XIHbox 6 is an early spatially restricted marker for molecular studies of neural induction. The sequence of the full-length XIHbox 6 protein is reported. An antibody raised against a β-galactosidase/XIHbox 6 fusion protein was used to analyze the expression of XIHbox 6 proteins during frog embryogenesis. The anterior border of XIHbox 6 expression lies just posterior of the hindbrain/spinal cord junction. Immunostalnlng extends the entire length of the spinal cord. A much weaker transient expression with a similar anterior border is observed in mesoderm. Almost all nuclei in the newly closed spinal cord contain XIHbox 6. The number of positive nuclei decreases over the next stages of development, until in later embryos XIHbox 6 is restricted to nuclei of the dividing neuroepithelium, and not the mantle or marginal zones of the spinal cord. When the limb buds begin to grow, there is a second burst of XIHbox 6 expression in proliferating neurons of the cervical and lumbar enlargements, where nerves arise that supply the limbs. The data suggest that XIHbox 6 expression is spatially and temporally restricted to immature neurons of the spinal cord, before their differentiation into mature neurons.

The early amphibian embryo is the classical embryological material for studying the generation of the three major germ layers of the vertebrate embryo ectoderm, mesoderm and endoderm. The mesoderm arises as a result of an interaction between endodermal cells and cells of the animal pole. The remaining animal pole cells either become epidermis or are further instructed by the mesodermal cells, as they invaginate underneath the ectoderm during gastrulation, to adopt a neural fate. A neural tube and neural crest cells are formed from these induced cells.

In this process, the mesoderm plays a critical role in establishing the positional character of the neuroectoderm and endoderm (reviewed in Spemann, 1938; Huxley and DeBeer, 1934), i.e. whether it develops as brain or spinal cord. The nature of the information controlling the nature of the mesoderm and the subsequent character of the neuroectoderm is at present unknown, although growth factors have been implicated (Smith, 1989; Dawid et al. 1989). Genes that respond to such signals and which are expressed in particular restricted regions of the embryo, rather than in a tissue-type or cell-type-specific manner, are therefore useful molecular markers for the study of initial differences along the anteroposterior axis of the embryo. They may also allow us to trace back and unravel the regulatory cascade determining the initial polarization of the axis after fertilization.

Homeobox-containing genes are good candidates for genes playing important roles in such processes. They bear protein motifs called homeodomains which are conserved in genes of Drosophila that are genetically proven to determine positional character in the developing embryo (Gehring, 1987; Akam, 1987). In both fruitflies and vertebrates, they probably respond to positional information (either directly or indirectly) and then interpret it to subdivide the embryo along the anteroposterior axis into defined bands, or fields, of morphogenetic potential. For example, the gene XlHbox 1 is expressed in nuclei of cells forming an aligned cervical band through the neural crest, CNS and mesoderm of the early Xenopus embryo (Wright et al. 1989a; De Robertis et al. 1989). Evidence is accumulating that the vertebrate homeobox genes may play roles in vertebrate embryogenesis similar to their Drosophila counterparts. Reductions in anterior-posterior axis development (Ruiz i Altaba and Melton, 1989a; 1989b) and specific changes in th e body plan which may be considered homeotic transformations (Wright et al. 1989c; Balling et al. 1989; M. Kessel and P. Gruss, personal communication) have been found as a consequence of altering the expression pattern of vertebrate homeobox genes.

The majority of vertebrate homeobox genes so far isolated have homeodomains that are most related to the Drosophila Antennapedia sequence, although other classes such as engrailed and paired have now been found (Scott et al. 1989). As is the case in Drosophila, it has been suggested that the linear order of the vertebrate homeobox genes along a cluster also reflects the relative domain of expression along the anteroposterior axis. Genes that lie more 5’ in the cluster are expressed more posteriorly in the embryo than genes lying at the 3’ end. Put another way, the mouse genes that are more similar to the posterior Drosophila gene Abd-B are expressed more posteriorly than genes that are similar to the anterior Drosophila gene Dfd (Gaunt et al. 1988; Boncinelli et al. 1988; Giampaolo et al. 1989; Graham et al. 1989; Akam, 1989; Duboule and Dolle, 1989). Intriguingly, the domains of expression of at least some vertebrate homeobox genes may be correlated with the expression boundaries of other genes encoding zinc finger proteins related to Drosophila gap gene products (Murphy et al. 1989; Wilkinson et al. 1989).

XlHbox 6 is a useful marker for neural induction in Xenopus (Sharpe et al. 1987; Sharpe et al. 1989). XlHbox 6 expression requires neural induction in ectoderm/mesoderm conjugates in vitro in order to detect a signal by Sl assay. It is restricted to the posterior third of the neural plate at neurula stages (Sharpe et al. 1987). This represents the tissue devoted to spinal cord development, whereas the anterior two-thirds produces the various regions of the brain. Moreover, Sharpe et al. used an XlHbox 6 probe to show that the ectoderm of the early gastrula displays a polarized predisposition for neural development in the dorsal region. This has recently been confirmed using other markers, leading to the conclusion that signals emanating from the dorsal lip region bias the ectoderm toward a non-epidermal pathway of differentiation before significant mesodermal invagination takes place (Savage and Phillips, 1989).

