INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Receptor tyrosine kinases (RTKs) play critical roles in translating cues gathered from the extracellular environment into biological responses such as cellular differentiation, proliferation, and migration events. RTKs such as the platelet-derived growth factor (PDGF), epidermal growth factor (EGF), and fibroblast growth factor (FGF) receptors define a family of single-pass transmembrane proteins with an intracellular tyrosine kinase domain (51). Binding of growth factor to the extracellular portions of these receptors promotes receptor dimerization and activation of the intracellular tyrosine kinase domain (43). Receptor activation leads to receptor autophosphorylation as well as phosphorylation of cytoplasmic signaling proteins. Specific phosphotyrosine sites on the receptor can serve to propagate signaling by recruiting Src homology domain 2 (SH2) and PTB domain-containing signaling proteins directly to the activated receptor (37, 38). For example, many RTKs can signal to the RAS/mitogen-activated protein kinase (MAPK) cascade via direct recruitment of the SH2/SH3 domain-containing adaptor protein GRB2 in complex with SOS, the guanine nucleotide exchange factor for RAS (31). EGF and PDGF RTKs can utilize direct recruitment of the GRB2/SOS complex to achieve RAS activation. This canonical pathway is well-conserved in Caenorhabditis elegans, Drosophila, and mammalian systems (37).

There is compelling evidence that the RAS/MAPK signaling pathway plays an important role in signaling via FGF receptors (FGFRs) as well (32). However, unlike the EGF and PDGF receptors, FGFRs do not appear to recruit GRB2/SOS directly. Instead, mammalian FGFR-1 makes use of the multisubstrate adaptor (MSA) FRS2/SNT1 to link receptor activation to GRB2/SOS recruitment and the RAS/MAPK cascade (25, 53). FRS2 consists of an N-terminal myristylation sequence followed by a PTB domain and an extended C-terminal tail with a number of tyrosine motifs. Upon FGFR activation, FRS2 becomes highly tyrosine phosphorylated, creating docking sites for the SH2 domain of GRB2 as well as other SH2 domain-containing signaling components.

FRS2 typifies one class of MSA proteins. A second, structurally distinct class of MSAs includes mammalian GAB1 and Drosophila DOS (38). This second class of MSAs differs from FRS2 in both structure and function. GAB1 and DOS consist of an N-terminal pleckstrin homology (PH) domain followed by a C-terminal tail that contains a number of tyrosine and polyproline motifs. Unlike FRS2, which appears to act specifically in FGF and nerve growth factor RTK signaling, GAB1 and DOS have been implicated in a number of RTK signaling pathways (18, 20, 21, 34, 41, 54). The precise molecular mechanisms by which the GAB1/DOS-type MSAs function to mediate signaling remain unclear. In Drosophila, DOS function is intimately linked to the cytoplasmic tyrosine phosphatase Corkscrew (CSW) (3, 18, 19). Similarly, GAB1 can associate with the mammalian ortholog of CSW, SHP2, and may represent a similar signaling module in mammalian systems (20, 45).

In C. elegans, egl-15 encodes the sole FGFR and plays a critical role in multiple aspects of development (11). Modulating the strength of signaling through EGL-15 can have profound phenotypic effects. Hyperactive EGL-15 signaling results in a distinctive Clear (Clr) phenotype. This Clr phenotype can be observed in transgenic animals bearing a constitutively active version of EGL-15 or in animals that have lost the CLR-1 receptor tyrosine phosphatase that normally attenuates EGL-15 signaling (23). Animals that have completely lost egl-15 function arrest early in larval development, demonstrating an essential role for this receptor. Less severe reduction of egl-15 function can confer multiple developmental defects. These defects include a Scrawny (Scr) body morphology, an egg-laying-defective phenotype, and suppression of the Clr phenotype of clr-1 mutants. We have called this last phenotype Soc for suppression of Clr and have used the Soc phenotype as an indication of reduced levels of EGL-15 signaling (11).

