ZYG-9, A Caenorhabditis elegans Protein Required for Microtubule Organization and Function, Is a Component of Meiotic and Mitotic Spindle Poles

The N2 Bristol strain was used as the wild-type strain. The zyg-9 alleles used in these analyses were the ethyl methane sulfonate–induced alleles, b244ts (Wood et al., 1980), b279, b288ts, b301, b307, ct10, and it3 (Kemphues et al., 1988), which were isolated in screens for maternal effect lethal mutations, the zyg-9 alleles it63 and it64, induced in the mut-2 mutator strain TR679 (Collins et al., 1989), and it152, isolated from a strain carrying the mutator-induced mutation sqt-1 (sc114). Other mutations used were sqt-1(sc114), dpy-10(e128), unc-4(e120), zyg-11(b272), rol-6(e187), mnC dpy-10 unc-52, and fP1, a new Tc1 polymorphism closely linked to zyg-9 in the strain zyg-9(it152) sqt-1(sc114).

zyg-9 was mapped relative to the cloned markers zyg-11, fP1, and rol-6 on linkage group (LG) II. The strain dpy-10 zyg-11 rol-6/fP1 zyg-9 (it152) was constructed and a standard multifactor recombination analysis was performed except that the presence of the fP1 marker was detected using the single worm PCR method of Williams et al. (1992). The recombinants obtained (290 between dpy-10 and zyg-11, 35 between zyg-11 and fP1, 5 between fP1 and zyg-9, and 62 between zyg-9 and rol-6) revealed that zyg-9 was located at 0.7 m.u. on the chromosome II physical map, ∼0.009 map units (m.u.) to the left of fP1. This corresponds to a physical distance of ∼24 kb, assuming a distance of 320 kb between fP1 and rol-6 (Waterston and Sulston, 1995).

K07D6, C45G11, or C28G11 cosmid DNA (15–40 ng/μl) were mixed with plasmid DNA carrying the dominant marker, rol-6(su1006) (200–300 ng/μl), and then injected into the distal arms of both gonads of KK44 [zyg-9 (b279)unc-4/mnC1 II] using the protocol of Mello et al. (1991). Rol non-Unc progeny were picked to separate plates to identify germline transformants. To assay for rescue of zyg-9, Unc Rol and Unc non-Rol segregants from germline-transformed lines were tested for the production of viable progeny.

Southern blot analyses of wild-type (N2) and zyg-9 mutant DNAs were performed according to standard protocols (Ausubel et al., 1995). Genomic DNA was prepared using the Puregene kit (Gentra Systems, Research Triangle Park, NC). Allele-specific DNA polymorphisms were detected using C28G11 cosmid DNA and yk28d8 cDNA as probes (The yk28d8 cDNA clone was a generous gift from Y. Kohara, National Institute of Genetics, Mishima, Japan). An open reading frame in the region of the gene-specific polymorphisms was identified using the genefinder program of ACeDB (Durbin and Thierry-Mieg, 1991). Additional cDNAs were isolated by screening a λgt11 mixed stage cDNA library (Okkema and Fire, 1994) according to standard protocols (Ausubel et al., 1995) using the yk28d8 cDNA as a probe. The largest clone, PO12.3, was used for subsequent analyses.

Antisense RNA corresponding to the cDNA yk28d8 was prepared as described by Guo and Kemphues (1995). Antisense RNA (∼1 μg/μl) was injected in the gonads of wild-type hermaphrodites as described above. Injected animals were picked to individual plates and scored for embryonic lethality 18 h after injection. Individual one-cell embryos were cut from injected animals, mounted on 5% agar pads for microscopy, and then videotaped to observe embryonic phenotypes during the first cell cycle.

The sequences of the zyg-9 cDNA PO12.3 were obtained using the dideoxy chain termination method (Sanger et al., 1977). The PO12.3 subclones were sequenced by the Cornell University Biotechnology program Automatic Sequencing Facility (Ithaca, NY). The 5′ end of the zyg-9 transcript was determined by PCR using the −300 primer in the PO12.3 cDNA sequence (5′GCGAAAGAAGAGAATTTG3′) and the SL1 primer (5′GGTTTAATTACCCAAGTTTGAG3′) containing the SL1 spliced leader sequence. DNA sequences were analyzed using the BLAST program (Altschul et al., 1990) and the Lasergene suite of programs (DNAstar, Inc., Madison, WI).

