Modulation of EGF receptor-mediated vulva development by the heterotrimeric G-protein Gαq and excitable cells in C. elegans

C. elegans vulva development is used as a model system to study the regulation of cell fate by growth factors. Normally, only three of six vulval precursor cells (VPCs) (P3.p-P8.p) adopt vulval cell fates (reviewed by Greenwald, 1997; Moghal and Sternberg, 2003b). At the end of the second larval stage of development, LIN-3(Hill and Sternberg, 1992), an EGF-like growth factor produced by the gonadal anchor cell induces P6.p to adopt a primary vulval fate. LET-23 (Aroian et al., 1990), an EGF receptor-like tyrosine kinase, and its downstream effectors, which include LET-60 (RAS)(Beitel et al., 1990; Han and Sternberg, 1990) and MPK-1 (MAP kinase) (Lackner et al.,1994; Wu and Han,1994), transduce the LIN-3 signal. Following LET-23 activation in P6.p, LIN-12 NOTCH-like receptors (Yochem et al., 1988) and LET-23 are stimulated on the adjacent P5.p and P7.p cells, which induces secondary vulval fates in these cells(Katz et al., 1995; Simske and Kim, 1995). A WNT signaling pathway acts parallel to the RAS pathway to promote vulval fates,but unlike the RAS pathway, WNT signaling is not absolutely essential for vulval fate specification. A null mutation in the β-catenin gene, bar-1, causes a partially penetrant vulvaless phenotype(Eisenmann et al., 1998), in which fewer than three VPCs adopt vulval fates, and a loss-of-function mutation in the axin-like gene pry-1(Korswagen et al., 2002; Maloof et al., 1999) causes some animals to have a multivulva phenotype because of more than three VPCs adopting vulval fates (Gleason et al.,2002).

Several genes have been identified whose mutation does not affect vulval induction under standard laboratory growth conditions, but affects vulva development in sensitized backgrounds. These include ksr-1(Kornfeld et al., 1995; Sundaram and Han, 1995), ksr-2 (Ohmachi et al.,2002), sur-8(Sieburth et al., 1998), sur-6 (Sieburth et al.,1999), ptp-2 (Gutch et al., 1998), unc-101(Lee et al., 1994), sli-1 (Jongeward et al.,1995; Yoon et al.,1995), gap-1 (Hajnal et al., 1997), ark-1(Hopper et al., 2000), lip-1 (Berset et al.,2001), dpy-22/sop-1(Moghal and Sternberg, 2003a), eor-1 and eor-2 (Howard and Sundaram, 2002), and the redundant class A and class B genes in the synthetic multivulva pathway (e.g. Ferguson and Horvitz, 1989). It is conceivable that the general absence of mutant phenotypes for these genes reflects roles under natural ecologic conditions that are not recapitulated in the laboratory. Therefore it is unclear whether vulval cell-fate specification is modulated by additional pathways in the wild, and in what context this modulation might occur.

The first report of a large-scale genetic screen to identify mutations affecting vulva development provided evidence that the environment and an animal's physiology can modulate the ability of growth factors to induce vulval fates (Ferguson and Horvitz,1985). The severity of the vulvaless phenotypes of certain let-23, lin-2, lin-7, lin-3, lin-24 and lin-33 alleles is reduced by starvation and exit from dauer. The dauer is an arrested,alternative third stage of larval development that occurs under conditions of high population density and reduced food supply (reviewed by Riddle and Albert, 1997). Both entry and exit from dauer are controlled by chemosensory cues involving neurons (Bargmann and Horvitz,1991; Shakir et al.,1993; Tabish et al.,1995), suggesting that excitable cells have the capacity to modulate vulval cell fate. Because the contribution of excitable cell activity to growth factor-dependent regulation of cell fate has not been studied in detail, we sought to analyze this in C. elegans.

We find that activation of the heterotrimeric Gαq protein EGL-30,normally associated with regulation of animal behavior, promotes vulval cell fates. Post-embryonic muscles, and in particular, muscle-expression of the EGL-19 α1 L-type voltage-gated-calcium channel subunit are required,suggesting that muscle excitation can promote development of vulval tissue. This pathway is sensitive to functional levels of BAR-1 (β-catenin), and can be stimulated by activation of EGL-30 in neurons, or by the EGL-30-dependent change in behavior that occurs when worms are grown in a liquid environment. On plates, ablation of the post-embryonic muscles, or egl-19 and egl-30 loss-of-function mutations, do not affect vulval induction, suggesting that this pathway might exist to modulate development in response to certain environmental conditions.

C. elegans were cultured at 20°C using standard protocols(Brenner, 1974). Alleles used were: egl-30(tg26gf) (Doi and Iwasaki, 2002), egl-30(ad805)(Brundage et al., 1996), unc-13(e51) (Brenner,1974), goa-1(n363)(Segalat et al., 1995),phm-2(ad538) (Avery,1993), unc-54(e190)(Epstein et al., 1974) on LGI; let-23(sy1) (Aroian and Sternberg,1991), let-23(sa62gf)(Katz et al., 1996),unc-4(e120) (White et al.,1992) on LGII; dpy-17(e164)(Brenner, 1974),sur-1(ku1) (Wu and Han,1994), unc-64(e246)(Brenner, 1974), pha-1(e2123ts) (Granato et al.,1994) on LGIII; unc-24(e138)(Riddle, 1978), lin-3(n378) (Ferguson and Horvitz, 1985), lin-3(e1417)(Ferguson and Horvitz, 1985); egl-19(n582) (Trent et al.,1983), egl-19(n2368gf)(Lee et al., 1997), mec-3(e1338) (Way and Chalfie,1989), let-60(n1046gf)(Beitel et al., 1990; Ferguson and Horvitz, 1985), let-60(sy95dn) (Han et al.,1990; Han and Sternberg,1991), dpy-20(e1362), dpy-20(e1282)(Hosono et al., 1982) on LGIV; him-5(e1490) (Hodgkin et al., 1979) on LGV; syIs1(Katz et al., 1995), bar-1(mu63) (Maloof et al.,1999), bar-1(ga80)(Eisenmann et al., 1998), and lin-15(e1763) (Ferguson and Horvitz, 1985) on LGX. Genetic balancers used were: mnC1[dpy-10(e128) unc-52(e444)](Herman, 1978; Sigurdson et al., 1984) on LGII; nT1[let(m435)] on LGIV and LGV(Rogalski and Riddle,1988).

