DNA Binding and Protein-Protein Interaction Sites in MutS, a Mismatched DNA Recognition Protein from Thermus thermophilus HB8*

The mismatch repair system repairs mismatched base pairs, which are caused by either DNA replication errors, DNA damage, or genetic recombination. Mismatch repair begins with the recognition of mismatched base pairs in DNA by MutS. Protein denaturation and limited proteolysis experiments suggest thatThermus thermophilus MutS can be divided into three structural domains as follows: A (N-terminal domain), B (central domain), and C (C-terminal domain) (Tachiki, H., Kato, R., Masui, R., Hasegawa, K., Itakura, H., Fukuyama, K., and Kuramitsu, S. (1998)Nucleic Acids Res. 26, 4153–4159). To investigate the functions of each domain in detail, truncated genes corresponding to the domains were designed. The gene products were overproduced inEscherichia coli, purified, and assayed for various activities. The MutS-MutS protein interaction site was determined by size-exclusion chromatography to be located in the B domain. The B domain was also found to possess nonspecific double-stranded DNA-binding ability. The C domain, which contains a Walker's A-type nucleotide-binding motif, demonstrated ATPase activity and specific DNA recognition of mismatched base pairs. These ATPase and specific DNA binding activities were found to be dependent upon C domain dimerization.

In living organisms, DNA damage often arises as a result of errors introduced by DNA replication, genetic recombination, and other processes (1). These DNA lesions can result in mutations, genetic diseases, and tumors. To remove such lesions, all organisms have developed DNA repair systems. The mismatch repair (MMR) 1 system is one of such repair systems and is conserved significantly throughout all organisms. In Escherichia coli, MutS, MutL, and MutH proteins are included in the MMR system (2). The pathogenic genes of human hereditary nonpolyposis colorectal cancer appear to share a high degree of homology (ϳ30%) with bacterial MutS and MutL (3,4). Moreover, MutS homologues have also been isolated from plants (5).
These observations suggest that the MMR system is essential for all living organisms from bacteria to eukaryotes. Recently, the three-dimensional structures of E. coli MutL and MutH have been reported, and the relationships between structures and functions have been addressed (6 -8).
MutS plays a key role in the early processes of MMR, mediating mismatched base pair recognition. Eukaryotic MutS homologues have been found to bind to mismatched/looped out DNA (9 -12). It has also been reported that E. coli MutS binds to mismatched DNA as a dimer, forming an ␣-shaped loop structure (13). This finding suggests that this protein has two DNA-binding sites, one of which binds to mismatched DNA and the other to homoduplex DNA. There have been many studies exploring the DNA-binding region of MutS. A C-terminal mutant of Salmonella typhimurium MutS was found to have reduced affinity for heteroduplex DNA (14). Furthermore, the C-terminal region of hMSH2, a human MutS homologue, has been shown to be sufficient for binding to mismatches in DNA (15). On the other hand, a phenylalanine residue (Phe- 39) in the N-terminal region of Thermus aquaticus MutS was found to be important for heteroduplex DNA binding (16). We have reported that the central region of Thermus thermophilus HB8 MutS binds homoduplex DNA (17). Recently, the N-terminal end and central regions of E. coli MutS were demonstrated to be required for binding of mismatched DNA (18).
Overproduced and purified MutS protein (91.4 kDa) from T. thermophilus HB8 possesses ATPase activity and binds specifically to mismatched DNA (19). T. thermophilus HB8 is an aerobic, rod-shaped, and non-sporulating Gram-negative eubacterium, which can grow at temperatures over 80°C (20). Generally, the proteins from T. thermophilus are suitable material for structural and functional analyses because they are stable under a wide range of conditions (19,(21)(22)(23)(24)(25)(26). T. thermophilus MutS protein is stable from pH 1.5 to 12 at 25°C and at a neutral pH up to 80°C (19). Results from protein denaturation and limited proteolysis experiments suggest that T. thermophilus MutS can be divided into three domains as follows: A (N terminus to residue 274), B (residues 275-570), and C (residue 571 to C terminus) (17). We have previously shown that the B domain interacts with double-stranded DNA (dsDNA); however, the assay system used in that study employed a mixture of all three domains. The functions of the other domains, A and C, have not been analyzed. To investigate these domains further, a more refined assay system was required. Truncated genes corresponding to the domains were designed based on the results of proteolytic analysis. Three truncated proteins corresponding to the A-C domains were overproduced in E. coli and purified. To analyze the functions of the individual domains, we assayed for MutS-MutS protein interaction, ATPase activity, and DNA-binding ability. The DNA binding activity of the B domain was confirmed in this study. The results also suggest that the MutS-MutS protein interaction site is located in the B domain, whereas the mismatched DNAbinding site, as well as the ATPase activity, is located in the C domain, and the ATPase site is also in the C domain. On the basis of these results, the relationships between the functions and the structures of MutS are discussed.

