The Drosophila orphan nuclear receptor Seven-up requires the Ras pathway for its function in photoreceptor determination

Transcription

1 Development 121, (1995) Printed in Great Britain The Company of Biologists Limited The Drosophila orphan nuclear receptor Seven-up requires the Ras pathway for its function in photoreceptor determination Gerrit Begemann 1, Anne-Marie Michon 1, Loesje v.d.voorn 1, Roger Wepf 2 and Marek Mlodzik 1, * 1 Differentiation and 2 Cell Biophysics Programme, EMBL, Meyerhofstrasse 1, D Heidelberg, Germany *Author for correspondence SUMMARY The Drosophila seven-up (svp) gene specifies outer photoreceptor cell fate in eye development and encodes an orphan nuclear receptor with two isoforms. Transient expression under the sevenless enhancer of either svp isoform leads to a dosage-dependent transformation of cone cells into R7 photoreceptors, and at a lower frequency, R7 cells into outer photoreceptors. To investigate the cellular pathways involved, we have taken advantage of the dosage sensitivity and screened for genes that modify this svp-induced phenotype. We show that an active Ras pathway is essential for the function of both Svp isoforms. Loss-of-function mutations in components of the Ras signal transduction cascade act as dominant suppressors of the cone cell transformation, whilst loss-of-function mutations in negative regulators of Ras-activity act as dominant enhancers. Furthermore, Svp-mediated transformation of cone cells to outer photoreceptors, reminiscent of its wildtype function in specifying R3/4 and R1/6 identity, requires an activated Ras pathway in the same cells, or alternatively dramatic increase in ectopic Svp protein levels. Our results indicate that svp is only fully functional in conjunction with activated Ras. Since we find that mutations in the Egfreceptor are also among the strongest suppressors of svpmediated cone cell transformation, we propose that the Ras activity in cone cells is due to low level Egfr signaling. Several models that could account for the observed svp regulation by the Ras pathway are discussed. Key words: neuron specification, eye development, cell-cell interactions, seven-up, Ras pathway INTRODUCTION Members of the steroid receptor superfamily regulate different aspects of development and differentiation in vertebrates and Drosophila (reviewed in Evans, 1988; Segraves, 1991). Although their biochemical function and molecular nature of action is well understood, much less is known about their functions during development. The least characterized group are the so-called orphan receptors, for which no ligands have yet been identified. Several members of this gene family have been characterized in Drosophila. With the exception of the Ecdysone receptor and Ultraspiracle proteins that form a heterodimer and bind ecdysone as ligand, all the Drosophila members are orphan receptors. Nevertheless most of them have been implicated in the ecdysone response (reviewed by Segraves, 1991). One exception is seven-up (svp), which is involved in neuronal specification in the embryonic central nervous system and in the eye (Mlodzik et al., 1990). Svp is part of a subclass of orphan receptors that includes the human proteins COUP and ARP-1 and several genes from zebrafish and other vertebrates (Wang et al., 1989; Ladias and Karathanasis, 1991; Fjose et al., 1993). All members of this subgroup share more than 90% amino acid identity in the DNAand putative ligand-binding domains. Biochemical studies suggest that these receptors are able to modify responses mediated by retinoic acid, other retinoids or even other ligands, most likely by competing for common DNA target sites (Kliewer et al., 1992; Tran et al., 1992). Interestingly, their requirement during nervous system development appears to have been conserved between Drosophila and vertebrates. Like svp in Drosophila, the zebrafish cognate of COUP and other vertebrate members of this subgroup are specifically expressed during nervous system and eye development (Fjose et al., 1993; Lutz et al., 1994). Nevertheless, due to the lack of identified ligands, regulation of these receptors is poorly characterized, in particular during developmental processes. The Drosophila compound eye provides a suitable model system for the analysis of mechanisms underlying neuronal cell fate specification. The eye consists of approximately 800 ommatidia, each of which contains 8 neuronal photoreceptor cells (R-cells), 4 lens-secreting non-neuronal cone cells and 8 accessory cells. During the development of the eye, short-range cell interactions guide ommatidial assembly via a series of inductive events whereby committed cells instruct their undetermined neighbors to adopt specific cell fates (reviewed in Wolff and Ready, 1993; Dickson and Hafen, 1993). The determination of the central photoreceptor R7 has been studied in most detail. Briefly, activation of the Sevenless (sev) receptor tyrosine kinase (RTK) via interaction with the Boss ligand presented on the neighboring R8 cell, leads to the activation of Ras which acts as the molecular switch between the

2 226 G. Begemann and others R7 cell and the non-neuronal cone cell fate (reviewed in Hafen et al., 1993; Zipursky and Rubin, 1994). The resulting kinase cascade ultimately leads to the phosphorylation of nuclear factors such as the ETS-domain proteins Pointed (pnt) and Yan (O Neill et al., 1994; Brunner et al., 1994a), as well as Jun (Bohmann et al., 1994). Sina encodes a nuclear protein that is also required for R7 development (Carthew and Rubin, 1990). The specification of the outer photoreceptors R1-6 is less well understood. However, the Ras pathway is also required for their development (Simon et al., 1991). It is thought that Ras is activated in these cells via a ligand-receptor interaction of the Drosophila EGF-receptor homologue, Egfr, and its putative ligand encoded by spitz (Xu and Rubin, 1993; Freeman, 1994). It is currently not clear how photoreceptor diversity is generated in response to Ras activation. Seven-up (svp) is expressed and required in a subset of outer photoreceptors, R3/4 and R1/6. It encodes two protein isoforms that differ in the C-terminal part of the putative ligand-binding domains. In svp mutant ommatidia, these four photoreceptors are transformed towards the fate of R7 (Mlodzik et al., 1990). Misexpression of either Svp isoform in any photoreceptor precursor (besides R1/6 and R3/4) or cone cells leads to aberrant ommatidial development suggesting that at least some genes involved in the Svp-mediated molecular pathway are also expressed in other cells within each ommatidium. Both isoforms display similar phenotypes in these experiments. In particular, ectopic Svp expression under the control of the sev-enhancer (sev-svp) causes a transformation of cone cells to R7 photoreceptors (Hiromi et al., 1993). Thus, although svp is required for outer photoreceptor development in wild type, both the loss-of-function and this gain-offunction phenotype of svp cause a switch in cell fates that leads to ectopic R7 development. This suggests that gene products required for normal Svp function (specification of outer photoreceptors) are limiting in cone cells. Nevertheless, Svp expression in cone cell precursors is sufficient to initiate neuronal development (i.e. induce R7) of these otherwise nonneuronal cells. Interestingly, sev-svp is phenotypically similar to the activation of the Ras pathway in cone cells (Basler et al., 1991; Fortini et al., 1992; Dickson et al., 1992; Brunner et al., 1994b), suggesting that similar mechanisms may be employed. To study the mechanism of svp function and to identify other components required for svp-induced cell fate commitment, we have attempted to screen systematically for genes that would affect the svp-mediated cone cell transformation. Here, we describe the identification of genes of the Ras pathway as essential components of the svp-mediated cell fate specification. All components of the Ras signal transduction cascade are required for Svp function acting as suppressors of the cone cell transformation. Moreover, negative regulators of Ras-activity act as dominant enhancers. Ras activity in cone cell precursors, normally without any phenotypic consequence, appears to be due to low level Egfr signaling. In addition, we show that ectopic activation of the Ras-pathway combined with Svp in cone cells leads to their transformation to outer photoreceptors, which is reminiscent of svp function in wild type. MATERIALS AND METHODS Fly strains and mutants The sev-svp1/2 strains are described in Hiromi et al. (1993). Mutant strains used in genetic interaction tests are all described in the respective references listed in the legend to Table 1. In these experiments, 2 sev-svp, a 3rd chromosome carrying two independent insertions, was combined with the respective mutations over wild type and reared at 25 C, where the 2 sev-svp phenotype is most stable. Histology Antibody stainings of eye imaginal discs and sections of adult retinae were performed as previously described (Tomlinson and Ready, 1987). For scanning electron microscopy, the eyes were prepared by critical point drying and coated with 2 nm of gold. Images were taken on a low-voltage SEM, stored digitally and processed with NIH- Image. The l(3)7842 enhancer detector insertion in svp providing nuclear β-galactosidase was used to monitor endogenous svp expression (Mlodzik et al., 1990). Enhancer detector insertion line H214 (Mlodzik et al., 1992) was used to detect R7 cells in 3rd instar discs. β-galactosidase was detected with a monoclonal antibody purchased from Promega. Generation of eye clones The svp e22 null allele was used to generate mutant eye clones with a P[w + ] at 90E as marker for mutant tissue. In an otherwise w background, clones were induced with X-rays (1000 rad) in transheterozygotes of the 3rd chromosomes mentioned above carrying either a wild-type 2nd chromsome as control or one with a sev-svp1/2 insertion. Clones were recovered with a frequency at 1/50 eyes. Tissue sections were examined with phase-contrast microscopy. RESULTS Dosage dependence of sev-svp induced cone cell transformation Expression of either Svp isoform under the control of the sevenhancer in R3/4, R1/6 and ectopically in R7, the mystery and cone cells causes a transformation of cone cell precursors to R7 neurons (Hiromi et al., 1993). The strength of the phenotype is dosage sensitive so that one copy of sev-svp2 causes a mild external eye roughening with on average less than 1 extra R7 per ommatidium (Fig. 1B,F), while flies carrying 2 sev-svp2 display rougher eyes and most ommatidia contain on average 2-3 extra R7 cells (Fig. 1C,G). Flies carrying two or more copies show an additional phenotypic feature: 10 to 20% of the ommatidia have a transformation of the original R7 precursor to an outer photoreceptor (Fig. 1G). Furthermore, with an increase in gene dosage, an increasing number of ommatidia have malformed and collapsed lenses (Figs 1C,D, 2A,B). A similar defect is observed upon expression of Rough in cone cells. Although Rough does not transform cone cells to photoreceptors, it causes differentiation defects leading to such collapsing lenses (Basler et al., 1990; Kimmel et al., 1990). A further increase in dosage (4 sev-svp2) causes a very severe external eye roughening. Ommatidia of such eyes contain multiple extra photoreceptor cells. Interestingly, many of the extra photoreceptors found in this genotype are of the outer R1-6 type, indicating that R7 and cone cell precursors are transformed to an R1-6 fate (Fig. 1D,H). In contrast to the lower doses that cause a cone cell to R7 transformation, this phenotypic effect corresponds to the wild-type svp function, specification of outer photoreceptors. However, due to the lack of markers, it is not possible to distinguish between the R2/5 and R3/4 or R1/6 cell fates.

3 Seven-up requires Ras activation 227 Fig. 1. Dosage sensitivity of sev-svp eye phenotypes. (A-D) SEM pictures (magnification: 100 ) and (E-H) sections of eyes of the respective genotypes (magnification: 600 ). (A,E) Wild type; R7 is indicated with arrowhead in one ommatidium. (B,F) 1 sev-svp2. Note slight external roughness in B and some ommatidia having an extra R7 (examples indicated with arrowheads). (C,G) 2 sev-svp2. Note that the eye is rougher than in B and often the lens is covered with black spots that are due to malformed lenses (see Fig. 3B for detail); almost all ommatidia contain additional R7 (examples indicated with arrowheads) and some contain 7 outer photoreceptors (indicated with arrows), due to an R7 to R1-6 transformation as judged from eye disc expression of R7 marker H214 (see also Fig. 4). (D,H) 4 sev-svp2. The eye is very rough with many malformed lenses (black spots) and most ommatidia contain additional outer photoreceptors (examples indicated with arrows). All eye sections shown here and following figures are at the R7 level, with R8 below the plane of section. Anterior is left, dorsal up. All phenotypic aspects are also observed in the equivalent sev-svp1 flies. However, the transformation frequency is slightly lower with type1, such that for example 3 sev-svp1 is phenotypically very similar to 2 sev-svp2 (Figs 1C,G, 3). Simultaneous expression of both protein isoforms has only an additive effect, e.g. 1 sev-svp1; 1 sev-svp2 look very similar to 2 sev-svp2 (or 3 sev-svp1), and 2 sev-svp1; 2 sev-svp2 is very similar to 4 sev-svp2 (not shown). This suggests that there is no synergism between the two isoforms and that the absolute protein levels are important. The quantitative and qualitative dosage sensitivity of the sev-svp phenotypes suggests that some products required for svp function are limiting in cone cells and can only be compensated for by strong overexpression of svp. Based on this dosage sensitivity, we suspected that mutations in genes that are rate limiting for the svp transformation of cone cells would act as dominant suppressors, whereas mutations in genes that repress svp activity might be dominant enhancers in this assay. If the assumption was correct, this genetic approach should allow the identification of other elements of svp-mediated cell fate specification. Such genetic modifiers could include genes that interact with Svp at the post-translational level or alternatively are transcriptional targets of svp. The 2-copy genotype appears suited for such a genetic approach, since it displays an intermediate phenotype and can be modified towards either wild-type or the 4-copy phenotype. Mutations in Ras pathway components are dominant suppressors of sev-svp It has previously been shown that the observed sev-svp-induced cell fate transformation is independent of boss and sev, but requires sina (Hiromi et al., 1993). To test the above hypothesis and to define further the genetic requirements for the svpmediated phenotype, we have tested many other genes known to be required for photoreceptor determination in the assay described above. Most of the genes that have been implicated in eye development do not interact with sev-svp (Table 1). However, mutations in components of the Ras signaling cascade required for R-cell development, drk, Sos, ras, raf and rl/mapk (reviewed in

