Analysis of Corkscrew Signaling in the Drosophila Epidermal Growth Factor Receptor Pathway During Myogenesis

Corresponding author: Lizabeth A. Perkins, Pediatric Surgical Research Labs, Rm. 2425, Massachusetts General Hospital, 114 16th St., Charlestown, MA 02129-9127., perkins{at}helix.mgh.harvard.edu (E-mail)

Communicating editor: T. SCHÜPBACH

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The Drosophila nonreceptor protein tyrosine phosphatase, Corkscrew (Csw), functions positively in multiple receptor tyrosine kinase (RTK) pathways, including signaling by the epidermal growth factor receptor (EGFR). Detailed phenotypic analyses of csw mutations have revealed that Csw activity is required in many of the same developmental processes that require EGFR function. However, it is still unclear where in the signaling hierarchy Csw functions relative to other proteins whose activities are also required downstream of the receptor. To address this issue, genetic interaction experiments were performed to place csw gene activity relative to the EGFR, spitz (spi), rhomboid (rho), daughter of sevenless (DOS), kinase-suppressor of ras (ksr), ras1, D-raf, pointed (pnt), and moleskin. We followed the EGFR-dependent formation of VA2 muscle precursor cells as a sensitive assay for these genetic interaction studies. First, we established that Csw has a positive function during mesoderm development. Second, we found that tissue-specific expression of a gain-of-function csw construct rescues loss-of-function mutations in other positive signaling genes upstream of rolled (rl)/MAPK in the EGFR pathway. Third, we were able to infer levels of EGFR signaling in various mutant backgrounds during myogenesis. This work extends previous studies of Csw during Torso and Sevenless RTK signaling to include an in-depth analysis of the role of Csw in the EGFR signaling pathway.

RECEPTOR tyrosine kinase (RTK) pathways have a role in cell growth, differentiation, and proliferation in organisms as diverse as Drosophila and humans. RTK pathways use a conserved collection of molecules to transduce their signals. Ligand binding to the RTK triggers dimerization and autophosphorylation of specific tyrosine residues on its cytoplasmic domain. The receptor phosphotyrosines enable interacting proteins to bind to the activated RTK through their Src homology 2 (SH2) domains. SH2 domain-containing proteins either transduce the RTK signal themselves or act as adapters that recruit signaling proteins lacking SH2 domains to the receptor. The adapter Drk binds the receptor via its SH2 domain while binding the guanine nucleotide exchange factor, Son of Sevenless (SOS), via its SH3 domains. SOS activates Ras1 by catalyzing GDP/GTP exchange. Gap1 and/or RasGAP negatively regulates RTK signaling by stimulating GTP/GDP exchange. Ras1 activates the serine/threonine (Ser/Thr) kinase D-Raf, which then activates the Ser/Thr kinase D-sor1 (MEK). MEK phosphorylates and activates Rl/mitogen-activated protein kinase (MAPK), which is then translocated into the nucleus to phosphorylate and activate target transcription factors (reviewed by CASCI and FREEMAN 1999 ; SCHLESSINGER 2000 ). Rl/MAPK is likely transported into the nucleus by interaction with the nuclear import protein DIM-7, encoded by the gene moleskin (msk; LORENZEN et al. 2001 ).

Along the signal transduction pathway, other molecules are also activated. Working within the Raf/MEK/MAPK cassette is kinase suppressor of ras (Ksr), a Ser/Thr kinase most similar to the Raf family of protein kinases. Genetically, Ksr is a positive transducer downstream of multiple RTK pathways. Daughter of Sevenless (DOS) encodes a multi-adapter molecule containing a pleckstrin homology domain, a poly-proline region, and tyrosine residues in consensus to bind SH2 domains of many proteins (THERRIEN et al. 1995 ; HERBST et al. 1996 ; RAABE et al. 1996 ).

Another important conserved signaling molecule active within several RTK pathways is the nonreceptor protein tyrosine phosphatase, Corkscrew (Csw). Csw was first discovered by genetic and developmental studies to be a positive signal transducer downstream of the Drosophila RTK Torso (PERKINS et al. 1992 ). Loss-of-function mutations in csw resemble partial loss-of-function mutations in torso. Additionally, genetic epistasis analysis places Csw function downstream of Torso (PERKINS et al. 1992 , PERKINS et al. 1996 ). It has been shown that upon Torso activation, Torso is phosphorylated at two major sites: tyrosine 630, Y⁶³⁰ and tyrosine 918, Y⁹¹⁸. Csw interacts with Y⁶³⁰ via its SH2 domain. The result of this interaction is phosphorylation of Csw on tyrosine 666, Y⁶⁶⁶, which is a binding site for the adapter protein Drk. Moreover, Torso does not appear to bind Drk directly and relies on Csw to act as an adapter. Additionally, the negative regulator of Ras, RasGAP, binds Y⁹¹⁸ on Torso and this site is dephosphorylated by Csw to maintain positive signaling by the Torso RTK pathway (CLEGHON et al. 1996 , CLEGHON et al. 1998 ).

Csw functions within the Drosophila epidermal growth factor receptor (EGFR) pathway, which is responsible for multiple developmental processes throughout development, including oogenesis, embryogenesis, and metamorphosis (reviewed by PERRIMON and PERKINS 1997 and SCHWEITZER and SHILO 1997 ). For this pathway there are three well-characterized activating ligands, Spitz (Spi), Vein (Vn), and Gurken (Grk), and one antagonistic ligand, Argos (Aos). Spi requires processing and presentation by two other membrane-bound proteins, Rhomboid (Rho) and Star. Vn is a neuregulin homolog made as a secreted protein. The EGFR requires Spi and Vn activity for multiple functions throughout development. Conversely, Grk is known to function specifically during oogenesis. Grk and Spi are both required during dorsal appendage patterning in follicle cells. The inhibitory ligand, Aos, a secreted molecule, is part of a negative feedback loop in EGFR signaling that is important in antagonizing the EGFR signal during oogenesis and throughout EGFR-dependent development (reviewed by CASCI and FREEMAN 1999 ).

Loss of Csw function affects the same tissues as those affected in embryos expressing loss-of-function mutations in other positive EGFR signaling pathway genes, including the EGFR itself (this report; PERKINS et al. 1996 ). We have also demonstrated that hypomorphic phenotypes associated with egfr mutations can be enhanced by reduction of one copy of wild-type csw (PERKINS et al. 1996 ). Additionally, during eye development, csw mutations presumably affect EGFR-dependent specification or differentiation of photoreceptors (ALLARD et al. 1996 ).

The Drosophila EGFR also plays an important role during embryonic mesoderm development. BUFF et al. 1998 demonstrated that loss of EGFR signaling results in loss of a subset of somatic muscle precursor cells, including the ventral acute 2 (VA2) muscle precursor. Analysis of VA2 formation, in different mutant backgrounds, provides a unique opportunity to study Csw function in the EGFR pathway at the resolution of a single cell. In this study, we show that embryos lacking Csw function variably lack VA2 precursor cells, demonstrating that Csw has a positive role in embryonic mesoderm development. Moreover, we show that expression of a gain-of-function csw construct, as shown during eye development (ALLARD et al. 1996 ), phenocopies gain-of-function mutations and constructs in other positive RTK pathway genes. Finally, the formation of VA2 precursor cells serves as a sensitive assay to infer the contribution of signaling by Csw and other molecules in the EGFR pathway.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Drosophila strains:
The following wild-type and mutant Drosophila stocks were used in this study: Oregon-R (OR), y w, csw⁶, csw^VA199, csw^LE120 (PERKINS et al. 1992 , PERKINS et al. 1996 ), spi^OE92 (RUTLEDGE et al. 1992 ), rho³⁸ (FREEMAN et al. 1992 ), pnt⁸⁸ (KLAMBT 1993 ), ksr^s628 (THERRIEN et al. 1995 ), DOS^R31 (RAABE et al. 1996 ), D-raf^11-29 (AMBROSIO et al. 1989 ), ras^C40B (LI et al. 1997 ), and msk⁵ (LORENZEN et al. 2001 ).

