Genetic Analysis of the Caenorhabditis elegans MAP Kinase Gene mpk-1

Corresponding author: Stuart K. Kim, Department of Developmental Biology, Stanford University School of Medicine, Stanford, CA 94305-5329., kim{at}cmgm.stanford.edu (E-mail).

Communicating editor: R. K. HERMAN

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The Caenorhabditis elegans mpk-1 gene encodes a MAP kinase protein that plays an important role in Ras-mediated induction of vulval cell fates. We show that mutations that eliminate mpk-1 activity result in a highly penetrant, vulvaless phenotype. A double mutant containing a gain-of-function mpk-1 mutation and a gain-of-function mek mutation (MEK phosphorylates and activates MPK-1) exhibits a multivulva phenotype. These results suggest that mpk-1 may transduce most or all of the anchor cell signal. Epistasis analysis suggests that mpk-1 acts downstream of mek-2 (encodes a MEK homolog) and upstream of lin-1 (encodes an Ets transcription factor) in the anchor cell signaling pathway. Finally, mpk-1 may act together with let-60 ras in multiple developmental processes, as mpk-1 mutants exhibit nearly the same range of developmental phenotypes as let-60 ras mutants.

MAP kinases are key mediators of cellular differentiation and proliferation in all animals, and they function in receptor tyrosine kinase/Ras signaling pathways (reviewed in MARSHALL 1994 ). MAP kinase plays an important role in the Ras signaling pathway because it can activate downstream substrates that directly mediate the cellular response to growth factors, suggesting that MAP kinase acts near or at the end of this signaling pathway (reviewed in TREISMAN 1996 ).

MAP kinases are activated when they become phosphorylated by the protein kinase MEK (MAP or ERK kinase; ADAMS and PARKER 1992 ; CREWS et al. 1992B ). The major known substrate for MEK is currently MAP kinase, suggesting that the predominant function of MEK may be to activate MAP kinase (SEGER et al. 1992 ). Once activated, a significant fraction of MAP kinase molecules translocate to the nucleus, and many important MAP kinase substrates are localized in the nucleus (e.g., the mammalian transcription factors Elk-1 and Ets-1 (reviewed in TREISMAN 1996 ). In addition to nuclear targets, some MAP kinase substrates are located in the cytoplasm, e.g., the guanine nucleotide exchange factor SOS and the protein kinase pp90Rsk (reviewed in FERRELL 1996 ).

Vertebrates have two MAP kinases (Erk1 and Erk2) that have overlapping substrates (reviewed in CREWS et al. 1992A and MARSHALL 1994 ). Saccharomyces cerevisiae has two MAP kinase genes, fus3 and kss1, that act in the pheromone and invasive growth signaling pathways, respectively (HERSKOWITZ 1995 ; MADHANI et al. 1997 ). In Drosophila, there is currently a single known MAP kinase, which is encoded by the rolled gene (BIGGS et al. 1994 ). A mutation (termed Sevenmaker) that activates rolled MAP kinase results in the formation of extra R7 cells, similar to the phenotype caused by constitutively activated Ras1 (BRUNNER et al. 1994 ).

In Caenorhabditis elegans, there is a single known MAP kinase that is encoded by the mpk-1/sur-1 gene (LACKNER et al. 1994 ; WU and HAN 1994 ). Previous studies have shown that mpk-1 plays a role in vulval induction in the C. elegans hermaphrodite. The vulva is induced from a set of six ectodermal blast cells, referred to as Pn.p cells (P3.p to P8.p), that lie in a row along the ventral midline of the developing larva (reviewed in EISENMANN and KIM 1994 ). All six Pn.p cells are capable of adopting any one of three potential cell fates (1°, 2°, or 3°), and these fates are predominantly determined by the sequential action of two signaling pathways. First, a signal from the anchor cell induces the nearest Pn.p cell (P6.p) to adopt the 1° cell fate, which is to divide three times and generate eight descendant cells that form the inner portion of the vulva. Then, a lateral signal from P6.p induces P5.p and P7.p to adopt the 2° cell fate (STERNBERG 1988 ; KOGA and OHSHIMA 1995B ; SIMSKE and KIM 1995 ), which is to divide three times to give seven descendant cells that form the outer portions of the vulva. In addition to the lateral signal, some experiments suggest that a graded signal from the anchor cell can sometimes induce Pn.p cells to adopt the 2° cell fate (STERNBERG and HORVITZ 1986 ; KATZ et al. 1995 ). The Pn.p cells furthest from the anchor cell (P3.p, P4.p, and P8.p) adopt uninduced (3°) cell fates, which is to divide once and then fuse with the hypodermal syncytium.

Genetic studies have shown that the anchor cell signal activates a receptor tyrosine kinase/Ras pathway (reviewed in KORNFELD 1997 ). The lin-3 gene encodes a protein similar to epidermal growth factor (EGF) that is the inductive signal secreted by the anchor cell (HILL and STERNBERG 1992 ). The let-23 gene encodes a receptor tyrosine kinase similar to EGF receptor that is thought to be the receptor for LIN-3 signal (AROIAN et al. 1990 ). Downstream of LIN-3 and LET-23 are a Grb2-related adapter protein (encoded by sem-5; CLARK et al. 1992 ), a Ras protein (encoded by let-60; HAN and STERNBERG 1990 ), a Raf homolog (encoded by lin-45; HAN et al. 1993 ), and a protein similar to the serine/threonine protein kinase MEK (encoded by mek-2; KORNFELD et al. 1995 ; WU et al. 1995 ). Loss-of-function mutations in any of these genes prevent P6.p from expressing the 1° cell fate and the lateral signal, so that all of the Pn.p cells express 3° cell fates, resulting in a vulvaless phenotype. Conversely, gain-of-function mutations in let-23 or let-60 cause the vulval precursor cells to adopt the 1° cell fate and to presumably express the lateral signal, resulting in a multivulva phenotype.

Previous studies have implicated mpk-1 in vulval induction. These studies showed that two mpk-1 alleles (n2521 and ku8) suppress the multivulva phenotype of the activated ras allele let-60(n1046) (LACKNER et al. 1994 ; WU and HAN 1994 ). This result indicates that mpk-1 most likely acts downstream of let-60 ras in the vulval signaling pathway. However, mpk-1(n2521) mutants did not exhibit any defect in vulval induction, and mpk-1(ku8) exhibited only a weak defect in vulval induction. Thus, these mpk-1 alleles do not result in a strong reduction of the anchor cell signaling pathway in P6.p. It is not clear whether these mpk-1 alleles eliminate or might only partially reduce mpk-1 activity, since both alleles are missense mutations and gene dosage studies have not been possible because of the lack of a deficiency that deletes the mpk-1 locus.

Because the mpk-1–null phenotype has not been established, an important but unanswered question is whether the vulval induction pathway is linear between let-23 RTK and mpk-1, such that mpk-1 transduces all of the signal from activated let-23. If the pathway is linear, then null mutations in mpk-1 should prevent vulval induction and result in a Vul phenotype. Alternatively, the signaling pathway might be branched in a way that other signaling molecules might act in parallel with mpk-1. For instance, mammalian EGF receptor and Ras are thought to transduce signals by activating targets in completely separate signaling branches (such as phospholipase C gamma or PI3 kinase; ROCHE et al. 1996 ) or by activating two MAP kinases (Erk1 and Erk2). If the C. elegans vulval signaling pathway is similarly branched, then mpk-1 null mutations should result in only a partial reduction in vulval induction because the mpk-1 mutations should eliminate one but not all of the signaling branches.

