Abstract
We have investigated the role of Caenorhabditis elegans RAD-51 during meiotic prophase and embryogenesis, making use of the silencing effect of RNA interference (RNAi). rad-51 RNAi leads to severe defects in chromosome morphology in diakinesis oocytes. We have explored the effect of rad-51 RNAi in mutants lacking fundamental components of the recombination machinery. If double-strand breaks are prevented by spo-11 mutation, rad-51 RNAi does not affect chromosome appearance. This is consistent with a role for RAD-51 downstream of the initiation of recombination. In the absence of MRE-11, as in the absence of SPO-11, RAD-51 depletion has no effect on the chromosomes, which appear intact, thus indicating a role for MRE-11 in DSB induction. Intriguingly, rad-51 silencing in oocytes that lack MSH-5 leads to chromosome fragmentation, a novel trait that is distinct from that seen in msh-5 mutants and in rad-51 RNAi oocytes, suggesting new potential roles for the msh-5 gene. Silencing of the rad-51 gene also causes a reduction in fecundity, which is suppressed by mutation in the DNA damage checkpoint gene rad-5, but not in the cell death effector gene ced-3. Finally, RAD-51 depletion is also seen to affect the soma, resulting in hypersensitivity to ionizing radiation in late embryogenesis.
THE RAD51 gene, homologous to the Escherichia coli RecA gene, was originally isolated in Saccharomyces cerevisiae (Aboussekhraet al. 1992; Basileet al. 1992; Shinoharaet al. 1992). Mutations in this gene confer on yeast cells an enhanced sensitivity to genotoxic agents, a reduction in mitotic recombination, and an impaired meiosis. Homologs and paralogs of the RAD51 gene have been found in all eukaryotes investigated so far. Among these, DMC1 encodes for a protein highly similar in sequence and function to Rad51, but is specifically required only in meiotic recombination (Bishopet al. 1992).
Much of our understanding of the enzymology and genetic control of meiotic recombination comes from studies in S. cerevisiae (reviewed in Paques and Haber 1999). Meiotic recombination includes two coupled processes: the formation and processing of double-strand breaks (DSBs). The SPO11 gene product, which is responsible for initiating recombination, induces enzymatic DNA cleavage resulting in DSBs (Caoet al. 1990; Bergeratet al. 1997; Keeneyet al. 1997). Mre11 is required for the induction and processing of DSBs (Johzuka and Ogawa 1995; Ogawaet al. 1995; Usuiet al. 1998). Rad51 catalyzes the strand-invasion and strand-exchange reaction between homologous DNA molecules (Sung 1994). A number of genes are involved in the control of meiotic crossing over, including MSH4 and MSH5, which are required for reciprocal exchange between, but not within, homologs (Ross-Macdonald and Roeder 1994; Hollingsworthet al. 1995).
Orthologs of SPO11, MRE11, MSH4, MSH5, and RAD51 have been identified in Caenorhabditis elegans (Dernburget al. 1998; Rinaldoet al. 1998; Takanamiet al. 1998; Zalevskyet al. 1999; Kellyet al. 2000; Chin and Villeneuve 2001). Surprisingly, the meiosis-specific DMC1 gene present in fungi, plants, and mammals is absent in C. elegans as well as in Drosophila melanogaster. Mutants in spo-11, mre-11, msh-4, and msh-5 are all defective in chiasma formation. It has been shown that spo-11 is required for the induction of meiotic recombination and that artificially induced breaks bypass the requirement for spo-11 (Dernburget al. 1998). Conversely, γ-irradiation of mre-11 null mutants does not lead to induction of chiasmata: after treatment, lethality is increased and the chromosomes in nuclei of oocytes at the diakinesis stage appear abnormal. mre-11 in C. elegans is therefore required for repair of radiation-induced DSBs during meiotic prophase (Chin and Villeneuve 2001). A further role of MRE-11 in DSB induction is hypothesized on the basis of the intact appearance of univalents in unirradiated mre-11 mutant oocytes, but this role has not yet been clearly proven. Treatment of msh-5 hermaphrodites with γ-rays does not induce chiasma formation, and therefore radiation-induced breaks do not bypass the requirement for msh-5 in chiasma formation. In contrast to what is observed in mre-11 mutants, after the γ-irradiation of msh-5 mutants,the chromosomes at diakinesis appear morphologically intact and lethality is not significantly increased; therefore, DNA repair is not affected by the msh-5 mutation (Kellyet al. 2000).
