Introduction

The faithful segregation of chromosomes during meiosis is vital for sexual reproduction, as missegregation leads to aneuploidy. Meiotic chromosome segregation involves two processes: loss of cohesion between chromosome homologs and the movement of chromosomes to opposite spindle poles. The molecular mechanisms that regulate meiotic chromosome cohesion are relatively well understood. Central to chromosome cohesion is the REC-8 cohesin complex, which holds chromosomes together from premeiotic S phase until anaphase (Petronczki et al., 2003). In budding and fission yeast, Rec8 is cleaved during meiosis I by separase in non-centromeric regions to allow the separation of chromosome homologs (Buonomo et al., 2000; Kitajima et al., 2003). In meiosis II, the remaining centromeric Rec8 is removed to allow sister chromatid separation, and this has been shown to require separase activity in fission yeast (Kitajima et al., 2003). In C. elegans, REC-8 is partially removed from chromosomes at anaphase I and is completely removed at anaphase II (Pasierbek et al., 2001). Inactivation of a C. elegans separase homolog leads to a failure of chromosome separation, suggesting that the cohesin degradation mechanism is conserved (Siomos et al., 2001). In contrast to the detailed information about chromosome cohesion and its dissolution, little is known about the mechanisms that regulate chromosome movement during meiotic anaphase.

In many metazoa, the anteroposterior (AP) body axis is defined at a very early stage, when patterning molecules are asymmetrically distributed in the embryo (Pellettieri and Seydoux, 2002). In the nematode C. elegans, AP polarity is initiated in the zygote immediately after meiosis by the sperm pronucleus/centrosome complex (SPCC) (Goldstein and Hird, 1996). In response to the SPCC, the PDZ-domain polarity protein PAR-6 becomes excluded from the posterior cortex. PAR-6 functions in a complex with the PDZ-domain protein PAR-3 and the atypical protein kinase C PKC-3 (Tabuse et al., 1998; Hung and Kemphues, 1999; Joberty et al., 2000). The exclusion of PAR-6 from the posterior allows the RING finger protein PAR-2 to accumulate on the posterior cortex (Watts et al., 1996; Boyd et al., 1996; Hung and Kemphues, 1999; Cuenca et al., 2003). This redistribution of PAR-2 and PAR-6 is required for subsequent AP asymmetries, including the polarized localization of maternal proteins to specify different cell fates and the asymmetric orientation of the mitotic spindle to generate unequal cell divisions (Schneider and Bowerman, 2003; Pellettieri and Seydoux, 2002).

Ubiquitin-mediated protein degradation is an essential aspect of many dynamic cellular processes, including cell cycle progression, signal transduction and transcription (Pickart, 2001). The covalent attachment of poly-ubiquitin chains to proteins can signal degradation by the 26S proteasome (Pickart, 2001). The addition of ubiquitin to proteins is highly regulated and requires the action of an ubiquitin-activating enzyme (E1), an ubiquitin-conjugating enzyme (E2) and an ubiquitin ligase (E3). The substrate specificity of ubiquitination derives from the recognition of the substrate by the ubiquitin ligase. A major class of E3s is the cullin/RING finger ubiquitin ligases (Tyers and Jorgensen, 2000). The cullin gene family comprises five major groupings in metazoa, CUL-1 through CUL-5 (Tyers and Jorgensen, 2000). In mammals, CUL2 functions in an E3 complex that contains the core components elongin C, elongin B and the RING-H2 finger protein RBX1/ROC1 (Kim and Kaelin, 2003).

Here, we show that CUL-2, as well as orthologs of elongin C and RBX1, is required for the initiation of meiotic anaphase II but not for anaphase I. The meiotic delay in cul-2 mutants causes a reversal of AP polarity. By titrating the length of the delay, we demonstrate that there is a tight linkage between meiotic timing and the placement of the AP axis. We also show that CUL-2 has an additional role to prevent the ectopic localization or spreading of PAR-2 on the cortex.

Materials and methods

Strains and alleles

C. elegans strains were cultured as previously described (Brenner, 1974). Strains and alleles used were: Bristol N2; AZ212 [unc-119(ed3); ruIs32 (pAZ132: pie-1 promoter/histone H2B::GFP)]; AZ244 [unc-119(ed3); ruIs57 (pAZ147: pie-1 promoter/β-tubulin::GFP; unc-119(+))]; KK866 [itIs153 (pie-1 promoter/PAR-2::GFP, pRF4, N2 genomic DNA)]; JH1461 [axEx1121 (pie-1 promoter/PAR-6::GFP, pRF4, N2 DNA)]; JH1473 [itIs153; ruIs57]; ET108 [cul-2(ek1)/unc-64(e246); ruIs32]; ET129 [unc-119(ed3); itIs153; ruIs32]; DG627 [emb-30(tn377ts)]; ET143 [zyg-11(mn40) unc-4(e120)/mnC1; cul-2(ek1)/unc-64(e246)]; ET144 [zyg-11(mn40) unc-4(e120)/mnC1; cul-2(ek1)/unc-64(e246); ruIs32]; SP152 [him-1(e879); zyg-11(mn40) unc-4(e120)/mnC1]; ET155 [zyg-11(mn40) unc-4(e120)/mnC1; ruIs32]; and ET113 [unc-119(ed3); ekIs2 (pie-1 promoter/CYB-1::GFP; unc-119(+))]. N2 and ET108 were maintained at 20°C, DG627 was maintained at 15°C, and all other strains were maintained at 24°C.

