INTRODUCTION

The understanding of the molecular mechanisms by which cellsrespond to extracellular stimuli to proliferate or differentiateis still fragmentary. Genetic studies in organisms such asDrosophila or C. elegans, coupled with biochemistry and tissueculture experiments, have provided a partial understanding ofthe molecular nature of signal transduction pathways, themolecules involved and potential crosstalk between pathways(e.g. McCormick, 1989; Greenwald and Rubin, 1992;Herskowitz, 1995; Hunter, 1995; Marshall, 1995).

The small GTPase p21Ras (or Ras) performs a centralfunction during many cellular responses, includingproliferation and differentiation (e.g. Marx and Kerr, 1993;Zipursky and Rubin, 1994; Marshall, 1995). In response tosignals from receptor tyrosine kinases (RTKs), Ras is activatedby guanine nucleotide exchange factors (GEFs), which convertthe inactive GDP-bound form to the active GTP-bound form.Activated Ras stimulates downstream effectors such as theserine/threonine kinase Raf, which is the first in a cascade ofkinases, the MAPK cascade, conveying the signal to thenucleus (Marshall, 1994). Ras is inactivated by GTPaseactivating proteins (GAPs), which stimulate conversion to theinactive GDP-bound form (McCormick, 1989).

The Drosophila eye is a valuable model system forelucidating Ras/MAPK signaling and identifying novelcomponents involved in the Ras pathway (reviewed inZipursky and Rubin, 1994; Wassarman et al., 1995; Freeman,1997). Ras is required for the differentiation of all Drosophilaphotoreceptors (Simon et al., 1991) and is activated by theRTKs EGF-receptor (Egfr) and Sevenless (Sev), via theadaptor Drk and the GEF Sos. Ras then recruits and activatesRaf, which initiates the MAPK cascade via Dsor1 (MEK) andthe MAPK/ERK Rolled (reviewed in Zipursky and Rubin,1994; Freeman, 1997).

Ras activity is negatively regulated during photoreceptordifferentiation by the Drosophila GAP, Gap1 (Gaul et al.,1992), but it is likely that other GAPs also negatively regulateRas. Gap1 apparently is not required during Ras signaling inthe wing, and the strongest mutations in Gap1 are only semi-lethal (Gaul et al., 1992; Diaz-Benjumea and Hafen, 1994). Incontrast, mutations in Ras cause embryonic lethality and arecell-lethal (Simon et al., 1991; homozygous clones cannot berecovered in imaginal discs). Drosophila Gap1 is homologousto vertebrate p120 RasGAP (subsequently referred to asRasGAP) only in the catalytic GTPase activating domain. Inaddition to this domain that binds Ras, RasGAP contains anSH2-SH3-SH2 domain that is required for its function and

1715Development 127, 1715-1725 (2000)Printed in Great Britain © The Company of Biologists Limited 2000DEV8674

The small GTPase Ras plays an important role in manycellular signaling processes. Ras activity is negativelyregulated by GTPase activating proteins (GAPs). It hasbeen proposed that RasGAP may also function as aneffector of Ras activity. We have identified andcharacterized the Drosophila homologue of the RasGAP-binding protein G3BP encoded by rasputin (rin). rinmutants are viable and display defects in photoreceptorrecruitment and ommatidial polarity in the eye. Mutationsin rin/G3BP genetically interact with components of theRas signaling pathway that function at the level of Ras andabove, but not with Raf/MAPK pathway components.These interactions suggest that Rin is required as an

effector in Ras signaling during eye development,supporting an effector role for RasGAP. The ommatidialpolarity phenotypes of rin are similar to those of RhoA andthe polarity genes, e.g. fz and dsh. Although rin/G3BPinteracts genetically with RhoA, affecting bothphotoreceptor differentiation and polarity, it does notinteract with the gain-of-function genotypes of fz and dsh.These data suggest that Rin is not a general component ofpolarity generation, but serves a function specific to Rasand RhoA signaling pathways.

Key words: RasGAP, Development, Ommatidial polarity, Ras, Rho,Drosophila

SUMMARY

Rasputin, the Drosophila homologue of the RasGAP SH3 binding protein,

functions in Ras- and Rho-mediated signaling

Cecilia Pazman1,*,‡, Caryl A. Mayes2,*, Manolis Fanto2,*, Susan R. Haynes1,‡ and Marek Mlodzik2,§

1LMG, NICHD, NIH, 6 Center Drive, MSC 2785, Bethesda, MD 20892, USA2EMBL, Developmental Biology Programme, Meyerhofstrasse 1, 69117 Heidelberg, Germany *These authors contributed equally and are co-first authors‡Present address: Department of Biochemistry and Molecular Biology, Uniformed Services University of the Health Sciences, 4301 Jones Bridge Road, Bethesda MD20814-4799, USA§Author for correspondence (e-mail: mlodzik@embl-heidelberg.de)

Accepted 27 January; published on WWW 21 March 2000

1716

binds a number of other proteins (Ellis et al., 1990; Moran etal., 1991; Duchesne et al., 1993; Pomerance et al., 1996).

Increasing evidence suggests that RasGAP is itself aneffector of Ras, not simply a negative regulator (reviewed by,for example, McCormick, 1989; Tocque et al., 1997). RasGAPis tyrosine-phosphorylated in fibroblast cells stimulated withEGF and in cells expressing inducible forms of v-Src or v-Fps(Ellis et al., 1990), and binds the PDGF receptor (Kazlauskaset al., 1990). Antibodies against the SH3 domain of RasGAPprevent activation of maturation-promoting factor byoncogenic Ras, without blocking the MAPK cascade(Pomerance et al., 1996) and block germinal vesiclebreakdown induced by oncogenic Ha-Ras in Xenopus oocyteswithout affecting Ha-Ras-GTPase stimulation by RasGAP(Duchesne et al., 1993). This suggests that Ras activates apathway involving the SH3 region of RasGAP, in addition tothe MAPK cascade.

Several proteins bind to the N-terminal region of RasGAPand are thus candidates for effectors. p190 binds to RasGAPSH2 domains in mitogenically stimulated and tyrosine kinasetransformed cells. RasGAP, p190 and another RasGAP-binding protein, p62, become tyrosine-phosphorylated upontransformation of cells by cytoplasmic and/or receptor tyrosinekinases (Ellis et al., 1990; Moran et al., 1991). p190 itself hasGAP activity on the Rho class of small GTPases, and inaddition has homology to a transcriptional repressor(Settleman et al., 1992).

A conformational change induced by binding of p190 couldmodulate RasGAP activity (Hu and Settleman, 1997). Bindingof two phosphorylated tyrosines on p190 to the two SH2domains of RasGAP brings these domains in close proximityto one another, allowing the SH3 domain to bind other proteins.RasGAP SH3 binding protein (G3BP) is currently the onlyprotein known to bind this region of RasGAP (Parker et al.,1996). G3BP binds to RasGAP only when Ras is active and istherefore an attractive candidate for an effector of Ras.Furthermore, G3BP possesses a phosphorylation-dependentexoribonuclease activity (Gallouzi et al., 1998), suggesting aninvolvement in the regulation of mRNA turnover in responseto RasGAP-mediated signaling.

