The Major Replicative Histone Chaperone CAF-1 Suppresses the Activity of the DNA Mismatch Repair System in the Cytotoxic Response to a DNA-methylating Agent*

The DNA mismatch repair (MMR) system corrects DNA mismatches in the genome. It is also required for the cytotoxic response of O6-methylguanine-DNA methyltransferase (MGMT)-deficient mammalian cells and yeast mgt1Δ rad52Δ cells to treatment with Sn1-type methylating agents, which produce cytotoxic O6-methylguanine (O6-mG) DNA lesions. Specifically, an activity of the MMR system causes degradation of irreparable O6-mG-T mispair-containing DNA, triggering cell death; this process forms the basis of treatments of MGMT-deficient cancers with Sn1-type methylating drugs. Recent research supports the view that degradation of irreparable O6-mG-T mispair-containing DNA by the MMR system and CAF-1-dependent packaging of the newly replicated DNA into nucleosomes are two concomitant processes that interact with each other. Here, we studied whether CAF-1 modulates the activity of the MMR system in the cytotoxic response to Sn1-type methylating agents. We found that CAF-1 suppresses the activity of the MMR system in the cytotoxic response of yeast mgt1Δ rad52Δ cells to the prototypic Sn1-type methylating agent N-methyl-N′-nitro-N-nitrosoguanidine. We also report evidence that in human MGMT-deficient cell-free extracts, CAF-1-dependent packaging of irreparable O6-mG-T mispair-containing DNA into nucleosomes suppresses its degradation by the MMR system. Taken together, these findings suggest that CAF-1-dependent incorporation of irreparable O6-mG-T mispair-containing DNA into nucleosomes suppresses its degradation by the MMR system, thereby defending the cell against killing by the Sn1-type methylating agent.

sion of the discontinuous daughter strand. After MutL␣ endonuclease incises the discontinuous daughter strand, the Pol ␦ holoenzyme uses a MutL␣-generated 3Ј end that is 5Ј to the mismatch to perform strand displacement DNA synthesis that removes the mismatch. The role of the Pol ␦ holoenzyme in the excision-independent MMR reaction may be unique because the replacement of the Pol ␦ holoenzyme with the Pol ⑀ holoenzyme abolishes the excision-independent MMR reaction (43).
One of the activities of the MMR system is required for the cytotoxic response of MGMT (O 6 -methyl guanine methyl transferase)-deficient mammalian cells and yeast mgt1⌬ rad52⌬ cells to S n 1-type methylating agents (46,47). This activity of the MMR system depends on MutS␣ and MutL␣ and is necessary for several therapies against MGMT-deficient cancers (3). The treatment involves dacarbazine, procarbazine, or temozolomide, each of which is an S n 1-type methylating drug that triggers death of MGMT-deficient cancer cells by activating the MMR system. O 6 -Methylguanine (O 6 -mG) is the cytotoxic product of treatment of the cell with the S n 1-type methylating agent (48). Normally, O 6 -mG lesions are removed by MGMT that protects the cell from killing by the S n 1-type methylating agent (49). However, a significant number of cancers are deficient in MGMT due to methylation of the MGMT promoter (50). MGMT-deficient cancer cells treated with the S n 1-type methylating agent accumulate O 6 -mG-T mispairs that are recognized by MutS␣ (51). Upon recognition of an O 6 -mG-T mispair, MutS␣ initiates its repair. If the O 6 -mG is on the discontinuous strand, it gets repaired (52). In contrast, if the O 6 -mG is on the continuous strand, it triggers futile cycles of MMR (52). The futile cycles of MMR lead to the formation of persistent strand breaks (53), which are converted in the next S phase into double strand breaks that cause cell cycle arrest followed by cell death (54,55). Consistent with this, double strand break repair defects sensitize eukaryotic cells to the killing effects of S n 1type methylating agents (47,56).
The heterotrimeric CAF-1 is the major histone chaperone for the assembly of nucleosomes onto the newly replicated DNA (57)(58)(59)(60)(61). CAF-1 loads histone (H3-H4) 2 tetramers onto DNA, producing tetrasomes (62,63). Each tetrasome is then converted into a nucleosome by the addition of two histone H2A-H2B dimers (57,64). CAF-1 interacts physically with PCNA, and this interaction is necessary for the action of CAF-1 on the newly replicated DNA (65,66). Recent research has indicated that postreplicative MMR coincides with CAF-1-dependent nucleosome assembly and that the two processes interact with each other (43,63,67). Furthermore, recent findings are consistent with the idea that the eukaryotic MMR system degrades irreparable O 6 -mG-T mispair-containing DNA when it is being packaged into nucleosomes by the CAF-1-dependent mechanism (53,63,67). We show here that CAF-1 suppresses the activity of the MMR system in the cytotoxic response to S n 1-type methylating agents.

