The Drosophila Ral GTPase Regulates Developmental Cell Shape Changes through the Jun NH₂-terminal Kinase Pathway

Degenerate primers were designed to amplify Ras-like genes by PCR. The sequences were: GGIGTIGGIAA(A/G)(A/T)(C/G)(A/C/G/T)GC(A/C/G/T)(C/T)T(A/C/G/T)AC and A(C/T)TC(C/T)TGICC(A/C/G/T)GC(A/C/G/T)GT(A/G)TC. PolyA⁺ mRNA was prepared from S2 cells using a Micro Fast Track kit (Invitrogen Corp.) and used as the template for synthesizing cDNAs using a first strand cDNA kit (Pharmacia Biotech, Inc.). The PCR procedure was: five cycles at 94°C for 1 min, 46°C for 1 min, and 72°C for 1 min, followed by 45 cycles at 94°C for 1 min, 52°C for 1 min, and 72°C for 1 min. PCR products were subcloned into the pT7 blue T vector (Novagen, Inc.), transformed into JM109 cells, and subjected to DNA sequencing according to standard protocols. The PCR product of DRal was ³²P-labeled and used as a probe to screen an eye imaginal disc cDNA library. Five 1.2-kb cDNAs were identified that contained the entire open reading frame encoding DRal, a 261-bp 5′ untranslated region and a 354-bp 3′ untranslated region. The nucleotide sequence encompassing the open reading frame was determined by sequencing the cDNAs from both directions.

RNA samples were prepared from eye imaginal discs of third-instar larvae according to the method previously described by Fisher-Vize et al. (1992). Northern blotting analysis was performed using standard methods. The DRal cDNA was labeled with ³²P-dCTP and used as a probe.

The DRal cDNA was labeled with digoxigenin using a random-primer kit (Boehringer Mannheim Corp.) and hybridized with squashed Polytene chromosomes, as described previously (Zuker et al., 1985). The chromosomes were incubated with alkaline phosphatase-coupled antidigoxigenin antibodies. The signal was developed according to the manufacturer’s instructions.

The DRal cDNA in pBluescript was used as the template for site-directed mutagenesis with QuickChange™ and Chameleon Kits (Stratagene). The constitutively active DRal^G20V mutation was created using an oligonucleotide with a base change from GGC to GTC, converting amino acid 20 from Gly to Val. The dominant negative DRal^S25N mutation was created using an oligonucleotide with a base change from TCC to AAC, converting amino acid 25 from Ser to Asn. Mutations were confirmed by DNA sequencing. The cDNA inserts with or without mutations were excised from pBluescript and then ligated into either pGEX (Pharmacia Biotech, Inc.), for the expression of glutathione S-transferase (GST)-fusion proteins in Escherichia coli, or into pUAST (Brand and Perrimon, 1993), for the generation of transgenic Drosophila lines and transfection of S2 cells.

To purify GST fusion proteins (GST-DRal, GST-DRal^G20V, GST-DRal^S25N, and GST-RalGDS) from E. coli, transformed E. coli were initially grown in Luria-Bertani’s broth at 37°C to an absorbance of 0.8 (OD = 600 nm), and subsequently transferred to 25°C. Isopropyl-1-β-d-thiogalactopyranoside was added to a final concentration of 100 μM and further incubation was carried out for 10 h at 25°C. The GST fusion proteins were purified from E. coli by glutathione Sepharose 4B, in accordance with the manufacturer’s instructions.

GST-DRal and GST-DRal mutants (8 pmol each) were preincubated for 10 min at 30°C in 20 μl of reaction mixture (50 mM Tris/HCl, pH 7.5, 2 μM [³H]GDP [1,500–3,000 dpm/pmol], 5 mM MgCl₂, 10 mM EDTA, 1 mM DTT, and 1 mg/ml BSA). After preincubation, 1 μl of 400 mM MgCl₂ was added. To this preincubation mixture, 29 μl of reaction mixture (50 mM Tris/HCl, pH 7.5, 170 μM GTP, and 1 mg/ml BSA) containing GST-Ral-GDS (10 pmol) was added, and the mixture was further incubated for 5–30 min at 30°C. Assays were quantified by rapid filtration on nitrocellulose filters (Albright et al., 1993).