EXPERIMENTAL PROCEDURES
Bacterial Strains, Media, Plasmids, and Chemicals-E. coli strains DH5␣ and BL21(DE3) carrying either a pLysE or pLysS plasmid were cultured in Terrific or LB broth at 37°C (27,28). DNA manipulation was performed using standard procedures (27). The enzymes and reagents used were purchased from the following sources: restriction enzymes from Toyobo, New England Biolabs, and Nippon Gene; Taq DNA polymerase from PerkinElmer Life Sciences; [␣-32 P]ATP and [␥-32 P]ATP were from ICN; plastic-backed polyethyleneimine-cellulose sheets (MN-Polygram CEL200PEI/UV) were from Machery and Nagel; glutathione-Sepharose 4B was from Amersham Pharmacia Biotech; and His-Bind resin was from Novagen. Factor Xa was kindly provided by Dr. Kato (National Cardiovascular Center, Japan). Plasmid pSS1, derived from a pET3a (Novagen), contains the complete T. thermophilus mutS gene under control of a T7 promoter (19). Plasmid pET16b (Novagen), which is designed to add 10 His residues at the N terminus of a target protein, was used for overproduction of the N-terminal and central domains of MutS proteins. Plasmid pGEX5X-1 (Amersham Pharmacia Biotech) was used to overexpress N-terminal glutathione S-transferase (GST)-fused C-terminal MutS. All the other chemicals and reagents used were obtained from commercial sources.
Construction of Truncated mutS Genes-DNA fragments expressing each domain of T. thermophilus MutS (Fig. 1) were generated by the polymerase chain reaction using pSS1 as a template. The following pairs of primers were used for amplification of each domain: 5Ј-AGAAAGCGCATATGGGGGGGTATGGCGGATGGCGGAGTTAA-G-3Ј and 5Ј-CTGGGGTCTAGGGTTCAAAGACCTCGAGGG-3Ј (A domain), 5Ј-CTCGAGGTCTTTGAACGCTAGCGTCGACATATGACCC-TCTTTGGCGTTC-3Ј and 5Ј-ACCGGATCCTACCCCGCCCTGATCCG-AAGCC-3Ј (B domain), and 5Ј-ACCCGGCCCCGCTTCGTTTAAAGAT-ATCATATGAGGGCGGGGCGCCACC-3Ј and 5Ј-GGGGGATCCTACC-CCTTCATGCTACCCAGGG-3Ј (C domain). All of the forward primers contained NdeI restriction sites at the first ATG codon (underlined). After subcloning into vectors, the nucleotide sequences of the amplified DNA fragments were confirmed by the dideoxy method (Applied Biosystems, Taq cycle sequencing system) using a model 373S ABI automated DNA sequencer. By using the NdeI restriction sites, the DNA fragments corresponding to the A and B domains were each ligated with pET16b, and the resulting plasmids were named pSS5A and pSS5B, respectively. The DNA fragment corresponding to the C domain was inserted into pGEX5X-1, and the resulting plasmid was named pSS2C. The His-tagged sequences and the GST-fused region were designed so that they could be cleaved from the domains by factor Xa protease.
Overproduction and Purification of the Full-length and Truncated MutS Proteins-Full-length T. thermophilus MutS protein was prepared as described previously (19). The pSS5A and pSS5B transformants were induced as described previously except the host E. coli strain BL21(DE3) pLysS was used, instead of pLysE. The E. coli strain DH5␣ was transformed with pSS2C, and the resultant transformant was induced by isopropyl-␤-D(Ϫ)-thiogalactopyranoside (IPTG).