RNAi

RNAi was performed either by injection of dsRNA into hermaphrodites or by feeding with bacteria expressing dsRNA, as described (Feng et al., 1999; Timmons et al., 2001). To produce a partial loss of CUL-2 function, embryos were observed between 4 and 12 hours post-injection of cul-2 dsRNA into adult hermaphrodites. The cul-2; rec-8 double RNAi experiment followed the protocol of Davis et al. (Davis et al., 2002), except that after injection of rec-8 dsRNA into adult hermaphrodites, L4-stage progeny were placed on cul-2 RNAi bacteria feeding plates for 24 hours prior to observation. Inactivation of cul-2 in double RNAi experiments was assessed by observation of the multinuclei and cytoplasmic extensions that occur in cul-2 mutant embryos (Feng et al., 1999). The effectiveness of cyb-1 RNAi was confirmed by the elimination of CYB-1::GFP signal.

Microscopy

Time-lapse movies were made of embryos in utero. Young adult hermaphrodites were anesthetized either with 0.1% tricaine, 0.01% tetramisole (Sigma) in M9 solution or 10 mM levamisole (Sigma) in M9 solution. To observe the initiation of PAR-2 cortical localization, zygotes were cut from the uterus into EBGM medium (Shelton and Bowerman, 1996) on 12×12 mm coverslips. The coverslip was mounted on 4% agarose pads with a thin layer of petroleum jelly applied between the coverslip and the pad to eliminate pressure on the zygotes. Automatic time-lapse imaging was performed with a Zeiss Axioplan microscope equipped with a Hamamatsu ORCA-ER CCD camera, LUDL hardware controller, automated filter wheels and shutters, and an Apple G4 computer running Openlab software (Improvision). Movies were made with pulsed 100 msec exposures for DIC and epifluorescence every 40-80 seconds; epifluorescent illumination was from a 100 W HBO mercury lamp that was filtered to 25% of normal levels. PAR-2::GFP cortical patches were only scored when they were of higher epifluorescent intensity than the interior cytoplasm, except in Table 2 where they were noted as being transient and weak. For the timing of meiosis I, the following number of embryos were observed: wild type, n=5; cul-2(RNAi), n=5; cul-2(ek1), n=5; elc-1(RNAi), n=3; rbx-1(RNAi), n=4; elb-1(RNAi), n=6; zyg-11(RNAi), n=9; zyg-11(mn40), n=3; and zyg-11(mn40); cul-2(ek1), n=5. For the timing of meiosis II, the following number of embryos were observed: wild type, n=5; cul-2(RNAi), n=5; cul-2(ek1), n=5; elc-1(RNAi), n=5; rbx-1(RNAi), n=4; elb-1(RNAi), n=9; zyg-11(RNAi), n=9; zyg-11(mn40), n=4; and zyg-11(mn40); cul-2(ek1), n=5. Quantitation of CYB-1::GFP levels was performed by measuring the fluorescent signal with OpenLab software and subtracting background fluorescence.

The following primary antibodies were used for immunofluorescence: anti-REC-8 (Pasierbek et al., 2001); anti-PAR-2 (Boyd et al., 1996); anti-PAR-6 (Hung and Kemphues, 1999); anti-P granules, OI1CD4 (Strome and Wood, 1982) and monoclonal 4G8 (J.L., S.V. and E.T.K., unpublished); anti-α-tubulin (N356, Amersham); and anti-GFP, 3E6 (Molecular Probes) and ab6556 (AbCam). Secondary antibodies used were: anti-mouse rhodamine (Cappel); and anti-rabbit Alexa Fluor 488 (Molecular Probes). Slides were incubated with 1 μg/ml DAPI to stain DNA prior to mounting in 90% glycerol/PBS with 1 mg/ml p-phenylenediamine (Sigma). The anti-α-tubulin and DAPI-stained meiotic spindle images were obtained as 0.3 μm z-sections and processed with Openlab multi-neighbor deconvolution software. For emb-30(tn377) immunofluorescence experiments, only zygotes that were arrested in meiosis I with the meiotic spindle associated with the anterior cortex and with no sperm asters were scored for PAR-2 and PAR-6 localization; this represents relatively younger arrest embryos prior to the wandering of the meiotic spindle or the formation of sperm asters. Images were processed with Adobe Photoshop software (6.0). The χ² test was used to determine statistical significance. Means are given with s.e.m.

Results

CUL-2 is required for meiotic anaphase II

Following fertilization in wild-type zygotes, the six maternal chromosome bivalents align in a pentagonal array at metaphase I (Albertson and Thomson, 1993) (Fig. 1A). In anaphase I, homologs segregate and a 2N polar body is extruded from the zygote. In metaphase II, the remaining six sister chromatids again adopt a pentagonal array (Fig. 1A). After the second polar body is extruded from the zygote, the maternal chromatids decondense, marking the completion of meiosis.

We examined meiotic chromosome dynamics using time-lapse movies of chromosomes visualized by transgenic expression of histone H2B::GFP. Zygotes homozygous for the cul-2(ek1) null allele (Feng et al., 1999) and zygotes depleted of CUL-2 by RNA-mediated interference (RNAi) (Fire et al., 1998) had indistinguishable phenotypes. They both completed meiosis I with normal chromosome dynamics approximately 15 minutes post-fertilization, similar to wild type (Fig. 1A,B). Significantly, while meiosis II lasted approximately 15 minutes in wild type, it took an average of 50 to 60 minutes in cul-2(RNAi) and cul-2(ek1) zygotes (Fig. 1A,B). Afterwards, the chromosomes generally congressed and individual chromosomes could not be discerned; this was followed by decondensation. Of 18 zygotes observed, ten never executed anaphase II (Fig. 1A), whereas in the other eight the DNA split immediately before or after chromosome decondensation, with DNA bridges observed in the majority of cases (7/8). RNAi depletion of elc-1 and rbx-1, the orthologs of mammalian CUL-2 complex components elongin C and RBX1/ROC1, respectively, produced metaphase II arrest phenotypes similar to that of cul-2 mutants (Fig. 1A,B). This suggests that CUL-2 normally promotes entry into anaphase in the context of a conserved ubiquitin ligase complex.

