Human MutY Homolog, a DNA Glycosylase Involved in Base Excision Repair, Physically and Functionally Interacts with Mismatch Repair Proteins Human MutS Homolog 2/Human MutS Homolog 6*

Adenines mismatched with guanines or 7,8-dihydro-8-oxo-deoxyguanines that arise through DNA replication errors can be repaired by either base excision repair or mismatch repair. The human MutY homolog (hMYH), a DNA glycosylase, removes adenines from these mismatches. Human MutS homologs, hMSH2/hMSH6 (hMutSα), bind to the mismatches and initiate the repair on the daughter DNA strands. Human MYH is physically associated with hMSH2/hMSH6 via the hMSH6 subunit. The interaction of hMutSα and hMYH is not observed in several mismatch repair-defective cell lines. The hMutSα binding site is mapped to amino acid residues 232–254 of hMYH, a region conserved in the MutY family. Moreover, the binding and glycosylase activities of hMYH with an A/7,8-dihydro-8-oxo-deoxyguanine mismatch are enhanced by hMutSα. These results suggest that protein-protein interactions may be a means by which hMYH repair and mismatch repair cooperate in reducing replicative errors caused by oxidized bases.

lesions caused by oxidative stress. In Escherichia coli, MutT, MutM, and MutY are involved in defending against the mutagenic effects of 8-oxoG lesions (4,5). Human cells have MutT, MutM, and MutY homologs named MutT homolog (hMTH1), 8-oxoG glycosylase (hOGG1), and MutY homolog (hMYH), respectively. The hMTH1 protein eliminates 8-oxo-dGTP from the nucleotide pool with its nucleoside triphosphatase activity (Fig. 1, reaction 1) (9 -11). The hOGG1 protein provides a second level of defense by removing both ring-opened purine lesions and mutagenic GO adducts (Fig. 1, reaction 2) (12)(13)(14)(15). Human MYH is involved in the removal of adenines mismatched with guanines or 8-oxoG that arise through DNA replication errors (Fig. 1,  reaction 3) (16 -20). After repair synthesis and hOGG1 repair ( Fig. 1, reaction 4), normal base pairs can be restored. Recently, we have demonstrated that hMYH is the major protein in human cell extracts that recognizes A/8-oxoG-containing DNA substrates (21). Therefore, together with hOGG1 and hMTH1, hMYH is the primary repair protein responsible for protecting the cell from the mutagenic effects of 8-oxoG. If 8-oxo-dGTP is not hydrolyzed by hMTH1 protein, it can be incorporated to pair with template adenine. In this case, if hMYH excises the adenine on the template strand ( Fig. 1, reaction 5), a higher A:T to C:G transversion will occur. Thus, directing hMYH to the daughter but not the parental strand is crucial for the fidelity of DNA replication involving GO adducts. It has been suggested that hMYH repair is coupled to DNA replication through docking with human proliferating cell nuclear antigen (PCNA) and replication protein A (RPA) (22,23). The mismatch repair system enhances the fidelity of DNA replication and genetic recombination. Mismatch repair enzymes are also involved in cell cycle arrest, transcription-coupled repair, and meiotic recombination (for reviews, see Refs. 24 -28). E. coli mismatch repair is dependent on dam methylation and requires specifically the MutH, MutL, and MutS proteins. The MutH endonuclease cleaves at the 5Ј-end of unmethylated GATC sequences. A homodimer of MutS recognizes base/base mismatches and short insertion-deletion loops. MutL enhances the activities of MutH, MutS, and DNA helicase II (29 -31). The eukaryotic mismatch repair system contains multiple MutS and MutL homologs; however, no MutH homolog has been identified. The hMSH2/hMSH6 (hMutS␣) heterodimer recognizes base/base mismatches and short insertiondeletion loops, whereas the hMSH2/hMSH3 (hMutS␤) recognizes longer insertion-deletion loops (32)(33)(34). A major unanswered question in eukaryotic mismatch repair is what dictates strand specificity. It has been suggested that strand breaks on the daughter DNA strands or a direct interaction between repair components and the replication machinery may provide discrimination between daughter and parental strands. Germ line mutations in human mismatch repair genes can lead to gene mutations and increase microsatellite instability, and they have been correlated with hereditary nonpolyposis colon cancer and various sporadic cancers (25,27,28). It has been demonstrated that mismatch repair-deficient cells are resistant to a variety of chemotherapeutic agents including alkylation agents, cisplatin, and 6-thioguanine (28). The mismatch repair system has also been shown to be involved in the repair of oxidative DNA damage. Cells defective in the MSH2 gene are impaired in transcription-coupled repair of oxidative damage (35). Furthermore, mouse embryonic stem cells carrying a defective Msh2 allele accumulate oxidized bases in their DNA (36). Recently, Msh2p/Msh6p of yeast Saccharomyces cerevisiae has been shown to bind A/GO mismatches and to be involved in repair of GO lesions in DNA (Fig. 1, reactions 3 and 6) (37). Because S. cerevisiae does not contain MutY and MutT homologs, it is unclear whether the role of MSH2/MSH6 in GO repair is also important in other organisms.