Transgenic strains:
The following Gal4-UAS stocks were used: 71B-Gal4 drives transgene expression to the imaginal discs (BRAND and PERRIMON 1993 ). 55B-Gal4 and CY2-Gal4 drive gene expression in the follicle cells (BRAND and PERRIMON 1993 ; QUEENAN et al. 1997 ). Mesodermal-specific expression was obtained by using a twist-Gal4 (twi-Gal4) driver (GREIG and AKAM 1993 ). The following UAS-gene X strains were used: UAS-EGFR^DNDER (dominant-negative EGFR; BUFF et al. 1998 ), UAS-EGFR^top (activated EGFR; QUEENAN et al. 1997 ), UAS-csw^WT (wild-type Csw), and UAS-csw^src90 (activated Csw), made as described below.

Molecular biology:
To make UAS-csw^WT, we started with a pBSK vector containing the csw open reading frame with a 24-nucleotide flag tag inserted immediately 3' of the initiating ATG of csw (kindly provided by V. Cleghon). We generated a BamHI fragment containing the flag-tagged csw cDNA and cloned it into the BglII site of the pUAST vector (BRAND and PERRIMON 1993 for pUAST vector). The pUAST-csw^WT (UAS-csw^WT) transformants were generated by using standard protocols (SPRADLING 1986 ). Stable lines were generated on the II and III chromosomes.

To make UAS-csw^src90, we obtained the pKB267-src-csw vector (ALLARD et al. 1996 ) and a KpnI fragment containing the csw^src90 cDNA was then cloned into the KpnI site in the pUAST vector. The pUAST-csw^src90 (UAS-csw^src90) transformants were generated using standard protocols (SPRADLING 1986 ). Stable lines were generated on the X, II, and III chromosomes. For the experiments described herein, we used UAS-csw^WT on II and UAS-csw^src90 on II and/or III. These UAS-csw^src90 lines are designated as UAS-csw^src90 (II) and UAS-csw^src90 (III), respectively.

Genetics:
Generation of germline mosaics: X-linked csw and D-raf germline clones were generated using the dominant female sterile technique as described in PERKINS et al. 1996 . ras1, ksr, and DOS autosomal germline clones were made as described in HOU et al. 1995 and CHOU and PERRIMON 1996 .

Expression of UAS-csw^src90 and twi-Gal4 in germline clone-derived embryos: To express UAS-csw^src90 in a csw/Y or D-raf/Y germline clone mosaic background, females of genotype csw FRT¹⁰¹/FM7 ftz-lacZ; UAS-csw^src90/UAS-csw^src90(II) or D-raf FRT¹⁰¹/FM7, ftz-lacZ; UAS-csw^src90/UAS-csw^src90(II) were generated using standard genetic crosses. These virgin females were crossed to males carrying the dominant female sterile mutation (ovo^D1 FRT ¹⁰¹), a heat-shock-inducible flipase (FLP³⁸), flipase recombination target (FRT) sequences, and a Balancer (Bal) on the II chromosome marked with lacZ. These males were made as follows: C(1)DX, y f/Y virgin females were crossed to +/Y; CyO, en 11 (wg-lacZ)/Sco flies. The resulting C(1)DX, y f/Y; CyO, en11 females were then crossed to ovo^D1 FRT¹⁰¹/Y flies to generate males of genotype ovo^D1 FRT¹⁰¹/Y; CyO, en11/+. Concurrently, C(1)DX, y f/Y virgin females were crossed to +/Y; nkd/MRKS, FLP⁹⁹ flies, a third chromosome balancer with a hs-FLP insertion (CHOU and PERRIMON 1996 ). Progeny of genotype C(1)DX, y f/Y; MRKS FLP⁹⁹/+ were crossed to the aforementioned ovo^D1 FRT¹⁰¹/Y; CyO, en11/+ males to yield the desired males of genotype ovo^D1 FRT¹⁰¹/Y; CyO, en11/+; MRKS, FLP⁹⁹/+. These males were crossed to either D-raf FRT¹⁰¹/FM7, ftz-lacZ; UAS-csw^src90/UAS-csw^src90 (II) or csw FRT¹⁰¹/FM7, ftz-lacZ; UAS-csw^src90/UAS-csw^src90 (II) virgin females. Germline clones were made as described elsewhere (PERKINS et al. 1996 ), and virgin females of somatic genotype csw FRT¹⁰¹/ovo^D1 FRT¹⁰¹; UAS-csw^src90/CyO, en11; MRKS, FLP⁹⁹/+ or D-raf FRT¹⁰¹/ovo^D1 FRT¹⁰¹; UAS-csw^src90/CyO, en11; MRKS, FLP⁹⁹/+ were crossed to males of genotype FM7, ftz-lacZ/Y; twi-Gal4/twi-Gal4. We selected germline clone-derived embryos of genotype csw FRT¹⁰¹/Y; UAS-csw^src90 (II)/twi-Gal4 or D-raf FRT¹⁰¹/Y; UAS-csw^src90 (II)/twi-Gal4 due to their lack of ß-galactosidase activity or lack of expression of the lacZ gene.

To express UAS-csw^src90 in autosomal germline clone mosaic backgrounds (i.e., ras1, ksr, and DOS), we first made females of genotype UAS-csw^src90/UAS-csw^src90 (II); mutation (m), FRT/TM3, ftz-lacZ by marking UAS-csw^src90/UAS-csw^src90 (II) on the III chromosome. Then we marked m, FRT/Bal on the II chromosome using the stock y w; Sp/CyO, ftz-lacZ; Dr/TM3, ftz-lacZ. UAS-csw^src90/UAS-csw^src90; Dr/TM3, ftz-lacZ flies were then crossed to Sp/CyO, ftz-lacZ; m, FRT/TM3, ftz-lacZ flies. Virgin females of genotype UAS-csw^src90/UAS-csw^src90; m, FRT/TM3, ftz-lacZ were then crossed to males carrying a flipase (FLP²²) on the X chromosome, a balancer expressing lacZ on II, and the m, FRT, ovo^D1/Bal, and lacZ on III, generated using standard genetic crosses. Germline clones were made as described elsewhere (HOU et al. 1995 ), and virgin females of somatic genotype FLP²²/+; UAS-csw^src90/CyO, ftz-lacZ; m, FRT/ovo^D1, FRT were crossed to males of genotype +/Y; twi-Gal4/twi-Gal4; m, FRT/TM3, ftz-lacZ.

Germline clone-derived embryos of genotype UAS-csw^src90/twi-Gal4; m, FRT/m, FRT were chosen by selecting embryos lacking lac Z activity or proteins.

Expression of mutations or transgenes on the third chromosome in a csw germline clone-derived background: Virgin female progeny of genotype FM7, ftz-lacZ/y w; Dr/+ were crossed to y w/Y; mutation (m) or [transgene]/Bal males. Male progeny of genotype FM7, ftz-lacZ/Y; TM3, ftz-lacZ/+ were crossed to csw FRT¹⁰¹/FM7, ftz-lacZ virgin females. Females of genotype csw FRT¹⁰¹; TM3, ftz-lacZ/+ were then crossed to males of genotype FM7, ftz-lacZ /Y; m or [transgene]/Dr.

Expression of UAS-csw^src90 and twi-Gal4 in nongermline clone-derived mutant embryos: Mutations (m) on the III chromosome that were previously marked on II were crossed to UAS-csw^src90/UAS-csw^src90; Dr/TM3, ftz-lacZ flies or twi-Gal4/twi-Gal4; Sb/TM6ß AbdB-lacZ flies to make UAS-csw^src90/UAS-csw^src90; m/TM3, ftz-lacZ or twi-Gal4/twi-Gal4; m/TM6ß AbdB-lacZ, respectively. Males or females or either genotype were crossed to each other to make embryos of genotype UAS-csw^src90/twi-Gal4; m/m. These embryos were chosen by selecting against the expression or activity of the lac-Z gene.

