labelled embryos were cleared in benzyl alcohol-benzyl benzoate. Histological sections (12.5 μm) were obtained by sectioning paraplast-embedded embryos. The sections were subsequently dewaxed in xylene and mounted in Permount.
Results
Isolation and structural features of Pintallavis
The Drosophila fork head and the rat HNF-3 genes are very similar in a region encoding their putative DNA-binding domains (Weigel and Jäckle, 1990; Lai et al., 1991). This permitted the design of degenerate oligonucleotide primers with which to screen for related Xenopus genes using the polymerase chain reaction. First-strand cDNA was prepared from a combined RNA sample isolated from the dorsal lip of early gastrula (stage 10) embryos and from the midline of mid-gastrula (stage ) embryos. The primers amplified a ∼240 bp fragment the sequence of which is highly similar to that of the conserved domain in the HNF-3 and fork head genes (Fig. 1A, B).
To isolate a full-length cDNA clone, a mid-neurula (stage 17) library (Kintner and Melton, 1987) was screened with the random-primed PCR product. Three recombinant lamda clones were isolated which contained unique EcoRI inserts of 1.6 to 1.8 kb in length. The longest cDNA clone was sequenced, revealing an open reading frame that encodes a protein of 399 amino acids that is similar to the HNF-3 and fork head proteins (Fig. 1A, B). The position of the first methionine in Pintallavis was assigned by sequence similarity with the NH2-terminal region of the HNF-3 proteins (Fig. 1B). A high degree of similarity in the putative DNAbinding domains was observed among all five proteins (positions 107 and 222 of Pintallavis; Fig. 1B). Although this region exhibits over 90% identity in all five proteins, Pintallavis is not significantly more closely related to any individual protein of this family. The region of Pintallavis that is NH2-terminal to the conserved putative DNA-binding domain has a very high methionine content as reported for the HNF-3 and fork head proteins (Weigel et al., 1989; Lai et al., 1991). Two other regions of similarity between HNF-3 and fork head are also found in Pintallavis, a domain between positions 316 and 341 containing the HPFSI motif, and a small domain from position 386 to the COOH terminus of the protein (Fig. 1A, B).
Pintallavis exhibits 47% overall identity with HNF-3 gamma, 46% identity with HNF-3 alpha, 45% identity with HNF-3 beta, and 42% identity with fork head. The NH2terminal region of Pintallavis is similar to that of the HNF3 genes; however, the rat genes are more closely related (∼60%) to each other than to the frog gene. Pintallavis is therefore unlikely to be the Xenopus homologue of the rat HNF-3 genes. The conservation of several distinct domains between Pintallavis, the three HNF-3 genes and fork head, and the distinct patterns of expression of the vertebrate genes (see below) suggests that Pintallavis is a new member of this gene family.
Transient embryonic expression of Pintallavis
The period over which Pintallavis is expressed during embryonic development was examined by RNase protection. RNA was isolated from embryos at different embryonic stages and the expression of Pintallavis compared with that of EF-1 alpha, a gene expressed at similar levels in all cells of the Xenopus embryo (Krieg et al., 1989). The expression of Pintallavis is first detected in the late blastula (stage 9) embryo and its expression peaks in mid-gastrula (stage 12) embryos (Fig. 2A). Pintallavis expression declines at later stages and is barely detectable in the late tailbud-early tadpole stage (stage ∼26). Thus, the transcription of the Pintallavis gene is zygotic and transient.
To determine the time of onset of Pintallavis expression more precisely, RNA samples were prepared at 1-hour intervals from batches of timed embryos from the same frog starting at the early blastula stage (stage 6) and analysed by RNase protection (Fig. 2B). There is a large increase in abundance of EF-1 alpha transcripts in the 1-hour interval between stages ∼8 and ∼9. This marks the mid-blastula transition, the time at which zygotic transcription commences (Newport and Kirschner, 1982; Krieg et al., 1989). Pin tallavis expression appears first at stage ∼8 and its levels increase markedly by stage ∼9. These result show that Pin tallavis is transcribed at the midblastula transition.