We have previously shown that hMYH is directly associated with human apurinic/apyrimidinic endonuclease, human PCNA, and human RPA, suggesting that hMYH plays a role in the long patch base excision repair pathway (23). In this report, we demonstrate that hMYH also interacts with the hMSH2/ hMSH6 (hMutS␣) heterodimer but not with hMSH2/hMSH3 (hMutS␤). More specifically, hMYH was found to interact physically with hMSH6 but not with hMSH2. Additionally, the hMSH2/hMSH6 heterodimer stimulates the DNA binding and glycosylase activities of hMYH with an A/GO mismatch. Interestingly, both hMYH and hMSH6 interact with PCNA and colocalize with PCNA to replication foci (22, 23, 38 -40). The interactions of PCNA with both hMYH and hMutS␣ suggest that PCNA may act as a coordinator of both repair pathways. Thus, the MYH-mediated base excision repair may cooperate with mismatch repair in defending against the mutagenic effects of 8-oxoG lesions.

EXPERIMENTAL PROCEDURES
Human Cell Extracts-TK6 and MT1 cells were kindly supplied by Guo-Min Li. LoVo and HCT15 cell lines were purchased from the American Type Culture Collection (ATCC) (Manassas, VA). TK6, MT1, and HCT15 were grown in RPMI 1640 medium (Invitrogen) supplemented with 10% fetal bovine serum (HyClone, Logan, UT) at 37°C in 5% CO 2 . LoVo cells were grown in Ham's F-12K medium (Invitrogen) with 10% fetal bovine serum with 1.5 g/liter NaHCO 3 , 2 mM L-glutamine, and 1% penicillin/streptomycin at 37°C in 5% CO 2 . Cells were grown to late log phase. The cell pellet from three T-75 flasks (ϳ3 ϫ 10 7 cells) was resuspended in 0.5 ml of buffer containing 50 mM potassium phosphate, pH 7.4, 50 mM KCl, 1 mM dithiothreitol, 0.1 mM EDTA, 0.1 mM phenylmethylsulfonyl fluoride, and 10% glycerol. An equal volume of 0.1-mm glass beads was added into the cell suspension, and cells were disrupted by vigorous vortexing for 10 s at 4°C and cooled on ice for 20 s. Vortexing was repeated 10 times. The mixture was then centrifuged at 12,000 ϫ g for 15 min. The supernatant was aliquoted and stored at Ϫ80°C. The protein concentration was determined by Bio-Rad protein assay (Bio-Rad).
Expression and Purification of the Recombinant Proteins-Recombinant hMYH protein expressed in E. coli was partially purified according to the procedures described by Gu and Lu (21). The percentage of hMYH in the preparation is about 10% as judged by SDS-PAGE. The concentration of hMYH in the preparation was estimated to be 10% of the total protein concentration. Two sources of hMutS␣ were tested in the ex-periments. The untagged hMutS␣ and hMutS␤ kindly provided by Dr. Josef Jiricny were purified according to Palombo et al. (41). The insect Sf9 cells were coinfected with baculovirus vectors carrying cDNA encoding hMSH2 and hMSH6 to express hMutS␣ and with baculovirus vectors carrying cDNA encoding hMSH2 and hMSH3 to express hMutS␤. Untagged hMSH2 complexed with His-tagged hMSH6 was purified according to the described procedures (42,43). In this system, the cDNAs for hMSH2/hMSH6 were cloned into viral expression vector pFastBacDual (Invitrogen) (a kind gift from Dr. Richard Fishel) to express hMutS␣.