cul-2 and zyg-11 mutants have a similar meiosis II arrest

The failure of cul-2 mutant embryos to initiate anaphase II was reminiscent of the zyg-11 mutant phenotype. In a study by Kemphues et al. (Kemphues et al., 1986), embryos homozygous for the hypomorphic allele zyg-11(bn2) were found to progress through meiosis I with normal timing, whereas the time for meiosis II was extended to 26 minutes. We analyzed embryos homozygous for the null allele zyg-11(mn40), as well as zyg-11(RNAi) embryos, and found a longer meiosis II delay of 50.2±4.0 minutes (n=5) and 43.6±3.1 minutes (n=9), respectively, comparable to the delay observed in cul-2 mutants (Fig. 1A,B). To test whether simultaneous inactivation of both genes increased the severity of the meiotic arrest, as might occur if the two genes affected meiotic progression through independent pathways, we created a double heterozygous mutant strain (ET144) using null alleles for each gene. Double homozygous zyg-11(mn40); cul-2(ek1) embryos from this strain remained in meiosis II for 53±4.5 minutes (n=5), equivalent to the timings seen upon individual inactivation of each gene (Fig. 1B). This result is consistent with the two genes functioning in the same pathway to regulate anaphase II.

In addition to defective meiosis II, zyg-11 and cul-2 mutant embryos share other phenotypes, including the presence of multinuclei and cytoplasmic extensions during mitotic division (Feng et al., 1999; Kemphues et al., 1986). One difference between the two mutants is the presence of G1-phase arrested germ cells in cul-2 mutants (Feng et al., 1999), which are not observed in zyg-11 mutants. The decreased number of germ cells in cul-2 mutants leads to a lower numbers of eggs. We sought to test whether combining the cul-2 mutant with the zyg-11 mutant would enhance the cul-2 egg production defect. zyg-11(mn40); cul-2(ek1) double homozygotes produced a low number of eggs (35.2±4.7; n=18), which was comparable to the egg numbers in cul-2(ek1) homozygotes (Feng et al., 1999) or hermaphrodites that are homozygous for cul-2(ek1) and heterozygous for zyg-11(mn40) (29.3±5.5; n=19). Conversely, animals homozygous for zyg-11(mn40) but heterozygous for cul-2(ek1) had egg numbers (333±3.2; n=10) similar to wild type, indicating that the loss of zyg-11 cannot induce a germline arrest phenotype in cul-2(ek1) heterozygotes. These results suggest that ZYG-11 is unlikely to function with CUL-2 in promoting germline proliferation.

CUL-2 is not required for the removal of REC-8 from sister chromatids

A failure to segregate chromosomes during meiosis could result from defects in either chromosome separation or chromosome movement. In yeast, cohesin degradation is sufficient for the initiation of anaphase chromosome segregation (Uhlmann et al., 2000). In C. elegans, RNAi depletion of the cohesion REC-8 produces chromosome separation at diakinesis (Pasierbek et al., 2001).

To determine whether REC-8 is removed from chromosomes in cul-2(RNAi) zygotes, we studied REC-8 localization using immunofluorescence with anti-REC-8 antibody (Pasierbek et al., 2001). We found that the anti-REC-8 staining pattern in cul-2(RNAi) zygotes was similar to that in wild type. For both wild-type and cul-2(RNAi) zygotes, REC-8 is located along the axes of sister chromatids during metaphase I, and is located at the junction of the sister chromatids during metaphase II (Fig. 2A) (Pasierbek et al., 2001). In wild type, REC-8 is completely lost from the separating wild-type chromatids during anaphase II (Fig. 2A). In cul-2(RNAi) zygotes, although chromosomes remain in a metaphase II pentagonal array for an extended period, the majority of metaphase chromosomes do not have anti-REC-8 staining (70%, n=26; Fig. 2A). When cul-2(RNAi) zygotes are in the later stages of the extended meiosis II, the meiotic spindle often wanders away from the anterior pole of the egg. None of these late stage meiosis II zygotes were observed to have anti-REC8 staining (data not shown).

The lack of REC-8 staining suggests that the chromosomes in cul-2(RNAi) zygotes are not bound together. Indeed, we occasionally observed separated sister chromatids during the extended metaphase II of cul-2 mutant zygotes (Fig. 1C). In 1/11 cul-2(RNAi) and 1/7 cul-2(ek1) mutants, approximately 12 individual chromatids were observed at the metaphase II plate rather than the normal six joined sister chromatids, indicating that cohesion had been lost between the sister chromatids, although chromosome segregation did not occur. A similar separation of sister chromatids during metaphase II is observed in zyg-11(mn40) mutant embryos (data not shown).

A further indication that a defect in chromosome cohesion was not responsible for the CUL-2 meiotic defects was obtained from experiments in which zygotes were depleted of REC-8. Inactivation of REC-8 produces a loss of chromatid cohesion from the meiotic diakinesis stage onwards (Pasierbek et al., 2001). In both rec-8(RNAi) and cul-2; rec-8 double RNAi zygotes, chromosomes segregate during anaphase I (Davis et al., 2002). However, in both zygotes we observed that chromosomes failed to segregate during meiosis II. Nevertheless, rec-8(RNAi) zygotes had normal meiosis II timing (16.1±0.7 minutes, n=9) (Fig. 2B). By contrast, cul-2; rec-8 double RNAi zygotes had a meiosis II delay comparable to that observed in cul-2(RNAi) zygotes (55.2±6.5 minutes, n=4), indicating that the cul-2 meiotic delay was not rescued by removing REC-8 (Fig. 2B).

