Two genes, originally identified in genetic screens for Caenorhabditis elegans mutants that arrest in metaphase of meiosis I, prove to encode subunits of the anaphase-promoting complex or cyclosome (APC/C). RNA interference studies reveal that these and other APC/C subunits are essential for the segregation of chromosomal homologs during meiosis I. Further, chromosome segregation during meiosis I requires APC/C functions in addition to the release of sister chromatid cohesion.
IN mitotically dividing cells, loss of sister chromatid cohesion during the metaphase-to-anaphase transition triggers the segregation of sisters to opposite poles (reviewed in Nasmyth 1999). During meiosis I, reductional segregation of chromosomal homologs is achieved by restricting the loss of sister chromatid cohesion to chromatid arms. During meiosis II, equational segregation of sister chromatids occurs as the remaining centromeric cohesions are lost (reviewed in Miyazaki and Orr-Weaver 1994). During mitosis, a multisubunit E3 ubiquitin ligase known as the anaphase-promoting complex or cyclosome (APC/C) facilitates both the loss of sister chromatid cohesion and mitotic exit by targeting several mitotic proteins for destruction (Kinget al. 1996; Zachariae and Nasmyth 1999). Among these targets is securin, whose destruction frees its normal binding partner, separin, to cleave the cohesion proteins that hold together sister chromatids (Uhlmannet al. 1999). Although the role of APC/C in separating sister chromatids during mitosis is thought to be universal, its role in separating chromosomal homologs during meiosis I remains controversial. Recently, genetic analysis of two different APC/C subunits in Caenorhabditis elegans revealed a requirement for APC/C in the meiosis I segregation of chromosomal homologs (Goldenet al. 2000; Furutaet al. 2000). However, conflicting studies in Xenopus oocytes suggested that APC/C may be dispensable during meiosis I and required for the meiosis II segregation of sister chromatids only (Peteret al. 2001; Taiebet al. 2001).
In previous studies, we identified 32 temperature-sensitive maternal-effect embryonic lethal mutants in C. elegans that arrest as fertilized one-cell embryos, which are blocked at metaphase of the oocyte's first meiotic division (Goldenet al. 2000). Seven of these mutants were new alleles of emb-27 and emb-30, which proved to encode CDC-16(APC-6) (Goldenet al. 2000) and APC-4 (Furutaet al. 2000), respectively. These findings not only demonstrated a role for several highly conserved APC/C subunits during meiosis I but also suggested that mutations in other APC/C subunits might be included within this phenotypic class. The remaining 25 mutants fell into three novel complementation groups, termed mat (metaphase-to-anaphase transition defective): mat-1, mat-2, and mat-3 (Goldenet al. 2000). Here, we show that mat-2 and mat-3 encode additional APC/C subunits and that RNA-mediated interference (RNAi) of most APC/C subunits results in a meiotic metaphase I arrest. In addition, we use rec-8 RNAi to demonstrate that the loss of sister chromatid cohesion is insufficient to bypass a requirement for APC/C during meiosis I chromosome segregation, a striking result that implies that APC/C has additional, potentially novel, meiosis I targets.
The mat-2 and mat-3 genes encode subunits of the anaphase-promoting complex: The mat-2 mutations mapped to a 0.25 map unit region of LG II (Goldenet al. 2000; see http://www.wormbase.org/db/gene/mapping_data?name=mat-2). In this region, one locus, W10C6.1, encodes a predicted ortholog of the Saccharomyces cerevisiae Apc1p protein (Figure 1). While S. cerevisiae Apc1p is required for the metaphase-to-anaphase transition (Zachariaeet al. 1996), and APC-1 orthologs are present in many eukaryotic organisms (Starborget al. 1994; Peterset al. 1996; Yamashitaet al. 1996; Jorgensenet al. 2001), the protein lacks any obvious structural motifs that might indicate its biochemical function.
