|Home | About | Journals | Submit | Contact Us | Français|
Most spontaneous DNA double-strand breaks (DSBs) arise during replication and are repaired by homologous recombination (HR) with the sister chromatid. Many proteins participate in HR, but it is often difficult to determine their in vivo functions due to the existence of alternative pathways. Here we take advantage of an in vivo assay to assess repair of a specific replication-born DSB by sister chromatid recombination (SCR). We analyzed the functional relevance of four structure-selective endonucleases (SSEs), Yen1, Mus81-Mms4, Slx1-Slx4, and Rad1, on SCR in Saccharomyces cerevisiae. Physical and genetic analyses showed that ablation of any of these SSEs leads to a specific SCR decrease that is not observed in general HR. Our work suggests that Yen1, Mus81-Mms4, Slx4, and Rad1, but not Slx1, function independently in the cleavage of intercrossed DNA structures to reconstitute broken replication forks via HR with the sister chromatid. These unique effects, which have not been detected in other studies unless double mutant combinations were used, indicate the formation of distinct alternatives for the repair of replication-born DSBs that require specific SSEs.
Double-strand breaks (DSBs) are among the most harmful DNA lesions. Failure to repair DSBs is often associated with apoptosis, aging, and cancer in metazoans and can lead to different types of genome instability in all organisms, including high mutation frequency, chromosome rearrangements, or chromosome loss. As a consequence, cells have developed a variety of specialized and complex mechanisms for DSB repair, defined as nonhomologous end joining (NHEJ) and homologous recombination (HR). In contrast to NHEJ, which works preferentially in nondividing cells at the G1 stage of the cell cycle, HR is the major DSB repair mechanism occurring at the S/G2 phase. In particular, HR is responsible for the repair of breaks that are associated with DNA replication (1). Understanding the mechanisms of HR and the proteins that catalyze these reactions is therefore central to our understanding of cell proliferation and associated pathological states and diseases.
A key step in HR is the resolution of crossed-stranded DNA structures formed during DNA strand exchange (28, 41, 53). D-loops formed by Rad51-mediated DNA strand exchange may lead to the formation of double Holliday junctions (HJs), which can be resolved in two ways (17, 28): (i) by dissolution catalyzed by the Sgs1/BLM-Top3 helicase-topoisomerase complex (54); (ii) by endonucleolytic cleavage mediated by structure-selective endonucleases (SSEs) (28, 32, 35, 38). Four conserved SSEs, Mus81-Mms4, Slx1-Slx4, Yen1, and Rad1-Rad10, have been identified in Saccharomyces cerevisiae, and their biochemical activities have been studied extensively (15, 20, 21, 31). The in vivo roles of Mus81-Mms4, Slx1-Slx4, and Yen1, however, remain to be determined. One complication has been their apparent functional overlap and the presence of alternative pathways by which recombination intermediates can be resolved.
Mus81-Mms4 (Mus81-Eme1 in Schizosaccharomyces pombe) is a conserved member of the XPF family of heterodimeric nucleases (9) that is required for maturation of recombination intermediates that lead to meiotic crossovers in different organisms (6, 30, 47) and for efficient DNA repair in mitotic cells and after replicative stress (14, 27). Mus81-Mms4 preferentially cleaves D-loops, 3′-flap structures, and nicked HJs (16, 23, 33, 39, 50). Mutations in Mus81-Mms4 are synthetically lethal with mutations in the Sgs1-Top3-Rmi1 complex, which is involved in dissolution of double Holliday junctions (4, 7, 17). The fact that this synthetic lethality is suppressed by a defect in HR (rad51, rad52, rad55, rad57, rad54) (4, 17) provides genetic evidence that Mus81-Mms4 cleaves a recombination-dependent joint molecule (29). Slx1-Slx4 preferentially cleaves 5′-flap structures in yeast (12, 21), but the human counterpart has been proposed to have HJ resolvase activity (3, 18, 37, 48). In Drosophila melanogaster and Caenorhabditis elegans, Slx1-Slx4 is required for meiotic crossover formation (44, 46, 55). Interestingly, Slx1-Slx4 was defined by mutations leading to synthetic lethality with sgs1 (36) in a screen in which mus81 and mms4 were also obtained, but the lethality was not rescued by rad51Δ mutations. As a consequence, despite its in vitro activities, it appeared unlikely that Slx1-Slx4 had a sole function in HR. Moreover, Slx4 seems to provide a platform for the association of various SSEs, including MUS81-EME1 in humans and the nucleotide excision repair (NER) endonuclease Rad1-Rad10 (18, 19), adding a further complication to understand the specific in vivo role of Slx4. Rad1-Rad10, on the other hand, is a 3′-flap endonuclease that, aside from its role in NER, participates in the single-strand annealing repair pathway of DSBs, and it also has in vitro SSE activity (40, 45).
