The Sox-domain containing gene Dichaete/fish-hook acts in concert with vnd and ind to regulate cell fate in the Drosophila neuroectoderm

How cells acquire specific and often unique fates as a function of their position in a developing cellular field is a central question in developmental biology. One of the best model systems in which to explore the link between pattern formation and cell-type specification is the Drosophila embryonic central nervous system (CNS). The Drosophila CNS develops from the reiterative divisions of neural stem cells (reviewed by Goodman and Doe, 1993). Neural stem cells, called neuroblasts (NBs), segregate from the neuroectoderm into the interior of the embryo in five temporally distinct waves (SI-SV) and form an invariant and roughly orthogonal pattern of 30 NBs per hemisegment. Based on its position, each NB acquires a unique fate and divides in a stem cell manner to produce a unique and nearly invariant family of neurons and/or glia. Cell transplantation and genetic experiments support the model that the fate of a NB is predetermined by the fate of the cells within the neural equivalence group from which it segregates (Chu-LaGraff and Doe, 1993; Skeath et al., 1995; Udolph et al., 1995). Thus, to link pattern formation to cell type specification, it is crucial to identify the genes and elucidate the genetic pathways that regulate pattern and cell fate in the neuroectoderm.

Work from many laboratories has begun to define the genetic mechanisms that establish pattern along the anteroposterior (AP) and dorsoventral (DV) axes of the Drosophila embryo. These studies show that segment polarity gene activity divides each segment along the AP axis into four parallel transverse rows (reviewed by Bhat, 1999). Along the DV axis, the activity of the Drosophila EGF receptor (Egfr), ventral nerve cord defective (vnd), intermediate neuroblasts defective (ind) and muscle segment homeobox (msh; Dr – FlyBase) genes initially divides the neuroectoderm into three parallel longitudinal columns (reviewed by Skeath, 1999). Superimposition of the expression patterns of these genes subdivides the neuroectoderm into a checkerboard pattern of cell clusters that corresponds to the pattern of neural equivalence groups from which individual NBs arise. Functional studies indicate that these genes regulate NB fate by controlling the precise combination of genes expressed within each cell cluster (reviewed by Bhat, 1999; Skeath, 1999). These data have led to the model that the coordinated action of the segment polarity genes, as well as of Egfr, vnd, ind and msh specify the fate of the cells in an equivalence group and with it the fate of the NB that segregates from the group.

We focus on the genetic regulatory mechanisms that control DV pattern and cell fate in the neuroectoderm. Previous research has shown that Egfr, vnd, ind and msh subdivide the early neuroectoderm into three longitudinal columns – medial, intermediate and lateral – from which early forming NBs arise. Active Egfr signaling occurs in the medial and intermediate columns prior to the first wave of NB formation and persists in the medial column throughout neurogenesis. Egfr activates ind expression in the intermediate column and promotes intermediate column NB fates. In addition, Egfr acts in the medial column to help specify the individual fates of medial NBs (Skeath, 1998; Udolph et al., 1998; von Ohlen and Doe, 2000; Yagi et al., 1998). vnd expression marks the medial neuroectodermal column throughout neurogenesis. vnd promotes medial column fates, at least in part by repressing ind expression. vnd also promotes SI and SII medial NB formation but appears to play only a limited role in SIII-SV medial NB formation (Chu et al., 1998; Jimenez and Campos-Ortega, 1990; McDonald et al., 1998; Skeath et al., 1994; Jimenez et al., 1995). ind expression marks the intermediate column during the first two, but not subsequent, waves of NB formation. ind promotes the formation and fate of SI and SII intermediate NBs, and represses msh expression in the intermediate column (Weiss et al., 1998). msh expression marks the lateral column during the first two waves of NB formation. However, msh does not affect lateral column gene expression and its function in the lateral column is not well defined (Buescher and Chia, 1997; Isshiki et al., 1997; Skeath, 1999). Despite the well-defined roles Egfr, vnd and ind play in regulating DV pattern and cell fate in the neuroectoderm, existing evidence suggests additional genes regulate these processes. For example, neither Egfr nor ind is active in the intermediate column after SII NB formation (Skeath, 1998; Weiss et al., 1998), and most late-forming medial column NBs develop normally in vnd mutant embryos (Chu et al., 1998; McDonald et al., 1998).

