DNA repair by a Rad22–Mus81-dependent pathway that is independent of R
http://www.100md.com
《核酸研究医学期刊》
Department of Biochemistry, University of Oxford, South Parks Road, Oxford OX1 3QU, UK
* To whom correspondence should be addressed. Tel: +44 1865 275192; Fax: +44 1865 275297; Email: matthew.whitby@bioch.ox.ac.uk
The authors wish it to be known that, in their opinion, the first two authors should be regarded as joint First Authors
ABSTRACT
In budding yeast most Rad51-dependent and -independent recombination depends on Rad52. In contrast, its homologue in fission yeast, Rad22, was assumed to play a less critical role possibly due to functional redundancy with another Rad52-like protein Rti1. We show here that this is not the case. Rad22 like Rad52 plays a central role in recombination being required for both Rhp51-dependent and -independent events. Having established this we proceed to investigate the involvement of the Mus81–Eme1 endonuclease in these pathways. Mus81 plays a relatively minor role in the Rhp51-dependent repair of DNA damage induced by ultraviolet light. In contrast Mus81 has a key role in the Rad22-dependent (Rhp51-independent) repair of damage induced by camptothecin, hydroxyurea and methyl-methanesulfonate. Furthermore, spontaneous intrachromosomal recombination that gives rise to deletion recombinants is impaired in a mus81 mutant. From these data we propose that a Rad22–Mus81-dependent (Rhp51-independent) pathway is an important mechanism for the repair of DNA damage in fission yeast. Consistent with this we show that in vitro Rad22 can promote strand invasion to form a D-loop that can be cleaved by Mus81.
INTRODUCTION
Problems associated with DNA damage can be compounded during S-phase. Single-strand breaks can act as run-off sites for polymerases resulting in replication fork collapse and the formation of one-ended double-strand breaks (DSBs). Bulky DNA lesions can block fork progress, and may in some instances result in replisome disassembly and the regression of the fork where the nascent strands are unwound and anneal to each other. Alternatively, the replisome may skip damaged sections of the genome leaving behind lesion-containing single-strand gaps. Problems such as these can be dealt with by homologous recombination, which promotes the repair of a break or a lesion-containing gap, enables template switching for polymerases to bypass DNA lesions, and helps restart replication at non-origin sites .
In Saccharomyces cerevisiae, recombination depends on the RAD52 epistasis group of genes . Typically a 3' tail, exposed by the resection of a DSB or unwinding of a single-strand gap, is bound by Rad51, which then catalyses strand invasion of a homologous DNA molecule to form a D-loop. The loading of Rad51 on to the DNA and catalysis of strand invasion are variously aided by proteins such as Rad52, Rad54, Rad55 and Rad57. In some situations Rad52 can promote the formation of a D-loop without Rad51 through its ability to anneal complementary ssDNAs (4). D-loops can act as sites for replisome reassembly, and therefore by promoting strand invasion recombination can act to restart replication at stalled or broken replication forks (1).
Various mechanisms have been proposed for RDR and most involve the formation and subsequent processing of one or two Holliday junctions (HJs) (1). One enzyme that has been implicated in resolving HJs in eukaryotes is Mus81–Eme1 (Mus81–Mms4 in S.cerevisiae). Mus81 is a member of the XPF family of endonucleases, and together with Eme1/Mms4 has been shown to cleave synthetic HJs in vitro. Evidence that it cleaves HJs in vivo has come from genetic studies which have shown that mus81 and eme1/mms4 mutant phenotypes are suppressed by the bacterial HJ resolvase RusA (5–9). However, the idea that Mus81–Eme1/Mms4 is a HJ resolvase has been questioned by in vitro data showing that recombinant enzyme, purified from Escherichia coli, exhibits only weak HJ cleavage activity, whilst cleaving fork structures and 3' flaps relatively well (8,10–12). In the case of Mus81–Mms4 this has lead to proposals that its prime function is to remove 3' flaps that are formed during the repair of DSBs and single-strand gaps by synthesis-dependent strand annealing (SDSA) (6). However, these models do not provide a satisfactory explanation why RusA suppresses mus81 and eme1/mms4 mutant phenotypes.
It has been shown recently that recombinant Mus81–Eme1 can resolve four-way DNA junctions that contain a single-strand nick or gap at or near the junction centre at least as well as a 3' flap (10,13,14). Such four-way junctions are formed during the regression of a replication fork and prior to the formation of a HJ at a D-loop. Based on these data we have proposed that Mus81–Eme1 may cleave regressed replication forks to promote RDR, and resolve the four-way DNA junction at a D-loop during RDR (Figure 1). In these models RusA is able to suppress mus81 mutant phenotypes not because it provides a direct substitute for Mus81–Eme1 but because it is able to resolve the HJs that will form in its absence.
Figure 1. Hypothetical model showing potential roles that Mus81–Eme1 might have in processing stalled and broken replication forks. Nascent strands are shown as red lines with arrowheads at 3' ends. The replication fork block is indicated by a solid circle, and the position of Mus81–Eme1 cleavage sites by the labelled arrows.
Here we have investigated the relationship of Mus81–Eme1 (for simplicity referred to as Mus81 throughout this paper) to recombinational pathways of DNA repair in Schizosaccharomyces pombe. Existing data suggest that the pathways of recombination in S.pombe are similar to those in S.cerevisiae, with the exception that Rad22 (the S.pombe Rad52 homologue) is not required for all recombination possibly due to functional redundancy with another Rad52-like protein Rti1 (2). However, we show here that this is not the case and that Rad22 is in fact required for most recombination just like Rad52 in S.cerevisiae. The original confusion in the literature appears to stem from the ease with which rad22 strains acquire suppressor mutations. Having established that there are Rhp51-dependent (the S.pombe Rad51 homologue) and -independent pathways of recombination, which are both dependent on Rad22, we show that Mus81 functions predominantly in the Rhp51-independent pathway. We also show that Mus81 can cleave D-loops that are formed by Rad22 in vitro. These data are consistent with our proposal that Mus81 resolves D-loops formed by Rad22 (without Rhp51) during the repair of broken replication forks.
MATERIALS AND METHODS
General techniques and strains
Procedures for S.pombe genetics were as described by Moreno et al. (15). Media for S.pombe have been described (16). The complete media was yeast extract medium (YES) and the minimal medium was Edinburgh minimal medium (EMM) each supplemented with appropriate amino acids. Hydroxyurea (HU), methyl-methanesulfonate (MMS) and camptothecin (CPT) were added to media as indicated.
Strains
The S.pombe strains used in this study are listed in Table 1. The rhp51arg3 and rad22ura4 mutants both contain a replacement of the entire open reading frame by the indicated marker, and were made by PCR-based gene targeting as described by B?hler et al. (17). Genuine deletion mutants were identified by genomic Southern blot analysis.
Table 1. S.pombe strains
Spot assays and quantitative survival curves
Cell cultures growing exponentially in liquid media were adjusted to a density of 1 x 107 cells/ml, serially diluted, and 10 μl aliquots of each dilution spotted onto media containing genotoxins as indicated. For ultraviolet (UV) exposure plates were irradiated using a Stratalinker (Stratagene) after spotting. Plates were incubated for 2–5 days at 30°C before being photographed. All spot tests were repeated at least once as an independent experiment to ensure reproducibility. For survival curves all data points represent the mean from at least three independent experiments.
Recombination assay
Mitotic recombination was assayed by the recovery of Ade+ recombinants from strains containing the intrachromosomal recombination substrate shown in Figure 3C. Spontaneous recombinant frequencies were measured as described by Osman et al. (18). Recombinant frequencies are the mean from at least three independent assays where five independent colonies were tested in each assay. Two sample t-tests were used to determine the statistical significance of differences in recombination frequencies.
Figure 3. Rhp51-dependent and -independent recombination and DNA repair. (A) Spot assay comparing the sensitivity of wild-type (MCW1221), rhp51– (MCW1088), rad22– (MCW1285), and rad22– rhp51– (MCW1335) strains to UV, HU, MMS and CPT. (B) Survival curves showing the UV and -ray sensitivity of the same strains as in ‘A’. (C) Schematic of the recombination substrate and recombinant products. Solid and open circles represent the ade6-L469 and ade6-M375 mutations, respectively. (D) Comparison of the spontaneous recombinant frequencies of wild-type (MCW429), rhp51– (MCW1224), rhp55– (MCW431), rhp54– (MCW1225), rad22– (MCW1494), rhp51– rad22– (MCW1038), rhp55– rad22– (MCW1227), rhp51– rhp55– (MCW1228) and rhp54– rhp55– (MCW1229) strains. Error bars represent standard deviations about the mean.
Proteins
Recombinant S.pombe Mus81–Eme1 and S.cerevisiae Mus81–Mms4 were expressed in E.coli and purified as described previously (8). The overexpression and purification of recombinant Rad22 is described below. Concentrations of Mus81–Eme1/Mms4 and Rad22 are in moles of heterodimer and monomer, respectively.
Overexpression and purification of recombinant Rad22
The rad22 gene was amplified by PCR from genomic DNA with NdeI- and BamHI-flanking restriction sites to enable cloning in-frame with the N-terminal hexahistidine tag in pET14b (Novagen). The resultant plasmid pMW601 expresses His-tagged Rad22 from a T7 phage ?10 promoter. E.coli BL21-RIL cells (Stratagene) containing pMW601 were grown with aeration at 25°C in a Luria–Bertani broth containing 125 mg/ml carbenicillin and 50 mg/ml chloramphenicol. At a cell density corresponding to an A600 of 0.5, His-tagged Rad22 was induced by adding isopropyl-1-thio-?-D-galactopyranoside to a final concentration of 1 mM, following which the cells were incubated for a further 2.5 h. The cells were then harvested by centrifugation, resuspended in Buffer H (50 mM potassium phosphate, pH 8.0, 10 mM ?-mercaptoethanol, 0.3 M NaCl, 10% glycerol) with protease inhibitors and 1% Triton X-100 and lysed by passage through a French pressure cell at 30 000 p.s.i. After centrifugation at 43 700 g for 50 min, the supernatant was loaded directly onto a 1 ml nickel-nitrilotriacetic acid Superflow column (Qiagen) that was washed with 60 ml of Buffer H and 25 mM imidazole before eluting bound Rad22 with 3 ml of Buffer H and 200 mM imidazole. This sample was then diluted with 6 ml of Buffer A (50 mM Tris–HCl, pH 8.0, 1 mM EDTA, 1 mM dithiothreitol, 10% glycerol) and loaded directly onto a 1 ml HiTrap heparin column (Amersham Pharmacia Biotech). The column was washed with 10 ml of Buffer A and 0.1 M NaCl before eluting bound Rad22 with an 18 ml gradient from 0.2 to 1.0 M NaCl. Rad22 eluting between 550 and 650 mM NaCl was pooled and dialyzed against Buffer A with 0.1 M NaCl before storing as aliquots at –80°C.
DNA substrates
X174 form I DNA was from New England BioLabs. The X174-based partial duplex was made from oligonucleotides oMW585 (5'-TGCCGAATTCTACCAGTGCACAAAGTAAGAGCTTCTCGAGCTGCGCAAGGATAGGTCGAATTTTCTCATTTTCCGCCAGCAGTCCACTTCGATTTAATTC-3') and oMW586 (5'-TGCACTGGTAGAATTCGGCA-3'), and the pBR322-based partial duplex from oMW592 (5'-TGCCGAATTCTACCAGTGCACGCCTCCATCCAGTCTATTAATTGTTGCCGGGAAGCTAGAGTAAGTAGTTCGCCAGTTAATAGTTTGCGCAACGTTGTTG-3') and oMW586. The procedures for partial duplex preparation have been described previously (19). Each substrate was labelled with 32P at the 5' end of one of its component oligonucleotides. For both partial duplexes the longer oligonucleotide was labelled. The concentration of partial duplexes was estimated by relating the specific activity of the labelled oligonucleotide to the activity of the purified substrate and is expressed in molar concentrations of DNA substrate. To purify D-loops from Rad22 and un-reacted partial duplex, deproteinized reaction mixtures were applied to 3.5 ml sepharose CL-2B columns, which were developed with 20 mM Tris–HCl (pH 8.0) and 0.5 mM MgCl2.