In this article, we have further characterized the role of mpk-1 in let-60 ras signaling pathways during C. elegans development. First, we present evidence suggesting that mpk-1 transduces most if not all of the signal from let-23 receptor and let-60 ras. We used three genetic screens to recover new mpk-1 alleles (including three putative null alleles) and found that mpk-1–null mutants exhibit a strong vulvaless phenotype. We also showed that activation of both MEK and MAP kinase activates the vulval signaling pathway, resulting in a multivulva phenotype. Thus, mpk-1 is necessary for vulval induction, and mek and mpk-1 together are sufficient to activate vulval induction. Second, genetic and molecular analyses suggest that mpk-1 acts downstream of all the signal transduction genes in the let-23 RTK vulval signaling pathway, but upstream of lin-1 Ets (a vulval repressor gene encoding a putative transcription factor). Finally, our results suggest that mpk-1 may act together with let-60 in many, but perhaps not all, of the tissues whose fates are determined by let-60 ras signaling.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

General methods:
Strain maintenance, genetic manipulation, and ethyl methanesulfonate (EMS) mutagenesis were performed as described by BRENNER 1974 . Animals were cultivated at 20° unless otherwise noted. The animals described as wild type were C. elegans, variety Bristol, strain N2. The following genes and alleles were used in this work [unless otherwise noted, alleles used are described by WOOD 1988 ]:

LGI: mek-2(n2678) (KORNFELD et al. 1995 ), sup-11(n403), dpy-5(e61), sur-2(ku9) (SINGH and HAN 1996 ), smg-1(r861) (PULAK and ANDERSON 1993 ), unc-54(r293) (PULAK and ANDERSON 1993 ).

LGIII: unc-79(e1068), mpk-1(ku8) (WU and HAN 1994 ), mpk-1(oz140) (CHURCH et al. 1995 ), dpy-17(e164), lin-12(n137sd), sDp3 (ROSENBLUTH et al. 1985 ), unc-119(ed3op) (MADURO and PILGRIM 1995 ).

LGIV: lin-1(sy254) (BEITEL et al. 1995 ), lin-45(sy96) (HAN et al. 1993 ), let-60(n1046gf) (BEITEL et al. 1990 ), dpy-20(e1282ts), unc-30(e191).

LGV: him-5(e1490), lin-25(e1446) (TUCK and GREENWALD 1995 ).

LGX: lin-15(n309), lin-3(syIs1) (gift from R. HILL, Cal Tech), unc-7(e5).

Transgenic arrays (transgenes; cotransformation markers): gaEx36 {hs-mpk-1(gf); hs-D-mek(gf); rol-6(d)}, gaEx72 {lin-31-mpk-1(gf); lin-31-D-mek(gf); unc-119(+)}, gaIs17{hs-mpk-1(gf); hs-D-mek(gf); unc-30(+)}; gaIs36 {hs-mpk-1(+); EF1alpha-D-mek(gf); unc-30(+)}.

Genetic mapping:
The oz140 allele was initially mapped to the left arm of chromosome III (T. SCHEDL, personal communication). To determine if it mapped to the same interval as the mpk-1 locus, we mapped it relative to unc-79 and dpy-17. Briefly, we picked Dpy non-Unc and Unc non-Dpy recombinants from a strain of the genotype oz140/unc-79(e1068) dpy-17(e164). The progeny of these recombinant animals were screened for sterile (Ste) animals (the oz140 mutant phenotype). Eighteen out of 19 Dpy non-Unc animals segregated Ste animals, whereas 1 out of 17 Unc non-Dpy animals segregated Ste animals. These data suggest that oz140 maps to the left of dpy-17 and very close or to the right of unc-79. This is the approximate map position of the mpk-1 locus. The SD344 strain [+ mpk-1(oz140) dpy-17(e164)/unc-79(e1068) + dpy-17 (e164)] was constructed by picking Dpy non-Unc recombinant progeny of hermaphrodites of genotype mpk-1(oz140)/unc-79(e1068) dpy-17(e164). All other newly isolated mpk-1 alleles were shown to be linked to unc-79 and dpy-17 by allowing hermaphrodites of genotype mpk-1(rf)/unc-79 dpy-17 to self, picking 20 Unc Dpy progeny and confirming that these hermaphrodites did not segregate sterile mpk-1(rf) animals (this experiment was performed at 25° for ga111).

Three mutations isolated in a clonal screen for sterile and vulvaless animals were mapped using STS polymorphisms (data not shown; WILLIAMS et al. 1992 ). STS mapping and two-factor mapping with dpy-5 and sup-11 indicated that ga180 and ga181 are linked to the left arm of chromosome I, close to mek-2. We next showed that both ga180 and ga181 fail to complement mek-2(n2678) for the Ste phenotype. STS mapping suggested that ga119 is linked to chromosome III. Next, we showed that ga119 fails to complement mpk-1(oz140) for the sterile phenotype.

Strain construction:
Double mutants containing mpk-1(ga117) and a Muv mutation (see Table 6) were all constructed using a common strategy, which is illustrated here for lin-15(n309). First, dpy-17(e164); lin-15(n309) was built. These dpy-17(e164); lin-15(n309) hermaphrodites were then mated with males of genotype mpk-1(ga117) dpy-17(e164); sDp3. sDp3 is a chromosomal duplication that complements mutations in mpk-1 and dpy-17. Non-Dpy cross-progeny of genotype ga117 e164/+ e164; sDp3; n309/+ were then allowed to self-fertilize. Muv non-Dpy progeny of this cross were then placed on individual plates and allowed to self-fertilize. Muv non-Dpy animals in which all the Dpy progeny were Ste were taken to be of genotype mpk-1(ga117) dpy-17(e164); lin-15(n309); sDp3.

The lin-45; gaEx36 strain was constructed by mating lin-45/+ males with gaEx36 (Rol) hermaphrodites. Cross-progeny hermaphrodites were cloned, and their self-progeny were screened for Vul Rol animals. Vul Rol animals were assumed to be of genotype lin-45; gaEx36. lin-25; gaEx36 and sur-2; gaEx36 strains were constructed similarly.

Gonad ablations:
Somatic gonad precursors Z1.aaa and Z4.ppp were ablated with a laser microbeam (Laser Science, Newton, MA) as described in AVERY and HORVITZ 1989 . Animals were then subjected to a 30-min heat shock of 33° as early L3 larvae and then scored as Muv or non-Muv.

Mutant sequence determination:
We used oligonucleotide primers (mpk-1f and mpk-12r in Table 1) in polymerase chain reactions to amplify a 3.2-kb genomic DNA fragment containing the entire mpk-1-coding region for the SLX2 form from wild-type strains and each mutant strain. The DNA sequence of the mpk-1 gene (exons 1–6 and the exon/intron borders) from each mutant was then determined by direct sequencing of this fragment in low-melting-temperature agarose (KRETZ et al. 1989 ). DNA from two separate single-worm PCR reactions was sequenced using the following primers: mpk-1f, mpk-2f, mpk-3r, mpk-4r, mpk-5r, mpk-6f, mpk-7f, mpk-8r, mpk-9f, mpk-10f, mpk-11f, mpk-12r, mpk-13f, mpk-14r, and mpk-15r (see Figure 1 and Table 1).