One candidate regulator of DV pattern in the neuroectoderm is the Sox-domain-containing gene Dichaete or fish-hook. (We refer to the gene as Dichaete.) Dichaete belongs to the conserved Sox family of high-mobility group domain DNA-binding proteins (Nambu and Nambu, 1996; Russell et al., 1996). Sox proteins regulate the transcription of target genes through their ability to bind DNA and to partner with a wide variety of different transcription factors (Kamachi et al., 2000). Many vertebrate Sox-domain-containing genes are expressed in the neural plate/tube (Cremazy et al., 2000; Wegner, 1999); however, their function in neural development remains unclear. In Drosophila, Dichaete is initially expressed in seven transverse pair-rule stripes in early embryos, where it regulates segmentation (Nambu and Nambu, 1996). Later, Dichaete is activated in the ventral half of the neuroectoderm during gastrulation. Late-stage Dichaete mutants exhibit severe defects in CNS development, consistent with Dichaete playing a role in DV neural patterning (Nambu and Nambu, 1996). However, the possibility that these defects arise because of a direct role for Dichaete in the DV patterning of the neuroectoderm or indirectly as a consequence of the role of Dichaete in segmentation was not investigated.

We demonstrate that Dichaete plays a crucial role to pattern the neuroectoderm along the DV axis. Our expression studies show that Dichaete is expressed in the medial and intermediate neuroectodermal columns throughout all waves of NB formation. Loss-of-function studies indicate that Dichaete regulates cell fate and neuroblast formation in the medial and intermediate column. Genetic interactions, as well as molecular epistasis tests, demonstrate that Dichaete functions in parallel to vnd and ind, and downstream of Egfr to regulate pattern and cell fate in the neuroectoderm. Work from vertebrates indicates that Sox-domain-containing proteins regulate transcription by partnering with different transcription factors (Kamachi et al., 2000). Together with results presented here, these data support a model whereby Dichaete physically associates with Vnd and Ind to regulate gene expression and NB formation within the medial and intermediate columns of the neuroectoderm.

Wild-type patterns of gene expression were examined in Oregon R embryos. Mutant lines used were: Egfr, allele flb^IK35 (Clifford and Schupbach, 1994); ind^RR108 and ind^16.2 (Weiss et al., 1998); vnd^Δ38 (Chu et al., 1998); Dichaete⁸⁷, Dichaete⁹⁶, provided by John R. Nambu (Mukherjee et al., 2000); and H162, an enhancer trap line inserted into the seven-up gene and referred to as svp-lacZ (Mlodzik et al., 1990). We used standard genetic means to create fly lines or embryos multiply mutant for the following genes: (1) Dichaete⁸⁷ svp-lacZ; (2) Dichaete⁸⁷ ind^RR108; (3) vnd^Δ38; Dichaete⁸⁷; (4), vnd^Δ38; flb^IK35; (5) vnd^Δ38; ind^RR108; and (6) vnd^Δ38; Dichaete⁸⁷ and ind^RR108.

Single- and double-label immunohistochemistry, and RNA in situ analysis were performed as described elsewhere (Skeath, 1998). For the active MAP kinase antibody, we used biotinyl tyramide (NEN Life Science Products) to amplify the signal following the manufacturer’s protocol. We used the following antibodies at the indicated dilutions: mouse anti-Achaete (1:3) (Skeath and Carroll, 1991); rabbit anti-Vnd (1:10) (McDonald et al., 1998); rabbit anti-Dichaete (1:1000) (Mukherjee et al., 2000); rat anti-Ind (1:250) (Weiss et al., 1998); rabbit anti-Msh (1:600) (Isshiki et al., 1997); rabbit anti-Eve (1:2000) (Frasch et al., 1986); mouse anti-Engrailed 4D9 (1:5) (Patel et al., 1989a); mouse anti-βgal (1:2000; Promega); mouse anti-Pros MR1A (1:3) (Spana and Doe, 1995); and mouse anti-Active MAP kinase (1:2000; Sigma) (Gabay et al., 1996).