Expression of Pintallavis in the dorsal lip and midline cells
To determine the regional pattern of expression of Pin tallavis RNA we performed in situ hybridization analyses by whole-mount labelling of albino embryos using RNA probes (Harland, 1991). Expression of Pintallavis RNA was first detected in the marginal zone of late blastulae (stage) in an arc of ∼30° (not shown). The level of expression of Pintallavis RNA then increases and by stage before the appearance of the dorsal lip, Pintallavis expression occupies about a 60°arc in the marginal zone (Fig. 3A). Labelling of pigmented embryos shows that the early expression of Pintallavis is confined to the dorsal, less pigmented, side (not shown). By stage 10 when the dorsal lip is first apparent there is a coincidence of the organizer and the expression of Pintallavis RNA (Fig. 3B).
After the onset of gastrulation (stage 10+–11) Pintallavis RNA is present at high levels in cells undergoing convergent-extension movements at the midline, and at lower abundance in more lateral regions (Fig. 3C). By stage 111 a few weakly labelled cells were detected all around the blastopore (not shown). At stage 12 Pintallavis RNA is expressed in the dorsal midline (Fig. 3D, E) including the prechordal region. Expression of Pintallavis RNA in the axial mesoderm is graded with the highest levels detectable in posterior regions (Fig. 3D, F, 4G). At this stage Pin tallavis RNA is expressed at very low levels in the paraxial mesoderm that gives rise to the somites (Fig. 3D and not shown). Expression of Pintallavis decreases at later stages of development and by the mid-late neurula stage (stage 18–20) there is very low expression in the midline (Fig. 3F). By the late tailbud-early tadpole stage (stage ∼28) Pintallavis RNA is expressed solely in the tailbud and at the ventral midline of the newly formed posterior spinal cord (Fig. 3G). Very low levels of Pintallavis RNA were also detected in the tailbud of older tadpoles (stage 36; not shown).
Sections of whole-mount labelled embryos at stage confirmed that Pintallavis RNA is present at high levels in the notochord (Fig. 3I) with lower levels in the prechordal plate mesoderm (Fig. 3J). These sections also revealed that Pintallavis is expressed in the midline of the neural plate overlying the notochord but not above the prechordal mesoderm (Fig. 3H–J). Moreover, transverse sections show that the expression of Pintallavis RNA in the midline of the neural plate appears to be restricted to deep layer cells (not shown). Pintallavis RNA is also expressed in the midline endodermal cells lining the archenteron (Fig. 3H, I). Thus, Pintallavis RNA is expressed by midline cells of the three germ layers.
Pintallavis expression in the dorsal lip requires dorsalizing signals
The presence of Pintallavis RNA in the dorsal lip suggests that the expression of this gene requires signals that confer dorsal axial properties. To examine this possibility we determined the expression of Pintallavis in embryos in which the development of axial structures had been modified. Treatment of blastula stage embryos with high doses of Li+ results in the acquisition of a dorsal character by all mesodermal cells (Kao and Elinson, 1986; Kao et al., 1988). Inversely, irradiation of fertilized eggs before first cleavage with high levels of UV light results in the acquisition of an exclusively ventral character by all mesodermal cells (Scharf and Gerhart, 1983; Cooke and Smith, 1987). To determine whether Pintallavis expression is modified in accordance with the dorso-ventral character of the mesoderm, embryos were treated with high doses of Li+ or irradiated with UV light. In Li+-treated embryos, the normal region of expression of Pintallavis at stage 9–10 in the dorsal lip (Fig. 4A) expands to include the entire marginal zone (Fig. 4B). Inversely, Pintallavis expression is completely absent in UV-treated embryos (Fig. 4C). These results suggest that Pintallavis expression in vivo requires signals that confer dorsal character to the mesoderm.
The early expression of Pintallavis is more restricted than that of the Xenopus Brachyury gene (Xbra; Smith et al., 1991). However, we observed that the expression of Xbra RNA in the marginal zone at stage 9–10 is also graded, with high levels in the dorsal lip region and low levels ventrally (Fig. 4D). Li+ treatment abolished this graded expression resulting in a high levels of Xbra in the entire marginal zone (Fig. 4E). In contrast, UV irradiation resulted in a low level of expression of Xbra RNA throughout the entire marginal zone (Fig. 4F).
