¹Department of Neuroanatomy, Biomedical Research Center, Osaka University Medical School, Suita, Osaka 565-0871, Japan

²Department of Biochemistry, Hiroshima University School of Medicine, 1-2-3 Kasumi, Minami-ku, Hiroshima, Hiroshima 734-0037, Japan

³CREST, Japan Science and Technology Corporation (JST), 2-6-15 Shiba Park, Minato-ku, Tokyo 105-0011, Japan

^aAuthor for correspondence

The small GTP-binding protein Ral is activated by RalGDS, one of the effector molecules for Ras. Active Ral binds to a GTPase activating protein for CDC42 and Rac. Although previous studies suggest a role for Ral in the regulation of CDC42 and Rac, which are involved in arranging the cytoskeleton, its in vivo function is largely unknown. To examine the effect of overexpressing Ral on development, transgenic Drosophila were generated that overexpress wild-type or mutated Ral during eye development. While wild-type Ral caused no developmental defects, expression of a constitutively activated protein resulted in a rough eye phenotype. Activated Ral did not affect cell fate determination in the larval eye discs but caused severe disruption of the ommatidial organization later in pupal development. Phalloidin staining showed that activated Ral perturbed the cytoskeletal structure and cell shape changes during pupal development. This phenotype is similar to that caused by RhoA overexpression. In addition, the phenotype was synergistically enhanced by the coexpression of RhoA. These results suggest that Ral functions to control the cytoskeletal structure required for cell shape changes during Drosophila development.

actin; cell shape; Drosophila; Ral

Introduction

Ral is a member of a family of Ras-like small GTPase proteins (reviewed by Feig et al., 1996). Like all other Ras-like GTPases (Bos, 1997), Ral cycles between GTP-bound active and GDP-bound inactive forms. Replacement of GDP with GTP, which is required for the activation of Ral, is promoted by at least three guanine-nucleotide-exchange factors (GEFs): RalGDS (Albright et al., 1993), RGL (Kikuchi et al., 1994; Murai et al., 1997), and RLF (Wolthuis et al., 1996). Interestingly, all these RalGEFs interact with the active form of Ras (Kikuchi et al., 1994; Hofer et al., 1994; Spaargaren and Bischoff, 1994). In addition, stimulation of cells with insulin or epidermal growth factor results in an increase in Ral-GTP through the activation of Ras (Kishida et al., 1997; Wolthuis et al., 1998). These findings suggest that activated Ral, which results the interaction of Ras with RalGEFs, mediates some of the downstream signaling from activated Ras. Indeed, RalGEFs and Ral have been implicated in Ras-induced DNA synthesis, gene expression, and oncogenic transformation (Urano et al., 1996; White et al., 1996; Okazaki et al., 1997; Miller et al., 1997; Wolthuis et al., 1997).

Recently, putative downstream targets for Ral have been identified. The binding of Ral to phospholipase D (PLD) is required for the activation of PLD downstream of activated Ras (Jiang et al., 1995). In addition, RalBP1 (also called RLIP or RIP1) binds to the effector domain of the active Ral protein (Cantor et al., 1995; Jullien-Flores, 1995; Park et al., 1995). RalBP1 contains a RhoGAP domain and acts as GTPase activating protein (GAP) for CDC42 and Rac, small GTP-binding proteins that regulate the actin cytoskeleton (Cantor et al., 1995; Jullien-Flores, 1995; Park et al., 1995). Thus, it is possible that RalGEFs and Ral function downstream of Ras to regulate the actin cytoskeleton through modification of the activities of CDC42 and/or Rac. However, the function of Ral in cytoskeletal regulation has not been determined.

During the development of multicellular organisms, proper regulation of the actin cytoskeleton is essential for precise cell shape changes. Members of the Rho family of small GTPases have been implicated in such processes (reviewed by Aelst and D'Souza-Schorey, 1997). The fruit fly, Drosophila melanogaster, provides a unique system to investigate the roles of the small GTP-binding proteins in development. The Drosophila homologues of RhoA, Rac1, Rac2 and CDC42 have been identified and implicated in several developmental processes including eye morphogenesis, axonal outgrowth, dorsal closure and hair development (Luo et al., 1994; Hariharan et al., 1995; Harden et al., 1995; Eaton et al., 1996; Barret et al., 1997; Strutt et al., 1997).

