During DNA replication, the extension of daughter strands is continuously impaired by a number of factors, such as proteins bound to the template, endogenously or exogenously induced DNA damage, and the presence of DNA secondary structures. If the replication fork stalls, and if the stalled fork is not processed to restore fork progression, disassembly of the replication complex can ensue. The stalled forks can also break, generating a double-strand break (DSB). Additionally, the presence of a DNA lesion, such as a single-strand nick in the template strand, can lead to a DSB. Consequently, a failure to repair the replication-associated lesions, and to then restart the stalled fork, will lead to chromosome loss or impairment of the integrity of the genome. Maintenance of the stability of the genome is critical for normal cell growth and cell viability.

To avoid genetic instability, cells have evolved a variety of mechanisms to rescue the stalled fork; extensive studies in both prokaryotes and eukaryotes suggest that homologous recombination plays a critical role in repair of the replication-associated DNA lesions, and in allowing the replication to continue 123. For example, DSBs arising as a result of replication defects can be repaired by homologous recombination, using the sister chromatid as a template. Similarly, a replication fork stalled due to the presence of a replication block can be reinitiated by a template-switching mechanism, before the replication block is removed. However, unscheduled recombination can be detrimental, leading to a higher rate of genetic instability, as observed in the cancer-prone Bloom, Werner, and Rothmund-Thomson syndromes, respectively due to mutations in the BLM, WRN, and RECQL4 genes 4. These three genes belong to a highly conserved family of RecQ DNA helicases, originally described in Escherichia coli as a component of the RecF recombination pathway 45.

BLM cells show a high rate of sister-chromatid exchange (SCE), and the sensitivity of both BLM and WRN cells to S-phase-specific inhibitors (e.g., camptothecin) suggests that these genes function during DNA replication 4. In addition, there is mounting evidence in yeast suggesting that replication does not proceed normally in the absence of RecQ helicases. Cells lacking the RecQ homolog Sgs1 in Saccharomyces cerevisiae exhibit an increased sensitivity to DNA-damaging agents (e.g., ultraviolet light, hydroxyurea, and methyl-methane sulphonate); an increased level of recombination between homologous sequences and between modestly divergent DNA sequences; gross chromosomal rearrangements; unequal SCE; and mitotic chromosome non-disjunction 6789101112. The Sgs1 protein closely associates with the replication fork and is thought to stabilize and restart the stalled fork 131415. In vitro studies have indicated that Sgs1, like its human counterpart, is a 3'-5' DNA helicase that can disrupt a variety of DNA structures, including cruciform structures that resemble the Holliday junction intermediate of the recombination process, suggesting its possible role in homologous recombination 16. Sgs1 physically interacts with type I topoisomerase I (Top3), and both genetic and biochemical studies indicate that the Sgs1/Top3 complex acts on Holliday junctions to suppress crossover outcomes 17181920.

Several synthetic lethal screens have been employed to identify the genes that are functionally related to Sgs1 21222324252627. The sgs1 mutation is synthetically lethal with a mutation in the SRS2 gene, which encodes another 3'-5' DNA helicase 26. Cells lacking Sgs1 and Srs2 are extremely sick, and the growth defect is suppressed by a mutation in any of the RAD51, RAD52, RAD55, and RAD57 genes involved in early stages of homologous recombination 21242728. Since the Sgs1 and Srs2 proteins function during DNA replication 429, it has been proposed that Sgs1 and Srs2 in wild-type cells regulate the accumulation of toxic recombination intermediates, during DNA replication. In vitro studies have shown that Srs2 possesses an anti-recombination activity; it displaces Rad51, a strand-annealing protein, from DNA filaments 3031, which is in agreement with Srs2's in vivo recombination-inhibiting activity 32.

The sgs1 mutation is also synthetically lethal with mus81, but a rad51 mutation suppresses the lethal effect of the double mutation 2224, suggesting that Sgs1 and Mus81 function in separate pathways. Mus81 acts in a complex with Mms4; the heterodimeric protein has been shown to cleave branched DNAs and has been implicated in DNA repair; it also functions during sporulation 333435. Results of several genetic studies have led to the proposal that DNA structures formed during replication are acted upon by recombination proteins, forming intermediates that are toxic unless processed further. Sgs1/Top3 and Mus81 are needed to process these intermediates, whereas Srs2 limits the numbers of such intermediates, by disrupting Rad51 filaments.

In eukaryotes, SCE occurs spontaneously, probably representing recombination events during replication. The factors that impair the normal progression of the replication fork are likely to increase the rate of spontaneous SCE. One of the factors that compromises the normal progression of the replication fork is the presence of inverted repeats (IRs) that can form secondary structures in single-stranded DNA, by intra-strand base pairing between complementary sequences. Consistently, IRs have been found to be associated with gross chromosomal rearrangements 3637. Previously, we constructed a recombination substrate to study the effect of IRs on unequal SCE in haploid S. cerevisiae 38. The presence of the repeated sequences increases spontaneous unequal SCE by about 10-fold 3839. While non-IR-mediated SCE events are independent of DSB-repair genes, IR-stimulated SCE events depend on DSB repair genes, suggesting that IR-associated SCEs occur by DSB repair 38.

