ABSTRACT
We have isolated a novel Xenopus homolog of the Drosophila achaete-scute genes, called XASH-3. XASH-3 expression is neural specific and is detected as early as stage 11, making it one of the earliest markers of neural induction so far described. Moreover, XASH-3 expression within the neural plate is regionally restricted. Transverse bands of XASH-3 mRNA mark discrete positions along the anteroposterior axis, while longitudinal bands mark a discrete position along the mediolateral axis. This latter site of XASH-3 expression appears to demarcate the prospective sulcus limitans, a boundary zone that later separates the functionally distinct dorsal (alar) and ventral (basal) regions of the spinal cord. In sandwich explants lacking any underlying mesoderm, XASH-3 is expressed in longitudinal stripes located lateral to the midline. This provides the first indication that planar or midline-derived inductive signals are sufficient to establish at least some aspects of positional identity along the mediolateral axis of the neural plate. By contrast, the transverse stripes of XASH-3 expression are not detected, suggesting that this aspect of anteroposterior neural pattern is lost or delayed in the absence of vertically passed signals. The restricted mediolateral expression of XASH-3 suggests that mediolateral patterning of the neural plate is an early event, and that this regionalization can be achieved in the absence of inducing signals derived from underlying mesoderm.
INTRODUCTION
The identification of genes that determine cell type and pattern within the nervous system is a fundamental problem of developmental neurobiology. In Drosophila, the achaetescute complex (AS-C) of four homologous genes (achaete, scute, lethal of scute and asense) is required for neuronal determination in both the central and peripheral nervous systems (Cabrera, 1992; Campuzano and Modolell, 1992). The AS-C genes are members of the basic helix-loop-helix (bHLH) family of transcription factors which includes myogenic determination factors such as MyoD (Weintraub et al., 1991). These tissue-specific proteins form heterodimers with more broadly expressed bHLH proteins such as mammalian E12 and its Drosophila homolog, daughter - less (Murre et al., 1989a). These heterodimers mediate their transcriptional regulatory function through a six base pair core consensus binding sequence, termed the E box element (Murre et al., 1989b).
Two mammalian homologs of the Drosophila achaete-scute genes, MASH1 and MASH2, have previously been isolated from a rat sympathoadrenal progenitor cell line (Johnson et al., 1990). These genes share approximately 80% identity to the Drosophila achaete-scute genes within their bHLH regions but diverge from the Drosophila genes, as well as from each other, outside of this region. Although the expression of MASH2 in vivo is not neural specific (Johnson et al., 1992b; K. Zimmerman, unpublished data), MASH1 expression is restricted to neural progenitor cells within discrete regions of the brain and spinal cord and to the developing sympathetic, parasympathetic and enteric ganglia in the peripheral nervous system (Guillemot and Joyner, 1993; Lo et al., 1991; L. Lo, unpublished observations). More recently, a Xenopus achaete-scute homolog has been isolated that shares 75% homology throughout its protein coding sequence and 100% identity within its bHLH domain with MASH1 (Ferreiro et al., 1992). XASH1 is not expressed at appreciable levels prior to neural tube closure (stage 20) but is then expressed in a regionally specific pattern within the developing brain and retina (Ferreiro et al., 1992). Unlike MASH1, XASH1 is not expressed in either the spinal cord or in any neural crest derivatives (Ferreiro et al., 1992).
The regionally and temporally restricted pattern of MASH1 and XASH1 expression suggested that additional achaete-scute homologs might be required at other places or times of neural development. Based on homology within the bHLH region, we have isolated a novel achaete-scute homolog from Xenopus, which we call XASH3. The expression of XASH3 is neural specific and regionally restricted. Expression is detected as early as the mid-gastrula stage (stage 11-11 ), making XASH3 one of the earliest regional markers of neural induction thus far described.