let-23(sa62gf)/+; syIs1/+ worms were obtained by first crossing N2 males into hermaphrodites carrying the X-linked syIs1 transgenic array. F1 males that were hemizygous for the X chromosome were used to transfer syIs1 into homozygous let-23(sa62gf) unc-4(e120)hermaphrodites. Non-Unc F1 hermaphrodite cross progeny were scored for vulval induction during the L4 stage. let-23(sa62gf)/+; let-60(sy95dn)/+;syIs1/+ animals were obtained by crossing syIs1-bearing males into let-23(sa62gf) unc-4(e120); unc-24(e138)let-60(sy95dn)/nT1[let(m435)] hermaphrodites, and picking non-Unc cross progeny. Because the F1 cross progeny consisted of both let-60(sy95dn)/+ and nT1[let(m435)]/+ genotypes, we scored vulval induction in all cross progeny, recovered the individual worms and identified the F1 genotype by examining the F2 generation. The let-60(sy95dn)/+ genotype was assigned to F1s segregating Vul animals and dead larvae in the F2 generation.

The plasmid pR30, used to overexpress wild-type egl-30, was constructed as follows. Full-length egl-30 genomic DNA was amplified by PCR from wild-type worms using the upstream primer 5′-ATGGCCTGCTGTTTATCCGAAGAG-3′ and the downstream primer 5′-TTACACCAAGTTGTACTCCTTCAGATTATGCTGTAGAAT-3′. The PCR product was blunt-end ligated into the BamHI site of the unc-119promoter plasmid pBY103 (a gift from M. Maduro), making pR30(unc-119::egl-30). To determine how the egl-30 introns contributed to the expression pattern of pR30, pR39 was constructed from pR30,and contained the gfp open reading frame cloned in-frame to egl-30. The same upstream egl-30 primer was used with the downstream primer 5′-CACCAAGTTGTACTCCTTCAGATTATGCTGTAGAATTG 3′that does not contain a stop codon, to amplify full-length egl-30genomic DNA by PCR. This egl-30 coding fragment was fused in-frame to gfp by blunt-end ligating the PCR fragment into the BamHI site of the promoterless GFP plasmid pPD95.75 (a gift of A. Fire), creating the construct pR36. A 5 kbp XbaI-ApaI fragment from pR36 was swapped between the NheI-ApaI sites of pR30 to create the unc-119::egl-30::gfp plasmid pR39. The myo-3::egl-19 and myo-3::egl-19::gfp constructs were generated by injection of ligation reactions directly into worms and have been described previously(Garcia et al., 2001). The aex-3::egl-30(tg26gf) plasmid, pTG100.1, which places the egl-30 cDNA containing the tg26 mutation under the control of the aex-3 promoter, has been described previously(Doi and Iwasaki, 2002). This egl-30 cDNA contains the last intron of egl-30 to aid with expression from this vector. To determine how the last intron of egl-30 affects expression from the aex-3 promoter, pAEXYFP was constructed. This plasmid contains the yfp coding region and unc-54 3′UTR inserted downstream of the aex-3 promoter and upstream of the last 58 bp of exon 7 to the end of exon 8 (including the last intron) of egl-30. The yfp coding region and unc-54 3′UTR were amplified by PCR from pSX95.77 (courtesy of S. Xu) with the primers 5′-GCCGCGGATCCAAATGAGTAAAGGAGAAGAACTTTTCAC-3′ and 5′-ACGTGCTGCGACAAACAGTTATGTTTGGTATATTGGGAATG-3′. pSX95.77 contains the yfp coding region and unc-54 3′UTR from pPD136.64(a gift of A. Fire) inserted into pBR322. The yfp::unc-54 PCR product was digested with BamHI and SalI, and ligated into BamHI/SalI-digested pTG100.1. pLIN31EGL30 was constructed by amplifying the egl-30(tg26gf) cDNA from pTG100.1 with the primers 5′-ATAAGAATGCGGCCGCAAAAATGGCCTGCTGTTTATCCGAAGAG-3′ and 5′-CCTGTAAAGCGGCCGCTTACACCAAGTTGTACTCCTTCAG-3′, and cloning the NotI-digested PCR product into the NotI site of the lin-31 expression vector, pB255(Tan et al., 1998). The unc-18::egl-30(tg26gf) plasmid, pUNC18tg26, was created as follows. First, the egl-30 cDNA containing the egl-30(tg26gf)mutation and the last intron was released from pTG100.1 by digestion with ApaI and BsrG1. The egl-30 fragment was made blunt,and subsequently cloned into the SmaI site of pBSKS to generate pBStg26. The unc-18 promoter (extending from the unc-18start ATG to the next 5′ gene F27D9.8) was amplified by PCR with the primers UNC18-3 5′-AGCCCAAGCTTTGAAGGACAATGAACTAGAGG GAC-3′ and UNC18-4 5′-AGCCCAAGCTTCCCATTTTTCAAAAATCCTCGTC GATGCACTCAC-3′,digested with HindIII, and cloned into HindIII-digested pBStg26 to yield pUNC18tg26. The unc-18::yfp::egl-30 reporter plasmid, pUNC18YFP, was created as follows. First, the yfp::egl-30intron fusion from pAEXYFP was released by digestion with ApaI and BamHI, the ends made blunt and the fragment cloned into the SmaI site of pBSKS to yield pBSYFPEGL30. The PCR-generated unc-18 promoter fragment was then cloned into the HindIII site of pBSYFPEGL30 to yield pUNC18YFP. The plasmid pLR1, which was used to express the egl-30(tg26gf) cDNA from the unc-4 promoter, was constructed as follows. A 3.0 kbp HindIII-Sma I fragment,which contains the unc-4 promoter and extends through part of the second exon (Miller and Niemeyer,1995), was cloned between the HindIII and SmaI sites in pBSKS to make the plasmid pBSUNC4 (courtesy of C. Van Buskirk). The primers 5′ ATGGCCTGCTGTTTATCCGAAGAG 3′ and 5′TCCCCCGGGGGATTTCGAGGTTAGCTTGATGGG 3′ were used to PCR amplify the egl-30(tg26gf) cDNA and its 3′ UTR sequences from pTG100.1. The∼2.3 kbp PCR fragment was then blunt-end ligated between the NsiI and SmaI sequences in the unc-4 expression vector. The cloning fuses the first egl-30(tg26gf) ATG 18 basepairs downstream of the unc-4 initiation codon.

