A Genetic Screen for Modifiers of Drosophila Src42A Identifies Mutations in Egfr, rolled and a Novel Signaling Gene

Corresponding author: Xiangyi Lu, Department of Molecular Biosciences, The University of Kansas, Lawrence, KS 66045., xlu{at}kuhub.cc.ukans.edu (E-mail)

Communicating editor: T. SCHÜPBACH

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Drosophila Src42A, a close relative of the vertebrate c-Src, has been implicated in the Ras-Mapk signaling cascade. An allele of Src42A, Su(Raf)1, dominantly suppresses the lethality of partial loss-of-function Raf mutations. To isolate genes involved in the same pathway where Src42A functions, we carried out genetic screens for dominant suppressor mutations that prevented Su(Raf)1 from suppressing Raf. Thirty-six mutations representing at least five genetic loci were recovered from the second chromosome. These are Drosophila EGF Receptor (Egfr), rolled, Src42A, and two other new loci, one of which was named semang (sag). During embryogenesis, sag affects the development of the head, tail, and tracheal branches, suggesting that it participates in the pathways of Torso and DFGF-R1 receptor tyrosine kinases. sag also disrupts the embryonic peripheral nervous system. During the development of imaginal discs, sag affects two processes known to require Egfr signaling: the recruitment of photoreceptor cells and wing vein formation. Thus sag functions in several receptor tyrosine kinase (RTK)-mediated processes. In addition, sag dominantly enhances the phenotypes associated with loss-of-function Raf and rl, but suppresses those of activated Ras1^V12 mutation. This work provides the first genetic evidence that both Src42A and sag are modulators of RTK signaling.

RECEPTOR tyrosine kinases (RTKs) form a large and important class of cell surface receptors that regulate cell proliferation, differentiation, survival, and numerous other biological processes. All known RTKs stimulate the highly conserved Ras-Mapk protein phosphorylation cascade. The cascade is initiated following the activation of an RTK, which recruits the Grb2-Sos complex to the site of Ras-GDP, thereby triggering the release of GDP and formation of Ras-GTP (PAWSON 1995 ). Ras-GTP is the active form of the protein and it binds to cellular targets, the most important of which is the serine/threonine kinase Raf. The binding of Raf with Ras-GTP brings Raf to the plasma membrane, where it appears to be further activated by unidentified factors (TRAVERSE et al. 1993 ; STOKOE et al. 1994 ). Activated Raf then phosphorylates and activates the dual-specificity kinase Mek (Map kinase kinase), which in turn phosphorylates and activates Mapk (Map kinase; MARSHALL 1994 ). Activated Mapk translocates into the nucleus where it phosphorylates nuclear factors, resulting in the alteration of gene expression patterns and the elicitation of cellular responses (KARIN 1994 ; TREISMAN 1994 ).

The current excitement in the field is the identification of many branch pathway components that feed into this seemingly obligatory sequential activation cascade from Ras to Mapk. For example, mutations in kinase suppressor of Ras (ksr), identified in both Drosophila and Caenorhabditis elegans, impair RTK signaling (KORNFELD et al. 1995 ; SUNDARAM and HAN 1995 ; THERRIEN et al. 1995 ). Ksr is a ceramide-activated protein kinase that phosphorylates Raf and activates Raf kinase activity toward its substrate Mek (ZHANG et al. 1997 ). Thus, although activation of the Mapk cascade by Ras-GTP appears to be sufficient to generate many RTK-mediated responses, an endogenous RTK pathway is likely to be modified by multiple branch pathways impinging on the cascade. These branch pathway components could potentially play very significant roles. They could mediate feedback regulation of individual members of the Mapk cascade. They could also be involved in the coupling of various previously thought "independent" pathways. For example, the ceramide-activated protein kinase Ksr links the sphingomyelin pathway with the Mapk cascade at the level of Raf, bypassing Ras (ZHANG et al. 1997 ).

To investigate other possible Ras-independent means of activating the Mapk cascade, we have isolated mutations that suppress the lethality of a Drosophila Raf mutation [also referred to as l(1) pole hole], Raf^C110, which cannot interact with Ras1 (LU et al. 1994 ). Raf^C110 contains a point mutation (Arginine²¹⁷ to Leucine) in the Ras1 interaction domain (MELNICK et al. 1993 ). This mutation completely abolishes the ability of the Raf^C110 mutant protein to bind with Ras1 as tested using in vitro binding and the yeast "two hybrid" assays (HOU et al. 1995 ). Interestingly, the phenotypes of Raf^C110 mutants are substantially weaker than those of Raf null mutants (MELNICK et al. 1993 ). This could be attributed to, in part, the regulation of Raf^C110 by Ras-independent factor(s). It has been shown that the Torso (Tor) RTK can activate wild-type Raf in the complete absence of Ras1 (HOU et al. 1995 ). Similar results have also been shown for the mammalian proteins (FABIAN et al. 1994 ). The Arginine-to-Leucine mutation, when introduced in the corresponding position in the mammalian Raf1 protein, disrupts the interaction with Ras and prevents the Ras-mediated but not tyrosine kinase-mediated enzymatic activation of Raf1 (FABIAN et al. 1994 ).

Raf transduces signals from several RTKs throughout Drosophila development (PERRIMON and PERKINS 1997 ; SCHWEITZER and SHILO 1997 ). Because Raf^C110 is a partial loss-of-function mutation, the mutant animals live beyond the embryonic stage and die at the pupal stage as pharate adults. This provides a sensitized genetic background whereby suppressor mutations, which allow Raf^C110 mutants to develop further into fertile adults, have been isolated (LU et al. 1994 ). Interestingly, some Suppressors of Raf^C110, Su(Raf), appear to function without restoring the binding between Raf^C110 and Ras1. For example, Su(Raf)3 is an intragenic suppressor containing a compensatory amino acid change next to the Ras1 interaction domain on Raf^C110 (LU et al. 1994 ). Su(Raf)3 does not restore the interaction with Ras1 even though it is the strongest suppressor and it is not an activating mutation (HOU et al. 1995 ). In this case, it is possible that a decreased affinity for Ras1 may be compensated for by an increased affinity for a member of the Ras1-independent pathway (HOU et al. 1995 ).

Six extragenic Su(Raf) loci have also been identified. These mutations not only suppress Raf^C110 but also other partial loss-of-function Raf alleles that do not impair Ras-Raf binding (LU et al. 1994 ). This suggests that the suppression of Raf^C110 by the extragenic Su(Raf) mutations does not necessarily involve the restoration of Ras-Raf binding. Developmental analyses have shown that all six extragenic Su(Raf) mutations promote signaling in the Sevenless (Sev) and Egfr RTK pathways (LU et al. 1994 ). Su(Raf)34B is a gain-of-function mutation in the Dsor1 locus that encodes the fly Mek (TSUDA et al. 1993 ). We recently showed that Su(Raf)1 encodes Src42A (previously referred to as Dsrc41 by TAKAHASHI et al. 1996 ; Y. LI and X. LU, unpublished results; see MATERIALS AND METHODS). Here we report the isolation of mutations that suppress the suppressor activity of Su(Raf)1. These mutations define two known genes, Egfr and rolled (rl; also referred to as Mapk) and two previously uncharacterized loci. In addition, two alleles of Src42A were also isolated in the screen, although these mutations are not true suppressors of Su(Raf)1.