RalGAP was partially purified from bovine brain cytosol as described previously (Hinoi et al., 1996). GST-DRal and GST-DRal mutants (3 pmol each) were preincubated for 10 min at 30°C in 9 μl of the preincubation mixture (50 mM Tris/HCl, pH 7.5, 2 μM γ[³²P]GTP [8,000–12,000 cpm/pmol], 5 mM MgCl₂, 10 mM EDTA, 1 mM DTT, and 1 μg/ml BSA). After preincubation, 1 μl of 340 mM MgCl₂ was added. To this preincubation mixture, 30 μl of reaction mixture (50 mM Tris/HCl, pH 7.5, 1.3 mM GTP, 0.3 mM MgCl₂, and 1 mg/ml BSA) containing RalGAP (7 μg of protein) was added, and the second incubation was performed for 15 min at 30°C. Assays were quantified by rapid filtration on nitrocellulose filters. The actual catalytic rates (K_cat) were calculated from the decrease in radioactive γ[³²P]GTP in the presence or absence of RalGAP (Higashijima et al., 1987).

The K_d values for GDP or GTPγS of, dissociation rate of GDP (K⁻¹) from, and the steady-state rate (K_ss) of GTP hydrolysis of the mutant forms of DRal were determined as described previously (Kikuchi et al., 1988; Shoji et al., 1989).

Plasmids were injected into the embryos of w¹¹¹⁸; Dr/TMS, Sb P[ry⁺, Δ2–3] (from the Bloomington stock center) to generate transgenic lines, as described previously (Sawamoto et al., 1994). w¹¹¹⁸ was used as the wild-type strain. GMR-GAL4 was provided by M. Freeman (MRC Laboratory of Molecular Biology, Cambridge, UK); GAL4-69B by R. Ueda (Mitsubishi-Kasei Institute of Life Sciences, Tokyo, Japan); sca-GAL4 by T. Hosoya (National Institute of Genetics, Mishima, Japan); mbt^P1 and mbt^P2 by T. Raabe (Universitaet Wuerzburg, Wuerzburg, Germany); da-GAL4 by F. Matsuzaki (Tohoku Univ., Sendai, Japan); RhoA^P2 by M. Mlodzik (EMBL, Heidelberg, Germany); cdc42¹ by R. Fehon (Duke Univ., Durham, NC); hep^r75, hep¹, bsk¹, and Df(2L)flp 147E by Y. Takatsu (National Institute for Basic Biology, Okazaki, Japan); actin-GAL4 and arm-GAL4 from M. Okabe (National Institute of Genetics, Mishima, Japan); pnr-GAL4 and LE-GAL4 from M. Tateno (Nagoya Univ., Nagoya, Japan); Ras1^e2F by D. Yamamoto (Waseda Univ., Tokyo, Japan); D-raf¹, rl^Su23, rl^EMS64, Dsor1^Gp158, and Dsor1^Su1 by Y. Nishida (Nagoya Univ., Nagoya, Japan); and Df(3L)emc5, Df(3L)pbl-X1, w¹¹¹⁸, and w¹¹¹⁸; Dr/TMS, Sb P[ry⁺, D2–3] by the Bloomington Stock Center. Fly crosses were performed at 25°C unless noted otherwise.

In situ hybridization to embryonic and larval tissues was performed as described by Tautz and Pfeifle (1989), using an antisense RNA probe encompassing the entire DRal cDNA. A sense probe was used in parallel as the control.

For scanning EM, adult flies or isolated wings were dehydrated in a graded ethanol series and dried using a critical point drier. The mounted samples were ion-coated and observed with a scanning electron microscope (Hitachi Instruments, Inc.).

For phalloidin staining, pupal wings were dissected away from the surrounding cuticle and fixed in 8% paraformaldehyde/PBS at room temperature for 20 min. The wing samples were washed in 0.1% Triton X-100/PBS (PBT) three times, then incubated in rhodamine-phalloidin/PBT (0.5 mg/ml; Sigma Chemical Co.) overnight at 4°C. After rinsing in three changes of PBT, the wings were mounted and examined with a confocal laser microscope (Olympus).

Pupal nota were dissected and fixed in 4% paraformaldehyde/PBS as described previously (Tilney et al., 1996). The nota were stained with rhodamine-phalloidin following the same protocol used for wing samples described above. For immunohistochemistry, the fixed and washed nota were incubated in 10% normal goat serum/PBT for 30 min at room temperature and then in 1:5 mAb 22C10 (obtained from S.C. Fujita, Mitsubishi Kasei Institute of Life Sciences, Tokyo, Japan) diluted in 10% normal goat serum/PBT overnight at 4°C. After rinsing in three changes of PBT, the nota were incubated in biotin-conjugated anti-mouse IgG at a dilution of 1:200 in 10% normal goat serum/PBT for 2 h at room temperature. The signal was developed using an ABC Elite Kit (Nycomed Amersham Inc.).