Cloning of the hMYH Gene and GST-hMYH Protein Constructs-The hMYH gene and three deletion constructs cloned into the pGEX-4T-2 vector (Amersham Biosciences, Inc.) to express fusion proteins of glutathione S-transferase and hMYH (GST-hMYH) were described previously (23). Four additional N-terminal deletion constructs of hMYH fused to glutathione S-transferase were made by the PCR method using the sense primers listed in Table I and the antisense primer (5Ј-TAT-TCTCGAGTTCACTGGGCTGCACTGT-3Ј). The PCR products were digested with BamHI and XhoI and ligated into the BamHI-XhoI-digested pGEX-4T-2 vector. The sequences of the cloned hMYH gene fragments were confirmed by DNA sequencing.
Expression and Immobilization of GST-hMYH Constructs-E. coli (BL21-CodonPlus-RIL/DE3) cells (Stratagene, La Jolla, CA) harboring the expression plasmids of the GST-hMYH constructs were grown in LB broth containing 100 mg/ml ampicillin at 25°C. Protein expression was induced at an A 590 of 0.6 by the addition of isopropyl-␤-D-thiogalactopyranoside to a final concentration of 0.4 mM, and the cells were harvested 16 h later by centrifugation at 10,000 ϫ g for 20 min. The cell paste, from a 500-ml culture, was resuspended in 9 ml of buffer G (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 2 mM EDTA) and treated with lysozyme (1 mg/ml) for 30 min at room temperature. After adding 0.2% Triton X-100 and 2 mM dithiothreitol, the solution was then centrifuged at 10,000 ϫ g for 20 min. To the supernatant (10 ml), 1 ml of a 50% slurry of glutathione-Sepharose 4B in buffer G was added and incubated overnight at 4°C. The GST-hMYH fusion proteins bound to the beads were pelleted at 1000 ϫ g for 5 min and incubated with 5% bovine serum albumin in buffer G overnight at 4°C. The beads were pelleted at 1000 ϫ g for 5 min and washed four times with 5 ml of buffer G. The beads were suspended in buffer G containing chymostatin, pepstatin, leupeptin (5 g/ml), 0.1% sodium azide, and a protease inhibitor mixture (Sigma) to form a 50% slurry and stored at 4°C.
GST-hMYH Pull-down Assay-Heterodimers hMSH2/hMSH6 (hMutS␣) and hMSH2/hMSH3 (hMutS␤) (200 ng) expressed in the baculovirus expression system in insect Sf9 cells were added to the appropriate GST-hMYH constructs (300 ng) immobilized on glutathione-Sepharose 4B (see above) and incubated overnight in 50 l of buffer G at 4°C. After centrifugation at 1000 ϫ g, the supernatant was saved, and the pellets were washed five times with 800 l of buffer G at 4°C. The pellets and supernatants (30 l) were fractionated on a 10% SDSpolyacrylamide gel, and Western blot analyses for hMSH2 and hMSH6 were performed as described (44) (see below). A control was run concurrently with immobilized GST alone. For co-precipitation of hMSH2 and hMSH6 with GST-hMYH from human cell extracts, 200 -300 ng of GST-hMYH or GST alone was added to extracts (1.0 mg) and incubated overnight at 4°C. After centrifugation at 1000 ϫ g, the supernatant (ϳ3% of total volume), and the pellet was treated as described above.
Western blotting was detected by the ECL analysis system from Amersham Biosciences according to the manufacturer's protocol.
Far Western Analysis-Recombinant hMSH2/hMSH6 (hMutS␣) (7 pmol) were separated on 10% SDS-PAGE and transferred to nitrocellulose membrane. The membrane was blocked with 5% low fat milk in phosphate-buffered saline for 1 h and then incubated with 3 g/ml partially purified recombinant hMYH (21) at 4°C overnight. After extensive washing with blocking solution, the membrane was incubated with anti-hMYH peptide antibody ␣516 and subjected to Western blotting as described above.