Recently, the S. cerevisiae Yen1 and human GEN1 proteins were identified and shown to resolve HJs by a symmetrical cleavage mechanism similar to that mediated by the bacterial RuvC resolvase (31, 42). Single yen1Δ mutants show no or very subtle defects in either mitotic or meiotic recombination in S. cerevisiae, and Schizosaccharomyces pombe lacks a Yen1 ortholog. However, S. cerevisiae mus81Δ yen1Δ double mutants display additional sensitivities to DNA-damaging agents and show reduced levels of crossovers compared to the single mutants, suggesting that Mus81-Mms4 and Yen1 might have possibly overlapping functions in DSB repair (5, 29, 49). Consistent with this, overexpression of human GEN1 can rescue the meiotic phenotype of an S. pombe mus81 mutant (34).
To assay the in vivo role of SSEs in HR as the main mechanism of DSB repair associated with replication, we have taken advantage of an in vivo assay that assesses the repair of a specific replication-born DSB by sister chromatid exchange or recombination (SCE or SCR). Using physical and genetic analyses, we show that ablation or inactivation of the catalytic domains of Mus81-Mms4 and Yen1, as well as ablation of Slx4 and Rad1, but not of Slx1, leads to a clear and specific decrease of SCR that is not observed in spontaneous general HR in budding yeast. Our work suggests that these nucleases act in vivo on specific DNA intermediates generated during the repair of replication-born DSBs and provides evidence for an independent function of these SSEs in DSB repair that may relate to the resolution of specific DNA intermediates.
Yeast strains used in this study are listed in Table 1. All strains were made in the W303 (RAD5+) genetic background. Mutants were obtained by gene replacement with the KanMX4, NatMX6, or HphMX4 cassettes as described and confirmed by PCR and Southern analyses. The mus81-dd (mus81-D414A,D415A) mutation was transplaced to the genome by exact gene replacement using mus81-D::URA3K. lactis and transformation with a linear fragment containing mus81-dd. Plasmid pRS316TINV containing the 24-bp mini-HO site inserted at the EcoRI site of one of the inverted LEU2 repeats was described previously (10, 24). Plasmids pAG414GPD-Yen1E193AE195A-HA and pAG414GPD-Yen1-HA, containing different alleles of YEN1 under the control of the GPD1 promoter, were generated by Gateway cloning (Invitrogen) using pAG414GPD-ccdB-HA (2) as the destination vector in a similar way as described previously (5). Plasmid pWDH800 (GST-Mms4/HIS10-FLAG-Mus81) containing Mus81-Mms4 under the GAL1,10 promoter was generated by insertion of a TRP1 marker into an NcoI site of plasmid pWDH595 (15). Plasmid pWDH815 GST-Mms4/HIS10-FLAG-mus81-D414A,D415A, carrying the catalytically deficient form of Mus81, was generated by site-directed mutagenesis using specific primers oIWSD374 (5′-TAGTTGAAAGAAAAAGGCTAGCCGCTTT-3′) and olWDH375 (5′-TCCCTTATACTTAAAGCTAAAGCGGCTA-3′).
