Two homologous regulatory genes, lin-12 and glp-1, have overlapping functions

Two Caenorhabditis elegans regulatory genes, lin-12 and glp-1, are remarkably similar in structure and function. Genetic and phenotypic analyses have demonstrated that these genes control cell fate decisions that are normally regulated by intercellular communication. The hn-12 gene is required in somatic tissues for lateral signalling between cells of equivalent developmental potential (Greenwald et al. 1983; Sternberg, 1988), while glp-1 is required both in the germ line and in the early embryo for inductive signalling between cells with different developmental potential (Austin and Kimble, 1987; Priess et al. 1987). Furthermore, each gene acts cell autonomously (Austin and Kimble, 1987; Seydoux and Greenwald, 1989). Therefore, lin-12 and glp-1 both act in the receiving cell to influence regulatory cell interactions, but each gene controls a distinct type of fate decision. More recently, molecular studies have revealed that the predicted protein products of lin-12 and glp-1 are similar both in amino acid sequence (50– 60% identical) and overall molecular organization (Yochem et al. 1988; Yochem and Greenwald, 1989; Austin and Kimble, 1989). Each gene encodes a putative transmembrane protein with multiple EGF-like repeats in the extracellular domain and six cdc10/ SWI6-like repeats in the intracellular domain (Yochem and Greenwald, 1989; Austin and Kimble, 1989).

Given the similarities between lin-12 and glp-1, we wondered whether the roles of these homologous genes might overlap during development. If either lin-12 or glp-1 were sufficient for a given process, no defect would be observed in either single mutant. Three previously reported results provided clues that lin-12 and glp-1 might indeed be redundant. First, an unusual allele of glp-1, q35, mimics one aspect of gain-of-function (gf) mutations of lin-12: both lin-12(gf)/ + and glp-l(q35)/ + hermaphrodites have multiple ectopic vulvae (the Muv phenotype) (Greenwald et al. 1983; Austin and Kimble, 1987). Second, some glp-1 RNA is present in somatic tissues (Austin and Kimble, 1989), where only lin-12 is known to function. And third, the expression of lin-12 by certain cells of the somatic gonad is required to prevent anchor cell-dependent mitotic proliferation of the proximal germ line (Seydoux et al. 1990), Seydoux et al. have proposed that the anchor cell produces a signal that can interact with either lin-12 or glp-1, and that this signal is ordinarily intercepted by lin-12 expressed by cells of the somatic gonad before it can reach glp-1 in the germ line.

One way to test the idea that lin-12 and glp-1 are redundant is to construct a double mutant that has a loss-of-function (lf) mutation in each gene. If the two genes function independently, the effects of ltn-12(lf) and glp-1 (lf) should be additive in the double mutant, but if they are redundant, a new phenotype might be observed. In this paper, we report that the hn-12 glp-1 double mutant has a novel phenotype, which we call the Lag phenotype (for lin-12 and glp-1). In addition, we report the isolation of mutations in two genes, lag-1 and lag-2, which also cause a Lag phenotype. We propose that the functions of lin-12 and glp-1 are overlapping and that lag-1 and lag-2 are required for the activities of both lin-12 and glp-1.

The following mutations were used (Hodgkin et al. 1988):LGI, dpy-5(e61); LGII, rol-l(e91); LGIII, dpy-19(el259ts), sma-2(e502), unc-32(el89), hn-12(nl37), unc-69(e587); LGIV, dpy-13(el84), unc-5(e53), unc-8(el5), LGV, unc-34(e315), unc-60(e723), unc-46(el77), dpy-11(e224), unc-42(e270), LGX lon-2(e678). In addition, rol-9(scl48) V (R. Edgar, personal communication) was obtained from the Caenorhabditis Genetics Center. Single mutant phenotypes of glp-1 alleles q46, q224ts and q231ts have been described (Austin and Kimble, 1987); postembryonic phenotypes of lin-12 (n941 ) and ltn-12(q269) have been described (Seydoux et al. 1990) We also used sDf50, a deletion that removes lag-2 (Johnsen and Baillie, 1988; Johnsen, 1990).

