In the mitotic cell cycle of the yeast cerevisiae, the sister chromatid is preferred over the homologous chromosome (non‐sister chromatid) as a substrate for DNA double‐strand break repair. However, no genes have yet been shown to be preferentially involved in sister chromatid‐mediated repair. We developed a novel method to identify genes that are required for repair by the sister chromatid, using a haploid strain that can embark on meiosis. We show that the recombinational repair gene RAD54 is required primarily for sister chromatid‐based repair, whereas TID1, a yeast RAD54 homologue, and the meiotic gene DMC1, are dispensable for this type of repair. Our observations suggest that the sister chromatid repair pathway, which involves RAD54, and the homologous chromosome repair pathway, which involves DMC1, can substitute for one another under some circumstances. Deletion of RAD54 in S.cerevisiae results in a phenotype similar to that found in mammalian cells, namely impaired DNA repair and reduced recombination during mitotic growth, with no apparent effect on meiosis. The principal role of RAD54 in sister chromatid‐based repair may also be shared by mammalian and yeast cells.
DNA repair is an essential component of genome maintenance in living cells. DNA lesions such as double‐strand breaks (DSBs) are repaired by homologous recombination (Petes et al., 1991; Liang et al., 1998). Template sequences for such repair are found either on the sister chromatid or on the homologous chromosome (non‐sister chromatid). DSB repair strongly depends on recombinational repair genes (RAD50, RAD51, RAD52, RAD54 and others) that are conserved from yeast to mammals (Petrini et al., 1997). In the yeast Saccharomyces cerevisiae, DSBs are naturally induced during meiosis, and repaired primarily via the homologous chromosome (Schwacha and Kleckner, 1997). In the mitotic cell cycle, the DSB repair machinery prefers the sister chromatid as a substrate for repair (Kadyk and Hartwell, 1992). Is the choice between sister chromatid and homologous chromosome in the two alternative processes genetically regulated?
Most recombinational repair genes that are required for DSB repair in mitotic cells are also required for DSB repair during meiosis (Game, 1993) and therefore are likely to play a role in homologous chromosome‐based repair as well as in sister chromatid‐based repair. However, a few genes are considered to function preferentially in homologous chromosome‐based repair. One of these genes was recently identified and named TID1 (or RDH54) (Dresser et al., 1997; Klein, 1997). TID1 functions in a diploid‐specific manner, promoting recombination between homologous chromosomes. Although TID1 is expressed in the mitotic cell cycle (Klein, 1997), it seems to play only a minor role in DNA repair during vegetative growth (Klein, 1997; Shinohara et al., 1997). Another gene that may be specifically involved in repair based on the homologous chromosome is DMC1. DMC1 was shown to be expressed only in meiosis (Bishop et al., 1992), and therefore may be (although not proved to be) involved strictly in recombination between homologous chromosomes. Dmc1p is structurally similar to bacterial RecA proteins. Mutant dmc1 cells arrest in late meiosis I prophase with DSBs accumulated to abnormally high levels (Bishop et al., 1992). Dmc1p interacts with Tid1p in the two‐hybrid assay (Dresser et al., 1997), suggesting that Dmc1p and Tid1p may work together as part of a complex.
DSB repair during the mitotic cell cycle was shown to depend largely on the sister chromatid as a template (Kadyk and Hartwell, 1992). However, no genes that are preferentially involved in sister chromatid‐based repair have yet been described. Accumulated indirect evidence led us to consider RAD54 as a candidate gene: mutant rad54 cells cannot repair DSBs induced by ionizing radiation in the mitotic cell cycle (Budd and Mortimer, 1982), but appear to handle meiotic DSBs fairly well (Shinohara et al., 1997; this study). Hence, RAD54 is specifically required for repair of mitotic DSBs. Our working hypothesis tested whether RAD54 is involved primarily in sister chromatid‐based repair. RAD54 and TID1 share significant sequence homology with one another (Dresser et al., 1997; Klein, 1997). Like Tid1p, Rad54p also interacts with a RecA homologue, in this case Rad51p (Jiang et al., 1996; Clever et al., 1997). Rad54p strongly stimulates Rad51p activity in homologous DNA pairing (Petukhova et al., 1998).