PCR fragments containing native egl-30 upstream sequences and coding region were too unstable to be maintained in worms when injected into the gonadal syncytium at concentrations even as low as 5 pgμl^-1. Transgenic worms were extremely hyperactive, slow growing and had low fertility, making transmittance of the extrachromosomal arrays not efficient for strain maintenance. To circumvent this problem, we fused the egl-30 genomic coding region to the unc-119 promoter to make the pR30 hybrid construct that led to more stable expression of egl-30. pR39, which has gfp cloned in-frame to egl-30 in pR30, was used to determine the expression pattern of the pR30 hybrid construct. Injection of pR39 at a concentration of 50 ngμl^-1 into worms resulted in expression of GFP in the nervous system, pharyngeal muscles, sex muscles, anal depressor muscles and epidermis(data not shown). Although the unc-119 promoter drives expression mainly in neurons, sequences in the egl-30 genomic DNA (specifically from the first intron) contribute to broad expression of the transgene. Consistent with this expression pattern, injection of pR30 at 50 pgμl^-1 rescued every behavioral phenotype caused by the loss-of-function mutation egl-30(ad805) {data not shown, rescuing array: syEx474 [myo-2:gfp (10 ng μl^-1); unc-119::egl-30 (50 pg μl^-1)]}. The extrachromosomal array that overexpresses wild-type egl-30, syEx532, was obtained by co-injecting pR30 (750 pg μl^-1) and pTG96 (sur-5::gfp)(Gu et al., 1998) (10 ngμl^-1) into egl-30(ad805); let-23(sy1) hermaphrodites. From the GFP-positive transgenic animals, the most hyperactive transmitting line was kept and scored for vulval development.

Experiments with the egl-30 cDNA driven from the unc-119promoter indicated that functional rescue of the egl-30(ad805)phenotypes could be obtained in the F1 generation, but could not be segregated into the F2 generation without the first intron of egl-30. Microinjection mixtures containing digested N2 genomic DNA as a source of carrier DNA were also used in an attempt to create complex arrays; however,this did not alleviate the generational silencing problem. These data suggested to us that the first intron of egl-30 prevents generational silencing of the transgene, and that high levels of EGL-30 activity in certain cell populations are toxic. Because of this problem, we were not able to achieve stable expression of the wild-type or tg26mutation-containing egl-30 cDNA from certain heterologous promoters. To analyze the consequences of driving egl-30(tg26gf) cDNA expression in neurons, we coinjected either pTG100.1 [aex-3::egl-30(tg26gf)] (50 ng μl^-1), or pUNC18tg26 [unc-18::egl-30(tg26gf)] (50 ngμl^-1), or pLR1 [unc-4::egl-30(tg26gf)] (50 ngμl^-1) with pPD118.33 (myo-2::gfp) (10 ngμl^-1) and pBSSK (120 ng μl^-1) into lin-3(n378) animals. Transgenic F1s were identified by expression of myo-2::gfp in the pharynx. The extrachromosomal arrays containing lin-31::egl-30(tg26gf) were generated by injecting pLIN31EGL30[lin-31::egl-30(tg26gf)] (50 ng μl^-1) with pBSSK (140 ng μl^-1) and pPD118.33 (myo-2::gfp) (10 ngμl^-1) into lin-3(n378) animals. The extrachromosomal arrays syEx570 and syEx594 were generated by injecting pAEXYFP (aex-3::yfp::egl-30) (200 ng μl^-1) with pBX-1(pha-1) (Granato et al.,1994) (100 ng μl^-1) or pUNC18YFP(unc-18::yfp::egl-30) (50 ng μl^-1) with pBX-1(pha-1) (100 ng μl^-1) and pBSSK (30 ngμl^-1) into pha-1(e2123ts) animals, respectively.

Vulval induction was scored during the L4 stage under Nomarski optics(Sternberg and Horvitz, 1986). The number of vulval nuclei is used to extrapolate how many of the Pn.p cells were induced to adopt vulval fates. A VPC that gives rise to seven or eight great granddaughters and no hyp7 tissue is scored as 1.0 cell induction. A VPC in which one daughter fuses with hyp7, and the other daughter divides to generate three or four great granddaughter cells is scored as 0.5 cell induction. In wild-type animals, P5.p, P6.p and P7.p each undergo 1.0 cell induction, whereas the other Pn.p cells do not adopt vulval fates, resulting in a total of 3.0 cell induction. Animals displaying more than 3.0 cell induction are multivulva and animals with less than 3.0 cell induction are vulvaless. Laser ablations were conducted using a standard protocol(Bargmann and Avery, 1995). M cell ablations were done at the L1 stage and were confirmed by the loss of M-derived coelomocytes (Sulston and Horvitz, 1977).

A solution of commercial bleach and 4N NaOH (1:1 v/v) was applied to gravid worms on NG plates without bacteria. The next day, starved, synchronized L1 worms were transferred to liquid M9 buffer(Brenner, 1974) in 1.5 ml Eppendorf tubes. On average, tubes contained 30-40 worms in a final volume of 25 μl and E. coli OP50 at a concentration of A600 nm=1.0. Worms were grown in a 20°C incubator without shaking until they reached mid-L4,at which time, vulval induction was scored.