The presence of Pintallavis in the dorsal lip together with the alteration of its expression pattern by Li+ and UV suggests that it is regulated by factors that induce dorsal mesodermal cell types (Smith, 1987; Smith et al., 1990; Thomsen et al., 1990). To test this possibility, Pintallavis expression was assayed in animal cap explants treated with activin at concentrations that induce dorsal mesoderm or with basic fibroblast growth factor (bFGF) at concentrations that induce posterior paraxial and ventral mesoderm. No expression of Pintallavis was detected in untreated animal caps or in caps treated with FGF as determined at stages 10 and 20 by in situ hybridization (not shown) and RNase protection assays (Fig. 2C, lane F). As a positive control, FGF was found to induce Xbra expression (not shown; Smith et al., 1991). By contrast, exposure of animal caps to activin resulted in a significant induction of Pintallavis mRNA (Fig. 2C, lane A). This result is consistent with the ability of activin but not FGF-treated caps to behave as an organizer (Ruiz i Altaba and Melton, 1989c; Cooke, 1989) and to induce notochord tissue (Smith, 1987; Slack et al., 1988; Kimmelman and Kirschner, 1988).
In situ hybridization analyses revealed that in activintreated animal caps expression of Pintallavis, unlike that of Brachyury, was usually restricted to a small region of the induced cap (not shown). This provides further evidence for the heterogeneity of animal cap cells that can be revealed by mesoderm-inducing factors (Ruiz i Altaba and Jessell, 1991a; Sokol and Melton, 1991).
The rapid expression of Pintallavis at the mid-blastula transition raised the possibility that its expression in the dorsal lip is a direct response to induction. To test this possibility animal caps were treated with activin in the presence of cycloheximide, to prevent protein synthesis (Cascio and Gurdon, 1987; Rosa, 1989). Treatment with cycloheximide did not change the levels of Pintallavis RNA induced by activin at stage 10 (Fig. 2C, lane AC). This suggests that the transcription of Pintallavis is a direct response to dorsal mesoderm-inducing factors.
Treatment of pre-gastrula Xenopus embryos with retinoic acid (RA) results in a dose-dependent truncation of axial structures in an anterior-to-posterior progression (Durston et al., 1989; Sive et al., 1990; Ruiz i Altaba and Jessell, 1991a). This effect is mediated, in part, by inhibiting the development of dorsal mesoderm (Ruiz i Altaba and Jessell, 1991a). Moreover, in animal caps treated with high concentrations of activin, RA modifies the character of the induced mesoderm from anterior-dorsal to posterior-ventral in a concentration-dependent manner (Ruiz i Altaba and Jessell, 1991a). The restricted expression of Pintallavis in dorsal-axial structures raised the possibility that its expression may be sensitive to RA. Exposure of activininduced animal caps to RA resulted in a decrease in the level of Pintallavis mRNA at stage 10 (Fig. 2C, lane AR). This effect appears to require protein synthesis, since exposure of RA and activin-treated caps to cycloheximide blocked the ability of RA to repress Pintallavis RNA (Fig. 2C, lane ARC).
Pintallavis is not expressed in exogastrula ectoderm
In Xenopus, cells that are located at the midline of the neural plate have been termed the notoplate. The notoplate is derived from cells of the dorsal NIMZ, a cell group adjacent to the dorsal lip mesoderm before gastrulation (Jacobson, 1981; Keller and Danilchik, 1988). The notoplate is formed by the convergent and extension movements of NIMZ cells during gastrulation. The neural expression of Pintallavis in midline neural plate cells could therefore result from its earlier expression in cells in the dorsal NIMZ, the notoplate precursors, or may require inductive signals from the notochord that are transmitted during gastrulation. However, the small size of the dorsal NIMZ makes it difficult to determine whether these cells express Pintallavis prior to gastrulation.