For mutations on the II chromosome, UAS-csw^src90 (II) or twi-Gal4 (II) were separately recombined onto the appropriate mutant chromosomes. Briefly, either UAS-csw^src90 (II) or twi-Gal4 females were crossed to mutation (m)/CyO males. The resulting Cy⁺ virgin female progeny were selected and crossed to w/Y; CyO/Sco males. Of the Cy males carrying the p[w+] transgene from either UAS-csw^src90 or twi-Gal4 transgenic stocks, 40 to 50 were then backcrossed to the females from the original m/CyO strain. If no Cy⁺ progeny were observed and the p[w+] transgene was still present, a stock was established. Presence of p[w+] confirmed that the stock contained either UAS-csw^src90(II) or twi-Gal4. Absence of Cy⁺ progeny confirmed that a lethal mutation was present. Additional confirmation of the presence of the correct mutation was done by complementation tests with the original stock from which the recombinant was made and by other lethal alleles of the same gene. Finally, confirmation of the presence of the mutation was established by analysis of homozygous mutant embryos obtained from the recombinant strain.

Immunohistochemistry and visualization:
Immunohistochemistry was performed by modifications of standard protocols (PATEL 1994 ). For some antibodies, the signal was enhanced by using the Tyramide Signal Amplification-Indirect protocol (New England Nuclear, Boston). Embryos were dehydrated in ethanol and cleared in methyl salicylate. Antibodies used (at dilutions ranging from 1:10 to 1:1000) were guinea-pig Krüppel (Kr) antisera, ß-galactosidase mouse monoclonal antibody, Even-skipped mouse monoclonal antibody, and biotin-conjugated secondary antibodies. X-gal staining was performed as described in YARNITZKY et al. 1997 . All embryos were analyzed and photographed with a Nikon FXA equipped with Nomarski optics.

Larval cuticles were prepared in Hoyer's mountant as described by VAN DER MEER 1977 . Cuticles were examined using dark-field illumination. Adult wings were dissected and mounted in Hoyer's mountant. Wings were visualized using bright-field illumination. Eggshells were collected on molasses plates and washed with PBS. They were then prepared in Hoyer's mountant and visualized using dark-field illumination.

Quantification of VA2 development:
VA2 precursor cells form in seven abdominal hemisegments. We assayed their presence or absence in stage 13 embryos. The development of VA2 precursor cells was quantified by determining the ratio of the number of hemisegments developing VA2 cells divided by the total number of hemisegments scored per genotype. On average we scored 191 hemisegments per genotype. We scored only hemisegments in which other EGFR-independent muscle precursor cells were present, particularly the lateral transverse (LT) 2 and 4 muscle precursors. This was done to ensure that the capacity for mesoderm development was still possible in various mutant backgrounds. Our statistical analyses took into account that we were measuring only VA2 precursors equal to or less than wild type; therefore we utilized the one-tailed z-test.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Expression of a csw gain-of-function construct phenocopies gain-of-function phenotypes in EGFR signaling during oogenesis and metamorphosis:
In an effort to examine Csw function in the context of EGFR signaling, we tested whether gain-of-function csw phenotypes could be generated using the UAS-Gal4 system for tissue-targeted gene expression (BRAND and PERRIMON 1993 ). The gain-of-function csw construct encodes an otherwise wild-type form of Csw that has the Src1 myristylation domain fused to the N terminus of Csw (ALLARD et al. 1996 ). The result of this fusion is the constitutive docking of Csw to the plasma membrane, the presumed effect of which is increased or sustained Csw activity. Expression of csw^src90 in wild-type eye imaginal discs using sevenless enhancer/heat-shock proximal promoter results in extra R7 photoreceptor cells (ALLARD et al. 1996 ). The csw^src90 cDNA was cloned into a pUAST transformation vector to make UAS-csw^src90. Multiple independent transgenic lines were generated, and in the presence of a Gal4 driver each produced essentially indistinguishable phenotypes (see below). These transgenic Csw proteins were also expressed to similar levels as determined by Western blot analysis (data not shown).

Dorsal/ventral (D/V) patterning of the eggshell and embryo is an EGFR-dependent process (reviewed in VAN BUSKIRK and SCHUPBACH 1999 ). Expression of UAS-csw^src90 in wild-type follicle cells using the CY2-Gal4, 55B-Gal4, or T155-Gal4 drivers results in dorsalized eggshells (Fig 1A) and embryos (Fig 1B), respectively. To examine the D/V patterning of these dorsalized embryos, the expression patterns of Twist and zen were visualized at blastoderm stages (data not shown). Expression of UAS-csw^src90 in wild-type follicle cells resulted in the loss of Twist expression from its normal position along the ventral side of blastoderm embryos. The opposite effect was observed with zen where ectopic expression extended to the ventral midline from its normal position along the dorsal side of the blastoderm embryo. These expression patterns of both Twist and zen suggest that expression of the csw^src90 construct in follicle cells is sufficient to dorsalize the embryo.

View larger version (69K): In this window In a new window Download PPT slide   Figure 1. Cswsrc90 hyperactivates the EGFR pathway during oogenesis and metamorphosis. (A) Wild-type (WT) eggshell (left) and eggshell in which UAS-cswsrc90 (II) is expressed specifically in the follicle cells using the CY2-Gal4 driver. Arrow points to a region of excess dorsal appendage material that is produced when the EGFR pathway is hyperactivated (see also QUEENAN et al. 1997 ). (B) Embryos from mothers expressing UAS-cswsrc90 in their follicle cells using the 55B-Gal4 driver are also dorsalized to varying extents. Cuticle on the far left is wild type (WT), followed by two cuticles of genotype UAS-cswsrc90 (II)/UAS-cswsrc90 (II) and 55B-Gal4/+, whose level of dorsalization varies from weak (middle cuticle) to strong (cuticle on right). Arrows point to the posteriorly located Filzkörper material. (C) Wild-type (WT) wing. (D) Wing derived from an adult in which UAS-cswsrc90 (III) was expressed in imaginal discs using the 71B-Gal4 driver. Arrows point to excess wing vein material, a phenotype seen when the EGFR pathway is hyperactivated (see also STURTEVANT et al. 1993 ).

Significantly, these UAS-csw^src90-induced effects in D/V patterning are indistinguishable from those seen by expression of a constitutively active construct of the EGFR (EGFR^top; QUEENAN et al. 1997 ), by overexpression of wild-type rho (RUOHOLA-BAKER et al. 1993 ), or by expression of a gain-of-function construct of D-raf (BRAND and PERRIMON 1994 ). Eggs and embryos from females expressing the Gal4 driver alone, UAS-csw^src90 alone, or wild-type csw (UAS-csw^WT) with a Gal4 driver were wild type in appearance (data not shown).

To examine the EGFR-dependent process of wing vein formation (reviewed in SCHWEITZER and SHILO 1997 ), the 71B-Gal4 driver was used to express UAS-csw^src90 in wild-type imaginal discs during metamorphosis. In all animals of genotype UAS-csw^src90 (III)/71B-Gal4, ectopic wing veins were observed (Fig 1C and Fig D). These ectopic veins resembled those seen in flies expressing the gain-of-function rl/MAPK mutation rolled^sevenmaker (BRUNNER et al. 1994B ) and in flies in which wild-type rho is overexpressed (STURTEVANT et al. 1993 ). Wing veins were wild type in flies expressing the Gal4 driver alone, UAS-csw^src90 alone, or UAS-csw^WT driven by 71B-Gal4 (data not shown).

In summary, in multiple EGFR-dependent contexts during oogenesis, metamorphosis, and embryogenesis (see further experiments below), the csw^src90 construct functions as predicted for a gain-of-function csw mutation.