Genetic analysis has begun to identify components that function downstream of EGL-15 in its various roles. Some of these roles, such as sex myoblast migration guidance and the essential function of EGL-15, appear to be functionally distinct and may involve different downstream signaling pathways. Genetic screens for suppressors of clr-1 (soc genes) have identified the most comprehensive set of downstream signaling components for EGL-15. These suppressor mutations include multiple alleles of four genes (egl-15, sem-5, soc-1, and soc-2/sur-8) as well as a single allele of let-341/sos-1 (11). Soc mutations in sem-5, the GRB2 ortholog in C. elegans (10); soc-2/sur-8, encoding a leucine-rich repeat protein that binds to LET-60 RAS (44, 46); and let-341/sos-1 (8) implicate the RAS/MAPK cascade in mediating this function of EGL-15 signaling. LET-60 RAS function has been implicated in other processes mediated by EGL-15 (9, 11, 49). Here we present direct evidence for the involvement of LET-60 RAS and members of the MAPK cascade in the Clr/Soc aspect of EGL-15 signaling. In addition, we have cloned soc-1 and present evidence that it functions in conjunction with the C. elegans SHP2 homolog, PTP-2 (13), to mediate a portion of the EGL-15 signal transduction cascade.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Isolation of let-60(ay75gf) A semiclonal screen for enhancers of the Clr phenotype of clr-1(e1745ts) was performed at 15°C, the permissive temperature for e1745ts. All manipulations were performed at 15°C. clr-1(e1745ts) hermaphrodites were mutagenized with ethyl methanesulfonate (EMS) as described previously (6). F1 progeny were picked to individual plates at five per plate, the F2 progeny were screened for Clr animals, and the mutations responsible for the Clr phenotype were recovered from F2 siblings of Clr animals. From animals representing approximately 6,000 mutagenized haploid genomes, two mutations that caused a recessive Clr phenotype at 15°C were identified. Besides the ay75 mutant, the other mutant isolated bore a second mutation in clr-1 that resulted in a nonconditional Clr phenotype (ay76). Although ay75 was obtained in a clr-1(e1745ts) background, it confers a Clr phenotype even in a clr-1(+) background. All subsequent characterization has been performed in a clr-1(+) background.

Isolation of let-60(ay75 ay100) Suppressors of ay75 were isolated based on its temperature-sensitive semidominant fertility defect. let-60(ay75) + unc-22(e66)/+ dpy-20(e1362) + hermaphrodites were mutagenized with EMS at 15°C, and the healthy F1 progeny (non-Dpy, non-Clr, non-Twi) were picked to a single plate and transferred to 25°C for 1 day. Healthy young adults were picked at one per plate; after an additional 2 days at 25°C, plates with large broods were identified. The broods from 1,678 F1 animals were screened, and two mutants (ay100 and ay101) were isolated, both of which segregated rod-like larval-lethal progeny characteristic of loss-of-function let-60 alleles. ay75 ay100 was backcrossed by mating N2 males into non-Dpy progeny [ay75 ay100 unc-22(e66)/dpy-20(e1362)]. The resulting males were crossed into dpy-20(e1362) hermaphrodites. Non-Dpy cross-progeny that segregated rod-like larval-lethal progeny were selected as the twice-backcrossed strain, and the process was repeated. The behavior of ay100 during the backcrossing protocol was consistent with it being an intragenic suppressor. ay101 was not characterized further.

Characterization of let-60(gf) phenotypes. In contrast to the Clr phenotype of clr-1 mutants, which can be observed at any postembryonic stage, let-60(ga89ts) and let-60(ay75ts) mutants display a late-onset Clr phenotype, becoming observably Clr only as adults. The let-60(ay75) mutant was not used for epistasis studies due to its low brood size. Instead, the let-60(ga89) mutant was used to assess the epistatic relationship between let-60 ras and the soc genes. Double mutant combinations between let-60(ga89) and the soc genes were assayed by shifting L4 hermaphrodites to 25°C and scoring for the Clr or Soc phenotype approximately 36 h later. Due to a severe egg-laying-defective phenotype, the Clr phenotype of let-60(ga89); sem-5(n1779) was assessed in gonad-ablated animals; gonads in L1 animals were ablated, and the animals were shifted to 25°C. The strain was severely developmentally delayed, and the Clr phenotype was apparent after 5 days at 25°C.

Cloning of soc-1 The soc-1 map position was refined by mapping with respect to restriction fragment length polymorphisms, which placed soc-1 to the left of the RW#L63 polymorphism (L. DeLong and M. J. Stern, personal communication).