To determine whether Pintallavis is expressed by notoplate cells in the absence of underlying notochord, we examined the expression of Pintallavis in complete exogastrulae. In exogastrulae, the mesoderm does not involute under the ectoderm but the extension movements of the notochord and notoplate still occur, resulting in the elongation but lack of apposition of these two midline structures (Ruiz i Altaba, 1992). In complete exogastrulae examined at stage , Pintallavis was present in the region that contains endoderm and mesoderm, where its expression was detected in the notochord and prechordal mesoderm (Fig. 4G–I). Within the mesoderm, Pintallavis expression exhibited a posterior-to-anterior gradient (Fig. 4G). In contrast, Pintallavis RNA was not detected in the ectodermal region of the exogastrulae, despite the presence of notoplate cells, as defined by their convergent extension movements and of extensive neural tissue (Fig. 4I; Ruiz i Altaba, 1992). Similarly, Xbra expression was absent from the ectoderm of exogastrulae (Fig. 4J). These results show that Pin tallavis is not expressed in the notoplate of exogastrulae. Expression of Pintallavis by cells at the midline of the neural plate in normal embryos may therefore be dependent on inductive or maintenance signals from the underlying notochord.
Injection of synthetic Pintallavis RNA perturbs axial development
To begin to examine the function of Pintallavis we injected synthetic RNA into fertilized eggs. The injected Pintallavis RNA is functional as its translation in vitro (see Materials and methods) yielded a protein of 40–45×103Mr (Fig. 2D), in agreement with its predicted size (Fig. 1A).
Pintallavis RNA was first injected into the equatorial region of a single cell of two-cell stage embryos. Rhodamine/lysine/dextran was routinely coinjected as a lineage tracer to identify the injected side later in development. Since the first cleavage plane defines the axis of bilateral symmetry in most embryos (Klein, 1987), the phenotypes observed in the injected and uninjected sides were compared. We first describe the general features of the phenotypes observed and then analyse mesodermal and neural tissues in more detail.
The injected halves of tadpole-stage embryos displayed a markedly shortened axis, a decrease in or absence of anterior neural structures and a widening of the neural folds of the future hindbrain and spinal cord (Fig. 5A-C; Table 1). In neurula embryos, the head folds and the eye primordia were usually absent from the injected side (Fig. 5B). The cement gland, an anterior ectodermal structure, was also markedly reduced in size (Fig. 5A, B). At tadpole stages, embryos lacked an eye (Fig. 5D) and exhibited a deformed hindbrain with malformed melanocytes (not shown).
The defects in axial development observed after injection of Pintallavis RNA were not observed in control embryos or in embryos injected with RNA encoding a frame-shift mutant, that is predicted to generate a mutated protein that lacks virtually the entire putative DNA-binding domain (Table 1A), or antisense Pintallavis (not shown). Similarly, injection of RNA encoding the rat Isl-1 homeobox gene (Thor et al., 1991) did not cause these defects (not shown). Moreover, injection of Wnt RNAs produced hyperdorsalized and twinned embryos (not shown, see McMahon and Moon, 1989; Smith and Harland, 1991; Sokol et al., 1991), phenotypes that are distinct from those obtained with Pintallavis RNA. Distinct phenotypes have also been obtained by injection of RNA encoding the homeobox gene Xhox3 (Ruiz i Altaba and Melton, 1989b; Ruiz i Altaba et al., 1991) or the helix-loop-helix gene MyoD (Hopwood and Gurdon, 1990). Thus, it appears that the phenotype obtained after Pintallavis injection is specific.
We only occasionally (<5% of injected embryos, see Table 1) observed secondary axes after injection of Pin tallavis RNA (not shown). This suggests that the action of Pintallavis differs from that of Goosecoid, another gene expressed transiently in the dorsal lip (Blumberg et al., 1991) since injection of Goosecoid RNA into the ventral region of early Xenopus embryos results in secondary axes (Cho et al., 1991).
Pintallavis affects axial development when injected into blastomeres that give rise to mesodermal and neural tissues
To determine the cell types in which Pintallavis overex-pression can induce axial defects, we injected Pintallavis RNA into single blastomeres of 32-cell embryos, the fates of which are known (Table 1C; Dale and Slack, 1987a,b; Moody, 1987). Injection of Pintallavis RNA into blas-tomeres that give rise to axial (C1) and paraxial (C2/C3) mesodermal structures resulted in embryos with axial defects (Table 1B). Injection of Pintallavis RNA into blastomeres that give rise to the dorsal equatorial region (C1) which forms the organizer and the axial mesoderm resulted in a high incidence of defective axes. In contrast, there was no obvious effect when Pintallavis RNA was injected into blastomeres that give rise predominantly to ventral mesoderm (C4; Table 1B, C).
