Identification of Heterochronic Mutants in Caenorhabditis elegans: Temporal Misexpression of a Collagen::Green Fluorescent Protein Fusion Gene

Corresponding author: Ann E. Rougvie, University of Minnesota, 250 BioScience Center, 1445 Gortner Ave., St. Paul, MN 55108, rougvie{at}biosci.cbs.umn.edu (E-mail).

Communicating editor: I. GREENWALD

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The heterochronic genes lin-4, lin-14, lin-28, and lin-29 specify the timing of lateral hypodermal seam cell terminal differentiation in Caenorhabditis elegans. We devised a screen to identify additional genes involved in this developmental timing mechanism based on identification of mutants that exhibit temporal misexpression from the col-19 promoter, a downstream target of the heterochronic gene pathway. We fused the col-19 promoter to the green fluorescent protein gene (gfp) and demonstrated that hypodermal expression of the fusion gene is adult-specific in wild-type animals and temporally regulated by the heterochronic gene pathway. We generated a transgenic strain in which the col-19::gfp fusion construct is not expressed because of mutation of lin-4, which prevents seam cell terminal differentiation. We have identified and characterized 26 mutations that restore col-19::gfp expression in the lin-4 mutant background. Most of the mutations also restore other aspects of the seam cell terminal differentiation program that are defective in lin-4 mutant animals. Twelve mutations are alleles of three previously identified genes known to be required for proper timing of hypodermal terminal differentiation. Among these are four new alleles of lin-42, a heterochronic gene for which a single allele had been described previously. Two mutations define a new gene, lin-58. When separated from lin-4, the lin-58 mutations cause precocious seam cell terminal differentiation and thus define a new member of the heterochronic gene pathway.

THE formation of an animal from a single cell requires a rigorously controlled schedule of cell division, differentiation and morphogenesis. For these events to be specified correctly throughout the developing animal, there must be integration of temporal information with appropriate spatial and sexual instructional cues.

The molecular mechanisms that control the timing of developmental events are beginning to be elucidated. Mutations have been identified in several organisms that cause alterations in the time of onset of certain developmental events and likely define genes with roles in the temporal progression of patterning. For example, in Dictyostelium, mutations in rde cause premature terminal differentiation of stalk and spore cells (SIMON et al. 1992 ), and in maize, mutations in the Teopod genes retard the transition between the expression of juvenile and adult characteristics in shoot development (POETHIG 1988 ), while mutations in glossy15 cause premature expression of adult characteristics (EVANS et al. 1994 ). In Drosophila, mutation of the ana gene causes certain neuroblasts to proliferate too early (EBENS et al. 1993 ). Our work is centered on the Caenorhabditis elegans heterochronic genes (AMBROS and HORVITZ 1984 ; see AMBROS 1997 , for review), which control the temporal execution of specific postembryonic developmental events.

One stage-specific event timed by the heterochronic genes is the terminal differentiation of lateral hypodermal seam cells, which occurs during the final molt, at the end of the fourth and final larval stage (L4). The seam cells are hypodermal stem cells arranged in rows along the left and right lateral midlines of the animal. The seam cells divide at each of the first three molts during larval development, maintaining a seam cell population and generating cells that join the hypodermal syncytium that covers most of the animal. During the L4-to-adult molt the seam cells switch to their adult developmental program as they terminally differentiate; they exit permanently from the cell cycle, fuse and synthesize a set of longitudinal lateral ridges termed adult alae that distinguish morphologically the adult cuticle from larval stage cuticles (SULSTON and HORVITZ 1977 ; SINGH and SULSTON 1978 ). The execution of this terminal differentiation program requires the transcription factor LIN-29, which accumulates in the hypodermis during the L4 stage (ROUGVIE and AMBROS 1995 ; BETTINGER et al. 1996 ). The proper restriction of LIN-29 to this particular stage in the hypodermis requires the action of the heterochronic genes lin-4, lin-14 and lin-28 (BETTINGER et al. 1996 ).

Loss-of-function lin-14 and lin-28 alleles produce a precocious phenotype; seam cells in these mutants terminally differentiate one and, in the case of lin-28, sometimes two stages early (AMBROS and HORVITZ 1984 ; AMBROS 1989 ). The precocious phenotype observed in these animals is probably the result of premature accumulation of LIN-29 in the hypodermis (BETTINGER et al. 1996 ). Loss-of-function lin-4 and lin-29 alleles and gain-of-function lin-14 alleles produce a retarded phenotype; seam cell terminal differentiation in these mutants is not observed. Animals bearing these mutations undergo extra molting cycles not observed in wild-type animals and synthesize a new larval-type cuticle during the extra molts (AMBROS and HORVITZ 1984 ). LIN-29 is not detected in the hypodermis of these retarded mutants (BETTINGER et al. 1996 ).

Although these four heterochronic genes are key players in restricting seam cell terminal differentiation to the L4-to-adult molt in wild-type animals, they are probably not the complete set. lin-14 is defined genetically as a negative regulator of lin-29 (AMBROS 1989 ), yet lin-14 protein, which is present during the L1 stage, becomes undetectable in the hypodermis by the early L2 stage (RUVKUN and GIUSTO 1989 ), significantly prior to the time of hypodermal LIN-29 accumulation during the L4 stage (BETTINGER et al. 1996 ). This observation suggests that LIN-14 functions indirectly in the control of lin-29 activity. Although lin-28 cannot be ruled out as a direct negative regulator of lin-29 activity because its protein accumulation patterns are not yet known, hypodermal expression of a lin-28-reporter gene fusion is generally not detected beyond the L2 stage (MOSS et al. 1997 ). The observed down regulation of lin-14 and lin-28 activity is mediated by the lin-4 gene product, a 22-nucleotide RNA predicted to act through complementary sequences in the 3' untranslated regions (UTRs) of lin-14 and lin-28 (LEE et al. 1993 ; WIGHTMAN et al. 1993 ; MOSS et al. 1997 ). The molecular analyses of these four heterochronic genes thus suggest that there are additional genes that act downstream of lin-14 and lin-28 to control seam cell terminal differentiation. In addition, since directed screens to identify genes that time seam cell differentiation have not been previously reported, heterochronic genes whose products act with or prior to LIN-14 and LIN-28 may also have eluded identification.

To understand more thoroughly how the timing of seam cell terminal differentiation is restricted to the L4-to-adult molt, we sought to identify additional members of the heterochronic gene pathway. We devised a genetic screen for mutants defective in the timing of seam cell terminal differentiation in which the desired class of mutants temporally misexpress a lin-29-dependent reporter gene fusion. We fused the promoter of col-19, a gene encoding a cuticle collagen normally expressed only in adult hypodermal cells (LIU et al. 1995 ; ROUGVIE and AMBROS 1995 ), to the green fluorescent protein (GFP) gene. The ease of assaying GFP expression in live worms (CHALFIE et al. 1994 ) affords a simple method for monitoring col-19::gfp expression and, by extension, lin-29 activity in vivo. We placed the col-19::gfp reporter gene in a lin-4 mutant background where it is not normally expressed and screened for restored expression. The isolated mutations bypass the need for lin-4 activity in the hypodermis.

In this study, we have identified and characterized 26 new mutations that were recovered solely on the basis of restored expression of a col-19::gfp reporter gene in a lin-4 background. These mutations include six new alleles of lin-14 and two new alleles of lin-28, thus demonstrating the effectiveness of this approach for the isolation of heterochronic genes. We have also isolated four new alleles of lin-42, a heterochronic gene for which a single allele had been described (LIU 1990 ). Finally, at least one new gene, lin-58, has been identified as a member of the heterochronic gene pathway.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

General procedures for the maintenance of C. elegans were as described by BRENNER 1974 . The wild-type strain of C. elegans was var. Bristol N2 (BRENNER 1974 ).