CAF-1 Suppressed the Activity of the MMR System in the Cytotoxic Response of Yeast mgt1⌬ rad52⌬ Cells to MNNG-
The cytotoxic response to S n 1-type methylating agents occurs in the chromatin environment (46,52,53,68). However, it has remained unknown whether the chromatin environment affects the cytotoxic response to S n 1-type methylating agents. It has also been unknown whether histone chaperones, proteins that are involved in the control of chromatin environment, influence the cytotoxic response to S n 1-type methylating agents. We wanted to study whether the major replicative histone chaperone CAF-1 impacts the cytotoxic response to an S n 1-type methylating agent. Previous research showed that MGMT-deficient mammalian cells and Saccharomyces cerevisiae mgt1⌬ rad52⌬ that lack Mgt (the yeast ortholog of MGMT (69)) and the recombination mediator Rad52 (70) are efficiently killed by the prototypic S n 1-type methylating agent MNNG in a manner that involves the MMR system (46,47,53). Thus, we utilized the yeast mgt1⌬ rad52⌬ cells to determine whether CAF-1 impacts the cytotoxic response to MNNG. In budding yeast, CAC1 encodes the largest subunit of CAF-1 (60). Accordingly, we investigated whether loss of the CAC1 gene increased the sensitivity of the yeast mgt1⌬ rad52⌬ cells to killing by MNNG. The use of an MNNG cytotoxicity assay permitted us to establish that the cac1⌬ mgt1⌬ rad52⌬ cells were more sensitive to treatment with MNNG than the mgt1⌬ rad52⌬ cells (Fig. 1A). A more detailed analysis revealed that ϳ1.7% of the mgt1⌬ rad52⌬ cells and only ϳ0.2% of the cac1⌬ mgt1⌬ rad52⌬ cells survived the treatment with MNNG (Fig. 1C). Thus, the surviving fraction of the MNNG-treated cac1⌬ mgt1⌬ rad52⌬ cells was ϳ9 times smaller than that of the MNNG-treated mgt1⌬ rad52⌬ cells. We then conducted experiments to determine whether MNNG killed the cac1⌬ mgt1⌬ rad52⌬ cells via an MMR system-dependent mechanism (Fig. 1C). The results showed that the deletion of the MMR system gene MLH1 rescued the sensitivity of the cac1⌬ mgt1⌬ rad52⌬ cells to the cytotoxic effect of MNNG. The experiments also demonstrated that the mlh1⌬ cac1⌬ mgt1⌬ rad52⌬ cells were as resistant to the MNNG treatment as the mlh1⌬ mgt1⌬ rad52⌬ cells. Based on these findings, we concluded that loss of CAC1 sensitizes the yeast mgt1⌬ rad52⌬ cells to MMR system-dependent killing by MNNG.
The cytotoxicity of S n 1-type methylators is mediated by replication-and MMR system-dependent double strand breaks that these agents form. A number of other DNA-damaging agents generate strand breaks that kill cells. Among them are camptothecin, hydroxyurea, and bleomycin. Camptothecin induces replication-dependent double strand breaks by stabilizing topoisomerase I-DNA covalent complexes (71), hydroxyurea causes double strand breaks by depleting the dNTP pools (72), and bleomycin creates double strand and single strand breaks by attacking DNA (73). Unlike the S n 1-type methylating agent, camptothecin, hydroxyurea, and bleomycin kill cells via MMR system-independent mechanisms. Nevertheless, there is a significant similarity between one of these drugs, camptothecin, and MNNG in that double strand breaks caused by these two agents are formed during DNA replication. The results of our genetic experiments (Fig. 1, A and C) raised the possibility that loss of CAC1 sensitized the yeast mgt1⌬ rad52⌬ cells to the cytotoxic effects of DNA-damaging agents that generate DNA breaks. To address this possibility, we studied the effect of CAC1 absence on the sensitivity of the mgt1⌬ rad52⌬ cells to camptothecin, hydroxyurea, and bleomycin (Fig. 1, D and E). An analysis of the data showed that the cac1⌬ mgt1⌬ rad52⌬ cells and the mgt1⌬ rad52⌬ cells had the same sensitivities to bleomycin, camptothecin, and hydroxyurea. Thus, CAC1 absence did not affect the sensitivity of the yeast mgt1⌬ rad52⌬ cells to bleomycin, camptothecin, and hydro-xyurea, drugs that kill cells via MMR system-independent mechanisms. Because MNNG kills the yeast mgt1⌬ rad52⌬ cells by activating an MMR system-dependent mechanism and bleomycin, camptothecin, and hydroxyurea kill the cells via other mechanisms, the results of our genetic experiments (Fig.   FIGURE 1. CAF-1 suppresses the activity of the MMR system in the cytotoxic response of yeast mgt1⌬ rad52⌬ cells to MNNG. Cytotoxicity assays were carried out as detailed under "Experimental Procedures." A and B, cytotoxic responses of cac1⌬ mgt1⌬ rad52⌬ and cac2⌬ mgt1⌬ rad52⌬ cells to treatment with 1 M MNNG. 10-Fold serial dilutions of yeast cultures that were treated or not treated with 1 M MNNG were spotted on the YPDAU plates. C, quantitative analysis of cytotoxic responses of cac1⌬ mgt1⌬ rad52⌬ and cac2⌬ mgt1⌬ rad52⌬ cells to treatment with 1 M MNNG. D, quantitative analysis of cytotoxic response of cac1⌬ mgt1⌬ rad52⌬ to treatment with 30 g/ml bleomycin (BLE). The data in C and D are averages Ϯ 1 S.D. (error bars) (n Ն 3). E and F, cytotoxic responses of cac1⌬ mgt1⌬ rad52⌬ and cac2⌬ mgt1⌬ rad52⌬ cells to treatments with 0.5 g/ml camptothecin and 10 mM hydroxyurea. 10-Fold serial dilutions of yeast cultures were spotted on the YPDAU plates, YPDAU plate with 0.5 g/ml camptothecin, and YPDAU plate with 10 mM hydroxyurea. G, cytotoxic responses of hht2-hhf2⌬ mgt1⌬ rad52⌬, hir2⌬ mgt1⌬ rad52⌬, and rtt106⌬ mgt1⌬ rad52⌬ cells to treatment with 1 M MNNG. 10-Fold serial dilutions of yeast cultures that were treated or not with 1 M MNNG were spotted on the YPDAU plates.