In the present study, we report the entire protein sequence of the XlHbox 6 homeodomain protein. The homeodomain of the frog gene XlHbox 6 is more similar to Drosophila Abd-B than to Antp. Sequence comparisons and the structures of its neighbouring genes suggest that XlHbox 6 may represent the frog homologue of the murine Hox-2.5 gene. Immunolocalization studies show that XlHbox 6 is expressed along the entire length of the spinal cord even to the most posterior regions, but not in the brain. We suggest that XIHbox 6, which is a useful marker for neural induction in Xenopus (Sharpe et al. 1987), is expressed at the beginning of the lifetime of the great majority, if not all, of the spinal cord neurons.

Sequencing and mapping

We present here the deduced amino acid sequence of the fulllength XlHbox 6 protein. There are some discrepancies between this and the previously reported partial XlHbox 6 sequence (Sharpe et al. 1987). In Fig. 1 of Sharpe et al. (1987) an EcoRI fragment, probably derived from an unrelated cDNA, was mistakenly positioned 5’ to the major XlHbox 6 EcoRI fragment. Thus, the nucleotide sequence upstream of the 5’-most EcoRI site (actually an EcoRI linker) is incorrect. The sequence presented here is entirely derived from a single genomic clone to make allowance for the presence in the Xenopus laevis genome of two genomically duplicated copies of the XlHbox 6 gene and also for allelic variation. The present sequence supercedes that previously shown in Fig. 1 of Sharpe et al. (1987) which was in error. E.D.R. wishes to apologize for any inconvenience this may have caused to other investigators. The EcoRI linker and 5’ Smal site in the previously reported XlHbox 6 cDNA would be located at positions corresponding to one nucleotide upstream of Ser29 and at Arg36, respectively. Another Smal site used to construct fusion protein 2 (Fig. 1) is at Pro106. A Pstl site is at Cys49 and EcoRI cuts in the homeobox at Glu185.

To clone the upstream portions of the XlHbox 6 gene a 70 bp sense strand RNA probe from the 5’ end of the partial cDNA clone (5’ EcoRI linker to the downstream Pstl site in Fig. 1 of Sharpe et al. 1987) was prepared and used to screen a library of Xenopus genomic DNA (partial Sau3AI-cut) in λ EMBL4 plated in DL231. Subcloning and genomic mapping using hybridization and restriction digestion was performed according to standard procedures.

All subcloned pieces of DNA were sequenced with a commercial kit (USB or Pharmacia) employing T7 DNA polymerase and the chain termination protocol of Sanger et al. (1977). Sequencing templates were either double-or single-stranded, and the sequence on both strands was determined.

Antibody production

AnEcoRI fragment from an XlHbox 6 cDNA was filled-in and ligated in frame to BamHI-cut, filled-in, pTRBO (Burglin and De Robertis, 1987) to generate Fusion Protein 1 (FPl; Fig. 7). Fusion proteins were grown and purified from E. coli strain F’11recA and used to immunize rabbits. Antiserum was depleted of anti-β-galactosidase and anti-E. coli antibodies on a matrix of β-galactosidase/E. coli proteins, followed by affinity purification on a β-galactosidase/XlHbox 6 fusion protein column as described previously (Wright et al. 1989b). A Fusion Protein 1-Sepharose matrix was normally used to affinity purify the depleted antiserum, but sometimes a smaller fusion protein containing less of the XlHbox 6 protein-coding sequence (and no homeodomain sequence) was used. In this case, the Smal fragment shown in Fig. 1 of Sharpe et al. (1987) was ligated in frame to pTRB0 to generate Fusion Protein 2 (FP2; Fig. 1). This was then purified and linked to Sepharose as above. Antibodies purified on either of these matrices gave the same immunostaining patterns.

Antibody staining

These were carried out as described previously (Wright et al. 1989b). For whole-mounts, albino embryos were used and the protocol of Dent et al. (1989) was followed.

The antibodies react only with XlHbox 6/β-galactosidase fusion proteins on Western blots of bacterial lysates. They do not react with β-galactosidase alone or a number of other homeodomain protein/ β-galactosidase fusion proteins (not shown). The immunostaining is competed away if the X1Hbox 6 fusion protein is included during the first antibody binding reaction. Moreover, the distribution of XIHbox 6 antigen determined here is congruent with the distribution of XIHbox 6 RNA, which was extensively studied using Sl nuclease assays (Sharpe et al. 1987).

We wish to further address the question of the specificity of these XIHbox 6 antibodies. The vertebrate homeobox genes are arranged in four major clusters that presumably arose by chromosomal duplication (e.g. Graham et al. 1989; Duboule and Dolle, 1989). The Hox-2.5 gene of the HOX-2 complex has three other very similar versions in each of the other three clusters (Fig. 2B). Presuming the cluster organization to be conserved from frog to mouse, it might seem possible that these XlHbox 6 antibodies could crossreact with protein products from other XIHbox 6-like genes. This may not be the case, however, because although Hox-2.5 and XIHbox 6 probably represent homologous genes, the protein sequence against which the antibody was raised is extremely diverged between mouse and frog (Robb Krumlauf, personal communication). Indeed, the present XIHbox 6 antibodies do not crossreact in mouse tissue (our own unpublished observations). This is unlike the antibodies raised against other frog homeodomain proteins which crossreact in a wide variety of vertebrate species (Oliver et al. 1988).