D-loop and nuclease assays
D-loop assays (25 μl) consisted of 1.2 nM X174 or pBR322 form I DNA and 0.2 nM of 32P-labelled partial duplex in reaction buffer (25 mM Tris–HCl, pH 8.0, 1 mM dithiothreitol, 100 μg/ml BSA, 6% glycerol) with 10 mM MgCl2, Rad22 and Mus81–Eme1 as indicated. Rad22 was typically pre-incubated with the partial duplex for 5 min at 30°C prior to the addition of the form I DNA and Mus81–Eme1 as indicated. Incubation was then continued for 90 min before stopping the reaction with 5 μl of stop mix (2.5% SDS, 200 mM EDTA, 10 mg/ml proteinase K) followed by a further 15 min at 30°C. Reactions were analysed by electrophoresis through 1% agarose gels in 1x TBE buffer at 3.75 V/cm with phosphorus imaging on a Fuji FLA3000. For the cleavage of purified D-loops by Mus81, reaction mixtures (25 μl) contained 0.01 nM D-loop and 0.05 nM of non-D-loop X174 DNA in reaction buffer with 10 mM MgCl2 and Mus81–Eme1 as indicated. Reactions were incubated at 30°C for 90 min before stopping and analysing as described above.
RESULTS
Rad22 plays a pivotal role in recombinational repair in S.pombe
In S.cerevisiae RAD52 is required for nearly all homologous recombination (2). However, in S.pombe genetic analyses suggest that Rhp51 can function without Rad22 (20,21). These data have been explained by the presence of another Rad52-like protein in S.pombe called Rti1 (or Rad22B), which could substitute in the absence of Rad22 (21–23).
To clarify the genetic interactions between rad22 and the rhp51 epistasis group we first constructed a new rad22 deletion strain by replacement of the entire rad22 open reading frame with an ura4+ marker. This strain grew much slower and was much more hypersensitive to UV light, HU, MMS and CPT than previously published rad22 mutant strains (Figure 2). These more extreme phenotypes were solely due to the deletion of rad22 because they could be fully complemented by a cloned copy of the gene expressed from a plasmid (data not shown). The discrepancy in phenotypes between these different rad22 mutant strains is explained by the presence of suppressor mutations, which will be analysed elsewhere.
Figure 2. Spot assay comparing the genotoxin sensitivity of different rad22 mutant strains. The strains used were MCW1221 (wild-type), MCW1285 (new rad22–), MCW1222 , MCW4 and MCW1501 . Cultures were serially diluted and spotted onto YES agar as described in Materials and Methods. The neat spot ‘1’ represents 105 cells plated.
Having obtained a non-suppressed rad22 mutant we could unambiguously establish rad22's relationship to the rhp51 epistasis group. A rad22 rhp51 double mutant strain was constructed and compared to its respective single mutant strains for sensitivity to different DNA damaging agents. The rad22 single mutant is more sensitive to UV, HU, MMS and CPT than a rhp51 single mutant but is similarly sensitive to gamma irradiation (Figure 3A and B). The double mutant exhibits slightly more sensitivity to most of these DNA damaging agents than the rad22 single mutant (Figure 3A and B). This enhanced sensitivity is particularly noticeable at high UV doses where the sensitivity of the rad22 strain appears to plateau off (Figure 3B). The majority of the rad22 mutant survivors at high UV doses had acquired a suppressor mutation, and, as suppression depends on rhp51+ (our unpublished data), this explains why the rad22 single mutant appears to be slightly less sensitive to DNA damage than the rad22 rhp51 double mutant. When taken together these data show that rad22 and rhp51 are epistatic for the repair/tolerance of DNA damage. For DSBs caused by gamma irradiation it would seem that Rhp51 and Rad22 mainly work together to promote repair since both rhp51 and rad22 single mutants show approximately the same sensitivity as a rhp51 rad22 double mutant. However, the greater sensitivity of a rad22 mutant compared to a rhp51 mutant to agents such as UV, HU, MMS and CPT, which variously cause replication fork stalling, blockage and breakage, suggests that the recombinational repair of damaged replication forks can be undertaken by Rad22 both with and without Rhp51.
Pathways of direct repeat recombination in S.pombe
To investigate the relationship between rhp51 and rad22 with respect to recombination we used strains containing a non-tandem direct repeat of ade6– heteroalleles to measure Ade+ recombinant frequencies (Figure 3C). Such recombinants arise from inter- or intra-chromatid recombination by a number of different pathways, which in part can be categorised using a his3+ gene placed between the repeats that allows Ade+ recombinants that are His– (deletion types) to be distinguished from those that are His+ (conversion types) (24). In wild-type strains 3.5 Ade+ recombinants are generated in every 104 viable cells of which 30% are conversion types and 70% are deletion types (Figure 3D). Mutants from the rhp51 epistasis group, including rhp51, rhp54 and rhp55 single mutants and rhp51 rhp55 and rhp54 rhp55 double mutants, are devoid of conversion type recombinants but exhibit a 4- to 5-fold increase in deletion type recombinants compared to wild-type (Figure 3D). The rad22 mutant is also devoid of conversion types, but unlike the rhp51, rhp54 and rhp55 mutants it exhibits a >3-fold reduction in recombinant frequency (<1 x 10–4) (Figure 3D). The same low level of recombinants and absence of conversion types is seen in both rhp51 rad22 and rhp55 rad22 double mutants. These data show that two Rad22-dependent pathways are responsible for the majority of recombination between repeated DNA sequences in S.pombe. One of these pathways involves the Rhp51 epistasis group of proteins, and is required for generating conversion type recombinants, whilst the other is independent of Rhp51 and only generates deletion type recombinants.
Mus81 functions in Rad22-dependent pathways of DNA repair
In S.cerevisiae Mus81/Mms4 functions in the Rad52-dependent pathway for the repair/tolerance of UV- and MMS-induced damage (5). Other genetic studies have shown that for UV-induced damage Mus81/Mms4 functions exclusively in the Rad54-dependent sub-pathway of Rad52-dependent repair (25). However, for MMS it appears to work in Rad52-dependent pathways that are both dependent and independent of Rad54 (25). Similar to S.cerevisiae, Mus81 appears to function exclusively in a Rhp51-dependent pathway for the response to UV-induced damage in S.pombe (26).
To more thoroughly establish the recombination repair pathways in which Mus81 functions in S.pombe, we compared the sensitivity of rhp51, rad22 and mus81 single, double and triple mutants to UV, HU, MMS and CPT. The mus81 rhp51 double mutant shows an increase in sensitivity to each of these agents compared to its respective single mutant strains (Figure 4A and B). In the case of HU, MMS and CPT the increase is synergistic, whereas for UV it is approximately additive. Since our data for UV sensitivity contradicted a previous study (26), we sought to confirm our data by constructing several independent mus81– rhp51– double mutant strains, made from two different rhp51 deletion strains. Each of these double mutant strains exhibits a similar increase in sensitivity to UV, HU, MMS and CPT as shown in Figure 4A–C (data not shown). These data indicate that Mus81 functions in a separate but overlapping pathway to Rhp51 for the repair of at least some types of DNA damage.
Figure 4. Mus81 functions in both Rhp51-dependent and -independent repair pathways depending on the type of DNA damage. (A) Spot assay showing the relative sensitivities to UV, HU, MMS and CPT of wild-type (MCW45), mus81– (MCW745), rhp51– (MCW3) and mus81– rhp51– (MCW892) strains (left panel), and wild-type (MCW1221), mus81– (MCW1502), rad22– (MCW1285) and mus81– rad22– (MCW1337) strains (right panel). (B) Survival curves showing the sensitivity of wild-type (MCW1221), mus81– (MCW1502), rhp51– (MCW1088), rad22– (MCW1285), mus81– rhp51– (MCW1235) and mus81– rad22– (MCW1337) strains to UV. (C) Survival curves showing the CPT sensitivity of wild-type (MCW42), mus81– (MCW745), rhp51– (MCW3) and mus81– rhp51– (MCW891) strains (left panel), and wild-type (MCW1221), mus81– (MCW1502), rad22– (MCW1285) and mus81– rad22– (MCW1337) strains (right panel). (D) Spot assay showing the relative growth and genotoxin sensitivity of rhp51– rhp54– (MCW1497), mus81– rhp51– (MCW1235), mus81– rhp54– (MCW1498) and mus81– rhp51– rhp54– (MCW1499) strains after 3 and 5 days growth as indicated. (E) Spot assay showing the relative sensitivities to UV, HU, MMS and CPT of MCW1335 (rad22– rhp51–), MCW1337 (rad22– mus81–), MCW1235 (rhp51– mus81) and MCW1478 (rad22– rhp51– mus81–).
We have recently reported that a mus81 rhp54 double mutant exhibits a synergistic increase in poor growth and reduced viability (27). Rhp54 is a homologue of S.cerevisiae Rad54, which is believed to promote recombination by aiding Rad51-mediated strand invasion and D-loop formation, remodelling chromatin, and dissociating Rad51-dsDNA filaments (28). We have observed that rhp51 and rhp54 mutants exhibit an epistatic relationship with respect to genotoxin sensitivity consistent with Rhp51 and Rhp54 proteins functioning together (unpublished data). It was conceivable therefore that the poor growth and viability of a mus81 rhp54 double mutant might be due to aberrant or unresolved recombination intermediates formed by Rhp51. To test this we compared a mus81 rhp51 rhp54 triple mutant with its respective double mutants for growth and viability. The triple mutant exhibited improved growth and viability compared to a mus81 rhp54 double mutant indicating that either Rhp54 or Mus81 are required to prevent Rhp51-mediated toxicity (Figure 4D, data not shown). Unexpectedly we also observed that the triple mutant grew slightly better than a mus81 rhp51 double mutant suggesting that Rhp54 can have a deleterious effect in the absence of both Mus81 and Rhp51. When we tested these strains for genotoxin sensitivity it was apparent that the deletion of rhp54 in a mus81 rhp51 mutant background also reduced sensitivity to UV, HU and MMS (Figure 4D, data not shown). In the case of UV, sensitivity is reduced almost to the level of a rhp51 rhp54 double mutant, which is the same as a rhp51 single mutant. These data show that the additive UV sensitivity of a mus81 rhp51 double mutant is mostly a consequence of Rhp54-mediated toxicity, and suggest that mus81 and rhp51 may in fact function in the same pathway for the response to UV-induced damage. However, a mus81 rhp51 rhp54 triple mutant is still synergistically more sensitive to HU, MMS and CPT than mus81 single and rhp51 rhp54 double mutants (Figure 4D, data not shown). This reaffirms our conclusion that Mus81 functions in a Rhp51-independent pathway for the repair of certain kinds of DNA damage.
To see if Mus81 functions in a Rad22-dependent Rhp51-independent pathway we compared mus81 and rad22 single and double mutants for growth and genotoxin sensitivity. Contrary to our expectation a mus81 rad22 double mutant grows much slower and is more sensitive to UV, HU, MMS and CPT than its respective single mutant strains (Figure 4A–C). A comparison of this double mutant with a mus81 rhp51 rad22 triple mutant revealed that the poor growth and enhanced sensitivity of the mus81 rad22 mutant strain is due mainly to rhp51+ (Figure 4E, data not shown). Seemingly in the absence of both Rad22 and Mus81, Rhp51 has a deleterious effect. When we compared the triple mutant to a rad22 rhp51 double mutant we observed that both strains are similarly sensitive to UV, MMS and CPT, and that the triple mutant is slightly more sensitive to HU (Figure 4E). Taken together these data indicate that Mus81 functions mainly in a Rad22-dependent pathway for the repair of damage induced by UV, HU, MMS and CPT. Moreover in most cases this pathway is independent of Rhp51.