EGL-30 is the ortholog of mammalian Gαq/Gα₁₁, and is the only C. elegans member of the Gαq class of heterotrimeric Gα proteins (Brundage et al.,1996; Jansen et al.,1999). During the course of behavioral studies with the egl-30(tg26) allele (Doi and Iwasaki, 2002), we noticed that rare hermaphrodites (0.5%, n=200) developed ectopic vulval tissue. The egl-30(tg26)allele causes a glutamine for arginine substitution (R243Q) in the α3 helix region of Gαq, a region implicated in GTP hydrolysis and effector binding (Itoh and Gilman,1991; Noel et al.,1993; Sprang,1997). In contrast to egl-30 loss-of-function alleles that decrease motility and egg-laying(Brundage et al., 1996), the tg26 allele induces semi-dominant hyperactive locomotion and egg-laying behaviors (data not shown), suggesting that the mutation confers gain-of-function properties to the protein. The rare occurrence of ectopic vulval tissue in egl-30(tg26gf) hermaphrodites suggests that activated EGL-30 might interact with the LET-23 pathway during vulval induction. To study interactions with the LET-23 pathway, we used the same strategy used to study other modulators of LET-23 signaling, which also do not cause penetrant phenotypes when mutated in isolation. We examined the effects of activation of EGL-30 in sensitized backgrounds. We made double mutant combinations of gain-of-function egl-30(tg26gf) with loss-of-function lin-3(n378), let-23(sy1), and dominant-negative let-60(sy95dn). egl-30(tg26gf) reduces the severity of the vulvaless phenotypes caused by all three mutant alleles (Table 1; see Table 4).

Because egl-30(tg26gf) causes several behavioral phenotypes, we tested whether modulation of vulval induction was caused by an indirect physiological consequence of the activated allele. egl-30(tg26gf)worms have hyperactive muscle behaviors and feeding problems. Thus, we tested other mutations that cause similar phenotypes for suppression of the let-23(sy1) vulvaless phenotype. phm-2(ad538) has a deletion removing the 3′ end of dys-1 (dystrophin-like), and goa-1(n363) has a deletion removing goa-1 (Gαo). Although both alleles cause muscle hyperactivity and feeding defects(Avery, 1993; Bessou et al., 1998; Segalat et al., 1995), neither suppresses the let-23(sy1) vulvaless phenotype(Table 1). Therefore,modulation of vulval induction by egl-30(tg26gf) is unlikely to be caused by general effects on movement or worm physiology.

We next tested whether the effects on vulval cell fate caused by the R243Q change encoded by egl-30(tg26gf) reflected elevated EGL-30 activity or a new property of EGL-30. We injected the wild-type egl-30 genomic coding region into an egl-30 loss-of-function; let-23loss-of-function double mutant. We were able to recover one transgenic line, egl-30(ad805); let-23(sy1); syEx532 [egl-30 XS], that displayed hyperactive behaviors similar to those observed in egl-30(tg26gf)animals (data not shown). In addition to conferring behavioral similarities with egl-30(tg26gf) animals, overexpression of wild-type EGL-30 also suppresses the vulvaless phenotype let-23(sy1)(Table 1). Thus, elevated EGL-30 activity promotes vulval induction.

Anti-EGL-30 antibody staining indicates that EGL-30 is most strongly expressed in neurons (C.B., M.S. and P.W.S., unpublished), and egl-30::gfp translational fusions also reveal strong expression in neurons, as well as muscle, and the differentiated secondary cells of the mature vulva (Lackner et al.,1999) (C.B., M.S. and P.W.S., unpublished). In view of its expression in excitable cells, and the defective locomotion and egg-laying behaviors associated with loss-of-function mutations in egl-30(Brundage et al., 1996), we asked whether neuronal signaling is required for activated EGL-30 to promote vulval induction. We tested whether mutations in either unc-13, which encodes a diacyglycerol-binding protein involved in EGL-30-mediated synaptic transmission (Lackner et al.,1999; Maruyama and Brenner,1991; Richmond et al.,1999), or unc-64, which encodes syntaxin 1a(Ogawa et al., 1998; Saifee et al., 1998), disrupt the ability of egl-30(tg26gf) to suppress the let-23(sy1)vulvaless phenotype. The non-null unc-64(e246) and unc-13(e51) alleles partially reduced the effect of activated EGL-30 on vulval induction, possibly indicating some involvement of neurons in this pathway (Table 2).

To determine the site(s) of action for EGL-30, we expressed the egl-30(tg26gf) cDNA under the control of muscle, neuronal and vulval-specific promoters. In general, stable expression of the egl-30 cDNA from heterologous promoters in transgenes was difficult to achieve (see Materials and Methods). When egl-30(tg26gf)expression was driven by the myo-3 muscle-specific promoter in lin-3(n378) animals (Okkema et al., 1993), high transgene doses failed to yield viable F1 transformants, and at low doses, F1s carrying the transgene did not display any behavioral phenotypes, and were not rescued for the vulvaless phenotype(data not shown).

