Suppressors of glp - I , a Gene Required for Cell Communication During Development in Caenorhabditis elegans , Define a Set of Interacting Genes Eleanor M . Maine

The glp-1 gene is essential for two cell interactions that control cell fate in Caenorhabditis elegans: induction of anterior pharynx in the embryo and induction of mitotic proliferation in the germ line. To identify other genes involved in these cell interactions, we have isolated suppressors of two temperature sensitive alleles of glp-1. Each of 14 recessive suppressors rescues both embryonic and germline glp-l(ts) defects. These suppressors are extragenic and define a set of six genes designated sog, for suppressor of glp-1. Suppression of glp-1 is the only obvious phenotype associated with sog mutations. Mutations in different sog genes show allele-specific intergenic noncomplementation, suggesting that the sog gene products may interact. In addition, we have analyzed a semidominant mutation that suppresses only the glp-1 germline phenotype and has a conditional feminized phenotype of its own. None of the suppressors rescues a glp-1 null mutation and therefore they do not bypass a requirement for glp-1. Distal tip cell function remains necessary for germline proliferation in suppressed animals. These suppressor mutations identify genes that may encode other components of the glp-1 mediated cell-signaling pathway or regulate glp-1 expression. T HE specification of certain cell fates in multicellular organisms depends on information that is received from the cellular environment. The development of a number of cell types in the nematode, Caenorhabditis elegans, depends on the presence of one or more neighboring cells (KIMBLE 198 1 ; KIMBLE and WHITE 198 1 ; SULSTON and WHITE 1980; SULSTON et al. 1983; STERNBERG and HORVITZ 1986; PRIESS and THOMSON 1987). The molecular mechanisms by which one cell influences the development of another cell, a process termed induction, are not understood. However, many genes have been identified that appear to mediate specific inductive events (reviewed in LAMBIE and KIMBLE 1991a). The glp-1 (for germ line proliferation defective) gene mediates at least two inductive cell-cell interactions during C. elegans development (AUSTIN and KIMBLE 1987; PRIESS, SCHNABEL and SCHNABEL 1987). One interaction occurs early in embryogenesis when descendants of one blastomere, PI, induce descendants of another blastomere, AB, to produce pharyngeal muscle (PRIESS and THOMSON 1987). IQ the absence of PI or maternal glp-1, AB does not produce pharyngeal muscle (PRIESS and THOMSON 1987; PRIESS, SCHNABEL and SCHNABEL 1987). A second interaction occurs post-embryonically when two somatic cells, the distal tip cells, induce mitotic proliferation of the germ line (KIMBLE and WHITE 1981). In the absence of the distal tip cells or glp-1, germ cells do not proliferate and hermaphrodites are sterile Genetics 135 101 1-1022 (December, 1993) (KIMBLE and WHITE 198 1 ; AUSTIN and KIMBLE 1987). In addition, maternal glp-1 function is essential for the formation of the embryonic hypodermis (PRIES, SCHNABEL and SCHNABEL 1987). Finally, in the absence of lin-12 (lineage defective) gene product, glp1 is required for formation of several cells or structures required for larval viability (LAMBIE and KIMBLE 1991b). To identify other genes involved in glp-l-mediated cell-signaling, we have isolated extragenic suppressors of two temperature sensitive (ts) glp-1 mutations. Molecular analysis has shown that both ts alleles are missense mutations within the cytoplasmic portion of the predicted glp-1 protein (KODOYIANNI, MAINE and KIMBLE 1992). We previously reported a group of extragenic glp-1 suppressors with morphological defects, including mutations in three genes now known to encode collagen (MAINE and KIMBLE 1989). Here, we report the isolation and genetic characterization of 14 additional suppressors that have no obvious phenotype other than glp-1 suppression. All 14 