The background staining present in embryos was monitored (especially in whole-mounts) by the use of an out-of-frame antibody raised against nonsense protein sequence (see Wright et al. 1989c). For example, the aligned myotome nuclei that may look weakly stained in some of the whole mounts presented here (e.g. Fig. 3, Fig. 7) are equivalently stained when the out-of-frame antibody is used as primary instead of anti-XIHbox 6 antibodies. In sectioned material, some weak staining for XIHbox 6 is seen in lateral plate mesoderm nuclei (see below), but myotome nuclei are not stained.

We observe heavy staining in the cement gland (indicated as cg in Fig. 3) but conclude that this does not represent XIHbox 6 protein for the following reasons. The staining is cytoplasmic and extracellular, and S1 nuclease protection assays have shown that XlHbox 6 transcripts are absent from all anterior parts of the embryo (Sharpe etal. 1987). This background could be due to non-specific binding or crossreaction to the large amounts of mucopolysaccharide secreted by this organ. The lack of cement gland staining in sectioned embryos fixed in Bouin’s fixative (not shown) supports our interpretation that this staining is artefactual. All other staining patterns are unchanged in Bouin’s-fixed embryos.

XlHbox 6 is similar to several vertebrate homeobox genes

The sequence of part of the XIHbox 6 gene has been reported previously (see Materials and methods). The length of the major XIHbox 6 transcript has been estimated from Northern blots to be approx. 1.8 kb (Sharpe et al. 1987). All of our cDNA clones were truncated at the 5’ end. We have now cloned the 5’ end of the protein coding sequence from a genomic clone and overlapped it with cDNA sequences to allow the deduction of the entire XIHbox 6 amino acid sequence, as shown in Fig. 2A. Because of sequence comparison with the mouse Hox-2.5 gene product (see below), we believe that we have correctly identified the initiator methionine of the XIHbox 6 protein. By standard restriction mapping, hybridization analysis and cDNA sequencing, we mapped the position of an intron that interrupts the protein coding sequence just in front of the homeobox. We do not presently know where transcription is initiated in the genomic DNA, or whether differential splicing occurs within this gene that might lead to the production of alternative protein products with or without the XIHbox 6 homeodomain. The deduced protein sequence is 232 amino acids long, with only 6 amino acids following the homeodomain. There are two genomically duplicated versions of XIHbox 6 in the tetraploid Xenopus laevis genome (Fritz et al. 1989). The version presented here contains a string of six (the other copy has five, as reported in Sharpe et al. 1987) glutamine residues about one-third into the protein. Note that the first 24 amino acids of the partial sequence reported previously by our group (Sharpe et al. 1987) were incorrect, and are hereby retracted (see Materials and methods).

As shown in Fig. 2B, the homeodomain bears significant similarities to several vertebrate homeodomain sequencesnamely mouse Hox-2.5, Hox-1.7, Hox-3.2, Hox-5.2, and a sea urchin homeobox gene called HB4. The four mouse genes have been proposed to be duplicated versions of each other occurring in analogous positions within their respective homeobox gene clusters (Graham et al. 1989; Duboule and Dolle, 1989). Interestingly, the homeodomains of this group of genes are more similar to that of Drosophila homeobox gene Abdominal-B than Antennapedia. This may be especially relevant when considering the spatial domain of expression of the XlHbox 6 gene, which is addressed in Results and Discussion.

The homeodomains of Hox-2.5 and XIHbox 6 are 100 % similar when allowing for conservative changes (Fig. 2). When more Hox-2.5 protein sequence (derived from cDNA) is compared to XlHbox 6, the similarity extends to many regions outside of the homeodomain (R. Krumlauf, personal communication; data not shown). XlHbox 6 and Hox-2.5 probably represent homologous genes from frog and mouse. This view is supported by the fact that the two genes located 3’ to XlHbox 6 also strongly resemble adjacent counterparts in the mouse HOX-2 complex (Fritz et al. 1989). Good similarity exists over the proposed amino termini of XlHbox 6 and Hox-2.5. Previous to the initiator methionine there is no significant similarity in any reading frame. Therefore, although we have derived the 5’ part of the nucleotide sequence from genomic DNA, we suggest that this does indeed represent the sequence of a full-length XlHbox 6 homeodomain protein product.

Anti-XlHbox 6 antibody preparation

Previous work has defined the XIHbox 6 gene as an early marker of neural differentiation (Sharpe et al. 1987). The detail of analysis was limited, however, because Sharpe et al. assayed RNA extracted from dissected embryo pieces. We have now made a polyclonal anti-XlHbox 6 antibody in rabbit using a galactosidase/XIHbox 6 fusion protein as the immunogen to allow the analysis of the distribution of XIHbox 6 antigens during embryogenesis. The staining patterns presented in this study are specific for the XIHbox 6 protein by a variety of criteria (see Methods).