To analyse the in vivo function of Ral in animal development, we took advantage of the genetic dissections of Drosophila compound eyes, described as follows. A stereotyped, three dimensional arrangement of cells in each ommatidium is formed through regulated cell shape changes during pupal development (Longley and Ready, 1995), providing us with an excellent model system to study the mechanisms that control cell shape. Moreover, signal transduction mechanisms, including the Ras pathway, involved in the development of the Drosophila eye have been extensively characterized using genetic strategies (Wassarman and Therrien, 1997). In this study, we generated transgenic flies in which activated Ral expression was induced in the eye, and used these strains to examine Ral's effect on cell shape changes and the actin cytoskeleton.

Results

Expression of Ral^G23V disrupts normal eye development

To target the expression of Ral in developing eyes, transgenic strains carrying a cDNA encoding wild-type Ral or the Ral^G23V mutant (constitutively-activated form; Frech et al., 1990) in the transformation vector pGMR were established. Two GMR-Ral and four GMR-Ral^G23V lines were established. We then examined the external morphology and internal structure of the compound eyes from each line. The compound eyes of flies carrying one or two copies of GMR-Ral were morphologically indistinguishable from the wild-type eyes (data not shown). On the other hand, each of the GMR-Ral^G23V strains displayed eyes with abnormal external morphology (Figure 1b c) compared with the regular array of ommatidia in the wild-type eye (Figure 1a). The effect of Ral^G23V on eye morphology appeared to be dependent on the gene dosage, since the phenotype of flies with two copies of the transgenes (Figure 1c) was more severe than that with one copy (Figure 1b). To characterize the defects in the internal structure of the ommatidia, we examined sections through the eyes. In the wild-type eye (Figure 1d), an ommatidium contains eight photoreceptors, seven of which can be observed in one tangential section (Wolff and Ready, 1993). The rhabdomeres, or light-sensitive organelles of the photoreceptors, are arranged in a characteristic trapezoid. Secondary and tertiary pigment cells form pigmented lattices surrounding each ommatidium. In GMR-Ral^G23V/+ eyes (Figure 1e), the orientation of ommatidia was irregular compared with wild-type (Figure 1d), although almost all of the ommatidia contained the normal number of photoreceptor rhabdomeres. In addition, a severe loss of rhabdomeres and pigmented lattices was observed in transgenic flies carrying two copies of the GMR-Ral^G23V (Figure 1f). These phenotypes are not likely to be caused by retinal degeneration after eclosion, since the examination of adults immediately after eclosion showed a similar level of defects (data not shown).

Cell fate determination is initiated normally despite the expression of activated Ral

Signal transduction through the Ras pathway plays a key role in cell fate determination in developing eye imaginal discs (reviewed by Zipursky and Rubin, 1994). Since Ral functions downstream of Ras (reviewed by Feig et al., 1996), Ral may be involved in this process. To examine the effect of overexpressing Ral on cell fate determination, developing eyes from third instar larvae and pupae were examined. In the wild-type eye discs from third instar larvae, neuronal differentiation occurs sequentially following the movement of the morphogenetic furrow (Tomlinson and Ready, 1987). The pattern of neuronal differentiation can be examined using an antibody against the neuronal RNA binding protein, Elav (Robinow and White, 1988). The Elav-immunostaining pattern of eye imaginal discs from GMR-Ral^G23V third-instar larvae (Figure 2b) was normal and indistinguishable from the wild-type pattern (Figure 2a), indicating that the initial stage of neuronal differentiation is not disturbed. At 40 h after puparium formation (APF), most of the GMR-Ral^G23V (Figure 2d) ommatidia contained eight Elav-positive nuclei but the normal spacing pattern of the ommatidia was disrupted compared with the wild-type pattern (Figure 2c). Differentiation of cone cells and primary pigment cells was examined using anti-Cut (Blochlinger et al., 1993) and anti-Bar (Higashijima et al., 1992) antibodies, respectively. The number of cells in the ommatidia of GMR-Ral^G23V eyes stained with these antibodies appeared to be normal: there were four Cut-positive cone cells (Figure 2e and f) and two Bar-positive primary pigment cells (Figure 2g and h) in both wild-type and GMR-Ral^G23V ommatidia. However, the arrangement of cells in the ommatidia (Figure 2e and g) was highly disorganized in GMR-Ral^G23V eyes (Figure 2f and h). We conclude that the ectopic expression of activated Ral did not grossly interfere with the initiation of cellular differentiation, but affected the arrangement of cells in the ommatidia.