Although the recombination machinery is highly conserved among eukaryotes, metazoa have developed a germ cell line distinct from the somatic cell line, requiring a different level of gene regulation compared to unicellular eukaryotes. In C. elegans physiological and damage-induced apoptosis has been described in the female germline of hermaphrodites (Gumiennyet al. 1999; Gartneret al. 2000). Unlike somatic cells, germ cells respond to genotoxic stress with programmed cell death and by inducing a transient cell cycle arrest, although these two responses are spatially separate. Mitotic germ cells respond with cell cycle arrest, whereas pachytene-stage meiotic germ cells undergo apoptosis. Interestingly, silencing of the rad-51 gene mediated by RNA interference (RNAi) also induces germ cell apoptosis in the same checkpoint-mediated fashion as ionizing radiation (IR). In both cases, cell death uses the same basic execution machinery (such as the ced-3, ced-4, and ced-9 genes) and requires other genes such as mrt-2 (Ahmed and Hodgkin 2000), rad-5, and him-7 that are likely to be involved in a meiotic DNA damage checkpoint. spo-11 mutation partially suppresses the increase in apoptosis caused by rad-51 RNAi, suggesting that oocyte precursor apoptosis may be triggered by the accumulation of recombination intermediates (Gartneret al. 2000).
RNA interference of the rad-51 gene in C. elegans leads to a number of visible phenotypes, such as (i) high levels of embryonic lethality (Takanamiet al. 1998), (ii) increase in the frequency of males (Gartneret al. 2000), (iii) reduced fertility (Takanamiet al. 1998), and (iv) hypersensitivity to γ-radiation in the germline (Takanamiet al. 2000) and in soma (this article).
In this article, the effects of rad-51 RNAi in several genetic backgrounds are analyzed in detail to dissect the pathways in which rad-51 is involved in C. elegans in the germline and to establish the role of this gene in the somatic cells of this nematode.
MATERIALS AND METHODS
Strains and maintenance: All the strains were maintained and cultured according to Sulston and Hodgkin (1988). Unless otherwise specified, all the experiments were performed at 18°.
The following C. elegans strains used in this work were kindly provided by the Caenorhabditis Genetics Center:
N2: wild-type strain (Bristol variety; Brenner 1974)
NL917: mut-7(pk204)III (Kettinget al. 1999)
AV106: spo-11(ok79)IV/nT1[unc-?(n754) let-?](IV;V) (Dernburget al. 1998)
MT1522: ced-3(n717)IV (Ellis and Horvitz 1986; Xueet al. 1996)
CB1392: nuc-1(e1392)X (Hevelone and Hartman 1988)
CB1256: him-3(e1256)IV (Hodgkinet al. 1979; Zetkaet al. 1999)
The strain NL936: mut-7(pk204)III, unc-32 was a gift from Ronald Plasterk. The strain SP506: rad-5(mn159)III (Hartman and Herman 1982) was a gift from Anton Gartner. The strain AV112: mre-11(ok179)V/nT1[unc-?(n754) let-?](IV;V) (Chin and Villeneuve 2001) and the strain msh-5(me23)IV/nT1 [unc-?(n754) let-?](IV;V) (Kellyet al. 2000) were gifts from Anne Villeneuve.