1, A and C-E) indicated that Cac1 increases the survival of the MNNG-treated mgt1⌬ rad52⌬ cells by being involved in a process that suppresses the cytotoxic activity of the MMR system.
Loss of HIR2, RTT106, or HHT2-HHF2 Did Not Change the Sensitivity of Yeast mgt1⌬ rad52⌬ Cells to MNNG-Histone H3-H4 chaperone HIR (Hir1-Hir2-Hir3-Hpc2 complex) plays a major role in replication-independent nucleosome assembly (74), and histone H3-H4 chaperone Rtt106 participates in replication-coupled nucleosome assembly (75). We analyzed whether the absence of HIR2 or RTT106 had an effect on the sensitivity of the mgt1⌬ rad52⌬ cells to MNNG. The experiments showed that lack of HIR2 or RTT106 did not change the sensitivity of the mgt1⌬ rad52⌬ cells to MNNG (Fig. 1G).
CAF-1-dependent Packaging of Irreparable O 6 -mG-T Mispair-containing DNA into Nucleosomes Suppressed Its Degradation by the MMR System-MMR system-dependent degradation of irreparable O 6 -mG-T mispair-containing DNA is involved in the cytotoxic response to the S n 1-type methylating drug (52,53,68). Our genetic experiments indicated that CAF-1 activity suppresses degradation of irreparable O 6 -mG-T mispair-containing DNA by the MMR system ( Fig. 1). To find evidence that CAF-1-dependent incorporation of irreparable O 6 -mG-T mispair-containing DNA into nucleosomes suppresses its degradation by the MMR system, we performed biochemical experiments that are summarized in Figs. 2-7; DNA substrates that we utilized in these experiments were 3Ј-nicked O 6 -mG-T (3Ј O 6 -mG-T), 3Ј-nicked G-T (3Ј G-T), and 3Ј-nicked A-T (3Ј A-T) DNAs. The substrates were made using a plasmid, pAH1A, as a starting material and differed from each other by 1-2 bases (52). The 3Ј-nicked O 6 -mG-T DNA contained a single O 6 -mG-T mispair, the 3Ј-nicked G-T DNA carried a single G-T mispair, and the 3Ј-nicked A-T DNA lacked a mispair. The 3Ј-nicked O 6 -mG-T DNA was irreparable by the MMR system because the O 6 -mG was on the continuous strand. In these biochemical experiments, we used a cytosolic extract prepared from human embryonic kidney cell line 293T that lacked MutL␣ and MGMT (77,78) and had a reduced level of CAF-1 ( Fig. 2A) (58). Although the 293T cytosolic extract lacks the MMR system due to the absence of MutL␣ (77), supplementation of the extract with purified MutL␣ reconstitutes the MMR system (78). In agreement with previous studies (37,52,63,(77)(78)(79), our control experiments showed that 1) the reconstituted MMR system failed to repair an O 6 -mG-T mispair on a nicked DNA (3Ј-nicked O 6 -mG-T DNA) but repaired a G-T mispair on a similar nicked DNA (the 3Ј-nicked G-T DNA) (Fig. 3), 2) supplementation of the 293T cytosolic extract with purified MutL␣-E705K endonuclease mutant (37) did not lead to reconstitution of the MMR system (Fig. 3), 3) the omission of We next utilized Southern hybridization to detect MMR system-dependent degradation of the irreparable O 6 -mG-T mispair-containing DNA that was reconstituted with the 293T cytosolic extract in the absence or presence of the purified CAF-1 (Figs. 4 -7). The initial experiments in this series analyzed degradation products that were separated on denaturing agarose gels (Figs. 4 -6). The data showed that the reconstituted MMR system degraded the discontinuous strand of an irreparable O 6 -mG-T mispair-containing DNA (the 3Ј-nicked O 6 -mG-T DNA) leading to the formation of a ϳ130-nt product (Figs. 4A and 5A, lane 10). Importantly, the ϳ130-nt product was not observed in the reaction mixture that contained endonuclease-deficient MutL␣-E705K instead of MutL␣ (Figs. 4A and 5A, lane 13) demonstrating that the endonuclease activity of MutL␣ is required for the degradation of the 3Ј-nicked O 6 -mG-T DNA. Additional experiments revealed that the ϳ130-nt product or a similar product was not formed from the 3Ј-nicked A-T DNA in the reaction mixture that contained the reconstituted MMR system (Figs. 4A and 5A, lane 17). The results of a time course analysis demonstrated that the ϳ130-nt product of degradation of the discontinuous strand of the 3Ј-nicked O 6 -mG-T DNA was observed in the reaction mixture that was incubated for 10 -120 min (Fig. 5B). This observation indicated that the reconstituted MMR system caused persistent degradation of the discontinuous strand of the 3Ј-nicked O 6 -mG-T DNA. In contrast, the products of degradation of the discontinuous strand of the 3Ј-nicked G-T DNA that were present in the reaction mixture incubated for 10 min (Fig. 4A, lane 3) were not detected in the same reaction mixture that was incubated for 60 min (Fig. 5A, lane 3). This finding is consistent with the view that the product of degradation of the discontinuous strand of the 3Ј-nicked G-T DNA is an intermediate of the MMR reaction (37).