suppressors rescue both the embryonic and germline phenotypes of glp-1. We also describe a unique mutation that specifically suppresses the g&-1 germline phenotype. MATERIALS AND METHODS Strains and culture methods: In general, worms were maintained on agar plates as described (BRENNER 1974). The wild-type strain C. elegans var. Bristol (N2) and most 1012 E. M. Maine and J. Kimble mutants are described in HODCKIN et al. (1988) except where indicated. Nomenclature follows the guidelines of HORVITZ et al. (1 979). Mutations used in this study were [ bli (blister), dpy (dumpy), fog (feminization of the germ line), glp (germ line proliferation defective), him (high incidence of males), lin (abnormal lineage), lon (long), rol (roller), sel (suppressor and/or enhancer of Zin-l2), sma (small), unc (uncoordinated)]: Linkage group I (LG I ) : dpy-5(e61), dpy-l4(e188ts), gldl(q268) (provided by T. SCHEDL), lin-lO(e1439), lin1 l(n566), unc-lj(e51, e1091), nDj25, ozDf5 (provided by T. SCHEDL). LG I I : roLl(e91), unc-4(e120). LG I I I dpy-l7(e164), dpy-l8(e364), glp-l(q35, q50, q158, q224ts, q231ts) (AUSTIN and KIMBLE 1987), lin-l2(q269), sel2(n655) (provided by G. SEYDOUX and I. GREENWALD), unc32(e189), unc-36(e251), unc-79(e1068), eT1, nDf11. LC I V dPy-l3(e184), dpy-2O(eI282ts), unc-5(e53), unc24(e138), eDfl8, eDfl9. LG V dpy-Il(e224), him-5(el467), unc-42(e270), sDf35. LG X : h-2(e678), unc-l(e719), unc-l8(e81). Isolation of recessive suppressors of glp-l(ts): Fourth larval stage (L4) hermaphrodites of genotype unc-32 glpl(ts) were raised at 15", mutagenized, and returned to plates at 15". Two strategies were used to isolate recessive suppressors. (1) To generate suppressors of both the germline and embryonic glp-1 phenotypes, F, progeny of mutagenized animals were picked (three animals per plate) and grown at 15". FP progeny were shifted to 20" as late embryos or L1 larvae. Plates were screened visually for viable Fs progeny. In this way, three mutations, q294, 9295 and q297, were isolated from 2100 glp-l(q224) F, animals and nine mutations, q298, q299, q300, q301, q303, q304, q305, q306 and q308, from 1900 glp-l(q231) Fl animals. (2) In an attempt to isolate germline-specific suppressors, animals were treated as described above except that FP animals were shifted back to permissive temperature soon after reaching adulthood. One suppressor, q345, was isolated from 600 FI glp-l(q224) animals shifted to 20"; no suppressors were recovered from an additional 600 F1 gZp-l(q224) animals shifted to 25". Upon testing, q345 proved to suppress the embryonic as well as the germline phenotype of One recessive suppressor, q309, was recovered in an F1 screen for dominant suppressors of glp-l(q231) that is reported in the accompanying paper by LISSEMORE et al (1 993). Upon retesting (see below), it proved to be recessive. Isolation of q162: q162 was isolated separately in a noncomplementation screen for new glp-1 alleles (AUSTIN 1989). The allele used in the screen was glp-l(q35). q162 acts as a semidominant suppressor of the glp-l(q35) loss of function phenotype. Recessivity tests: To remove extraneous mutations from the genome, suppressed lines were outcrossed to wild-type (N2) and fertile unc-32 glp-l(ts) animals were recovered in the F2. Here, suppressors are designated sup@). To test whether a given suppressor was strictly recessive, unc-32 glp-l(ts); sup(x) hermaphrodites were mated to N2 males (at 20°), heterozygous cross-progeny were isolated, and fertile Unc animals were recovered in the FP. For all alleles except q303, 25% or fewer of the Unc FP were fertile, suggesting that (1) suppressed animals are homozygous for the sog mutation and (2) the sog mutations are unlinked to glp-1 (see RESULTS). In most cases where fewer than 25% of the FP were suppressed, later tests indicated that the suppression phenotype was not completely penetrant (see RESULTS). As q303 proved to be unlinked (see RESULTS), it was tested for semidominance by mating