The SRS2 gene of Saccharomyces cerevisiae encodes a DNA helicase that is active in the postreplication repair pathway and homologous recombination. srs2 mutations are lethal in a rad54Δ background and cause poor growth or lethality in rdh54Δ, rad50Δ, mre11Δ, xrs2Δ, rad27Δ, sgs1Δ, and top3Δ backgrounds. Some of these genotypes are known to be defective in double-strand break repair. Many of these lethalities or poor growth can be suppressed by mutations in other genes in the DSB repair pathway, namely rad51, rad52, rad55, and rad57, suggesting that inhibition of recombination at a prior step prevents formation of a lethal intermediate. Lethality of the srs2Δ rad54Δ and srs2Δ rdh54Δ double mutants can also be rescued by mutations in the DNA damage checkpoint functions RAD9, RAD17, RAD24, and MEC3, indicating that the srs2 rad54 and srs2 rdh54 mutant combinations lead to an intermediate that is sensed by these checkpoint functions. When the checkpoints are intact the cells never reverse from the arrest, but loss of the checkpoints releases the arrest. However, cells do not achieve wild-type growth rates, suggesting that unrepaired damage is still present and may lead to chromosome loss.
THE SRS2 gene was first identified through dominant mutations that suppressed trimethoprim sensitivity of rad6 and rad18 mutants (Lawrence and Christensen 1979). It was again recovered as a suppressor of damage sensitivity of the rad18 mutant (Aboussekhraet al. 1989) and as a mutant that showed increased spontaneous mitotic gene conversion (Ronget al. 1991). These studies placed SRS2 in the RAD6 epistasis group for DNA repair, in an error-free postreplication repair pathway. The sequence of the SRS2 gene revealed high homology to the bacterial repair gene uvrD, encoding helicase II (Aboussekhraet al. 1989). Subsequently, it was demonstrated that the Srs2 protein has DNA helicase activity (Rong and Klein 1993). While the biochemical activity of SRS2 indicates a function in DNA metabolic activity, the precise role of the gene has remained enigmatic.
First, it is not clear why srs2 suppressor alleles are semidominant for the suppressor effect of rad6 or rad18 DNA damage sensitivity phenotypes. Second, SRS2 haploid mutants have a 5- to 10-fold increased sensitivity to UV damage (Aboussekhraet al. 1989; Palladino and Klein 1992), which is significant but is not of the magnitude of other UV-sensitive mutants. Haploid srs2 mutants are not sensitive to X-rays while diploid mutants are sensitive (Aboussekhraet al. 1989). Third, all the mitotic DNA damage sensitivities are enhanced in mutant diploid strains as compared to mutant haploid strains (Aboussekhraet al. 1989). This is unusual and has been interpreted as lethal recombination events occurring between homologous chromosomes in the absence of Srs2 function (Aboussekhraet al. 1989). Thus in the mutant the haploid has a hyperrecombination phenotype while in the diploid some of the interhomolog recombination events are lethal, while other recombination events such as intragenic recombination are increased. To make the situation even more complicated, the suppression of rad6 and rad18 mutants by srs2 mutation is thought to occur by the channeling of repair substrates into the RAD52 recombination repair pathway (Aboussekhraet al. 1989; Ronget al. 1991).
Fourth, in meiosis the gene is required for full spore viability, and in its absence spore viability is ∼50% and map distances for the LEU2-HIS4 interval and the HIS4-MAT interval are reduced 2-fold (Palladino 1991). Further clues as to the functional role of SRS2 have come from identification of double mutant combinations with srs2 mutants that are lethal or exhibit extremely poor growth and from suppressors of those srs2 mutant phenotypes. srs2 rad50 double mutants grow very poorly (Ronget al. 1991; Palladino and Klein 1992). This most likely reflects an essential role for RAD50 in recombinational repair involving sister chromatids (Saekiet al. 1980) and suggests that SRS2 functions in G2 repair in haploids. The diploid DNA damage sensitivity and lethal recombination of the srs2 mutant is suppressed by semidominant mutations in RAD51 (Aboussekhraet al. 1992), which encodes a RecA-like protein that functions in recombinational repair (Shinoharaet al. 1992), again placing SRS2 in a step in homologous recombination that may occur after the RAD51-mediated step, although an alternate interpretation involves negative regulation of homologous recombination pathways by SRS2. srs2 mutants also suppress nonnull semidominant alleles of RAD51 (Milneet al. 1995; Chanetet al. 1996). This has been interpreted as a function for the Srs2 helicase in reversing abortive recombination intermediates.
