A component of the transcriptional mediator complex inhibits RAS-dependent vulval fate specification in C. elegans

Activation of signal transduction pathways by growth factors is frequently used in development as a mechanism to specify distinct cell fates. In the case of the C. elegans hermaphrodite, vulval cell fate specification(reviewed by Greenwald, 1997;Sternberg and Han, 1998)begins at the end of the second larval stage with the secretion of LIN-3(Hill and Sternberg, 1992), an EGF-like growth factor, from the gonadal anchor cell. Six Pn.p cells,P3.p-P8.p, comprise the vulval precursor cells (VPCs). These cells express LET-23 (Aroian et al., 1990;Kaech et al., 1998;Simske et al., 1996), an EGF receptor-like protein, and are competent to respond to overexpressed LIN-3 by adopting vulval fates (Hill and Sternberg,1992; Katz et al.,1995). However, in the presence of physiological levels of LIN-3,only P6.p adopts a primary vulval fate. This response is transduced through LET-60 (RAS) (Beitel et al.,1990; Han and Sternberg,1990) and SUR-1/MPK-1 (MAP kinase)(Lackner et al., 1994;Wu and Han, 1994), and is accompanied by the activation of LIN-12, a NOTCH-like molecule(Yochem et al., 1988), on adjacent P5.p and P7.p, which causes these cells to adopt secondary vulval fates (Simske and Kim, 1995). The remaining VPCs, P3.p, P4.p and P8.p, do not adopt vulval fates, and fuse with the hypodermal syncytium, hyp7.

Given the likely diffusible nature of LIN-3, mechanisms must exist to ensure that normally only a subset of the VPCs responds to the growth factor. A ligand-independent gain-of-function allele of let-23, sa62, confers ectopic vulval fate transformations in P3.p and P4.p at high frequency(Katz et al., 1996),indicating that receptor activation is sufficient to drive cell differentiation. Under physiological conditions, positional selectivity for cell fate determination might be achieved by dependency on a threshold level of LET-23 pathway activation, which is normally achieved only by the VPC colsest to the anchor cell, the source of LIN-3. Consistent with this model,P6.p is positioned closest to the anchor cell and invariantly responds to LIN-3.

Genetic studies of negative regulation of vulval development indicate that additional mechanisms operate to ensure the selective response of P6.p to LIN-3. Although ectopically activated LET-23 induces vulval fate transformations at high frequency in P3.p and P4.p, the posteriorly expressedmab-5 homeobox gene inhibits this response in the most posterior Pn.p cell, P8.p (Clandinin et al.,1997). Another mechanism of restricting the response to LIN-3 involves two classes of genes that function redundantly to inhibit vulval fates. Animals that harbor loss-of-function mutations both in a class A and class B `synmuv' gene display a synthetic multivulva phenotype because of the adoption of vulval fates by most of the VPCs(Ferguson and Horvitz, 1989). This vulval induction is independent of the LIN-3-producing anchor cell, but dependent on LET-23 and its downstream effectors(Ferguson et al., 1987;Huang et al., 1994;Lu and Horvitz, 1998), raising the possibility that this pathway exists to repress low, but functional,levels of ligand-independent activity by the LET-23 pathway. Molecular identification of some of the synmuv genes and RNA interference experiments suggest that this pathway comprises components of a histone deacetylase complex, which represses LET-23-dependent gene expression(Chen and Han, 2001;Lu and Horvitz, 1998;Solari and Ahringer,2000).

Several genes have been identified whose properties do not fully resemble the synmuv genes, but nevertheless, function as negative regulators of vulval induction. These include unc-101(AP47 medium chain of trans-Golgi AP-1 complexes) (Lee et al.,1994), sli-1(c-CBL)(Yoon et al., 1995),gap-1 (Hajnal et al.,1997), ark-1 (ACK-related tyrosine kinase)(Hopper et al., 2000) andlip-1 (MAP kinase phosphatase)(Berset et al., 2001). Mutations in these genes suppress loss-of-function mutations in thelet-23 pathway that cause vulvaless phenotypes, and in different genetic backgrounds, they enhance the frequency of multivulva phenotypes,indicating they affect all six VPCs, similar to the synmuv genes. These negative regulators may function to raise the requirement for the amount of LET-23 pathway activity necessary to drive a functional response.

To address whether these are the only mechanisms and points of negative regulation of the response to LIN-3, we performed a genetic screen for mutations that enhance the frequency of ectopic vulval fate transformations in the presence of gain-of-function let-23(sa62). We isolated two new alleles of sop-1 (Zhang and Emmons, 2000), which has recently been found to be allelic to the older locus, dpy-22 (Meneely and Wood, 1987) (H. Sawa, personal communication). DPY-22 is most closely related to human TRAP230 (Ito et al., 1999; Nagase et al.,1996; Philibert et al.,1998), a component of the transcriptional mediator complex(Ito et al., 1999), and has been shown to be an inhibitor of WNT-dependent ray formation in the C. elegans male tail (Zhang and Emmons,2000). We describe some of the phenotypes of our newdpy-22 alleles, and present evidence that DPY-22 also is an inhibitor of RAS-dependent vulval fate specification, independent of its role in regulating WNT signaling.