Axial defects were also observed after injection of Pin tallavis into blastomeres that normally give rise to neural tissue (A1/B1, B2; Table IB). The occurence of axial defects after injection both into blastomeres that give rise to mesodermal or neural cells suggests that Pintallavis overexpression affects the development of the neural axis indirectly by actions on mesoderm and directly by actions on ectoderm.
Injection of Pintallavis RNA does not transform paraxial into axial mesoderm
The extension of the embryo along the A–P axis is thought to be dependent on forces generated by the paraxial mesoderm (Wilson et al., 1989). The shortened axis observed in injected embryos (Fig. 5C) could therefore be due to a change in fate of paraxial mesodermal cells. However, injected embryos had segmented somites and the diameter of the notochord appeared to be only slightly larger than in control embryos (Fig. 6; and not shown). In addition, the formation of muscle in activin-induced animal caps as detected by the expression of the 12/101 antigen (Kintner and Brockes, 1984) was similar in caps derived from Pin tallavis-injected (n=45) and control embryos (not shown). These observations provide evidence that Pintallavis does not transform the paraxial somitic mesoderm into notochord. Thus, overexpression of Pintallavis, unlike Xhox3 (Ruiz i Altaba and Melton, 1989a), results in a defective axis with apparently normal axial mesoderm. The defects that occur after injection of Pintallavis RNA are partly similar to those observed after treatment of early neurula-stage embryos with RA (Ruiz i Altaba and Jessell, 1991b).
Together, these results suggest that Pintallavis expression does not produce an obvious change in the fate of mesodermal cells even though it appears to alter their extension properties.
Pintallavis overexpression affects neural differentiation
Embryos injected at the 2-cell stage with Pintallavis RNA exhibited defects in anterior neural structures. At the neu-rula stage, the anterior neural folds were absent (Fig. 5B), and by the tadpole stage, anterior neural structures, including the eye were missing (Fig. 5D, 6C). Labelling of affected (n=10) and control embryos with a neural-specific marker, mAb Xen1 (Ruiz i Altaba, 1992) showed that forebrain structures are reduced in size (not shown). In contrast to the decrease in size of forebrain structures, there was an increase in the amount of posterior neural tissue (Fig. 6D, E, G, H), which sometimes formed multiple tubular structures (Fig. 6D, E).
This expansion of posterior neural tissue together with the ability of Pintallavis RNA to cause axial defects when injected into blastomeres that give rise to neural ectoderm (Table 1B, C) raised the possibility that Pintallavis causes neural differentiation when expressed in ectodermal cells. To test this possibility, explanted animal caps derived from embryos injected with Pintallavis RNA into the animal pole of one blastomere at the 2-cell stage were assayed for the presence of neural tissue. Animal caps from injected embryos did not differentiate neural tissue as assessed by the lack of Xen1 immunoreactivity (n=68; not shown). Thus, the overproduction of neural tissue observed in injected embryos is unlikely to result from Pintallavis conferring neural properties to recipient ectodermal cells.
In chick embryos, floor plate grafts placed next to the dorsal neural tube induce ectopic ventral neurons while repressing dorsal neuronal differentiation (Yamada et al., 1991; Placzek et al., 1991). The normal expression of Pin tallavis by midline neural plate cells, which later become the floor plate, raises the possibility that expression of Pin tallavis in dorsal regions affects adjacent neural tissue in a manner analogous to that of floor plate grafts. Since floor plate and notochord grafts repress dorsal neuronal differentiation (Yamada et al., 1991), we examined whether ectopic Pintallavis expression affects the differentiation of dorsal neurons in the spinal cord. At the early tailbud stage (stage ∼24–26) Rohon-Beard sensory neurons (Hughes, 1957) can be identified in the dorsal spinal cord by their morphology and by expression of the HNK-1 epitope (Figs 5E, 6F; Nordlander, 1989). Embryos injected into one side with Pintallavis RNA displayed a unilateral reduction or complete loss of Rohon-Beard neurons (n=10; Figs 5E, F, 6F-H). In addition, in injected embryos there was a decrease in the number of melanocytes and dorsal fin mesenchymal cells which derive from the neural crest (not shown).