Genes, alleles, rearrangements and strains:
The mutations identified in this work are listed in Table 1. Other genes and alleles used in this work are as follows and are described by HODGKIN 1997 unless otherwise noted: Linkage group (LG) I, dpy-5(e61), lin-28[n719, n947, ga54 (MOSS et al. 1997 )]; LG II, lin-42(n1089), dpy-25(e817), lin-4(e912), unc-4(e120), lin-29(n836); LG IV, him-8(e1489); LG V, dpy-11(e224), lon-3(e2175), him-5(e1467), unc-76(e911), dpy-21(e428); and LG X, dpy-7(e1324ts), dpy-6(e14), lin-14(n179ts, ma154ts, n355n679ts, n536sd).

The chromosomal deficiencies ccDf11 II (CHEN et al. 1992 ), yDf4 V and yDf8 V (DELONG et al. 1993 ) were used. The LG II balancer mnC1[dpy-10(e128) unc-52(e444)] and the reciprocal translocation balancers szT1(X;I) and nT1[unc-?(n754) let-?](IV;V) were also used (EDGLEY et al. 1995 ).

Construction of col-19::gfp expression plasmid and description of transgenic strains:
pJA1 contains an 866-bp HindIII-BamHI fragment extending from -845 to +21 relative to the ATG of the col-19 promoter in a translational fusion with GFP in the TU#62 vector (M. CHALFIE, personal communication). Transgenic animals were generated using standard methods (MELLO et al. 1991 ). The injection mix contained pJA1 and pRF4, each at a concentration of 100 ng/µl. pRF4 contains the semidominant rol-6(su1006) allele that acts as a scorable marker for the identification of transgenic animals by causing animals to move abnormally (the Rol phenotype; MELLO et al. 1991 ). We generated five independent transformants containing extrachromosomal arrays of these plasmids. Each line showed adult-specific expression of GFP in the hypodermis. The extrachromosomal array in one of these strains was integrated into the genome by gamma irradiation (MELLO and FIRE 1995 ). Two independent homozygous integrated arrays were obtained. They showed indistinguishable patterns of gfp expression, which were similar to that observed for the extrachromosomal arrays, except they were less mosaic. One of the integrated arrays, veIs13, was chosen for subsequent analyses. veIs13 is tightly linked to unc-76 V.

Microscopy:
Light microscopy was performed with a Microphot-FXA (Nikon, Inc., Melville, NY) equipped with Nomarski and fluorescence optics. Techniques used for Nomarski differential interference contrast microscopy were as described by SULSTON and HORVITZ 1977 . To assess gfp expression, worms were examined using FITC or B2A filters (Chroma Technology Corp., Brattleboro, VT). Genetic screens were performed using a modified Wild MZ8 dissecting microscope (Leica, Inc., Deerfield, IL) with an attached 100-W mercury light source and Chroma Technology filter set #32001A.

Isolation of lin-4 suppressors:
lin-4; veIs13 animals were mutagenized with ethyl methane sulfonate (EMS) as described by SULSTON and HODGKIN 1988 or -irradiated with 4500 r from a ¹³⁷Cs source. After mutagenesis, late L4 and young adult stage hermaphrodites were picked to 60-mm agar plates seeded with Escherichia coli OP50, and their F₁ progeny were subsequently picked to fresh plates (30 per plate). Putative mutants expressing col-19::gfp were picked from the F₁ and F₂ generations.

Outcrossing of mutants:
Animals expressing col-19::gfp were outcrossed to assess the phenotype of each new mutation in the absence of lin-4 and veIs13. Mutants possessing a functional vulva were crossed to wild-type males, and their F₂ progeny were analyzed for mutant phenotypes. The remaining mutants, which were egg-laying defective (Egl), were outcrossed by one of two techniques. For the majority of the mutants, an artificial vulva was created by piercing the animal's cuticle near the vulva with a microinjection needle. The hermaphrodites were then mated with wild-type males, and non-Egl cross-progeny were identified. Crosses that did not yield non-Egl animals were followed to the F₂ generation to distinguish between failed crosses and lin-4 suppressors that caused a dominant Egl phenotype. In no case did we observe non-Egl animals in the F₂ but not in the F₁ generation. Non-Egl F₁ animals were allowed to self fertilize, and their progeny were screened for mutant phenotypes.

We reasoned that a subset of animals might be Egl due solely to the lin-4 mutation. Such mutants should become non-Egl upon transformation rescue with a lin-4(+) plasmid. We injected a lin-4 rescuing plasmid, pSAL6 (LEE et al. 1993 ) at 100 ng/µl into the syncytial gonads of mutant adult hermaphrodites. Resulting non-Egl hermaphrodite F₁ progeny rescued by the lin-4 transgene were crossed to wild-type males. Non-Egl F₂ cross-progeny were identified based on the segregation of mutant phenotypes in their self-progeny. Some of these lines could still bear the lin-4 rescuing array; however, less than 10% of lin-4-rescued F₁ animals transmitted the extrachromosomal array to their progeny, and therefore we had no trouble isolating F₂ heterozygotes that did not contain the lin-4(+) array. Five mutants (lin-4; veIs13 in combination with ve24, ve30, ve16, ve46, and ve37) were tested for rescue of the egg-laying defect by lin-4(+) injection. Of these, only the ve16 lin-4; veIs13 mutant was rescued for the egg-laying defect.

Mapping of new mutations and complementation tests:
The initial outcrossing step allowed many mutations to be assigned to linkage groups. X-linked mutations were identified by examining the F₁ males for phenotypic defects. Other mutations were assigned to LG II or LG V based on linkage to lin-4 or veIs13, respectively. The remaining mutations were mapped to chromosomes by the STS mapping procedure of WILLIAMS et al. 1992 or by standard two-factor crosses as described by BRENNER 1974 .

Complementation tests were performed when mutants resembled previously identified heterochronic mutants and mapped to the same chromosome. The egg-laying (EULING and AMBROS 1996 ) and male mating defects associated with lin-14 and lin-28 mutations are suppressed by development through the dauer pathway, enabling us to use post-dauer homozygous mutant animals in these crosses.

In complementation tests and in the construction of double mutants described in the following section, we scored the Lin phenotype of heterochronic mutants by morphological features. The following specific morphologies were scored as indicating a Lin phenotype: Lin-4(lf) and Lin-14(gf), long, vulvaless and lacking adult alae; Lin-14(lf) and Lin-28(lf), semi-dumpy (Dpy) and Egl with precocious vulval protrusions; Lin-29, Egl with vulval protrusions and lack of adult alae; Lin-42, semi-Dpy with precocious alae; and Lin-58, precocious adult alae.

lin-14: ve17, ve19, ve21, ve23, ve34 and ve35 each mapped to LG X and were thus candidate lin-14 alleles. lin-14(ma154ts) dpy-7; him-5 post-dauer males grown at 15° were crossed to post-dauer putative lin-14 hermaphrodites at 25°. Only Lin cross progeny were obtained from crosses to ve17, ve19, ve21, ve23 and ve35, and in each case the hermaphrodites segregated approximately 3/4 Lin and 1/4 Lin Dpy animals, indicating a failure to complement lin-14(ma154ts). ve34 homozygotes are unhealthy and are maintained over the translocation balancer szT1. Post-dauer lin-14(n179ts); him-5 males were mated to ve34/szT1 hermaphrodites at 25°. Lin and wild-type cross-progeny were obtained. All wild-type hermaphrodites contained the szT1 balancer and the majority of these were cross-progeny as evidenced by their segregation of lin-14(n179ts) animals. The remaining animals were Lin, indicating the failure of ve34 and lin-14(n179ts) to complement.

lin-28: ve24 and ve30 mapped to LG I. Post-dauer hermaphrodites from these strains were crossed to N2 males. Resultant heterozygous male progeny were mated to post-dauer lin-28(n719) dpy-5; him-5 hermaphrodites. Non-Dpy Lin and non-Dpy non-Lin cross-progeny were picked, and their progeny were scored. Each member of the non-Dpy Lin class segregated approximately 3/4 non-Dpy Lin and 1/4 Dpy Lin progeny, whereas the non-Dpy non-Lin class segregated 1/4 Dpy Lin and 3/4 non-Dpy non-Lin, indicating that ve24 and ve30 are lin-28 alleles.