We have identified mutations in a Drosophila homologue ofG3BP, encoded by the rasputin (rin) gene, and characterizedits function during eye development. Phenotypic analysisindicates a role for rin in ommatidial polarity generation andin Ras signaling. Genetic interactions suggest that it functionsdownstream of Ras but upstream or independently of theRaf/MAPK cascade. The ommatidial polarity phenotype of rinis similar to that of the RhoA GTPase (Strutt et al., 1997). Itsspecific genetic interaction with RhoA suggests that it acts asa link between Ras and Rho signaling.

MATERIALS AND METHODS

Fly strains and geneticsP-element line P4957 was isolated from the EMBL lethal collection(Guichet et al., 1997). The rin2 allele was generated by impreciseexcision of P4957 using standard methods. Putative excision alleleswere screened for female sterility over a small deletion in 87F,Df(3R)urd, since genetic studies suggested that rin was also requiredfor fertility (C. P. and S. R. H., unpublished; rin mutants and anti-Rin

antibody are available from S. R. H., e-mail: shaynes@usuhs.mil orM. M., e-mail: mlodzik@embl-heidelberg.de). DNA corresponding tothe rin locus was analyzed by PCR and genomic Southern blotting.One line, rin2, had a deletion of the entire coding region. The deletionextended for approx. 14 kb and removed most of the P-element, thenearby Tsr gene and part of the Hrb87F gene (Haynes et al., 1997).

The tub-rin transgene was generated by inserting a 3.2 kb cDNAcontaining the complete ORF into pCaSpeR downstream from a 2.4kb fragment containing the tubulin promoter. Transgenic lines wereobtained by standard procedures (Spradling and Rubin, 1982). Onecopy of this construct rescued the eye phenotype of several alleliccombinations. UAS-rin was generated by inserting the same 3.2 kbcDNA into pUAST (Brand and Perrimon, 1993); transformant lineswere crossed to various GAL4 driver lines.

The following Ras signaling pathway strains were used for geneticinteraction crosses: sevS11 (Basler et al., 1991), sevE4;SosJC2 (Roggeet al., 1991), sevRasV12 (Karim et al., 1996), sevRaftorY9 (Dickson etal., 1992), rlSEM (Brunner et al., 1994), Gap11-16 (Gaul et al., 1992),sevGAL4, UAS-Gap1 (gift of U. Gaul), sE-JunAsp and sE-JunAla

(Treier et al., 1995) and sevGAL4, UAS-RhoA (Strutt et al., 1997).

Histological analysisEye sections and antibody stainings of eye discs were performed asdescribed (Tomlinson and Ready, 1987). The following antibodies wereused: rat anti-ELAV (gift from G. Rubin), rat anti-Sal-r (gift from R.Barrios), rabbit anti-Bar (gift of K. Saigo), mouse anti-β-gal (Promega),all at 1:200 dilution, and fluorescently labelled secondary antibodies(Jackson Labs). Embryos were fixed and stained as described (Patel,1994), with rabbit anti-Rin (diluted 1:500; 1:200 for discs), and an HRP-coupled donkey anti-rabbit secondary antibody (diluted 1:100, fromAmersham). Pictures were processed with Adobe Photoshop.

Molecular techniquesGenomic DNA adjacent to the P-element was isolated by plasmidrescue. Using the corresponding fragment as a probe, cDNAs wereisolated from a whole disc cDNA library (gift from K. Basler), andusing portions of a rin (Nts) cDNA from R. Kelley (Kelley, 1993) asa probe from an ovarian cDNA library (Steinhauer et al., 1989).Representative cDNAs from both libraries and the cDNA from R.Kelley were sequenced by standard molecular methods. Portions of arin cDNA encoding amino acids 1-171 and 494-589 (correspondingto the conserved N-terminal domain and the RRM, respectively)were amplified by PCR using Vent DNA polymerase (NEB) andcloned separately into pET-28a (Novagen). Clones were verified bysequencing. Recombinant proteins were produced and purified onHis-Bind resin (Novagen) as described by supplier. Purified proteinswere used to generate rabbit polyclonal antisera. Antisera against bothproteins recognized the same size band on western blots; the anti-RRM antiserum gave a stronger response, and was used for allsubsequent experiments.

For western blotting, samples were resolved on 8%-16% Tris-glycine gradient gels (Novex) and transferred to nitrocellulose usinga Novex transblot apparatus. Blots were blocked in PBS, 0.05%Tween 20 and 5% nonfat dry milk, and immunoblotted with anti-Rinantibody diluted 1:5000 in PBS, 0.05% Tween 20, then incubated withHRP-conjugated donkey anti-rabbit secondary antibody (Amersham)diluted 1:50,000 in PBS, 0.05% Tween 20. Signals were detected withenhanced chemiluminescence.

RESULTS

Isolation of a mutation in rasputin as a dominantsuppressor of sev-svpThe nuclear receptor Seven-up (Svp) is expressed in andrequired for the correct specification of a subset of outer

C. Pazman and others

1717Rin/G3BP and eye development

photoreceptors in the developing eye (Mlodzik et al., 1990).Ectopic expression of svp in any photoreceptor precursor or incone cells (sev-svp) causes cell fate transformations and arough eye phenotype that is dosage-sensitive (Hiromi et al.,1993). Loss-of-function mutations in all components of theRas/MAPK pathway act as dominant suppressors of thisphenotype, whereas negative regulators of the pathway act asdominant enhancers (Begemann et al., 1995; Kramer et al.,1995). To identify other components involved in Ras signalingduring photoreceptor determination or, more specifically, genesrequired for outer photoreceptor specification we used a sev-svp strain to screen for second site modifiers. A collection ofapproximately 4,500 P-element insertions on the second andthird chromosomes (Guichet et al., 1997) was screened (notshown). Strain P4957 was identified as a dominant suppressor.The single P-element is inserted at chromosomal band 87F3-6within a known transcription unit adjacent to the squid gene(originally referred to as Next-to-squid; Kelley, 1993). We havenamed this gene rasputin (rin) and this P-allele is referred toas rin1. Both the homozygous phenotype of rin1 (see below)and the dominant suppression of sev-svp are revertable uponexcision of the P-element (see Materials and Methods).

To generate additional alleles of rin we mobilized the P-element and selected for excision events. PCR and Southernanalysis of the rin2 allele showed that the entire open readingframe had been deleted; thus rin2 is a null allele (see Materialsand Methods). Homozygous or transheterozygous rin1 and rin2

flies are viable and exhibit a mild rough eye phenotype. Whentransheterozygous to a small deficiency uncovering the region87F1-15 (Df(3R)urd), both rin alleles result in a similarphenotype (Fig. 1).

rin is required for photoreceptor recruitment andommatidial rotationDetailed analysis of the eyes of rin mutant alleles revealeddefects in two aspects of eye development. Externally sucheyes are mildly rough (not shown) with defects in bothphotoreceptor recruitment and ommatidial polarity, as apparentin sections (Fig. 1). Many ommatidia have fewer than the wild-type complement of eight photoreceptors, and some have extraphotoreceptors (Fig. 1E). Both inner and outer photoreceptorscan be affected, indicating a general role for rin duringphotoreceptor specification.