Effect of mus81 on the frequency of direct repeat recombination
As shown in Figure 3D Rad22-dependent (Rhp51-independent) recombination generates only deletion-type recombinants. Consistent with Mus81's involvement in this pathway the formation of deletion types is suppressed in a mus81 mutant, whereas the frequency of conversion types remains unaffected (Figure 5). Although the reduction in deletion types is modest it is statistically significant (P = 0.02). Furthermore it accounts for approximately half of the deletion types that are dependent on Rad22. Deletion of mus81 also slightly reduces the elevated frequency of deletion types in a rhp55 mutant (Figure 5). Again the effect is modest but significant (P = 0.02). Finally, the residual level of deletion types in a rad22 single or rad22 rhp55 double mutant is unaffected by mus81 mutation showing that Mus81 promotes deletion type formation in a Rad22-dependent manner.
Figure 5. Effect of mus81– on the frequency and type of recombinants in different mutant backgrounds. The strains used were MCW429 (wild-type), MCW988 (mus81–), MCW431 (rhp55–), MCW992 (rhp55– mus81–), MCW1494 (rad22–), MCW1500 (rad22– mus81–), MCW1227 (rad22– rhp55–) and MCW1039 (rad22– rhp55– mus81–).
Cleavage of Rad22-dependent D-loops by Mus81 in vitro
Our genetic data indicates that Mus81 can function with Rad22 independently of Rhp51 to promote the repair of DNA damage. Human (Hs) Rad52 can promote the formation of D-loops in vitro independently of Rad51 (4). This reaction probably depends on Rad52 promoting annealing between the invading single-stranded DNA and transiently unwound duplex DNA. It has also been shown recently that both human and budding yeast Rad52 can promote DNA strand exchange (29,30). Therefore one-way in which Rad22 and Mus81 could function together to promote DNA repair without Rhp51, is if Rad22 forms D-loops that Mus81 cleaves. To test this possibility in vitro we first purified recombinant Rad22 and found that it could promote D-loop formation similar to HsRad52 (data not shown). Standard D-loop assays measure homologous pairing between ssDNA fragments and superhelical dsDNA substrates, with reaction products being resolved by agarose gel electrophoresis. In such assays the ssDNA typically shares absolute homology to a region of the dsDNA and therefore can form a D-loop with no extruding flaps. However, the repair of a collapsed replication fork would involve strand invasion from a detached arm of the replication fork (i.e. a duplex DNA with a single-stranded tail). This would generate a D-loop with a four-way DNA junction containing three duplex arms. Such D-loops are ideal substrates for Mus81 in vitro (13).
In order for Rad22 to promote the formation of the type of D-loop that Mus81 favours for cleavage, we used a radiolabelled partial duplex consisting of a 20 bp heterologous duplex region and an 80 nt 3'-ended single-stranded region with homology to X174 DNA (Figure 6A). Incubation of these two DNAs in the absence of Rad22 results in trace amounts of two retarded radiolabelled products. Both products migrate slightly behind the position of the unlabelled X174 supercoiled DNA, and the slightly faster migrating product runs approximately at the position of linear X174 DNA (Figure 6B, lane a; data not shown). When the partial duplex is pre-incubated with stoichiometric amounts of Rad22 prior to the addition of the supercoiled DNA there is >10-fold increase in the amount of these retarded species (Figure 6B, lanes b–d). Optimal formation of these species occurs at approximately a 1:1 ratio of monomeric Rad22 and nucleotides of single-stranded tail, which is the same as for D-loop formation by HsRad52 (Figure 6B and C, data not shown) (4). To see if these species have the expected properties of the type of D-loop formed by RecA or HsRad52 we first looked to see if they were dependent on homology. To do this we made a second radiolabelled partial duplex that shares homology to a region of pBR322. Pre-incubation of this partial duplex with Rad22, followed by the addition of supercoiled pBR322 DNA, results in the formation of a D-loop (Figure 6C, lanes e–h). However, if X174 is used instead of pBR322 then no D-loop is formed (Figure 6C, lanes b and c). Similarly, combining the X174-based partial duplex with pBR322 also results in no D-loop (Figure 6C, lanes i and j). These data show that the D-loops formed by Rad22 are homology-dependent. A second characteristic of a bona fide D-loop is that it can be dissociated by spontaneous branch migration. Branch migration induced either by heat treatment or through the release of superhelical tension, by cutting the X174 DNA away from the D-loop with a restriction enzyme, resulted in the dissociation of the D-loops formed by Rad22 (data not shown). Finally, similar to HsRad52, pre-incubation of Rad22 with the X174 DNA inhibited its ability to promote D-loop formation (Figure 6B, lanes e–g) (4). This suggests that a Rad22–X174 DNA complex is unable to form a D-loop.
Figure 6. Rad22-promoted D-loop formation and D-loop cleavage by Mus81. (A) Schematic of D-loop formation. The asterisk indicates the position of the 5'-32P-end-label on the partial duplex. (B) Rad22-promoted D-loop formation. Reactions are described in Materials and Methods and contained Rad22 (6 nM, lanes b and e; 12 nM, lanes c and f; and 24 nM, lanes d and g) and 10 mM MgCl2. Rad22 was pre-incubated with partial duplex or X174 DNA as indicated. (C) Dependence on homology for D-loop formation. Reactions are described in Materials and Methods. (D) Effect of Mus81 on Rad22-promoted X174-based D-loop formation. Reactions contained 12 nM Rad22 and 14 nM Mus81–Eme1 as indicated. (E) Cleavage of purified X174-based D-loops by Mus81–Eme1. Reactions contained 7 nM (lane c) or 14 nM (lanes d and e) Mus81–Eme1 and 10 mM MgCl2 as indicated. The D-loop (lane a) was dissociated by heat treatment at 96°C for 2 min.
Having established that the D-loops generated by Rad22 are genuine we looked to see whether Mus81 could cleave them. Reactions in which Mus81 was added just after the supercoiled DNA showed a Mg2+-dependent decrease in the amount of D-loop together with an increase in the smear of radiolabelled material migrating between the D-loop and the free partial duplex (Figure 6D, compare lanes c–e). In these reactions Mus81 preferentially affects the more retarded D-loop band. These results are consistent with cleavage of the D-loop by Mus81, which would release the superhelical tension in the X174 DNA thereby enabling the D-loop to dissociate by branch migration.
To provide further evidence of Mus81's ability to target D-loops, we purified the D-loops from a Rad22 reaction and incubated them with Mus81 (Figure 6E). In the presence of Mg2+ Mus81 dissociated the purified D-loops, again preferentially targeting the more retarded D-loop band (Figure 6E, lanes c and d). A comparison with D-loops dissociated by heat treatment (lane a), shows that Mus81 dissociates much of the D-loop back to free partial duplex. Analysis of these reaction products by polyacrylamide gel electrophoresis revealed very little cleavage of the partial duplex DNA (data not shown). Incubation of Mus81 with supercoiled X174 DNA also revealed relatively little non-specific nicking activity (data not shown). Furthermore, incubation of Mus81 with synthetic D-loops under the conditions used here, confirms that Mus81 cleaves predominantly at sites just 5' of the four-way DNA junction in the strand that is complementary to the invading 3' single-stranded tail (Supplementary Material). These data indicate that Mus81 is causing the dissociation of the D-loop by specifically nicking the X174 DNA at the D-loop.
DISCUSSION
Epistasis analysis of the S.pombe Rad52 homologue Rad22 originally suggested that, unlike in S.cerevisiae, Rad22 and Rhp51 worked independently of each other (20). This conclusion was questioned by data showing that Rad22 and Rhp51 physically interact (31). A further characterization of rad22 mutant phenotypes, using a new rad22 deletion mutant, also suggested that earlier conclusions were misguided by the use of a presumed hypomorphic allele of rad22 (22). However, these studies still did not observe the same kind of dramatic effect on recombination that mutation of RAD52 has in S.cerevisiae (21,22). It was therefore concluded that Rad22 played a less critical role in S.pombe because of functional redundancy with another Rad52-like protein called Rti1 (2,31). We have shown here that this is not the case. Deletion of rad22 has a profound effect on survival following exposure to a range of different genotoxins, which, depending on the genotoxin, is either equivalent or greater than that of a rhp51 mutant. Direct repeat recombination is also dramatically reduced in a rad22 mutant, including a total ablation of conversion type recombinants. Finally, these rad22 mutant phenotypes show epistasis with those of a rhp51 mutant indicating that recombination is driven by Rad22-dependent mechanisms that are both dependent and independent of Rhp51. This is just like the relationship between RAD52 and RAD51 in S.cerevisiae. The reason that the earlier studies of rad22 were misguided appears to be due to the presence of suppressor mutations in the strains that were used. The identity and characterisation of the suppressors will be presented elsewhere.
Direct repeat recombination in S.cerevisiae and S.pombe
From our analyses it appears that the pathways of spontaneous direct repeat recombination are essentially the same in S.pombe as those that have been described in S.cerevisiae (32). In both yeast there are RAD52/rad22-dependent and -independent pathways. Although the exact mechanisms of recombination are unknown, it has been proposed that the major mechanism of RAD52-independent recombination is single-strand annealing (SSA) (32,33). In contrast RAD52-dependent recombination may be a mixture of SSA, RAD51-dependent gene conversion/reciprocal exchange, and BIR (32,33). In the absence of RAD51/rhp51 the frequency of gene conversion/reciprocal exchange is reduced or abolished, but the overall frequency of recombination increases (32,34). Like in S.cerevisiae this hyper-recombination in S.pombe is dependent on rad22, and may stem from un-repaired lesions (e.g. single-strand gaps) being converted into DSBs that are then repaired by SSA or BIR.
Mus81 functions in a recombinational repair pathway that is independent of Rhp51
We have shown that Mus81 generally functions in a Rad22-dependent pathway for DNA repair. Furthermore, for damage induced by HU, MMS and CPT it can promote repair independently of Rhp51. These genotoxins can variously cause stalling, blockage and breakage of the replication fork. It is difficult therefore to ascertain the exact type of lesion that is repaired by the Rad22–Mus81-dependent pathway. The exquisite hypersensitivity of mus81 mutant strains to CPT, however, provides perhaps the best clue as to what Mus81 may be doing in mitotic cells. CPT traps Top1 cleavage complexes thereby generating single-strand breaks at which replication forks collapse by ‘replication run-off’ (35). Cleavage complexes may also block DNA replication if present in the lagging strand template (36). Therefore Mus81 would appear to play an important role in the repair of blocked and/or broken replication forks (8). Suppression of this hypersensitivity by RusA indicates that Mus81's prime role here is in the processing of reversed replication forks and/or recombination intermediates (8). From our epistasis analysis we know that there are Rhp51-dependent and independent pathways for repairing CPT-induced damage. Furthermore, we know that Mus81 functions in the Rhp51-independent pathway, and based on our observation that mus81 rhp51 and rad22 rhp51 double mutants exhibit similar sensitivities to CPT, it seems that this pathway is largely dependent on Mus81. As mus81 and rhp51 single mutants exhibit similar sensitivities to CPT, we suspect that the Rhp51-independent pathway deals with approximately half of the replication fork problems induced by CPT.
We have found that CPT induces mostly deletion recombinants in the direct repeat substrate used in our studies (unpublished data). This is consistent with data showing that fork collapse gives rise to deletions between direct repeats, which occurs by strand invasion from the broken sister chromatid into the ‘wrong’ repeat (37). Interestingly approximately 50% of spontaneous Rad22-dependent deletion type recombinants are dependent on Mus81. We suspect that, like CPT-induced deletion types, these spontaneous events are due to replication fork collapse. The observation that half of these deletion types depend on Mus81, correlates well with our supposition that the Rhp51-independent pathway deals with approximately half of the replication fork problems induced by CPT. The hyper-levels of Rad22-dependent deletion types in a rhp55 mutant (equivalent to a rhp51 mutant), which are largely independent of Mus81, probably arise by SSA.