View larger version (19K): In this window In a new window Download PPT slide   Figure 1. Sequence alterations in mpk-1 mutants. The boxes represent exons 2–7 of the mpk-1 gene, slanted lines indicate introns, and arrows indicate locations in the coding sequence of the mpk-1 mutations isolated in this study. mpk-1 expresses two mRNAs, which differ in that one mRNA (1 x 2) includes exon 1 while the other mRNA (SL x 2) contains the SL1 trans-spliced leader joined to exon 2 (LACKNER et al. 1994 ). The DNA sequence alteration and predicted protein coding change of each mutation is shown. Relative locations and orientations of all primers used to amplify and sequence the mpk-1-coding region from mutants are indicated below the boxes (Table 1). Arrows pointing to the right indicate that a given primer anneals to the antisense strand, while arrows pointing to the left indicate that a given primer anneals to the sense strand. Primers mpk-1f and mpk-12r were used in PCR experiments to amplify a 3.2-kb genomic fragment from single worms containing 292 bp of sequence 5' to the coding region, the complete coding region, and 259 bp of 3' untranslated region sequence. The 3.2-kb fragments from the wild type and each mutant were then sequenced using the internal primers shown, which span the entire coding region and all the intron-exon boundaries.

Germ line transformation experiments:
Germ line transformation experiments were done according to the method of FIRE and WATERSTON 1989 , as modified by MELLO et al. 1991 . DNAs were injected at concentrations of 100 µg/ml. Cotransformation markers were rol-6(su1006), an unc-30(+) plasmid and an unc-119(+) plasmid injected into N2, unc-30(e191), and unc-119(ed3op) hermaphrodites, respectively. For each type of array, at least two transgenic lines were obtained, and data from a representative line are shown. To integrate extrachromosomal arrays, L4 larvae containing the array of interest were subjected to gamma irradiation from a Cesium 137 source (4000 rad). Strains with integrated arrays were backcrossed at least once to dpy-20; him-5.

Manipulation of DNA and RNA:
Molecular biological techniques were done essentially as described by SAMBROOK et al. 1989 , with minor modifications. Heat shock mpk-1(+) was constructed by PCR amplifying the mpk-1-coding region contained in cDNA pML1 using primers mpk-20f and mpk-21r. These primers were used to amplify a 1.2-kb fragment starting at the mpk-1 start codon and ending at the stop codon. Amplified fragments were cloned into the heat shock promoter-containing vector JK465 (ROEHL and KIMBLE 1993 ) to make plasmid pML54. To construct hs-mpk-1(gf), we used the unique site elimination method to engineer the D324N substitution in pML1 using primers xba-1 and gain-of-function-1 (DENG and NICKOLOFF 1992 ). We then PCR amplified the insert of pML14 using primers mpk-20 and mpk-21 and ligated this fragment in the heat shock promoter vector JK465 to create construct pML40. hs-D-mek constructs were kindly provided by M. KOGA (KOGA and OHSHIMA 1995A ). lin-31-mpk-1(gf) was constructed by amplifying the mpk-1-coding region from pML14 with primers mpk-27 and mpk-28 (Table 1) and then ligating the resulting fragment into the lin-31 promoter-containing vector pB255. lin-31-D-mek(gf) was constructed by amplifying the D-mek-coding region from construct Dsor1(su1) (M. KOGA, personal communication) with primers mek-1 and mek-2 (Table 1) and then ligating the resulting fragment into pB255.

Phenotypic characterization:
When examined with a dissecting microscope, worms homozygous for strong mpk-1 alleles have a characteristic appearance in which the gonad appears mostly clear and devoid of embryos, with several clumps of dark granular material present in the gonad. They do not produce progeny.

Pn.p cell lineages shown in Table 2 and Table 3 were determined by direct observation of cells or cell divisions using Nomarski optics, as described by SULSTON and HORVITZ 1977 .

Male mating assays were conducting by mating 20 individual mpk-1(ga117) dpy-17 males with two unc-119 hermaphrodites each. None produced non-Unc cross-progeny. Nine out of 10 control matings of dpy-17 males with unc-119 hermaphrodites gave cross-progeny.

Western blotting:
Western blots were prepared essentially as described in HOSKINS et al. 1996 , with minor modifications. For each lane, 50 adult worms were picked into Laemmli buffer and boiled for 5 min before loading. Filters were immunoblotted overnight at 4° in anti-Erk2 peptide antibody K-23 (Santa Cruz Biotechnology, Santa Cruz, CA) diluted 1:500.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Isolation of new mpk-1 alleles:
We wanted to determine whether all of the anchor cell signal is transduced by mpk-1 (in which case elimination of mpk-1 activity should prevent all vulval induction caused by the anchor cell) or whether the anchor cell signaling pathway is branched (in which case elimination of mpk-1 activity should cause defects only in the mpk-1 branch of the signaling pathway). To accomplish this, we isolated new mpk-1 alleles in genetic screens, used molecular characterization to determine if any of them were likely to be null alleles, and then determined whether elimination of mpk-1 activity completely prevented the expression of vulval cell fates.

We used three approaches to isolate additional alleles of mpk-1. The first approach was to screen for mutations that failed to complement the egg-laying defective (Egl) phenotype of mpk-1(ku8). This screen should identify mpk-1 alleles that predominantly affect vulval development, but would miss null alleles if these alleles cause sterile or lethal phenotypes when heterozygous with ku8. Mutagenized N2 (wild-type) males were mated with mpk-1(ku8) dpy-17; unc-7 hermaphrodites, and the cross-progeny hermaphrodites were screened for an Egl phenotype. From a screen involving 7500 mutagenized haploid genomes, we recovered two mutations (ga110 and ga111) that mapped near to mpk-1 in the central region of chromosome III and failed to complement the mpk-1(ku8) Egl phenotype (data not shown). An additional allele, mpk-1(oz140), was provided to us by T. SHEDL and we subsequently showed that this allele results in the same DNA sequence change as mpk-1(ga110) (see Figure 1). Based on the molecular analyses described below, neither ga110, ga111, nor oz140 are likely to be null alleles.

Our second approach was to screen for mutations that failed to complement the sterile phenotype of mpk-1(oz140) (CHURCH et al. 1995 ). This screen could identify mpk-1 null alleles if they resulted in sterility, but would miss null alleles if they cause a lethal phenotype when heterozygous with oz140. We screened individual males for those that might be heterozygous for a new mpk-1 allele and, hence, sire sterile cross-progeny when mated to oz140/+ hermaphrodites. him-5 (him-5 mutations result in a high incidence of male progeny) hermaphrodites were mutagenized and allowed to self-fertilize. F₁ male progeny (potentially heterozygous for a new mpk-1 mutation, new/+) were mated individually to two hermaphrodites of genotype mpk-1(oz140) dpy-17/unc-79 dpy-17. Non-Dpy cross-progeny on each plate were then screened for the presence of approximately one-quarter sterile animals, and the new mpk-1 alleles were recovered from these plates by picking the wild-type siblings (genotype mpk-1(new)/+) of the sterile mutants (genotype mpk-1(new)/mpk-1(oz140)). From a screen involving 6008 haploid genomes, two mutations (ga117 and ga118) were isolated that mapped near to mpk-1 in the central region of chromosome III and failed to complement the sterile phenotype of mpk-1(oz140). As described below, our molecular analyses indicate that both ga117 and ga118 may severely reduce and possibly eliminate mpk-1 activity.