To test whether W10C6.1 was a mat-2 candidate, its in vivo function was analyzed using RNAi. Wild-type hermaphrodites were injected with double-stranded RNA (dsRNA) corresponding to exon 4 of W10C6.1. Under these conditions, dsRNA often depletes the maternal levels of the corresponding mRNA (Fireet al. 1998), and mothers receiving a specific dsRNA usually produce embryos with strong loss-of-function phenotypes (Guo and Kemphues 1996). Mothers injected with W10C6.1 dsRNA produced embryos that arrested at the meiotic one-cell stage with a phenotype similar to that of mat-2 (Table 1).
DNA sequencing of the apc-1 gene from all seven mat-2 strains revealed that each contained a single nucleotide substitution that altered the predicted amino acid sequence (Table 2). The mat-2 mutations were generally scattered throughout the protein sequence. However, the lesions in ax76 and or170 were in adjacent residues, and ax143 and or224 carried identical mutations. Each change occurred in a residue that was conserved in APC-1 orthologs from at least one other species (Figure 1). Therefore, mat-2 encodes the C. elegans APC-1. We hereafter refer to this gene as mat-2/apc-1.
Similarly, mat-3 mapped to the left arm of LG III (Goldenet al. 2000; see http://www.wormbase.org/db/gene/mapping_data?name=mat-3), where one locus, F10C5.1, encodes an ortholog of Cdc23p(APC8), an APC/C subunit in S. cerevisiae (Figure 2A; Doi and Doi 1990; Sikorskiet al. 1990; Zachariaeet al. 1996) and many other eukaryotic organisms (Figure 2A; Yuet al. 1998; Zhaoet al. 1998; Yamashitaet al. 1999).
As do all other CDC-23 orthologs, C. elegans CDC-23 contains nine 34-amino-acid degenerate repeats known as tetratricopeptide repeats (TPRs). TPRs were first described in S. cerevisiae Cdc23p (Sikorskiet al. 1990), but have since been discovered in many proteins with diverse functions (reviewed in Blatch and Lassle 1999). TPRs share little primary sequence identity but are predicted to form amphipathic α-helices (Sikorskiet al. 1990) that mediate protein-protein interactions (Lambet al. 1994; Blatch and Lassle 1999).
mat-3 proved to encode an ortholog of CDC-23. Specifically, RNAi of F10C5.1 caused mothers to produce meiotic one-cell arrested embryos (Figure 3; Table 1), and DNA sequencing of the cdc-23 gene from the 12 mat-3 strains revealed that each contained a single nucleotide substitution (Table 2). We hereafter refer to this gene as mat-3/cdc-23. mat-3/cdc-23 lesions occurred throughout the CDC-23 protein sequence (Figure 2A), but, of the 12 mutations, only 9 are unique (Table 2). Seven of the 9 unique mutations occurred within TPRs (Figure 2, A and B), 3 within TPR8.
In addition to the oocyte meiotic defects shared by all mat alleles (Goldenet al. 2000), some mat-3/cdc-23 alleles exhibit defects in germline proliferation (Table 2; Goldenet al. 2000). Interestingly, the four mutants with glutamic acid to lysine changes all exhibit germline defects, suggesting that the CDC-23 protein is sensitive to charge alterations. Furthermore, the E404K substitution in ax79 and ax136 is identical to the change found in the cdc23-37 temperature-sensitive mutant in S. cerevisiae (Sikorskiet al. 1993).