Activated Ral protein disrupts the cell shape changes required for normal ommatidial development

Until the mid-pupal stage, the morphology of each retinal cell in the ommatidia undergoes dramatic changes (Wolff and Ready, 1993). As Ral has been implicated in regulating CDC42 and Rac activities, it is possible that the expression of activated Ral perturbs the cytoskeletal regulation required for the precise cell shape changes that occur during eye development. To pursue this possibility, the pupal retinas were stained with cobalt sulfide. This technique enables one to visualize the outlines of cells at the apical surface of the pupal retina (Wolff and Ready, 1991). The wild-type retina stained with cobalt sulfide showed the stereotyped pattern of cellular arrangement in each ommatidium (Figure 3a). However, the cellular arrangements were disorganized in GMR-Ral^G23V flies (Figure 3b). Notably, secondary and tertiary pigment cells were misshapen compared with the wild-type cells.

CDC42 and Rac, the putative downstream effectors of Ral, play important roles in the regulation of the actin cytoskeleton (reviewed by Aelst and D'Souza-Schorey, 1997). Then, we next examined the structure of the actin cytoskeleton in the retinal cells. The filamentous actin in the retina at 40 h APF was labeled with rhodamine-conjugated phalloidin and examined with a confocal microscope at several distinct focal planes. Phalloidin labels the cell outlines as well as the structure of the actin cytoskeleton in the cells. At the apical plane of the wild-type retina, the outlines of cone and primary pigment cells were clearly observed (Figure 4a). Filamentous structures composed of actin were also observed in this optical section. In the GMR-Ral^G23V/GMR-Ral^G23V retina, more filamentous actin bundles were observed, especially in the pigment cells (Figure 4b). Figure 4c is an optical section through a wild-type pupal retina at the intermediate level. Eight photoreceptor cells (R1-R8) were arranged in a trapezoid, forming the rhabdomere. Outlines of the pigment cells were not recognizable, since they were spread long and thin. In the GMR-Ral^G23V/GMR-Ral^G23V retina (Figure 4d), the shapes of photoreceptor cells and their rhabdomeres were disrupted. In addition, swollen pigment cells were visible between the photoreceptor cells, suggesting that the spreading of cells was impaired. At the basal surface of the wild-type retina (Figure 4e), secondary pigment cells elongated radially and formed a petal pattern with tertiary pigment cells and bristles. On the floor of the GMR-Ral^G23V/GMR-Ral^G23V retina (Figure 4f), the petal pattern was highly disorganized. Thus, constitutive activation of Ral affected the organization of the actin cytoskeleton and thus the changes in cell shape.

Ral^G23V synergistically enhances the phenotype caused by expression of RhoA

To search for other genes that might be associated with Ral-induced cytoskeletal defects, GMR-Ral^G23V flies were crossed to a number of fly stocks with transgenes or mutations in various genes involved in the Ras pathway, cytoskeletal regulation, or tissue porality. The resulting F1 progenies were scored for modification of the rough eye phenotype (see Materials and methods). As a result, we identified a transgenic line GMR-RhoA (Hariharan et al., 1995) as an enhancer of the eye phenotype of GMR-Ral^G23V. In the GMR-RhoA flies, Drosophila RhoA is overexpressed specifically in the developing eyes. It has been reported that the phenotype of GMR-RhoA is similar to that of GMR-Ral^G23V described above, i.e., the actin cytoskeleton is disorganized while cell fates are unaffected (Hariharan et al., 1995).