We have used XASH3 as a marker to examine the relative contributions of planar and vertical inductive signals to axial patterning of the neural plate. Previously, it had been shown that markers of anteroposterior position along the neuraxis exhibit appropriate regional localization in explants or exogastrulae lacking vertical inducing signals (Doniach, 1992; Papalopulu and Kintner, 1993; Ruiz i Altaba, 1990, 1992). Other studies have shown that some neuronal cell types characteristic of both dorsal and ventral regions of the spinal cord can differentiate in the absence of vertical inducing signals (Ruiz i Altaba, 1992). However, these latter studies leave open the issue of whether planar signals are sufficient to specify positional information correctly, independent of cell type, along the prospective dorsoventral axis. XASH3 expression defines a unique position along this axis, demarcating the boundary between the presumptive dorsal and ventral halves of the neural tube. Analysis of XASH3 expression in explants indicates that this boundary can form in the absence of vertical inducing signals. Taken together, these data support the idea (Yamada et al., 1993) that the inductive signals that determine at least some aspects of positional identity along the prospective dorsoventral axis operate very early, at the neural plate stage.
MATERIALS AND METHODS
Molecular cloning of XASH3 cDNA
Random primed cDNA was prepared from embryonic stage 12-14 Xenopus total RNA. Fully degenerate oligonucleotide primers, which corresponded to the regions indicated in Fig. 1, were synthesized. Inosine residues were substituted at all positions of 4-fold degeneracy and a BamHI or EcoRI restriction site was included on the 5′ or 3′ primer respectively. The sequence of the 5′ and 3′ primers were ccggatccAA(C/T)(A/C)GIGTIAA(A/G)(C/T)TIGTIAA and ccgaattcC(G/T)(A/G/T)AT(A/G)TA(C/T)TCIACIGC, respectively. PCR amplification conditions were as recommended by the manufacturer (Perkin Elmer/Cetus) with an annealing temperature of 45°C. PCR products were digested with BamHI/EcoRI, subcloned into pGEM-3 and sequenced using standard methods (Sequenase/USB). The XASH3 PCR amplified cDNA was used to probe a stage 17 neural plate library (Kintner and Melton, 1987). The two cDNA clones isolated, X3.A and X3.B, were subcloned in pBS-KS and the coding region of each was sequenced in its entirety on both strands.
Electrophoretic mobility shift and transcriptional activation assays
In vitro transcription, translation and electrophoretic mobility shift assays (EMSA) were according to previous experiments (Johnson et al., 1992a). The X3.A cDNA was linearized with HindIII or NotI and transcribed with T7 or T3 RNA polymerase to generate the sense and antisense transcripts, respectively. The sequence of the top strand of the oligonucleotide used in gel shift assays was as follows, GATCCCCCCAACACCTGCTGCCTGA. Transcriptional activation assays were as previously described (Johnson et al., 1992a). X3.A was subloned into the pRSVSV40 expression vector in both orientations (Johnson et al., 1992a). The pRSVMASH1, (2E)80MCKCAT and pRSVL constructs were as previously described (de Wet et al., 1987; Johnson et al., 1992a).
In situ hybridization analysis
The X3.A cDNA was linearized with NotI or HindIII and transcribed with T3 or T7 RNA polymerase to generate antisense and sense XASH3 probes, respectively. The full-length X3.A antisense probe was used for all in situ hybridization analyses shown. For control experiments, the X3.B cDNA was linearized with XbaI or HindIII and transcribed with T3 or T7 RNA polymerase to generate antisense and sense probes, respectively. In addition, each cDNA was linearized with PvuII and transcribed with T3 RNA polymerase to generate antisense probes, which lacked the bHLH domain. A 500 bp 3′ BamHI/EcoRI fragment of X3.A and a 600 bp internal BamHI fragment from X3.B, which lacked the bHLH domain, were also used as probes in control experiments. The N-CAM cDNA, pTN1, has been described previously (Krieg et al., 1989).