4 228 G. Begemann and others Fig. 2. Genetic interactions with 2 sev-svp2. SEM pictures of 2 sev-svp2 eyes of the respective genotypes are shown. (A) 2 sev-svp in an otherwise wild-type background. (B) Higher magnification of a 2 sev-svp eye. Note malformed and collapsing lenses in several ommatidia (indicated with arrowheads) that are probably due to differentiation defects of cone cells. (C-H) 2 sev-svp in drk /+ (C); Sos /+ (D); ras1 /+ (E); rl /+ (F); yan /+ (G) and Egfr /+ (H); for alleles see Table 1. Note suppression of rough eye phenotype in C,D,E,F and H as compared to A, and enhancement in G. A very similar enhancement as shown in G is observed in 2 sev-svp2; Gap1 /+ eyes. Magnification: 100 in A and C-H; 600 in B; anterior is left and dorsal up. Hafen et al., 1993; Zipursky and Rubin, 1994), act as dominant suppressors of sev-svp1/2 (Figs 2, 3 and Table 1). Similarly, E(sev)1A, which has been shown to be required in all photoreceptors and possibly part of the Ras pathway (Simon et al., 1991), acts as a suppressor. Other genes that were found to be required for sevenless signaling, but do not seem to be part of the Ras pathway, Hsp83, cdc37and E(sev)3D (Simon et al., 1991), do not interact with sev-svp (Table 1). Moreover, when chromosomal deficiencies covering about 60% of the genome were crossed to the 2 sev-svp2 genotype, only a few deficiencies were found to interact with sev-svp, implying the existence of a small number of specific dominant suppressors and enhancers of this genotype and further demonstrating the specificity of the assay. Several deficiencies that are dominant suppressors of sev-svp remove Ras pathway genes (e.g. Df(3)by10 and Df(3R)by62 uncover Ras itself; Table 1). Interestingly, of the two phenotypic features displayed in 2 sev-svp2 eyes, cone cell to R7 transformation and R7 to R1-6 transformation (see above), only the cone cell transformation, but not the R7 to R1-6 transformation is suppressed by mutations in Ras pathway components (Fig. 3 and Table 2, see also Discussion). Negative regulators of Ras are enhancers of sev-svp Several genes have been identified as negative regulators of photoreceptor differentiation. Gap1, encoding a GTPase activating protein, is a negative regulator of Ras and is required to inhibit cone cell precursors from developing as R7 (Gaul et al., 1992). Yan encodes a nuclear Ets-related protein that inhibits photoreceptor determination and is downregulated by Ras (O Neill et al., 1994; Lai and Rubin, 1992). Both Gap1 and yan act as strong enhancers of sev-svp1/2. Both the external roughness, including the collapsing lens phenotype, and the number of transformed cone cells are enhanced (Figs 2, 3). The cone cell transformation is also affected qualitatively, such that cone cell precursors develop in part as outer photoreceptors (Fig. 3F). None of the other genes known to restrict photoreceptor determination to R-cell precursors (argos, fat facets; Freeman et al., 1992b; Fischer- Vize et al., 1992) displayed an interaction with sev-svp1/2 (Table 1). Thus mutations in antagonists of Ras activation act as dominant enhancers of sev-svp1/2. Taken together, these results imply that both Svp isoforms require Ras activity for their function in cone cell transformation.

5 Seven-up requires Ras activation 229 Table 1. Genetic interactions with sev-svp Tester-strain Tester-strain Genotype Reference 2 sev-svp1 2 sev-svp2 Comments Ras pathway: drk E(sev)2B /+ 1 Su Su 2 new alleles recovered in screen Sos e4g /+ 1, 2 Su Su Sos JC2 /+ 2 E E gain-of-function allele ras1 e2f /+ 1 Su Su 1 new allele recovered Df(3R)by10 [ras1 ]/+ Su Su Df(3R)by62 [ras1 ]/+ Su Su raf EA75 /+ 3 Su weak Su raf C110 /+ 3 weak Su weak Su rl/mapk EMS698 /+ 4 Su Su Df(2R)rl 10a /+ 4 Su Su rl/mapk Sem /+ 4 E E gain-of-function allele E(sev)1A e0p 1 /+ 1 Su Su allelic to corkscrew? Gap /+ 5 E E Gap1 ri533 /+ 5 E E Nuclear proteins: sina A16 /+ 6 sina P21 /+ 6 yan P /+ 7 E E 3 new alleles recovered in screen yan 1 /+ 7 E E pnt 88 /+ 8, 9 also member of Spitz-group se-jun bzip 10 dom. neg. protein in R- and cone cells phyl odin3g6 /+ 11 Su Su 1 allele recovered in screen phyl odin17l1 /+ 11 Su Su Df(2R)trix [phyl ]/+ Su Su ro 1 /+ 12 ro X63 /+ 13 svp e22 /+ 14 svp H162 /+ 14 gl 2 /+ 15 gl 3 /+ 15 Spitz-group: spi A14 /+ 16 rho rev37 /+ 17 n.d. Star KP347 /+ 18, 20 Su Su 3 new alleles recovered in screen Df(2L)ast2 [Star ]/+ Su Su Egfr flb3c81 /+ 19, 20 Su Su 5 new alleles recovered in screen Egfr top1 /+ 19, 20 weak Su weak Su Egfr ElpB1 /+ 19, 20 Neurogenic genes: N 55e11 /+ 21 Dl F10 /+ 22 mam 33 /+ 22 Df(3R)E(spl) RB251 /+ 22 deletes entire E(spl)-C Other eye genes: argos gild15 /+ 23 argos gila254d4 /+ 23 sine oculis 1 /+ 24 faf BX4 /+ 25 Hsp83 e1d /+ 1, 26 formerly called E(sev)3A cdc37 e1e /+ 1, 26 formerly called E(sev)3B E(sev)3D e0q /+ 1 Df(3R)pl 3 [atonal ]/+ 27 sca BP2 /+ 28 usp VE653 /+ 29 Genetic interactions between 2 sev-svp1/2 and other genes required during eye development and photoreceptor specification. In all interactions shown, the different genotypes were tested for dominant suppression or enhancement of the respective sev-svp phenotype as visible by the external eye morphology and in sections. Interactions are marked with Su for suppression, E for enhancement and for no detectable interaction (n.d.: not done). References where mutant eye phenotypes are described: (1) Simon et al., 1991; (2) Rogge et al., 1991; (3) Dickson et al., 1992; (4) Brunner et al., 1994a; (5) Gaul et al., 1992; (6) Carthew and Rubin, 1990; (7) Lai and Rubin, 1992; (8) Brunner et al., 1994b; (9) O Neill et al., 1994; (10) Bohmann et al., 1994; (11) Chang et al., Dickson et al., unpublished data; (12) Tomlinson et al., 1988; (13) Kimmel et al., 1990; (14) Mlodzik et al., 1990; (15) Moses et al., 1989; (16) Freeman, 1994; (17) Freeman et al., 1992a; (18) Heberlein et al., 1993; (19) Baker and Rubin, 1992; (20) Xu and Rubin, 1993; (21) Cagan and Ready, 1989; (22) Dietrich and Campos-Ortega, 1984; (23) Freeman et al., 1992b (24) Milani, 1941; (25) Fischer-Vize et al., 1992; (26) Cutforth and Rubin, 1994; (27) Jarman et al., 1994; (28) Baker et al., 1990; (29) Oro et al., 1992.