RNA interference: Young adults were injected with double-strand RNA (dsRNA) [corresponding to nucleotides (nt) 513–943 of the rad-51 cDNA, GenBank accession no. AF061201] at a concentration of 500 μg/ml in proximity of the gonad (Fireet al. 1998). Injected worms (P0) were individually cloned, left to lay eggs overnight and then transferred every 12 hr onto fresh seeded plates, and the progeny were screened until only unfertilized oocytes were laid. Eggs laid during the first night after injection were discarded because they were unlikely to be affected. F1 hermaphrodites were individually cloned, transferred every 24 hr onto fresh seeded plates, and the progeny were screened. To rule out any mechanical or chemical stress due to the RNA injection procedure as a possible cause of the phenotype observed in P0, the same sample of RNA was also injected into mut-7 worms, which are known to be RNA-interference resistant (Kettinget al. 1999). The brood size, the percentage of unhatched eggs, and the percentage of male progeny produced by the mut-7 P0 were comparable to those of untreated mut-7 hermaphrodites (Tables 1 and 2).
Injection of rad-5 hermaphrodites (which are temperature sensitive) was performed taking further precautions: worms were placed on a cool (4°) clean plate and injected one at a time. After performing the injection, a cold recovery buffer was immediately added.
For each strain used, between 10 and 40 hermaphrodites were injected and an equal number of matching age hermaphrodites were individually cloned to provide a control group and their progeny were screened as described.
In analyzing the brood size, the total number of (hatching and unhatching) eggs laid for each given strain was considered. The mean values and standard deviations of control and rad-51 RNAi worms are reported in Table 3.
Genetic cross: Eighteen hours after dsRNA injection, five mut-7 unc-32 P0 hermaphrodites were crossed with three or four wild-type (N2) males, the parents were passed onto fresh seeded plates every 12 hr, and the F1 and F2 progenies were screened and cloned as described above. Five uninjected mut-7 unc-32 P0 hermaphrodites were similarly crossed with wild-type males, and the F1 and F2 progenies were screened in the same way. We also injected mut-7 unc-32 P0 hermaphrodites with a control dsRNA corresponding to an open reading frame (ORF) in which RNAi leads to embryonic lethality due to zygotic effect before the ventral enclosure stage (Maedaet al. 2001; C. Rinaldo, unpublished results). While the self-fertilized F1 progeny of such hermaphrodites show no embryonic lethality, when the injected mut-7 unc-32 P0 hermaphrodites are crossed with the wild-type males, the heterozygous F1 progeny turn out to be 100% unviable.
Imaging of meiotic chromosomes at diakinesis: Adult hermaphrodites 48 hr after dsRNA injection (P0) and matching age syngenic control worms were microdissected in cold PBS (phosphate-buffered saline), 0.25 mm levamisole. Worms were sliced with 25-gauge needles near the head and tail, and the gonads were released. Samples were then fixed with cold acetone; incubated for 25 min in PBS, 1 μg/ml 4’,6-diamidino-2-phenylindole (DAPI); washed in PBS; mounted on a dried 2% agarose pad; and observed under a Zeiss Axiovert 10 fluorescence microscope. Between 50 and 100 oocytes were observed for each genetic combination.
Effects of rad-51 RNAi on embryo viability and brood size
γ-Ray sensitivity assay: Microinjected worms (P0) were individually cloned, left to lay eggs overnight, then transferred every 12 hr onto fresh seeded plates. In C. elegans embryogenesis takes >800 min; only early embryogenesis normally takes place in the uterus; embryos in well-fed hermaphrodites are normally laid during gastrulation, 2–3 hr after fertilization. rad-51 RNAi F1 embryos were collected every 12 hr and treated with 20 Gy of γ-rays from a Cs137 source (4.285 Gy/min), and the phenotypes were analyzed 48–60 hr after treatment under a Zeiss Axiovert 10 fluorescence microscope equipped with Nomarski optics. The matching age control embryos were collected every 12 hr, treated with 20 Gy or 120 Gy of γ-radiation, and analyzed as described above.