Many rin mutant ommatidia with the normal photoreceptorcomplement display polarity defects (Fig. 1), reminiscent ofthe phenotypes of tissue polarity mutations such as frizzled(Adler, 1992; Zheng et al., 1995). In wild type the arrangementof photoreceptors in ommatidia in the ventral and dorsal fieldsare mirror images (Ready et al., 1976). This polarity isestablished in the imaginal disc, concurrent with cell fatespecification. Developing ommatidial preclusters begin torotate in opposite directions on either side of the dorso-ventralmidline, the equator, finally reaching an angle of 90° relativeto their starting position. At the same time the R3/R4precursors move into asymmetric positions, giving rise toommatidia of opposite chirality on either side of the equator(Fig. 1A; Ready et al., 1976; Reifegerste and Moses, 1999;Mlodzik, 1999). rin affects all aspects of this process: thedirection and degree of rotation, and the generation of thechiral forms (see Fig. 1).

To analyze these defects during early development we

examined rin mutant eye discs. Both the recruitment andpolarity defects are evident in third instar imaginal discs (Fig.2), indicating an early requirement for rin in both processes.Developing photoreceptor cells are often missing in mutantdiscs (as visualized with anti-Elav, anti-Bar or anti-Sal-rstaining, Fig. 2). Similarly, the wild-type arrangement andorientation of ommatidial preclusters (Fig. 2C) is affected (Fig.2B,D), indicating that these are primary defects.

In addition to the eye defects (Figs 1 and 2), most rin mutantadults hold their wings out and null mutants are sterile (notshown; will be described elsewhere). It is not clear whether rinis required for embryonic or early larval functions. Northernand western analyses (see below) indicate that large amountsof maternal rin mRNA and protein are deposited in the egg,and this may be sufficient for early development.

rin encodes the Drosophila G3BP homologueGenomic DNA flanking the P-element insert in rin1 wasisolated (see Materials and Methods) and used to screen animaginal disc cDNA library. Seven independent overlappingcDNAs were recovered. Also, the original rin cDNA (referredto as Nts; Kelley, 1993) was used to isolate cDNA clones froman ovarian library. The complete nucleotide sequence of thecDNAs and the exon-intron boundaries within the genomicDNA were determined (Fig. 3A). Multiple rin transcripts weredetectable at all stages during development (Fig. 3C). Thehighest transcript levels were seen in 0-3 hour embryos,suggesting a maternal contribution to the embryo (Fig. 3C).The different transcripts result from the use of multiplepolyadenylation sites within the 3′ untranslated region (UTR)and the use of an alternative exon in the 5′ UTR. Comparisonof the sequence surrounding the P insertion with those of thecDNAs revealed that the P-element is inserted within thealternative exon in the 5′ UTR.

Conceptual translation of the ORF present in the cDNAspredicts a protein of 690 residues (Fig. 3B), sharing extensivehomology with the vertebrate p120 RasGAP SH3 bindingprotein (G3BP) (Parker et al., 1996). Mammalian G3BP wasisolated due to its ability to bind the SH3 domain of p120RasGAP in ER22 cells when Ras is in its active, GTP-boundform. Two forms of G3BP, differing mostly in their centraldomains, are found in humans and mice, and a homolog hasbeen found in S. pombe. Rin shares 40% amino acid identityand 60% homology with human G3BP1 and G3BP2 over theirentire lengths (Fig. 3B) and is therefore likely to be theDrosophila homologue of G3BP. The S. pombe protein is lessclosely related to the others (26% identity and 36% homologyto Rin), but the sequence conservation is significant. Theproteins can be divided into four domains, which areconserved to different extents. Highest conservation,suggesting an important role, is found in the N terminus in aregion with no predicted function. Each protein has amoderately well-conserved RNA recognition motif (RRM).RRM-type RNA binding domains are present in many knownRNA binding proteins (Birney et al., 1993). The RNA-bindingsurface of the RRM is a four stranded β-sheet; residues inadjacent loops may help to recognize specific RNA sequences.The RRM in Rin has two short inserts relative to the humanG3BP proteins; these are predicted to be within loop regions(Birney et al., 1993). An arginine/glycine-rich (RGG) domainis located at the C terminus of human and Drosophila Rin.

1718

RGG sequences, which have been implicated in RNA binding(Kiledjian and Dreyfuss, 1992), are found in the humanproteins The central portion of the protein is the mostdivergent, both between and within species. This acidic,

proline-rich region contains the SH3 binding domain in G3BP,and this region consists mostly of proline and glutamineresidues. Although we do not know whether Rin binds an SH3domain, it contains several PXXP motifs (double overliningin Fig. 3B), which are characteristic of SH3-binding proteins(Feller et al., 1994).

To confirm that the eye phenotypes described above (Figs 1,2) are due only to lesions in rin, we expressed a full-lengthcDNA under the control of the tubulin promoter (tub-rin) andasked whether this transgene can rescue these phenotypes. Theeye phenotypes of both rin alleles and transheterozygous

C. Pazman and others

Fig. 1. Adult eye phenotype of rin mutations. Panels show tangentialeye sections (with a schematic representation reflecting ommatidialpolarity beneath each section); anterior is left, and dorsal is up in allpanels. (A) Wild type, (B) rin1/rin1, (C) rin1/Df(3R)urd, (D) rin2/rin2.Other allelic combinations show very similar phenotypes. In wild type(A) all ommatidia are correctly oriented with respect to the equator,ventral ommatidia (red arrows) are mirror images of dorsal ones(black). In rin mutants polarity defects are evident; both the degreeand direction of rotation as well as chirality are affected.Photoreceptor recruitment defects are also observed. Examples ofhow arrows are drawn for a wild-type dorsal ommatidium and a non-chiral ommatidium are shown in (E). Black arrows, dorsal chirality;red arrows, ventral identity; green arrows, non-chiral symmetricalommatidia; black circles, ommatidia with recruitment defects (inthese polarity cannot be scored). Examples of some such ommatidiaare also shown at high magnification in E; photoreceptors of both theinner and outer subtypes are affected and are either missing or, lessfrequently, ectopic. (F) rin2/rin1; tub-rin/+. The rin eye phenotype isfully rescued by the transgene.

Fig. 2. Ommatidial polarity and photoreceptor recruitment defectsarise early in development in rin mutants. Panels show 3rd instar eyeimaginal discs; anterior is left and dorsal up. Ventral fields are shownexcept A, which shows an equatorial region; the mid-furrow (MF) isindicated with white triangle. (A) rin1 homozygous disc stained forElav (in red; marking photoreceptor nuclei), revealing defects inphotoreceptor number and organization. Preclusters are disorganisedand often have a reduced number of Elav-positive cells (arrowheads).(B) rin2 disc stained with anti-Elav (red) and anti-Bar (green; visibleas yellow in overlay; marks the R1/R6 pair), highlighting theorientation of the clusters. Note misoriented clusters; wild-typeexamples are indicated with blue arrows, whereas some clusters withabnormal polarity are marked with white arrows. Note also theclusters with missing Bar-positive cells (arrowheads). (C) Wild-typedisc and (D) rin1/Df(3R)urd disc stained with anti-Spalt-related (Sal-r, in red) marking the R3/R4 precursor pair and highlightingommatidial orientation from its earliest appearance. Blue arrowsindicate the axis of wild-type clusters (in C,D), white arrows markmisoriented examples (in D). The green channel (in A and D) is anti-Rin staining in the rin1 allele: note that Rin expression is very muchreduced (compare to Fig. 4 for wild type; due to the low Rin amountsno apical enrichment as seen in wild type is detected in the mutantdiscs, and the Rin amount in these discs is 10% or less).

