ABSTRACT
The generation of asymmetry in the one-cell embryo of Caenorhabditis elegans is necessary to establish the anterior-posterior axis and to ensure the proper identity of early blastomeres. Maternal-effect lethal mutations with a partitioning defective phenotype (par) have identified several genes involved in this process. We have identified a new gene, par-6, which acts in conjunction with other par genes to properly localize cytoplasmic components in the early embryo. The early phenotypes of par-6 embryos include the generation of equal-sized blastomeres, improper localization of P granules and SKN-1 protein, and abnormal second division cleavage patterns. Overall, this phenotype is very similar to that caused by mutations in a previously described gene, par-3. The probable basis for this similarity is revealed by our genetic and immunolocalization results; par-6 acts through par-3 by localizing or maintaining the PAR-3 protein at the cell periphery. In addition, we find that loss-of-function par-6 mutations act as dominant bypass suppressors of loss-of-function mutations in par-2.
INTRODUCTION
A dramatic reorganization of cytoplasm occurs in the C. elegans egg after fertilization (Nigon et al., 1960). During this reorganization, the anterior and posterior poles of the egg display several different characteristics. Foci of filamentous actin accumulate transiently at the anterior pole, and anterior contractions result in an incomplete cleavage called a psuedocleavage. Cytoplasm associated with the outermost cortex of the egg flows toward the anterior pole, and non-cortical cytoplasm streams toward the posterior pole (Hird and White, 1993). Cytoplasmic P granules, which later associate with the germline cells, become localized to the posterior pole (Strome and Wood, 1982, 1983). The maternal and paternal pronuclei join in the posterior and migrate to the center of the cell. The first mitotic spindle becomes asymmetrically positioned in the cell and the first division is unequal (Albertson, 1984; Kemphues et al., 1988b).
After division, the anterior daughter, called AB, differs markedly from the posterior daughter, P1 (Sulston et al., 1983). For example, only AB expresses GLP-1, a membrane receptor that is required for anterior cell fates in the early embryo, and only P1 contains P granules. SKN-1, a transcription factor that specifies the fate of ventral blastomeres in the early embryo, is much more abundant in the P1 nucleus than in the AB nucleus (Bowerman et al., 1993). AB and P1 also have different cell-cycle periods and cleavage patterns; at cleavage the AB spindle is transverse, and the P1 spindle is longitudinal, with respect to the long axis of the egg.
To determine the genetic basis for anterior/posterior differences in the C. elegans embryo, mutants have been isolated in which these differences are reduced or absent (Kemphues, 1989; Kemphues et al., 1988b). Mutations in five par genes (partitioning defective) can result in embryos in which AB and P1 have equal distributions of P granules, equal cell cycles, and similar spindle alignments (Kemphues, 1989; Kemphues et al., 1988b). Many of the early phenotypes of par mutants, such as improper P granule localization, equal division of the fertilized zygote, and incorrect spindle alignments in AB and P1 can be phenocopied by a brief pulse of cytochalasin D during a critical period of the first cell cycle (Hill and Strome, 1990), indicating that the par gene products could be regulating microfilaments or interacting with them.
Three of the par genes have been cloned and the intracellular distributions of their protein products have been determined. par-1 encodes a putative serine/threonine kinase that is localized to the posterior cortex of the one-cell embryo, and continues to be asymmetrically distributed in the P lineage during subsequent divisions (Guo and Kemphues, 1995). The par-2 gene encodes a novel protein with a myosin-like ATP binding site and a conserved cysteine-rich domain (Levitan et al., 1994) and has a localization pattern similar to PAR-1 (see accompanying paper by Boyd et al., 1996). The par-3 gene encodes a novel protein that is localized to the anterior periphery of the one-cell embryo; a distribution that is the reciprocal of PAR-1 and PAR-2 (Etemad-Moghadam et al., 1995). In subsequent stages PAR-3 is present uniformly at the periphery of somatic cells but in the germ line lineage it maintains its reciprocal relationship with PAR-1 and PAR-2.