We have previously reported that srs2 mutations are lethal in a rad54 null mutant background (Palladino and Klein 1992). This lethality is suppressed by mutations in RAD51, RAD52, RAD55, and RAD57 (Schild 1995). These genes function in the recombinational repair pathway of yeast and are required for homologous recombination (Saekiet al. 1980). These results implicate SRS2 in recombinational repair, although epistasis analysis does not place SRS2 in the RAD52 repair pathway. Nonetheless, there is an interaction between SRS2 and RAD52 as null alleles of SRS2 are able to suppress nonnull alleles of RAD52 (Kaytoret al. 1995; Milneet al. 1995; Schild 1995). Although the mechanism of suppression is not understood, it may reflect a need for the Srs2 protein to stabilize and promote recombination between substrates with limited homology (Paques and Haber 1997). The rad52 alleles may attempt recombination between sequences of reduced homology and this reduced fidelity recombination requires SRS2. In the absence of functional Srs2p, aberrant recombination would be prevented and the rad52 alleles would be suppressed.
Additional evidence for SRS2 functioning in recombination comes from studies of the interaction between SRS2 and SGS1. SGS1 encodes a DNA helicase related to the RecQ family of helicases (Gangloffet al. 1994; Wattet al. 1995; Luet al. 1996). sgs1 mutant strains are characterized by increased genomic instability (Wattet al. 1996; Yamagataet al. 1998). Although the single null allele mutant strains exhibit normal growth, the double mutant srs2 sgs1 is inviable or grows extremely poorly (Leeet al. 1999; Gangloffet al. 2000). The growth defect of the srs2 sgs1 double mutant can be suppressed by mutations in the RAD51, RAD55, or RAD57 genes (Gangloffet al. 2000), indicating that the growth defect is caused by attempted homologous recombination that cannot be completed when both Srs2p and Sgs1p are missing.
To gain additional insight into the biological role of SRS2, we have examined other DNA recombination functions for synthetic interactions with a srs2 null allele strain. We have also asked what types of mutants suppress srs2 phenotypes. The results link SRS2 to homologous recombination that is associated with DNA replication and DNA damage that is sensed by DNA damage checkpoints.
MATERIALS AND METHODS
Yeast strains: Parental strains used in this study are listed in Table 1. All strains are isogenic and are in the W303 background (W303 genotype: leu2-3, 112 his3-11, 15 ade2-1 ura3-1 trp1-1 can1-1 RAD5). sgs1 disruption strains were a gift from Rodney Rothstein. The mre11 and exo1 strains were a gift from Lorraine Symington. The rad9, rad17, rad24, and mec3 strains were a gift from Ted Weinert. Strains, if needed, were converted to the RAD5 version of W303 through crosses. The rad54 point mutation strains have been described (Petukhovaet al. 1999). All double and triple mutant strains were derived from crosses using the parental single mutant strains listed in Table 1. All recombination and checkpoint mutants are null alleles with the exception of the two indicated rad54 point mutants and the one indicated mre11 point mutant.
Media and growth conditions: Standard media were prepared as described (Shermanet al. 1986). All strains were grown on solid media at 30°.
srs2 and rad54: We have previously reported that the srs2 rad54 double null mutant is lethal (Palladino and Klein 1992). This lethality is suppressed by loss of the homologous recombination functions of RAD51, RAD52, RAD55,or RAD57 (Schild 1995 and this study). However, not all mutations of genes involved in homologous recombination or repair of a double-strand break rescue the srs2 rad54 lethality. The RAD1 gene encodes an endonuclease that is active in nucleotide excision repair and some recombination reactions (Klein 1988; Schiestl and Prakash 1988; Thomas and Rothstein 1989; Siedeet al. 1993; Tomkinsonet al. 1993; Ivanov and Haber 1995; Kirkpatrick and Petes 1997). However, a rad1 null mutation was not able to rescue the srs2Δ rad54Δ strain.
