Interplay of Mre11 Nuclease with Dna2 plus Sgs1 in Rad51-Dependent Recombinational Repair

The Mre11/Rad50/Xrs2 complex initiates IR repair by binding to the end of a double-strand break, resulting in 5′ to 3′ exonuclease degradation creating a single-stranded 3′ overhang competent for strand invasion into the unbroken chromosome. The nuclease(s) involved are not well understood. Mre11 encodes a nuclease, but it has 3′ to 5′, rather than 5′ to 3′ activity. Furthermore, mutations that inactivate only the nuclease activity of Mre11 but not its other repair functions, mre11-D56N and mre11-H125N, are resistant to IR. This suggests that another nuclease can catalyze 5′ to 3′ degradation. One candidate nuclease that has not been tested to date because it is encoded by an essential gene is the Dna2 helicase/nuclease. We recently reported the ability to suppress the lethality of a dna2Δ with a pif1Δ. The dna2Δ pif1Δ mutant is IR-resistant. We have determined that dna2Δ pif1Δ mre11-D56N and dna2Δ pif1Δ mre11-H125N strains are equally as sensitive to IR as mre11Δ strains, suggesting that in the absence of Dna2, Mre11 nuclease carries out repair. The dna2Δ pif1Δ mre11-D56N triple mutant is complemented by plasmids expressing Mre11, Dna2 or dna2K1080E, a mutant with defective helicase and functional nuclease, demonstrating that the nuclease of Dna2 compensates for the absence of Mre11 nuclease in IR repair, presumably in 5′ to 3′ degradation at DSB ends. We further show that sgs1Δ mre11-H125N, but not sgs1Δ, is very sensitive to IR, implicating the Sgs1 helicase in the Dna2-mediated pathway.


Introduction
Double-strand breaks (DSBs) are a major result of endogenous DNA replication errors and of exogenous DNA damaging agents, and failure to repair such breaks leads to the gross chromosomal rearrangements characteristic of tumor cells. In yeast, homologous recombination (HR), carried out by proteins encoded by the RAD52 epistasis group, is used to repair DSBs induced by ionizing radiation (IR). The time course of IR repair in vivo has been assayed by observing the post-irradiation appearance of foci of the various HR repair proteins tagged with green fluorescent protein [1]. Mre11, Rad50, and Xrs2 form a complex (MRX) and are the first arrivals after DSB formation. The MRX complex regulates the processing of the end of the DSB, resulting in resection of the 59 end and leaving a 39 single-stranded overhang [2]. RPA binds the single-stranded DNA, followed by Rad52 and Ddc2/Mec1 binding. Ddc2/Mec1 is the apical checkpoint-activating protein kinase. Rad52 can mediate the exchange of RPA for the Rad51 strand exchange protein, generating the 39 terminal filament that initiates invasion into the intact sister or homologous chromosome [3]. Rad51 association is followed by the binding of Rad54 and additional strand exchange proteins, Rad55 and Rad57.
Mechanistically, the initial steps are poorly understood. Our studies address the function of the MRX complex, which has ATP binding, ATPase, adenylate kinase, nuclease, and DNA endbridging activity [4,5]. Mre11 houses the nuclease activity in a domain containing five phosphodiesterase motifs homologous to the E. coli sbcD exonuclease. Mre11 has 39 to 59 exonuclease activity on double-stranded DNA and single-strand endonuclease activity on DNA hairpins [6][7][8]. Rad50 is an ATP binding protein, with the Walker A motif in the N terminus and Walker B motif in the C terminus separated by a long heptad repeat domain [9]. A signature (arginine finger) region in the C terminus is also part of the ATP binding motif [10]. Mutations in the ATP binding motif result in strains with meiotic recombination null phenotypes [9,11]. The Xrs2 protein (known as Nbs1 in mammals) contains a forkhead domain at the N terminus and a binding domain for Tel1 at the C-terminus. Tel1 is a kinase involved in telomere maintenance and checkpoint signaling. Mre11/Rad50/Nbs1 has weak ATP-stimulated DNA unwinding activity [12].