6 230 G. Begemann and others Fig. 3. Sections of 2 sev-svp1/2 eyes in suppressed and enhanced genotypes. (A-C) 2 sev-svp1; +/+ (A), 2 sev-svp1; ras1 /+ (B), 2 sev-svp1; Gap1 /+ (C). (D-H) 2 sev-svp2; +/+ (D), 2 sev-svp2; ras1 /+ (E), 2 sev-svp2; Gap1 /+ (F), 2 sev-svp2; Sos /+ (G), 2 sev-svp2; Egfr /+ (H). For reference of wild type, see Fig. 1D. Note that in panels B, E, G and H the presence of supernumerary R7 cells is suppressed and thus many ommatidia have wild-type appearance (indicated with arrowheads), but that the R7 to R1-6 transformation observed in 2 sev-svp2 is not suppressed leading to ommatidia with 7 outer R-cells and no R7(examples indicated with arrows). A similar R7 to R1-6 transformation is observed in 3 sev-svp1. (C,F) The enhancement due to gene dosage reduction in Gap1 (C,F) leads to all ommatidia having extra R-cells that in 2 sev-svp2 eyes (F) are often of the outer R1-6 type (indicated by triangle, compare to Fig. 4). Again 3 sev-svp1 is very similar to 2 sev-svp2 in this situation. Anterior is left. Fig. 4. Phenotypes of sev-svp combined with constitutively active Ras pathway components. (A-C) Eye sections of parental genotypes: (A) 2 sev-svp2, (B) sevs11, (C) rl/mapksevenmaker. (D-F) Genetic combinations of sev-svp2 with activated signaling proteins: (D) SosJC2 (gain-of-function allele of Sos); 2 sev-svp2, (E) sevs11; 2 sev-svp2, (F) rl/mapksevenmaker; 2 sev-svp2. (D-F) The cone cell transformation is qualitatively affected leading to supernumerary R1-6 cells and thus ommatidia with up to 11 R1-6 present (examples indicated with arrows, compare to parental genotypes in A-C). Note that frequency of R7 to R1-6 transformation is also increased. Combination of activated forms of Ras (sev-rasv12) and Raf (seraftory9) with sev-svp leads to the same effect as shown in E,F.

7 Seven-up requires Ras activation 231 Table 2. Mutations in Ras components suppress 2 sev-svp2 induced cone cell to R7 but not R7 to R1-6 transformation Ommatidia with Ommatidia with Total R7 R1-6 cone cell R7 ommatidia transformation transformation analyzed in Genotype (%) (%) 4 flies 2 sev-svp2; +/+ 15.3± ±4.4* sev-svp2; Sos /+ 13.1± ±4.1* sev-svp2; ras1 /+ 15.6± ±3.6* 651 *Note that the average number of ectopic R7 is not reflected in % of ommatidia with cone cell transformation. In 2 sev-svp2, there are on average 2.2 cone cells/ommatidium transformed as compared to the suppressed genotypes where only 1 extra R7 is found in almost all cases and thus only about 0.2 cone cells/ommatidium are affected. The numbers in the table were compiled by analyzing 4 independent eyes of the respective genotypes. The data were further confirmed by the analysis of the R7 marker H214 in developing eye imaginal discs. Interaction with nuclear proteins required for photoreceptor determination Besides svp there are several nuclear factors required as positive regulators in photoreceptor development. Some are required in all R-cells, like glass (Moses et al., 1989), others, like svp, in specific subtypes, e.g. rough in R2/5 (Tomlinson et al., 1988), and sina in R7 (Carthew and Rubin, 1990). Furthermore, pointed (pnt) (Brunner et al., 1994a; O Neill et al., 1994) and d-jun (Bohmann et al., 1994) have been implicated as nuclear phosphorylation targets of Ras activation, either in a subset or in all photoreceptor precursors. Of all the nuclear proteins only phyllopod (phyl; Chang et al., unpublished data; Dickson et al., unpublished data) shows a dosage-dependent interaction with sev-svp1/2 and acts as a suppressor (Table 1; see Discussion). It is worthwhile to note that, while sina is required for the svp mediated cone cell transformation (Hiromi et al., 1993), no dosage-sensitive effect is observed. Ras-activity in cone cells is due to low level Egfrsignaling To identify other genes interacting with sev-svp, we have screened ~50,000 genomes mutagenized with EMS and about 2300 lethal P-insertions (Török et al., 1993). Besides having identified a few novel loci that interact specifically with sevsvp (our unpublished results), we recovered additional mutant alleles in drk, ras1, yan and phyl (Table 1). Furthermore, several alleles of the Drosophila EGF-receptor tyrosine kinase gene (Egfr) were isolated. All new Egfr alleles are strong suppressors of sev-svp1/2 and constitute the largest complementation group (Figs 2, 3 and Table 1). Previously known alleles of Egfr, such as the strong allele Egfr flb3c81 or the viable hypomorphic allele Egfr top1, show the same or slightly weaker suppression of sev-svp1/2, respectively (Table 1). Another complementation group recovered as strong suppressors corresponds to Star (3 alleles; Table 1). All Star alleles, as well as deficiencies uncovering Star, show a very similar suppression (Table 1). Star encodes a transmembrane protein and appears to serve an accessory function in Egfr signaling (Heberlein et al., 1993; Freeman, 1994; Kolodkin et al., 1994). These results prompted us to test whether other activating components of Egfr signaling would interact with sev-svp. The spitz (spi) and rhomboid (rho) genes have been implicated in Fig. 5. Transient expression of Svp1 or 2 is sufficient to partially rescue the svp phenotype. All panels show homozygous mutant clones of svp e22 (null allele) marked by the lack of pigment. (A) svp clone. Note that mutant ommatidia contain up to 5 R7s, but only 3-4 outer photoreceptors (examples indicated by arrows). (B) svp clone; 1 sev-svp1. (C) svp clone; 1 sev-svp2. Note partial rescue in B,C, documented by the presence of ommatidia that have 6 outer photoreceptors and a reduced number of R7 as compared to A (examples indicated by arrows) and mosaic ommatidia with wildtype appearance lacking endogenous svp in R-cells that require its function (indicated by arrowheads to respective R3/4 or R1/6 cells). Some ommatidia in B-C contain ectopic R7 cells due to sev-svp induced cone cell transformation. To keep this effect low and better visualize the rescue, we have used transformants with a low cone cell transformation frequency and probably also lower than wild-type expression levels of Svp.