To distinguish between sister chromatid and homologous chromosome‐based repair pathways, we took advantage of a unique feature of S.cerevisiae, namely the return‐to‐growth (RTG) procedure. Yeast cells that are induced to enter meiosis initiate the recombination process with a high level of DNA double‐strand breakage. When switched to mitotic growth conditions (RTG), the cells complete interhomologue recombination with high meiotic frequencies, but undergo mitotic cell division and remain diploid (Ganesan et al., 1958; Sherman and Roman, 1963). Recently, we have established that upon RTG there is a rapid disappearance of meiotic features (Zenvirth et al., 1997); the synaptonemal complex and its related structures are rapidly dismantled, and DSBs are rapidly and efficiently repaired. DSBs were shown to be repaired during RTG in the absence of a homologue in haploid cells that initiated meiosis. In diploids undergoing RTG, meiotic DSBs were mostly repaired by a homologue‐dependent pathway that involved recombination. DSBs were repaired in diploid dmc1Δ, but not rad51Δ mutants, as if a recombinational repair machinery, with characteristics similar to those found in mitotic cells, came into play during RTG (Zenvirth et al., 1997). From these findings we concluded that at least two pathways of DSB repair can operate during RTG. One pathway involves interhomologue recombination, whereas the other involves sister chromatid recombination. By using mutants that differentially block one of these repair pathways, we exploited the power of RTG experiments to differentiate between sister chromatid and homologous chromosome‐based DSB repair, in a situation where all DSBs are of the same type. RTG experiments offer an additional advantage: meiotic DSBs induced in RTG experiments are produced naturally in a controlled manner and are efficiently handled by the cell, whereas radiation and chemicals, which are commonly used to create DSBs in mitotic cells, cause other damage in addition to DNA damage.
To examine sister chromatid‐based DSB repair directly we used haploid cells. In haploids, the exclusive template available for DSB repair is the sister chromatid, whereas in diploid cells DNA sequences for repair may also be found on the chromatids of the homologous chromosome. We used haploids carrying the sir3 mutation, which enables the cells to initiate meiosis and thereby to generate a high level of DNA DSBs (Gilbertson and Stahl, 1994). Mutations in the SIR genes were shown not to alter the normal frequency, timing and position of meiotic DSBs, nor the kinetics of their repair (de Massy et al., 1994; Gilbertson and Stahl, 1994). If such meiotic cells are returned to mitotic growth conditions, they are found to be viable and produce colonies. Thus meiosis and RTG of sir3 haploids provide us with a new tool to test the involvement of various gene products in sister chromatid‐dependent DNA repair.
We show that RAD54 is required for sister chromatid‐based repair, whereas its homologue TID1 and the meiotic gene DMC1 are dispensable for this type of repair. Under certain circumstances RAD54 is involved to some extent in recombinational repair by non‐sister chromatids. Our data provide evidence that meiotic DSBs can be repaired by the DNA repair machinery that usually operates in mitotic cells. In diploid cells, DSBs may be repaired by either of the two pathways, the one depending on Rad54p and the other depending on Dmc1p and/or Tid1p.
A novel assay for detecting sister chromatid repair
To study DSB repair by the sister chromatid, a haploid strain was used. Haploids carry only one set of chromosomes, and therefore DSBs can be repaired only by using the sister chromatid as a substrate. In order to generate, in a natural manner, a high level of chromosomal DSBs, meiosis was induced. Meiosis in haploid was achieved using a sir3 mutation which relieves transcriptional silencing at the normally silent mating type cassettes (Rine and Herskowitz, 1987). The haploid sir3Δ strain (strain 3531) was induced into meiosis by transferring the culture to sporulation medium (SPM). In order to assess cell viability, samples withdrawn from SPM at various time points were returned to mitotic growth by plating aliquots on a vegetative growth medium (yeast extract/peptone/dextrose; YEPD). The number of colonies obtained from 0 h in SPM was considered as 100% cell viability, and at the following time points cell viability was calculated accordingly. To monitor the fate of DSBs upon RTG, samples were transferred after 4 h in SPM to liquid YEPD medium, and incubation was continued at 30°C. Aliquots were taken at 1 and 2 h after transfer to YEPD, and assayed for DSBs (as explained in Materials and methods). Full viability was maintained up to 8 h in the haploid sir3Δ strain (Figure 1A). During meiosis, DSBs were accumulated, and following a switch to mitotic growth conditions, the number of DSBs declined rapidly (Figure 1A and B). At 4 h in meiosis, 15% of the population of chromosome III molecules were broken. Considering the high level of chromosomal DSBs, the maintenance of full viability upon RTG indicates that the broken DNA was faithfully repaired rather than degraded. Since no homologous chromosome was present, the meiotic DSBs must have been repaired using the sister chromatid.
The haploid sir3dmc1 strain (3533) served to test this methodology further. The gene DMC1 is known to play a major role in recombination between homologous chromosomes (non‐sister chromatids) in meiosis (Bishop et al., 1992), and therefore the haploid mutant sir3Δdmc1Δ strain was not expected to be different from haploid sir3ΔDMC1+ in the sister chromatid repair assay. Indeed, these two strains behaved similarly, namely there was no reduction in cell viability for up to 8 h (Figure 1C), and DSBs were rapidly repaired upon switching to RTG (Figure 1C and D). At 4 h in meiosis ∼45% of chromosome III molecules were broken in strain 3533, and after 1 h in RTG conditions no DSBs were detected. DMC1 is therefore not required for sister chromatid repair during RTG.