The majority of neurons born in the posterior CNS have an early transient phase of XIHbox 6 expression

Fig. 3 shows the distribution of XIHbox 6 antigens at various stages of Xenopus development immunostained in whole-mount. Sl assays by Sharpe et al. (1987) showed that XlHbox 6 transcripts could first be detected at early neurula (stage 13; all stages here are according to Nieuwkoop and Faber, 1%7). We could not reproducibly detect XIHbox 6 antigens in early neurulae, either in sectioned or whole-mouint material. In late neurulae (stage 19-20), nuclei in the future spinal cord, but not brain, are faintly imrnunostained with our antibody. These are not shown because they did not reproduce photographically although the pattern could be seen by careful observatioin under the microscope while racking the focus up and down. The anterior border in late neurulae is the same as in later stages, with immunostaining extending posteriorly to the tip of the presumptive spinal cord tissue. We could not see any distinguish any mesodermal staining at this embryonic stage (see below).

Thus there appears to be a lag period,of about 6 h (stage 13 until stage 19) between the time of XIHbox 6 mRNA appearance and the accumulation of detectable amounts of the protein product, at least with this antibody. A similar lag has been reported between XIHbox 1 mRNA and protein (Oliver et al.. 1988). This may be a function of th e sensitivity of these polyclonal antibodies compared to the RNA detection methods, but could also represent an important phenomenon of delayed translation of homeobox gene mRNAs.

There is a sharp increase in the intensity of staining over the next few stages. Fig. 3A and 3A’ shows two views of an embryo at approximately stage 22-23 of development, or elongating neurula. XIHbox 6 antigens are detected in most nuclei in the newly formed spinal cord. Since glia make their appearance much later in development, we conclude that XIHbox 6 expression occurs in most, if not all, of the proliferative neuronal precursors. XIHbox 6 is expressed over a very broad anteroposterior region of the embryo, stretching back to nuclei at the posterior-most tip of the spinal cord. The intensity of staining per nucleus is uniform over the whole of this region. A fairly sharp anterior border is observed, which lies just posterior of the hindbrain/spinal cord junction. We defined this border in two ways. First, in relation to the abrupt morphological change where the wide ventricular cavity of the hindbrain turns into the narrow ependymal canal of the spinal cord. Second, by comparison with the staining patterns obtained with anti-XlHbox 1 antibodies, which do not immunostain the hindbrain but decorate a narrow band of the anterior spinal cord (antibody C, directed against the long version of the: XlHbox 1 homeo protein; Oliver et al. 1988).

Tadpoles were analyzed at many stages of development up to stages with advanced limb buds. The anterior border of XIHbox 6 expression is respected throughout these stages. That is, immunopositive nuclei are never found in the brain or anterior mesodermal derivatives but are found down the whole liength of the spinal cord. We have not yet found XlHbox 6 staining in nuclei of the spinal cord or brain dissected from adult Xenopus.

We have also reconstructed the distribution of XIHbox 6 from serial transverse sections of entire tailbud tadpoles. Immunostaining is always restricted to posterior parts of the embryo and is clearly absent from the brain (Fig. 4A). By far the major site of XIHbox 6 expression is in the spinal cord. In mesoderm/ectodenn conjugates it is clear that neural induction is required to detect XIHbox 6 expression by Sl nuclease protection (Sharpe et al. 1987). For this reason, it was concluded that XIHbox 6 expression was probably restricted to the posterior tissue of the neural plate which produces the spinal cord. As reported below, however, this is not absolutely true, because weaker XlHbox 6 immunostaining is observed in the lateral plate mesodenn.

Fig. 5 depicts a stage 24 embryo immunostained for XIHbox 6. Comparison of Fig. SA with SB shows that nuclear fluorescence is quenched in XIHbox 6-positive cells (indicated by arrowheads) when the immunostained sections are counterstained with Hoechst 33258. Hoechst fluorescence is not quenched in non-immunostained nuclei. From this kind of analysis, we conclude that almost every spinal cord nucleus at this stage of development (stage 24) contains the XIHbox 6 homeodomain protein. At this stage, the population of cells expressing XIHbox 6 homeodomain proteins therefore includes proliferating and immature neurons.

Fig. 6 shows a high-power magnification of the dorsal region of a slightly later tailbud tadpole (approx. stage 28) immunostained for XIHbox 6 and compared to the same section stained with Hoechst 33258 dye. Note the absence of immunostaining in the myotome nuclei. From this section, it is apparent that not all spinal cord nuclei contain XIHbox 6 protein. Most, if not all, of the cells laying next to the central spinal canal are immunopositive. In addition, several nuclei that are displaced laterally away from the centre of the spinal cord also contain XlHbox 6. The more peripheral nuclei, which correspond to differentiated primary neurons that were born earlier, no longer express XIHbox 6 homeodomain protein (compare the number of Hoechst-stained fluorescent nuclei with the number that are immunostained). The most ventral cells of the spinal cord (the floor plate) also become devoid of XIHbox 6 antigens at this stage.

At later stages of development when most neuronal differentiation occurs (stage 35 onwards), the spinal cord becomes well differentiated into concentric shells referred to as ependymal layer (the inner mitotic zone next to the narrow ependymal canal of the spinal cord), mantle layer (containing postmitotic neuroblast, immature neurons, and neurons at various stages of differentiation) and marginal layer (presumptive white matter).