We have shown that Rad22 can promote the strand invasion of a double-stranded DNA molecule by a 3' single-stranded tailed duplex. Therefore, the Rhp51-independent pathway for dealing with CPT-induced damage could involve the repair of one-ended DSBs by Rad22-promoted strand invasion. Mus81 could be required to generate these one-ended DSBs by cleaving stalled replication forks. However, it is our contention that its primary role is to resolve the D-loops that are formed by Rad22 (Figure 1A). Certainly our in vitro data shows that Mus81 has this ability. The idea that a Rhp51-independent mechanism could be a prominent way of repairing collapsed replication forks is consistent with data showing that the accumulation of HJs in the rDNA array in S.cerevisiae depends on RAD52 and not RAD51 (38).
In S.cerevisiae Rad52 can repair two-ended DSBs by BIR independently of Rad51. BIR involves strand invasion from one of the broken ends into an intact sister chromatid or homologue (3). This primes DNA synthesis allowing chromosomal regions on the other side of the break to be recovered by DNA replication. However, the RAD51-independent pathway of BIR is relatively inefficient, and a recent study has shown that >95% of BIR events involving a single-end invasion depend on RAD51 (39). The inefficiency of the RAD51-independent pathway is probably due to the dependence of Rad52-promoted strand invasion on the donor duplex being transiently unwound (39). This study questions our idea that a Rad22–Mus81-dependent (Rhp51-independent) pathway could be a significant mechanism for one-ended DSB repair in S.pombe. However, studies of BIR in S.cerevisiae measure recombination between one broken DNA end and an intact sister chromatid or homologous chromosome. In contrast, replication fork collapse, by ‘replication run-off’ or by Mus81 cleavage of a reversed fork, generates one broken DNA end and a sister chromatid with a single-stranded gap . We think that this gap may provide access for a DNA helicase, which could generate the necessary unwound donor duplex for efficient Rad22-dependent strand invasion/annealing. We are currently investigating the validity of this assertion.
SUPPLEMENTARY MATERIAL
Supplementary Material is available at NAR Online.
ACKNOWLEDGEMENTS
This work was supported by a project grant (065278/Z/01/Z) and a Senior Research Fellowship from the Wellcome Trust awarded to M.C.W.
REFERENCES
McGlynn,P. and Lloyd,R.G. ( (2002) ) Recombinational repair and restart of damaged replication forks. Nature Rev. Mol. Cell. Biol., , 3, , 859–870.
Symington,L.S. ( (2002) ) Role of RAD52 epistasis group genes in homologous recombination and double-strand break repair. Microbiol. Mol. Biol. Rev., , 66, , 630–670.
Paques,F. and Haber,J.E. ( (1999) ) Multiple pathways of recombination induced by double-strand breaks in Saccharomyces cerevisiae. Microbiol. Mol. Biol. Rev., , 63, , 349–404.
Kagawa,W., Kurumizaka,H., Ikawa,S., Yokoyama,S. and Shibata,T. ( (2001) ) Homologous pairing promoted by the human Rad52 protein. J. Biol. Chem., , 276, , 35201–35208.
Odagiri,N., Seki,M., Onoda,F., Yoshimura,A., Watanabe,S., Enomoto,T., De Los Santos,T., Hunter,N., Lee,C., Larkin,B. et al. ( (2003) ) Budding yeast mms4 is epistatic with rad52 and the function of Mms4 can be replaced by a bacterial Holliday junction resolvase. DNA Repair (Amst.), , 2, , 347–358.
Bastin-Shanower,S.A., Fricke,W.M., Mullen,J.R. and Brill,S.J. ( (2003) ) The mechanism of Mus81–Mms4 cleavage site selection distinguishes it from the homologous endonuclease Rad1–Rad10. Mol. Cell. Biol., , 23, , 3487–3496.
Boddy,M.N., Gaillard,P.-H.L., McDonald,W.H., Shanahan,P., Yates,J.R.,III and Russell,P. ( (2001) ) Mus81–Eme1 are essential components of a Holliday junction resolvase. Cell, , 107, , 537–548.
Doe,C.L., Ahn,J.S., Dixon,J. and Whitby,M.C. ( (2002) ) Mus81–Eme1 and Rqh1 involvement in processing stalled and collapsed replication forks. J. Biol. Chem., , 277, , 32753–32759.
Chen,X.B., Melchionna,R., Denis,C.M., Gaillard,P.H., Blasina,A., Van de Weyer,I., Boddy,M.N., Russell,P., Vialard,J. and McGowan,C.H. ( (2001) ) Human Mus81-associated endonuclease cleaves Holliday junctions in vitro. Mol. Cell, , 8, , 1117–1127.
Whitby,M.C., Osman,F. and Dixon,J. ( (2003) ) Cleavage of model replication forks by fission yeast Mus81–Eme1 and budding yeast Mus81–Mms4. J. Biol. Chem., , 278, , 6928–6935.
Ciccia,A., Constantinou,A. and West,S.C. ( (2003) ) Identification and characterization of the human Mus81–Eme1 endonuclease. J. Biol. Chem., , 278, , 25172–25178.
Kaliraman,V., Mullen,J.R., Fricke,W.M., Bastin-Shanower,S.A. and Brill,S.J. ( (2001) ) Functional overlap between Sgs1-Top3 and the Mms4-Mus81 endonuclease. Genes Dev., , 15, , 2730–2740.
Osman,F., Dixon,J., Doe,C.L. and Whitby,M.C. ( (2003) ) Generating crossovers by resolution of nicked Holliday junctions: a role for Mus81–Eme1 in meiosis. Mol. Cell, , 12, , 761–774.
Gaillard,P.H., Noguchi,E., Shanahan,P. and Russell,P. ( (2003) ) The endogenous Mus81–Eme1 complex resolves Holliday junctions by a nick and counternick mechanism. Mol. Cell, , 12, , 747–759.
Moreno,S., Klar,A. and Nurse,P. ( (1991) ) Molecular genetic analysis of fission yeast Schizosaccharomyces pombe. Methods Enzymol., , 194, , 795–823.
Gutz,H., Heslot,H., Leopold,U. and Loprieno,N. ( (1974) ) Schizosaccharomyces pombe. In King,R.C. (ed.), Bacteria, Bacteriophages and Fungi, Plenum Press, New York, NY, Vol. 1. pp. 395–446.
Bahler,J., Wu,J.Q., Longtine,M.S., Shah,N.G., McKenzie,A.,III, Steever,A.B., Wach,A., Philippsen,P. and Pringle,J.R. ( (1998) ) Heterologous modules for efficient and versatile PCR-based gene targeting in Schizosaccharomyces pombe. Yeast, , 14, , 943–951.
Osman,F., Adriance,M. and McCready,S. ( (2000) ) The genetic control of spontaneous and UV-induced mitotic intrachromosomal recombination in the fission yeast Schizosaccharomyces pombe. Curr. Genet., , 38, , 113–125.
Whitby,M.C. and Dixon,J. ( (1998) ) Substrate specificity of the SpCCE1 holliday junction resolvase of Schizosaccharomyces pombe. J. Biol. Chem., , 273, , 35063–35073.
Khasanov,F.K., Savchenko,G.V., Bashkirova,E.V., Korolev,V.G., Heyer,W.D. and Bashkirov,V.I. ( (1999) ) A new recombinational DNA repair gene from Schizosaccharomyces pombe with homology to Escherichia coli RecA. Genetics, , 152, , 1557–1572.
van den Bosch,M., Zonneveld,J.B., Vreeken,K., de Vries,F.A., Lohman,P.H. and Pastink,A. ( (2002) ) Differential expression and requirements for Schizosaccharomyces pombe RAD52 homologs in DNA repair and recombination. Nucleic Acids Res., , 30, , 1316–1324.
van den Bosch,M., Vreeken,K., Zonneveld,J.B., Brandsma,J.A., Lombaerts,M., Murray,J.M., Lohman,P.H. and Pastink,A. ( (2001) ) Characterization of RAD52 homologs in the fission yeast Schizosaccharomyces pombe. Mutat. Res., , 461, , 311–323.
Suto,K., Nagata,A., Murakami,H. and Okayama,H. ( (1999) ) A double-strand break repair component is essential for S phase completion in fission yeast cell cycling. Mol. Biol. Cell, , 10, , 3331–3343.
Doe,C.L., Dixon,J., Osman,F. and Whitby,M.C. ( (2000) ) Partial suppression of the fission yeast rqh1(–) phenotype by expression of a bacterial Holliday junction resolvase. EMBO J., , 19, , 2751–2762.
Interthal,H. and Heyer,W.D. ( (2000) ) MUS81 encodes a novel helix–hairpin–helix protein involved in the response to UV- and methylation-induced DNA damage in Saccharomyces cerevisiae. Mol. Gen. Genet., , 263, , 812–827.
Boddy,M.N., Lopez-Girona,A., Shanahan,P., Interthal,H., Heyer,W.D. and Russell,P. ( (2000) ) Damage tolerance protein Mus81 associates with the FHA1 domain of checkpoint kinase Cds1. Mol. Cell. Biol., , 20, , 8758–8766.
Doe,C.L. and Whitby,M.C. ( (2004) ) The involvement of Srs2 in post-replication repair and homologous recombination in fission yeast. Nucleic Acids Res., , 32, , 1480–1491.
Tan,T.L., Kanaar,R., Wyman,C., Hsiang,Y.H., Lihou,M.G. and Liu,L.F. ( (2003) ) Rad54, a Jack of all trades in homologous recombination. DNA Repair (Amst.), , 2, , 787–794.
Bi,B., Rybalchenko,N., Golub,E.I. and Radding,C.M. ( (2004) ) Human and yeast Rad52 proteins promote DNA strand exchange. Proc. Natl Acad. Sci. USA, , 101, , 9568–9572.
Kumar,J.K. and Gupta,R.C. ( (2004) ) Strand exchange activity of human recombination protein Rad52. Proc. Natl Acad. Sci. USA, , 101, , 9562–9567.
Tsutsui,Y., Khasanov,F.K., Shinagawa,H., Iwasaki,H. and Bashkirov,V.I. ( (2001) ) Multiple interactions among the components of the recombinational DNA repair system in Schizosaccharomyces pombe. Genetics, , 159, , 91–105.
Klein,H.L. ( (1995) ) Genetic control of intrachromosomal recombination. Bioessays, , 17, , 147–159.
Prado,F. and Aguilera,A. ( (1995) ) Role of reciprocal exchange, one-ended invasion crossover and single-strand annealing on inverted and direct repeat recombination in yeast: different requirements for the RAD1, RAD10, and RAD52 genes. Genetics, , 139, , 109–123.
Aguilera,A. ( (1995) ) Genetic evidence for different RAD52-dependent intrachromosomal recombination pathways in Saccharomyces cerevisiae. Curr. Genet., , 27, , 298–305.
Pommier,Y., Redon,C., Rao,V.A., Seiler,J.A., Sordet,O., Takemura,H., Antony,S., Meng,L., Liao,Z., Kohlhagen,G. et al. ( (2003) ) Repair of and checkpoint response to topoisomerase I-mediated DNA damage. Mutat. Res., , 532, , 173–203.
Vance,J.R. and Wilson,T.E. ( (2002) ) Yeast Tdp1 and Rad1–Rad10 function as redundant pathways for repairing Top1 replicative damage. Proc. Natl Acad. Sci. USA, , 99, , 13669–13674.
Bierne,H., Ehrlich,S.D. and Michel,B. ( (1997) ) Deletions at stalled replication forks occur by two different pathways. EMBO J., , 16, , 3332–3340.
Zou,H. and Rothstein,R. ( (1997) ) Holliday junctions accumulate in replication mutants via a RecA homolog-independent mechanism. Cell, , 90, , 87–96.
Davis,A.P. and Symington,L.S. ( (2004) ) RAD51-dependent break-induced replication in yeast. Mol. Cell. Biol., , 24, , 2344–2351.
Ostermann,K., Lorentz,A. and Schmidt,H. ( (1993) ) The fission yeast rad22 gene, having a function in mating-type switching and repair of DNA damages, encodes a protein homolog to Rad52 of Saccharomyces cerevisiae. Nucleic Acids Res., , 21, , 5940–5944.