Statistical analysis: Brood-size variations between rad-51 RNAi and untreated populations were analyzed with the z-statistics.
RESULTS
Embryonic lethality mediated by RNAi of rad-51 is due to maternal effect: The embryonic lethality arising in the late offspring of the rad-51 dsRNA-injected hermaphrodites (P0; Table 1, Figure 1) and reinforced in the surviving F1 progeny (F2) is consistent with a maternal effect. However, a pronounced maternal effect might obscure a milder zygotic effect. To discriminate between maternal and zygotic effects, rad-51 RNAi was performed in the double mutant mut-7 unc-32 and an appropriate genetic cross was set up (Figure 2). The mut-7 is a recessive mutation that confers resistance to RNAi (Grishoket al. 2000); unc-32 is a high-penetrance behavioral marker linked to mut-7. Homozygous mut-7 unc-32 hermaphrodites injected with rad-51 dsRNA were crossed with wild-type males. Parental mut-7 unc-32 hermaphrodites are not affected by RNAi, but F1 heterozygotes derived from such a cross will, however, inherit interfering agents and be sensitive to rad-51 RNAi. Embryonic lethality is expected to arise only if expression of the rad-51 gene is required during F1 embryonic development. A low level of embryonic lethality (4%) is observed throughout the F1 offspring, comparable to that observed in the untreated control mut-7 unc-32 hermaphrodites (Figure 2).
In the following generation, F1 heterozygous hermaphrodites (in which RNAi is active) are allowed to self-fertilize and the progeny are screened. About one-quarter of the offspring (F2) will be homozygous for mut-7 and insensitive to any residual effect of dsRNA injection. Embryonic lethality in this subpopulation can reflect only a parental defect. A 93% embryonic lethality is observed in the F2. The ratio between Unc and non-Unc worms among the survivors is the same as in the control experiment, indicating that the fate of the zygotes does not depend on their genotype but only on the interaction between interference and genotype in the parent.
Kinetics of appearance of embryonic lethality in the rad-51 RNAi F1. Injected P0 wild-type hermaphrodites were passed onto fresh plates every 12 hr. At each step the laid eggs were counted and screened for dead embryos (unhatched eggs indicated by solid circles) after a further 12 hr. The numbers on the x-axis indicate the end of the 12-hr egg-laying interval. The embryos die at different developmental stages. The offspring of the wild-type hermaphrodites of corresponding ages was screened in the same way (unhatched eggs are indicated by open circles).
mut-7 unc-32 homozygous hermaphrodites were injected with rad-51 dsRNA and crossed with wild-type males. mut-7 unc-32 homozygous hermaphrodite parents are not affected by interference (RAD-51+). mut-7 unc-32/++ F1 heterozygotes derived from the cross are sensitive to rad-51 RNAi (RAD-51−). In the control cross mut-7 unc-32 homozygous hermaphrodites were not injected with dsRNA and therefore mut-7 unc-32/++ F1 heterozygotes express RAD-51 (RAD-51+). The level of F1 embryonic lethals in the experiment and in the control (4 and 4.5% respectively) is comparable. mut-7 unc-32/++ F1 heterozygous hermaphrodites carrying interfering agents (RAD-51−) were allowed to self-fertilize and the progeny were screened. We expect one-quarter of the offspring (F2) to be homozygous for mut-7 unc-32 (RAD-51+). We observe 93% embryonic lethality in the F2 population. One-quarter of the surviving F2 progeny show an Unc phenotype.
Embryonic lethality is a common feature among meiotic mutants (Dernburget al. 1998; Zalevskyet al. 1999; Zetkaet al. 1999; Kellyet al. 2000; Chin and Villeneuve 2001) and mutations in these genes also result in an increase in the percentage of male progeny. In C. elegans, a high incidence of X0 males (known as the Him phenotype) is indicative of a defect in X chromosome segregation (Hodgkinet al. 1979). rad-51 RNAi leads to a Him phenotype in the F2 (Gartneret al. 2000 and Table 2).