To drive activated EGL-30 expression in neurons, we first used the aex-3 (Iwasaki et al.,1997) and unc-18(Gengyo-Ando et al., 1993)promoters, which drive expression in multiple neurons, including head, tail and ventral cord motor neurons. Because the egl-30 cDNA we used contained the last intron of egl-30 to promote splicing, we verified that this intron did not cause ectopic expression in muscle, the Pn.p cells or the anchor cell. When placed downstream of the yfp coding region, the last intron of egl-30 did not alter the activity of the aex-3 or unc-18 promoters. In both cases, YFP accumulated in the ventral cord and head and tail neurons, but not in muscle, the anchor cell or the Pn.p cells (Fig. 1A-Fand Fig. 2). We co-injected the aex-3::egl-30(tg26gf) or unc-18::egl-30(tg26gf) plasmids with pPD118.25, which drives GFP expression in the pharynx, into lin-3(n378) vulvaless animals. We observed two types of GFP-positive transgenic F1 animals: those that displayed a hyperactive phenotype similar to that of egl-30(tg26gf) mutants, and those that looked behaviorally wild-type. None of the F1s displaying a strong behavioral phenotype transmitted this phenotype to subsequent generations, suggesting that transgenic expression of egl-30 was being lost, and that high levels of EGL-30 activity can be toxic (see Materials and Methods). Therefore, we only examined F1 animals for rescue of vulval induction defects. We found that aex-3 or unc-18-driven EGL-30 (R243Q) could only rescue the vulvaless phenotype of lin-3(n378) when expressed in cells at levels sufficient to confer a hyperactive phenotype(Table 2). Animals that were GFP-positive, but did not show a behavioral phenotype, did not have more vulval induction than uninjected worms(Table 2). This tight correlation between the effects of the aex-3::egl-30(tg26gf) and unc-18::egl-30(tg26gf) transgenes on behavior and rescue of the lin-3(n378) vulvaless defect, suggests that activation of EGL-30 in motor neurons promotes vulval induction. To examine this possibility in more detail, we used the unc-4 promoter(Miller and Niemeyer, 1995) to drive expression of activated EGL-30 in the A-type motor neurons, which include the VAs and DAs in ventral cord, and the SABs in the retrovesicular ganglion. We obtained three stable transgenic lines, with one line demonstrating clear functional rescue of the lin-3(n378) vulvaless phenotype (Table 2). Together,these results suggest that activation of EGL-30 in ventral cord motor neurons promotes vulval induction.

To examine the effects of activating EGL-30 in the vulval precursor cells,we cloned the egl-30(tg26gf) cDNA downstream of the lin-31promoter, which only drives expression in the Pn.p cells(Tan et al., 1998). When injected into lin-3(n378) vulvaless animals, this construct was not able to rescue the vulvaless phenotype in either transgenic F1s or stable lines (Table 2). Thus,activation of EGL-30 in excitable cells, rather than the vulval precursor cells, promotes vulval cell fates.

Because of the hyperactive movement phenotype displayed by egl-30(tg26gf) mutants, the correlation between hyperactive behavior and enhancement of vulval induction in the aex-3::egl-30(tg26gf) and unc-18::egl-30(tg26gf) transgenic animals, and the ability of activated EGL-30 to promote vulval development from the A-type motor neurons,we considered the possibility that muscle excitation might be necessary for EGL-30 to promote vulval cell fates. We therefore used an egl-19loss-of-function mutation to compromise muscle excitation(Jospin et al., 2002; Lee et al., 1997). egl-19 encodes the worm homolog of the L-type voltage-gated calcium channel α1 subunit, and although it is expressed in both neurons and muscles, site of action experiments have thus far only demonstrated function in muscles (Garcia et al.,2001; Jospin et al.,2002; Lee et al.,1997). Reducing EGL-19 activity with the egl-19(n582)allele did not affect vulval induction by itself, but it strongly reduced the ability of egl-30(tg26gf) to suppress the let-23(sy1)vulvaless phenotype (Table 3). However, hyperactivation of EGL-19 by the gain-of-function egl-19(n2368gf) allele, which induces hypercontraction of muscles(Lee et al., 1997), did not rescue the vulvaless phenotype of let-23(sy1), suggesting that EGL-30 has additional targets besides EGL-19(Table 3).

To confirm that muscle-expression of EGL-19 is required for EGL-30 to promote vulval induction, we used a transgene that expresses the genomic coding region of egl-19 from the muscle-specific myo-3myosin promoter (Okkema et al.,1993). When the entire egl-19 genomic coding region was fused in-frame to gfp, the myo-3 promoter directed high levels of GFP expression to muscle, but not to neurons, the anchor cell or the vulval precursor cells (Fig. 1G-L), confirming the specificity of this promoter and the absence of regulatory elements in egl-19 introns. We found that wild-type EGL-19 expressed from the myo-3 promoter restores vulval induction to egl-30(tg26gf); let-23(sy1); egl-19(n582) triple mutants, indicating that EGL-30 requires muscle-expressed EGL-19 for modulating vulval induction(Table 3).

To identify muscles involved in the EGL-30-stimulated pathway, we ablated the M mesoblast in egl-30(tg26gf); let-23(sy1) L1 larvae. The M mesoblast gives rise to 14 post-embryonic body-wall muscles prior to the onset of vulval induction, and 16 sex muscles after vulval induction has occurred(Sulston and Horvitz, 1977). Removal of the M mesoblast does not enhance the vulvaless phenotype of animals containing the weak ku1 mutation in the worm MAP kinase gene mpk-1 (Wu and Han,1994), demonstrating that ablation of the M lineage does not generally exacerbate mutation-induced defects in vulval induction(Table 3). However, ablation of the M cell reduces egl-30(tg26gf)-mediated suppression of the let-23(sy1) vulvaless phenotype(Table 3). Because the sex-muscles are not formed until after the period of vulval induction, and the myo-3 promoter is not expressed in undifferentiated sex myoblasts(data not shown), the post-embryonic body-wall muscles are the muscles that probably transduce some of the EGL-30-modulating activity to the VPCs. Furthermore, these muscles only contribute to vulval induction under certain conditions, such as in the presence of activated EGL-30, which causes muscle hyperactivity.