Albino Xenopus laevis embryos were staged according to Nieuwkoop and Faber (Nieuwkoop and Faber, 1967). Wholemount in situ hybridization analysis was performed using previously described techniques (Harland, 1991; Hemmati-Brivanlou et al., 1990). Modification to the published procedure included preblocking the alkaline phosphatase-conjugated digoxigenin antibody. Xenopus embryos were rehydrated, minced and incubated in PTw (PBS + 0.1% Tween-20) containing 10% sheep serum at room temperature for 1 hour. The liquid was then removed from the embryos and a 1:200 dilution of the digoxigenin antibody in PTw containing 1% sheep serum was added. The antibody was preblocked for greater than 3 hours prior to use. The ratio of embryo to antibody volume during this preblock was approximately 1:5. In addition, the washing steps following antibody incubation were extended to seven 1 hour washes. Reactions were stopped by fixation in MEMFA. Fixation was carried out at 4°C overnight. The embryos were then dehydrated in methanol and cleared in a 2:1 mixture of benzyl benzoate:benzyl alcohol. Selected embryos were mounted in paraplast and sectioned at 10 μm thickness.
Preparation of explants
‘Closed face’ sandwich explants were prepared exactly as described by Keller and colleagues (Keller and Danilchik, 1988; Keller et al., 1992). All explants were made to include a radial sector of no more than 100° centered on the dorsal midline of stage 10+ embryos. While the entire height of the neural plate is included in the explants, the animal cap is excluded. Although the explants are made by initially apposing two similar pieces of tissue, extensive cell mixing between these two sheets eliminates this initial mirror duplication (Keller et al., 1992), so that the explant behaves like a single, integrated piece of tissue. Explants were analyzed at stage 17, a stage at which the longitudinal and transverse domains of XASH3 expression are clearly visible in normal embryos (Fig. 7A). In contrast, previous studies using Keller sandwich explants to examine aspects of anteroposterior patterning have typically analyzed the explants at stage 26 (Doniach et al., 1992; Ruiz i Altaba, 1992). In our experiments, sibling explants from the same batches allowed to develop to stage 26 exhibited the typical, narrow constriction of the neurectoderm indicative of successful convergent extension and an absence of invading mesoderm, as has been extensively documented in earlier studies (Doniach, 1992; Keller and Danilchik, 1988).
RESULTS
Isolation of achaete-scute homologs from Xenopus
We were interested in isolating achaete-scute homologs involved in early stages of neural determination in Xenopus laevis. We therefore prepared RNA and synthesized cDNA from embryos at the late gastrula/early neurula stages (stage 12-14) of development. PCR reactions were carried out using fully degenerate primers which represented conserved regions of the 5′ and 3′ helical regions of the Drosophila achaete-scute and mammalian MASH genes. Characterization of the products of these PCR reactions revealed two groups of sequences with homology to the achaete-scute genes. The first group was identical to MASH1 in its protein coding sequence and was identical to the previously characterized XASH1 gene (Ferreiro et al., 1992) in nucleotide sequence.
A second group of clones shared approximately 70% nucleotide sequence and 75% protein sequence homology with MASH1 and MASH2 within the region amplified (Fig.1, sequence between primers). The equal divergence of this gene from both MASH1 and MASH2, as well as the isolation of a Xenopus MASH1 homolog with 100% protein homology, suggests that this is not a Xenopus homolog of a previously isolated mammalian gene but is instead a unique vertebrate achaete-scute homolog. We have named this gene XASH3. The same gene has been isolated independently by D. Turner and H. Weintraub (D. Turner, personal communication). Attempts to identify a mammalian XASH3 homolog using PCR and low-stringency Southern blot analysis have been unsuccessful (K. Zimmerman, unpublished data).
In order to characterize more fully the XASH3 gene, a stage 17 neurula cDNA library (generously provided by C. Kintner and B. Harris) was screened. Two independent cDNA clones were isolated. These clones, designated X3.A and X3.B, were approximately 2.4 and 1.4 kb in length, respectively. The protein coding regions of both cDNA clones were sequenced. The X3.A clone contains a 160 amino acid open reading frame downstream of an in-frame methionine (Fig. 2). We are currently unable to determine whether this is the actual start codon for the X3.A protein as the cDNA clone extends only 81 bp 5′ to this methionine and no stop codons are located upstream. However, the region upstream of this potential start codon shares 5/6 nucleotides with the Kozak translation initiation consensus sequence (consensus CC(A/G)CCAUG, XASH3 CCGCGAUG) including the important −3 position suggesting that this may be the actual start site of the XASH3 protein (Kozak, 1984). The second cDNA clone, X3.B, begins 156 nucleotides 3′ of the X3.A clone and therefore contains no start codon.