Effects of rad-51 RNAi in mutants lacking components of the meiotic recombination machinery: The transparent gonad of the C. elegans hermaphrodite presents a spatial succession of regions containing cells undergoing subsequent meiotic events. In the proximal part of the gonad, next to the spermatheca, oocyte nuclei are in late meiotic prophase (diakinesis). At this stage, homologous chromosomes are already desynapsed but remain attached by chiasmata, temporary physical links established as a result of reciprocal recombination events completed in earlier stages. Although six DAPI-stained bodies (corresponding to six bivalents) are detectable in oocytes of wild-type worms (Figure 3A, left), highly unshaped and poorly condensed chromosomes, often grouped in bunches, are observed in all the oocytes of rad-51 RNAi worms (Takanami et al. 1998, 2000; and Figure 3A, right).
To dissect the pathways in which rad-51 is involved in C. elegans meiosis, we explored the epistatic relationships between rad-51 and three well-characterized meiotic genes: spo-11, involved in DSB induction (Dernburget al. 1998); mre-11, involved in DSB processing (Chin and Villeneuve 2001); and msh-5, promoting crossovers (Kellyet al. 2000). Mutants in all three meiotic genes display diakinesis oocytes containing 12 individual DAPI-stained bodies corresponding to achiasmate univalent chromosomes (Dernburget al. 1998; Chin and Villeneuve 2001; Kellyet al. 2000; and Figure 3, B–D, left).
When rad-51 interference is performed in a spo-11 background, we observe 12 DAPI-stained bodies corresponding to properly condensed univalents (Figure 3B, right) such as those displayed by the spo-11 oocytes in the presence of a functional rad-51 (Dernburget al. 1998; Figure 3B, left). In the above experiment we do not observe any defective chromosome structures like those observed in RAD-51-depleted oocytes in a wild-type background (Figure 3A, right). This observation shows that, in C. elegans, rad-51 acts downstream of spo-11 during meiotic prophase and is consistent with the results on germline apoptosis in rad-51 RNAi/spo-11 background described by Gartneret al. 2000.
Effects of rad-51 RNAi on incidence of males
rad-51 RNAi, in a mre-11(ok179) background, leads to the appearance of DAPI-stained bodies corresponding to properly condensed univalents in diakinesis nuclei (Figure 3C, right). This observation provides evidence that mre-11, known to be required for the repair of DSBs during meiosis (Chin and Villeneuve 2001), is also necessary in C. elegans in the initial steps of recombination for DSB induction.
The MSH-5 protein is required for both normal and radiation-induced meiotic crossing over, but is not required for DNA repair during meiosis (Kellyet al. 2000). rad-51 RNAi in a msh-5(me23) background results in the appearance of chromatin that distinctly differs both from that seen in msh-5 mutants (where 12 intact-appearing univalents are observed) and from the poorly condensed appearance of the chromatin in rad-51 RNAi oocytes. While we did not detect any DNA condensation defects, we did detect, in addition to a partial aggregation of bodies, very small DAPI-stained spots, pointing to the occurrence of DNA fragmentation (Figure 3D, right). Chromosome fragmentation is not observed in rad-51 RNAi oocytes in a wild-type background.
Brood-size reduction by rad-51 RNAi requires rad-5: In an attempt to understand which pathways lead to the reduced fertility of the rad-51 RNAi hermaphrodites (Takanamiet al. 1998; Table 1), we compared broodsize variations induced by rad-51 RNAi in genetic backgrounds affecting meiosis, DNA damage checkpoints, and apoptosis (Table 3).
Gartner et al. (2000) have demonstrated that in a rad-51-interfered background there is a net increase in checkpoint-dependent cell death of oocyte precursors. In the ced-3 mutant germ cells, both physiological and DNA damage-induced apoptosis are inhibited. In the meiotic checkpoint mutant rad-5, physiological germ cell death occurs, but DNA damage-induced apoptosis is absent.