In bacterial recombinational repair, the strand invasion intermediate is generated by the RecBCD helicase/nuclease in a complicated reaction requiring 59 to 39 helicase, 39 to 59 helicase, 59 to 39 nuclease, 39 to 59 nuclease, and a specific DNA sequence called chi. A counterpart to this complex has not been identified in eukaryotic cells. The most highly radioresistant organism known, Micrococcus radiodurans, does not have a RecBCD helicase/nuclease, but does possess a complete RecF pathway [13]. A major question has been what pathway is found in eukaryotes. MRE11 is a nuclease involved in DSB repair and mre11D strains are extremely sensitive to irradiation. After 17 krads the survival of an mre11D strain is about 0.06%. If one assumes 2 krads results in 1 DSB per haploid cell [14], then a survival of about 0.02% would be expected in the absence of repair. Therefore, repair is nearly absent in a mre11D strain. mre11-D65N and mre11-H125N are more sensitive to IR than MRE11 strains but much more resistant than mre11D strains. At a high dose of 70 krads about 75 DSBs are induced, which would predict a survival value of e 235 or 10 213 % in the complete absence of repair. The survival of mre11-D56N and mre11-H125N mutants at 70 krads is about 5 to 10 fold less than MRE11 strains (3% survival), which is much higher than the predicted survival in the absence of DSB repair [15,16]. In addition, paradoxically, Mre11 has 39 to 59 exonuclease activity rather than the 59 to 39 nuclease expected for 59 resection [6][7][8]. Furthermore, resection is not reduced in mre11 mutants defective in the nuclease active site but proficient in the other functions of Mre11 [17,18]. This has led to the proposal that other nucleases are required and that they can compensate for the absence of the Mre11 nuclease. Sae2 is another nuclease that appears to be associated with the MRX complex, but also has 39 to 59 activity [19]. Exo1, a 59 to 39 nuclease of the Fen1 family involved in mismatch repair, may play a role. Overproduction of Exo1 indeed increases the IR resistance of an mre11D strain [20,21]. However, exo1D mre11D and exoD mre11-nuclease deficient double mutants are viable, are no more sensitive to IR than mre11D or mre11-nuclease deficient mutants, respectively, and can carry out gene conversion as efficiently in vivo as EXO1 MRE11 strains [20]. This suggests that yet another nuclease must compensate for Mre11 nuclease deficiency.
Dna2 is a candidate for this role. Dna2 is a 59 to 39 helicase, 59 to 39 exo/endonuclease, 39 to 59 exo/endonuclease, and has single strand DNA annealing and strand exchange activity [22][23][24][25][26][27][28]. Dna2 plays a role in Okazaki fragment processing (OFP), assisting FEN1 in RNA primer removal [26,29]. dna2 mutants, however, are extremely sensitive to bleomycin and IR, consistent with an additional role in DSB repair [30][31][32]. Also, we showed many years ago that dna2-2 rad50-5 mutants are synergistically defective in IR repair [31]. Finally, Dna2 forms a hub in a genetic network of 322 proteins that preserve genome stability [33]. Since the DNA2 gene is essential, however, it has been difficult to test the hypothesis that Dna2 nuclease participates in generation of the 39 overhang during DSB repair. Recently, we found that the inviability of dna2D mutants is efficiently suppressed by deletion of Pif1, a helicase orthologous to E. coli RecD helicase, that functions in DNA replication and telomere homeostasis [34,35]. Using a dna2D pif1D strain and mre11 nuclease defective mutants, we show that the nucleases of Dna2 and Mre11 are required in repair of IR-induced damage, presumably in 59 resection, and that they can compensate for each other's loss. In addition, we show that the helicase of Dna2 is dispensable but that the Sgs1 helicase is required for Dna2 nuclease to be able to compensate for lack of Mre11 nuclease.