1719Rin/G3BP and eye development

animals were fully rescued (Fig. 1F and not shown), indicatingthat this cDNA encodes the gene affected in the mutants.

Rin is cytosolic and localised apically inphotoreceptor precursorsWe cloned sequences encoding the Rin RRM domain into anexpression vector and generated polyclonal antisera (seeMaterials and Methods). Rin is expressed throughout

development (Fig. 4A). Consistent with the abundant maternaltranscripts, Rin is highly expressed in early embryos (Fig.4A,B). In rin1 adults and eye imaginal discs protein levels aresignificantly reduced as compared to wild type (compare Fig.4A, lanes 7 and 6, and Figs 4E,F with 2A,D). Rin is cytosolicand exluded from the nucleus in all tissues analyzed (Fig. 4and not shown). In addition, Rin is localised apically in thedeveloping cells in the eye imaginal disc (Fig. 4D), similar to

Fig. 3. rin sequence and expression duringdevelopment. (A) Organization of the rinlocus. The genomic region of rin in 87F andthe locations of the rin1 and rin2 mutationsare shown. The 0 point corresponds to 0 onthe genomic map of Kelley (1993);orientation is reversed relative to that map.The splicing pattern of rin transcripts isshown; coding and noncoding regions areindicated by thick and thin lines,respectively. The splicing pattern wasdetermined by comparison of cDNA clonesand a Drosophila EST clone (GenBank#AA440649) with the genomic DNA.Arrow indicates the location and directionof the sqd transcription unit. Dashed lineindicates DNA deleted in rin2.(B) Comparison of the Drosophila Rinprotein sequence with two forms of G3BPfrom humans and a related sequence from S.pombe. The sequences were aligned usingthe ClustalW1.7 software (Thompson et al.,1994), and optimized by eye. Identicalamino acids are indicated by shading. Theconserved N-terminal region is boxed, theRRM is underlined, and the PxxP motifs inthe center of the Drosophila protein aremarked by double overlining. Rin contains690 residues; G3BP1 (Parker et al., 1996)466 residues and G3BP2 (GenBank #AF053535) 449 residues; S. pombe 434residues. (C) Northern blot probed with aportion of a rin cDNA clone showing fourtranscripts, ranging in size from 2.3 to 4.7kb. Each lane contains 1 µg of poly(A)+

RNA from the indicated stages.Approximate sizes are indicated on left; thebroadness of the bands suggests that each isa mixture of transcripts containing andlacking the 167 nt alternative exon.Equivalence of loading was determined byprobing with the ribosomal protein generp49. All lanes were approximately equal,with the exception of late pupae and adultfemales, for which less RNA was loaded.

B

A

C

1720

the localisation of Ras pathway proteins such as Drk, Sos andDos (Olivier et al., 1993; Karlovich et al., 1995; Raabe et al.,1996; see below).

Overexpression of Rin causes defects in ommatidialpolarity and photoreceptor recruitmentTo further characterize the role of rin during eye development,we overexpressed Rin in the developing eye using theGAL4/UAS system (Brand and Perrimon, 1993). UAS-rin wasexpressed under the control of the sevenless-GAL4 driver (inR3/4, R1/6, R7, the mystery and cone cells; Basler et al., 1989)

(Fig. 5). At 25°C such eyes showed extensive roughening, andphotoreceptor recruitment was affected in many ommatidia(Fig. 5B); there were missing and extra photoreceptors of boththe inner and outer sub-types. In ommatidia in whichorientation can be scored (those with the normal photoreceptorcomplement) polarity is severely affected. Many ommatidia aremisrotated and/or display a non-chiral arrangement. Inaddition, defects in rhabdomere morphology are often present.Consistent with the temperature sensitivity of GAL4-drivenexpression, sev-GAL4, UAS-rin flies raised at 18°C had lessseverely rough eyes, and all phenotypic aspects are weaker inflies raised at 18°C (Fig. 5A). In such eyes the equator is stillevident, with mostly the correct chiral forms in each eye field,but many ommatidia are misrotated (Fig. 5A).

To test whether these phenotypes are primary defects, weanalyzed sevGAL4, UAS-Rin eye discs. The recruitment andpolarity defects are evident in early third instar discs (Fig. 6;visualized with anti-Elav and anti-β-gal in an svp-lacZ line; svpis expressed in R3/R4 and later also in R1/R6; Mlodzik et al.,1990; Fanto et al., 1998). Developing photoreceptor cells areoften missing (examples are indicated with arrowheads) andthe regular wild-type arrangement and orientation ofommatidial preclusters in mutant discs is affected (asvisualized with svp-lacZ, Fig. 6). These data indicate that thesephenotypes are primary defects.

The observation that the loss- and gain-of-function polarityphenotypes of rin are similar is reminiscent of the tissuepolarity genes, where either loss of a protein or overexpression

C. Pazman and others

Table 1. rin interactions with components of the Raspathway

Genotype rin1 rin2

sevS11 Su SusevE4;SosJC2 Su SusevRasV12 Su1 Su1

sevRaftorY9 −2 −2

rlSEM −3 −3

Gap11-16 − n.d.sevGAL4, UASGap1 − −

Su, suppression; –, no visible interaction; n.d., not done.Flies were heterozygous for the indicated rin mutation and the dosage-

sensitive gene of interest, from crosses grown at constant temperature. Onaverage 3-7 sectioned eyes were analyzed statistically for each genotype.Mutations in rin dominantly suppress activated sevenless (sevS11): averagenumber of R7s per ommatidium in sevS11;+ is 3.38±0.26 (mean ± s.d, n=6eyes), in sevS11; rin1 is 2.71±0.23 (n=7 eyes) and in sevS11; rin2 is 2.32±0.11(n=5 eyes). rin mutations relieve the suppression of R7 loss in sevE4;SosJC2.The percentage of ommatidia with an R7 cell is 28.7% in sevE4; SosJC2; +/+(n=411 ommatidia), 13.9% in sevE4; SosJC2/+; rin1/+ (n=293 ommatidia) and18.2% in sevE4; SosJC2/+; rin2 /+ (n=545 ommatidia).

1The sevRasV12 allele used was CR2 (Karim et al., 1996); suppression of adifferent sevRasV12 alleles was also observed. No statistical analysis waspossible with these genotypes as the unsuppressed phenotype is too severe toallow photoreceptor counting.

2No suppression of sevRaftorY9 was seen either externally or upon statisticalanalysis.

3A very weak, statistically insignificant, effect on rlSEM was observed uponanalysis of sections [average number of R7s per ommatidium 3.47±0.24 inrlSEM/+ (n=3 eyes) as compared to 3.13±0.22 in rlSEM/+; rin1/+ (n=3 eyes)and 3.42±0.3 in rlSEM/+; rin2/+ (n=3 eyes)]; no suppression was obviousexternally.