In this paper we describe the characterization of par-6, a new gene involved in the generation of asymmetry in the early embryo. We show that mutations in par-6 result in the disruption of many asymmetries, including P granule localization, SKN-1 localization, and spindle orientations at the second division. We show that par-6 mutations act as dominant bypass suppressors of par-2 mutations, and can enhance weak par-3 mutations. Finally, in par-6 mutants the PAR-1, PAR-2, and PAR-3 proteins are improperly localized in early embryos. We propose that the par-6 gene product is necessary to localize or maintain PAR-3 protein at the periphery of early embryos.
MATERIALS AND METHODS
Strains and maintenance
C. elegans strains were maintained as described by Brenner (1974). The genetic markers, deficiencies, and balancer chromosomes used are listed by linkage group (LG) as follows: LGI, dpy-5(e61), unc-101(m1), unc-13(e1091), lin-11(n386), unc-75(e950), par-6(zu170), par-6(zu174), par-6(zu222), hT2(I’; III), hIn1 unc-54, hDf15(h1486), hDf16(h1487), hDf17(h1488); LGII, bli-2(e768); LGIII, par-2(it5ts), par-2(lw32), daf-5(e1372ts), par-3(e2074), par-3(it71), lon-1(e185), qC1, dpy-1(e1), sC1; LGIV, unc-5(e53); LGV, dpy-11(e224); X, lon-2(e678). Most mutations are described by Wood (1988). The LGI balancers hT2 (I; III),hIn1 unc-54 are described by Edgley et al. (1995). The deficiencies hDf15, hDf16, and hDf17 were gifts from Fred Ho and Ann Rose and the deficiency sy216 was a gift from Paul Sternberg (Lee et al., 1994). The term ‘par-6 embryos’ refers to embryos produced by homozygous par-6 mothers.
All strains were grown at 20 °C except for temperature-sensitive alleles, which were maintained at 15 °C, the permissive temperature, and shifted to the restrictive temperature, 25 °C, for analysis of the mutant phenotypes.
Mapping and complementation
The par-6 mutations were mapped to LGI by standard linkage tests. Two- and three-factor crosses showed that par-6 maps between unc-101 and glp-4, 0.7 map units to the right of unc-101. Mapping data are available from the Caenorhabditis Genetics Center, University of Minnesota.
For complementation tests with hDf15, hDf16, hDf17 and sy216, deficiencies that delete unc-101, hermaphrodites of the genotype Df/hIn1 were crossed to par-6 unc-101/hIn1 males. F1 Unc-101 progeny were picked individually and tested for maternal effect lethality. hDf15, hDf16, and sy216 were found to delete the par-6 locus.
Observing early development in live embryos
Embryos were dissected out of gravid hermaphrodites in water and transferred to a polylysine coated slide (Kirby et al., 1990), mounted under a coverslip supported by enough petroleum jelly to preclude contact with the embryos. The embryos were examined with Nomarski optics under a Zeiss microscope equipped with a video camera, and the period from pronuclear migration to the four-cell stage was videorecorded. Spindle orientations were scored as described by Cheng et al. (1995).
Indirect immunofluorescence
P granules were visualized with OIC1D4 antibodies (gift from S. Strome) according to the procedure of Strome and Wood (1983). Spindles were visualized using a monoclonal antibody against Drosophila alpha-tubulin (gift from M. Fuller, Stanford University) according to procedures previously described (Albertson, 1984; Kemphues et al., 1986). Staining with the monoclonal antibody FA2 against SKN-1 protein followed the procedures of Bowerman et al. (1993). Immunofluorescence assays for terminal differentiation markers followed the method of Kemphues et al. (1988), using the monoclonal antibodies 5.6 (body wall myosin; Miller et al., 1983), 9.2.1 (pharyngeal myosin; Epstein et al., 1982), and MH27 (adherens junctions; Francis and Waterston, 1985). Staining with anti-PAR-1 antibodies was performed as described by Guo and Kemphues (1995), PAR-2 staining followed Method II from Waddle et al. (1994) as described by Boyd et al. (1996), and PAR-3 staining was performed as described by Etemad-Moghadam et al. (1995).
Western blot analysis
Approximately 300 gravid wild-type and homozygous unc-13 par-6(zu222) hermaphrodites were hand picked and washed with M9 buffer. Embryo isolation and western blot preparation followed the method of Etemad-Moghadam et al. (1995). The blot was probed with purified anti-PAR-3 antibody or monoclonal anti-tubulin and horse-radish peroxidase-linked anti-rabbit IgG (anti-mouse for tubulin) and detected using the ECL system (Amersham).
