General methods and strains:

C. elegans strains were cultured as described by Brenner (1974) and grown at 20° unless otherwise noted. The wild-type strain and parent of all mutant strains was C. elegans variety Bristol strain N2. Mutations used in this study are described by Riddle et al. (1997) and are as follows: linkage group (LG)III—lin-12(n137n460ts), lin-12(n950); LGIV—unc-5(e53), bli-6(sc16), dpy-20(e1282), lin-1(e1026, e1275, n176) (Horvitz and Sulston 1980), lin-1(e1777, n303, n304, n383, n431, n746, n753, n757, n1047, n1054, n1140) (Ferguson and Horvitz 1985), lin-1(ar147, m546, n381) (Beitel et al. 1995), lin-1(n2374) (Thomas et al. 2003), lin-1(ga56, ga68) (Eisenmann and Kim 2000), lin-1(he117, he119, n1814, n1815, n1816, n1817, n1848, n2692, n2693, n2694, n2696, n2701, n2704, n2705, n2714, n2750, n3000, n3443, n3522, sy254, sy289, sy321, sy618) (this study); LGV—let-341(n1613ts), him-5(e1490); LGX—lin-15(n767), lon-2(e678). We used standard genetic techniques to separate the lin-1(lf) mutations from lin-12(n137n460ts), lin-12(n950), let-341(n1613ts), or lin-15(n767), to backcross these mutations to N2, to position these mutations on the genetic map, and to perform complementation tests (Brenner 1974).

Identification of lin-1(lf) mutations:

In general, genetic screens were conducted as described by Brenner (1974) using ethyl methanesulfonate (EMS) as a mutagen. To identify mutations that suppress the rod-like larval lethality caused by let-341(n1613ts), we mutagenized let-341(n1613ts) hermaphrodites with EMS, raised F₁ hermaphrodites at the permissive temperature of 20° until the L4 larval stage, and then transferred the animals to the nonpermissive temperature of 25°. The F₂ self-progeny were screened for survivors. We screened 40,000 haploid genomes and identified five mutations that caused a Muv phenotype and failed to complement the Muv phenotype of lin-1(e1275), indicating that these are alleles of lin-1: n1814, n1815, n1816, n1817, and n1848. We also isolated one let-60(n1849gf) Muv mutation, five linked mutations that are likely to be intragenic revertants, and several unlinked suppressors that did not cause a Muv phenotype. In a related screen, we mutagenized let-341(n1613ts) hermaphrodites with EMS and screened for mutations that suppressed both the larval lethal and vulvaless phenotypes at 25°. An F₁ screen of 1.3 × 10⁶ haploid genomes identified 21 Muv mutations and 60 non-Muv mutations. An F₂ screen of 1.2 × 10⁵ haploid genomes identified 33 Muv mutations and 50 non-Muv mutations. These Muv mutations included the following 10 lin-1 alleles: n2692, n2693, n2694, n2696, n2701, n2704, n2705, n2714, n2750, and n3000.

To identify class B synthetic Muv mutations, we mutagenized a strain containing the class A mutation lin-15(n767) with EMS and conducted a clonal screen for F₂ progeny that displayed a Muv phenotype. We screened 6760 haploid genomes and identified two mutations, n3443 and n3522, that caused independent Muv phenotypes. n3443 and n3522 displayed linkage to dpy-20(IV) and failed to complement the Muv phenotype of lin-1(e1275).

To identify enhancers of the lin-12(gf) Muv phenotype, we conducted two screens. First, we mutagenized lin-12(n137n460) hermaphrodites using trimethylpsoralen (TMP) with ultraviolet (UV) light (Yandell et al. 1994) and screened F₂ self-progeny for extra vulval precursor cell (VPC) divisions. We screened 10,480 haploid genomes and identified the Muv mutation he117. Second, we mutagenized lin-12(n950); him-5(e1490) hermaphrodites with EMS and screened F₂ self-progeny for extra VPC divisions. We screened 2450 haploid genomes and identified the Muv mutation he119. he117 and he119 animals were backcrossed four times to N2. Each mutation displayed linkage to unc-5(IV). he117 and he119 failed to complement the Muv phenotype of lin-1(e1777) and lin-1(n304).

To identify mutations that cause a protruding vulva, we used EMS to mutagenize wild-type hermaphrodites and screened F₂ self-progeny for animals with a protruding vulva. We screened ∼7000 haploid genomes and identified 11 mutations, one of which, sy618, was Muv and mapped to chromosome IV. The sy618 strain was backcrossed twice to N2, and sy618 failed to complement the Muv phenotype of lin-1(e1777). To identify mutations that cause vulval defects that result in an extruded gonad at the L4 molt, we used EMS to mutagenize wild-type hermaphrodites and conducted an F₁ clonal screen. We identified the mutation sy289 and backcrossed sy289 twice to N2. To identify mutations that cause a Muv phenotype, we mutagenized wild-type hermaphrodites with EMS and isolated the Muv mutation sy321. sy254 was isolated as an F₂ Muv after TMP and UV mutagenesis of N2.

To confirm that the eight alleles that contained the wild-type sequence in the lin-1 coding region are indeed alleles of lin-1, we performed an additional complementation test. Males containing the mutant alleles ga68, he119, n1054, n1814, n1817, n2693, n2705, and n3443 were mated to bli-6(sc16) lin-1(e1275) hermaphrodites. Non-Bli, Muv hermaphrodites were observed in each case, indicating that these eight mutations failed to complement the lin-1(e1275) Muv phenotype.