Molecular identification of additional APC/C genes in C. elegans: As it became apparent that mat genes could encode APC/C subunits (Table 1), we asked whether additional APC/C subunits function during the metaphase-to-anaphase transition of C. elegans meiosis I. Of the 8–12 known APC/C subunits in yeast and vertebrates, orthologs of several have been predicted on the basis of C. elegans genomic sequence data (Table 1). In particular, K06H7.6 and F35G12.9 were predicted as orthologs of the subunits APC-2 (Yuet al. 1998; Zachariaeet al. 1998) and APC-11 (Ohtaet al. 1999), respectively. We confirmed the gene predictions of the C. elegans Sequencing Consortium by sequencing the cDNAs yk268d10 (apc-2), yk566h7 (apc-11), and a reverse transcription-PCR clone of apc-11 (data not shown). The amino acid sequence of APC-2 exhibits 32% identity with S. cerevisiae Apc2p within the critical cullin domain but shows little similarity along the rest of the protein. Conversely, APC-11, the key catalytic subunit of the APC/C (Leversonet al. 2000), exhibits 44% identity along the entire protein. M163.4 was reported to encode APC-5 (Zachariaeet al. 1998) and has 11% identity and 20% similarity with S. cerevisiae Apc5p. BLAST searches revealed F15H10.3 as a convincing ortholog for APC-10 (27% identity, 45% similarity).
Other APC/C subunits required for the metaphase-to-anaphase transition of meiosis I: In a series of RNAi experiments, mothers injected with either apc-2 or apc-11 dsRNA produced embryos that arrested uniformly at the meiotic one-cell stage within 12–18 hr postinjection (Figure 3, c–e; Table 1). In contrast, mothers injected with either the apc-5 or apc-10 candidate dsRNAs, during the same time interval and/or at a higher incubation temperature, failed to produce meiotic one-cell arrested embryos; rather, they produced a reduced clutch of viable and inviable embryos (Table 1). However, after longer time intervals, they exhibited severe germline maintenance defects and became sterile. Importantly, earlier analysis of sterility in mat mutants indicated that they were associated with defects in the mitotic divisions of the germline nuclei (Goldenet al. 2000). In the current studies, mothers injected with dsRNA corresponding to other APC/C genes exhibited similar sterility defects after producing an initial small clutch of one-cell arrested embryos (data not shown). Although these results raise the possibility that apc-5 and apc-10 do not function as part of a meiotic APC/C, it is also possible that the rapid onset of RNAi-associated germline defects is masking subsequent defects in oocyte meiosis.
To distinguish whether these one-cell embryos were arresting in metaphase of meiosis I, meiosis II, or mitosis, the structures of the DNA and spindles were examined by 4′,6-diamidino-2-phenylindole (DAPI) staining and tubulin immunofluorescence. Under these conditions, the arrested embryos exhibited four critical features: (1) The oocyte chromosomes were aligned in a pentagonal array on a morphologically normal, barrel-shaped meiotic spindle; (2) meiotic polar bodies were absent; (3) the sperm chromatin mass remained highly condensed (Figure 3c); and (4) the embryos possess incompletely hardened, osmotically sensitive eggshells. These four features define an arrest in metaphase of meiosis I (Albertson and Thomson 1993; Goldenet al. 2000). Importantly, the same RNAi phenotype has been observed for not only apc-1, apc-2, cdc-23, and apc-11 (this study), but also cdc-16 (Goldenet al. 2000), apc-4 (Furutaet al. 2000), and cdc-27 (J. Schumacher, M. Abdolurasulnia, D. Shakes and A. Golden, unpublished results). Thus, with the possible exception of the apc-5 and apc-10 candidates, most of the APC/C subunits, which are essential for mitotic anaphase in other organisms, are also required to separate chromosomal homologs during meiosis I in C. elegans. Given that these predicted APC/C orthologs not only share a common RNAi phenotype but also are the only C. elegans genes with any significant similarity to their corresponding APC/C subunits in biochemically analyzed systems, they too are likely to function together in a complex.
APC/C is required for anaphase I functions beyond the separation of homologs: In mitotically dividing yeast, the loss of cohesion between sister chromatids is sufficient to trigger the metaphase-to-anaphase transition. In fact, some of the first cohesin mutants were isolated as suppressors of temperature-sensitive APC/C mutants; in the absence of sister chromatid cohesion, mitotically dividing double mutants bypass the APC/C-associated metaphase block and instead arrest with two distinct masses of DNA (Michaeliset al. 1997). Could the disruption of sister chromatid cohesion likewise bypass the meiosis I metaphase arrest of APC/C(RNAi) embryos or is the degradation of additional factors required for the segregation of paired, homologous chromosomes? Or, more specifically, do specialized features of either the meiotically telocentric chromosomes of C. elegans or its anastral, acentriolar spindle (Albertsonet al. 1997) require regulation by additional APC/C substrates?