Northern blots containing poly A mRNA isolated from gravid wild-type worms were prepared according to standard protocols (Ausubel et al., 1995) and probed with the cDNA yk28d8.

Fusion protein containing the carboxy-terminal 462 amino acids of the ZYG-9 protein was prepared for polyclonal antibody production. A fragment containing 1,385 base pairs from the 3′ end of the zyg-9 cDNA was cloned into the expression vector QE30 (QIAGEN, Inc., Santa Clarita, CA). ZYG-9–6HIS fusion proteins were expressed in bacteria, inclusion bodies were isolated (Harlow and Lane, 1988), and then the expected 51-kD protein was purified by SDS-PAGE followed by electroelution and dialysis. Rabbit polyclonal antibodies were generated at the Cornell Research Animal Resources Facility (Ithaca, NY). ZYG-9–specific antibodies were purified from crude serum by column chromatography using ZYG-9 fusion protein coupled to CNBr-activated Sepharose 6MB beads (Pharmacia Biotech., Inc., Piscataway, NJ).

The specificity of purified antibodies for endogenous ZYG-9 protein was assayed by Western blot analysis. Proteins were prepared from embryos by picking 150 gravid adults into dH₂O. Animals were washed and transferred to 20 μl of dH₂O. 180 μl of a solution of four parts 0.66 M KOH to one part commercial bleach was added to each sample. After worms dissolved (∼3 min), samples were spun at 6,000 rpm for 1 min to pellet the embryos. Embryos were washed twice with 200 μl dH₂O, resuspended in 20 μl of sample buffer, boiled for 1 min, and then homogenized (100 strokes) with the narrow end of a Pasteur pipette that had been closed and rounded by heating. Samples were then boiled an additional 3 min. Proteins were resolved by SDS-PAGE, transferred to Immobilon membranes (Millipore Corp., Waters Chromatography, Milford, MA) and probed with purified antiserum. Antibodies were detected using the enhanced chemiluminescence protein detection kit (Amersham Pharmacia Biotech., Inc., Piscataway, NJ).

Embryos were stained according to the protocol of Albertson (1984) with minor modifications. Adult worms were washed in dH₂O, resuspended in PBS, and then transferred with 7–10-μl drops of PBS to polylysine-coated slides. Worms were cut open at the vulva to release embryos, covered with a coverslip, and then frozen immediately on dry ice. Coverslips were removed from frozen slides using a razor blade. Slides were immediately immersed in MeOH (−20°C) for 15 min and then transferred to PBS at room temperature for 5 min. Fixed embryos were incubated in 30 μl of goat serum for 1 h at room temperature. Goat serum was removed and replaced by 10 μl of affinity-purified anti–ZYG-9 antibodies diluted 1:50 in PBS or by mouse monoclonal anti-Drosophila α-tubulin antibodies (4A-1, gift of M. Fuller, Stanford University, Stanford, CA) diluted 1:40 in PBS. Slides were incubated in a moist chamber at room temperature for 1 h and then washed for 10 min in PBS-T (1× PBS, 0.5% Tween [Sigma Chemical Co., St. Louis, MO]) followed by two 10-min washes in PBS. 30 μl of secondary antibody (diluted 1:100 in PBS) was added and slides were incubated at room temperature for 1 h. Slides were then washed in PBS-T for 10 min followed by 10 min in PBS; 1 μg/ml of 4′,6-diamidino-3-phenylindole dihydrochloride (DAPI),¹ 10 min in PBS-T, and then 10 min in PBS. Coverslips were mounted with Vectashield (Vector Labs Inc., Burlingame, CA) to decrease photobleaching.

To disrupt microtubules in embryos before fixation, a protocol adapted from the laboratory of P. Mains (University of Calgary, Calgary, Canada) was used. Embryos were cut out of gravid hermaphrodites in PBS containing 30 μg/ml nocodazole (Sigma Chemical Co.) as described above. Coverslips were placed over embryos and gentle pressure was applied to break the eggshells. Slides were incubated at room temperature for 12 min and then frozen and fixed as described above.