Comparison of the X3.A and X3.B protein coding regions indicates that these clones encode related but distinct proteins sharing 75% amino acid homology (Fig. 2). The most obvious point of divergence is the C terminus of the protein where the X3.B coding region extends 25 amino acids beyond the X3.A stop codon due to a single nucleotide insertion in the X3.B clone. If this region is not included in the homology calculation, the degree of amino acid conservation is approximately 90%. The conservation of sequences outside of the bHLH domain, a region not conserved between XASH3 and XASH1, suggests that the X3.A and X3.B genes may result from the duplication of the Xenopus genome rather than representing two unique achaete-scute homologs. The homology between XASH3 and the MASH1 and MASH2 proteins within the bHLH region is approximately 90% (Fig. 1). XASH3 shares only 60% homology with the Drosophila scute protein, the most related member of the achaete-scute family, but the divergence between these two proteins is largely accounted for by the additional ten amino acids within the scute loop region (Fig. 1). Both XASH3 proteins diverge from the Drosophila achaete-scute genes and their previously characterized vertebrate homologs outside of the bHLH region.
XASH3 functions as a transcriptional regulator
In vitro characterization of the binding sites of a number of bHLH-containing proteins has established a consensus binding sequence of 6 nucleotides (CANNTG) termed the E box element (Murre et al., 1989b). To assess the function of XASH3, we first tested its ability to bind an E box sequence in a gel shift assay, using protein synthesized by in vitro transcription and translation from a cloned cDNA template. Previous studies with the achaete-scute proteins as well as with MASH1 and MASH2 have demonstrated that they bind E box sequences only as heterodimers with another member of the bHLH family, the Drosophila daughterless protein and its vertebrate homolog, E12, respectively (Johnson et al., 1992a; Murre et al., 1989b). The combination of the in vitro-synthesized XASH3 and E12 proteins also binds an E box-containing oligonucleotide, and the resulting DNA-protein complex is equal in intensity to that observed for the combination of the MASH1 and E12 proteins (Fig. 3A, lanes 3 and 4). This result not only confirms that XASH3 is a DNA-binding protein, but also establishes that the protein encoded within the X3.A cDNA clone, even if not full length, is sufficient to carry out this function.
We next tested the ability of XASH3 to act as a transcriptional regulator. MASH1 and MASH2 have been shown to activate specifically transcription in transfected 10T cells via an E box sequence located within the muscle creatine kinase (MCK) promoter (Johnson et al., 1992a). We employed the same assay to test the transcriptional regulatory capacity of the XASH3 protein. 10T cells were cotransfected with a XASH3 expression construct and an MCK/CAT reporter construct, which contained a minimal MCK promoter with two upstream E box-containing oligonucleotide sequences. In XASH3-transfected cells, an approximately 10-fold increase in CAT activity was detected (Fig. 3B, lane 4). A similar enhancement was seen in MASH1-transfected cells (Fig. 3B, lane 2). These results indicate that XASH3, like other members of the bHLH family, is a DNA-binding, transcriptional regulatory protein.
The pattern of XASH3 expression during development
We have characterized the expression of XASH3 during Xenopus embryonic development using the technique of whole-mount in situ hybridization (Hemmati-Brivanlou et al., 1990). For most studies, the entire X3.A cDNA clone was used as a probe. To confirm that the expression pattern that we observed with this probe was not due to cross-hybridizing genes, we also used two non-bHLH-containing probes from both the X3.A and X3.B genes. All probes tested yielded a similar expression pattern indicating that the hybridization pattern that we observe is specifically that of the XASH3 gene rather than related bHLH-containing genes. We were not able to distinguish whether both the X3.A and X3.B genes are expressed in a similar pattern or if the signal that we observe is due to cross-hybridization, as all regions of the two genes cross-hybridize at high stringency in plasmid blotting assays. For this reason, the expression pattern described will be referred to as that of XASH3 without reference to the specific gene. Comparison of the signal intensity of XASH3 to that of control probes such as Krox-20, actin and Xtwist indicated that XASH3 mRNA is a very low abundance transcript, and may explain our failure to detect it by northern analysis (data not shown).