Eyes of GMR-Ral^G23V/+ (Figure 5a) and GMR-RhoA/+ (Figure 5b) show a mild rough-eye phenotype. On the other hand, eyes of the flies carrying both GMR-Ral^G23V and GMR-RhoA revealed synergistically enhanced defects (Figure 5c). The GMR-RhoA/GMR-Ral^G23V eyes were extremely rough and had a glossy appearance due to fusion of the facets (Figure 5c).

To characterize the cellular abnormalities in these mutants, their retinas were sectioned and examined. Figure 5d shows a longitudinal retinal section of the GMR-Ral^G23V/+fly. The GMR-Ral^G23V/+ retina showed a nearly normal pattern, i.e., pigment cells extending along and surrounding the rhabdomeres of the photoreceptor cells (Figure 5d). At the floor of the retina, a layer of pigment cell feet was observed (Figure 5d). Interestingly, overexpression of RhoA has been shown to inhibit the pigment cell elongation (Hariharan et al., 1995). Indeed, pigment cells failed to extend to the retinal floor and the layer of pigment cell feet could not be recognized in the section of the GMR-RhoA/+ retina (Figure 5e). The GMR-RhoA/ GMR-Ral^G23V sections displayed a more enhanced phenotype (Figure 5f). This retina contained a thin layer of pigment cells at the apical region and lacked recognizable ommatidia (Figure 5f). This phenotype is similar to that caused by two copies of the GMR-RhoA transgenes (Hariharan et al., 1995). Thus, the similarity of the Ral^G23V phenotype with that of RhoA, and the synergy seen with the simultaneous expression of these two genes, suggest that Ral^G23V and RhoA might affect the actin cytoskeletal organization by common mechanisms.

Discussion

Ras mediates its diverse biological functions by activating multiple downstream targets (Marshall, 1996). Guanine nucleotide exchange factors for Ral GTPases including RalGDS, RGL and RLF have been recently identified as Ras effector molecules. In this work, we used the developing Drosophila eye as a model system to elucidate the in vivo function of Ral. Although we used human transgenes in this study, a Drosophila Ral homologue that has high sequence homology with mammalian Ral proteins has recently been identified (K Sawamoto et al., unpublished data). The Drosophila eye has been used successfully to study the function of human genes (de Nooij and Hariharan, 1995; Warrick et al., 1998). The advantage of using the human gene in this study is that the biochemical properties of Ral^G23V have been well characterized (Frech et al., 1990; Hinoi et al., 1996). The human Ral has also been used to study the effects on migration of border cells during Drosophila ovary development (Lee et al., 1996) We believe that Ral recognizes the correct target protein(s) in Drosophila, since the amino acid sequences in the effector domains of human and Drosophila Ral are completely identical with each other.

Drosophila possesses a single homologue of the mammalian N-ras, K-ras, and H-ras genes, called Ras1 (Simon et al., 1991). Ras1 is activated in response to the activation of multiple receptor tyrosine kinases and plays critical roles in various developmental processes. In particular, the function of Ras1 in ommatidial development has been well documented (Wassarman and Therrien, 1997). It has been shown that the expression of a constitutively activated Ras1 (Ras1^G12V) in developing eyes has several distinct effects on various developmental stages, i.e., transformation of non-neuronal cone cells into ectopic R7 cells (Fortini et al., 1992), induction of hyperplastic growth (Karim and Rubin, 1998) and inhibition of programmed cell death (Sawamoto et al., 1998; Miller and Cagan, 1998). These functions of Ras1 are likely to be mediated by one of the Ras effectors, D-raf/Raf. Two Drosophila genes coding for possible Ras effectors Canoe/AF6 (Miyamoto et al., 1995; Matsuo et al., 1997) and phosphoinositide 3-kinase (Leevers et al., 1996) have also been identified, but their roles in the Ras pathway have not yet been determined. In the present work, the novel phenotype caused by another downstream-signaling molecule, Ral, indicates that Ral has a function distinct from that of the Raf/ERK pathway.