The ϳ130-nt product was identified by Southern hybridization with a 32 P-labeled probe that was complementary to a discontinuous strand sequence located 2 nt downstream from the mismatched T (Figs. 4A and 5A). However, Southern hybridization with a 32 P-labeled probe that was complementary to a discontinuous strand sequence located 2 nt upstream from the mismatched T did not detect the ϳ130-nt product (Fig. 5C). This finding indicated that the 5Ј end of the ϳ130-nt product was located at or near the mismatch. To determine whether a different part of the 3Ј-nicked O 6 -mG-T DNA contained a MutL␣ endonuclease-dependent strand break, we carried out Southern hybridizations with two other 32 P-labeled probes (Fig. 5, D and E). One of the probes was complementary to a discontinuous strand sequence that was downstream from the preexisting strand break (Fig. 5D) and the other to a continuous strand sequence (Fig. 5E). These two Southern hybridizations did not identify any additional MutL␣ endonuclease-dependent strand break in the 3Ј-nicked O 6 -mG-T DNA.
The addition of purified CAF-1 to the 293T extract-and MutL␣-containing reaction mixture led to packaging of the 3Ј-nicked O 6 -mG-T DNA into nucleosomes (Fig. 2B, lane 9). The presence of purified CAF-1 in the 293T extract-and MutL␣-containing reaction mixture also caused a significant decrease in the yield of the ϳ130-nt product of degradation of the irreparable O 6 -mG-containing DNA (Fig. 5, A and B). The effect of CAF-1 on the yield of the ϳ130-nt product was especially pronounced in the reaction mixture in which the concentration of the 3Ј-nicked O 6 -mG-T DNA was decreased 4 times (Fig. 5F). In this case, the addition of purified CAF-1 reduced the yield of the ϳ130-nt degradation product 5-fold. In the above experiments, we included 2.4 pmol of purified CAF-1 in the 293T extract-and MutL␣-containing reaction mixtures. Further experiments showed that the addition of 0.3 pmol of purified CAF-1 was sufficient to cause a significant decrease in the yield of the ϳ130-nt product (Fig. 6, A and B). The simplest interpretation of these results is that CAF-1-dependent packaging of irreparable O 6 -mG-T mispair-containing DNA into nucleosomes reduces its degradation by the MMR system.
We also utilized Southern hybridization to analyze the degradation products that were separated on native agarose gels (Fig. 7). The data revealed that in the 293T extract-and MutL␣containing reaction mixture, a significant fraction of the 3Ј-nicked O 6 -mG-T DNA was converted into a gapped product (Fig. 7). A gap was formed in the discontinuous but not continuous strand of the 3Ј-nicked O 6 -mG-T DNA (Fig. 7, B and C, lane 7). The same or a similar gapped product was largely absent in the 293T extract-and MutL␣-E705K-containing mixture (Fig. 7B, lane 10). Although some of the 3Ј-nicked G-T DNA products in the 293T extract-and MutL␣-containing reaction mixture were also gapped (Fig. 7B, lane 2), their yield was ϳ2.5 times lower than that of the gapped product of the 3Ј-nicked O 6 -mG-T DNA (Fig. 7D). As expected, 3Ј-nicked A-T DNA products that were formed in the 293T extract-and MutL␣-containing reaction mixture lacked gaps (Fig. 7, B (lane 12) and C). Supplementation of the 293T extract-and MutL␣containing reaction mixture with 0.6 or 2.4 pmol of purified CAF-1 caused a significant reduction in the yield of the gapped product of the 3Ј-nicked O 6 -mG-T DNA (Fig. 7, B-E). This finding supports the view that CAF-1-dependent packaging of irreparable O 6 -mG-T mispair-containing DNA into nucleosomes suppresses its degradation by the MMR system.

Degradation of an Irreparable O 6 -mG-T Mispair-containing DNA by the Activated MutL␣ Endonuclease in a Defined
System-Our biochemical experiments (Figs. 4 -7) have implicated MutL␣ endonuclease activity in MMR system-dependent degradation of the discontinuous strand of irreparable O 6 -mG-T mispair-containing DNA. However, these experiments did not provide a clear view of how MutL␣ endonuclease activity is involved in this process. To better understand the involvement of MutL␣ endonuclease activity in MMR system-dependent degradation of the discontinuous strand of irreparable O 6 -mG-T mispair-containing DNA, we carried out additional experiments summarized in Fig. 8. In these experiments, we studied degradation of the discontinuous strand of the 3Ј-nicked O 6 -mG-T DNA in a defined system (37). Purified proteins that were included in the defined system were MutL␣ endonuclease, the mismatch recognition factor MutS␣, the PCNA clamp, the RFC clamp loader, and the single-stranded DNA-binding protein RPA. It can be seen that the discontinuous strand of the 3Ј-nicked O 6 -mG-T DNA was degraded in the defined system in a MutL␣ endonuclease concentration-dependent manner (Fig. 8, A (lanes 13-17) and B). The level of degradation of the discontinuous strand of the 3Ј-nicked O 6 -mG-T DNA was 5-6 times higher than the degradation level of the discontinuous strand of the control 3Ј-nicked A-T DNA (Fig.  8B). No degradation of the discontinuous strand of the 3Ј-nicked O 6 -mG-T DNA was observed in the defined system in which the endonuclease-deficient MutL␣-E705K substituted for MutL␣ (Fig. 8, A (lanes 6, 12, and 18) and C). This informa- The experiments were carried out as described in the legend to Fig. 3. The recovered DNAs were digested with ClaI, separated on alkaline agarose gels, transferred onto nylon membranes, and hybridized with the 32 P-labeled probes ts154 and ts1, which are complementary to the discontinuous strands. A and B, DNA species visualized with the ts154 probe and the ts1 probe, respectively. Each of the diagrams outlines relative positions of the probe (a bar with an asterisk), the unique ClaI site, the strand break, and a mismatch. The distance between the ClaI site and the mismatch is 145 bp.