lin-42: ve11, ve16, ve20 and ve25 mapped to LG II. Males of the genotype lin-42(n1089)/+ were mated to homozygous ve11 unc-4 hermaphrodites. Non-Lin non-uncoordinated (Unc) and Lin non-Unc cross-progeny were picked. The Lin non-Unc class segregated approximately 3/4 Lin non-Unc and 1/4 Lin Unc animals, whereas the non-Lin non-Unc class segregated 1/4 Lin Unc and 3/4 non-Lin non-Unc animals, indicating ve11 is a lin-42 allele. Similar crosses demonstrated that ve16, ve20 and ve25, fail to complement lin-42.

lin-42(ve20) was analyzed over the LG II deficiency ccDf11 (CHEN et al. 1992 ). Post-dauer lin-42(ve20); him-8 males were crossed to ccDf11/dpy-25 hermaphrodites. The dpy-25 allele causes a semi-Dpy phenotype when heterozygous and a Dpy phenotype when homozygous. Wild-type F₁ cross-progeny were not observed. lin-42/ccDf11 F₁ hermaphrodites were Lin as judged by alae synthesis during the third molt. In addition, approximately half of the F₁ males had malformed tails in which the tail spike was not fully retracted, rays were often missing, and spicules were either missing or short and crumpled. These males are probably ccDf11/lin-42 because similar male tail abnormalities are observed in lin-42(ve20); him-8 and lin-42(n1089); him-8 males. The remaining males, presumably lin-42/dpy-25, possessed well-developed tails, containing rays and spicules, and were semi-Dpy. A large percentage of lin-42/ccDf11 hermaphrodites were sterile. This phenotype may be due to a hypomorphic second mutation tightly linked to ve20 that is also uncovered by ccDf11.

lin-58: ve12 and ve33 are recessive suppressors of lin-4 and were tested for complementation in a lin-4 mutant background. lin-4/mnC1; veIs13 lin-58(ve12) hermaphrodites were mated with wild-type males. The resulting males were crossed to dpy-11 veIs13 lin-58(ve33) hermaphrodites and non-Dpy cross-progeny were picked. Of these, those animals segregating Lin-4 and no non-Rol animals were identified. These animals were of the genotype lin-4/+; veIs13 lin-58(ve12)/dpy-11 veIs13 lin-58(ve33). Non-Dpy Egl animals were picked from their progeny and were scored for adult alae synthesis and col-19::gfp expression. As adults, all of these animals had alae on their adult cuticles and expressed col-19::gfp. Approximately two-thirds of these animals gave rise to Dpy and non-Dpy progeny indicating they were of the genotype lin-4; veIs13 lin-58(ve12)/dpy-11 veIs13 lin-58(ve33) and that the ve12 and ve33 alleles fail to complement each other.

Both ve33 and ve12 map to LG V and are tightly linked to veIs13. To be certain that the ve12 and ve33 lesions are independent of the integrated array, we screened progeny of lon-3 unc-76/veIs13 ve33 animals for rare Lon non-Unc recombinants that failed to roll or express col-19::gfp. We identified such animals and they segregated Lon animals that maintained the heterochronic phenotype. A similar strategy was used to separate ve12 from veIs13.

lin-58(ve12) was analyzed over the LG V deficiencies yDf4 and yDf8 (DELONG et al. 1993 ). yDf4/dpy-11 unc-76 hermaphrodites were mated with veIs13 lin-58(ve12)/+ males and cross-progeny expressing col-19::gfp were picked. yDf4/veIs13 lin-58(ve12) animals were identified by the segregation of dead eggs and no Dpy Unc progeny. Animals from this strain were scored for precocious alae synthesis (see Table 2) during the L3 molt and their genotypes inferred by progeny testing.

yDf8/nT1 hermaphrodites were crossed to veIs13 lin-58(ve12)/+ males and non-Unc progeny expressing col-19::gfp were picked and scored for precocious alae synthesis during the L3 molt (see Table 2). The genotypes of these animals were confirmed by analysis of the subsequent generation.

lin-58 was mapped to the left of unc-76 in a three factor cross. Lon animals were picked from the progeny of lon-3 lin-58(ve12)/unc-76 heterozygotes. Thirteen of these animals contained a Lon Unc recombinant chromosome, and 10 of these chromosomes contained lin-58(ve12), placing lin-58 to the left of unc-76.

Construction of double mutants:
The double mutants used in the genetic epistasis analysis were constructed by the following crosses:

lin-42 lin-4: Males of genotype lin-4/mnC1 were mated to post-dauer lin-42 hermaphrodites. Hermaphrodites of genotype + lin-4/lin-42 + were identified, and Lin-42 and Lin-4 self-progeny were picked. Animals of the Lin-4 class were identified which gave progeny with adult alae, and these were used to establish the doubly mutant strain.

lin-42; lin-14(n536sd): Post-dauer dpy-6 lin-14(n536sd)/szT1 hermaphrodites were mated with post-dauer lin-42(ve20); him-8 males. Lin-14sd animals were picked, and animals of genotype dpy-6 lin-14/++; lin-42/+; him-8/+ were identified by their segregation of Lin-42 progeny. Non-Dpy Egl animals were picked and an animal of genotype dpy-6 lin-14/++; lin-42 was identified by progeny testing. dpy-6 was removed by recombination prior to making lin-14 homozygous.

lin-42 lin-29: Males of genotype lin-29/mnC1; him-5 were mated to post-dauer lin-42 hermaphrodites. Animals of genotype + lin-29/lin-42 +; him-5/+ were identified, and Lin-42 and Lin-29 progeny were picked. Only animals of the Lin-42 class gave rise to progeny of the opposite phenotype, and they were used to establish the doubly-mutant strain.

lin-4; lin-58: Males of genotype lin-4/mnC1 were mated to lin-58 hermaphrodites, and cross-progeny were identified by their segregation of lin-4 or mnC1 mutant animals. Lin-4 progeny were picked, and their progeny were scored for adult alae in order to identify the presence of lin-58.

lin-58; lin-14(n179ts): Post-dauer him-5; lin-14(n179ts) males were crossed to veIs13 lin-58(ve33) hermaphrodites, and the resultant transheterozygotes were allowed to self-fertilize at 25°. Lin-14 progeny that expressed col-19::gfp were identified and transferred to 15°. Animals with precocious alae were picked from their progeny to establish the lin-58; lin-14 strain.

lin-58; lin-14(n355n679ts): Males of genotype veIs13 lin-58(ve33)/++ were mated with post-dauer lin-14(n355n679ts) hermaphrodites. Resultant non-Lin transheterozygotes were picked by virtue of col-19::gfp expression and allowed to self-fertilize at 25°. Lin-14 animals were picked, and the Rol phenotype was used to homozygose the veIs13 lin-58 chromosome. Several independent lines were generated; each showed suppression of the retarded Lin-14 phenotype at 15°, confirming the presence of lin-58.

lin-58; lin-14(n536): Males of genotype veIs13 lin-58/++ were crossed to post-dauer dpy-6 lin-14(n536sd)/szT1 hermaphrodites. Rol Lin-14 transheterozygotes were picked, and a progeny of genotype dpy-6 lin-14/++; veIs13 lin-58 was identified on the basis of its segregation of Egl and non-Egl animals. dpy-6 was removed by recombination prior to making lin-14 homozygous.

lin-28; lin-58: Males of genotype veIs13 lin-58(ve33)/++ were crossed to post-dauer lin-28(ga54) or lin-28(ve24) hermaphrodites, and non-Lin transheterozygotes were picked on the basis of col-19::gfp expression. Lin-28 animals were picked from among the self-progeny of these animals, and the Rol phenotype was then used to homozygose the veIs13 lin-58 chromosome. Both strains exhibited an enhanced precocious phenotype not seen in lin-28; veIs13 animals.