Results
Dna2 nuclease can compensate for the loss of Mre11 nuclease in DSB repair and vice versa The relative importance of repair proteins can be measured by determining the survival of mutants defective in the respective proteins after increasing doses of a DNA damaging agent. We have used assays that measure survival after an acute dose of IR followed by growth in the absence of the DNA damaging agent. We were surprised that the dna2D pif1D mutant was resistant to Xrays ( Figure 1, the sensitivity of mre11D and mre11D pif1D are shown as controls), since several dna2 hypomorphic mutants are sensitive to IR [31,36]. This suggested that in the absence of Pif1, another protein might carry out the role of Dna2 in repair. Both hypomorphic dna2 mutants and the dna2D pif1D strain are synthetically lethal with mutants affecting three nuclease mutants defective in DSB repair, mre11D, sae2D, and exo1D [31,33,34]. We reasoned that the Mre11 nuclease might be compensating for the loss of Dna2 during X-ray repair in the dna2D pif1D strain.
To test the hypothesis that Mre11 and Dna2 had compensatory functions, we wished to compare the X-ray sensitivity of dna2D pif1D and dna2D pif1D mre11-nuclease defective strains. We first created dna2D pif1D mre11D strains carrying MRE11, mre11-D56N or mre11-H125N on centromeric plasmids. The D56N and H125N mutations lie in the conserved phosphodiesterase motif of Mre11, and the mutant proteins have no in vitro endonuclease activity [18,37]; however, they are able to form a complex with Rad50 and Xrs2. As shown in Figure 1. X-ray resistance of the dna2D pif1D strain compared to sensitivity of mre11D pif1D. The strains MB120-5A (WT), MB121-pif1D, MB161B-dna2D pif1D, MB122-17C-mre11D, MB124-2D-mre11D pif1D were harvested in mid-log phase, resuspended in water, irradiated, serially diluted (1:10), and plated. doi:10.1371/journal.pone.0004267.g001 Figure 2, both wild-type and nuclease-defective alleles of MRE11 suppress the lethality of dna2D pif1D mre11D strains. As predicted, however, the strains deficient in both Dna2 and the Mre11 nuclease are X-ray sensitive, while strains expressing wild-type Mre11 are resistant to IR. Thus, in the absence of Dna2, Mre11 nuclease is required for X-ray repair. By contrast, the X-ray sensitivity of pif1D mre11D, in which there is a functional Dna2 nuclease, is complemented not only by wild-type MRE11 but also by nuclease defective mre11-D56N or mre11-H125N. These results suggest that Dna2 nuclease can function in X-ray repair in the absence of Mre11 nuclease, but not in the complete absence of Mre11. It also says that Mre11 nuclease can function in X-ray repair in the absence of Dna2.
To insure that the X-ray sensitivities of the nuclease-dead mre11 mutants were not the result of inefficient expression due to employing extrachromosomal plasmids to express the mre11 alleles, the mre11 nuclease defective genes were integrated into the chromosome by two-step gene replacement. The dna2D pif1D mre11-D56N and dna2D pif1D mre11-H125N clones were as X-ray sensitive as the mre11D pif1D strain (compare Figure 3A and Figure 3B to Figure 1). Importantly, in both dna2D pif1D mre11-D56N and dna2D pif1D mre11-H125N X-ray sensitive clones, X-ray resistance is restored by introduction of a plasmid expressing either DNA2 or MRE11 ( Figure 3A,B). The helicase activity of Dna2 was not essential for complementation, since the X-ray resistance of the dna2D pif1D mre11-H125N clone was restored by plasmids expressing Dna2 with either of two mutations in the helicase domain: pRS314dna2K1080E plasmid, which has a mutation in the conserved Walker P loop but has a functional nuclease, and pRS314dna2R1253Q (dna2-2) ( Figure 3B). We conclude that it is the nucleolytic activities of Mre11 and Dna2 that are complementary. mre11-D56N and mre11-H125N mutants are not sensitive enough to DNA damaging agents to allow direct scoring in genetic crosses. Therefore, to compare the DNA2 PIF1 mre11 nuclease-minus with the dna2D pif1D mre11 nuclease-minus mutants, triple mutants were also constructed by introduction of a genetically tagged copy of the mre11 nuclease mutant genes into the chromosome, as described in the legend to Figure 4. dna2D pif1D MRE11 were X-ray resistant (see Figure 1). As before, dna2D pif1D mre11D::NatR::306 mre11-D56N::URA3 or dna2D pif1D mre11D::NatR::306 mre11-H125N::URA3 were highly sensitive to IR compared to the mre11D::NatR::306 mre11D56N::URA3 and the mre11D::NatR::306 mre11H125N::URA3 strains (Figure 4 A,B). Clearly, in the absence of Dna2 nuclease, Mre11 nuclease is essential for repair of IR-induced damage.