Determination of DNA sequences of lin-1 alleles:

We determined the DNA sequence of 28 lin-1 alleles (Table 1). Genomic DNA was derived from homozygous mutant adult hermaphrodites and amplified by polymerase chain reaction (PCR) according to Williams et al. (1992). lin-1 contains six exons (Beitel et al. 1995). We PCR amplified fragments containing each exon and adjacent introns as shown in Figure 1A and as described by Jacobs et al. (1998) with the following modifications: GBO6f (5′-CCTTTCGAAATCGCTTCAAATC) and GBO14r were used to amplify a 514-bp fragment containing 109 bp of intron 2, exon 3, and 69 bp of intron 3. GBO15f (5′-GTGATAACAAACATTTTGTCAGTTG) and GBO17r were used to amplify a 538-bp fragment containing 121 bp of intron 3, exon 4, and 137 bp of intron 4. We purified the PCR-amplified DNA fragments and determined the complete sequences of both strands using an automated ABI Prism 377 DNA sequencer or an ABI Prism 3100 genetic analyzer (Applied Biosystems, Foster City, CA).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.—

LIN-1 gene and protein. (A) The lin-1 locus is diagrammed with exons (numbered boxes), introns (lines), and the start (ATG) and stop (TAG) codons of the open reading frame (Beitel et al. 1995). The beginning of exon 1 is defined by the position of an SL1 trans-spliced leader in lin-1 mRNA, and the end of exon 6 is defined by the position of a poly(A) tail in lin-1 mRNA. Arrows show the positions of pairs of oligonucleotides that were used to amplify exons and splice sites for DNA sequencing (see materials and methods; Jacobs et al. 1998). (B) Diagram of the LIN-1 protein with predicted PEST domains (P; amino acids 11–27, 176–199, and 216–252); the ETS domain (amino acids 60–145); the D domain (D; amino acids 277–291), and the FXFP motif (F; amino acids 382–385) that are ERK MAP kinase docking sites; and serine/threonine proline motifs that are potential ERK phosphorylation sites (asterisks). Lines indicate the positions of lin-1 missense mutations (above) and nonsense and splice site mutations (below). (C) ClustalW alignment of the ETS domains of C. elegans LIN-1 (amino acids 60–145; Beitel et al. 1995), human ETS-1 (amino acids 331–415; Watson et al. 1988), human ETS-2 (amino acids 359–443; Watson et al. 1988), human Net (amino acids 1–85; Giovane et al. 1994), Drosophila Yan (amino acids 392–479; Lai and Rubin 1992), and human Elk-1 (amino acids 1–86; Rao et al. 1989). Identities are shown between two to four proteins (shaded background) and five to six proteins (solid background). Substitutions caused by lin-1(lf) missense mutations are shown above. Solid and open triangles indicate residues of human Elk-1 that interact with nucleotides in the core binding motif (GGAA) and nucleotides outside the core binding motif, respectively, on the basis of the crystal structure of Elk-1 bound to DNA (Mo et al. 2000). α-helices (boxes) and β-sheets (arrows) of human Elk-1 are indicated.

Phenotypic analysis:

The 10 lin-1 alleles described in Table 2 were backcrossed twice to N2. Phenotypes were analyzed by placing one egg per petri dish for strains that did not display a highly penetrant egg-laying (Egl) defect or one L1 larva per petri dish for strains that displayed a highly penetrant Egl defect. Animals were allowed to develop at 20° and inspected daily using a dissecting microscope to identify: (1) animals that died during larval development or at the transition from L4 to young adult, (2) the number of pseudovulval protrusions on adults, and (3) the ability to generate live progeny during the first 2 days of adulthood. To analyze temperature sensitivity, we placed 15 young adult hermaphrodites on petri dishes, allowed the F₁ progeny to develop at 15°, 20°, and 25°, and scored the percentage of the F₁ population that displayed the Muv phenotype. To further analyze lin-1(ga56), we placed one egg per petri dish at 15°, 20°, and 25° and scored the Muv phenotype of adults.

Expression and purification of LIN-1 proteins:

A fragment of a lin-1 cDNA encoding LIN-1 residues 1–278 (Beitel et al. 1995) was ligated to pGEX-2T or pGEX-KG (Pharmacia LKB Biotechnologies, Piscataway, NJ) to generate plasmids encoding the fusion protein glutathione-S-transferase (GST):LIN-1(1–278). Standard in vitro mutagenesis techniques (Sambrook et al. 1989) were used to modify the lin-1 coding region to create the following plasmids: pDG91 encodes GST:LIN-1 (1–278 M114K), pDG92 encodes GST:LIN-1(1–278 Y127F), pDG93 encodes GST:LIN-1(1–278 Y126F), pDG94 encodes GST:LIN-1(1–278 R124Q), pDG95 encodes GST:LIN-1(1–278 R121K), pDG96 encodes GST:LIN-1(1–278 E92K), pDG97 encodes GST:LIN-1(1–278 G106R), and pDG112 encodes GST:LIN-1(1–278 L70F). We verified the entire LIN-1 coding region of each plasmid by DNA sequencing. The GST:LIN-1(1–278 M114K), GST:LIN-1(1–278 L70F), and GST:LIN-1(1–278 E92K) proteins did not express efficiently in bacteria (data not shown). To express these proteins in Sf9 insect cells, we ligated the coding regions for GST and LIN-1 into the vector pFastBac1 (Invitrogen, San Diego) to create the following plasmids: pDG104 encodes GST:LIN-1(1–278 WT), pDG105 encodes GST:LIN-1(1–278 M114K), pDG106 encodes GST:LIN-1(1–278 E92K), and pDG122 encodes GST:LIN-1(1–278 L70F).

