DOI: 10.1534/genetics.109.113373
Caenorhabditis elegans
Fibroblast Growth Factor Receptor Signaling Can
Occur Independently of the Multi-Substrate Adaptor FRS2
Te-Wen Lo
,*
,1Daniel C. Bennett
,*
,2S. Jay Goodman
†and
Michael J. Stern
*
,3*Department of Genetics and†Department of Cell Biology, Yale University School of Medicine, New Haven, Connecticut 06520-8005
Manuscript received December 21, 2009 Accepted for publication March 14, 2010
ABSTRACT
The components of receptor tyrosine kinase signaling complexes help to define the specificity of the effects of their activation. TheCaenorhabditis elegans fibroblast growth factor receptor (FGFR),EGL-15, regulates a number of processes, including sex myoblast (SM) migration guidance and fluid homeostasis, both of which require a Grb2/Sos/Ras cassette of signaling components. Here we show thatSEM-5/Grb2 can bind directly toEGL-15to mediate SM chemoattraction. A yeast two-hybrid screen identifiedSEM-5as able to interact with the carboxy-terminal domain (CTD) ofEGL-15, a domain that is specifically required for SM chemoattraction. This interaction requires theSEM-5SH2-binding motifs present in the CTD (Y1009and Y1087), and these sites are required for the CTD role ofEGL-15in SM chemoattraction.SEM-5, but not the SEM-5binding sites located in the CTD, is required for the fluid homeostasis function ofEGL-15, indicating thatSEM-5can link toEGL-15through an alternative mechanism. The multi-substrate adaptor protein FRS2 serves to link vertebrate FGFRs to Grb2. InC. elegans, an FRS2-like gene,rog-1, functions upstream of a Ras/ MAPK pathway for oocyte maturation but is not required forEGL-15function. Thus, unlike the vertebrate FGFRs, which require the multi-substrate adaptor FRS2 to recruit Grb2,EGL-15can recruitSEM-5/Grb2 directly.
F
IBROBLAST growth factors (FGFs) play important roles in many developmental and physiological processes, including cell migration, angiogenesis, pro-liferation, differentiation, and survival (Ornitz and Itoh 2001; Polanska et al. 2009). Mammals have a battery of both FGF ligands and high-affinity receptors to carry out this diverse set of important functions. These ligands and their receptors are generated from a set of 18 genes encoding FGFs and 4 genes encoding their receptors (Eswarakumar et al. 2005). Upon li-gand binding, fibroblast growth factor receptors (FGFRs) dimerize, activating their intrinsic tyrosine kinase activ-ity, which causes both autophosphorylation on intracel-lular tyrosine residues and phosphorylation of additional substrates (Eswarakumaret al. 2005). These phosphor-ylation events lead to the assembly of a signaling complex around the activated receptor, ultimately promoting various downstream signaling pathways (Eswarakumar et al. 2005).A large portion of mammalian FGFR signaling is mediated by the multi-substrate adaptor protein FRS2/ snt-1 (Kouhara et al. 1997; Hadari et al. 1998, 2001; Laxet al. 2002; Gotohet al. 2005). FRS2 constitutively associates with the juxtamembrane region of the FGFR via its amino-terminal PTB domain (Xuet al. 1998;Ong et al. 2000). Upon FGFR activation, FRS2 becomes heavily phosphorylated, allowing it to recruit Grb2 and Shp2 via their SH2 domains (Kouharaet al. 1997; Eswarakumar et al. 2005). Since these components cannot associate with the receptor in the absence of FRS2 (Hadariet al. 2001), FRS2 serves as an essential link between the activated receptor and many down-stream signal transduction pathways.
The understanding of FGF-stimulated signal trans-duction pathways has been aided by the study of FGF signaling in model organisms. Powerful genetic screens and the reduced complexity of the set of FGFs and their receptors in bothDrosophila melanogasterand Caenorhab-ditis elegans have helped promote an understanding of the conserved aspects of FGF signaling pathways (Huangand Stern2005; Polanskaet al. 2009). InC. elegans, FGF signaling is mediated by two FGF ligands, EGL-17andLET-756, and a single FGF receptor,EGL-15 (DeVoreet al. 1995; Burdineet al. 1997; Roubinet al. 1999). TheEGL-15FGFR is structurally very similar to mammalian FGF receptors, with the highest level of sequence conservation found within the intracellular Supporting information is available online athttp://www.genetics.org/
cgi/content/full/genetics.109.113373/DC1.
1Present address:Department of Molecular and Cell Biology, University of California, 131 Koshland Hall, Berkeley, CA 94720-3204.
2Present address:Department of Therapeutic Radiology, Yale University, 15 York St., HRT 210, New Haven, CT 06510-3221.
3Corresponding author:College of Graduate Studies, MH230, University of Central Florida, P.O. Box 160112, 4000 Central Florida Blvd., Orlando, FL 32816-0112. E-mail: [email protected]
tyrosine kinase domain and the three extracellular immunoglobulin (IG) domains.
Similar to mammalian FGFRs, alternative splicing also generates functionally distinctEGL-15isoforms. A ma-jor structural difference between EGL-15 and other FGFRs lies in an additional domain located between the first IG domain and the acid box ofEGL-15. This EGL-15-specific insert is encoded by a pair of mutually exclusive fifth exons, generating twoEGL-15isoforms, 5A and 5B, with different functions (Goodman et al. 2003). Alternative splicing also affects the sequence at the very end of the carboxy-terminal domain (CTD) of EGL-15 (Goodman et al. 2003), giving rise to four distinct C-terminal isoform types, referred to as types I–IV (see Figure 1A).
EGL-15, like its mammalian orthologs, is also involved in a large variety of functions, including cell migration guidance affecting the sex myoblast (SM) (Sternand Horvitz 1991; DeVore et al. 1995; Goodman et al. 2003) and CAN cells (Fleminget al. 2005), muscle arm extension (Dixon et al. 2006), a number of processes controlling terminal axon morphology (Bulow et al. 2004), muscle protein degradation (Szewczyk and Jacobson 2003), and fluid homeostasis (Huang and Stern 2004). A conserved FGFR signaling pathway in C. elegans was established by identifying the genes necessary for the role ofEGL-15in fluid homeostasis, and much of this same pathway is utilized in other functions ofEGL-15(DeVoreet al. 1995; Borlandet al. 2001; Huangand Stern2004, 2005). The identification of these genes was facilitated by a temperature-sensitive mutation affecting a receptor tyrosine phosphatase, CLR-1(CLeaR), which functions to negatively regulate EGL-15signaling (Kokelet al. 1998). Mutations inclr-1 abolish this regulatory constraint onEGL-15, resulting in fluid accumulation in the pseudocoelomic cavity due to hyperactive EGL-15 signaling. The buildup of this clear fluid, resulting in the Clr phenotype, is easily scored and can be used to identify suppressors (soc, suppressor of Clr) that reduce EGL-15 signaling effi-ciency. These suppressors define an EGL-15signaling pathway necessary for fluid homeostasis, which involves the activation of the Ras/MAPK cascade via theSEM-5/ Grb2 adaptor protein, thelet-341/sos-1Sos-like guanine nucleotide exchange factor, and a PTP-2-SOC-1/Shp2-Gab1 cassette (Borlandet al. 2001).