Csw affects EGFR and Heartless-dependent processes in the mesoderm:
The EGFR and Heartless (Htl), one of two known Drosophila FGFR homologs, both function during myogenesis (GISSELBRECHT et al. 1996 ; BUFF et al. 1998 ; MICHELSON et al. 1998 ). Presumptive pericardial cells whose specification requires Htl, and the presumptive dorsal acute muscle 1 (DA1) precursor cells whose specification requires Htl and the EGFR, can be visualized using antibodies to the pair-rule transcription factor Even-skipped (Eve; FRASCH et al. 1987 ; BUFF et al. 1998 ; MICHELSON et al. 1998 ). At stage 11, dorsal mesodermal progenitor cells are visible (Fig 2A). By stage 13, the pericardial and the DA1 precursor cells can be seen as two distinct groups of cells along the dorsal limits of the embryo (Fig 2B).

View larger version (107K): In this window In a new window Download PPT slide   Figure 2. Csw affects mesoderm development. (A, C, E, G, and I) Stage 11 embryos. (B, D, F, H, and J) Stage 13 embryos labeled with antibodies to the transcription factor Eve. (A and B) Wild-type (WT) embryos. (A) At stage 11 Eve is expressed in clusters of dorsal mesodermal cells that by stage 13 (B) are distinguishable as the pericardial precursor cells (PC) and the somatic dorsal acute muscle 1 precursor cells (DA1). (C and D) Expression of one copy of UAS-cswsrc90 in the mesoderm using the twist-Gal4 (twi-Gal4) driver results in increased numbers and/or duplication of Eve-positive cells. (E and F) More Eve-positive cells are seen, some outside of their normal boundary, when two copies of UAS-cswsrc90 are expressed using the twi-Gal4 driver. (G and H) Expression of an activated EGFR mutation (UAS-EGFRtop) using twi-Gal4 also results in excess Eve-positive cells, thereby implicating Csw in EGFR-dependent mesoderm development. (I and J) Loss of Csw function results in severe loss of Eve-positive mesodermal cells. Clusters of dorsal mesodermal cells (stage 11) or PC and DA1 precursor cells (stage 13) are enlarged in the right corner of A–J. In A–J, anterior is to the left and ventral is down.

To examine whether Csw^src90 would affect the formation of these mesodermal cells, UAS-csw^src90 was expressed in the mesoderm using the twist-Gal4 (twi-Gal4) driver (GREIG and AKAM 1993 ) in an otherwise wild-type background. UAS-csw^src90(II)/twi-Gal4 embryos have increased numbers of Eve-positive mesodermal cells, most likely due to duplication of the DA1 muscle presursors (Fig 2C and Fig D), a phenotype indistinguishable from that seen by expression of UAS-EGFR^top (BUFF et al. 1998 and Fig 2G and Fig H) or wild-type rho (UAS-rho^WT; BUFF et al. 1998 ; MICHELSON et al. 1998 ) using the same twi-Gal4 driver. Interestingly, expression of two copies of UAS-csw^src90 results in the formation of Eve-positive cells outside of their normal boundary (Fig 2E and Fig F), a phenotype usually seen if the Wg and activated Ras1 pathways are overexpressed simultaneously in the mesoderm (HALFON et al. 2000 and see DISCUSSION).

Conversely, stage 11 and 13 embryos lacking Csw function fail to stain for Eve in many hemisegments, suggesting that these embryos are missing dorsal mesodermal precursor cells (Fig 2I and Fig J). The phenotype of a csw hemizygous mutant embryo at stage 11 is nearly as strong as the phenotype seen in a htl (null) mutant embryo (GISSELBRECHT et al. 1996 ). Later in development, it is interesting to note that if any dorsal mesodermal cells do form in csw hemizygous mutant embryos, they are pericardial precursor cells; i.e., the DA1 muscle is never observed alone (Fig 2J and data not shown), suggesting, perhaps, that the EGFR pathway is more sensitive to loss of Csw function than is the Htl pathway. Alternatively, since Htl is required for both migration, as well as specification, of pericardial cells and the DA1 muscle (GISSELBRECHT et al. 1996 ; MICHELSON et al. 1998 ), the later presence of only pericardial cells may be due to a more prominent role for Csw in Htl-dependent migration rather than Htl-dependent pericardial cell specification.

In summary, our results with both loss-of-function mutations and gain-of-function constructs in csw further support its positive role in RTK signaling, in general, and in mesoderm development, in particular.

EGFR-dependent VA2 formation serves as a sensitive assay to study the EGFR pathway:
Development of the DA1 precursor cells requires signaling by both Htl and the EGFR (BUFF et al. 1998 ; MICHELSON et al. 1998 ). However, only the EGFR is essential in the formation of the somatic VA2 precursor cell (BUFF et al. 1998 ). Antibodies to the segmentation transcription factor Krüppel (Kr) allow visualization of the VA2 precursor cells in addition to other somatic muscle precursors (Fig 3A; GAUL et al. 1987 ; RUIZ-GOMEZ et al. 1997 ). Virtually no hemisegments form VA2 precursor cells in spi mutant embryos (BUFF et al. 1998 and Fig 4A and Fig G) whereas VA2 precursor cells do form in 98% of htl mutant embryos (GISSELBRECHT et al. 1996 ; see MATERIALS AND METHODS for description of quantification of VA2 precursor cell formation).

View larger version (52K): In this window In a new window Download PPT slide   Figure 3. Csw and DOS affect VA2 precursor cell formation. (A–C) Stage 13 embryos stained with antibodies to the transcription factor Kr. (A) Kr protein is expressed in several somatic muscle precursor cells in wild-type embryos, including the ventral acute muscle precursor cells (VA2) and the lateral transverse 2 and 4 (LT2, -4) precursor cells. (B) Germline clone-derived embryos lacking Csw function (genotype cswVA199/Y) do not form EGFR-dependent precursor cells in many hemisegments, including the VA2 precursor cells. The LT2, -4 precursor cells, which are not affected by EGFR signaling, are still present in csw null mutant embryos. (C) Expression of UAS-cswsrc90 using twi-Gal4 in csw hemizygous mutant embryos results in rescue of VA2 precursor cells. (D) Quantification of the number of hemisegments in which VA2 precursor cells are formed in wild-type (WT) embryos, embryos of genotype UAS-cswsrc90(II)/twi-Gal4 where the genetic background is wild type, hemizygous csw mutant embryos (cswVA199/Y), cswVA199/Y embryos in which UAS-cswsrc90 is expressed using the twi-Gal4 driver, embryos hemizygous mutant for cswLE120 (cswLE120/Y), cswLE120/Y embryos in which UAS-cswsrc90 is expressed with twi-Gal4, DOS mutant embryos, and DOS mutant embryos in which UAS-cswsrc90 is expressed using the twi-Gal4 driver. Quantification was done by determining the ratio of the number of hemisegments in which VA2 precursor cells formed to the total number of hemisegments counted. The numbers within the graph represent the percentage of hemisegments in which VA2 precursor cells formed for a given genotype. Both WT embryos and embryos of genotype UAS-cswsrc90/twi-Gal4 form VA2 precursor cells in seven hemisegments on each side of the embryo. Rescue of csw mutant embryos by Cswsrc90 (45%) is statistically significant at P < 0.001 using the one-tailed z-test. Rescue of DOS mutant embryos by Cswsrc90 (41%) is statistically significant at P < 0.001 using the same statistical analysis. The extent of rescue by Cswsrc90 is the same for DOS mutant embryos as it is for cswVA199/Y mutant embryos (P < 0.001). The sample size and standard error of the mean (SEM; where applicable) for each genotype are as follows: WT: n = 441 hemisegments; UAS-cswsrc90/twi-Gal4: n = 350, cswVA199/Y, n = 350, SEM = 0.025; cswVA199/Y; UAS-cswsrc90/twi-Gal4: n = 287, SEM = 0.029; cswLE120/Y: n = 309, SEM = 0.028; cswLE120/Y; UAS-cswsrc90/twi-Gal4: n = 16, SEM = 0.061; DOS/DOS: n = 133, SEM = 0.038; UAS-cswsrc90/twi-Gal4; DOS/DOS: n = 217, SEM = 0.034. Note: On this and subsequent figures, error bars were too small to be marked on the graph.