For bacterial expression we transformed plasmids into Escherichia coli strain BL21, induced expression with 0.05 mm isopropyl thiogalactoside, disrupted the cells with sonication in 1× PBS buffer containing 1× complete protease inhibitor cocktail (PIC by Roche) and 0.2 m dithiothreitol (DTT), and prepared protein extracts. GST:LIN-1 fusion proteins were either soluble in 1× phosphate buffered saline (PBS) or solubilized in 1× PBS containing 7 m urea and then dialyzed overnight against PBS. Sf9 insect cells were prepared according to Ausubel (1987) and the manufacturer's instructions (Invitrogen). We transformed plasmids into E. coli DH10Bac cells (Invitrogen), isolated the recombinant bacmid DNA, transfected Sf9 cells, harvested the virus, and infected Sf9 cells. Infected Sf9 cells were scraped from the dish, washed with 1× PBS, pelleted, frozen, and treated with insect cell lysis buffer [50 mm Tris-HCl, pH 8.0, 100 mm NaCl, 5 mm EDTA, 0.5% NP-40, 1 mm DTT, 10 mm PMSF, 1× complete PIC (Roche)]. To purify GST:LIN-1 fusion proteins, we incubated protein extracts with glutathione-sepharose 4B, extensively washed, eluted bound proteins using 10 mm glutathione (in 50 mm Tris-HCl, pH 8.0) or 100 mm triethylamine, dialyzed the eluted proteins against kinase assay buffer (36 mm Tris, 0.1 mm EGTA, pH 7.08), and then dialyzed overnight against 1× gel shift buffer (200 mm Tris-HCl, pH 7.5, 50 mm NaCl, 1 mm MgCl₂, 0.5 mm EDTA, 10% glycerol). Protein samples were stored at 4° and used within 24 hr or stored at −80° after addition of 50% glycerol (Tanaka et al. 2002). To determine the amount of intact fusion protein in a partially purified sample, we separated the protein sample by SDS-PAGE with a 4% stacking/10% resolution gel, stained with Coomassie blue, and estimated the amount of fusion protein by comparison to the staining of known amounts of bovine serum albumin in adjacent lanes.

Electrophoretic mobility shift assay:

Double-stranded DNA was created by annealing two single-stranded oligonucleotides. Wild-type oligonucleotides were 5′-CTAGAGCTGAATAACCGGAAGTAACTAT and 5′-CATGATAGTTACTTCCGGTTATTCAGCT. Mutant oligonucleotides were 5′-CTAGAGCTGAATAAGCTACTGTAACTAT and 5′-CATGATAGTTACAGTAGCTTATTCAGCT. DNA was radioactively labeled by using DNA polymerase I Klenow (New England Biolabs, Beverly, MA) to fill the 3′ single-stranded overhang with [α-³²P]dATP (Ausubel 1987). Radiolabeled DNA was purified by phenol:chloroform extraction and a Sephadex G-50 column (Ausubel 1987; Sambrook et al. 1989). We incubated ∼30,000 cpm of radiolabeled DNA, various amounts of protein, and 1× reaction buffer (5 mm Tris-HCl, pH 7.5, 50 mm NaCl, 1 mm MgCl₂, 0.5 mm EDTA, 5% glycerol) in a total volume of 12.5 μl for 20 min on ice or at room temperature. Samples were fractionated using a 6% polyacrylamide gel (30% acrylamide/0.8% bis-acrylamide) with 0.5× TBE (Ausubel 1987; Sambrook et al. 1989). Gels were dried and exposed to film or a Kodak PhosphorScreen (Molecular Dynamics, Sunnyvale, CA). The PhosphorScreen was scanned using a Molecular Imager FX (Bio-Rad, Richmond, CA) and analyzed with Quantity One 4.1.0 (Bio-Rad) software.

Identification of 23 lin-1 alleles:

To identify mutations in genes that regulate vulval development, we conducted several different genetic screens. Here we focus on 23 lin-1 alleles that were identified in these screens. The let-341 gene encodes a guanine nucleotide exchange factor that is necessary for Ras-mediated signaling; let-341(lf) mutations cause rod-like, larval lethality and a vulvaless defect (Johnsen and Baillie 1991; Clark et al. 1992; Chang et al. 2000). lin-1(lf) mutations suppress the let-341(lf) lethal phenotype, and the lin-1(lf) Muv phenotype is epistatic to the let-341(lf) Vul phenotype, indicating that let-341 functions upstream of lin-1 (Clark et al. 1992). We identified 15 alleles of lin-1 as suppressors of the let-341(n1613ts) lethal phenotype, including n1814, n1815, n1816, n1817, n1848, n2692, n2693, n2694, n2696, n2701, n2704, n2705, n2714, n2750, and n3000. lin-15A is a synthetic multivulva gene, and lin-15A(n767) hermaphrodites display normal vulval development (Ferguson and Horvitz 1989; Sternberg and Horvitz 1989). We identified the lin-1 alleles n3443 and n3522 in a screen for multivulva hermaphrodites after mutagenizing lin-15A(n767) hermaphrodites. lin-12 encodes a receptor similar to Notch (Yochem et al. 1988). lin-12 is activated in P5.p and P7.p to promote the 2° cell fate, and lin-12(gf) mutations cause a Muv phenotype characterized by ectopic 2° cell fates. We identified the lin-1 alleles he117 and he119 as enhancers of the lin-12(gf) Muv phenotype. We identified four lin-1 alleles by mutagenizing wild-type animals and screening for vulval defects. The allele sy618 was identified by screening for animals with a protruding vulva, the allele sy289 was identified by screening for animals that died at the L4 molt because of an extruded gonad, and the alleles sy254 and sy321 were identified by screening for animals with a multivulval phenotype.