No interaction was observed with either Gap11-16 or with the phenotypecaused by overexpression of Gap1, sevGAL4, UASGap1. No interaction wasobserved with any of the putative nuclear targets of Ras signaling.Fig. 4. Expression of Rin. (A) Western blot analysis of Rin. Lanes 1-5,

developmental expression profile; (1) 0-12 hour embryo, (2) 1st and2nd instar larvae, (3) early pupae, (4) late pupae, (5) adult fly. Aprotein of about 75 kDa is recognized, in agreement with the predictedsize (74895). Lanes 6-9, Rin expression in adult flies of the followinggenotypes: (6) wild type, (7) rin1, (8) rin2, (9) rin2/Df(3R)urd. Rinprotein is reduced in rin1 and absent in rin2 flies. Balanced loading wasverified by determining protein concentration (Biorad assay) andstaining the blot with Ponceau S before antibody incubation. Each lanecontains total protein equivalent to one whole adult male. (B) Rinembryonic expression, detected with an HRP-coupled secondaryantibody). Note the ubiquitous cytoplasmic expression.(C-F) Immunofluorescence images of a wild-type 3rd instar eyeimaginal discs. Rin (green channel) and Elav (C,F) or BP104 (D) areshown (red channel; marking all photoreceptors). Rin is expressed inall cells of the disc (C) including cells of the peripodial membrane;(E,F) at higher magnification show the cytoplasmic localisation of Rin(E), which is excluded from the nuclei (overlay with Elav staining, F).An XZ-section through a disc (D) highlights the subcellular apicalconcentration of Rin.

1721Rin/G3BP and eye development

cause similar defects (Krasnow et al., 1995; Strutt et al., 1997).Similarly, the rhabdomere morphology defects are also presentin the stronger RhoA gain- and loss-of-function alleles (Struttet al., 1997). The photoreceptor recruitment defects observedin the gain-of-function situation are consistent with a role inRas-mediated photoreceptor induction (see below).

Mutations in rin genetically interact with Ras andRho signalingThe high homology of Rin to vertebrate G3BPs, the defects inphotoreceptor recruitment and the subcellular apicalenrichment of Rin (Fig. 4D) suggested an involvement of rinin Ras signaling. To test this further we analyzed geneticinteractions between Ras pathway components and rin.Constitutive activation of Ras signaling components during eyedevelopment, by expression of activated alleles under thecontrol of the sev-enhancer, causes the recruitment of ectopicphotoreceptors (mainly a transformation of cone cellprecursors to R-cells) and an externally rough eye (Basler etal., 1991; Dickson et al., 1992; Fortini et al., 1992; Brunner etal., 1994). Reduction of the dose of a gene required for thiscell-fate transformation often reduces the number of extraphotoreceptors and visibly alters thephenotype. This approach has been usedsuccessfully in genetic screens fordownstream targets of known componentsof Ras signaling (reviewed in Zipurskyand Rubin, 1994; Wassarman et al., 1995;Freeman, 1997).

Strikingly, rin interacts with thosemembers of the pathway that function atthe level of Ras and above and not withdownstream components such as Raf (Fig.7, Table 1). Loss of one copy of rindominantly suppresses the phenotypescaused by the sevenless RTK (sevS11)and sevRasV12 (Fig. 6A,B and C-F,respectively), and relieves the partialrescue of sevE4 by SosJC2 (Table 1), a gain-of-function mutation in the GEF (Roggeet al., 1991). In contrast, rin dosagedoes not affect the phenotype causedby activated Raf (sevRaftorY9) or otherdownstream components analysed (Table1). Additionally, there is no dosage effectof rin mutations on the phenotype causedby mutations in Gap1 or on the phenotypecaused by overexpression of Gap1(sevGAL4, UAS-Gap1; our unpublisheddata). These data support the notionthat Rin functions in the context ofRas signaling. The specificity of theinteractions suggests that Rin actsupstream of or in a parallel pathway to Raf(see Discussion).

The defects in ommatidial polarity inrin mutants and upon Rin overexpressionsuggested an involvement of Rin in planarpolarity (Fz/Dsh) signaling. Polarityspecific gain-of function genotypes suchas sev-Fz and sev-Dsh interact genetically

with other components of planar polarity signaling (Strutt etal., 1997; Boutros et al., 1998). Also it has been shown thatNotch (N) signaling plays a specific role in polarityestablishment in the eye downstream of Frizzled (Cooper andBray, 1999; Fanto and Mlodzik, 1999). To ask whether rininteracts genetically with these genotypes, we tested whetherrin mutations affect the sev-Fz, sev-Dsh, sev-N* and sev-RhoAphenotypes. Whereas rin mutations do not affect the gain-of-function polarity phenotypes of fz, dsh and N (neitherexternally nor in sections, not shown), they dominantlyenhance the eye phenotype caused by overexpression of wild-type and activated RhoA (sev-RhoA and sev-RhoAV14; Fig. 8and not shown). rin mutants affect both aspects of the RhoAphenotype, photoreceptor differentiation and polaritygeneration. These data suggest that rin is not a commoncomponent of polarity signaling but appears to have a generalrequirement for RhoA function.

DISCUSSION

We have characterized the rasputin (rin) gene, encoding the

Fig. 5. Overexpression of Rin causes recruitment and polarity defects. Tangential sections ofeyes from sevGAL4, UAS-rin flies, which were raised at 18°C (A) or 25°C (B), are shown onleft with schematic diagrams on the right. Arrows are as in Fig. 1; green circles representommatidia with recruitment defects. Anterior is left and dorsal is up. The defects are moresevere in the anterior region of the eye.

1722

Drosophila homologue of vertebrate G3BP, a p120 RasGAPSH3 domain binding protein. Mutations in rin lead to defectsin eye development, affecting both photoreceptor recruitmentand ommatidial polarity, resembling phenotypes of Ras1 andRhoA mutants. These observations and geneticinteraction data support a role for Rin in Ras-mediated signaling. Mutations in rin dominantlysuppress phenotypes caused by activatedcomponents of the Ras pathway, acting at the levelof Ras and above. We originally isolated a rin allelein a screen for genes interacting with sev-svp. Allcomponents of the Ras pathway also modify sev-svp (Begemann et al., 1995), thus this interaction isconsistent with a role for rin in Ras signaling.

Rin as a potential Ras effectorRas activation is required for the development ofall photoreceptors (for reviews see Wassarman etal., 1995; Freeman, 1997). The cytoplasmic

components of Ras signaling colocalise apically indifferentiating photoreceptors (Olivier et al., 1993; Karlovichet al., 1995; Raabe et al., 1996). We find that Rin/G3BP is alsoenriched apically, consistent with a role in Ras signaling.

The Drosophila GAP, Gap1, negatively regulates Ras duringeye development (Gaul et al., 1992). However, Gap1 does notcontain the SH2-SH3-SH2 domains found in the vertebratep120 RasGAP. Therefore if Rin associates with the SH3domain of RasGAP, it does so with a GAP protein other thanGap1. Consistent with this, our genetic studies provide noevidence for an interaction between rin and Gap1: rinmutations affect neither the Gap1 mutant phenotype nor thatcaused by Gap1 overexpression. A candidate for another GAPis the recently described Drosophila RasGAP gene, whichcontains an SH2-SH3-SH2 region similar to that found in themammalian protein (Feldmann et al., 1999). Mutations in thisgene are not currently available, however, and thus the role ofthis RasGAP remains to be elucidated.