The number of XIHbox 6-positive nuclei decreases greatly, in agreement with the decline in XlHbox 6 RNA reported by Sharpe et al. (1987). Immunostaining becomes progressively excluded from the outer two layers and restricted to the ependymal layer (not shown), although the floor plate cells are not immunostained. The expression of XIHbox 6 in the proliferative ependymal layer is somewhat intriguing. The expression of most other homeobox gene products in the CNS seems to be in localized regions of the spinal cord mantle layer that contain differentiating, or differentiated, neurons (e.g. Oliver et al. 1988).

At stages 38-45, the only XlHbox 6-positive nuclei occur in a very small ventral area of the ependymal layer. In transverse sections of the same stages, these cells abut the floor plate and appear as one or two immunopositive cells per 10-20 micron section (not shown). Fig. 4B shows that in sagittal sections or whole-mounts they are visualized as a periodic row of dark dots. We do not yet understand the significance of XIHbox 6 being found in this small population of cells, although it is conceivable that they represent stem cells with proliferative capability.

XlHbox 6 is weakly expressed in lateral plate mesoderm

Some lightly immunostained nuclei are visible in the LPM (lateral plate mesoderm) of early embryos. This is consistent with the results reported by Sharpe et al. (1987) in that in some experiments a minor amount of XIHbox 6 mRNA could be detected in ventral mesectoderm. Fig. SC shows an immunostained tailbud embryo (approx. stage 26) that was overdeveloped during the DAB reaction. This serves to show more clearly that nuclei in the lateral plate mesoderm (LPM) contain XIHbox 6 antigens, but at a much lower level than in the CNS (see arrowheads in Fig. 6A). Serial reconstructions show that the anterior borders in LPM and CNS are aligned at the boundary of somite 3 and 4, and that LPM expression also extends to the very tip of the embryo’s tail (also see Fig. 3C). The LPM staining seems to be fairly transient, as we have difficulty observing immunopositive nuclei in embryos later than stage 35. Most, if not all, of the LPM nuclei are stained, and so these cells are probably not neural crest cells that are migrating through the LPM.

This is slightly different from the situation in the murine Hox-2.5 gene. By in situ analysis, Hox-2.5 is also expressed over a broad posterior region of the embryo with an anterior border in the central nervous system that is almost identical to that of the frog XIHbox 6 gene. Hox-2.5 expression in embryonic mouse mesoderm is also transient, but the expression by RNA detection in situ is higher and the anterior border in this germ layer appears displaced posteriorly by two somites (Bogarad et al. 1989).

XlHbox 6 is re-expressed in spinal cord regions innervating the limb buds

We also immunostained older tadpoles seeking to analyze XIHbox 6 expression in embryos as they approach and begin the process of metamorphosis. The data are summarized in Fig. 7, which shows CNS and attached somitic regions dissected from a whole-mount stained tadpole at about stage 49 of development. At this stage, the forelimb and hindlimb buds have made their appearance. The spinal cord of these embryos now begins a second phase of XIHbox 6 expressiion. A large number of intensely immunopositive nuclei form two condensations that correspond to the c,ervical and lumbar enlargements of the spinal cord. The: nerves that supply the growing limbs arise from these two regions. In the region of these condensations, both ependymal and mantle layer nuclei appear equally heavily stained over the dorsal-ventral axis of the spinal cord. Compared to immediately previous stages, additional XIHbox 6-positive cells also arise in the intermediate region of the CNS between the cervical and lumbar enlargements. Transverse sections through these intermediate regions show that the whole ependymal layer is stained, and that in the mantle layer immunopositive nuclei appear displaced toward the dorsal side. By comparison with the background immunostaining when the out-of-frame control antibody is used, it is possible that the spinal ganglia (not shown) and sympathetic ganglia (Fig. 7B) represent additional sites of XIHbox 6 homeodomain protein expression. This new pattern of XlHbox 6 staining is maintained through stages 52-55.

As shown in Fig. 2B, the homeodomain of XIHbox 6 is also similar to another mouse gene, Hox-5.2. In addition to its expression in spinal cord, Hox-5.2 is expressed in the distal parts of the very early forelimb and hindlimb bud mesoderm (Dolle and Duboule, 1989; Oliver et al. 1989). Using RNAse protection assays on RNA extracted from 80 forelimb or hindlimb buds, we could not detect XIHbox 6 mRNA in limb buds although transcripts could be detected easily in controls using RNA extracted from whole tadpoles (our unpublished observation).

We conclude that XIHbox 6 has an early and a late phase of expression in the CNS, and that this expression is probably restricted to postmitotic neuroblasts and immature or proliferative neurons of the spinal cord.

In this paper, we have analyzed the expression of XIHbox 6 protein during Xenopus development using a polyclonal antibody, thereby extending the analysis made previously at the RNA level. XIHbox 6 expression is restricted to the spinal cord but is expressed down its entire length, with an anterior boundary just posterior of the hindbrain/spinal cord junction. This anterior boundary is maintained throughout embryonic development (Fig. 3 and Fig. 4). By in situ hybridization, the mouse gene Hox-2.5, which may be the homolog of XIHbox 6, has been shown to have a similar spatial expression pattern (Bogarad et al. 1989).