We showed that overexpression of the GTP-bound form of Ral interfered with normal cell-shape changes during Drosophila eye development. The precise mechanisms of how Ral^G23V affects cell shape are unknown. However, identification of RalBP1, a GAP protein for CDC42 and Rac, as a possible effector of Ral (Cantor et al., 1995; Jullien-Flores et al., 1995; Park and Weinberg, 1995) suggests that Ral may function to activate or inactivate the activity of CDC42 and Rac, which control the actin cytoskeleton. Indeed, several data from this study support the idea that the GMR-Ral^G23V phenotype was caused by inhibition of cell shape changes due to aberrant organization of the actin cytoskeleton. First, developmental defects became manifest in pupal stages, when the cell shape changes occur for normal eye morphogenesis. Second, the GMR-Ral^G23V phenotype was highly similar to that caused by overexpression of RhoA. Third, coexpression of RhoA synergistically enhanced the phenotype caused by activated Ral. Thus, there is a possibility that activated Ral might affect the actin cytoskeleton through regulating members of the Rho family of GTPases. Alternatively, it is also possible that Ral affected the actin cytoskeleton independently of CDC42 and Rac. To study if changes in CDC42 or Rac activities are responsible for the Ral^G23V-induced phenotype, we examined effects of coexpression of dominant negative CDC42 or Rac on the phenotype of GMR-Ral^G23V. However, the phenotype caused by dominant negative CDC42 or Rac was too severe to observe the modification of the phenotype of GMR-Ral^G23V (data not shown). Identification and characterization of Ral and RalBP1 homologues from Drosophila are required to further study Ral function in regulation of CDC42 and Rac, and in actin-dependent cell shape changes during development.

In addition to its role in inducing cellular proliferation, Ras is known to have effects on the actin cytoskeleton. Microinjection of activated Ras protein induces membrane ruffling or actin stress-fiber formation in cultured cells (Bar Sagi and Feramisco, 1986; Ridley and Hall, 1992). These defects are due to rearrangement of the actin cytoskeleton, mediated by the Rho family of GTPases (Ridley and Hall, 1992). In Drosophila, Ras1 has been implicated in cell migration during development of tracheal system and ovary (Reichman-Fried et al., 1994; Lee et al., 1996). It is possible that Ras1 controls cytoskeleton to regulate the cell motirity and movement during these development processes.

Which Ras effector protein is important for the cytoskeletal rearrangement? Raf is important for oncogenic proliferation, but is not directly involved in regulating the cytoskeleton (Joneson et al., 1996). Other Ras effectors include phosphoinositide 3-kinase (Marshall, 1996). Rodriguez-Viciana et al. (1997) have reported that activation of the phosphoinositide 3-kinase is essential in Ras-induced cytoskeletal reorganization. In fission yeast, Ras has been shown to directly regulate a GEF for CDC42 (Chang et al., 1994), although no such connection has yet emerged in mammalian systems. The data presented here indicate that Ral, which is activated by another family of Ras-effector RalGEFs, is also able to influence the actin cytoskeleton in the developing Drosophila eye. The regulated activation of multiple effectors of Ras and cross-talking among them may be important for the precise actin rearrangement during transformation as well as for the cell shape changes that occur during development.

Materials and methods

Drosophila stocks and culture

All fly crosses were performed at 25°C on standard cornmeal medium. The GMR-RhoA strain was kindly provided by Dr Iswar Hariharan (Hariharan et al., 1995). Canton-S or w¹¹¹⁸ were used as the wild-type strain. Mutations tested that displayed no interaction with GMR-Ral^G23V included alleles of RhoA, cdc42, nemo, hemipterous, basket, Ras1, D-raf, rolled and Dsor1.