An interesting observation was that in the majority of rec-8(RNAi) zygotes (7/8), a monopoleward movement of chromosomes toward the cortex was observed at the time when the second polar body would be extruded in wild type (Fig. 2B). The mechanistic basis of this monopoleward chromosome movement is not known, but it is similar to the chromosome behavior reported in air-2(RNAi) zygotes, which have a failure to remove REC-8 from meiotic chromosomes (Rogers et al., 2002). Interestingly, cul-2; rec-8 double RNAi zygotes did not exhibit monopoleward chromosome movement (n=9), indicating that this aspect of meiotic chromosome movement was also defective (Fig, 2B).

Cyclin B1 is not degraded during meiosis in cul-2(RNAi) zygotes

The cyclin B/CDK1 complex is essential for entry and completion of mitosis and meiosis (Nebreda and Ferby, 2000). A failure to degrade cyclin B1 blocks the metaphase to anaphase transition in mitotically-dividing mammalian cells and is required for meiotic homolog disjunction in mice (Stemmann et al., 2001; Hagting et al., 2002; Chang et al., 2003; Herbert et al., 2003). To analyze whether the degradation of cyclin B1 (CYB-1), was normal in cul-2(RNAi) embryos, we used the pie-1 promoter to express CYB-1::GFP in the germline. In wild type, CYB-1::GFP fluorescence was detected in mature oocytes, but rapidly disappeared during meiosis I and II following fertilization (Fig. 3A,B). By contrast, in cul-2(RNAi) embryos, CYB-1::GFP signal decreased only modestly during meiosis and remained at elevated levels in mitotic-stage embryos (Fig. 3A,B). Elevated levels of CYB-1 were also detected in cul-2(RNAi) embryos with an anti-CYB-1 antibody kindly provided by Sander van den Heuvel (MGH Cancer Center, USA) (data not shown). A similar defect in CYB-1::GFP degradation was observed in zyg-11(RNAi) embryos (Fig. 3A).

Given the requirement for cyclin B1 degradation for entry into anaphase in mammals, we tested whether the failure to degrade cyclin B1 was responsible for the cul-2 metaphase II arrest phenotype. RNAi depletion of cyb-1 does not affect the timing of meiosis II (Fig. 3C). If the elevated level of CYB-1 in cul-2 mutants was solely responsible for the metaphase arrest then depletion of CYB-1 would be expected to allow cul-2 mutants to proceed through anaphase. However, this is not the case, as chromosome segregation was not observed in six out of 12 of the cul-2(ek1); cyb-1(RNAi) embryos, a ratio similar to that observed in cul-2 mutants alone. Interestingly, in cul-2(ek1) homozygous embryos depleted for cyb-1, chromosomes were observed to split in three directions at anaphase I in three out of 12 embryos, and in all 12 embryos examined, chromosomes did not form a pentagonal array in meiosis II (data not shown). This latter result is surprising as pentagonal arrays form in both cyb-1(RNAi) and cul-2(ek1) homozygous meiosis II-stage embryos (Fig. 1A; data not shown). Although the anaphase defect was not rescued, cyb-1 RNAi did produce a partial rescue of the cul-2 meiotic delay phenotype. In seven out of 12 cul-2(ek1); cyb-1(RNAi) embryos, meiosis II length was similar to that in cul-2(ek1) embryos; however, the remaining five embryos had significantly shorter meiosis II timings, three of which were indistinguishable from wild type (Fig. 3C).

The anaphase-promoting complex/cyclosome (APC/C) ubiquitin ligase promotes cyclin B degradation during mitosis in yeast and metazoa (Peters, 2002). APC/C also promotes cyclin B degradation during meiosis in Xenopus and mouse (Peter et al., 2001, Nixon et al., 2002). In C. elegans, loss of APC/C produces a one-cell embryonic arrest at metaphase of meiosis I (Furuta et al., 2000; Davis et al., 2002). Levels of CYB-1::GFP were stabilized in apc-11(RNAi) embryos (Fig. 3B), suggesting that both CUL-2 and APC/C are required for CYB-1 degradation.

The meiotic spindle maintains normal morphology for an extended period in cul-2(RNAi) zygotes

We studied the morphology and dynamics of the meiotic spindle in cul-2(RNAi) zygotes by both anti-α-tubulin antibody staining and live β-tubulin::GFP movies. In meiosis I, the meiotic spindles of cul-2(RNAi) and wild-type embryos were indistinguishable (data not shown). Meiosis II spindles in cul-2(RNAi) embryos maintained a normal barrel-shaped structure for an extended period of time (Fig. 4A). Eventually, the cul-2(RNAi) meiotic spindle shrunk and lost its barrel structure. On average, the cul-2(RNAi) meiosis II spindle lasted three times as long as in wild type (36.2±2.5 versus 12.2±0.7 minutes, n=6 for each) (Fig. 4A). Meiotic spindle morphology characteristic of anaphase II (Fig. 4B) was never observed in cul-2(RNAi) embryos.