RAD54 is essential for repair by the sister chromatid, whereas TID1 is not
Although RAD54 and TID1 are both recombinational repair genes that share significant sequence homology with one another, they differ in several characteristics. Mutant rad54Δ cells are extremely sensitive to the alkylating agent methyl methane sulfonate (MMS), while tid1Δ mutants do not show great sensitivity. In meiosis, the diploid rad54Δ mutant produces viable spores at a reasonable frequency. The tid1Δ mutant, on the other hand, shows poor sporulation and reduced spore viability (Klein, 1997; our data not shown). The differences in behaviour between these mutants suggested to us a basic difference in DNA repair specificity between the proteins produced by these two genes. Thus, Rad54p may be involved primarily in repair by the sister chromatid, whereas Tid1p is largely involved in DNA repair by the homologous chromosome.
To determine the effect of RAD54 and TID1 on sister chromatid‐based repair, we constructed a haploid sir3Δtid1Δ strain (3534) and a haploid sir3Δrad54Δ strain (3532), and assayed these mutants in our sister chromatid repair assay. During meiosis of the haploid sir3Δtid1Δ mutant, DSBs were accumulated to a high level (Figure 2B). Following a switch to mitotic growth conditions, most DSBs in cells of strain 5334 disappeared after 1 h. At 4 h in meiosis 43% of the population of chromosome III molecules were broken, and after 1 h in RTG only 9% remained broken. These residual DSBs disappeared by 2 h in RTG (Figure 2A). The DSBs disappearance was associated with high maintenance of cell viability (Figure 2A), indicating that the DSBs were faithfully repaired. These results indicate that TID1 has no essential role in sister chromatid‐mediated repair. In the haploid sir3Δrad54Δ strain, on the other hand, although DSBs were accumulating during meiosis to a lesser extent than in the haploid sir3Δtid1Δ, these breaks were maintained upon RTG, even 2 h after the switch (Figure 2D). The maintenance of DSBs was associated with a massive reduction in cell viability upon RTG (Figure 2C). Only 32% of the cells were viable after 6 h. These findings suggest that repair of DSBs by the sister chromatid is largely dependent on RAD54, and that this mode of repair does not require TID1.
RAD54 and DMC1 represent the two principal pathways for homologous repair of DSBs
A haploid strain deleted for the RAD54 gene was unable to repair DSBs upon RTG. We further examined whether diploid rad54Δ cells are capable of DSB repair under these conditions. A diploid strain homozygous for a deletion in RAD54 was constructed (3510). Samples were taken from this diploid rad54Δ sporulating culture after 0, 3.5 and 4 h and assayed for DSBs in meiosis. In addition, samples were taken from the same diploid rad54Δ sporulating culture after 3.5 and 4 h and returned‐to‐growth by transferring the cell cultures to liquid YEPD. To monitor DSBs upon RTG, samples were further taken from YEPD at 0.5 and 1 h after the switch. The amount of DSBs obtained at 3.5 h (Figure 3A) and at 4 h (data not shown) during the meiotic time course, was largely decreased after 1 h in RTG conditions. Upon 3.5 h in meiosis 18% of the population of chromosome III molecules were broken, and after 1 h in RTG only 2% of these DSBs remained. These residual DSBs disappeared completely by 2 h in RTG. The disappearance of DSBs was accompanied by a high (although not full) maintenance of cell viability (Figure 3B, compare rad54Δ with wild‐type). RAD54 is therefore not essential for DSB repair upon RTG in a diploid strain. This result raises the following possibility. In diploid rad54Δ cells, the sister chromatid‐dependent repair pathway is blocked due to a deletion of the RAD54 gene, but the non‐sister chromatid of the homologous chromosome can serve as a template for repair of DSBs upon RTG. If this is indeed the case, then diploid rad54Δ cells are not expected to be defective in non‐sister chromatid‐based repair. Therefore, recombination frequencies between the homologous chromosomes in the rad54Δ strain are expected to be similar to those in the wild‐type strain upon RTG. Recombination levels between homologous chromosomes were examined in a diploid rad54Δ strain (3510) and compared with a diploid wild‐type strain. Both strains were heteroallelic for his4 mutations at the his4::LEU2 locus on chromosome III (Cao et al., 1990). Thus, only recombination between the two homologous chromosomes III could yield His+ cells. Samples were taken from meiotic cultures in SPM at various time points, up to 10 h, and aliquots were plated on YEPD, as well as on defined vegetative growth medium lacking histidine. Recombination frequencies were estimated from the frequencies of His+ colonies produced. In the wild‐type strain (2982), the percentage of His+ recombinants normally increased 200–500 fold (to 0.7–1%) during the course of the experiment (Figure 3C). In the rad54Δ mutant (3510), the level of His+ recombinants was high (Figure 3C; Shinohara et al., 1997), and in some experiments even higher than the wild‐type level. This indicates that RAD54 is not essential for recombination between homologous chromosomes in meiosis and in RTG.