The pattern of XIHbox 6 homeo protein expression, with a sharp anterior boundary and expression back to the most posterior parts of the embryo, is different from other homeodomain protein distributions such as XIHbox 1 (Wright et al. 1989a), XIHbox 8 (Wright et al. 1989b), or the frog engrailed-related antigen (Hemmati Brivanlou and Harland, 1989), which form relatively narrow bands with defined anterior and posterior boundaries. XIHbox 6 has a homeodomain sequence that places it in the Abd-B class (Scott et al. 1989). In Drosophila, Abd-B proteins are expressed in a broad band covering posterior segments 4 to 9 of the embryo (Celniker et al. 1989). The distribution pattern of XIHbox 6 could be consistent with the model elaborated from the mouse homeobox system in which the order of genes along the cluster is colinear with their relative anterior boundaries of expression. Genes that encode homeodomains more similar to Abd-B lie at the 5’ end of the cluster and are expressed more posteriorly than genes at the 3’ end (Gaunt et al. 1988; Boncinelli et al. 1988; Giampaolo et al. 1989; Graham et al. 1989; Akam, 1989; Duboule and Dolle, 1989).

Temporally, XIHbox 6 expression is biphasic. One major peak of immunostaining occurs just after neural tube closure (stages 20-28; Fig. 3), followed by a decline in intensity and number of stained nuclei. At stage 35 only a very few immunopositive cells are observed, lying next to the floor plate, but extending along the whole spinal cord (Fig. 4B). At around stage 48-49, when the mesoderm begins to contribute to the formation of the limb buds, a second burst of XIHbox 6 expression becomes apparent as two condensations of heavily immunostained nuclei at the cervical and lumbar enlargements (Fig. 7). We have not yet seen any reproducible nuclear signal in the adult spinal cord or brain. The engrailed-related antigen of frogs is expressed as a narrow band in the future midbrain region of the Xenopus neural plate (Hemmati Brivanlou and Harland, 1989). With the present antibodies, engrailed is therefore expressed significantly earlier than XIHbox 6 at the protein level.

In preliminary experiments designed to address the possible role of XIHbox 6 in neural differentiation, we microinjected Xenopus embryos with the affinity-purified antibodies described here (Wright et al. 1989c). Despite an extensive series of experiments, we failed to obtain any phenotypic defects resulting from this treatment. This may be because only antibodies against other parts of the XIHbox 6 protein (for example the amino terminal portions) will be effective in interfering with XIHbox 6 function.

It is interesting to compare XIHbox 6 expression in relation to the periods of mitotic activity in the spinal cord. The period after the first mitosis of the neural plate when primary neurons are born (stages 17-18) is characterized by a period of quiescence within the neural tissue. Mitotic activity is resumed at stage 20 and dividing nuclei, which are mostly confined to the ependymal layer, correspond to secondary precursors which between stages 20 and 35 undergo one or two additional divisions (Hartenstein, 1989). This period corresponds to the first phase of XIHbox 6 expression. The growth of the limbs and their innervation is a major developmental change at metamorphosis. The later peak of XIHbox 6 expression, especially in the cervical and lumbar enlargements that innervate the limbs, most probably occurs as neurons proliferate to accommodate this process.

There is a paucity of markers for the early neuronal precursor cell. One useful reagent that has been reported, an antibody called Rat-401, specifically labels the proliferating neuronal precursor population in rat CNS (Frederiksen and McKay, 1988). Rat-401 expression is not confined to spinal cord, but it is interesting to note that the regions of expression of Rat-401 and XIHbox 6 in the spinal cord are strikingly similar. In future, it would be worthwhile to determine the possible relationship between XlHbox 6 expression and neuronal proliferation in the spinal cord. Similarly, one might ask whether XIHbox 6 expression is increased in the cells of the cervical and lumbar enlargements after amputation of their target tissue, the limb buds. One would also predict that XIHbox 6 should be re-expressed in regenerating tadpole tail tissue because Xenopus tadpoles can effectively regenerate the entire tail after its removal.

Most experiments using the XIHbox 6 gene as a probe in neural induction experiments have analyzed RNA extracted from dissected embryo pieces, in which complex cell-cell interactions may confuse the analysis. This antibody might be useful in experiments on neural induction involving single, or a few, cell(s), and will certainly facilitate the analysis of individually manipulated embryos by whole-mount. Expression constructs encoding the entire XlHbox 6 homeo protein will allow experiments testing the effect of overexpressing this gene in microinjected frog embryos. In addition, the detailed study of the XIHbox 6 promoter may allow the dissection of the molecular mechanisms involved in the activation of this gene following neural induction.

We thank Jane Hardwicke for expert photographic work. We acknowledge the laboratory of Robb Krumlauf for communicating the sequence of their Hox-2.5 clone with us prior to publication. We also thank Mike Farrell for help in the initial stages of this work. We thank Volker Hartenstein and Ron McKay for discussions before publication, and Dennis Bittner, Bruce Blumberg and Ken Cho for comments on the manuscript. This work was supported by NIH grant HD21502-05. C.V.E.W. was an American Cancer Society (California Division) Senior Fellow. E.A.M. was the recipient of USPHS National Research Service Award GM-07104.