Purification methods differed for each truncated protein. All purification steps were carried out at 4°C except for the heat treatment and the elution steps from the His-Bind resin (Novagen) and glutathione-Sepharose 4B (Amersham Pharmacia Biotech). The A domain, which carries the N-terminal region of the protein, was purified according to the following procedure. Frozen cells (20 g) carrying pSS5A were suspended in 200 ml of Bind buffer (20 mM Tris-HCl, 500 mM NaCl, 5 mM imidazole, pH 7.9) and were disrupted by sonication on ice. Nonidet P-40 was added to a final concentration of 0.1% (w/v), and the cell extract was stirred for 30 min at 4°C. Following this, the cell lysate was incubated at 70°C for 10 min and then centrifuged (39,000 ϫ g) for 30 min at 4°C. The supernatant containing the A domain protein was filtered through a 0.22-m membrane (Millex-GV, Millipore) and loaded onto a His-Bind resin column (bed volume 5 ml) that was equilibrated with Bind buffer. The column was subsequently washed with 50 ml of Bind buffer and 30 ml of Wash buffer (20 mM Tris-HCl, 500 mM NaCl, 60 mM imidazole, pH 7.9), and eluted in four steps with 15 ml each of 100, 300, 500, and 1000 mM imidazole in Elute buffer (20 mM Tris-HCl, 500 mM NaCl, pH 7.9). The fraction containing the A domain protein was eluted with 300 -500 mM imidazole. Ammonium sulfate was added to the sample to a final concentration of 30% saturation and loaded onto a phenyl-Toyopearl 650M (Tosoh) column (bed volume 70 ml) that was equilibrated with 30% saturated ammonium sulfate in buffer I (50 mM Tris-HCl, 5 mM ␤-mercaptoethanol, 1 mM EDTA, 10% (v/v) glycerol, pH 7.5). The column was washed with 210 ml of the same buffer, after which the protein was eluted with 700 ml of a gradient of 30 -0% saturated ammonium sulfate in buffer I. Fractions containing the protein were loaded onto a Sephacryl S-300HR (Amersham Pharmacia Biotech) column (bed volume of 450 ml) equilibrated with buffer I. The protein was eluted with buffer I. The protein fractions were collected, concentrated, and stored at 4°C. The amount of the purified A domain protein was 42 mg, and the yield was 2.1 mg of protein per 1 g wet cell. The A domain protein was designated as MutS-A.
For purification of the B domain protein, which carries the central region of MutS, E. coli cells carrying pSS5B were induced by IPTG and harvested. Frozen cells (20 g) were disrupted and processed by the same method as described for the MutS-A without the 70°C incubation. The B domain protein elution fractions from the His-Bind resin column, containing 300 -500 mM imidazole, were dialyzed three times against buffer I and loaded onto a CM-Sepharose CL-6B (Amersham Pharmacia Biotech) column (bed volume of 40 ml) previously equilibrated with buffer I. The column was washed with 120 ml of the same buffer, and the protein was eluted with 400 ml of 0 -3 M NaCl gradient in buffer I. Fractions containing the B domain protein were collected and loaded onto a Sephacryl S-300HR column. The protein fractions were collected and dialyzed three times against 10-fold volumes of buffer I. Because the purified B domain protein had a tendency to precipitate at concentrations higher than 10 M, its concentration was kept below 10 M during purification and storage. The amount of the purified B domain protein was 10 mg, and the yield was 0.5 mg of protein per 1 g wet cells. The B domain protein was designated as MutS-B.