The MMR System and CAF-1 DECEMBER 30, 2016 • VOLUME 291 • NUMBER 53 tion indicated that MutL␣ provided the endonuclease activity that degraded the discontinuous strand of an irreparable O 6 -mG-T mispair-containing DNA in the presence of MutS␣, PCNA, RFC, and RPA. Importantly, the ϳ130-nt fragment that was detected in the cell extract system (Fig. 5A, lane 10) was not a preferred product of the endonuclease reaction in the defined system (Fig. 8A, lanes 15-17). This is an indication that the defined system lacks one or more factors that are involved in the formation of the ϳ130-nt fragment in the extract system. While degrading the discontinuous strand, MutL␣ endonuclease did not incise the continuous strand of the 3Ј-nicked O 6 -mG-T DNA (Fig. 8E, lanes 13-17). An inspection of the data in Fig. 8A showed that MutL␣ endonuclease incised the discontinuous strand of the 3Ј-nicked O 6 -mG-T DNA at random sites, displaying no significant site or sequence specificity. These findings suggested that MutL␣ endonuclease contributes to MMR system-dependent degradation of irreparable O 6 -mG-T mispair-containing DNA by introducing strand breaks at random sites on the discontinuous strand. In addition, these findings suggested that processing of the MutL␣ endonuclease-incised discontinuous strand of the 3Ј-nicked O 6 -mG-T DNA in the extract system led to the formation of the ϳ130-nt product (Fig. 4A, lane 10). In agreement with this idea, the pattern of degradation of the discontinuous strand of the 3Ј-nicked O 6 -mG-T DNA in the extract system that was supplemented with the DNA polymerase inhibitor aphidicolin (Fig. 4A, lane 14) is similar to the pattern of degradation of the discontinuous strand of the 3Ј-nicked O 6 -mG-T DNA in the defined system (Fig. 8A, lanes 13-17).

Discussion
Mammalian MGMT-deficient and yeast mgt1⌬ rad52⌬ cells are efficiently killed by low doses of S n 1-type methylating agents (46, 47, 78, 80 -82). Several anticancer therapies exploit the marked sensitivity of MGMT-deficient cells to S n 1-type methylating agents. The marked sensitivity of mammalian MGMT-deficient and yeast mgt1⌬ rad52⌬ cells to S n 1-type methylating agents is a result of the MMR system-dependent cytotoxic response (46, 47, 78, 80 -82). Previous research showed that the cytotoxic response to the S n 1-type methylating drug involves MMR system-dependent degradation of irreparable O 6 -mG-containing DNA that leads to the formation of lethal double strand breaks (47,53,56). The MMR system starts to degrade irreparable O 6 -mG-containing nascent DNA behind the replication fork (53). At the same time, this DNA is incorporated into nucleosomes in the CAF-1-orchestrated process (57)(58)(59)(60)(61). It was previously unknown whether concomitant CAF-1-dependent nucleosome assembly affects degradation of irreparable O 6 -mG-containing DNA by the MMR system. We have found that CAF-1 suppresses the activity of the MMR system in the cytotoxic response of yeast mgt1⌬ rad52⌬ cells to MNNG (Fig. 1). We have also found that in an MGMT-deficient extract system, CAF-1-dependent incorporation of an irreparable O 6 -mG-T mispair-containing DNA into nucleosomes (Fig. 2B) correlates with a substantial decrease in degradation of this DNA by a MutL␣ endonuclease-dependent mechanism (Figs. 5-7). These findings imply that CAF-1-dependent packaging of irreparable O 6 -mG-T mispair-containing DNA into nucleosomes suppresses its degradation by the MMR system, therefore defending the cell against killing by the S n 1type methylating drug. Consistent with this, we have established that loss of CAF-1 does not affect the sensitivity of yeast mgt1⌬ rad52⌬ cells to bleomycin, camptothecin, and hydroxyurea, DNA-damaging drugs that kill cells in an MMR system-independent manner (Fig. 1). It is known that 1-2 dou-  The MMR System and CAF-1 DECEMBER 30, 2016 • VOLUME 291 • NUMBER 53 ble strand breaks are sufficient to kill the yeast rad52 cell (83). Therefore, the fact that the MNNG treatment kills the cac1⌬ mgt1⌬ rad52⌬ and cac2⌬ mgt1⌬ rad52⌬ cells more efficiently than the mgt1⌬ rad52⌬ cells indicates that CAF-1 loss increases the fraction of MNNG-treated cells that experience at least 1-2 double strand breaks.
In addition to CAF-1, S. cerevisiae cells contain several other histone chaperones, including HIR and Rtt106 (75). Our results indicated that loss of HIR2 or RTT106 does not increase the sensitivity of mgt1⌬ rad52⌬ cells to MNNG (Fig. 1G). Thus, these results imply that neither of these histone chaperones has a non-redundant function that provides a protection for mgt1⌬ rad52⌬ cells from the cytotoxic activity of the MMR system. We determined that decreasing histone H3-H4 gene dosage by deletion of the HHT2-HHF2 locus in the mgt1⌬ rad52⌬ cells does not change their sensitivity to MNNG (Fig. 1G). A previous report described that hht2-hhf2⌬ does not affect the level of the chromatin H3-H4 histones but decreases the level of the soluble H3-H4 histones 2-fold (84). Thus, it appears that a small decrease in the level of the soluble histones H3-H4 does not affect the cytotoxic activity of the MMR system.