Sgs1 is Required for the Dna2-mediated Repair Pathway
Based on the above results, we hypothesized that Dna2 nuclease might play a role in resection of the 59 end at IR-induced breaks. Since the Dna2 helicase was not required, we asked what helicase might participate to convert the DSB into a substrate recognized by Dna2, which recognizes only single-stranded termini. Sgs1, which encodes a helicase of the RecQ family, was a likely candidate, since helicase defective dna2-2 mutants are defective in X-ray repair but the double sgs1D dna2-2 mutant is significantly more sensitive to IR than dna2-2, even though sgs1D itself is X-ray resistant [31,38]. Furthermore, we also recently demonstrated that human BLM, a RecQ family helicase, can suppress the MMS and . dna2D pif1D mre11D (pRS414mre11-D56N) and dna2D pif1D mre11D (pRS414mre11-H125N) strains exhibit similar sensitivity to X-rays as mre11D pif1D. These strains were created by dissecting tetrads from the following crosses: MATa dna2D pif1D trp1D bar1D/ MATa mre11D trp1D (pRS414MRE11), MATa dna2D pif1D trp1D bar1D/ MATa mre11D trp1D (pRS414mre11-D56N) and MATa dna2D pif1D trp1D bar1D/ MATa mre11D trp1D (pRS414mre11-H125N) diploids. Trp + spores were replica plated, and irradiated. Control strains mre11D pif1D (pRS414MRE11) and dna2D pif1D mre11D (pRS414MRE11) were identified from the respective cross and were X-ray resistant. dna2D pif1D mre11D (pRS414mre11D56N) or dna2D pif1D mre11D (pRS414mre11H125N), from the two respective crosses were radiation sensitive. To compare the radiation sensitivity of the dna2D pif1D mre11D (pRS414mre11-D56N) and the dna2D pif1D mre11D (pRS414mre11-H125N) (bottom panel) to mre11D pif1D (pRS414mre11-D56N) and to mre11D pif1D (pRS414mre11-H125N) strains (top panel), serially diluted cells were irradiated with 26 krads on a plate and allowed to grow for three days. Note that a dna2D pif1D mre11D strain carrying only the pRS vector is inviable, so an mre11D pif1D strain carrying the pRS vector alone is shown as the control, as indicated by the * in the bottom panel. doi:10.1371/journal.pone.0004267.g002 bleomycin sensitivity of dna2-2 mutants [39,40]. We therefore investigated whether yeast Sgs1 might be involved in the Dna2 repair pathway. We found, as have others, that sgs1D mutants are as resistant to IR as wildtype ( Figure 5). We then constructed an sgs1D mre11-H125N double mutant. As we predicted, the double mutant was as sensitive to IR as the dna2D mre11-H125N mutant (compare Figures 3 and 4 with Figure 5). We conclude that Sgs1 is required in order for Dna2 to compensate for the loss of Mre11 nuclease during repair of IR-induced damage.