Our in vivo analyses suggest a number of possible

C. Pazman and others

Fig. 6. sevGAL4, UAS-Rin induces early polarity and recruitmentdefects. 3rd instar eye maginal discs double-stained for anti-Elav(red) and anti-β-gal in svp-lacZ (green) are shown; svp is expressedin the R3/4 pair and later also in R1/R6. Anterior is left and dorsalup; a field from the dorsal half of a disc is shown in both panels, andthe equator is below panel edge. (A) shows a field through the entirepost-furrow (MF indicated with white triangle) region of a disc,(B) shows a smaller field at higher magnification. The overlay ofElav (red) and β-gal (green) staining is yellow. Two clusters withnormal photoreceptor complement are circled in B and in one the R-cells are marked with the appropriate numbers. Ommatidial clustersare often missing photoreceptor cells (examples are indicated witharrowheads). The regular arrangement and orientation of ommatidialpreclusters is affected (as visualized with svp-lacZ, green channel).Examples with wild-type orientation are indicated with yellowarrows; examples with abnormal orientation are marked with whitearrows.

Fig. 7. rin mutations dominantly interact with Rassignaling. (A-F) Eye phenotypes of activated Raspathway components in the presence and absence ofrin mutations. (A) sevS11 (gain-of-functionsevenless), (B) sevS11; rin1/+, (C,E) sevRasV12/+;(D,F) sevRasV12/+; rin1/+. Activated Sev and Ras aredominantly suppressed by rin. For statistical analysissee legend to Table 1.

1723Rin/G3BP and eye development

scenarios for how Rin/G3BP is involved in Ras signaling.First, the association of Rin with RasGAP could simplyinterfere with the catalytic RasGAP activity on Ras.Alternatively, Rin could be an effector of a signal that comesfrom Ras via RasGAP and is required for Raf activation. Ourdata support a more complex third model in which Rin wouldact as an effector of Ras via RasGAP in a manner that isindependent of the Raf/MAPK cascade (Fig. 9). We speculatethat Rin binds the SH3 domain of RasGAP when Ras is active(as does human G3BP) and exerts an effect itself, or itinterferes with the association of RasGAP with other effector

proteins. This model is supported by the data: (1) rin doesnot interact with the Raf/MAPK cascade downstream of Ras,yet does suppress the effect of Ras activation; (2) RasV12 isinsensitive to GAP activity (John et al., 1988) and thus rinshould not have an effect on sevRasV12 if its only effect wason RasGAP catalytic activity; (3) both Ras and Raf arerequired for proliferation of imaginal disc cells, whereas rin−

tissue proliferates. Although our results do not rule out anyof these models, we favour the last model, because of theassociation of G3BP with RasGAP in mammalian cells(Parker et al., 1996), the effector function proposed forRasGAP and the above mentioned differences in Raf and Rinrequirements.

The role of Rin in Rho signalingMutations in rin and RhoA (Strutt et al., 1997) cause similardefects during eye development, and they interact genetically.The enhancement of RhoA gain-of-function by rin is veryspecific (no other gene tested interacted with RhoA) andaffects both aspects of the RhoA phenotype, photoreceptordifferentiation and planar polarity. Moreover, rin does notinteract with other genotypes tested, including the sev-Fz andsev-Dsh polarity specific phenotypes, and sev-N*, whichaffects both photoreceptor recruitment and polarity (Fortini etal., 1993; Cooper and Bray, 1999; Fanto and Mlodzik, 1999).Thus, the interactions of rin with Ras and RhoA are veryspecific and restricted to these factors.

Could these interactions share a common reason? Polaritydetermination and ommatidial rotation occur concurrently withphotoreceptor recruitment. The respective signals are beingtransduced in the same cells within a short period of time. It istherefore conceivable that some genes are involved in bothpathways or are necessary to insulate them from each other.RhoA itself is required for both processes: strong alleles showdefects in photoreceptor recruitment and differentiation, andpolarity (Strutt et al., 1997). Although components of Rassignaling do not interact with the Fz/Dsh polarity pathway(Strutt et al., 1997; Boutros et al., 1998), it remains possible

Fig. 8. rin dominantly enhances both aspects of the gain-of-functionRhoA phenotype. Tangential sections of adult eyes are shown;anterior is left, and dorsal up. (A) sev-RhoA/+ (weak), (B) sev-RhoA/rin2 (weak). The bottom panels schematize ommatidialorientation; symbols are as in Fig. 1 with the green arrowrepresenting ommatidia that are close to 180° misaligned. Noteenhancement of polarity defects. (C) sev-RhoA/+ (moderate),(D) sev-RhoA/rin2 (moderate). Note enhancement of general disorderand photoreceptor loss. Other sev-RhoA and sev-RhoAV14 genotypeswere also enhanced by rin.

Fig. 9. Schematic presentation for the potential roles of Rin in Rasand Rho signaling. See text for details.

p190

RasGAPRin/

G3BP

RasRhoA

MAPKcascade

RTKactivation

Gap1

?

response to Rasactivation

?

Rho effectors

?

?

1724

that some aspects of Ras signaling might have an influence onthe rotational aspect of planar polarity.

What could be the molecular mechanisms? Vertebrate p190that binds to the SH2 domains of p120 RasGAP is itself a GAPacting on Rho/Rac GTPases (Settleman et al., 1992). Theinvolvement of Rin in polarity establishment may thereforearise from a positive or negative interaction with a p120/p190protein complex (Fig. 9). Rin might be required for the actionof p190 on RhoA, or may interfere with binding of otherproteins required in this process. Although no interaction hasbeen detected between Ras and RhoA (not shown), rin appearsto have a role in both Ras and Rho signaling, and might act asa link between these two signaling molecules.

In this context it will be interesting to determine whether Rinbinds RNA, as suggested by its homology, and how this couldaffect Ras or RhoA signaling. Rin contains motifs found inRNA binding proteins: an RRM and an arginine-glycine-richregion (Birney et al., 1993). The effector function of Rin couldtherefore be in regulating trafficking, stability or translation ofRNAs encoding factors required during Ras/Rho signaling.Other putative RNA binding protein are also implicated insignal transduction. HnRNP-K, found in complexes withnuclear pre-mRNA (Matunis et al., 1992) and also involved intranscriptional regulation of c-myc (Michelotti et al., 1996),binds to the SH3 domains of p95vav and Src (Hobert et al.,1994; Taylor and Shalloway, 1994). Src also associates withanother RNA binding protein, p68 (Taylor and Shalloway,1994). Thus RNA binding proteins may emerge as moregeneral effectors of signal transduction.

We are most grateful to R. Kelley and U. Gaul for sharingunpublished materials and information. We thank R. Barrios, K.Basler, U. Gaul, E. Hafen, F. Kafatos, G. Rubin and U. Weber for flystocks and reagents, R. Kelley for the genomic map and cDNA clone.We are grateful to Ann-Mari Voie for embryo injections, AnnaCyrklaff for chromosome in situs, and Monica Cooper for technicalassistance. C. Blaumueller, T. Bouwmeester, L. Kockel, N. Paricio, K.Weigmann and U. Weber made helpful comments on the manuscript.C.A.M was supported by a grant from the Boehringer IngelheimFonds.