Germ line transformation of ptp-2 Due to the recessive sterility of ptp-2(op194), ptp-2-rescuing activity was assayed by injecting tester DNA at 20 ng per µl with 50 to 100 ng of pRF4 DNA per µl into gravid clr-1(e1745ts) ptp-2(op194)/mIn1 hermaphrodites, establishing transformed lines, and scoring the clr-1(e1745ts) ptp-2(op194) homozygous progeny for a Clr or Soc phenotype. mIn1 is an inversion on chromosome II that balances clr-1 and ptp-2. mIn1 is homozygous viable and contains a mutation in dpy-10 that confers a Dpy phenotype in mIn1 homozygotes (M. Edgley, personal communication). Injected animals were allowed to lay eggs at 15°C; transformed F1 Rol progeny were transferred to 25°C. As for soc-1 rescue, ptp-2-rescued animals are Clr at 25°C; nonrescued animals retain the Soc phenotype at 25°C. Approximately one-third of the non-Dpy transformed progeny are of the genotype clr-1(e1745ts) ptp-2(op194) and are relevant to assay for rescuing activity of the transgene; the other non-Dpy progeny are heterozygous for ptp-2(op194) and non-Clr at all temperatures regardless of the transgene. Heterozygotes were identified by segregating 1/4 Dpy mIn1 progeny. Stably transformed lines were isolated from these transformed heterozygotes and scored as described for F1 transformants. A wild-type genomic 5.3-kb PstI-XbaI fragment was isolated from the cosmid F59G1 that contained the region of overlap from the ptp-2-rescuing cosmids (13). This construct (NH#772) confers robust rescue of the Soc phenotype but very weak rescue of the fertility defect of ptp-2(op194). Rescue of the fertility defect was not assayed in the mutant constructs.

soc-1 cDNA isolation and allele sequencing. A soc-1 cDNA was isolated using reverse transcription-PCR and 5' and 3' rapid amplification of cDNA ends. Primers were designed based on the exon-intron structure of the gene F43F1.2 predicted by the program GENEFINDER. These were used to amplify a soc-1 cDNA from a mixed-stage C. elegans cDNA library. A 1,350-bp cDNA with an exon-intron structure similar, although not identical, to that of F43F1.2 (GenBank accession no. AF419335) was isolated. BLAST searches of nonredundant genomic DNA databases at the National Center for Biotechnology Information using the soc-1 cDNA sequence identified significant homology only to the PH domain of GAB1.

soc-1 and ptp-2 mutant constructs. Tyrosine-to-phenylalanine changes in SOC-1 were introduced into a soc-1 cDNA (NH#693) using the Transformer site-directed mutagenesis kit (Clontech). To facilitate testing of these constructs in the germ line transformation rescue assay, restriction sites were introduced into both the soc-1 cDNA and genomic rescuing constructs: an AgeI restriction site was introduced just upstream of the endogenous stop codon, and a NotI site was introduced just downstream of the stop codon. The AgeI site was positioned to allow GFP-coding sequences from pD95.67 (kindly provided by A. Fire, J. Ahnn, G. Seydoux, and S. Xu) to be added in frame to any modified construct. A C-terminal portion of the cDNA containing these site-directed mutations was cloned into the soc-1 genomic rescuing construct using an endogenous SphI site just downstream of the PH domain and the engineered AgeI restriction site.

sem-5 noncomplementation screen. Null alleles of sem-5 were isolated based on two known characteristics: (i) noncomplementation of the Soc phenotype of sem-5(n1779) and (ii) the rod-like larval-lethal phenotype of strong alleles of sem-5 (10). clr-1(e1745ts); sem-5(n1779)/0 males were mated into EMS-mutagenized clr-1(e1745ts); lon-2(e678) hermaphrodites at 15°C. F1 progeny were transferred to 25°C to select for Soc cross-progeny; those which segregated rod-like larval lethals were characterized further. Two new mutations, ay73 and ay74, were isolated from animals representing approximately 12,000 mutagenized genomes. ay53 was isolated from a similar screen that did not include the lon-2(e678) marker, representing approximately 8,000 additional mutagenized genomes. Strains homozygous for ay53 could be maintained, while those with either ay73 or ay74 were maintained as balanced heterozygotes using the szT1[lon-2(e678)] balancer. Sequence determination revealed that ay53 corresponds to a molecular lesion that is identical to that found in the n1619 allele (P49L), ay73 corresponds to a nonsense mutation at Q10, and ay74 corresponds to a mutation in the SH2 domain (R87Q). Both ay73 and ay74 are partially penetrant rod-like larval-lethal mutations; the surviving homozygotes are extremely Scr and have no live self-progeny. ay73 is used as a sem-5 null allele due to its strong phenotype and its molecular lesion. ay74 displays a more penetrant rod-like larval-lethal phenotype, possibly due to dominant negative characteristics interfering with the maternal contribution of sem-5.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

let-60 ras is critical for EGL-15 signaling. In a screen for additional Clr mutants (see Materials and Methods), a single mutation, ay75, that was not an allele of clr-1 was isolated. The original characterization of this mutant was performed at 15°C, where ay75 confers a recessive late-onset Clr phenotype (Fig. 1D and 2A). On the basis of its recessive Clr phenotype at 15°C, ay75 was mapped by standard sequence-tagged site and multifactor mapping to a region of linkage group IV which contains the let-60 ras gene. Gain-of-function mutations in let-60 ras can confer a spectrum of phenotypes similar to those displayed by ay75 mutants (see below), leading us to test whether ay75 was associated with a lesion in the let-60 ras gene. Sequence analysis of the let-60 ras gene in ay75/ay75 homozygotes revealed a G₁₇₈-to-A transition which results in a G60R alteration, demonstrating that ay75 is an allele of let-60.