lin-29; veIs13 lin-58: lin-29/mnC1; him-5 males were mated with veIs13 lin-58 hermaphrodites. An animal of genotype lin-29/+; him-5/+; veIs13 lin-58/++ was identified by its segregation of Lin-29 progeny. Lin-29 and Lin-58 animals were picked and allowed to self-fertilize. Only the Lin-58 class segregated animals of the opposite phenotype; these animals were used to establish the doubly mutant strain. The Rol phenotype produced by veIs13 was used to ensure homozygosity of the lin-58(ve33) mutant chromosome.

lin-42; lin-58: Males of genotype veIs13 lin-58(ve33)/++ were mated with lin-42 hermaphrodites. Hermaphrodites expressing col-19::gfp were picked and allowed to self-fertilize. Lin-42 animals expressing col-19::gfp were picked, and in all cases segregated animals that were egg-laying defective, unlike either parent. These animals were homozygous for the veIs13 lin-58 chromosome, as judged by the segregation of veIs13.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Construction of a col-19::gfp reporter gene and its control by the heterochronic gene pathway:
We set out to isolate heterochronic mutants in a genetic screen that detects temporal misexpression from a cuticle collagen gene promoter. We constructed a translational fusion gene between col-19 and the GFP gene (col-19::gfp; see MATERIALS AND METHODS) and took advantage of the ability to use the GFP as a reporter molecule in live animals (CHALFIE et al. 1994 ). The col-19 fragment we used was shown previously to program adult-specific, heterochronic gene-dependent expression of a lacZ reporter gene (LIU et al. 1995 ) and is probably a direct target of LIN-29 (ROUGVIE and AMBROS 1995 ). We generated a transgenic strain bearing an integrated array of col-19::gfp and rol-6(su1006sd) (pRF4; MELLO et al. 1991 ). The Rol phenotype conferred by rol-6(su1006sd) provides an assay for the array even in genetic backgrounds where the col-19 promoter is inactive (see MATERIALS AND METHODS). We refer to the integrated col-19::gfp + pRF4 array as veIs13. Comparison of Figure 1A and Figure B demonstrates that the col-19::gfp fusion is under tight temporal regulation. When veIs13 is placed in a wild-type background, GFP is detected in the hypodermis of adult animals (Figure 1A) but is not detected in embryos or larvae.

View larger version (88K): In this window In a new window Download PPT slide   Figure 1. col-19::gfp expression is adult-specific in wild-type animals and is regulated by the heterochronic gene pathway. In each pair of panels, a fluorescence micrograph is on the left and the same field viewed by Nomarski optics is on the right. In this and subsequent figures, anterior is to the left and ventral is down. Bar, 100 µm. (A and B) GFP expression in veIs13 animals. col-19::gfp is expressed in adults but not larvae or eggs. The twisting of the hypodermis is a consequence of the rol-6(su1006sd) gene present in the integrated veIs13 array as a marker. (C and D) Adult stage lin-4(e912); veIs13 animals. col-19::gfp expression is not detected in the lin-4 background. (E and F) L3 molt stage veIs13; lin-14(n179ts) at 25° demonstrating precocious col-19::gfp expression. (G and H) L3 molt stage lin-28(n947); veIs13. col-19::gfp is expressed precociously. (I and J) "L5" stage lin-29(n836); veIs13. By sexual maturity this animal is an adult; however, it is referred to as an L5 stage larvae to reflect the failure of the hypodermis to terminally differentiate. col-19::gfp expression is not observed in the hypodermis of lin-29 mutants, although GFP is occasionally detected in a few cells in the vulval region.

The col-19::gfp fusion is regulated by the heterochronic gene pathway. Loss-of-function mutations in lin-14 or lin-28 cause precocious hypodermal accumulation of LIN-29 (BETTINGER et al. 1996 ), which results in precocious seam cell terminal differentiation (AMBROS and HORVITZ 1984 ; AMBROS 1989 ). lin-14(n179ts); veIs13 animals express the col-19::gfp fusion gene one stage early, beginning during the third molt, when reared at the restrictive temperature (25°) (Figure 1E). Similarly, lin-28(n947); veIs13 animals express the col-19::gfp fusion gene during the third molt (Figure 1G). In contrast, col-19::gfp is not expressed in lin-4 and lin-29 mutants (Figure 1C and Figure I), in agreement with the failure of these retarded mutants to accumulate LIN-29 in the hypodermis (BETTINGER et al. 1996 ) and their failure to execute the seam cell terminal differentiation program (CHALFIE et al. 1981 ; AMBROS and HORVITZ 1984 ). This analysis demonstrates that the col-19::gfp fusion is regulated by the heterochronic gene pathway, by lin-29 in particular, consistent with the identification of the col-19 promoter as a target of LIN-29 (ROUGVIE and AMBROS 1995 ). Assessment of col-19::gfp expression thus provides a convenient in vivo assay for lin-29 activity. A mutagenesis strategy based on identification of temporally altered col-19::gfp expression patterns should permit isolation of heterochronic mutants.

Identification of mutants:
We screened progeny of mutagenized lin-4; veIs13 animals to identify heterochronic mutations that restore col-19::gfp expression, in a background of animals with no expression. Mutant animals isolated in these screens should contain lin-4 suppressor mutations that allow the hypodermal requirement for lin-4 to be bypassed. Since the lin-4 mutation we used is a deletion of the lin-4 locus (LEE et al. 1993 ), intragenic revertants should not be obtained.

The design of the screen allows isolation of dominant or recessive suppressors of lin-4 on any chromosome. We screened ~30,000 haploid genomes mutagenized with EMS and isolated 20 mutations that restore col-19::gfp expression (Table 1; Figure 2). We screened an additional 6,400 haploid genomes mutagenized by -irradiation and isolated six additional mutations. Although several animals expressing the col-19::gfp fusion were picked in the F₁ generation, they were either sterile or produced progeny that failed to express col-19::gfp. The mutants described here were each isolated from the F₂ generation, and were recessive with respect to restoration of col-19::gfp expression in the lin-4 background.

View larger version (119K): In this window In a new window Download PPT slide   Figure 2. Representative mutants as isolated. In each pair of panels, a fluorescence micrograph is shown on the left and the same field viewed by Nomarski optics is on the right. All animals shown are sexually mature and contain fertilized eggs. (A and B) lin-4; veIs13; lin-14(ve23). (C and D) lin-28(ve33); lin-4; veIs13. (E and F) lin-42(ve25) lin-4; veIs13. (G and H) lin-4; veIs13 lin-58(ve33). (I and J) lin-4; veIs13; ve37. Bar, 100 µm.

The phenotypes of the mutants as isolated, i.e., with the new mutation in the presence of lin-4 and veIs13, are listed in Table 1. For clarity in Table 1 and the following discussion, gene names, where known, have been included. The justification of these assignments appears in subsequent RESULTS sections. The levels of GFP expression resulting from these new mutations varied from low and variable (e.g., strains containing ve38 or ve42), to levels indistinguishable from that observed with veIs13 in a wild-type background (e.g., strains containing lin-14(ve23), lin-14(ve34), or ve37). We examined each mutant for evidence of more complete restoration of seam cell terminal differentiation by scoring for the presence of adult alae on the cuticle. lin-4 mutants can undergo supernumerary molts and additional larval stages (L5, etc.; CHALFIE et al. 1981 ), necessitating the examination of cuticles for alae during the fourth and fifth molts as appropriate. Adult alae were detected on the cuticles of 19/26 mutant strains (Table 1). Overall, the restoration of alae synthesis varied in penetrance; in each strain we observed some animals with incomplete alae, indicating a failure of certain seam cells to terminally differentiate. In the majority of mutant strains, alae synthesis occurred during the L4-to-adult molt, as in wild-type animals (Table 1). However, three mutations, [lin-14(ve17), lin-14(ve21) and lin-14(ve35)], produced a highly penetrant precocious phenotype in the retarded lin-4 mutant background, with adult alae synthesis during the third molt.