glp-l(q23l);him-5 males to g1p1 . q303;unc-32 glp-l(q23l) hermaphrodites and examining the fertility and progeny viability of non-Unc cross progeny at 20". The brood produced by 38 cross-progeny was 48 animals; on average two survived and produced viable progeny. Thus, q303 is only very weakly semidominant. Genetic mapping and complementation tests: Linkage and complementation were determined by standard tests (see Tables 2 and 3 in RESULTS). Complementation and mapping were done on the basis of suppression of glp-l(ts) in all cases except for the mapping of q162, which was done on the basis of its visible phenotype. Mapping of suppressors of the germline and embryonic phenotypes: The recessive suppressors were mapped to a linkage group by one of two methods. (1) Strains containing glpl(q231) and one marker mutation on each of three chromosomes were used to generate animals that were dpy-5(e61/ +);rol-l(e91/+);unc-32(e189/+) glp-l(ts);sog-?(+/-) or unc5(e53/+);dpy-11 (e224/+);lon-2(e678/+);sog-?(+/-);glp-l(ts). From these heterozygous animals, progeny were isolated that were homozygous for one of the markers; they were tested at 20" for the presence of the suppressor. Fertile animals homozygous for a particular marker mutation were not recovered if the marker was located close to the suppressor. (2) glp-l(ts) males carrying a suppressor were mated to strains containing two markers on a single chromosome and cross-progeny were isolated. For example, dpy-5 unc13; glp-l(ts) was used for mapping on LC I . From dpy-5 unc13/++;glp-l(ts);sog-?(-/+) mothers, Dpy Unc [ glp-l(ts);dpy5 unc-131 hermaphrodites were recovered and tested for fertility. If the sog mutation in question is located on LC I , it should be difficult to recover fertile Dpy Unc animals. Seven mutations, q295, q298, q303, q305, q308, q309, q345, mapped to LG I , and three-factor mapping with dpy5 unc-13 placed them in a common position close or to the right of unc-13; additional mapping of one allele, q298, with dpy-14 unc-13 confirmed this location. All seven alleles fail to complement each other and are designated sog-1. More precise mapping of two alleles, sog-l(q295) and sog-l(q298), was done with unc-13gld-1 and unc-13 lin-10. One mutation, q299, mapped to LC I I and is designated sog-2; it was threefactor mapped with unc-4 rol-1. One mutation, q294, mapped to LC N a n d is designated sog-3; it was three-factor mapped with unc-5 and dpy-20. Two alleles, q301 and q304, mapped to LC V; they fail to complement and are designated sog-4. They were three-factor mapped with dpy-11 and unc42. One allele, q297, mapped to LC X and is designated sog5; it was three-factor mapped with lon-2 and unc-18. Finally, two alleles, q300 and q306, mapped to LC IV; they fail to complement and are designated sog-6. They were threefactor mapped with unc-5 and dpy-20 to a position distinct from sog-j(q294). Mapping of germline-specijc suppressor: The germline-specific suppressor, q162, was three-factor mapped on the basis of its feminized germline (Fog) phenotype using unc-93 dpy17: Dpy and Unc recombinants from unc-93 dpy-l7/q162 were picked at 20" and their progeny were examined at restrictive temperature (1 2 ") for a Fog phenotype. Determination of brood size and percent hatching: L4 hermaphrodites of genotype sog;glp-l(ts) were picked from stocks grown at restrictive temperature (20"), placed individually on Petri dishes and transferred every -24 hr to a fresh plate. The total number of embryos produced by each hermaphrodite was counted; embryos were scored for viability -36 hr after the hermaphrodite had been transferred. Hatched progeny were counted once they had achieved at least the L3 stage of development. As a control, brood sizes and percent hatching of glp-l(q231) and glp-l(q224) were determined at 20". Heterozygous q162 animals were tested Suppressors of glp-1 in C. elegans 1013 by crossing unc-l(e719j;q162 glp-l(q224) hermaphrodites to glp-l(q224j;him-5 males. Brood sizes of the