Previous genetic mapping studies placed zyg-9 at the center of LG II between the genetic markers zyg-11 and rol-6 (Kemphues et al., 1988) and to the right of the deficiency mnDf 104, an interval of ∼0.18 m.u. We further mapped zyg-9 within this interval using the physical marker fP1, a transposable element closely linked to the zyg-9 allele, it152. This additional mapping data placed zyg-9 0.11 m.u. to the left of rol-6, ∼24 kb to the right of the fP1 marker. These results indicated that zyg-9 was located within the region covered by the cosmids K07D6, C45G11, and C28G11 (Fig. 1).

To identify the location of the zyg-9 gene, we tested whether any one of these three cosmids could rescue zyg-9 mutants. Germline transformants were generated according to the protocol of Mello et al. (1991) as described in Materials and Methods. We were able to rescue zyg-9(b279) animals only with the cosmid C28G11. 4 out of 17 C28G11 lines produced a small number of fertile homozygous zyg-9 transformants (Rol Unc) (Table 1). Fertile animals produced between 1 and 10 progeny. Most of these progeny arrested as L1 larvae, but a few developed to adults that produced equally low numbers of viable progeny.

Three lines of evidence suggested that the weak fertility of the zyg-9 C28G11 transformants corresponded to rescue. First, the fertility of these correlated with the presence of injected DNA. All fertile homozygous zyg-9 animals analyzed were phenotypically Rol-6, whereas all of the homozygous zyg-9 non-Rol segregants from the same lines produced only dead embryos (Table 1, columns 3 and 4). Second, it is unlikely that the low level of fertility of these transformants was the result of “leakiness” of the b279 allele because none of the 650 b279 homozygous progeny from uninjected animals were fertile (Table 1, row 5). Finally, injection of the pRF4 plasmid alone was not responsible for the observed fertility, as none of 117 b279 homozygous transformants carrying only pRF4 DNA were fertile (Table 1, row 4). Together, these results suggest that the zyg-9 gene is contained within the cosmid C28G11 sequence.

To further test this possibility, we analyzed the zyg-9 mutants it64 and it152, both isolated in mutator backgrounds, for the presence of restriction fragment length polymorphisms in the region covered by the cosmid C28G11. Using this cosmid as a probe in Southern blot analyses, we detected polymorphisms within a 7.7-kb HindIII fragment in both alleles (data not shown).

Using the published DNA sequence from this region (Waterston and Sulston, 1995) and the genefinder program of ACeDB, three predicted open reading frames were identified within this 7.7-kb HindIII fragment. A 1.6-kb cDNA, yk28d8, corresponding to one of these open reading frames, had previously been identified as an expressed sequence tag (Kohara, Y., personal communication). Additional Southern blot analyses using fragments of the yk28d8 cDNA as probes revealed allele-specific novel bands consistent with the possibility that the lesions in the alleles it64 and it152, as well as a third mutator-induced allele, it63, are the results of insertions within this transcription unit (Fig. 2,a). These observations strongly suggested that the yk28d8 cDNA corresponds to the zyg-9 gene. We verified this using the method of RNA interference (Rocheleau et al., 1997). This technique has been used successfully to phenocopy numerous C. elegans maternal effect mutations (Guo and Kemphues, 1995; Lin et al., 1995; Draper et al., 1996; Hunter and Kenyon, 1996; Mello et al., 1996). Antisense yk28d8 RNA was injected into the gonads of wild-type worms as described in Materials and Methods. The results of this analysis are shown in Table II and Fig. 3.