The third approach used to isolate new alleles was to perform a general screen to identify heterozygous animals that segregated approximately one-quarter sterile, vulvaless self-progeny, and then to determine which were new mpk-1 alleles. This screen could identify mutations that would be missed in the first screen (because it could identify mutations in any gene that cause a sterile, vulvaless phenotype), and it is simpler to perform than the second screen (because it does not require individual matings). Individual F₁ progeny of mutagenized hermaphrodites were placed on plates and allowed to self-fertilize, and then these plates were screened for the presence of one-quarter sterile F₂ progeny. We recovered three mutations from a screen involving 4020 mutagenized genomes. Two of these mutations are alleles of mek-2 because they map near to mek-2 on the left region of chromosome I and fail to complement the sterile phenotype caused by mek-2(n2678). One mutation (ga119) is a new allele of mpk-1 because it maps to the central region of chromosome III, fails to complement the sterile phenotype of mpk-1(oz140), and contains a DNA sequence change in the mpk-1 gene (see below).

Molecular characterization of mpk-1 alleles:
We used molecular analysis of the new mpk-1 alleles to determine if any were likely to be null mutations. (Figure 1). DNA from mpk-1(ga117) mutants had an AG to AA nucleotide substitution in the splice acceptor site upstream of exon 3 (Figure 1), and this G nucleotide is present in all known splice acceptor sites in C. elegans (BLUMENTHAL and STEWARD 1997 ). This result suggests that ga117 results in aberrant mRNA splicing and would be predicted to eliminate expression of MPK-1 sequences after amino acid 24 of the 376-amino-acid protein. We used anti-MAP kinase antibodies in Western blotting experiments to examine MPK-1 expression levels in extracts from adult wild-type worms and mpk-1(ga117) mutants (Figure 2). The antibodies detect one pair of bands near 45 kD and another pair near 55 kD in protein extracts from wild-type worms. As shown below, each pair of bands corresponds to phosphorylated and unphosphorylated forms of one MPK-1 isoform. The sizes of the lower band in each pair are the same as those predicted to result from translation of the two types of mpk-1 RNA (SLX2 and 1X2, respectively; LACKNER et al. 1994 ). Neither pair of bands is present in extracts from mpk-1(ga117) mutants, suggesting that mpk-1(ga117) reduces expression of wild-type MPK-1 to undetectable levels as a result of improper RNA splicing.

View larger version (51K): In this window In a new window Download PPT slide   Figure 2. Western blot analysis of MPK-1. Lysates were prepared from 50 adult worms from wild-type (N2), mpk-1(ga117), and mek-2(n2678) strains. The extracts were analyzed by Western blotting using 10% SDS-PAGE and anti–MAP kinase antibodies (Santa Cruz Biochemical). The positions and sizes (in Mr x 10-3) of prestained molecular weight standards (Bio-Rad, Richmond, CA) are indicated on the left. The positions of the two MPK-1 species predicted from analysis of cDNAs and Northern blots (of Mr 45 and 55 kD, respectively) are indicated by arrows.

DNA from mpk-1(ga118) mutants had a T to C transition that would change an arginine residue to a cysteine residue (R197C; Figure 1). This residue is conserved in all known MAP kinases and is in a region referred to as the phosphorylation loop, which contains the MEK phosphorylation site (TEY) and the kinase catalytic site (APE; ZHANG et al. 1995 ). An arginine to alanine substitution at the analogous residue in human ERK1 (R208A) showed increased tyrosine phosphorylation but decreased threonine phosphorylation by MEK, and it completely abolished kinase activity of Erk1 toward its substrate, myelin basic protein (BUTCH and GUAN 1996 ). Thus, the R197C substitution is likely to eliminate MPK-1 kinase activity.

DNA from mpk-1(ga119) mutants had a G to A transition that would change the glutamic acid in the kinase catalytic site (APE) to a lysine (APK) (Figure 1). A glutamic acid at this site is conserved in all known protein kinases (HANKS et al. 1988 ). An alanine substitution at the analogous residue in yeast cyclic AMP-dependent protein kinase resulted in a mutant protein with <0.9% of the activity of the wild-type protein (GIBBS and ZOLLER 1991 ). Similarly, the analogous mutation (E238K) in the C. elegans mek-2 gene results in a severe loss-of-function or null phenotype (KORNFELD et al. 1995 ). Thus, the mpk-1(ga119) mutation may nearly eliminate MPK-1 kinase activity.

DNA from mpk-1(ga110) and mpk-1(oz140) mutants had the identical DNA sequence alteration, a G to A transition predicted to change tryptophan at amino acid 351 to an amber stop codon (Figure 1). This substitution would result in truncation of the C-terminal 25 amino acids of the MPK-1 protein. These amino acids are not conserved in other MAP kinase proteins, suggesting that these mutations may reduce but not eliminate gene activity. Premature termination of translation often results in mRNA instability, so the oz140 phenotype might result from lower mRNA levels, decreased protein kinase activity, or both. smg-1 mutations increase the stability of nonsense messages (PULAK and ANDERSON 1993 ), and we found that the phenotypes caused by mpk-1(oz140) are suppressible by a smg-1 mutation (Table 2). Thus, C-terminal-truncated MPK-1 has MAP kinase activity when the levels of mpk-1(oz140) mRNA are increased by the smg-1 mutation. This result indicates that ga110 and oz140 partially reduce mpk-1 activity by decreasing mRNA levels rather than by affecting protein activity.

DNA from mpk-1(ga111) mutants has a T to C transition that would be predicted to alter amino acid 148 from valine to glycine (Figure 1). This valine is conserved in other MAP kinase proteins but not in other types of protein kinases (HANKS et al. 1988 ). This site is near the predicted MEK-binding site (BRUNET and POUYSSEGUR 1996 ; ZHANG et al. 1995 ), suggesting that MPK-1(ga111) may be defective in activation by MEK.

Vulval phenotypes of mpk-1 alleles:
Our first finding is that mpk-1 is required for vulval induction (Table 2). Using the putative null mutation ga117, we determined the cell lineages of the Pn.p cells in 20 mutant hermaphrodites and found that P5.p, P6.p, and P7.p expressed the uninduced 3° cell fate rather than induced 1° or 2° cell fates in every animal (Figure 3, Table 2 and Table 3). Next, we determined the number and types of cells that were present at the L4 stage of development in 20 homozygous mpk-1(ga117), mpk-1(ga118), and mpk-1(ga119) animals and found that all the vulval precursor cells had expressed uninduced 3° cell fates in each case (Table 2). These results show that mpk-1 null mutations cause a highly penetrant vulvaless phenotype.