We named and characterized one of the novel suppressor loci semang (sag). sag is required during both embryonic and imaginal disc development. Mutations in sag cause zygotic lethality. To identify developmental pathways where sag functions, we have examined the phenotypes associated with sag mutations with particular attention to those processes controlled by known Drosophila RTKs. The results of these analyses show that sag participates in the Torso (Tor) and Drosophila DFGF-R1 RTK pathways during embryonic development. During imaginal disc development, sag mutations affect two processes known to require Egfr signaling: the recruitment of photoreceptor cells and wing vein formation. Thus sag functions broadly in several RTK-mediated processes. This role of sag in RTK signaling is further supported by the genetic interaction between sag and other known RTK signaling genes. sag dominantly enhances the phenotypes caused by reductions of RTK signaling in loss-of-function Raf or rl mutants. Consistent with this, sag dominantly suppresses the formation of supernumerary R7 cells caused by the activated sev-Ras1^V12 mutation (FORTINI et al. 1992 ). The sag mutations analyzed are likely loss-of-function mutations. These results suggest that sag may have a positive role in RTK signaling.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The genetics of Su(Raf)1:
This mutation was isolated as an extragenic suppressor of Raf^C110 located on the right arm of the second chromosome (LU et al. 1994 ). Flies with the genotype Raf^C110/Y die whereas those with the genotype Raf^C110/Y; Su(Raf)1/+ live. Su(Raf)1/+ heterozygotes are viable and have no obvious phenotypes. Flies with the genotype Su(Raf)1/Su(Raf)1 or Su(Raf)1/Df are lethal. Two deficiencies, Df(2R)nap⁹ (missing 42A1-2 to 42E56-F1) and Df(2R)bw^vDe2LCy^2R (missing 41A-B to 42A2-4; KERNAN et al. 1991 ), failed to complement Su(Raf)1 for viability. Thus the Su(Raf)1 locus lies within 42A1-4, a polytene interval common for these deficiencies. Within this interval, a cDNA named Src42A was isolated using a DNA fragment flanking a P-element-induced allele, l(2)k10115 (TOROK et al. 1993 ; Berkeley Genome Project), which failed to complement Su(Raf)1 for viability. The Src42A cDNA was able to rescue the lethality of both Su(Raf)1/Su(Raf)1 and Su(Raf)1/Df flies using a ubiquitous promoter (Y. LI and X. LU, unpublished results). Thus Su(Raf)1 encodes Src42A. Src42A is the same gene as Dsrc41, which was misplaced by polytene in situ hybridization (TAKAHASHI et al. 1996 ). Our results indicate that Su(Raf)1 has the characteristics of a loss-of-function mutation. However, the suppression of Raf^C110 may be attributed to a dominant negative effect of the mutation since Raf^C110/Y; Df(2R)nap⁹/+ flies are not viable.

The genetic screen:
The screen was designed to isolate suppressors of Su(Raf)1 on the second chromosome. Suppressor mutations induced on the third chromosome were not saved. Approximately 20,000 F₁ progeny were screened and 36 mutations were obtained. All mutations isolated were recessively lethal and were classified into five lethal complementation groups. Each group was then tested for complementation with known RTK signaling mutations on the second chromosome.

Figure 1 shows the genetic scheme for the mutagenesis of marked second chromosome P(w⁺, FRT)^42B (HOU et al. 1995 ). Briefly, young males (1–4 days posteclosion) were fed 20 mM ethyl methanesulfonate (EMS; Sigma Chemical Co., St. Louis) for 14 hr (LEWIS and BACHER 1968 ). The mutagenized males were mated en masse with w; Sco/CyO female virgins. Next, F₁ male progeny of genotype P(w⁺, FRT)^42B */CyO (* stands for a newly induced mutation) were crossed individually to six female virgins of genotype w Raf^C110; Su(Raf)1/+ to test if the mutagenized chromosome carried a suppressor mutation. If a particular test vial contained a large excess of the CyO-carrying F₂ males due to the death of sibling males of genotype w Raf^C110/Y; Su(Raf)1/P(w⁺, FRT)^42B *, the P(w⁺, FRT)^42B * chromosome carried, by definition, a suppressor mutation of Su(Raf)1. The suppressor-carrying chromosome was then recovered from sibling females by selecting for w⁺ on the mutagenized P(w⁺, FRT)^42B chromosome. In a separate mutagenesis, the above scheme was also used to isolate suppressors on the b pr cn wx bw chromosome.

View larger version (24K): In this window In a new window Download PPT slide   Figure 1. The rationale (A) and genetic scheme (B) for isolating suppressor mutations of Su(Raf)1. The screen used the adult viability as a selection. Su(Raf)1 suppresses the lethality of RafC110/Y. By introducing a suppressor of Su(Raf)1, Su[Su(Raf)1], which functions in common processes as those of Raf and/or Su(Raf)1, RafC110/Y; Su(Raf)1/Su[Su(Raf)1] flies are no longer viable. The screen followed the scheme shown in B using P[w+, FRT]42B (HOU et al. 1995 ) as the parental chromosome. In separate mutagenesis attempts, two other chromosomes, b pr cn wx bw and P[w+, FRT]42B Su(Raf)1, were also mutagenized (see details in MATERIALS AND METHODS).

In an attempt to isolate intragenic revertants of Su(Raf)1, males of genotype Su(Raf)1 P(w⁺, FRT)^42B/CyO were treated with EMS and mated en masse with w; Sco/CyO female virgins. Next, male progeny of genotype Su(Raf)1 P(w⁺, FRT)^42B */CyO were crossed individually to six female virgins of genotype FM7/y Raf^C110. If no y Raf^C110/Y; Su(Raf)1 P(w⁺, FRT)^42B */+ males lived, a suppressor mutation on the Su(Raf)1 P(w⁺, FRT)^42B parental chromosome was then recovered from the female siblings. However, only extragenic suppressors were isolated.

The suppressor mutations of Su(Raf)1:
Two novel suppressor loci were identified in addition to Egfr and rl. One of these was named semang (sag; Chinese for color-blind) because it affects the development of a subset of photoreceptors in the eye. The other novel locus was referred to as Su[Su(Raf)1]IV. Two Src42A alleles were also isolated; however, they are not true suppressors because Su(Raf)1/Su(Raf)1 or Su(Raf)1/Df are lethal. Suppressor mutations that did not fall in any one of the above loci were designated as Su[Su(Raf)1] followed by an individualized number (see Table 2).

The mapping and genetics of sag:
Deficiency chromosomes uncovering sag are not available. sag was mapped using a series of P(w⁺) insertions around polytene position 54B and was found to lie between two insertions l(2)k07110 (54B1-2) and l(2)k14517 (54B4-5) (TOROK et al. 1993 ; Berkeley Genome Project). The 7 sag alleles shown in Table 2 are EMS-induced alleles identified as suppressors of Su(Raf)1. Four additional alleles (2 induced by X ray and 2 induced by P-element excisions) have been identified on the basis of failing to complement sag^13L for viability (not shown). These new alleles isolated by this nonbiased noncomplementation selection also suppressed Su(Raf)1 (not shown). Because all 11 sag alleles, including 2 alleles induced by P-element excisions, can suppress Su(Raf)1, these sag mutations are most likely loss-of-function rather than gain-of-function mutations. Other evidence for this is that all embryonic and pupal phenotypes associated with sag are recessive. Because chromosome deficiencies uncovering the sag region are not available, standard genetic tests could not be performed at this time to verify that the sag mutations isolated are indeed loss-of-function mutations.