View larger version (76K): [in this window] [in a new window]   FIG. 1.   let-60(ay75gf) confers a Clr phenotype. (A) Wild type. (B) clr-1(n1992) null mutant, which displays a severe Clr phenotype. (C) clr-1(e1745ts) animal shifted from 15 to 25°C at mid-L4; these animals develop a dramatic Clr phenotype within a normal body shape. (D) let-60(ay75gf) homozygous animals raised at 15°C display a Clr phenotype similar to that of the clr-1(e1745ts) animal shown in panel C. (E) let-60(ay75)/+ heterozygous animals raised at 25°C display a dominant Clr phenotype. This animal also displays a protruding excretory pore and vulva. (F) let-60(ay75 ay100)/+ heterozygous animals raised at 25°C are non-Clr, demonstrating the cis-dominant effect of the intragenic suppressor, ay75 ay100. Bar, 200 µm.

View larger version (37K): [in this window] [in a new window]   FIG. 2.   Penetrance of the Clr (A) and Muv (B) phenotypes associated with let-60(ay75gf). Each phenotype was scored individually in a set of animals (>100) for each of the following genotypes: wild type, ay75/+ heterozygotes raised at 15 and 25°C, ay75/ay75 homozygotes raised at 15°C, ay75 ay100/+ raised at 25°C, and the canonical gain-of-function C. elegans ras allele, let-60(n1046gf), for comparison at 15 and 25°C. The penetrance of the Clr or Muv phenotype is represented as a percentage of the total population examined; error bars represent 95% confidence intervals. For clarity, error bars are not shown for penetrances that are less than 1%; all of these values have 95% confidence intervals that are also less than 1%.

The MAPK cascade is required for the Clr phenotype. Many functions of LET-60 RAS are mediated by the MAPK cascade, which consists of the serine/threonine kinases LIN-45 RAF, MEK-2 MAPKK, and MPK-1 MAPK (17, 24, 26, 56). We tested loss-of-function mutations in the genes encoding MAPKK and MAPK for their ability to suppress the Clr phenotype of clr-1(e1745ts). While mutations in these genes were not isolated in the original Soc screen, alleles of both mek-2 (n2678) and mpk-1 (ku1 and ga117) were found to confer a Soc phenotype when examined in a clr-1(e1745ts) background (Table 2), implicating this MAPK cascade in EGL-15 signaling. Due to interfering pleiotropies, we were unable to test lin-45 raf directly, but the involvement of let-60, mek-2, and mpk-1 strongly implicate LIN-45 RAF as a component of this EGL-15 pathway.

Epistasis analysis of soc genes using let-60(gf) We used the Clr phenotype conferred by let-60(ga89) to analyze the epistatic relationship between the soc genes and let-60 ras. Consistent with the well-characterized roles that these soc genes play in RTK signaling, let-60 is epistatic to egl-15, sem-5, and let-341/sos-1, while mpk-1 is epistatic to let-60 (Table 2). let-60 is also epistatic to soc-1, suggesting that soc-1 acts either between EGL-15 and RAS or independently of LET-60 RAS. To understand the function of soc-1 better, we have molecularly and genetically characterized its role in EGL-15 signaling pathways.

soc-1 encodes an MSA. soc-1 was cloned by a germ line transformation rescue assay using yeast artificial chromosomes (YACs) and cosmids corresponding to its genetic map position. The soc-1-rescuing region was initially defined by rescue with the YAC Y47D10 (L. DeLong and M. J. Stern, unpublished data). Subsequently, a single cosmid, T04H9, that efficiently rescues the Soc phenotype of clr-1(e1745ts); soc-1(n1789) was identified. The rescuing activity of T04H9 was pared down to a 10-kb XhoI-to-SacII fragment (Fig. 3A). This rescuing fragment contains a single open reading frame according to GENEFINDER predictions. We cloned a soc-1 cDNA using reverse transcription-PCR and 5' and 3' rapid amplification of cDNA ends. The resulting 1,350-bp cDNA contains a single open reading frame with an in-frame stop codon preceding the first potential ATG start site. Using the germ line transformation rescue assay, we have demonstrated that this cDNA can restore wild-type soc-1 function (Fig. 4). Similar, although not identical, to GENEFINDER predictions, this cDNA encodes a protein of 430 amino acids with an N-terminal PH domain. PH domains are found in a variety of signaling molecules with and without enzymatic activity and are thought to mediate targeting of these proteins to the cell membrane through interactions with phosphoinositides and/or interactions with other proteins (27, 33). The PH domain of SOC-1 is most similar to the PH domains of the MSA proteins Drosophila DOS and mammalian GAB1 (Fig. 3D) (20, 41). MSA proteins link RTKs with downstream pathways by recruiting SH2 and SH3 domain-containing signaling components to assemble a signaling complex at the plasma membrane near the activated receptor (37, 38). The C-terminal two-thirds of SOC-1 does not share significant sequence identity with any other known protein but has 12 potential tyrosine phosphorylation sites (Fig. 3B). In addition, there are two polyproline motifs in the C terminus of SOC-1 that could function as binding sites for SH3 domain-containing proteins. Despite lacking a definitive sequence ortholog, its overall structural organization, PH domain similarity, and tyrosyl and polyprolyl motifs strongly suggest that SOC-1 is an MSA protein most similar to DOS and GAB1.