non-Unc crossprogeny were counted. Temperature shift experiments: Shqt down: Homozygous q162 animals grown at 25" were picked to fresh plates preincubated at 25 " to generate progeny for shifting to 1 5 O . Progeny of the specified stages were picked and shifted to restrictive temperature (1 0 ") on preincubated plates. Shij up: Homozygous q162 escaper hermaphrodites from a stock grown at 10" were picked to fresh plates preincubated at 10" to generate progeny for shifting up to 25 O . Progeny of the specified stages were shifted to permissive temperature ( 2 5 " ) on plates preincubated at 25". Scoring shgted animals: Shifted animals were picked to individual plates and assayed for the production of selfprogeny. In addition, they were examined by Nomarski optics if one gonad arm appeared to have a different phenotype than the other, or when sterile. Dosage studies: For four genes, sog-1, sog-4, sog-6 and sog-10, one or more deficiencies (Of) exist that are predicted to remove the gene based on its genetic map position. It should be noted that either one or both LG Z Dfs should delete sog-1 whereas both LG ZV Dfs should delete sog-6. Deficiencies are maintained in a variety of ways: over a balancer chromosome, over a chromosome containing visible markers that flank or are included within the DJ or under a duplication. To test the visible phenotype of each sog/Df combination, sog;glp-1;him males were crossed to hermaphrodites carrying the Df. FI hermaphrodites were cloned out and their phenotypes examined. Those animals that segregated small, misshapen, dead embryos (Df/Df) and did not segregate the balancer or marked chromosome in the F S were presumed to be sog/Djglp-l/+. To determine whether the Glp phenotype is suppressed in sog/DJglp-l animals, strains were constructed and tested as follows. In all cases, crosses were done at 15", and progeny were shifted to 20" as embryos or L1 larvae. (1) For sDf35, a doubly balanced strain, unc-32glp-l/eTl;sDf35/ eT1, was constructed and crossed to sog-4;glp-1;him males. The fertility and progeny viability of unc-32 glp-I/+ glpl;sDf35/sog-4 cross-progeny was examined. (2) For deficiencies maintained over double marker chromosomes (eDfl9, nDf25), homozygous unc-32 glp-1 strains were constructed that carried the Df over the double marker chromosome; hermaphrodites were crossed to sog;glp-1;him males, and the fertility and progeny viability of non-Unc sog/Djunc-32 glpl/+glp-1 cross-progeny were examined. eDfl9lsog-6 animals were distinguished from unc-24 dpy-2O/sog-6 siblings by their small body size and low fertility. nDj25/sog-l were distinguished from unc-13 lin-1 l/sog-I siblings by their production of Df/Df F:! embryos. (3) For ozDf5, an ozDf5/ ozDf5;nDp4;unc-32 glp-1 strain was constructed and hermaphrodites were crossed to sog-1;glp-1;him-5 males. Embryo counts indicated that 15% (331218) of progeny from ozDf51 ozDf5;nDp4 hermaphrodites die, presumably from loss of nDp4. The fertility and progeny viability of non-Unc ozDf5/ sog-1;unc-32 glp-l/+glp-l cross-progeny were tested; at least 15% of cross-progeny were assumed to have lost nDp4. (4) nDfl1 was not tested for suppression of glp-I since sog-lo/ nDfl1 animals could not be recovered. Distal tip cell ablations: Ablation experiments were done by the method of SULSTON and WHITE (1 980) using a laser microbeam system similar to that described by STERNBERG (1988). Prior to ablation, the number of germ cells was counted for later comparison. Animals were maintained at 20" both prior to and after ablation. The distal tip cell was identified by its location at the tip of the developing gonad arm and by its characteristic morphology (KIMBLE and WHITE 1981). Typically, ablations were done in L2 and/or L3 hermaphrodites. Cell death was verified -2 hr after ablation. Animals were examined -24 hr later to determine whether germ cells in the operated arm continued to divide. The unoperated arm was used as a control for proper germline growth.