XASH3 expression is first detected at the mid-gastrula stage (stage 11-11 ) of development in two symmetric patches of expression lateral to the midline (Fig. 4A, closed arrow). In neurula stage embryos (stage 14-20), these patches become condensed and extended, appearing as two bilaterally symmetric stripes (Fig. 4B and C, closed arrow). These stripes run the length of the presumptive spinal cord and terminate anteriorly within the hindbrain region. A transverse band of XASH3 expression is observed at the anterior end of each stripe, first detectable as a tangential ‘eyebrow’ like zone of expression at stage 11-11 (Fig. 4A-C, open arrow). Cohybridization of late neurula embryos (stage 18) with a mixture of Krox-20 and XASH3 probes indicates that the ‘eyebrow’ stripes of XASH3 expression are located immediately anterior to the third rhombomere (data not shown). A fainter horizontal stripe of XASH3 expression is observed in late neurula stage embryos (stage 17+) just anterior to the ‘eyebrow’ stripes, near the presumptive hindbrain/midbrain junction region (Fig. 4C, arrowhead). Diffuse staining of XASH3 is observed in more anterior regions of the developing nervous system with more intense signal being detected at the anterior most end of the prosencephalon (Figs 4C, 5A). In addition, faint staining is detected in scattered cells located immediately lateral to the cephalic neural plate (Fig. 5A, open arrow). These cells may represent a subset of prospective cranial neural crest cells.
Comparison of the XASH3 expression pattern to that of the neural cell adhesion molecule, N-CAM, at stage 13 indicates that the XASH3 expression domain is restricted within the region of the neural plate defined by N-CAM expression (Fig. 5A,D). To define more clearly the mediolateral boundaries of XASH3 expression within the neural plate, we examined cross-sections of early neurula stage embryos (stage 13/14) that had been hybridized with the XASH3 probe. Within the region of the embryo where XASH3 appears as bilaterally symmetric longitudinal stripes, expression is restricted to a group of cells located near the midpoint of the medial and lateral extremes of the neural plate, above the center of the somites (Fig. 5B). By contrast, N-CAM mRNA is more broadly distributed along the mediolateral axis and is less abundant at this stage (Fig. 5E). In the region containing the ‘eyebrow’ stripes, the XASH3 expression domain is expanded along the mediolateral axis although expression is absent from the ventral midline (Fig. 5C).
Following neural tube closure (stage 20+), XASH3 expression persists in the neural tube as two bilaterally symmetric longitudinal stripes. Cross-sections of embryos following neural tube closure reveal that XASH3 expression is restricted to a group of cells located at the midline of the dorsoventral axis, the sulcus limitans (Fig. 6B,C, closed arrows). Expression continues within this region at least until stage 38, the latest stage which we have examined. A second, more ventral, stripe of XASH3 expression is also observed in the hindbrain region at later stages (stage 20+) of development (Fig. 6A, closed arrow). At stage 25+, this ventral stripe of XASH3 expression can be clearly resolved in whole-mount specimens into three longitudinally oriented dashes. These dashes are longer than the length of a single rhombomere by comparison to Krox-20 expression in whole mounts (data not shown). Cross-sections within the hindbrain region reveal that this second stripe of XASH3 expression is located in a region of the neural tube immediately adjacent to the floor plate (Fig. 6C, open arrow).
XASH3 continues to be expressed in a diffuse pattern throughout midbrain and forebrain regions in all embryonic stages examined. XASH3 expression is also detected in the developing retina (Fig. 6A). In the eye vesicle, XASH3 appears as early as stage 21 and continues to be expressed within the retinal marginal zone as late as stage 38 (data not shown). Later stages of development have not yet been examined.