Further experiments are required to determine whether Pintallavis expression mimicks a floor plate graft in inducing ectopic ventral neurons in adjacent neural ectoderm.
These results provide evidence that Pintallavis overexpression represses the differentiation of anterior neural structures but increases the amount of neural tissue in the posterior neural tube. In addition, there appears to be a suppression of neuronal differentiation in the dorsal spinal cord.
Discussion
The frog gene Pintallavis is related to the rat HNF-3 and fly fork head genes. Pintallavis is transiently expressed by midline cells in all three germ layers of gastrula and neurula embryos. The phenotype detected after overexpression provides no evidence that Pintallavis is involved in repressing segmentation in midline structures but suggests a function in the development of the neural axis.
Pintallavis is expressed in the organizer
Pintallavis is expressed in the dorsal mesoderm soon after the mid-blastula transition. Like Brachyury (Smith et al., 1991), Goosecoid (Cho et al., 1991) and Xlim1 (Taira et al., 1992), three other genes expressed in the organizer, Pin tallavis transcription is a direct response to mesodermal induction and requires dorsalizing signals. The coincident expression of Pintallavis, Brachyury, Goosecoid and Xlim1 in the dorsal lip raises the possibility that these genes interact to determine the different cell types that derive from the organizer and to control their patterning along the A-P axis. Moreover, the differing lateral extents of the graded expression of Pintallavis and Brachyury observed in mesodermal cells adjacent to the organizer may be important in patterning the mesoderm. Indeed, studies of the mouse Brachyury gene suggest that different levels of its function contribute to mesodermal patterning along the A-P axis (Yanagisawa, 1990; Hermann, 1991).
At later stages of development, the tail bud has organizer-like properties, producing patterned mesodermal and neural tissues (Spofford, 1948). Pintallavis, like Xhox3 (Ruiz i Altaba and Melton, 1989a; Ruiz i Altaba et al., 1991) and Brachyury (not shown), is expressed in this region, consistent with the combined function of these genes in the induction and patterning of axial tissues.
Pintallavis is expressed in the midline
After the onset of gastrulation Pintallavis is expressed in axial mesodermal cells of the prechordal plate and notochord, and in the endoderm of the archenteron roof. Expression of Pintallavis in midline mesodermal cells is graded with lowest levels in the prechordal plate and highest levels in the posterior notochord. Pintallavis is also expressed by cells at the midline of the neural plate that gives rise to the floor plate. A comparison of Pintallavis expression in normal and exogastrula embryos suggests that this gene is not expressed in notoplate cells in the absence of inducing signals from the notochord. Thus, Pintallavis is expressed in midline mesodermal structures derived from the organizer and may be rapidly induced in the overlying neural plate. The expression of Pintallavis in both the notochord and the midline cells of the neural plate may reflect the common properties of these two midline axial structures (Jessell et al., 1988; Wagner et al., 1990; Jessell and Dodd, 1992). It remains unclear whether the notochord induces the expression of Pintallavis in endodermal cells of the archenteron roof.
The expression patterns of Pintallavis and XFKH1 (Dirksen and Jamrich, 1992) are similar but differ in two aspects. First, Pintallavis is expressed before stage 9 in a restricted region (∼30° of arc) of the dorsal marginal zone, whereas XFKH1 is transcribed at similar stages over a much wider area (>90°) and only later becomes restricted. By stage 10, the onset of gastrulation, the expression of both genes is coincident with the site of dorsal lip formation. Second, at the late neurula stage, Pintallavis expression decreases rapidly and anterior regions exhibit little or no Pintallavis mRNA. In contrast, XFKH1 mRNA is still evident in midline cells that will form the floor plate of the hindbrain.
The identity of mesodermal cells is not affected by Pintallavis overexpression
Injection of Pintallavis RNA into cells that give rise to axial and paraxial mesoderm results in embryos that display pronounced axial defects. However, there are no overt defects in the posterior mesoderm. Expression of high levels of Pin tallavis RNA in presumptive somitic mesodermal cells may disturb their normal movements without affecting their specification. Elevated levels of Pintallavis RNA could also cause anterior mesodermal cells to acquire a more posterior character. In this case, the embryo might elongate less, since posterior axial cells appear to undergo less pronounced extension movements than their anterior counterparts (Symes and Smith, 1987; Keller and Danilchik, 1988). In either case, it appears that expression of Pintallavis in presumptive somitic mesoderm affects only some properties of the recipient cells without altering their overall determination. This is similar to the result observed after ectopic expression of myoD in Xenopus, where non-muscle cells acquire some muscle properties; for example, the expression of the alpha actin gene, in the absence of myogenesis (Hopwood and Gurdon, 1990).