Assay of Protein-DNA Binding Complexes-The reaction mixtures for hMYH and hMutS␣ binding activities on A/8-oxoG-containing DNA substrate were similar to that of the glycosylase assay, except that 30 mM NaCl and 0.5 g/ml poly(dI-dC) were added to the reactions. The reactions were carried out at 37°C for 30 min and were supplemented by adding 2 l of 50% glycerol. The protein-DNA complexes were then resolved on a 4% polyacrylamide gel in 50 mM Tris borate buffer containing 2.5% glycerol. The electrophoresis was carried out in a cold room with 20-mA current with 50 mM Tris borate buffer. The gel was dried and exposed to a PhosphorImager screen, and the percentages of bound DNA were analyzed by Molecular Dynamics ImageQuant.

Physical Interaction of hMYH and hMutS␣ in Cell Extracts-
Due to the fact that both hMYH base excision repair and mismatch repair can remove adenines mispaired with gua-nines or 8-oxoG lesions as a result of DNA replication errors ( Fig. 1, reaction 3), we suspect that some relationship may exist between these two repair pathways. To investigate whether hMYH interacts with the human MutS homologs, GST-hMYH fusion protein immobilized on glutathione-Sepharose was added to TK6 lymphoblastoid cell extracts to pull down mismatch repair proteins. As shown in Fig. 2A (lane 2 in the top two panels), both hMSH2 and hMSH6 were found to bind to the GST-hMYH beads. However, hMSH3 did not interact with immobilized GST-hMYH ( Fig. 2A, lane 2 in the bottom panel), and none of the above proteins bound to immobilized GST alone ( Fig. 2A, lane 1 in all panels). This result indicates that hMYH interacts with the hMSH2/hMSH6 (hMutS␣) heterodimer but not with the hMSH2/hMSH3 (hMutS␤) heterodimer.
To determine which subunit of the hMSH2/hMSH6 heterodimer interacts with hMYH, we performed similar experiments with extracts from several mismatch repair-deficient cell lines, MT1, LoVo, and HCT15. MT1 cells, derived from TK6 cells, are resistant to killing by alkylating agents (49). Both alleles of the human MSH6 gene in the MT1 cell contain missense mutations (50). LoVo cells, with a large deletion in the human MSH2 gene, cannot express stable hMSH2 and hMSH6 (51). The human colon tumor line HCT15 carries a frameshift mutation in the human MSH6 gene (52) and has no hMSH6 but has normal expression of hMSH2 and hMSH3 (53). First, the protein levels of hMSH2, hMSH6, and hMYH in these cells were confirmed by Western blotting. As shown in Fig. 2B, TK6 cells had high expression levels of hMSH2 and hMSH6 (lane 1 in panels 1 and 2) while MT1 cells contained a similar level of hMSH2 but a slightly lower level of hMSH6 (lane 2 in panels 1 and 2) as compared with TK6. No hMSH2 and hMSH6 were detected in LoVo cell extracts, and HCT15 cell extracts had no hMSH6 expression as reported previously (Fig. 2B, lanes 3 and 4 in panels 1 and 2). The protein expression levels of hMYH were similar in all four cell lines (Fig. 2B,  third panel). Thus, mismatch repair deficiency has no effect on hMYH protein expression and stability. As shown in Fig. 2A, hMSH2 and hMSH6 were not detected in the GST-hMYH beads when LoVo and HCT15 cell extracts were used ( Fig. 2A,  lanes 4 and 5 in panels 1 and 2) and were weakly pulled down by immobilized GST-hMYH from MT1 cell extracts (Fig. 2A,  lane 3 in panels 1 and 2). Because HCT15 and MT1 cells contain mutant hMSH6 and LoVo cells do not express hMSH6, the physical interaction between hMYH and hMSH2/hMSH6 heterodimer (hMutS␣) is probably mediated through hMSH6. Similar results were obtained by co-immunoprecipitation experiments using an antibody against an hMYH peptide with the above four cell extracts and followed by Western blotting with antibodies against hMSH2 and hMSH6 (data not shown).