Plasmid construction and P-element mediated germline transformation

Human cDNAs (Hinoi et al., 1996) coding for wild-type RalB and the RalA^G23V mutant (in which Gly-23 was changed to Val) were ligated into the EcoRI site of pGMR (Hay et al., 1994). The resulting plasmids, pGMR-Ral and pGMR-Ral^G23V, were injected into embryos from the w¹¹¹⁸; Dr/TMS, Sb P[ry⁺2 - 3] stock to establish the transgenic lines as described previously (Sawamoto et al., 1994). Multiple independent transformant lines were established and analysed for each construct. All data presented in this paper are from a single strain.

Phenotypic analyses

Flies were prepared for scanning electron microscopy as described by Kimmel et al. (1990). Semi-thin sections of adult eyes were prepared as described in Sawamoto et al. (1994). Cobalt sulfide staining of developing retinas was performed as described previously (Wolff and Ready, 1991). Immunohistochemistry was carried out essentially as described (Tomlinson and Ready, 1987). Phalloidin staining was performed using the method of Hariharan et al. (1995).

Acknowledgements

We thank Kaoru Saigo, Seymor Benzer, Iswar Hariharan, Yuh Nung Jan, Tetsuya Kojima, Liqun Luo, Marek Mlodzik, Rick Fehon, Yasuyoshi Nishida, Gerald Rubin, the Developmental Studies Hybridoma Bank and the Bloomington Stock Center for the fly stocks and antibodies used in this study. We are grateful to Ritsuko Shimamura and Sachiyo Miyao for their help in maintenance of fly strains. This work was supported by grants from the Japan Ministry of Education, Science and Culture to KS, AK and HO.

Aelst LV and D'Souza-Schorey C. (1997). Genes Dev. 11, 2295-2322. MEDLINE

Albright CF, Giddings BW, Liu J, Vito M and Weinberg RA. (1993). EMBO J. 12, 339-347. MEDLINE

Bar Sagi D and Feramisco JR. (1986). Science 233, 1061-1068. MEDLINE

Barrett K, Leptin M and Settleman J. (1997). Cell 91, 905-915. MEDLINE

Blochlinger KL, Jan LY and Jan YN. (1993). Development 117, 441-450. MEDLINE

Bos JL. (1997). Biochem. Biophys. Acta 1333, M19-M31.

Cantor SB, Urano T and Feig LA. (1995). Mol. Cell. Biol. 15, 4578-4584. MEDLINE

Chang EC, Barr M, Wang Y, Jung V, Xu HP and Wigler MH. (1994). Cell 79, 131-141. MEDLINE

de Nooij JC and Hariharan IK. (1995). Science 270, 983-985. MEDLINE

Eaton S, Wept R and Simons K. (1996). J. Cell Biol. 135, 1277-1289. MEDLINE

Feig LA, Urano T and Contor S. (1996). Trends Biochem. Sci. 21, 438-441. MEDLINE

Fortini ME, Simon MA and Rubin GM. (1992). Nature 355, 559-561. MEDLINE

Frech M, Schlichting I, Wittinghofer A and Chardin P. (1990). J. Biol. Chem. 265, 6353-6359. MEDLINE

Harden N, Loh HY, Chia W and Lim L. (1995). Development 121, 903-914. MEDLINE

Hariharan IK, Hu K-Q, Asha H, Quintanilla A, Ezzel RM and Settleman J. (1995). EMBO J. 14, 292-302. MEDLINE

Hay BA, Wolff T and Rubin GM. (1994). Development 120, 2121-2129. MEDLINE

Higashijima S, Kojima T, Michiue T, Ishimaru S, Emori Y and Saigo K. (1992). Genes Dev. 6, 50-60. MEDLINE

Hinoi T, Kishida S, Koyama S, Ikeda M, Matsuura Y and Kikuchi A. (1996). J. Biol. Chem. 271, 19710-19716. MEDLINE

Hofer F, Fields S, Schneider C and Martin GS. (1994). Proc. Natl. Acad. Sci. USA 91, 11089-11093. MEDLINE