RAD54 encodes a protein of the SNF2/SW12 family and contains DNA helicase consensus motifs (Eisenet al. 1995). The protein has demonstrated in vitro ATPase activity and promotes Rad51-mediated strand exchange (Petukhovaet al. 1998) via a change in DNA double helix conformation (Petukhovaet al. 1999). Two point mutations in the first helicase consensus domain, rad54K341R and rad54K341A, have been shown to abolish ATP hydrolysis and Rad51-promoted homologous DNA pairing as well as mitotic recombination (Petukhovaet al. 1999). Both of these mutations in a srs2Δ background are lethal, showing that the essential function provided by Rad54 protein in a srs2Δ strain requires the ATPase activity.
Our studies on rad54 rdh54 diploid double mutants had suggested that the double mutant poor growth was due to the accumulation of DNA damage through attempted recombination that was sensed by the DNA damage checkpoint functions. This finding prompted us to speculate that the srs2 rad54 lethality might be due to the accumulation of DNA damage from attempted recombination that was sensed by the DNA damage checkpoint functions. Therefore the role of the checkpoint functions in preventing srs2 rad54 cells from progressing through the cell cycle was assessed by determining the growth phenotype of triple mutants. Triple mutants were segregated in crosses. An example of such a cross is shown in Figure 1. In Figure 1A growth of colonies of the genotype srs2 rad54 rad17, srs2 rad54 rad24, or srs2 rad54 mec3 is compared to a wild-type strain. While the triple mutant strains grow slower than wild type and throw off colonies of varying sizes, these genotypes are viable. In contrast, a rad9 mutation is unable to rescue the srs2 rad54 double mutant (Figure 1B). Figure 1, B and C compares the ability of a rad51 mutation vs. a rad17 mutation to suppress the srs2 rad54 lethality. It can be seen that rad51 is more effective in suppressing the srs2 rad54 lethality, suggesting that the checkpoint mutants suppress the double mutant by release from growth arrest, but DNA damage is still present. Further support for this statement comes from plating efficiencies. One hundred unbudded cells of the srs2 rad54 rad17 and srs2 rad54 rad51 genotypes were micromanipulated to fresh YPD plates. A total of 50/100 srs2 rad54 rad17 cells grew to visible colonies whereas 93/100 srs2 rad54 rad51 cells grew to visible colonies.
srs2 and rdh54: The RAD54 and RDH54 genes have related sequences, but the single mutants have unique effects on mitotic recombination and meiotic viability (Klein 1997; Shinoharaet al. 1997). We have previously reported that, while in contrast to the srs2 rad54 haploid mutant lethality the srs2 rdh54 haploid mutant shows normal growth, lethality is exhibited in the srs2 rdh54 diploid (Klein 1997). Moreover, this lethality is suppressed by mutation in any one of the recombination repair genes RAD51, RAD52, RAD55, or RAD57 (Klein 1997). Similar to the suppression of srs2 rad54 by DNA damage checkpoint mutations, the srs2 rdh54 diploid lethality is suppressed by mutations in RAD17, RAD24, or MEC3, but not by mutation of the RAD9 gene (Table 2).
srs2 and rad50/xrs2/mre11: The Rad50, Xrs2, and Mre11 proteins function in a complex to protect DNA ends in end-joining reactions and in meiotic recombination (for a review see Haber 1998). The RAD50, XRS2, and MRE11 genes are not required for mitotic homolog recombination, but are required for recombination between sister chromatids (Saekiet al. 1980). The srs2 null allele mutant has a synthetic growth interaction with null allele mutations in RAD50, XRS2, or MRE11 (see Figure 2A) and this is suppressed by mutations in the recombination repair genes (Table 3). However, the srs2 rad50 poor growth was not suppressed by mutation of the DNA damage checkpoint function RAD17 (Figure 2B and Table 3). The plating efficiency of 100 micromanipulated unbudded cells from a spore colony of the srs2 rad50 genotype was 49/100 while the plating efficiency of 100 micromanipulated unbudded cells from a spore colony of the srs2 rad50 rad17 genotype was 34/100. All colonies that grew were extremely small compared to a wild-type strain.
