Developmental defects observed in hypomorphic anaphase-promoting complex mutants are linked to cell cycle abnormalities

The anaphase-promoting complex or cyclosome (APC/C) is an evolutionarily conserved, multi-subunit E3 ubiquitin ligase that irreversibly drives cells through and subsequently out of mitosis(Harper et al., 2002;Peters, 2002). Like other E3 ubiquitin ligases, the APC/C functions by poly-ubiquitinating its specific target proteins, thus marking them for destruction by the 26S proteasome. Models of APC/C function, based largely on studies in unicellular eukaryotes,suggest that the APC/C acts at two key points during mitosis. Initially, the APC/C drives cells from metaphase to anaphase by poly-ubiquitinating securin,the inhibitory binding partner of separase(Cohen-Fix et al., 1996;Funabiki et al., 1996b). Liberated separase can, in turn, proteolytically cleave cohesin proteins between sister chromatids, thus enabling microtubule-associated forces to pull the separated sisters to opposing spindle poles(Uhlmann et al., 1999;Uhlmann et al., 2000). The APC/C then promotes mitotic exit by ubiquitinating the M-phase cyclins(reviewed by Morgan, 1999). More recently, the APC/C has been shown to function in separating paired homologs during meiosis I (Cooper et al.,2000; Davis et al.,2002; Furuta et al.,2000; Golden et al.,2000) and sister chromatids during meiosis II(Peter et al., 2001;Taieb et al., 2001). To date,SnoN, a negative regulator of TGFβ signaling(Stroschein et al., 2001;Wan et al., 2001) is the only documented non-cell cycle target of the APC/C. In contrast, SCF, the E3 ligase that drives the G1 to S transition, has a wide variety of non cell-cycle targets (reviewed by Deshaies,1999; Hershko and Ciechanover,1998).

Nonetheless, the APC/C is likely to have several other targets in addition to securin and the M-phase cyclins. For instance, the APC/C clearly regulates spindle and chromosome dynamics beyond disrupting sister chromatid cohesion since its known spindle-related substrates include Ase1(Juang et al., 1997;Visintin et al., 1997)Aurora-A (Castro et al., 2002;Taguchi et al., 2002) and several specialized kinesins (Funabiki and Murray, 2000; Gordon and Roof,2001). Recent studies have also implicated developmental roles for the APC/C during spore formation in yeast(Asakawa et al., 2001;Blanco et al., 2001) and embryonic axis formation in C. elegans(Rappleye et al., 2002). However, in the absence of identified developmentally relevant substrates, it is unclear whether the APC/C regulates these developmental processes directly or indirectly via its known cell cycle substrates.

Several features make C. elegans an excellent model system not only for analyzing the role of the APC/C in cell cycle progression but also for investigating potential links between the cell cycle and development. For instance, the developmental impact of cell cycle mutants that truncate or alter the timing of cell lineages can be studied in the context of a known,invariant cell lineage and well-understood developmental signaling interactions (reviewed by Lambie,2002). In fact, the late-developing everted vulva phenotype inemb-30/apc-4 mutants, and presumably other APC/C mutants, has been shown to be associated with extended M-phase delays and variable lineage truncations within the vulva cell lineage(Furuta et al., 2000). In a different multicellular context, the C. elegans germline is a highly proliferative tissue, which like the germline and imaginal discs inDrosophila, is particularly sensitive to cell cycle defects(Albertson et al., 1978;Glover, 1989). Given that the 1000+ cells that compose the mature, syncytial germline arise from just two cells within hatching L1 larva (Hirsh et al., 1976), it is not surprising that many APC/C mutants develop severely reduced germlines due to mitotic defects(Furuta et al., 2000;Golden et al., 2000).

Lastly, the powerful combination of excellent cytology, well-established genetics and RNA interference (RNAi) methodology can be used to analyze how the APC/C functions at the cellular level to support cell cycle progression and development of the C. elegans zygote. The one-cell stage of these large, transparent embryos encompasses both meiotic divisions of the oocyte chromosomes and the subsequent events leading up to the first mitotic division. Thus, these zygotes can be used to study common but poorly understood modifications of the standard cell cycle including how cells transition between meiosis I (MI) and meiosis II (MII) in the absence of full M-phase exit (reviewed by Abrieu et al.,2001), and how cells exit meiosis II and enter pre-mitotic S phase in the apparent absence of G1 (King et al., 1994; Vidwans and Su,2001). In C. elegans, meiotic exit is accompanied by a striking change in microtubule dynamics. During the meiotic phase, oocyte chromosomes segregate on small, anastral, acentriolar spindles(Albertson and Thomson, 1993),but, upon meiotic exit, the sperm centrosomes begin to nucleate microtubules as nearby cytoplasmic microtubules disappear(Clandinin and Mains, 1993). These developing sperm asters also specify the position of the embryonic posterior (Goldstein and Hird,1996; O'Connell et al.,2000; Sadler and Shakes,2000; Wallenfang and Seydoux,2000) and thus microtubules may serve as a common link between cell cycle progression and the zygote's developmental program.

The role of the APC/C in the various cell cycle and developmental events of the one-cell C. elegans embryo is just beginning to be elucidated. When any one of several APC/C subunits is significantly depleted, the affected embryos experience both a cell cycle block in metaphase of meiosis I and a corresponding developmental block (Davis et al., 2002; Furuta et al.,2000; Golden et al.,2000; Kitagawa et al.,2002). To date, this metaphase I block has precluded the analysis of C. elegans APC/C functions in meiosis II, meiotic exit or the first mitotic division of the embryo.

Here, we demonstrate that mat-1, a gene known to be involved in meiosis I metaphase to anaphase transition(Golden et al., 2000), encodes the CDC27/APC3 subunit of the APC/C. To analyze late and possibly novel APC/C functions, we focused on specific hypomorphic defects exhibited by seven temperature-sensitive mat-1 alleles grown under a variety of temperature-shift regimes. These studies, in conjunction with RNAi dosage studies and combinations of APC/C double mutants, demonstrate that different levels of APC/C activity result in distinct meiotic and post-embryonic phenotypes. Our analysis not only reveals new roles for the C. elegans APC/C in meiotic spindle dynamics, meiosis II chromosome separation, and meiotic exit, but it also extends our understanding of the male tail defects in APC/C mutants. Importantly, specific APC/C-related cell cycle defects were found to correlate with predictable and distinct developmental abnormalities suggesting that at least some of the previously reported developmental defects (Lyczak et al., 2002; Rappleye et al.,2002) might be secondary consequences of meiotic progression and exit abnormalities.