Wang,S.W., Goodwin,A., Hickson,I.D. and Norbury,C.J. ( (2001) ) Involvement of Schizosaccharomyces pombe Srs2 in cellular responses to DNA damage. Nucleic Acids Res., , 29, , 2963–2972.(Claudette L. Doe, Fekret Osman, Julie Di)
* To whom correspondence should be addressed. Tel: +44 1865 275192; Fax: +44 1865 275297; Email: matthew.whitby@bioch.ox.ac.uk
The authors wish it to be known that, in their opinion, the first two authors should be regarded as joint First Authors
ABSTRACT
In budding yeast most Rad51-dependent and -independent recombination depends on Rad52. In contrast, its homologue in fission yeast, Rad22, was assumed to play a less critical role possibly due to functional redundancy with another Rad52-like protein Rti1. We show here that this is not the case. Rad22 like Rad52 plays a central role in recombination being required for both Rhp51-dependent and -independent events. Having established this we proceed to investigate the involvement of the Mus81–Eme1 endonuclease in these pathways. Mus81 plays a relatively minor role in the Rhp51-dependent repair of DNA damage induced by ultraviolet light. In contrast Mus81 has a key role in the Rad22-dependent (Rhp51-independent) repair of damage induced by camptothecin, hydroxyurea and methyl-methanesulfonate. Furthermore, spontaneous intrachromosomal recombination that gives rise to deletion recombinants is impaired in a mus81 mutant. From these data we propose that a Rad22–Mus81-dependent (Rhp51-independent) pathway is an important mechanism for the repair of DNA damage in fission yeast. Consistent with this we show that in vitro Rad22 can promote strand invasion to form a D-loop that can be cleaved by Mus81.
INTRODUCTION
Problems associated with DNA damage can be compounded during S-phase. Single-strand breaks can act as run-off sites for polymerases resulting in replication fork collapse and the formation of one-ended double-strand breaks (DSBs). Bulky DNA lesions can block fork progress, and may in some instances result in replisome disassembly and the regression of the fork where the nascent strands are unwound and anneal to each other. Alternatively, the replisome may skip damaged sections of the genome leaving behind lesion-containing single-strand gaps. Problems such as these can be dealt with by homologous recombination, which promotes the repair of a break or a lesion-containing gap, enables template switching for polymerases to bypass DNA lesions, and helps restart replication at non-origin sites .
In Saccharomyces cerevisiae, recombination depends on the RAD52 epistasis group of genes . Typically a 3' tail, exposed by the resection of a DSB or unwinding of a single-strand gap, is bound by Rad51, which then catalyses strand invasion of a homologous DNA molecule to form a D-loop. The loading of Rad51 on to the DNA and catalysis of strand invasion are variously aided by proteins such as Rad52, Rad54, Rad55 and Rad57. In some situations Rad52 can promote the formation of a D-loop without Rad51 through its ability to anneal complementary ssDNAs (4). D-loops can act as sites for replisome reassembly, and therefore by promoting strand invasion recombination can act to restart replication at stalled or broken replication forks (1).
Various mechanisms have been proposed for RDR and most involve the formation and subsequent processing of one or two Holliday junctions (HJs) (1). One enzyme that has been implicated in resolving HJs in eukaryotes is Mus81–Eme1 (Mus81–Mms4 in S.cerevisiae). Mus81 is a member of the XPF family of endonucleases, and together with Eme1/Mms4 has been shown to cleave synthetic HJs in vitro. Evidence that it cleaves HJs in vivo has come from genetic studies which have shown that mus81 and eme1/mms4 mutant phenotypes are suppressed by the bacterial HJ resolvase RusA (5–9). However, the idea that Mus81–Eme1/Mms4 is a HJ resolvase has been questioned by in vitro data showing that recombinant enzyme, purified from Escherichia coli, exhibits only weak HJ cleavage activity, whilst cleaving fork structures and 3' flaps relatively well (8,10–12). In the case of Mus81–Mms4 this has lead to proposals that its prime function is to remove 3' flaps that are formed during the repair of DSBs and single-strand gaps by synthesis-dependent strand annealing (SDSA) (6). However, these models do not provide a satisfactory explanation why RusA suppresses mus81 and eme1/mms4 mutant phenotypes.
It has been shown recently that recombinant Mus81–Eme1 can resolve four-way DNA junctions that contain a single-strand nick or gap at or near the junction centre at least as well as a 3' flap (10,13,14). Such four-way junctions are formed during the regression of a replication fork and prior to the formation of a HJ at a D-loop. Based on these data we have proposed that Mus81–Eme1 may cleave regressed replication forks to promote RDR, and resolve the four-way DNA junction at a D-loop during RDR (Figure 1). In these models RusA is able to suppress mus81 mutant phenotypes not because it provides a direct substitute for Mus81–Eme1 but because it is able to resolve the HJs that will form in its absence.
Figure 1. Hypothetical model showing potential roles that Mus81–Eme1 might have in processing stalled and broken replication forks. Nascent strands are shown as red lines with arrowheads at 3' ends. The replication fork block is indicated by a solid circle, and the position of Mus81–Eme1 cleavage sites by the labelled arrows.
Here we have investigated the relationship of Mus81–Eme1 (for simplicity referred to as Mus81 throughout this paper) to recombinational pathways of DNA repair in Schizosaccharomyces pombe. Existing data suggest that the pathways of recombination in S.pombe are similar to those in S.cerevisiae, with the exception that Rad22 (the S.pombe Rad52 homologue) is not required for all recombination possibly due to functional redundancy with another Rad52-like protein Rti1 (2). However, we show here that this is not the case and that Rad22 is in fact required for most recombination just like Rad52 in S.cerevisiae. The original confusion in the literature appears to stem from the ease with which rad22 strains acquire suppressor mutations. Having established that there are Rhp51-dependent (the S.pombe Rad51 homologue) and -independent pathways of recombination, which are both dependent on Rad22, we show that Mus81 functions predominantly in the Rhp51-independent pathway. We also show that Mus81 can cleave D-loops that are formed by Rad22 in vitro. These data are consistent with our proposal that Mus81 resolves D-loops formed by Rad22 (without Rhp51) during the repair of broken replication forks.
MATERIALS AND METHODS
General techniques and strains
Procedures for S.pombe genetics were as described by Moreno et al. (15). Media for S.pombe have been described (16). The complete media was yeast extract medium (YES) and the minimal medium was Edinburgh minimal medium (EMM) each supplemented with appropriate amino acids. Hydroxyurea (HU), methyl-methanesulfonate (MMS) and camptothecin (CPT) were added to media as indicated.
Strains
The S.pombe strains used in this study are listed in Table 1. The rhp51arg3 and rad22ura4 mutants both contain a replacement of the entire open reading frame by the indicated marker, and were made by PCR-based gene targeting as described by B?hler et al. (17). Genuine deletion mutants were identified by genomic Southern blot analysis.
Table 1. S.pombe strains
Spot assays and quantitative survival curves
Cell cultures growing exponentially in liquid media were adjusted to a density of 1 x 107 cells/ml, serially diluted, and 10 μl aliquots of each dilution spotted onto media containing genotoxins as indicated. For ultraviolet (UV) exposure plates were irradiated using a Stratalinker (Stratagene) after spotting. Plates were incubated for 2–5 days at 30°C before being photographed. All spot tests were repeated at least once as an independent experiment to ensure reproducibility. For survival curves all data points represent the mean from at least three independent experiments.
Recombination assay
Mitotic recombination was assayed by the recovery of Ade+ recombinants from strains containing the intrachromosomal recombination substrate shown in Figure 3C. Spontaneous recombinant frequencies were measured as described by Osman et al. (18). Recombinant frequencies are the mean from at least three independent assays where five independent colonies were tested in each assay. Two sample t-tests were used to determine the statistical significance of differences in recombination frequencies.
Figure 3. Rhp51-dependent and -independent recombination and DNA repair. (A) Spot assay comparing the sensitivity of wild-type (MCW1221), rhp51– (MCW1088), rad22– (MCW1285), and rad22– rhp51– (MCW1335) strains to UV, HU, MMS and CPT. (B) Survival curves showing the UV and -ray sensitivity of the same strains as in ‘A’. (C) Schematic of the recombination substrate and recombinant products. Solid and open circles represent the ade6-L469 and ade6-M375 mutations, respectively. (D) Comparison of the spontaneous recombinant frequencies of wild-type (MCW429), rhp51– (MCW1224), rhp55– (MCW431), rhp54– (MCW1225), rad22– (MCW1494), rhp51– rad22– (MCW1038), rhp55– rad22– (MCW1227), rhp51– rhp55– (MCW1228) and rhp54– rhp55– (MCW1229) strains. Error bars represent standard deviations about the mean.
Proteins
Recombinant S.pombe Mus81–Eme1 and S.cerevisiae Mus81–Mms4 were expressed in E.coli and purified as described previously (8). The overexpression and purification of recombinant Rad22 is described below. Concentrations of Mus81–Eme1/Mms4 and Rad22 are in moles of heterodimer and monomer, respectively.
Overexpression and purification of recombinant Rad22
The rad22 gene was amplified by PCR from genomic DNA with NdeI- and BamHI-flanking restriction sites to enable cloning in-frame with the N-terminal hexahistidine tag in pET14b (Novagen). The resultant plasmid pMW601 expresses His-tagged Rad22 from a T7 phage ?10 promoter. E.coli BL21-RIL cells (Stratagene) containing pMW601 were grown with aeration at 25°C in a Luria–Bertani broth containing 125 mg/ml carbenicillin and 50 mg/ml chloramphenicol. At a cell density corresponding to an A600 of 0.5, His-tagged Rad22 was induced by adding isopropyl-1-thio-?-D-galactopyranoside to a final concentration of 1 mM, following which the cells were incubated for a further 2.5 h. The cells were then harvested by centrifugation, resuspended in Buffer H (50 mM potassium phosphate, pH 8.0, 10 mM ?-mercaptoethanol, 0.3 M NaCl, 10% glycerol) with protease inhibitors and 1% Triton X-100 and lysed by passage through a French pressure cell at 30 000 p.s.i. After centrifugation at 43 700 g for 50 min, the supernatant was loaded directly onto a 1 ml nickel-nitrilotriacetic acid Superflow column (Qiagen) that was washed with 60 ml of Buffer H and 25 mM imidazole before eluting bound Rad22 with 3 ml of Buffer H and 200 mM imidazole. This sample was then diluted with 6 ml of Buffer A (50 mM Tris–HCl, pH 8.0, 1 mM EDTA, 1 mM dithiothreitol, 10% glycerol) and loaded directly onto a 1 ml HiTrap heparin column (Amersham Pharmacia Biotech). The column was washed with 10 ml of Buffer A and 0.1 M NaCl before eluting bound Rad22 with an 18 ml gradient from 0.2 to 1.0 M NaCl. Rad22 eluting between 550 and 650 mM NaCl was pooled and dialyzed against Buffer A with 0.1 M NaCl before storing as aliquots at –80°C.
DNA substrates
X174 form I DNA was from New England BioLabs. The X174-based partial duplex was made from oligonucleotides oMW585 (5'-TGCCGAATTCTACCAGTGCACAAAGTAAGAGCTTCTCGAGCTGCGCAAGGATAGGTCGAATTTTCTCATTTTCCGCCAGCAGTCCACTTCGATTTAATTC-3') and oMW586 (5'-TGCACTGGTAGAATTCGGCA-3'), and the pBR322-based partial duplex from oMW592 (5'-TGCCGAATTCTACCAGTGCACGCCTCCATCCAGTCTATTAATTGTTGCCGGGAAGCTAGAGTAAGTAGTTCGCCAGTTAATAGTTTGCGCAACGTTGTTG-3') and oMW586. The procedures for partial duplex preparation have been described previously (19). Each substrate was labelled with 32P at the 5' end of one of its component oligonucleotides. For both partial duplexes the longer oligonucleotide was labelled. The concentration of partial duplexes was estimated by relating the specific activity of the labelled oligonucleotide to the activity of the purified substrate and is expressed in molar concentrations of DNA substrate. To purify D-loops from Rad22 and un-reacted partial duplex, deproteinized reaction mixtures were applied to 3.5 ml sepharose CL-2B columns, which were developed with 20 mM Tris–HCl (pH 8.0) and 0.5 mM MgCl2.