In an earlier section we showed that the TID1 gene is not essential for sister chromatid‐based DNA repair. Klein (1997) demonstrated that TID1 is specifically required for gene conversion events that occur between homologous chromosomes. We therefore examined the ability of a diploid tid1Δ strain 3536 to repair meiotic DSBs upon RTG and to undergo recombination (Figure 3). In this strain the amount of DSBs obtained at 3.5 h (20%) was reduced after 1 h in RTG conditions (9%), and after 2 h in RTG no DSBs could be detected (Figure 3D). Cell viability was high, with a small reduction at 6–10 h (similar to the results obtained with the diploid rad54Δ strain), indicating that most of the DSBs were faithfully repaired upon RTG. In contrast to the diploid rad54Δ strain, the level of His+ recombinants obtained in the tid1Δ strain remained very low (0.16% at 10 h; see Figure 3F). Thus, efficient repair of meiotic DSBs in the diploid tid1Δ strain occurs mainly by a repair mechanism that is distinct from the homologous chromosome‐based repair pathway, probably by sister chromatid repair.
DMC1 was previously shown to affect recombination between homologous chromosomes (Bishop et al., 1992). In the dmc1Δ strain (NKY1879), this type of recombination is dramatically reduced upon RTG (Figure 4C, compare with the wild‐type on Figure 3C). In addition, we showed that RAD54 is required preferentially for sister chromatid‐mediated repair upon RTG. Do these two genes represent the two main pathways for DSB repair? A diploid strain mutated for both rad54Δ and dmc1Δ was constructed (3521). Cell viability and the fate of DSBs upon RTG were examined and compared with the diploid single mutants, deleted for either dmc1 or for rad54. During meiosis of the diploid double mutant rad54Δdmc1Δ, DSBs accumulated to an extremely high level (at 7 h in meiosis 73% of the chromosome III molecules were broken at least once). After the shift to RTG conditions most of the DSBs were not repaired, up to 2 h (Figure 4A). Cell viability in both rad54Δ and dmc1Δ single mutants (diploids 3510 and NKY 1879, respectively) was highly maintained upon RTG (Figures 3B and 4B, respectively). In the rad54Δdmc1Δ double mutant, however, cell viability was severely reduced (Figure 4B). Only 13% of the cells were viable after 10 h in meiosis. These results suggests that RAD54 and DMC1 represent the two principal pathways for DSB repair in S.cerevisiae.
The rad54Δdmc1Δ mutant was arrested in meiosis, and no spores were produced after 24 h in SPM. Meiotic arrest was previously shown for the diploid dmc1Δ single mutant (Bishop et al., 1992), whereas rad54Δ diploids generally complete meiosis. This indicates an epistatic effect of DMC1 over RAD54 in this respect.
RAD54 is also involved in some recombinational repair by the homologous chromosome (non‐sister chromatid)
Recombination between homologous chromosomes in the diploid dmc1Δ mutant is not completely eliminated but is greatly decreased (Figure 4C; Bishop et al., 1992). The residual recombination events were almost abolished in a rad54Δdmc1Δ double mutant, heteroallelic for his4 (Figure 4C). In addition, recombinant DNA molecules (which represent intermediate products in the process of recombination between homologous chromosomes) were almost absent in the rad54Δdmc1Δ strain in meiosis (data not shown). These observations suggest that, in the absence of DMC1, the RAD54 pathway may be responsible for some recombinational repair between homologous chromosomes in meiosis and possibly also in RTG.
We thus examined whether RAD54 is involved in recombination between homologous chromosomes in the mitotic cell cycle. We used the same heteroallelic his4 strains to examine spontaneous recombination frequencies between homologous chromosomes in mitotically dividing cells (Table I). The level of His+ recombinants was 9‐fold lower in the rad54Δ strain (3510) compared with the wild‐type strain (2982). In similar experiments by Shinohara et al. (1997), a 12‐fold reduction was found. Much to our surprise, a further 28‐fold reduction in the level of His+ recombinants was observed in the rad54Δdmc1Δ strain (3521), compared with the rad54Δ single mutant. DMC1 was shown to be a meiosis‐specific gene and therefore its absence was not expected to further reduce the recombination level in mitotically dividing cells. As we have observed that SK1 strains of S.cerevisiae enter meiosis at high cell density, some of the cells in the above experiment could have embarked on meiotic recombination in YEPD. We suspected that the DMC1‐dependent recombination that was observed in the mitotic culture of strain 3510 (rad54Δ) in fact reflected a small fraction of cells that were induced to enter meiosis. To ascertain this possibility and to ensure that no meiotic recombination events would mask the spontaneous mitotic recombination, a spo11Δ mutation was introduced into the diploid strains. SPO11 is a meiosis‐specific gene which is required for meiotic DSB formation (Cao et al., 1990). SPO11 transcripts are absent from vegetatively growing cells and the gene is transcribed specifically in early meiotic prophase (Giroux et al., 1989). In spo11 mutants, spontaneous mitotic recombination occurs at a normal level and no meiotic recombination is observed (Klapholz et al., 1985). We used the heteroallelic his4 site to examine spontaneous mitotic recombination frequencies between homologous chromosomes in a diploid spo11Δ strain (3546), a diploid spo11Δrad54Δ strain (3543) and a diploid spo11Δrad54Δdmc1Δ strain (3551). The level of His+ recombinants in the spo11Δ‐deleted strain was 5‐fold lower than in the wild‐type strain (Table I), suggesting that the occurrence of SPO11‐dependent recombination events (most probably meiotic events) is not negligible. In the spo11Δrad54Δ strain the level of His+ recombinants was 26‐fold lower than the level observed in the spo11Δ single mutant. Thus RAD54 is responsible for >95% of the mitotic recombination events. However, in the spo11Δrad54Δdmc1Δ strain the recombination level was similar to the level observed in the spo11rad54 strain (the two values were found not to be significantly different by t‐test of the log‐transformed values). This indicates that DMC1 is not involved in recombination between homologous chromosomes in the mitotic cell cycle. The nature of the remaining RAD54‐independent recombination events (occurring in ∼1×10−7 of the cells) remains unexplained.