Each of these 23 newly identified mutations caused a multivulval phenotype that was similar in appearance to the multivulval phenotype caused by lin-1(lf) mutations. Furthermore, these mutations were positioned on chromosome IV like the lin-1 gene (see materials and methods). To determine if these were alleles of lin-1, we performed complementation tests. Each mutation failed to complement the Muv phenotype caused by a lin-1(lf) allele (see materials and methods). These results suggest that these are alleles of lin-1.

Previous studies have identified 20 loss-of-function lin-1 mutations and 7 gain-of-function lin-1 mutations. These lin-1(lf) mutations include e1026, e1275, n176 (Horvitz and Sulston 1980), e1777, n303, n304, n383, n431, n746, n753, n757, n1047, n1054, n1140 (Ferguson and Horvitz 1985), ar147, m546, n381 (Beitel et al. 1995), ga56, ga68 (Eisenmann and Kim 2000), and n2374 (Thomas et al. 2003). Therefore, the 23 mutations described above bring the total number of lin-1(lf) mutations to 43. The 7 lin-1(gf) mutations include ky54, n1761, n1790, n1855, n2515, n2525 (Jacobs et al. 1998), and cs50 (Rocheleau et al. 2002).

Molecular characterization of lin-1 alleles:

The 7 lin-1(gf) mutations have been characterized molecularly (Jacobs et al. 1998; Rocheleau et al. 2002), and 16 lin-1(lf) mutations have been characterized molecularly by Beitel et al. (1995). To identify residues and domains that are necessary for the function of the lin-1 gene, we investigated the molecular lesions of the 27 lin-1(lf) mutations that had not been characterized molecularly. These include 20 of the newly identified mutations and 7 previously identified mutations. The lin-1 gene contains six exons (Beitel et al. 1995). We determined the DNA sequence of these six exons and each splice junction, a region that includes the entire lin-1 coding region, using DNA from each of these 27 mutants (Figure 1A). A single base pair change compared to the wild-type sequence was detected in 19 alleles. Table 1 shows these nucleotide substitutions and the predicted amino acid substitutions. Seven newly identified alleles and two previously identified but molecularly uncharacterized alleles have missense changes: n1816, n1848, n2696, n2704, n2714, n2750, sy289, ga56, and n753 (Table 1). Beitel et al. (1995) previously identified missense changes in n303 and n1047. Thus, a total of 11 lin-1(lf) alleles have missense changes (Figure 1, B and C). The alleles n303 and n1816 have the identical base pair change and result in the substitution R121K; only the allele n1816 was further characterized. The alleles n1047 and n2696 have the identical base pair change and result in the substitution Y126F; only the allele n1047 was further characterized. The mutations n1848 and sy289 affect adjacent base pairs and change the same codon resulting in the substitutions G106R and G106E, respectively. Thus, these 11 independently derived mutations affect eight different amino acids.

Seven newly identified mutations and three previously identified but molecularly uncharacterized mutations are nonsense changes: he117, n3000, n1815, sy321, e1026, n746, n2374, n2694, n3522, and sy618 (Table 1). Beitel et al. (1995) showed that six mutations are nonsense changes: n757, e1275, n176, n383, n431, and e1777 (Table 1). Therefore, a total of 16 lin-1(lf) alleles contain nonsense changes. These mutations cause premature stop codons at nine different amino acid residues (Figure 1B). The alleles n3000 and he117 contain the identical base pair change and result in the substitution E99Ochre. Codon 105 is mutated in three alleles: n757 and n1815 have the identical base pair change that results in an amber stop codon whereas sy321 affects the adjacent base pair and results in an opal stop codon. The alleles n431 and e1777 contain the identical base pair change and result in the substitution Q309Amber. The alleles n2374, n2694, n3522, and sy618 contain the identical base pair change and result in the substitution Q390Amber.

Eight lin-1(lf) alleles contain rearrangements and/or deletions of the lin-1 locus (Beitel et al. 1995; Table 1). This group includes the alleles n2692, n2701, and sy254 that were identified in the genetic screens described above. Beitel et al. (1995) used Southern blots to show that lin-1(n1140) contains a deletion that was estimated to be ∼100 bp. We determined the sequence of the n1140 allele; n1140 contains a 61-bp deletion that includes 9 bp of intron 3 and 52 bp of exon 4 (Table 1).

Six newly identified alleles and two previously identified alleles contained the wild-type sequence of DNA in the lin-1 coding region and the portions of the introns that were analyzed: ga68, he119, n1054, n1814, n1817, n2693, n2705, and n3443 (Table 1). To confirm that these mutations affect the lin-1 locus, we performed an additional complementation test. Each allele failed to complement the Muv phenotype of lin-1(e1275) (see materials and methods). These findings suggest that the molecular lesions in these lin-1 alleles are outside the coding regions, and these mutations are likely to affect the expression or processing of the lin-1 mRNA.