To address these questions, we used rec-8 RNAi to disrupt sister chromatid cohesion. Rec8 is a meiosis-specific cohesion protein, first discovered in fission yeast, whose absence leads to aberrant, equational division of chromosomes during meiosis I (Molnaret al. 1995; Krawchuket al. 1999). Notably, in fission yeast, rec8 mutants complete both meiotic divisions and form four, albeit abnormal, spores. In C. elegans, the F1 progeny of mothers injected with rec-8 dsRNA exhibit striking defects in their diakinetic oocytes (Pasierbeket al. 2001). While diakinetic oocytes from wild-type mothers contain six DAPI-staining structures corresponding to the six paired homologs (Figure 4a), both sister and homolog cohesion is disrupted in rec-8(RNAi) oocytes (Figure 4b). In our hands, two-thirds of F1 rec-8(RNAi) hermaphrodites contained oocytes with 6 frayed DAPI-staining entities. The remaining one-third contained oocytes with 12 or more DAPI-staining entities, suggesting that the disruption of homolog cohesion was complete but that, as previously reported (Pasierbeket al. 2001), sister chromatids remained loosely associated (Figure 4b). Following the fertilization of these more severely affected oocytes, further detachment of the sister chromatids occurred as the chromosomes attempted to align in an aberrant metaphase I plate. Such plates lacked the normal pentagonal organization characteristic of wild-type embryos (compare Figure 4d to 4e), but, during anaphase, the affected chromosomes nonetheless segregated into two apparently equal-sized masses (Figure 4h). Importantly, fertilized rec-8(RNAi) oocytes not only completed meiosis, albeit abnormally, but formed multicellular, morphogenetic embryos. In fact, quantitative analysis of all embryos that were laid on growth plates revealed that 26% (n = 1415) hatched into larvae.
To analyze the double mutant phenotype for evidence of suppression, F1 rec-8(RNAi) mothers were soaked in apc-11 dsRNA. Such animals produced oocytes with the expected rec-8(RNAi) defects (Figure 4c), and, following fertilization, the disjoined chromosomes aligned in an aberrant metaphase plate (Figure 4f). However, unlike rec-8(RNAi) embryos, rec-8(RNAi); apc-11(RNAi) embryos failed to segregate their chromosomes into two distinct chromatin masses (Figure 4i) that would have indicated progression to or through anaphase and suppression of the apc-11(RNAi) metaphase I arrest. Instead, these embryos (n = 105) remained arrested with their chromosomes locked in a meiosis I metaphase-like state (Figure 4i), and, except for having disjoined sister chromatids, they were otherwise indistinguishable from the apc-11(RNAi) controls. Importantly, some rec-8(RNAi); apc-11(RNAi) embryos exhibited complete detachment of sister chromatids into >24 DAPI-staining chromatids and chromatid fragments during their prolonged metaphase arrest (data not shown). Notably, these one-cell arrested embryos also lacked several key indicators of meiotic progression including the decondensation of the sperm chromatin mass, the formation of polar bodies, and the formation of an impermeable eggshell. In parallel quantitative studies, only 0.5% of embryos (n = 1123) laid on growth plates from pooled rec-8(RNAi); apc-11(RNAi) mothers ever hatched into larvae. Similar qualitative and quantitative results were obtained in our analysis of mat-3(or180ts); rec-8(RNAi) embryos grown at the restrictive temperature (data not shown).