In the present study we have developed a new method based on RTG of meiotic cells containing DSBs, which enables us to identify genes that are required for sister chromatid‐based repair. Our results show that the gene RAD54 is required preferentially for DNA repair by the sister chromatid, whereas the genes TID1 and DMC1 are dispensable for this type of repair. The principal evidence leading to this conclusion is drawn from the behaviour of haploid rad54 cells, which cannot repair DSBs upon RTG, because sister chromatids are the only templates available for repair. As a consequence, cell viability is severely reduced in such mutants. In contrast, haploid tid1 and haploid dmc1 cells showed efficient DSB repair and full viability upon RTG. We further show that DMC1 and RAD54 represent the two principal pathways for homologous repair of DSBs. These two pathways are able to substitute for one another in some circumstances.
RAD54 is required preferentially for sister chromatid‐mediated repair
We showed that a haploid strain deleted for RAD54 was unable to repair DSBs upon RTG, whereas diploid rad54Δ cells were able to repair DSBs efficiently under the same conditions. The simplest interpretation would be that RAD54 is specifically required for repair by the sister chromatid, and in rad54Δ‐deleted cells this type of repair is blocked. In the haploid rad54Δ strain, the sister chromatid, which is the only available template for repair, cannot serve for DSB reconstruction due to the rad54Δ mutation. Thus, a massive cell death is observed. In diploid rad54Δ cells, although the sister chromatid‐dependent pathway cannot operate, the homologous chromosome‐dependent pathway is functioning. As a result, DSBs are repaired and high viability is maintained. Indeed, in diploid rad54Δ cells, repair of DSBs occurs with high levels of recombination between homologous chromosomes and meiosis is almost normal. This indicates that RAD54 is not required for DSB repair by the non‐sister chromatids of the homologous chromosomes in meiosis as well as in RTG.
In normal wild‐type meiosis, some events of sister chromatid‐based repair occur. The frequencies of such repair events are estimated to be between 30% (Game et al., 1989) and <10% (Schwacha and Kleckner, 1994) of the total DSB repair during meiosis. Interestingly, in diploid rad54Δ, cell viability was high, although not fully maintained. A 35% reduction in cell viability was observed after 10 h of meiosis (Figure 3B). As mentioned above, recombination between homologous chromosomes was not affected. A possible explanation of these results is that in diploid rad54Δ meiotic cells, a number of DSBs were committed to be repaired by the sister chromatid, and since this repair pathway was blocked, some reduction in cell viability was observed.
Other evidence supports the notion that RAD54 is specifically required for sister chromatid‐based repair. Upon RTG of the diploid dmc1Δ mutant, full viability was maintained (Figure 4B), as a consequence of efficient repair of meiotic DSBs (Zenvirth et al., 1997). The reduced level of His+ recombinant colonies produced by this mutant, 35% of the wild‐type level (Figure 4C), indicates that this repair is largely based on the sister chromatid. Indeed, in a two‐dimensional gel electrophoresis assay, only inter‐sister joint molecules were detected in the dmc1Δ mutant, upon RTG (Schwacha and Kleckner, 1997). We further suggest that RAD54 has a major role in sister chromatid‐based repair in the dmc1Δ mutant, since in our diploid double mutant rad54Δdmc1Δ, a dramatic reduction in cell viability was observed, compared with dmc1Δ, and DSBs were not repaired upon RTG. Our interpretation of these results is that in the double mutated strain rad54Δdmc1Δ, both principal pathways for DSBs repair are blocked (i.e. the homologous chromosome‐dependent pathway and the sister chromatid‐dependent pathway). Therefore, no recombinational repair of DSBs could take place.
Another gene that was proposed to have a preference for repair from the sister chromatid is RAD50 (Ivanov et al., 1992). However, the ubiquitous phenotypes of rad50Δ mutant in DNA metabolism and chromatin structure (for representative references see Moore and Haber, 1996; Kironmai and Muniyappa, 1997) suggest a much more complex role for this gene in DNA repair.