Experiments in cultured mammalian cells have demonstrated that heterotrimeric G-protein signaling can promote EGFR and MAP kinase activation(reviewed by Gschwind et al.,2001; Lowes et al.,2002). In some instances, G-proteins promote metalloprotease-dependent shedding of the EGFR ligand, HBEGF, whereas in other cases G-proteins promote MAP kinase activation through stimulation of signaling molecules such as protein kinase C, c-SRC and PYK2. We therefore tested whether activated EGL-30 acts by promoting LET-23 (EGFR) activation. We constructed double mutants with dominant-negative let-60(sy95dn) and either the gain-of-function let-23(sa62gf) allele or egl-30(tg26gf). let-23(sa62gf) encodes a constitutively active receptor that induces ectopic vulval fates in more than 90% of animals(Katz et al., 1996)(Table 4). Despite the ability of constitutively active LET-23 to cause a much stronger multivulva phenotype than egl-30(tg26gf), the let-23(sa62gf) allele did not suppress the let-60(sy95dn) mutation as well as egl-30(tg26gf) (Table 4). We also examined whether overexpression of LIN-3 can suppress the vulvaless phenotype conferred by dominant-negative LET-60 (RAS). In this experiment, we also included one copy of let-23(sa62gf) to further enhance the amount of pathway activation upstream of RAS. Despite the fact that multiple copies of the lin-3 gene in the form of the integrated transgenic array syIs1 (Katz et al., 1995) also confer a much stronger multivulva phenotype than egl-30(tg26gf) (Table 4), egl-30(tg26gf) is still a much better suppressor of the dominant-negative let-60(sy95dn) mutation. These data indicate that EGL-30 pathway activity is not correlated with functional levels of LIN-3 or LET-23 activation, and that EGL-30 acts either downstream or parallel to LET-60 (RAS).

To determine whether EGL-30 acts directly on some component downstream of LET-23 signaling, we made use of the observation that egl-30(tg26gf)suppresses the vulvaless phenotypes of lin-3(n378) and let-23(sy1) single mutations(Table 1). If EGL-30 acts downstream of receptor activation, then activation of EGL-30 might be expected to suppress a let-23(sy1); lin-3(n378) double mutant (Table 4). We find that egl-30(tg26gf) cannot suppress the vulvaless phenotype of let-23(sy1); lin-3(n378) double mutants(Table 4). Taken together, our data suggest that EGL-30 might act parallel to the LET-23 pathway.

WNT signaling is the only other pathway known to act parallel to the RAS pathway during vulval induction. A null mutation in the β-catenin bar-1 results in a partially penetrant vulvaless phenotype, partly because of a reduction in lin-39 hox gene expression(Eisenmann et al., 1998), also a target of the RAS pathway (Maloof and Kenyon, 1998). Moreover, hyperactivation of WNT signaling through a mutation in pry-1, an axin-like inhibitor of the WNT pathway, can suppress let-23 pathway vulvaless mutations(Gleason et al., 2002). Based on these observations, we explored the possibility that activated EGL-30 might promote vulval induction by elevating WNT pathway activity. We found that unlike the effects on the RAS pathway mutations, egl-30(tg26gf) did not suppress the vulvaless phenotype of the bar-1(ga80) allele(Table 4). In addition, we examined the sensitivity of the EGL-30 pathway to bar-1(β-catenin) levels. Although the weak loss-of-function bar-1(mu63) allele is phenotypically wild-type by itself(Maloof et al., 1999), it reduces the output of elevated WNT signaling. In particular, bar-1(mu63) suppresses the ectopic mab-5 expression, and polyray and vulval phenotypes conferred by pry-1(mu38)(Maloof et al., 1999; Moghal and Sternberg, 2003a). bar-1(mu63) also reduces the ability of egl-30(tg26gf) to suppress the vulvaless phenotype of let-23(sy1)(Table 4). The bar-1(mu63) mutation does not, however, affect the ability of a mutation in the RAS pathway component dpy-22 to comparably suppress let-23(sy1) (Moghal and Sternberg, 2003a). Thus, bar-1(mu63) reduces WNT and EGL-30 pathway activity, but not RAS pathway activity. These data are consistent with a model in which EGL-30 acts through BAR-1, rather than on RAS signaling, or a novel third pathway, to promote vulval induction.

Because excitable cell populations can respond to changing environmental conditions, we thought the EGL-30 pathway might modulate vulval induction in response to certain environmental conditions. We therefore searched for an environmental condition that might promote vulval induction in an EGL-30-dependent manner. Previous work has shown that the vulvaless phenotypes of lin-3(n378) and let-23(n1045) can be partially suppressed by exit from dauer and starvation, respectively(Ferguson and Horvitz, 1985). We found that these conditions only weakly suppressed let-23(sy1),and did not suppress the dominant-negative let-60(sy95dn) mutation(data not shown). Besides being grown on standard NG Petri plates, C. elegans can be grown in liquid media. When we grew lin-3(n378),let-23(sy1) and let-60(sy95dn) single mutants in liquid M9 buffer, instead of on Petri plates, we found that in all cases, animals consistently had a higher number of VPCs adopting vulval cell fates compared with animals grown on standard NG plates(Fig. 3, Table 5). Thus, growth of worms in a liquid environment promotes vulval induction.

C. elegans behave differently in liquid media than on NG plates. Instead of crawling in a sinusoidal fashion, the worms vigorously thrash their bodies back and forth. Because activation of EGL-30 promotes hyperactive locomotion on plates, and loss-of-function mutations in egl-30 slow movement (Brundage et al.,1996), it is conceivable that in a liquid environment, endogenous wild-type EGL-30 is strongly activated. Because muscles and the EGL-19 calcium channel only contribute to vulval induction under conditions in which EGL-30 is strongly activated, we predicted that an egl-30 loss-of-function mutation would not affect vulval induction on standard NG plates, but would block the ability of a liquid environment to promote vulval cell fates. Consistent with this model, we found that on NG plates, the loss-of-function egl-30(ad805) allele does not cause a vulvaless phenotype on its own, nor does it suppress the multivulva phenotype caused by excessive activation of the RAS or WNT pathways(Table 5). egl-30(ad805) does not suppress the gain-of-function mutations let-23(sa62) and let-60(n1046), or the loss-of-function mutations lin-15(e1763) and pry-1(mu38). Furthermore, on NG plates, egl-30(ad805) does not enhance the vulvaless phenotype of a weak loss-of-function mutation in the MAP kinase gene mpk-1(ku1)(Table 5). In contrast, egl-30(ad805) blocks the ability of liquid growth to suppress let-23(sy1) (Table 5). Thus, EGL-30 specifically modulates vulval induction under certain environmental conditions, as mimicked by growth in a liquid environment.