Mutations inegl-15have been identified on the basis of their effects on either fluid homeostasis or the guidance of the migrating SMs (DeVoreet al. 1995; Goodmanet al. 2003). Most of these alleles fall into an allelic series and affect the general aspects ofEGL-15signaling (Goodman et al. 2003). The strongest of these alleles confers an early larval arrest phenotype, whereas weaker alleles confer either a scrawny body morphology (Scr) or just the ability to suppress the Clr phenotype (Soc). While the weakest of these alleles does not affect egg laying, more highly compromised mutants show an egg-laying defect due to
the mispositioning of the SMs. Fouregl-15alleles specif-ically affect SM migration; when homozygous, these mutations cause dramatic mispositioning of the SMs, but do not cause a Soc phenotype (Goodman et al. 2003). Three of these are nonsense mutations in exon 5A and eliminate the 5AEGL-15isoform. The phenotype of these mutants highlights the specific requirement of the 5A isoform for SM migration guidance. The fourth mutation in this class, egl-15(n1457), is a non-sense mutation that truncates the carboxy-terminal domain, specifically implicating the CTD in SM migra-tion guidance.
Immediately following their birth at the end of the first larval stage (L1), the two bilaterally symmetric SMs undergo anteriorly directed migrations to final positions that flank the precise center of the gonad (Sulston and Horvitz 1977). In the middle of the third larval stage, the SMs divide to generate 16 cells that differentiate into the egg-laying muscles. Multiple mechanisms help guide the migrations of the SMs (Chen and Stern 1998; Branda and Stern 2000), including a chemoattraction mediated byEGL-15that guides the SMs to their precise final positions (Burdine et al. 1998). The EGL-17 FGF serves as the chemo-attractive cue, emanating from central gonadal cells (Branda and Stern 2000). In the absence of this chemoattraction, SMs are posteriorly displaced (Stern and Horvitz 1991). While mispositioned SMs still generate sex muscles, these muscles end up too far posterior to attach properly, causing the animal to be defective in egg laying (Egl).
The signal transduction pathway downstream ofEGL-15 that mediates SM chemoattraction is not well estab-lished. Several lines of evidence implicateSEM-5/Grb2, LET-341/Sos, andLET-60/Ras in SM chemoattraction. The roles of components in the Ras-MAPK cascade in this event are less clear (Sundaram et al. 1996; Chen et al. 1997; Chenand Stern1998). A crucial gap in our understanding lies in the link between activatedEGL-15 and the downstream signaling components. Here we show thatSEM-5, theC. elegansGRB2 ortholog, appears to bind directly to SH2 binding sites within the carboxy terminal tail ofEGL-15. These interactions are required for SM chemoattraction, but not for the essential function ofEGL-15.
MATERIALS AND METHODS
Genetic manipulations: All strains were derived from C. elegansvar. Bristol, strainN2, using standard genetic protocols (Brenner1974) and standard genetic manipulations (H er-man1988). Nematode strains were grown and maintained on
NGM agar plates and raised at 20°unless otherwise indicated. Transgenic assays were conducted as previously described for the genomic and cassette assays (Loet al. 2008) and for the
deletion and truncation assays (Goodmanet al. 2003), using
Bioresource Project forC. elegans(Tokyo). This allele deletes a critical portion of the PTB domain (supporting information, Figure S1), thereby destroying the sole structure on which the putative interaction with EGL-15 is based. tm1031 mutants were maintained as balanced heterozygotes using the semi-dominantsup-9(n1550sd)allele, which lies 3 map units from rog-1 on chromosome II (Levin and Horvitz 1993).
sup-9(n1550sd) causes heterozygotes to display a characteristic suite of phenotypes, including Lon (Long), Unc (Unc oordi-nated), Rbr (Rubberband), and Egl (Egg-laying defective); n1550homozygotes are inviable. Mutants were identified as Rbr animals, and recombinants were identified by non-Rbr non-Unc animals that had broods of wild-type size. PCR identification oftm1031 mutants was by duplex PCR, which can differentiate wild-type, mutant, and heterozygote animals by the presence or absence of both alleles. RNA interference (RNAi) was carried out according to standard protocols (Fire
et al. 1998).
Determination and representation of sex myoblast posi-tion:SM final positions were determined with respect to the underlying hypodermal Pn.p cells (Pn.p, the posterior daugh-ters of cells P1–P11) as previously described (Thomaset al.
1990) and are depicted using box-and-whisker plots (Moore
and McCabe1993) aligned to a schematic representation of
the Pn.p cell positions (see Figure 2, Figure 3, andFigure S4). In brief (Goodmanet al. 2003), each set of SMs is ordered
according to anteroposterior position and divided into quar-tiles. The ‘‘box’’ includes the positions of SMs within the two central quartiles. An additional vertical line within the box indicates the median SM position at the boundary between the second and third quartiles. Overlap of the line representing the median position with a right or left border of the box is depicted as a thickening of that line. A quartile length (1Q) is determined on the basis of the range of positions covered by the 2Q box length. Bars (‘‘whiskers’’) of up to 1.5Q length extend from the edges of the box to additional data points. As whisker length does not extend beyond the range of data points, these bars may be shorter than a 1.5Q length or even absent. Data points beyond the edge of the bars (‘‘outliers’’) are indicated by individual hatch marks. This representation of SMs therefore depicts the overall range of SMs as well as their general distribution and median position. To score SMs of tm1031 sterile mutant homozygotes, a large population of tm1031/n1550heterozygotes was grown to adulthood and then cloned to individual NGM agar plates. The mixed progeny of these heterozygotes were grown to mid-L3 stage at 20° and individual non-Rbr animals were numbered and mounted, and their SMs were scored using Nomarski optics. To eliminate nonmutant recombinant progeny, these individually scored animals were then recovered to PCR tubes and individually genotyped for the tm1031 deletion via duplex PCR, and recombinants were excluded from the data set.
Plasmid construction: All plasmids constructed using PCR were confirmed by sequencing. Details of plasmid construc-tion are available upon request. NH#112 is the wild-typeegl-15 genomic rescuing fragment (DeVoreet al. 1995). NH#838
egl-15(Q948STOP)was generated by introducing the nonsense muta-tion in theEGL-15CTD that corresponds to theegl-15(n1457) mutation, thereby truncating the CTD from allEGL-15 iso-forms expressed from this construct. The genomic constructs used for the experiments the results of which are summarized in Figure 2 are derivatives of NH#112. NH#934, the EGL-15(Intra I) yeast two-hybrid bait, was created by PCR amplifi-cation of the intracellular domain ofEGL-15from theEGL-15 type I cDNA (NH#324). The resulting PCR product was cloned into theHpaI sites of the yeast two-hybrid ‘‘bait’’ vec-tor pBTM116. The following constructs are all derivatives of NH#934: NH#726, egl-15(K672A); NH#1285, egl-15(Y1009F);
NH#1286,egl-15(Y884F); and NH#1287,egl-15(Y1009F, Y884F). NH#1354, the EGL-15(Intra IV) yeast two-hybrid bait, was created by two-round PCR amplification, using NH#934 for the common upstream portion ofEGL-15(Intra IV) and the EGL-15type IV cDNA for the unique downstream portion of EGL-15(Intra IV). The resulting PCR product was cloned into the PstI site of the yeast two-hybrid ‘‘bait’’ vector pBTM116. The following constructs are derivatives of NH#1354: NH#1355, egl-15(Intra IV, Y1009F); NH#1360, egl-15(Intra IV, Y1009F, Y1087F). NH#1321, the backbone vector for our cassette assays, was generated from the NH#112 genomic rescuing construct as follows. The C-terminal domain ofEGL-15, downstream of the uniqueHpaI site at amino acid L932, was replaced with the unc-54 39 UTR from pPD136.61. Various fragments were cloned into thisHpaI site to generate transgenes that corre-spond to specific portions of the alternatively spliced CTD isoforms. Mutations in these constructs were generated using standard PCR methods. Full-lengthced-2andptp-2were cloned into theXhoI site in pACT2 to generateCED-2andPTP-2yeast two-hybrid constructs.