View larger version (62K): In this window In a new window Download PPT slide   Figure 4. cswsrc90 interacts with the EGFR, spi, and rho during mesoderm development. (A–F) Stage 13 embryos stained with antibodies to Kr. Loss-of-function mutations in spi (A), rho (B), and expression of a dominant-negative EGFR mutation (UAS-EGFRDNDER) using (in an otherwise wild-type background) twi-Gal4 (C) result in loss of VA2 precursor cells. The EGFR-independent LT2, -4 precursor cells are present in each mutant background. Expression of UAS-cswsrc90 with twi-Gal4 in spi (D), rho (E), or UAS-EGFRDNDER backgrounds results in rescue of VA2 precursor cell formation. (G) Quantification of the formation of VA2 precursor cells in spi, rho, and UAS-EGFRDNDER backgrounds in the presence or absence of UAS-cswsrc90 expressed with the twi-Gal4 driver. Cswsrc90 rescues spi, rho, and EGFRDNDER mutant embryos to the same extent, 70% (P < 0.001). Sample size and SEM for each genotype are as follows: spi/spi: n = 441, SEM = 0.005; spi, UAS-cswsrc90/spi, twi-Gal4: n = 330, SEM = 0.015; rho/rho: n = 214, SEM = 0.010; UAS-cswsrc90/twi-Gal4; rho/rho: n = 163, SEM = 0.036; +/twi-Gal4; UAS-EGFRDNDER/UAS-EGFRDNDER: n = 163, SEM = 0.034; UAS-cswsrc90/twi-Gal4; UAS-EGFRDNDER/UAS-EGFRDNDER: n = 195, SEM = 0.032.

Loss of Csw function reduces the number of VA2 and other EGFR-dependent precursor cells as determined using two different strong csw mutations (Fig 3B and Fig D). The csw^VA199 allele is observed by Western blot analysis to make a truncated protein consisting of just two SH2 domains (J. LORENZEN, M. MELNICK and L. A. PERKINS, unpublished observations). Twenty-nine percent of hemisegments form VA2 precursor cells in germline clone-derived embryos of genotype csw^VA199/Y (Fig 3D). The csw^LE120 allele by Western blot analysis has never been observed to make a Csw protein and is therefore believed to be a protein null mutation (J. LORENZEN, M. MELNICK and L. A. PERKINS, unpublished observations). Interestingly, hemizygous germline clone-derived embryos carrying csw^LE120 form VA2 precursor cells in 53% of hemisegments. These results support the idea that the mutant Csw protein produced by the csw^VA199 mutation interferes with VA2 precursor formation by acting as a dominant-negative protein (see DISCUSSION).

Expression of the activating csw^src90 protein in a csw^VA199 mutant background improves formation of VA2 precursor cells such that 45% of hemisegments form VA2 precursor cells in embryos of genotype csw^VA199/Y; UAS-csw^src90(II)/twi-Gal4 (Fig 3C and Fig D; rescue significant at P < 0.001). Likewise, Eve-positive muscle progenitors and the subsequent pericardial and DA1 precursor cells are recovered in csw^VA199/Y; UAS-csw^src90(II)/twi-Gal4 embryos compared to csw^VA199/Y embryos (data not shown). However, embryos of genotype csw^LE120/Y; UAS-csw^src90(II)/twi-Gal4 form VA2 precursor cells in 94% of hemisegments, which approaches wild-type levels (Fig 3D).

In summary, expression of csw^src90 can rescue loss-of-function mutations associated with csw itself, suggesting that csw^src90 is able to substitute for wild-type Csw function. Further, we conclude that the formation of VA2 precursor cells can be used as a single cell assay system to monitor signaling in the EGFR pathway.

Csw^src90 rescues loss of VA2 precursor cells in embryos mutant for the EGFR, rho, and spi genes:
Embryos mutant for the EGFR ligand spi essentially delete VA2 precursor cells (BUFF et al. 1998 and Fig 4A). When quantified, 1.1% of hemisegements from spi embryos form VA2 precursor cells (Fig 4G). Expression of UAS-csw^src90 with twi-Gal4 in a mutant spi background significantly rescues VA2 precursor cell formation such that 67% of hemisegments in embryos of genotype spi, UAS-csw^src90(II)/spi, twi-Gal4 form VA2 precursor cells (Fig 4D and Fig G; P < 0.001). The formation of VA2 precursor cells is also significantly reduced in rho embryos (2.3% of hemisegments form VA2 precursor cells; Fig 4B and Fig G). As with its interaction with spi, expression of UAS-csw^src90 with twi-Gal4 significantly rescues the rho phenotype such that 69% of hemisegments form VA2 precursor cells in embryos of genotype UAS-csw^src90(II)/twi-Gal4; rho/rho (Fig 4E and Fig G; P < 0.001).

Embryos expressing a dominant-negative EGFR construct (UAS-EGFR^DNDER) with twi-Gal4 in an otherwise wild-type background also have reduced numbers of hemisegments that form VA2 precursor cells (26%; BUFF et al. 1998 and Fig 4C and Fig G). In this background expression of UAS-csw^src90 with twi-Gal4 was able to rescue VA2 formation such that 71% of hemisegments form VA2 precursor cells (Fig 4F and Fig G). As a control for all of the experiments presented here, we confirmed that there was no effect on VA2 formation when one copy of UAS-csw^src90 was expressed in the absence of a Gal4 driver in these or other mutant backgrounds (data not shown). Additionally, expression of UAS-csw^src90 with twi-Gal4 in an otherwise wild-type background shows formation of VA2 precursor cells in all seven hemisegments (Fig 3D).

In summary, in embryos mutant for signaling components acting upstream or at the level of the receptor, expression of UAS-csw^src90 with twi-Gal4 significantly rescues, to a similar extent (70%), VA2 precursor cell formation (see DISCUSSION).

DOS mutant embryos phenocopy csw^VA199 mutant embryos and both mutations are rescued to the same extent by expression of UAS-csw^src90 with twi-Gal4:
DOS encodes a multi-adapter protein that interacts with and is a putative substrate for Csw (HERBST et al. 1996 ; RAABE et al. 1996 ). It has been reported that a direct physical interaction between DOS and Csw is important for signaling in the Sevenless (Sev) pathway (HERBST et al. 1999 ; BAUSENWEIN et al. 2000 ). Interestingly, our genetic data are consistent with a close interaction between DOS and Csw on the basis of their similar phenotypes. For example, VA2 precursor cells form in 25% of hemisegments in mutant embryos expressing the putative dominant-negative DOS mutation, DOS^R31 (BAUSENWEIN et al. 2000 ), which is in the same range statistically as the percentage of VA2 precursor formation in csw^VA199/Y embryos (29%; Fig 3D). Likewise, expression of UAS-csw^src90 with twi-Gal4 in these DOS mutant embryos increases VA2 development to 41%, a statistically significant rescue (P < 0.001) that places Csw function downstream of DOS (Fig 3D). Further, this rescue is in the same range as Csw^src90-induced rescue of csw^VA199 mutant embryos (45%).

In summary, these data are consistent with a close association between Csw and DOS and suggest that Csw function is required downstream of DOS function (see DISCUSSION).

Csw^src90 partially rescues loss of Ras1, Ksr, and D-Raf functions:
Ras1 is a key molecule in RTK pathways as its activation transduces the RTK signal, leading to activation of D-Raf, MEK, and Rl/MAPK (reviewed by ROMMEL and HAFEN 1998 ). ras1 null embryos have VA2 precursor cells in 9% of hemisegments (Fig 5C), suggesting that a small amount of the EGFR signal is independent of Ras1, since in the absence of Spi, virtually no hemisegments form VA2 precursor cells (Fig 4A and Fig G). In a ras1 null background, expression of UAS-csw^src90 with twi-Gal4 partially suppresses the ras1 phenotype in that VA2 precursor cells form in 21% of hemisegments (P < 0.001; Fig 5C), placing some of the Csw function downstream of Ras1.