Fourteen mutants retained the supernumerary molting phenotype associated with lin-4 mutations (Table 1). These mutants can be divided into three classes based on the percent of L5 stage animals that show alae synthesis. In Class I mutants, which contain the alleles lin-58(ve12), lin-58(ve33) and ve38, adult alae were detected on greater than 75% of the animals. In Class II mutants, containing the alleles ve36, ve37, ve40 and ve46, a relatively low percent (25–40) of animals synthesized alae even during the fifth molt. These alleles are weak suppressors of the lin-4 hypodermal defect. Finally, seven mutants, containing the alleles ve15, ve39, ve41, ve42, ve43, ve44 and ve45, comprise Class III. These mutants fail to synthesize an adult-type cuticle even during the L5-to-L6 molt, despite the observation that two of them, those containing the ve41 and ve44 alleles, express col-19::gfp at moderate levels and with high penetrance. The Class III mutants could contain very weak mutant alleles of heterochronic genes, mutations in minor players in the pathway, or mutations that allow col-19::gfp expression independent of the heterochronic pathway. These Class III mutations have not been analyzed further.

lin-4 mutants exhibit retarded defects in lineages other than the lateral hypodermal seam (CHALFIE et al. 1981 ; AMBROS and HORVITZ 1984 ). For example, vulva precursor cell divisions are usually blocked or delayed. As a result, the animals are vulvaless (the Vul phenotype) and unable to lay eggs (CHALFIE et al. 1981 ; EULING and AMBROS 1996 ). Five mutations, lin-14(ve17), lin-14(ve19), lin-14(ve21), lin-14(ve23), and lin-14(ve35) suppressed the lin-4 vulval cell lineage defect to the extent that egg-laying was restored (Table 1). Several other mutations resulted in partial suppression of the Vul phenotype (Table 1). In these animals (e.g., lin-14(ve34) ve37 and lin-42(ve11), some vulva precursor cells divide, but proper vulvae fail to form and egg-laying defective animals with ventral protrusions often result.

Development through the alternative L3 larval stage, the dauer larva, occurs in response to lack of food, overcrowding and high temperatures (CASSADA and RUSSELL 1975 ). Dauer formation has been shown previously to depend on the heterochronic gene pathway, and, in particular, the initiation of dauer formation requires lin-4-mediated downregulation of lin-14 (LIU and AMBROS 1989 ). Dauer larvae can be selected from starved populations by virtue of their resistance to 1% sodium dodecyl sulfate (SDS) (CASSADA and RUSSELL 1975 ). We tested whether any of the mutations restored dauer forming ability to lin-4 animals. Five mutations suppressed the lin-4 dauer formation defect, based upon the formation of SDS-resistant animals that were morphologically dauer larvae, as judged by microscopic examination (Table 1). These five mutations are the same five that suppressed the lin-4 egg-laying defect, and as described below, all five proved to be new lin-14 alleles.

Separation of new mutations from lin-4 and genetic mapping:
The 19 mutants which exhibited suppression of the lin-4 alae defect were outcrossed to separate the new mutations from veIs13 and lin-4(e912) (see MATERIALS AND METHODS) to assess whether the mutations produce a phenotype alone. Fourteen of the outcrossed mutations cause precocious seam cell terminal differentiation, as judged by the synthesis of adult alae during the third molt. These mutations were mapped, and complementation tests were performed as described below and in MATERIALS AND METHODS. We did not detect a phenotype associated with ve36, ve37, ve38, ve40 or ve46 on their own, and consequently we have not yet pursued the analysis of these five apparently silent suppressors.

New alleles of lin-14 and lin-28:
Five mutations, ve17, ve19, ve21, ve23, and ve35, map to the X chromosome and exhibit phenotypic characteristics similar to lin-14(lf) alleles (AMBROS and HORVITZ 1984 ). These animals synthesize alae during the third molt, are semi-Dpy, Unc, often contain one or two vulval protrusions (Pvul) and are Egl. The vulval and lateral hypodermal defects caused by loss-of-function lin-14 mutations have been shown previously to be largely suppressed by development through the dauer pathway (EULING and AMBROS 1996 ; LIU and AMBROS 1991 ). Post-dauer suppression of these defects is also observed for animals bearing the ve17, ve19, ve21, ve23 and ve35 alleles (data not shown). ve34 also maps to LG X and produces a precocious alae phenotype, but it fails to suppress the Egl and dauer defective phenotypes of lin-4. All six of these mutations failed to complement lin-14(ma154ts) or lin-14(n179ts) at the restrictive temperature (see MATERIALS AND METHODS). The six new lin-14 mutations varied with respect to their suppression of the lin-4 alae defect (Table 1). The alleles ve17, ve21 and ve35 were epistatic to lin-4 in this respect; seam cells of the double mutant terminally differentiated precociously, during the third molt, as do animals bearing lin-14 null mutations. The other three lin-14 alleles, ve19, ve23 and ve34, exhibited mutual suppression in combination with lin-4. While they restored seam cell terminal differentiation to lin-4 animals, they did not cause it to occur precociously (Table 1). We suggest that ve17, ve21 and ve35 are probably null alleles of lin-14, and that ve19, ve23 and ve34 are likely to be hypomorphic alleles.

Two mutations, ve24 and ve30, map to LG I and produce phenotypes which resemble lin-28(lf) mutants (AMBROS and HORVITZ 1984 ). Seam cell terminal differentiation occurs during the third molt in these mutants, they are semi-Dpy, have vulval protrusions and are unable to lay eggs. Both of these mutations were shown to be new lin-28 alleles by their failure to complement lin-28(n719) (see MATERIALS AND METHODS).

The isolation of new alleles of lin-14 and lin-28, well characterized heterochronic genes known to be suppressors of lin-4, underscores the utility of this screen for isolating mutations that affect terminal differentiation of the hypodermis.

Multiple alleles of lin-42:
We have isolated four alleles of lin-42, a heterochronic gene for which a single allele had been described (LIU 1990 ). These mutations, ve11, ve16, ve20 and ve25, map to LG II and fail to complement the original lin-42 allele, n1089.

The phenotypes associated with the five lin-42 alleles do not differ dramatically (Table 2). As observed through the dissecting microscope, lin-42 mutants have a variable semi-Dpy appearance. Nomarski microscopy reveals that some seam cells in the lin-42 mutants terminally differentiate one stage early, during the third molt (Figure 3; Table 2). Instead of dividing and synthesizing larval-type cuticle, these seam cells exit permanently from the cell cycle, and synthesize adult-type cuticle containing lateral alae (Figure 3A and Figure B). ve16, ve20 and n1089 are the strongest alleles with respect to adult alae synthesis during the third molt (Table 2), while ve25 and ve11 appear less penetrant in this respect. To assess the timing of seam cell fusion in these mutants we visualized the outlines of hypodermal cells in fixed animals with the antibody MH27, which recognizes an antigen in adherens junctions (FRANCIS and WATERSTON 1991 ). Figure 3C shows that seam cell fusion occurs precociously in lin-42 mutants. Together, these results show that wild-type lin-42 activity is required to prevent seam cell terminal differentiation during the L3-to-L4 molt.