We observed that rad-5 is the only genetic background (besides mut-7) in which the rad-51 RNAi does not have a significant effect on brood size. Reduced fecundity is the only rad-51 RNAi-induced phenotype suppressed by the rad-5 mutation: rad-51 RNAi in the rad-5 mutant strain leads to abnormal chromosome appearance in diakinesis oocytes (Figure 3E, right) and embryonic lethality.
A significant, although modest, reduction in brood size is caused by rad-51 interference in a ced-3 background (ced-3 being a cell death effector gene; Xueet al. 1996), suggesting a very partial involvement, if any, of apoptosis in reducing fertility in rad-51-interfered worms. rad-51 RNAi has a modest effect on the brood size also in a spo-11 genetic background.
We also analyzed the effects of rad-51 RNAi on fertility in two control strains respectively mutated in the genes nuc-1 (involved in apoptosis well downstream from ced-3 when commitment to death is irreversible; Wuet al. 2000) and him-3 (acting in meiotic recombination downstream of rad-51; Zetkaet al. 1999; C. Rinaldo, unpublished results). In both cases rad-51 RNAi effectively reduces fertility.
rad-51-interfered worms are hypersensitive to IR in soma: We have investigated the effect of γ-radiation on somatic cells during late embryonic development of rad-51-interfered worms.
Wild-type worms subjected to 20 Gy of γ-rays as embryos and RAD-51-depleted worms that have not been subjected to γ-rays exhibit a body morphology that is indistinguishable from wild-type untreated worms. Most F1 rad-51 RNAi adults that have been irradiated as embryos display gross abnormalities in the gonads and in the vulva (88% of all the adults). These defects appear only sporadically in the wild-type population treated with 120 Gy of γ-rays (<1% of all the treated worms). The gonad defects (observed in 75% of the γ-treated RAD-51-depleted worm population) might in part be a consequence of defects in germline growth or survival, although anomalies in arm migration and gonad elongation are likely to be due to somatic defects. That somatic tissues depleted of RAD-51 are hypersensitive to γ-radiation is demonstrated by the vulva defects: 19% of a RAD-51-depleted population, which had been treated with γ-radiation, shows protruding vulva, bursting at the vulva (Figure 4B), and egg-laying defects (Figure 4C).
DAPI staining of oocyte nuclei in the proximal portion of the gonad corresponding to the diakinesis stage. Chromosome morphology in wild type (N2) (A), spo-11(ok79) (B), mre-11(ok179) (C), msh-5(me23) (D), and rad-5(mn159) (E) is shown on the left. Chromosome morphology in the corresponding strains 48 hr after injection with rad-51 dsRNA is shown on the right. Some univalents in B and C are out of focus as they are located on different focal planes (see asterisks). Arrows indicate DAPI-stained spots with a diameter smaller than the average univalent (chromosome fragments). Bar, 5 μm.
DISCUSSION
Embryonic lethality resulting from rad-51 RNAi is due to maternal effect: Embryonic lethality can be explained if most embryos are aneuploid, since proper chromosome disjunction has not taken place during meiosis I in the affected parent, and if they also carry chromosome aberrations. An increase in X chromosome nondisjunction is confirmed by the observed Him phenotype. However, the incidence of males observed in the progeny of RAD-51-depleted F1 (6.5%, see Table 2) is lower than that observed in RNAi or null mutations of other meiotic genes such as him-3 (20%, Zetkaet al. 1999) and spo-11 (50%, Dernburget al. 1998). One possible explanation is that, since males are hemizygous for the X-linked genes, aberrant X chromosomes may be subjected to strong negative selection. Although meiotic defects are sufficient to account for the embryonic lethality observed as a result of RAD-51 depletion, we do not rule out a likely effect in mitosis: it is conceivable that RNAi in the hermaphrodite parent might prevent transmission of the RAD-51 protein, which fertilized eggs may normally inherit and which might be required for DNA repair during the first cell divisions. However, it should be borne in mind that the occurrence of DSBs in C. elegans may be less deleterious during mitotic cell division in soma (where transmission of broken chromosomes could be allowed by the holocentric organization) than in meiosis (where chromosomes adopt a monocentric organization). Studies on null mutants of the rad-51 gene will be necessary to rule out any requirement for rad-51 during early embryogenesis.