Yeast two-hybrid screen selection procedure: The con-structs in the yeast two-hybrid vectors, pBTM116 and pACT or pACTII, were used to transform theSaccharomyces cerevisiaeL40 (MATa,trp1,leu2,his3,LYSTlexA-HIS3,URA3TlexA-LacZ) and AMR70 (MATa,trp1,leu2,his3,URA3TlexA-LacZ) yeast strains, respectively. L40 yeast was sequentially transformed with the EGL-15(Intra I) plasmid, NH#934, and with aC. eleganscDNA library using the lithium acetate method and treated as described (Roseet al. 1990). Transformations were plated to
synthetic medium lacking histidine, leucine, and tryptophan. The plates were incubated at 30°for 3 days. Colonies that grew were patched on selective plates to test for regrowth. Colonies that regrew were assayed forb-galactosidase activity by a filter assay (Daltonand Treisman1992), and colonies that turned
blue in 24 hr at 30°were retained. Library plasmid DNA was recovered via yeast miniprep and retested via a mating assay and filter-lift b-galactosidase assays. Library plasmids that passed both retests were sequenced for identification.
Yeast two-hybrid interaction assays:Dilution series for yeast matings were set up in the following manner. Individual cultures for dilution interaction assays were grown in the appropriate selective liquid medium cultures overnight at 30°. Samples were diluted to OD600¼0.2, 5ml was dotted on plates 1 cm apart undiluted and at dilutions of 1:100, 1:1000, and 1:10,000, and the plate was incubated for 3 days at 30°.
Quantitativeb-galactosidase assays:Mated yeast cells were grown overnight in TRPLEUmedia at 30°. An OD
600was taken, and the pellet of 1.0 ml of the overnight culture was resuspended in 1.0 ml of Z buffer (8.52 g anhydrous Na2HPO4, 4.8 g anhydrous NaH2PO4, 0.12 g anhydrous MgSO4, 0.74 g KCl dissolved in 1.0 liter water, and pH was adjusted to 7.0 with HCl), pelleted, and resuspended in 150ml Z buffer supple-mented with 2-mercaptoethanol (27ml of 2-mercaptoethanol/ 10 ml Z buffer); 50ml of chloroform and 20ml of 0.1% SDS were added, and the sample was vortexed for 15 sec. A total of 700 ml of prewarmed (30°) ONPG (ortho-nitrophenyl-B(beta)-galactoside; 1 mg/ml in Z buffer with 2-mercapto-ethanol; Sigma) was added, and the reactions were incubated at 30°until the solutions turned a yellow color. The reactions were checked at 1 and 5 min and then every 5 min until 15 min. Reactions were stopped by adding 0.5 ml 1 mNa2CO3 and
microcentrifuged for 10 min to pellet debris. OD420was taken for each of the reactions. Miller units were calculated using the following equation: (OD42031000)/OD6003time (in min)3 volume (in ml). All assays were repeated at least three times.
animals were picked and observed 48 hr later for the presence of accumulated eggs in the uterus. For interactors in which a genetic mutant was not available, we utilized standard RNAi techniques (Fireet al. 1998). To determine if the interactors
played a role in the essential function pathway, we screened the mutants for the ability to suppress the Clr phenotype. We used RNAi (against the interactors) in aclr-1(e1745ts)background. L4 animals were placed on RNAi feeding clones, and their progeny were shifted to 25°(the restrictive temperature) and examined for the presence of Soc animals.egl-15and L4440 RNAi were used as a positive and negative control, respectively.
RESULTS
A yeast two-hybrid screen reveals an important interaction between the C. elegans Grb2 homolog, SEM-5, and EGL-15: To identify potential signaling components that might bind directly to EGL-15, we performed a yeast two-hybrid screen using the intracel-lular domain of EGL-15. Since the identification of interactors might be dependent on tyrosine autophos-phorylation, the entire intracellular domain, including the juxtamembrane domain, the full kinase domain, and the CTD, was inserted in a two-hybrid bait vector. This EGL-15 ‘‘bait’’ construct (NH#934) contains the type I CTD [EGL-15(Intra I)], since it is the most prevalent C-terminal isoform (Goodman et al. 2003), and preliminary analysis indicated that this CTD isoform could mediate SM chemoattraction (Branda 2001). This bait is tyrosine-phosphorylated in yeast (Figure S2). We screened 2.03108transformants and identified 104 clones that specifically interact withEGL-15(Intra I). These clones define 10 putative interactors (Table 1). Six of the putative interactors have an SH2 domain, including theC. eleganshomologs of Grb2 (SEM-5), Vav (VAV-1), Abl (ABL-1), and Csk (CSK-1) and the p85 sub-unit of PI3K (AAP-1). The sixth SH2 domain-containing interactor is a protein of unknown function (F13B12.6);
clones of this gene represented the majority of isolates from this screen (76/104).
A number of phenotypic tests were conducted to determine whether any of the genes corresponding to the identified interactors shared functions withEGL-15. These tests (see materials and methods) measured effects on either fluid homeostasis or SM chemoattrac-tion when the interactor genes were compromised using RNA interference or existing mutant alleles. The only assays that revealed egl-15-like phenotypes were those conducted with sem-5 (data not shown); these assays confirmed the common functions shared between SEM-5andEGL-15(DeVoreet al. 1995).
An alternative way to authenticate the yeast two-hybrid interactors was to show the specificity of the observed interaction. Because SH2 domains have a well-defined interaction mechanism, it was possible to test the SH2 domain-containing interactors further for the mecha-nism by which they bound EGL-15. Phosphorylated tyrosines, like those present in theEGL-15intracellular domain, often serve as high-affinity binding sites for SH2 domain-containing proteins (Pawson2004). We tested whetherEGL-15-interactor binding activity was depen-dent on the kinase activity ofEGL-15(Intra I). Mutating the ATP-binding site (K672A) of the kinase domain within the EGL-15(Intra I) bait construct abolishes EGL-15autophosphorylation in yeast (Figure S2). This mutation is also known to abolish the phenotypic effects of otherwise hyperactivated egl-15 constructs (Kokel et al. 1998). When tested with thisEGL-15(K672A Intra I) bait (NH#726), all 10 interactors failed to show yeast two-hybrid binding activity (Figure 1B and data not shown), indicating that all 10 require an active kinase domain to interact withEGL-15(Intra I).