The lin-1(lf) missense mutations form an allelic series:

To determine how the amino acids affected by the lin-1(lf) missense mutations contribute to the activity of LIN-1, we analyzed the phenotypes caused by these mutations. To analyze larval development, we placed one egg or L1 larva on a petri dish and used a dissecting microscope to analyze viability. lin-1 mutants occasionally died during the L1, L2, and L3 larval stages. lin-1 mutants often died during the transition from the L4 larval stage to the young adult stage, and these dead hermaphrodites frequently displayed an extruded gonad, suggesting that the lethality is related to defective vulval morphogenesis. Table 2 shows the phenotypes of nine lin-1 missense mutants, wild type, and lin-1(sy254), a putative null mutation that causes a rearrangement and deletion of the lin-1 locus. The lin-1(sy254) strain displayed 6% lethality during the L1 to L3 larval stages and 54% lethality during the L4 to young adult transition for a total of 60% lethality (Table 2, line 2). The nine lin-1 missense mutant strains displayed a wide range of lethality. The penetrance of all larval lethality ranged from 48% for lin-1(n1047) to 2% for lin-1(ga56) (Table 2, lines 3–11). These results indicate that these nine missense mutations reduce lin-1 activity to different extents and can be ordered in an allelic series.

To analyze adult phenotypes, we used a dissecting microscope to determine the number of pseudovulval protrusions of animals that survived to the adult stage. The lin-1(sy254) strain displayed a Muv phenotype that was 100% penetrant, and these mutants had an average of 3.6 pseudovulvae (range of 3–6; Table 2, line 2). Eight strains containing lin-1 missense mutations displayed a Muv phenotype that was 98–100% penetrant and had an average of 3.2–3.9 pseudovulvae (Table 2, lines 3–10). The lin-1(ga56) mutant strain displayed a Muv phenotype with a penetrance of 86%, and these mutants had a smaller average number of 2.5 pseudovulvae (range of 0–4; Table 2, line 11). We analyzed the overall function of the egg-laying system. lin-1(sy254) and five lin-1 missense mutants displayed a highly penetrant egg-laying defect and died from internally hatched progeny without depositing eggs on the petri dish. By contrast, four lin-1 missense mutants laid abundant eggs on the petri dishes, demonstrating the presence of a functional vulval passageway (Table 2).

We analyzed reproductive function by determining the number of animals that survived to adulthood and did not produce live progeny. The cellular basis for this lin-1 sterile phenotype has not been established. The lin-1(sy254) strain displayed 16% sterility; seven of the lin-1(lf) missense mutants displayed a similar value (Table 2, lines 2–9). By contrast, the lin-1(n753) and lin-1(ga56) mutant strains had a lower value of 2% sterility (Table 2, lines 10–11). The analysis of the adult phenotypes, like the analysis of the larval phenotypes, indicated that the missense mutations reduce lin-1 activity to different extents and clearly identified lin-1(ga56) as the weakest loss-of-function mutation.

lin-1(ga56) causes a temperature-sensitive phenotype:

The lin-1(e1275 R175Opal) mutation causes a Muv phenotype that is temperature sensitive (Beitel et al. 1995). To determine if any of the lin-1(lf) missense mutations cause a Muv phenotype that is temperature sensitive, we analyzed strains containing the missense alleles at 15°, 20°, and 25°. The Muv phenotype caused by the seven missense mutations n753, n1047, n1848, n1816, n2704, n2714, and n2750 displayed a similar penetrance at these three temperatures (data not shown). By contrast, the Muv phenotype caused by lin-1(ga56) displayed a penetrance of 53% at 15° (N = 167), 83% at 20° (N = 171), and 87% at 25° (N = 166), demonstrating that the lin-1(ga56) Muv phenotype is heat sensitive. For comparison, the penetrance of the Muv phenotype caused by lin-1(e1275 R175Opal) was 65% at 15° (N = 972), 91% at 20° (N = 911), and 100% at 25° (N = 863).

Wild-type LIN-1 binds DNA:

The predicted LIN-1 protein contains a highly conserved ETS domain (Figure 1C). The ETS domains of a variety of proteins have been demonstrated to bind DNA that contains the core motif GGAA/T (Karim et al. 1990). To determine if the LIN-1 ETS domain binds DNA, we developed an electrophoretic mobility shift assay (EMSA). Purified LIN-1 protein was obtained by generating a plasmid that encodes a fusion protein containing GST and the N-terminal 278 residues of LIN-1, GST:LIN-1(1–278) (Figure 2A). The protein was expressed in E. coli and partially purified by GST-affinity chromatography. The concentration of purified protein was estimated by resolving the partially purified extract by SDS-PAGE, staining with Coomassie blue, and comparing the intensity of GST:LIN-1(1–278) to known concentrations of bovine serum albumin.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.—

LIN-1 has sequence-specific DNA-binding activity. (A) Diagram of the fusion protein containing GST (amino acids 1–222; not drawn to scale) and LIN-1 amino acids 1–278 (numbered above). (B) Sequences of the wild-type DNA derived from the Drosophila E74 gene and the mutant DNA. The core GGAA motif necessary for ETS domain binding is in boldface type, and the mutated nucleotides are boxed. (*) denotes ³²P-labeled adenine. (C) Phosphorimage of an EMSA. Lanes are numbered above. The concentration of GST:LIN-1(1–278) protein is indicated below: lane 1, no protein; lanes 2 and 5, 8 ng of protein; lanes 3 and 6, 16 ng of protein; lanes 4 and 7, 32 ng of protein. Wild-type (WT) DNA was included in lanes 1–4 and mutant (M) DNA in lanes 5–7. The arrowhead indicates unbound DNA; the arrow indicates DNA that is bound to GST:LIN-1(1–278) and displays retarded mobility. The lower, less prominent band is also likely a complex of protein and DNA. To quantify binding, we measured the band indicated by the arrow.