vnd, ind and Egfr are key factors that regulate pattern and cell fate along the DV axis of the neuroectoderm. To ask if Egfr pathway activity depends on vnd or ind, we assayed MAPK activity in homozygous vnd or ind single mutant embryos. In both backgrounds, the initial activation of Egfr signaling in the medial and intermediate columns is normal. Thus, in the early neuroectoderm, Egfr acts either upstream or in parallel to vnd and ind. To investigate whether Egfr acts upstream of vnd or ind, we assayed vnd and ind expression in embryos homozygous mutant for the Egfr null allele flb^IK35 (referred to as Egfr mutant embryos). ind expression is absent in Egfr mutant embryos, indicating that Egfr activates ind expression in the intermediate column (data not shown) (von Ohlen and Doe, 2000). By contrast, vnd expression in Egfr mutant embryos appears normal through the onset of stage 8 (Fig. 1). However, during stage 8, vnd expression begins to dissipate in medial column cells, and by early stage 10 these cells no longer express vnd (Fig. 1) (Gabay et al., 1996). Conversely, medial column NBs that form in Egfr mutant embryos express vnd normally and retain vnd expression throughout embryogenesis (Fig. 1). Thus, Egfr functions to maintain vnd expression in the neuroectoderm but is dispensable for vnd expression in NBs. These data indicate that Egfr resides atop the genetic hierarchy known to subdivide the neuroectoderm along the DV axis.

Our results suggest that Egfr patterns the neuroectoderm, at least in part, through its regulation of vnd and ind. To determine if additional genes act downstream of Egfr in this process, we compared the phenotypes of embryos singly mutant for Egfr and ind. We reasoned that if Egfr patterns the intermediate column solely through regulation of ind, then Egfr and ind mutant embryos should exhibit identical intermediate column phenotypes. To compare the early CNS phenotypes of Egfr and ind, we carried out a precise analysis of msh expression and the NB pattern. In both cases, Egfr exhibits a more severe phenotype than ind (Fig. 2). msh expression expands more medially in Egfr mutant embryos than in ind mutant embryos (Fig. 2). In addition, lateral NBs are most often separated from medial NBs by a gap in ind mutant embryos, while lateral NBs develop immediately adjacent to medial NBs in Egfr mutant embryos (Fig. 2). These data indicate a greater disruption to the intermediate column in Egfr mutant embryos than in ind mutant embryos. These phenotypic differences are consistent with the presence of additional genes acting downstream of Egfr and in parallel to ind to control cell fate in the intermediate column. However, Egfr maintains vnd expression in the neuroectoderm; thus, these data do not exclude the possibility that the differences in phenotype between Egfr and ind arise due to the late regulation of vnd expression by Egfr.

To test whether the phenotypic differences between ind and Egfr mutant embryos are an indirect result of the regulation of vnd expression by Egfr, we asked whether these differences were equalized in double mutants where vnd function is also removed. In vnd; ind mutant embryos, msh is expressed throughout the neuroectoderm, although its expression is higher in the lateral column relative to the medial column (Fig. 2). By contrast, msh is expressed at uniformly strong levels throughout the neuroectoderm in vnd; Egfr mutant embryos (Fig. 2). Thus, removal of vnd and Egfr causes a stronger derepression of msh in the neuroectoderm than loss of vnd and ind. These results suggest that additional gene(s) act downstream of Egfr and in parallel to vnd and ind to regulate DV pattern in the neuroectoderm. They also suggest that in the absence of vnd and Egfr function, the entire neuroectoderm acquires a lateral column fate.

Based on its restricted expression pattern in the ventral region of the neuroectoderm (Fig. 3) (Nambu and Nambu, 1996), we identified the Sox-domain-containing gene Dichaete as a likely candidate to act downstream of Egfr to regulate DV pattern in the neuroectoderm. To investigate whether Dichaete contributes to neuroectodermal patterning, we first determined the precise limits of Dichaete expression in the neuroectoderm using the expression of msh and achaete (ac) to mark different longitudinal columns (Fig. 3 and data not shown). msh is expressed in the lateral column; ac is expressed in neural equivalence groups (proneural clusters) in the medial and lateral columns of rows 3 and 7. Within the neuroectoderm, Dichaete expression begins during stage 7. Dichaete is expressed uniformly in the ventral region of the neuroectoderm with a lateral expression boundary that precisely abuts the medial limit of msh and ac expression in the lateral column (Fig. 3; data not shown). Within the neuroectoderm, Dichaete expression is restricted to the medial and intermediate columns through late stage 12, at which point Dichaete expression expands to include the entire neuroectoderm (data not shown). Thus, in contrast to the transient presence of Egfr and ind activity in the intermediate column, Dichaete is expressed in the intermediate and medial columns throughout all waves of NB formation.