Six alleles of mat-1 were isolated in a large-scale genetic screen for temperature-sensitive mutants that arrest during the MI metaphase to anaphase transition (Golden et al.,2000), while a seventh allele (ye121) was subsequently isolated in an independent screen for osmotically sensitive mutants(Rappleye et al., 2002). To facilitate the analysis of male phenotypes, genetic doubles of mat-1alleles were constructed with him-5(e1490) (V) orhim-8(e1489) (IV); all male phenotypes were confirmed inhim/+ strains.

Doubly marked mat-1(ax227ts); mat-x or mat-1(ax227ts);emb-x strains were constructed as follows: unc-38(x20)mat-1(ax227ts) or mat-1(ax227ts) dpy-5(e61) L4 hermaphrodites were mated with N2 or him-8(e1489) males. Non-Unc and non-Dpy male progeny from these crosses were mated with L4 hermaphrodites of the following genotypes: mat-2(or170ts) unc-4(e120), mat-3(ax148ts) dpy-1(e1),mat-3(or180ts) dpy-1(e1), emb-27(g48ts) unc-4(e120), andemb-30(g53ts) dpy-17(e164). Five to ten non-Unc or non-Dpy cross progeny were picked and from the next generation, the phenotypes of the DpyUnc double mutant animals were determined. In all cases, double mutant strains could be identified, though many of the double mutant strains could not be maintained. All of the above crosses were carried out at the permissive temperature of 15°C.

For the mat-1 allele ax212ts, double mutants were constructed with an unmarked ax212ts allele. mat-1(ax212ts);him-5(e1490) males were mated with marked L4 hermaphrodites listed above. Non-Unc or non-Dpy F₁ progeny were picked, and then 16 F₂ Unc or Dpy animals were picked to separate plates (thus homozygous for the second mat or emb loci). One quarter of such animals should also be homozygous for ax212ts. These plates were examined for the presence of candidate double mutants. Again, these experiments were carried out exclusively at the permissive temperature(15°C).

For the mat-1(ye121ts); mat-3(or180ts) double mutant,mat-1(ye121ts); him-8 (e1489) males were mated withmat-3(or180ts) dpy-1(e1) L4 hermaphrodites. From the progeny of non-Dpy F₁ animals, 51 F₂ Dpy Mat-3 animals were picked to separate plates. One quarter of such animals should also be homozygous forye121ts. These plates were examined for the presence of candidate double mutants at the permissive temperature. The same cross was performed and F₂ embryos (from non-Dpy F₁ mothers) were shifted to 25°C. mat-1; mat-3 double mutants were then identified among the Dpy F₂ adults.

To construct the mat-1(ax144ts); HIS2B::GFP line, unc-74(x19)mat-1(ax144ts) heterozygous males were mated into AZ212:unc-119(ed3); ruIs32 [pAZ132] III. This strain contains an integrated histone (H2B) GFP transgene expressed under the pie-1 promoter(Praitis et al., 2001). Unc-74 animals were examined in subsequent generations for GFP expression until a line segregating all GFP-positive animals was found. This unc-74(x19)mat-1(ax144ts); ruIs32 line was used for the images inFig. 1.

PCR primers containing T7 polymerase binding sites were used to amplify an∼850 bp fragment of the cDNA clone yk466b3, an EST clone corresponding to Y110A7A.17. This PCR fragment was used to synthesize dsRNA using the T7 Megascript Transcription kit (Ambion Inc., Austin, TX) according to manufacturer's instructions. dsRNA was injected into either wild-type animals(N2 strain) or animals containing the integrated histone H2B::GFP transgene(strain AZ212) (Praitis et al.,2001). After a 16- to 20-hour recovery period, gravid AZ212 mothers were directly examined for GFP expression in their embryos. In other experiments, gravid N2 or AZ212 mothers were dissected and their progeny were either examined directly for GFP expression or were fixed and stained for immunofluorescence according to previously published protocols(Golden et al., 2000). Similar results were observed when RNAi experiments were performed by standard feeding protocols (Fraser et al.,2000; Timmons et al.,2001) using a construct created by Wallenfang and Seydoux(Wallenfang and Seydoux,2000). To observe hypomorphic mat-1 phenotypes(Fig. 1F), L4 animals were fed at 15°C and progeny were examined 20-24 hours later. For sep-1RNAi feeding, the bacterial strain expressing sep-1 dsRNA(Zipperlen et al., 2001) was used to feed the C. elegans strain AZ212 at 20°C and progeny were examined 12-16 hours later.

The open reading frame of Y110A7A.17 was confirmed by sequencing the yk10h2 and yk466b3 cDNAs. To obtain the complete 5′ UTR sequence, cDNA was PCR-amplified from wild-type animals using an SL1 5′ primer and a primer within the coding sequence. DNA sequencing revealed a cDNA identical to the Y110A7A.17 coding region predicted by the C. elegans sequencing consortium(http://www.wormbase.org/). cDNA synthesis and DNA sequencing of Y110A7A.17 from the variousmat-1 alleles was performed essentially as described by Golden et al.(Golden et al., 2000). The GenBank accession number for the full-length cDNA is AY081955.

For immunohistochemical analysis of embryos, germline and somatic tissue,adult animals were transferred to an 11.5 μl drop of egg buffer(Edgar, 1995) on a poly-l-lysine-subbed Color Frost Plus slide (Sigma-Aldrich, Fischer Scientific). Embryos and gonads were extruded using a 27.5 gauge needle,covered with a 24×50 mm SuperSlip coverslip (Fischer Scientific), freeze cracked and processed for antibody staining. For tubulin and phospho-histone H3 antibody staining, similar protocols and reagents were used as previously described (Golden et al.,2000).