Animals injected with yk28d8 antisense RNA exhibited zyg-9–specific defects. Within 18 h of the injection of antisense yk28d8 RNA, 88% of worms produced more than 50% dead embryos. An analysis of one-cell embryos from injected animals revealed that 70% exhibited defects identical to those observed in the embryos of zyg-9 mutants. These defects included failure of pronuclear migration, transverse spindle orientation, and a characteristic abnormal first cleavage (Fig. 3). Injection of par-1 antisense RNA also resulted in a high percentage of embryonic lethality (60%). However, these embryos displayed a Par-1 mutant phenotype that is distinct from that of zyg-9 (Guo and Kemphues, 1995). Taken together, the rescue of zyg-9 by the cosmid C28G11, the identification of allele-specific polymorphisms, and RNA interference phenocopy allows us to conclude that the yk28d8 cDNA corresponds to the zyg-9 gene.

Northern blot analyses revealed that the 1.6-kb yk28d8 cDNA hybridized to a message of ∼4.9 kb in wild-type worms (refer to Fig. 2 b). To identify the full-length zyg-9 transcript, we screened a λgt11 cDNA library derived from mixed stage animals (Okkema and Fire, 1994) using the yk28d8 cDNA as a probe. A clone of 4,509 base pairs, PO12.3, was subcloned and sequenced. The sequence of the missing 5′ end of the message was determined as described in Materials and Methods. The 5′ end of the zyg-9 transcript is transspliced to the SL1 leader three base pairs upstream of the initiating ATG. The complete ZYG-9 sequence has been submitted (EMBL/GenBank/DDBJ accession number AF035197).

Database searches for proteins similar to the predicted ZYG-9 sequence revealed similarity to the human protein ch-TOG and, to a lesser extent, two yeast proteins: Schizosaccharomyces pombe p93^Dis1 and Saccharomyces cerevisiae Stu2p. ch-TOG is the product of a gene that is overexpressed in colonic and hepatic tumors (Charrasse et al., 1995). The protein associates with microtubules in vitro and is the putative homologue of the Xenopus microtubule assembly protein, XMAP215 (Charrasse et al., 1998). The p93^Dis1 protein is a microtubule-binding protein involved in sister chromatid separation (Nabeshima et al., 1995; Nakaseko et al., 1996; Ohkura et al., 1988). STU2 was identified as a suppressor of tub2-423, a cold-sensitive allele of the β-tubulin gene TUB2, and encodes a protein that associates with microtubules in vitro (Wang and Huffaker, 1997).

A schematic representation of the similarities between ZYG-9 and these three proteins is shown in Fig. 4. All four proteins share sequence similarity in a domain of ∼220 amino acids. Within this region, ZYG-9 is 35% identical to p93^Dis1, 31% identical to Stu2p, and 48% identical to ch-TOG. This domain is present in two tandem divergent copies in ZYG-9. These repeats are 52% identical to one another. ZYG-9 and ch-TOG are also 39% identical in a second region of ∼280 amino acids.

To better understand the function of ZYG-9 protein during early development, we determined its subcellular distribution in early embryos by immunofluorescence microscopy using ZYG-9–specific antibodies. Rabbit polyclonal antibodies were raised against a fusion protein containing the carboxy-terminal 462 amino acids of ZYG-9 as described in Materials and Methods. To determine the specificity of the antibodies raised against these fusion proteins, we performed Western blot analyses of proteins from wild-type embryos and embryos from six zyg-9 mutants using affinity-purified antisera. A single band of 155 kD, the expected size for the ZYG-9 protein, is detected in wild-type embryos (Fig. 5). A band of 155 kD was also detected in the temperature-sensitive zyg-9 allele b244, but not in the four strong alleles, b301, b279, it3, and it152, or in the temperature-sensitive allele b288.

We studied the distribution of ZYG-9 protein in early embryos by in situ immunofluorescence microscopy. Wild-type and zyg-9 mutant embryos were fixed and stained with purified polyclonal ZYG-9 antibodies. In certain cases, embryos were also stained for α-tubulin using a mouse monoclonal anti-Drosophila α-tubulin antibody (refer to Materials and Methods). We found that ZYG-9 exhibits a dynamic cell cycle-dependent distribution pattern. The protein localizes to the meiotic spindle and spindle poles and to the mitotic centrosomes. In metaphase and early anaphase stages, ZYG-9 can be detected over the central mitotic spindle region as well. During interphase, ZYG-9 protein is distributed throughout the cytoplasm. Specificity of this staining pattern was confirmed by its absence in embryos homozygous for any of three strong zyg-9 alleles that lacked the 155-kD protein (Figs. 6 h and 7 q).