View larger version (42K): [in this window] [in a new window]   FIG. 3.   soc-1 encodes an MSA protein. (A) Cloning of soc-1 by germ line transformation rescue. The genetic map position of soc-1 is shown with respect to flanking genes and Bristol/Bergerac restriction fragment length polymorphisms. The clones tested for soc-1-rescuing activity are placed with respect to the rescuing YAC, Y47D10. Germ line transformation rescue results for the cosmids (T04H9 and W06H9) and derivatives of the rescuing cosmid T04H9 are shown as the number of stably transformed lines capable of rescuing the Soc phenotype of clr-1(e1745ts); soc-1(n1789) animals; soc-1 rescue with Y47D10 was tested in clr-1(e1745ts); soc-1(n1833) animals. A break in the bar designates a deletion. The position of the minimal rescuing fragment is indicated by shading. Relevant restriction sites: A, ApaI; P, PstI; S, SacII; X, XhoI. (B) Schematic representation of SOC-1. The open box represents the N-terminal PH domain. The positions of allele lesions are noted at the top. The sequences of the C-terminal tyrosyl and polyprolyl motifs are listed at the bottom; 4 of the 12 tyrosines within the C terminus of SOC-1 are found within recognizable SH2 domain binding motifs (47). Three of these tyrosine motifs correspond to predicted docking sites for two SH2 domain-containing soc genes: Y202NNL (SEM-5/GRB2), Y340RNL (SEM-5/GRB2), and Y408ISV (PTP-2/SHP2). In addition, Y181HRM is a predicted phosphatidylinositol 3-kinase SH2 domain binding motif. Arrowheads indicate the 3' extent of the C-terminal truncations (see Fig. 5). (C) Mutational lesions. soc-1 alleles are listed with their corresponding DNA lesions, numbered according to the soc-1 cDNA starting at the initiation codon. All mutations were induced using EMS mutagenesis, except for n1789, n1833, and n1834, which were induced with gamma rays (11). (D) Alignment of PH domains from SOC-1, Drosophila DOS, and human GAB1. Amino acid sequences corresponding to the PH domain as defined by the Pfam protein motif finder from SOC-1 (positions 8 to 132), DOS (4 to 112), and GAB1 (6 to 115) were aligned using the CLUSTAL method and DNAstar sequence analysis software. Sequence identities are shown in black, and similarities are shaded. Subdomain locations are indicated with a bar and a roman numeral below the alignment (33); the invariant tryptophan found in all PH domains is indicated with an asterisk above the alignment. Amino acid positions with respect to the SOC-1 sequence are noted above the alignment.

View larger version (23K): [in this window] [in a new window]   FIG. 4.   Functional analysis of potential protein binding sites in the C-terminal tail of SOC-1. (A) Canonical SEM-5 binding sites in SOC-1 are dispensable for function. (B) The C-terminal Y408 is critical for SOC-1 function. (C) Fusing an active phosphatase domain from PTP-2 to SOC-1 can enhance signaling by SOC-1(Y408F). WT, wild type. The SOC-1 PH domain is depicted by solid rectangles, and the SOC-1 C-terminal tail is indicated by a thick line. Injected constructs contain a single intron between exons one and two. Tyrosine-to-phenylalanine and proline-to-alanine changes are depicted above each construct. The C-terminal SH2 domain (C-SH2) and phosphatase domain (PTP) of PTP-2 are depicted as shaded boxes. PTP C518S bears a cysteine-to-serine mutation within the catalytic core of PTP-2. Data shown are for untagged constructs in panels A and C and for GFP-tagged constructs in panel B. Stably transformed lines were isolated as described for Fig. 3 and tested for rescue of the Soc phenotype of clr-1(e1745ts); soc-1(n1789) by shifting L4 transformants to the nonpermissive temperature. The extent of rescue for each of the lines was assigned to one of five categories: (i) complete, animals display a Clr phenotype similar to that seen when clr-1(e1745ts) L4 animals are shifted to the nonpermissive temperature; (ii) strong, animals are sterile and display a reduced but obvious Clr phenotype; (iii) intermediate, animals are not sterile and display an obvious Clr phenotype; (iv) weak, animals are fertile and only subtly Clr (defect confirmed by Nomarski microscopy); or (v) none, animals are completely Soc and indistinguishable from the injected strain clr-1(e1745ts); soc-1(n1789).