Construction of par-2; par-6, par-2; par-6/+ and par-2; hDf15/+
daf-7 par-2 (it5ts) III hermaphrodites were mated with unc-13 par-6/hT2 (I; III) dpy-5 males at 15 °C. Cross progeny were mated to their F1 siblings and the F2 progeny were grown at 25°C. F2 Dafs were picked individually to 15 °C to recover daf-7 par-2 homozygotes. Worms that segregated both Dpy and Unc progeny were presumed to be daf-7 par-2; unc-13 par-6/hT2 dpy-5. (I); daf-7 par-2 (III). (The hT2 dpy 5(I); daf-7 par-2 (III) translocation is abbreviated as hT2 par-2 below). Unc progeny were picked individually to 15 °C to confirm the presence of par-6. The Dafs (daf-7 par-2; unc-13 par-6/hT2, par-2) and Dpy Dafs (daf-7 par-2; hT2 par-2) were picked individually to 25 °C and scored for maternal-effect lethality (Mel) to confirm the presence of par-2. The Dpy Dafs were Mel, indicating that the strain was homozyous for par-2, but due to the suppression described in the Results section, the non-Dpy Dafs (daf-7 par-2; par-6/hT2 par-2) produced a large number of hatching embyros. The par-2; par-6 double homozygotes were identified as Daf Unc animals.
To better quantify the extent of suppression without the lethality caused by heterozygosity for the balancer hT2, worms of genotype daf-7 par-2 (it5ts); unc-101 par-6 (zu170)/+ were constructed. We crossed daf-7 par-2 (it5ts) hermaphrodites to unc-101 par-6/hIn1 males at 15 °C. F1 cross progeny were allowed to self at 25 °C and F2 Dafs were plated individually at 25 °C. Daf nonUncs that recovered from dauer were scored for Mel. Two classes were identified, as expected. One class had low viability (daf-7 par-2; +/+); a second class had 70 –90% viable embryos and segregated Unc progeny (daf-7 par-2; +/unc-101 par-6).
Non-balanced strains were also constructed using the other par-6 alleles zu174 and zu222. However, these strains were not marked with unc-101. In these two constructs, we inferred the genotype based on the suppression of par-2. This was possible because the progeny of daf-7 par-2; par-6/+ always gave a bimodal distribution, with half the worms giving greater than 48% viable embryos (daf-7 par-2; par-6/+) and the other half giving less than 24% viable embryos, the highest value seen among control daf-7 par-2 homozygotes (daf-7 par-2; par-6/par-6 or +/+).
Worms of genotype dpy-1 par-2(lw32); par-6/+ were constructed in a similar manner as described above.
To construct dpy-1 par-2(lw32); hDf15/+ worms, dpy-1par-2(lw32)/sC1 hermaphrodites were mated to hDf15/hIn1 males. Cross progeny were plated individually and Dpy progeny of dpy-1 par-2/+; hDf15/+ mothers were picked individually. Many Dpy worms gave survival rates higher than 16%, the highest control value for dpy-1 par-2(lw32), indicating suppression of par-2 (lw32) by hDf15. However, these worms did not give the same bimodal distribution seen with zu170 and zu222. The worms showed a continuous range of survival from 1 –63%, making it impossible to deduce the genotype of each worm. To obtain the data in Table 2 only those worms with greater than 18% viable progeny were included. The number of total progeny from dpy-1 par-2(lw32); hDf15/+ mothers in Table 2 was adjusted for the lethality of hDf15 homozygotes.
Construction of par-3; par-6 double mutants
lon-1par-3(e2074)/qC1 hermaphrodites were mated to unc-101 par-6(zu170)/hIn1 males. Cross progeny were mated to each other and the progeny of this F1 × F1 cross were plated individually and screened for the presence of all markers. A strain of the genotype unc-101 par-6(zu170)/hIn1; lon-1 par-3(e2074)/++ was obtained and double homozygotes were recognized as Lon Unc animals.