The in utero defects of L1 upshifted animals were analyzed either with DIC optics or epifluorescence (for H2B::GFP transgenic animals) of living adult animals (Praitis et al., 2001;Sulston and Horvitz, 1977). The germline defects of L1 upshifted animals were analyzed by UV epifluorescence in whole mount, DAPI(4′,6′-diamidino-2-phenylindole)-stained animals that were fixed with Carnoy II fixative (6:3:1 ethanol/acetic acid/chloroform).

Animals were grown on a lawn of E. coli (strain OP50) on MYOB plates (Church et al., 1995). The analysis of mat-1 male phenotypes was carried out in the mutant background him-8(e1489), a mutation that increases the frequency of males by specifically increasing non-disjunction of the X chromosomes(Hodgkin et al., 1979); but which does not significantly enhance the mat-1 defects under restrictive conditions (Table 1). For the analysis of semi-synchronous L1 upshifted males and hermaphrodites, ten or more mat-1; him-8 hermaphrodites of a given allele were allowed to lay embryos on MYOB agar plates for 8-12 hours at 16°C before being removed. As soon as 75% of these embryos had hatched,the plate of largely L1 larvae was shifted to 25°C, and the animals were subsequently analyzed 42 hours later as young adults. Alternatively,developmentally arrested L1 larvae were collected from freshly starved plates and transferred to 25°C plates with food. In late-stage upshift experiments, animals were semi-synchronized by selecting L4 larvae from mixed-stage, 15°C populations and shifting them to the indicated temperature either as L4 larvae or as young adults for a specified duration.

The upshifted ax72 and ax520 homozygotes reported in this analysis were collected from the progeny of balanced heterozygotes, since these homozygotes produce dead embryos even when raised at 15°C.

To ensure that our analysis of embryos was confined to the mat-1maternal effects and not its paternal effects(Golden et al., 2000;Sadler and Shakes, 2000), themat-1 hermaphrodites in these studies were mated to wild-type males. Importantly, with the exception of ax520 and ax144ts, mat-1embryos exhibited similar defects regardless of whether their hermaphroditic parents had been upshifted before, during, or after the L4 larval period, whenC. elegans hermaphrodites generate their full complement of sperm.

For L4 upshift studies, 10-20 young L4 stage animals that were reared at 15°C were picked onto a fresh plate and shifted to specific temperatures ranging from 20 to 25°C for specified time intervals. For analysis,animals were dissected, fixed and stained for immunohistochemistry. The 15°C and 20°C incubations were done in a Precision Scientific Low Temperature Incubator 815 (±0.3-0.5°C). The 24°C studies were done in an Echotherm (TM) Bench Top Chilling Incubator (±0.01°C). The 25°C studies were done in a Percival Scientific 30 Series Incubator(±0.2°C). The temperatures were monitored with a Barnant RTD(Platinum) Datalogger.

As four other genes associated with Mat mutant phenotypes(Davis et al., 2002;Furuta et al., 2000;Golden et al., 2000) were subsequently shown to encode subunits of the APC/C, it seemed likely thatmat-1 also encoded an APC/C related gene. To test this, we searched the C. elegans sequence database for candidate genes within the genetic interval known to contain mat-1(Golden et al., 2000). Within this region is the single C. elegans homolog ofcdc-27/apc-3. To determine whether this cdc-27/apc-3 homolog(Y110A7A.17) would produce a metaphase I arrest phenotype, we performed RNAi with a corresponding EST clone (Fire et al., 1998; Hunter,1999). While the uteri of wild-type control hermaphrodites were filled with rapidly dividing embryos of many different ages(Fig. 1E,I), the uteri of hermaphrodites that had been injected with double stranded Ce-cdc-27RNA were filled with one-cell embryos (Fig. 1G). These cdc-27(RNAi) embryos looked like those frommat-1 mothers when examined either in utero(Fig. 1H) or as individual fixed specimens that had been processed for DAPI staining and tubulin immunofluorescence (Fig. 1B-D). The mat-1 and cdc-27(RNAi) embryos accumulated in an arrested but otherwise normal metaphase I state(Fig. 1A). In all cases, the oocyte chromosomes were aligned in a pentagonal metaphase plate on a barrel-shaped meiotic spindle, and the sperm chromosomes remained highly condensed. Such embryos also lacked polar bodies (the discarded products of the meiotic divisions) and had incompletely hardened eggshells.

To confirm that mat-1 encoded the C. elegans cdc-27/apc-3homolog, the Y110A7A.17 ORF from each of the seven mat-1 mutant alleles was PCR-amplified and sequenced. The sequencing data revealed unique missense mutations in the cdc-27 ORF for each mat-1 mutant allele (Fig. 2), indicating that mat-1 does indeed encode the C. elegans cdc-27 homolog. The seven mutations are scattered throughout the coding sequence; five mutations alter highly conserved residues while those in ax520 andye121 lie immediately adjacent to highly conserved regions. None of the mutations affect potential phosphorylation sites, but five occur within the conserved tetratricopeptide repeats (TPR) that are thought to mediate protein-protein interactions in a wide variety of proteins including three other APC/C components (Lamb et al.,1995).