Recombination frequencies are the median values of fluctuation tests performed with six independent yeast colonies each from each transformant analyzed. For every genotype, the fluctuation test was repeated three times with three different yeast transformants. The final frequency shown for each genotype corresponds to the mean value of the three median frequencies obtained from the tests. For the analysis of HO-mediated DSB recombination mid-log-phase yeast cells carrying the HO gene under the control of GAL1 were obtained from SC–3% glycerol, 2% lactate liquid cultures and split into halves. One-half was maintained in liquid SC–3% glycerol, 2% lactate (no HO expression), and the other was cultured in SC–2% galactose for 5 h for transient expression of HO, before performing fluctuation tests. In all cases doxycycline was used to avoid leu2-HOr expression, because we previously found that HO cuts more efficiently under these conditions (25). Recombinants were selected on SC-leu-ura containing 2% glucose.
To assay the impact of Mus81-Mms4 and Yen1 in SCR induced by a replication-born DSB, we used pRS316-TINV, a plasmid containing the inverted TINV system carrying a 24-bp mini-HO site (Fig. 1A) (24). We previously showed that HO endonuclease cleaves this site preferentially in one strand, leading to a replication-induced DSB (10). Due to the low efficiency of HO cleavage at this site, the DSB occurs in only one chromatid, which may then be repaired by HR with the intact sister. We previously showed that the measurement of unequal SCR with the inverted repeat in the sister chromatid is a reliable measurement of SCR, as a valid and direct measurement of equal SCR (24). Genetically, the TINV system can detect Leu+ recombination events, which occur mainly by recombination with the sister chromatid, even though other mechanisms can also occur at a lower efficiency (25). However, spontaneous recombination, which can be initiated in different ways and at different sites, may occur at high frequencies by multiple intrachromatid events, including reciprocal exchange, gene conversion, and break-induced replication (BIR) plus single-strand annealing (SSA) events (41). Therefore, DSB-induced Leu+ events can be taken as an approximate genetic measurement of SCR, whereas spontaneous Leu+ events measure different types of recombination events, many of which do not involve the sister chromatid (24).
To determine the effect of mus81Δ and mms4Δ on SCR, we first performed a physical analysis of DSB formation and SCR in the TINV system. We performed a kinetic analysis of SCR in a time frame of 9 h under HO induction (Fig. 1B). DSBs accumulated to similar levels in the wild type, mus81Δ and yen1Δ single mutants, and the mus81Δ yen1Δ double mutant. However, DSB levels remained high for longer periods in the mutants than in the wild type, suggesting a lower capacity to repair them. Direct analysis of SCR by quantification of the 4.7-kb SCR band revealed that SCR was significantly impaired in both mutants. SCR impairment was stronger in mus81Δ than in yen1Δ cells, with a further but relatively small additional defect observed in the double mutant.
For the genetic analysis of Leu+ recombinants, DSBs were induced for only 5 h to avoid saturation of recombinants, as previously described (25). As shown in Fig. 1C, whereas spontaneous recombination was unaffected in the single mutants, a slight 3-fold increase was observed in the mus81Δ yen1Δ double mutant. This might have been a consequence of either replication failures or the SCR defect itself, given that a lower capacity to use the sister chromatid as repair template choice could channel repair into other recombination events. DSB-induced SCR was significantly decreased in both the mus81Δ (38.6-fold) and yen1Δ (6.4-fold) single mutants and in the double mutant (58.8-fold) with respect to the wild type. The difference between double mus81Δ yen1Δ and single mus81Δ mutants, however, was not statistically significant (Student's t test, P = 0.087). Altogether, these results indicate an inability of the mutant cells to promote efficient DSB repair by SCR.
Although our results suggested that the double mutant mus81Δ yen1Δ showed a stronger reduction in genetic SCR products than the single mutant mus81Δ, the difference was not statistically significant to allow the conclusion that Yen1 can substitute for the function of Mus81 in SCR. This contrasts with the results reported for DNA damage sensitivity or mitotic recombination between homologs, in which only double mutant combinations reveal a phenotype for yen1Δ (5, 29). To assay the level of functional overlap of both proteins in SCR, we determined the capacities of Mus81-Mms4 and Mus81dd-Mms4 to substitute for the role of Yen1 (where Mus81dd indicates the catalytic mutant D414A,D415A). For this, we overexpressed the wild-type Mus81-Mms4 and the catalytically inactive Mus81dd-Mms4 in wild-type and yen1Δ cells as well as in mus81Δ controls and analyzed SCR. Physical analysis of SCR showed that, whereas Mus81-Mms4 overexpression complemented the SCR defect of both yen1Δ and mus81Δ strains, Mus81dd-Mms4 overexpression did not (Fig. 2A; see also Fig. S1 in the supplemental material). Notably, Mus81dd-Mms4 overexpression decreased SCR in both wild-type and yen1Δ cells, consistent with Mus81dd-Mms4 interfering with endogenous Mus81.