View larger version (85K): In this window In a new window Download PPT slide   Figure 3. Phenotypes caused by mpk-1 alleles. (A) Wild-type hermaphrodite in the middle of the L4 larval stage. The 22 descendants of P5.p, P6.p, and P7.p have been generated and are beginning vulval morphogenesis (arrow). (B) mpk-1(ga110) hermaphrodite in the middle of the L4 larval stage. The posterior daughter of P6.p has failed to divide, and the undivided cell is seen adhering to the cuticle (arrowhead). The 2° cell lineages occurred normally in this animal. (C) mpk-1(ga117) hermaphrodite in the middle of the L4 larval stage. All the Pn.p cells have expressed the 3° cell fate (indicated by lines), and no vulval tissue has been formed. (D) Tail of a wild-type adult male. The arrow indicates a spicule, the male copulatory organ. (E) Tail of an mpk-1(ga117) adult male. The arrow indicates a crumpled spicule. mpk-1(ga117) males are unable to mate (see MATERIALS AND METHODS). All animals were viewed with Nomarski optics. Scale bar is 10 µm.

mpk-1(ga110) and mpk-1(oz140) have the same DNA sequence alteration that results in a weak vulvaless phenotype that is partially temperature sensitive. At 25°, 84% of mpk-1(oz140) animals exhibited a Vul phenotype in which all six vulval precursor cells expressed 3° cell fates, and the remaining 16% showed a partial Vul phenotype (Table 2). At 20°, none of mpk-1(oz140) mutants showed a complete Vul phenotype, 60% showed partial vulval defects, and the remaining 40% had a wild-type pattern of vulval cell lineages (Figure 3 and Table 2). We determined the vulval cell lineage pattern of nine mpk-1(ga110) animals at 20°. Four animals showed defects in the expression of the 1° cell fate, and the remaining five animals showed wild-type cell lineage patterns (Table 3). Gene dosage experiments suggest that oz140 is a partial reduction-of-function mutation because oz140 homozygotes have a weaker phenotype than heterozygotes carrying oz140 in trans to a null allele, ga117 (Table 2).

mpk-1(ga111) results in vulval phenotypes similar to the phenotypes previously observed for mpk-1(n2521) (Table 2). Homozygous mpk-1(ga111) animals exhibit wild-type vulval development at all temperatures (Table 2). However, mpk-1(ga111) is a recessive suppressor of the multivulva phenotype caused by the gain-of-function ras allele let-60(n1046). Specifically, let-60(n1046) single mutants have a Muv phenotype, but mpk-1(ga111); let-60(n1046) double mutants have a wild-type vulval phenotype (data not shown).

Interestingly, mpk-1(ku8) might have a neomorphic activity in addition to partially reducing mpk-1 activity. We found that mpk-1(ga111)/mpk-1(ga117) heterozygotes are wild-type, but that mpk-1(ga111)/mpk-1(ku8) heterozygotes are Egl although they have a normal pattern of vulval cell fates (data not shown). The cause of the Egl phenotype in these heterozygotes has not yet been identified. These results indicate that mpk-1(ku8) can have a more severe phenotype than putative mpk-1(null) alleles in trans to mpk-1(ga111).

Activation of the vulval pathway with MEK and MAP kinase is sufficient to cause vulval differentiation:
Because MAP kinase is the primary substrate for MEK in vertebrate cells (SEGER et al. 1992 ), and because mek-2 and mpk-1 act at adjacent points in the vulval genetic pathway (see below), we have made the assumption that activating MEK results specifically in an increase in MPK-1 activity. If the vulval induction pathway is linear, then activating MEK and MAP kinase should give full vulval induction. Alternatively, if the pathway is branched, activating only MEK and MAP kinase should give rise to only partial induction because other branches should not be activated. We have addressed this issue by observing the effects of activated D-mek and mpk-1 on vulval development (Table 4).

All experiments with MEK were performed using a Drosophila MEK cDNA containing the gain-of-function Dsor1 mutation (referred to as D-mek(gf)) (TSUDA et al. 1993 ). Drosophila mek and C. elegans mek-2 are 55% identical (WU et al. 1995 ), suggesting that D-mek may be functionally homologous to mek-2. We expressed D-mek(gf) under the control of a C. elegans heat shock promoter (hs-D-mek(gf)) or under the control of the lin-31 promoter, which predominantly drives expression in the vulval precursor cells (lin-31-D-mek(gf)) (P. TAN, unpublished results). Expression of hs-D-mek(gf) in transgenic worms does not result in a multivulva phenotype (Table 4).

The amino acid substitution D334N is a gain-of-function mutation of Drosophila MAP kinase (BRUNNER et al. 1994 ). We used site-directed mutagenesis to engineer the analogous substitution (D324N) into an mpk-1 cDNA clone. This cDNA construct was then placed under control of a C. elegans heat shock promoter (hs-mpk-1(gf)) or the lin-31 promoter (lin-31-mpk-1(gf)). We found that expression of hs-mpk-1(gf) in transgenic worms did not cause a multivulva phenotype (Table 4).

We next constructed transgenic lines containing both hs-mpk-1(gf) and hs-D-mek(gf) and found that heat shock expression of mpk-1(gf) and D-mek(gf) resulted in 90% of animals exhibiting a multivulva phenotype (Table 4). We followed the Pn.p cell lineages of seven animals transgenic for hs-mpk-1(gf) and hs-D-mek(gf) to determine the pattern of cell fates expressed by the Pn.p cells (Table 3). In these seven animals, there were 21 vulval precursor cells that would normally express the 3° cell fate (P3.p, P4.p, or P8.p in each animal), and we found that these cells expressed either the 1° cell fate or a hybrid 1°/2° cell fate in 13 cases.

We also constructed transgenic animals that express mpk-1(gf) and mek(gf) from the lin-31 promoter and found that 90% of these animals exhibited a Muv phenotype (Table 4). We then followed the Pn.p cell lineages of five such animals (Table 3). In four animals, all six vulval precursor cells expressed induced (1° and 2°) cell fates. In one animal, five out of six Pn.p cells expressed induced cell fates. We observed 12 cases in which adjacent Pn.p cells both expressed 1° cell fates (out of 25 possible cases). These results suggest that expression of mek(gf) and mpk-1(gf) is sufficient to induce vulval precursor cells to express the 1° cell fate in most cases.

mpk-1 acts downstream of signal transduction genes but upstream of transcription factor genes in the vulval signaling pathway:
We have used several complementary approaches to determine the position of mpk-1 in the genetic pathway specifying vulval development. Our results suggest that MPK-1 acts downstream of other conserved signal transduction molecules and upstream of transcription factors. The experiments are presented in the order in which the tested gene (or signaling event) is thought to act in the anchor cell signaling pathway.

To determine whether MEK and MAP kinase act downstream of the anchor cell signal, we used laser microsurgery to ablate the entire somatic gonad, including the anchor cell, and then induced expression of hs-D-mek(gf) and hs-mpk-1(gf). We observed that P3.p, P4.p, or P8.p expressed vulval cell fates leading to ectopic vulval tissue in L4 stage animals and a Muv phenotype in adults (Figure 4, Table 5). The vulval phenotype observed in these gonad-ablated animals was similar to that of mock-operated control animals. This result indicates that the Muv phenotype caused by activation of MEK and MAP kinase is independent of the anchor cell signal.