During DNA replication, the lagging strand is expected to contain a higher level of single-stranded regions than does the leading strand, due to the discontinuous nature of DNA synthesis. The single-stranded regions facilitate the formation of a secondary structure at an IR. The secondary structures are also substrates for structure-specific nucleases in vivo. Cleavage of the secondary structure at the stalled fork will lead to the formation of a DSB that can be repaired by either gene conversion or break-induced replication 1, using the sister chromatid as a template. In the present study, we analyzed IR-stimulated unequal SCE in cells lacking Sgs1 and/or functionally related enzymes that are believed to function at the stalled replication forks. Our results showed that the sgs1 mutation increases unequal SCE for both IR-containing and non-IR-containing substrates. However, IR-stimulated SCE events in the sgs1 background are significantly reduced when defects in the mismatch repair (MMR) gene MSH2 are also present. Additionally, we showed that SGS1 and EXO1 regulate SCE in two distinct pathways.

Inverted repeats that have the potential to form secondary structures provide an excellent system in which to study the consequences of replication block due to the presence of secondary structures. A replication block at the secondary structure may cause disassembly of the replication complex, exposing the newly synthesized strand, which then becomes the recruiting center for the recombinational proteins. Alternatively, an endonucleolytic cleavage of the secondary structure can result in the formation of a DSB. In mitotic cells, the sister chromatids are preferentially used for recombination repair 52. Accordingly, the presence of an IR increases the rate of spontaneous SCE; these SCE events occur by DSB repair 38. In this study, we analyzed the effect of the sgs1 mutation and of mutations in genes that are functionally related to SGS1 on IR-associated spontaneous unequal SCE.

Sgs1 is functionally related to Srs2. Results from several genetic studies have suggested that Sgs1 and Srs2 deal with toxic recombination intermediates that arise during normal DNA replication by two separate mechanisms 24. While Sgs1 resolves the toxic recombination intermediates, Srs2 limits the formation of such intermediates. A null mutation in either SGS1 or SRS2 is synthetically lethal with a mutation in several genes involved in DNA replication 25, and both SGS1 and SRS2 are implicated in the intra-S damage checkpoint mechanism 1429. The rates of spontaneous SCE events for both substrates in the srs2 background were increased. However, the fact that the rates were about 4–5 fold lower than those observed in the sgs1 background (Table 2) suggests that replication defects in srs2 cells are either repaired by Sgs1 or channeled into alternative repair pathways. For example, the otherwise toxic intermediates can be processed by the action of another helicase, such as Rrm3 5354.

Mus81 is believed to act in pathways parallel to those involving Sgs1 to resolve recombination intermediates generated during DNA replication 24. The mus81 mutation exhibited a modest effect on SCE, suggesting that in mus81 cells, intermediates that are normally metabolized by Mus81 are either resolved by Sgs1 or else are repaired by pathways not involving SCE. It should be noted here that repair of a replication defect by SCE can occur by either equal or unequal SCE. Equal SCE is the predominant DSB repair mechanism 55. In our system, we can detect only the unequal events. Therefore, we cannot rule out the possibility that the modest effect of the srs2 and mus81 mutations on unequal SCE is due to repair of replication defects by equal SCE.

The nature of the substrate that is metabolized by Sgs1 is not clear. During replication, DSBs are normally not generated in wild-type cells; recombination events are likely to be initiated on single-stranded gaps that form at stalled replication forks 24. Accordingly, spontaneous SCE events are independent of genes involved in DSB repair; the rate of SCE remained close to the wild-type level in rad51 cells 3856. The IR-associated SCE events are, however, dependent on the DSB-repair enzymes, because IR-associated SCEs occur by DSB repair (38; Fig. 1). Since recombination enzymes function upstream of Sgs1, we expected spontaneous SCE to be reduced to the rad51 level, in the sgs1 rad51 double mutant. The spontaneous SCE rates in the sgs1 rad51 background for both substrates were reduced in the double mutant, but they nevertheless remained higher than the rate in the rad51 cells (Table 2). Similar results were also obtained by Spell and Jinks-Robertson 11, who found that homologous recombination in the rad51 sgs1 cells was reduced relative to the rate in sgs1 single mutant, although it remained higher than the wild-type level. These results suggest that both Rad51-dependent and Rad51-independent SCE events occur in the sgs1 background. In S. cerevisiae, some homologous recombination events are RAD51-independent but RAD59-dependent, but both types of event are RAD52-dependent 28, suggesting that Rad51-independent SCE events can occur by the Rad59 pathway.