Genetic analysis of Leu+ recombination events revealed that spontaneous recombination was not significantly affected in any of the strains analyzed (Fig. 2B). Consistent with the physical analysis, the defect in HO-induced SCR Leu+ events in mus81Δ cells was complemented by overexpression of Mus81-Mms4 but not by Mus81dd-Mms4. Furthermore, Mus81-Mms4 overexpression, but not Mus81dd-Mms4 overexpression, suppressed the SCR defect of yen1Δ strains, raising the levels of Leu+ recombination to wild-type levels. Mus81dd-Mms4 overexpression did not affect the levels of Leu+ recombinants in wild-type and yen1Δ cells, which may be explained based on the genetic assay measuring other recombination events in addition to SCR. Altogether, these results suggest that the catalytic activity of Mus81-Mms4 can replace the loss of Yen1 function.
As overexpression of Mus81-Mms4 can replace Yen1 in the SCR assay, we wondered whether the reciprocal was also the case. For this, we overexpressed Yen1 and its catalytically inactive Yen1-ee form in mus81Δ strains as well as in the wild type and yen1Δ control. Physical analysis of SCR showed that overexpression of the active form of Yen1, but not of inactive Yen1-ee, suppressed the mus81Δ mutant defect (Fig. 3A). Interestingly, the overexpression of Yen1-ee in wild-type and mus81Δ cells did not result in a decrease of SCR levels, as it occurred with Mus81dd-Mms4 overexpression in wild-type and yen1Δ cells. This difference can be explained by the distinct relevance of both endonucleases for SCR, as the defect of yen1Δ cells is weaker than that of mus81Δ (Fig. 1B), or by the ability of high overexpression of Mus81dd to mimic the mus81Δ phenotype. Consistently, as shown in Fig. 3B, no effect was observed in spontaneous Leu+ recombination in any of the strains tested, with the exception of mus81Δ. This may be consistent with the fact that Yen1-ee retains DNA binding activity and when overexpressed can displace the wild-type Yen1 from the repair site (5). However, while the frequency of Leu+ SCR events after Yen1-ee overexpression had no effect in wild-type, mus81Δ, or yen1Δ strains, the overexpression of catalytically active Yen1 elevated the frequency of Leu+ recombinants to wild-type levels in both yen1Δ and mus81Δ strains (Fig. 3B). This result indicates that the role of Yen1 in SCR is mediated by its nuclease activity and that overexpression of Yen1 can compensate for loss of Mus81-Mms4 function in SCR as a mechanism of repair for replication-born DSBs.
Next, we assayed whether Slx4 was also required for the repair of replication-born DSBs by SCR. Physical analysis of the TINV inverted repeat system showed that the levels of DSBs reached in different slx4Δ mutant combinations were the same as in wild-type cells after 3 h of HO induction but remained higher during the time course, consistent with a repair deficiency. Importantly, SCR was heavily impaired in the single slx4Δ mutants (Fig. 4A). The levels were equally low in double and triple mutant combinations of slx4Δ with mus81Δ and yen1Δ. Therefore, we conclude that Slx4 is required for SCR independently of the presence or absence of Yen1 and/or Mus81-Mms4.