To test the specificity of these interactions further, we investigated whether the six SH2-containing proteins
TABLE 1 EGL-15 interactors
Genea Function/structure Strength of interaction (MU)b Mutant phenotype
SH2-containing interactors
F13B12.6 (76) SH2 domain protein 330061118 Wild type
sem-5(2) Grb2 homolog 210061050 Vul, Let, Mig
vav-1(3) Vav proto-oncogene 19006398 Let
abl-1(13) Abl homolog 12006724 Wild type
aap-1(1) AGE-1 adapter protein 38611 Wild type
csk-1(3) C-terminal Src kinase 46652 Let, Ste
Non-SH2-containing interactors
tag-153(1) NOT2/NOT3/NOT5 domain protein 13006360 Wild type
tdc-1(1) Tyrosine decarboxylase 10006547 Wild type
T05E11.3 (3)c Endoplasmic reticulum protein 33613 Wild type
Y48G8AL.9 (1)c C
2H2-type zinc-finger protein 32612 Wild type
a
The number of screen isolates of each gene is shown in parentheses. b
Strength of interaction is measured by a quantitativeb-gal assay and reported in Miller units (MU) as the mean6SD.
c
identified in our screen [of 61 predicted from the C. elegansgenome sequence (http://www.wormbase.org)] indicated binding specificity in these assays. Two addi-tional SH2-containing constructs, which encode CED-2/CrkII andPTP-2/Shp2, were built and tested for their ability to interact withEGL-15(Intra I). BothCED-2and PTP-2fail to interact withEGL-15(Intra I), despite being expressed and able to interact with another EGL-15 isoform (Figure S3; see below). Taken together, these data indicate that the mode of these interactions is specific and suggest that they may be mechanistically relevant.
The migration-specific phenotype of egl-15(n1457) suggests that the CTD might be an important site for protein–protein interactions required for SM chemo-attraction. Thus, we tested whether deletion of the C-terminal domain of theEGL-15(Intra I) bait [EGL-15(DCTD); NH#1278] would abolish the binding of any
of the 10 interactors. Deletion of the C-terminal domain had a variable effect on the strength of the interactions between EGL-15and all of the interactors (Table S1). However, 2 of the 10 interactors, SEM-5 and CSK-1, showed the most dramatic reductions, consistent with their interactions requiring the C-terminal domain of EGL-15(Intra I). Of these 2 interactors, onlySEM-5has been observed to exhibit egl-15-like phenotypes, with established roles in both the essential and the SM migration functions ofEGL-15(Clarket al. 1992; Stern et al. 1993; Chenet al. 1997).
Within the C-terminal domain of EGL-15, there are two consensus binding sites (YXNX) for the SH2 do-main of SEM-5(see Figure 1A; Songyanget al. 1993). Y1009(YCND) lies in the amino-terminal portion of the CTD within a fragment that is common to all EGL-15 isoforms, including the type I isoform that was used for the yeast two-hybrid screen. Y1087(YYNT) lies within a fragment that is unique to the type IV isoform. An additional consensus SEM-5 binding site sequence is found at Y884(YANL) within the kinase domain of EGL-15. We tested the relevance of each of these potential binding sites by mutating these tyrosines to phenyl-alanines, both individually and in combination, and by testing the constructs in yeast two-hybrid assays (Figure 1B,Table S1). Mutating Y1009 to phenylalanine in the type I construct abolishes the EGL-15TSEM-5
interac-tion. By contrast, the Y884F mutation does not, which is consistent with the CTD being necessary for this in-teraction. The type IV isoform contains both Y1009 and Y1087 (Figure 1A). Mutating both of these sites is necessary to abolish the yeast two-hybrid interaction between EGL-15 andSEM-5. These data indicate that SEM-5can interact withEGL-15via either Y1009or Y1087. The two SEM-5 binding sites in the EGL-15 CTD redundantly mediate SM chemoattraction:Two types of transgenic rescue assays were used to assess the mech-anism by which the CTD mediates SM chemoattraction. In the first of these, a genomic rescuing construct was used to identify CTD sites that are necessary for SM chemoattraction. The wild-type EGL-15 genomic con-struct (NH#112) rescues both the larval arrest and the SM migration defects of egl-15(null) animals when in-troduced in transgenic assays (Figure 2A). By contrast, a construct that contains the mutation found in egl-15(n1457)that truncates the CTD rescues only the larval arrest defect and fails to rescue the SM migration defect, resulting in severely posteriorly displaced SMs (Figure 2B). The results of a series of CTD deletion and trun-cation constructs tested using this assay indicated the presence of a site in the alternatively spliced region that can efficiently mediate SM chemoattraction (Figure S4). These results are consistent with a possible role of Y1087 in SM chemoattraction.
The second transgenic rescue assay tested whether portions of the CTD are sufficient for mediating SM chemoattraction. This assay has the benefit of eliminat-Figure1.—The SEM-5TEGL-15 interaction is dependent
ing the complexities due to redundancy and alternative splicing. We created a CTD-sufficiency ‘‘cassette’’ in which we deleted all egl-15 sequences downstream of the kinase domain within the genomic rescuing frag-ment and appended the commonly used 39 UTR se-quences from theunc-54gene (Figure 3). This deleted theEGL-15CTD and all C-terminal alternative splicing from theegl-15genomic rescuing fragment. Fragments could be tested for sufficiency by introducing them into a uniqueHpaI site located just after the kinase domain. The ‘‘empty cassette’’ (without any fragments inserted), when expressed in anegl-15(null)strain, can rescue the essential function, but not the SM chemoattraction function ofEGL-15(Figure 3A). Thus, short sequences can be inserted into this ‘‘cassette’’ to serve as an artificial CTD to test whether they are sufficient to restore proper SM chemoattraction. To distinguish this assay from the normal genomic rescuing assay, we refer to this as the ‘‘cassette assay.’’ All constructs tested in this assay rescue the essential function ofEGL-15.
Individual DNA fragments spanning the Y1009SEM-5 binding site (Figure 3B) and the Y1087 SEM-5 bind-ing site (Figure 3D) are each capable of efficient SM chemoattraction rescue. SMs in these rescued lines are centered, unlike the empty vector control where the SMs are posteriorly displaced. To test the relevance of the SEM-5binding sites within these fragments, we intro-duced tyrosine-to-phenylalanine mutations into the re-spective fragments. The mutated constructs fail to rescue SM migration (Figure 3, C and E), resulting in SMs that are posteriorly displaced to a similar extent to those observed in the empty cassette control lines. Thus, Y1009and Y1087are specifically required for these fragments to rescue, suggesting thatSEM-5binding at these sites can provide the links toEGL-15that mediate SM chemoattraction.
Removal of both C-terminal domain SEM-5 binding sites abolishes SM chemoattraction:To test the roles of the Y1009CND and Y1087YNT binding sites in SM chemo-attraction, we introduced tyrosine-to-phenylalanine mu-tations into the genomic rescuing construct, singly and in combination, and determined the effects on SM migration guidance. Mutation of Y1009and Y1087together abolishes SM chemoattraction rescue (Figure 2H). By contrast, individual mutations do not show the same dramatic effect on SM positions (Figure 2, D and E). Consistent with the inability of Y884to bindSEM-5in the yeast two-hybrid assay, the presence of Y884 does not permit SM chemoattaction (Figure 2H), and mutation of Y884to phenylalanine does not significantly affect SM positions, either alone or in combination with other Figure2.—Genomic assay results of the EGL-15 CTD role
in SM chemoattraction. EGL-15 Y1009 and Y1087 are func-tionally redundant. Sex myoblast distributions for three lines for each construct are shown. The indicated transgenes were all tested in anegl-15(null)background. (A) Wild-type EGL-15 (NH#112). (B) A truncated version of EGL-15 that correlates to theegl-15(n1457)mutant (NH#838). (C) The single Y884F mutation fails to disrupt sex myoblast migration (NH#1379). (D and E) The single Y1009F (NH#818) or Y1087F (NH#1382) mutations do not account for the posterior SM positions of DCTD. (F and G) Double mutations with Y884 (Y884F/ Y1009F, NH#841; Y884F/Y1087F, NH#1380) are not more affected than the respective single mutations. (H and I)
tyrosine-to-phenylalanine mutations (Figure 2, C, F, G, and I). Thus, our yeast two-hybrid data show thatSEM-5 can bind directly toEGL-15at both Y1009and Y1087. Both Y1009and Y1087are consensus Y-X-N-XSEM-5binding sites. Analyses of these sites in the cassette assay demonstrate that both are sufficient for SM chemoattraction, while the genomic assays demonstrate that the sites act redundantly to mediate SM chemoattraction.
