We found that the srs2Δ
mutant has several novel synthetic lethal interactions, which include the mrc1Δ
, and csm3Δ
mutations; these genes encode proteins that are proposed to form a complex that acts during DNA replication at the replication fork. Mrc1 and Tof1 have been shown directly to be associated with replication forks (17
), and Csm3 has been shown to interact with Tof1 by yeast two-hybrid analysis and by coimmunoprecipitation (15
). Mrc1, Tof1, and Csm3 have been shown to be required for DNA damage checkpoint activation in response to replication blocks (2
). Although Srs2 is required for full Rad53 activation and regulation of S-phase progression in response to induced intra-S damage, Srs2 is not essential for a normal S phase, and there is no spontaneous activation of Rad53 (27
). Thus, it is unlikely that the srs2Δ mrc1Δ
synthetic lethality is due to a failure to activate Rad53 under normal growth conditions.
Moreover, an srs2Δ mutant does not require Rad53 for survival in undamaged conditions and the mrc1Δ lethality with srs2Δ can be separated from the function of Mrc1 in the checkpoint DNA damage signaling. The role of Srs2 during S phase appears to be the promotion of gap repair by the RAD18 postreplication repair pathway, through mechanisms that do not involve homologous recombination and double-strand break formation. In the absence of Srs2, more gaps are repaired by the homologous recombination pathway. We do not know if the homologous recombination events are promoted by double-strand breaks or initiated by single-strand gaps. It is also possible that single-strand gaps are processed to double-strand breaks once a commitment to homologous recombination is made by having a Rad51 filament form on the single-strand gap. Whatever the case, we believe that double-strand breaks are not formed in the srs2Δ mrc1Δ rad51Δ mutant, as this mutant does not show any cell cycle arrest or reduced growth compared to mrc1Δ, srs2Δ, or rad51Δ single mutants.
We propose that the spontaneous lesions in the srs2Δ mrc1Δ mutant that stall replication forks are channeled into a homologous recombination repair pathway instead of a gap-filling pathway due to loss of the Srs2 antirecombinase action against Rad51. Normally, this would be tolerated, but when Mrc1 is also absent, this becomes a lethal situation. The lethality arises from attempting homologous recombination without the correct molecular scaffold set up at the point of replication stalling. We suggest that this scaffold is sister chromatid cohesion that is established specifically at a point of fork stalling. Sister chromatid cohesion may stabilize the fork and may promote sister chromatid recombination, which would be necessary to reestablish a replication fork if it becomes collapsed by a double-strand break. The increased spontaneous deletion rate that we observed in the mrc1Δ mutant but not the mrc1-AQ mutant (Fig. ) indicates that more double-strand breaks form in the absence of the Mrc1 protein. Although this level of double-strand breaks can be tolerated when Srs2 is present (although it probably accounts for the slower growth of the mrc1Δ deletion mutant), in its absence this becomes lethal.
We suggest that the role of Mrc1 and its functionally associated partners Tof1 and Csm3 is to recruit the Ctf18-RFC-like complex to the site of fork stalling. The Ctf18-RFC-like complex would then recruit or activate a cohesion complex as replication occurs (29
), even if the newly replicated DNA is gapped. This model would account for the srs2Δ
synthetic lethality with ctf18Δ
and components of the Ctf18-RFC-like complex as well as synthetic sickness with structural maintenance of chromosome (SMC) mutants defective in condensin and cohesin subunits (31
). In this model, the double-strand breaks would not inhibit S-phase progression, and indeed double-strand breaks are not inhibitory to S-phase progression (A. Pellicioli and M. Foiani, personal communication). Rather, when they are engaged by the homologous recombination pathway in G2
, sister chromatid cohesion is required for their repair (39
Inactivating homologous recombination rescues the srs2Δ mrc1Δ synthetic lethality because the homologous recombination pathway is no longer active and cells can now use the gap-filling pathways to repair the single-strand gaps. Thus, the postreplication repair pathway becomes a mechanism to avoid inappropriate homologous recombination events that may result in genome instability.
Srs2 and Mrc1 are involved in the S-phase checkpoint pathway (2
). The sister chromatid cohesion factor SMC1 has also been linked to a branch of the S-phase checkpoint pathway involving ATM and NBS1 in mammalian cells (19
). Indeed, the yeast smc1-259
mutation has been shown to affect DNA damage response, causing sensitivity to ionizing radiation, UV, and MMS treatment (19
). However, the mammalian ATM serine target sites in SMC1 are not present in the yeast Smc1 protein, so it is not known if Smc1 of S. cerevisiae
is involved in the S-phase checkpoint pathway.
It is possible that the srs2Δ mrc1Δ
synthetic lethality arises from loss of two checkpoint signaling effectors that converge on Smc1. We do not believe that this is the explanation for the synthetic lethality. First, mrc1Δ
shows a synthetic sickness, not lethality, with rad50Δ
but not with mre11Δ
). Rad50, Xrs2, and Mre11 form a complex in S. cerevisiae
), and Mre11 is the key player in checkpoint signaling (7
). Second, in mammalian cells, the SMC1-dependent S-phase checkpoint activation does not affect the ability of SMC1 to bind to chromatin (55
), and both phosphorylated and unphosphorylated SMC1 appear to be competent for sister chromatid cohesion (55
). Third, loss of the Mec1 checkpoint signal does not rescue the srs2Δ mrc1Δ
growth defect. In this situation, the cells can grow for one to two additional generations but still cannot sustain further growth. Fourth, specific loss of the Mrc1 checkpoint signaling function does not lead to lethality in an srs2Δ
Thus, we propose that sister chromatid cohesion is set up at a site of DNA damage when the cell engages in mitotic homologous recombination in the context of replication stalling. We do not know whether Mrc1 interacts with homologous recombination components or the Ctf18-RFC-like complex, nor do we know when homologous recombination occurs, in S phase or G2, when gap repair is inhibited. Nonetheless, our results highlight the importance of the postreplication repair pathway in maintaining cell survival and repair of spontaneous lesions by the correct pathway to avoid genome instability and lethal homologous recombination events.