C. elegans were cultured at 20°C using standard protocols(Brenner, 1974). Alleles used in this work were: pry-1(mu38)(Maloof et al., 1999) on LGI;let-23(sy1) (Aroian and Sternberg,1991), let-23(sa62gf)(Katz et al., 1996) andunc-4(e120) (White et al.,1992) on LGII; pha-1(e2123ts)(Granato et al., 1994) on LGIII; let-60(n1046gf) (Beitel et al., 1990; Ferguson and Horvitz, 1985), unc-31(e169)(Avery et al., 1993) anddpy-4(e1166) (Cox et al.,1980) on LGIV; him-5(e1490)(Hodgkin et al., 1979) on LGV;and lon-2(e678) (Brenner,1974), bar-1(ga80)(Eisenmann et al., 1998),bar-1(mu63) (Maloof et al.,1999), dpy-6(e14)(Brenner, 1974),dpy-22(bx93) (Zhang and Emmons,2000), egl-15(n484)(Trent et al., 1983) andunc-9(e101) (Brenner,1974) 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). All let-23(sa62) strains carried the linked mutationunc-4(e120), and heterozygous let-23(sa62)/+ strains were balanced with mnC1. let-60(n1046)/+ strains carried the linked markerunc-31(e169) and were balanced by nT1[let(m435)].

bar-1(mu63) was previously reported to have a mutation causing a L130F change (Maloof et al.,1999), which we did not detect in the extant bar-1(mu63)strains. In order to determine the molecular lesion in bar-1(mu63),we amplified 1 kb regions of bar-1 genomic DNA frombar-1(mu63) animals by PCR, and directly sequenced the products. Primer sets used included BAR1-6 5′-ttc agt tct act tgt cta ttg gtg tgc-3′ and BAR1-7 5′-cac atg gta gtc cgc gac ttg tac-3′;BAR1-8 5′-cga gaa ttg acc agc tcc aga aga g-3′ and BAR1-9 5′-gc tgc tta ctg atg aag ccg gtg-3′; and BAR1-10 5′-gc ttt gtg cac aac ctc ctg taa g-3′ and BAR1-11 5′-ct ctt cat ccg gca gac aaa tcg-3. After sequencing 36% of the bar-1 genomic locus, we identified a C to T mutation at position 39108 of C54D1 inbar-1(mu63) animals. This change was detected on both strands, and not in N2 animals. This mis-sense mutation is predicted to cause a G524D change in the BAR-1 protein. Linkage of bar-1(mu63) todpy-22(sy622) was confirmed by DNA sequencing. Linkage ofbar-1(ga80) to dpy-22(sy622) was confirmed by PCR and digestion with MseI, which detects the MseI restriction site created by the ga80 C to T point mutation.

let-23(sa62); him-5(e1490) animals were subjected to standard mutagenesis with ethylmethanesulfonate(Brenner, 1974). One thousand F1 offspring were picked to individual plates, and those segregating adult males with ventral protrusions were saved.

Full-length dpy-22 used for PCR-based rescue was amplified by PCR(Expand Long, Roche) from positions 21115 to 6068 of cosmid F47A4 with the primers 5′-gtc ccg tta tga taa cgt atc tcc aag-3′ and 5′-caa gcg tta tct tga tga cgc ggt c-3′. The PCR fragment was injected at 10 ng/μl with pPD118.33 (myo-2::gfp) (10 ng/μl) and pBSSK(Stratagene) (160 ng/μl) into dpy-6(e14) dpy-22(sy622); stDp2animals. Rescuing arrays were subsequently crossed into dpy-22(sy622)single mutants.

A full-length dpy-22 gene spanning 20759 to 6464 of cosmid F47A4 was reconstructed in pBR322 as follows. First, an XhoI fragment from F47A4 (20759-10290) was cloned into the SalI site of pBR322 to yield pBRF47A4Xh. This plasmid was digested with AgeI, and an AgeI fragment from F47A4 (10406-6464) was introduced into this vector, to generate pBRsop1FL, which harbors a full-length dpy-22 gene. dpy-22::gfp,dpy-22 2548::gfp and dpy-22 2141::gfp transgenes were constructed using overlap extension PCR (High Fidelity, Roche) to make C-terminal in-frame translational fusions between appropriately truncateddpy-22 fragments and gfp. In all cases, codons encoding two glycine residues were placed between the two genes, and the fusions were made to the codon encoding the first Ser residue of GFP. dpy-22 fragments were amplified by PCR from F47A4 using the 5′ primer sop1-17 5′-ct tat gtt cca cgg tat cat caa tcc-3′ and the 3′ primer sop1-28 5′-c ttc tcc ttt act tcc tcc gta ctg att tgg tgg ttg ttg gtt g-3′;sop1-26 5′-c ttc tcc ttt act tcc tcc ttt ctg ctg ctc cac aag ttg ttg atg g-3′; or sop1-35 5′- c ttc tcc ttt act tcc tcc gaa cat tct gaa ctt cca tcc gcc-3′, for dpy-22::gfp, dpy-22 2548::gfp anddpy-22 2141::gfp, respectively. gfp fragments included theunc-54 3′ untranslated region, and were amplified by PCR from pPD95.79 with the 5′ primers sop1-34 5′-aat cag tac gga gga agt aaa gga gaa gaa ctt ttc act gg-3′, sop1-33 5′-cag cag aaa gga gga agt aaa gga gaa gaa ctt ttc act gg-3′ or sop1-32 5′-aga atg ttc gga gga agt aaa gga gaa gaa ctt ttc act gg-3′, for dpy-22::gfp,dpy-22 2548::gfp, and dpy-22 2141::gfp, respectively, and the 3′ primer unc54-5 5′-a taa gaa tgc ggc cgc aaa cag tta tgt ttg gta tat tgg gaa tg-3′. The individual fragments were purified by agarose gel electrophoresis, mixed in the appropriate combinations, and then subjected to PCR in the presence of the 5′ primer sop1-17, and the 3′ primer unc54-5. The dpy-22::gfp fusion was digested with NotI and cloned into NotI-digested pBRsop1FL. dpy-22 2548::gfp anddpy-22 2141::gfp truncations were digested with BstEII andNotI, and cloned into BstEII/NotI-digested pBRsop1FL. dpy-22::gfp and dpy-22 2548::gfp transgenes were injected at 10 ng/μl with pPD118.33 (myo-2::gfp) (10 ng/μl) and 160 ng/μl of pBSSK into dpy-6(e14) dpy-22(bx93)/dpy-22(sy622)animals. dpy-22 2141::GFP was injected at 12.5 ng/μl with pBX-1(Granato et al., 1994) (100 ng/μl) and pBSSK (37.5 ng/μl) into pha-1(e2123ts) animals.