An additional factor is implicated in the EGL-15-SEM-5 fluid homeostasis pathway: SEM-5is a required EGL-15 signaling component not only for SM chemo-attraction, but also for fluid homeostasis. While muta-tions insem-5result in a Soc phenotype, theegl-15(n1457) truncation of the EGL-15CTD, which is predicted to eliminate both SEM-5binding sites, does not confer a Soc phenotype (Goodmanet al. 2003). The phenotype of this mutant thus indicates that an additional site must be able to recruit SEM-5to EGL-15, either directly or indirectly.
Y884is another predicted direct binding site for the SEM-5SH2 domain, even though it cannot bindSEM-5 in the yeast two-hybrid system. To test formally whether it might function to recruitSEM-5to mediate fluid home-ostasis, theEGL-15(DSEM-5) transgene, which mutates all three potential SEM-5 SH2 binding sites, was ex-pressed in a clr-1(e1745ts); egl-15(null) background. clr-1(e1745ts); egl-15(null); ayEx[EGL-15(DSEM-5)] animals are Clr at the nonpermissive temperature forclr-1(e1745ts), formally demonstrating thatEGL-15(DSEM-5) efficiently mediates fluid homeostasis. Thus, EGL-15 transduces its signal toSEM-5independently of all of its canonical SEM-5 SH2 binding sites, implicating another mecha-nism that linksEGL-15toSEM-5to mediate the essential/ fluid homeostasis function.
One possible alternate mechanism could utilize an additional adaptor protein. The yeast two-hybrid EGL-15interactors might provide this additional link to SEM-5. These were tested using either RNAi or existing viable mutations in a background that truncates the SEM-5 binding sites in theEGL-15CTD. None of the nine other putative interactors confers a Soc phenotype when tested in a clr-1(e1745ts); egl-15(n1457) background, suggesting that none provides the link toSEM-5(data not shown).
Another candidate that could serve as the adaptor that linksEGL-15toSEM-5isrog-1, theC. elegansFRS2 homolog. FRS2 is a key component of the mammalian FGFR signaling pathway that acts as the main adaptor protein linking FGF receptors to downstream signaling molecules (Hadariet al. 2001). FRS2 is constitutively complexed with FGF receptors. Upon FGF receptor activation, tyrosines present within FRS2 are phosphor-ylated, enabling the recruitment of additional signaling components. Phosphorylation of FRS2 on Y196, Y306, Y349, and Y392allows Grb2 to bind (Kouharaet al. 1997), thus linking activated FGF receptors to Grb2. InC. elegans,
rog-1 encodes a 599-amino-acid protein containing an Figure3.—The cassette assay (‘‘sufficiency’’) results. (Top)
N-terminal PTB domain that is 34% identical and 60% similar to the PTB domain of murine FRS2a; the carboxy-terminal 350 amino acids are less similar (Figure S1) (Matsubaraet al. 2007).
We tested whetherrog-1inC. elegansplays a similar role to FRS2 in mammals and serves as an adaptor protein linkingSEM-5toEGL-15. Mutation ofrog-1alone does not confer anyegl-15-like phenotypes (Table S2) (Matsubara et al. 2007; Bennett 2009). Both rog-1(RNAi) and the putative null allele,rog-1(tm1031), are sterile (Table S2) and have been shown to have defects in a Ras/MAP kinase pathway that mediates germline progression (Matsubara et al. 2007; Bennett2009). These tests did not result in either a Soc or SM migration phenotype (Table S2). However, for the fluid homeostasis function, this might be because of redundancy with the direct binding of SEM-5via the EGL-15CTD. To circumvent this poten-tial redundancy, we tested whether compromisingrog-1in aclr-1(e1745ts);egl-15(n1457)background could suppress the Clr phenotype (Soc). clr-1(e1745ts); rog-1(RNAi);
egl-15(n1457)animals are not Soc, indicating thatROG-1 does not play a role in a redundant mechanism. These animals were sterile, similar to therog-1(tm1031) single mutant, confirming the effectiveness of therog-1(RNAi). These data support the model that ROG-1 is not the adaptor that linksSEM-5toEGL-15for the fluid homeo-stasis pathway.
DISCUSSION
The C. elegans Grb2 ortholog,SEM-5, was originally identified in screens for EGF receptor signaling pro-cesses and the fluid homeostasis function mediated by the FGF receptor, EGL-15(Clark et al. 1992). SEM-5 binding sites on theC. elegansLET-23EGF receptor itself mediate some of the functions of this receptor tyrosine kinase, similar to the mechanism by which the EGF receptor and GRB2 function in mammalian systems (Lesaand Sternberg1997). By contrast, Grb2 is linked to the activated FGF receptor signaling complex via the FRS2 adaptor protein (Hadari et al. 2001). Here we show thatSEM-5can bind directly toEGL-15to mediate SM chemoattraction and that SEM-5 associates with EGL-15via a different mechanism for the signal trans-duction mechanism that regulates fluid homeostasis.
The direct association betweenSEM-5andEGL-15was initially indicated by the results of a yeast two-hybrid screen designed to identifyC. elegansproteins with the potential to interact directly with the intracellular domain ofEGL-15. This screen identified 10 potential direct interactors, includingSEM-5and 5 other inter-actors with SH2 domains. Although a number of these potential interactors have known functions that offer tantalizing possible connections to migration guidance or other signaling functions of EGL-15, none, in ad-dition to SEM-5, could be corroborated with genetic
evidence linking them either to SM chemoattraction or fluid homeostasis. For example, Abl has a well-studied role in actin remodeling (Koleskeet al. 1998; Plattner et al. 1999; Grevengoedet al. 2001; Sini et al. 2004), while Csk plays a role in integrin-mediated cell adhesion and migration in human colon cancer cells (Rengifo -Camet al. 2004). Vav, like other Dbl homology-containing Rho-GEFs, has a cytoskeletal role and also has known roles in F-actin polymerization (Fischeret al. 1998) and Rac-mediated lamellipodia formation (Crespo et al. 1996, 1997). TheAGE-1adaptor protein,AAP-1, serves to linkC. elegansPI3-kinase (Paradiset al. 1999) to the DAF-2 insulin receptor and potentially could recruit AGE-1 to EGL-15. PI3 kinase has been shown to act downstream of vertebrate FGFRs; however, no role for PI3 kinase has been established in EGL-15 functions inC. elegans.As tantalizing as these possibilities seem, a variety of genetic tests failed to corroborate the associ-ation indicated by the yeast two-hybrid results. It is possible that some of these potential interactors are involved in some of the other functions ofEGL-15.