Perdurance of the Meiosis II spindle is associated with polarity reversal

In wild-type embryos after meiosis, PAR-3 and PAR-6 are expelled from the posterior cortex near the sperm asters, and, concomitantly, PAR-2 stably accumulates on the posterior cortex (Hung and Kemphues, 1999; Boyd et al., 1996) (Fig. 5A). In contrast to wild type, cul-2(RNAi) zygotes stably accumulate cortical PAR-2 while still in meiosis II. We analyzed the PAR-2 localization pattern in meiosis II-stage cul-2(RNAi) zygotes using immunofluorescence with anti-PAR-2 antibodies. In the majority of meiosis II-stage cul-2(RNAi) embryos (76%, 50/67), strong cortical PAR-2 staining was observed on the anterior cortex, a reversal of the wild-type pattern (Fig. 5A; Table 1). zyg-11(RNAi) embryos similarly had a high percentage of anterior-localized PAR-2 (Fig. 5C; Table 1). PAR-3 and PAR-6 localizations were also reversed, with exclusion of both proteins from the anterior cortex in the majority of cul-2(RNAi) embryos (23/33 and 8/10, respectively; Fig. 5A).

We analyzed whether downstream aspects of polarity were reversed in cul-2(RNAi) embryos. After meiosis is completed in wild type, dispersed P granules move toward the posterior and are completely in the posterior half by the pronuclei meeting stage (Strome and Wood, 1983) (Fig. 5D). In cul-2(RNAi) zygotes, P granules can localize to the anterior, posterior, center or lateral cytoplasm. The abnormal localization is coupled to the altered placement of PAR-2, as P granules co-localize with anterior or lateral cortical PAR-2 in the majority of mitotic cul-2(RNAi) embryos (10/11; Fig. 5D).

In wild type, maternal pronuclei migrate from the anterior to meet the sperm pronucleus in the posterior (Fig. 5E). This aspect of polarity is also reversed in cul-2(RNAi) embryos. In the majority of cul-2(RNAi) embryos (8/10), the sperm pronucleus migrates further than the maternal pronucleus (Fig. 5E). Pronuclei often meet in the anterior of the embryo rather than in the posterior, and the anterior daughter cell is smaller than the posterior cell in over half of cul-2(RNAi) embryos (16/31; Fig. 5B). Similar polarity defects were observed in elc-1 and rbx-1 RNAi embryos (Fig. 5B; data not shown).

To study whether the polarity reversal correlated with the extended meiosis II delay in cul-2(RNAi) zygotes, we varied the penetrance of cul-2 RNAi depletion by altering the length of time adult hermaphrodites were exposed to cul-2 dsRNA prior to fertilization. The partial cul-2 RNAi treatment produced meiosis II timing ranging from that of wild type to delays similar to that of cul-2(ek1) null mutants (Table 2). In wild type, meiosis II lasted 16.2±1.3 minutes and PAR-2::GFP accumulated on the posterior cortex approximately 2 minutes after the completion of meiosis II (18.4±0.9 minutes after meiosis I; Table 2). In two of five wild-type zygotes, transient, weak, anterior cortical PAR-2::GFP was observed during meiosis II, similar to previous reports (Boyd et al., 1996; Cuenca et al., 2003) (Table 2). Transient, weak, anterior PAR-2 localizations were also observed in two cul-2(RNAi) embryos that had meiosis II timing similar to wild type (Table 2). cul-2(RNAi) zygotes with meiosis II timing of ≥28 minutes developed intense, stable PAR-2::GFP patches on anterior or lateral cortexes (Table 2). An extended meiotic delay coupled with an anterior PAR-2::GFP patch was associated with a reversal in the longitudinal placement of the mitotic spindle, so that the first division produced a posterior cell larger than the anterior cell (Table 2).

Ectopic lateral PAR-2 localization is associated with loss of CUL-2

We were intrigued by the observation in cul-2(RNAi) zygotes of PAR-2 patches (8/67 embryos), and exclusions of PAR-3 (3/33) and PAR-6 (2/10), on the lateral cortex that were often distant from both the meiotic spindle and the SPCC (Fig. 5A; Table 1). As microtubule (MT) organizing foci have been implicated in PAR-2 localization (O'Connell et al., 2000; Wallenfang and Seydoux, 2000), we sought to determine whether the lateral PAR-2 arose from prior transient association of the meiotic spindle or SPCC with the lateral cortex. To address this, we followed the initiation of cortical PAR-2 in living embryos expressing either a combination of PAR-2::GFP and β-tubulin::GFP (to visualize the meiotic spindle), or PAR-2::GFP and histone H2B::GFP (to visualize oocyte and sperm DNA). In the two strains, PAR-2 patches appeared to initiate distant from the meiotic spindle or distant from both oocyte and sperm DNA (5/20 and 3/10 embryos, respectively) (Fig. 6A). With the PAR-2::GFP strains, we observed a higher percentage of lateral accumulation of PAR-2 upon cul-2 RNAi than was observed for cul-2(RNAi) embryos probed with anti-PAR-2 antibodies (Table 1). In particular, we observed a substantial percentage of embryos in which PAR-2::GFP spread extensively on the cortex to form a circular band around the embryo (Fig. 6B; Table 1). One third of the circular PAR-2 patterns were observed distant from the meiotic spindle (Fig. 6B; Table 1). These observations of PAR-2 accumulation distant from the meiotic spindle suggest that in cul-2(RNAi) embryos, the initiation or spreading of cortical PAR-2 is under less stringent control.

In wild-type zygotes, PAR-2 and PAR-6 cortical localizations are mutually antagonistic (Watts et al., 1996; Hung and Kemphues, 1999; Cuenca et al., 2003). It has been proposed that the normal accumulation of PAR-2 on the posterior cortex occurs after the exclusion of PAR-6 (Cuenca et al., 2003). With a combination of anti-PAR-6 antibodies and PAR-2::GFP, we observed that ten out of twelve lateral or circular PAR-2 patches that were distant from the meiotic spindle had overlapping cortical PAR-6, suggesting that the lateral PAR-2 localization can occur without the prior removal of PAR-6 (Fig. 6C).