To verify that an active Dna2 nuclease was necessary for IR repair in the absence of Mre11 nuclease, we compared the growth and X-ray sensitivity of dna2D pif1D strains transformed with centromeric plasmids carrying DNA2, dna2K1080E, and dna2E675A, or dna2K1080E,E675A. The dna2K1080E,E675A is doubly deficient in both helicase and nuclease. When the dna2D pif1D transformants were incubated at 23uC on tryptophan deficient plates all five transformants grew, as illustrated in Figure 6. At 30uC, the transformants containing the plasmids pRS314, pRS314DNA2, pRS314dna2K1080E (helicase minus, nuclease plus) formed viable colonies. Thus, pif1D suppresses the inviability of a dna2D PIF1 (pRS314dna2-K1080E) strain, and in fact the dna2D pif1D (RS314dna2-K1080E) strain even grows at 37uC, unlike the dna2D pif1D strain [34]. Interestingly, the pRS314dna2E675A transformant (helicase plus, nuclease minus) did not form colonies at 30uC ( Figure 6). However, the pRS314dna2K1080E,E675A (helicase minus, nuclease minus) did support viability at 30uC. We conclude that at this temperature the absence of the nuclease was deleterious to normal DNA replication, and furthermore that it is the Dna2 helicase activity that generates a need for the Dna2 nuclease.
Next, the dna2D pif1D transformants were irradiated after plating and incubated at 23uC. The pRS314, pRS314DNA2, pRS314dna2K1080E, and pRS314dna2K1080E,E675A transformants were resistant to IR, whereas the transformant containing the pRS314dna2E675A plasmid was sensitive to IR. Thus, as with Figure 5. Comparison of the X-ray sensitivity of strain sgs1D with sgs1D mre11-H125N and sgs1D sae2D. Strains MB137 sgs1D, MB138 sgs1D mre11-H125N and MB135 sae2D and MB136 sgs1D sae2D were constructed in this study and treated as in Figure 1. doi:10.1371/journal.pone.0004267.g005 Figure 6. Expression of nuclease deficient Dna2 is toxic in a dna2D pif1D strain. Strain MB21-dna2D pif1D carrying either pRS314DNA2, Hel + , Nuc + ; pRS314, Vector; pRS314dna2K1080E, Hel 2 Nuc + ; pRS314dna2E675A, Hel + Nuc 2 or pRS314dna2K1080E,dna2E675A, Hel 2 Nuc 2 plasmids was grown, serially diluted 10 fold and plated on trptophan deficient plates with or without 26 krads IR, and allowed to grow for four days at the temperatures indicated. doi:10.1371/journal.pone.0004267.g006 growth at the restrictive temperature of 30uC, after IR, the Dna2 nuclease minus strain is sensitive to damage, but only in the presence of an active Dna2 helicase. We conclude that the activities of the Dna2 helicase and nuclease must be properly coordinated for both DNA replication and repair.
The requirement for coordination of the Dna2 helicase and nuclease was further examined in the dna2D pif1D mre11-H125N mutant. The dna2D pif1D mre11-H125N (pGAL::DNA2) was transformed with the same set of Dna2 mutant plasmids used in Figure 6, to yield double transformants, dna2D pif1D mre11-H125N (pGAL::DNA2) (pRS414-dna2 mutant). Each double transformant was grown in tryptophan deficient liquid media to select for the pRS314dna2 mutant and to allow loss of the URA3 containing the pGAL::DNA2 plasmid. The resulting cultures were spotted on tryptophan-deficient or tryptophan-deficient media containing 5-FOA and grown at 23uC to determine the ability of the dna2 mutants to support growth. dna2D pif1D mre11-H125N containing the plasmids pRS314, pRS314DNA2, pRS314dna2K1080E grew, but strains containing the plasmids pRS314dna2E675A (nuclease minus) did not grow on 5-FOA-containing plates (Figure 7). In this Mre11-nuclease deficient background, there is also reduced viability of the transformant containing the pRS314dna2-K1080E,E675A (nuclease minus, helicase minus). We conclude that the viability of the Dna2-nuclease defective dna2D pif1D (pRS314dna2E675A) at 23uC shown above (Figure 6) depends on a functional Mre11 nuclease, suggesting their complementarity not only during IR repair but also during DNA replication or repair of replication errors. This is consistent with a model in which Dna2 helicase makes a potentially toxic structure during DNA replication that is preferentially and coordinately processed by the Dna2 nuclease, but that can also be processed by Mre11 nuclease. Inactivation of Dna2 helicase is sufficient to remove the requirement for nucleolytic processing in the absence Dna2 alone but is insufficient in the absence of both Dna2 and Mre11 nuclease. In the dna2D pif1D mre11-H12N mutant, the residual MRX helicase may now contribute to generating a structure needing nuclease processing.