The distribution of ZYG-9 is similar in meiosis I and meiosis II spindles. ZYG-9 staining is present throughout the metaphase spindle (Fig. 6, b and e). This localization pattern is similar, but not identical to that of tubulin (Fig. 6, a, b, d, and e). ZYG-9 appears more concentrated than tubulin at the spindle poles (arrows) and more diffuse in the central spindle. This distribution persists through anaphase with a progressive decrease in the relative strength of the signal from the central spindle (data not shown).

After the second meiotic division, ZYG-9 localizes to two small spots adjacent to the male pronucleus (Fig. 7,b). Embryos at this stage costained with antitubulin antibodies revealed that these spots correspond to the foci of the astral microtubules (Fig. 7 a). This localization pattern suggests that ZYG-9 is a component of the centrosome. We were unable to compare the spatial and temporal distribution of ZYG-9 to the distribution of known centrosomal proteins such as γ-tubulin (Stearns et al., 1991) or pericentrin (Doxsey et al., 1994), because available antibodies did not cross-react with C. elegans proteins or reacted with proteins that were not specifically localized to the centrosomes.

ZYG-9 protein localizes to the centrosomes throughout the first mitosis, but is also detected in the central mitotic spindle region between late prometaphase and early anaphase (Fig. 7, e and h). Between telophase and interphase, ZYG-9 staining at the centrosome becomes increasingly diffuse (Fig. 7,n) and is eventually undetectable (data not shown). ZYG-9 protein exhibits the same distribution pattern in all mitotic cells of the developing embryo (Fig. 8,a), in mitotic cells of the gonad (Fig. 8,c), and in dividing spermatocytes (data not shown). The level of ZYG-9 protein in the cytoplasm appears to be reduced in mitotic cells as compared with interphase cells suggesting that the ZYG-9 protein may be cycling between the centrosome and cytoplasm in a cell cycle-dependent manner (Fig. 8 a).

To determine whether the localization of ZYG-9 proteins to the centrosome was dependent upon microtubules, we treated cells with nocodazole before immunostaining for ZYG-9 and tubulin. Under our experimental conditions (refer to Materials and Methods) meiotic and mitotic spindle microtubules, including astral microtubules, were undetectable in most cells after nocodazole treatment, although a small amount of tubulin staining was present at the centrosome (Fig. 9,a). Under these conditions, significant amounts of ZYG-9 protein were detected at centrosomes (Fig. 9 b). These results suggest that long microtubules are not required to maintain ZYG-9 at the centrosome. It is likely, therefore, that ZYG-9 is an integral component of the centrosome rather than a constituent transported there along microtubules. However, because we were unable to completely eliminate tubulin from the centrosomes, we cannot rule out the possibility that tubulin or short microtubules are required for centrosomal localization of ZYG-9.

These experiments also revealed that spindle disassembly led to an accumulation of ZYG-9 in the region of the chromosomes in both mitotic and meiotic cells (Fig. 9, b, e, and h). In meiotic cells, the protein forms a halo around individual bivalents (Fig. 9,h), whereas in mitotic cells the protein exhibits a more diffuse distribution (Fig. 9 b).

It appears that this accumulation around chromatin is not a passive consequence of spindle disassembly. Treatments of one-cell embryos with nocodazole before pronuclear migration can result in zyg-9 phenocopies in which pronuclear migration fails and the mitotic apparatus forms in the posterior and includes only the paternal chromosomes (Albertson et al., 1984). We observed that in such embryos (Fig. 9, d–f), ZYG-9 accumulated not only around the paternal chromosomes that had been incorporated into the spindle apparatus before its disassembly (Fig. 9,e, small arrow), but also around the condensed chromosomes of the female pronucleus that had never associated with the spindle (Fig. 9,e, large arrow). Because we saw a similar distribution of ZYG-9 protein around the dispersed chromosomes in the disorganized meiotic spindles of mei-1 mutant embryos (see below and Fig. 9 j), we believe that the localization pattern observed in nocodazole-treated embryos is not the result of nonspecific effects of nocodazole.