Structure-function analysis of SOC-1. soc-1 and sem-5/GRB2 are both soc genes that behave similarly in epistasis tests (Table 2) (44), suggesting the possibility that SEM-5 might associate directly with SOC-1 to mediate EGL-15 signal transduction. SOC-1 has several motifs in its C terminus with the potential to interact directly with SEM-5. These include two YXNX motifs and two polyproline motifs that are potential binding sites for SEM-5 via its SH2 and SH3 domains, respectively (Fig. 3B) (22, 30, 47). To assess the importance of these motifs for SOC-1 function, we tested variants of the soc-1-rescuing fragment in which these motifs were specifically mutated (Fig. 4A). soc-1-rescuing fragments with altered YXNX or polyproline motifs were introduced into clr-1(e1745ts); soc-1(n1789) mutant animals, and stably transformed progeny were scored for rescue of the Soc phenotype. For both the YXNX and polyproline motifs, the mutated variants rescued the soc-1(n1789) defect as well as the wild-type soc-1 construct, suggesting that neither of these motifs on its own is essential for SOC-1 function.

View larger version (14K): [in this window] [in a new window]   FIG. 5.   Both the PH domain and C-terminal tail are required for SOC-1 function. Injected constructs are diagrammed as described for Fig. 4, and amino acid positions are noted above the diagram. The PH domain mutation, W124A, and its wild-type (WT) control were tested in an intronless construct without GFP. When GFP was appended to monitor expression, these constructs showed a pattern of expression that was indistinguishable from that of the wild-type control. The extent of the depicted truncations with respect to the tyrosyl and polyprolyl motifs within the C terminus of SOC-1 is indicated by arrowheads in Fig. 3B.

SOC-1-PTP-2 association appears to be critical for EGL-15 signaling. SOC-1 is structurally similar to the Drosophila MSA DOS. An important function of DOS appears to be recruitment of the tyrosine phosphatase CSW (3, 18, 19). An ortholog of vertebrate SHP2, CSW functions in a number of Drosophila RTK signaling pathways, including the Breathless FGFR pathway (39, 40). SHP2 in vertebrates is also known to function in signaling via FGFRs, for example, during mesoderm induction in Xenopus development (50) and in the differentiation of PC12 cells in culture (14). These findings suggest that the SHP2 ortholog in C. elegans might function in EGL-15 FGFR signal transduction.

SOC-1 does not mediate all aspects of EGL-15 signaling. In addition to conferring a Soc phenotype in a clr-1 mutant background, soc-1 null alleles confer a Scr body morphology in a wild-type background (Fig. 6C). This phenotype is similar to the Scr defect observed in animals homozygous for hypomorphic alleles of egl-15 (11). Similar to the case for these egl-15 alleles, the Scr phenotype of soc-1 null mutants is predominantly suppressed in a clr-1(e1745ts) background, resulting in an animal that appears to be essentially wild type (Fig. 6D). The ability of clr-1(e1745ts) to suppress the Scr phenotype of soc-1 null mutants even at the permissive temperature for clr-1(e1745ts) suggests that soc-1 null mutants reduce signaling to a level that is sensitive to small perturbations in pathway activity. Distinct from the phenotype of soc-1(n1789) null mutants, egl-15 null mutants display a more severe phenotype and arrest in the first larval stage (Fig. 6B) (11). This result indicates that some signaling activities of EGL-15 are independent of SOC-1.