Discussion
We have shown that the Dna2 59 to 39 nuclease can function in IR repair in the absence of an active Mre11 nuclease. The key result is that the dna2D pif1D strain is resistant to X-rays but the dna2D pif1D mre11-H125N mutant is as sensitive to X-rays as an mre11D strain, showing that in the absence of both the Mre11 nuclease and Dna2 nuclease, cells are nearly blocked in repair. Thus, strikingly, either nuclease can support repair efficiently ( Figure 3). The most direct mechanistic inference is that Dna2 and Mre11 are performing at least partially overlapping nucleolytic functions in processing the ends of DSBs. Since Mre11 has 39 to 59 nuclease rather than 59 to 39 activity in vitro, the question arises as to how it can perform 59 to 39 nucleolytic degradation in the absence of Dna2. Mre11 may achieve this in the absence of Dna2 by cleaving hairpin structures in ends that are unwound by an unknown helicase [8]. Another interpretation is that Mre11 nuclease and Dna2 function sequentially and that a second nuclease, exonuclease 1, can substitute for Dna2 nuclease, when Mre11 is present. In the absence of Mre11 nuclease, however, Dna2 but not Exo1, can overcome the requirement for the Mre11 nucleolytic step. We favor the latter interpretation (see below).
While our studies show that the Dna2 nuclease can compensate entirely, in terms of repair and survival, for the absence of Mre11 nuclease, Dna2 cannot compensate for the absence of Mre11. Mre11 is clearly required to form the MRX end processing complex.
The interaction between Mre11 and Dna2 at DSB damage is paralleled by the fact that Dna2, like Mre11, is found at telomeres, the linear ends of chromosomes that resemble DSBs [32]. The Mre11 nuclease is required during the de novo synthesis of telomeres at a DSB with a telomere seed sequence, though an mre11 nuclease deficient mutant does not have short telomeres [42]. Dna2 also appears to be required for extending a short telomere, and its nucleolytic function may be at play [32,43,44]. Interestingly, dna2D pif1D mutants have short telomeres compared to pif1D mutants [34].
Unexpectedly, our study has also supplied genetic evidence that supports biochemical studies suggesting that the Dna2 helicase and nuclease activities are coupled. dna2D pif1D expressing a Dna2-E675A (helicase plus, nuclease minus) protein have a reduced maximum permissive growth temperature and they are sensitive to IR. However, dna2D pif1D expressing a helicase minus, nuclease minus double Dna2 mutant protein has no growth or repair defect, i.e. inactivation of the helicase removes the requirement for the nuclease (Figure 6). This suppression occurs only if Mre11  pif1D mre11-H125N strain. A dna2D pif1D mre11-H125N trp1D (pGAL::DNA2URA3) strain was transformed with the Trp + plasmids pRS314DNA2, Hel + , Nuc + ; pRS314, Vector; pRS314dna2K1080E, Hel 2 Nuc + ; pRS314dna2E675A, Hel + Nuc 2 or pRS314dna2K1080E,dna2E675A, Hel 2 Nuc 2 . The Trp + Ura + transformants were grown in tryptophan deficient media and spotted onto tryptophan minus (left) or tryptophan minus, 5-FOA containing (right) media at 23uC and photographed after 4 days at 23uC. doi:10.1371/journal.pone.0004267.g007 nuclease is functional (Figure 7). Therefore, we propose that the Dna2 helicase creates a structure during DNA replication and DNA repair that is potentially toxic unless coordinately processed by the nuclease of either Dna2 or, in its absence, Mre11. Further, the observed lethality at all temperatures and in the absence of DNA damage in mre11-nuclease minus dna2D pif1D expressing nuclease defective Dna2 leads to the hypothesis that Mre11 nuclease may play a role in Okazaki fragment processing or in repair of lethal endogenous DNA damage. A role for Mre11 in OFP is consistent with previous results showing that mre11-H125N rad27 mutants are inviable [18].