The genetic analysis showed that slx4Δ has a minor impact (2-fold increase above wild-type levels) on spontaneous Leu+ recombination levels in the same TINV system (Fig. 4B). The same levels were obtained in double mutants with yen1Δ and mus81Δ, as well as in the triple mutant lacking the three SSEs, implying that their roles are independent. HO-induced Leu+ recombination was reduced 7- to 8-fold in the single slx4Δ mutant and the slx4Δ mus81Δ and slx4Δ yen1Δ double mutants, consistent with a defect in SCR. Interestingly, slx4Δ suppressed the strong defect in generating Leu+ recombinants observed in mus81Δ (38.6-fold for mus81Δ [Fig. 1C] and 7.4-fold for mus81Δ slx4Δ [Fig. 4B]). However, no effect on hydroxyurea (HU) sensitivity and only a slight suppression in methyl methanesulfonate (MMS) sensitivity were observed (see Fig. S2 in the supplemental material). Triple mutants yen1Δ mus81Δ slx4Δ showed the same strong reduction in HO-induced Leu+ recombinants (56.7-fold) as the double yen1Δ mus81Δ (compare Fig. 4B and and1C),1C), while MMS and HU sensitivities were higher in the triple mutant (see Fig. S2). These results are consistent with the physical analysis, which showed that Slx4 is required for the repair of replication-born DSBs via SCR but does not uncover an additional SCR pathway beyond those defined by Mus81-Mms4 and Yen1.
To determine whether overexpression of Yen1 and Mus81-Mms4 suppressed the SCR defect of slx4Δ, we determined the effect on SCR by genetic analysis. As shown in Fig. 5, the spontaneous increase in recombination of slx4Δ cells was reduced by overexpression of either Yen1 or Mus81-Mms4. Accordingly, the same occurred in the double and triple mutants, suggesting that Yen1 and Mus81-Mms4 facilitate repair with alternative donors. However, when DSB-induced SCR was analyzed, overexpression of Yen1 suppressed the defects of single, double and, to a lesser extent, triple mutants. These results showed that Yen1 overexpression can almost completely suppress the effects provoked by the absence of Mus81-Mms4 or Slx4 (Fig. 5). Mus81-Mms4 overexpression had little or no effect on single slx4Δ or double mus81Δ slx4Δ mutants. A partial rescue of the defect of the yen1Δ slx4Δ double and mus81Δ yen1Δ slx4Δ triple mutants was observed, as expected from the capacity of Mus81-Mms4 overexpression to suppress yen1Δ defects (Fig. 2B). These data showed that overexpression of Yen1, but not of Mus81-Mms4, can suppress the SCR defect in slx4Δ mutants.
Slx4 may work as a platform for the action of SSEs, such as Rad1-Rad10 (XPF-ERCC1), or together with Slx1 as a protein complex with a function in DNA junction cleavage (18, 19). Although there is no evidence yet that this happens in yeast, it was important to determine whether Slx4 action on SCR could be due to a possible function as a Slx4-Slx1 resolvase, and whether Rad1, which has been reported to have in vitro cleavage activity on HJs (26), could have a role in SCR. As can be seen in Fig. 6A, physical analysis showed that the DSB accumulation kinetics was essentially similar in wild-type, rad1Δ, and slx4Δ rad1Δ cells but slightly enhanced in slx4Δ cells. Importantly, SCR levels were clearly diminished in rad1Δ and to similar levels in slx4Δ rad1Δ cells, and further in slx4Δ cells (Fig. 6A). However, slx1Δ cells did not show the enhanced DSB accumulation kinetics and decreased SCR kinetics of slx4Δ, while the slx1Δ rad1Δ strain showed similar DSB and SCE kinetics to those of rad1Δ cells (Fig. 6B). These data indicate that Rad1, but not Slx1, has a role in SCR. The physical analysis was extended to double mutant combinations of slx1Δ and rad1Δ with mus81Δ, and revealed that double and triple mutants accumulated DSBs above the wild-type levels. SCR was reduced in all mutant combinations to levels comparable to mus81Δ, with the lowest being those of mus81Δ rad1Δ and mus81Δ slx1Δ rad1Δ (Fig. 6C). These results confirmed that while Rad1 has an effect on SCR as prominent as that of Mus81, Slx1 does not have a major role in this process, since slx1Δ mutants only displayed a very mild decrease in SCR when combined with the absence of Mus81 and Rad1.