Preparation and analysis of embryonic cuticle were performed as described previously (Wieschaus and Nüsslein-Volhard, 1986).

pWAGAL4 was a kind gift from Dr. Yasushi Hiromi (National Institute of Genetics, Japan). S2 cells were grown on 24-well plates to 60–80% confluence in Schneider’s medium (Sigma Chemical Co.) supplemented with 10% FBS and 0.5% peptone (Difco Laboratories Inc.). The cells were transfected with pWAGAL4 (200 ng) alone, pWAGAL4 (200 ng) plus pUAST-DRal (1 μg), or pWAGAL4 (200 ng) plus pUAST-DRal^G20V (1 μg) using Cell Fectin reagent (GIBCO BRL) according to the manufacturer’s instructions. After 24 h, the cells were incubated in Drosophila serum-free medium (GIBCO BRL) for 30 min, then treated with 500 mM d-sorbitol for 5 min. Cells were lysed in 40 μl of SDS-PAGE sample buffer containing phosphatase inhibitors (100 nM okadic acid, 200 μM sodium orthovanadate, and 50 mM sodium fluoride), heated at 100°C for 3 min and spun at 10,000 g for 10 min. The resulting supernatant fractions were subjected to SDS-PAGE (12.5% gel) and transferred to a nylon membrane. After blocking in 5% dry milk in PBS + 0.1% Tween-20 overnight at 4°C, the membranes were incubated with anti-ACTIVE JNK (Promega Corp.) or anti-JNK1 (Santa Cruz Biotechnology) antibodies for 1 h at room temperature, and then with HRP-conjugated anti-rabbit IgG (Jackson ImmunoResearch Laboratories, Inc.) for 1 h at room temperature. Signals were detected by ECL reagents (Nycomed Amersham Inc.).

The known GTPase genes of the Ras family share significant homology in several structurally and functionally important regions. To search for novel Ras-like GTPases in Drosophila, we designed degenerative PCR primers that recognize the nucleotide binding and effector regions of the known GTPases of the Ras family and that were also likely to amplify novel Ras-like GTPases. Using these primers to perform reverse transcriptase PCR, we isolated a number of cDNAs encoding Ras-like GTPases. Some were known genes, such as Ras1 (Simon et al., 1991), Ras2 (Neuman-Silberberg et al., 1984), roughened (Hariharan et al., 1991), and Ric (Wes et al., 1996), and others represented a novel gene similar to Ral. We used a PCR fragment with a sequence homologous to Ral as a probe to screen an eye imaginal disc cDNA library and isolated a 1.2-kb cDNA clone. The size of the cDNA was similar to that of the transcript detected by Northern analysis of RNA from eye antennal discs in third-instar larvae using the same cDNA as a probe (data not shown).

The sequence of the cDNA indicated a single open reading frame encoding a protein of 201 amino acids with a predicted molecular mass of 21 kD. The deduced amino acid sequence shared high homology with all of the mammalian Ral proteins (Fig. 1). The amino acid sequence in the putative effector domain was identical to that of the mammalian Ral proteins. The CAAX motif at the COOH terminus required for geranylgeranylation (Hinoi et al., 1996) was also conserved. Based on the sequence similarity, we named the gene DRal.

To determine the cytological map position of the DRal gene, we performed in situ hybridization with chromosomes from Drosophila salivary glands using the DRal cDNA as the probe. A single signal was detected in the 3E region on the X chromosome (data not shown).

To examine the spatiotemporal expression pattern of the DRal mRNA during development, in situ hybridization analysis was performed at various stages of development using a DRal antisense RNA probe. Widespread expression of the DRal transcripts was detected throughout embryogenesis (Fig. 2, A–C). In the third-instar larval stages, DRal mRNA was also broadly expressed in the brain hemispheres and ventral nerve cords (Fig. 2 D), leg discs (Fig. 2 E), eye discs (Fig. 2 F), and wing discs (Fig. 2 G). The sense control probe did not hybridize to these tissues (data not shown).