View larger version (126K): In this window In a new window Download PPT slide   Figure 3. Phenotypic characterization of lin-42 mutants. (A) Nomarski micrograph of the left lateral cuticle of an L3-to-L4 molt stage lin-42(ve16) hermaphrodite. (B) Nomarski micrograph of the anterior gonad arm of animal shown in (A). The animal was executing a molt as judged by lack of pharyngeal pumping and cuticle morphology at the nose (SINGH and SULSTON 1978 ) and the gonad had just reflexed (white arrow), indicative of the L3-to-L4 molt. (C) Fluorescence micrograph of the left lateral seam of a wild-type L3 molt stage hermaphrodite stained with the antibody MH27, to illustrate the individual cells of the unfused lateral seam. (D) Hoechst staining of animal in (C) to demonstrate stage. White arrow indicates the tip of a just reflexed gonad arm. (E) An L3 molt stage lin-42(ve20) hermaphrodite stained with the antibody MH27. The cells of the lateral seam have fused precociously. (F) Hoechst staining of the gonad of the animal shown in (C) to illustrate stage. (G) Fluorescence micrograph of mid-L4 stage lin-42(ve20) showing precocious col-19::gfp expression. (H) Nomarski micrograph of animal shown in (E). Bars, 20 µm.

Although lin-42 mutants execute seam cell terminal differentiation during the third molt, they fail to exit the molting cycle; instead, lin-42 mutant animals enter a fourth molt and synthesize a new cuticle also containing adult alae. lin-42 mutants often fail to execute fully this final L4-to-adult molt. Many animals appear to be unable to shed the fourth stage cuticle, and they eventually burst after accumulating fertilized eggs. Those animals that do complete the fourth molt are able to lay eggs. The continuation of molting cycles is observed in strains bearing each of the five lin-42 alleles and, as described in more detail below, several of these alleles are likely to be null. In contrast, strong loss-of-function alleles for the other precocious mutants, lin-14 and lin-28, undergo only three molts (AMBROS and HORVITZ 1984 ; AMBROS 1989 ).

In addition to the precocious alae phenotype, a vulval protrusion is observed at very low penetrance in all four lin-42 mutants despite repeated outcrosses (Table 2). The penetrance of the vulval phenotype is near 1% but is difficult to quantify due to the tendency of lin-42 mutants to form bags-of-worms as a result of failure to shed the L4 stage cuticle during the final molt. Whether this low penetrant vulval defect is due to precocious vulval cell divisions is not known.

As observed for lin-14 and lin-28 mutants, the precocious seam cell terminal differentiation phenotype of lin-42 mutants is suppressed by development through the dauer stage (Table 2). Additionally, lin-42 mutants are sensitive to environmental conditions that induce dauer formation. We have observed phenotypic suppression of the precocious alae defect in animals that did not form dauer larvae but that developed in conditions of low or restricted food supply (on crowded plates or in a bag-of-worms) or at high temperature (25°). These observations are in agreement with the work of LIU 1990 , who showed that lin-42(n1089) animals were phenotypically suppressed by development through the distinctive predauer stage (L2d); suppression did not require complete dauer larva formation. Thus, hypersensitivity to environmental conditions is a distinctive property of lin-42, and as a consequence, our analyses of continuous development in lin-42 single and double mutant combinations have been performed on animals that hatched from eggs into a lawn of E. coli and developed on uncrowded plates at 20°.

We compared the phenotypes of animals homozygous for lin-42(ve20), a -ray induced allele, to lin-42(ve20)/ccDf11 hemizygotes (see MATERIALS AND METHODS). The transheterozygotes are viable and resemble lin-42(ve20) homozygotes in phenotype. lin-42(ve20)/ccDf11 hermaphrodites synthesize alae during the third molt and are semi-Dpy as adults. The extent of alae synthesis in lin-42(ve20)/ccDf11 animals resembled that of lin-42(ve20) homozygotes; although the majority of seam cells terminally differentiated during the third molt, some did not, and as a result, gaps were detected in the alae (Table 2). Moreover, lin-42/ccDf11 animals did not become more precocious; alae synthesis was not observed during the second molt. The precocious alae phenotypes produced by the ve16 and n1089 alleles are similar to ve20 mutants, and thus the phenotype produced by these alleles likely represents the null, or nearly null, phenotype of lin-42.

lin-42 epistasis analysis:
The lin-42 mutations were separated from lin-4, outcrossed a minimum of three additional times and, along with n1089, were reconstructed as double mutants with lin-4. The hypodermal phenotype of each lin-42 lin-4 double mutant is less severe than either single mutant from which it was constructed (Table 2). While seam cell terminal differentiation is precocious in lin-42 mutants and retarded in lin-4 mutants, it usually occurs with the correct temporal specificity in the lin-42 lin-4 double mutants. Thus, lin-4 and lin-42 mutations mutually suppress each other with respect to their alae defects. These results suggest that there is not a simple linear regulatory relationship between lin-4 and lin-42 with respect to hypodermal terminal differentiation. Other aspects of the lin-4 phenotype are not suppressed by lin-42 mutations. The reconstructed lin-42 lin-4 double mutants are vulvaless and unable to lay eggs (Table 2), as are lin-4 mutants. In addition, lin-42 does not suppress the dauer formation defect of lin-4 mutants.

We tested whether the precocious execution of seam cell terminal differentiation in lin-42 mutants requires lin-29 activity. We constructed lin-42(ve11); lin-29(n836) and lin-42(ve20); lin-29(n836) double mutant animals and scored for adult alae synthesis during the L4-to-adult molt. Adult alae synthesis was not observed in the double mutants. The seam cells continued to divide, and the animals underwent supernumerary molting cycles characteristic of lin-29 mutants (AMBROS and HORVITZ 1984 ). This result suggests that in wild-type animals lin-42 acts, directly or indirectly, to inhibit lin-29 activity during the third molt and thereby prevents early seam cell terminal differentiation.

These analyses place lin-42 in the heterochronic gene pathway upstream of lin-29, a position shared with lin-14 and lin-28. At present, these three genes cannot be ordered unambiguously because their loss-of-function phenotypes are the same with respect to the time of alae synthesis. There are gain-of-function lin-14 alleles (AMBROS and HORVITZ 1984 ; WIGHTMAN et al. 1991 ), that produce a retarded hypodermal phenotype similar to lin-4(lf) mutations, with no alae synthesis during the fourth molt. To test whether lin-42 alleles could suppress a gain-of-function lin-14 phenotype, we analyzed lin-42(ve20); lin-14(n536sd) double mutants. As in the case of lin-42 lin-4 double mutants, we observed mutual suppression of the hypodermal defects in these mutants. The lin-42(ve20) mutation restores terminal differentiation to the fourth molt in the lin-14(n536sd) mutant background (Table 2). Ninety percent of seam cells examined in L4 stage lin-42(ve20); lin-14(n536sd) double mutants synthesized alae. Thus, the retarded phenotype of lin-14(gf) requires lin-42 activity, and the precocious phenotype of lin-42 mutants requires downregulation of lin-14 activity.

Identification of a new heterochronic gene:
Two of the new mutations, ve33 and ve12, define a novel member of the heterochronic gene pathway. Both map to the same region of LG V, to a position where no heterochronic genes have been previously placed (see below). ve33 suppresses lin-4 such that most double mutant animals express col-19::gfp at moderate levels and synthesize adult alae at the L4-to-adult molt (Table 1). ve33 does not suppress the execution of supernumerary molts that is seen in lin-4 mutant animals nor does it appear to alter the Vul or Daf phenotypes of lin-4 mutants. The suppression of lin-4 by ve12 resembles lin-4 suppression by ve33, but is weaker with respect to suppression of the alae defect (Table 1 and Table 2).

When removed from the lin-4 background, ve33 and ve12 produce a precocious alae phenotype (Table 2). Alae are detected one stage early, on cuticles of animals in the L3-to-L4 molt (Figure 4A). The alae appear less distinct than the alae observed in loss-of-function lin-14, lin-28, or lin-42 mutants and often do not extend the entire length of the worm. We assessed the temporal specificity of seam cell fusion in ve33 and ve12 animals using the MH27 antibody and found that seam cell fusion occurs precociously in both mutants (Figure 4C). We quantitated this defect for the stronger allele, ve33, and found that 57% of seam cells (n = 835) fuse during the third molt. In addition, every animal examined contained at least one pair of precociously fused cells. We also monitored division of the seam nuclei in ve33 and ve12 animals. In ve33 animals, approximately 80% of seam nuclei in cells with precocious alae failed to divide by the end of the third molt (n = 9). The remainder did divide, but only after an ~30-min delay relative to the other seam nuclei. The ve12 allele appeared weaker relative to this phenotype: 38% failed to divide by the end of the molt, 43% divided 30–45 min late and 14% divided approximately at their normal time (n = 21). Thus, these alleles appear weaker than the previously described precocious mutants (AMBROS and HORVITZ 1984 ) with respect to premature exit from the nuclear division cycle in the seam.