Jiang H, Luo J-Q, Urano T, Frankel P, Lu Z, Foster DA and Feig LA. (1995). Nature 378, 409-412. MEDLINE

Joneson T, White MA, Wigler MH and Bar Sagi D. (1996). Science 271, 810-812. MEDLINE

Jullien-Flores V, Dorseuil O, Romero F, Letourneur F, Saragosti S, Berger R, Tavitian A, Gacon G and Camonis HJ. (1995). J. Biol. Chem. 270, 22473-22477. MEDLINE

Karim FD and Rubin GM. (1998). Dev. 125, 1-9.

Kikuchi A, Demo SD, Ye Z-H, Chen YW and Williams LT. (1994). Mol. Cell. Biol. 14, 7483-7491. MEDLINE

Kimmel BE, Heberlein U and Rubin GM. (1990). Genes Development 4, 712-727. MEDLINE

Kishida S, Koyama S, Matsubara K, Kishida M, Matsuura Y and Kikuchi A. (1997). Oncogene 15, 2899-2907. MEDLINE

Lee T, Feig L and Montell DJ. (1996). Development 122, 409-418. MEDLINE

Leevers SL, Weinkove D, MacDougail K, Hafen E and Waterfield MD. (1996). EMBO J. 15, 6584-6594. MEDLINE

Longley Jr RL and Ready DF. (1995). Dev. Biol. 171, 415-433. Article MEDLINE

Luo L, Liao YJ, Jan LY and Jan YN. (1994). Genes Dev. 8, 1787-1802. MEDLINE

Marshall CJ. (1996). Curr. Opin. Cell Biol. 8, 197-204. MEDLINE

Matsuo T, Takahashi K, Kondo S, Kaibuchi K and Yamamoto D. (1997). Dev. 124, 2671-2681.

Miller MJ, Prigent S, Kupperman E, Rioux L, Park SH, Feramisco JR, White MA, Rutkowski JL and Meinkoth JL. (1997). J. Biol. Chem. 272, 5600-5605. MEDLINE

Miller DT and Cagan RL. (1998). Development 125, 2327-2335. MEDLINE

Miyamoto H, Nihonmatsu I, Kondo S, Ueda R, Togashi S, Hirata K, Ikegami Y and Yamamoto D. (1995). Genes Dev. 9, 612-625. MEDLINE

Murai H, Ikeda M, Kishida S, Ishida O, Okazaki-Kishida M, Matsuura Y and Kikuchi A. (1997). J. Biol. Chem. 272, 10483-10490. MEDLINE

de Nooij JC and Hariharan IK. (1995). Science 270, 983-985. MEDLINE

Okazaki M, Kishida S, Hinoi T, Hasegawa T, Tamada M, Kataoka T and Kikuchi A. (1997). Oncogene 14, 515-521. MEDLINE

Park SH and Weinberg RA. (1995). Oncogene 11, 2349-2355. MEDLINE

Reichman-Fried M, Dickson B, Hafen E and Shilo BZ. (1994). Genes Dev. 8, 428-439. MEDLINE

Ridley A and Hall A. (1992). Cell 70, 389-399. MEDLINE

Robinow S and White K. (1988). Dev. Biol. 126, 294-303. MEDLINE

Rodriguez-Viciana P, Warne PH, Khwaja A, Marte BM, Pappin D, Das P, Waterfield MD, Ridley A and Downward J. (1997). Cell 89, 457-467. MEDLINE

Sawamoto K, Okano H, Kobayakawa Y, Hayashi S, Mikoshiba K and Tanimura T. (1994). Dev. Biol. 164, 267-276. MEDLINE

Sawamoto K, Taguchi A, Hirota Y, Yamada C, Jin M-h and Okano H. (1998). Cell Death Differ. 5, 262-270. MEDLINE

Simon MA, Bowtell DL, Dodson GS, Laverty TR and Rubin GM. (1991). Cell 67, 701-716. MEDLINE

Spaargaren M and Bischoff JR. (1994). Proc. Natl. Acad. Sci. USA 91, 12609-12613. MEDLINE

Strutt ID, Weber U and Mlodzik M. (1997). Nature 387, 292-295. MEDLINE

Tomlinson A and Ready DF. (1987). Dev. Biol. 123, 264-275.