Genetic analysis of Leu+ recombinants showed that spontaneous recombination was not significantly affected in slx1Δ or rad1Δ single mutants or in double mutant combinations with mus81Δ, yen1Δ, or each other, except in the case of rad1Δ mus81Δ (Fig. 7). Consistent with the physical analysis, HO-induced SCR was unaffected in slx1Δ cells but significantly diminished in rad1Δ (3.3-fold). This effect was the same in double, triple, and quadruple combinations with mus81Δ, slx1Δ, and/or yen1Δ. Interestingly, mus81Δ was epistatic to rad1Δ, whereas yen1Δ caused a slightly greater decrease in SCR (Fig. 7), consistent with the results shown in Fig. 1 and with MMS and HU sensitivities (see Fig. S2 in the supplemental material). Altogether, the results indicate that whereas Slx1 has no detectable role in SCR, Rad1 functions in SCR mainly in the same pathway as Mus81. Interestingly, Slx1 was required for SCR in the absence of Rad1. This suggests that Slx1 may cleave the same substrates (or their processing products) as Rad1 in SCR, whether in junction processing or in cleavage of putative short 5′-end flaps that can be generated during strand exchange by the invading heterologous HO site sequences. Further studies on the role of Slx1 nuclease are required to establish its role in vivo. However, in contrast to the idea that Slx1 and Slx4 act together in a resolvase complex, our study reveals that Slx4 has a more central role in SCR, likely in junction processing, that is independent of Slx1.
Using genetic and physical analyses, we found that the repair of a specific replication-born DSB by SCR is dependent upon three endonucleases, Mus81-Mms4, Yen1, and Rad1, as well as Slx4. Deletion of any of these activities led to a specific and independent impairment of SCR. In contrast, mus81Δ, yen1Δ, and slx4Δ single mutants showed no defects in general spontaneous recombination. Slx1, on the other hand, did not play a major role in SCR in our assay, unless Rad1 was absent, which suggests that Slx1 may cleave the same substrates or their processing products as Rad1. Studies with the wild-type and catalytically inactive forms of Yen1 and Mus81-Mms4 overexpressed in different mutant background combinations of mus81Δ, yen1Δ, and slx4Δ suggest that each SSE can process the replication-born recombination intermediates that accumulate in the absence of the other two SSEs, albeit to different extents. The three proteins, and in particular the recently discovered Yen1 resolvase, have independent and nonredundant functions in the repair of replication-born DSBs via SCR, acting on different SCR intermediates, one processed by Yen1 and the other by Mus81-Mms4, together with Slx4. Importantly, the fact that slx1Δ has no effect in SCR indicates that the contribution of Slx4 does not occur in the context of the Slx1-Slx4 complex, which suggests that in yeast both proteins may not function as a resolvase unit. This is consistent with the proposed role of Slx4 as a platform for other nucleases, including Rad1-Rad10 (18).
The in vivo impact of mutations in different HR genes on the general mechanism of DSB repair is usually determined in meiotic assays, in which HR events between homologous chromosomes are initiated by Spo11-mediated DSB formation (35), or in vegetative assays that examine either spontaneous or DSB-induced HR between homologous chromosomes or between ectopic DNA repeats (40). Despite the essential contributions of these assays to our understanding of the in vivo functions of HR proteins, the large variety of mitotic HR events and the availability of alternative DSB repair pathways, such as the double Holliday junction pathway, synthesis-dependent strand annealing (SDSA), break-induced replication (BIR), single-strand annealing (SSA), SCR, etc., can limit our capacity to discern the in vivo role for many HR proteins. In other words, some factors involved in recombinational repair cannot be linked to a detectable effect in most mitotic HR assays. A role for cohesins and Smc5-6 proteins, for example, has been demonstrated in vivo only by studying SCR induced by replication-born DSBs (10, 13). The SCR assay used here is sensitive enough to be able to differentiate between the roles of various SSEs, which has not been possible in other general repair and HR assays unless double mutants were analyzed (5, 29).