Since no mutants of the DRal gene were available, we designed constitutively active and dominant negative DRal mutants based on structure–function studies of human Ral (Hinoi et al., 1996). Point mutations at two positions in DRal, G20V and S25N, were generated. The DRal^G20V (Val-20 for Gly-20) mutation corresponds to Ras^G12V, which was originally identified in an oncogenic form of Ras and is shown to render Ras constitutively active as a result of defective GTPase activity (reviewed by Barbacid, 1987). The DRal^S25N (Asn-25 for Ser-25) mutation corresponds to Ras^S17N, which was originally identified by its preferential binding to GDP over GTP (Feig and Cooper, 1988). Ras^S17N may function as a dominant negative mutant by sequestering the GEF (Schweighoffer et al., 1993).

Previously, we characterized the biochemical activities of human wild-type Ral and its mutants (Hinoi et al., 1996). To examine the biochemical characteristics of the DRal mutants used here, we inserted wild-type DRal and the two DRal mutants (DRal^G20V and DRal^S25N) into bacterial expression vectors and purified them as GST fusion proteins. The characterization of these DRal mutants is summarized in Table I. The K_d values of wild-type DRal for GDP and GTPγS were similar (∼14 and 31 nM, respectively). DRal^G20V also showed similar K_d values for both GDP and GTPγS. The K_d values of DRal^S25N for GDP and GTPγS were larger than those of wild type, and its affinity for GDP was four- to fivefold higher than for GTPγS. The GDP dissociation constants (K⁻¹) of wild-type DRal, DRal^G20V, and DRal^S25N were 0.009, 0.006, and 0.09, respectively. RalGDS stimulated the dissociation of GDP from DRal four- to fivefold. RalGDS stimulated the dissociation of GDP from DRal^G20V threefold, but did not affect that from DRal^S25N. The steady-state rates (K_ss) of the GTPase activity of DRal, DRal^G20V, and DRal^S25N were 0.007, 0.003, and 0.004, respectively. RalGAP stimulated the actual GTPase K_cat of wild-type DRal eightfold, but not that of DRal^G20V. The biochemical characteristics of DRal^G20V and DRal^S25N were almost identical to those of human Ral^G23V and Ral^S28N, respectively. These results indicate that Ser-25 of DRal is important for the action of RalGDS, that Gly-20 is important for the action of Ral-GAP, and that DRal^G20V and DRal^S25N are constitutively active and dominant negative forms of DRal, respectively.

To gain insight into the function of DRal in Drosophila development, we examined the effects of overexpressing the dominant mutants described in a specific tissue using the GAL4/UAS (upstream activation sequence) system (Brand and Perrimon, 1993). The cDNAs encoding the wild-type, constitutively active (G20V), and dominant negative (S25N) DRal proteins were subcloned into the transformation vector pUAST (Brand and Perrimon, 1993), and used to generate transgenic lines. The pUAST vector contains the UAS that is responsive to the yeast transcription factor GAL4. We then crossed these transgenic flies to several GAL4 lines. For all of the experiments in this study, at least two independent UAS-DRal lines were examined and found to show similar phenotypes.

Overexpression of the wild-type DRal protein did not cause any visible phenotype. On the other hand, overexpression of DRal^G20V and DRal^S25N resulted in a variety of phenotypes that depended on the GAL4 line used. In this study, we focused on the effect of DRal^S25N on the development of two cell types that have highly specialized structures, hair and bristles, since the phenotypes were obvious and easy to analyze. The development of these structures is dependent on the proper regulation of the cytoskeleton (Adler, 1992; Wulfkuhle et al., 1998).

Each epithelial cell of the Drosophila wing forms a hair by extending a single process from its apical membrane during pupal development (Mitchell et al., 1983; Adler, 1992; Fristrom et al., 1993; Wong and Adler, 1993). At ∼35 h after puparium formation (APF), F-actin accumulates on the distal side of the epithelial cells. Subsequently, outgrowth of prehairs is initiated from the distal side (Wong and Adler, 1993). To examine if DRal is involved in hair outgrowth, wild-type and dominant negative DRal proteins were misexpressed in developing wings, using the Gal4 line 69B (Brand and Perrimon, 1993). To observe the fine structure of the hair, wing samples were examined with a scanning electron microscope (Fig. 3, A–E). The wild-type wing hairs were evenly spaced with distal polarity (Fig. 3 A). The hairs on the wings overexpressing the wild-type DRal protein were morphologically indistinguishable from the wild-type hair (data not shown). On the wings expressing DRal^S25N, the cells often formed multiple wing hairs (Fig. 3, B, D, and E). The hairs were also shorter than in wild type (Fig. 3 B). Moreover, some hairs were forked, curved, or twisted (Fig. 3, B and D). The abnormal appearance of the hairs suggests that the organization of the actin cytoskeleton in the hairs may have been defective.

