Mismatch repair, but not heteroduplex rejection, is temporally coupled to DNA replication

Abstract

In eukaryotes, it is unknown if mismatch repair (MMR) is temporally coupled to DNA replication and how strand-specific MMR is directed. Here we fused Saccharomyces cerevisiae MSH6 with cyclins to restrict the availability of the Msh2-Msh6 mismatch recognition complex to either S-phase or G2/M. The Msh6-S cyclin fusion was proficient for suppressing mutations at three loci that replicate at mid-S-phase, whereas the Msh6-G2/M cyclin fusion was defective. However, the Msh6-G2/M cyclin fusion was functional for MMR at a very late-replicating region of the genome. In contrast, the heteroduplex rejection function of MMR during recombination was partially functional during both S-phase and G2/M. These results indicate a temporal coupling of MMR, but not heteroduplex rejection, to DNA replication.

DNA mismatch repair (MMR) recognizes mispairs in DNA caused by DNA replication errors, and targets the newly synthesized DNA strand for repair by an excision and resynthesis mechanism (1-3). MMR repair is initiated by the recognition of the mispair by the MutS homodimer in bacteria and the MutS-related Msh2-Msh6 and Msh2-Msh3 heterodimers in eukaryotes, which possess differing mispair-binding specificities (4, 5). Identification of the newly synthesized DNA strand in Escherichia coli MMR is mediated by nicking of the newly synthesized, unmethylated strand by MutH at hemi-methylated GATC sites in response to a mispaired base (2). The critical hemi-methylated sites are generated by replication of methylated chromosomal DNA and provide a temporal post-replication window for repair that closes when the Dam methylase modifies the newly synthesized DNA strand. Whether MMR in eukaryotes is coordinated with DNA replication is less clear. The replication clamp PCNA functions in MMR (6) and both Msh6 and Msh3 contain N-terminal binding sites for PCNA providing the potential for coupling of mismatch recognition to the replication machinery (7, 8). Despite this, msh6 and msh3 mutations that specifically eliminate PCNA binding cause very weak MMR defects (7-9), which in the case of Msh6 reflects the functional redundancy of the Msh6 PCNA binding site with other regions of the Msh6 N-terminus whose roles are not yet understood (9). To directly test temporal coupling of MMR with DNA replication, we restricted the availability of the Msh2-Msh6 mispair recognition complex to specific stages of the cell cycle of S. cerevisiae msh3Δ mutants.

Functional Msh6 fusions with controlled cell cycle expression were generated by fusing different cyclins to the Msh6 N-terminus at the MSH6 chromosomal locus and replacing the MSH6 promoter with the cell cycle-regulated cyclin promoter (Figure 1A). A fusion of MSH6 with a fragment of CLB6 encoding the first 195 residues under control of the CLB6 promoter, S-MSH6, restricted Msh6 expression primarily to between 20 and 50 minutes after release from alpha-factor arrest (Figure 1B). This expression period corresponded to S-phase as demonstrated by FACS (Figure 1B). We also constructed a fusion of MSH6 to a fragment of CLB2 encoding the first 181 residues under control of the CLB2 promoter and encoding the L26A mutation that prevents nuclear export (10) similar to the CLB2 fusions used to study post-replication repair (11). This fusion, G2/M-MSH6, restricted Msh6 expression to primarily after 40 minutes after release from alpha-factor arrest, which was during G2/M-phase based on FACS analysis (Figure 1B). We were unable to construct a G1 phase restricted MSH6 gene that expressed sufficient Msh6 to allow testing on MMR in G1.

Figure 1.

A. Diagrams of tagged MSH6 constructs. B. (Top) Expression level of Msh6 revealed by Western blots of whole cell lysates with anti-Myc antibodies for log-phase cells (log), alpha-factor arrested cells (α-F), and cells released from alpha-factor arrest for the indicated times. An anti-Pgk1 blot is shown as a loading control. (Bottom) FACS profiles of cells showing the cell cycle distribution at the indicated time points. The FACS profile displayed is for the MSH6-9MYC msh3Δ strain; however, all of the strains used had essentially identical profiles.