The physical and genetic assay used in this study relies on a single-stranded break that leads to a DSB during DNA replication. This leads to a situation in which a spontaneous DSB occurs naturally as a consequence of replication fork collapse. Reestablishment of the interaction between the break end and the sister chromatid (SCR) is the major mechanism for repair of the break and restoration of the replication fork (11, 24, 25). Regardless of whether a DSB occurs at the replication fork as a consequence of its collapse or distal to the fork after the fork has passed the lesion (and therefore regardless of whether the break is single ended or double ended), DNA strand invasion into the sister chromatid is needed to generate the D-loop intermediate (Fig. 8). The Mus81-Mms4 endonuclease may process such D-loops or other junctions that have not yet been ligated (nicked HJ, nicked dHJ) (15, 45), which explains why mus81 mutants show SCR defects. This is consistent with the requirement for Mus81-Eme1 in S. pombe for the repair of DSBs originating from a nick (43). Yen1 has a weak but significant effect on SCR as detected both at the physical and genetic level. To date, this represents the only evidence showing an effect of yen1Δ single mutants on DNA repair and recombination. Based on the biochemical specificity of Yen1 (31), it is possible that Yen1 acts after Mus81-Mms4 on HJs that have been ligated. At the physical level, the effect is better observed in a rad5-G535R mutant background (see Fig. S3 in the supplemental material), carrying a point mutation in the ATPase domain of the postreplicative repair gene RAD5 that confers a weak DNA repair defect, consistent with a role of Yen1 in the resolution of an intermediate that arises during replication.
It has recently been shown that depletion of both MUS81 and GEN1 reduces the high levels of cytologically detectable sister chromatid exchanges found in BLM−/− human cell lines (52), implying a role of SSEs in crossover HR. However, the increased level of sister chromatid exchanges in BLM−/− human cells, as well as in yeast sgs1 cells (8, 22, 51), may alternatively be explained by an increased number of DNA breaks that arise in BLM-defective cells during replication. Such breaks presumably would not be observed if they were mediated by SSEs (17). The present results therefore provide important new insights by unambiguously confirming that the SSE proteins assist in the repair of broken replication forks via SCR, rather than acting to suppress DSB formation. The biochemical properties of Yen1/GEN1 (31) suggest an interesting possibility, that Yen1 may cleave specific HJs formed when the double-ended DSB occurs in the lagging versus the leading strand (Fig. 8). Our data indicate that Yen1 and Mus81-Mms4 are required to process SCR intermediates, but they cannot easily interchange their roles, unless overexpressed. Our results therefore confirm the previously reported in vivo roles of Yen1 and GEN1 in HR in yeast and mammals (29, 52), and they extend the conclusions further to assign a specific in vivo role for each SSE.
SCR is also strongly reduced in slx4Δ cells, as detected by physical kinetic analysis (Fig. 4), confirming a relevant role for Slx4 in SCR. Interestingly, the physical effect is as strong as in mus81Δ mutants, but not the genetic end point effect (Leu+ recombinants) (Fig. 1 and and4).4). This difference between the physical and genetic end point analyses was expected, because molecular assays are more sensitive and genetic end point assays are not reflective of repair kinetics. The slx4Δ mus81Δ double mutant showed suppression of the much stronger mus81Δ defect, whereas the triple mutant lacking the three SSEs showed the low recombination and SCR efficiency of the double mus81Δ yen1Δ mutants. These results indicate that in DSB-induced SCR, Slx4 plays an important role and possibly functions to inhibit Yen1 or Mus81 action. Thus, we wondered whether the absence of Slx4 could rescue the inviability of mus81Δ sgs1Δ double mutants, by allowing Yen1 to resolve the toxic recombination intermediates accumulated in the absence of Mus81 and Sgs1 (17). However, this does not seem to be the case, as the mus81Δ sgs1Δ slx4Δ triple mutant was also inviable (see Fig. S4 in the supplemental material). It has been proposed that Slx4 may serve as a platform for the action of the Slx1 