Neurogenic genes control gene expression at the transcriptional level in early neurogenesis and in mesectoderm specification

Neuroblasts, the progenitor cells of the central nervous system (CNS) in insects, delaminate from the embryonic neuroectoderm of Drosophila in a stereotyped manner that involves several waves of segregation (Hartenstein and Campos-Ortega, 1984; Doe, 1992). It has been shown, at least for thirteen neuroblasts that segregate in each hemisegment during the first two waves (S1 and S2), that their formation is initiated by the acquisition of neural potential by small groups of ectodermal cells, called proneural clusters (see Campos-Ortega, 1993, for a recent review). Proneural clusters in the embryonic neuroectoderm are defined by the expression of achaete (ac), scute (sc) and lethal of scute (l’sc), genes of the achaete-scute complex (AS-C) (Cabrera et al., 1987; Romani et al., 1987; Martín-Bermudo et al., 1991; Skeath and Carroll, 1992; Ruiz-Gómez and Ghysen, 1993). During segregation, each neuroblast accumulates the highest levels of AS-C proteins, whereas in the remaining cells of the proneural cluster these proteins gradually dissapear. This pattern of expression and the analysis of mutant phenotypes indicates that the function of the AS-C genes is to confer upon cells of the clusters the capacity to become neuroblasts (reviewed by Campuzano and Modolell, 1992; Campos-Ortega, 1993). The restriction of the neural fate to a single cell of a proneural cluster is mediated by neurogenic genes, through an inhibitory process that involves intercellular communication (reviewed by Ghysen et al., 1993). According to Cabrera (1990, 1992), the molecular outcome of the inhibitory process would be the accumulation of an active, unphosphorylated form of the AS-C proteins exclusively in the presumptive neuroblasts. This contrasts with another report which suggests that the control of the AS-C proteins is largely transcriptional (Skeath and Carroll, 1992).

It has been proposed that neurogenic genes constitute a functional cassette that mediates cell interactions in many developmental processes (Ruohola et al., 1991). One such process could be the specification of the mesectoderm in the early embryo, for the analysis of mesectoderm-specific gene expression in different mutants has lead to the suggestion that mesectoderm specification requires an inductive signal from the mesoderm (see Nambu et al., 1993, and references therein). Additional evidence for the induction derives from the ectopic transplantation of mesodermal cells, which results in the expression in the surrounding ectoderm of single-minded (sim), a master regulatory gene for mesectodermal development (Leptin and Roth, 1994; Nambu et al., 1991).

In this paper, we first attempt to clarify the discrepancies existing in the control of proneural gene expression by neuro-genic genes. Our results indicate that neurogenic genes principally regulate l’sc expression at the transcriptional level, without affecting the domains of proneural gene expression. Secondly, we analyse whether neurogenic genes intervene in the transmission of the inductive signal that leads to mesectoderm specification. The results indicate that N and neuralized (neu) are required for the establishment of the mesectodermal fate and suggest also that Dl and members of the Enhancer of split gene complex (E(spl)-C) may be involved in the same process.

The following Drosophila strains were used: wild-type Oregon-R, N^55e11, neu^IF65, Df(3R)Dl^FX3, Df(3R)E(spl)^RB25.1 (the latter three mutations balanced over a TM3 lacZ chromosome), ovo^D1 (Lindsley and Zimm, 1992), P[w⁺, sim-lacZ] (the sim promoter fused to lacZ, a gift from S. Crews) (Nambu et al., 1991), FRT¹⁰¹ and hsFLP³⁸ (Chou and Perrimon, 1992).

Female germ-line clones were induced in N^55e11/ovo^D1 larvae by Xirradiation (Jiménez and Campos-Ortega, 1982), or using the FRT/FLP technique (Chou and Perrimon, 1992), as described by Menne and Klämbt (1994). Virgins of the appropriate genotype were crossed either to wild-type males or to P[w⁺, sim-lacZ] males, to study sim expression. Heterozygous N^55e11/+ female embryos were recognized with an anti-Sex lethal antibody.

Two classes of anti-L’sc polyclonal antibodies were used. One class is a rat antibody raised against a fusion protein that contains the entire L’sc sequence, except the first fifteen amino acids (Martín-Bermudo et al., 1991). A second class are different antibodies raised against the synthetic peptide DDEELLDYISSWQE, corresponding to the C terminus of the translated sequence of l’sc (Alonso and Cabrera, 1988; Martín-Bermudo et al., 1993). Of this latter class, a rabbit serum was a gift from the late C. Cabrera. In addition, we raised and affinity purified similar antibodies in rats and in one rabbit, following essentially the protocol described by Cabrera (1990). Other antibodies used were: anti-Ac (from J. Skeath and S. Carroll), anti-β-gal (Cappel), anti-Sex lethal (from L. García-Alonso), anti-Hunchback (from P. Macdonald) and anti-Snail (from A. Alberga). Embryos were fixed and stained essentially as described previously (Martín-Bermudo et al., 1991, 1993). Staged embryos were used in most instances, and were collected at 20 minutes (early stage 8), or 35–40 minutes (late stage 8) after the onset of gastrulation at 25°C. In some cases, embryos embedded in Epon were sectioned at 5 μm.