The allele sag^13L was derived from the Su[Su(Raf)1]13 chromosome that contained two recessive lethal mutations, one in the rl locus (referred to as rl^13R) and another at the sag locus (referred to as sag^13L). When separated by recombination, sag^13L showed 100% of the suppressor activity; rl^13R did not have any detectable suppressor activity. rl^13R homozygotes are viable with rough eyes weaker than rl¹ homozygotes (EBERL et al. 1993 ). Flies carrying rl^13R over a rl deficiency Df(2R)MS2¹⁰ (EBERL et al. 1993 ) have rougher eyes.

The suppressor mutation rl^41-1 failed to complement known rl mutations for viability. Flies carrying the rl^41-1 chromosome over any sag allele showed a rough eye phenotype (see RESULTS) even though rl^41-1/+ or sag/+ flies had normal eyes. To rule out the possibility that the chromosome carrying rl^41-1 may also contain a sag allele, 38 crossover products (pr rl¹ cn⁺ or pr⁺ rl^41-1 cn) were generated from females of genotype pr rl¹ cn/+ rl^41-1 +. Because sag is located at 54B1-5, these crossover products should contain either rl^41-1 or sag, but not both. These crossover products can be identified by eye colors and differences between rl¹ and rl^41-1. Flies of genotype rl¹/Df(2R)MS2¹⁰ are viable whereas those of rl^41-1/Df(2R)MS2¹⁰ are lethal. rl¹ does not suppress Su(Raf)1, but rl^41-1 does. By testing each crossover product for complementation with sag^13L and suppression of Su(Raf)1, it was found that both the rl^41-1-associated suppressor activity and rl¹ map to 55.6 cM, a location very similar to the previously reported genetic position of rl. The locus responsible for the rough eye phenotype observed in rl^41-1/sag flies cosegregated with the rl^41-1-associated suppressor activity and no sag allele was present on the rl^41-1 chromosome. Thus rl^41-1 is an unusual rl allele. Chromosomes and mutations that are not described in the text can be found in LINDSLEY and ZIMM 1992 .

Scanning electron microscopy and plastic sections:
To prepare scanning electron microscopy (SEM) samples, flies (stored in 70% ethanol at 4°) were dehydrated in 80 and 95% ethanol for 1 hr each followed by two changes of 100% ethanol for 1 hr each. Flies were then incubated with a 1:1 mixture of 100% ethanol and hexamethyldisilazane (HMDS) for 15 min, followed by two changes of pure HMDS for 15 min each. The samples were poured onto filter paper in glass petri dishes and allowed to air dry under the hood for 2 hr before mounting onto SEM stubs for examination. Plastic sections of adult eyes were performed using procedures adopted from TOMLINSON and READY 1987 . Briefly, fresh adult eyes were fixed in 0.1 M PO₄ buffer (pH 7.2) containing 2.5% glutaraldehyde and 2% OsO₄, dehydrated, embedded in EM-BED-812 kit (Cat. no. 01412; Electron Microscopy Sciences, Fort Washington, PA) and sectioned into 1.5-µm slices.

Mosaic clones:
Mosaic females carrying sag mutant germline clones were induced by the FLP-DFS technique (CHOU and PERRIMON 1992 , CHOU and PERRIMON 1996 ). Briefly, progeny from a cross between P(w⁺, FRT)^42B sag/CyO females and y w P(ry⁺; hs-FLP)¹²; P(w⁺, FRT)^42B P(w⁺, Ovo^D1)^32X9/CyO males were heat shocked at 37° for 1 hr to induce sag mutant clones in the germline. sag mutant clones in somatic tissues were also induced by the FLP technique (XU and RUBIN 1993 ). The wing clones were induced in ies of genotype y w P(ry⁺; hs-FLP)¹²; P(ry⁺; hs-neo; FRT)^42D sag/P(ry⁺; hs-neo; FRT)^42D P(ry⁺; y⁺)^44B. Yellow bristles were found at the wing margins in wings that were missing the L4 veins. The eye clones were induced in flies of genotype y w P(ry⁺; hs-FLP)¹²; P(ry⁺; hs-neo; FRT)^42D sag/P(ry⁺; hs-neo; FRT)^42D P(ry⁺; w⁺)^47A. Small patches of unpigmented sag homozygous cells were found to form disorganized ommatidial arrays. No clones in the eye or wing were recovered in control experiments either without heat-shock treatment or with heat-shock treatment in the absence of P(ry⁺; hs-FLP)¹².

Distinction between different genotypic classes of embryos:
To distinguish sag/sag from sibling sag/+ embryos derived from females carrying sag/sag germline clones, the mosaic females were crossed to males carrying sag over a balancer chromosome that contained a lacZ gene. The lacZ gene was fused to either a hunchback (hb) or an engrailed (en) promoter. Thus embryos that expressed the lacZ gene, as detected either by in situ hybridization or antibody staining, were sag/+ in genotype. Those that did not express the lacZ gene are sag/sag embryos. For antibody staining of eye imaginal discs, a sag/SM6-TM6B Tb stock was used, where SM6-TM6B is a balancer carrying Tubby (Tb). Tb causes shorter larvae and pupae, allowing sag/sag eye discs to be isolated from Tb⁺ larvae or pupae.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The specificity of the genetic screen:
Our genetic screen was designed to isolate suppressor mutations of a Src42A allele Su(Raf)1 (MATERIALS AND METHODS). Su(Raf)1 suppresses the lethality of the X-linked Raf^C110 mutation. Raf^C110/Y; +/+ mutants normally die as late-stage pupae or a few hours following emergence. In the presence of one copy of Su(Raf)1, flies of genotype Raf^C110/Y; Su(Raf)1/+ survive and are fertile (LU et al. 1994 ). A suppressor mutation of Su(Raf)1, designated as Su[Su(Raf)1], prevented Su(Raf)1 from suppressing Raf^C110 and caused the death of Raf^C110/Y; Su(Raf)1/Su[Su(Raf)1] flies (Figure 1).

Su(Raf)1 was used for the screen because it provided an intermediate level of suppression that was ideal for genetic modifying screens. Raf^C110/Y pharate adults have rough eyes with the R7 cells missing in 88% of the ommatidia. The rescued Raf^C110/Y; Su(Raf)1/+ flies still have rough eyes with the R7 cells missing in 36% of the ommatidia. Many genetic dosage-sensitive screens have been based on modification of the external appearance of the eye (SIMON et al. 1991 ; DICKSON et al. 1996 ; KARIM et al. 1996 ). Because we used lethality as the selection criteria, our screen should recover genes required in all cells and tissues that are important for the survival of the organism.