SSE, which functions with Slx4 as a heterodimer, as well as other SSEs, such as MUS81-EME1 in humans and Rad1-Rad10 in S. cerevisiae (18, 19). Interestingly, our data indicated that yeast Slx1 has no role in SCR unless Rad1 is absent in the cell. This result argues against the possibility that yeast Slx1 and Slx4 form a protein complex responsible for HJ resolution in SCR in a wild-type scenario. Additionally, it suggests that the role of Slx4 in SCR may be exerted as part of a platform for other SSEs. In contrast, the rad1Δ mutant showed a clear defect in SCR (Fig. 6). One possibility is that Rad1-Rad10 functions in SCR by using the Slx4 platform. However, given the role of Rad1 nuclease in cleaving overhanging DNA structures formed during recombination (40), it is difficult to establish whether the effects observed in this study indicate a specific role of Rad1-Rad10 in junction cleavage. Alternatively, the Rad1-Rad10 major contribution could be to remove the short nonhomologous tail of the mini-HO site after DNA strand invasion, allowing priming of DNA synthesis and completion of the recombination event (Fig. 8). Either way, further studies are required to determine at which SCR step Rad1 acts. However, the epistatic relationship of rad1Δ with all other mutations tested in HO-induced SCR events (Fig. 7) can be explained without the need of invoking a role for rad1 in junction resolution. This is clear from the results showing a similar frequency of HO-induced SCR events in rad1Δ mus81Δ yen1Δ cells (6.4 × 10−4 [Fig. 7]) and yen1Δ mus81Δ cells (8.2 × 10−4 [Fig. 1C]), suggesting that Rad1 participates in a common HR intermediate regardless of whether the recombination event is resolved by Yen1 or by Mus81 (Fig. 8).
Interestingly, only overexpression of Yen1 has a significant effect on slx4Δ mutants (Fig. 5). No effect was observed by overexpression of Mus81-Mms4 in slx4Δ cells, as suppression of the HR defect observed in mus81Δ slx4Δ and mus81Δ yen1Δ double mutants could be explained by the capacity of Mus81-Mms4 overexpression to complement or suppress the mus81Δ or the yen1Δ mutation (Fig. 1 and and2).2). This suggests that Mus81-Mms4 and Slx4 functions are not interchangeable and likely involve different types of DNA junction substrates. In mammals, it has been shown that SLX4 immunoprecipitates contain a HJ resolvase activity in association with Slx1 (3, 18, 37, 48), but analysis in yeast has shown that HJ cleavage appears biologically insignificant and that the complex is most active on 5′-flap structures (12, 21).
The fact that the slx4Δ mus81Δ double mutant (Fig. 4) shows higher levels of SCR than single mutant mus81Δ cells (Fig. 1) suggests that the presence of Slx4 at the junction substrate prevents the access or function of Yen1 at the D-loop. Mus81-Mms4 together with Slx4 would prevent Yen1 action. In their absence, Yen1 could access the junction, yielding partial suppression of the mus81Δ phenotype.
In summary, this work shows that Yen1 on the one hand and Mus81-Mms4 and Slx4 on the other hand cleave different intercrossed DNA structures to reconstitute broken replication forks via HR with the sister chromatid, with Rad1 having a role in cleavage of either the junction or the 3′-end overhanging tail (Fig. 8). This unique effect, which has previously been detected only in double mutants by measuring general HR events or sensitivity to DNA-damaging agents (5, 29), indicates that distinct alternatives for the repair of replication-born DSBs require specific, nonoverlapping functions of Mus81-Mms4, Slx4, Yen1, and Rad1-Rad10.
We thank U. Rass and F. Cortés-Ledesma for critical reading of the manuscript, M. J. Martín-Niclos for technical assistance, B. Pardo for providing yeast strains, and D. Haun for style supervision.
This work was supported by grants from the Spanish Ministry of Science and Innovation (BFU2010-16372 and Consolider Ingenio 2010 CSD2007-015), Junta de Andalucía (BIO102 and CVI4567), and FEDER to A.A., the U.S. National Institutes of Health (GM58015) to W.D.H., and the ERC, Cancer Research UK, and the Louis-Jeantet Foundation to S.C.W. S.M.-G. and E.K.S. were supported by CSIC and HHMI-IMBS predoctoral training grants, respectively.
We declare that we have no conflict of interest.
Published ahead of print 21 February 2012
Supplemental material for this article may be found at http://mcb.asm.org/.