During our analysis of Dichaete expression in the neuroectoderm, we noticed that most medial and intermediate NBs do not express Dichaete at detectable levels. To determine the pattern of Dichaete expression in neuroblasts we co-labeled wild-type embryos for Dichaete and hunchback, a marker of all neuroblasts (Kambadur et al., 1998). We observe, in general, that newly formed medial and intermediate column NBs express weak levels of Dichaete but that most older NBs in these domains do not express Dichaete. Exceptions to this exist, as two conspicuous NBs express Dichaete – one in the medial column of row 4 and one in the intermediate column of row 3 (data not shown). These data suggest that newly formed medial and intermediate NBs retain residual Dichaete expression from the neuroectoderm but that Dichaete expression is downregulated in most medial and intermediate NBs once they form.

In contrast to medial and intermediate NBs, many lateral NBs activate Dichaete at specific points in their lineages. NB 7-4 is the first lateral NB to activate Dichaete expression during late stage 10 (Fig. 3). Dichaete expression in lateral NBs is dynamic. NBs 5-6, NB 2-5 and eventually NB 3-5 express Dichaete (Fig. 3; data not shown for NB 3-5). Thus, all medial and intermediate column neuroectodermal cells express Dichaete but most medial and intermediate NBs do not express Dichaete. Conversely, lateral NBs but not neuroectodermal cells express Dichaete. These data are consistent with Dichaete regulating cell fate in the medial and intermediate neuroectodermal columns, and at specific points in the lineage of lateral NBs. We focus on the role Dichaete plays to regulate DV pattern and cell fate in the medial and intermediate neuroectodermal columns.

The restricted expression of Dichaete in the medial and intermediate columns suggests that Dichaete regulates cell fate and NB formation in this region. However, Dichaete mutant embryos exhibit AP patterning defects, owing to an early requirement in segmentation (Nambu and Nambu, 1996). The segmental defects are largely restricted to the abdominal segments; thoracic segments appear largely normal (Soriano and Russell, 1998). These segmentation defects could obscure a role for Dichaete during neuroectodermal patterning. Thus, we restricted our analysis of Dichaete function in the neuroectoderm to thoracic segments.

To investigate whether Dichaete patterns the neuroectoderm, we followed early neural development in embryos mutant for the Dichaete⁸⁷ and Dichaete⁹⁶ null alleles (Nambu and Nambu, 1996). We first tested whether Dichaete regulates gene expression in the neuroectoderm by following ac and msh expression. Normally, ac is expressed in the medial and lateral, but not intermediate, proneural clusters of rows 3 and 7 during the first wave of NB formation (Fig. 4) (Skeath and Carroll, 1992). In Dichaete mutant embryos, we observe a partial derepression of ac expression in the intermediate column (Fig. 4). We find that roughly 50% of the cells within the intermediate column of rows 3 and 7 express ac. ac expression in the medial column appears normal, as do the AP limits of ac expression in the thoracic segments. In contrast to ac, we detect no obvious alterations to msh expression in the neuroectoderm (data not shown). As ac is a key determinant of neural fate, we interpret its derepression in the intermediate column to indicate that Dichaete regulates cell fate in this column. However, our msh results indicate that lateral fates are specified normally in Dichaete mutant embryos.

ind normally represses ac expression in the intermediate column, because in ind mutant embryos, ac expression is completely derepressed within rows 3 and 7 of the intermediate column (Weiss et al., 1998). The Dichaete and ind phenotypes demonstrate that both genes are necessary for intermediate column fates. To determine if Dichaete and ind function in a linear pathway to regulate intermediate cell fates, we followed ind expression in Dichaete mutant embryos and Dichaete expression in ind mutant embryos. ind expression is normal in Dichaete mutant embryos (Fig. 4) and Dichaete expression is normal in ind mutant embryos (data not shown). Thus, ind and Dichaete are regulated independently of each other.

Double labeling Dichaete mutant embryos for ac and ind, and double labeling ind mutant embryos for ac and Dichaete revealed an interdependent relationship between Dichaete and ind. In Dichaete mutant embryos, a significant number of row 3 and 7 intermediate column cells and NBs co-expressed ac and ind – an occurrence never observed in wild-type embryos (Fig. 4). Thus, the ability of ind to repress ac in the intermediate column requires Dichaete activity. Reciprocally, in ind mutant embryos, all row 3 and 7 intermediate column cells co-express ac and Dichaete (data not shown). Thus, the ability of Dichaete to repress ac in the intermediate column requires ind activity.