sy622 and sy665 were placed on LGX by crossing N2 males into mutant hermaphrodites and observing that 100% of F1 males were small/dumpy (Dpy) and had abnormal tails (Mab). Three-factor mapping was carried out following the small/Dpy and egg-laying-defective (Egl) phenotypes of sy622. sy622 was placed between lon-2 and unc-9as 6/11 Lon non-Unc and 4/11 Unc non-Lon recombinants picked up the mutation.sy622 was placed between dpy-6 and unc-9 as 6/13 Dpy non-Unc and 8/10 Unc non-Dpy recombinants picked up sy622. sy622was placed between lon-2 and egl-15 as 23/26 Lon-non-Egl and 4/33 Egl non-Lon recombinants picked up sy622. Using sy622 egl-15 double mutants and CB4856, a Hawaiian isolate of C. elegans, Egl-non sy622 recombinants were generated that allowed us to analyze the positions of crossovers by the absence or presence of single nucleotide polymorphisms (SNPs) from CB4856(Wicks et al., 2001).sy622 was mapped to the left of the SNP at 36555 in F47A4. Usingdpy-6 sy622 double mutants and CB4856, and picking Dpy non-sy622 and sy622 non-Dpy recombinants, sy622 was mapped to the right of the SNP at 19169 of F15G9. Owing to the unhealthiness of sy622 animals, cosmids from this region initially were coinjected with pPD118.33 (myo-2::gfp) and pBX-1 into pha-1(e2123ts)animals, and stable extrachromosomal arrays were crossed into sy622animals to test for rescue. Initial attempts at rescue failed using this strategy. Later, cosmid F47A4 (30 ng/μl) was co-injected with pPD118.33(myo-2::gfp)(10 ng/μl) and pBSSK (140 ng/μl) intopal-1(e2091); him-5(e1490); dpy-22(bx93) animals. Arrays conferring functional rescue in this background were crossed into sy622 animals and found to rescue all of the sy622 phenotypes. dpy-22genomic DNA was amplified in 1 kb pieces from sy622 andsy665 worms, and the products directly sequenced. Mutations were confirmed by sequencing both strands, and comparing the sequencing data from N2, sy622 and sy665 animals in the appropriate regions.

Exon 17 from dpy-22 was amplified by PCR from the cosmid F47A4 with the primers 5′-tta ata cga ctc act ata ggg aga cat tcg aac tag ctc cag aga aac-3′ and 5′-tta ata cga ctc act ata ggg aga atc aaa tgg gta ctt ccc agc ttc-3′, which introduce a T7 bacteriophage promoter at both ends of the fragment. An intronless GFP gene was amplified from the plasmid pPD79.44, with the primers 5′-tta ata cga ctc act ata ggg aga tga gta aag gag aag aac ttt tca c-3′ and 5′-tta ata cga ctc act ata ggg aga cta ttt gta tag ttc atc cat gcc atg-3′, which also add a T7 promoter to both ends of the PCR product. dsRNA was synthesized in vitro using the MEGAscript T7 kit (Ambion). The presence of dsRNA was confirmed by agarose gel electrophoresis, and quantified by spectrophotometry. L1 stage hermaphrodites were incubated in 12 μl M9 buffer containing 1.5 mg/ml total RNA and OP50 (A600 nm=1.0) for 24 hours at 20°C. After incubation, worms were recovered and placed on standard NG plates with OP50, and allowed to develop to the mid-L4 stage, at which time they were examined by Nomarski.

Vulval development was scored during the L4 stage under Nomarski optics(Sternberg and Horvitz, 1986). Nuclei in the ventral region of the worm that were not of hypodermal, neuronal or muscle descent were counted. In wild-type animals, 22 nuclei arise from vulval fates. The number of vulval nuclei is used to extrapolate how many of the Pn.p cells were induced to adopt vulval fates. A vulval precursor cell(VPC) in which both daughter cells divide one more time, and both granddaughters divide to generate seven or eight great granddaughters and no hypodermal tissue, is scored as 1.0 cell induction. A VPC in which one daughter fuses with the hypodermis, and one daughter continues to divide over the next two generations, resulting in four great granddaughter cells is scored as 0.5 cell induction. In wild-type animals, P5.p, P6.p and P7.p each undergo the equivalent of 1.0 cell induction, whereas the other Pn.p cells do not adopt vulval fates, resulting in an overall cell induction of 3.0. Animals displaying a cell induction of more than 3.0 are multivulva, and animals with a cell induction less than 3.0 are vulvaless. Laser ablations were conducted using a standard protocol (Bargmann and Avery, 1995). Gonadal cells (Z1, Z2, Z3 and Z4) were ablated during the L1 stage.

To identify new negative regulators of LET-23-dependent vulval fate specification, we employed a screen using a gain-of-function allele oflet-23, sa62. let-23(sa62) encodes a receptor with a mis-sense mutation changing C359Y in the extracellular domain(Katz et al., 1996). In wild-type hermaphrodites, only three VPCs, P5.p-P7.p, adopt vulval fates,whereas in 98% of hermaphrodites homozygous for let-23(sa62), one additional Pn.p, usually P3.p and/or P4.p, adopts a vulval fate(Table 1; seeTable 4)(Katz et al., 1996). This ectopic induction, as well as the normal induction displayed by P5.p-P7.p is independent of the LIN-3-producing anchor cell in the gonad, suggesting that when expressed at these levels, the receptor is ligand independent (seeTable 4)(Katz et al., 1996). Male gonads do not give rise to an anchor cell, and their Pn.p cells do not normally adopt vulval fates. However, despite the ligand-independence of LET-23(C359Y), this activated receptor is unable to cause male Pn.p cells to adopt vulval fates at high frequency (data not shown). We hypothesized that vulval tissue formation in homozygous let-23(sa62) males could be used as a sensitized assay to identify inhibitors of vulval fate specification. We mutagenized let-23(sa62); him-5(e1490)hermaphrodites with EMS, and looked for F1 animals that segregated a high percentage of progeny males exhibiting vulval tissue. From a pilot screen of 1000 haploid genomes, three mutations were isolated. Two alleles,sy622 and sy665, resembled each other in that mutant worms shared additional phenotypes (see below), whereas the third mutation seemed to affect only Pn.p fates.