View larger version (53K): In this window In a new window Download PPT slide   Figure 5. Cswsrc90 partially rescues loss of Ras1, Ksr, and D-Raf, but not Pnt functions. (A and B) Stage 13 embryos stained with antibodies to Kr. (A) D-raf hemizygous (D-raf/Y) mutant embryos lack VA2 precursor cells in virtually all hemisegments. Not all hemisegments are shown due to the twisted morphology of the embryo. Arrow points to the EGFR-independent LT2, -4 precursor cells. (B) Expression of UAS-cswsrc90 using twi-Gal4 in a D-raf/Y background rescues formation of VA2 precursor cells. (C) ras1 and ksr mutant embryos also variably lack VA2 precursor cells. However, in these backgrounds, as for D-raf, expression of UAS-cswsrc90 with twi-Gal4 rescues the formation of VA2 precursor cells by 10% above basal levels (P < 0.001). Formation of VA2 precursor cells in pnt mutant embryos and in UAS-cswsrc90/twi-Gal4; pnt/pnt mutant embryos is statistically the same (P < 0.001), placing Csw function upstream of Pnt. The sample size and SEM for each genotype is as follows: ras1/ras1: n = 209, SEM = 0.017; UAS-cswsrc90/twi-Gal4; ras1/ras1: n = 155, SEM = 0.033; ksr/ksr: n = 714, SEM = 0.008; UAS-cswsrc90/twi-Gal4; ksr/ksr: n = 126, SEM = 0.027; D-raf/Y: n = 252, SEM = 0.007; D-raf/Y; UAS-cswsrc90/twi-Gal4: n = 266, SEM = 0.019; pnt/pnt: n = 142, SEM = 0.032; UAS-cswsrc90/twi-Gal4; pnt/pnt: n = 156, SEM = 0.031.

D-raf null mutant embryos develop VA2 precursor cells in only 1.2% of hemisegements, significantly less than in ras1 mutant embryos (Fig 5A and Fig C).

D-raf/Y embryos are rescued to 10.5% VA2 precursor cell formation by expression of UAS-csw^src90 with twi-Gal4 (Fig 5B and Fig C; P < 0.001). As with its genetic interaction with ras1, csw^src90 interaction with D-raf places some Csw function downstream of D-Raf.

Ksr function provides another mechanism to regulate D-Raf. Loss of Ksr function markedly reduces the formation of VA2 precursor cells (4.5%; Fig 5C). This direct phenotypic evidence demonstrating a role for Ksr in the Drosophila EGFR pathway correlates well with the recent isolation of the EGFR in a ksr modifier screen (THERRIEN et al. 2000 ). csw^src90 suppresses the ksr phenotype because UAS-csw^src90(II)/twi-Gal4; ksr/ksr embryos have 14% of hemisegments with VA2 precursor cells (P < 0.001; Fig 5C).

In summary, it was found that some EGFR signaling occurs in ksr and ras1 mutant embryos derived from females bearing germline clones, whereas no EGFR signaling is detected in hemizygous mutant D-raf embryos derived from females bearing germline clones. In all three mutant backgrounds, expression of UAS-csw^src90 with twi-Gal4 rescued formation of VA2 precursor cells to the same extent above basal levels (10%; see DISCUSSION).

Loss of DIM-7 suppresses Csw^src90 in the mesoderm:
The DIM-7 protein, encoded by the gene moleskin (msk), is a nuclear import protein that physically interacts with Csw, and mutations in msk genetically interact with mutations in rl/MAPK (LORENZEN et al. 2001 ). DIM-7 likely transports activated Rl/MAPK into the nucleus since embryos carrying loss-of-function msk mutations significantly reduce nuclear localized activated Rl/MAPK (LORENZEN et al. 2001 ). The context of mesoderm development was used to determine how msk and csw interact genetically. Loss-of-function msk mutant embryos develop VA2 precursor cells in nearly all of the appropriate hemisegments (data not shown), presumably because the msk mutant embryos were not derived from mothers carrying msk germline clones. Therefore, it was not possible to assay formation of VA2 formation in msk mutant embryos that were also expressing UAS-csw^src90 with twi-Gal4. However, it was possible to determine if loss of DIM-7 function affected Csw^src90-induced formation of extra dorsal mesodermal cells, as determined by labeling stage 11 embryos with antibodies to Eve. Compared to otherwise wild-type embryos in which UAS-csw^src90 is expressed with twi-Gal4 (Fig 2C), expression of UAS-csw^src90 with twi-Gal4 in msk homozygous loss-of-function mutant embryos results in the formation of a wild-type number of Eve-positive dorsal mesodermal cells (Fig 6A). Moreover, msk can dominantly suppress csw^src90 as removal of one copy of msk in a UAS-csw^src90/twi-Gal4 background results in the formation of fewer Eve-positive cells (Fig 6B; compare to Fig 2C).

View larger version (48K): In this window In a new window Download PPT slide   Figure 6. DIM-7 functions downstream of Csw during myogenesis. (A and B) Stage 11 embryos stained with antibodies to Eve. (A) Loss of DIM-7 in homozygous msk5/msk5 mutant embryos in UAS-cswsrc90/twi-Gal4 embryos suppresses the formation of extra dorsal mesodermal cells relative to wild type (see Fig 2C). (B) Loss of one copy of msk (genotype UAS-cswsrc90/twi-Gal4; msk5/TM3, ftz-lacZ) suppresses the extent of formation of extra dorsal mesodermal cells. This embryo is also expressing ftz-lacZ on the III chromosome. Clusters of dorsal mesodermal cells are enlarged in the bottom right corner of A and B.

In summary, these results demonstrate a genetic interaction between DIM-7 and Csw, which is consistent with its putative role as a nuclear transporter of activated Rl/MAPK.

Csw^src90 does not suppress the pointed mutant phenotype:
One known downstream target of EGFR signaling is the transcription factor Pointed (Pnt), a Rl/MAPK target (KLAMBT 1993 ; BRUNNER et al. 1994A ; O'NEILL et al. 1994 ). A total of 82% of pnt mutant hemisegments form VA2 precursor cells (Fig 5C), suggesting that Pnt plays only a minor role in VA2 precursor formation. Our results demonstrate that in the context of EGFR-dependent muscle formation, functionally redundant or overlapping transcription factors work to ensure proper VA2 precursor cell formation. Expression of UAS-csw^src90 in a pnt background was not able to suppress the pnt phenotype (Fig 5C).

In summary, in a pnt mutant background, 82% of VA2 precursors form, which supports the idea that at least one additional, positive-acting transcription factor must function within this developmental context. Since the identity of this additional transcription factor is presently unknown, our data do not predict where Csw function falls relative to one or more additional positive transcription factors. However, with respect to Pnt, the simplest interpretation of the data is that all Csw function is upstream of this transcription factor.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

It is well established that Csw and its homologs are positive signal transducers downstream of several RTKs, including the EGFR. We had previously established a genetic link between Csw and the Drosophila EGFR pathway in a variety of tissues on the basis of phenotypic analyses of csw mutations and direct genetic interaction between csw and a dominant-negative EGFR construct (PERKINS et al. 1996 ). In this report, we extend our previous analyses by genetic dissection of Csw function in EGFR signaling in various genetic backgrounds during formation of VA2 muscle precursor cells. A variety of genetic interaction experiments between gain- and loss-of-function mutations and/or constructs in genes involved in EGFR signaling has resulted in three principal findings. First, and consistent with findings in the developing retina (ALLARD et al. 1996 ), Csw^src90 functions like a bona fide gain-of-function protein in several EGFR-initiated developmental processes during oogenesis, embryogenesis, and metamorphosis. Second, Csw plays a positive role in EGFR signaling during myogenesis. Third, tracking the formation of VA2 precursor cells serves as a sensitive assay to infer levels of EGFR signaling in various mutant genetic backgrounds.