View larger version (139K): [in this window] [in a new window]   FIG. 6.   Interactions between genes involved in egl-15 signaling. Bright-field photographs comparing the Scr phenotypes of worms carrying mutations in genes involved in egl-15 signaling are shown. All animals were photographed at a magnification of ×10; bar, 200 µm. The animals pictured are representative of nonoverlapping phenotypic classes where wild-type or non-Scr and Scr are clearly distinguishable. (A) Wild type (WT). (B) egl-15(n1456) null mutant arrested larva. (C) soc-1(n1789) Scr hermaphrodite. (D) clr-1(e1745ts); soc-1(n1789) Soc, non-Scr hermaphrodite. (E) ptp-2(op194) non-Scr hermaphrodite. (F) clr-1(e1745ts) ptp-2(op194); soc-1(n1789) non-Scr hermaphrodite. (G) sem-5(ay73) Scr hermaphrodite. (H) sem-5(n1781) non-Scr hermaphrodite. (I) clr-1(e1745ts); soc-1(n1789); sem-5(n1781) extremely Scr hermaphrodite. The animals shown in panels B, E, F, and G are homozygotes derived from mothers heterozygous for either egl-15(n1456), ptp-2(op194), or sem5(ay73).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

While several components that are critical to EGL-15 signaling have been described previously (8, 10, 11, 44), we have now identified many additional components which allow us to begin to assemble an EGL-15 signaling pathway. We present evidence supporting the involvement of LET-60 RAS and the MAPK cascade in EGL-15 FGFR signal transduction. We isolated a new allele of let-60 ras, ay75, and showed that it is a gain-of-function allele that confers a Clr phenotype, characteristic of hyperactivation of the EGL-15 FGFR. Additionally, we showed that loss of LET-60 RAS can confer a Soc phenotype, supporting a normal role for LET-60 RAS in this Clr/Soc pathway. Mutations in mek-2 and mpk-1 are also Soc, implicating the MAPK cascade in mediating this aspect of EGL-15 signaling. Thus, our data strongly suggest that EGL-15 signals via SEM-5/SOS-1 to the RAS/MAPK pathway. Exactly how the signal is transduced from EGL-15 to this cascade remains unresolved.

To further our understanding of EGL-15 signal transduction, we cloned soc-1 and demonstrated that it has both structural and functional similarities to Drosophila DOS. Like DOS, SOC-1 has an N-terminal PH domain followed by a C-terminal tail with tyrosine residues and polyproline motifs, suggesting that it functions as an MSA protein. In Drosophila RTK signal transduction pathways, DOS function is closely linked to the tyrosine phosphatase CSW (3, 18, 19). We have shown that the C. elegans ortholog of CSW, ptp-2, like soc-1 is also a soc gene, supporting a similar relationship between SOC-1 and PTP-2 in EGL-15 signaling.

Our genetic and structure-function analyses support the hypothesis that SOC-1 and PTP-2 function as a signaling cassette to transduce the EGL-15 signal. First, we found that SOC-1 Y₄₀₈, a predicted SHP2/PTP-2 binding site, is critical for SOC-1 function. Experiments with Drosophila have demonstrated that a similar C-terminal CSW binding site is also critical for DOS function (3, 19). Second, PTP-2 requires an intact SH2 domain(s) to mediate EGL-15 signaling. Third, a chimeric protein that fuses the C-terminal SH2 and phosphatase domains of PTP-2 to SOC-1(Y408F) can restore some of the lost signaling potential of SOC-1(Y408F). Fourth, although both SOC-1 and PTP-2 function to mediate EGL-15 signaling, ptp-2; soc-1 double mutants do not reduce EGL-15 pathway activity to a greater extent than the single mutants. These data support a model in which SOC-1 and PTP-2 function together to mediate EGL-15 signaling. More specifically, in response to EGL-15 activation, SOC-1 is likely phosphorylated at Y₄₀₈, enabling SOC-1 to recruit PTP-2 via a direct interaction between the C-terminal tail of SOC-1 and the SH2 domain(s) of PTP-2.

Two paradigms from other signaling systems provide a context for thinking about how a SOC-1/PTP-2 cassette might function in EGL-15 signaling. The first of these paradigms rests on the high degree of structural similarity between SOC-1 and the MSAs GAB1 and DOS. These proteins all have N-terminal PH domains followed by a series of tyrosyl and polyprolyl motifs in their C-terminal tails. These domains not only share a common structure but also appear to share common functions. The SOC-1 PH domain is most similar to the GAB1 PH domain, and the PH domains of GAB1 and DOS are functionally interchangeable, suggesting that a common function is carried out by these PH domains (3). The C-terminal tails of SOC-1 and DOS perform a common function to recruit PTP-2/CSW. This function is likely to be conserved in mammals, since the association of GAB1 with SHP2 plays an important role during epithelial morphogenesis (29, 42).

In addition to its association with SHP2, GAB1 has a well-established relationship with GRB2. Initially identified as the GRB2-associated binder 1, GAB1 associates with GRB2 by a phosphotyrosine-independent mechanism (20). We tested a possible association between SOC-1 and SEM-5 and found that neither the canonical SEM-5 SH2 binding sites nor the canonical SEM-5 SH3 binding sites in SOC-1 are essential for SOC-1 function. More recently, the association between GAB1 and GRB2 has been shown to be mediated in part by a noncanonical GRB2 SH3 domain binding site (termed GBS1) as well as a canonical GRB2 SH3 binding site (termed GBS2) (42). This newly identified GBS1 motif is found in both DOS and SOC-1, further supporting the structural homology among these proteins. The GBS1 motif in SOC-1 presents an alternative mechanism by which SOC-1 may associate with SEM-5 in a manner similar to that described for GAB1.