Because the pathway stimulated by activated EGL-30 on NG plates is sensitive to functional levels of BAR-1(Table 4), we tested whether the liquid growth-stimulated pathway displayed the same sensitivity to WNT pathway mutations. Similar to the results with egl-30(tg26gf) animals growing on NG plates, liquid growth was not able to suppress the vulvaless phenotype of bar-1(ga80) (Table 5). Furthermore, as with egl-30(ad805), the weak loss-of-function bar-1(mu63) mutation blocked the ability of liquid growth to suppress the let-23(sy1) vulvaless phenotype(Table 5). This result is also similar to the sensitivity of egl-30(tg26gf) to bar-1(mu63)on NG plates (Table 4). Thus,liquid growth-mediated effects on vulval induction are strongly dependent on EGL-30 and BAR-1 (β-catenin) signaling.

Genetic analyses performed under standard laboratory conditions have provided much information regarding the mechanisms by which growth factor signaling regulates cell fates in vivo. However, the paradigms worked out under these conditions might not fully explain the mechanisms underlying animal development, especially with regards to how development occurs in the wild. By analyzing a gain-of-function allele of the heterotrimeric Gαq gene egl-30, we have identified a connection between the activity of excitable cells and the responsiveness of epithelial precursor cells to EGF in C. elegans. Activation of EGL-30 by either a point mutation or overexpression causes hyperactive forward and backward locomotion and suppresses the vulvaless phenotype of loss-of-function mutations in the let-23 pathway (Table 1). egl-30 is strongly expressed in neurons(Lackner et al., 1999) (C.B.,M.S. and P.W.S., unpublished), and consistent with this expression pattern, we find that transgenic expression of activated EGL-30 in neurons, especially those that regulate motor output, including the SABs, and the VA and DA ventral cord motor neurons, also drives this pathway(Table 2). These results suggest that activation of neurons that innervate body-wall muscle, and whose cell bodies are close to the vulval precursor cells, can promote vulval cell fates (Fig. 4). However,mutations in unc-64 and unc-13, which reduce synaptic transmission, only weakly impair the ability of endogenous activated EGL-30 to promote vulval induction (Table 2). This observation could reflect the non-null nature of these alleles, or it could indicate a role for the neurons that is not heavily dependent on synaptic transmission. Although the vulval induction-promoting activity of EGL-30 can be mediated by excitation of motor neurons, hyperactive locomotion, per se, is not required. egl-30(tg26gf); let-23(sy1)animals are severely paralyzed in the presence of the unc-64 and unc-13 mutations, yet still display enhanced vulval development.

Despite the absence of a requirement for hyperactive locomotion for EGL-30 to promote vulva development, muscle excitation appears to be necessary. Disruption of muscle-expressed EGL-19 L-type voltage-gated calcium channel activity, which affects muscle excitation(Garcia et al., 2001; Jospin et al., 2002; Lee et al., 1997), blocks the EGL-30 pathway (Table 3). Because EGL-30 is also expressed in muscle, it is possible that EGL-30 has a second site of action, in muscle, to regulate vulva development(Fig. 4). In this scenario, the egl-30(tg26gf) genetic mutation might cause sufficient cell-autonomous excitation of the muscles to bypass the requirement for UNC-64- and UNC-13-mediated synaptic transmission. This model would be consistent with data that L-type voltage-gated calcium channels can be positively modulated by protein kinase C(Linn, 2000), and diacylglycerol and/or inositol-1, 4, 5-trisphosphate, two second messengers downstream of Gαq (Boyer et al.,1994; Jiang et al.,1994). However, because we find that the egl-19(n2368gf)gain-of-function allele does not phenocopy the effect of activated EGL-30 on vulval induction (Table 3), it is possible that EGL-30 has targets other than EGL-19. Alternatively, because the egl-19(n2368gf) mutation does not cause the same hyperactive behavior observed in egl-30(tg26gf) mutants, the G365R change conferred by the egl-19(n2368gf) mutation might not mimic the effect of activated EGL-30 on the biophysical properties of EGL-19, and the biological properties of muscles.

The M cell lineage is required for the EGL-30 pathway, and myo-3-driven egl-19, which is only expressed in differentiated muscle, can promote pathway activity(Table 3). Because the M-derived body-wall muscles are formed during L2, prior to vulval induction,whereas the M-derived sex muscles are not formed until the L4 stage, after vulval induction is completed (Sulston and Horvitz, 1977), it is probable that post-embryonic body-wall muscles promote vulva development. These muscles occupy positions posterior to the gonadal primordium, both dorsally and ventrally, and are used for locomotion, similar to their embryonically derived counterparts(Garcia et al., 2001; Sulston and Horvitz, 1977). The most anterior of the muscles is close to P7.p., which might be close enough to affect vulval induction directly(Fig. 4). However, because the unc-4-expressing VA and DA motor neurons innervate both ventral and dorsal body-wall muscles, respectively(White et al., 1986), it is also possible that the post-embryonic muscles can affect vulva development from a distance. At present, there is no reason to think that the post-embryonic body-wall muscles are functionally distinct from the other body-wall muscles. Therefore, we favor a model in which ablation of the M cell simply reduces the number of muscles below a critical threshold, so that the EGL-30 pathway can no longer fully promote vulva development.

Several models could explain the form of communication between the neurons,muscles and vulval precursor cells that promotes vulval induction. In all cases, excitation of body-wall muscles is crucial. In a simple model(Fig. 4), EGL-30-driven activation of motor neurons stimulates body-wall muscle, which in turn,signals to the VPCs. In contrast, in C. elegans, body-wall muscles have been shown to modulate neuronal synaptic development via the C2 domain protein AEX-1 and the AEX-5 prohormone convertase(Doi and Iwasaki, 2002). Therefore, it is also possible that neurons may directly modulate vulval induction in a manner that is not strongly dependent on synaptic transmission,but requires retrograde signaling from the muscle to the neurons. Finally,it's possible that two parallel signals are sent by the neurons and body-wall muscle to the vulval precursor cells.