Understanding the composition and function of the APC/C during meiosis: Using forward and reverse genetics in C. elegans, we have demonstrated that several APC/C subunits, known to play essential mitotic roles in other organisms, are also required for the metaphase-to-anaphase transition during oocyte meiosis I. Because of the mitotic requirements for APC/C, this analysis was carried out using a combination of conditional alleles and RNAi. Null mutants in all of these genes are expected to develop into sterile adult hermaphrodites with highly reduced germlines as well as protruding or everted vulvae, as is the case for null alleles of emb-30/apc-4 (Furutaet al. 2000) and mat-3/cdc-23 (D. Garbe and M. Sundaram, personal communication). In the absence of conditional alleles, RNAi proved to be the best method for examining the role of C. elegans maternal products in the completion of the oocyte meiotic divisions following fertilization and the only viable approach for studying the potential meiotic role of apc-2 and apc-11. Importantly, the meiotic defects displayed by the APC/C(RNAi) embryos proved to be no more severe than those of our non-null mat mutants, a result that indicates that, because of either dosage or tissue specificity, these temperature-sensitive mutants behave as nulls during oocyte meiosis I.
Recent genome-wide RNAi screens have expanded the number of genes required for meiotic progression. In addition to APC/C subunits, this class includes numerous subunits of the proteosome (Gonczyet al. 2000; Zipperlenet al. 2001), the protein complex that degrades proteins ubiquitinated by the APC/C. Taken together, these genetic and RNAi studies suggest that, in C. elegans, the APC/C drives the meiosis I segregation of homologs by degrading one or more maternal products. Importantly, the RNAi and genetic approaches used in this study can be used to effectively deplete functional APC/C levels during the earliest stages of meiosis. Thus, an alternative explanation for seemingly contradictory studies in Xenopus (Peteret al. 2001; Taiebet al. 2001) is that, by the time Xenopus oocytes reach late prophase, the meiosis I functions of their preassembled and localized APC/Cs are protected from disruption by APC/C inhibitors.
In rec-8(RNAi); apc-11(RNAi) studies, the loss of sister chromatid cohesion proved insufficient to bypass the metaphase I block of APC/C-depleted embryos. This result indicates that, in C. elegans, the metaphase-to-anaphase transition during meiosis I requires an early function for APC/C beyond the separation of sister chromatids. Although the nature of these additional APC/C targets awaits further investigation, intriguing possibilities include both cyclin A (Parry and O'Farrell 2001) and chromokinesin (Funabiki and Murray 2000) as well as potentially novel, meiotic-specific APC/C substrates.
We thank all members of the Shakes and Golden laboratories, especially D. Chase and J. Schumacher, for sharing unpublished information. Thanks also to Andrew Page for helpful suggestions during the early stages of this work; Yuji Kohara for providing cDNAs; and Mike Krause for providing lab space, reagents, and advice during the cloning of apc-2 and apc-11. We also thank Mike Krause and Tommy Brodigan for their help and advice in sequencing using the ABI 310 Analyzer. For comments on the manuscript, we thank Mike Krause and the anonymous reviewer who suggested the rec-8 experiments. This work was supported by grants to D.S. from the Jefferess Memorial Trust (J-387) and National Institutes of Health (R15 GM60359-01) and to A.G. from National Institutes of Health Intramural funds. L.W. received support from both the Beckman Foundation as an Undergraduate Beckman Scholar and Howard Hughes Medical Institute (HHMI) as part of the HHMI Educational Grant through the College of William and Mary. Note added in proof: As first noted by David Greenstein (personal communication), the mat-2 alleles fail to complement the previously identified mutant evl-22 (ar104), which is sterile and has an everted vulva (G. Seydoux, C. Savage and I. Greenwald, 1993, Isolation and characterization of mutations causing abnormal eversion of the vulva in Caenorhabditis elegans. Dev. Biol. 157: 423–436). Our sequencing of this allele revealed that the mutation in ar104 results in a Pheto-Leu change at amino acid 931 of APC-1.
Communicating editor: B. J. Meyer
- Received September 5, 2001.
- Accepted December 6, 2001.
- Copyright © 2002 by the Genetics Society of America