Interestingly, the percentage of lateral and circular PAR-2 was significantly lower in zyg-11(RNAi) embryos, with no examples of PAR-2 patches distant from the meiotic spindle observed (Table 1). Although this may reflect a reduced effectiveness of the RNAi procedure in depleting ZYG-11 relative to CUL-2, it nevertheless suggests that the lateral accumulation of PAR-2 does not arise directly from the meiotic delay, as RNAi of both genes gave comparable meiotic lengths but only cul-2(RNAi) embryos had lateral patches distant from the meiotic spindle.

To further test whether the accumulation of PAR-2 distant from the meiotic spindle was a secondary consequence of the meiotic delay or reflected the loss of a separable CUL-2 function, we analyzed PAR-2 localization in an emb-30 mutant background. emb-30 encodes the APC/C component APC-4 (Furuta et al., 2000). At the restrictive temperature of 24°C, emb-30(tn377ts) embryos arrest at the metaphase I stage of meiosis (Furuta et al., 2000). As described above, CUL-2 is not required for progression through meiosis I; emb-30(tn377); cul-2(RNAi) embryos have the same metaphase I-arrest phenotype and spindle morphology as emb-30(tn377) embryos (Fig. 6E; data not shown). Therefore, any differences in PAR-2 localization between the two strains should reflect the loss of CUL-2 functions that are independent of its role in meiotic progression. As expected from previous work on emb-30 by Wallenfang and Seydoux (Wallenfang and Seydoux, 2000), we observed that a high percentage of both emb-30(tn377) and emb-30(tn377); cul-2(RNAi) embryos had anterior PAR-2 (89%, n=54, and 83%, n=47, respectively). In emb-30(tn377) zygotes, only a small minority of embryos had lateral or posterior PAR-2 patches (3.7%, 2/54) (Fig. 6E), as was also observed by Wallenfang and Seydoux (Wallenfang and Seydoux, 2000). In emb-30(tn377); cul-2(RNAi) zygotes, however, the percentage of embryos with lateral or posterior PAR-2 patches was significantly higher (23%, 11/47; P<0.01); most of these embryos also had anterior PAR-2 patches (Fig. 6E). Similarly, 27% of emb-30(tn377); cul-2(RNAi) zygotes (n=56) exhibited an exclusion of PAR-6 on lateral or posterior cortexes compared with 4% of emb-30(tn377) zygotes (n=28) (significance of P<0.05) (Fig. 6E). There was no difference in the MT distributions observed in emb-30(tn377) and emb-30(tn377); cul-2(RNAi) embryos, both of which had disorganized spindles at comparable percentages (43% and 33%, respectively). These results provide evidence that CUL-2 has a separable function to prevent the ectopic initiation or spreading of cortical PAR-2.

Our observations suggested that the lateral PAR-2 patches could occur distant from the meiotic spindle or sperm DNA. To rigorously test whether PAR-2 can localize to the cortex in the absence of the meiotic spindle in cul-2(RNAi) embryos, we disrupted MTs by depleting the β-tubulin gene tbb-2, which had the effect of eliminating the meiotic spindle and visible MTs during meiosis. The elimination of MTs made distinguishing the two meiotic stages problematic because of the loss of chromosome and spindle dynamics. Therefore, all meiotic embryos were analyzed rather than just those in meiosis II, the stage in which ectopic PAR-2 localization generally occurs. We observed that in tbb-2; cul-2 double RNAi embryos, the overall percentage of cortical PAR-2 was markedly reduced, particularly in the anterior where PAR-2 is normally associated with the meiotic spindle (Table 1). However, discrete PAR-2 patches were still observed in 11.2% of meiotic embryos, most of which was localized to the lateral cortex (Fig. 6D; Table 1). Wild-type embryos subjected to tbb-2 RNAi had no meiotic PAR-2 staining (Table 1). These results indicate that PAR-2 localization can occur in cul-2(RNAi) embryos in the absence of the meiotic spindle. However, the reduction in the percentage of lateral PAR-2 localization in tbb-2; cul-2 RNAi embryos relative to cul-2(RNAi) embryos suggests that the presence of MTs potentiates the ectopic PAR-2 localization.

Interestingly, depletion of tbb-2 alone produced a high percentage of post-meiotic embryos with condensed mitotic prometaphase chromosomes and cortical PAR-2 localized in the same half of the embryo. In 13 out of 15 of these embryos, α-tubulin staining was observed in one or two small circles of approximately 0.3 μm diameter near the condensed chromosomes, presumably the centrosomes. For tbb-2, cul-2 RNAi embryos with mitotic prometaphase chromosomes, 8 out of 12 had one or two similar smallα -tubulin circles, with PAR-2 either localized on the cortical half of the embryo with the chromosomes and α-tubulin circles (9/12) or on lateral cortexes distant from α-tubulin circles (3/12). The observation of α-tubulin at the centrosome suggests that even though visible MT filaments were eliminated after tbb-2 RNAi, a residual capacity for tubulin to be organized into the centrosome still exists and potentially this structure is capable of localizing PAR-2. Therefore, although our results demonstrate that the loss of CUL-2 during meiosis can localize cortical PAR-2 in the absence of the meiotic spindle, we cannot conclude that the PAR-2 localization is independent of all MT activity.