To obtain purified the C domain protein, which is composed of the C-terminal domain, E. coli cells carrying pSS2C were induced by IPTG and harvested. Frozen cells (20 g) were suspended in 20 ml of PBS buffer (10 mM Na 2 HPO 4 , 1.8 mM KH 2 PO 4 , 140 mM NaCl, 2.7 mM KCl, pH 7.3). Lysozyme was added to the cell suspension to a final concentration of 0.1 mg/ml, and the mixture was incubated for 15 min at 37°C. Following this, 180 ml of PBS buffer was added to the solution, and the cells were disrupted by sonication on ice. Triton X-100 was added to a final concentration of 1% (w/v), and the cell extract was stirred for 30 min at 4°C, and then was centrifuged (39,000 ϫ g) for 30 min at 4°C. After centrifugation, most of the C domain protein (about 70 -80%) was found in the insoluble fraction; however, a considerable amount of protein remained in the supernatant. The supernatant was filtered through a 0.22-m membrane (Millex-GV, Millipore) and was loaded onto a glutathione-Sepharose 4B column (bed volume 2 ml) that was equilibrated with PBS buffer. The column was washed with 100 ml of the same buffer, and the GST-fused C domain was then eluted with 10 ml of glutathione elution buffer (50 mM Tris-HCl, 10 mM glutathione, pH 8.0) at room temperature. The elution fractions containing the C domain protein were loaded onto a SuperQ-Toyopearl (Tosoh) column (bed volume of 10 ml) equilibrated with buffer I. The column was washed with 30 ml of the same buffer, and the protein was eluted with a 100-ml gradient of 0 -1 M NaCl in buffer I. Fractions containing the C domain protein were collected and loaded onto a Sephacryl S-300HR column. The protein fractions were collected and dialyzed three times against 10-fold volumes of buffer I. The purified C domain protein also had tendency to precipitate at concentrations over 10 M. Therefore, the concentration of the protein was kept below 10 M during purification and storage. The amount of the purified C domain protein was 1.0 mg, and the yield was 0.05 mg of protein per 1 g wet cells. The GST-fused C domain protein was designated as GST-C.
To remove the His tag or the GST region from the purified fused proteins, 100 l of 10 M fusion protein was incubated with 5 l of 0.36 mg/ml factor Xa at 37°C overnight. The factor Xa was removed from the reaction mixture by the following procedure. The reaction solution was loaded onto a Superdex 200HR column (Amersham Pharmacia Biotech) that had been equilibrated with buffer I. Then the column was eluted with the same buffer at room temperature and at a flow rate of 0.5 ml/min. The N-terminal amino acid sequences of the purified proteins with and without cleavage by factor Xa were confirmed by an automated protein sequencer (ABI, model 473A). The cleaved products from GST-C were designated as GST-N and MutS-C (Fig. 1). The molar extinction coefficients of each purified truncated protein were calculated to be 19,500 (MutS-A), 51,000 (MutS-B), and 23,400 M Ϫ1 cm Ϫ1 (GST-C), using a procedure described previously (29). CD measure-ments were carried out in a sample solution containing 50 mM Tris-HCl, 100 mM KCl, 0.1 mM dithioerythritol, pH 7.5, and 1 M of each protein using a 0.1-cm cell. The residue molar ellipticity [] was defined as 100 obs (lc) Ϫ1 , where obs is the observed molar ellipticity; l is the length of the light path in centimeters, and c is the residue molar concentration of each protein. CD measurements were performed using a Jasco spectropolarimeter, model J-720W.
Size-exclusion Chromatography-In order to analyze the relationship between the apparent molecular mass and the protein concentration of full-length MutS, 50 l of the protein sample at various concentrations was injected onto a Superdex 200HR column (inner diameter, 1 ϫ 30 cm, Amersham Pharmacia Biotech) that had been equilibrated with buffer comprising of 50 mM Tris-HCl, 100 mM KCl, 5 mM ␤-mercaptoethanol, 0.1 mM EDTA, pH 7.5. The measurements were carried out at room temperature at a flow rate of 0.5 ml/min, and the absorption was detected at 280 nm. The following proteins were used as molecular size markers: thyroglobulin (669 kDa), apoferritin (443 kDa), ␤-amylase (200 kDa), alcohol dehydrogenase (150 kDa), albumin (66 kDa), carbonic anhydrase (31 kDa), and cytochrome c (12.8 kDa).
In order to measure the molecular mass of each truncated MutS protein, 100 l of each purified protein (3 M) was injected onto a Superdex 200HR column that was equilibrated with buffer I or buffer II (50 mM Tris-HCl, 1 M NaCl, 5 mM ␤-mercaptoethanol, 1 mM EDTA, 10% glycerol, pH 7.5). As the nonspecific interaction between the MutS-B and the column resin was rather strong, the measurement of MutS-B was carried out in the presence of 1 M NaCl. The following proteins were used as molecular size markers: phosphorylase b (97 kDa), albumin (66 kDa), ovalbumin (45 kDa), carbonic anhydrase (31 kDa), and cytochrome c (12.8 kDa).