An earlier study has implicated MutL␣ endonuclease activity in the cytotoxic response to the S n 1-type methylating drug in the yeast and mammalian cells (68). We have now shown that activated MutL␣ endonuclease degrades the discontinuous strand of an irreparable O 6 -mG-T mispair-containing DNA in an extract system and in a purified system (Figs. 4 -8). This information implies that MutL␣ endonuclease-dependent deg-FIGURE 7. CAF-1-dependent suppression of the formation of a gap in an irreparable O 6 -mG-T mispair-containing DNA in an extract system. The experiments were performed and analyzed as described under "Experimental Procedures." Each reaction mixture contained a DNA substrate (41 fmol) and other indicated components and was incubated for 60 min at 37°C. The recovered DNA products were digested with AhdI, separated on native agarose gels, transferred onto nylon membranes, and hybridized with the 32 P-labeled probes c154 and ts154. A, outline of the assay. B and C, DNA species visualized with the 32 P-labeled probes c154 and ts154, respectively. The c154 probe is complementary to the continuous strand, and the ts154 probe is complementary to the discontinuous strand. Each diagram depicts the relative positions of the 32 P-labeled probe (a bar with an asterisk), AhdI sites, the strand break, and a mismatch. radation of the discontinuous strand of irreparable O 6 -mG-T mispair-containing nuclear DNA is involved in the cytotoxic response to the S n 1-type methylating drug. MutL␣ endonuclease degrades the 3Ј-nicked O 6 -mG-T and G-T DNA substrates in the presence of MutS␣, PCNA, RFC, and RPA in a very similar way (Fig. 8A, lanes 3-5 and lanes 15-17), suggesting that the same mechanism activates MutL␣ endonuclease in MMR and in the cytotoxic response to the S n 1-type methylating drug. The mechanism of activation of MutL␣ endonuclease in MMR requires the presence of MutS␣ and a mismatch (37,39,42) (Fig. 8). The O 6 -mG-T mispair is one of many mispairs recognized by MutS␣ (21). A crystallographic study determined that MutS␣ has the same structure in the MutS␣-G-T DNA and MutS␣-O 6 -mG-T DNA crystals (85). This information and the results of our analysis of the defined reactions (Fig.  8) suggest that adoption of the same structure by MutS␣ on the G-T mispair and on the irreparable O 6 -mG-T mispair permits the protein to convey the same activating signal to MutL␣ FIGURE 8. MutL␣ endonuclease-dependent degradation of an irreparable O 6 -mG-T mispair-containing DNA in a defined system. The experiments were performed and analyzed as described under "Experimental Procedures." Each reaction mixture contained a DNA substrate (81 fmol) and the indicated proteins and was incubated for 10 min at 37°C. When MutS␣, PCNA, RFC, and RPA were present in the reaction mixtures, their concentrations were 40, 24, 4, and 40 nM, respectively. The recovered DNAs that either were not digested (A and C) or were digested (E) with ClaI were resolved on alkaline agarose gels and hybridized with the 32 P-labeled probes ts154, ts1, and c29. The ts154 and ts1 probes are complementary to the discontinuous strands, and the c29 probe is complementary to the continuous strands. A and C, DNA species visualized with the 32 P-labeled probes ts154 and ts1, respectively. B, degradation of the discontinuous strands of the 3Ј-nicked DNA substrates as a function of MutL␣ endonuclease concentration. The data were obtained by quantification of images generated with the 32 P-labeled probe ts154. One of the images is shown in A. D, degradation of the discontinuous strands of the 3Ј-nicked DNA substrates in the presence of the MutL␣-E705K mutant. The data were obtained by quantification of images generated with the 32 P-labeled probe ts1. One of the images is shown in C. The data in B and D are averages Ϯ 1 S.D. (error bars) (n ϭ 3). E, DNA species visualized with the 32 P-labeled probe c29. Each of the diagrams depicts the relative positions of the 32 P-labeled probe (a bar with an asterisk), the strand break, and a mismatch.
The MMR System and CAF-1 DECEMBER 30, 2016 • VOLUME 291 • NUMBER 53 endonuclease during MMR and the cytotoxic response to the S n 1-type methylating drug. Although the major product of persistent MutL␣ endonuclease-dependent degradation of the 3Ј-nicked O 6 -mG-T DNA in the extract system has a size of ϳ130 nt (Figs. 4A and 5A (lane 10) and 6A (lane 8)), the products of MutL␣ endonuclease-dependent degradation of the 3Ј-nicked O 6 -mG-T DNA in the extract system that was supplemented with aphidicolin have sizes in the range of 100 -2,000 nt (Fig. 4, A and B, lane 14). Because aphidicolin is an inhibitor of the biosynthetic activities of Pol ␦ and Pol ⑀, these findings suggest that DNA synthesis and ligation that occur on the MutL␣-degraded DNA in the extract system in the absence of aphidicolin are necessary to remove the majority of the 100 -2,000-nt products.