Akam
,
M.
(
1987
).
The molecular basis for metameric pattern in the Drosophila embryo
.
Development
101
,
1
22
.
Akam
,
M.
(
1989
).
Hox and HOM: homologous gene clusters in insects and vertebrates
.
Cell
51
,
347
349
.
Balling
,
R.
,
Mu’iter
,
G.
,
Gruss
,
P.
and
Kessel
,
M.
(
1989
).
Craniofacial abnormalities induced by ectopic expression of the homeobox gene Hox-1.1 in transgenic mice
.
Cell
58
,
337
347
.
Bogarad
,
L. D.
,
Utsit
,
M. F.
,
Awgulewitsch
,
A.
,
Mikj
,
T.
,
Hart
,
C. P.
Andruoole
,
F. H.
(
1989
).
The Developmental expression pattern of a new murine homeo box gene: Hox-2.5
.
Devi Biol
.
133
,
537
549
.
Boncinelli
,
E.
,
Somma
,
R.
,
Acampora
,
D.
,
Pannese
,
M.
,
D’esposito
,
M.
,
Faiella
,
A.
and
Simeone
,
A.
(
1988
).
Organization of human homeobox genes
.
Hum. Reprod
.
3
,
880
886
.
Breier
,
G.
,
Dressler
,
G.R.
Andgruss
,
P.
(
1988
).
Primary structure and developmental expression pattern of Hox-3.1, a member of the murine Hox-3 homeobox gene cluster
.
EMBO J
.
7
,
1329
1336
.
Burglin
,
T.R.
and
Derobertis
,
E. M.
(
1987
).
The nuclear migration signal of Xenopus laevis nucleoplasmin
.
EMBO J
.
6
,
2617
2625
.
Celniker
,
S. E.
,
Keelan
,
D. J.
and
Lewis
,
E.B.
(
1989
).
The molecular genetics of the bithorax complex of Drosophila: characterization of the products of the Abdominal-B domain
.
Genes Dttv
.
3
,
1424
1436
.
Da\Vid
,
I.B.
,
Sargent
,
T. D.
Androsa
,
F.
(
1989
).
The role of growth factors in embryomc induction in amphibrnns
.
Current Topics in Developmental Biology (in press)
.
Dent
,
J. A.
,
Poulson
,
A.G.
and
Klymkowsky
,
M. W.
(
1989
).
A whole-mount immunocytochemical analysis of the expression of the intermediate filament protein vimentin in Xenopus
.
Dttvelopmttnt
105
,
61
74
.
Dolecki
,
G. J.
,
Wang
,
G.
and
Humphreys
,
T.
(
1988
).
Stage- and tissue-specific expression of two homeo box genes in sea urchin embryos and adults
.
Nuc. Acids. Res
.
16
,
11 543
11 557
.
Dolle
,
P.
and
Dueoule
,
D.
(
1989
).
Two gene members of the murine HOX-5 complex show regional and cell-type specific expression in developing limbs and gonads
.
EMBO J
.
8
,
1507
1515
.
Duboule
,
D.
and
Dolle
,
P.
(
1989
).
The structural and functional organization of the murine HOX gene family resembles that of Drosophila homeotic genes
.
EMBO J
.
8
,
1497
1505
.
Fredericksen
,
K.
and
Mckay
,
R. D.G.
(
1988
).
Proliferation and differentiation of rat neuroepithelial precursor cells in vivo
.
J. Neurosci
.
8
,
1144
1151
.
Fritz
,
A. F.
,
Cho
,
K. w. Y.
,
Wright
,
C. V. E.
,
Jegalian
,
B.G.
and
Derobertis
,
E. M.
(
1989
).
Duplicated homeobox genes in Xenopus
.
Devl Biol
.
131
,
584
588
.
Gaunt
,
S. J.
,
Sharpe
,
P. T.
and
Dueoule
,
D.
(
1988
).
Spatially restricted domains of homeogene transcription in mouse embryos: relation to a segmented body plan
.
Development 104 {Supplement), 169-179
.
Gehring
,
W. J.
(
1987
).
Homeoboxes in the study of development
.
Scittnce
236
,
1245
1252
.
Giampaolo
,
A.
,
Acampora
,
D.
,
Zappavigna
,
V.
,
Pannese
,
M.
,
D’esposito
,
M.
,
Care
,
A.
,
Faiella
,
A.
,
Stornaiuolo
,
A.
,
Russo
,
G.
,
Slmeone
,
A.
,
Boncinelli
,
E.
and
Peschle
,
C.
(
1989
).
Differential expression of human HOX-2 genes along the anterior-posterior axis in embryonic central nervous system
.
Differentiation
40
,
191
197
.
Graham
,
A.
,
Papoldpulu
,
N.
and KRUMLAUF,R
. (
1989
).
The murine and Drosophila homeobox gene complexes have common features of organization and expression
.
Cell
57
,
367
378
.
Hartenstein
,
V.
(
1989
).
Early neurogenesis in Xmopus: the spatio-temporal pattern of proliferation and cell lineages in the embryonic spinal cord
.
Neuron
3
,
399
411
.