8 232 G. Begemann and others Egfr signaling as the putative ligand and a transmembrane protein that is required for ligand induced activation, respectively (Bier et al., 1990; Rutledge et al., 1992; Sturtevant et al., 1993; Freeman, 1994). However, neither spi nor rho display an interaction with sev-svp1/2 (Table 1). Ellipse, proposed to be a ligand-dependent gain-of-function allele of Egfr (Baker and Rubin, 1992; Freeman, 1994), does not affect sev-svp (Table 1). These results indicate that Egfr plays an important role in the sev-svp-mediated phenotype and, since the strength of its suppression is very similar to components of the Ras pathway, it may cause the Ras activity present in cone cells (see Discussion). Activated Ras-pathway components combined with Svp in cone cells lead to a transformation to outer photoreceptors The above results indicate that svp requires the Ras pathway in cone cell precursors. Nevertheless, it does not seem to fulfil its wild-type function, specification of R3/4 and R1/6 (Mlodzik et al., 1990). Instead, when expressed in cone cells (1-2 sevsvp), Svp induces the R7 fate. Only a high dosage (4 sev-svp) transforms cone cells to R1-6. However, when the dosage of Gap1, a negative regulator of Ras, is reduced, resulting in a putative increase in Ras activity, 2 sev-svp2 is at least partially capable of causing cone cell to R1-6 transformation (Fig. 3F). To test if a qualitative change in svp activity can be obtained in cells where components of the Ras signal transduction cascade are activated, we combined 2 sev-svp with Sos JC2, a gain-of-function allele in the Ras activating guanine nucleotide exchange factor, that is normally without phenotypic consequence, but leads to enhanced signaling in sensitized backgrounds (Rogge et al., 1991). Sos JC2 strongly enhances sev-svp leading to ommatidia with many extra outer photoreceptors (Fig. 4D). Similar experiments were conducted using flies expressing activated forms of either Sev, Ras or Raf in cone cells, sev S11 (Basler et al., 1991), sev-ras V12 (Fortini et al., 1992) and se-raf tory9 (Dickson et al., 1992), respectively, or the constitutive gain-of-function allele of rl/mapk, Sevenmaker (Sem; Brunner et al., 1994b). When these genotypes are combined with sev-svp, a strong enhancement of either rough eye phenotype is observed, which is similar in external morphology to 4 sev-svp (not shown). All ommatidia in such eyes contain extra photoreceptors and cone cells are now transformed to outer R1-6 cells giving rise to ommatidia with up to 11 outer photoreceptors (Fig. 4). This transformation is reminiscent of the function of svp in wild type and indicates that, when the Ras pathway is activated, lower levels of Svp are sufficient to specify outer photoreceptors. Transient expression of Svp is sufficient for its function In wild-type eye development, svp is expressed in R3/4 and R1/6 from its first appearance behind the morphogenetic furrow, throughout the 3rd instar imaginal eye disc as determined by in situ hybridization and the detection of β-galactosidase in svp enhancer trap alleles (Mlodzik et al., 1990). Also, as revealed in the enhancer detectors, it is still expressed in R3/4 in the adult retina (unpublished). Sev-enhancer driven Svp expression is, however, transient and restricted to only a few columns of ommatidial assembly. Fig. 6. svp expression in 2 sev-svp eye discs. 3rd instar eye imaginal discs showing β-galactosidase expression in a svp enhancer detector line in wild-type (A) and 2 sev-svp2 background (B). No ectopic endogenous svp expression and autoregulation are detected in B. 2 sev-svp1 or any combination of sev-svp1/2 behave the same. The disorder and varying nuclear positions in B are due to cone cell transformation and development of ectopic R-cells. The morphogenetic furrow is indicated with arrowheads and svp expressing photoreceptors are numbered according to their identity in one cluster in each panel. Anterior is left. Note that the equator is located between A and B so that A shows dorsal and B ventral part of disc, respectively. Nevertheless, such transient expression of Svp is sufficient, in conjunction with an activated Ras pathway in the same cell, to transform cone cell precursors to outer photoreceptors. To test whether transient expression is also sufficient for svp function in R3/4 and R1/6 precursor determination, we have induced svp clones in a 1 sev-svp1/2 background. We find that the sevenhancer driven transient expression of either Svp isoform is sufficient to partially rescue the svp eye phenotype in homozygous clones (Fig. 5). Moreover, either Svp isoform can rescue either cell type (R3/4 and R1/6) that requires svp function in wild type (Fig. 5). To test whether transient expression of Svp can activate endogenous svp via a mechanism of autoregulation, we have monitored the expression of svp enhancer trap insertions in sevsvp1/2 flies. Neither Svp1 nor Svp2 activate transcription of the endogenous gene, even when multiple copies of the respective constructs are present (Fig. 6). Similarly, ectopic expression of Svp1/2 via a heat-shock promoter does not cause activation of the endogenous gene during imaginal development (not shown). Taken together, these results suggest that neither Svp

9 Seven-up requires Ras activation 233 isoform has an autoregulatory feature and that svp is only required for a restricted period of time during early stages of ommatidial assembly for photoreceptor determination. DISCUSSION Cone cell precursors can be induced to differentiate as R7 neurons by expressing constitutively active Sev, Ras or other components of its pathway in these cells (Basler et al., 1991; Fortini et al., 1992; Dickson et al., 1992; Brunner et al., 1994b). Similarly, expression of Seven-up (Svp) in cone cells (sev-svp) is sufficient to induce their neuronal differentiation as R7. However, this phenotype is in stark contrast to the wild-type function of svp, specification of R3/4 and R1/6. Thus, in the cone cell environment Svp does not function as in wild type, suggesting that some components of its pathway are limiting. Interestingly, this functional deficit may be overcome by an increase in sev-svp dosage: 4 sev-svp does not only cause an increase in cone cell transformation, but also affects qualitatively the neuronal cell type, e.g. 4 copies of sev-svp cause cone cell to R1-6 transformation. Genetic interactions between sev-svp and the Ras pathway To define the mechanisms underlying