MRE11 encodes a protein with single-stranded endonuclease activity (Moreauet al. 1999). Mutation of consensus motifs in the phosphodiesterase domains abolishes the nuclease activity, but has little effect on mitotic growth, recombination, end-joining or telomere length (Moreauet al. 1999). The effect of mre11-nuclease-deficient alleles on growth of srs2 strains was examined (Figure 2A). The Mre11 nuclease activity is not required for viability of srs2 mutants. Since the double null allele growth defect is rescued by recombination repair mutations, this suggests that the poor growth is due to attempted recombination. The fact that the srs2 mre11-nuclease-deficient strain is viable indicates that recombinational repair is not impaired in this strain.
RAD50, XRS2, and MRE11 are also required for telomere length control and in the mutants telomeres are shortened (Kironmai and Muniyappa 1997; Leet al. 1999). Since double mutants of srs2 combined with either rad50, xrs2, or mre11 show poor growth, one possibility is that telomere length is greatly altered in the double mutant. This was excluded by first showing that srs2 strains have normal telomere lengths and then observing that srs2 rad50 strains have telomeres of the rad50 characteristic length (H. Klein and R. Boguslavsky, data not shown).
srs2 and exo1: The EXO1 gene encodes an exonuclease that functions in mitotic recombination (Fiorentiniet al. 1997). The exo1 mutant when combined with a mutant in nuclease involved in DNA replication is lethal (Tishkoffet al. 1997a). This property is shared with genes involved in recombinational repair (Symington 1998). Since the srs2 mutant also has genetic interactions with the same set of genes, the phenotype of the double mutant srs2 exo1 was determined. The double mutant srs2 exo1 was fully viable, showing that neither gene is required for any essential function when the other gene is defective.
srs2 sgs1 and srs2 top3: SGS1 encodes a DNA helicase of the same polarity as the Srs2 DNA helicase (Luet al. 1996; Bennettet al. 1998). It has been reported that the double mutant srs2 sgs1 is lethal or grows extremely poorly (Leeet al. 1999; Gangloffet al. 2000). This growth phenotype is due to attempted recombination as mutations in the recombination repair genes RAD51, RAD52, RAD55,or RAD57 rescue the growth defect. We have confirmed these observations. sgs1 cells have been found to be partially defective for an intra-S checkpoint for DNA damage that acts in parallel with the RAD24 checkpoint function (Frei and Gasser 2000). Therefore, it was of interest to examine the phenotype of the srs2 sgs1 rad24 triple mutant. We found that the rad24 mutant did not suppress the lethality or poor growth of the srs2 sgs1 double mutant (Figure 3). Suppression was also measured by plating efficiency of micromanipulated unbudded cells from spore colonies onto fresh growth medium. The srs2 strain had a high plating efficiency (47/50) while the sgs1 mutant plating efficiency was moderate (36/50). Although spores of the srs2 sgs1 genotype were viable, the growth was slow and the plating efficiency was 0/50. The presence of a rad24 mutation in this genotype increased the plating efficiency to 31/100 (srs2 sgs1 rad24 genotype), but the colonies grew very slowly and never reached wild-type size.
The Sgs1 protein of yeast interacts physically with the Top3 protein, which has type I topoisomerase activity. Moreover, the slow growth phenotype of the top3 mutant is suppressed by a sgs1 mutation (Gangloffet al. 1994). Therefore we examined the phenotype of the srs2 top3 double mutant. The double mutant is lethal (Figure 4) and this lethality is also suppressed by loss of the recombination repair pathway through mutations in RAD-51, RAD52, RAD55, or RAD57 (Figure 4 and Table 3).
srs2 and rad27: RAD27 encodes an exonuclease that is involved in the completion of lagging strand synthesis (Harrington and Lieber 1994; Reaganet al. 1995; Sommerset al. 1995). rad27 mutants have elevated rates of recombination between direct repeats (Sommerset al. 1995; Tishkoffet al. 1997b; Symington 1998) among the multiple mutant phenotypes and rad27 mutants are lethal when combined with mutations in genes involved in double-strand break repair and homologous recombination (Sommerset al. 1995; Vallen and Cross 1995; Tishkoffet al. 1997b; Symington 1998). Therefore the phenotype of the double mutant srs2 rad27 was determined. The double mutant was inviable (Figure 5), undergoing 8–10 cell divisions before arresting as enlarged doublet cells or lysing, providing additional support for a role of SRS2 in homologous recombination and double-strand break repair. A rad51 mutation was not able to suppress the double mutant lethality, but this is most likely due to the lethality of the rad51 rad27 genotype (Tishkoffet al. 1997b; Symington 1998). The DNA damage checkpoint gene RAD17 also was not able to rescue the srs2 rad27 double mutant lethality when mutant, but this could be due to lethality of the rad27 rad17 genotype. RAD17 is part of the DNA damage checkpoint that also includes the RAD9 and MEC1 genes. rad27 rad9 and rad27 mec1 double mutants have been reported to be lethal (Vallen and Cross 1995).