When adult hermaphrodites of any of the seven mat-1 alleles were shifted to the restrictive temperature of 25.5°C for 12-24 hours as young adults, they produced metaphase I-arrested embryos that remained developmentally blocked in a meiotic one-cell state(Golden et al., 2000)(ye121, this study). This metaphase I-arrest phenotype is identical to the most extreme RNAi phenotype of any APC/C subunit(Davis et al., 2002;Furuta et al., 2000;Golden et al., 2000) and is likely to reflect the earliest function of the APC/C. However because this early block precludes the analysis of later functions, we reasoned that additional functions of the APC/C in either M-phase exit and/or early embryonic development could be studied in temperature-sensitive (ts)mat-1 hermaphrodites raised at various intermediate temperatures. In these experiments, L1 larvae were shifted to various temperatures and their resulting clutches were scored for overall embryonic lethality and whether the affected embryos were dying as meiotic one-cell stage or multicellular embryos(Table 1). Multicellular embryonic lethality was previously reported in a subset of what are now known to be APC/C mutant alleles (Golden et al.,2000; Rappleye et al.,2002), but our finding that multicellular embryonic lethality is associated with all seven mat-1 alleles at intermediate temperatures(Tables 1 and2), confirms this as a general,rather than an allele-specific, phenotype. Similar phenotypes were observed when the levels of maternal mat-1 mRNA were directly lowered by RNAi(Fig. 1F). When fed bacteria expressing mat-1/cdc-27 dsRNA, affected wild-type hermaphrodites initially produced viable offspring, but subsequently produced a brief burst of multicellular dead embryos (20-24 hours) before producing only one-cell arrested embryos. These results, along with the observation of similar phenotypes in other temperature-sensitive APC/C mutants(Golden et al., 2000;Rappleye et al., 2002),suggest that this phenotype is associated with intermediate levels of APC/C activity rather than the specific disruption of either MAT-1 or a particular MAT-1 subdomain.

Since C. elegans embryos presumably require the APC/C for both their meiotic and mitotic divisions, it was not obvious whether multicellular embryonic lethality, like the loss-of-function phenotype, would be specifically associated with defects in meiosis I or, alternatively with later defects in meiosis II or mitosis. To determine whether intermediate levels of APC/C activity were affecting a particular stage of early development, we scored the pattern of embryos within the uteri of mat-1hermaphrodites raised at various temperatures (see Materials and Methods). Because C. elegans oocytes are ovulated sequentially, the bilobed gonad of wild-type hermaphrodites produces a linear array of temporally ordered embryos within each half uterus. In wild-type hermaphrodites (andye121 hermaphrodites at 15°C), each half uterus typically contained 0-1 meiotic stage embryos followed by a single one-cell mitotic embryo and a series of progressively older multicellular embryos(Fig. 1I;Table 1). In contrast, the uteri of mat-1 hermaphrodites contain a striking increase in the number of meiotic embryos. At permissive temperatures, mat-1hermaphrodites produced primarily viable embryos and contained 1-2 (±1)meiotic embryos per half uterus (blue inTable 1). Since new oocytes are ovulated every 20 minutes (McCarter et al., 1999), this result suggested that normal embryonic development can tolerate minor (20-40 minute) delays in meiotic progression. Under semi-permissive conditions, mat-1 hermaphrodites accumulated 3-6 times as many meiotic embryos as same-temperature controls (green inTable 1;Fig. 1J,K) and, under fully restrictive conditions, mat-1 hermaphrodites produced only meiotically arrested one-cell embryos (MeOC; red inTable 1;Fig. 1L). Ovulation rates were unaffected and could not account for the observed accumulation of meiotic embryos (data not shown).

To test whether these meiotic embryos were accumulating specifically in meiosis I or, alternatively, at various stages throughout meiosis, control and mutant embryos were dissected from wild-type or mutant mothers and prepared for DNA (DAPI) staining and anti-tubulin immunofluorescence. Embryos isolated from the uteri of wild-type mothers contained approximately equal numbers of meiosis I (MI) and meiosis II (MII) embryos indicating that these stages are normally of equal duration (Table 2). Under permissive conditions, mat-1 hermaphrodites contained more meiotic embryos than wild-type controls (blue inTable 1), but roughly equal numbers of MI and MII embryos (Table 2). However, increasingly restrictive temperatures were associated with an increasing skew in the MI:MII ratios (green in Tables1 and2), and, at fully restrictive temperatures, only MI arrested embryos were observed.

Although semi-permissive temperatures were associated with significant mother to mother phenotypic variability (Table1 and2), inspection of the overall MI:MII ratios and polar body numbers revealed two phenotypic subgroups amongst multiple mat-1 alleles. Less affected mothers produced MI, MII, and multicellular embryos, but owing to an extended MI, the MI:MII ratio was 3:1. Older embryos within such mothers had two positionally distinct polar bodies(2PB), indicating that both meiotic divisions had occurred. More severely affected mothers produced only MI and multicellular embryos. Whether examined by DAPI staining or DIC optics, these older embryos had only a single polar body (1PB), suggesting that they had exited meiosis after completing only a single meiotic division. Likewise, pronuclear stage embryos lacked the extra maternal pronucleus that would be expected if a second meiotic division had occurred in the absence of cytokinesis. Under conditions favoring the production of 1PB embryos, a few mothers within the population produce MII embryos, resulting in an overall MI:MII ratio of 9:1(Table 2), however such variation was not observed amongst sibling embryos from the same mother. For clarity, we will hereafter refer to individual embryos as belonging to the`1PB class' or `2PB class'.

In contrast to previous reports(Rappleye et al., 2002),hypomorphic mat-1 embryos proved to exhibit striking meiotic defects. In wild-type C. elegans embryos, there are four basic stages to each meiotic division (Fig. 3A-D). In mature MI oocytes, nuclear envelope breakdown occurs just prior to fertilization. Fertilization is followed by the rapid assembly of an anastral meiotic spindle and the congression and alignment of paired oocyte homologs on a metaphase plate (Fig. 3A). During early anaphase I (Fig. 3B) the homologs separate as the spindle shortens and then rotates to a position perpendicular to the cortex(Albertson, 1984). By late anaphase I (Fig. 3C), a prominent microtubule bundle lies almost entirely between the separated homologs and possesses a morphology reminiscent of a cinched haystack. During telophase I (Fig. 3D) the individual chromosomes coalesce into two opposing chromatin masses as the spindle splits during the highly asymmetric cell division that forms the first polar body. This sequence is reiterated during MII but is immediately followed by nuclear envelope formation and a round of DNA synthesis. In the absence of a functional APC/C, embryos remained blocked in metaphase I; with time, the meiotic spindle disassembles but nuclear envelopes never reassemble(Golden et al., 2000).