RESULTS
Identification and genetic characterization of par-6
During a screen for maternal effect lethal mutations in a mutator strain (Mello et al., 1994), we recovered several mutations that exhibited equal first cleavage and synchronous second cleavages with longitudinal spindle orientations in both blastomeres. Because this phenotype resembled that of par-3 mutants (Cheng et al., 1995; Kemphues et al., 1988b), we carried out complementation tests with par-3 mutations. Three mutations (zu170, zu174, and zu222) complemented par-3 and mapped to linkage group I, 0.7 map units to the right of unc-101. The mutations fail to complement the deletions hDf15, hDf16 and sy216, placing them in a region devoid of other known mutations (A. Rose, personal communication). Thus, the mutations define a new gene that we call par-6.
Each of the par-6 mutations causes a strict maternal effect, with no evidence of zygotic or paternal effects. Hermaphrodites of the genotype unc-101 par-6(zu170)/++ produce viable self-progeny at a frequency of 98.5% (n=1,398), similar to unc-101/+ hermaphrodites (99%, n=1,056), indicating that par-6 (zu170) is recessive and that homozygous par-6(zu170) embryos are viable when the mother is heterozygous. Less than 1% of embryos produced by hermaphrodites homozygous for unc-13 par-6 hatched (1/7241 for zu222, 0/5115 for zu222/hDf15, 1/8465 for zu174, 4/11,694 for zu170), and these escapers arrested in larval stages or became agametic adults. To determine if the maternal effect lethality is rescuable by wild-type male sperm, we crossed plg-1; him-5 males to par-6(zu222)/hDf15 hermaphrodites. Mated hermaphrodites were scored for the presence of a vulval plug (Kemphues et al., 1988a) and none of the embryos laid by these animals hatched (>400). Expression of par-6(+) is not required for male fertility since dpy-5 unc-101 hermaphrodites produced viable cross progeny when mated to males of the genotype unc-13 par-6(zu222)/hDf15.
Each of the par-6 mutations cause similar embryonic phenotypes. All par-6 embryos arrest as amorphous masses of differentiated cells. 98% of late stage embryos are positive for staining with monoclonal antibodies which specifically stain body wall myosin or pharyngeal myosin. 98% also show staining with monoclonal antibody MH27, which stains the adherens junctions of the hypodermal and intestinal cells. Staining is present both internally and on the embryo surface, consistent with the presence of both intestine and hypodermal cells. To assess the presence of intestinal cells, we scored for birefringent gut granules under polarized light (Laufer et al., 1980). par-6 embryos are variable in their production of gut granules; 51% to 70% of embryos produce gut granules, depending on the allele examined (Table 1). This seems to be a sensitive marker for the severity of the alleles, and suggests that par-6(zu222) is the strongest allele. Furthermore, since both par-6(zu222)/hDf15 and par-6(zu170)/hDf15 embryos produce intestine slightly less frequently than embryos of par-6(zu222) or par-6(zu170) homozygotes (Table 1), it is possible that these are not complete loss-of-function mutations. We conclude that par-6, like the other par mutants, produce differentiated cell types but in abnormal numbers and patterns.
Early markers of polarity are improperly distributed in par-6 embryos
In par-6 embryos, several aspects of P-granule localization are abnormal, although some polarity remains. Fig. 1 depicts the most common P-granule staining patterns observed in the one-, two-, and four-cell wild-type and par-6 mutant embryos. In the wild-type one-cell embryo, P granules become localized posteriorly during pseudocleavage as the pronuclei condense and migrate toward each other, such that by the time the pronuclei have met, virtually all embryos show complete posterior P-granule localization (Fig. 1A) (Rose et al., 1995; Strome and Wood, 1983). In many par-6 mutant embryos, P granules remain in the anterior half of the embryo until late in the one-cell stage (Fig. 1B). By the time the one-cell embryo enters metaphase of mitosis, however, most par-6 embryos display P-granule staining only in the posterior (see Fig. 1 legend). At the two-cell stage, most par-6 embryos showed P granules only in the posterior blastomere (Fig. 1D), yet, surprsingly, four-cell embryos often displayed P granules in all four blastomeres (see Fig. 1 legend, and Fig. 1F). In late stage embryos (>50 cells), no P granule staining is detectable.