To determine whether sy622 and sy665 affected vulval fate specification only in male Pn.p cells, we introduced these mutations into a sensitized hermaphrodite background. Hermaphrodites heterozygous forlet-23(sa62) have mostly wild-type vulvae, but occasionally some animals display ectopic vulval tissue(Table 1). sy622 andsy665 enhance the frequency of ectopic vulval fate specification inlet-23(sa62)/+ animals (Table 1), indicating that the mutations also affect the response of hermaphrodite Pn.p cells to activated LET-23, and that some regulators of LET-23 signaling are shared between male and hermaphrodite Pn.p cells.

We tested whether the multivulva-enhancing effect of sy622 was unique to let-23(sa62), or reflected an interaction with activated RAS. let-60(n1046) is a gain-of-function let-60 allele that encodes a RAS protein with a G13E change(Beitel et al., 1990). Seventy-three percent of let-60(n1046) homozygotes display ectopic vulval cell fate differentiation (Table 1). However, similar to let-23(sa62)/+ animals,let-60(n1046)/+ heterozygotes have mostly wild-type vulvae(Table 1). We therefore usedlet-60(n1046)/+ as a sensitized background to examine specifically the interaction of sy622 with activated RAS. sy622 enhanced ectopic vulval fate specification in a let-60(n1046)/+ background(Table 1), consistent with thesy622 mutation increasing the response to RAS activation.

We also tested if sy622 and sy665 could affect vulval fate specification in the central VPCs, P5.p-P7.p, which are normally specified to form vulval tissue. We found that both sy622 andsy665 could restore the ability of P6.p to adopt a vulval fate in the presence of the lin-3(n378) or let-23(sy1)reduction-of-function mutations, which reduce RAS signaling(Table 1). However, many of these double mutants still do not display wild-type vulval induction,indicating sy622 and sy665 do not bypass the requirements for RAS signaling.

In addition to enhancing vulval fate specification in the presence of activated LET-23, the sy622 and sy665 mutations cause other phenotypes. sy622 and sy665 hermaphrodites have a 90%penetrant egg-laying defect (Table 2), and, as young adults, are dumpyish and only about 70% of the size of wild-type animals (Fig. 1A-C; Table 2). In addition, 95% of sy622 and sy665 adult males have abnormal ray development in the tail (Fig. 1D-F; Table 2). Given the shortened body length of sy622 and sy665 animals,we asked whether the effect on ectopic vulval fate specification might indirectly result from a reduced distance between the anchor cell and P3.p,P4.p and P8.p. We, therefore, introduced the dpy-4(e1166) mutation,which comparably reduces body length (Table 2), into let-23(sa62)/+ animals. Unlike sy622and sy665, dpy-4(e1166) did not significantly enhance ectopic vulval fate specification in this background, indicating that the effect ofsy622 and sy665 on vulval induction is not due to a general reduction in body size (Table 1).

Based on the similarity in phenotypes between sy622 andsy665, and their common linkage to the X chromosome, we tested whether they were allelic. These recessive alleles failed to complement each other for egg-laying and body size defects, suggesting that they define a single locus (data not shown). We three-factor mapped sy622 betweendpy-6 and egl-15 on the genetic map, and using single nucleotide polymorphisms in CB4856, a Hawaiian isolate of C. elegans,we placed sy622 between the cosmids F15G9 and F47A4 on the physical map. Cosmid F47A4 rescued the egg-laying, body size and male tail defects(Table 2). A PCR fragment encompassing the complete predicted coding region of only F47A4.2,sop-1 (Zhang and Emmons,2000) (Fig. 2A),also rescued the egg-laying, body size and male tail defects(Table 2).F47A4.2/sop-1 has recently been found to be allelic to an older genetic locus, dpy-22 (Meneely and Wood, 1987) (H. Sawa, personal communication). When crossed into a let-23(sa62)/+; sy622 background, extra-chromosomal arrays containing either F47A4 or the dpy-22 PCR product suppressed the ectopic vulval fate specification observed in these animals(Table 3). These results suggest that all of the sy622 phenotypes result from mutation of a single gene, dpy-22. We then sequenced dpy-22 genomic DNA from sy622 and sy665 animals. We found that relative to wild-type animals, sy622 animals harbored a C to T mutation at position 13,100 of F47A4, which changes a CAG glutamine codon in exon 12 to an Amber STOP codon. sy665 animals harbored a C to T mutation at position 11,718 of F47A4, which changes a CAA glutamine codon in exon 13 to a TAA Ochre STOP codon. The mutant DNA in sy622 and sy665animals is predicted to truncate the DPY-22 protein after amino acids 1697 and 2141, respectively (Fig. 2B).

dpy-22 was cloned as sop-1, by virtue of the isolation of non-Dpy alleles that suppress the ray loss phenotype conferred by a regulatory region mutation in the homeobox gene pal-1(Zhang and Emmons, 2000). Thedpy-22 alleles that suppress pal-1(e2091) consist of one splicing mutation and three nonsense mutations, with the strength of the allele correlated with the extent of the predicted C-terminal truncation.dpy-22(bx93) and dpy-22(bx92) are the strongest alleles, and they are expected to make proteins that are truncated after amino acids 2548 and 3165, respectively. RNAi experiments indicate that all of thesedpy-22 alleles are reduction-of-function mutations, raising the possibility that dpy-22(sy622) and dpy-22(sy665) are more severe loss-of-function alleles, because they would cause even earlier truncations than would dpy-22(bx93). Consistent with this notion,recent RNAi experiments against dpy-22, and double mutant analyses using the dpy-22 alleles that suppress pal-1(e2091) and either unc-37 or sur-2 also have revealed a ray loss phenotype (Zhang and Emmons,2002), similar to that observed in dpy-22(sy622) anddpy-22(sy665) single mutants (Fig. 1). This observation suggests that ray development is ultimately compromised when DPY-22 pathway activity is reduced below a certain threshold. In addition, we find that the loss-of-function allele dpy-22(bx93)weakly promotes ectopic vulval fate transformations in alet-23(sa62)/+ background (Table 3). Moreover, dpy-22(bx93) fails to complementdpy-22(sy622) for enhancing ectopic vulval fate specification in the presence of let-23(sa62)/+ (Table 3), providing further evidence that DPY-22 negatively regulates vulval development.