The rad-51 gene is involved in DNA damage response in the soma: Eukaryotic RAD51 genes are involved in the cellular response to genotoxic agents and in particular in the DSB repair pathway in somatic cells. In S. cerevisiae, rad51 mutants are viable although hypersensitive to genotoxic agents (Game and Mortimer 1974; Shinoharaet al. 1992). In the mouse, null mutations in RAD51 are lethal during early embryogenesis and this arrest in development can be explained as the result of a failure in the repair of DNA damage (Lim and Hasty 1996; Tsuzuki 1996). We show that C. elegans rad-51 RNAi, followed by γ-ray treatment during late embryogenesis, causes several developmental defects mostly affecting those postembryonic cell lineages leading to body structures dispensable for survival, such as the vulva and the gonads. This effect is likely to result from defective repair of radiation-induced DSBs. We conclude that rad-51 is involved in resistance to IR during embryonic growth.
Brood size of untreated RAD-51(+) and corresponding RAD-51-depleted worms
Effect of γ-radiation on development of F1 rad-51 RNAi worms. (A) rad-51 RNAi adult hermaphrodite that has not been treated with γ-radiation (vulva in particular). (B) rad-51 RNAi adult hermaphrodite that has been treated with 20 Gy of γ-rays during embryogenesis (bursting at the vulva). (C) rad-51 RNAi adult hermaphrodite that has been treated with 20 Gy of γ-rays during embryogenesis (vulva showing defects in egg laying). Bar, 10 μm.
New roles of mre-11, rad-51, and msh-5 in C. elegans: Chromosomes at the diakinesis stage, when RAD-51 is depleted, appear poorly condensed and associated in bundles. However, in this context if DSBs are inhibited by the absence of SPO-11, the chromosomes appear intact and properly condensed. Therefore, we can conclude that the abnormal morphology of chromosomes in rad-51 RNAi oocytes is a consequence of a defect in repair of meiotic DSBs. This is consistent with RAD-51 playing a role downstream of DSB induction. Interestingly, in mre-11 mutant oocytes IR induces the appearance of large chromosome aggregates (Chin and Villeneuve 2001) partially reminiscent of the cytological phenotype of rad-51 RNAi untreated oocytes. It is conceivable that MRE-11 and RAD-51 act in contiguous steps of recombination (such as the 5′ to 3′ resection of DSBs and strand invasion), which are critical for both crossover and noncrossover, and that the chromosome aggregates may be the result of abortive attempts to repair DSBs using alternative pathways.
We have shown here that depletion of RAD-51 in an mre-11 background (exactly as in a spo-11 background) leads to the appearance of properly condensed univalents at diakinesis, indicating that DSBs are not formed in the mre-11 mutant. We therefore demonstrate that MRE-11 is required for initiation of meiotic recombination in C. elegans and thus the dual function of MRE-11, i.e., meiosis-specific induction of programmed DSBs and DNA repair, is conserved in evolution from fungi to metazoa.
Although DNA repair is not affected by the msh-5 mutation (Kellyet al. 2000), rad-51 RNAi in a msh-5 mutant (Figure 3D) leads to chromosome fragmentation in oocytes. The RAD-51 depletion is therefore the cause of such fragmentation. We suggest that when a functional RAD-51 protein is available in the msh-5 mutants, DSBs are resolved as noncrossovers, for which MSH-5 is not required, resulting in undamaged, properly condensed univalents. In the absence of RAD-51 in a msh-5 background, DSBs cannot be resolved either as crossovers or as noncrossovers and therefore chromosomes are fragmented.