ATPase Assay-The ATPase activity of each protein was analyzed using a TLC method (19). The reaction mixture contained 1 M of each protein, 1 mM [␣-32 P]ATP, 10 mM MgCl 2 , 100 mM KCl, 1 mM DTT, and 50 mM Tris-HCl, pH 7.5. Reactions were carried out at 25°C for 10 min and stopped by addition of 0.5% SDS. The samples were spotted on PEI cellulose sheets and eluted with 0.75 M potassium phosphate, pH 3.4. The sheets were placed in contact with an imaging plate, and the resultant hydrolyzed products were analyzed by using a BAS2000 image analyzer (Fuji photo film).
Surface Plasmon Resonance (SPR) Measurements-DNA-protein interactions were measured using an SPR sensor system, BIAcore 1000 (Amersham Pharmacia Biotech). Single-stranded oligonucleotides were immobilized on the surface of the streptavidin-coated sensor chips (SA5-chip, Amersham Pharmacia Biotech), according to the following method. One microliter of 5Ј-biotinylated TAKA1 (0.5 M) oligonucleotide in buffer A (50 mM Tris-HCl, 100 mM KCl, 1 mM MgCl 2 , 1 mM DTT, 0.1 mM EDTA, pH 7.5) was applied at a flow rate of 5 l/min over the sensor chip at 25°C. Following this, 10 l of TAKA2 or TAKA4 (5 M) oligonucleotide was applied over the TAKA1-immobilized sensor chip at the same flow rate to make double-stranded oligonucleotide probes.
From the intensities of the resonance unit before and after the annealing of the second oligonucleotide, it was estimated that about 90% of the second oligonucleotide was annealed to the first oligonucleotide. After immobilization of the oligonucleotide, 30 l of a solution containing 0, 0.2, 0.4, 0.7, 1.0, or 2.0 M of each protein was applied at a flow rate of 20 l/min at 25°C over flow cells in sensor chips, either without oligonucleotide, with TAKA1, with TAKA1-TAKA2, or with TAKA1-TAKA4. Overlay plots were drawn from the data obtained between 10 and 90 s after injection. The data were averaged and corrected by the blank controls, obtained with eluent buffer A and the sensor chip without DNA. Data calculation was carried out using the BIAevaluation software (Amersham Pharmacia Biotech) with algorithms for numerical integration (30). Apparent rate constants (k obs ) were plotted against protein concentration, and the association rate constants (k on ) and dissociation rate constants (k off ) were calculated from the slope and the intercept of these plots, respectively. The value of dissociation constant, K d , was calculated from those of k on and k off .

Construction of Truncated Proteins Corresponding to Each
Domain-As we have already shown, MutS protein can be divided into at least three domains based on the results of limited proteolysis and the denaturation (17). In order to analyze the functions of each domain in more detail, truncated proteins that correspond to the proteolytic fragments were designed (Fig. 1). Each truncated gene was inserted into an expression vector for overexpression in E. coli as described under "Experimental Procedures." These proteins were designated MutS-A (A domain), MutS-B (B domain), and GST-C (C domain), respectively (Fig. 1). MutS-A and MutS-B contain a short His repeat on their N termini, and GST-C was constructed as a fusion protein with GST on the N terminus.
MutS-A, MutS-B, and GST-C were purified as described under "Experimental Procedures" (Fig. 2). GST-C could be cleaved by factor Xa protease into two products, GST-N (GST region) and MutS-C (C domain region) (Fig. 1). CD measurements in the region between 200 and 250 nm showed that the purified proteins were retained the secondary structures (Fig.  3). MutS-A, but not MutS-B or GST-C, was stable at temperatures up to 80°C, similar to the full-length T. thermophilus MutS (data not shown). This suggests that the N-terminal region of MutS contributes to the stability of the full-length protein.