The skn-1 gene encodes a putative transcription factor that is required for the fate of the EMS blastomere (Bowerman et al., 1992). In wild-type two-cell embryos, SKN-1 protein is present in the nuclei of both blastomeres; however, it is more abundant in the nucleus of the P1 blastomere than in the nucleus of AB (Fig. 2A) (Bowerman et al., 1993). In four-cell embryos it is more abundant in EMS and P2 nuclei than in the nuclei of the AB daughters. We examined early par-6(zu222) embryos stained with SKN-1 antibody F2A and found that SKN-1 protein was present in equal amounts in both nuclei of about half of the two-cell embryos examined (56%, n=9) (Fig. 2B) or in all four nuclei in over half of four-cell embryos examined (65%, n=17).
Mutations in par-6 and par-3 cause similar defects in early cleavage patterns
At the first division of the C. elegans embryo, the mitotic spindle is aligned parallel to the long axis of the egg. In wild-type embryos, this spindle is positioned asymmetrically such that the AB daughter is larger than the P1 daughter. In par-6 embryos, like most par mutants (Kemphues et al., 1988b), the first mitotic spindle is located in the center of the fertilized egg, and the AB and P1 daughters have similar sizes.
At the second cleavage of a wild-type embryo, the AB and P1 blastomeres have different centrosome morphologies and have distinct mitotic spindle orientations. The AB centrosome is spherical, with an axial ratio (the ratio of the transverse axis to the longitudinal axis of the centrosome) of about 1.2. The P1 centrosome is disc-shaped, with an axial ratio of about 4.4. We find that AB and P1 centrosomes in par-6 mutant embryos are both disc-shaped with axial ratios of 2.6 and 2.8, respectively (n=7). In wild-type mitosis, the AB spindle becomes aligned transversely with respect to the first division axis. In P1, however, the centrosomal-nuclear complex undergoes a rotation such that the spindle aligns longitudinally, or parallel to the first division axis (Fig. 3A) (Hyman and White, 1987). In par-6 embryos, the centrosomal-nuclear complex rotates in both AB and P1, such that both blastomeres have longitudinal spindles (Fig. 3B)
Two general classes of par mutations have been distinguished previously by their effects on the centrosome morphology and spindle alignment in the AB and P1 blastomeres. In most par-2 and par-5 mutant embryos, and in about 20% of par-1 and par-4 mutants, both the AB and P1 spindles align transversely (Kemphues et al., 1988b; Morton et al., 1992). In par-3 mutants, both AB and P1 have disc-shaped centrosomes with axial ratios of about 2.2, and both the spindles of AB and P1 align longitudinally (Cheng et al., 1995). Thus par-6 mutants closely resemble par-3 mutants in centrosome morphology and spindle alignment.
Mutations in par-6 and par-3 cause similar mislocalization of PAR-1 and PAR-2 proteins
In wild-type embryos, both PAR-1 and PAR-2 proteins are localized to the posterior periphery of the one-cell embryo (Guo and Kemphues, 1995; Boyd et al., 1996). This asymmetric distribution depends upon par-3 activity; in one-cell embryos from par-3 mutants, both of these proteins are still peripherally localized but their distribution is uniform and the peripheral signal is weaker than in wild type (Etemad-Moghadam et al., 1995; Boyd et al., 1996). To determine whether par-6 function was also required, we examined PAR-1 and PAR-2 distributions in par-6 embryos. In more than 20 embryos examined for each, the distributions of PAR-1 and PAR-2 are similar to those seen in par-3 mutants. (Fig. 4).
PAR-3 protein is not properly localized in par-6 mutants
The striking similarity in the phenotypes of par-6 and par-3 mutants raised the possibility that the two genes might act independently in the same process, or that one gene might regulate the expression of the other. To investigate the second possibility, we examined the expression pattern of PAR-3 protein in par-6 mutant embryos. We first tested if PAR-3 protein was present in par-6 mutants by western blot analysis of embryonic proteins, and found that par-6 mutants had wild-type levels of PAR-3 (Fig. 5).