All of the double mutant interactions with srs2 and suppression by defects in homologous recombination or the DNA damage checkpoint are summarized in Tables 2 and 3. The findings would indicate that Srs2p acts in recombination after commitment to repair via homologous recombination. If homologous recombination is prevented by inhibiting the initial steps of forming a Rad51 filament on single-stranded DNA, then srs2 rad54 or srs2 rad50/xrs2/mre11 lethality/poor growth can be prevented. This suggests, first, that there exist other pathways in the cell to repair the spontaneous damage that occurs in these double mutants and, second, that once a cell is committed to repair damage through homologous recombination, through action of Rad51p, there is no alternate repair option available. Experiments on in vivo repair of substrates containing a double-strand break indicate an important role for the Srs2 helicase in the repair process (Paques and Haber 1997). The lethality of haploid strains bearing srs2 and rad54 mutations shows that Srs2p can have an important role in repair of spontaneous DNA damage, but in srs2 single mutants there is some substrate specificity as to the action of Srs2p in repair as diploid srs2 mutants show a greatly increased damage sensitivity and it is the attempted homologous recombination between homolog chromosomes, not sister chromatids, that is lethal in the absence of Srs2 helicase (Aboussekhra et al. 1989, 1992).
Studies of srs2 mutants as suppressors of nonnull alleles of RAD51 and RAD52 have provided an alternative interpretation of the role of Srs2p in homologous recombination (Kaytoret al. 1995; Milneet al. 1995; Chanetet al. 1996). These articles have suggested two regulatory functions for SRS2. First, the Srs2 helicase is proposed to antagonize the action of Rad51p and Rad52p in homologous recombination. Through such an activity Srs2p would have an antirecombinase function and could also control use of the homologous recombination pathway to repair a spontaneous damaged substrate. Srs2p would determine which lesions could be processed through homologous recombination. Therefore Srs2p would inhibit homologous recombination prior to the commitment to repair by homologous recombination. In this scheme one would then suggest that, in the absence of Srs2p, substrates are inappropriately targeted for repair through the homologous recombination pathway. When Rad54p is also absent, the processing of substrates through homologous recombination becomes an irreversible lethal action. Rescue through loss of Rad51p, Rad52p, Rad55p, or Rad57p is thought to happen by inhibiting the initial steps of homologous recombination, prior to the Rad54p-mediated step (Petukhovaet al. 1998). Rescue through loss of DNA damage checkpoints implies that the cells are arrested due to the accumulation of unrepaired DNA damage that is sensed by these checkpoint genes.
Other studies have suggested that the Srs2 helicase can act at a later step in recombination (Ronget al. 1991; Paques and Haber 1997). Whether the helicase acts in concert with Rad54 protein is not known. However, under either model the basic scenario for suppression remains the same, namely, preventing recombination at an early step rescues lethal mutant combinations.
The second regulatory function for Srs2p has been proposed to be repression of a repair pathway that parallels the RAD52 recombination repair pathway (Kaytoret al. 1995). This remains conjectural, although the recent finding of multiple repair pathways and the RAD59 gene, which is related to RAD52 (Rattray and Symington 1995; Bai and Symington 1996), provides support for this idea.
The arrest of haploid cells through loss of Srs2 and Rad54 proteins triggers an irreversible arrest that is sensed by DNA damage checkpoint functions. Loss of the checkpoint functions allows cells to grow, but full viability as measured by plating efficiency and colony size is not restored, suggesting that unrepaired damage remains or that chromosome loss is occurring. This cannot be measured in haploid cells for essential chromosomes. The data thus far show that mutations in RAD17, RAD24, and MEC3 can partially suppress the growth arrest, but loss of RAD9 function cannot. Why there is a differential suppression by this group of DNA damage checkpoint genes is not clear, but for some types of DNA damage RAD9 acts additively to RAD17/RAD24/MEC3 in sensing DNA damage (Lydall and Weinert 1995; de la Torre-Ruizet al. 1998). However, in the DNA damage arrest of rad54 rdh54 mutants, no additive effect of a rad9 and a rad24 mutation has been seen in suppression of poor growth (H. Klein, unpublished observations).