To determine if the purified LIN-1 protein can bind DNA, we chose to use a DNA fragment from the Drosophila E74 gene that contains the core motif GGAA and displays high-affinity binding to several ETS proteins (Figure 2B; Shore and Sharrocks 1995; Shore et al. 1996). A radiolabeled, 32-bp DNA fragment was incubated with purified GST:LIN-1(1–278) protein, resolved on a nondenaturing polyacrylamide gel, and visualized. The DNA displayed similar mobility with no added protein and GST alone, indicating that GST does not bind the DNA (Figure 2C, lane 1, and data not shown). The addition of GST:LIN-1(1–278) caused the appearance of a slowly migrating band (Figure 2C, lane 2). The intensity of this band was proportional to the amount of GST:LIN-1(1–278) added to the reaction (Figure 2C, lanes 2–4). The band was eliminated by a mutation of the core GGAA motif (Figure 2C, lanes 5–7). These results indicate that the band is a complex between GST:LIN-1(1–278) and wild-type DNA and suggest that LIN-1 protein has a high-affinity, sequence-specific DNA-binding activity.

LIN-1 mutant proteins displayed reduced DNA binding:

The lin-1(lf) missense mutations affect conserved residues in the ETS domain (Figure 1C). To investigate how these amino acid substitutions affect DNA binding, we used site-directed mutagenesis to engineer eight lin-1(lf) missense mutations into plasmids encoding GST:LIN-1(1–278). Mutant proteins were expressed in bacteria or insect cells and partially purified by affinity chromatography. Binding experiments with mutant GST:LIN-1(1–278) were conducted in parallel with wild-type GST:LIN-1(1–278) for comparison. To determine how the protein concentration affects DNA binding, we used six different protein amounts in the binding reactions: 0, 1.56, 3.1, 6.25, 12.5, 25, and 50 ng (Figure 3, left). The results of the EMSA were quantified by using a phosphorimager to measure the radioactivity in the band corresponding to the protein:DNA complex (Figure 3, right).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.—

Missense changes reduce or eliminate LIN-1 DNA binding. A–H have a similar organization. We performed an EMSA analysis using wild-type and mutant proteins in parallel. (Left) A phosphorimage of the gel with lanes numbered above. Each reaction contained radiolabeled wild-type DNA. Lane 1 was a control reaction containing no protein (0). Lanes 2–7 contained 1.56, 3.13, 6.25, 12.5, 25, and 50 ng of wild-type GST:LIN-1(1–278), respectively. Lanes 8–13 contained 1.56, 3.13, 6.25, 12.5, 25, and 50 ng of the indicated GST:LIN-1 mutant protein, respectively. Each LIN-1 mutant protein contained LIN-1 residues 1–278. The band in lanes 2–7 is the complex of GST:LIN-1(1–278) and the radiolabeled DNA and corresponds to the band indicated by an arrow in Figure 2C. We quantified the intensity of the band in each lane using a phosphorimager. The signals were adjusted by subtracting the background intensity of lane 1 (0 ng of protein) and normalized by setting the signal in lane 7 equal to 100. These data are graphed on the right side. Squares show wild-type GST:LIN-1(1–278) from lanes 2–7, and circles show mutant GST:LIN-1 from lanes 8–13. The wild-type and mutant LIN-1 proteins in A, B, and D were expressed in insect cells. The wild-type and mutant LIN-1 proteins in C and E–H were expressed in E. coli. Wild-type LIN-1(1–278) derived from E. coli and insect cells displayed similar DNA binding.

Six mutant proteins displayed no detectable DNA binding at any of the protein concentrations analyzed (Figure 3, B–G). The results suggest that these six mutant proteins have a dramatic reduction in DNA-binding affinity. The GST:LIN-1(1–278 Y127F) protein displayed no DNA binding in reactions containing 1.56, 3.13, 6.25, and 12.5 ng of protein and significantly diminished DNA binding in reactions containing 25 and 50 ng of protein (Figure 3H). The GST:LIN-1(1–278 L70F) protein displayed reduced but detectable DNA binding in reactions containing 6.25, 12.5, 25, and 50 ng of protein (Figure 3A). Therefore, these two amino acid substitutions reduce but do not eliminate the DNA-binding activity of LIN-1. The lin-1(ga56L70F) mutation results in a mutant protein that retains the most DNA-binding activity in this group of eight mutant LIN-1 proteins.

Identification of lin-1 alleles:

A large collection of lin-1 alleles has been generated by conducting many different genetic screens in many different laboratories. Because these alleles were created by random chemical mutagenesis and isolated by screening for altered nematode morphology, this collection represents a large scale and relatively unbiased investigation of the residues and domains of LIN-1 that are important for function in an animal. Here we describe the isolation of 23 lin-1 alleles in genetic screens for the Muv phenotype, the protruding vulva phenotype, the extruded gonad phenotype, and the suppression of the let-341 larval lethal phenotype. Twenty lin-1(lf) mutations were previously identified in screens for the Muv and protruding vulva phenotypes (Horvitz and Sulston 1980; Ferguson and Horvitz 1985; Beitel et al. 1995; Eisenmann and Kim 2000; Thomas et al. 2003). Thus, a total of 43 lin-1(lf) mutations have now been described. Seven lin-1 gain-of-function mutations were previously identified in screens for suppression of Muv, enhancement of Vul, or larval lethal phenotypes (Jacobs et al. 1998; Rocheleau et al. 2002).