In wild-type embryos the PAR-3 protein has a complex and dynamic distribution (Etemad-Moghadam et al., 1995). Briefly, PAR-3 is localized to the anterior periphery of a one-cell embryo; after the first division, PAR-3 surrounds the AB blastomere, but is present only at the anterior periphery of P1. This pattern of expression is reciprocal to that of the PAR-1 and PAR-2 proteins, which are present at the posterior periphery of P1 (Guo and Kemphues, 1995; Boyd et al., 1996). We stained par-6 embryos with antibodies that recognize PAR-3, and found similar abnormalities in PAR-3 localization in each of the strains tested (par-6(zu170), par-6(zu222), and par-6(zu222)/hDf15). In par-6 mutants PAR-3 peripheral staining is reduced or absent (Fig. 6). In addition, when staining is detectable, it is not always asymmetric. Because par-6 mutants appear to have wild-type levels of PAR-3 protein, we conclude that the wild-type pattern of PAR-3 localization requires par-6(+) activity.
The presence of detectable PAR-3 protein in par-6 mutants was cell cycle dependent, with a higher percentage of pre-metaphase embryos showing staining than embryos in metaphase, anaphase, or telophase (see Fig. 6 legend). When peripheral staining was detectable, it was much fainter around the embryonic periphery than is observed in wild-type embryos. However, in two-cell and four-cell embryos, staining at the membranes between the blastomeres was quite pronounced.
Because par-2(+) activity is required to restrict PAR-3 protein to the anterior (Boyd et al., 1996), it was possible that the absence of PAR-3 protein at the periphery of par-6 mutants could have been the indirect result of PAR-2 protein mislocalization. To test this hypothesis we stained par-2; par-6 double mutants with PAR-3 antibody. We found that these embryos were indistiguishable from par-6 embryos, with very little PAR-3 staining detectable at the cell periphery (data not shown).
par-6 mutations act similarly to par-3 mutations in genetic tests
The immunolocalization results suggest that par-6(+) acts by mediating the peripheral localization of the PAR-3 protein. Genetic results are consistent with this model. We found that embryos from the double homozygote par-3(it71); par-6(zu170) are indistinguishable in phenotype from par-3(it71) embryos. Furthermore, par-6 mutations and a weak par-3 mutation exhibit synergistic effects in double mutant combinations. As shown in Table 1, only 9% of double homozygote par-6(zu170); par-3(e2074) embryos produce gut granules compared to 57% of par-3(e2074) embryos and 70% of par-6(zu170) embryos.
Mutations in par-2 and par-3 show opposite effects on the spindle orientations of the second cleavage. In par-3 embryos, both blastomeres show longitudinal divisions, with the spindles oriented parallel to the long axis of the egg. In par-2 embryos, both blastomeres divide transversely, with the spindles oriented perpendicular to the long axis (Fig. 3C). In par-2; par-3 double mutants, AB and P1 both have longitudinal oriented spindles, indicating that par-3 is epistatic to par-2 (Cheng et al., 1995). If par-6 acts by localizing PAR-3, then we expected that par-6 mutations would also be epistatic to par-2. Indeed, we find that the AB and P1 spindles also are longitudinal in par-2(lw32); par-6(zu222) and par-2(it5ts); par-6(zu222) mutant embryos (Fig. 3D).
par-6 suppresses par-2 alleles
In constructing the par-2; par-6 double mutant we discovered that worms carrying one mutant copy of par-6 bypass the need for a functional par-2 gene. Hermaphrodites that are homozygous for the temperature-sensitive allele par-2(it5ts) produce only 8% hatching progeny at 25°C. Most of these hatching progeny become sterile adults lacking mature gametes. However, a homozygous par-2 hermaphrodite that is heterozygous for any of the par-6 mutations produces 81–85% hatching progeny, with 78–93% of these progeny becoming fertile adults (Table 2). Thus, the par-6 mutations act as dominant suppressors of both the embryonic lethality and the adult sterility of par-2(it5ts).
To test if this suppression was limited to the temperature-sensitive allele of par-2, we asked if par-6 mutations could suppress other par-2 mutations. par-2(lw32) allele is a nonsense mutation in the first third of the coding sequence (Levitan et al., 1994) and has a stronger phenotype than it5. We found that par-2(lw32) homozygotes that were heterozygous for either par-6(zu170) or par-6(zu222) produced 77% and 76% hatching progeny, respectively. We also observed some suppression in the adult sterility of the hatched progeny, with about 5% becoming fertile adults (Table 2).