It is important to note that a previous work described that MutL␣-dependent processing of the 3Ј-nicked O 6 -mG-T DNA in the HCT116BBR nuclear extract-containing system leads to the formation of a ϳ130-nt product (52) that does not appear to be different from the one that we have detected in our extract system . This observation implies that reactions that occur in different extract systems generate the same product of degradation of the 3Ј-nicked O 6 -mG-T DNA. We have observed that MutL␣ endonuclease-dependent degradation of the 3Ј-nicked O 6 -mG-T DNA in our extract system leads to the formation of a gapped product (Fig. 7). The gap was generated in the discontinuous strand in the presence of dNTPs and was located downstream from the mispaired T. Thus, the MMR system-dependent processing of the 3Ј-nicked O 6 -mG-T DNA in the cell extract generates two kinds of DNA products; one of them carries the gap (Fig. 7), and the other contains the ϳ130-nt fragment (Figs. 5A and 6A). It is likely that the gap is formed when the excision step is not blocked by the O 6 -mG, whereas the ϳ130-nt fragment is formed when the excision step is blocked by the lesion. The presence of a gap in the irreparable DNA is in good agreement with a previous study that documented that gaps are formed behind replication forks in response to treatment of both mammalian MGMTdeficient cells and yeast mgt1⌬ rad52⌬ cells with MNNG (53). The fact that the gap was generated downstream from the mismatched T (Fig. 7) suggests that O 6 -mG is a stronger block for the DNA polymerization reaction than for the excision reaction.
The S n 1-type methylating drug temozolomide is used for treatment of glioblastoma patients. However, recurrent glioblastomas often arise during post-treatment period (86). This indicates that an approach that increases the sensitivity of MGMT-deficient tumors to treatment with the S n 1-type methylating drug might suppress recurrent cancers. More research is needed to determine whether defective replication-coupled nucleosome assembly increases the sensitivity of MGMT-deficient cancer cells to treatment with the S n 1-type methylating drug.
MNNG and Bleomycin Cytotoxicity Assays-Liquid YPDAU medium (1% yeast extract, 2% Bacto-peptone, 2% dextrose, 60 mg/liter adenine, 60 mg/liter uracil), YPDAU plates, 1 mM stock solutions of MNNG (Wako Chemicals USA) in DMSO, a 2 mg/ml stock solution of bleomycin (Santa Cruz Biotechnology) in DMSO, and DMSO were used in the assays. Yeast cultures were grown to saturation in liquid YPDAU for ϳ20 h at 30°C. The saturated cultures were diluted 10-fold in fresh medium and grown for 4 h at 30°C. The cultures were then diluted with fresh medium to A 600 ϭ 1.3, and aliquots of the diluted cultures were treated with 1 M MNNG, 0.1% DMSO (vehicle control in the MNNG toxicity assay), 30 g/ml bleomycin, and 1.5% DMSO (vehicle control in the bleomycin cytotoxicity assay) for 2 h at 30°C. After treatment, the cultures were diluted, and appropriate dilutions of the cultures were spread on YPDAU plates. The plates were incubated for 3-4 days at 30°C, and colonies were counted. A somewhat different MNNG cytotoxicity assay was also used in this study. In this assay, 10-fold serial dilutions of the treated cultures were made and spotted on YPDAU plates. The plates were incubated for 2 days at 30°C and photographed.
Camptothecin and Hydroxyurea Cytotoxicity Assays-Yeast cultures were grown to saturation as described above and diluted to A 600 ϭ 1.4 with sterile water. 10-fold serial dilutions of the cultures were prepared and spotted on YPDAU plates, YPDAU plates containing 0.5 g/ml camptothecin (Enzo Life Science), and YPDAU plates containing 10 mM hydroxyurea (US Biological). After incubation for 2 days at 30°C, the plates were photographed.
Western Blotting-Samples of the purified CAF-1, 293T cytosolic extract (15 g), and 293T nuclear extract (15 g) were separated on a denaturing SDS gel and transferred onto a PVDF membrane. After the protein transfer step, the membrane was incubated with ␣-CAF-1 p150 antibodies (catalog no. sc-10772, lot E2004, rabbit polyclonal IgG, Santa Cruz Biotechnology, Inc.) and then with ECL HRP-conjugated secondary antibodies (catalog no. NA934V, lot 389592, donkey antibody, GE Healthcare). Immune complexes were visualized utilizing ECL2 Western blotting substrate (Thermo Fisher Scientific) and a CCD camera. Amounts of CAF-1 in the samples of the 293T cytosolic and nuclear extracts were measured by quantification of the data with the ImageJ software.
DNA Substrates, Oligonucleotides, and 32 P-Labeled Hybridization Probes-3Ј-Nicked O 6 -mG-T, 3Ј-nicked G-T, and 3Ј-nicked A-T DNAs were essentially prepared as described previously (52), except that after DNA ligation, unligated material was degraded by exonuclease III (New England Biolabs), and the remaining DNA was purified by chromatography on BND-cellulose (Sigma). The O 6 -mG-containing oligonucleotide that was used to prepare the 3Ј-nicked O 6 -mG-T DNA was synthesized by Midland Certified Reagent Co. All other oligonucleotides used in this study were synthesized by IDT. To prepare a 32 P-labeled hybridization probe, the oligonucleotide was labeled at the 5Ј end with 32 P by T4 polynucleotide kinase in the presence of [␥-32 P]ATP. The sequences of oligonucleotides used to construct the DNA substrates and prepare 32 P-labeled hybridization probes are shown in Table 1.