As in the fully restrictive meiotic arrest class, mat-1 embryos of the 1PB class assembled their paired homologs into a normal metaphase I spindle (37/37; Fig. 3E). However, the subsequent steps of spindle rotation and polar body formation occurred in the absence of homolog separation(Fig. 3F′). As a result,anaphase and telophase-like spindles had prominent chromosome bridges between the two DNA masses (Fig. 3G,H). Such polar bodies were stable but frequently contained misshapen and fragmented DNA, suggesting that the process of polar body formation `cuts'these DNA bridges in half, as happens during the mitotic divisions of hypomorphic S. pombe APC/C mutants(Funabiki et al., 1996a). Consistent with meiotic chromosome segregation defects, the maternal pronuclei of early post-meiotic embryos also were frequently misshapen (data not shown). Similar defects were observed in both meiotic divisions of 2PB class embryos. Analysis of MII embryos revealed relatively normal metaphase figures(Fig. 3I,I′), yet MII spindle rotation and second polar body formation occurred in the absence of sister chromatid separation (Fig. 3J), resulting in a MII `cut' phenotype. In addition, mutant spindles were frequently longer than wild type(Fig. 3K′). These MII defects suggest a role for the C. elegans APC/C in either spindle shortening or MII, although our data do not preclude the possibility that many or all of these MII defects are a secondary consequence of MI abnormalities. Importantly, we identified no semi-permissive temperature conditions that resulted in either metaphase II arrest or an accumulation of MII embryos.

In wild-type C. elegans embryos, there is a predictable and tight correlation between the developmental and cell cycle events of the one-cell stage (Sadler and Shakes,2000). Key developmental events include the formation of an impermeable, three layer eggshell during the late meiotic phase(Chitwood and Chitwood, 1974)and the post-meiotic establishment of the embryo's anterior-posterior (AP)axis (for reviews, see Gotta and Ahringer,2001; Lyczak et al.,2002; Pellettieri and Seydoux,2002). Importantly, AP polarization appears to be triggered, in part, by growth of the sperm-derived microtubule asters(O'Connell et al., 2000;Sadler and Shakes, 2000;Wallenfang and Seydoux, 2000). AP polarization culminates in a highly asymmetric first mitotic cell division that results in the formation of sister blastomeres that differ in size,synchrony, and the orientation of their cell divisions.

Metaphase I arrested mat-1/cdc-27 (RNAi) embryos fail to develop either an impermeable eggshell or a stable A-P axis, indicating that the early cell cycle block is coupled with a corresponding developmental block(Golden et al., 2000;Wallenfang and Seydoux, 2000)(this study). To determine the effect of a partial APC/C depletion, embryos from mat-1 mothers grown under semi-permissive conditions were analyzed for eggshell and polarization defects. Mutant embryos with two distinct polar bodies made eggshells that were impermeable to the lipid soluble DNA dye Hoechst 33248 whereas mutant embryos with a single polar body made Hoechst-permeable eggshells. To assess potential polarization defects,dividing sister blastomeres of either mutant or wild-type two-cell embryos were scored for their relative size, synchrony and spindle orientations(Table 3,Fig. 4). Compared to wild-type controls, hypomorphic mat-1 embryos either exhibited increased variability in the positioning of their first cleavage plane(Table 3) or, under increasingly stringent conditions, divided symmetrically. Symmetric divisions were strongly correlated with the presence of a single polar body (14/14;Fig. 4N,O). In general, sister blastomeres of such 1PB class embryos divided synchronously with both spindles oriented perpendicular to the long axis of the embryo(Fig. 4N,O). In contrast, those of 2PB class embryos were variably intermediate in their relative size,synchrony, and cleavage orientations (15/15;Fig. 4I,J). Lastly, although germline-specific P-granules segregated normally in most 2PB class embryos,P-granule segregation failed in both MI-arrested embryos and in symmetrically and synchronously dividing 1PB class embryos (data not shown). In summary, 1PB class embryos exhibited consistently more severe defects in the symmetry,timing, and cleavage orientations of their early cell divisions when compared with either 2PB class embryos or wild-type controls.

To determine whether the defects observed in dividing two-cell embryos were being presaged by abnormalities during the earlier pronuclear stage, we compared pronuclear-stage mutant embryos with same-stage control embryos using either DAPI/tubulin staining or DIC microscopy. During the pronuclear stage of wild-type embryos, the replicated sperm centrosomes expand and separate to lie on opposite sides of the male pronucleus, and the female and male pronuclei migrate to join in the posterior half of the embryo (seeFig. 4A-C for an early migration stage). In 12/12 2PB class mat-1 embryos(Fig. 4F-H), sperm aster maturation and pronuclear migration remained unaffected. In contrast, 1PB class mat-1 embryos exhibited numerous defects(Fig. 4K-M). In wild-type embryos, increased cytoplasmic flow rates immediately prior to the initiation of pronuclear migration promote the rapid relocation of the male pronucleus to the presumptive posterior end (Fig. 4A) (see also Goldstein and Hird, 1996). Such flows were absent in 1PB mutant embryos, and in 15/15 same-stage embryos, the male pronuclei failed to relocate from their initial cortical position to one of the two embryonic ends(Fig. 4K). In addition, both the expansion and separation of the sperm asters was delayed(Fig. 4L,M).

Displaced male pronuclei have also been reported in embryos partially depleted of separase, the key effector of metaphase APC/C activity(Rappleye et al., 2002). Our own observations revealed that wild-type mothers fed for a specific short time on sep-1 dsRNA-producing bacteria produce embryos that specifically fail to relocate their male pronuclei. However, in striking contrast to the earlier report, such defects were always associated with severe meiotic defects, namely the presence of a single polar body (12/12;Fig. 5).

To determine if the pronuclear and cleavage defects in 1PB classmat-1 embryos could be a secondary consequence of even earlier meiotic exit defects, we compared the process of meiotic exit in wild-type and 1PB class embryos. In mitotic cells of other organisms, the APC/C is known to drive M-phase exit by poly-ubiquitinating B-type cyclins and activating the mitotic exit network (McCollum and Gould,2001). In one-cell C. elegans embryos, exit out of MII is complex because it involves both a direct M to S transition (no apparent G1)and the final conversion of the fertilized oocyte into a zygote. This M to S phase transition is marked by several events including the disappearance of M-phase cell-cycle markers such as phosphorylated histone H3 (p-H3)(Fig. 6), the extensive remodeling of the sperm chromatin, the disassembly of the meiotic spindle, and the duplication and nucleation of the paternally derived centrosomes(Albertson and Thomson, 1993;Sadler and Shakes, 2000).