Csw^src90 phenotypically reflects Csw function in the EGFR pathway:
Expression of UAS-csw^src90 in several tissues phenocopies gain-of-function mutations and constructs in positive signaling genes in the EGFR pathway. Moreover, tissue-specific expression of csw^src90 is able to rescue VA2 precursor cell formation in loss-of-function csw mutant embryos. However, there are important considerations to be made regarding use of the csw^src90 construct to study Csw function in RTK pathways. csw^src90, being a synthetic mutation, may have neomorphic activity, the result of which is an artificial, nonspecific phenotype not correlating with wild-type Csw function. For instance, in embryos expressing two copies of UAS-csw^src90 in the mesoderm, Eve-positive cells formed outside of the normal boundaries previously prepatterned by Wg signaling (CARMENA et al. 1998 and Fig 2E). This phenotype resembles the effect seen by simultaneous overexpression of UAS-wingless, UAS-twist (Twist is a downstream target of Wg signaling), and activated ras1 (UAS-ras1^ACT) in the embryonic mesoderm (HALFON et al. 2000 ), but not by expression of UAS-ras1^ACT alone (MICHELSON et al. 1998 ). This result, seen with two copies of UAS-csw^src90, might reveal a nonphysiological ability for Csw^src90 to bypass the need for Wg signaling at the transcriptional level during myogenesis (HALFON et al. 2000 ).

While we cannot rule out the possibility that Csw^src90 exhibits some neomorphic properties, it is notable that, in all developmental contexts that we examined, the phenotypes resulting from expression of one copy of UAS-csw^src90 never differed from what was expected for a gain-of-function csw mutation (this report; ALLARD et al. 1996 ). Therefore, we analyzed phenotypes only in embryos in which one copy of UAS-csw^src90 was expressed.

Furthermore, the phenotypes do not reflect promiscuous phosphatase activity because membrane-targeted expression solely of the Csw phosphatase domain is embryonic lethal and results in cuticle phenotypes not reflecting a predicted gain-of-function csw mutation (M. R. JOHNSON HAMLET, M. MELNICK and L. A. PERKINS, unpublished observations).

Interestingly, no phenotypes were observed when wild-type csw (UAS-csw^WT) was expressed using twi-Gal4 in various genetic backgrounds. While this could be due to the extent to which UAS-csw^WT was expressed, on the basis of what is known about the regulation of its vertebrate functional homolog SHP-2 (PERKINS et al. 1996 ; HOF et al. 1998 ), an alternative explanation is that simply adding more wild-type Csw in an otherwise wild-type background is not sufficient to increase its activity.

The crystal structure SHP-2 has revealed that the N-terminal SH2 domain binds to the catalytic domain, which keeps SHP-2 inactive. Engagement of the N-terminal SH2 domain with a tyrosine-phosphorylated protein releases the block of the catalytic domain, resulting in SHP-2 activation (HOF et al. 1998 ). Thus, if the molecules that engage the SH2 domain of Csw are limiting in amount, exogenously expressed wild-type Csw protein would not be able to release the N-terminal SH2 domain from the catalytic domain, thereby keeping the exogenous wild-type Csw protein in an inactive state. However, the myristylated and thereby membrane-targeted Csw^src90 protein is already in an active state, which results in hyperactivation of the RTK pathway. Csw^src90 is hence insensitive to the normal downregulation of the RTK signal that occurs. The mechanism of action of Csw^src90 is unknown, but it is possible that membrane localization either provides constitutive access to substrates or changes the conformation of Csw^src90 such that the N-terminal SH2 domain is unable to bind to the catalytic domain to block its function. Nevertheless, the phenotypes produced by csw^src90 are consistent with those we expect for a gain-of-function csw mutation.

Synopsis of Csw function in the EGFR pathway:
Interaction between csw and spi, rho, and the EGFR: Within the context of VA2 precursor cell formation, our results enable us to infer the relative contribution of gene function to the EGFR signal. For example, complete loss-of-function mutations in spi, rho, and D-raf essentially eliminate VA2 precursor cells, supporting the idea that these proteins are absolutely essential for the propagation of the EGFR signal.

As we previously observed, the phenotype of csw loss-of-function mutant embryos is not as severe as the phenotypes of loss-of-function mutations in other positive RTK transducers, suggesting that Csw, unlike spi, rho, and D-raf, is not needed to transduce the entire RTK signal. Further support for this finding comes from the similar levels, although <100%, to which Csw^src90 rescues VA2 precursor cell formation in spi, rho, and twi-Gal4/+; UAS-EGFR^DNDER mutant embryos. This latter finding places the interaction of Csw^src90 with these upstream signaling components in a separate category from that of the other genes we analyzed.

Interaction between csw and DOS: Our genetic interaction data between csw and DOS are consistent with a model whereby a direct interaction between Csw and DOS is essential for Drosophila EGFR signaling. When the predominant Tyr residues are mutated in DOS, only DOS protein lacking the phosphorylated Tyr (pTyr) site(s) in consensus to bind the Csw SH2 domain significantly abrogated Sev signaling (HERBST et al. 1999 ; BAUSENWEIN et al. 2000 ). Conversely, a DOS protein containing only the pTyr sites that bind to the Csw SH2 domains is sufficient to provide wild-type DOS function (HERBST et al. 1999 ). A vertebrate DOS homolog, Gab1, and SHP-2 associate upon activation of the vertebrate EGFR, the result of which is an increase in MAPK signaling (HOLGADO-MADRUGA et al. 1996 ; LEHR et al. 1999 ; SHI et al. 2000 ).

The readout from the putative DOS dominant-negative mutant embryos is in the same range as that of dominant-negative csw mutant embryos. The identical genetic interaction of csw and DOS with csw^src90 places their function in a category separate from that of the other signaling genes we analyzed and suggests that they both function at the same level in the EGFR pathway.

Interestingly, DOS mutant embryos phenocopy the putative dominant-negative csw mutant embryos but not the protein null csw mutant embryos. These results suggest that the dominant-negative csw mutant phenotype reflects loss of DOS function. Since the csw^VA199 mutation generates a truncated Csw protein where only the SH2 domains are expressed, perhaps the SH2 domains still bind to and sequester DOS function away from the signaling pathway.

Interaction between csw and ras1, ksr, and D-raf: Loss-of-function mutations derived from females bearing germline clones in ras1, ksr, and D-raf result in 9, 4.5, and 1.2%, respectively, of hemisegments in which VA2 precursor cells form. As mentioned above, the D-raf and spi mutant phenotypes are nearly the same, suggesting that Spi and D-raf are absolutely essential for EGFR signal propagation. However, the ras1 protein null phenotype is not as strong as the D-raf protein null phenotype, suggesting that Ras1 transduces <100% of the EGFR signal. These results correlate well with phenotypic analyses of ras1 and D-raf in the Torso pathway where loss-of-function ras1 mutant embryos maintain a low level of Torso signaling, whereas loss-of-function mutations in D-raf abolish Torso signaling (HOU et al. 1995 ). Hence, it can be inferred from our studies that in the EGFR pathway, as perhaps in the Torso pathway, there is also a Ras1-independent mechanism to activate D-Raf.

The loss-of-function ksr mutant phenotype suggests that Ksr contributes more function to the EGFR pathway than Ras1 but less than D-Raf. Similarly, in the Torso pathway, the ksr loss-of-function mutant phenotype is more severe than the ras1 loss-of-function mutant phenotype (HOU et al. 1995 ; THERRIEN et al. 1995 ). These data suggest that loss of Ksr function is more detrimental to transducing an RTK signal than is loss of Ras1 function. Ksr is thought to function as a scaffolding protein that binds Raf1, MEK, Rl/MAPK, and other signaling molecules to regulate a given RTK pathway (reviewed by RAABE 2000 ). Therefore, the phenotype of embryos lacking Ksr function is more severe than that from loss of Ras1 because Ksr directly regulates not only Raf1, but also other crucial downstream molecules such as Rl/MAPK. It has been proposed that the scaffold function of Ksr may be analogous to the budding yeast scaffolding protein Ste5, which binds the Raf, MEK, and MAPK yeast homologs to facilitate MAPK-induced signaling in the mating response pathway (reviewed by GARRINGTON and JOHNSON 1999 ).