To examine directly the effects of reducing DPY-22 levels on vulval development, we used RNAi against dpy-22. In an otherwise wild-type background, dpy-22 RNAi did not affect vulval development (data not shown). However, in a sensitized background consisting of thelet-23(sy1) loss-of-function mutation, dpy-22 RNAi could partially suppress the vulvaless phenotype of let-23(sy1), similar to the dpy-22 alleles sy622 and sy665(Table 3). By contrast, controlgfp dsRNA did not suppress let-23(sy1)(Table 3). Together, these data indicate that sy622 and sy665 are stronger loss-of-function alleles of dpy-22.

DPY-22 is most closely related to human TRAP230(Ito et al., 1999;Nagase et al., 1996;Philibert et al., 1998), a component of human mediator complexes, and KOHTALO(Treisman, 2001), a regulator of cell fate in the Drosophila eye. The fly and human proteins have similar length, and display 35% overall identity. DPY-22 is larger than both proteins by ∼1000 amino acids, and BLAST analysis indicates that its homology with TRAP230 and KOHTALO is spread over three regions(Fig. 2B). These include small regions of identity at the N and C termini, and a larger region of identity in the middle of the protein. In all three proteins, the C terminus is rich in glutamine. Thirty-three percent of the terminal 781 and 589 amino acids are glutamine in DPY-22 and KOHTALO, respectively, and 41% of the 278 terminal amino acids in TRAP230 are glutamine. Of the previously isolated nonsense mutations in dpy-22, bx93 and bx92 are the strongest.bx93 results in the greatest truncation, removing all the amino acids C-terminal to 2548, including the entire glutamine-rich region. Asdpy-22(sy622) and dpy-22(sy665) cause a number of highly penetrant phenotypes (Fig. 1;Tables 1,2) not observed in the other nonsense mutations, domains N-terminal to the glutamine-rich region must be important for DPY-22 function. The nonsense mutation in dpy-22(sy622)would be predicted to truncate DPY-22 in the largest central region of identity; however, dpy-22(sy665) is predicted to truncate DPY-22 after amino acid 2141, close to the end of this region(Fig. 2B). This observation suggests that another important functional domain exists between amino acids 2141 and 2548 that does not have identity with TRAP230 or KOHTALO.

PROSITE analysis of DPY-22 indicated the presence of two putative nuclear localization signals in this region. To test whether the stronger phenotypes observed in dpy-22(sy622) and dpy-22(sy665) animals are correlated with a mislocalized, non-nuclear form of DPY-22, we constructed transgenes in which dpy-22 DNA was fused in-frame to the green fluorescent protein (GFP) open reading frame(Fig. 2C). dpy-22 has been reported to be expressed in vulval cells, during the later stages of vulval development (Zhang and Emmons,2000). We found that a rescuing (see last section of the Results)fusion of wild-type DPY-22 to GFP was expressed in the VPCs, the anchor cell,and hyp7 nuclei at the time of vulval fate specification, and continued to be expressed in these cells throughout vulval development(Fig. 3A-F). Wild-type DPY-22 and DPY-22 truncated after amino acid 2548, as generated indpy-22(bx93) mutants, directed nuclear expression of GFP(Fig. 3A-H). By contrast,DPY-22 truncated after amino acid 2141 could not direct GFP into the nucleus,and instead, caused GFP to accumulate in the cytoplasm(Fig. 3I,J). These data indicate that the predicted nuclear localization sequences (NLSs) in this region are functional, and suggest that the severity of thedpy-22(sy622) and dpy-22(sy665) phenotypes may result from mislocalization of the DPY-22 protein, and/or loss of other functional regions of the protein.

let-23(sa62)/+ and let-60(n1046) animals are responsive to LIN-3 (Chang et al., 2000;Katz et al., 1996;Sundaram and Han, 1995)(Table 3), and suppression of the lin-3(n378) and let-23(sy1) hypomorphic mutations could in principle occur through elevated production of LIN-3. As DPY-22::GFP is expressed in the anchor cell and the VPCs(Fig. 3), we tested whetherdpy-22 affects vulval fate specification by modulating the production of LIN-3 from the gonad. We ablated the gonadal primordium inlet-23(sa62)/+ and let-23(sa62)/+; dpy-22(sy622)early L1 larvae. Gonad-ablated let-23(sa62)/+ animals displayed very little vulval fate specification in any of the Pn.p cells(Table 4). By contrast,dpy-22(sy622) increased the frequency of vulval fate specification in all six VPCs in this background (Table 4), indicating that DPY-22 does not act primarily by regulating LIN-3 production from the gonad, and that the fates of all the VPCs are regulated by DPY-22.

In males, a bar-1(β-catenin) null allele, ga80, does not affect ray development on its own(Zhang and Emmons, 2000). However, it reduces the degree to which dpy-22(bx92) can restore wild-type ray development in a pal-1(e2091) mutant background(Zhang and Emmons, 2000). This property has led to the proposal that in the presence of thepal-1(e2091) regulatory region mutation, DPY-22 inhibits BAR-1-dependent regulation of the pal-1 gene(Zhang and Emmons, 2000). WNT/β-catenin signaling also promotes vulval fate specification, parallel to the LET-23-LET-60 pathway. A null mutation in bar-1 results in an incompletely penetrant vulvaless phenotype(Table 5)(Eisenmann et al., 1998), and a loss-of-function mutation in the axin-like inhibitor of WNT signaling,pry-1 (Korswagen et al.,2002; Maloof et al.,1999), causes ectopic vulval fate specification(Table 5)(Gleason et al., 2002). Furthermore, hyperactivated WNT signaling can bypass loss-of-function mutations in the let-23 pathway(Gleason et al., 2002). We therefore tested whether relief of inhibition of BAR-1 might account for the interactions between our dpy-22 alleles and mutations in the LET-23 pathway during vulval development.