REFERENCES

Adler, P. N. (1992). The genetic control of tissue polarity in Drosophila.BioEssays 14, 735-741.

Basler, K., Christen, B. and Hafen, E. (1991). Ligand-independent activationof the sevenless receptor tyrosine kinase changes the fate of cells in thedeveloping Drosophila eye. Cell 64, 1069-1081.

Basler, K., Siegrist, P. and Hafen, E. (1989). The spatial and temporalexpression pattern of sevenless is exclusively controlled by gene-internalelements. EMBO J. 8, 2381-2386.

Begemann, G., Michon, A. M., van der Voorn, L., Wepf, R. and Mlodzik,M. (1995). The Drosophila orphan nuclear receptor Seven-up requires theRas pathway for its function in photoreceptor determination. Development121, 225-235.

Birney, E., Kumar, S. and Krainer, A. R. (1993). Analysis of the RNA-recognition motif and RS and RGG domains: Conservation in metazoan pre-mRNA splicing factors. Nucleic Acids Res. 21, 5803-5816.

Boutros, M., Paricio, N., Strutt, D. I. and Mlodzik, M. (1998). Dishevelledactivates JNK and discriminates between JNK pathways in planar polarityand wingless signaling. Cell 94, 109-118.

Brand, A. H. and Perrimon, N. (1993). Targeted gene expression as a meansof altering cell fates and generating dominant phenotypes. Development 118,401-15.

Brunner, D., Oellers, N., Szabad, J., Biggs, W. H., Zipursky, S. L. and

Hafen, E. (1994). A gain of function mutation in Drosophila MAP kinaseactivates multiple receptor tyrosine kinase signaling pathways. Cell 76, 875-888.

Cooper, M. T. D. and Bray, S. J. (1999). Frizzled regulation of Notchsignalling polarizes cell fate in the Drosophila eye. Nature 397, 526-529.

Diaz-Benjumea., F. J. and Hafen, E. (1994). The sevenless signalling cassettemediates Drosophila EGF receptor function during epidermal development.Development 120, 569-78.

Dickson, B., Sprenger, F., Morrison, D. and Hafen, E. (1992). Raf functionsdownstream of Ras1 in the sevenless signal transduction pathway. Nature360, 600-603.

Duchesne, M., Schweighoffer, F., Parker, F., Clerc, F., Frobert, Y., Thang,M. N. and Tocqué, B. (1993). Identification of the SH3 domain of GAP asan essential sequence for Ras-GAP-mediated signaling. Science 259, 525-528.

Ellis, C., Moran, M., McCormick, F. and Pawson, T. (1990).Phosphorylation of GAP and GAP-associated proteins by transforming andmitogenic tyrosine kinases. Nature 343, 377-380.

Fanto, M., Mayes, C. A. and Mlodzik, M. (1998). Linking cell-fatespecification to planar polarity: determination of the R3/R4 photoreceptorsis a prerequisite for the interpretation of the Frizzled mediated polaritysignal. Mech. Dev. 74, 51-58.

Fanto, M. and Mlodzik, M. (1999). Asymmetric Notch activation specifiesphotoreceptors R3 and R4 and planar polarity in the Drosophila eye. Nature397, 523-526.

Feldmann, P., Eicher, E. N., Leevers, S. J., Hafen, E. and Hughes, D. A.(1999). Control of growth and differentiation by Drosophila RasGAP, ahomolog of p120 Ras-GTPase-Activating protein. Mol. Cell Biol. 19, 1928-1937.

Feller, S. M., Ren, R., Hanafusa, H. and Baltimore, D. (1994). SH2 and SH3domains as molecular adhesives: the interactions of Crk and Abl. TrendsBiochem. Sci. 19, 453-458.

Fortini, M. E., Rebay, I., Caron, L. A. and Artavanis-Tsakonas, S. (1993).An activated Notch receptor blocks cell-fate commitment in the developingDrosophila eye. Nature 365, 555-557.

Fortini, M. E., Simon, M. A. and Rubin, G. M. (1992). Signalling by thesevenless protein tyrosine kinase is mimicked by Ras1 activation. Nature355, 559-561.

Freeman, M. (1997). Cell determination strategies in the Drosophila eye.Development 124, 261-270.

Gallouzi, I. E., Parker, F., Chebli, K., Maurier, F., Labourier, E., Barlat,I., Capony, J. P., Tocque, B. and Tazi, J. (1998). A novel phosphorylation-dependent RNase activity of GAP-SH3 binding protein: a potential linkbetween signal transduction and RNA stability. Mol. Cell Biol. 18, 3956-3965.

Gaul, U., Mardon, G. and Rubin, G. M. (1992). A putative Ras GTPaseactivating protein acts as a negative regulator of signaling by the sevenlessreceptor tyrosine kinase. Cell 68, 1007-1019.

Greenwald, I. and Rubin, G. M. (1992). Making a difference: the role ofcell-cell interactions in establishing seperate identities for equivalent cells.Cell 68, 271-282.

Guichet, A., Copeland, J. W. R., Erdelyi, M., Hlousek, D., Zavorszky, P.,Ho, J., Brown, S., Percival-Smith, A., Krause, H. M. and Ephrussi, A.(1997). The nuclear receptor homologue Ftz-F1 and the homeodomainprotein Ftz are mutually dependent cofactors. Nature 385, 548-552.

Haynes, S. R., Cooper, M. T., Pype, S. and Stolow, D. T. (1997). Involvementof a tissue-specific RNA recognition motif protein in Drosophilaspermatogenesis. Mol. Cell Biol. 17, 2708-2715.

Herskowitz, I. (1995). MAP kinase pathways in yeast: for mating and more.Cell 80, 187-197.

Hiromi, Y., Mlodzik, M., West, S. R., Rubin, G. M. and Goodman, C. S.(1993). Ectopic expression of seven-up causes cell fate changes duringommatidial assembly. Development 118, 1123-1135.

Hobert, O., Jallal, B., Schlessinger, J. and Ullrich, A. (1994). Novelsignaling pathway suggested by SH3 domain-mediatedp95vav/heterogeneous ribonucleoprotein K interaction. J. Biol. Chem. 269,20225-20228.

Hu, K.-Q. and Settleman, J. (1997). Tandem SH2 binding sites mediate theRasGAP-RhoGAP interaction: a conformational mechanism for SH3domain regulation. EMBO J. 16, 473-483.

Hunter, T. (1995). Protein kinases and phosphatases: the Yin and Yang ofprotein phosphorylation and signaling. Cell 80, 225-236.

John, J., Frech, M. and Wittinghofer, A. (1988). Biochemical properties ofHa-ras encoded p21 mutants and mechanisms of the autophosphorylationreaction. J. Biol. Chem. 263, 11792-11799.

C. Pazman and others

1725Rin/G3BP and eye development

Karim, F. D., Chang, H. C., Therrien, M., Wassarman, D. A., Laverty, T.and Rubin, G. M. (1996). A screen for genes that function downstream ofRas1 during Drosophila eye development. Genetics 143, 315-329.

Karlovich, C. A., Bonfini, L., Rogge, R. D., Daga, A., Czech, M. P. andBanerjee, U. (1995). In vivo analysis of the Ras exchange factor son ofsevenless. Science 268, 576-579.