The lin-12(q269) glp-l(q231) double mutant chromosome was constructed by picking Unc non-Sma recombinants from the strain sma-2(e502) unc-32(el89) glp-l(q231) unc-69(e587)/lin-12(q269) grown at 15°. None of the recombinants segregated progeny with the combination of lin-12 and glp-1 phenotypes expected if the effects of lin-12 and glp-1 were additive. However, several recombinants segregated LI lethals (Lags), suggestive of a synergistic effect of the lin-12 and glp-1 mutations. Complementation testing confirmed that these recombinants had the genotype lin-12 glp-1 unc-69/sma-2 unc-32 glp-1 unc-69. The unc-69 marker was removed from the double mutant chromosome prior to its use in experiments.

The lin-12(n941) glp-1 (q46) double mutant chromosome was obtained by crossing males of genotype dpy-19(el259) glp-1 (q46)/ltn-12(n941) unc-69(e587) with hermaphrodites of genotype dpy-19(el259) unc-69(e587) and picking non-Dpy, non-Unc recombinant progeny The desired recombinant had the genotype lin-12 glp-1/dpy-19 unc-69. Complementation testing was performed with isolates that segregated Lag progeny to confirm the presence of the double mutant chromosome.

The percentage of viability among lin-12 homozygous progeny of a heterozygous mother was estimated as follows. To calculate the number of unhatched eggs resulting from background lethality, the total number of eggs from lin-12 heterozygotes was multiplied by the fraction of unhatched eggs from homozygous wild-type mothers The number of unhatched eggs in excess of background was divided by the total number of homozygous mutant progeny (estimated as 1/4 of the total number of eggs laid) to obtain the fraction of unhatched homozygous mutant eggs. To estimate the percentage of viability among progeny of lin-12 homozygous mothers, eggs were dissected from gravid worms with protruding vulvae, transferred to Petri plates, counted and scored at appropriate intervals.

Heterozygous hermaphrodites were allowed to lay eggs at 20° for 1 – 2 days, then worms were rinsed off the plate and transferred to a microfuge tube to allow the larger animals to settle out. After a few minutes, worms remaining in suspension were transferred to a new tube and centrifuged briefly. Larvae were mounted for Nomarski microscopy (Sulston and Hodgkin, 1988) and data were recorded for those with Lag characteristics.

Indirect immunofluorescence was performed as described by Press and Hirsh (1986). Images were taken using an MRC-500 confocal microscope.

All alleles of lag-1 and lag-2 were obtained after mutagenesis with ethylmethane sulfonate (EMS) (Brenner, 1974). Wildtype L4 hermaphrodites were mutagenized with EMS and allowed to self. Gravid F₁ progeny were picked to individual plates at 25° and their offspring were screened by dissecting scope for the presence of LI lethals. LI lethals were inspected by Nomarski to find those with a Lag phenotype. New isolates were crossed with wild-type males and the resulting heterozygotes mated with appropriate tester or balancer stocks

Mapping and complementation of lag-1 and lag-2 mutations were done using standard genetic techniques (Sulston and Hodgkin, 1988).

Three factor mapping with the strain, lag-l(q385)/dpy-13(el84) unc-5(e53), indicated that lag-l(q385) is either to the right of, or immediately to the left of, unc-5. 22/22 Dpy non-Unc and 0/18 Une non-Dpy recombinants segregated Lag progeny Further mapping with dpy-13(el84) lag-l(q385)/ unc-8(el5) indicated that lag-l(q385) is either to the left of, or immediately the right of, unc-8. 25/25 Dpy non-Lag recombinants segregated Une progeny. Subsequently isolated alleles of lag-1 were identified based on their tight linkage to unc-5(e53) and their failure to complement lag-l(q385).

lag-2(q387) was mapped to the left arm of chromosome V using lag-2(q387) / unc-34(e315) rol-9(scl48) 18/21 Rol non-Unc and 1/18 Une non-Rol recombinant F₁ clones segregated Lag progeny. The refined position of lag-2(sl486) on the left arm of V was determined by deficiency mapping performed by R Johnsen (Johnsen, 1990). Subsequently isolated alleles were assigned to lag-2 based on their location on the left arm of chromosome V and their failure to complement q387.