XASH3 expression in explant cultures
The expression of XASH3 in the neural plate appears to demarcate the presumptive sulcus limitans, a major landmark of dorsoventral pattern in the spinal cord. As other aspects of dorsoventral pattern have been shown to be dependent on vertical inducing signals, we examined whether the pattern of XASH3 expression was similarly dependent on such signals (Ruiz i Altaba, 1992; Ruiz i Altaba and Jessell, 1992). We chose to address this question using sandwich explants (Keller and Danilchik, 1988). The absence of vertical mesoderm-ectoderm contact in properly made explants is well documented and therefore the specific effects of planar inducing signals can be examined in the absence of vertical signalling (Keller and Danilchik, 1988; Keller et al., 1992). We examined two components of XASH3 expression: the longitudinal stripes in the spinal cord/hindbrain region (Fig. 7A, dark arrow) and the ‘eyebrow’ stripes in the anterior hindbrain (Fig. 7A, open arrow). These aspects of XASH3 pattern are readily visible in stage 17 embryos, the stage at which the explants were examined (Fig. 7A). In 12/13 explants, XASH3 expression was observed in the neuroectodermal region of the explant. Moreover, XASH3 expression was restricted with respect to the mediolateral axis of the neural plate. In some cases (4/13), two parallel longitudinal stripes similar to those seen in normal embryos were observed (Fig. 7B and C, dark arrows).
Several aspects of the XASH3 expression pattern in explant cultures differed from those observed during normal development. In a number of explants (8/13), particularly those in which convergence and extension of the neuroectoderm was asymmetric, XASH3 expression was not bilaterally symmetric (Fig. 7D, dark arrow). The mediolateral position of XASH3 expression also differed from that seen in normal embryos. XASH3 expression no longer defined the midpoint of the mediolateral axis of the neuroectoderm but was instead located more laterally, particularly within the posterior ‘neck’ of the explant. Another striking difference in the pattern of XASH3 expression in explant cultures was the consistent absence of the ‘eyebrow’ stripes of XASH3 expression within the hindbrain. XASH3 expression was not detected at this position in any explant examined (0/13). Thus in sandwich explants lacking an underlying source of mesoderm-derived vertical inducing signals (see Materials and Methods), the longitudinal and transverse components of the XASH3 expression pattern can be dissociated from one another. The longitudinal component is retained while the transverse component is apparently lost.
DISCUSSION
XASH3 expression
The Drosophila achaete-scute proteins function as neuronal determination factors in both the central and peripheral nervous systems (Cabrera, 1992; Campuzano and Modolell, 1992). Two mammalian homologs of these genes, MASH1 and MASH2, have previously been characterized and a Xenopus homolog of MASH1, called XASH1, has also been identified (Ferreiro et al., 1992; Johnson et al., 1990). We have isolated a novel Xenopus homolog of the Drosophila achaete-scute genes, which we call XASH3. Expression of the XASH3 gene is neural specific and appears in a regionally restricted population of central nervous system precursor cells at the time of neural induction. The early and restricted expression of XASH3 superficially parallels the spatially restricted and progenitor-specific expression patterns observed for the AS-C genes during Drosophila neural development (Campuzano and Modolell, 1992).
Possible role of XASH3 in patterning the neural plate
The expression of XASH3 in mid-gastrula stage embryos makes it one of the earliest markers of neural regionalization thus far described. Moreover, XASH3 is expressed in a regionally restricted manner with respect to both the antero-posterior and mediolateral axes of the neural plate. The mediolateral restriction is particularly striking because it appears to correspond to the presumptive sulcus limitans, a boundary that separates the functionally distinct alar and basal plates of the spinal cord. Emerging data from several organisms suggest that diffusible signals produced at both the midline and prospective dorsal margins of the neural plate may control patterning of the dorsoventral axis of the developing spinal cord (Basler et al., 1993; Placzek et al., 1993; Yamada et al., 1991, 1993). Later in development, regulatory genes such as Pax-3 and Pax-6 are expressed in a non-graded manner within the alar or basal plates, respectively, and terminate abruptly at the emerging sulcus limitans (Goulding et al., 1991; Gruss and Walther, 1992; Walther and Gruss, 1991) (Fig. 8C). This suggests that graded, continuous information produced at the prospective dorsal and ventral extremes of the neural tube may first be converted into several discontinuous mediolateral zones of regulatory gene expression, prior to the specification of cell type (Fig. 8C,D). Such a mechanism would be analagous to the early stages of anteroposterior axis patterning in Drosophila (Nüsslein-Volhard, 1991).