To label the F-actin, the developing wings were dissected from pupae at 30–36 h APF and stained with rhodamine-conjugated phalloidin. In the wild-type pupal wings, a single large bundle of F-actin, termed the prehair, is formed in each wing cell (Wong and Adler, 1993; Fig. 3 F). In the developing wings expressing DRal^S25N, cells with two or three prehairs were occasionally seen (Fig. 3 G). In addition, the morphology of the prehairs was irregular (Fig. 3 G). These data suggest that DRal is required for regulation of the initiation process during hair development.

The development of sensory bristles provides another excellent model system to study how the cytoskeleton controls cell shape changes. Each external sense organ consists of four cells: the neuron, the sheath, the tormogen (socket forming cell), and the trichogen (shaft forming cell; Hartenstein and Posakony, 1989). During pupal development, a cytoplasmic extension of the trichogen becomes the bristle shaft. To induce expression of the DRal proteins in the developing trichogens, UAS-DRal flies were crossed to the sca-GAL4 line. The bristles of flies expressing the wild-type DRal protein were indistinguishable from those of wild-type (Fig. 4, A and C): their length, morphology, and orientation were normal (data not shown). On the other hand, the expression of DRal^S25N resulted in the loss of bristles on the nota (Fig. 4, B and D). In some cases, DRal^S25N affected both shafts and sockets, suggesting that DRal may be involved in the development of these structures. Both macrochaetes (large bristles) and microchaetes (small bristles) were affected by the DRal^S25N expression.

The absence of bristles on the nota expressing DRal^S25N could be due to failure in the process of shaft initiation from the trichogen cells. Alternatively, overexpression of the dominant negative DRal protein could disrupt the formation of the trichogen cells. To distinguish between these two possibilities, developing nota from pupae at 26–32 h APF were stained with the antibody mAb 22C10 (Fujita et al., 1982). At this stage, mAb 22C10 labeled at least two cells in each macrochaete and microchaete on the wild-type nota: a neuron sending out axons and a trichogen cell producing a shaft (Fig. 5, A and C). On the nota expressing DRal^S25N, the neuron and trichogen were stained with mAb 22C10 in each macrochaete and microchaete, but the developing shafts appeared to be malformed (Fig. 5, B and D). These data indicate that DRal^S25N perturbed shaft initiation, but not the recruitment of trichogen cells.

To visualize the F-actin in the developing bristles, pupal nota at 26–32 h APF were dissected and stained with rhodamine-conjugated phalloidin. Fig. 6 A shows developing microchaetes in wild-type nota. Developing shafts containing F-actin were observed at this stage. On the nota expressing DRal^S25N, initiation of shafts was often inhibited (Fig. 6 B). Fig. 6 C shows a wild-type macrochaete. The developing shaft was filled with well-organized actin bundles that ran parallel to the long axis of the bristle. At the tip, patches of F-actin were observed. On the nota expressing DRal^S25N, the development of actin structures in the macrochaetes appeared to be interrupted at the initiation of extension (Fig. 6 D).

Next, we used the GAL4-69B line to examine the effects of expressing wild-type DRal and DRal^S25N in the trichogen and hair cells on the nota. The phenotype of the hairs on the nota was similar to that of the wing hairs (Fig. 3). They were often shortened, forked, twisted, duplicated, or triplicated (Fig. 7, A and C). As for the bristles, the GAL4-69B line expressing DRal^S25N resulted in a similar phenotype to that caused by sca-GAL4 (i.e., some of the bristles were lost; Fig. 7, A and C). We expected that these phenotypes were caused by a dominant negative effect on the endogenous DRal protein. To address whether these phenotypes could be caused by decreased function of DRal, wild-type DRal protein was expressed with the dominant negative DRal^S25N protein. The loss of bristles and morphological defects resulting from DRal^S25N expression were largely rescued by coexpression of the wild-type DRal protein (Fig. 7, B and D). Therefore, these phenotypes are likely to have resulted from decreased DRal function.