Direct Interaction of hMYH with Purified hMutS␣ Protein-Since both hMYH and hMutS␣ interact with PCNA (23,38,40) and DNA, it is possible that an indirect association between hMYH and hMutS␣ may occur. To demonstrate a direct physical interaction of these two proteins, we performed affinity binding experiments with the GST-hMYH and highly purified recombinant hMutS␣ and hMutS␤ proteins expressed in the baculovirus insect cell system. Because both hMutS␣ and hMutS␤ contain hMSH2, an antibody against hMSH2 was used in the Western blot analyses. Two different forms of hMutS␣ were tested in the experiments. The initial experiments used untagged hMutS␣, while untagged hMSH2/N-terminal His 6 -tagged hMSH6 was used in subsequent experiments. Both preparations of hMutS␣ heterodimer produced similar results. As shown in Fig. 3A, GST-hMYH could pull down hMutS␣ (lane 3 in upper panel), but not hMutS␤ (lane 3 in middle panel). hMutS␣ was not detected in the pellet of GST alone (Fig. 3A, lane 3 in lower panel). In similar experiments, hMYH did not interact with hMLH1/human PMS2 (hMutL␣) and hMLH1/human PMS1 (hMutL␤) (data not shown). Therefore, the physical interaction of hMYH and hMutS␣ was confirmed.
The lack of interaction between hMYH and hMutS␤ (hMSH2/hMSH3) implies that hMYH interacts with hMSH6 but not with hMSH2 and hMSH3. Another possibility is that hMYH recognizes the interface between hMSH2 and hMSH6 heterodimer rather than the peptide motif(s) of hMSH6. We therefore conducted Far Western analysis in which the individual proteins of hMutS␣ were separated by denaturing poly-acrylamide gel electrophoresis and transferred onto a membrane. Both hMSH2 and hMSH6 proteins were transferred onto the membrane (Fig. 3B, lane 2). The membrane was then incubated with partially purified recombinant hMYH (21). Western blotting with an antibody generated against an hMYH peptide demonstrated that hMYH interacted with hMSH6 but not with hMSH2 (Fig. 3B, lane 3). This result shows that a direct interaction exists between hMYH and hMSH6 and is consistent with the finding that no interaction between hMYH and hMSH2 is observed in HCT15 cell extracts.
Mapping the Region of hMYH Involved in the Interaction with hMutS␣-By using constructs containing different portions of hMYH fused to GST, we determined the regions of hMYH engaged in the physical interactions with hMSH6. The results are shown in Fig. 4A and summarized in Fig. 4B. The hMSH6-interacting domain was localized to the region that includes residues 232-254 of hMYH due to the fact that construct 5 (⌬231) retained interaction (Fig. 4A, lane 5), but construct 6 (⌬254) exhibited no interaction (Fig. 4A, lane 6) with hMutS␣. This region is very conserved among the MutY family members including mouse MYH (54), Schizosaccharomyces pombe MYH (55), and E. coli MutY (Fig. 5).
Functional Interactions of hMYH and hMutS␣-Ni et al. (37) showed that Msh2p/Msh6p of yeast S. cerevisiae could form a specific complex with A/GO-containing DNA. To test whether hMSH2/hMSH6 has similar A/GO binding capability, we performed gel mobility binding assays with hMSH2/hMSH6 and DNA fragments containing an A/GO mismatch or C:G base pair. Human MutS␣ (at 4 nM) formed a weak complex with the homoduplex (Fig. 6A, lane 1). Two complexes were formed between hMutS␣ and the A/GO mismatch (Fig. 6A, lane 3). The major fast migrating complex was not observed with homoduplex DNA and thus is referred to as specific complex, whereas the slow migrating complex is referred to as nonspecific complex. These data are consistent to the findings obtained by Ni et

FIG. 2. Binding of hMutS␣ to GST-hMYH bound to glutathione-Sepharose in human cell extracts.