We linked two bar-1 mutations to dpy-22(sy622). bar-1(mu63) encodes a weak loss-of-function mutation(Maloof et al., 1999) that does not confer defects in vulval development on its own(Table 5). The mis-sense mutation in this allele results in a protein with a G524D change (see Materials and Methods). We hypothesized that under conditions where a phenotype was dependent on elevated BAR-1 activity, bar-1(mu63)should quantitatively reduce it. Consistent with this notion,bar-1(mu63) suppresses the ectopic mab-5 expression and poly ray phenotypes observed in pry-1(mu38) animals(Maloof et al., 1999). As a control, we tested whether bar-1(mu63) also could reduce vulval induction in a background where WNT signaling was specifically hyperactivated. We built a pry-1(mu38); let-23(sy1) double mutant and, as recently reported (Gleason et al.,2002), found that excess WNT signaling could strongly suppress this let-23 reduction-of-function allele(Table 5). bar-1(mu63)reduced the ability of pry-1(mu38) to suppress let-23(sy1)(Table 5), indicating thatbar-1(mu63) can be used to define conditions under which WNT signaling is hyperactivated during vulval development. Althoughdpy-22(sy622) is a weaker suppressor of let-23(sy1) than ispry-1(mu38), its ability to suppress the let-23 mutation was not reduced in the presence of the bar-1(mu63) mutation(Table 5). These data suggest that DPY-22 does not interfere with vulval development through inhibition of BAR-1. We also examined the body size, egg-laying behavior and male rays inbar-1(mu63) dpy-22(sy622) animals, and found none ofdpy-22(sy622) phenotypes to be suppressed by the bar-1mutation (Table 2).

As our data suggested that DPY-22 might primarily inhibit the output of the RAS pathway, rather than the WNT pathway, during vulval development, we tested whether the dpy-22(sy622) mutation was strong enough to compensate for the absence of BAR-1 in bar-1(ga80) null mutants. bar-1(ga80)dpy-22(sy622) double mutants were sick, but still displayed thebar-1(ga80) vulvaless phenotype(Table 5). Because the gain-of-function let-60(n1046) allele, which is far stronger thandpy-22(sy622) [(Table 1) 73% Muv and 4.0 cell induction versus 3% Muv and 3.0 cell induction, respectively] only suppresses bar-1(ga80) to wild-type levels, we hypothesized that failure of dpy-22(sy622) to suppressbar-1(ga80) might result from insufficient activation of the RAS pathway. Therefore, we used a more sensitive assay to test further whether DPY-22 acts independently of BAR-1 during vulval development.let-60(n1046gf); bar-1(ga80) double mutants are mostly wild type and are mutually suppressed with regards to the let-60(n1046) multivulva phenotype and the bar-1(ga80) vulvaless phenotype(Table 5). Asdpy-22(sy622) is a strong enhancer of the gain-of-functionlet-60(n1046) allele (Table 1), we asked whether dpy-22(sy622) could still enhance activated RAS even in the absence of BAR-1 protein. let-60(n1046);bar-1(ga80) dpy-22(sy622) triple mutants were very sick, but survived to L4 and displayed much more vulval induction than did let-60(n1046);bar-1(ga80) double mutants (Table 5). The enhanced vulval induction in the triple mutants was abrogated by the presence of the dpy-22-rescuing cosmid, F47A4,demonstrating that enhancement in the triple mutant was indeed due to loss of DPY-22 activity (Table 5). These data indicate that DPY-22 can regulate vulval induction independently of any effects on BAR-1 activity.

The absence of body size, egg-laying, and ray defects, and the presence of only a weak vulval phenotype in dpy-22(bx93) mutants, suggest that the glutamine-rich region is dispensable for the function of DPY-22 in these processes. Because it is possible that small amounts of wild-type DPY-22 might be produced through translational readthrough in dpy-22(bx93)mutants, we directly examined the functional properties of a DPY-22 protein that completely lacks the glutamine-rich region. We injected a construct consisting of dpy-22 DNA ending with the codon for amino acid 2548,fused in-frame to the GFP open reading frame, into dpy-22(sy622)animals (Fig. 2C). This transgene should produce a protein similar to that made indpy-22(bx93) animals, which ends after amino acid 2548, before the beginning of the glutamine-rich region. Wild-type and glutamine-deleted DPY-22 fully rescued the body length, egg-laying and male tail defects observed indpy-22(sy622) mutants (Fig. 1G,H; Table 2). Furthermore, glutamine-deleted DPY-22 also rescued the ectopic vulval fate specification observed in a let-23(sa62)/+; dpy-22(sy622)background, comparable with wild-type DPY-22(Table 3). These results demonstrate that the glutamine-rich region is not essential for the majority of functions performed by DPY-22.

We have used a genetic approach to study mechanisms by which a diffusible growth factor can invariantly trigger a response in a particular cell within a field of competent cells. Initially using vulval tissue formation by male Pn.p cells as an assay, we identified two mutations in dpy-22 that generally promote the activity of LET-23 in Pn.p cells, regardless of the sex of the animal. Our genetic analysis using the dpy-22(sy622),dpy-22(sy665) and dpy-22(bx93) alleles, as well as RNAi againstdpy-22, indicate that DPY-22 is normally an inhibitor of vulval fate specification (Tables 1,3). As DPY-22 can act as an inhibitor in the absence of the LIN-3-producing anchor cell(Table 4), and DPY-22 is expressed in the vulval precursor cells during the time of vulval fate specification (Fig. 3B), we favor the idea that part of its function is executed in the VPCs. The ability of our dpy-22 mutations to promote vulval fate specification in P5.p-P7.p, under conditions where RAS signaling is reduced (Tables1,4), indicates that DPY-22 acts as an inhibitor in all the VPCs, not just the VPCs that never adopt vulval fates (P3.p, P4.p and P8.p). The dpy-22(sy622) anddpy-22(sy665) mutations do not bypass the RAS pathway as they cannot fully suppress lin-3 and let-23 reduction-of-function mutations (Table 1). This suggests that DPY-22 acts directly on the RAS pathway or a pathway that cooperates with the RAS pathway and converges on a common target(s).