DMC1 and TID1 are preferentially involved in repair from the non‐sister chromatid and are not required in sister chromatid‐based repair
DMC1 is a meiosis‐specific gene that was shown to affect recombination between homologous chromosomes in meiosis (Bishop et al., 1992). Yet, in dmc1Δ mutants, this type of recombination is dramatically reduced also upon RTG (Figure 4C). It is not clear whether DMC1 is expressed in RTG conditions, or whether the residual Dmc1p, which was induced during meiosis, persists and functions in RTG. A Northern blot analysis should be performed in order to distinguish between these two possibilities. Alternatively, DMC1 is neither expressed nor remaining active in RTG, but its action during meiosis had committed the cell to a certain mode of recombinational repair, which is pronounced also in the following RTG conditions. In any case, the action of Dmc1p is conspicuous upon RTG.
During meiosis, the homologous chromosome (non‐sister chromatid) is preferred over the sister chromatid as a substrate for DSB repair, but some level of repair by the sister chromatid does occur. Is DMC1 involved in both types of recombinational repair during meiosis or is it specifically required in homologous chromosome‐based repair? We show that DMC1 is unable to repair DSBs using the sister chromatid, and therefore is preferentially involved in homologous chromosome‐based repair. In the haploid sir3Δdmc1Δ strain, DSBs are fully and efficiently repaired upon RTG, indicating that DMC1 is not essential for sister chromatid‐mediated repair. Moreover, in the haploid sir3Δrad54Δ strain, the intact DMC1 gene cannot serve to repair DSBs utilizing the sister chromatid, and as a consequence a severe reduction in cell viability is observed.
While this study was in progress, a new RAD54 homologue was identified in S.cerevisiae, named RDH54 or TID1 (Dresser et al., 1997; Klein, 1997). The TID1 gene was shown to be specifically required for recombination between homologous chromosomes. The observation that mitotic tid1Δ cells show no increased sensitivity to DNA damage at standard MMS concentrations has led Klein (1997) to suggest that TID1 is not normally involved in recombinational repair by the sister chromatid. We explored this speculation and showed directly that TID1 is indeed not required for sister chromatid‐based repair. Haploid sir3Δtid1Δ cells are able to fully repair DSBs upon RTG, and this efficient repair by the sister chromatid is translated into maintenance of high cell viability.
In the diploid tid1Δ strain, although recombinational repair utilizing the homologous chromosome occurred at a very low level, high cell viability was maintained during meiosis and RTG. This might be a result of the functional RAD54 gene product that replaces the missing Tid1p during the repair of meiotic DSBs, using the sister chromatid as a substrate.
Alternative pathways for homologous repair of DNA DSBs
In the yeast S.cerevisiae, meiosis and the mitotic cell cycle are alternative developmental pathways. RTG conditions reflect a unique situation in which both meiotic and mitotic DSB repair pathways are largely functional (Zenvirth et al., 1997). Thus, RTG is a convenient tool to monitor the mutual association between these repair pathways. Upon RTG in a diploid cell, two main pathways can promote the repair of meiotically induced DSBs: the homologous chromosome repair pathway, which depends mainly on the DMC1 gene; and the sister chromatid repair pathway, which requires the RAD54 gene. These two pathways are able to substitute for one another, at least in part, since DSBs are efficiently repaired upon RTG in the absence of either DMC1 (Zenvirth et al., 1997) or RAD54 (Figure 3A), and as can be deduced from the high viability of cells lacking either genes upon RTG (Figures 3B and 4B). These two pathways might be able to substitute for each other also during meiosis. This conclusion is drawn from the following findings. First, DSBs that are induced in rad54Δ cells during meiosis (Figures 2D and 3A) disappear after 8–10 h (data not shown, and Shinohara et al., 1997), indicating that the main DSB repair machinery in meiosis, which is based on the homologous chromosome as a substrate for repair, does not require RAD54. Hence, in meiosis, the DMC1 pathway can handle most of the DSBs in the absence of RAD54. Secondly, the mutation dmc1Δ was shown to cause a meiotic cell arrest at the pachytene stage, before DSBs are repaired (Bishop et al., 1992; Rockmill et al., 1995). In search of high copy suppressors of the meiotic arrest and spore inviability phenotypes of the dmc1Δ mutation, RAD54 was isolated (D.K.Bishop, personal communication), suggesting that the RAD54 pathway may be able to repair DSBs also in meiosis. A third observation supporting this notion comes from the effect of a red1Δ mutation on the dmc1Δ arrest (Xu et al., 1997). A red1Δdmc1Δ double mutant does not arrest in meiosis, and DSBs are repaired primarily by sister chromatid recombination (Schwacha and Kleckner, 1997) (the RED1 gene appears to channel recombination repair in meiosis towards the non‐sister chromatid of the homologous chromosome, and in the red1Δ mutant this bypass is abolished). However, in the triple mutant red1Δdmc1Δrad54Δ, DSBs remain unrepaired and there is no meiotic arrest (D.K.Bishop, personal communication); i.e. repair from the sister chromatid cannot take place.