The Sgs1 helicase is a component of the S-phase checkpoint response acting upstream of the Rad53 kinase (Frei and Gasser 2000). This response has been shown to act in parallel to the checkpoint response that requires RAD24 (Frei and Gasser 2000). The Srs2 helicase has also been shown to be involved in an S-phase checkpoint response that is dependent on RAD17 and RAD24 (Liberiet al. 2000). The observation that the srs2 sgs1 double mutant poor growth is suppressed by mutations in recombination repair functions, but not by a mutation in the RAD24 checkpoint function, may indicate that the double mutant defect is due to attempted recombination, which can trigger a checkpoint response. Recombination may be an attempt to repair DNA damage that has its origin in a replication defect (Frei and Gasser 2000). Mutation of RAD24 in the srs2 sgs1 double mutant may result in a defect in two parallel checkpoint response pathways, with the result of continued poor growth. Alternatively, loss of the RAD24 checkpoint function may permit the srs2 sgs1 cells to progress through the cell cycle without repairing DNA lesions that have become lethal. Since a rad24 mutation can suppress the lethality of the srs2 rad54 mutant, but not the srs2 sgs1 mutant, this would suggest that the DNA intermediate, presumably related to attempted recombination, that occurs in each of these cell types is different. The difference may be the presence of an unrepaired double-strand break or a structure that can be shunted into an alternate repair pathway.
It has been suggested that the Sgs1 helicase can act on different substrates, controlled by the action of the topoisomerases I and III and the Srs2 DNA helicase (Dunoet al. 2000). In this regard the srs2 sgs1 poor growth/lethality is proposed to be similar to the srs2 top3 lethality as the alternative helicase/topoisomerase activities of Sgs1/Top3 and Srs2/Top1 are disrupted in these double mutants. What determines the specificity of the target substrate is not known, but must be linked to recombinational repair.
Srs2 helicase has some interaction with Rad50p/Xrs2p/Mre11p although whether this is merely genetic or also has a physical basis is not known. We have not detected any interaction with these proteins from twohybrid screens (H. Klein, unpublished observations). However, the genetic interaction does not involve the Mre11p nuclease function, suggesting that Srs2p may be targeted to open up broken ends by the Rad50p/Xrs2p/Mre11p complex. The failure of a rad17 mutation to suppress the poor growth phenotype of the srs2 rad50 mutant can be interpreted in two ways. Loss of the RAD17 checkpoint function may permit srs2 rad50 cells to progress through the cell cycle with unrepaired damage, with continued poor growth due to some type of genomic instability. Alternatively, the DNA damage that accumulates in the srs2 rad50 mutant may not be a target for the DNA damage checkpoints that act through RAD17.
We have no information as to when the spontaneous damage is occurring in the srs2 rad54 strain and in the other double mutant combinations that exhibit poor or no growth. SRS2 is induced in expression at the beginning of S phase and is induced in G2 by DNA damage (Heudeet al. 1995), suggesting that these phases of the cell cycle may be more prone to DNA damage resulting from loss of Srs2 protein. In the case of the srs2 rad27 lethality, it is likely that the damage occurs during S phase as RAD27 is required for DNA replication (Harrington and Lieber 1994; Reaganet al. 1995; Sommerset al. 1995). If the damage that is targeted for repair by Srs2 helicase, either as a catalytic function in repair or through regulation of which substrate is repaired by the DSB repair pathway, occurs during S phase, this would reinforce the close link between DNA replication and recombination and also bring into play a link between DNA replication and nonrecombination (non-DSB repair) repair modes.
I thank Rodney Rothstein, Lorraine Symington, and Ted Weinert for strains; and Revekka Boguslavsky for technical assistance. This work was supported by U.S. Public Health Service grant GM53738.
Communicating editor: L. S. Symington
- Received September 5, 2000.
- Accepted November 8, 2000.
- Copyright © 2001 by the Genetics Society of America