We introduced the S-MSH6 and G2/M-MSH6 fusion constructs into a strain containing an msh3Δ mutation to eliminate any repair mediated by the partially redundant Msh2-Msh3 heterodimer (4, 5). We then measured the rates of reversion of the hom3-10 and lys2-10A frameshift mutations and inactivation of the CAN1 gene (Figure S1A), which are elevated in MMR-defective mutants. The mutation rates of the S-MSH6 msh3Δ strain were substantially lower than the fully MMR-defective msh6Δ msh3Δ strain in all three MMR assays, indicating that this fusion construct was functional in vivo (Table 1). The mutation rates of the S-MSH6 msh3Δ strain were, however, slightly higher than that of the msh3Δ strain, possibly due to the reduced expression of S-Msh6 relative to Msh6 (Figure 1B). In contrast, the mutation rates of the G2/M-MSH6 msh3Δ strain were equivalent to that of the MMR-defective msh6Δ msh3Δ strain in all three MMR assays (Table 1), indicating that restricting Msh6 expression to G2/M-phase prevented MMR at the loci tested.

Table 1. MMR repair activity during S and G2/M stages of the cell cycle.

		Mutation rate (fold increase)*
Relevant genotype	Strain RDKY	Thr +	Lys +	CanR
msh3 Δ	7672	5.1 [3.4-6.4] × 10−8 (1)	7.8 [6.0-11.3] × 10−8 (1)	8.0 [6.2-16.8] × 10−8 (1)
msh3Δ msh6Δ	7674	5.5 [3.6-8.6] × 10−6 (108)	3.3 [2.2-4.6] × 10−5 (423)	2.4 [1.6-3.8] × 10−6 (30)
G2/M-MSH6 msh3 Δ	7676	4.3 [3.3-5.5] × 10−6 (84)	2.9 [1.7-4.7] × 10−5 (372)	2.9 [2.1-4.3] × 10−6 (36)
S-MSH6 msh3 Δ	7678	6.9 [4.8-8.9] × 10−7 (13)	2.6 [1.8-4.0] × 10−6 (33)	4.1 [3.1-5.1] × 10−7 (5)
G2/M-degmut-MSH6 msh3 Δ	7680	2.5 [1.1-3.1] × 10−7 (5)	3.9 [1.5-5.0] × 10−7 (5)	2.1 [1.6-3.4] × 10−7 (3)
msh3Δ lys2-10ALATE	7681	4.8 [2.3-5.9] × 10−8 (1)	1.1 [0.6-2.5] × 10−7 (1)	9.7 [5.2-12.5] × 10−8 (1)
msh3Δ msh6Δ lys2-10ALATE	7688	4.8 [2.5-10.8] × 10−6 (93)	7.3 [2.1-11.1] × 10−5 (932)	2.1 [1.0-2.8] × 10−6 (26)
G2/M-MSH6 msh3Δ lys2-10ALATE	7682	3.4 [2.5-4.2] × 10−6 (68)	8.8 [5.3-17.9] × 10−7 (11)	1.8 [1.0-2.4] × 10−6 (22)
S-MSH6 msh3Δ lys2-10ALATE	7684	8.1 [5.6-10] × 10−7 (16)	1.2 [0.8-1.8] × 10−5 (155)	3.2 [1.9-4.4] × 10−7 (4)

^*Median rates of hom3-10 (Thr⁺) and lys2-10A (Lys⁺) reversion and inactivation of CAN1 (Can^R) with 95% C.I. in square brackets and fold increase relative to the msh3Δ strain in parentheses.