The emerging picture is that two principal homologous recombinational DSB repair pathways exist in S.cerevisiae (Figure 5). One recombinational repair pathway employs the sister chromatid as a template for repair. This pathway, which involves RAD54 and most probably RAD51, is the main mechanism for DSB repair during the mitotic cell cycle. (The extreme sensitivity to ionizing radiation seen in rad54Δ and in rad51Δ cells is due to their important role in sister chromatid‐mediated DNA repair.) The other pathway employs the non‐sister chromatid of the homologous chromosome, and involves the Dmc1p–Tid1p complex (and also Rad51p). This is the main pathway for DSB repair in meiosis. The two pathways can substitute for one another in some circumstances (as discussed above). Yet, minor repair pathways do exist both in mitosis and in meiosis. In the mitotic cell cycle, minor pathways employing the homologous chromosome are DMC1‐independent, as our spo11Δrad54Δdmc1Δ diploid showed the same level of recombination between homologous chromosomes as the spo11Δrad54Δ diploid (Table I). In addition, it is known that DMC1 is not expressed in mitotic cells. One of these DMC1‐independent pathways is RAD54‐dependent, as we found that in the rad54Δ diploid strain the level of recombination between the homologous chromosomes was lower than in the wild‐type strain (Table I). The other pathway is RAD54‐independent and TID1‐dependent (Klein, 1997). In meiosis, a minor DSB repair pathway based on the sister chromatid also exists. As was shown by the limited loss of viability of meiotic rad54Δ cells, RAD54 may be involved in this DSB repair pathway. RAD51 has been shown to be required for repair of DNA damage in the mitotic cell cycle as well as in meiosis (Shinohara et al., 1992), and therefore may act as a key gene in both pathways.
Testing sister chromatid‐based repair
Detection of recombinational repair events that occur between non‐sister chromatids is relatively easy because they differ from each other at various sites. A new linkage relationship between two or more heterozygous markers is usually analyzed. In contrast, recombinational repair between sister chromatids is difficult to detect because exchange between two identical DNA molecules does not alter linkage relationships. Our new way to evaluate sister chromatid‐based DSB repair uses haploids, because in haploids the sister chromatid is the only template available for homologous DSB repair. High level of DSBs is generated upon induction of meiosis. To induce meiosis and meiotic DSBs in haploid strains, we have employed the sir3Δ mutation, in addition to mutations in DNA repair genes. DSBs in our method are naturally induced, and are not associated with other damage to DNA (e.g. single‐strand nicks that are commonly produced by ionizing radiation) or to other cellular components. The fate of these DSBs upon RTG is assayed physically, simultaneously with cell viability. If the amount of DSBs is reduced and cell viability is maintained, the conclusion is that efficient DSB repair has occurred. If the high level of DSBs persists upon RTG and cell viability is reduced, the conclusion is that DSB repair could not have taken place. This method of testing repair of DSBs in haploids enabled us to identify genes that are preferentially involved in sister chromatid‐based repair.
Several types of approaches have been previously used to detect sister chromatids recombination (Petes and Pukkila, 1995). One is the genetic detection of unequal sister chromatid exchanges between repeated genes on the chromosome (Jackson and Fink, 1985). Unequal sister chromatid exchange reflects an ectopic type of recombination and therefore might represent atypical features of recombination. The second approach involves the detection of topological changes resulting from recombination between sister chromatids of circular DNA molecules (Haber et al., 1984; Game et al., 1989). The third method is a physical analysis that detects joint molecules generated during sister chromatid recombination by two‐dimension gel electrophoresis (Collins and Newlon, 1994; Schwacha and Kleckner, 1994). Our method of using RTG cultures of sir3Δ haploids with meiotic DSBs is an important addition to the ways in which sister chromatid repair can be assayed.