The distribution of ZYG-9 within the meiotic spindle is similar to that of MEI-1, an ATPase required for meiotic spindle formation (Mains et al., 1990b; Clark-Maguire and Mains, 1994,a,b). In mei-1 loss of function mutant embryos, meiotic spindles do not form or are disorganized, but mitotic spindles function normally. A gain of function mei-1 mutation (ct46) and a gain of function mutation in another gene, mel-26(ct61), do not affect meiosis but lead to mislocalization of the MEI-1 protein to the microtubules and centrosomes of the mitotic spindle and produce a Zyg-9–like phenotype (Mains et al., 1990a; Clark-Maguire and Mains, 1994b). To better understand the relationship between ZYG-9 and MEI-1, we studied the distribution of ZYG-9 protein in the mitotic spindles of gain of function and loss of function mei-1 mutant embryos. In mei-1(ct46) gain of function embryos, the distribution of ZYG-9 appeared normal during both meiotic and mitotic stages indicating that the presence of MEI-1 protein in the mitotic spindle does not prevent the association of ZYG-9. A decrease in the quantity of ZYG-9 protein at the poles, however, might not have been detected in this analysis. In mei-1(ct46ct101) loss of function mutant embryos, in which no meiotic spindle forms, the ZYG-9 protein was found to accumulate around meiotic chromosomes (Fig. 9, j and k), similar to its distribution after nocodazole treatment.

The zyg-9 mutant phenotypes of failure in both pronuclear migration and spindle rotation have been interpreted as direct consequences of the observed decrease in microtubule length (Albertson, 1984; Kemphues et al., 1986). Our results, together with the zyg-9 mutant phenotype, lead us to propose that ZYG-9 functions within the spindle apparatus during early embryogenesis to increase the length of the microtubules. Such a role would be required for generating the exceptionally long microtubules necessary for cell division and spindle positioning in the large blastomeres of the early embryo. Because meiotic spindles, which do not require long microtubules, are also disorganized in zyg-9 mutants, the protein must also have some role in spindle organization. Based on available data, we propose two models for zyg-9 action.

Our observation that ZYG-9 protein localizes predominantly to the spindle poles suggests that it may influence microtubule length by affecting microtubule stability at the spindle poles. In this model, ZYG-9 antagonizes other constitutively present centrosomal factors that normally facilitate microtubule disassembly at the minus ends. Such factors have been described; microtubule-severing ATPases have been identified in Saccharomyces purpuratus (McNally and Vale, 1993) and Xenopus eggs (Vale, 1991), and the S. purpuratus protein has been shown to localize to the centrosome (McNally et al., 1996). The S. cerevisiae kinesin-like motor protein, Kar3p, which localizes to the spindle poles of the preanaphase spindle, is thought to depolymerize or otherwise destabilize microtubules during vegetative growth (Saunders et al., 1997).

The meiotic defects associated with zyg-9 mutations indicate that ZYG-9 protein is also essential for meiotic spindle organization and function. The pattern of ZYG-9 localization during meiosis is consistent with this role, but it is paradoxical that a protein required to produce long astral microtubules would also be essential for the organization of the very short meiotic spindle. One possible reconciliation of the two phenotypes is that the activity of ZYG-9 is required to counterbalance the activity of a factor that acts to limit the length of meiotic spindle microtubules. One candidate for such a factor is the MEI-1 ATPase. This protein is required for spindle assembly, exhibits a pattern of localization in the meiotic spindle which is similar to that of ZYG-9 and results in the formation of short microtubules when mislocalized to the mitotic spindle by gain of function mutations (Mains et al., 1990b; Clark-Maguire and Mains, 1994 a,b). Our observation that mitotic ZYG-9 protein is not displaced by MEI-1 protein in gain of function mei-1 mutants indicates that if the two proteins do act antagonistically, they probably do not do so by competing for a common anchoring site.