By contrast, not only doesSEM-5have well-established roles in the functions mediated byEGL-15, but a large amount of evidence points to the specificity of the yeast two-hybrid interaction. First, not all SH2-containing proteins interact with the yeast two-hybrid ‘‘bait’’ EGL-15(Intra I), which was used in this screen. C. elegans CED-2and PTP-2each have SH2 domains that fail to interact with EGL-15(Intra I), although they interact strongly withEGL-15(Intra-IV) (Figure S3). This dem-onstrates some degree of specificity of the screen itself. Second, the interaction between SEM-5 and EGL-15 requires theEGL-15CTD, the domain required for SM chemoattraction. Third, the binding toEGL-15(Intra I) is specific to Y1009, and the binding toEGL-15(Intra-IV) specifically requires a pair of tyrosine residues, Y1009and Y1087; the conservative mutation of these tyrosines to phenylalanines completely abolishes the interaction. Fourth, the interaction is dependent onEGL-15kinase activity and thus likely requires the tyrosine phosphor-ylation known to be essential for SH2 binding. These characteristics of the yeast two-hybrid interaction in-dicate its relevance to the physiological functions of EGL-15andSEM-5.
These two sites function redundantly for SM chemo-attraction; only when both the Y1009and the Y1087binding sites are mutated is SM chemoattraction abolished. The presence of both sites within the type IV isoform suggests that this isoform may play a particularly important role in SM chemoattraction. The conservation of this site within Caenorhabditis briggsae EGL-15 provides further support for its importance. Thus, despite the prevalence of type I transcripts within mixed cDNA isolates, it is possible that the type IV isoform might be more highly expressed in the migrating SMs. The average relative abundance of the isoforms might be accounted for by a skew toward type I expression within the hypodermis, which is a large syncytium that contains many nuclei.
The intracellular domain of EGL-15 contains an additional YXNX motif at Y884that could act as another binding site for the SH2 domain ofSEM-5. However, Y884 does not appear to play a role in recruitingSEM-5or in SM chemoattraction. First, SEM-5 does not associate with EGL-15 via this site by yeast two-hybrid analysis. Second, mutation of the two CTDSEM-5binding sites abolishes SM chemoattraction, indicating that Y884is not sufficient to carry out this recruiting function. The inability of Y884 to recruit SEM-5 is likely due to the structural role that it plays within the highly conserved G a-helix of the EGL-15 kinase domain. The structural importance of this residue is underscored by the fact that the corresponding residue in all human FGFRs is phenylalanine instead of tyrosine. By contrast, Y1009and Y1087are C-terminal to the kinase domain and are likely to be in flexible segments that can be readily phosphor-ylated to triggerSEM-5recruitment.
Other EGL-15 functions require a different mecha-nism to signal to SEM-5: Similar to the egl-15(n1457) mutation that truncates theEGL-15CTD, disruption of the two CTD SEM-5 binding sites, Y1009 and Y1087, specifically affects SM chemoattraction, leaving intact other common functions shared bySEM-5andEGL-15. One of these shared roles is the SM chemorepulsion
mechanism. AlthoughSEM-5is required for the gonad-dependent repulsion, removing the CTD binding sites does not compromise this function. Therefore, the mechanism by which SEM-5 and EGL-15mediate SM chemorepulsion has yet to be determined. Similarly, SEM-5andEGL-15also have a shared role in the fluid homeostasis/essential function. Mutations inSEM-5can suppress the Clr phenotype; however, theegl-15(DSEM-5) mutation cannot, suggesting that SEM-5must link to EGL-15 by a different mechanism for this function. While it is possible that SEM-5 could interact directly with EGL-15 via its SH3 domains to mediate these functions, the lack of two-hybrid interactions in the absence of the SH2 binding sites in the CTD suggests that this is not the case. Rather, these functions may require an additional adaptor protein to transduce signals betweenEGL-15andSEM-5. We have identified a number of potential candidates that could play this role, including therog-1-encoded FRS2-like adaptor and the nine remaining yeast two-hybrid interactors. Genetic tests failed to reveal the involvement of these genes in EGL-15-mediated processes. The multi-substrate adap-tor protein SOC-1 is known to function in EGL-15-mediated fluid homeostasis (Schutzmanet al. 2001). It is possible that it serves to link SEM-5 to EGL-15 independently of the CTD direct binding sites, although the mechanism by which SOC-1itself links toEGL-15 remains to be elucidated.
Conclusions: Many signal transduction components have been conserved throughout the evolution of metazoa. Nevertheless, there is considerable flexibility in how they are used (Figure 4). Data inC. elegans in-dicate that the FRS2-like protein,ROG-1, is not involved in FGF signal transduction, but rather is involved in a different RAS/MAP kinase pathway that is critical for germline progression.D. melanogasteralso has an FRS2-like protein, but, FRS2-likeC. elegans, no evidence links it to FGF signaling. Instead, the Dof protein is critical to FGF signal transduction in Drosophila (Michelsonet al. Figure 4.—Comparison of FGFR
1998; Vincentet al. 1998; Imamet al. 1999).C. elegans does not have a Dof-like gene in its genome and thus must utilize yet other mechanisms. Here we show thatC. elegansuses direct binding to linkSEM-5toEGL-15for SM chemoattraction, obviating the need for the FRS2-like multi-substrate adaptor protein for this function. For other EGL-15 functions, an additional, yet to be discovered, mechanism linksSEM-5to theEGL-15 sig-naling complex. Identifying that mechanism will further show the breadth of evolutionary flexibility in these signal transduction systems.
We thank J. Schutzman for initiating the EGL-15 yeast two-hybrid experiments, I. Sasson for initiating the ROG-1 experiments, Y. Kohara for cDNAs, S. Hubbard for discussions about the structure of the EGL-15 kinase domain, and members of our lab for critical reading of this manuscript. Some nematode strains used in this work were provided by the Caenorhabditis Genetics Center, which is funded by the National Institutes of Health (NIH) National Center for Research Resources.tm1031was provided by the National Bioresource Project forC. elegans(Tokyo). This work was supported by NIH grant GM50504 (M.J.S.) and by the following NIH training grants: T32 GM07223 (T-W.L. and D.C.B) and T32 DK07259 (S.J.G.).
LITERATURE CITED
Bennett, D. C., 2009 Identifying interactors of C. elegans FGFR
sig-naling. Ph.D. Thesis, Yale University, New Haven, CT.
Borland, C. Z., J. L. Schutzmanand M. J. Stern, 2001 Fibroblast
growth factor signaling in Caenorhabditis elegans. Bioessays23:
1120–1130.
Branda, C. S., 2001 Mechanisms of sex myoblast migration inC.
ele-ganshermaphrodites. Ph.D. Thesis, Yale University, New Haven, CT. Branda, C. S., and M. J. Stern, 2000 Mechanisms controlling sex
myoblast migration in Caenorhabditis elegans hermaphrodites. Dev. Biol.226:137–151.
Brenner, S., 1974 The genetics ofCaenorhabditis elegans. Genetics 77:71–94.
Bulow, H. E., T. Boulinand O. Hobert, 2004 Differential
func-tions of the C. elegans FGF receptor in axon outgrowth and maintenance of axon position. Neuron42:367–374.
Burdine, R. D., E. B. Chen, S. F. Kwokand M. J. Stern, 1997 egl-17
encodes an invertebrate fibroblast growth factor family member required specifically for sex myoblast migration in Caenorhabdi-tis elegans. Proc. Natl. Acad. Sci. USA94:2433–2437.