The phenotype of lag-2(q387)/sDf50 was assessed by crossing males of genotype lag-2(q387)/ + with hermaphrodites of genotype sDf50 unc-46(el77)/dpy-ll(e224) unc-42(e270) sDf50 homozygotes almost invariably arrest prior to hatching, so Lag offspring were known to be lag-2(q387)/ sDf50 These were scored by Nomarski optics for the various Lag traits. In this and other experiments, it was found that males and hermaphrodites exhibit essentially identical Lag phenotypes (males were identified by the presence of the blast cell B in the tail region)

The lin-12 glp-1 double mutant has a novel phenotype The single mutant phenotypes of hn-12(lf) and glp-1 (lf) are distinct (Table 1). hn-12(lf) mutants usually survive to adulthood; however, they exhibit various morphological abnormalities (the Lin-12 phenotype) due to numerous cell fate transformations in somatic tissues (Greenwald et al. 1983; Seydoux et al. 1990). In general, the cells affected by mutations in lin-12 are members of equivalence groups, i.e. cells of equivalent developmental potential that adopt distinct fates as the result of lateral signalling (Greenwald et al. 1983; Sternberg, 1988). The simplest equivalence group consists of two cells that cooperate so that one assumes a preferred (primary) fate and the other assumes a secondary fate. The primary fate is defined as that adopted by the surviving member of an equivalence group when its counterpart has been removed by laser microsurgery (Sulston and White, 1980; Kimble, 1981; Sulston et al. 1983). In hn-12(lf) mutants, both equivalent cells typically follow the primary fate, whereas in lin-12(gf) mutants both follow the secondary fate (Greenwald et al. 1983). Therefore, lin-12 activity appears to be required for adoption of the secondary fate. There is some indication that lin-12 may be involved in cell fate decisions required for viability since a low frequency of lethality among lin-12(lf) homozygotes has been observed (I. Greenwald and P. Sternberg, personal communication); however, the basis for this lethality has not been investigated.

glp-1(lf) mutants derived from a glp-1(lf)/ + mother always survive to adulthood; however, these animals are sterile due to a defect in germline proliferation (the Glp-1 phenotype) (Austin and Kimble, 1987; Press et al. 1987). Germline proliferation normally depends on a single somatic cell, the distal tip cell (DTC) (Kimble and White, 1981). If the DTC is removed by laser microsurgery, germline nuclei that are normally mitotic enter meiosis This is the same phenotype that is seen in glp-1 (lf) mutants (Press et al. 1987; Austin and Kimble, 1987). Genetic mosaic analysis has shown that mitotic proliferation of the germ line requires expression of glp-1 in the germ line, but not in the DTC (Austin and Kimble, 1987). Some of the glp-1 expressed in the germ line is contributed to developing oocytes, and subsequently performs essential functions during embryogenesis. Analyses of temperature-sensitive glp-1 mutants have revealed that both the glp-1 /glp-1 and glp-1/+ progeny of a glp-1/glp-1 mother die as embryos that are defective in the induction of the anterior portion of the pharynx (Press et al. 1987; Austin and Kimble, 1987). Thus, glp-1 is required for two inductive interactions, both of which depend on germline expression.

We constructed the hn-12(lf) glp-1 (If) double mutant in order to determine the developmental consequences of removing the activities of both genes. We find that, unlike either single mutant, the lin-12 glp-1 double mutant invariably arrests in the first larval stage (LI) (Table 1). The same effect is observed using either putative null alleles or weaker alleles for each gene. The lin-12(q269) allele causes a slightly less severe phenotype than the putative null allele Hn-12(n941) (Seydoux et al. 1990), suggesting that Itn-12(q269) retains some lin-12 activity. Further, the glp-1 (q231) allele is temperature sensitive, whereas the putative null allele glp-l(q46) is non-conditional (Austin and Kimble, 1987). The phenotypes of Hn-12(q269) glp-l(q231) and lin-12(n941) glp-l(q46) are virtually identical (Tables 1, 2). Henceforth, each allele combination will be referred to simply as lin-12 glp-1.