Overexpression of Pintallavis affects neural development
Pintallavis overexpression has distinct neural phenotypes. First, the development of anterior neural structures, such as the eye, is repressed. Second, there is an overproduction of posterior neural tissue. Third there is a repression of dorsal neuronal differentiation. The effects of overexpression of Pintallavis on the differentiation of anterior neural structures could be due to an increase in the low level of Pin tallavis RNA that is normally present in the prechordal plate. As a consequence, these cells may acquire a more posterior character eliminating their ability to induce forebrain structures.
The increase in posterior neural tissue observed after injection of Pintallavis RNA could reflect an enhanced neural-inducing activity of paraxial mesoderm. In Xenopus, neural-inducing activity is concentrated in the notochord, where Pintallavis expression is high, but is also present at much lower levels in the paraxial mesoderm (Jones and Woodland, 1989), where Pintallavis expression is very low. Indeed, preliminary experiments suggest that presumptive somitic mesoderm isolated from embryos injected with Pin tallavis RNA has a greater ability to induce neural tissue in recombinates with animal cap ectoderm than equivalent regions of mesoderm from control embryos (not shown).
The neural phenotype observed after Pintallavis overexpression could also be related to its expression at the midline of the neural plate, which later gives rise to the floor plate. Since the floor plate is a ventral neural structure present in the hindbrain and spinal cord but absent from the forebrain, the neural phenotypes observed after injection of Pintallavis RNA could result from the partial acquisition of floor plate properties by neural cells. The repression of the differentiation of anterior neural cells could result from the acquisition of floor plate properties in the anterior neural tissue. Similarly, repression of the differentiation of dorsal neurons could reflect a dorsal expansion of floor plate properties (Yamada et al., 1991).
Pintallavis and neural induction
The pattern of expression of Pintallavis RNA in midline cells with inductive properties, together with the neural phenotypes that result from its injection suggest an involvement of Pintallavis in neural induction.
Pintallavis alone, however, does not appear to be able to direct neural differentiation, since animal cap ectoderm isolated from injected embryos does not develop into neural tissue. Moreover, Pintallavis is not expressed by all neural plate cells which might be expected if its expression were required in ectodermal cells that become neural. Indeed, the lateral extent of the neural plate is thought to be regulated by the propagation, from cell to cell, of a midline-derived neuralizing signal combined with a gradual loss of competence of the responding ectoderm (Leussink, 1970; Nieuwkoop and Albers, 1990; Servetnick and Grainger, 1991). In addition, Pintallavis expression in exogastrula embryos is not detected in the notoplate, a cell group implicated in neural induction (Ruiz i Altaba, 1992).
Neural induction may depend on the combined action of Pintallavis and other mesodermal genes, such as Brachyury. Such genes, expressed in the dorsal lip, could initiate the production of a neural-inducing signal. The spread of this signal, from the dorsal lip to adjacent ectoderm, and its subsequent propagation through the ectoderm (Leussink, 1970) may, however, be independent of the mesodermal genes responsible for the initiation of the signal. Thus, cells of the notoplate may acquire neural-inducing properties (Ruiz i Altaba, 1992) in the absence of Pintallavis expression.
The graded expression of Pintallavis along the A-P axis of the mesodermal midline could contribute to the ability of the notochord to induce a floor plate in the overlying neural ectoderm. The expression of Pintallavis by midline neural plate cells, the floor plate precursors, may then be involved in specifying the floor plate and its signalling properties.
ACKNOWLEDGEMENTS
We thank Richard Harland for providing the whole-mount in situ protocol prior to publication, Jim Smith for the Xbra clone, Cliff Hume for inducing factors, Andy Furley for help with the computer programs and Eseng Lai for communicating the corrected HNF-3 beta and gamma sequences. A.R.A. is an associate and T.M.J an investigator of the Howard Hughes Medical Institute.