Besides the RAS pathway, LIN-12/NOTCH and WNT promote vulval fates. However, LIN-12 and RAS antagonize each other with respect to the type of vulval fate that is induced (Berset et al.,2001; Wang and Sternberg,1999). High levels of LIN-12 signaling promote secondary vulval fates (Sternberg and Horvitz,1989), and high levels of RAS signaling promote primary vulval fates (Katz et al., 1995). In our double mutants consisting of dpy-22 and let-23 pathway mutations, P6.p induction is restored(Table 1), and all animals displaying full vulval induction have correctly patterned vulvae consisting of 2°-1°-2° fates for P5.p-P7.p. These data indicate thatdpy-22 mutations cooperate with RAS to induce the primary fate in P6.p, rather than antagonize it, and suggest that DPY-22 does not act primarily by inhibiting the LIN-12 pathway.

DPY-22 has previously been described as an inhibitor of BAR-1-dependent WNT signaling (Zhang and Emmons,2000). In contrast to the LIN-12-RAS relationship, RAS and WNT signaling are known to converge on at least one common target, necessary for vulval development, lin-39(Eisenmann et al., 1998;Maloof and Kenyon, 1998). Thus, DPY-22 inhibition likely occurs through interference with RAS and/or WNT signaling. However, our triple mutant analysis with a bar-1 null mutation indicates that DPY-22 can inhibit activated RAS even in the absence of BAR-1 protein (Table 5). Together with our molecular data that the glutamine-rich region, which is necessary for inhibition of BAR-1-dependent ray development(Zhang and Emmons, 2000), is dispensable for inhibition of vulval development(Table 3), we propose that DPY-22 largely acts on the RAS pathway, rather than the WNT pathway, to inhibit vulval development. The strong gain-of-function let-60(n1046)allele only suppresses the bar-1(ga80) null mutation to wild type,and is itself suppressed for its multivulva phenotype by thebar-1(ga80) mutation (Table 5). Furthermore, our dpy-22(sy622) mutation which does not cause a strong multivulva phenotype, partially suppresses the moderate vulvaless phenotype of let-23 pathway mutations(Table 1), but not the weak vulvaless phenotype of bar-1(ga80)(Table 5). These results suggest that shared targets between RAS and WNT signaling are unequally activated by the two pathways. In addition, it is likely that the WNT pathway has targets that are not shared with the RAS pathway.

Most inhibitors of vulval development affect all six VPCs. This observation is consistent with a model in which negative regulators of the RAS pathway raise the requirement for the amount of pathway activation needed to generate a biological response. Although the LIN-3 growth factor might diffuse outside of the source anchor cell, only P6.p, because of its proximity to the anchor cell, achieves sufficient pathway activation to adopt a primary vulval fate. A surprising number of points in the pathway must be negatively regulated,presumably to achieve invariant positional specificity for the response to LIN-3. SLI-1(c-CBL) (Jongeward et al.,1995; Yoon et al.,1995), GAP-1 (Hajnal et al.,1997) and LIP-1 (MAPK phosphatase)(Berset et al., 2001) appear to directly regulate LET-23 (EGFR), LET-60 (RAS) and SUR-1/MPK-1(MAP kinase),respectively, while ARK-1 might control some aspect of SEM-5(GRB2)-dependent RAS activation (Hopper et al.,2000).

Ultimately, the end point for growth factor signaling can be considered to be a change in RNA Pol II activity on specific promoters. Thus, in principle,sequence-specific and global regulators of transcription also might play important roles in regulating the output of the RAS pathway. Genetic studies have identified LIN-1, an Ets-domain sequence-specific DNA-binding protein as an inhibitor of vulval fate specification(Beitel et al., 1995). Global regulators of transcription broadly include chromatin remodeling proteins, RNA Pol II and the general transcription factors, and components of the mediator complex, which promote the activity of sequence-specific activators and repressors. Mutations in the synmuv genes lin-35(RB) andlin-53(RBAP48) have suggested that a histone deacetylase complex is inhibitory towards vulval fate specification(Lu and Horvitz, 1998). In support of this, experiments that directly examine HDAC-1 and components of the NURD complex have provided some evidence for their inhibitory roles(Chen and Han, 2001;Lu and Horvitz, 1998;Solari and Ahringer, 2000). Our work with dpy-22, which is most related to humanTRAP230(Ito et al., 1999;Nagase et al., 1996;Philibert et al., 1998), now indicates that a specific component of the mediator also can act as an inhibitor of a RAS-dependent response.

At least 20 components of the mediator complex have been identified in yeast. Ten are essential for viability and seven of these appear to have related counterparts in higher species including C. elegans (for reviews, see Boube et al.,2002; Gustafsson and Samuelsson, 2001; Myers and Kornberg, 2000; Woychik and Hampsey, 2002). By contrast, the remaining yeast mediator components do not have clear orthologs in higher species. This has led to the proposal that these components and their counterparts in other species might not peform core mediator functions but, instead, specifically integrate information from rapidly evolving transcription factor families. Consistent with this model, mutation or RNA interference against the conserved components in C. elegans result in early lethality, or multiple phenotypes in the rare animals escaping lethality (Kwon et al., 1999; Kwon et al.,2001; Kwon and Lee,2001). By contrast, the metazoan-specific mediator, sur-2(Boyer et al., 1999;Singh and Han, 1995), is not essential for viability, and single mutants have a strong vulvaless phenotype,resembling a RAS pathway mutation (Singh and Han, 1995). Similarly, prior to the discovery thatsop-1 was allelic to dpy-22, it was thought that mutations in C. elegans TRAP230, another metazoan-specific mediator component,do not affect viability and cause one major phenotype, which is the relief of inhibition on BAR-1-dependent regulation of pal-1 expression in males(Zhang and Emmons, 2000).