In wild-type embryos, late anaphase II can be distinguished from post-meiotic S phase on the basis of DAPI/tubulin/p-H3 staining patterns(Fig. 6). During late anaphase II, the highly condensed, maternally derived sister chromatids stain brightly with anti-p-H3 (Fig. 6C,E). In contrast, the highly condensed, paternally derived chromatin mass fails to stain with anti-p-H3, and its associated centrosome remains quiescent(Fig. 6B,D,F). During M-phase exit, each chromatin set decondenses and becomes enclosed within a pronuclear envelope (Fig. 6I,J). At the same time, p-H3 staining is rapidly lost from the female pronucleus while the chromatin within the second polar body continues to stain brightly(Fig. 6K). Also at this stage,the newly formed second polar body is associated with a compact, flattened spindle remnant at the cortical surface(Fig. 6G), while the sperm centrosome duplicates and begins to nucleate microtubules(Fig. 6H).

Analysis of M-phase exit in 1PB class embryos revealed several striking abnormalities. In particular, the post-meiotic maternal chromosomes continued to stain with p-H3 antibodies, albeit unevenly, even after nuclear envelope reformation (10/10; Fig. 6Q). In addition, the male pronucleus sometimes stained aberrantly with p-H3 (5/10;Fig. 6R) even after centrosome duplication and the initiation of aster formation(Fig. 6N). In addition, such embryos frequently contained large, disorganized meiotic spindle remnants that, unlike their wild-type counterparts, appeared to extend into the cortical interior (Fig. 6M). Such defects suggest that the aberrant meiotic exit of the 1PB class directly out of MI disrupts the normal coordination of the various cellular events that accompany the process of meiotic exit and that set the stage for the proper development of the zygote.

In previous studies, a subset of temperature-sensitive mat alleles were found to exhibit defects not only in the meiotic divisions of the post-fertilization embryo but also in the mitotically proliferating germline,male tail and hermaphrodite vulva (Furuta et al., 2000; Golden et al.,2000). Amongst mat-1 alleles, only two alleles(ax144 and ax520) proved to result in significant germline proliferation defects (Golden et al.,2000) (Table 4;Fig. 7) and only three alleles(ax144, ax520 and ax212) produced significant defects in the male tail and hermaphrodite vulva (Table 4; Fig. 8). In contrast, other alleles either resulted in no mitotic defects or moderate defects in only a subset of the population(Table 4). Although such results could suggest that the other mat-1 alleles were meiotic specific, the discovery that the mat-1 molecular lesions were scattered throughout the protein (Fig. 2) suggested that the meiotic divisions of the oocyte may simply be more sensitive to a partial loss of APC/C function. If the apparent allele-specificity was merely a matter of dosage, additional mat-1alleles might be expected to display mitotic defects in double mutant combinations with other APC/C genes.

To test whether mutations in other APC/C subunits could enhance the phenotype of the relatively unaffected alleles and thus uncover `latent' APC/C mitotic defects, we selected ax212 and ax227 as appropriate Mel (maternal-effect lethal) alleles. At 15°C ax212 and ax227 hermaphrodites are fully fertile,and at 25.5°C they produce clutches of meiotic one-cell embryos but do not exhibit the sterility (ste) associated with significant germline and somatic defects. In these studies, we constructed doubles of mat-1(ax212) andmat-1(ax227) with alleles of mat-2/apc-1, mat-3/cdc-23,emb-27/cdc-16 and emb-30/apc-4 that are known to be Mel but not sterile in L1 upshift experiments (Golden et al., 2000). At 15°C, all of the double mutants developed to adulthood, but some of these double mutant mothers produced clutches of dead embryos (Table 5) suggesting that, even at 15°C, APC/C activity in mat-1(ax227) andmat-1(ax212) animals is not completely normal. To specifically test whether these double mutant combinations would exhibit `latent' germline defects, ax227 double mutants from heterozygous parents were shifted to the restrictive temperature as embryos or L1 larvae. While all of the double mutants developed through the adult stage, many were sterile(Table 5). Likewise,mat-1(ye121); mat-3(or180) doubles were also sterile at 25°C. Taken together, these results indicate that the molecular lesions inax212, ax227 and ye121 are not meiotic-specific, even though, as single mutants, these alleles provide sufficient APC/C function to support normal germline proliferation. Although we did not observe enhancement of male tail and hermaphrodite vulva defects in these mat-1 double mutant experiments, these somatic mitotic defects were enhanced in the double APC/C Mel combination of emb-27(g48) and emb-30(g53) (P.S.,unpublished).

In previous studies, defects in the mitotic proliferation of germline stem cells were associated with mitotic metaphase I blocks and/or M-phase delays(Furuta et al., 2000;Golden et al., 2000) as were everted vulva (Evl) defects in emb-30/apc-4 mutants(Furuta et al., 2000). Format-1 Evl phenotype, see Fig. 8I,J. However, no such direct link had been made between the male tail phenotype and cell cycle defects. To test whether the mat-1 male tail defects (Fig. 8A-D) also correlated with mitotic cell cycle defects, L1 larvae from either the severely affected ax144; him-8 strain or him-8 controls were shifted to 25.5°C for periods ranging from 18 to 36 hours. The animals were dissected to separate the gonad from the soma, and then processed for immunostaining with anti-p-H3 antibody. The somatic staining patterns in theax144 samples were indistinguishable from the controls during the 18-and 24-hour time points. However, a dramatic excess of p-H3 staining cells was found in the posterior of ax144 L4 male larvae collected 30 to 36 hours (L4) after upshift (Fig. 8H,K) when compared to the wild-type controls(Fig. 8F,K). Importantly, only some of these p-H3-positive cells were in metaphase, suggesting that the APC/C defect is prolonging more than just the metaphase to anaphase transition. Because both the final cell divisions and morphogenetic events of male tail formation occur during this same L4 period, it is likely that at least some of these p-H3-positive cells were progenitors of male tail cells. Thus, as in the vulva, the morphological defects in the mutant male tails are likely to stem from a combination of M-phase delays and potential truncations within these late-dividing cell lineages.