An alternative model for ZYG-9 function is suggested by the finding that ZYG-9 localizes to the region surrounding chromosomes as well as to centrosomes and that this localization is enhanced after nocodazole treatment. One interpretation of this distribution that is consistent with the meiotic spindle defect is that ZYG-9 functions in chromosome-mediated spindle organization. Several studies have shown that, in the absence of centrosomes, spindle microtubules form and elongate in the region surrounding chromatin and are subsequently reorganized into a bipolar array with microtubule minus ends focused at the spindle poles (for review see Hyman and Karsenti, 1996). ZYG-9 is not required for the nucleation of microtubules in the meiotic spindle, since disorganized arrays of microtubules are detected around meiotic chromosomes in strong zyg-9 alleles (Mains et al., 1990b). This protein may be recruited to the region surrounding the meiotic chromosomes, however, where it, along with other proteins, promotes the elongation and rearrangement of microtubules into an ordered bipolar array.

The progressive accumulation of ZYG-9 protein at the poles of the meiotic spindle is consistent with such a role. Interestingly, the distribution pattern of γ-tubulin in the Xenopus meiotic spindles exhibits certain similarities to the distribution of ZYG-9 during meiosis (Gard et al., 1995). γ-Tubulin associates with the minus ends of microtubules and has been shown to play a role in microtubule nucleation (Li and Joshi, 1995; Zheng et al., 1995). During the early stages of meiosis in Xenopus embryos, γ-tubulin surrounds meiotic chromosomes. By metaphase, it localizes within meiotic spindles and is enriched at poles. Finally, at late stages of meiosis, γ-tubulin is present primarily at the spindle poles (Gard et al., 1995). It has been suggested that this changing pattern of γ-tubulin distribution in the meiotic spindle may reflect the proposed reorganization of the minus ends of microtubules into the spindle poles (Gard et al., 1995).

Although we have shown that ZYG-9 localizes to the mitotic apparatus throughout embryogenesis and is present in the centrosomes of mitotic germ line cells, genetic data indicates that this protein is essential only during early embryogenesis (Kemphues et al., 1986; Mains et al., 1990a). One interpretation of these observations is that ZYG-9 activity is only necessary for the elongation of microtubules in the relatively large cells of the early embryo. After the onset of gastrulation, ZYG-9 may be inactive, or active but functionally redundant.

The relationship between ZYG-9 and the microtubule-associated proteins ch-TOG, p93^Dis1, and Stu2p remains to be determined. Interestingly, like ZYG-9, ch-TOG (Charrasse et al., 1998), p93^Dis1 (Nabeshima et al., 1995), and Stu2p (Wang and Huffaker, 1997) all localize to the spindle poles. In addition, the localization of ch-TOG (Charrasse et al., 1998), and p93^Dis1 (Nabeshima et al., 1995), like ZYG-9, is cell cycle dependent. As the sequence similarity among these proteins is restricted to a domain of 220 amino acids, it is possible that this region is involved in the localization of these proteins to the spindle poles. However, Nakaseko et al. (1996) have shown that the carboxy-terminal region of p93^Dis1, which does not include the conserved domain, is necessary and sufficient for localization of this protein to the spindle poles. Interestingly, truncated p93^Dis1 proteins lacking the amino-terminal region, including the conserved domain, localize to the spindle poles throughout the cell cycle, suggesting that the conserved domain may play a role in cell cycle–dependent changes in the localization of p93^Dis1 protein (Nakaseko et al., 1996). Perhaps this domain has a similar function in ZYG-9 and ch-TOG. The identification of the components of the centrosome and chromosomes with which ZYG-9 interacts may provide further insight into the function of ZYG-9 in regulating microtubule organization.

The authors thank A. Coulson and J. Sulston (both from Sanger Centre, Cambridge, UK) for cosmid strains, and Y. Kohara for providing the cDNA clone yk28d8. Some genetic strains were provided by the Caenorhabditis Genetics Center, which is supported by the National Institutes of Health [NIH] National Center for Research Resources, Bethesda, MD).

This work was supported by a NIH grant (HD27689), a Cornell University Sigma XI grant in Aid of Research, a Mario Enoudi grant for International Studies, and a Bourse Chateaubriand Fellowship from the Department of Science and Technology of The Embassy of France (Washington, D.C.).