The G2/M-MSH6 construct could be inactive due to the nature of the fusion protein or due to the restricted expression of Msh6 in G2/M. To distinguish between these possibilities, a version of the Clb2-Msh6 fusion, G2/M-deg_mut-MSH6, was generated that was expressed throughout the cell cycle (Figure 1B). Three modifications of the G2/M-MSH6 construct were required to obtain expression throughout the cell cycle: replacement of the CLB2 promoter with the MSH6 promoter, deletion of the destruction box (D-box) located at amino acids 25-33 and mutation of the KEN₁₀₀-box degron that control ubiquitin-mediated degradation at the end of mitosis and in G1 (12, 13) (Figure 1A). The G2/M-deg_mut-MSH6 msh3Δ strain had mutation rates that were similar to the msh3Δ strain and were substantially lower than the msh6Δ msh3Δ strain (Table 1). The slightly higher mutation rates of the G2/M-deg_mut-MSH6 msh3Δ strain relative to the S-MSH6 msh3Δ strain is consistent with observed expression levels of the G2/M-deg_mut-Msh6 protein and the S-Msh6 protein (Figure 1B). These results indicate that the N-terminal Clb2 fusion per se does not compromise the function of Msh6 and that the failure of G2/M-Msh6 to support MMR is due to its G2/M restricted expression. Thus, expression of Msh6 during S-phase is required for functional MMR at the hom3-10, lys2-10A and CAN1 loci, and the capability of these loci to be repaired by MMR is lost some time between 40 and 60 minutes after release from alpha-factor arrest.

The average times of replication for the hom3-10, lys2-10A, and CAN1 loci are 32, 39, and 33 minutes, respectively, after release from alpha-factor arrest (14). These times are within the time window of the bulk of genome replication (Figure S2) and during the peak expression of S-MSH6 but not G2/M-MSH6. We reasoned that if the difference between S-MSH6 and G2/M-MSH6 in the MMR assays were due to temporal coupling between MMR and DNA replication, then G2/M-MSH6 might be functional during MMR in late replicated regions of the genome. We chose one of the latest replicating non-telomeric regions of the genome (14), which is on chromosome IV adjacent to the POL3 and DUN1 genes with an average replication time of 49 minutes (Figure S2) and moved the lys2-10A assay to this location in an msh3Δ strain, generating lys2-10A_LATE. We did not move the other MMR assays as hom3-10 is linked to an origin of replication and the CAN1 inactivation assay could identify other types of gene-inactivating events, like chromosomal rearrangements, depending on chromosomal context (15). Expression of the wild-type MSH6 allele resulted in the same low mutation rate in both the lys2-10A_LATE and the lys2-10A frameshift reversion assays (Table 1) (p=0.14); in contrast, the G2/M-MSH6 allele was completely defective for suppressing the frameshift reversion rate of lys2-10A (372-fold increase in mutation rate) (p=0.5) but almost completely suppressed the frameshift reversion rate of lys2-10A_LATE (11-fold increase in mutation rate). Additionally, the S-MSH6 allele, which had weak Msh6 expression at 50 minutes (Figure 1B), was significantly more defective for MMR at the lys2-10A_LATE locus than the G2/M-MSH6 allele (p=0.0001), but was not completely defective compared to the msh3Δ msh6Δ double mutant (p=0.002). As expected, the mutation rates of the native-positioned and earlier replicating hom3-10 and CAN1 loci were reduced to a low level by the S-MSH6 allele but not G2/M-MSH6 allele in the lys2-10A_LATE strain. Thus, these data indicate that the G2/M-MSH6 allele was functional and could mediate MMR at late replicating loci but not at early replicating loci, further supporting a temporal coupling of MMR with DNA replication.

In addition to roles in MMR, Msh2-Msh6 and Msh2-Msh3 suppress recombination between closely related, but non-identical (homeologous) substrates, a process termed heteroduplex rejection (16-18). We introduced the G2/M-MSH6 and S-MSH6 alleles into msh3Δ strains containing an assay, which monitors the formation of His⁺ prototrophs mediated by recombination between 100% identical (homologous) or between 91% identical (homeologous) substrates (16) (Figure S1B). With wild-type MSH6, recombination was strongly biased towards homologous recombination indicating that heteroduplex rejection was functional (Table 2). In contrast, this bias was strongly reduced when MSH6 was deleted, indicating that the msh6Δ mutation caused a defect in heteroduplex rejection. Both the G2/M-MSH6 and S-MSH6 alleles were partially functional and equivalent in the suppression of homeologous recombination, despite the fact that the assay was integrated at the URA3 locus, which has an average replication time of 28 minutes (14). Similarly, both the G2/M-MSH6 and S-MSH6 alleles were able to suppress the formation of translocations mediated by homeologous recombination between a segmental duplication region on chromosome V and targets on chromosomes IV, X, and XIV (18) (Figure S1C). In this assay, the G2/M-MSH6 allele resulted in a greater decrease in the rate of translocations mediated by homeologous recombination relative to that caused by the S-MSH6 allele, although the 95% confidence intervals of the two rates overlapped (Table 3). These results indicate that heteroduplex rejection is mediated to an equivalent extent by Msh6 expressed in both S and G2/M and does not appear to be dependent upon replication timing, in contrast to MMR. This is similar to prior observations of MMR acting on mispaired bases in meiotic and mitotic recombination intermediates after DNA replication has been completed and in mitotic recombination intermediates formed prior to DNA replication (1, 19).