To determine if this suppression results from loss-of-function at the par-6 locus, we tested whether par-2 worms heterozygous for a deficiency which deletes par-6 would also be suppressed. We constructed a par-2(lw32); hDf15/+ strain (see Materials and Methods) and scored the percentage of hatching embryos. 52% of the worms produced more viable embryos than ever observed for par-2(lw32) worms, with an average of 54% of embryos hatching (Table 2), indicating that hDf15 also suppresses par-2. This suppression is less robust than suppression observed with the par-6 alleles. This could indicate that suppression is sensitive to levels of par-6 activity, could reflect differences in genetic background between the strains, or could mean that hDf15 deletes other genes that play a role in the processes mediated by par-6 and par-2.
DISCUSSION
We have described the characterization of par-6, a new gene necessary for the generation of polarity in early C. elegans embryos. In par-6 mutants, the first cleavage results in equalsized blastomeres that undergo nearly synchronous divisions, have altered aster morphologies, and divide along the long axis. Late embryos arrest as amorphous masses of differentiated cells and the rare survivors of maternal effect lethality are agametic. This phenotype is very similar to that produced by mutations in a previously characterized gene, par-3. A probable explanation for this similarity is suggested by our genetic and immunolocalization results. We propose that par-6(+) acts through par-3(+) by localizing or maintaining the PAR-3 protein at the cell periphery.
The protein product of the par-3 gene is localized to the anterior periphery of one-cell embryos (Etemad-Moghadam et al., 1995) and the products of the par-1 and par-2 genes are localized to the posterior periphery (Guo and Kemphues, 1995; Boyd et al., 1996). These patterns represent the earliest known molecular markers of anterior/posterior polarity in the C. elegans embryo, and understanding how these patterns are generated should provide insight into the first steps of embryogenesis. PAR-3 plays a role in determining the distributions of PAR-1 and PAR-2, because in par-3 mutants these proteins are uniformly present at the periphery of the embryo (Etemad-Moghadam et al., 1995; Boyd et al., 1996).
In this paper, we provide evidence that the role of the par-6 gene is to localize or maintain the PAR-3 protein at the cell periphery. Our genetic analysis indicates that par-3 and par-6 function in a common process. The mutant phenotypes are similar; strong par-3 mutants are epistatic to par-6 mutants, weak par-3 mutants are enhanced by par-6 mutants, and both par-3 and par-6 are epistatic to par-2. Our immunofluorescence results reveal at least one aspect of the relationship between the two genes: in the absence of par-6(+) activity, PAR-3 protein is greatly reduced or absent from the cell periphery. Western blots show that this is the result of failure to localize the protein rather than a consequence of reduction in synthesis or stability of the protein. The patchy distribution of PAR-3 in early stages of the cell cycle could reflect residual par-6(+) activity in the mutants or could mean that par-6 is not required for initial localization of PAR-3. Furthermore, the failure of the patchy staining to be restricted to the anterior suggests that par-6(+) may play a role in the asymmetric distribtution of PAR-3.
An unexpected result from our double mutant analysis of par-6 and par-2 was the finding that reducing the amount of wild-type par-6 can suppress par-2 mutations. This is an example of bypass suppression since the suppression is not allele-specific with respect to par-6 or par-2 (Hartman and Roth, 1973). At least one other dominant bypass suppressor of par-2 has been isolated (Cheng, 1991), demonstrating that the par-6/+ background is not the only genetic background in which the requirement for the par-2 gene product is reduced.
The suppression of par-2 mutations can be explained in light of the proposed role for par-6 in localizing or maintaining the localization of PAR-3. In wild-type one-cell embryos, the PAR-3 distribution is sharply defined, with high levels in the anterior and no detectable PAR-3 in the posterior. In par-2 mutants, PAR-3 distribution becomes graded, remaining at high concentration in the anterior but now showing detectable protein in the posterior (Etemad-Moghadam et al., 1995). We propose that the abnormalities in par-2 mutants result primarily from inappropriate posterior distribution of PAR-3. Reducing the level or activity of PAR-3 in a par-2 mutant (via removal of one copy of par-6(+)) might prevent PAR-3 function in the posterior, but permit PAR-3 to function in the anterior, and thus suppress embryonic lethality.