To explore other genes associated with the DRal-induced defects described above, flies carrying both sca-GAL4 and UAS-DRal^S25N were crossed to a number of mutants for genes known to be involved in the Ras pathway and cytoskeletal regulation. The resulting F1 progenies were scored for modification of the bristle-loss phenotype caused by DRal^S25N (Table II). No effect was seen as a result of halving the dosages of the genes coding for the proteins of the Ras/Raf/ERK pathway or the Rho family of small GTPases, i.e., Ras1 (Simon et al., 1991), D-raf (Nishida et al., 1988), Dsor1 (encoding an MEK; Tsuda et al., 1993), rolled (encoding an ERK; Brunner et al., 1994), mbt (encoding a PAK; Melzig et al., 1998), RhoA (Hariharan et al., 1995; Strutt et al., 1997), DCdc42 (Luo et al., 1994; Fehon et al., 1997), DRac1 (Luo et al., 1994), and DRac2 (Luo et al., 1994; data not shown). However, we found that mutations of the genes encoding JNKK and JNK dominantly suppressed the DRal^S25N-induced bristle phenotype (Fig. 8). Two alleles of the hemipterous (hep) gene (encoding a JNKK; Glise et al., 1995) and three alleles of the basket (bsk) gene (encoding a JNK; Riesgo-Escovar et al., 1996; Sluss et al., 1996) acted as dominant suppressors of the DRal^S25N-induced bristle phenotype (Table II). These genetic interactions suggest that DRal functions in a common signal pathway with JNKK and JNK in an antagonistic fashion.

It has been shown that both the bsk and hep mutations disrupt the process of dorsal closure during embryonic development (Glise et al., 1995; Riesgo-Escovar et al., 1996; Sluss et al., 1996). Dorsal closure is a morphogenetic process in which the two sheets of lateral epidermis are elongated along their dorsoventral axes. On meeting at the dorsal midline, the two leading edges suture. If DRal functions to downregulate the JNK pathway, the constitutively active DRal protein should affect the process of dorsal closure. To induce the expression of DRal^G20V in the embryonic epidermis, UAS-DRal^G20V lines were crossed to a number of GAL4 lines such as actin-GAL4, 69B-GAL4, and arm-GAL4. Embryonic expression of DRal^G20V resulted in lethality (data not shown). In some cases, the cuticle of the embryos showed defects on the dorsal surface (data not shown), indicating that dorsal closure was defective. To test whether expression of activated DRal in the leading edge is sufficient to induce the dorsal cuticle phenotype autonomously, DRal^G20V was expressed using pnr-GAL4 (Fig. 9) and LE-GAL4 (data not shown), which target expression specifically to the leading edge. The cuticle patterns in these GAL4 lines were normal and indistinguishable from those of the wild-type embryos (data not shown). Expression of DRal^G20V in the leading edge specifically caused the appearance of large holes in the anterior or dorsal epidermis (Fig. 9, A and B). Some embryos showed a severe dorsal-open phenotype (Fig. 9 C) similar to the phenotypes caused by bsk and hep mutations (Glise et al., 1995; Riesgo-Escovar et al., 1996; Sluss et al., 1996). These results suggest that activated DRal inhibited the activation of JNK in the leading edge.

The genetic data suggested that DRal could act as a negative regulator of the JNK pathway in vivo. We next examined the ability of the constitutively active DRal protein to inhibit JNK activation when overexpressed in tissue culture cells. JNK is activated by phosphorylation on both threonine and tyrosine residues in the Thr-X-Tyr sequence within the catalytic core of the enzyme. Therefore, the level of Bsk/JNK activation in cells was evaluated on Western blots using an antibody that specifically recognizes phosphorylated JNK. S2 cells were transfected with pUAST-DRal or pUAST-DRal^G20V together with a plasmid that expresses GAL4 under control of the actin5C promoter, pWAGAL4 (Hiromi, Y., unpublished observations). DRal did not affect the basal level of Bsk phosphorylation in untreated S2 cells (data not shown). It has been shown that JNK is activated by osmotic shock (Galcheva-Gargova et al., 1994). In fact, treatment of the cells with 0.5 M d-sorbitol for 5 min resulted in an increase in Bsk phosphorylation compared with the untreated control (data not shown). Whereas expression of the wild-type DRal protein did not affect Bsk activation, the constitutively active mutant significantly inhibited the phosphorylation of Bsk (Fig. 10). To test whether the difference in the signals determined using the anti-ACTIVE JNK antibody were due to differences in the levels of Bsk protein loaded onto each lane, the blots were probed with an antibody that recognizes total JNK protein (both active and inactive forms), which showed similar signals in each lane (Fig. 10). These results suggest that DRal is an upstream negative regulator of Bsk/JNK in tissue culture cells.