The lin-12 glp-1 double mutant exhibits three major anatomical defects (Fig. 1). (1) There is no detectable excretory cell or excretory duct and a small protrusion is present at the normal location of the excretory pore (Fig. 1A, B). (2) The nose of the worm is often twisted sideways or backwards (Fig. 1C). (3) The rectum is undetectable and a protrusion is present at the normal location of the anal opening (Fig. 1D,E). The frequency at which each of these defects is observed is presented in Table 2. We refer to this combination of anatomical defects as the Lag phenotype (for lin-12 and glp-1). Henceforth, the term Lag will be used to describe animals that exhibit one or more of these anatomical characteristics. Lag animals also have a behavioral phenotype: They are inactive and do not eat (although they can live for several days after hatching). This behavior may result from the lack of the excretory cell, which is known to be essential for viability (Nelson and Riddle, 1984), or it may be due to some unknown anatomical or physiological defect.

We examined the epithelia and muscles of lin-12 glp-1 double mutants in more detail. Using an antibody that recognizes epithelial belt desmosomes, we observed a duplicated structure near the excretory pore (Fig. 2). The presence of a second pore cell could explain the protrusion occurring at this location in Lag animals (Fig. IB). With the same antibody, we also saw that certain cells were missing in the posterior region. The cells of the intestinal/rectal valve (virR and virL) and the anterior rectum (rect D, rect VL and rect VR) appeared to be present as usual, but one or both of the next two pairs of rectal cells (K,K’ and F,U) appeared to be missing (data not shown). Using an antibody that recognizes muscle myosin, we found that the anal depressor muscle and at least one intestinal muscle were absent in the double mutant (Fig. 3A,B). In-triguingly, in some double mutant animals, a muscle resembling the anal depressor was present at the normal location of the intestinal muscle (Fig. 3C).

We also examined hn-12 and glp-1 single mutants to ask if they ever exhibit a Lag phenotype. For lin-12, we found that a fraction (<10%) of both Un-12(q269) and lin-12(n941) single mutants are Lag (Tables 1, 2). However, the twisted nose typical of the double mutant is not observed in lin-12 single mutants (Table 2). The low frequency of Lag animals among lin-12 single mutant progeny raised the possibility that some animals might be rescued by maternal lin-12(+) product. We therefore compared the percentage of lin-12 homozygotes that are Lag when derived from hn-12/ + versus lin-12/hn-12 mothers, and found that the maternal genotype of hn-12 has little effect on the viability of homozygous lin-12 offspring (Table 3).

For glp-1, we found that Lag worms could be observed among single mutant homozygotes, but only under special circumstances. For example, when progeny from a homozygous glp-l(q224ts) hermaphrodite are shifted to restrictives temperatures during early embryogenesis, a small fraction (<1%) develop into larvae with twisted noses or lacking both the excretory cell and the rectum (data not shown). In contrast, Lag animals are never observed among the progeny of wildtype mothers (Table 3). Thus, the absence of either glp-1 or lin-12 can result in a Lag phenotype, but in the absence of both genes, it occurs at high frequency.

The distinctive phenotype of the lin-12 glp-1 double mutant allowed us to screen for genes required for the activities of both lin-12 and glp-1- A loss-of-function mutation in such a gene might result in a Lag phenotype. We screened 13000 EMS mutagenized haploid genomes for zygotically acting recessive mutations with a Lag phenotype (see Materials and methods). From this screen, we isolated 11 Lag mutations, all of which are fully recessive.