Protein-Protein Interaction-We previously showed that the apparent molecular mass of T. thermophilus MutS at 86 M was about 330 kDa, which corresponds to a trimer or tetramer (19). With T. aquaticus MutS, a similar apparent molecular mass of 280 kDa was reported by size-exclusion chromatography (16,31); however, by other physicochemical methods, the molecular mass was estimated to be about 180 kDa, which corresponds to a dimer (31). Recently, it has been reported that E. coli MutS exists as mixture of dimers and higher order forms  (17). The N-terminal and central domains were constructed as His-tagged proteins, and the resultant proteins were designated as MutS-A and MutS-B, respectively. The C-terminal domain was constructed as a GST fusion protein, and the resultant protein was designated as GST-C. The GST-C protein can be cleaved into two products, GST-N and MutS-C, by the sequence-specific endoprotease, factor Xa (arrow). (32). In order to elucidate the oligomerization state of T. thermophilus MutS in more detail, size-exclusion chromatography experiments at various protein concentrations were carried out. As shown in Fig. 4A, the apparent molecular mass of T. thermophilus MutS at lower concentrations (1 or 5 M MutS) was estimated to be 220 kDa, which corresponds to a dimer. As the protein concentration increased, the apparent molecular mass increased to about 350 kDa. This suggests that the minimal unit of T. thermophilus MutS is a dimer. The highly oligomerized state of MutS at higher protein concentrations may have some physiological functions that need to be elucidated.
In order to determine where the MutS-MutS interaction site is located, the molecular sizes of the truncated proteins in solution were measured using size-exclusion chromatography. The apparent molecular mass of MutS-A was calculated to be 30 kDa (Fig. 4B, open circle), which is almost identical to the size predicted by the amino acid sequence (32.6 kDa). The MutS-B eluted at a volume corresponding to apparent molecular size of 65 kDa (Fig. 4B, square) which is two times larger than the predicted molecular mass (35.8 kDa). The GST-C, which is the GST-fused C-terminal region of MutS, eluted at a volume corresponding to 110 kDa (Fig. 4B, diamond) which is two times larger than the predicted molecular size (54.6 kDa). Since GST alone forms a dimer (33), we measured the molecular masses of GST-N and MutS-C, which were derived from cleavage of GST-C by factor Xa. The molecular mass of GST-N was calculated to be 60 kDa (Fig. 4B, inverted triangle), which is two times larger than the predicted size (26.3 kDa), and that of MutS-C was calculated to be 25 kDa (Fig. 4B, triangle), which is almost identical to the predicted size (28.3 kDa). In ATPase Activity of Truncated Proteins-MutS contains a Walker's A-type nucleotide-binding motif (34) and shows weak ATPase activity (14,19,(35)(36)(37). It has been known that this motif is essential for ATP hydrolysis. In our construction, the C domain contains the motif. In order to determine which domain is responsible for ATPase activity, the activity of each truncated protein was measured using the TLC method. The ATPase activity of T. thermophilus MutS was maximal at 80°C (19); however, the assay was carried out at 25°C because MutS-B and GST-C were not heat-stable. As shown in Fig. 5, GST-C showed ATPase activity, but MutS-A and MutS-B did not. Interestingly, MutS-C, a product of factor Xa cleavage of GST-C, showed no ATPase activity. Since it has not been reported that GST shows any ATPase activity, the loss of the ATPase activity of MutS-C may be related to the lack of dimer formation without the GST fusion.
DNA Binding Activity of the Truncated Proteins-T. thermophilus MutS binds with higher affinity to G/T mismatched heteroduplex DNA than to homoduplex DNA (19). To clarify whether or not the truncated proteins bind to DNA or not, the binding activity of each domain to a 32 P-labeled 37-base pair dsDNA with or without a G/T mismatched base pair was examined by a gel retardation assay. Although the full-length MutS bound with higher affinity to mismatched DNA than to homoduplex DNA, MutS-A could not bind to either DNA, whereas MutS-B could bind to both (Fig. 6, upper panel). In contrast, GST-C bound with higher affinity to heteroduplex DNA, in a similar manner to the full-length MutS (Fig. 6, lower  panel). These results suggest that there are two DNA-binding sites in MutS, a nonspecific site in the B domain and a specific site in the C domain. Interestingly, GST-N and MutS-C, which are the cleaved products from GST-C, could not bind to either mismatched or complementary DNA (Fig. 6, lower panel). This observation suggests that the specific binding to mismatched DNA by GST-C is related to dimer formation.