There was concern that using lethality as a genetic selection might run the risk of isolating any mutations that, as heterozygotes, decrease the general health of the organism. To test the effectiveness and the specificity of the screen, we tested many known signaling genes for suppression of Su(Raf)1 (Table 1). Our results show that the screen is highly selective for general components of the RTK pathway. Egfr, drk, and rl showed significant suppressor activities. A Ras1 deficiency did not suppress Su(Raf)1, but Ras1 point mutations did (see possible explanation in DISCUSSION; note that Dsor1 was not tested as it is on the same chromosome carrying Raf^C110). The Egfr ligand spitz (spi; RUTLEDGE et al. 1992 ) and two other genes, Star (S) and rhomboid (rho), which have been implicated in the processing of Spi (GOLEMBO et al. 1996 ), also showed significant suppressor activities. Another ligand of Efgr, vein (vn; SCHNEPP et al. 1996 ), did not show any suppressor activity. This may be because Vn is only a moderate activator of Efgr compared to Spi (SCHNEPP et al. 1998 ). Some suppressor mutations of activated tor RTK, Su(tor) (DOYLE and BISHOP 1994 ), suppressed Su(Raf)1 and some did not. The Tor RTK is a maternal component required for the development of embryonic termini. Those Su(tor) mutations that had no effect on Su(Raf)1 also did not interact with the Egfr^Elp mutation (DOYLE and BISHOP 1994 ), suggesting that their functions may be more restricted to the tor pathway. Three mutations in the TGFß/Dpp pathway did not show any suppressor activity (Table 1). Two ligands of the Notch receptor, Delta and Serrate, showed the suppressor activity, but the meaning of this interaction is not clear because none of the other neurogenic mutations tested showed any suppressor activity (Table 1).

The mutagenesis screen:
Thirty-six suppressors of Su(Raf)1 were isolated from a total of 20,000 mutagenized F₁ flies (average frequency 0.18%). All the suppressors isolated were apparently recessive lethal and most of these fell into five complementation groups (Table 2). Multiple alleles for each of the five complementation groups were isolated and all showed high suppressor activities. Seven suppressor mutations did not fall into any of the five groups and were designated as Su[Su(Raf)1] followed by an individualized number (Table 2). Some suppressors in this category showed complex complementation patterns or very low suppressor activities. For example, Su[Su(Raf)1]2 complemented all alleles of sag except sag^32-3. These single-hit mutations were not characterized further.

The classification of the suppressors of Su(Raf)1:
Eleven Su(Raf)1 suppressors failed to complement loss-of-function Egfr. All mutations of this group cause embryonic lethal phenotypes that are characteristic of loss-of-function Egfr, such as failure of germband retraction and narrowed ventral dentical bands (PRICE et al. 1989 ).

Six Su(Raf)1 suppressors failed to complement strong loss-of-function rl for viability. Transheterozygotes of any mutation in this group over the viable rl¹ allele showed bent wings and rough eyes that are characteristic of the rl locus (EBERL et al. 1993 ).

Seven Su(Raf)1 suppressors belong to a novel locus named sag located at 54B1-5. Mutations in this group complemented all known RTK mutations on the second chromosome including phyllopod (phyl), which maps to a nearby location. These are most likely loss-of-function mutations (see MATERIALS AND METHODS). The two strongest alleles (sag^13L and sag^32-3) show a 100% suppression activity, i.e., no Raf^C110/Y; Su(Raf)1 +/+ sag^13L males survive. Homozygotes or heteroallelic combinations of strong alleles die at late pupal stages. The gene sag⁺ appears to be expressed maternally. Removal of maternal sag⁺ product from the oocytes advances the lethality of homozygous mutants from the late pupal to embryonic stages. However, maternal sag⁺ function is fully replaceable by a paternal sag⁺ copy.

The fourth suppressor locus, Su[Su(Raf)1]IV, contained two alleles that caused a near complete suppression of Su(Raf)1. This locus may also be a novel signaling gene because it complemented all known RTK mutations on the second chromosome.

Two Src42A alleles, Src42A^15-1 and Src42A^18-2, were also isolated in the screens. These mutations are not true suppressors because Su(Raf)1 homozygotes and Su(Raf)1/Df hemizygotes are lethal. Like Su(Raf)1/Df hemizygotes, Src42A^15-1 and Src42A^18-2 homozygotes die as first instar larvae. The lethality associated with these two alleles was also rescued by Src42A cDNA under the control of the polyubiquitin promoter (Y. LI and X. LU, unpublished results).

Su(Raf)1 suppressors are not Su(Raf)1-specific:
We tested whether Su(Raf)1 suppressors also suppress two other strong suppressors of Raf^C110, Su(Raf)34B and Su(Raf)43B (Table 3). Su(Raf)34B encodes a partially activated fly Mek (LU et al. 1994 ). Su(Raf)43B is an uncharacterized gene. Both of these mutations were shown to upregulate RTK signaling in the sev and Egfr signaling pathways (LU et al. 1994 ). Our results showed that Egfr, rl, sag, and Su[Su(Raf)1]IV were able to suppress both Su(Raf)34B and Su(Raf)43B. Because sag and Su[Su(Raf)1]IV are not specific suppressors of Su(Raf)1, it is unlikely that these two genes encode factors specific for Src42A function, such as enzymes involved in Src myristylation. We hypothesize that the two novel suppressor loci encode general factors involved in modulating RTK signaling efficiency.

sag participates in embryonic processes controlled by several RTK pathways:
To provide evidence that the novel suppressor locus sag directly affects RTK signaling, we characterized the embryonic defects associated with sag/sag embryos derived from females carrying sag germline clones (GLC) crossed to sag/+ males. The sibling sag/+ embryos from the same cross are viable with no phenotypes. To distinguish these two classes of sibling embryos, second chromosome balancers carrying lacZ genes were used to mark the sag/+ embryos (see MATERIALS AND METHODS). The embryonic phenotypes described below were associated with sag/sag embryos derived from sag GLC eggs.

We focused on processes known to be controlled by RTKs. At the beginning of embryogenesis, the Tor RTK pathway specifies the embryonic terminal cell fates (PERRIMON et al. 1995 ). Activation of Tor at the embryonic poles triggers the Ras-Mapk signaling cascade, resulting in the expression of two transcription factors, tailless (tll) and huckebein (hkb). tll and hkb in turn activate genes required for terminal development. Whenever there are reductions of tll and hkb expression due to reduced levels of Tor signaling, deletions of terminal structures occur. In the absence of Tor, the mutant embryos lack the anterior acron and all structures posterior to abdominal segment seven (A7). In sag/sag embryos derived from sag GLC eggs crossed to sag/+ males, terminal defects similar to that of the Raf^PB26 allele (MELNICK et al. 1993 ) were observed. The head skeletal structure was collapsed; the tail region contained a partial deletion of the abdominal segment eight; the size of the anal pads was reduced and associated structures appeared abnormal (arrow in Figure 2B).

View larger version (90K): In this window In a new window Download PPT slide   Figure 2. sag mutations affect the embryonic head and tail. Dark-field photographs of cuticular pattern elements (A and B) and the in situ hybridization patterns of tll (C and D) and hkb (E and F). All embryos with in situ staining are oriented with the anterior to the left and dorsal up. The domains of tll and hkb expression are indicated as percentages of the total egg length (EL) with 0% EL at the posterior end. (A, C, and E) Wild-type embryos; note the well-differentiated cephalopharyngeal head skeleton (CS), eight abdominal segments with the eighth segment indicated as A8 and posterior spiracle (PS). (B, D, and F) sag mutant embryos lacking both the maternal and paternal sag+ gene products. Note the abnormal head skeleton, and posteriorly partial deletions of A8, the anal pad and associated structures (arrow in B). Correlated with these cuticular defects, there was ~30% reduction of both tll and hkb at the posterior end.