the molecular action of svp, we have investigated the genetic interactions of the dosage-sensitive Svp-dependent cone cell transformation phenotype. In this assay, the cone cell transformation is quantitatively and qualitatively affected by the strength of Ras signaling. Thus, Svp and activated Ras not only have similar effects on cone cell precursors, but also mutations in components of the Ras pathway act as dominant suppressors of sevsvp. Gene dosage reduction of Ras pathway components almost completely abolishes the sev-svp-mediated cone cell transformation. Such dosage-sensitive interactions suggest that the respective genes function in a common genetic pathway. This effect appears very specific, since we find only a few loci displaying an interaction with sev-svp. Only components of Ras and Egfr signaling pathways display such dosage-sensitive interactions with sev-svp (Table 1). In contrast, increased Ras-pathway activity generated by gene dosage reduction in Gap1, a negative regulator of Ras, or yan enhances the sev-svp phenotype. Yan is a negative regulator of photoreceptor determination encoding a nuclear protein of the Ets-family that is downregulated following activation of Ras (O Neill et al., 1994). Since hypomorphic yan alleles are enhanced or suppressed by gene dosage reduction in Gap1 and ras1, respectively (Lai and Rubin, 1992), the interaction observed between sev-svp and yan is comparable to the one detected with Gap1. Interestingly, such an enhancement does not only affect transformation frequency, but also qualitatively the cell type induced so that many of the ectopic photoreceptors are of the outer R1-6 type. Moreover, when 2 sev-svp is combined with the gain-of-function allele Sos JC2 or activated components of the Sev/Ras pathway, it induces cone cell to outer photoreceptor transformation. This effect is reminiscent of the svp wild-type function, specification of R3/4 and R1/6. Taken together, these results indicate that Svp requires an activated Ras pathway in the same cell to perform its normal function. The dosage-sensitive interactions with Ras pathway components are only observed in cone cell precursors where Ras has not been activated and signaling is at a low level. The Svpmediated transformation of R7 to R1-6 is not sensitive to dosage reduction in these genes (Fig. 4; Table 2) indicating that, upon Ras activation, e.g. via the Boss-Sev interaction (Hafen et al., 1993; Zipursky and Rubin, 1994), components of its pathway are not limiting for Svp function. Nevertheless, the frequency of Svp-induced R7 to R1-6 transformation is rather low (<20%), suggesting that in R7 other gene products necessary for svp function are rate limiting. Alternatively, R7 may contain an inhibitory mechanism for svp function, thus keeping a large fraction of these cells committed to the R7 developmental program. However, Rough-mediated R7 to R2/5 transformation is (almost) fully penetrant (Basler et al., 1990; Kimmel et al., 1990) and thus, if such an inhibitory mechanism exists, it cannot interfere with the Rough-mediated transformation. In this context, it is worth noting that in a sev background it is easier to induce a neuronal fate in cone cells than in the endogenous R7 precursor as judged from the phenotypes of sev; Gap1 (Gaul et al., 1992) or sev; sev-svp (Hiromi et al., 1993) double mutants. Thus the fate of the R7 precursor appears to be under a tighter control than the cone cells. Low level Ras activity in cone cells is due to Egfr signaling Several lines of evidence suggest that there is a low level activity of Ras in cone cell precursors which in a wild-type situation is without consequence for cell fate specification. For example, homozygous mutants in Gap1, a negative regulator of Ras, cause a transformation of cone cell precursors to R7 cells (Gaul et al., 1992), indicating that even without expression of activated forms of Ras pathway components (Basler et al., 1991; Fortini et al., 1992; Dickson et al., 1992; Brunner et al., 1994b) there is considerable Ras activity in these cells. Both Gap1 and the nuclear antagonist Yan must be present in cone cells to prevent their differentiation as R7 (Gaul et al., 1992; Lai and Rubin, 1992). The phenotypic effect of Svp expression in these cells, transformation to R7, also depends on this Ras activity. Indeed, mutants in both genes act as dominant enhancers of the Svp-mediated cone cell transformation, probably by causing higher levels of activated Ras (Gap1) or reducing the threshold (yan). Our results suggest that Egfr is a likely candidate for this low level Ras activity in cone cells. The dominant suppression observed with loss-of-function Egfr alleles is indistinguishable from the one detected with Ras pathway components. The gene encoding the putative Egfr ligand, spitz (Rutledge et al., 1992), does not display any interaction with sev-svp, nor does rhomboid, a transmembrane protein that appears to be required to restrict ligand-dependent Egfr activation to specific cells throughout development (Bier et al., 1990; Freeman et al., 1992a; Sturtevant et al., 1993; Freeman, 1994). Star acts like Egfr as a dominant suppressor in this assay. It also encodes a transmembrane protein and has been implicated with an accessory role, but it is unknown how its function relates to Egfr signaling (Sturtevant et al., 1993; Freeman, 1994). Moreover, Star has also been implicated in sev signaling (Kolodkin et al., 1994) and thus could serve a more general function in RTK signaling. Since Egfr activates Ras, we propose that the low level Ras activity in cone cells is