View larger version (101K): In this window In a new window Download PPT slide   Figure 4. Characterization of lin-58(ve33). (A) Nomarski micrograph of the left lateral cuticle of an L3-to-L4 molt stage lin-58(ve33) hermaphrodite illustrating precocious alae formation (black arrows). (B) Nomarski micrograph of the posterior gonad arm of animal shown in (A) to illustrate stage. (C) Fluorescence micrograph of left lateral seam of a lin-58(ve33) hermaphrodite stained with the antibody MH27 to demonstrate precocious seam cell fusion. (D) Hoechst staining showing the gonad of the animal in (C) to illustrate stage. In B and D the white arrow indicates the tip of the developing gonad which has just reflexed, indicative of the L3-to-L4 molt stage. Bars, 20 µm.

The ve12 allele appears recessive to wild type. In contrast, the ve33 allele is semi-dominant; precocious alae are detected on the cuticles of ve33/+ animals with a frequency and appearance similar to that of ve33 homozygotes (Data not shown). However, ve33 appears recessive in its suppression of lin-4. Adults of genotype lin-4; ve33 veIs13/++ lack alae and do not express col-19::gfp. We were thus able to test for complementation between ve33 and ve12 in a lin-4 mutant background. Adults of genotype lin-4; ve33 veIs13 dpy-21/ve12 veIs13 + express col-19::gfp and synthesize adult alae. The failure of ve33 and ve12 to complement each other, and the mapping data described below indicate that ve33 and ve12 are allelic.

Recombination mapping experiments place ve12 and ve33 in the lon-3-unc-76 interval (see Figure 5). There is not a candidate gene previously mapped to this interval likely to correspond to the ve12 and ve33 mutations. The gene defined by these alleles has been named lin-58 to reflect the cell lineage defect resulting from precocious terminal differentiation of seam cells.

View larger version (6K): In this window In a new window Download PPT slide   Figure 5. Genetic map position of lin-58. The lin-58 gene is located in the ~2.9 map unit interval between lon-3 and unc-76 on chromosome V in the region indicated by the dashed line. The chromosomal deficiency yDf8 fails to complement lin-58(ve12), indicating that lin-58 is to the right of him-5. yDf4 complements lin-58(ve12).

ve33 and ve12 could be loss-of-function or gain-of-function alleles of lin-58. We suspect that they are loss-of-function alleles based primarily on their frequency of isolation. Gain-of-function alleles tend to be induced at a much lower frequency than are loss-of-function alleles, and yet we isolated independently two alleles of lin-58 in our EMS screens. This frequency is equal to that with which we recovered EMS-induced loss-of-function lin-28 and lin-42 alleles. Thus, one possibility is that ve12 and ve33 are loss-of-function alleles and that the lin-58 locus is haploinsufficient. Alternatively, ve33 may encode a mutant product that acts in a dominant negative manner, for example binding to interaction partners and rendering them inactive.

We tested whether deficiencies in the unc-76 region uncover lin-58 by placing them in trans to ve12 and examining the phenotype of the heterozygotes. yDf4 and yDf8 both break in the unc-76-dpy-21 interval. yDf4 extends rightward removing dpy-21, while yDf8 removes unc-76 and extends leftward through unc-61 (DELONG et al. 1993 ) (Figure 5). ve12/yDf4 animals resemble ve12/+ animals and precocious alae are not detected indicating that the yDf4 chromosome is lin-58(+) (Table 2). In contrast, adult alae are detected during the third molt of ve12/yDf8 animals, indicating that yDf8 removes at least a portion of the lin-58 locus. The penetrance of the precocious alae defect in yDf8/ve12 animals resembles that observed in ve12 homozygotes. We did not detect precocious alae on yDf8/+ animals, arguing against a model involving haploinsufficiency of the lin-58 locus.

An alternative possibility is that ve33 and ve12 are gain-of-function lin-58 alleles, and a single copy of either allele is insufficient to suppress the lin-4 mutant phenotype. Although we favor the model that they are loss-of-function alleles, further work is necessary to determine the precise nature of these alleles.

lin-58 epistasis analysis:
Although the lin-58 alleles are likely to be loss-of-function alleles, and ve12 behaves as a null, or nearly null allele when placed in trans to a deficiency, the actual degree of loss of gene function is not known with certainty. With this qualification in mind, we examined the epistasis relationships of lin-58 with the previously identified heterochronic genes. We performed most of these analyses with both ve33 and ve12 and found that they behave similarly.

We reconstructed lin-4; lin-58 double mutants with multiply-outcrossed lin-58 alleles and quantitated the extent of alae formation. We found that 10–13% of seam cells in lin-4; lin-58 double mutant combinations terminally differentiate during the L4 molt (Table 2). lin-58(ve33) and lin-58(ve12) mutations thus behave as weak suppressors of lin-4 and partially restore the temporal specificity of alae synthesis to the L4-to-adult molt.

To test if the precocious phenotype of lin-58 mutants depends on lin-29 activity, we constructed lin-29; lin-58 double mutants and assayed for seam cell terminal differentiation. We found that these double mutants resemble lin-29 mutants; their seam cells fail to terminally differentiate during the fourth molt (Table 2). Thus, lin-29 is epistatic to lin-58(ve33) and lin-58(ve12), indicating that in wild-type animals lin-58 acts through lin-29 to prevent precocious alae synthesis at the L3-to-L4 molt.

To assess whether lin-58 mutations can suppress a lin-14 gain-of-function lesion, we examined lin-58(ve33); lin-14(n355n679ts) animals. This lin-14 allele contains two lesions. It contains a deletion of the lin-4 RNA binding sites in the lin-14 3'UTR (WIGHTMAN et al. 1991 ) and is therefore insensitive to lin-4 down regulation. In addition, it contains a temperature sensitive loss-of-function mutation. Thus, lin-14(n355n679ts) animals exhibit a loss-of-function phenotype at the nonpermissive temperature (25°) and a weak gain-of-function phenotype at the permissive temperature (16°). As a control we examined alae synthesis in lin-14(n355n679ts) animals at 16° and 25°. We observed that 72% of seam cells terminally differentiate precociously when animals were raised at 25°. In contrast, precocious alae were not observed on lin-14(n355n679ts) animals reared at 16°, and 46% of seam cells in these animals exhibited a retarded phenotype, failing to terminally differentiate during the fourth molt (Table 2). The behavior of seam cells in lin-58(ve33); lin-14(n536n679ts) double mutants grown in parallel differed dramatically. At 16°, 39% of seam cells terminally differentiated precociously as judged by alae synthesis at the third molt, and 99% of seam cells of animals in the fourth molt were terminally differentiated. Thus, lin-58 mutations can suppress a weak lin-14 gain-of-function allele.

Similar results were obtained with the lin-14(n536sd) allele, a gain-of-function allele that deletes five of the seven lin-4-complementary sites in the lin-14 3'UTR and thereby produces a strong retarded phenotype (AMBROS and HORVITZ 1984 ; WIGHTMAN et al. 1991 ). In contrast to the failure of seam cells to terminally differentiate in n536 mutants, 98% and 88% of seam cells in lin-58(ve33); lin-14(n536) and lin-58(ve12); lin-14(n536) double mutants, respectively, synthesize adult alae during the fourth molt. Thus, the inhibition of seam cell terminal differentiation caused by the lin-14 gain-of-function allele is blocked by the ve33 and ve12 mutations, indicating that the retarded alae phenotype of lin-14 gain-of-function animals requires lin-58(+).