The expressions of tll and hkb at the posterior embryonic pole are solely activated by tor signaling, whereas the anterior expressions are also activated by the bicoid morphogene (PERRIMON et al. 1995 ). In wild-type cellular blastoderm embryos, tll is expressed posteriorly from 0 to 15% egg length (EL; 0% EL is at the posterior pole; Figure 2C); hkb is expressed posteriorly from 0 to 9% EL (Figure 2E). In sag mutant embryos, posterior tll expression was reduced to 10% EL (Figure 2D); posterior hkb expression was reduced to 6% EL (Figure 2F). In other words, there was an ~30% reduction of the posterior expressions of both tll and hkb in sag mutants. The anterior expression of these two genes appeared grossly normal, with perhaps a slight broadening of tll and a slight reduction of hkb (Figure 2D and Figure F). Consistent with this, the anterior head defect appeared variable and was only observed in 50% of the embryos. These results suggest that sag is involved in tor signaling, although the mutation blocks tor signaling to a lesser extent than a Ras1 gene deletion mutation (HOU et al. 1995 ). In embryos lacking maternal Ras1⁺, the posterior tll expression domain is reduced to 5% EL and hkb is not expressed at the posterior (HOU et al. 1995 ). The residual tll expression in the Ras1 mutant embryos reflects the functioning of the Ras1-independent pathway that activates the Mapk cascade (HOU et al. 1995 ).

Several other RTKs function following the activation of Tor at the blastoderm stages. At the late embryonic stages examined, sag/sag embryos derived from sag GLC eggs lacked gut constrictions (compare Figure 3B with 3A). This feature was used in addition to the lacZ marker to confirm the genotype of sag/sag embryos (see MATERIALS AND METHODS).

View larger version (132K): In this window In a new window Download PPT slide   Figure 3. sag mutations affect the gut, tracheal branches, and PNS neurons. (A, C, and E) Wild-type embryos; (B, D, and F) sag embryos lacking both the maternal and paternal sag+ gene products. (A and B) DIC pictures showing the normal gut constrictions (A) and the lack of them in the sag mutants (B). (C and D) The tracheal branches were visualized by antibody 2A12 that reacts with tracheal luminal antigen (MANNING and KRASNOW 1993 ). Note the normal tracheal branches of the wild-type embryo (C) were absent or disconnected in the sag embryos (D). (E and F) The PNS neurons were visualized by anti-HRP antibody. Note that most or all of the chordotonal neurons (CH) were missing in the sag embryo (arrow points to one remaining CH neuron in F). Some of the external sensory neurons (ES) were also missing. Orientation: (A and B) The anterior is to the left and dorsal is facing toward the viewer. (C and D) The anterior is to the left and dorsal up. (E and F) The anterior is down and dorsal to the left.

The DFGF-R1 RTK (breathless; REICHMAN-FRIED et al. 1994 ) is involved in directing the migration of tracheal cells to form tracheal branches. Monoclonal antibody 2A12, which reacts with an unknown tracheal luminal antigen (MANNING and KRASNOW 1993 ), was used to visualize the branches. In contrast to those of wild-type embryos (Figure 3C), the mutant tracheal branches were fragmented with incomplete connections (Figure 3D). This truncation of tracheal branches could be explained by reduced Fgfr signaling in the sag mutants.

The Egfr pathway is involved in specifying ventral ectodermal cell fates. Reduction of Egfr signaling causes deletion of the ventral-most cell types (SCHWEITZER and SHILO 1997 ). Consequently, the width of ventral dentical bands is shortened and the central nervous system (CNS) is disrupted. In sag/sag embryos derived from sag GLC eggs, the width of ventral dentical bands as well as the distance between Keilin's organs appeared grossly normal (not shown). Anti-HRP antibody staining, which labels all neurons and their processes, showed that the CNS axon tracks were also grossly normal (not shown). However, severe defects were observed in the peripheral nervous system (PNS). The PNS axons appeared to be reduced in number; most if not all chordotonal neurons (CH) were missing; and neurons of the external sensory organs (ES) were also abnormal in number and in their arrangement in clusters (compare Figure 3F with 3E).

sag is involved in Egfr RTK signaling during eye development:
RTK signaling mediates cell proliferation of imaginal discs. In strong sag zygotic mutants (sag^13L or sag^32-3) that die at pupal stage, eyes were rough and contained only ~50% of the normal number of ommatidia (Figure 4B). Third instar larval eye discs dissected from these mutants were also smaller in size (not shown), suggesting that sag may have a role in cell proliferation of the eye disc.

View larger version (92K): In this window In a new window Download PPT slide   Figure 4. sag mutations affect the eye and wing vein. (A and B) SEM and (C, D, and E) apical tangential sections of the eyes. (F and G) Pictures of wings. (A and C) Wild-type eyes; (B and D) Eyes derived from sag13L zygotic pharate adults; note that the mutant eye contained disorganized ommatidial units (B) and two to three photoreceptor cells per ommatidium were missing (D). (E) A section through a mosaic eye carrying a sag/sag clone (unpigmented) in a sag/+ background (pigmented); note that the clone showed similar photoreceptor defects as in D. (F) A wing-type wing. (G) A wing carrying sag/sag clones; note the distal L4 vein was missing.

The development of all cells in the eye requires a normal level of RTK signaling. FREEMAN 1996 has shown that several pulses of Egfr RTK signaling are involved in the sequential recruitment of all photoreceptor cells (except R8) and subsequently the cone cells and pigment cells. In addition, activation of the Sev RTK in the R7 precursors is required for the formation of the R7 cells. In partial loss-of-function signaling mutants, such as Raf^HM7 or Dsor1^XS520, reduced RTK signaling levels result in the reduction of both the R7 and outer photoreceptors (R1–6; MELNICK et al. 1993 ; KARIM et al. 1996 ). The eyes derived from sag zygotic mutants (pharate adults) were examined in cross section. All sag mutant ommatidia lacked the normal complement of photoreceptor cells and these cells were often in the state of degeneration (compare Figure 4D with 4C). To observe the eye phenotypes in the absence of cellular degeneration, mitotic clones of sag homozygous cells were generated in a sag/+ genetic background. Inside the unpigmented patches of sag^13L mutant clones, no wild-type ommatidia were found; 75% of the ommatidia lacked the R7 cell and 94% of the ommatidia lacked from one to three outer photoreceptors (Figure 4E). No degeneration was observed for the photoreceptor cells inside the clones in mosaic eyes after aging the flies for several weeks. A similar result was obtained for the sag^32-3 allele. Thus the eye defect associated with sag mutations is similar to that observed in partial loss-of-function signaling mutants, such as Raf^HM7 or Dsor1^XS520 (MELNICK et al. 1993 ; KARIM et al. 1996 ).

To examine how sag affects the development of photoreceptor cells, anti-Elav antibody was used to visualize all photoreceptor neurons as they were recruited into the preommatidial cell clusters. In third instar sag mutant eye discs, the initial formation of the R8 cell appeared normal. Subsequently, more mature clusters contained reduced numbers of Elav-positive cells (compare Figure 5B with 5A). This defect was more easily seen in the mutant pupal eye disc where ommatidial spacing became more irregular, but no further loss of photoreceptor cells was observed (Figure 5D). These results indicate that some of the photoreceptor precursor cells failed to be recruited into the clusters at the third instar stage. Since preommatidial cluster formation requires several rounds of Egfr-mediated inductive recruitment (FREEMAN 1996 ), the most likely explanation for the observed phenotype is that sag mutations impair Egfr signaling, resulting in fewer photoreceptor precursors being recruited into the ommatidia.

