Epidermal Progenitor Cells

The Enhancer of split complex [E(spl)-C] of Drosophila melanogaster is located in the 96F region of the third chromosome and comprises at least seven structurally related genes, HLH-m6, HLH-my, HLH-ma, HLH-m3, HLH-m5, HLH-m7 and E(sp1). The functions of these genes are required during early neurogenesis to give neuroectodermal cells access to the epidermal pathway of development. Another gene in the 96F region, namely groucho, is also required for this process. However, groucho is not structurally related to, and appears to act independently of, the genes of the E(sp1)-C; the possibility is discussed that groucho acts upstream to the E(sp1)-C genes. Indirect evidence suggests that a neighboring transcription unit (m4) may also take part in the process. Of all these genes, only gro is essential; m4 is a dispensable gene, the deletion of which does not produce detectable morphogenetic abnormalities, and the genes of the E(spl)-C are to some extent redundant and can partially substitute for each other. This redundancy is probably due to the fact that the seven genes of the E(sp1)-C encode highly conserved putative DNA-binding proteins of the bHLH family. The genes of the complex are interspersed among other genes which appear to be unrelated to the neuroepidermal lineage dichotomy. S INCE Enhancer ofsplit [E(spl)] was first recognized to be one of the neurogenic loci of Drosophila melanogaster (LEHMANN et al. 1983), genetic analyses have uncovered a number of unusual features of this locus (VASSIN, VIELMETTER and CAMPOS-ORTEGA 1985; KNUST et al. 1987; ZIEMER et al. 1988). Thus, it was shown that E(sp1) acts downstream of all the other neurogenic loci in the genetic network that controls the segregation of epidermal and neural progenitor cells (VASSIN, VIELMETTER and CAMPOS-ORTEGA 1985; DE LA CONCHA et al. 1988). The finding that mutants lacking this locus express their phenotype cell autonomously led to the conclusion that the functions of the E(@) locus are needed either to receive or to process signals that regulate the epidermal pathway of development within the neuroectoderm (TECHNAU and CAMPOS-ORTEGA 1987). A number of lethal mutations in the E(@) region showed a complex pattern of heteroallelic complementation and provided evidence suggesting that the neurogenic function ascribed to E(@) is in fact carried out by a gene complex rather than by a single gene (ZIEMER et al. 1988). In particular, the observation that increasingly large deletions cause neurogenic phenotypes of increasing severity indicated that several functions participate in the neuroepidermal lineage dichotomy (KNUST et al. 1987; ZIEMER et al. 1988). The cloning of the E(@) genomic region (KNUST, TIETZE and Genetics 1 3 2 481-503 (October, 1992) CAMPOS-ORTEGA 1987) provided molecular evidence for genetic complexity, as mutations phenotypically related to E(sp1) were found to affect different genes. Indeed, the DNA in the region studied contains 13 major transcription units, which were called m6, my, m/3, ma, ml, m2, m3, m4, m5, m6, m7, m8 and m9/m10. During the period of segregation of the neuroblasts, RNA products of seven of these transcription units (m6, my, m/3, m4, m5, m7 and m8) exhibit nearly identical distribution patterns, which correlate with epidermoblasts (KNUST, TIETZE and CAMPOS-ORTEGA 1987; KNUST et al. 1992). This pattern of expression of course supports a role for the genes of the E(sp1) region in controlling epidermal development of neuroectodermal cells, as had already been inferred from phenotypical studies. KLAMBT et al. (1989) showed that m5, m7 and m8 encode proteins of the bHLH family. Recently, further four transcription units in the region, m3, m/3, my and m6, have also been found to encode proteins of the bHLH family (KNUST et al. 1992). Since seven neighboring genes are structurally and functionally related, we propose that they form a gene complex, called henceforth E(sp1)-C, probably arisen by duplication and mutation of a primordial gene. We shall see in this report that, except for m8, which encodes the E(sp1) function itself (see below), no visible phenotypes are known to be associated with mutations in any of these transcription units; there482 H. Schrons, E. Knust and J. A. Campos-Ortega fore, we use the helix-loop-helix motif to designate the other six genes encoding bHLH proteins, i.e., HLH-ma, HLH-my, HLH-mP, HLH-m3, HLH-m5 and HLH-m7. The gene grouch (gro) lies immediately adjacent to the E(sp1)-C, and is also required for segregation of neuroblasts and epidermoblasts. gro was named for a mutant with an increased number of supraorbital bristles (LINDSLEY and GRELL 1968). Various genetic analyses (KNUST et al. 1987; ZIEMER et al. 1988; PREISS, HARTLEY and ARTAVANIS-TSAKONAS 1988) have shown that the phenotype of gro is not complemented by any of several revertants of the E(sp1)D mutation. Hence, it was proposed that gro is allelic to E(sp1). However, it later emerged that the genetic variants used for complementation tests