Discussion

CUL-2 and ZYG-11 are required for the initiation of anaphase II and meiotic progression

In zygotes devoid of CUL-2, anaphase II either fails to occur or is severely delayed, and meiosis II lasts three to four times longer than in wild type. The delay in meiotic exit may be more than just a secondary effect of the failure of anaphase II, as air-2(RNAi) and rec-8(RNAi) zygotes have normal meiotic timing even though the former lacks anaphase I and II, and the latter lacks anaphase II (Rogers et al., 2002) (this study).

Three lines of evidence indicate that the meiosis II defects in cul-2 mutants do not arise from a failure of chromosome separation. First, the cohesin REC-8, which is required to hold sister chromatids together (Pasierbek et al., 2001), dissociates from metaphase II chromosomes in cul-2 mutants. Second, a small percentage of cul-2 mutant zygotes are observed in which sister chromatids separate during meiosis II but still fail to move to the spindle poles. And third, in cul-2(RNAi); rec-8(RNAi) zygotes, all chromatids separate from each other, yet exit from meiosis II is still delayed.

Mitotic anaphase chromosome movement has been extensively analyzed. Both MT spindle dynamics and MT motors are required for mitotic anaphase poleward force generation (Wittmann et al., 2001). By contrast, meiotic anaphase has not been extensively studied, and it is unclear to what extent similar mechanisms will function to move chromosomes during meiosis. Interestingly, the morphology and dynamics of the C. elegans meiotic spindle are quite different from the mitotic spindle. The mitotic spindle is composed of long polar MTs and relatively short midzone MTs. Ablation of the mitotic midzone MTs increases the velocity and distance of chromosome separation, suggesting that they are not required to `push' chromosomes apart but instead function to limit the speed and/or extent of chromosome movement (Grill et al., 2001). By contrast, during meiotic anaphase, the MTs near the meiotic spindle poles appear to depolymerize, whereas the amount of midzone MTs between the separating chromosomes increases dramatically (Albertson and Thomson, 1993) (Fig. 4B). During late anaphase, all MTs appear to be located between the separating chromosomes (Fig. 4B). The reorganization of MTs to the midzone that occurs during meiotic anaphase suggests that meiotic chromosomes are `pushed' apart by MT polymerization between the chromosomes rather than being `pulled' to the spindle poles as occurs during mitosis. In cul-2 mutants, the meiotic spindle maintains its normal size and shape for an extended period of time during metaphase II, indicating a defect in the initiation of MT depolymerization at the spindle poles and polymerization between sister chromatids. CUL-2 therefore appears to function upstream of the dramatic MT rearrangements that occur during meiotic chromatid segregation.

CUL-2 and ZYG-11 are required for the degradation of Cyclin B1 during meiosis

Using a CYB-1::GFP reporter, we observed that maternally-provided cyclin B1 is degraded upon fertilization. In both cul-2(RNAi) and zyg-11(RNAi) embryos, CYB-1::GFP is stabilized during both meiosis I and II. This is not a secondary consequence of the meiosis II arrest, as the failure to degrade CYB-1::GFP precedes the arrest. Surprisingly, depletion of CYB-1 by RNAi in cul-2(ek1) mutants produced a synthetic phenotype of loss of the pentagonal chromosome array in meiosis II, indicating that cul-2 mutant zygotes have additional meiotic defect(s) that are exacerbated by loss of CYB-1. The depletion of CYB-1 partially rescued the meiosis II delay of cul-2(ek1) embryos, suggesting that the failure to degrade CYB-1 may be contributing to the meiotic delay.

The APC/C ubiquitin ligase promotes cyclin B degradation during mitosis in yeast and metazoa (Peters, 2002). APC/C has also been shown to promote meiotic cyclin B degradation in budding yeast, fission yeast, Xenopus and mouse (Cooper et al., 2000; Blanco et al., 2001; Peter et al., 2001; Nixon et al., 2002). In C. elegans, inactivation of the APC component apc-11 also stabilized cyclin B1 levels. Further research will be required to assess the roles of CUL-2 and APC/C in regulating cyclin B1 levels.

The spindle dynamics of anaphase I and II appear to be similar to each other in wild type (Albertson and Thomson, 1993). Therefore it is surprising that in cul-2(ek1) embryos, anaphase I is normal but anaphase II is severely delayed or abolished. This is not due to the perdurance of cul-2 maternal product through meiosis I because these embryos arose from cul-2(ek1) homozygous hermaphrodite parents; furthermore, cul-2(ek1) embryos completely lack anti-CUL-2 staining (Feng et al., 1999). The initiation of the metaphase to anaphase transition in the two meiotic stages therefore appears to be differentially regulated.

CUL-2 ubiquitin ligase complex components share meiotic and polarity functions

In mammals, elongin C has been shown to interact with CUL2 and CUL5, but not with other cullins (Pause et al., 1999; Kamura et al., 2001). CUL2 and CUL5 share homology in the N-terminal region, which is required for elongin C binding (Pause et al., 1999; Pintard et al., 2003b). Inactivation of CUL-5 in C. elegans does not produce visible phenotypes (Kamath et al., 2003), suggesting that it is not involved in the meiotic or polarity defects described in this work. RNAi depletion of elc-1 phenocopies all visible cul-2 mutant phenotypes, including the G1 arrest of germ cells (Hui Feng and E.T.K., unpublished), as well as the degradation of embryonic CCCH proteins (DeRenzo et al., 2003), suggesting that ELC-1 participates in all known CUL-2 functions. RBX-1 depletion in C. elegans phenocopies multiple cullin loss-of-function phenotypes (Sasagawa et al., 2003) (data not shown), consistent with its interaction with multiple cullins in mammals (Ohta et al., 1999). The observation of meiotic and polarity defects upon inactivation of homologs of both core CUL-2 complex components suggests that CUL-2 functions in the context of a conserved ubiquitin ligase complex.