View larger version (155K): In this window In a new window Download PPT slide   Figure 5. Preommatidial clusters in sag mutants lacked the normal numbers of photoreceptor cells. Anti-Elav immunostaining of the eye imaginal discs from third instar larvae (A and B) and 24-hr pupae (C and D). The anti-Elav antibody stains the nuclei of all eight photoreceptors as they begin neural differentiation in an ordered sequence in the developing clusters. The clusters are progressively more mature posteriorly behind the morphogenic furrow (black arrowheads). (A and C) Wild-type discs. The mature clusters contain eight cells although only five cells (R3/4, R1/6, and R7) were seen on the focal plane. (B and D) sag mutant discs; note that two to three Elav-positive cells were missing (arrow in B; arrowhead in D). At the pupal stage, irregular spacing between ommatidial units became obvious (compare D with C), but no further cell loss in the clusters was observed in the pupal disc as compared to the third instar disc.

In sag/+ flies carrying clones of sag/sag cells in somatic tissues, defects were found in only two tissues, the eye (described above) and wing. sag clones in the wing caused a deletion of the L4 vein (Figure 4G). This vein phenotype is similar to that observed in viable loss-of-function Egfr mutants (STURTEVANT et al. 1993 ), suggesting that sag may participate in Egfr-mediated vein formation.

sag interacts genetically with Ras1, Raf, and rl:
To provide further evidence that sag has a role in RTK signaling, sag was tested for genetic interaction with other known signaling genes. First, sag dominantly enhanced the eye phenotype of weak rl mutants. In the eyes derived from rl¹/rl^13R flies, 65% of the ommatidia were normal and the remaining 35% of the ommatidia lacked the R7 cell, while outer photoreceptors were all normal (Figure 6A). In contrast, in the eyes derived from rl¹ +/rl^13L sag^13L flies, all ommatidia lacked the R7 cell and 77% of these also lacked from one to three outer photoreceptor cells (Figure 6B). Second, although eyes from rl^41-1/+ or sag^13L/+ flies are normal (not shown), the eyes derived from rl^41-1+/+ sag^13L flies exhibited disorganized ommatidia (true for all sag alleles in Table 2; Figure 6C). Sections showed that only 41% of the ommatidia were normal and the remaining 59% of the ommatidia lacked the R7 cells and/or R1–6 cells (Figure 6D).

View larger version (117K): In this window In a new window Download PPT slide   Figure 6. sag dominantly enhanced the eye phenotype of rl, but suppressed that of sev-Ras1V12. (A, B, D, E, and F) Apical tangential sections and (C) SEM of eyes. (A) rl1/rl13R; note that the R7 cell was missing in only 35% of the ommatidia and all outer photoreceptor cells were present. (B) rl13R sag13L/rl1 +; note that all ommatidia lacked the R7 cells and 77% of them also lacked one to three outer photoreceptor cells. (C and D) rl41-1+/+ sag13L; note that ommatidial units were disorganized (C). The cross section showed that 37 and 50% of the ommatidia lacked the R7 cell and outer photoreceptor cells, respectively (D). rl41-1/+ and sag13L/+ flies have normal eyes (not shown). (E) sev-Ras1V12/+. The average number of R7 cells was 2.4 ± 0.12 cells per ommatidium (n = 400). (F) sag13L/+; sev-Ras1V12/+. The average number of R7 cells was reduced to 1.48 ± 0.14 cells per ommatidium (n = 450). sag32-3 showed similar interactions with rl and sev-Ras1V12.

Third, the expression of a constitutively activated form of Ras1 under the control of the sev promoter/enhancer sequences (sev-Ras1^V12) mimics the effects of sev RTK activation (FORTINI et al. 1992 ). Because the promoter sequence also directs transcription in cone cell precursors and the mystery cells, these cells are transformed into R7-like cells in flies carrying the activated construct. Similar to mutations in many known RTK signaling genes, sag dominantly suppressed the formation of these supernumerary R7 cells caused by sev-Ras1^V12 (compare Figure 6E with 6F). The number of R7 cells was reduced from an average of 2.4 ± 0.12 cells per ommatidium in sev-Ras1^V12/+ flies to 1.48 ± 0.14 cells per ommatidium in sag^13L/+; sev-Ras1^V12/+ flies.

Fourth, Raf^HM7 mutants survive at 18–22°, but die as pupae at 29° (MELNICK et al. 1993 ). sag dominantly enhanced the lethality of Raf^HM7 and caused the complete death of Raf^HM7/Y; sag/+ flies at the permissive temperatures (Table 4). This result suggests that sag interacts genetically with RTK signaling mutations not only in the eye, but also in tissues required for survival. As expected for a gene involved in RTK signaling, sag enhances loss-of-function Raf and rl, but suppresses the activated sev-Ras1^V12 mutation. These genetic interactions provide further evidence that sag plays an important role in RTK signaling.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

RTK signaling regulates cell proliferation, cell fate determination, cell migration, and many other processes throughout Drosophila development (PERRIMON and PERKINS 1997 ; SCHWEITZER and SHILO 1997 ; KARIM and RUBIN 1998 ). Normal RTK signaling is required for the viability of the organism. The level of RTK signaling in Raf^C110 mutants is slightly below what is required for viability. The lethality of Raf^C110 is suppressed in Su(Raf)1/+ heterozygotes because the Su(Raf)1 mutation upregulates RTK signaling enough to compensate for the loss of Raf activity in Raf^C110 mutants (LU et al. 1994 ). Here we report the identification of at least four genetic loci on the second chromosome that prevent the suppression of Raf^C110 by Su(Raf)1. Two of these loci are known genes Egfr and rl, and two others are novel.

The effectiveness and target genes of the genetic screen:
To examine the types of genes targeted by the screen, we tested many existing mutations in the RTK, Notch, and TGFß/Dpp pathways. Only the genes in the RTK pathway showed consistent suppression of Su(Raf)1. This suggests that the screen is highly selective for mutations involved in RTK signaling. Interestingly, the screen appears to detect mutations in genes throughout the entire pathway. For example, mutations acting both upstream of Raf (e.g., Egfr, spi, S) and downstream of Raf (e.g., rl) showed suppression activities. It appears that our screen should recover most RTK genes whose products are limited by gene dosage. For genes whose products are abundant in the cell, a >50% reduction of the gene activity may be needed to show a suppressive effect. These genes would be missed in our screen unless relatively rare dominant-negative mutations were induced. For example, hemizygosity at the Egfr or rl locus does not cause a complete suppression (~60% suppression activity). However, some EMS-induced Egfr and rl mutations isolated in the screen show a near 100% suppression (Table 2). In the case of Egfr, it is possible that a strong Egfr suppressor allele may produce mutant receptor molecules that form nonfunctional dimers with the wild-type Egfr molecules, thereby showing greater suppressive activity than a mere 50% reduction of the gene dosage. Because of our decision to keep strong suppressors, some mutations isolated, such as rl^41-1, appear to be dominant-interfering alleles. Overall, the screen has high specificity and a broad range of gene targets.