The APC/C is widely recognized as the key E3 ubiquitin ligase that drives cells through and subsequently out of mitosis(Peters, 2002). In the present study, we have employed C. elegans temperature-sensitive mutants in the APC/C subunit MAT-1/CDC-27 to reveal a wide range of defects at both the organ and cellular level that can result from partial depletions of APC/C activity. When MAT-1 was significantly depleted by either shiftingmat-1(ts) mothers to 25.5°C or injecting wild-type mothers with double-stranded mat-1/cdc-27 RNA, the affected oocytes were fertilized but arrested as one-cell embryos with their paired homologs locked in metaphase of meiosis I. In contrast, when various temperature shift regimes were used to create a graded MAT-1 depletion series, the resulting one-cell embryos exhibited an extended meiotic phase as well as several dosage-dependent meiotic defects ranging from incomplete chromosome separation and abnormal meiotic spindle dynamics to a complete bypass of the second meiotic division. While metaphase I-arrested mat-1 embryos exhibit a corresponding developmental block at the meiotic one-cell stage, hypomorphicmat-1 embryos exhibited dosage-dependent meiotic defects that correlated with specific defects in early development(Fig. 9). These developmental defects included altered pronuclear migration patterns, aberrant polarization,and alterations in the normal timing and orientation of the embryo's mitotic cell divisions. When mat-1 mutants were upshifted as L1 larvae,additional somatic and germline defects were observed in three of the seven alleles; other alleles could be induced to exhibit these phenotypes in double mutant combinations with other APC/C subunits.

Sequence analysis of the seven mutant alleles revealed molecular lesions predominantly within the TPR domains but otherwise scattered throughout themat-1 coding region, and our subsequent phenotypic analysis revealed no obvious correlation between the location of these lesions and the severity of the mutant defects. In particular, the mutations in two phenotypically similar alleles, ax161 and ax227, are well separated within the linear sequence of MAT-1 whereas two phenotypically dissimilar, alleles,ax520 and ax212, are almost adjacent. These results combined with the previous findings that Mat embryos contain sufficient functional,maternal APC/C to develop into sterile adults even when upshifted to 25°C as two-cell embryos (Golden et al.,2000), suggest that all of these alleles, except ax72 andax520, are temperature sensitive for protein folding. In such mutants, high temperatures destabilize an intermediate in the folding pathway without altering the functions of the folded state(King et al., 1996). None of the seven mat-1 alleles are molecular nulls, however, we believe that the mat-1 metaphase I arrest phenotype reflects an absence of APC/C function within the oocyte since RNAi embryos of seven different APC/C subunits display an identical phenotype (this study)(Davis et al., 2002). Contrary to our initial expectations, none of these mat-1 alleles, except perhaps ax72, proved to be meiotic specific. Perhaps mutations in TPR domains are poor candidates as single function lesions if the TPR domains in MAT-1/CDC-27 serve primarily to glue the complex together. Theoretically,mutations of phosphorylation sites might cause single function defects, but no such lesions are represented within our current mutant collection.

Although none of our mat-1 alleles are molecular nulls, homozygous null embryos of other APC/C subunits survive embryogenesis and develop into sterile hermaphrodites with everted vulva [EMB-30/APC-4(Furuta et al., 2000),MAT-2/APC-1 (Davis et al.,2002), MAT-3/CDC-23 (D. Garbe and M. Sundaram, personal communication), and APC-11 (A.G., unpublished)]. Furthermore, similar phenotypes have been described for null mutants of other cell cycle genes including cyclin D (Boxem and van den Heuvel, 2001; Park and Krause,1999) cyclin E (Fay and Han,2000) CDK-4 (Boxem and van den Heuvel, 2001; Park and Krause,1999) and CDK-1 (Boxem et al.,1999). Presumably these cell cycle mutants survive embryogenesis and larval development because of high levels of persistent maternal mRNA or protein stores, and, in the case of temperature-sensitive mutants, the resistance of previously synthesized and complexed proteins to unfold at restrictive temperatures. The biology of C. elegans may also explain the surprisingly mild phenotypic defects. Since C. elegans embryos develop in the absence of net growth, maternal stores of cell cycle proteins may persist at sufficient concentrations to support cell divisions throughout embryogenesis. Likewise, surprisingly normal larval development can occur in the complete absence of larval cell divisions, albeit the resulting adults are sterile and uncoordinated (Albertson et al., 1978). Similarly, many Drosophila cell cycle mutants(Gatti and Baker, 1989),including those in APC5 (Bentley et al.,2002) survive embryogenesis and larval development before dying during the prepupal stage with underdeveloped imaginal discs.

The germline, vulva, and male tail may be particularly sensitive to defects in the APC/C and other cell cycle genes because all three are generated through mitotic proliferation of embryonic blast cells, with the majority of divisions occurring during the last two larval stages. In the case of the germline, two germline progenitor cells proliferate more than any other blast cells in the hatching L1 larvae. The cell lineages giving rise to the developing male tail and hermaphrodite vulva do not involve more cell divisions than those giving rise to other somatic structures, but these divisions are sufficiently late that maternal stores may be inadequate. In addition, proper development of these organs requires a high degree of coordinated signaling interactions(Lambie, 2002) and thus may be particularly sensitive to M-phase delays and/or abnormalities that result in either truncation or altered timing of the cell lineages. Interestingly,mat-1 mutants also exhibit gonad migration defects at low frequency(M.A. and D.S., unpublished), which may stem from cell signaling defects during gonadal development. Taken together, these studies suggest that, in multicellular organisms, the sensitivity of a particular tissue to hypomorphic APC/C levels will depend on total cell proliferation and the degree to which coordinated cell division is linked to morphogenesis.