were in fact deletions that affected both gro and the E(sp1)-C. The following evidence shows that gro and E(sp1) are in fact two different genes. KLAMBT et al. (1989) unambiguously demonstrated that the enhancement of split that defines the E(sp1) gene (WELSHONS 1956) is due to molecular lesions in the transcription unit m8. This identification was based on genetic transformation experiments, showing that the transcription unit m8 derived from the E(sp1)D strain is both necessary and sufficient to enhance the phenotype of split (KLAMBT et al. 1989; TIETZE, OELLERS and KNUST 1992). Therefore, using the normal nomenclature convention of Drosophila genetics according to which a gene should be assigned a name that is descriptive of the mutant phenotype (LINDSLEY and ZIMM 1985; ASHBURNER 1989), transcription unit m8 is E(@). However, the E(sP1)D chromosome carries, in addition to the E(@) mutation, other lesions which cause phenotypic traits unrelated to the enhancement of split. One of these additional mutations is responsible for lethality of heterozygotes carrying E(sp1)D over large deletions of the region (LEHMANN et al. 1983; KNUST et al. 1987; PREISS, HARTLEY and ARTAVANIS-TSAKONAS 1988; TIETZE, OELLERS and KNUST 1992); this phenotype is probably caused by the insertion of a middle repetitive element in an intron of transcription unit m9/m10 (KNUST, TIETZE and CAMPOS-ORTEGA 1987; PREISS, HARTLEY and ARTAVANIS-TSAKONAS 1988). A transgenic fragment comprising m9/mlO from the E(sp1)D genome does not at all modify the split phenotype (TIETZE, OELLERS and KNUST 1992); therefore, the insertion is unrelated to the enhancement of split. PREISS, HARTLEY and ARTAVANIS-TSAKONAS (1 988) unambiguously identified transcription unit m9/mIO as gro, as they showed that a genomic fragment containing only the wild-type m9/mIO transcription unit rescues the bristle defects that characterize the gro mutation; in addition, the same authors showed that this transgenic wild-type m9/mIO rescues the maternally conditioned lethality of heterozygotes of E(sp1)D with large deletions of the region. TIETZE, OELLERS and KNUST (1 992) have shown that the corresponding transgenic m9/mlO fragment from the E(sp1)D strain fails to rescue gro lethal mutations. Hence, the insertion in transcription unit m9/m10 of the E(sp1)D strain causes a hypomorphic, visible mutation in the gro gene. The gro protein exhibits weak similarity to the P-subunit of transducin, a G-protein (HARTLEY, PREISS and ARTAVANIS-TSAKONAS 1988), and is therefore structurally unrelated to the genes of the E(sp1)-C, which encode bHLH proteins. Therefore, the E(sp1)D chromosome carries at least two different mutations: a gain-of-function mutation in E($) (ma) (KLAMBT et al. 1989) and a hypomorphic mutation in gro (rn9lmlO) (TIETZE, OELLERS and KNUST 1992). The continued denomination of m9/ m10 as E(sp1) (HARTLEY, PREISS and ARTAVANIS-TSAKONAS 1988; PREISS, HARTLEY and ARTAVANIS-TSAKONAS 1988; SHEPARD, BROVERMAN and MUSKAVITCH 1989; ARTAVANIS-TSAKONAS and IMPSON 199 1 ;DELIDAKIS et al. 199 1) has created much confusion in the literature. We show below that gro is actually a distinct neurogenic gene that happens to be in the neighborhood of, but is ancestrally unrelated to, the genes of the E(sp1)-C. All the genes under discussion are located in the 96F region of the third chromosome; hence, we refer to the E(sp1)-C and adjacent genes as the 96F region. Over the years, a large number of genetic variants, including deletions of various sizes and transgenic animals carrying wild-type copies of many of the genes of the 96F region, have been collected. This material has allowed us to carry out a comprehensive genetic analysis of the organization of this region, the results of which are presented below. These results lead to three main conclusions. (i) There is evidence that eight genes in the region, HLH-ma, HLH-my, HLH-mP, HLH-m3, HLH-m5, HLH-m7, E(@) and gro, are required to allow epidermal development of the neuroectodermal cells. Weaker indications suggest the participation of m4 in this process. (ii) The genes of the E(sp1)-C act as a functional unit composed of redundant genes which can partially substitute for each other. (iii) gro is itself a neurogenic gene showing prominent maternal expression. MATERIALS AND METHODS Genetic variants: The genetic variants of the 96F region used in our study were obtained from various sources. D ~ ( ' ~ R ) E ( S ~ ~ ) R ~ ~ ' I and Dji'3R)E(~pl)~-*~.' were recovered as Xray-induced revertants of E(spl)D (see KNUST et al. 1987). Dji'3R)boss" is a lethal allele of boss given to us by L. ZIPURSKY. D f 3 R ) g r 0 ~ Y ~ ~ ~ ~ ~ ~ . ~ , Dj('3R)gr01y78R-r17'.t Dj('3R)g,.orY78R-r72.' and Dj('3R)grorY78R-r8.1 (this work) resulted from imprecise excisions of a P[ry+-lacZ]-element insertion (kindly given to us by CHRISTIAN KLAMBT and COREY GOODMAN), located in the 5' noncoding region of gro (this work); we have named this line Is(3R)grory7", es P[ry+ lacZ]. Putative Organization of the E(spl)-C