Interpretation of the screen results:
Like the other five extragenic suppressors of Raf^C110, Su(Raf)1 not only suppresses Raf^C110 but also other partial loss-of-function Raf alleles that do not impair Ras-Raf binding. This suggests indirectly that the suppression of Raf^C110 by Su(Raf)1 does not involve the restoration of Ras-Raf binding. This would be consistent with Su(Raf)1 functioning downstream of a RTK on a branch pathway parallel to Ras1. If this were true, suppression of Raf^C110 by Su(Raf)1 would not require Ras1, but would still require a cell surface RTK such as Egfr.

On the basis of the above hypothesis, two types of genes could be mutated to cause the suppression of Su(Raf)1: (1) Genes that operate upstream of Su(Raf)1: ligand, receptor, and factors that feed into Su(Raf)1. Egfr, spi, and S fit in this class. Even though spi and S were not isolated in the screen, some spi and S alleles tested suppress Su(Raf)1 very well. (2) Genes that act downstream of Su(Raf)1 or genes that work together with Su(Raf)1 in contributing to the suppression of Raf^C110. rl, which encodes Drosophila Mapk, fits in this class. If Su(Raf)1 acts on a pathway parallel to Ras1, mutations in genes directly involved in activation of Ras1 would not cause suppression of Su(Raf)1. This could explain why Ras1 deficiency caused no suppression of Su(Raf)1 and why the Ras1 activator Sos is a very poor suppressor. However, certain Ras1 point mutations did cause significant degrees of suppression. This apparent contradiction could be because the Ras1 point mutant proteins somehow "clog up" the normal flow of signal in the pathway (i.e., a dominant interfering effect). An example of this was shown by KARIM et al. 1996 , where Raf gene deletion and simple loss-of-function mutations did not work as suppressors of activated sev-Ras1^V12 in the eye. However, a putative dominant-negative form of Raf mutation very efficiently suppressed sev-Ras1^V12.

The role of sag in RTK signal transduction:
sag suppresses not only Su(Raf)1, but also Su(Raf)34B and Su(Raf)43B. Su(Raf)34B encodes a partially activated form of Mek. While it is not known what Su(Raf)43B encodes, Su(Raf)43B has been shown to upregulate signaling levels in both the sev and Egfr pathways during eye development and oogenesis, respectively (LU et al. 1994 ). The Su(Raf)1 suppressor mutations in Egfr and rl loci are loss-of-function mutations based on their associated phenotypes. Although it is difficult to provide further verification without chromosome deficiencies, the sag mutations isolated are most likely loss-of-function mutations. This is because all the phenotypes associated with sag are recessive. All alleles isolated so far (>10 alleles), including those induced by P-element excisions (not shown), can suppress Su(Raf)1. This suggests that sag⁺ may be a positive regulator of RTK signaling, and sag mutations isolated in the screen cause reductions of RTK signaling efficiency. Consistent with this, temperature-sensitive Raf^HM7 mutants, which normally live at 18–20°, die in sag/+ genetic backgrounds at the permissive temperatures (Table 4).

Phenotypic analyses of sag have provided evidence that sag has a role in RTK signaling. Zygotic sag mutants die at late pupal stage. However, removing both maternal and zygotic sag⁺ function results in embryonic lethality. These dead embryos showed terminal defects similar to that of the partial loss-of-function Raf^PB26 allele with reduced abdominal segment eight, anal pads, and associated structures (MELNICK et al. 1993 ). In the case of strong sag alleles that fully suppress Su(Raf)1, the posterior expression domains of the two genes induced by the tor pathway, tll and hkb, were reduced by ~30%. At later embryonic stages, abnormal fragmented tracheal branching was also observed. These phenotypes suggest that sag participates in both Tor and DFGF-R1 RTK-mediated processes. However, sag does not appear to affect cell fate determination of the ventral ectoderm and the CNS, although the PNS neurons and axons were either missing or abnormal in their arrangement. Because sag is involved in two other processes controlled by Egfr signaling in the eye and wing (see below), one possible explanation for the lack of phenotypes in the ventral ectoderm could be that sag exhibits pathway specificities. The sag function is more critical for some RTK-mediated processes than for others.

The role of sag during imaginal disc development was examined by inducing mitotic sag clones in adult structural primordia. Two obvious structural defects were observed in the wing and eye. First, the distal portion of the L4 wing vein was often missing. This phenotype is similar to that observed in weak Egfr mutants (STURTEVANT et al. 1993 ). Activation of Egfr and its signaling cascade is necessary and sufficient for inducing wing veins (STURTEVANT et al. 1993 ; DIAZ-BENJUMEA and HAFEN 1994 ). The L4 vein is especially sensitive to the reductions of RTK signaling levels. It is possible that sag mutations affect vein formation by reducing Egfr signaling.

In the eye, sag mutant clones formed disorganized ommatidial arrays that contained reduced numbers of photoreceptors. Anti-Elav staining showed that two to three photoreceptor precursors failed to be recruited during preommatidial assembly at the third instar larval stage. This suggests that sag affects photoreceptor cell recruitment by reducing Egfr signaling. Mosaic analysis showed that sag is required in a cell autonomous fashion for the formation of the normal complement of photoreceptor cells. Furthermore, sag dominantly enhances the loss of the R7 cells in the viable rl¹/rl^13R flies, but suppresses the formation of supernumerary R7 cells caused by sev-Ras1^V12. These results suggest that sag is likely involved in both Egfr and Sev RTK-mediated pathways of photoreceptor cell development. Overall, the phenotypes associated with sag mutations support the conclusion that sag⁺ is required in Tor, Fgfr, and Egfr RTK pathways.

Implication of Src42A in signal transduction:
Drosophila has two other Src family members, Src64 and Tec29, both of which are involved in ring canal development during oogenesis (GUARNIERI et al. 1998 ; ROULIER et al. 1998 ). Src64 does not affect viability when mutated (GUARNIERI et al. 1998 ). The mammalian Src gene family has nine members, one of which, c-Src, has been shown to be activated by receptor tyrosine kinases (KYPTA et al. 1990 ; COURTNEIDGE et al. 1993 ). However, there is as yet no loss-of-function evidence that demonstrates the role of any Src family member in RTK signaling (BROWN and COOPER 1996 ). The isolation of Su(Raf)1 as a mutation in Src42A that restores the viability of Raf mutants and the isolation of Egfr, rl, and sag as extragenic suppressors of Su(Raf)1 provide the first in vivo evidence that both Src42A and sag are modulators of RTK signaling.

At this moment, we still do not know where Src42A and sag fit into the known RTK signaling cascade. Our unpublished results show that a Src42A cDNA driven by a ubiquitously expressing promoter rescues the lethality of both Su(Raf)1 homozygotes and Su(Raf)1/Df hemizygotes. Based on this, Su(Raf)1 has loss-of-function characteristics, suggesting that Src42A is, unexpectedly, a negative modulator of RTK signaling. On the other hand, the genetics of Su(Raf)1 suggest that the suppression of Raf^C110 may be attributed to a dominant-interfering effect because the Raf^C110 lethality was not suppressed in Src42A hemizygotes of genotype Df(2R)nap⁹/+ (see MATERIALS AND METHODS). Because of this, the role of Src42A in RTK signaling is still being investigated. However, the genetic interaction as revealed by the modifying screen suggests that Egfr and other RTKs may possibly regulate Src42A and sag, which in turn modulate the Mapk cascade.