In situ hybridization essentially followed the protocol of Tautz and Pfeifle (1989). After staining, embryos were dehydrated first through an ethanol series, then two times in acetone for a total of 3–5 minutes, transferred to a mixture of acetone:Epon (1:1) and inmediately spread over a microscope slide. Once the acetone had evaporated, the embryos were dissected and mounted in Epon.

The detection of AS-C proteins during early neurogenesis with different antibodies led to different hypothesis about the control of AS-C expression (Cabrera, 1990, 1992; MartínBermudo et al., 1991; Skeath and Carroll, 1992). To clarify this issue, we first analysed in detail l’sc RNA and protein expression patterns during wildtype segregation of neuroblasts in stage 8. For simplicity, we restrict our description to the ventral-most region of the neuroectoderm, which gives rise to four S1 neuroblasts (2 –2, MP2, 5 –2 and 7 –1) per hemisegment (Doe, 1992).

l’sc RNA is found at early stage 8 in the proneural clusters of neuroblasts 2 –2, 5 –2 and 7 –1 (Fig. 1A). Neuroblast MP2 arises from a cluster expressing ac and sc but not l’sc (Martín-Bermudo et al., 1991; Skeath and Carroll, 1992; Ruiz-Gómez and Ghysen, 1993). At late stage 8, l’sc transcription has decayed in the neuroectoderm (Fig. 1B) but it is still active in neuroblasts 2–2, 5–2 and 7–1, which have just segregated (Fig. 1C). The pattern of L’sc protein accumulation closely follows that of transcription, as detected either with an antibody raised against the entire protein (Fig. 1D-F; see also MartínBermudo et al., 1991), or with any of those raised against a Cterminal peptide (Fig. 1G-I). These results strongly suggest that the control of the spatial distribution of L’sc is largely transcriptional.

To analyse, at the same level of resolution as above, how the expression of proneural genes is modified in neurogenic mutants, we studied the RNA and protein patterns of l’sc in mutants for neu, Dl, N, and the E(spl)-C. The analysis was restricted to the zygotic lack-of-function situation, except for N. In this case, N⁻ embryos were derived from homozygous N⁻ female germ line clones, for we found it necessary to eliminate the maternal contribution in order to detect mutant phenotypes during stage 8.

Proneural clusters in the four mutants, revealed by l’sc transcription, are basically normal (Fig. 2A). The only difference from the wild type is that the bilaterally symmetric clusters in neu⁻ (Fig. 2A) and N⁻ (not shown) embryos tend to fuse at the midline (see below). At late stage 8, l’sc transcription does not become restricted to the neuroblast and persists in entire clusters, which otherwise maintain their initial size (Fig. 2B).

As in the wild type, the pattern of L’sc protein accumulation closely follows that of transcription in neurogenic mutants. Mutant proneural clusters look essentially normal, as shown by double staining for L’sc and Ac. In the wild-type control (Fig. 3A), S1 clusters occupy the entire ventral-most region of the neuroectoderm, expressing either ac, l’sc, or both. The precise cell-by-cell apposition of those clusters that express either ac or l’sc suggests that in this region a cell never belongs simultaneously to more than one cluster. Proneural clusters in N⁻ embryos are also not intermingled (Fig. 3B), indicating that they are not enlarged, at least along the anteroposterior axis. Unlike the wild type, most proneural cells in late stage 8 mutants still bear high levels of L’sc (Figs 2C-F and 3C) and already show the characteristic morphology and patterns of gene expression of neuroblasts (not shown; Jiménez and Campos-Ortega, 1990; Campos-Ortega and Haenlin, 1992). As already noticed for the RNA, the domain of L’sc accumulation is the same as in earlier stages.