Although structurally distinct from SOC-1, the MSA protein FRS2 offers a second paradigm to consider as a functional analog for SOC-1. FRS2 plays a key role in FGFR signaling by linking the FGFR to GRB2 and the RAS/MAPK cascade (25, 53). In addition to recruiting GRB2 directly, FRS2 can associate with SHP2, offering an indirect means to recruit GRB2 (14). This recruitment could be mediated via known GRB2 binding sites within the C terminus of SHP2 (5). Our results indicate that SOC-1 does not act similarly. First, since its phosphotyrosine motifs predicted to bind SEM-5 are not essential for SOC-1 function, SOC-1 does not appear to recruit SEM-5 directly. Second, although our data suggest a direct interaction between PTP-2 and SOC-1 that is important for SOC-1 function, SOC-1 is unlikely to recruit SEM-5 via PTP-2, since PTP-2 does not have canonical GRB2 binding sites within its C-terminal tail. Thus, if SOC-1 recruits SEM-5 to activate RAS signaling, it must accomplish this by a different mechanism.

If SOC-1 does not play a direct role in linking EGL-15 signaling to the activation of the RAS/MAPK cascade, how might this link be made? Unlike mammalian FGFRs, EGL-15 may be able to associate directly with GRB2 through phosphotyrosyl motifs within its intracellular region. EGL-15 contains two potential phosphotyrosine motifs predicted to bind the SH2 domain of SEM-5, although these do not appear to be essential for EGL-15 function (C. Branda and M. J. Stern, unpublished data). Alternatively, a C. elegans predicted gene, F54D12.6 (7), that has similarity to FRS2 may perform an FRS2-like function to recruit GRB2/SOS/RAS during EGL-15 signaling. The role of this gene in EGL-15 signaling is currently being investigated.

It is not yet clear which of these two paradigms SOC-1 follows most closely. Similar to FRS2, SOC-1 appears to be a relatively specific adaptor for EGL-15 signaling. This is in contrast to GAB1 and DOS, which have been implicated in a number of RTK pathways and may represent more promiscuous adaptor proteins. Mechanistically, however, SOC-1 appears to function most like DOS in that one of its critical functions is to recruit PTP-2/CSW during signal transduction. Although DOS has not been implicated in FGFR signaling to date, a role for GAB1 was recently identified in linking FGFR activation with phosphatidylinositol 3-kinase signaling (34). The identification of SOC-1 as an important component of EGL-15 FGFR signaling implicates GAB1 or DOS in FGFR signaling in their respective systems as well.

Our data establish that the Clr/Soc EGL-15 pathway utilizes a RAS/MAPK signaling cascade and also define a role for the SOC-1/PTP-2 cassette in EGL-15 FGFR signaling. Several lines of evidence suggest that SOC-1/PTP-2 may act in a parallel signaling branch that has a modulatory effect on RAS signaling downstream of EGL-15 (Fig. 7). First, we have observed a strong synergism between sem-5 mutations and null alleles of soc-1. Second, a supporting parallel pathway might explain why two different let-60 gain-of-function mutations confer only a late-onset Clr phenotype rather than the full Clr phenotype observed in clr-1 mutants. Third, a parallel-pathway model can help explain why so many of these soc genes, including let-60, ptp-2, and soc-1, require strong mutations to confer a Soc phenotype. Lastly, this model is supported by data for Drosophila that suggest a complex interdependence of the RAS and DOS/CSW pathways (1, 2). Unraveling the precise molecular mechanisms by which SOC-1/PTP-2 contributes to signaling will be crucial to a more complete understanding of RTK signal transduction.

View larger version (15K): [in this window] [in a new window]   FIG. 7.   Model of C. elegans EGL-15/FGFR signaling. EGL-15 activation leads to the phosphorylation of SOC-1 on tyrosine 408; this event triggers PTP-2 recruitment and/or activation. In addition, EGL-15 activates the canonical SEM-5/SOS-1/RAS/MAPK pathway. Potential targets of the SOC-1/PTP-2 cassette and/or its role in modulating activity of the RAS/MAPK cascade are indicated by dashed lines. Precisely how activation of EGL-15 triggers SOC-1 and SEM-5 to participate in EGL-15 signal transduction remains unresolved (indicated by dashed lines from the receptor).