The binding specificity of T. thermophilus MutS for mismatched DNA increases with temperature (19). Small differences between homo-and heteroduplex DNA-binding abilities were observed by the gel retardation experiments shown in Fig.  6; however, the experiments could not be carried out at higher temperatures because of the protein temperature instability. To confirm the DNA-binding specificity with more detail, SPR measurements were carried out. Single-stranded and doublestranded oligonucleotides were immobilized on the surface of the sensor chip, and the sample solution containing full-length MutS or each truncated protein was applied over the chip. As shown in Fig. 7A, increments of resonance units corresponding to binding of full-length MutS to dsDNAs were observed. When the full-length MutS flowed over the chip, higher resonance units were observed for heteroduplex DNA than for homoduplex DNA. This suggests that the amount of MutS bound to heteroduplex DNA was greater than the amount of MutS bound to homoduplex DNA. On the other hand, effective interaction of full-length MutS with single-stranded DNA (ssDNA) was not observed. Next, MutS-B or GST-C was applied over the DNA-immobilized sensor chip, and the changes in the resonance units were observed. As shown in Fig. 7B, MutS-B could interact with both dsDNAs but not ssDNA. When MutS-B was applied onto the sensor chip, the resonance units increased immediately and reached plateaus at similar values for both homo-and heteroduplex DNA. Due to fast association and dissociation of MutS-B and dsDNAs, the rate constants for association and dissociation could not be determined. The sensorgrams in Fig. 7C showed that GST-C interacts with the dsDNA with the mismatch specifically but not with ssDNA. No interaction was observed between GST alone and any DNA (data not shown). Thus, the specific interaction between GST-C and heteroduplex DNA is purely due to the C-terminal region of MutS and not from GST. This result indicates that the C-terminal region has the mismatch-specific dsDNA binding activity.
In order to estimate the kinetic parameters, the SPR measurements were carried out at various concentrations of each protein. Dissociation constant (K d ) were calculated to be 0.52 (homoduplex) and 0.45 M (heteroduplex) for full-length MutS, and 4.2 (homoduplex) and 3.2 M (heteroduplex) for GST-C, respectively. These results indicate that the binding specificity of MutS between homo-and heteroduplex DNA is rather small. We found that the C-terminal region of MutS (GST-C) could interact with heteroduplex DNA; however, the affinity was about 1 order lower than that of full-length MutS. This difference may be caused by a nonspecific DNA binding activity of the B domain.

DISCUSSION
Based on protein denaturation and limited proteolysis studies, we proposed a structural organization of T. thermophilus MutS into three structural domains as follows: A (N terminus to residue 274), B (residues 275-570), and C (residue 571 to C terminus) (17). Truncated proteins corresponding to the domains were constructed and purified. Each purified protein maintained its secondary structure and demonstrated characteristic activities. Our results show that the three structural domains of T. thermophilus MutS correspond to functional domains. Purified MutS-A was stable up to 80°C like the full-length T. thermophilus MutS; however, the other two truncated proteins were not heat-stable. This suggests that the A domain may contribute to increase the protein stability of T. thermophilus MutS. The B and C domains may be stabilized by protein-protein and/or domain-domain interaction.
T. thermophilus MutS, as well as the other MutS homologues, possesses weak ATPase activity (14,19,(35)(36)(37). It has been reported that the activity requires the Walker's nucleo-tide-binding motif which is located on the C-terminal region of the protein (14,15,18). A C-terminal mutant of S. typhimurium MutS containing a modified Walker's A-type motif bound to heteroduplex DNA with reduced affinity (14). The C-terminal region of the human MutS homologue, which corresponds to the C domain of T. thermophilus MutS, was required for both ATPase activity and binding to mismatched heteroduplex DNA (15). In addition, our results show that the C domain of T. thermophilus MutS interacts specifically with mismatched DNA. These findings suggest that the C-terminal region is important not only for ATPase activity but also for mismatched DNA recognition and protein-DNA interactions.
Other candidate DNA-binding sites have been reported. A phenylalanine residue  in the N-terminal region of T. aquaticus MutS was affinity-labeled with a single base, looped out duplex DNA (16). Moreover, N-terminal and central regions of E. coli MutS are required for binding to mismatched DNA (18). These findings suggest that not only the C-terminal region of MutS but also the other regions are involved in mismatched DNA interactions. In our study, the gel retardation and SPR experiments indicate that the B domain interacts with DNA nonspecifically. We have already shown that the B domain of T. thermophilus MutS binds to homoduplex DNA (17). These results suggest that the B domain is also involved in the DNA binding. The nonspecific DNA binding ability of the B domain may help increase the total binding affinity of MutS to DNA.