Mapping and complementation tests (see Materials and methods) revealed that the Lag mutations define two separate loci, each of which is distinct from lin-12 and glp-1 (Fig. 4). Five Lag mutations map to chromosome IV and all fail to complement lag-1 (q385). We call this newly identified gene lag-1. The six other Lag mutations map to chromosome V (Fig. 4) and fail to complement lag-2(387), but do complement lag-l(q385). Mutations of the locus on V had previously been isolated: One recessive allele, sl486, was isolated as a lethal (Johnsen, 1990) and two dominant alleles were isolated as lin-12(gf) suppressors (F. Tax, J. Thomas and R. Horvitz, personal communication). We designate this locus lag-2 because of its distinctive Lag phenotype and the argument (see below) that the Lag phenotype is the loss-of-function phenotype of this gene.

Phenotypic characterization of lag-1 and lag-2 mutants We compared the lag-1 and lag-2 mutant phenotypes to the Lag phenotype of the lin-12 glp-1 double mutant. For lag-1, the five original alleles plus lag-l(q426), which was not isolated based on a Lag phenotype (see below), were inspected. For lag-2, the six original alleles, plus lag-2(sl486), were examined. Based on the relative severity of their mutant phenotypes, the lag-1 and lag-2 alleles can be ranked according to strength (Table 2). For lag-1, q385 is the strongest allele: 100% of q385 homozygotes arrest as LI larvae (data not shown). Nearly all q385 homozygotes examined lacked the excretory cell and were defective in the rectal/anal region; however, only about 10 % of them had a twisted nose. Iag-l(q416) is classified as the weakest allele because, unlike other alleles, some q416 homozygotes are viable and fertile at 20°. lag-1 (q426) is an unusual allele that exhibits a low frequency of larval lethality.

For lag-2, q387 is the strongest mutation: 100% of q387homozygotes arrest as LI larvae (data not shown). AU lag-2(q387) homozygotes lacked both the excretory cell and the rectum/anus and most had a twisted nose. This phenotype was not enhanced when lag-2 (q387) was placed in trans to a deficiency for the lag-2 locus (Table 2). lag-2(q420) is classified as the weakest allele because some q420 homozygotes are viable and fertile at 20°.

We propose that the Lag phenotype of both lag-1 and lag-2 results from a decrease in function at each locus. This hypothesis is based on the recessiveness of the lag-1 and lag-2 alleles, the relatively high frequency at which they were isolated (approximately 5 × 10⁻⁴) and the allelic series available for each gene. Loss-of-function mutations in other genes occur at a similar frequency (Brenner, 1974; Greenwald and Horvitz, 1980). In addition, it seems likely that lag-2(q387) is a null allele: It is the most severe and its phenotype does not change when placed over a deficiency.

The epithelia and muscles of lag-1 (q385) and lag-2(q420) were examined by immunofluorescence as described above for the lin-12 glp-1 double mutant. In each case, the phenotype of lag animals resembled that seen in the lin-12 glp-1 double mutant. In the case of lag-2(q420), occasional animals were found in which there appeared to be two anal depressor muscles, one ectopic (at the position of the intestinal muscle) and one at the normal location in the tail (data not shown).

Since some alleles of lag-1 and lag-2 permit homozygotes to survive to adulthood, we could observe the postembryonic phenotypes of these mutants. For weak lag-1 and lag-2 mutants, adults were often found in which the germ line had failed to proliferate. In these animals, there was no mitotic germ line; instead germ cells that would normally be in mitosis entered meiosis and differentiated into sperm (Fig. 5). This phenotype is identical with that of glp-1 single mutants. The lag-1 allele, q426, is particularly striking. Few lag-1 (q426) animals are Lag (<10 %), and the rest have a glp-1-like germ line. In the case of lag-2 (but not lag-1), adults often possessed a protruding vulva similar to that of lin-12(lf) mutants. This vulval defect is associated with the production of two anchor cells (Greenwald et al. 1983). When examined during the L3 or L4 stages, some lag-2(q389) and lag-2(q420) animals were found to have two anchor cells (Fig. 6). Finally, mature lag-l(q416) homozygotes often display an unusual expansion of the anal region (Fig. 5). This phenotype may be caused by the absence of the anal depressor muscle, since antimyosin staining of such animals revealed that this muscle was missing. The absence of the intestinal muscle may also be involved in the production of this phenotype. We also examined glp-1 and lin-12 single mutants for this phenotype of tail expansion. It was found occasionally in lin-12 single mutants, and less frequently among the progeny of glp-1 (q224ts) homozygotes shifted to restrictive temperature during embryogenesis (data not shown).