Burdine, R. D., C. S. Brandaand M. J. Stern, 1998 EGL-17(FGF)
expression coordinates the attraction of the migrating sex myo-blasts with vulval induction in C. elegans. Development 125:
1083–1093.
Chen, E. B., and M. J. Stern, 1998 Understanding cell migration
guidance: lessons from sex myoblast migration in C. elegans. Trends Genet.14:322–327.
Chen, E. B., C. S. Brandaand M. J. Stern, 1997 Genetic enhancers
of sem-5 define components of the gonad-independent guidance mechanism controlling sex myoblast migration in Caenorhabdi-tis elegans hermaphrodites. Dev. Biol.182:88–100.
Clark, S. G., M. J. Sternand H. R. Horvitz, 1992 C. elegans
cell-signalling gene sem-5 encodes a protein with SH2 and SH3 do-mains. Nature356:340–344.
Crespo, P., X. R. Bustelo, D. S. Aaronson, O. A. Coso, M. Lopez
-Barahona et al., 1996 Rac-1 dependent stimulation of the
JNK/SAPK signaling pathway by Vav. Oncogene13:455–460. Crespo, P., K. E. Schuebel, A. A. Ostrom, J. S. Gutkindand X. R.
Bustelo, 1997 Phosphotyrosine-dependent activation of Rac-1
GDP/GTP exchange by the vav proto-oncogene product. Nature
385:169–172.
Dalton, S., and R. Treisman, 1992 Characterization of SAP-1, a
protein recruited by serum response factor to the c-fos serum re-sponse element. Cell68:597–612.
DeVore, D. L., H. R. Horvitzand M. J. Stern, 1995 An FGF
recep-tor signaling pathway is required for the normal cell migrations
of the sex myoblasts in C. elegans hermaphrodites. Cell83:611– 620.
Dixon, S. J., M. Alexander, R. Fernandes, N. Rickerand P. J. Roy,
2006 FGF negatively regulates muscle membrane extension in Caenorhabditis elegans. Development133:1263–1275. Eswarakumar, V. P., I. Laxand J. Schlessinger, 2005 Cellular
sig-naling by fibroblast growth factor receptors. Cytokine Growth Factor Rev.16:139–149.
Fire, A., S. Xu, M. K. Montgomery, S. A. Kostas, S. E. Driver
et al., 1998 Potent and specific genetic interference by double-stranded RNA in Caenorhabditis elegans. Nature391:
806–811.
Fischer, K. D., Y. Y. Kong, H. Nishina, K. Tedford, L. E. Marengere
et al., 1998 Vav is a regulator of cytoskeletal reorganization me-diated by the T-cell receptor. Curr. Biol.8:554–562.
Fleming, T. C., F. W. Wolfand G. Garriga, 2005 Sensitized
ge-netic backgrounds reveal a role for C. elegans FGF EGL-17 as a repellent for migrating CAN neurons. Development 132:
4857–4867.
Goodman, S. J., C. S. Branda, M. K. Robinson, R. D. Burdineand
M. J. Stern, 2003 Alternative splicing affecting a novel domain
in the C. elegans EGL-15 FGF receptor confers functional spec-ificity. Development130:3757–3766.
Gotoh, N., K. Manova, S. Tanaka, M. Murohashi, Y. Hadariet al.,
2005 The docking protein FRS2alpha is an essential compo-nent of multiple fibroblast growth factor responses during early mouse development. Mol. Cell. Biol.25:4105–4116.
Grevengoed, E. E., J. J. Loureiro, T. L. Jesse and M. Peifer,
2001 Abelson kinase regulates epithelial morphogenesis in Drosophila. J. Cell Biol.155:1185–1198.
Hadari, Y. R., H. Kouhara, I. Lax and J. Schlessinger,
1998 Binding of Shp2 tyrosine phosphatase to FRS2 is essential for fibroblast growth factor-induced PC12 cell differentiation. Mol. Cell. Biol.18:3966–3973.
Hadari, Y. R., N. Gotoh, H. Kouhara, I. Laxand J. Schlessinger,
2001 Critical role for the docking-protein FRS2 alpha in FGF receptor-mediated signal transduction pathways. Proc. Natl. Acad. Sci. USA98:8578–8583.
Herman, R. K., 1988 Genetics, pp. 17–45 inThe Nematode
Caenorhab-ditis elegans, edited by W. B.Wood. Cold Spring Harbor
Labora-tory Press, Cold Spring Harbor, NY.
Huang, P., and M. J. Stern, 2004 FGF signaling functions in the
hy-podermis to regulate fluid balance in C. elegans. Development
131:2595–2604.
Huang, P., and M. J. Stern, 2005 FGF signaling in flies and worms:
more and more relevant to vertebrate biology. Cytokine Growth Factor Rev.16:151–158.
Imam, F., D. Sutherland, W. Huang and M. A. Kransnow,
1999 stumps, a Drosophila gene required for fibroblast growth factor (FGF)-directed migrations of tracheal and mesodermal cells. Genetics152:307–318.
Kokel, M., C. Z. Borland, L. DeLong, H. R. Horvitz and M. J.
Stern, 1998 clr-1 encodes a receptor tyrosine phosphatase that
negatively regulates an FGF receptor signaling pathway in Caeno-rhabditis elegans. Genes Dev.12:1425–1437.
Koleske, A. J., A. M. Gifford, M. L. Scott, M. Nee, R. T. Bronson
et al., 1998 Essential roles for the Abl and Arg tyrosine kinases in neurulation. Neuron21:1259–1272.
Kouhara, H., Y. R. Hadari, T. Spivak-Kroizman, J. Schilling, D.
Bar-Sagiet al., 1997 A lipid-anchored Grb2-binding protein
that links FGF-receptor activation to the Ras/MAPK signaling pathway. Cell89:693–702.
Lax, I., A. Wong, B. Lamothe, A. Lee, A. Frostet al., 2002 The
docking protein FRS2alpha controls a MAP kinase-mediated neg-ative feedback mechanism for signaling by FGF receptors. Mol. Cell10:709–719.
Lesa, G. M., and P. W. Sternberg, 1997 Positive and negative
tissue-specific signaling by a nematode epidermal growth factor recep-tor. Mol. Biol. Cell8:779–793.
Levin, J. Z., and H. R. Horvitz, 1993 Three new classes of mutations
in theCaenorhabditis elegansmuscle genesup-9.Genetics135:53–70. Lo, T. W., C. S. Branda, P. Huang, I. E. Sasson, S. J. Goodmanet al.,
Matsubara, Y., I. Kawasaki, S. Urushiyama, T. Yasuda, M. Shirakata
et al., 2007 The adaptor-like protein ROG-1 is required for ac-tivation of the Ras-MAP kinase pathway and meiotic cell cycle progression in Caenorhabditis elegans. Genes Cells12:407– 420.
Michelson, A. M., S. Gisselbrecht, E. Buff and J. B. Skeath,
1998 Heartbroken is a specific downstream mediator of FGF re-ceptor signalling in Drosophila. Development125:4379–4389. Moore, D. S., and G. P. Mccabe, 1993 Introduction to the Practice of
Statistics. Freeman and Company, New York.
Ong, S. H., G. R. Guy, Y. R. Hadari, S. Laks, N. Gotoh et al.,
2000 FRS2 proteins recruit intracellular signaling pathways by binding to diverse targets on fibroblast growth factor and nerve growth factor receptors. Mol. Cell. Biol.20:979–989.
Ornitz, D. M., and N. Itoh, 2001 Fibroblast growth factors.