D-loop and nuclease assays
D-loop assays (25 μl) consisted of 1.2 nM X174 or pBR322 form I DNA and 0.2 nM of 32P-labelled partial duplex in reaction buffer (25 mM Tris–HCl, pH 8.0, 1 mM dithiothreitol, 100 μg/ml BSA, 6% glycerol) with 10 mM MgCl2, Rad22 and Mus81–Eme1 as indicated. Rad22 was typically pre-incubated with the partial duplex for 5 min at 30°C prior to the addition of the form I DNA and Mus81–Eme1 as indicated. Incubation was then continued for 90 min before stopping the reaction with 5 μl of stop mix (2.5% SDS, 200 mM EDTA, 10 mg/ml proteinase K) followed by a further 15 min at 30°C. Reactions were analysed by electrophoresis through 1% agarose gels in 1x TBE buffer at 3.75 V/cm with phosphorus imaging on a Fuji FLA3000. For the cleavage of purified D-loops by Mus81, reaction mixtures (25 μl) contained 0.01 nM D-loop and 0.05 nM of non-D-loop X174 DNA in reaction buffer with 10 mM MgCl2 and Mus81–Eme1 as indicated. Reactions were incubated at 30°C for 90 min before stopping and analysing as described above.
RESULTS
Rad22 plays a pivotal role in recombinational repair in S.pombe
In S.cerevisiae RAD52 is required for nearly all homologous recombination (2). However, in S.pombe genetic analyses suggest that Rhp51 can function without Rad22 (20,21). These data have been explained by the presence of another Rad52-like protein in S.pombe called Rti1 (or Rad22B), which could substitute in the absence of Rad22 (21–23).
To clarify the genetic interactions between rad22 and the rhp51 epistasis group we first constructed a new rad22 deletion strain by replacement of the entire rad22 open reading frame with an ura4+ marker. This strain grew much slower and was much more hypersensitive to UV light, HU, MMS and CPT than previously published rad22 mutant strains (Figure 2). These more extreme phenotypes were solely due to the deletion of rad22 because they could be fully complemented by a cloned copy of the gene expressed from a plasmid (data not shown). The discrepancy in phenotypes between these different rad22 mutant strains is explained by the presence of suppressor mutations, which will be analysed elsewhere.
Figure 2. Spot assay comparing the genotoxin sensitivity of different rad22 mutant strains. The strains used were MCW1221 (wild-type), MCW1285 (new rad22–), MCW1222 , MCW4 and MCW1501 . Cultures were serially diluted and spotted onto YES agar as described in Materials and Methods. The neat spot ‘1’ represents 105 cells plated.
Having obtained a non-suppressed rad22 mutant we could unambiguously establish rad22's relationship to the rhp51 epistasis group. A rad22 rhp51 double mutant strain was constructed and compared to its respective single mutant strains for sensitivity to different DNA damaging agents. The rad22 single mutant is more sensitive to UV, HU, MMS and CPT than a rhp51 single mutant but is similarly sensitive to gamma irradiation (Figure 3A and B). The double mutant exhibits slightly more sensitivity to most of these DNA damaging agents than the rad22 single mutant (Figure 3A and B). This enhanced sensitivity is particularly noticeable at high UV doses where the sensitivity of the rad22 strain appears to plateau off (Figure 3B). The majority of the rad22 mutant survivors at high UV doses had acquired a suppressor mutation, and, as suppression depends on rhp51+ (our unpublished data), this explains why the rad22 single mutant appears to be slightly less sensitive to DNA damage than the rad22 rhp51 double mutant. When taken together these data show that rad22 and rhp51 are epistatic for the repair/tolerance of DNA damage. For DSBs caused by gamma irradiation it would seem that Rhp51 and Rad22 mainly work together to promote repair since both rhp51 and rad22 single mutants show approximately the same sensitivity as a rhp51 rad22 double mutant. However, the greater sensitivity of a rad22 mutant compared to a rhp51 mutant to agents such as UV, HU, MMS and CPT, which variously cause replication fork stalling, blockage and breakage, suggests that the recombinational repair of damaged replication forks can be undertaken by Rad22 both with and without Rhp51.
Pathways of direct repeat recombination in S.pombe
To investigate the relationship between rhp51 and rad22 with respect to recombination we used strains containing a non-tandem direct repeat of ade6– heteroalleles to measure Ade+ recombinant frequencies (Figure 3C). Such recombinants arise from inter- or intra-chromatid recombination by a number of different pathways, which in part can be categorised using a his3+ gene placed between the repeats that allows Ade+ recombinants that are His– (deletion types) to be distinguished from those that are His+ (conversion types) (24). In wild-type strains 3.5 Ade+ recombinants are generated in every 104 viable cells of which 30% are conversion types and 70% are deletion types (Figure 3D). Mutants from the rhp51 epistasis group, including rhp51, rhp54 and rhp55 single mutants and rhp51 rhp55 and rhp54 rhp55 double mutants, are devoid of conversion type recombinants but exhibit a 4- to 5-fold increase in deletion type recombinants compared to wild-type (Figure 3D). The rad22 mutant is also devoid of conversion types, but unlike the rhp51, rhp54 and rhp55 mutants it exhibits a >3-fold reduction in recombinant frequency (<1 x 10–4) (Figure 3D). The same low level of recombinants and absence of conversion types is seen in both rhp51 rad22 and rhp55 rad22 double mutants. These data show that two Rad22-dependent pathways are responsible for the majority of recombination between repeated DNA sequences in S.pombe. One of these pathways involves the Rhp51 epistasis group of proteins, and is required for generating conversion type recombinants, whilst the other is independent of Rhp51 and only generates deletion type recombinants.
Mus81 functions in Rad22-dependent pathways of DNA repair
In S.cerevisiae Mus81/Mms4 functions in the Rad52-dependent pathway for the repair/tolerance of UV- and MMS-induced damage (5). Other genetic studies have shown that for UV-induced damage Mus81/Mms4 functions exclusively in the Rad54-dependent sub-pathway of Rad52-dependent repair (25). However, for MMS it appears to work in Rad52-dependent pathways that are both dependent and independent of Rad54 (25). Similar to S.cerevisiae, Mus81 appears to function exclusively in a Rhp51-dependent pathway for the response to UV-induced damage in S.pombe (26).
To more thoroughly establish the recombination repair pathways in which Mus81 functions in S.pombe, we compared the sensitivity of rhp51, rad22 and mus81 single, double and triple mutants to UV, HU, MMS and CPT. The mus81 rhp51 double mutant shows an increase in sensitivity to each of these agents compared to its respective single mutant strains (Figure 4A and B). In the case of HU, MMS and CPT the increase is synergistic, whereas for UV it is approximately additive. Since our data for UV sensitivity contradicted a previous study (26), we sought to confirm our data by constructing several independent mus81– rhp51– double mutant strains, made from two different rhp51 deletion strains. Each of these double mutant strains exhibits a similar increase in sensitivity to UV, HU, MMS and CPT as shown in Figure 4A–C (data not shown). These data indicate that Mus81 functions in a separate but overlapping pathway to Rhp51 for the repair of at least some types of DNA damage.
Figure 4. Mus81 functions in both Rhp51-dependent and -independent repair pathways depending on the type of DNA damage. (A) Spot assay showing the relative sensitivities to UV, HU, MMS and CPT of wild-type (MCW45), mus81– (MCW745), rhp51– (MCW3) and mus81– rhp51– (MCW892) strains (left panel), and wild-type (MCW1221), mus81– (MCW1502), rad22– (MCW1285) and mus81– rad22– (MCW1337) strains (right panel). (B) Survival curves showing the sensitivity of wild-type (MCW1221), mus81– (MCW1502), rhp51– (MCW1088), rad22– (MCW1285), mus81– rhp51– (MCW1235) and mus81– rad22– (MCW1337) strains to UV. (C) Survival curves showing the CPT sensitivity of wild-type (MCW42), mus81– (MCW745), rhp51– (MCW3) and mus81– rhp51– (MCW891) strains (left panel), and wild-type (MCW1221), mus81– (MCW1502), rad22– (MCW1285) and mus81– rad22– (MCW1337) strains (right panel). (D) Spot assay showing the relative growth and genotoxin sensitivity of rhp51– rhp54– (MCW1497), mus81– rhp51– (MCW1235), mus81– rhp54– (MCW1498) and mus81– rhp51– rhp54– (MCW1499) strains after 3 and 5 days growth as indicated. (E) Spot assay showing the relative sensitivities to UV, HU, MMS and CPT of MCW1335 (rad22– rhp51–), MCW1337 (rad22– mus81–), MCW1235 (rhp51– mus81) and MCW1478 (rad22– rhp51– mus81–).
We have recently reported that a mus81 rhp54 double mutant exhibits a synergistic increase in poor growth and reduced viability (27). Rhp54 is a homologue of S.cerevisiae Rad54, which is believed to promote recombination by aiding Rad51-mediated strand invasion and D-loop formation, remodelling chromatin, and dissociating Rad51-dsDNA filaments (28). We have observed that rhp51 and rhp54 mutants exhibit an epistatic relationship with respect to genotoxin sensitivity consistent with Rhp51 and Rhp54 proteins functioning together (unpublished data). It was conceivable therefore that the poor growth and viability of a mus81 rhp54 double mutant might be due to aberrant or unresolved recombination intermediates formed by Rhp51. To test this we compared a mus81 rhp51 rhp54 triple mutant with its respective double mutants for growth and viability. The triple mutant exhibited improved growth and viability compared to a mus81 rhp54 double mutant indicating that either Rhp54 or Mus81 are required to prevent Rhp51-mediated toxicity (Figure 4D, data not shown). Unexpectedly we also observed that the triple mutant grew slightly better than a mus81 rhp51 double mutant suggesting that Rhp54 can have a deleterious effect in the absence of both Mus81 and Rhp51. When we tested these strains for genotoxin sensitivity it was apparent that the deletion of rhp54 in a mus81 rhp51 mutant background also reduced sensitivity to UV, HU and MMS (Figure 4D, data not shown). In the case of UV, sensitivity is reduced almost to the level of a rhp51 rhp54 double mutant, which is the same as a rhp51 single mutant. These data show that the additive UV sensitivity of a mus81 rhp51 double mutant is mostly a consequence of Rhp54-mediated toxicity, and suggest that mus81 and rhp51 may in fact function in the same pathway for the response to UV-induced damage. However, a mus81 rhp51 rhp54 triple mutant is still synergistically more sensitive to HU, MMS and CPT than mus81 single and rhp51 rhp54 double mutants (Figure 4D, data not shown). This reaffirms our conclusion that Mus81 functions in a Rhp51-independent pathway for the repair of certain kinds of DNA damage.
To see if Mus81 functions in a Rad22-dependent Rhp51-independent pathway we compared mus81 and rad22 single and double mutants for growth and genotoxin sensitivity. Contrary to our expectation a mus81 rad22 double mutant grows much slower and is more sensitive to UV, HU, MMS and CPT than its respective single mutant strains (Figure 4A–C). A comparison of this double mutant with a mus81 rhp51 rad22 triple mutant revealed that the poor growth and enhanced sensitivity of the mus81 rad22 mutant strain is due mainly to rhp51+ (Figure 4E, data not shown). Seemingly in the absence of both Rad22 and Mus81, Rhp51 has a deleterious effect. When we compared the triple mutant to a rad22 rhp51 double mutant we observed that both strains are similarly sensitive to UV, MMS and CPT, and that the triple mutant is slightly more sensitive to HU (Figure 4E). Taken together these data indicate that Mus81 functions mainly in a Rad22-dependent pathway for the repair of damage induced by UV, HU, MMS and CPT. Moreover in most cases this pathway is independent of Rhp51.