The expression of XASH3 in the presumptive sulcus limitans may play an important role in this conversion, by establishing or maintaining an early zone of discontinuity along the mediolateral axis of the neural plate (Figs 5B, 8B). One possibility is that this zone may form a boundary that sharpens the transition point from dorsal to ventral character of the neural tube, for example by perturbing gradients of diffusible signals (see Fig. 8C). Alternatively, XASH3 could be a consequence rather than a cause of boundary formation, and might determine cell fates appropriate to that region. Whatever the case, the zone of XASH3 expression could identify one of the first responses of neural plate cells to dorsal or ventral midline-derived gradient signals, and may define an important early event in converting this continuous information to discontinuous domains of cellular differentiation. While we have thus far failed to detect XASH3 homologs in higher vertebrates, a homeobox gene specifically expressed at the sulcus limitans has been identified in both chick (Rangini et al., 1991) and mouse (Lu et al., 1992), supporting the idea that molecular specification of this
Planar and vertical inductive interactions
Classical experiments suggest that both planar and vertical inducing signals derived from the mesoderm may control the patterning of the amphibian neural plate (Spemann, 1938). In exogastrulae and explant cultures, two experimental systems in which no vertical contact between mesoderm and ectoderm is established, planar interactions are sufficient for inducing both general neural properties such as N-CAM expression (Dixon and Kintner, 1989; Kintner and Melton, 1987) and a number of markers of anteroposterior position (Doniach et al., 1992; Papalopulu and Kintner, 1993; Ruiz i Altaba, 1990, 1992). Vertical interactions are important for the patterning of the ventral spinal cord in Xenopus. In both exogastrula and UV-irradiated embryos (which lack a notochord), the floor plate, a ventral midline structure, does not develop and the number of ventral motor neurons is greatly reduced (Clarke et al., 1991; Ruiz i Altaba, 1992; Ruiz i Altaba and Jessell, 1992).
The appearance of XASH3 in bilaterally symmetric stripes in our explants indicates that vertical signals are not essential for inducing at least some aspects of mediolateral patterning. In these explants, XASH3 may be induced exclusively by planar signals derived from adjacent mesoderm; alternatively planar inductive signals may derive from midline sources within the neurectoderm such as the notoplate, which has been suggested to account for axial patterning in Xenopus exogastrulae (Ruiz i Altaba, 1992). In a number of explants, XASH3 was not expressed at equivalent levels on both sides. This lack of bilateral symmetry was particularly apparent in cases where convergence and extension movements were asymmetric (Fig. 7D). A similar asymmetry is observed in N-CAM staining in such explants (J. Shih, unpublished data) suggesting a general failure to achieve symmetric neural induction. This variability may reflect an insufficient concentration and/or transmission of mesoderm-derived signals or an asymmetric orientation of the notoplate in these explants, and is consistent with the idea that in vivo supplementary vertical signals ensure consistent and symmetric induction (Spemann, 1938).
The apparent absence of the transverse ‘eyebrow’ stripes of XASH3 expression was surprising, given that other markers of anteroposterior position, such as Krox-20, are expressed in explants assayed at later stages (Doniach et al., 1992; Ruiz i Altaba, 1992). While we cannot rule out that the transverse stripes were simply distorted or reoriented longitudinally, such an explanation is not supported by maps of the distortions that occur in anterior regions of Keller sandwiches (Keller et al., 1992). Whether this result indicates an absolute requirement of XASH3 for vertical inducing signals, or rather a delay in the onset of its expression in explants, remains to be determined. Preliminary data suggest that Krox-20 induction in explants may in fact be delayed relative to its normal time of onset in vivo (K. Zimmerman and J. Shih, unpublished data). Such a delay could reflect the time required for the transmission of planar inducing signals to the anterior region of the explant. This delay would not occur in vivo because the underlying axial mesoderm would provide a local source of vertical inducing signals.