Also worth emphasizing is the fact that our findings suggest an additional role for Dna2, besides 59 resection, in X-ray repair. dna2D pif1D strains are as resistant to IR as wildtype (this work), whereas dna2D PIF1 overexpressing FEN1 to maintain viability is as X-ray sensitive as rad52D [31]. Therefore, during IR repair, Pif1 may contribute to formation of a structure that is lethal unless processed by Dna2, and this function is not complemented by MRE11, EXO1, or SGS1. Furthermore, a dna2-2 PIF1 mutant (nuclease plus, helicase defective) is 100 times more sensitive to IR than DNA2 PIF1 (at 75 krads) [31]. However, the dna2-2 gene complements the X-ray sensitivity of the dna2D pif1D mre11-H125N strain ( Figure 3B). These latter two results appear at odds but could be explained if Dna2-2 protein can perform resection but not some other function necessary for X-ray resistance in the presence of PIF1, either during resection or during strand exchange.
Mechanistic studies that were published after our work was completed show that dna2 mutants as well as sgs1 mutants exhibit significantly delayed long-range resection in a Rad51-independent single-strand annealing (SSA) repair reaction at an enzymaticallyinduced DSB [45,46]. Based on these studies and the enzymatic activities of Dna2 and Sgs1, we propose that our results imply an interaction between Mre11 nuclease and Dna2/Sgs1 during resection in Rad51-dependent homologous recombination. At the same time, however, our new results also raise a conundrum. The reduced exonuclease resection in the dna2D pif1m2 mutant observed in the SSA studies does not appear to cause a defect in subsequent Rad51-dependent repair [46], since we show that the dna2D pif1D strain is as resistant to IR as a wild-type strain after 70 krads IR (35 DSB or 2 per 17 chromosomes). Similarly, sgs1D strains, though they show reduced SSA, are as resistant to IR as wild-type strains, even after 75 krads. Thus, the mechanisms at work during the longrange resection in SSA may not be entirely interchangeable with the resection pathway leading to strand exchange. In this regard it is interesting that Exo1, another 59 to 39 nuclease, appears able to compensate for Dna2 or Sgs1 deficiency in SSA [45,46], but does not appear to be able to do so in IR repair, since dna2 mre11-H125N and sgs1 mre11-H125N mutants are as sensitive to IR as mre11D, even though Exo1 is present at normal levels and since mre11 nuclease-defective exo1 double mutants are not significantly more sensitive to IR than mre11 nuclease deficient mutants. Previous studies showed that only 100-300 bp of homology are required for Rad51dependent gene conversion [47,48], which differs from SSA in that gene conversion requires strand invasion of a homologous sister chromatid or homolog. We propose that the Dna2/Sgs1 pathway is the major pathway of resection in homologous recombination.
The observed IR sensitivity of the sgs1 mre11-H125N mutant we report here, combined with the defect in resection in sgs1D demonstrated by others, may help explain both some of our previous and our current results. We have shown that helicase defective dna2-2 mutants are very defective in X-ray repair, but the double sgs1D dna2-2 mutant is significantly more sensitive to IR than dna2-2, even though sgs1D itself is X-ray resistant [31]. Thus, Dna2 may compensate for the absence of Sgs1 helicase, but Sgs1 does not fully compensate for a defective Dna2 helicase. We have previously shown that BLM, a human counterpart of Sgs1 helicase, complements the MMS sensitivity of dna2-2 mutants as well as the lethality of dna2-1 mutants [39], and this suppression can now be attributed to suppression of defects in DSB processing. We expect that overexpression of BLM should also suppress the IR-sensitivity of dna2-2 mutants. Genetic evidence suggests Pif1 helicase is an inhibitor of the Sgs1 helicase [49]. Another possible role for Dna2 may be to counteract Pif1 inhibition of Sgs1 during the 59 to 39 processing step.