The principal role of RAD54 in sister chromatid‐mediated DSB repair may be shared by yeast and by vertebrates
Recently, it has been shown that repair of DSBs in mammalian cells is highly dependent on homologous sequences (Liang et al., 1998). RAD51, RAD54 and other recombinational repair genes have been shown to be conserved through evolution from yeast to mammals (Petrini et al., 1997). Moreover, the human homologue of RAD54 (hHR54) can substitute functionally for the S.cerevisiae gene (Kanaar et al., 1996). Disruption of RAD54 in S.cerevisiae and in mouse ES cells results in a qualitatively similar phenotype, namely cells are sensitive to ionizing radiation and to MMS (Essers et al., 1997). RAD54 was shown to have only a minor role in meiosis of S.cerevisiae (Figure 3, above, and Game, 1993). Mice with a disruption of the RAD54 gene are fertile, indicating no essential function for RAD54 in mammalian meiosis as well (Essers et al., 1997). Homologous integration is greatly reduced in mouse, chicken and fission yeast (Schizosaccharomyces pombe) rad54 cells (Bezzubova et al., 1997; Essers et al., 1997; Muris et al., 1997) as well as in S.cerevisiae (see our difficulties in the construction of the haploid sir3Δrad54Δ strains, in Materials and methods). Recent findings imply that RAD54 is required for sister chromatid‐mediated repair not only in S.cerevisiae but also in higher eukaryotes (Takata et al., 1998): chicken DT40 cells deficient in the RAD54 gene were extremely sensitive to γ‐ray irradiation in G2, as well as in the G1 phase of the cell cycle. In contrast, wild‐type DT40 cells showed increased radiation resistance in the G2 phase relative to the G1 phase. At the G1 stage of the cell cycle, DNA replication has not yet occurred. Therefore, homologous DSB repair at this stage can only use the non‐sister chromatid as a template. Since the non‐sister chromatid is not normally involved in DSB repair in the mitotic cell cycle, wild‐type cells at G1 phase show great sensitivity to DNA damage induced by γ‐rays. At the G2 stage of the cell cycle, the sister chromatids of each chromosome are present, and serve as the main template for DSB repair. Thus, wild‐type cells show increased radiation resistance in G2 phase. In the chicken rad54 cell line, the sister chromatid‐dependent pathway is blocked, and therefore rad54 G2 cells show great sensitivity to γ‐ray irradiation, like rad54 G1 cells. These results reinforce our conclusion from the yeast haploid mutants, that RAD54 is required for DSB repair that is mediated preferentially by sister chromatids.
We therefore propose that the role of RAD54 in sister chromatid‐mediated DSB repair in yeast described in this work implies a parallel function for its mammalian and chicken homologues. Thus, vertebrate RAD54 may have a principal role in repair of DSBs in somatic cells through recombination with the sister chromatid.
Materials and methods
Yeast strains and media
The yeast strains used in this study are listed in Table II. All strains were constructed in SK1 genetic background. Yeast strains were maintained according to standard techniques and media were YEPD, YEPA and SPM, as described previously (Kassir and Simchen, 1991). Gene disruption in yeast was obtained by one step replacement using the electroporation technique (Becker and Guarente, 1991). All strain manipulations were verified by Southern blot analysis. We experienced great difficulties in introducing the sir3Δ mutation into rad54Δ cells. We had therefore to create a sir3Δ‐disrupted strain first (3531), and then to introduce the rad54 deletion, in order to create a haploid sir3Δrad54Δ double mutant (strain 3532). Thus haploid 3532 is isogenic to strain 3531, except for the newly introduced rad54Δ::URA3 deletion.
Sporulation and RTG
Sporulation was performed by a three‐step procedure. Yeast were grown overnight in YEPD, resuspended and diluted in YEPA for further growth overnight, then washed and transferred to SPM. Haploid cultures were also sonicated for 10 min in a water bath prior to their transfer to SPM, in order to separate aggregated cells. At intervals, samples were spun down and resuspended in liquid YEPD medium. These were the RTG sub‐cultures.
Cell viability and recombination frequencies
For assessment of cell viability and the frequency of His+ recombinants during meiosis, 0.1 ml samples were removed from the SPM culture at various times, diluted and plated on YEPD plates and on plates with defined medium lacking histidine. The frequency of His+ recombinant colonies was determined by dividing the number of colonies growing on the histidine‐deficient plates by the number of colonies on the complete‐medium plates, YEPD, which grew after 72 h incubation at 30°C.
To determine cell viability and frequency of His+ recombinants in mitotically growing cultures, 10 colonies from each strain were grown for 18 h in liquid YEPD medium, washed and plated at the appropriate dilutions on YEPD plates and on plates lacking histidine. Appropriate log transformations of the frequencies were made before calculating the means and standard deviations, and these were used in standard t‐tests.
For DSB analysis, 20 ml samples of cells were washed and stored in 50 mM EDTA at 4°C, until chromosomal DNA plugs were prepared, as previously described (Gerring et al., 1991; Zenvirth et al., 1992). DNA pulsed‐field gel electrophoresis was in 1% agarose and 0.5× TBE buffer (Maniatis et al., 1982), on a CHEF‐DR™II apparatus (Bio‐Rad). Pulsed‐field electrophoresis conditions were: 5–35 s pulses, 200 V, 18 h. Gels were blotted onto Hybond‐N nylon membranes (Amersham), which were then hybridized to 32P‐labeled probe according to Maniatis et al. (1982). Fragments of chromosome III were detected using labeled 2.2 kb EcoRV–EcoRV fragment from plasmid pSG315, as probe (Goldway et al., 1993), or a 1.55 kb XhoI–BglII fragment from the HIS4 gene, excised from plasmid B294 (provided by G.R.Fink). The total number of DSBs was quantified by scanning densitometry.
This research was supported by the Austrian friends of the Hebrew University and by the Israel Science Foundation. We thank Vardit Lavi for excellent technical assistance, and Shoshana Klein for critically reading the manuscript.
- Copyright © 1999 European Molecular Biology Organization