To exploit these lin-1 alleles for a structure/function study, we analyzed the molecular lesions. Here we describe the molecular analysis of 28 lin-1 alleles. As a result of these studies and previous investigations, each of these 50 loss-of-function and gain-of-function lin-1 alleles has now been characterized molecularly. All 11 missense mutations that cause a loss of function cluster in the ETS domain, and all four missense mutations that cause a gain of function cluster in the C-terminal region. The 18 nonsense mutations are distributed from codon 99 to codon 390. One mutation affects a splice site. Eight alleles have deletions or rearrangements of the locus. Eight alleles have the wild-type sequence in the coding region, suggesting that these mutations affect the regulation of lin-1 expression.

The analysis of these mutations indicates that many of the chemically inducible single base pair mutations that can significantly affect the function of lin-1 have been identified. In seven cases, the identical mutation was isolated twice (R121K, Y126F, P384L, E99Ochre, W105Amber, Q309Amber, R352Opal), and in one case, the identical mutation was isolated four times (Q390Amber). In three cases, two different mutations that affect the same codon were isolated (G106R/E, P384S/L, W105Amber/Opal). Overall, 9 of the 20 codons affected by nonsense or missense mutations were mutated in two or more independently identified strains. The identification of the same molecular lesion in two independently derived lin-1 alleles strongly supports the conclusion that the phenotype is caused by the molecular lesion and does not require a second mutation in the genome. Furthermore, the finding that nearly half of the codons affected by single base changes were affected in two or more strains indicates that molecular lesions of lin-1 have been heavily sampled during these different genetic screens. EMS, the mutagen used in most of these experiments, causes primarily GC-to-AT changes and therefore primarily affects certain codons (Coulondre and Miller 1977). It is likely that most EMS-induced mutations of lin-1 that cause a strong Muv or Vul phenotype have been identified.

lin-1 missense mutations identify two important functions of LIN-1—regulation by ERK and DNA binding—and identify eight residues of the LIN-1 ETS domain that are necessary for DNA binding:

The lin-1 missense mutations form two clusters. The four gain-of-function mutations affect the C terminus of lin-1. Three mutations affect proline 384, which is part of the FXFP motif. Biochemical studies demonstrated that the FXFP motif functions as a docking site for ERK MAP kinase (Jacobs et al. 1998, 1999). The proline 384 mutations diminish the binding affinity of LIN-1 and ERK MAP kinase, resulting in LIN-1 protein that is not effectively regulated and is constitutively active as an inhibitor of the 1° vulval cell fate. The lin-1(cs50gf) mutation affects proline 315 and may affect phosphorylation of serine 314 by ERK (Rocheleau et al. 2002). Together, the lin-1(gf) mutations identify a C-terminal domain of LIN-1 that binds ERK MAP kinase and is necessary for regulation of LIN-1 and induction of the 1° vulval cell fate in P6.p.

The 11 loss-of-function lin-1 missense mutations cluster in the ETS domain. Several ETS domains have been shown to bind DNA with the core motif GGAA/T (Karim et al. 1990; Shore and Sharrocks 1995; Shore et al. 1996). The structure of ETS domains bound to DNA has been characterized using X-ray crystallography (Kodandapani et al. 1996; Mo et al. 1998, 2000). ETS domains exhibit a winged helix-loop-helix topology. The α3 DNA recognition helix is embedded in the major groove of the DNA and has base-specific interactions with the GGA core motif (Figure 1C). The β3-turn-β4 region, termed the winged segment, and the turn between the α2 and α3 helices make extensive contacts to the phosphate backbone of the DNA. The ETS domain of LIN-1 has been highly conserved during evolution and displays extensive similarity to the ETS domain of human Elk-1. Here we demonstrate that the LIN-1 ETS domain has sequence-specific DNA-binding activity that requires the GGAA core motif. Therefore, LIN-1 and vertebrate Elk-1 share a conserved sequence and a conserved function of DNA binding.

We analyzed how the lin-1(lf) missense substitutions affect DNA binding; eight missense substitutions eliminate or significantly reduce DNA binding. The lin-1(n303 R121K) and lin-1(n2714 R124Q) mutations affect two arginine residues that are in the α3 recognition helix of Elk-1 and make extensive contacts with the GGA core motif (Figure 1C, solid arrowheads). The mutant proteins displayed no detectable DNA binding. The lin-1(n1047 Y126F) and lin-1(n2704 Y127F) mutations affect two tyrosine residues that are in the α3 recognition helix of Elk-1 and make contacts with nucleotides outside the GGA core (Figure 1C, open arrowheads). The LIN-1(Y126F) protein displayed no detectable DNA binding, and the LIN-1(Y127F) protein displayed very weak DNA binding. The lin-1(n753 M114K) mutation affects a methionine in the turn region of Elk-1 that makes contacts with the GGA core (Figure 1C, solid arrowheads). This mutant protein displayed no detectable DNA binding. These results are consistent with the model that the ETS domain of LIN-1 adopts a structure that is similar to the structure of Elk-1. These studies contribute to an understanding of the ETS domain by demonstrating that residues predicted to contact the DNA are essential for DNA binding.