APC/C dosage studies revealed a hierarchy of APC/C functions. Upon partial depletion of APC/C levels, chromosome separation is affected more than spindle shortening, and spindle shortening is more affected than either spindle rotation or polar body formation. A similar hierarchy has been observed in fission yeast; APC/C null mutants arrest in metaphase whereas most ts mutants exhibit a `cut' phenotype in which cytokinesis proceeds in the absence of chromosome segregation (Chang et al.,2001). To date, the basis for this differential sensitivity remains unclear. In other systems, APC/C substrate specificity is regulated by its associated WD repeat proteins. Current models suggest that securin is targeted primarily by APC^Cdc20(Cohen-Fix et al., 1996;Funabiki et al., 1996b)whereas cyclin B is targeted primarily by APC^Cdh1, although recent studies suggest that the specific roles of these WD proteins may be more variable (Hsu et al., 2002;Sigrist and Lehner, 1997;Stegmeier et al., 2002;Yamaguchi et al., 2000;Yeong et al., 2000). Recent studies also indicate that APC^Cdc20/Fizzy is the sole form of the APC/C during C. elegans meiosis; fzy-1(RNAi) embryos arrest in metaphase I (Kitagawa et al.,2002) whereas cdh-1/fzr-1(RNAi) embryos develop into sterile adults (Fay et al.,2002). If the CDH1/FZR-1 ortholog does not, in fact, function in meiotic C. elegans embryos, the observed hierarchy may reflect a differential affinity of APC^Cdc20 for its various targets. Alternatively, if C. elegans has additional, yet undiscovered meiotic specificity factors (WD repeat proteins), perhaps only the higher affinity complexes function at low APC/C levels.

Our studies have also revealed a potentially novel APC/C function in late anaphase meiotic spindle shortening. In mitotic cells, APC/C contributes to changes in anaphase spindle morphology in part by ubiquitinatingAse-1 (Juang et al.,1997; Visintin et al.,1997) and various kinesins(Gordon and Roof, 2001). Whether meiotic spindle shortening is directed through the same or unique APC/C substrates will be the subject of future studies.

One of the more intriguing findings of this study was the differential effect that APC/C depletions had on the two meiotic divisions(Fig. 9). While partial depletions of APC/C activity disrupt the separation of paired homologs in MI and sister chromatids in MII, extended delays and/or arrest of the metaphase to anaphase transition occurred only during MI. While both the 1PB and 2PB class embryos experience extended MI delays, it was surprising that the MII chromosome separation defects in 2PB class embryos were never coupled with a metaphase II arrest or a prolonged MII cell cycle. One explanation for this result is that meiosis I is as sensitive, or more sensitive, than meiosis II to decreased APC/C activity, and thus it may not be possible experimentally to observe a MII arrest (since such temperatures result in a MI arrest). Alternatively, the differential impact reflects true differences between the two meiotic divisions. More specifically, oocytes normally transition between MI and MII without fully exiting M-phase; whereas exit from MII absolutely requires the destruction of M-phase cyclins and is accompanied by nuclear envelope reformation (Kobayashi et al.,1991; Minshull et al.,1991). While mitotically dividing cells drive M-phase with APC^Cdh1 and the G1 to S transition with SCF(Deshaies, 1999), one-cellC. elegans embryos apparently lack a G1 stage and thus the final stages of MII exit in C. elegans and potentially other embryos could be driven not by the APC/C, but rather by SCF or another cullin/RING finger complex. Consistent with this model, RNAi depletion of a CUL-2-containing complex results in severe MII, but not MI, delays (E. Kipreos, personal communication).

In previous studies, we and others have reported that complete and/or severe depletions of APC/C activity that block cell cycle progression past metaphase I also block the developmental processes of eggshell maturation(Golden et al., 2000) and the establishment of a stable A-P axis(Wallenfang and Seydoux,2000). Subsequent analysis of separase, the cohesin-cleaving protease that is indirectly activated by APC^Cdc20 activity,indicates that the depletion of separase activity alone is sufficient to block these two developmental events (Siomos et al., 2001; Rappleye et al.,2002) (this study). Importantly, our current results call into question the proposal that either the APC/C or separase function as direct regulators of either eggshell hardening or A-P polarization activity. In our APC/C and separase studies, defects in A-P polarization and eggshell hardening could only be generated under conditions in which the embryos exhibited severe meiotic defects (Fig. 9). In our more extensive APC/C dosage studies, not only did these 1PB class embryos fail to separate their homologs during metaphase I, but they skipped meiosis II altogether. More importantly, this drastic alteration in normal cell cycle progression disrupted the normally clean transition from meiosis to zygotic S phase. Notable defects include slow disassembly of the meiotic spindle and delayed maturation of the sperm asters. Given that the polarity machinery is thought to cue on microtubule free-ends(Wallenfang and Seydoux,2000), the co-existence and aberrant placement of these two microtubule structures might be sufficient to disrupt polarity. Although our studies cannot specifically rule out a direct connection between separase function and polarity, the observed transition defects provide an explanation of the APC/C-associated polarity defects that does not require the existence of novel, non-cell cycle roles for the APC/C and/or separase.

We would like to thank Yuji Kohara (National Institute of Genetics, Japan)for providing cdc-27 cDNA clones, Matt Wallenfang and Geraldine Seydoux for the cdc-27 RNAi feeding construct, Raffi Aroian for theye121 allele, Judith Austin for the integrated histone H2B::GFP fusion transgenic line, the C. elegans Genetics Center (University of Minnesota), funded by the NIH National Center for Research Resources, for providing strains, and Michael Krause and Thomas Brodigan for assistance with sequencing. We also are grateful to Meera Sundaram, Risa Kitagawa, Ann Rose and Edward Kipreos for sharing unpublished data. Special thanks go to Barry Chestnut and James LaRue for technical assistance, and Lizabeth Allison, Orna Cohen-Fix, April Robbins, Ritu Argawal, Karen Ross, Edward Davis, Barry Chestnut, Kevin O'Connell, Geraldine Seydoux, and members of the Bethesda/DC Worm Club for feedback and advice during the course of these studies. This work was supported in part by NIH grant GM060359-01 (to D.C.S.) and a College of William and Mary Faculty Research Assignment Grant (to D.C.S.).