10 234 G. Begemann and others Svp Rl/MAPK activated Ras pathway P + P Svp P Svp target genes Ras target Svp Fig. 7. Models for Ras-dependent Svp activation. Note that the downregulated repressor in scenario (2) could interact with any of the other mechanisms shown. Any combination of the four scenarios is possible. See text for details. Egfr dependent. It is not clear if a Egfr ligand (spi or other) is required. Ras activation is required for Svp function The results discussed above indicate that Ras activity is necessary for Svp function. However, the genetic interactions do not reveal the nature of the mechanisms employed. Thus, several scenarios (or a combination thereof) could account for this effect (Fig. 7). (1) A co-factor/partner of Svp is modified, possibly phosphorylated by Rl/MAPK, (2) a negative regulator is repressed by such a modification, (3) Svp itself is target of Ras-dependent activation and (4) a Ras-dependent transcriptional upregulation of genes that are required together with Svp in photoreceptor specification. Of the nuclear phosphorylation targets of Ras activation implicated in photoreceptor determination, pointed, yan (Brunner et al., 1994a; O Neill et al., 1994) and Jun (Bohmann et al., 1994), only yan displays a dosage-sensitive effect in this assay. However, the lack of an interaction with pointed and jun does not exclude their requirement nor can the presence of other such factors be excluded. The third scenario, that Svp itself is a target of Ras induced modification(s), e.g. phosphorylation, cannot be excluded. Although neither Svp isoform contains a perfect MAPK target site (PXS/TP), the C termini of both isoforms are serine/threonine rich with several potential lower affinity target sites (S/TP). In particular, Svp2 contains 9 such sites. These motifs are located to the isoform-specific parts, which have been shown to be required for Svp function in sev-svp or hssvp flies (Hiromi et al., 1993). Phosphorylation has been demonstrated for the human homologue COUP: it was found that dopamine can increase transcriptional activity of COUP via a second messenger pathway and phosphorylation in its putative ligand binding domain (Power et al., 1991). To date there is one candidate for the fourth scenario, Rasdependent transcriptional activation of genes required in photoreceptor development. The recently isolated phyllopod (phyl) gene is a dominant suppressor of activated Ras and Raf and it also suppresses sev-svp. It is expressed and required in R1/6 and R7. However, its overexpression in cone cells leads to their transformation to R7, suggesting that it may be a transcriptional target of activated Ras (Chang et al., unpublished data; Dickson et al., unpublished data). Since phyl is also required in R1/6, there is an overlap with svp in this cell type and thus Phyl could qualify as a candidate for a Svp partner. The observation that 4 sev-svp can compensate for the lack of Ras activation might favour the second scenario, in which a negative regulator is repressed by Ras activation, since high overexpression of Svp could compete with a putative repressor to overcome its action. However, a biochemical analysis will be required to distinguish between the different scenarios. In summary, our results demonstrate that (1) Svp requires Ras activity in the same cell for its normal function, (2) transient expression of Svp appears sufficient, (3) both isoforms appear to have the same function and interact genetically with the same genes and (4) either isoform appears largely sufficient to perform the wild-type svp function as also judged from the svp e22 ; sev-svp rescue. We are grateful to Ursula Weber for help with the deficiency screen, Yasushi Hiromi for stimulating discussions and sharing unpublished results, and Tibor Török, Gabriella Tick and Istvan Kiss for allowing us to screen their collection of lethal P-element insertions. We thank Henry Chang, Barry Dickson, Ernst Hafen, Gerry Rubin and Tanya Wolff for sharing unpublished results, Utpal Banerjee, Barry Dickson, Janice Fischer-Vize, Matthew Freeman, Ulrike Gaul, Ernst Hafen, Christian Klämbt, Kevin Moses, Norbert Perrimon, Gerald Rubin, Stefan Schneuwly and Michael Simon for fly strains, and Steve Cohen, David Strutt, Ursula Weber, Volker Wiersdorff and Martin Zeidler for helpful comments on the manuscript. G. B. was supported in part by a fellowship from the Fonds der Chemischen Industrie and L. vdv. by a long-term EMBO fellowship. REFERENCES Baker, N. E., Mlodzik, M. and Rubin, G. M. (1990) Spacing differentiation in the developing Drosophila eye: a fibrinogen-related lateral inhibitor encoded by scabrous. Science 250, Baker, N. E. and Rubin, G. M. (1992) Ellipse mutations in the Drosophila homologue of the EGF receptor affect pattern formation, cell division and cell death in eye imaginal discs. Dev. Biol. 150, Basler, K., Christen, B. and Hafen, E. (1991) Ligand-independent activation of the sevenless receptor tyrosine kinase changes the fate of cells in the developing Drosophila eye. Cell 64, Basler, K., Yen, D., Tomlinson, A. and Hafen, E. (1990) Reprogramming cell fate in the developing Drosophila retina: transformation of R7 cells by ectopic expression of rough. Genes Dev. 4, Bier, E., Jan, L. Y. and Jan, Y. N. (1990) rhomboid, a gene required for the dorsoventral axis establishment and peripheral nervous system development in Drosophila melanogaster. Genes Dev. 4, Bohmann, D., Ellis, M. C., Staszewski, L. M. and Mlodzik, M. (1994) Drosophila Jun mediates Ras-dependent photoreceptor determination. Cell 78, Brunner, D., Dücker, K., Oellers, N., Hafen, E., Scholz, H. and Klämbt, C. (1994a) The ETS domain protein Pointed-P2 is a target of MAP kinase in the Sevenless signal transduction pathway. Nature 370, Brunner, D., Oellers, N., Szabad, J., Biggs, W. H., III, Zipursky, S. L. and Hafen, E. (1994b) A gain of function mutation in Drosophila MAP kinase activates multiple receptor tyrosine kinase signaling pathways. Cell 76,

11 Seven-up requires Ras activation 235 Cagan, R. L. and Ready, D. F. (1989) Notch is required for successive cell decisions in the developing Drosophila retina. Genes Dev. 3, Carthew, R. W. and Rubin, G. M. (1990) seven in absentia, a gene required for specification of R7 cell fate in the Drosophila eye. Cell 63, Cutforth, T. and Rubin, G. M. (1994) Mutations in Hsp83 and cdc37 impair signaling by the Sevenless receptor tyrosine kinase in Drosophila. Cell 77, Dickson, B. and Hafen, E. (1993). Genetic dissection of eye development in Drosophila. In The Development of Drosophila melanogaster (ed. M. Bate and A. Martinez-Arias), pp Cold Spring Harbor: Cold Spring Harbor Laboratory Press. Dickson, B., Sprenger, F., Morrison, D. and Hafen, E. (1992) Raf functions downstream of Ras1 in the sevenless signal transduction pathway. Nature 360, Dietrich, U. and Campos-Ortega, J. A. (1984) The expression of neurogenic loci in imaginal epidermal cells of Drosophila melanogaster. J. Neurogenet. 