Three missense substitutions affect residues that do not directly contact DNA but are highly conserved nonetheless. lin-1(n2750 E92K) affects a glutamic acid in the β2 turn and lin-1(n1848 G106R) affects a glycine in the α2 helix (Figure 1C). Both of these mutant LIN-1 proteins had no detectable DNA binding. The lin-1(ga56 L70F) mutation affects a leucine in the α1 helix. This mutant LIN-1 protein had reduced but still detectable DNA binding. These data indicate that these residues are important for the conformation of the ETS domain and may be necessary to correctly position the residues that directly contact the DNA.

DNA binding is necessary for LIN-1 function in animals:

In addition to determining how the lin-1(lf) substitutions affect DNA binding, we determined how these mutations affect lin-1 function in animals. These mutations reduce the activity of lin-1 at multiple developmental stages and can be arranged in an allelic series. The strongest mutations cause a phenotype that is similar to the phenotype caused by a lin-1 null mutation, whereas the lin-1(ga56 L70F) mutation causes a milder phenotype, indicating that it partially reduces lin-1 activity. An analysis of the DNA-binding activity of the mutant proteins demonstrated that each mutant protein had a profound defect in DNA binding and that the LIN-1(L70F) mutant retained the most DNA-binding activity. These results establish a correlation between the DNA-binding activity of the mutant LIN-1 protein and the function of the mutant lin-1 allele in animals, indicating that the defects in DNA binding are the cause of mutant phenotypes. The expression level of the mutant LIN-1 proteins was not examined; some of these mutations may also affect LIN-1 protein stability. The similarity between the strong missense mutations and the lin-1 null mutation suggests that DNA binding is necessary for all of the functions of lin-1.

Although the structures of ETS domains have been analyzed extensively, few studies have investigated the function of specific residues for DNA binding and activity in animals. Golay et al. (1988) identified one missense mutation in the v-ets oncogene that reduces the ability of the virus to transform cells. The analogous substitution in Elk-1 (R74D) reduced DNA affinity ∼10-fold (Janknecht and Nordheim 1992). These findings indicate that DNA binding is important for transformation activity of the viral ETS gene. Residues of PU.1 that are necessary for DNA binding were identified by making glycine substitutions (Kodandapani et al. 1996). Tootle et al. (2003) introduced two analogous substitutions in Drosophila Yan (W439G and K443G) and demonstrated that the mutant Yan protein was mislocalized to the cytoplasm of cultured cells. These findings indicate that DNA binding is important for nuclear localization of Yan. The results presented here show that eight different lin-1 missense mutations that result in reduced DNA binding also reduce or eliminate function in animals. Thus, DNA binding is necessary for the function of LIN-1, and this may be the conserved property of ETS proteins.

On the basis of these genetic and biochemical studies, we propose a model of LIN-1 function shown in Figure 4. Newly synthesized LIN-1 protein enters the nucleus, and the ETS domain binds to GGA motifs in the promoters of target genes (Figure 4, parts 1 and 2). These target genes are likely to promote the 1° vulval cell fate, and LIN-1 may repress the transcription of these target genes by interacting with corepressors and chromatin-remodeling enzymes that affect histones bound to the promoter and condense the chromatin (Figure 4, part 3). When the anchor cell signals P6.p, resulting in the activation of LET-23 RTK, LET-60 Ras, and MPK-1 ERK, phosphorylated MPK-1 enters the nucleus. MPK-1 interacts with the D domain and the FXFP motif of LIN-1 and phosphorylates one or more serine/threonine residues (Figure 4, parts 4 and 5). Phosphorylation may cause LIN-1 to activate transcription of target genes, and phosphorylated LIN-1 might recruit coactivators and/or chromatin-remodeling enzymes that decondense the chromatin (Figure 4, part 6). It is also possible that phosphorylation prevents LIN-1-mediated repression but does not cause LIN-1 to activate transcription. The genetic and biochemical studies of lin-1(lf) missense mutations support the importance of the ETS domain interaction with DNA shown in parts 1 and 2 of Figure 4. The genetic and biochemical analyses of lin-1(gf) missense mutations support the importance of interaction with MPK-1 (Figure 4, parts 4 and 5). It is notable that domains of LIN-1 that interact with corepressors, coactivators, and chromatin-remodeling enzymes were not defined by the collection of LIN-1 missense mutations. Thus, the mechanisms of LIN-1 transcriptional regulation shown in Figure 4, parts 3 and 6, have yet to be well defined. Biochemical screens for proteins that interact with LIN-1 are a promising approach that may complement the genetic screens described here and illuminate these aspects of LIN-1 function.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.—

A model of LIN-1 regulation of transcription. LIN-1 is diagrammed with the ETS domain (diagonal lines), D domain (solid), FXFP motif (open), and serine/threonine residues that can be phosphorylated by ERK (S/T). A LIN-1 target gene that promotes the 1° vulval cell fate (1° fate) contains the nucleotide sequence GGA. An interacting histone can vary between silencing (S) and activating (A). The size of the arrow indicates the level of target gene transcription. MPK-1 ERK MAP kinase is diagrammed with the kinase domain, a CD domain that interacts with the D domain (Tanoue et al. 2001), and a FD domain that interacts with the FXFP motif. The position of the FD domain on ERK has yet to be defined.