The mesectoderm forms during gastrulation, when two singlecell wide rows join at the ventral midline. In a stage 8 wildtype embryo these two rows are still devoid of proneural gene expression (Fig. 1), but already express sim (Fig. 4A; Thomas et al., 1988). As shown in Fig. 2, the presumptive mesectoderm appears to be almost absent in N⁻ and neu⁻ embryos. In addition, defective sim expression suggests that most presumptive mesectodermal cells have not adopted their normal fate (Fig. 4B,C). Slight defects in sim expression are also detected in Dl⁻ and in E(spl)-C⁻ embryos (Fig. 4D,E). In cross sections, mesectodermal cells are readily observable in the wild type (Fig. 5A), but seem to have adopted a neuroectodermal fate in N⁻ (Fig. 5B) and neu⁻ embryos (not shown). Comparison of Figs 1A and 2A, or of Fig. 3A and 3B, suggests that the medial proneural clusters of the two mutants might be ventrally enlarged. Alternatively, the position of the entire neuroectoderm may be ventrally shifted by one cell diameter, similar to the more extensive shift that it undergoes when the mesoderm is removed (Rao et al., 1991).

In the wild-type CNS, the proneural proteins become restricted from a cluster of cells to a single delaminating neuroblast. As for ac (Skeath and Carroll, 1992), we have shown that the restriction of l’sc is mainly accomplished at the transcriptional level. In neurogenic mutants, the transcriptional repression of l’sc fails to occur (see also Brand and Campos-Ortega, 1988) and all proneural cells accumulate high levels of L’sc protein, as is the case for ac (Skeath and Carroll, 1992; Ruiz-Gómez and Ghysen, 1993). A different mode of control of l’sc activity was proposed by Cabrera (1992), using an antibody against a C-terminal peptide of L’sc that contains a putative tyrosine phosphorylation site (Cabrera, 1990). With that antibody, L’sc was apparently detected only in neuroblasts, leading to the postulation that two forms of the protein exist. One form, accumulating in neuroblasts, would be unphosphorylated and functionally active. The other form, present in every cell that transcribes l’sc, would be phosphorylated and inactive. Accordingly, the anti-peptide antibody was found to stain a larger number of cells in Dl⁻ and N⁻ embryos, suggesting that neurogenic genes control l’sc at the posttranscriptional level (Cabrera, 1990, 1992). Our observations that proneural cells are similarly stained with the two available classes of anti-L’sc antibodies would suggest that both states of phosphorylation of L’sc, if present, would coexist in proneural cells. However, our observations are equally compatible with the possibility that antibodies against the C-terminal peptide cannot discriminate between different protein forms. In this context, it has been shown that neither the deletion of the C-terminal region, nor the substitution of the tyrosine in this domain by a residue that cannot be phosphorylated, have pronounced influence on the proneural activity of L’sc assayed in imaginal discs (Hinz et al., 1994).

Our analysis of embryos double stained for Ac and L’sc demonstrate, as already noted (Brand and Campos-Ortega, 1988; Ruiz-Gómez and Ghysen, 1993), that neurogenic genes do not play a role in the emergence of proneural clusters in the embryo, except for a possible slight ventral enlargement of the medial clusters in Nand neumutants. However, the possibility that the domains of proneural gene expression later become enlarged in neurogenic mutants has been a controversial issue (Brand and Campos-Ortega, 1988; Campos-Ortega, 1993; Ruiz-Gómez and Ghysen, 1993). In this regard, our results also support the view that, in neurogenic mutants, the domains of proneural gene expression in the CNS do not become enlarged at the stage when S1 neuroblasts normally delaminate.

As indicated in the Introduction, several lines of evidence suggest that mesectodermal specification partially requires an inductive signal from the mesoderm. Our observation that N is required for the activation of sim (see also Menne and Klämbt, 1994) and for the early repression of proneural genes in the presumptive mesectoderm suggests that mesectoderm induction involves cell communication. We have also found that neu participates in the same process, raising the possibility that the family of neurogenic genes participate in the specification of the mesectoderm, in a way analogous to their role during neuroblast formation (de la Concha et al., 1988). Thus, expression of neu and Dl in the mesoderm (Boulianne et al., 1991; Price et al., 1993; Kopczynski and Muskavitch, 1989; Haenlin et al., 1990) would correlate with their participation in the emission of the inductive signal. Similarly, specific expression of four E(spl)-C genes in the mesectoderm (Knust et al., 1992) suggests that they function to implement the reception of the signal. Maternal expression of Dl and of m3, a fifth E(spl)-C gene (Kopczynski and Muskavitch, 1989; Haenlin et al., 1990; Knust et al., 1987), could explain the lack of a strong mesectodermal phenotype in zygotic Dl⁻ and E(spl)-C⁻ mutants.

We thank S. Campuzano, A. Martínez Arias and J. Modolell for helpful comments on the manuscript; and A. Alberga, S. Crews, L. García Alonso, P. Macdonald, A. Martínez Arias and J. Skeath for antibodies and fly stocks. M. D. M.-B. and A. C. were predoctoral fellows of Ministerio de Educación y Ciencia. This work was supported by grant PB90-0082 from DGICIT to F. J. and an institutional grant from Fundación Ramón Areces.