In this paper, we present two lines of evidence that lin-12 and glp-1 function by similar mechanisms. First, lin-12 and glp-1 are functionally redundant during embryonic development. Whereas both lin-12 and glp-1 single mutants usually reach adulthood (Greenwald et al. 1983; Seydoux et al. 1990; Austin and Kimble, 1987; Press et al. 1987), the lin-12 glp-1 double mutant invariably arrests in the first larval stage, apparently as the result of one or more transformations in cell fate. There are two models that might explain this result. One possibility is that hn-12 and glp-1 act in separate cells during embryogenesis and that these cells function redundantly in controlling specific cell fates. Alternatively, lin-12 and glp-1 could act in the same cells during embryogenesis and function redundantly within them to control cell fates. Given the known functions and molecular similarity of the two genes, the second model seems more likely. Thus, the simplest interpretation is that these two homologous genes encode interchangeable proteins that mediate a common function during embryogenesis.

Second, we have identified two genes required for both lin-12 and glp-1 functions. The hn-12 glp-1 double mutant dies with a distinctive combination of defects, the Lag phenotype (for lin-12 and glp-1). Single mutants of lag-1 and lag-2 have this same Lag phenotype. Because mutations in both lag genes result in a decrease in gene function (see Results), these genes are likely to be required during embryogenesis for both lin-12 and glp-1 activities. In addition, because weak mutations of each lag gene exhibit phenotypes diagnostic of lin-12 and/or glp-1 single mutants, we suggest that the lag genes are also involved in postembryonic functions of lin-12 and/or glp-1. Furthermore, a role for lag-2 in postembryonic lin-12 activity is supported by the identification of gain-of-function (gf) mutations of lag-2 that suppress postembryonic defects in lin-12(gf) mutants (F. Tax, J. Thomas and R. Horvitz, personal communication).

The role of each lag gene in the functioning of lin-12/glp-1 is not understood. One or both might be a positive regulator of hn-12/glp-1, act together with lin-12/glp-1, or function downstream of lin-12 / glp-1. The suppression of lin-12(gf) by lag-2(gf) indicates an interaction between hn-12 and lag-2 (F Tax, J. Thomas and R. Horvitz, personal communication), but the nature of that interaction is unknown. Preliminary epistasis experiments show that lin-12(gf); lag double mutants are Lag (E. Lambie, unpublished observations.). Therefore, lin-12(gf) does not circumvent the lag gene requirement, but this result does not distinguish between the possible relationships among these genes. A further understanding of the relationship of the lag genes to lin-12/glp-1 must await the results of genetic mosaic and molecular analyses of these genes.

The cellular defects observed in Lag animals (either the lin-12 glp-1 double or strong lag mutants) may result from transformations in cell fate during embryogenesis. Certain cells are missing (e.g. the excretory cell and the anal depressor muscle) and others appear to be duplicated (e.g. the excretory pore). Most missing cells would normally arise in the lineage of blastomere AB.pl (Sulston et al. 1983). In lin-12 single mutants, one AB.pl descendant, called G₂, is often transformed into its AB.pr homolog, W (Greenwald et al. 1983). The effect of lin-12 mutations on the G2/W equivalence group is notable because in this case hn-12 is required for the primary fate (G2), unlike other cases, in which hn-12 is required for specification of the secondary fate. A similar atypical transformation could explain the coincident absence of the excretory duct cell, and apparent duplication of its homolog, the excretory pore cell (G1). Laser ablation studies have shown that the duct cell and the pore cell constitute an equivalence group, with the duct cell fate being primary (Sulston et al. 1983). Other cell fate transformations in Lag animals are less certain. The excretory cell (AB. plpappaap) may be transformed into a neuron, its AB.pr homolog (Sulston et al 1983), but it is difficult to detect one extra neuron. The anal depressor muscle does not appear to be transformed into the intestinal-rectal sphincter muscle, its AB.pr homolog (Sulston et al. 1983).