We have identified a Drosophila gene, DRal, that encodes a protein with strong homology to mammalian Ral GTP-ases. The Ral proteins identified in mammals so far are easily classified into two types, RalA and RalB, based on their amino acid sequences. Although the amino acid sequence of DRal is more homologous to that of RalA, some residues of DRal, e.g., Glu-103 and Pro-135, are identical to RalB, but not to RalA. Therefore, we could not classify DRal as a homologue of either RalA or RalB. The COOH-terminal region of DRal contains a basic amino acid repeat and a CAAX motif, which are important for post-translational modifications and membrane localization. DRal may be localized to the membrane with Ras and activated by RalGDS, as shown in mammals (Hinoi et al., 1996). The sequences of the effector domains of the Drosophila (from Tyr-40 to Tyr-48) and mammalian Rals are identical, suggesting that the target molecules of Ral are also conserved. There are four domains conserved in all the small GTPases, called I, II, III, and IV. I and II are important for GTPase activity, whereas II, III, and IV are important for nucleotide binding. The sequence of DRal in domains I (from Gly-18 to Lys-24), II (from Asp-65 to Gly-68), III (from Asn-124 to Asp-127), and IV (from Glu-152 to Lys-156) are well conserved with those of mammalian Rals. The structural similarity suggests that DRal is biochemically similar to the mammalian Rals. In fact, DRal bound to GTP and GDP with high affinities and showed a low intrinsic GTPase activity. DRal responded well to mammalian RalGDS and RalGAP. Moreover, a GTPase-deficient protein that is constitutively active could be made by introducing a mutation found in human Ral (Hinoi et al., 1996). Likewise, a dominant negative mutant that displays preferential affinity for GDP could be generated by introducing the mutation at the same position as in human Ral (Hinoi et al., 1996).

Much of our knowledge about the functions of small GTPases has been obtained from studies using dominant active and dominant negative mutants. In Drosophila, ectopic expression of wild-type or mutant proteins has been successfully used to study the roles of small GTPases in development (see Luo et al., 1994; Harden et al., 1995; Hariharan et al., 1995; Eaton et al., 1996; Barrett et al., 1997; Strutt et al., 1997; Hacker and Perrimon, 1998). Since no mutant flies exist that affect DRal function at present, we have used overexpression of a dominant negative protein to investigate the biological function of DRal. The advantage of this approach is that we can control the effect of the DRal mutation spatiotemporally using the GAL4/UAS system (Brand and Perrimon, 1993). The substitution of asparagine for glycine at amino acid 17 in Ras inhibits GTP binding and sequesters the GEFs from the endogenous Ras protein (Farnsworth and Feig, 1991). Therefore, the DRal^S25N protein may also function to sequester RalGDS, thereby inhibiting the activation of the endogenous DRal protein, although a RalGDS-like protein has not been identified in Drosophila. The DRal^S25N-induced phenotype reported in this paper is likely to be due to a reduction in the activity of the endogenous DRal protein, because the phenotype is rescued by coexpression of the wild-type DRal protein.

Development of wing hairs is controlled by both actin and microtubules (Turner and Adler, 1998). A number of genes involved in wing hair formation have been identified. For example, wing cells of mutants for the tissue polarity genes such as frizzled, disheveled, prickle, fuzzy, and multiple wing hair extend more than one prehair (Wong and Adler, 1993). These genes may play an important role in restricting the initiation site for hair outgrowth. Expression of DRal^S25N also resulted in the extension of multiple prehairs from a single cell (Fig. 3). Therefore, it is possible that DRal functions to regulate prehair initiation. Moreover, close examination of the hairs with a scanning electron microscope revealed that the expression of DRal^S25N affected their structure. Wing cells that expressed DRal^S25N produced hairs that were deformed and stunted. We conclude that DRal plays essential roles in both the initiation of hair outgrowth and hair extension.

Another structure examined in this work is the external sensory bristle. The development of bristles is also an excellent model system for studying the role of the cytoskeleton in cell shape changes. The trichogen cell extends and forms a bristle shaft during early pupal development (Hartenstein and Posakony, 1989). This cytoplasmic extension of the trichogen cell contains a central core of microtubules surrounded by F-actin bundles (Overton, 1967; Appel et al., 1993; Tilney et al., 1996). Mutations in the genes encoding actin binding proteins result in aberrant bristle formation, suggesting that the actin cytoskeleton plays an important role in bristle development (Cant et al., 1994; Petersen et al., 1994; Verheyen and Cooley, 1994). Microtubules also have roles in bristle development (Turner and Adler, 1998). DRal may regulate the cytoskeleton organization in developing bristles, since the dominant negative DRal protein inhibited the initiation of bristles.