A, interaction of hMutS␣ with hMYH is affected by mutation in hMSH6. GST-hMYH or GST protein alone bound to glutathione-Sepharose was added to extracts of TK6 to pull down hMSH2, hMSH3, and hMSH6 that are detected by Western blotting (lanes 1 and 2 in panels 1-3). Extracts of MT1, LoVo, and HCT15 were used to bind GST-hMYH, and pull-down pellets were detected for hMSH2 and hMSH6 by Western blotting (lanes 3-5 in panels 1 and 2). B, Western blot analysis for hMSH2, hMSH6, hMYH, and actin was performed with extracts from TK6, MT1, LoVo, and HCT15 cells. Cell extracts (20 g of protein) were resolved on a 10% polyacrylamide gel containing SDS and transferred to a nitrocellulose membrane for Western blot analysis. FIG. 3. hMSH6 in the hMutS␣ complex binds directly to hMYH. A, pull-down of purified hMutS␣ but not hMutS␤ by GST-hMYH bound to glutathione-Sepharose. Purified untagged hMutS␣ (upper panel) and hMutS␤ (middle panel) (200 ng; expressed in the baculovirus expression system in insect Sf9 cells) were added to bind GST-hMYH (300 ng) immobilized on glutathione-Sepharose 4B as described under "Experimental Procedures." Purified untagged hMutS␣ was added to bound GST alone (lower panel). The input proteins (I), a fraction (ϳ40%) of supernatants (S), and pellets (P) were fractionated by a 10% SDSpolyacrylamide gel followed by Western blot analysis with the antibody against hMSH2 that is present in both hMutS␣ and hMutS␤. B, far Western analysis of hMYH interaction with hMSH2 and hMSH6. Lane 1, purified untagged hMSH2/His-tagged hMSH6 complex (MutS␣) (1 pmol, expressed in the baculovirus expression system in insect Sf9 cells) were separated by 10% SDS-PAGE and visualized by silver stain. Lane 2, the separated proteins (10 pmol) on the gel similar to those on lane 1 were transferred onto a polyvinylidene difluoride membrane and stained with 0.025% Coomassie Blue R-250 in 40% methanol. Lane 3, the separated proteins (7 pmol) on the gel similar to those in lane 1 were transferred onto a nitrocellulose membrane, incubated with partially purified hMYH, and then probed with an antibody against an hMYH peptide (␣-516). Lane 4, similar to lane 3 but incubated with bovine serum albumin instead of hMYH.
al. (37). Thus, hMutS␣ can specifically recognize and form a complex with A/GO mismatches.
To determine whether hMutS␣ and hMYH can affect each other in respect to DNA substrate binding, we added hMutS␣ to the binding reaction of hMYH with the A/8-oxoG-containing DNA substrates. When hMYH is about 0.25 nM in the reaction, excess hMutS␣ could enhance the hMYH specific binding affinity by 8-fold with the A/8-oxoG-containing DNA substrates (comparing lanes 4 and 5 in Fig. 6A). The concentration of bovine serum albumin in the reactions (ϳ1 M) far exceeded that of either hMYH or hMutS␣. Therefore, enhanced hMYH binding in the presence of hMutS␣ was due to specific interactions between both proteins. However, the binding of hMutS␣ with the A/8-oxoG-containing DNA substrates was not affected by the addition of hMYH (comparing lanes 3 and 5 in Fig. 6A). No hMutS␣-hMYH-DNA tertiary complex was detected by the gel mobility assay (Fig. 6A, lane 5).