Genetic interactions of lin-58 with lin-14, lin-28 and lin-42 with respect to hypodermal function:
We analyzed the phenotypes produced by lin-58 in double mutant combinations with lin-14(lf), lin-28 and lin-42. These combinations of mutations cannot be assayed for epistatic relationships with respect to the time of alae synthesis because they do not cause opposing phenotypes. Instead, we examined the double mutants for novel phenotypes and for enhancement of precocious seam cell terminal differentiation. We detected enhanced alae defects in all three mutant combinations.

The most striking enhancement of alae defects was observed in lin-28; lin-58 double mutants. Terminal differentiation of the seam cells does not occur until the third molt in lin-58 mutants. In strong lin-28 mutants, seam cell terminal differentiation is occasionally observed during the L2 molt (AMBROS 1989 ). For example, we found that 32% of seam cells in animals bearing lin-28(ga54), a probable null allele (MOSS et al. 1997 ), terminally differentiate during the second molt. The percent of alae synthesis during the second molt increases nearly threefold, to 91%, in the lin-28(ga54); lin-58(ve33) double mutant (Table 2). A similar enhancement of the percent seam cells synthesizing alae at the L2 molt is seen in lin-28(ve24); lin-58(ve33) and lin-28(ga54); lin-58(ve12) double mutants (Table 2). Thus, the simultaneous loss of lin-28 and lin-58 produces a more severe precocious alae phenotype than does loss of either gene alone.

We also observed an enhancement of the loss-of-function phenotype of lin-14 in two genetic backgrounds. At the nonpermissive temperature of 25°, 100% of seam cells terminally differentiated during the third molt in lin-58(ve33); lin-14(n536n679ts) animals, relative to 72% in lin-14(n355n679ts) and 70% in ve33 animals. In lin-14(n179ts) animals at the permissive temperature of 16°, only 4% of seam cells terminally differentiate precociously. In contrast, 93% of seam cells in lin-58(ve33); lin-14(n179ts) animals synthesize alae during the third molt when reared at 16°.

We constructed lin-42; lin-58 double mutants to assess the hypodermal phenotype resulting from concomitant loss of these two gene activities. Precocious seam cell terminal differentiation in lin-42(ve20), lin-42(n1089) and lin-58(ve33) animals is observed with frequencies of 89, 96 and 70%, respectively. The lin-42(ve20); lin-58(ve33) and lin-42(n1089); lin-58(ve33) double mutants exhibit a fully penetrant L3 molt alae phenotype, with alae that appear distinctly formed as in lin-42 mutant animals. The penetrance of this phenotype is slightly higher than that observed in lin-42 single mutants, suggesting a mild enhancement of the phenotype.

Genetic interaction of lin-42 and lin-58 in vulva formation:
In addition to the rather mild enhancement of the hypodermal phenotype observed in lin-42; lin-58 double mutants, examination of these animals revealed a striking vulval defect. lin-58 mutants lay eggs normally, and their vulvae appear wild-type. lin-42 mutants, when they shed their cuticles properly at the L4-to-adult molt, also have apparently wild-type vulvae. Surprisingly, lin-42(ve20); lin-58(ve33) mutants are unable to lay eggs, and 86–94% have a ventral protrusion at the site of vulva formation (Figure 6). We refer to this phenotype as a synthetic abnormally everted vulva phenotype, or SynEvl. The SynEvl phenotype is not allele specific. It is observed in all three mutant combinations tested, involving multiple alleles of each gene [(n1089; ve33); (ve20; ve33) and (ve20; ve12); see Table 2; Figure 6].

View larger version (129K): In this window In a new window Download PPT slide   Figure 6. Synthetic everted vulval phenotype of lin-42; lin-58 double mutants. Nomarski micrographs of the midventral region of adult animals are shown. A functional vulva (white arrowheads) forms in (A) lin-42(ve20) and (B) lin-58(ve33) animals. In contrast, 100% of (C) lin-42(ve20); lin-58(ve33) animals are egg-laying defective and 92% develop vulval protrusions as shown (white arrow). Bar, 20 µm.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

We developed a screen to identify additional temporal regulators of lateral hypodermal cell terminal differentiation. Our goal was to screen large numbers of mutagenized animals efficiently for a relatively subtle phenotype, the difference between larval and adult hypodermis. At the level of light microscopy, the stage-specific distinction in the hypodermis is manifested as a morphological difference in the overlying cuticle. The adult cuticle contains alae, a set of lateral ridges that extend the length of the animal and which are best viewed by Nomarski microscopy (see Figure 3A). The adult alae are commonly used as an indicator of hypodermal seam cell terminal differentiation. To circumvent the need to score the presence or absence of adult alae as a primary screening criterion, we designed a col-19::gfp fusion gene that is expressed adult-specifically in wild-type worms (Figure 1A and Figure B), is under heterochronic gene control (Figure 1, C–J), and can be easily assayed in live animals using a dissecting microscope equipped with fluorescence optics.

We describe a set of mutations identified solely on the basis of their ability to restore hypodermal expression of col-19::gfp in a genetic background where it has been inactivated by preventing hypodermal terminal differentiation with a mutation in lin-4. There have been two major outcomes of our screens to date. First, they have focused our attention on lin-42 as an important member of the heterochronic gene pathway. Second, they have resulted in the identification of a new heterochronic gene, lin-58.

Isolation and characterization of multiple lin-42 alleles:
Among the 26 mutations isolated and characterized are 12 new alleles of the previously identified heterochronic genes, lin-14, lin-28 and lin-42. Of these, the four new lin-42 alleles are particularly important, as only a single allele of lin-42 had been described previously (LIU 1990 ). The measured frequency for loss- or reduction-of-function alleles for an average gene in C. elegans is approximately 5 x 10^-4 per EMS-mutagenized gamete (ANDERSON 1995 ). Our identification of two EMS-induced lin-42 alleles from ~30,000 haploid genomes screened (6.7 x 10^-4/gamete) is consistent with this frequency. Moreover, the lin-42 allele isolation frequency is identical to the frequency with which the same EMS-based screen yielded loss-of-function lin-28 alleles. These observations and the fact that all five lin-42 alleles produce similar phenotypes, suggest that the lin-42 alleles are loss-of-function and that the original allele, n1089, is not a special allele.

In wild-type animals, lin-42 activity is required to prevent early expression of the adult hypodermal program, probably by inhibiting lin-29 activity. Indeed, LIN-29 accumulates one stage early in lin-42(n1089) animals (BETTINGER et al. 1996 ). This molecular epistasis, as well as analysis of lin-42 lin-29 double mutants (Table 2), positions lin-42 with lin-14 and lin-28, upstream of lin-29 in the heterochronic pathway (Figure 7). Our isolation of lin-42 alleles as suppressors of the lin-4 hypodermal defect demonstrates that lin-42 is required for the lin-4 retarded phenotype. Analysis of lin-42 lin-4 double mutants reveals that mutations in these two genes mutually suppress each other with respect to their hypodermal phenotypes (Table 2). This result is not likely a reflection of hypomorphic lin-42 alleles because ve20 behaves as a null allele by genetic tests. We interpret the mutual suppression between lin-4 and lin-42 mutations to indicate that these two genes are not members of a simple linear regulatory pathway.

View larger version (9K): In this window In a new window Download PPT slide   Figure 7. The heterochronic gene pathway. lin-42 and lin-58 have been added to the heterochronic gene pathway devised by AMBROS 1989 for the control of lateral hypodermal cell terminal differentiation. Briefly, lin-29 is the most direct regulator of the switch between the larval and adult developmental programs. lin-14, lin-28, lin-42 and lin-58 all act as negative regulators of lin-29 activity, which places them upstream of <