Experiments with cultured mammalian cells have demonstrated that G-protein-coupled receptor activation can lead to metalloprotease-stimulated shedding of HB-EGF, an EGFR ligand (Daub et al., 1996; Prenzel et al.,1999), which can then act in an autocrine and paracrine manner. However, we find that expression of activated EGL-30 in the vulval precursor cells does not promote vulval induction(Table 2), nor is the activity of EGL-30 correlated with functional levels of LIN-3 or LET-23 activation(Table 4). Instead, we find that activated EGL-30 acts downstream or parallel to the LET-60 (RAS)(Table 4). WNT signaling is the only other pathway known to act parallel to RAS signaling during vulval induction, with one convergence point being the lin-39 hox gene(Eisenmann et al., 1998). We find that although activation of EGL-30 can suppress RAS pathway mutations, it cannot suppress the partially penetrant vulvaless phenotype of the bar-1(ga80) null allele (Table 4). Moreover, although the bar-1(mu63) loss-of-function allele does not affect vulval induction on its own, it reduces the ability of egl-30(tg26gf) to suppress the vulvaless phenotype of let-23(sy1) (Table 4). This sensitivity of egl-30(tg26gf) to bar-1(mu63) is specific because suppression of let-23(sy1) by the RAS pathway mediator component dpy-22 is not affected by bar-1(mu63)(Moghal and Sternberg, 2003a). These data suggest that one possible model in which EGL-30 promotes vulval induction is by upregulating bar-1 (β-catenin) signaling(Fig. 4). In support of this model, elevated bar-1 (β-catenin) signaling resulting from a loss-of-function mutation in the axin-like gene, pry-1, or from overexpression of a non-degradable form of BAR-1 suppresses the vulvaless phenotype of let-23(sy1) (Gleason et al., 2002). Mammalian cell culture experiments have suggested that Gαq may be a downstream component of WNT receptors(Liu et al., 2001; Liu et al., 1999). However,because transgenic expression of activated EGL-30 in neurons, but not in the vulval precursor cells, promotes vulval induction(Table 2), this model is unlikely to explain our results. Furthermore, BAR-1 is expressed in the vulval precursor cells, but not in muscles and neurons(Eisenmann et al., 1998), and we have shown that the latter cells mediate the effects of EGL-30.

Excitable cells can act as sensors for an animal's environment. Thus, the existence of pathways by which excitable cells can contribute to the developmental fates of cells may be generally important in ensuring that correct developmental decisions are made under a wide range of conditions. Accordingly, we find that when animals are removed from NG plates, and are placed in a liquid environment, an EGL-30-dependent pathway is activated which promotes vulval induction (Table 5). Because the egl-30(ad805) mutation does not affect vulval induction under conditions in which the RAS and WNT pathways are hyperactivated or compromised by genetic mutation(Table 5), EGL-30-mediated regulation of vulval induction is specific to conditions affecting animal behavior. Similar to our studies with activated EGL-30 on NG plates, we find that growth in liquid media suppresses vulvaless mutations in the RAS pathway,but not the bar-1(ga80) null mutation, and that the bar-1(mu63) loss-of-function mutation blocks the liquid-stimulated pathway (Table 5). Thus, one model for the liquid growth enhancement of vulval induction could also involve the indirect stimulation of BAR-1 (β-catenin) in the vulval precursor cells by EGL-30 (Fig. 4).

The link between the environment and the EGL-30 modulatory pathway suggests that some of the other positive and negative regulators of vulval induction,which have no phenotypes under standard growth conditions on NG plates, might also play important roles under different environmental conditions. It has recently been reported that the G-protein-coupled receptor, SRA-13, and the heterotrimeric Gα protein, GPA-5, are inhibitors of vulva development(Battu et al., 2003). Although mutations in these genes can affect vulva development on standard NG plates,SRA-13 is also necessary for starvation-mediated inhibition of vulva development. sra-13 and gpa-5 are both expressed in chemosensory neurons, and sra-13 is additionally expressed in body-wall muscle (Battu et al.,2003; Jansen et al.,1999). Thus, in conjunction with our results, it appears that C. elegans can use its neuromuscular system to both promote and inhibit vulva development, depending on the environmental context.

Although RAS and WNT signaling are both important for inducing vulval cell fates, hypomorphic mutations in the let-23 pathway cause more severe vulvaless phenotypes than a bar-1 null mutation. This difference raises questions as to why two different signaling pathways are used to specify vulval cell fates, and why they are used disproportionately. If a cell fate must be induced at a particular time in development, it might be best accomplished by robust activation of a pathway that is largely insensitive to environmental changes. However, should conditions arise that are deleterious to that pathway, the existence of a second pathway that is modulated by the environment would ensure that development remains invariant. Under stressful conditions, the hermaphrodite may use a behavioral response elicited by excitable cells and Gαq signaling to promote the activity of the WNT pathway, and ultimately enhance RAS-dependent vulval cell differentiation. The interplay between the environment, neurons, muscles and these signaling pathways adds a new dimension to the existing paradigms by which growth factors trigger cell fate changes during animal development.

Some nematode strains used in this work were provided by the Caenorhabditis Genetics Center, which is funded by the NIH National Center for Research Resources (NCRR). We thank A. Fire, S. Xu and C. Van Buskirk for providing plasmids, and B. Neel, C. Bastiani, R. Lee and G. Schindelman for discussions. This research was supported by the Howard Hughes Medical Institute of which P.W.S. is an investigator, and L.R.G. was an associate, and by grants from the Ministry of International Trading and Industry to K.I. N.M. was supported by postdoctoral fellowships from the Leukemia and Lymphoma Society, and the California Breast Cancer Research Program. L.R.G. was supported by a Department of Health and Human Services National Research Award (GM18857).