To test whether hMutS␣ can affect the DNA glycosylase activity of hMYH, we added increasing amounts of hMutS␣ to the hMYH glycosylase reactions. With molar ratios of hMutS␣ to hMYH in the range of 1-16-fold, there is a weak enhancement of hMYH adenine glycosylase activity toward the A/8-oxoG-containing DNA (Fig. 6B, lanes 3-7). From the results of several experiments, we observed that hMutS␣ in 0.25-4 nM ranges could enhance the hMYH glycosylase activity by ϳ2fold. The effect of hMutS␣ on the hMYH glycosylase activity was weaker than that on the hMYH binding affinity toward A/8-oxoG-containing DNA substrates (2-fold increase in glycosylase activity versus 8-fold increase in binding). However, when hMutS␣ was at 5 nM and in about a 20-fold molar excess over hMYH, hMYH adenine glycosylase was not stimulated (Fig. 6B, compare lanes 2 and 8). When hMutS␣ was at 6 nM, hMYH adenine glycosylase was slightly inhibited by hMutS␣ (reduced by ϳ15%) (Fig. 6B, compare lanes 2 and 9). DISCUSSION The human MYH is involved in immediate postreplicative removal of adenines or 2-hydroxyadenines misincorporated with template guanines or 8-oxoG lesions (16 -20). Its function in increasing the fidelity of DNA replication is similar to that of hMutS␣ that removes misincorporated base/base mismatches during replication (24 -28). We demonstrate here, for the first time, that hMYH directly interacts with hMSH6 in the hMSH2/hMSH6 heterodimer. Human MYH does not interact with the hMSH2/hMSH3 heterodimer. This is consistent with the finding that hMutS␤ does not bind to base/base mismatches. It has been shown that hMSH2 and hMSH6 play distinct roles in mismatch repair, with only hMSH6 being cross-linked to a mismatch-containing DNA after UV irradiation (56). Although both hMYH and hMSH6 can bind to their DNA substrates, the interaction between both proteins can occur in the absence of DNA. It is interesting to note that the interaction of hMYH with hMSH6 is substantially weaker in the MT1 cells that express wild-type hMSH2 and missense mutant hMSH6 as compared with the parental TK6 cells. The hMSH6 mutations in MT1 cells are on separate alleles; one is an Asp to Val mutation at codon 1213, and the other is a Val to Ile mutation at codon 1260. Both mutations are at the Cterminal region of hMSH6 downstream of the ATP binding site. The result suggests that the C-terminal region of hMSH6 is important for its interaction with hMYH. However, direct contact of hMYH with the C-terminal region of hMSH6 remains to be determined. MT1 cell extracts are deficient in repair of all eight base-base mismatches (57), but hMutS␣ in this cell can still bind mismatched DNA (58). It has been suggested that the mutations in hMSH6 in MT1 cells may affect its ATPase activity (56).
A functional interaction between hMYH and hMutS␣ was also detected. Human MutS␣ was found to enhance the binding affinity of hMYH for A/8-oxoG-containing DNA substrates. However, hMYH does not have an effect on hMutS␣ binding with A/8-oxoG-containing DNA; nor did we observe an hMutS␣-hMYH-DNA tertiary complex by gel mobility assay. The tertiary structure may not form or may be too unstable to be detected. Thus, hMYH base excision repair may efficiently target the A/GO mismatches with assistance from mismatch repair enzymes. Recently, we have demonstrated that hMYH forms the major UV cross-linked protein-DNA complex in human cell extracts, indicating that hMYH but not hMutS␣ is the major protein to recognize A/8-oxoG-containing DNA substrate (21). This is consistent with the weak binding affinity of hMutS␣ with A/8-oxoG containing DNA. The hMYH adenine glycosylase activity on A/8-oxoG-containing DNA can be slightly stimulated by hMutS␣ with molar ratios of hMutS␣ to hMYH in the range of 1-16-fold. When concentration of hMutS␣ is at a 24-fold excess over that of hMYH, hMYH adenine glycosylase was slightly inhibited. Although the reason for the inhibition is unclear, this high ratio is probably not physiologically relevant. Additionally, at higher hMutS␣ concentrations, hMutS␣ may compete with hMYH upon binding to DNA substrate and thus inhibit the hMYH glycosylase activity.
The effect of hMutS␣ on hMYH binding affinity was more significant than that on hMYH glycosylase activity with A/8-oxoG-containing DNA substrates. It appears that hMutS␣ facilitates the hMYH glycosylase activity by enhancing the DNA recognition of hMYH. This additional function of mismatch repair enzymes may have biological significance. The increased FIG. 7. Models showing the coupling of MYH base excision repair with DNA replication and mismatch repair. MYH, MutS␣, and replication enzymes including DNA polymerases and PCNA are shaded in gray, dotted, and white, respectively. Both MYH-and MutS␣-dependent pathways are coupled to DNA replication to ensure that both repair pathways are targeted to the daughter DNA strands but not to the parental strands. A, when adenines are misincorporated to template 8-oxoG (G O ), MYH repairs the misincorporated adenines on the daughter strands (Fig.  1, reaction 3). MutS␣ may be indirectly involved in this type of repair by enhancing the substrate recognition and glycosylase activity of MYH. B, MutS␣ recognizes A/8-oxoG mismatch, where A is on the parental strand (Fig. 1, reaction 6). In this case, MYH reaction should be prohibited, because such a reaction (Fig. 1, reaction 5), if allowed to proceed, will cause a mutation. levels of C:G to A:T transversions in mismatch repair-deficient cells may be due to an inefficient hMYH base excision repair. It is interesting to note that the spontaneous mutation spectra in the HPRT gene of MT1 cells contain 5% C:G to A:T transversions compared with 13% A:T to G:C transitions and 13% one-guanine insertions (59). It has been reported that the DNA binding and glycosylase activities of mouse MYH can be stimulated by the associated human apurinic/apyrimidinic endonuclease (21,23,54). Thus, protein-protein interactions are important in modulating MYH base excision repair efficiency.