Hemmatibrivanlou
,
A.
and HARLAND,R. M
. (
1989
).
Expression of an engrailed-related protein is induced in the anterior neural ectoderm of early Xttnopus embryos
.
Dttvelopment
106
,
611
617
.
Huxley
,
J. S.
and DEBEER,G.R
. (
1934
).
Elemttnts of Erpuimelltal Embryology
(
Cambridge
:
Cambridge University Press
).
Murphy
,
P.
,
Davidson
,
D.R.
and HILL,R. E
. (
1989
).
Segment-specific expression of a homeobox-containing gene in the mouse hindbrain
.
Narure
341
,
156
159
.
N1Euwkoop
,
P. D.
and
Faber
,
J.
(
1967
).
Nonnal Table of Xenopus laevis (Daudi11) (Amsterdam: North Holland)
.
Oliver
,
G.
,
Sidell
,
N.
,
Fiske
,
W.
,
Heinzmann
,
C.
,
Mohandas
,
T.
,
Sparkes
,
R. S.
and
Derobertis
,
E. M.
(
1989
).
Complementary homeo protein gradients in developing limb buds
.
Genes Dev
.
3
,
641
650
.
Oliver
,
G.
,
Wright
,
C. V. E.
,
Hardwicke
,
J.
and
Derobertis
,
E. M.
(
1988
).
Differential antero-posterior expression of two proteins encoded by a homeobox gene in Xenopus and mouse embryos
.
EMBO J
.
7
,
3199
3209
.
Rubin
,
M.R.
,
King
,
W.
,
Toth
,
L. E.
,
Sawczuk
,
I. s.
,
Levine
,
M. S.
,
D’eustachio
,
P.
and
Nguyen-Huu
,
C.
(
1987
).
Murine Hox-1.7 homeo-box gene: cloning, chromosomal location, and expression
.
Malec. cell. Biol
.
7
,
3836
3841
.
Ruiz I Altaba
,
A.
and
Melton
,
D. A.
(
1989a
).
Involvement of the Xenopus homeobox gene Xhox3 m pattern formation along the anterior-posterior axis
.
Ctill
57
,
317
326
.
Ruiz I Altaba
,
A.
and
Melton
,
D. A.
(
1989b
).
Interaction between peptide growth factors and homeobox genes in the establishment of antero-posterior polarity in frog embryos
.
Nature
341
,
33
38
.
Sanger
,
F.
,
Nlcklen
,
S.
and
Coulson
,
A.R.
(
1977
).
DNA sequencing with chain-terminating inhibitors
.
Proc. nal11. Acad. Sci. U.S.A
.
74
,
5463
5467
.
Savage
,
R.
and
Phillips
,
C.R.
(
1989
).
Signals from the dorsal blastopore lip region during gastrulation bias the ectoderm toward a non-epidermal pathway of differentiation in Xenopus lattvis
.
Devi Biol
.
133
,
157
168
.
Scotr
,
M. P.
,
Tamkun
,
J. W.
and HARTZELL,G. W
. (
1990
).
The structure and function of the homeodomain
.
BBA Rttvitt1vs 011 Cancttr. BBA
989
,
25
48
.
Sharpe
,
C.R.
,
Fritz
,
A.
,
Derobertis
,
E. M.
Andgurdon
,
J.B.
(
1987
).
A homeobox-containing marker of posterior neural differentiation shows the importance of predetermination in neural induction
.
Cell
50
,
749
758
.
Sharpe
,
C.R.
,
Pluck
,
A.
Andgurdon
,
J.B.
(
1989
).
XIF-3, a Xenopus peripherin gene, requires an inductive signal for enhanced expression in anterior neural tissue
.
Development
107
,
701
714
.
Smith
,
J. C.
(
1989
).
Mesoderm induction and mesoderm-inducing factors in early amphibian development
.
Development
105
,
665
677
.
Spemann
,
H.
(
1938
).
Embryonic Development and Induction
.
Yale University Press
,
New Haven (reprinted 1967
).
Wilkjnson
,
D.G.
,
Bhatt
,
S.
,
Cook
,
M.
,
Boncinelli
,
E.
and
Krumlavf
,
R.
(
1989
).
Segmental expression of Hox-2 homoeobox-containing genes in the developing mouse hindbrain
.
Nature
341
,
405
409
.
Wright
,
C. V. E.
,
Cho
,
K. w. Y.
,
Hardwicke
,
J.
,
Collins
,
R. H.
and DEROBERTIS
(
1989c
).
Interference with function of a homeobox gene in Xenopus embryos produces malformations of the anterior spinal cord
.
Cell
59
,
81
93
.
Wright
,
C. V. E.
,
Cho
,
K. W. Y.
,
Oliver
,
G.
and
Derobertis
,
E. M.
(
1989a
).
Vertebrate homeodomain proteins: families of region-specific transcription factors
.
Trends Biochem. Sci
.
14
,
52
56
.
Wright
,
C. V. E.
,
Schnegelsberg
,
P.
and DEROBERTIS
(
1989b
).
XlHbox 8: a novel Xe11opus homeo protein restricted to a narrow band of endoderm
.
Development
105
,
787
794
.