However, our alleles of dpy-22, which cause more severe truncations of the protein and mislocalize it(Fig. 3I), result in multiple developmental and behavioral phenotypes(Fig. 1,Table 2) that do not appear to be restricted to one specific signaling pathway. Some of these include an overall reduction in body length by 30%, a Dpy appearance, partial sterility(data not shown), abnormal ray development in the male tail and an egg-laying defect that does not result from the absence of vulval tissue or sex muscles,or a functional uterine-vulval connection (data not shown). We have also found that DPY-22 antagonizes vulval fates specified by activated RAS, and that the major target(s) of this inhibition is distinct from BAR-1(Table 5). These findings are consistent with the original description of dpy-22(e652) phenotypes,and the observation that dpy-22(e652) enhances the multivulva phenotypes of lin-53(n833); lin-15(n767) and lin-15(n765)males (Meneely and Wood,1987). Because we observe effects of dpy-22 mutations on vulval development with multiple non-X-linked alleles in the RAS pathway, we do not think that DPY-22 acts on vulval development through regulation of dosage compensation from the X-chromosome, as originally proposed(Meneely and Wood, 1987). Neither loss-of-function nor gain-of-function mutants in the RAS pathway result in the body size or ray defects observed in our mutants, andbar-1(mu63) and bar-1(ga80) do not suppress at least the body size phenotype of our dpy-22 mutants(Table 2, data not shown). These results suggest that transcription factors downstream of other signaling pathway also must be regulated by DPY-22. Similarly, recent data withdpy-22(bx92) and sur-2 double mutants have shown a synthetic loss of rays (Zhang and Emmons,2002), indicating SUR-2 also can function in pathways distinct from those using RAS and BAR-1. Thus, it is likely that DPY-22 and SUR-2 relay information from multiple activator/promoter contexts to RNA Pol II, but how they do it, and their relative contribution at a given promoter may be different.

The molecular identification of our new dpy-22 alleles provides some insight into how a mediator component such as DPY-22 can have general and specific functions, depending on the promoter context. Althoughdpy-22(bx93), which deletes the entire glutamine-rich domain of DPY-22, has a weak ability to promote ectopic vulval fate specification in the presence of activated LET-23, it has a profound ability to alleviate inhibition on BAR-1-dependent regulation of the pal-1 gene(Zhang and Emmons, 2000). Furthermore, whereas a transgene harboring nonsense mutations affecting the glutamine-rich domain cannot restore inhibition to the pal-1 gene as well as a wild-type dpy-22 transgene(Zhang and Emmons, 2000), we find that a transgene completely lacking the glutamine-rich domain has the same activity as a wild-type transgene in rescuing vulval fate specification(Table 3), ray development(Fig. 1), body size and egg-laying defects (Table 2). The glutamine-rich region thus defines a domain that probably distinguishes promoter/activator-dependent activity for DPY-22. Inhibition of BAR-1-dependent pal-1 expression is critically dependent on this domain, but not the activity of DPY-22 in the other pathways studied in this work. BLAST analysis (Boube et al.,2002; Gustafsson and Samuelsson, 2001) indicates the N-terminal region of TRAP230 is similar to both DPY-22 and SRB8(Hengartner et al., 1995), a yeast mediator component without a clear ortholog in metazoans. SRB8 is responsible for recruiting (Myer and Young, 1998) the SRB10/SRB11 cyclin/cdk complex(Liao et al., 1995) that represses Pol II-dependent transcription on many promoters(Holstege et al., 1998). In cases where DPY-22/TRAP230 represses transcription, this function might be dependent on the SRB8-like domain, but modulated by the metazoan-specific domains. TRAP230 and PCQAP (Berti et al.,2001), two glutamine-rich components of human mediator complexes have been implicated in human disease. TRAP230 has been reported to undergo a 12 bp expansion in a region encoding the glutamine-rich domain in a subset of mentally retarded individuals (Philibert et al., 1998), and PCQAP is in the 22q11 deletion associated with DiGeorge syndrome (Berti et al.,2001). If TRAP230 and PCQAP contribute to these conditions, one would expect them to arise out of loss of specific and general functions,respectively.

Although compared with sequence-specific DNA-binding proteins, proteins involved in nucleosome displacement, histone acetylation and deacetylation,and components of the mediator complex appear to be global regulators of transcription, these global regulators may not act equally at all promoters. In particular, multiple chromatin remodeling and histone acetylation and deacetylation complexes with distinct subunit composition, substrate specificity and activator preferences have been described (reviewed byEmerson, 2002;Narlikar et al., 2002). Although it is still controversial whether distinct mediator complexes are associated with the RNA Pol II holoenzyme at a given promoter, mutations in different components have different effects on gene expression in yeast(Holstege et al., 1998;Myers et al., 1999). Some mediator components may perform unique functions for distinct activators or repressors, while others may be used to varying degrees or redundantly with other components. Finally, as in the case of DPY-22, a single component may use distinct domains to regulate transcription on different promoters. Genetic studies thus demonstrate that a specific component of the mediator provides an important point for regulating the output of the RAS pathway, and that specific mutations can be generated with distinct properties, which might contribute to disease in humans.

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 S. Emmons and C. Kenyon for providing strains, and A. Fire for plasmids. We thank J. Maloof and C. Kenyon for help in identifying the bar-1(mu63) mutation; H. Sawa for communicating unpublished results; and L. R. Garcia, R. Lee, B. P. Gupta, H. Yu, G. Schindelman and C. Van Buskirk for helpful discussions. This research was supported by the Howard Hughes Medical Institute, of which P. W. S. is an investigator. N. M. was supported by postdoctoral fellowships from the Leukemia and Lymphoma Society,and the California Breast Cancer Research Program.