View larger version (144K): In this window In a new window Download PPT slide   Figure 4. The mpk-1(gf)-D-mek(gf) Muv phenotype is gonad independent. (A) Nomarski photomicrograph of the midbody region of an hs-D-mek(gf) hs-mpk-1(gf) (expressed from the integrated array gaIs17) adult hermaphrodite that was subjected to a heat shock (33° for 30 min) as an L2 larva. Syncytial gonad nuclei are apparent, and the animal has a multivulva phenotype. (B) Nomarski photomicrograph of the midbody region of a gaIs17 adult hermaphrodite in which the somatic gonad precursor cells (Z1.aaa and Z4.ppp) were ablated at the L1 larval stage with a laser microbeam and that was subjected to a heat shock (33° for 30 min) as an L2 larvae. No traces of the gonad are visible, and the animal has a multivulva phenotype.

lin-3 encodes a protein similar to EGF that is likely to be the inductive signal produced by the anchor cell, and overexpression of lin-3 results in a Muv phenotype (HILL and STERNBERG 1992 ). The vulvaless phenotype of mpk-1(ga117) was found to be epistatic to the Muv phenotype caused by overexpression of lin-3 because mpk-1(ga117); lin-3(d) double mutants have a vulvaless phenotype (Table 6 and Table 7). Furthermore, lin-3 is expressed in the anchor cell, and a genetic mosaic analysis shows that mpk-1 acts in the vulval precursor cells, showing that mpk-1(+) does not regulate lin-3 transcription (HILL and STERNBERG 1992 ; LACKNER et al. 1994 ). Thus, the Vul phenotype of mpk-1(ga117); lin-3(d) mutants does not result from loss of lin-3 expression, but it indicates that mpk-1 activity is necessary for the action of the lin-3 gene and suggests that mpk-1 functions downstream of lin-3.

Gain-of-function let-60 ras mutations result in a Muv phenotype. Previous work has shown that partial reduction-of-function alleles of mpk-1 suppress the Muv phenotype caused by a gain-of-function mutation of let-60. However, partial reduction-of-function alleles of mpk-1 do not block expression of the 1° cell fate in P6.p (LACKNER et al. 1994 ; WU and HAN 1994 ), so it was not clear whether activation of let-60(+) in P6.p required mpk-1 activity. We found that mpk-1(ga117); let-60(gf) double mutants are Vul, indicating that mpk-1 activity is necessary for let-60–mediated expression of vulval cell fates by any of the Pn.p cells (Table 6 and Table 7). This result confirms and extends previous work indicating that mpk-1 functions downstream of let-60.

lin-45 is a homolog of the proto-oncogene raf, and loss-of-function lin-45 mutations cause a Vul phenotype (HAN et al. 1993 ). We found that the Muv phenotype caused by D-mek(gf) and mpk-1(gf) is epistatic to the Vul phenotype caused by lin-45(lf), suggesting that mek and mpk-1 function downstream of lin-45 (Table 6).

To order the activity of mpk-1 relative to mek-2, we determined whether a mek-2 null mutation prevents phosphorylation of MPK-1. mpk-1 expresses two RNAs (1X2 and SLX2) predicted to encode proteins of 45 and 55 kD, respectively (LACKNER et al. 1994 ). Based on studies of vertebrate MAP kinase, phosphorylation of each of the MPK-1 protein isoforms by MEK would be expected to result in the appearance of a slightly slower migrating band on Western blots. As expected, Western blots of protein extracts from wild-type worms immunoblotted with anti-MAP kinase antibodies detect two pairs of bands, with one pair migrating at ~45 kD and the other at 55 kD (Figure 2). The upper bands of both the 45- and 55-kD doublets are missing in extracts from a mek-2 null mutant, suggesting that MPK-1 is not phosphorylated in these mutants. As phosphorylation is required for MAP kinase activation, this result suggests that MPK-1 is not activated in a mek-2 mutant and, hence, that mpk-1 is likely to function downstream of mek-2.

Genetic epistasis experiments suggest that lin-25 and sur-2 may function at a step between let-60 ras and transcription factors such as lin-1 Ets (TUCK and GREEWALD 1995; SINGH and HAN 1996 ). Both lin-25 and sur-2 encode novel gene products, and animals homozygous for a null mutation in either gene exhibit decreased levels of vulval differentiation and protruding vulva (Pvl) phenotypes. We found that the Pvl phenotype caused by either a lin-25 or a sur-2 mutation was epistatic to the Muv phenotype caused by a D-mek(gf); mpk-1(gf) array. Specifically, lin-25; hs-mek(gf); hs-mpk-1(gf) and sur-2; hs-mek(gf); hs-mpk-1(gf) animals have an Egl and Pvl phenotype (indicating reduced expression of the 1° cell fate by P6.p), but not a Muv phenotype (which would indicate increased expression of vulval cell fates by P3.p, P4.p, or P8.p; Table 6 and data not shown). Thus, the activity of lin-25 and sur-2 are necessary for the expression of vulval cell fates caused by mek-2(gf) and mpk-1(gf), indicating that lin-25 and sur-2 may act downstream of mek-2 and mpk-1 or in a parallel pathway.

Next, we ordered the activity of mpk-1 relative to lin-1, which encodes a protein with a region of similarity to the Ets DNA-binding domain (BEITEL et al. 1995 ). lin-1 loss-of-function mutations cause a Muv phenotype. mpk-1 (ga117); lin-1(sy254) double-mutant worms also exhibited a Muv phenotype (Table 6), indicating that lin-1 likely acts downstream of mpk-1.

In addition to the let-23 RTK/let-60 ras/mpk-1 signaling pathway, at least two other signaling pathways are involved in specifying the wild-type pattern of vulval cell fates. The first pathway is an inhibitory pathway postulated to arise from the surrounding hypodermis. lin-15 acts in this pathway and encodes two novel proteins that function to produce an inhibitory signal in the hypodermis (HEDGECOCK 1990; CLARK et al. 1994 ; HERMAN and HUANG et al. 1994). The Vul phenotype of mpk-1(ga117) was found to be epistatic to the Muv phenotype caused by a lin-15 null mutation (Table 6 and Table 7). This result suggests that mpk-1 functions downstream of lin-15 or in a parallel pathway.

The second pathway is a lateral signaling pathway in which a signal from the 1° cell induces adjacent Pn.p cells to adopt the 2° cell fate. lin-12 encodes a protein with significant homology to the Notch family of transmembrane receptors (YOCHEM and GREENWALD 1989 ), and is likely to be the receptor for the lateral signal. lin-12(n137) is a gain-of-function mutation that causes a Muv phenotype in which all six Pn.p cells express the 2° cell fate (YOCHEM and GREENWALD 1989 ). We found that an mpk-1(ga117) lin-12(n137) double-mutant strain exhibited a Muv phenotype, suggesting that most or all of the vulval precursor cells expressed the 2° cell fate (Table 6). This result indicates that mpk-1 does not act downstream of lin-12 in the specification of the 2° cell fate.

Constitutive activation of the anchor cell signaling pathway does not cause precocious vulval induction:
The vulval cell divisions begin ~29 hr after egg laying when wild-type animals are raised at 25°. The timing of vulval induction might be controlled by the time at which the vulval signaling pathway is activated. For example, some component of the signaling pathway, such as the anchor cell signal, might not be expressed until 29 hr after egg laying. Alternatively, the vulval precursor cells might not be competent to respond to activation of the let-23/let-60/mpk-1 signaling pathway until 29 hr after egg laying. We have constitutively activated the anchor cell signaling pathway by using animals that ubiquitously express D-mek(gf) from the EF1a promoter and constitutively express hs-mpk-1(+) when grown at 25°. As shown above, constitutive activation of mek and mpk-1 should be sufficient to induce vulval cell fates even if upstream components of the signaling pathway are not expressed before 29 hr after egg laying. In these transgenic worms, the vulval cell divisions began ~29–30 hr after the eggs were laid, as they do in wild-type worms (Table 8). The transgenes resulted in activation of the anchor cell signaling pathway because 90% of these animals exhibited a Muv phenotype. This result indicates that the vulval precursor cells are not competent to divide in response to MEK and MAP kinase signaling until at least 29 hr after egg laying.

mpk-1 acts in the let-60 ras signaling pathway in multiple cell types:
let-60 ras functions in at least six cell fate decisions because let-60 loss-of-function mutations affect at least six different cell types. An important question is whether let-60 signals through mpk-1 in all six cell types. If so, then mpk-1 mutants should show defects in all the cell types that are affected in let-60 mutants.