The results of DNA-binding experiments show that T. thermophilus MutS and MutS-C bind specifically to mismatched DNA. In the case of both proteins, the values of the K d for DNA with and without mismatches were significantly different. In our study, the SPR experiments were performed at 25°C because of mechanical limitations and protein instability; however, the growing temperature of T. thermophilus is about at 75°C (20). As previously shown, the ability of T. thermophilus MutS to specifically bind mismatched DNA decreases as the temperature decreases (19). The difference of K d with and without a mismatch may be greater at higher temperatures.
There were some inconsistent observations concerning the oligomerization state of MutS. Previously, we have shown by size-exclusion chromatography that the apparent molecular mass of T. thermophilus MutS in solution is about 330 kDa which corresponds to a trimer or tetramer (19). A similar size, 280 kDa, has also been reported by the same method using T. aquaticus MutS (16,31). However, studies using cross-linking, light scattering, mass spectrometry, and equilibrium sedimentation measurements concluded that the protein is a dimer (31). In a study on E. coli MutS using size-exclusion chromatography, it was reported that the protein forms dimers or oligomers (32,38). We conclude from this study that T. thermophilus MutS exhibits concentration-dependent oligomerization, forming dimers at low concentrations and higher order oligomers at higher concentrations (Fig. 4A). Similar observations have been reported from studies on T. aquaticus MutS using equilibrium sedimentation analysis (31) and on E. coli MutS using size-exclusion chromatography (32). Recently, we observed that T. thermophilus MutS forms a dimer in solution using small angle x-ray scattering measurements. 2  activities, nonspecific DNA-binding and MutS-MutS protein interaction.
Among the truncated MutS proteins, the mismatch-specific DNA-binding ability and the ATPase activity were observed only in the GST-fused C domain. These activities were lost when GST-C was cleaved by factor Xa. Size-exclusion chromatography experiments showed that the GST-fused C domain exists as a dimer, but the C domain digested by factor Xa exists as a monomer. Since GST forms dimers (33), the dimerization of GST-C may be mediated by its GST domain. Thus, it is possible that although the C domain dimers possess DNA binding and ATPase activities, monomers do not. These results suggest that the minimal functional unit of MutS is a dimer. It has been demonstrated that E. coli MutS (13) and T. aquaticus MutS (31) bind to DNA as a dimer. It has been reported that eukaryotic MutS homologues form heterodimers in solution; MSH2 and MSH6 form MutS␣ (9,11), and MSH2 and MSH3 form MutS␤ (11). These observations suggest that the bacterial MutS and its eukaryotic homologues function as dimers and that the molecular mechanism of MutS is similar among prokaryotes and eukaryotes.
Finally, we propose the following model of MutS function at the initial step of MMR. In solution, MutS forms a dimer which is the minimal functional unit. Dimerization of MutS is mediated by the B domain of the protein. The B domain binds dsDNA after which the mismatched region in the dsDNA is recognized by the C domain. Electron microscopy revealed that E. coli MutS binds to mismatched DNA with an ␣-shaped loop structure, suggesting two DNA-binding sites, one of which binds to mismatched DNA and the other to homoduplex DNA (13). One possibility is that the two DNA-binding sites correspond to the C domain and the B domain, respectively. Another possibility is that the DNA-binding ability of each subunit of the dimer is different. The latter possibility is not likely because MutS appears to have DNA binding ability only upon dimerization. The effects of adenine nucleotides on MutS binding to DNA should be considered as these cofactors have been shown to change the DNA-binding mode of MutS (9, 13, 35, 39 -41). Although the nucleotide-binding site is located on the C domain of MutS, the DNA-binding sites are located on the B and C domains. Moreover, the C domain has ATPase activity only upon dimerization, presumably with the help of the B domain. Therefore, changes in the DNA-binding mode induced by adenine nucleotides may be due to the effect of the C and B domain. Nucleotide binding may cause a local environmental change around the nucleotide-binding site in the C domain, which could then propagate to other domains. According to this hypothesis, structural rearrangement of domains and/or subunits may accompany nucleotide binding. It has been reported that a large conformational change of MutL is induced by nucleotide binding (8). It is interesting that both MutS and MutL, which work together in the early steps of MMR, may undergo a conformational change upon adenine nucleotide binding. Such changes may affect not only DNA binding but also protein-protein interaction, including the MutS-MutL association.