Nucleosome Assembly Reactions in 293T Cytosolic Extractcontaining Mixtures-The nucleosome assembly reactions were carried out at 37°C in 40-l mixtures that contained 20 mM HEPES-NaOH (pH 7.4), 100 mM KCl, 8 mM MgCl 2 , 2 mM DTT, 0.2 mg/ml BSA, 0.1 mM each of the four dNTPs, 3 mM ATP, 20 mM creatine phosphate, 0.02 mg/ml creatine phosphokinase, 1% glycerol (v/v), 75 g of 293T cytosolic extract, purified CAF-1 (0 or 1.2 pmol), MutL␣ (0 or 1.6 pmol), and 81 fmol (0.1 g) of a 3Ј-nicked DNA (the 3Ј-nicked O 6 -mG-T DNA, the 3Ј-nicked G-T DNA, or the 3Ј-nicked A-T DNA). After 60 min of incubation, a 35-l fraction of each reaction mixture was mixed with a 5-l mixture containing 20 mM HEPES-NaOH (pH 7.4), 64 mM CaCl 2 , 8 units/l micrococcal nuclease, and 0.02 mg/ml RNase A and incubated for 20 min at 21-23°C. Each reaction mixture was then supplemented with a 10-l solution containing 0.5% SDS, 150 mM EDTA, 40% glycerol, and 2 mg/ml Proteinase K, followed by incubation of the mixture for 20 min at 50°C. Proteinase K in the mixtures was inactivated by the addition of PMSF to a final concentration of 0.8 mM. The nucleosome assembly products were separated on native 1.7% agarose gels, transferred onto nitrocellulose membranes, and hybridized with 32 P-labeled probe c154. Indirectly labeled DNA species were visualized with a Typhoon biomolecular imager (GE Healthcare) and quantified with ImageQuant software.
Mismatch-provoked Reactions in an Extract System-The mismatch-provoked and control reactions in the extract system were carried out at 37°C in 80-l mixtures containing 20 mM HEPES-NaOH (pH 7.4), 100 mM KCl, 8 mM MgCl 2 , 2 mM DTT, 0.2 mg/ml BSA, 0.1 mM each of the four dNTPs, 3 mM ATP, 20 mM creatine phosphate, 0.02 mg/ml creatine phosphokinase, 1% glycerol (v/v), 150 g of 293T cell extract, MutL␣ (0 or 3.2 pmol), MutL␣-E705K (0 or 3.2 pmol), purified CAF-1 (0, 0.3, 0.6, or 2.4 pmol), and 41 or 162 fmol of a 3Ј-nicked DNA (the 3Ј-nicked O 6 -mG-T DNA, the 3Ј-nicked G-T DNA, or the 3Ј-nicked A-T DNA). The reaction mixtures were incubated for 10 -120 min. Reactions in each mixture were stopped by the addition of a 60-l solution containing 0.35% SDS, 0.4 M NaCl, 13 mM EDTA, 0.33 mg/ml Proteinase K, and 1 mg/ml glycogen, and the resulting mixtures were incubated for 20 min at 50°C, followed by their extraction with phenol/chloroform mixture. DNAs present in the supernatants were recovered by isopropyl alcohol precipitation. To detect whether the repair of O 6 -mG-T or G-T mispairs occurred in a reaction mixture, a fraction of the recovered DNA was cleaved with BanI and XhoI and separated on a 1.2% agarose gel, followed by staining of the gel with ethidium bromide and quantification of DNA species with the ImageJ software. To detect whether a 3Ј-nicked DNA was degraded in the reaction, a fraction of the recovered DNA was cleaved with ClaI, separated on an alkaline 1.4% agarose gel, transferred onto a nylon membrane, and hybridized with a 32 Plabeled probe. To detect whether gapped DNA was produced in a reaction mixture, a fraction of the recovered DNA was digested with AhdI, resolved on a native 1.4% agarose gel, transferred onto a nylon membrane, and hybridized with an indicated 32 P-labeled probe. Indirectly labeled DNA species were visualized and quantified as described above.
Cleavage of DNA by Activated MutL␣ Endonuclease in a Defined System-The reactions were performed in 40-l mixtures that contained 20 mM HEPES-NaOH (pH 7.4), 120 mM KCl, 5 mM MgCl 2 , 3 mM ATP, 2 mM DTT, 0.2 mg/ml BSA, 2% glycerol (v/v), and 2 nM (81 fmol) of a 3Ј-nicked DNA (the 3Ј-nicked O 6 -mG-T DNA, the 3Ј-nicked G-T DNA, or the 3Ј-nicked A-T DNA). When indicated, MutS␣ (40 nM), MutL␣ (2, 6, or 20 nM), MutL␣-E705K (20 nM), PCNA (24 nM), RFC (4 nM), and RPA (40 nM) were included in the reaction mixtures. The reaction mixtures were incubated for 10 min at 37°C, and each reaction was terminated by the addition of a 30-l mixture containing 0.35% SDS, 0.4 M NaCl, 13 mM EDTA, 0.33 mg/ml Proteinase K, and 2 mg/ml glycogen, followed by incubation of the mixtures for 15 min at 50°C. The mixtures were then extracted by phenol/chloroform, and the DNAs were recovered by isopropyl alcohol precipitation. Recovered DNAs that were digested or not with ClaI were separated on alkaline 1.4% denaturing agarose gels, transferred onto nylon membranes, and hybridized with 32 P-labeled probes. Indirectly labeled DNA species were visualized and quantified as described above.
Author Contributions-L. K., B. D., and F. K. performed experiments and analyzed data. F. K. and L. K. designed experiments, contributed reagents, prepared the figures, and wrote the paper. Oligonucleotide sequence AH1A A-T 5Ј-GCTACCGTCCTCGAAGCTTCCGCATCGGAGTCGACG-3Ј AH1A G-T 5Ј-GCTACCGTCCTCGAGGCTTCCGCATCGGAGTCGACG-3Ј AH1A O 6 -mG-T 5Ј-GCTACCGTCCTCGAO 6 5Ј-CGAGGACGGTAGCGAGAGACTCGA-3Ј a The first three oligonucleotides were used to prepare the 3Ј-nicked DNA substrates, and the remaining oligonucleotides were used to prepare 32 P-labeled hybridization probes.
The MMR System and CAF-1