Next, we assayed whether Dichaete regulates NB formation in the medial and intermediate columns. To do this, we followed the development of individual NBs using a panel of molecular markers that identify specific NBs or their progeny (Doe, 1992; Kambadur et al., 1998; Patel et al., 1989b). We used Svp-lacZ to label the medial column SI NBs 5-2 and 7-1, as well as SIII NB 4-1; castor expression to label the medial column SIII NB 6-1; and eve expression to label the first-born progeny of SI medial column NBs 1-1 and 7-1, and of SII intermediate column NB 4-2. In Dichaete mutant embryos, NBs 1-1 (99% formation; n=200), 5-2 (99.6%; n=226) and 7-1 (97.8%; n=226) develop normally. Thus, SI medial column NBs form normally in the absence of Dichaete function. By contrast, we observe defects in the formation of SII and SIII NBs in Dichaete mutants (Fig. 5; Table 1). For example, we fail to detect a Svp-lacZ-positive NB 4-1 in 30.9% of thoracic hemisegments (n=256) or a Castor-positive NB 6-1 in 7.9% of hemisegments (n=240). In addition, we fail to detect an Eve-positive RP2 neuron in 12.9% of hemisegments (n=210), suggesting the absence of NB 4-2 in these hemisegments. Together with our expression analyses, these phenotypic studies demonstrate that Dichaete acts in the neuroectoderm to promote the formation of late-forming NBs in the medial and intermediate columns.

Our loss of function analyses identify Dichaete as a regulator of DV pattern and cell fate in the neuroectoderm. To place Dichaete within the known genetic regulatory hierarchy that governs DV pattern in the neuroectoderm, we performed systematic molecular epistasis tests for Dichaete, ind, vnd and Egfr. Initially, we assayed vnd and ind expression, as well as Egfr activity in Dichaete mutant embryos. Dichaete mutant embryos exhibit no obvious defects to the expression of vnd or ind, or the activity of Egfr (Fig. 4; data not shown). Thus, Egfr, vnd and ind function upstream or in parallel to Dichaete.

To investigate whether Egfr, vnd or ind regulate Dichaete, we assayed Dichaete expression in embryos mutant for each gene. We observe no alterations to the initial pattern of Dichaete expression in vnd or ind mutants, or in embryos doubly mutant for vnd and ind (Fig. 6). Dichaete expression remains normal in ind mutant embryos throughout embryogenesis. However, by stage 11 in vnd and vnd; ind mutant embryos Dichaete expression narrows inappropriately to an irregularly patterned stripe two-to-four cells wide immediately adjacent to the ventral midline (Fig. 6). These results show that Dichaete is regulated independently of ind and is activated independently of vnd, but that vnd helps maintain Dichaete expression in the neuroectoderm.

In contrast to vnd and ind, the initial pattern of Dichaete in Egfr mutant embryos is greatly reduced in the intermediate column and moderately reduced in the medial column during early neurogenesis (stage 8; Fig. 6). By stage 11, Dichaete expression narrows inappropriately to a thin and irregular stripe zero-to-three cells wide immediately adjacent to the ventral midline; Dichaete expression in the ventral midline is normal (Fig. 6). These data identify Egfr as a key positive regulator of Dichaete in the neuroectoderm, and indicate that at least one other gene acts with Egfr to activate Dichaete expression in the medial column.

To investigate whether vnd acts with Egfr to promote Dichaete expression in the medial column, we followed Dichaete expression in vnd; Egfr mutant embryos. The initial pattern of Dichaete in these embryos is the same as that observed in Egfr mutant embryos (Fig. 6). However, by stage 11, Dichaete expression is completely absent from the neuroectoderm, although Dichaete expression is normal in the ventral midline. These results indicate that vnd and Egfr collaborate to maintain Dichaete expression in the neuroectoderm.

To determine if Egfr activity is sufficient to activate Dichaete expression, we used the GAL4/UAS system (Brand and Perrimon, 1993) to activate Egfr signaling throughout the early Drosophila embryo. Ubiquitous Egfr signaling activates Dichaete expression throughout the neuroectoderm but not in the dorsal ectoderm (data not shown). Thus, Egfr is necessary and sufficient to activate Dichaete in the neuroectoderm. However, in the dorsal ectoderm, either factors exist that inhibit the ability of Egfr to activate Dichaete or this domain lacks co-factors required for Egfr to activate Dichaete. Our molecular epistasis tests place Egfr upstream of Dichaete and indicate that vnd, ind and Dichaete function largely in parallel to regulate pattern and cell fate in the neuroectoderm.