The above conclusion, that various mechanisms operate in SGS1 cells to suppress spontaneous SCE, was supported by the result that 50% of the SCE events observed for his3-SCS_pal140 in the sgs1 mutant are MSH2-dependent (Table 2). The role of Msh2 in generating IR-associated SCE events is not known. Myung et al. observed a reduction by slightly over 2-fold in the rate of homologous recombination with an IR substrate in the sgs1 msh2 double mutant 10. In a separate study, Spell and Jinks-Robertson 11 found no difference in the rate of homologous recombination between the sgs1 single mutant and the sgs1 msh2 double mutant. However, neither of these results is directly comparable with our findings, because the substrate used by the two groups was not a perfect IR, but was interrupted by non-repeated sequences.

Only MSH2 and MSH3 of the MMR pathway are necessary for IR-associated SCE events 39. IR-associated SCE events occur by DSB repair. If DSBs occur within the secondary structures, then the 3' end, after DSB formation and exonucleolytic processing, will contain non-homologous tails that must be removed for generating a wild-type HIS3 gene by unequal SCE. The Rad1/Rad10 endonuclease aided by the Msh2/Msh3 complex removes the non-homologous ends. The rate of SCE remains unaffected in the rad1 background (39), suggesting that non-homologous tails are removed by a mechanism that does not involve the Msh2/Msh3 complex. The Msh2/Msh3 complex is involved in loop repair 45. The Msh2/Msh3 complex may recruit the processing enzyme to the stem-loop structure after DSB formation, or else Msh2/Msh3 could bind at the secondary structure, and then recruit the endonclease to generate the DSB.

These observations raise another issue: what enzyme is responsible for generating the break at the secondary structure? IR-associated SCE events are reduced in the rad50 and mre11 backgrounds 38. Rad50 and Mre11 are components of the Mre11-Rad50-Xrs2 (MRX) complex that is required for DSB formation during meiosis, both at normal meiosis-specific sites and at IRs 2857. In mitotic cells, the MRX complex is required for DSB repair and for maintenance of the genome integrity 28. Rad50 and Mre11 respectively show significant homology with the SbcC and SbcD proteins of E. coli 58. The SbcCD complex is known to cleave hairpin structures 59. Mre11 has been shown to cleave hairpin structures in vitro 6061. Therefore, one likely scenario is that the DSBs at the repeated sequences are generated via Mre11's hairpin cleavage activity. Our results on meiotic DSB formation in mre11-nuclease deficient cells (unpublished observation), and the results obtained by Resnick and his coworkers 62, suggest that Mre11's nuclease activity is not required for formation of DSBs during either meiosis or mitosis, but that it is necessary for processing of the DSBs at the secondary structure.

Exo1, like Sgs1, has been shown to act on the stalled replication fork 50. The exo1 mutation is synthetically lethal with rad27, and overexpression of EXO1 suppresses several rad27 defects, suggesting that EXO1 is functionally redundant with RAD27 (FEN1) for Okazaki fragment processing 51. However, the role of EXO1 in Okazaki fragment processing is unclear, because exo1 cells do not exhibit any growth defects or other phenotypes exhibited by the rad27 cells 51. Perhaps, the observed increase in the rate of SCE in exo1 cells is due to inefficient processing of the stalled replication fork. The SCE rates in the sgs1 exo1 double mutant were synergistically increased for both substrates (Table 2), suggesting that SGS1 and EXO1 regulate spontaneous SCE by independent mechanisms. However, further studies are necessary to understand the underlying mechanisms of each of these two pathways. It is also not clear from the available data whether the increased level of SCE in exo1 cells occurs by the same mechanism for the two substrates. These events may occur by separate mechanisms, analogous to our results in sgs1 cells.

IRs have the potential to form secondary structures, which are known to attenuate the normal progression of the replication fork. A block in the progression of the replication fork is likely to increase the rate of SCE. In this report, we studied the effects of mutations in SGS1 and in functionally related genes on IR-stimulated spontaneous unequal SCE. We conclude that 1) in wild-type cells, both IR and non-IR-associated spontaneous SCE events are suppressed by Sgs1; 2) the IR-associated SCE events in the sgs1 background are partially MSH2-dependent; 3) the increased level of SCE events in the sgs1 background arises via both RAD51-dependent and RAD51-independent pathways; however, most of the SCE events in the sgs1 background are RAD52-dependent; and 4) Sgs1 and Exo1 regulate spontaneous SCE events by independent mechanisms.

IR: inverted repeat; DSB: double-strand break; SCE: sister-chromatid exchange, CSM: complete synthetic medium; SCS: sister-chromatid recombination substrate.

DKN designed the experiments, SJC and DKN performed the experiments and analyzed the data, DKN wrote the paper. SJC read and approved the final manuscript.