First, let-60 acts in the germ line, as loss-of-function mutations cause a sterile phenotype by preventing oocyte nuclei from exiting the pachytene stage of meiosis (CHURCH et al. 1995 ). The mpk-1(oz140) allele has previously been shown to cause similar sterility in homozygous animals (CHURCH et al. 1995 ). We report here that the mpk-1(ga117) null allele also causes this sterile phenotype and that the mpk-1(ga111) allele causes a temperature-sensitive sterile phenotype (Table 2). These results suggest that let-60 and mpk-1 may act in the same signaling pathway in meiotic prophase progression.

Second, let-60 is required for larval viability, as let-60 null alleles cause developmental arrest in the L1 stage with a characteristic rod-like lethal phenotype. We have found that mpk-1 is also required for larval viability, but that maternal mpk-1(+) can rescue lethality in homozygous mutant larvae. We found that mpk-1(ga117) progeny segregated from mpk-1(ga117)/+ hermaphrodites do not exhibit lethality (Table 9). We reasoned that maternal mpk-1(+) might rescue larval lethality in these animals, so we asked whether larval lethality was observed in a strain with less maternal mpk-1(+) activity. We examined homozygous mpk-1(ga117) animals segregated from mothers heterozygous for a partial loss-of-function mpk-1 allele and a null allele, i.e., of genotype ga111/ga117. The heterozygous mpk-1(ga111)/mpk-1(ga117) mothers are not sterile, though they do show reduced fecundity (data not shown). We observed that 97% of mpk-1(ga117)/mpk-1(ga117) progeny segregated from mpk-1(ga111)/mpk-1(ga117) parents arrested in the L1 stage as rod-like larvae (Table 9). The remaining 3% of mpk-1(ga117)/mpk-1(ga117) animals were non-Let and grew into sterile, vulvaless adults. Similar results were obtained with another allele, mpk-1(ga119), which may strongly reduce mpk-1 function (Table 9). These results suggest that let-60 and mpk-1 may act in the same signaling pathway in early larval development.

Third, mutations in let-60 have been shown to affect cell fate decisions in the male tail (CHAMBERLIN and STERNBERG 1993 , CHAMBERLIN and STERNBERG 1994 ) that result in defective spicule structure and defective male mating. mpk-1(ga117) homozygous males also have crumpled spicules and are unable to mate (Figure 3, MATERIALS AND METHODS), suggesting that mpk-1 and let-60 may act in the same signaling pathway during spicule development.

Fourth, let-60 mutants have defects in the migration of the sex myoblasts; the partial loss-of-function mutation mpk-1(ku1) causes weak defects in the migration of the sex myoblasts (SUNDARAM et al. 1996 ). However, we did not observe migration defects in mpk-1(ga117) null mutants. One possibility is that mpk-1 has a weak role in sex myoblast migration and that an mpk-1 null mutation causes a defect that is too subtle for us to score. Another possibility is that mpk-1(ku1) may be neomorphic or antimorphic, such that the migration defect might reflect interference from the mutant MPK-1(ku1) protein rather than a requirement for wild-type MPK-1(+). Thus, it is not clear whether mpk-1(+) acts in sex myoblast migration.

Fifth, let-60 mutants have defects in the expression of the P12.p cell fate (SUNDARAM et al. 1996 ). In contrast, mpk-1(ga117) mutants segregated from mpk-1(ga117)/mpk-1(ga111) mothers do not appear to have defects in the expression of the P12.p cell fate. One possibility is that maternal mpk-1 might be active in P12.p, so mpk-1 mutants would not exhibit cell fate defects in these cells because of maternal rescue of mpk-1(+) from their heterozygous mothers. Another possibility is that let-60 ras transduces signals independently of mpk-1 in this cell, possibly involving another C. elegans MAP kinase homolog.

Sixth, we have not observed mpk-1 phenotypes other than the ones mentioned above. This result suggests that mpk-1 may be required only in let-60 ras signaling pathways because all the mpk-1 phenotypes are also observed in let-60 ras mutants.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The anchor cell signaling pathway may be linear:
Previously isolated mpk-1 alleles have weak vulval phenotypes, whereas loss-of-function alleles in other genes in the let-23 RTK/let-60 ras signaling pathway have a strong vulvaless phenotype. These results raised the possibility that the let-23 signaling pathway may branch in a way that another branch of the signaling pathway could transduce the anchor cell signal in mpk-1 mutants. We have tested this possibility and have shown that mpk-1 null mutations cause a highly penetrant, vulvaless phenotype. Thus, there is no evidence for a signaling branch acting in parallel to mpk-1, and our data are consistent with the idea that the anchor cell signaling pathway is linear through the mpk-1 step. However, our results do not rule out the possibility that the pathway may be branched in a way that mpk-1 transduces the majority of the anchor cell signal and another branch has a minor role. In the absence of mpk-1 activity, the minor branch may not be capable of mediating observable amounts of vulval induction.

The case for a linear signaling pathway is strengthened by our finding that coexpression of D-mek(gf) and mpk-1(gf) results in a strong multivulva phenotype. Both genes act at the end of the anchor cell signal transduction pathway, and the primary known function for MEK is to activate MAP kinase. These results indicate that the pathway is not branched upstream of mek and mpk-1, as mek-2 and mpk-1 are both required for vulval induction and coexpression of mek(gf) and mpk-1(gf) is sufficient for vulval induction.

However, expression of either gene alone [either mek(gf) or mpk-1(gf)] did not cause a Muv phenotype. One possibility is that these two genes act in one linear pathway and that gain-of-function mutations at two sequential steps are required to fully activate the pathway. It could be that full activation of MPK-1 requires both the D324N missense mutation (analogous to Drosophila Sevenmaker) and expression of D-MEK(gf). One finding that supports this idea is that a strong gain-of-function allele of mek-2 results in a Muv phenotype in single mutants (WU et al. 1995 ), showing that mek-2 is sufficient to activate vulval cell fates, even though the mek(gf) allele used in these studies does not cause full activation of mek. Another possibility is that these two genes act in separate signaling branches and both branches need to be activated to cause a Muv phenotype.

mpk-1 acts downstream of signal transduction genes but upstream of transcription factor genes:
We have found that the MAP kinase homolog mpk-1 acts downstream of the Ras homolog let-60, the Raf homolog lin-45, and the MEK homolog mek-2, consistent with the biochemical position of MAP kinase in vertebrate signaling pathways and the genetic position of the Drosophila rolled MAP kinase in the R7 photoreceptor signaling pathway (BIGGS et al. 1994 ; BRUNNER et al. 1994 ; TREISMAN 1996 ).