The parallel genetic activities of Dichaete, vnd and ind, the co-expression of Dichaete with vnd and ind, and the similarity of the early Dichaete CNS phenotype to those of vnd and ind (Chu et al., 1998; McDonald et al., 1998; Weiss et al., 1998) led us to test whether Dichaete interacted genetically with vnd and ind. To ascertain whether Dichaete interacted with vnd we made vnd; Dichaete double mutants and assayed the formation of medial column SIII NBs 4-1 and 6-1. In Dichaete mutant embryos, NBs 4-1 and 6-1 formed in 69.1% (n=256) and in 92.1% (n=240) of hemisegments, respectively (Fig. 5; Table 1). Although a previous report indicated little to no effect of vnd function on SIII-SV NB formation (Chu et al., 1998), in vnd mutant embryos we found that NBs 4-1 and 6-1 formed in 39.3% (n=168) and 35.5% (n=276) of hemisegments, respectively (Fig. 5; Table 1). In vnd; Dichaete mutant embryos NBs 4-1 and 6-1 formed in 10.8% (n=102) and 9.1% (n=132) of hemisegments, respectively (Fig. 5; Table 1). The increased defects in NB formation in vnd; Dichaete mutant embryos relative to either single mutant confirms that Dichaete and vnd do not act in a linear pathway to regulate NB formation, rather they demonstrate that Dichaete and vnd function in parallel to control NB formation in the medial column.

We should note that defects in NB formation in vnd; Dichaete mutant embryos are more severe than would be expected if these genes function independently. For example, if two genes act independently to promote NB formation then the frequency of NB formation in the double mutant would be the product of the individual probabilities that the indicated NB will form in each single mutant. Thus, if vnd and Dichaete function independently, we would expect NB 4-1 to form 27.2% of the time (0.393×0.691=0.272) and NB 6-1 to form 32.7% of the time (0.355×0.921=0.327) in vnd; Dichaete mutant embryos. However, NBs 4-1 and 6-1 form ∼10% of the time in vnd; Dichaete mutant embryos – roughly threefold more severe than predicted for independently acting genes. These results reveal a genetic interaction between Dichaete and vnd. Furthermore, we interpret these results to suggest that the activities of vnd and Dichaete are more convergent than parallel with respect to NB formation.

Next, we tested for genetic interactions between Dichaete and ind. The partial derepression of ac expression and the incomplete loss of an Eve-positive RP2 neuron are the most sensitive assays for Dichaete function in the intermediate column. However, strong alleles of ind cause a complete derepression of ac expression, and a complete loss of RP2 neurons in this domain (Weiss et al., 1998). Thus, an analysis of Dichaete ind double mutant embryos using these markers would be uninformative. To circumvent this problem, we tested whether ind dominantly enhances the Dichaete intermediate column ac and RP2 phenotypes. Embryos heterozygous for ind exhibit wild-type ac expression and RP2 formation. However, Dichaete ind/Dichaete + mutant embryos exhibit enhanced derepression of ac expression and an approximately threefold enhancement of the RP2 loss phenotype relative to Dichaete mutant embryos (Fig. 4; Table 1). The dominant enhancement of the Dichaete phenotype by ind reveals a genetic interaction between Dichaete and ind.

Our initial interest in Dichaete arose from our observation that vnd; Egfr mutant embryos exhibit a more severe neuroectodermal phenotype than vnd; ind mutant embryos. This suggested that at least one other gene acts downstream of Egfr, and in parallel to vnd and ind to pattern the early neuroectoderm and led to our analysis of Dichaete. To determine if the continued function of Dichaete in vnd; ind mutant embryos can explain the phenotypic differences between vnd; ind and vnd; Egfr mutant embryos, we followed msh expression in vnd; Dichaete ind triple mutant embryos. In this background, we observe a complete and uniform derepression of msh expression throughout the neuroectoderm (Fig. 2). The msh phenotype of vnd; Dichaete ind embryos is essentially identical to that of vnd; Egfr embryos, and more severe than that of vnd; ind embryos (Fig. 2). Thus, with respect to msh expression the difference between the vnd; ind and vnd; Egfr mutant phenotypes appears to result from the persistent function of Dichaete in vnd; ind mutant embryos.