The absence of AB.pl-derived muscles in Lag mutants is striking in light of the maternal requirement for glp-1 to induce the AB. a-derived muscles of the anterior pharynx (Press et al. 1987). Although there are no known cell interactions that influence the development of the intestinal muscles (one of which is derived from MS) or the anal depressor muscle, an induction that is mediated by either lin-12 or glp-1 may be required for the production of AB-derived muscles in the posterior part of the animal.

The structural similarity of lin-12 and glp-1 indicates that these two genes are probably divergent products of an ancient duplication event (Yochem and Greenwald, 1989; Austin and Kimble, 1989). One enticing possibility is that the ancestral gene was functionally equivalent to the Notch gene of Drosophila. Like lin-12 and glp-1, Notch is involved in numerous cell fate decisions that depend on intercellular signalling (Poulson, 1940; Shellenbarger and Mohler, 1975; Campos- Ortega, 1988; Doe and Goodman, 1985). Furthermore, Notch is similar in sequence and overall molecular organization to lin-12 and glp-1 (Wharton et al. 1985; Kidd et al. 1986). However, unlike lin-12 and glp-1, Notch appears to be unique in the Drosophila genome. No other Drosophila gene has been identified with all the molecular features of the lin-12, Notch and glp-1 family. If lin-12 and glp-1 together execute the worm equivalent of Notch function, which seems likely, the lag genes may be homologs of other neurogenic genes of Drosophila (Lehman et al. 1983). It is difficult to guess which neurogenic genes might be the most likely candidates because the roles played by lag-1 and lag-2 in lin-12/glp-1 function are not understood (see above).

The lin-12 and glp-1 genes are poised at an interesting point in evolution from their ancestral gene. They have overlapping functions during embryogenesis, but have acquired distinct roles during postembryonic development. One way to achieve this situation would be to place proteins with identical biochemical activities under separate controls of gene expression. From analyses of single mutants (Greenwald et al. 1983; Seydoux and Greenwald, 1989; Austin and Kimble, 1987; Press et al. 1987) and the tissue specificity of lin-12 and glp-1 RNAs (Austin and Kimble, 1987), it appears that lin-12 is expressed zygotically and preferentially in somatic tissues, while glp-1 is expressed preferentially in the germ line and has maternal effects. Therefore, the separate postembryonic roles of lin-12 and glp-1 would be achieved by somatic expression of lin-12 to govern fate decisions in equivalence groups and germline expression of glp-1 to control the decision between mitosis and meiosis and to provide maternal glp-1 product to the embryo. An additional mechanism that could generate the postembryonic specificity of lin-12 and glp-1 function depends on tissue-specific regulatory proteins that interact differentially with lin-12 and glp-1. Such genes may emerge from the current searches for tissue-specific suppressors or enhancers of hn-12 or glp-1 phenotypes. Further analyses of the similarities and differences in regulation and function of hn-12 and glp-1 should provide one of the clearest examples of the evolution of separate regulatory functions by homolagous and functionally similar genes.

We acknowledge the superb assistance of Phil Balandyk in generating all lag mutations described. Also, we thank Sean Carroll, Diane Church, Bill Dove and Ron Ellis for critical reading of this manuscript and Bob Johnsen, David Baillie, Susan Strome, Iva Greenwald and the Caenorhabditis Genetics Center for providing strains, Jim Thomas, Franz Tax, Bob Horvitz, John Sulston and John White for communicating unpublished results, and David Miller and Robert Waterston for gifts of antibodies. We are indebted to Steve Paddock and the UW-IMR for use of the confocal microscope. Leanne Olds is responsible for the excellent technical illustrations This research was supported by NIH grant GM31816 E.J.L. was supported by grant DRG-989 from the Damon Runyon-Walter Winchell Cancer Fund.