Our genetic and biochemical data suggest that DRal regulates cell shape changes through the inhibition of the JNK pathway (Figs. 8, 9, and 10; Table II). The JNK pathway has been implicated in cell shape changes and in the regulation of tissue polarity (for review see Noselli, 1998). The precise mechanisms for the regulation of JNK signaling by DRal are unknown. However, identification of RalBP1 as a putative effector protein of mammalian Ral may provide a clue to the mechanism (Cantor et al., 1995; Jullien-Flores et al., 1995; Park and Weinberg, 1995). RalBP1 acts as a GAP for CDC42 and Rac (Cantor et al., 1995; Jullien-Flores et al., 1995; Park and Weinberg, 1995). In mammalian cells, Cdc42 and Rac upregulate the JNK pathways via PAK (Coso et al., 1995). Similarly, DCdc42 and DRac1 are upstream activators of Hep/JNKK and Bsk/JNK in Drosophila (Glise and Noselli, 1997). The Drosophila homologues of PAK, DPAK and Mbt, may transduce the signal from DRac1 to the JNK pathway (Harden et al., 1996; Melzig et al., 1998). We have shown that the DRal^S25N-induced phenotype could be suppressed by halving the dosages of Hep/JNKK and Bsk/JNK. Expression of constitutively active DRal^G20V in the leading edge caused dorsal closure defects similar to those seen in JNK pathway mutants, supporting our idea that activation of DRal leads to downregulation of the JNK pathway. We also provided biochemical evidence showing that DRal could act as an upstream negative regulator of JNK activation. Consistently, the dorsal open phenotype of the bsk null mutants was not affected by expression of the dominant negative and constitutively active DRal mutants (data not shown). It is possible that DRal activates a GAP for the Cdc42 and Rac families of GTPases, resulting in a negative effect on the JNK signaling. It has recently been reported that Ral-GEFs suppress the neurite outgrowth of PC12 cells through inhibition of CDC42 and Rac (Goi et al., 1999). However, we could not detect any modifying effect of the mutations of DCdc42, DRac1, DRac2, and RhoA on the DRal^S25N-induced phenotype. One explanation for this result is that the multiple GTPases of the Rho family may have redundant functions for activating the JNK pathway. Alternatively, DRal may negatively regulate the JNK pathway independently of the known members of the Rho family.

Ras mediates its diverse biological functions by activating multiple downstream targets including GEFs for Ral (Vojtek and Der, 1998). Ras mediates its effects on cellular proliferation in part by activating Raf (reviewed by Bos, 1997; Vojtek and Der, 1998). Ras is also known to have effects on the cytoskeleton (Bar Sagi and Feramisco, 1986; Ridley and Hall, 1992). Rodriguez-Viciana et al. (1997) have reported that activation of the phosphoinositide 3-kinase, one of the Ras effectors, is essential in Rasinduced cytoskeletal rearrangement. Our data suggest that Ral, which is activated by another family of Ras effectors, the RalGEFs, also regulates the cytoskeleton through the JNK pathway, and thus plays a role in the cell shape changes that occur in animal development.

possible_ack_statement_here

We would like to thank Norbert Perrimon, Kozo Kaibuchi, and Beth Stronach for comments on the manuscript and valuable discussions. We are grateful to Richard G. Fehon, Mathew Freeman, Toshihiko Hosoya, Fumio Matsuzaki, Marek Mlodzik, Makoto Nakamura, Yasuyoshi Nishida, Masataka Okabe, Norbert Perrimon, Thomas Raabe, Gerald M. Rubin, Kuniaki Takahashi, Yoshihiro Takatsu, Minoru Tateno, and Ryu Ueda. The Bloomington Stock Center for fly stocks, Shinobu C. Fujita for the mAb 22C10, Yasushi Hiromi for the pWAGAL4 plasmid, Hiroko Kouike for sequencing, Ritsuko Shimamura for making the fly medium, and Kenichi Kimura, Ryusuke Niwa, Masataka Okabe, Kuniaki Takahashi, Minoru Tateno, and Tadashi Uemura for technical instructions.