	ACKNOWLEDGMENTS

We are grateful to the following people for their kind help: R. Kreber, D. Eberl, F. Karim, T. Schupbach, B. Dickson, E. Hafen, K. Matthew, T. Laverty, J. Rusconi, and V. Corbin for various fly stocks; G. Rubin for anti-Elav antibody. We also thank M. Melnick and D. Ruden for helpful comments on the manuscript and B. Culter at the University of Kansas for SEM operation. This work was supported by a research project grant (BE-263) from the American Cancer Society, a Basil O'Connor Starter Scholar Research Award (5-FY95-1132) from the March of Dimes Birth Defect Foundation, and a General Research Fund and an EPSCoR program from the University of Kansas to X.L.

Manuscript received June 5, 1998; Accepted for publication October 28, 1998.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ALTON, A. K., K. FECHTEL, A. L. TERRY, S. B. MEIKLE, and M. A. MUSKAVITCH, 1988  Cytogenetic definition and morphogenetic analysis of Delta, a gene affecting neurogenesis in Drosophila melanogaster.. Genetics 118:235-245[Abstract/Free Full Text].

BRAND, M. and J. A. CAMPOS-ORTEGA, 1990  Second-site modifiers of the split mutation of Notch define genes involved in neurogenesis in Drosophila melanogaster.. Roux's Arch. Dev. Biol. 198:275-285.

BROWN, M. T. and J. A. COOPER, 1996  Regulation, substrates and functions of Src. Biochim. Biophys. Acta 1287:121-149[Medline].

CHOU, T. B. and N. PERRIMON, 1992  Use of a yeast site-specific recombinase to produce female germline chimeras in Drosophila. Genetics 131:643-653[Abstract].

CHOU, T. B. and N. PERRIMON, 1996  The autosomal FLP-DFS technique for generating germline mosaics in Drosophila melanogaster.. Genetics 144:1673-1679[Abstract].

COURTNEIDGE, S. A., R. DHAND, D. PILAT, G. M. TWAMLEY, and M. D. WATERFIELD et al., 1993  Activation of Src family kinases by colony stimulating factor-1, and their association with its receptor. EMBO J. 12:943-950[Medline].

DIAZ-BENJUMEA, F. J. and E. HAFEN, 1994  The sevenless signalling cassette mediates Drosophila EGF receptor function during epidermal development. Development 120:569-578[Abstract].

DICKSON, B., J. A. STRATEN, M. DOMINGUEZ, and E. HAFEN, 1996  Mutations modulating Raf signaling in Drosophila eye development. Genetics 142:163-171[Abstract].

DOYLE, H. J. and M. J. BISHOP, 1994  Torso, a receptor tyrosine kinase required for embryonic pattern formation, shares substrates with the Sevenless and EGF-R pathways in Drosophila. Genes Dev. 7:633-646[Abstract/Free Full Text].

EBERL, D. F., J. B. DUFFY, and A. J. HILLIKER, 1993  The role of heterochromatin in the expression of a heterochromatic gene, the rolled locus of Drosophila melanogaster.. Genetics 134:277-292[Abstract].

FABIAN, J. R., A. B. VOJTEK, J. A. COOPER, and D. K. MORRISON, 1994  A single amino acid change in Raf-1 inhibits Ras binding and alters Raf-1 function. Proc. Natl. Acad. Sci. USA 13:5982-5986.

FLEMING, R. J., T. N. SCOTTGALE, R. J. DIEDERICH, and S. ARTAVANIS-TSAKONAS, 1990  The gene Serrate encodes a putative EGF-like transmembrane protein essential for proper ectodermal development in Drosophila melanogaster.. Genes Dev. 4:2188-2201[Abstract/Free Full Text].

FORTINI, M. E., M. A. SIMON, and G. M. RUBIN, 1992  Signaling by the sevenless protein tyrosine kinase is mimicked by Ras1 activation. Nature 355:559-561[Medline].

FREEMAN, M., 1996  Reiterative use of the EGF receptor triggers differentiation of all cell types in the Drosophila eye. Cell 87:651-660[Medline].

GOLEMBO, M., E. RAZ, and B. Z. SHILO, 1996  The Drosophila embryonic midline is the site of Spitz processing, and induces activation of the EGF receptor in the ventral ectoderm. Development 122:3363-3370[Abstract].

GUARNIERI, D. J., G. S. DODSON, and M. A. SIMON, 1998  Src64 regulates the localization of a Tec-family kinase required for Drosophila ring canal growth. Mol. Cell 1:831-840[Medline].

HOU, S. S., T. B. CHOU, M. MELNICK, and N. PERRIMON, 1995  The Torso receptor tyrosine kinase can activate Raf in a Ras-independent pathway. Cell 81:63-71[Medline].

KARIM, F. D. and G. M. RUBIN, 1998  Ectopic expression of activated Ras1 induces hyperplastic growth and increases cell death in Drosophila imaginal tissues. Development 125:1-9[Abstract].

KARIM, F. D., H. C. CHANG, M. THERRIEN, D. A. WASSARMAN, and T. LAVERTY et al., 1996  A screen for genes that function downstream of Ras1 during Drosophila eye development. Genetics 143:315-329[Abstract].

KARIN, M., 1994  Signal transduction from the cell surface to the nucleus through the phosphorylation of transcription factors. Curr. Opin. Cell Biol. 6:415-424[Medline].

KERNAN, M. J., M. I. KURODA, R. KREBER, B. S. BAKER, and B. GANETZKY, 1991  napts, a mutation affecting sodium channel activity in Drosophila, is an allele of mle, a regulator of X chromosome transcription. Cell 66:949-959[Medline].

KNUST, E., K. A. BREMER, H. VASSIN, A. ZIEMER, and U. TEPASS et al., 1987  The enhancer of split locus and neurogenesis in Drosophila melanogaster.. Dev. Biol. 122:262-273[Medline].

KORNFELD, K., D. B. HOM, and H. R. HORVITZ, 1995  The ksr-1 gene encodes a novel protein kinase involved in Ras-mediated signaling in C. elegans.. Cell 83:903-913[Medline].

KYPTA, R. M., Y. GOLDBERG, E. T. ULUG, and S. A. COURTNEIDGE, 1990  Association between the PDGF receptor and members of the Src family of tyrosine kinases. Cell 62:481-492[Medline].

LEHMANN, R., F. JIMENEZ, U. DIETRICH, and J. A. CAMPOS-ORTEGA, 1983  On the phenotype and development of mutants of early neurogenesis in Drosophila melanogaster.. Roux's Arch. Dev. Biol. 192:62-74.

LEWIS, E. B. and F. BACHER, 1968  Method of feeding ethyl methanesulphonate to Drosophila males. Dros. Inf. Serv. 43:193-194.

LINDSLEY, D., and G. ZIMM, 1992 The Genome of Drosophila melanogaster. Academic Press/Harcourt Brace Jovanovich, New York.

LU, X., B. MELNICK, J. C. HSU, and N. PERRIMON, 1994  Genetic and molecular analyses of mutations involved in Drosophila raf signal transduction. EMBO J. 13:2592-2599[Medline].

MANNING, G., and M. A. KRASNOW, 1993 The development of Drosophila tracheal system, pp. 609–686 in The Development of the Drosophila melanogaster, edited by MICHAEL B