Urano T, Emkey R and Feig LA. (1996). EMBO J. 15, 810-816. MEDLINE

Warrick JM, Paulson HL, Gray-Board GL, Bui QT, Fishbeck KH, Pittman RN and Bonini NM. (1998). Cell 93, 939-949. MEDLINE

Wassarman DA and Therrien M. (1997). Adv. Dev. Biol. 5, 1-41.

White MA, Vale T, Camonis JH, Schaefer E and Wigler MH. (1996). J. Biol. Chem. 271, 16439-16442. MEDLINE

Wolff T and Ready DF. (1991). Development 113, 825-839. MEDLINE

Wolff T and Ready DF. (1993). In: The development of Drosophila melanogaster. Bate M and Arias AM. (eds). Cold Spring Harbor Laboratory Press.,

Wolthuis RMF, Bauer B, van't Veer LJ, de Vries-Smits AMM, Cool RH, Spaargaren M, Wittinghofer A, Burgering BMT and Bos JL. (1996). Oncogene 13, 353-362. MEDLINE

Wolthuis RMF, de Ruiter ND, Cool RH and Bos JL. (1997). EMBO J. 22, 6748-6761.

Wolthuis RMF, Zwartkruis F, Moen TC and Bos JL. (1998). Curr. Biol. 8, 471-474. MEDLINE

Zipursky SL and Rubin GM. (1994). Ann. Rev. Neurosci. 17, 373-397. MEDLINE

Figure 1 (a - c) Scanning electron micrographs of adult compound eyes. (a) Wild-type. (b) GMR-Ral^G23V/+. (c) GMR-Ral^G23V/GMR-Ral^G23V. (d - f) Sections of adult compound eyes. Arrows indicate the ommatidial orientations. (d) Wild-type. (e) GMR-Ral^G23V/+. (f) GMR-Ral^G23V/GMR-Ral^G23V

Figure 2 Staining of developing eyes with cell-type specific antibodies. (a and b) Eye imaginal discs from third instar larvae stained with anti-Elav antibody. (a) wild-type. (b) GMR-Ral^G23V/GMR-Ral^G23V. (c - h) Developing eyes from pupae at 40 h APF stained with anti-Elav (c and d), anti-Cut (e and f) or anti-Bar (g and h) antibodies. (c, e, g) Wild-type. (d, f, h) GMR-Ral^G23V/GMR-Ral^G23V

Figure 3  Cobalt sulfide staining of retinas at 40 h APF. (a) Wild-type. (b) GMR-Ral^G23V/GMR-Ral^G23V. C: cone cell; P: primary pigment cells; S: secondary pigment cells; T: tertiary pigment cells

Figure 4 Phalloidin staining. Retinas at 40 h APF from the wild-type (a, c, e) and GMR-Ral^G23V/GMR-Ral^G23V(b, d, f) flies were stained with rhodamine-conjugated phalloidin and examined using a confocal microscope. (a and b) Apical plane. (c and d) Intermediate plane. (e and f) Basal plane. C: cone cell; P: primary pigment cell; 1: R1; 2: R2; 3: R3; 4: R4; 5: R5; 6: R6; 7: R7; 8: R8; S: secondary pigment cell; T: tertiary pigment cell; B: bristle. Some of the photoreceptor cells (R) and pigment cells (*) are marked in (d)

Figure 5 Coexpression of RhoA synergistically enhances the phenotype caused by activated Ral. (a - c) Scanning electron microscopy of adult eyes. (d - f) Longitudinal section of adult eyes. (a, d) GMR-Rho1/+. (b, e) GMR-Ral^G23V/+. (c, f) GMR-Rho1/GMR-Ral^G23V. Arrows indicate the postions of the pigment cell feet