Prior work has underlined the pivotal role Egfr, vnd and ind play to regulate DV pattern and cell-fate in the neuroectoderm (reviewed by Skeath, 1999). The results in this paper indicate that additional genes act with this genetic trio to pattern the neuroectoderm. We identified Dichaete as a key regulator of DV pattern in the neuroectoderm. Dichaete is expressed in the medial and intermediate columns and regulates cell fate and NB formation in these domains. Within the neuroectoderm, Dichaete acts downstream of Egfr and in parallel to vnd and ind (Fig. 7). Together with biochemical research on Sox-domain-containing genes in vertebrates (reviewed by Kamachi et al., 2000) our work supports a model (Fig. 7) in which Dichaete protein physically associates with Vnd and Ind to regulate target gene expression and NB formation in distinct neuroectodermal columns.

Our interest in Dichaete arose owing to our observation that removal of vnd and Egfr function caused a stronger derepression of msh expression in the neuroectoderm than removal of vnd and ind function. These results contrast slightly with previous research that did not identify a phenotypic difference between vnd; ind and vnd; Egfr mutant embryos (von Ohlen and Doe, 2000). This work analyzed msh expression in the neuroectoderm at a later stage (late stage 9) than ours. At late stage 9, we also observe identical alterations to msh expression in vnd; ind mutant embryos relative to vnd; Egfr mutant embryos. However, the msh expression pattern is dynamic – rapidly changing from uniform expression in the lateral column during stage 8 to a segmentally modulated pattern of cell clusters located within the lateral half of the neuroectoderm by stage 10. We attribute the differences in our observations to the different stages used to assay the effects of vnd, ind and Egfr on neuroectodermal development in the two studies.

Dichaete is expressed and regulates cell fate in the medial and intermediate neuroectodermal columns. However, Dichaete carries out distinct functions in each domain: Dichaete represses ac expression in the intermediate column but has no effect on ac expression in the medial column where Dichaete and ac are co-expressed.

How might Dichaete exhibit region specific effects on putative target genes? Work from vertebrate systems suggests that individual Sox-domain-containing proteins exhibit a widespread ability to partner with different transcription factors (reviewed by Kamachi et al., 2000). Thus, Dichaete protein could exhibit column-specific functions via its association with different transcription factors in different domains. The formation of distinct protein complexes containing Fish could alter the output of Fish activity in at least two ways. Different protein complexes that contain Fish could exhibit different effects on transcription: repression versus activation. Alternatively, different Fish-containing protein complexes could exhibit distinct DNA-binding properties and therefore bind distinct recognition sites. These two possibilities are not mutually exclusive, and different Fish-containing protein complexes may both bind different recognition sites and exert different transcriptional effects on target genes.

Examples of both forms of regulation are known. In the early Drosophila embryo, the transcription factor Dorsal activates one set of target genes ventrally and represses a distinct set dorsally (Jiang et al., 1993). On its own, Dorsal functions as a transcriptional activator. However, in the dorsal region of the embryo, the interaction of Dorsal with a co-factor that binds to adjacent sites on target promoters converts Dorsal to a repressor. Although less well-defined mechanistically, the vertebrate Sox2 protein appears capable of activating or repressing target gene expression depending on cell-type and the target promoter (Botquin et al., 1998). In addition, work on vertebrate Sox domain proteins indicates that the composition of Sox-protein containing complexes modulates the DNA-binding specificity of these complexes. For example, in lens cells, Sox2 interacts with the DNA-binding factor δEF3 and binds to a bipartite recognition site on the δ-crystallin enhancer (Kamachi et al., 1998; Kamachi et al., 1999). In embryonic stem cells, Sox2 interacts with Oct3/4 and binds to a different recognition site in the Fgf4 minimal enhancer (Ambrosetti et al., 1997). In both enhancers, Sox2 binds to the same individual sequence. However, the specificity for the entire recognition site in one enhancer over the other arises as a consequence of the interaction of Sox2 with different transcription factors in different cell types and the distinct DNA-binding preferences of the entire complex.