Table 2. MMR suppression of homeologous recombination during S and G2/M.

	Inverted repeat recombination rate (His+) (fold increase)*
	100% homologous substrate		91% homeologous substrate
Relevant genotype	Strain RDKY	Rate	Strain RDKY	Rate	Ratio 91/100†
msh3 Δ	7672	3.9 [3.0-5.2] × 10−6 (1)	7673	2.6 [1.9-3.0] × 10−7 (1)	0.07
msh3Δ msh6Δ	7674	3.6 [3.3-4.8] × 10−6 (0.9)	7675	1.4 [1.3-1.6] × 10−6 (5.4)	0.39
G2/M-MSH6 msh3 Δ	7676	5.8 [4.4-9.1] × 10−6 (1.5)	7677	7.8 [5.0-10.2] × 10−7 (3)	0.13
S-MSH6 msh3 Δ	7678	4.7 [3.8-6.2] × 10−6 (1.2)	7679	7.9 [6.3-11.8] × 10−7 (3)	0.17

^*Inverted repeat recombination rate (His⁺) with 95% C.I. in square brackets. The number in parentheses represents the fold increase relative to the msh3Δ strain rate for each respective recombination substrate.

^†The inverted repeat recombination rate for the 91% homeologous substrate divided by the inverted repeat recombination rate for the 100% homologous substrate.

Table 3. MMR suppression of duplication mediated GCRs during S- and G2/M.

Relevant genotype	Strain RDKY	GCR rate* (fold increase)
wild-type	6678	4.13 [3.1-6.6] × 10−8 (1)
msh6 Δ	6714	2.18 [1.6-2.7] × 10−7 (5.3)
G2/M-MSH6	7685	4.00 [2.8-7.4] × 10−8 (1.0)
S-MSH6	7686	9.11 [5.7-16.4] × 10−8 (2.2)

^*Rate of 5FOA^R Can^R progeny. The number in parentheses represents the fold increase relative to the wild-type GCR rate.

Restriction of Msh6 expression during the cell cycle through fusion to cyclins has revealed a strong temporal coupling between DNA replication and mismatch repair, but a lack of temporal coupling for heteroduplex rejection. This temporal coupling is likely due to the regulation or appearance of a signal for repair and is not due to the regulation of the MMR proteins. We have recently shown that Msh2-Msh6 dependent MMR involves two mispair recognition pathways, one in which Msh2-Msh6 is stably recruited to replication factories through the Msh6-PCNA interaction and a second Msh6-PCNA interaction independent pathway (20). Furthermore, Msh2-Msh6 dependent MMR is completely dependent on the PCNA coupling of Msh2-Msh6 to replication factories in exo1 mutants (20). Thus PCNA may provide a physical signal for coupling of one of the MMR pathway to the replication fork, whereas the signal for the second MMR pathway is unknown. Other possibilities for the repair signals are single-stranded gaps and nicks transiently generated by DNA replication (21), which is consistent with the requirement for a nick in MMR in vitro (22, 23). The signals triggering MMR on mitotic and meiotic recombination intermediates and for initiating heteroduplex rejection, which differs from MMR due to modified protein requirements (16, 24, 25), are presently unclear but could be nicks, gaps, branched structures or the mispairs in the DNA intermediates.