Effect of mus81 on the frequency of direct repeat recombination
As shown in Figure 3D Rad22-dependent (Rhp51-independent) recombination generates only deletion-type recombinants. Consistent with Mus81's involvement in this pathway the formation of deletion types is suppressed in a mus81 mutant, whereas the frequency of conversion types remains unaffected (Figure 5). Although the reduction in deletion types is modest it is statistically significant (P = 0.02). Furthermore it accounts for approximately half of the deletion types that are dependent on Rad22. Deletion of mus81 also slightly reduces the elevated frequency of deletion types in a rhp55 mutant (Figure 5). Again the effect is modest but significant (P = 0.02). Finally, the residual level of deletion types in a rad22 single or rad22 rhp55 double mutant is unaffected by mus81 mutation showing that Mus81 promotes deletion type formation in a Rad22-dependent manner.
Figure 5. Effect of mus81– on the frequency and type of recombinants in different mutant backgrounds. The strains used were MCW429 (wild-type), MCW988 (mus81–), MCW431 (rhp55–), MCW992 (rhp55– mus81–), MCW1494 (rad22–), MCW1500 (rad22– mus81–), MCW1227 (rad22– rhp55–) and MCW1039 (rad22– rhp55– mus81–).
Cleavage of Rad22-dependent D-loops by Mus81 in vitro
Our genetic data indicates that Mus81 can function with Rad22 independently of Rhp51 to promote the repair of DNA damage. Human (Hs) Rad52 can promote the formation of D-loops in vitro independently of Rad51 (4). This reaction probably depends on Rad52 promoting annealing between the invading single-stranded DNA and transiently unwound duplex DNA. It has also been shown recently that both human and budding yeast Rad52 can promote DNA strand exchange (29,30). Therefore one-way in which Rad22 and Mus81 could function together to promote DNA repair without Rhp51, is if Rad22 forms D-loops that Mus81 cleaves. To test this possibility in vitro we first purified recombinant Rad22 and found that it could promote D-loop formation similar to HsRad52 (data not shown). Standard D-loop assays measure homologous pairing between ssDNA fragments and superhelical dsDNA substrates, with reaction products being resolved by agarose gel electrophoresis. In such assays the ssDNA typically shares absolute homology to a region of the dsDNA and therefore can form a D-loop with no extruding flaps. However, the repair of a collapsed replication fork would involve strand invasion from a detached arm of the replication fork (i.e. a duplex DNA with a single-stranded tail). This would generate a D-loop with a four-way DNA junction containing three duplex arms. Such D-loops are ideal substrates for Mus81 in vitro (13).
In order for Rad22 to promote the formation of the type of D-loop that Mus81 favours for cleavage, we used a radiolabelled partial duplex consisting of a 20 bp heterologous duplex region and an 80 nt 3'-ended single-stranded region with homology to X174 DNA (Figure 6A). Incubation of these two DNAs in the absence of Rad22 results in trace amounts of two retarded radiolabelled products. Both products migrate slightly behind the position of the unlabelled X174 supercoiled DNA, and the slightly faster migrating product runs approximately at the position of linear X174 DNA (Figure 6B, lane a; data not shown). When the partial duplex is pre-incubated with stoichiometric amounts of Rad22 prior to the addition of the supercoiled DNA there is >10-fold increase in the amount of these retarded species (Figure 6B, lanes b–d). Optimal formation of these species occurs at approximately a 1:1 ratio of monomeric Rad22 and nucleotides of single-stranded tail, which is the same as for D-loop formation by HsRad52 (Figure 6B and C, data not shown) (4). To see if these species have the expected properties of the type of D-loop formed by RecA or HsRad52 we first looked to see if they were dependent on homology. To do this we made a second radiolabelled partial duplex that shares homology to a region of pBR322. Pre-incubation of this partial duplex with Rad22, followed by the addition of supercoiled pBR322 DNA, results in the formation of a D-loop (Figure 6C, lanes e–h). However, if X174 is used instead of pBR322 then no D-loop is formed (Figure 6C, lanes b and c). Similarly, combining the X174-based partial duplex with pBR322 also results in no D-loop (Figure 6C, lanes i and j). These data show that the D-loops formed by Rad22 are homology-dependent. A second characteristic of a bona fide D-loop is that it can be dissociated by spontaneous branch migration. Branch migration induced either by heat treatment or through the release of superhelical tension, by cutting the X174 DNA away from the D-loop with a restriction enzyme, resulted in the dissociation of the D-loops formed by Rad22 (data not shown). Finally, similar to HsRad52, pre-incubation of Rad22 with the X174 DNA inhibited its ability to promote D-loop formation (Figure 6B, lanes e–g) (4). This suggests that a Rad22–X174 DNA complex is unable to form a D-loop.
Figure 6. Rad22-promoted D-loop formation and D-loop cleavage by Mus81. (A) Schematic of D-loop formation. The asterisk indicates the position of the 5'-32P-end-label on the partial duplex. (B) Rad22-promoted D-loop formation. Reactions are described in Materials and Methods and contained Rad22 (6 nM, lanes b and e; 12 nM, lanes c and f; and 24 nM, lanes d and g) and 10 mM MgCl2. Rad22 was pre-incubated with partial duplex or X174 DNA as indicated. (C) Dependence on homology for D-loop formation. Reactions are described in Materials and Methods. (D) Effect of Mus81 on Rad22-promoted X174-based D-loop formation. Reactions contained 12 nM Rad22 and 14 nM Mus81–Eme1 as indicated. (E) Cleavage of purified X174-based D-loops by Mus81–Eme1. Reactions contained 7 nM (lane c) or 14 nM (lanes d and e) Mus81–Eme1 and 10 mM MgCl2 as indicated. The D-loop (lane a) was dissociated by heat treatment at 96°C for 2 min.
Having established that the D-loops generated by Rad22 are genuine we looked to see whether Mus81 could cleave them. Reactions in which Mus81 was added just after the supercoiled DNA showed a Mg2+-dependent decrease in the amount of D-loop together with an increase in the smear of radiolabelled material migrating between the D-loop and the free partial duplex (Figure 6D, compare lanes c–e). In these reactions Mus81 preferentially affects the more retarded D-loop band. These results are consistent with cleavage of the D-loop by Mus81, which would release the superhelical tension in the X174 DNA thereby enabling the D-loop to dissociate by branch migration.
To provide further evidence of Mus81's ability to target D-loops, we purified the D-loops from a Rad22 reaction and incubated them with Mus81 (Figure 6E). In the presence of Mg2+ Mus81 dissociated the purified D-loops, again preferentially targeting the more retarded D-loop band (Figure 6E, lanes c and d). A comparison with D-loops dissociated by heat treatment (lane a), shows that Mus81 dissociates much of the D-loop back to free partial duplex. Analysis of these reaction products by polyacrylamide gel electrophoresis revealed very little cleavage of the partial duplex DNA (data not shown). Incubation of Mus81 with supercoiled X174 DNA also revealed relatively little non-specific nicking activity (data not shown). Furthermore, incubation of Mus81 with synthetic D-loops under the conditions used here, confirms that Mus81 cleaves predominantly at sites just 5' of the four-way DNA junction in the strand that is complementary to the invading 3' single-stranded tail (Supplementary Material). These data indicate that Mus81 is causing the dissociation of the D-loop by specifically nicking the X174 DNA at the D-loop.
DISCUSSION
Epistasis analysis of the S.pombe Rad52 homologue Rad22 originally suggested that, unlike in S.cerevisiae, Rad22 and Rhp51 worked independently of each other (20). This conclusion was questioned by data showing that Rad22 and Rhp51 physically interact (31). A further characterization of rad22 mutant phenotypes, using a new rad22 deletion mutant, also suggested that earlier conclusions were misguided by the use of a presumed hypomorphic allele of rad22 (22). However, these studies still did not observe the same kind of dramatic effect on recombination that mutation of RAD52 has in S.cerevisiae (21,22). It was therefore concluded that Rad22 played a less critical role in S.pombe because of functional redundancy with another Rad52-like protein called Rti1 (2,31). We have shown here that this is not the case. Deletion of rad22 has a profound effect on survival following exposure to a range of different genotoxins, which, depending on the genotoxin, is either equivalent or greater than that of a rhp51 mutant. Direct repeat recombination is also dramatically reduced in a rad22 mutant, including a total ablation of conversion type recombinants. Finally, these rad22 mutant phenotypes show epistasis with those of a rhp51 mutant indicating that recombination is driven by Rad22-dependent mechanisms that are both dependent and independent of Rhp51. This is just like the relationship between RAD52 and RAD51 in S.cerevisiae. The reason that the earlier studies of rad22 were misguided appears to be due to the presence of suppressor mutations in the strains that were used. The identity and characterisation of the suppressors will be presented elsewhere.
Direct repeat recombination in S.cerevisiae and S.pombe
From our analyses it appears that the pathways of spontaneous direct repeat recombination are essentially the same in S.pombe as those that have been described in S.cerevisiae (32). In both yeast there are RAD52/rad22-dependent and -independent pathways. Although the exact mechanisms of recombination are unknown, it has been proposed that the major mechanism of RAD52-independent recombination is single-strand annealing (SSA) (32,33). In contrast RAD52-dependent recombination may be a mixture of SSA, RAD51-dependent gene conversion/reciprocal exchange, and BIR (32,33). In the absence of RAD51/rhp51 the frequency of gene conversion/reciprocal exchange is reduced or abolished, but the overall frequency of recombination increases (32,34). Like in S.cerevisiae this hyper-recombination in S.pombe is dependent on rad22, and may stem from un-repaired lesions (e.g. single-strand gaps) being converted into DSBs that are then repaired by SSA or BIR.
Mus81 functions in a recombinational repair pathway that is independent of Rhp51
We have shown that Mus81 generally functions in a Rad22-dependent pathway for DNA repair. Furthermore, for damage induced by HU, MMS and CPT it can promote repair independently of Rhp51. These genotoxins can variously cause stalling, blockage and breakage of the replication fork. It is difficult therefore to ascertain the exact type of lesion that is repaired by the Rad22–Mus81-dependent pathway. The exquisite hypersensitivity of mus81 mutant strains to CPT, however, provides perhaps the best clue as to what Mus81 may be doing in mitotic cells. CPT traps Top1 cleavage complexes thereby generating single-strand breaks at which replication forks collapse by ‘replication run-off’ (35). Cleavage complexes may also block DNA replication if present in the lagging strand template (36). Therefore Mus81 would appear to play an important role in the repair of blocked and/or broken replication forks (8). Suppression of this hypersensitivity by RusA indicates that Mus81's prime role here is in the processing of reversed replication forks and/or recombination intermediates (8). From our epistasis analysis we know that there are Rhp51-dependent and independent pathways for repairing CPT-induced damage. Furthermore, we know that Mus81 functions in the Rhp51-independent pathway, and based on our observation that mus81 rhp51 and rad22 rhp51 double mutants exhibit similar sensitivities to CPT, it seems that this pathway is largely dependent on Mus81. As mus81 and rhp51 single mutants exhibit similar sensitivities to CPT, we suspect that the Rhp51-independent pathway deals with approximately half of the replication fork problems induced by CPT.
We have found that CPT induces mostly deletion recombinants in the direct repeat substrate used in our studies (unpublished data). This is consistent with data showing that fork collapse gives rise to deletions between direct repeats, which occurs by strand invasion from the broken sister chromatid into the ‘wrong’ repeat (37). Interestingly approximately 50% of spontaneous Rad22-dependent deletion type recombinants are dependent on Mus81. We suspect that, like CPT-induced deletion types, these spontaneous events are due to replication fork collapse. The observation that half of these deletion types depend on Mus81, correlates well with our supposition that the Rhp51-independent pathway deals with approximately half of the replication fork problems induced by CPT. The hyper-levels of Rad22-dependent deletion types in a rhp55 mutant (equivalent to a rhp51 mutant), which are largely independent of Mus81, probably arise by SSA.
We have shown that Rad22 can promote the strand invasion of a double-stranded DNA molecule by a 3' single-stranded tailed duplex. Therefore, the Rhp51-independent pathway for dealing with CPT-induced damage could involve the repair of one-ended DSBs by Rad22-promoted strand invasion. Mus81 could be required to generate these one-ended DSBs by cleaving stalled replication forks. However, it is our contention that its primary role is to resolve the D-loops that are formed by Rad22 (Figure 1A). Certainly our in vitro data shows that Mus81 has this ability. The idea that a Rhp51-independent mechanism could be a prominent way of repairing collapsed replication forks is consistent with data showing that the accumulation of HJs in the rDNA array in S.cerevisiae depends on RAD52 and not RAD51 (38).