Based on these data, we expect Dichaete to associate with different transcription factors in the medial and intermediate columns to carry out its column-specific effects on target genes. The results in this paper identify Vnd and Ind as excellent candidates to be column-specific factors that associate with Dichaete and enable Dichaete to regulate transcription in a region specific manner. First, Dichaete is co-expressed with Vnd in the medial column and Ind in the intermediate column. Second, the neuroectodermal Dichaete mutant phenotype is similar to those of vnd and ind. Third, Dichaete functions in parallel to vnd and ind in the neuroectoderm. Fourth, Dichaete exhibits dose-sensitive interactions with ind and genetic interactions with vnd – consistent with these proteins interacting physically. Based on these data, we speculate that physical interactions between Dichaete and Vnd in the medial column and Dichaete and Ind in the intermediate column mediate the ability of distinct Dichaete protein complexes to bind to and to activate or to repress distinct target genes (Fig. 7). Validation of this model awaits the determination of whether Dichaete associates with Vnd or Ind, and how these proteins regulate target gene activity. However, recent results provide precedence for the model as genetic interactions between Dichaete, single-minded and drifter during midline development in the Drosophila CNS led to experiments that showed Dichaete physically associates with the Single-minded and Drifter proteins (Ma et al., 2000).

Our results place Dichaete within the known genetic regulatory hierarchy that controls pattern and cell fate along the DV extent of the neuroectoderm (Fig. 7). In the future, we expect many additional genes to join this pathway. For example, the Sox-domain-containing gene sox-neuro is expressed throughout the entire neuroectoderm (Cremazy et al., 2000) and it may exhibit region-specific effects in the neuroectoderm in a manner similar to that proposed by us for Dichaete. In addition, the Ras-pathway antagonist yan is expressed in the lateral half of the neuroectoderm during early neurogenesis and may help regulate pattern and cell fate in this domain (G. Z. and J. B. S., unpublished). A complete understanding of the genetic and molecular mechanisms that pattern the neuroectoderm requires the identification of all such genes and the elucidation of how these genes interact to regulate cell fate along the DV axis of the neuroectoderm.

As first noted by D’Alessio and Frasch (D’Alessio and Frasch, 1996), there is a remarkable conservation of gene expression patterns along the DV axis of the Drosophila neuroectoderm and the vertebrate neural tube. Members of the vertebrate vnd/Nkx2.2 gene family are expressed and control cell fate within the ventral/medial domain of the neural tube (Pabst et al., 1998; Price et al., 1992). Gsh1 and Gsh2, the vertebrate homologs of ind, are expressed in an intermediate position in the neural tube (Hsieh-Li et al., 1995; Valerius et al., 1995), while murine orthologs of msh are expressed in the most lateral region of the neural tube (Davidson, 1995).

The expression patterns of Sox-domain-containing genes in the Drosophila neuroectoderm and the vertebrate neural tube are also similar. For example, most vertebrate Sox genes are expressed throughout the entire neural plate (Wegner, 1999) similar to the expression of sox-neuro throughout the Drosophila neuroectoderm (Cremazy et al., 2000). In addition, chick Sox21 expression is expressed in the ventral half of the early vertebrate neural tube (Rex et al., 1997) – reminiscent of the Dichaete expression pattern. In the vertebrate neural tube, Sox domain proteins are likely to exhibit region specific effects via their interaction with transcription factors expressed in spatially restricted patterns. The proteins encoded by the Nkx2.2, Gsh1/2 and Msx genes are excellent candidates to interact with Sox proteins in this context. Future research in both flies and vertebrates will identify the partners of Sox proteins and the precise molecular mechanisms through which Sox-protein-containing complexes regulate pattern and cell fate in developing nervous systems. As this research progresses, it will be exciting to see the extent of conservation as well as divergence between the molecular logic employed by Sox proteins and their cohorts in Drosophila and vertebrates to regulate neural development.

We are especially grateful to John Nambu for helpful discussions, and for providing Dichaete mutant alleles and anti-Dichaete antibody. We thank Alan Michelson, Ward Odenwald, Eric Spana, Harald Vaessin, Tonia von Ohlen, Chris Doe, Dervla Mellerick-Dressler, Kalpana White, Daniel St Johnston and the Developmental Studies Hybridoma bank for antibodies and/or fly stocks. As always, we greatly appreciate the help received from Kathy Matthews and the Bloomington Stock Center. This work was supported by a grant from NINDS (RO1-NS36570) and by a Howard Hughes Medical Institute Research Resources Program for Medical Schools, #76296-538202, Junior Faculty Award.