Furthermore, we show that mpk-1 acts upstream of lin-1, which encodes a putative DNA binding protein similar to mammalian Ets-1 and Drosophila Yan (BEITEL et al. 1995 ). mpk-1 also acts upstream of lin-31, which encodes a winged helix transcription factor similar to mammalian HNF-3 and Drosophila forkhead (MILLER et al. 1993 ; TAN et al. 1998 ). Both LIN-1 and LIN-31 are phosphorylated by MAP kinase in vitro and in vivo (TAN et al. 1998 ). Unphosphorylated LIN-1 and LIN-31 bind to each other and form a protein complex that inhibits vulval induction. Phosphorylation of these proteins disrupts the LIN-1/LIN-31 protein complex and, thus, prevents inhibition of vulval cell fates. Phosphorylation of LIN-31 may also create a transactivation domain and permit LIN-31 to act as a transcriptional activator, thereby promoting expression of vulval cell fates.

An interesting point is that the anchor cell signaling pathway may be predominantly linear between lin-3 and mpk-1, but may branch downstream of mpk-1. mpk-1 null mutations have a highly penetrant vulvaless phenotype, but null mutations in genes that act downstream of mpk-1 (lin-1, lin-25, lin-31, and sur-2) neither completely activate nor completely prevent vulval induction. Specifically, null mutations in lin-25 and sur-2 result in a partial vulvaless phenotype (TUCK and GREEWALD 1995; SINGH and HAN 1996 ), lin-1 null mutations cause expression of partial or hybrid 1° or 2° cell fates (BEITEL et al. 1995 ), and lin-31 null mutations result in partially penetrant Muv and Vul phenotypes (MILLER et al. 1993 ; TAN et al. 1998 ). These results suggest that these genes may function on separate branches downstream of MAP kinase. Null mutations in any one gene may not completely block or activate vulval induction as they may affect only a subset of the branches downstream of MAP kinase.

mpk-1 and let-60 ras may act together in many developmental processes:
We have shown that mpk-1 may function with let-60 ras in one signaling pathway that regulates many developmental processes. One issue that has yet to be resolved regards signaling specificity of the let-23 RTK/let-60 ras/mpk-1 signaling pathway. How can activation of this signaling pathway in different cell types and times lead to different developmental outcomes? One possibility is that different levels of activation might lead to different cell fates. For example, it has been proposed that different levels of MPK-1 activity might cause vulval precursor cells to adopt either the 1° or 2° vulval cell fate. This graded signal hypothesis is consistent with some experiments (STERNBERG and HORVITZ 1986 ; THOMAS et al. 1990 ; KATZ et al. 1995 ), but is not consistent with others (GREENWALD et al. 1983 ; STERNBERG and HORVITZ 1989 ; KOGA and OHSHIMA 1995B ; SIMSKE and KIM 1995 ). Another possibility is that the mpk-1 signaling pathway might act combinatorially with other signaling pathways, such that activation of different combinations of signals could specify distinct cell fates. A third possibility is that different cell types intrinsically differ in how they respond to activation of MPK-1. MPK-1 might phosphorylate different targets in different cell types, leading to distinct cellular responses. For example, LIN-31 appears to be a target for MPK-1 in the vulval precursor cells and not in other cell types (TAN et al. 1998 ). Phosphorylation of LIN-31 in these cells may allow them to express vulval-specific target genes. In this case, signaling specificity reflects the range of MAP kinase targets expressed within the cell itself before activation of MAP kinase. Thus, MAP kinase activation may be a trigger to elicit a cellular response, but the manner in which the cell responds may be determined in a cell-intrinsic fashion.

	FOOTNOTES

¹ Present address: Department of Molecular and Cell Biology, University of California, Berkeley, CA 94720.

	ACKNOWLEDGMENTS

We thank D. EISENMANN, A. HAJNAL, A. VILLENEUVE, and current members of the Kim lab for discussions and critical comments on the manuscript. We thank M. KOGA, R. HILL, M. MADURO, and T. SCHEDL for reagents and/or strains and K. KORNFELD for communicating unpublished data on mpk-1(oz140). Many of the strains used in this work were supplied by the Caenorhabditis elegans Genetics Center. This work was supported by a National Institutes of Health (NIH) predoctoral training grant to M.L. and an NIH grant to S.K.K.

Manuscript received March 17, 1998; Accepted for publication May 26, 1998.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ADAMS, P. D. and P. J. PARKER, 1992  Activation of mitogen-activated protein (MAP) kinase by a MAP kinase-kinase. J. Biol. Chem. 267:13135-13137[Abstract/Free Full Text].

AROIAN, R. V., M. KOGA, J. E. MENDEL, Y. OHSHIMA, and P. W. STERNBERG, 1990  The let-23 gene necessary for Caenorhabditis elegans vulval induction encodes a tyrosine kinase of the EGF receptor subfamily. Nature 348:693-699[Medline].

AVERY, L. and H. R. HORVITZ, 1989  Pharyngeal pumping continues after laser killing of the pharyngeal nervous system of C. elegans. Neuron 4:473-485.

BEITEL, G. J., S. G. CLARK, and H. R. HORVITZ, 1990  Caenorhabditis elegans ras gene let-60 acts as a switch in the pathway of vulval induction. Nature 348:503-509[Medline].

BEITEL, G. J., S. TUCK, I. GREENWALD, and H. R. HORVITZ, 1995  The Caenorhabditis elegans gene lin-1 encodes an ETS-domain protein and defines a branch of the vulval induction pathway. Genes Dev. 9:3149-3162[Abstract/Free Full Text].

BIGGS, W. H., III, K. H. ZAVITZ, B. DICKSON, A. VAN DER STRATEN, and D. BRUNNER et al., 1994  The Drosophila rolled locus encodes a MAP kinase required in the sevenless signal transduction pathway. EMBO J. 13:1628-1635[Medline].

BLUMENTHAL, T., and K. STEWARD, 1997 RNA processing and gene structure, pp. 117–146 in C. elegans II, edited by D. RIDDLE, T. BLUMENTHAL, B. MEYER and J. PRIESS. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

BRENNER, S., 1974  The genetics of Caenorhabditis elegans. Genetics 77:71-94[Abstract/Free Full Text].

BRUNET, A. and J. POUYSSEGUR, 1996  Identification of MAP kinase domains by redirecting stress signals into growth factor responses. Science 272:1652-1655[Abstract].

BRUNNER, D., N. OELLERS, J. SZABAD, W. H. I. BIGGS, and S. L. ZIPURSKY et al., 1994  A gain-of-function mutation in Drosophila MAP kinase activates multiple receptor tyrosine kinase signaling pathways. Cell 76:875-888[Medline].

BUTCH, E. R. and K.-L. GUAN, 1996  Characterization of ERK1 activation site mutants and the effect on recognition by MEK1 and MEK2. J. Biol. Chem. 271:4230-4235[Abstract/Free Full Text].

CHAMBERLIN, H. M. and P. W. STERNBERG, 1993  Multiple cell interactions are required for fate specification during male spicule development in Caenorhabditis elegans. Development 118:297-324[Abstract].

CHAMBERLIN, H. M. and P. W. STERNBERG, 1994  The lin-3-let-23 pathway mediates inductive signalling during male spicule development in Caenorhabditis elegans. Development 120:2713-2721[Abstract].

CHURCH, D. L., K.-L. GUAN, and E. J. LAMBIE, 1995  Three genes of the MAP kinase cascade, mek-2, mpk-1/sur-1 and let-60 ras, are required for meoitic cell cycle progression in Caenorhabditis elegans. Development