Using various GST constructs, we defined the region within hMYH responsible for the interaction with hMSH6 (Fig. 4). The hMSH6 binding site is at a region including residues 232-254 of hMYH. The corresponding region of E. coli MutY is between 148 and 170 and consists of the loop-␣9 structure (60). This region is just upstream of the adenine recognition motif and is very conserved among the MutY family members including mouse MYH (54), S. pombe MYH (55), and E. coli MutY (Fig. 5). It will be interesting to see whether MutY or MutY homologs interact with MutS or MutS homologs in other organisms.
The interaction between hMYH and hMutS␣ provides a new interplay among various DNA repair pathways. Because hMYH excises adenine from A/8-oxoG mismatch, the removal of misincorporated adenine mismatched with template 8-oxoG (Fig. 1, reaction 3) by hMYH can increase the fidelity of DNA replication. On the other hand, if 8-oxo-dGTP is misincorporated onto template adenine and hMYH removes the adenine from the parental DNA strand (Fig. 1, reaction 5), an A:T to C:G transversion will occur. Thus, such a reaction should be prohibited. To increase DNA replication fidelity, hMYH must target the daughter strands but not the parental strands. However, the mechanism to control hMYH to target the daughter DNA strand is unclear. Based on our data, we propose a model in which MYH repairs misincorporated adenines on the daughter strand (Fig. 7A and Fig. 1, reaction 3) and MYH does not carry out reaction 5 in Fig. 1 through the interactions with MutS␣ and replication proteins. MutS␣ can be involved in the repair of misincorporated adenines on the daughter strand ( Fig. 1, reaction 3) directly or indirectly by enhancing the substrate recognition and glycosylase activity of MYH. Because hMYH has a higher affinity to A/GO mismatches than hMutS␣, hMutS␣ probably plays an indirect role in A/GO repair by enhancing hMYH activity in normal cells. A/GO mismatches where adenines are on the parental strand are mainly recognized by MutS␣ (Fig. 7B and Fig. 1, reaction 6). In this case, MYH does not react, because such a reaction (reaction 5 of Fig.  1), if allowed to proceed, will cause a mutation. With this functional cooperation between MYH and MutS␣, replication errors caused by oxidized bases can be reduced.
It has been suggested that coupling to DNA replication ensures that both MYH base excision repair and mismatch repair are targeted to the daughter DNA strands but not the parental strands. These models are supported by data showing that both hMYH and hMSH6 interact with PCNA and colocalizes with PCNA to replication foci (22, 23, 38 -40). In addition, hMYH interacts with RPA (23), and PCNA interacts with MSH3 and MLH1 (61-63). Bowers et al. (64) suggested that PCNA interacts with the MutS␣ and MutL␣ complex formed on DNA mispairs in S. cerevisiae. The interactions of PCNA with both hMYH and mismatch repair enzymes suggest that PCNA may act as a coordinator of both repair pathways. We hypothesize that proteins involved in DNA replication, mismatch repair, and base excision repair may exist as a multiple-protein complex and that hMYH may be orientated in the replication fork to recognize 8-oxoG on the parental strands and to excise misincorporated A on the daughter strand. A large complex called BRCA1-associated genome surveillance complex has been identified to contain DNA repair and replication proteins, including BRCA1, MSH2, MSH6, MLH1, ATM, RAD50, and DNA replication factor C (65). It has been suggested that PCNA may act as a molecular adaptor, coordinating and regulating the actions of DNA replication, DNA repair, and cell cycle control (66,67). However, the mechanism by which PCNA selects the appropriate partners remains unclear.