The diffused appearance of chromosomes resulting from RAD-51 depletion is somehow dependent on a functional MSH-5; in fact it is not observed when rad-51 RNAi is performed in the msh-5 mutant. MSH-5 may be involved in the stabilization of crossover intermediates (but not of noncrossover intermediates) and, in the absence of RAD-51, MSH-5 may improperly recognize some alternative substrate and indirectly contribute to the abnormal chromosome morphology characteristic of rad-51 RNAi oocytes. Our findings may also suggest a physiological role for MSH-5 protein in regulating chromatin organization during meiotic recombination. For instance, it might locally release a constraint promoting an “open” state of the chromatin that would normally regress once the exchange has been completed.
rad-51 cross-talk with the meiotic checkpoint gene rad-5 and fertility determinants: In self-fertilizing C. elegans hermaphrodites, spermatogenesis is completed after the L4/adult molt. Subsequent differentiation of germ cells gives rise exclusively to oocytes, which are produced in great excess. The number of spermatocytes normally acts as the limiting factor in determining the brood size. Oocyte precursor apoptosis could account for the reduction in fertility induced by rad-51 RNAi if it resulted in the spermatocytes outnumbering the residual oocytes. Alternatively, apoptosis would have to affect hermaphrodite spermatogenesis as well as oogenesis. rad-51 RNAi affects the brood size of the P0 as well as that of the F1 (Table 1). In spite of the dramatic increase in oocyte precursor cell death described by Gartner et al. (2000), we observed that, at the end of fertile life, the gonads of the P0 rad-51 RNAi hermaphrodites still contain a large number of developing oocytes, the spermathecae are devoid of sperms, and unfertilized oocytes are laid (data not shown). Therefore, although spermatogenesis in these worms should not be affected by RNAi because the dsRNA is injected after spermato-genesis has been completed, oocytes still outnumber the spermatocytes. Furthermore, rad-51 RNAi reduces fertility in a ced-3 background. Therefore apoptosis does not seem to be the cause of the reduced fecundity observed. However, since rad-5 brood size is not affected by rad-51 RNAi, we propose that, in response to rad-51 RNAi, RAD-5 is able to activate not only apoptosis but also alternative pathways contributing to brood-size contraction. Surprisingly, we observed that rad-51 interference has an effect on brood size in the spo-11 genetic background also. This observation leads us to envisage rad-51 performing other functions during gametogenesis: for example, besides acting downstream of spo-11 during homologous recombination, it may also play a role in the premeiotic S phase.
Acknowledgments
We thank Ronald Plasterk for the NL936 strain, Anne Villeneuve for the AV112 strain and the msh-5 null mutant strain, and Anton Gartner for the SP506 strain. All the other strains used in this work were provided by the Caenorhabditis Genetics Center, which is funded by the National Institutes of Health, National Center for Research Resources. We thank Salvatore Arbucci for his technical assistance, John Pulitzer for his critical appraisal of the manuscript, and Nancy Kleckner and Aurora Storlazzi for their helpful discussions throughout the course of this work. Cinzia Rinaldo is recipient of a predoctoral fellowship cofinanced by the “Fondo Sociale Europeo” (UE) and by the Consiglio Nazionale delle Ricerche; this work is in partial fulfillment of the requirements for her doctoral degree in Genetics at the University of Naples “Federico II.” This work was supported in part by Ministero dell'Universita' e della Ricerca Scientifica e Tecnologica, lex 488, cluster 2 “Biotecnologia applicata all'uomo” (project no. 1, “Genetica Funzionale Umana comparata e manipolazioni geniche in sistemi modello”).
Footnotes
-
Communicating editor: A. Nicolas
- Received April 26, 2001.
- Accepted November 26, 2001.
- Copyright © 2002 by the Genetics Society of America