We have demonstrated that in the EGFR pathway Csw functions downstream of or parallel to Ras1, Ksr, and D-Raf. Introduction of Csw^src90 into ras1, ksr, and D-raf loss-of-function mutant embryos derived from females bearing germline clones rescues each mutation to the same extent above basal levels. These levels of rescue are much lower than that for spi, rho, and EGFR mutant embryos. One reason for these lower levels of rescue might be that since D-Raf is the major feed-in molecule at this level of the signaling pathway, its absence or the absence of one or more of its activators will severely block any downstream signaling. Nevertheless, our results suggest that a portion of the EGFR signal requires Csw downstream of, or parallel to, Ras1, Ksr, and D-Raf.

The similar genetic interactions of ras1, ksr, and D-raf with csw^src90 place their functions in a category separate from that of the other signaling genes we analyzed and suggest roles for Csw both upstream and downstream of these intermediate signaling components.

Since Csw^src90 is able to function downstream of D-Raf, it is possible that Csw^src90 is able to facilitate Ras1-dependent, D-Raf-independent signaling, as is proposed to happen during RTK-dependent border cell migration (LEE et al. 1996 ). Alternatively, a portion of the Csw signal may contribute to a pathway functioning parallel to the D-Raf/MEK/MAPK pathway, perhaps by facilitating activation of other MAPK homologs, such as p38/MAPK. Mutations in licorne, a p38/MAPKK homolog, can phenocopy loss-of-function EGFR mutations and might affect Grk activity during oogenesis (SUZANNE et al. 1999 ), implicating a role for p38/MAPK signaling in the EGFR pathway.

Interaction between csw and msk: It is possible that Csw can function downstream of D-Raf at the level of Rl/MAPK. Csw physically interacts with the nuclear import protein DIM-7, a member of the importin family of nuclear import proteins, which is thought to transport Rl/MAPK to the nucleus (LORENZEN et al.. 2001 ). In this study, we show a genetic interaction between csw and msk, the gene encoding DIM-7, as loss of DIM-7 suppresses the phenotype associated with Csw^src90. This result is consistent with DIM-7 functioning downstream of Csw, as well as with DIM-7-dependent transport of Rl/MAPK into the nucleus.

Interaction between csw and pnt: Pnt is a downstream target of Rl/MAPK signaling and functions as a transcriptional activator in many RTK pathways, including the Drosophila EGFR pathway (reviewed by RAABE 2000 ). Deletion of both pnt transcripts (P1 and P2) results in 82% of hemisegments in which VA2 precursor cells form. This result suggests that Pnt contributes a small amount to the EGFR signal in this developmental context and that there are other Rl/MAPK target transcription factors whose activities are also required for proper VA2 precursor cell formation. The same partial pnt mutant phenotype is also seen in the context of Eve muscle progenitor specification (HALFON et al. 2000 ). Moreover, it has been reported that pnt mutant embryos primarily lack the lateral longitudinal muscle 1 and several dorsal oblique muscles (DO3, DO4, and DO5; KLAMBT 1993 ), suggesting that certain muscle precursor cells are more sensitive to loss of Pnt function. Nevertheless, Csw^src90 was unable to rescue loss of VA2 precursor cell formation in pnt mutant embryos, suggesting that all Csw function is upstream of Pnt and thereby placing Pnt function in a category separate from that of the other signaling genes we analyzed. It should be noted that these data do not allow the placement of Csw function relative to the unidentified, positive transcription factors in this pathway.

Model of the EGFR pathway during myogenesis:
On the basis of work presented here and elsewhere (reviewed by CASCI and FREEMAN 1999 ; HERBST et al. 1999 ; BAUSENWEIN et al. 2000 ; RAABE 2000 ), we propose a model for Csw function in the EGFR pathway during myogenesis (Fig 7). Activation of the EGFR pathway by Spi binding to the receptor results in an association between Csw and DOS. The Csw/DOS complex might interact with the receptor either via DOS, as it has been demonstrated that the DOS homolog Gab1 binds to the vertebrate EGFR (RODRIGUES et al. 2000 ), or via Csw, as there is a binding site on the Drosophila EGFR in consensus to bind the N-terminal SH2 domain of Csw (M. R. JOHNSON HAMLET and L. A. PERKINS, unpublished observations). Also contributing to the positive signal is the adapter protein Shc. Subsequently, the majority of Ras1 function leads to activation of D-Raf. However, on the basis of the ras1 null mutant phenotype, other molecules are capable of contributing to D-Raf activation. One of these molecules is likely Ksr, which binds to and regulates the Raf, MEK, and Rl/MAPK signaling cassette.

View larger version (25K): In this window In a new window Download PPT slide   Figure 7. Model of Csw function in the EGFR pathway. RTK signaling molecules tested in this study are shown in boldface type and all capital letters. See text for details.

Csw is also able to interact with the nuclear import protein DIM-7, which might provide a means by which Csw can function downstream of D-Raf. Activated Rl/MAPK is then transported to the nucleus, likely via DIM-7. Once in the nucleus, Rl/MAPK negatively regulates Yan and positively regulates Pnt, as well as one or more additional transcriptional activators that are required for formation of VA2 mesodermal cells. The EGFR signal is also downregulated by the inhibitory ligand Argos (Aos), the intracellular proteins Cbl and Sprouty (Sty), and the transmembrane protein Kekkon (Kek).

We have presented a model whereby different components of the EGFR pathway contribute differentially to the signal required for VA2 precursor cell formation. These inferences were made on the basis of our observations that VA2 precursor formation in single mutants alone or in conjunction with Csw^src90 fall into distinct categories. While it is difficult to quantify the precise levels of signal contributed by each component solely on the basis of its mutant phenotype or genetic interactions, we found the overall consistency of the data compelling.

Perspectives:
Collectively, the work presented in this article furthers our understanding of the EGFR signaling pathway during embryonic mesoderm development. We have shown not only where in the signaling hierarchy Csw functions relative to other signaling pathway genes, but we have also inferred the signaling strength contributed by several key molecules in the EGFR pathway. Knowing the relative contribution of a specific pathway component to the overall signal could be the first step in the design of therapeutics to regulate hyperactive RTK pathways that are common in many disease and/or oncogenic states.

	ACKNOWLEDGMENTS

For fly stocks, we thank the Bloomington Stock Center as well as M. Freeman, E. Hafen, A. Michelson, N. Perrimon, G. Rubin, T. Schüpbach, B. Z. Shilo, and M. Therrien. For cDNAs and vectors, we thank A. Brand, V. Cleghon, and M. Simon. For antibodies, we thank D. Kosman, A. Michelson, N. Patel, N. Perrimon, J. Reinitz, and C. Rushlow. Other antibodies were obtained from Boehringer Mannheim and Jackson ImmunoResearch Labs. We thank F. Denhez, S. Gisselbrecht, A. Michelson, N. Perrimon, and G. Ruvkun for helpful discussions and A. Michelson and N. Perrimon for critical comments on the manuscript. M.R.J.H. was supported by a Minority Predoctoral Fellowship from the National Institutes of Health, grant no. GM-18903. L.A.P. is supported by the National Science Foundation, grant no. IBN-9904606, and by the Department of Surgery at the Massachusetts General Hospital.

Manuscript received May 30, 2001; Accepted for publication August 14, 2001.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ALLARD, J., H. C. CHANG, R. HERBST, H. MCNEILL, and M. A. SIMON, 1996  The SH2-containing tyrosine phosphatase corkscrew is required during signali