We found that PAR-1 and PAR-2 proteins fail to be restricted to the posterior in par-6 mutants. This is also consistent with a role for par-6(+) in maintaining PAR-3 at the cell periphery, since in par-3 mutants the asymmetrical distribution of PAR-1 and PAR-2 is also disrupted. PAR-1 is not involved in the anterior localization of PAR-3, since PAR-3 distribution in par-1 mutants is normal (Etemad-Moghadam, 1995). However, PAR-2 is necessary for the restriction of PAR-3 to the anterior (Boyd et al., 1996), raising the possiblity that the presence of PAR-2 at the anterior of par-6 mutant embryos is the basis for absence of peripheral PAR-3. This is ruled out by the finding that par-2; par-6 double mutants also have very little PAR-3 at the cell periphery.
Mutants in par-6 display aberrant SKN-1 localization. SKN-1 is a putative transcription factor that is necessary for the proper fate of the posterior blastomere, EMS (Bowerman et al., 1992) and is asymmetrically distributed at the two- and four-cell stages, such that the posterior blastomeres contain more SKN-1 than do anterior blastomeres (Bowerman et al., 1993). We show that in par-6 embryos SKN-1 protein is sometimes localized to both the AB and P1 blastomeres at the two-cell stage and to all four blastomeres at the four-cell stage. Consistent with this distribution, some par-6 embryos produce extra pharyngeal tissue, as is the case with par-1 and mex-1, maternal effect lethal mutants that also mislocalize SKN-1 (Bowerman et al., 1993; Mello et al., 1992).
Germ-line specific P granules are abnormally localized in par-6 embryos. The P-granule mislocalization in par-6 mutants is consistent with a recent proposal that P granule localization in one-cell and two-cell embryos takes place by two mechanisms (Hird et al., 1996). The major mechanism is physical translocation of the granules to the posterior, a process mediated by cortical flow. A second mechanism is degradation or disaggregation of P granules that remain in the anterior. In wild-type embryos, P granules are always localized to the posterior by the time the male and female pronuclei meet. At this stage in par-6 embryos, P granules are often still present in the cytoplasm of the anterior half of the embryo. However, by the time the zygote nucleus enters metaphase of the first cell cycle, few P granules remain in the anterior.
This phenotype may reflect an uncoupling of the proposed two-step process for P granule localization. We suggest that in par-6 one-cell embryos, the mechanism for physical translocation of the granules is less effective than in wild-type embryos, causing many P granules to remain in the anterior, but the degradation mechanism remains functional such that by the end of the cell cycle most P granules still present in the anterior are degraded. It is also possible that the progressive restriction to the posterior in par-6 embryos reflects a longer period during which migration of P granules can take place. In the P1 blastomere of par-6 two-cell embryos, in contrast to one-cell embryos, P granules are evenly distributed to both daughters indicating a complete disruption of both mechanisms.
At the four-cell stage, P granules reappear in AB daughters of many par-6 embryos, even though they are rarely observed in the AB cell at the two-cell stage. We suggest that the reappearance of granules at the four-cell stage might reflect the deactivation of the normal system for eliminating mislocalized P granules. This could be an indirect consequence of the overall disruption of embryonic polarity. Early disappearance and subsequent reappearance of P granules also occurs in par-1, par-3 and par-4 embryos (Guo and Kemphues, 1995; K. J. Kemphues, unpublished results).
ACKNOWLEDGEMENTS
We thank Susan Strome, David Miller, Joel Rothman and Margaret Fuller for antibodies, and Fred Ho, Ann Rose, and Paul Sternberg for LGI balancer chromosomes and deficiency strains. Some strains used in this study were provided by the Caenorhabditis Genetics Center. We thank Tak-June Hung for help with mapping, Mona Hassab for media preparation, and Kathryn Baker for technical assistance. This work was funded by NIH grant HD27689. J.L.W. was supported by an NSF predoctoral fellowship. J.R.P. was supported by the Howard Hughes Medical Institute (HHMI) and a grant from the NIH. We are grateful to Lesilee Rose and Mariana Wolfner for useful comments on the manuscript.