What do our results tell us about the role of Dna2 in endogenous damage repair? We have previously identified the DNA replication fork pause at the replication fork barrier (RFB) in the ribosomal DNA as a specific site of endogenous DNA damage requiring the combined activities of Dna2 and Sgs1 for repair. In the presence of Fob1, DNA replication forks pause at the RFB and a DSB arises, the occurrence of which is elevated in dna2-2 and sgs1D mutants [40,50]. Interestingly, Pif1 is also required for normal levels of pausing at the RFB and subsequent chromosomal breakage [51]. These endogenous DSBs and the inability to repair them contribute to the synthetic sickness/lethality in the dna2-2 sgs1D strain, since the sickness/lethality is suppressed by the fob1D mutation [40,50]. A similar situation occurs with Ctf4, a replication/cohesion protein, in which the dna2-2 ctf4D mutant is lethal, but the dna2-2 ctf4D fob1D is viable and slow growing. The dna2-2 ctf4D fob1D mutant is significantly more sensitive to IR than either dna2-2 fob1D or ctf4 fob1D mutants [33]. Presumably dna2-2 ctf4D inviability also results from failure to repair endogenous DSBs at the RFB. Thus, Dna2 plays a role in repairing endogenous breaks in collaboration with Sgs1 and Ctf4 and this is likely a reflection of the role of Dna2 and Sgs1 in resection. It is probably this role of Dna2 that accounts for the remarkable observation that increasing the gene dosage of Dna2 by one copy leads to partial genome stabilization and is absolutely required for viability of a mec1-21 strain [52].
In conclusion, while our results and those of Zhu et al. [46] clarify the multiple events occurring early in repair at DSBs, especially the previously unappreciated centrality of Dna2, they also highlight several fundamental questions for the future: First, how much resection is sufficient for repair and what is the ratelimiting step in repair? Second, what is the mechanism of resection per se? How do Mre11 and Dna2, which may form a link between DNA replication and repair, coordinate repair of endogenous DSBs occurring due to replication fork failure?

Strains and Plasmids
The plasmids pRS314DNA2, pRS314dna2K1080E, pRS314dna2E675A, and pRS314dna2K1080E,E675A have the 6 kb EcoRI fragment containing Dna2 or Dna2 with the indicated mutations cloned into pRS314CEN TRP at the EcoR1 site. The construction of the Dna2 containing fragment and further construction of the mutants was described previously [25]. The plasmid pSEY18GALDNA2, referred to herein as pGAL::DNA2, was described previously [25], except that in this work the DNA2 gene carried an additional FLAG tag, MDYKDDDK, at the N terminus of the protein.
The plasmids pRS414MRE11, pRS414mre11-D56N, and pRSmre11-H125N (gift of K. Lewis, Texas State University, San Marcos, TX) contain the MRE11 genes with about 800 bp 59 to the ATG and 136 bp 39 to the TAG contained on a Kpn1 Not1fragment. The KpnI NotI fragments containing mre11-D56N and mre11-H125N were cloned into the KpnI NotI site of the integrating plasmid, pRS306. The pRS306mre11-D56N and pRS306mre11-H125N plasmids were cut with SphI and transformed into yeast.
All MRE11 and DNA2 alleles used in this study are expressed under the control of their native promoters.
Strains used are described in Table 1.

X-ray Treatment
Cells were treated with X-rays as previously described using a Pantak MKII X-ray machine operated at 20 mA and 70 kev [31]. The dosimetry was determined using a Radca l ; 0.6 cubic centimeter ion chamber (model 10X5-0.6, serial no. 9352) connected to a Radiation Monitor Controller (model 9010; serial no. 90-2910).