Previous studies have indicated that specific neuronal cell types characteristic of dorsal and ventral spinal cord can form in the absence of vertical inducing signals (Ruiz i Altaba, 1992). The restricted expression in explants of XASH3, a marker of proliferating neuroepithelial precursor cells, suggests that planar or notoplate-derived signals are also sufficient to determine aspects of positional identity, as well as cell type, along the mediolateral axis of the neural plate. The apparent requirement for underlying mesoderm to induce the most medial markers of the neural plate, such as pintallavis (Ruiz i Altaba and Jessell, 1992), suggests that different components of the mediolateral pattern may require different amounts and/or types of inducing signals. Furthermore, the mediolateral restriction of XASH3 expression in explants which presumably lack Pintallavis expression (Ruiz i Altaba and Jessell, 1992), indicates that Pintallavis itself is not normally responsible for exclusion of XASH3 expression from the midline region.
Control of XASH3 expression in embryos and explants
How is the XASH3 expression domain positioned precisely at the midpoint of the medial and lateral limits of the neural plate? Recent studies in the chick and rat spinal cord have identified diffusible signals deriving from both the medial and lateral margins of the neural plate as important determinants of pattern (Basler et al., 1993; Placzek et al., 1993; Yamada et al., 1993; Yamada et al., 1991). The site of XASH3 expression could be determined by a particular scalar value of a single gradient (Fig. 8A, right) (Yamada et al., 1991), or by the intersection of two opposing signal gradients (Fig. 8A, left). Such models could also account for the lateral displacement of XASH3 expression observed in sandwich explants. The tissue excised to generate the explant (see Materials and Methods) excludes the prospective sulcus limitans (Fig. 8E, see legend). Thus this explant should also exclude the prospective domain of XASH3 expression (Fig. 8F). The recovery of XASH3 expression suggests that the explant undergoes some degree of proportion regulation along the mediolateral axis. This is consistent with previous studies that have documented proportion regulation in explants (Holtfreter, 1965). As the edge of the explant progressively acquired more lateral character (Fig. 8G), it would be expected to achieve positional values that induce XASH3 expression before achieving values characteristic of an extreme lateral margin. XASH3 would therefore first appear at the edge of the explant. Alternatively, the edge of the explant might undergo incomplete lateralization to a value permitting XASH3 expression, and then stop. Distinguishing these possibilities by examining XASH3 expression in later stage explants is complicated by the extreme narrowing of the posterior neural plate. Nevertheless, these considerations illustrate how models that invoke gradient signals to explain the positioning of XASH3 in normal embryos (Fig. 8A) predict that XASH3 should be detected at the edge of explants that undergo lateralization (Fig. 8F-H), consistent with our experimental observations.
In summary, we have identified a novel vertebrate homolog of achaete-scute that, like the Drosophila genes, is specifically expressed in the developing nervous system. The expression pattern of this gene suggests that neural induction at its earliest stages is accompanied by subdivision of the neural plate along its mediolateral axis, in particular by the establishment of a boundary between the prospective alar and basal plates of the spinal cord. Experiments in explants suggest that this aspect of presumptive dorsoventral pattern, like some components of anteroposterior pattern, can be induced in the absence of vertical signals deriving from underlying mesoderm. Such observations suggest that regionalization along the prospective dorsoventral axis of the neural tube occurs at a much earlier stage of development than previously anticipated.
ACKNOWLEDGEMENTS
The authors are particularly grateful to Drs Beatriz Ferreiro, William Harris and Chris Kintner for providing cDNA libraries, technical advice and scientific discussion. The authors would also like to thank Drs Akiko Kumagai, Bill Dunphy, Paul Skoglund and Ariel Ruiz i Altaba for sharing their technical expertise at various stages of this project. We especially thank Dr Scott Fraser for support (grant HD 29304) and the use of his laboratory facilities. We also thank Drs David Wilkinson and Chris Kintner for providing the Xenopus Krox-20 and N-CAM probes respectively, David Turner for sharing his results prior to publication, and Adam Strassberg for help with computer graphics. The authors are also grateful to Drs Ariel Ruiz i Altaba and Tom Jessell for many helpful discussions, and to Bill Harris, Tom Jessell, Barbara Wold and Scott Fraser for their critical comments on the manuscript. K. Z. was supported by an NIH post-doctoral fellowship and is currently an Associate of the Howard Hughes Medical Institute. D. J. A. is an Associate Investigator of the Howard Hughes Medical Institute.