1, Evans, R. M. (1988) The steroid and thyroid hormone receptor superfamily. Science 240, Fischbach, K. F., Barleben, F., Boschert, U., Dittrich, A. P. M., Gschwander, B., Hoube, B., Ramos, R. G. P. and Schlosser, G. (1989). Developmental studies on the optic lobe of Drosophila using structural brain mutants. In Neurobiology of Sensory Systems (ed. R. N. Singh and N. J. Strausfeld), pp New York: Plenum Press. Fischer-Vize, J. A., Rubin, G. M. and Lehmann, R. (1992) The fat facets gene is required for Drosophila eye and embryo development. Development 116, Fjose, A., Nornes, S., Weber, U. and Mlodzik, M. (1993) Functional conservation of vertebrate seven-up related genes in neurogenesis and eye development. EMBO J. 12, Fortini, M. E., Simon, M. A. and Rubin, G. M. (1992) Signalling by the sevenless protein tyrosine kinase is mimicked by Ras1 activation. Nature 355, Freeman, M., Kimmel, B. E. and Rubin, G. M. (1992a) Identifying targets of the rough homeobox gene of Drosophila: evidence that rhomboid functions in eye development. Development 116, Freeman, M., Klämbt, C., Goodman, C. S. and Rubin, G. M. (1992b) The argos gene encodes a diffusible factor that regulates cell fate decisions in the Drosophila eye. Cell 69, Freeman, M. (1994) The spitz gene is required for photoreceptor determination in the Drosophila eye where it interacts with the EGF receptor. Mech. Development (in press). Gaul, U., Mardon, G. and Rubin, G. M. (1992) A putative Ras GTPase activating protein acts as a negative regulator of signaling by the sevenless receptor tyrosine kinase. Cell 68, Hafen, E., Dickson, B., Brunner, T. and Raabe, T. (1993) Genetic dissection of signal transduction mediated by the sevenless receptor tyrosine kinase in Drosophila. Phil. Trans. R. Soc. Lond. B. 340, Heberlein, U., Hariharan, I. and Rubin, G. M. (1993) Star is required for neuronal differentiation in the Drosophila retina and displays dosage sensitive interactions with Ras1. Dev. Biol. 160, Hiromi, Y., Mlodzik, M., West, S. R., Rubin, G. M. and Goodman, C. S. (1993) Ectopic expression of seven-up causes cell fate changes during ommatidial assembly. Development 118, Jarman, A. P., Grell, E. H., Ackermann, L., Jan, L. Y. and Jan, Y. N. (1994) atonal is the proneural gene for Drosophila photoreceptors. Nature 369, Kimmel, B. E., Heberlein, U. and Rubin, G. M. (1990) The homeo domain protein rough is expressed in a subset of cells in the developing Drosophila eye where it can specify photoreceptor cell subtype. Genes Dev. 4, Kliewer, S. A., Umesono, K., Heyman, R. A., Mangelsdorf, D. J., Dyck, J. A. and Evans, R. M. (1992) Retinoid X receptor-coup-tf interactions modulate retinoic acid signaling. Proc. Natl. Acad. Sci. USA 89, Kolodkin, A. L., Pickup, A. T., Lin, D. M., Goodman, C. S. and Banerjee, U. (1994) Characterization of Star and its interactions with sevenless and EGF receptor during photoreceptor cell development in Drosophila. Development 120, Ladias, J. A. A. and Karathanasis, S. K. (1991) Regulation of the apolipoprotein AI gene by ARP-1, a novel member of the steroid receptor superfamily. Science 251, Lai, Z.-C. and Rubin, G. M. (1992) Negative control of photoreceptor development in Drosophila by the product of the yan gene, an ets domain protein. Cell 70, Lutz, B., Kuratani, S., Cooney, A. J., Wawersik, S., Tsai, S. Y., Eichele, G. and Tsai, M.-J. (1994) Developmental regulation of the orphan receptor COUP-TFII gene in spinal motor neurons. Development 120, Mlodzik, M., Hiromi, Y., Weber, U., Goodman, C. S. and Rubin, G. M. (1990) The Drosophila seven-up gene, a member of the steroid receptor gene superfamily, controls photoreceptor cell fates. Cell 60, Mlodzik, M., Hiromi, Y., Goodman, C. S. and Rubin, G. M. (1992) The presumptive R7 cell of the developing Drosophila eye receives positional information independent of sevenless, boss and sina. Mech. Dev. 37, Moses, K., Ellis, M. C. and Rubin, G. M. (1989) The glass gene encodes a zinc-finger protein required by Drosophila photoreceptor cell. Nature 340, O Neill, E. M., Rebay, I., Tjian, R. and Rubin, G. M. (1994) The activities of two Ets-related transcription factors required for Drosophila eye development are modulated by the Ras/MAPK pathway. Cell 78, Oro, A. E., McKeown, M. and Evans, R. M. (1992) The Drosophila retinoid X receptor homolog ultraspiracle functions in both female reproduction and eye morphogenesis. Development 115, Power, R. F., Lydon, J. P., Conneely, O. M. and O Malley, B. W. (1991) Dopamine activation of an orphan of the steroid receptor superfamily. Science 252, Rogge, R. D., Karlovich, C. A. and Banerjee, U. (1991) Genetic dissection of a neurodevelopmental pathway: son of sevenless functions downstream of the sevenless and EGF receptor tyrosine kinases. Cell 64, Rutledge, B. J., Zhang, K., Bier, E., Jan, Y. N. and Perrimon, N. (1992) The Drosophila spitz gene encodes a putative EGF-like growth factor involved in dorsal-ventral axis formation and neurogenesis. Genes Dev. 6, Segraves, W. A. (1991) Something old, some things new: The steroid receptor superfamily in Drosophila. Cell Simon, M. A., Bowtell, D. D. L., Dodson, G. S., Laverty, T. R. and Rubin, G. M. (1991) Ras1 and a putative guanine nucleotide exchange factor perform crucial steps in the signaling by the sevenless protein tyrosine kinase. Cell 67, Sturtevant, M. A., Roark, M. and Bier, E. (1993) The Drosophila rhomboid gene mediates the localized formation of wing veins and interacts genetically with components of the EGF-R signaling pathway. Genes Dev. 7, Tomlinson, A. and Ready, D. F. (1987) Neuronal differentiation in the Drosophila ommatidium. Dev. Biol. 120, Tomlinson, A., Kimmel, B. E. and Rubin, G. M. (1988) rough, a Drosophila homeobox gene required in photoreceptors R2 and R5 for inductive interactions in the developing eye. Cell 55, Török, T., Tick, G., Alvarado, M. and Kiss, I. (1993) P-lacW insertional mutagenesis on the second chromosome of Drosophila melanogaster: isolation of lethals with different overgrowth phenotypes. Genetics 135, Tran, P., Zhang, X.-K., Salbert, G., Hermann, T., Lehmann, J. M. and Pfahl, M. (1992) COUP orphan receptors are negative regulators of retinoic acid response pathways. Mol. Cell Biol. 12, Wang, L.-H., Tsai, S. Y., Cook, R. G., Beattie, W. G., Tsai, M.-J. and O Malley, B. W. (1989) COUP transcription factor is a member of the steroid receptor superfamily. Nature 340, Wolff, T. and Ready, D. F. (1993). Pattern formation in the Drosophila retina. In The Development of Drosophila melanogaster (ed. M. Bate and A. Martinez-Arias), pp Cold Spring Harbor: Cold Harbor Laboratory Press. Xu, T. and Rubin, G. M. (1993) Analysis of genetic mosaics in developing and adult Drosophila tissues. Development 117, Zipursky, S. L. and Rubin, G. M. (1994) Determination of neuronal cell fate: lessons from the R7 neuron of Drosophila. Annu. Rev. Neurosci. 17, (Accepted 23 September 1994)