A substrate recognition component (SRC) is required for the core CUL-2 complex to bind substrates. A number of mammalian CUL2 SRCs have been identified that function in distinct CUL2 E3 complexes, including the von Hippel-Lindau tumor suppressor protein (VHL) (Kim and Kaelin, 2003; Kamizono et al., 2001; Brower et al., 2002). Inactivation of the C. elegans VHL ortholog produces no embryonic phenotypes (Epstein et al., 2001), indicating that SRCs other than VHL are required for the CUL-2 meiotic and embryonic functions.

Currently, the only genes in C. elegans known to specifically affect the meiosis II metaphase-to-anaphase transition are CUL-2 complex components and zyg-11. Combining homozygous null alleles of both cul-2 and zyg-11 in the same embryo did not enhance the meiotic phenotype, which is consistent with both genes functioning in the same pathway. Both genes also share other embryonic phenotypes, including multiple nuclei, cytoplasmic extensions, and embryonic arrest with approximately 24 cells (Kemphues et al., 1986; Feng et al., 1999) (data not shown). ZYG-11 is a leucine-rich repeat protein and this protein-interaction motif is found in a number of SRCs for CUL-1-based SCF E3 complexes (Tyers and Jorgensen, 2000). It is therefore possible that ZYG-11 functions as an SRC in one of the CUL-2-based E3 complexes to promote meiosis; although our data suggests that ZYG-11 does not function with CUL-2 to promote germ cell proliferation. As the localization of PAR-2 independently of MT-foci was not observed in zyg-11(RNAi) embryos, currently there is no evidence that ZYG-11 functions to restrain PAR-2 localization in regions distant from MT-organizing centers.

Meiotic timing and AP polarity

During the extended meiosis II in cul-2 mutants, the majority of embryos accumulate stable PAR-2 on the anterior cortex near the meiotic spindle. It has been shown that in APC/C mutant zygotes arrested in meiosis I, the meiotic spindle can initiate the localization of PAR-2 onto the anterior cortex (Wallenfang and Seydoux, 2000). The disorganized `frayed' appearance of the meiotic spindle observed in older meiosis I-arrested embryos was proposed to initiate the PAR-2 reversal (Wallenfang and Seydoux, 2000). By contrast, our study found that stable anterior PAR-2 localizations occur with meiosis II spindles of normal morphology and with timing similar to the appearance of PAR-2 on the posterior cortex in wild type. In addition, we observed that P granules segregate toward the mislocalized PAR-2, that the SPCC migrates further than the meiotic pronucleus, and that the cell division plane is often asymmetrically positioned to produce a reversal in daughter cell size. None of these downstream polarity changes are observed in meiosis I-arrested APC/C mutants (Wallenfang and Seydoux, 2000), suggesting either that the meiosis II spindle is fundamentally different than the meiosis I spindle in effecting polarity or that these changes can only manifest upon entry into interphase.

Transient and weak PAR-2 patches occasionally arise on the anterior cortex during meiosis II in wild-type zygotes, which suggests that the intact meiosis II spindle has an intrinsic ability to direct PAR-2 localization (Table 2) (Boyd et al., 1996; Cuenca et al., 2003). We observed that the longer cul-2(RNAi) zygotes remained in meiosis, the more stabilized the anterior PAR-2 became and the less likely that a posterior PAR-2 patch would form. With only a modest meiosis II delay, a full reversal of PAR localization was observed, with PAR-2 exclusively on the anterior and the asymmetric cleavage of the zygote reversed (Table 2). The failure of PAR-2 to localize to the posterior in these cul-2(RNAi) embryos may derive in part from the aberrant migration of the SPCC to the anterior, which begins during meiosis (Fig. 5E). In cul-2(RNAi) embryos, sperm asters generally form during meiosis but are very small until meiosis is completed, presumably because of the effect of the MT-severing katanin MEI-1, which is degraded after meiosis (Pintard et al., 2003a). Therefore, the failure of the SPCC to form full mitotic asters while in the posterior may contribute to the lack of PAR-2 in that region. In total, our results suggest that only a short window of meiotic timing is compatible with the proper placement of the AP axis.

CUL-2 limits the inappropriate localization of PAR-2

The establishment of AP polarity in the C. elegans zygote has been proposed to be a MT-directed process (O'Connell et al., 2000; Wallenfang and Seydoux, 2000). In cul-2(RNAi) embryos, PAR-2 was observed to initiate cortical localization distant from the meiotic spindle or sperm DNA. Upon elimination of the meiotic spindle by RNAi depletion of the β-tubulin TBB-2, lateral cortical PAR-2 was still observed in cul-2(RNAi) embryos at significant although somewhat reduced levels. This suggests that the aberrant PAR-2 localization is not dependent on the presence of a MT organizing center, although MTs may potentiate the localization. The ectopic PAR-2 localization in cul-2 mutants does not appear to arise as a secondary consequence of the meiosis II delay, as the depletion of CUL-2 produces a significant increase in ectopic PAR-2 localization even in an emb-30 mutant background in which embryos are arrested in metaphase I. Furthermore, although ZYG-11 is also required for meiotic progression, the localization of PAR-2 distant from the meiotic spindle was not observed in zyg-11(RNAi) embryos. We propose that CUL-2 has a separate function to restrain ectopic PAR-2 association with the cortex during meiosis. In combination with the recently described role for CUL-2 in degrading germ cell determinants in anterior cells of the early embryo (DeRenzo et al., 2003), our results indicate that CUL-2 promotes the initiation of AP polarity through multiple mechanisms.