Kazlauskas, A., Ellis, C., Pawson, T. and Cooper, J. A. (1990). Binding ofGAP to activated PDGF receptors. Science 247, 1578-1581.

Kelley, R. L. (1993). Initial organization of the Drosophila dorsoventral axisdepends on an RNA-binding protein encoded by the squid gene. GenesDev.7, 948-960.

Kiledjian, M. and Dreufuss, G. (1992). Primary structure and binding activityof the hnRNP U protein: binding RNA through RGG box. EMBO J. 7, 2655-2664.

Kramer, S., West, S. R. and Hiromi, Y. (1995). Cell fate control in theDrosophila retina by the orphan receptor Seven-up: its role in the decisionsmediated by the Ras signaling pathway. Development 121, 1361-1372.

Krasnow, R. E., Wong, L. L. and Adler, P. N. (1995). dishevelled is acomponent of the frizzled signalling pathway in Drosophila. Development121, 4095-4102.

Marshall, C. J. (1994). MAP kinase kinase kinase, MAP kinase kinase andMAP kinase. Curr. Opin. Genet. Dev. 4, 82-89.

Marshall, C. J. (1995). Specificity of receptor tyrosine kinase signaling:transient versus sustained extracellular signal-regulated kinase activation.Cell 80, 179-185.

Marx, J. and Kerr, R. A. (1993). Forging a path to the nucleus. Science 260,1588-1590.

Matunis, M. J., Michael, W. M. and Dreyfuss, G. (1992). Characterizationand primary structure of the poly(C)-binding heterogeneous nuclearribonucleoprotein complex K protein. Mol. Cell. Biol. 12, 164-171.

McCormick, F. (1989). ras GTPase activating protein: signal transmitter andsignal terminator. Cell 56, 5-8.

Michelotti, E. F., Michelotti, G. A., Aronsohn, A. I. and Levens, D. (1996).Heterogeneous nuclear ribonucleoprotein K is a transcription factor. Mol.Cell. Biol. 16, 2350-2360.

Mlodzik, M., Hiromi, Y., Weber, U., Goodman, C. S. and Rubin, G.M. (1990). The Drosophila seven-up gene, a member of the steroidreceptor gene superfamily, controls photoreceptor cell fates. Cell 60, 211-224.

Mlodzik, M. (1999). Planar polarity in the Drosophila eye: a multifaceted viewof signaling specificity and crosstalk. EMBO J. 18, 6873-6879

Moran, M. F., Polakis, P., McCormick, F., Pawson, T. and Ellis, C. (1991).Protein-tyrosine kinases regulate the phosphorylation, protein interactions,subcellular distribution, and activity of p21ras GTPase-activating protein.Mol. Cell Biol. 11, 1804-1812.

Olivier, J.-P., Raabe, T., Henkemeyer, M., Dickson, B., Mbamalu, G.,Margolis, B., Schlessinger, J., Hafen, E. and Pawson, T. (1993). ADrosophila SH2-SH3 adaptor protein implicated in coupling the sevenlesstyrosine kinase to an activator of Ras guanine nucleotide exchange, Sos. Cell73, 179-191.

Parker, F., Maurier, F., Delumeau, I., Duchesne, M., Faucher, D.,Debussche, L., Dugue, A. S., F. and Tocque, B. (1996). A Ras-GTPase-activating protein SH3-domain-binding protein. Mol. Cell. Biol. 16, 2561-2569.

Patel, N. H. (1994). Imaging neuronal subsets and other cell types in whole-

mount Drosophila embryos and larvae using antibody probes. Methods CellBiol. 44, 445-487.

Pomerance, M., Thang, M. N., Tocque, B. and Pierre, M. (1996). The Ras-GTPase-activating protein SH3 domain is required for Cdc2 activation andMos induction by oncogenic Ras in Xenopus oocytes independently ofmitogen-activated protein kinase activation. Mol. Cell. Biol. 16, 3179-3186.

Raabe, T., Riesgo-Escovar, J., Liu, X., Bausenwein, B. S., Deak, P., Maröy,P. and Hafen, E. (1996). DOS, a novel pleckstrin homology domain-containing protein required for signal transduction between sevenless andRas1 in Drosophila. Cell 85, 911-920.

Ready, D. F., Hanson, T. E. and Benzer, S. (1976). Development of theDrosophila retina, a neurocrystalline lattice. Dev. Biol. 53, 217-240.

Reifegerste, R. and Moses, K. (1999). The genetics of epithelial polarity andpattern in the Drosophila retina. BioEssays 21, 275-285.

Rogge, R. D., Karlovich, C. A. and Banerjee, U. (1991). Genetic dissectionof a neurodevelopmental pathway: son of sevenless functions downstreamof the sevenless and EGF receptor tyrosine kinases. Cell 64, 39-48.

Settleman, J., Narasimhan, V., Foster, L. C. and Weinberg, R. A. (1992).Molecular cloning of cDNAs encoding the GAP-associated protein p190:Implications for a signaling pathway from Ras to the nucleus. Cell 69, 539-549.

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 factorperform crucial steps in the signaling by the sevenless protein tyrosinekinase. Cell 67, 701-716.

Spradling, A. C. and Rubin, G. M. (1982). Transposition of cloned P-elements into Drosophila germ line chromosomes. Science 218, 341-347.

Steinhauer, W. R., Walsh, R. C. and Kalfayan, L. J. (1989). Sequence andstructure of the Drosophila melanogaster ovarian tumor gene and generationof an antibody specific for the ovarian tumor protein. Mol. Cell. Biol. 9,5726-5732.

Strutt, D. I., Weber, U. and Mlodzik, M. (1997). The role of RhoA in tissuepolarity and Frizzled signalling. Nature 387, 292-295.

Taylor, S. J. and Shalloway, D. (1994). An RNA-binding protein associatedwith Src through its SH2 and SH3 domains in mitosis. Nature 368, 867-874.

Thompson, J. D., Higgins, D. G. and Gibson, T. J. (1994). CLUSTAL W:improving the sensitivity of progressive multiple sequence alignmentthrough sequence weighing, position specific gap penalties and weightmatrix choice. Nucleic Acids Res. 22, 4673-4680.

Tocque, B., Delumeau, I., Parker, F., Maurier, F., Multon, M.-C. andSchweighoffer, F. (1997). Ras-GTPase activating protein (GAP): A putativeeffector for Ras. Cell. Signal. 9, 153-158.

Tomlinson, A. and Ready, D. F. (1987). Cell fate in the Drosophilaommatidium. Dev. Biol. 123, 264-275.

Treier, M., Bohmann, D. and Mlodzik, M. (1995). JUN cooperates with theETS domain protein Pointed to induce photoreceptor R7 fate in theDrosophila eye. Cell 83, 753-760.

Wassarman, D. A., Therrien, M. and Rubin, G. M. (1995). The Rassignaling pathway in Drosophila. Curr. Opin. Genet. Dev. 5, 44-50.

Zheng, L., Zhang, J. and Carthew, R. W. (1995). frizzled regulates mirror-symmetric pattern formation in the Drosophila eye. Development 121, 3045-3055.

Zipursky, S. L. and Rubin, G. M. (1994). Determination of neuronal cellfate: lessons from the R7 neuron of Drosophila. Annu. Rev. Neurosci. 17,373-397.