In S.cerevisiae Rad52 can repair two-ended DSBs by BIR independently of Rad51. BIR involves strand invasion from one of the broken ends into an intact sister chromatid or homologue (3). This primes DNA synthesis allowing chromosomal regions on the other side of the break to be recovered by DNA replication. However, the RAD51-independent pathway of BIR is relatively inefficient, and a recent study has shown that >95% of BIR events involving a single-end invasion depend on RAD51 (39). The inefficiency of the RAD51-independent pathway is probably due to the dependence of Rad52-promoted strand invasion on the donor duplex being transiently unwound (39). This study questions our idea that a Rad22–Mus81-dependent (Rhp51-independent) pathway could be a significant mechanism for one-ended DSB repair in S.pombe. However, studies of BIR in S.cerevisiae measure recombination between one broken DNA end and an intact sister chromatid or homologous chromosome. In contrast, replication fork collapse, by ‘replication run-off’ or by Mus81 cleavage of a reversed fork, generates one broken DNA end and a sister chromatid with a single-stranded gap . We think that this gap may provide access for a DNA helicase, which could generate the necessary unwound donor duplex for efficient Rad22-dependent strand invasion/annealing. We are currently investigating the validity of this assertion.
SUPPLEMENTARY MATERIAL
Supplementary Material is available at NAR Online.
ACKNOWLEDGEMENTS
This work was supported by a project grant (065278/Z/01/Z) and a Senior Research Fellowship from the Wellcome Trust awarded to M.C.W.
REFERENCES
McGlynn,P. and Lloyd,R.G. ( (2002) ) Recombinational repair and restart of damaged replication forks. Nature Rev. Mol. Cell. Biol., , 3, , 859–870.
Symington,L.S. ( (2002) ) Role of RAD52 epistasis group genes in homologous recombination and double-strand break repair. Microbiol. Mol. Biol. Rev., , 66, , 630–670.
Paques,F. and Haber,J.E. ( (1999) ) Multiple pathways of recombination induced by double-strand breaks in Saccharomyces cerevisiae. Microbiol. Mol. Biol. Rev., , 63, , 349–404.
Kagawa,W., Kurumizaka,H., Ikawa,S., Yokoyama,S. and Shibata,T. ( (2001) ) Homologous pairing promoted by the human Rad52 protein. J. Biol. Chem., , 276, , 35201–35208.
Odagiri,N., Seki,M., Onoda,F., Yoshimura,A., Watanabe,S., Enomoto,T., De Los Santos,T., Hunter,N., Lee,C., Larkin,B. et al. ( (2003) ) Budding yeast mms4 is epistatic with rad52 and the function of Mms4 can be replaced by a bacterial Holliday junction resolvase. DNA Repair (Amst.), , 2, , 347–358.
Bastin-Shanower,S.A., Fricke,W.M., Mullen,J.R. and Brill,S.J. ( (2003) ) The mechanism of Mus81–Mms4 cleavage site selection distinguishes it from the homologous endonuclease Rad1–Rad10. Mol. Cell. Biol., , 23, , 3487–3496.
Boddy,M.N., Gaillard,P.-H.L., McDonald,W.H., Shanahan,P., Yates,J.R.,III and Russell,P. ( (2001) ) Mus81–Eme1 are essential components of a Holliday junction resolvase. Cell, , 107, , 537–548.
Doe,C.L., Ahn,J.S., Dixon,J. and Whitby,M.C. ( (2002) ) Mus81–Eme1 and Rqh1 involvement in processing stalled and collapsed replication forks. J. Biol. Chem., , 277, , 32753–32759.
Chen,X.B., Melchionna,R., Denis,C.M., Gaillard,P.H., Blasina,A., Van de Weyer,I., Boddy,M.N., Russell,P., Vialard,J. and McGowan,C.H. ( (2001) ) Human Mus81-associated endonuclease cleaves Holliday junctions in vitro. Mol. Cell, , 8, , 1117–1127.
Whitby,M.C., Osman,F. and Dixon,J. ( (2003) ) Cleavage of model replication forks by fission yeast Mus81–Eme1 and budding yeast Mus81–Mms4. J. Biol. Chem., , 278, , 6928–6935.
Ciccia,A., Constantinou,A. and West,S.C. ( (2003) ) Identification and characterization of the human Mus81–Eme1 endonuclease. J. Biol. Chem., , 278, , 25172–25178.
Kaliraman,V., Mullen,J.R., Fricke,W.M., Bastin-Shanower,S.A. and Brill,S.J. ( (2001) ) Functional overlap between Sgs1-Top3 and the Mms4-Mus81 endonuclease. Genes Dev., , 15, , 2730–2740.
Osman,F., Dixon,J., Doe,C.L. and Whitby,M.C. ( (2003) ) Generating crossovers by resolution of nicked Holliday junctions: a role for Mus81–Eme1 in meiosis. Mol. Cell, , 12, , 761–774.
Gaillard,P.H., Noguchi,E., Shanahan,P. and Russell,P. ( (2003) ) The endogenous Mus81–Eme1 complex resolves Holliday junctions by a nick and counternick mechanism. Mol. Cell, , 12, , 747–759.
Moreno,S., Klar,A. and Nurse,P. ( (1991) ) Molecular genetic analysis of fission yeast Schizosaccharomyces pombe. Methods Enzymol., , 194, , 795–823.
Gutz,H., Heslot,H., Leopold,U. and Loprieno,N. ( (1974) ) Schizosaccharomyces pombe. In King,R.C. (ed.), Bacteria, Bacteriophages and Fungi, Plenum Press, New York, NY, Vol. 1. pp. 395–446.
Bahler,J., Wu,J.Q., Longtine,M.S., Shah,N.G., McKenzie,A.,III, Steever,A.B., Wach,A., Philippsen,P. and Pringle,J.R. ( (1998) ) Heterologous modules for efficient and versatile PCR-based gene targeting in Schizosaccharomyces pombe. Yeast, , 14, , 943–951.
Osman,F., Adriance,M. and McCready,S. ( (2000) ) The genetic control of spontaneous and UV-induced mitotic intrachromosomal recombination in the fission yeast Schizosaccharomyces pombe. Curr. Genet., , 38, , 113–125.
Whitby,M.C. and Dixon,J. ( (1998) ) Substrate specificity of the SpCCE1 holliday junction resolvase of Schizosaccharomyces pombe. J. Biol. Chem., , 273, , 35063–35073.
Khasanov,F.K., Savchenko,G.V., Bashkirova,E.V., Korolev,V.G., Heyer,W.D. and Bashkirov,V.I. ( (1999) ) A new recombinational DNA repair gene from Schizosaccharomyces pombe with homology to Escherichia coli RecA. Genetics, , 152, , 1557–1572.
van den Bosch,M., Zonneveld,J.B., Vreeken,K., de Vries,F.A., Lohman,P.H. and Pastink,A. ( (2002) ) Differential expression and requirements for Schizosaccharomyces pombe RAD52 homologs in DNA repair and recombination. Nucleic Acids Res., , 30, , 1316–1324.
van den Bosch,M., Vreeken,K., Zonneveld,J.B., Brandsma,J.A., Lombaerts,M., Murray,J.M., Lohman,P.H. and Pastink,A. ( (2001) ) Characterization of RAD52 homologs in the fission yeast Schizosaccharomyces pombe. Mutat. Res., , 461, , 311–323.
Suto,K., Nagata,A., Murakami,H. and Okayama,H. ( (1999) ) A double-strand break repair component is essential for S phase completion in fission yeast cell cycling. Mol. Biol. Cell, , 10, , 3331–3343.
Doe,C.L., Dixon,J., Osman,F. and Whitby,M.C. ( (2000) ) Partial suppression of the fission yeast rqh1(–) phenotype by expression of a bacterial Holliday junction resolvase. EMBO J., , 19, , 2751–2762.
Interthal,H. and Heyer,W.D. ( (2000) ) MUS81 encodes a novel helix–hairpin–helix protein involved in the response to UV- and methylation-induced DNA damage in Saccharomyces cerevisiae. Mol. Gen. Genet., , 263, , 812–827.
Boddy,M.N., Lopez-Girona,A., Shanahan,P., Interthal,H., Heyer,W.D. and Russell,P. ( (2000) ) Damage tolerance protein Mus81 associates with the FHA1 domain of checkpoint kinase Cds1. Mol. Cell. Biol., , 20, , 8758–8766.
Doe,C.L. and Whitby,M.C. ( (2004) ) The involvement of Srs2 in post-replication repair and homologous recombination in fission yeast. Nucleic Acids Res., , 32, , 1480–1491.
Tan,T.L., Kanaar,R., Wyman,C., Hsiang,Y.H., Lihou,M.G. and Liu,L.F. ( (2003) ) Rad54, a Jack of all trades in homologous recombination. DNA Repair (Amst.), , 2, , 787–794.
Bi,B., Rybalchenko,N., Golub,E.I. and Radding,C.M. ( (2004) ) Human and yeast Rad52 proteins promote DNA strand exchange. Proc. Natl Acad. Sci. USA, , 101, , 9568–9572.
Kumar,J.K. and Gupta,R.C. ( (2004) ) Strand exchange activity of human recombination protein Rad52. Proc. Natl Acad. Sci. USA, , 101, , 9562–9567.
Tsutsui,Y., Khasanov,F.K., Shinagawa,H., Iwasaki,H. and Bashkirov,V.I. ( (2001) ) Multiple interactions among the components of the recombinational DNA repair system in Schizosaccharomyces pombe. Genetics, , 159, , 91–105.
Klein,H.L. ( (1995) ) Genetic control of intrachromosomal recombination. Bioessays, , 17, , 147–159.
Prado,F. and Aguilera,A. ( (1995) ) Role of reciprocal exchange, one-ended invasion crossover and single-strand annealing on inverted and direct repeat recombination in yeast: different requirements for the RAD1, RAD10, and RAD52 genes. Genetics, , 139, , 109–123.
Aguilera,A. ( (1995) ) Genetic evidence for different RAD52-dependent intrachromosomal recombination pathways in Saccharomyces cerevisiae. Curr. Genet., , 27, , 298–305.
Pommier,Y., Redon,C., Rao,V.A., Seiler,J.A., Sordet,O., Takemura,H., Antony,S., Meng,L., Liao,Z., Kohlhagen,G. et al. ( (2003) ) Repair of and checkpoint response to topoisomerase I-mediated DNA damage. Mutat. Res., , 532, , 173–203.
Vance,J.R. and Wilson,T.E. ( (2002) ) Yeast Tdp1 and Rad1–Rad10 function as redundant pathways for repairing Top1 replicative damage. Proc. Natl Acad. Sci. USA, , 99, , 13669–13674.
Bierne,H., Ehrlich,S.D. and Michel,B. ( (1997) ) Deletions at stalled replication forks occur by two different pathways. EMBO J., , 16, , 3332–3340.
Zou,H. and Rothstein,R. ( (1997) ) Holliday junctions accumulate in replication mutants via a RecA homolog-independent mechanism. Cell, , 90, , 87–96.
Davis,A.P. and Symington,L.S. ( (2004) ) RAD51-dependent break-induced replication in yeast. Mol. Cell. Biol., , 24, , 2344–2351.
Ostermann,K., Lorentz,A. and Schmidt,H. ( (1993) ) The fission yeast rad22 gene, having a function in mating-type switching and repair of DNA damages, encodes a protein homolog to Rad52 of Saccharomyces cerevisiae. Nucleic Acids Res., , 21, , 5940–5944.
Wang,S.W., Goodwin,A., Hickson,I.D. and Norbury,C.J. ( (2001) ) Involvement of Schizosaccharomyces pombe Srs2 in cellular responses to DNA damage. Nucleic Acids Res., , 29, , 2963–2972.(Claudette L. Doe, Fekret Osman, Julie Di)