Ge-nome Biol.2:REVIEWS3005.
Paradis, S., M. Ailion, A. Toker, J. H. Thomasand G. Ruvkun,
1999 A PDK1 homolog is necessary and sufficient to transduce AGE-1 PI3 kinase signals that regulate diapause in Caenorhabdi-tis elegans. Genes Dev.13:1438–1452.
Pawson, T., 2004 Specificity in signal transduction: from
phospho-tyrosine-SH2 domain interactions to complex cellular systems. Cell116:191–203.
Plattner, R., L. Kadlec, K. A. DeMali, A. Kazlauskasand A. M.
Pendergast, 1999 c-Abl is activated by growth factors and
Src family kinases and has a role in the cellular response to PDGF. Genes Dev.13:2400–2411.
Polanska, U. M., D. G. Fernigand T. Kinnunen, 2009 Extracellular
interactome of the FGF receptor-ligand system: complexities and the relative simplicity of the worm. Dev. Dyn.238:277–293. Rengifo-Cam, W., A. Konishi, N. Morishita, H. Matsuoka, T. Yamori
et al., 2004 Csk defines the ability of integrin-mediated cell adhe-sion and migration in human colon cancer cells: implication for a potential role in cancer metastasis. Oncogene23:289–297. Rose, D. M., F. Winstonand P. Hieter, 1990 Laboratory Course
Man-ual for Methods in Yeast Genetics. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY.
Roubin, R., K. Naert, C. Popovici, G. Vatcher, F. Coulieret al.,
1999 let-756, a C. elegans fgf essential for worm development. Oncogene18:6741–6747.
Schutzman, J. L., C. Z. Borland, J. C. Newman, M. K. Robinson, M.
Kokelet al., 2001 TheCaenorhabditis elegansEGL-15 signaling
pathway implicates a DOS-like multisubstrate adaptor protein in fibroblast growth factor signal transduction. Mol. Cell. Biol.
21:8104–8116.
Sini, P., A. Cannas, A. J. Koleske, P. P. Di Fiore and G. Scita,
2004 Abl-dependent tyrosine phosphorylation of Sos-1 mediates growth-factor-induced Rac activation. Nat. Cell Biol.6:268–274. Songyang, Z., S. E. Shoelson, M. Chaudhuri, G. Gish, T. Pawson
et al., 1993 SH2 domains recognize specific phosphopeptide se-quences. Cell72:767–778.
Stern, M. J., and H. R. Horvitz, 1991 A normally attractive cell
in-teraction is repulsive in two C. elegans mesodermal cell migra-tion mutants. Development113:797–803.
Stern, M. J., L. E. Marengere, R. J. Daly, E. J. Lowenstein, M. Kokel
et al., 1993 The human GRB2 and Drosophila Drk genes can functionally replace the Caenorhabditis elegans cell signaling gene sem-5. Mol. Biol. Cell4:1175–1188.
Sulston, J. E., and H. R. Horvitz, 1977 Post-embryonic cell lineages
of the nematode, Caenorhabditis elegans. Dev. Biol.56:110–156. Sundaram, M., J. Yochemand M. Han, 1996 A Ras-mediated signal
transduction pathway is involved in the control of sex myoblast mi-gration in Caenorhabditis elegans. Development122:2823–2833. Szewczyk, N. J., and L. A. Jacobson, 2003 Activated EGL-15 FGF
receptor promotes protein degradation in muscles of Caeno-rhabditis elegans. EMBO J.22:5058–5067.
Thomas, J. H., M. J. Sternand H. R. Horvitz, 1990 Cell
interac-tions coordinate the development of the C. elegans egg-laying system. Cell62:1041–1052.
Vincent, S., R. Wilson, C. Coelho, M. Affolterand M. Leptin,
1998 The Drosophila protein dof is specifically required for FGF signaling. Mol. Cell2:515–525.
Xu, H., K. W. Leeand M. Goldfarb, 1998 Novel recognition motif
on fibroblast growth factor receptor mediates direct association and activation of SNT adapter proteins. J. Biol. Chem. 273:
17987–17990.
Supporting Information
http://www.genetics.org/cgi/content/full/genetics.109.113373/DC1
Caenorhabditis elegans Fibroblast Growth Factor Receptor
Signaling Can Occur Independently of the Multi-Substrate
Adaptor FRS2
Te-Wen Lo, Daniel C. Bennett, S. Jay Goodman and Michael J. Stern
Copyright © 2010 by the Genetics Society of America
T-W. Lo et al.
2 SI
FIGURE S1The structure of the rog-1 gene. (A) ROG-1 is encoded by two similar cDNAs of either 1800 or 1809 bp,
T-W. Lo et al. 3 SI
FIGURE S2EGL-15 DNA binding domain baits are expressed in yeast. -LexA western blot (left) shows that both the
EGL-15(IntraI) bait used in the two-hybrid screen (WT, NH#934) and the kinase-compromised version (K672A, NH#726) are expressed in yeast (open arrowhead). The LexA (empty vector, pBTM116) control is also expressed (solid arrowhead).
T-W. Lo et al.
4 SI
T-W. Lo et al. 6 SI
FIGURE S4Genomic assay of CTD deletion/truncation constructs. The indicated transgenes were all tested in an egl-15(null)
background. The portion of the CTD denoted in gray is common to all CTD isoforms and spans E994-Q1026. (A) Controls. Wild-type EGL-15 restores SM chemoattraction (NH#112), while the negative control, that correlates to the egl-15(n1457)
mutation (NH#838), does not. (B) Deletion constructs that retain the unique regions for types 2-4 all restore SM chemoattraction (NH#1166, M934-Y993; NH#1186, M934- C1010; NH#1187, M934- I1020). All deletions start after the kinase domain at M934. The region that was deleted is indicated by a short bent line. The first amino acid after the deletion is indicated with an
arrowhead. These results indicate the presence of a site in the alternatively-spliced region that can efficiently mediate SM chemoattraction. (C) Conversely, truncation mutations that exclude the alternatively-spliced region compromise rescue activity. The amino acid that was mutated to a STOP is indicated with an arrowhead (NH#1072, Q1026STOP; NH#1184, N1011STOP; NH#1071, E994STOP). The E994STOP mutation truncates EGL-15 upstream of Y1009 and any of the sites for alternative splicing. Although the N1011STOP mutation leaves Y1009 intact, it removes the N1011 residue, which is critical for the binding of the SH2 domain of SEM-5 (SONGYANG et al. 1993). It is unclear why the Q1026STOP truncation rescues less well than the
common fragment alone in the cassette assay (Fig. 3B); this fragment leaves intact the entire common fragment, including the Y1009 SEM-5 binding site. It is possible that the premature translational stop due to the Q1026STOP mutation within the
T-W. Lo et al. 7 SI
FIGURE S5Amino acid sequences of EGL-15 CTD isoforms. The structures of each of the four EGL-15 CTD isoforms,
T-W. Lo et al.
8 SI
TABLE S1
SEM-5 and CSK-1 interactions require the EGL-15 CTD
a Strength of interaction is measured by a quantitative ß-gal assay, and reported in Miller Units (MU) as the mean. b Comparison between the strengths of interaction between 15(Intra I) and either 15(CTD) or
T-W. Lo et al. 9 SI
TABLE S2
ROG-1 does not participate in EGL-15 signaling pathways
1 n 30 for all genotypes except rog-1(tm1031) where n=12.
