Functional Significance of the Conserved Residues for the 23-Residue Module among MTH1 and MutT Family Proteins*

Human MTH1 and Escherichia coli MutT proteins hydrolyze 7,8-dihydro-8-oxo-dGTP (8-oxo-dGTP) to monophosphate, thus avoiding the incorporation of 8-oxo-7,8-dihydroguanine into nascent DNA. Although only 30 amino acid residues (23%) are identical between MTH1 and MutT, there is a highly conserved region consisting of 23 residues (MTH1, Gly36–Gly58) with 14 identical residues. A chimeric protein MTH1-Ec, in which the 23-residue sequence of MTH1 was replaced with that of MutT, retains its capability to hydrolyze 8-oxo-dGTP, thereby indicating that the 23-residue sequences of MTH1 and MutT are functionally and structurally equivalent and constitute functional modules. By saturation mutagenesis of the module in MTH1, 14 of the 23 residues proved to be essential to exert 8-oxo-dGTPase activity. For the other 9 residues (40, 42, 44, 46, 47, 49, 50, 54, and 58), positive mutants were obtained, and Arg50 can be replaced with hydrophobic residues (Val, Leu, or Ile), with a greater stability and higher specific activity of the enzyme. Indispensabilities of Val39, Ile45, and Leu53 indicate that an amphipathic property of α-helix I consisting of 14 residues of the module (Thr44–Gly58) is essential to maintain the stable catalytic surface for 8-oxo-dGTPase.

There are various causes for mutations of genomes of living organisms. The oxidation of nucleic acids by oxygen radicals, generated during normal cellular metabolism such as oxygen respiration, is considered to be one of major causes of spontaneous mutation (1)(2)(3). Oxygen radicals attack free nucleotides as well as DNA, and it has been established that 7,8-dihydro-8-oxo-dGTP (8-oxo-dGTP), 1 an oxidized form of dGTP, is highly mutagenic and a main endogenous source for spontaneous mutagenesis. During DNA replication, 8-oxo-dGTP can be inserted into the nascent strand opposite adenine and cytosine in the template with almost equal efficiency, an event that leads to an A:T to C:G transversion mutation (4 -6).
MutT protein hydrolyzes 8-oxo-dGTP to the monophosphate form, thus eliminating mutagenic substrates from the DNA precursor pool (4). Lack of the mutT gene increases the spontaneous occurrence of A:T to C:G transversion 1000-fold over the wild type level (7)(8)(9). Moreover, it has been shown that Escherichia coli RNA polymerase misinserts 8-oxo-GTP, an oxidized form of GTP, into mRNA, yielding mutant forms of proteins known as nongenomic mutations in mutT-deficient cells. MutT protein efficiently hydrolyzes 8-oxo-GTP and thus minimizes errors caused by misincorporation of oxidized guanine nucleotides into RNA (10).
Enzymatic activity similar to that of MutT protein has been identified in human cells (11,12). Based on partial amino acid sequences obtained from a purified preparation of human 8-oxo-dGTPase, cDNA and the gene for the human enzyme were isolated and named MTH1 (mutT homolog-1) (12,13). Because 30 amino acid residues (23%) are identical between MutT and MTH1 proteins, expression of the human enzyme in mutT Ϫ E. coli cells suppresses the elevated level of spontaneous mutation frequency to almost normal, indicating that the human enzyme has the same antimutagenic capacity as does the E. coli MutT protein. MTH1 mRNA is abundant in the human thymus, testis, and the embryonic tissues. In peripheral blood lymphocytes, expression of MTH1 mRNA is induced after proliferative activation, suggesting that MTH1 expression is up-regulated in proliferative tissues (14). MTH1 protein is localized mostly in the cytoplasm with some in mitochondria (15), which means that MTH1 proteins are involved in sanitization of nucleotide pools, both for nuclear and mitochondrial genomes.
Genes for MutT homolog proteins with dGTPase or 8-oxo-dGTPase activity were identified in Proteus vulgaris and Streptococcus pneumoniae, bacteria distantly related to E. coli (16,17). All three MutT homolog proteins have molecular masses ranging from 13 to 18 kDa, and the degree of identity ranges from 19 to 40%. Most of the identical residues are in a region corresponding to the 23 residues from Gly 37 to Gly 59 of E. coli MutT, known as the MutT signature (18). Homologs of human MTH1 protein were identified in mice and rats when isolating their cDNA (19,20), and as the degree of identity ranges from 85 to 90%, mammalian MTH1 proteins are highly diverged from bacterial counterparts. Structural analyses of the E. coli MutT protein revealed that a region containing the MutT signature constitutes the active center of dGTPase, with a loop and ␣-helix structure (21,22).
The 23-residue sequence is a sole conserved sequence among all MutT and MTH1 homologs with 8-oxo-dGTPase, and of the many other proteins with the MutT signature so far identified, some hydrolyze various nucleotide derivatives, such as dATP, diadenosine oligophosphates, NADH, ADP-ribose, and GDPmannose (23)(24)(25). Thus enzymes with the MutT signature are designated as belonging to the Nudix hydrolase family, and the signature is known as Nudix box (18). Furthermore, a diphosphoinositol polyphosphate phosphohydrolase that hydrolyzes the non-Nudix substrate, diphosphoinositol polyphosphate, also contains the 23-residue sequence (26). It has been sug-gested that the 23-residue sequence may constitute a functional module essential for hydrolysis of a phosphodiester bond in such Nudix or diphosphoinositol derivatives; however, there has been no direct evidence indicating that the 23-residue sequence constitutes a functional module.
We now provide the first evidence that the 23-residue sequences of human MTH1 and E. coli MutT proteins are exchangeable so that they constitute functionally equivalent modules for 8-oxo-dGTPase, designated as the phosphohydrolase module. We further carried out a saturation mutagenesis of the module of human MTH1 protein to reveal the role of each amino acid residue.
Oligonucleotides-Oligonucleotide primers listed in Tables I and II were obtained from Greiner Labortechnik Co. Ltd. (Tokyo, Japan). Nineteen of 27-mer oligonucleotides carrying every possible combination of nucleotides for a target codon (Table II) were used as primers for saturation mutagenesis.
Construction of a Chimeric cDNA for Chimeric Protein, MTH1-Ec-To replace the 23-residue module between MTH1 and MutT proteins, two unique restriction sites were introduced into human MTH1 cDNA; BalI site at the 36th codon (Gly 36 ) and SmaI site at the 58th codon (Gly 58 ), and into E. coli mutT gene; SmaI site at the 37th codon (Gly 37 ) and HincII site at the 59th codon (Gly 59 ) of mutT gene, by polymerase chain reaction-mediated mutagenesis using mutagenesis primers shown in Table I (29). The BalI-SmaI fragment of human MTH1 cDNA cloned into pTT100 was replaced with the SmaI-HincII fragment from mutT gene, yielding a plasmid (pTT100:MTH1-Ec) encoding the chimeric protein, designated MTH1-Ec.
Saturation Mutagenesis-Saturation mutagenesis was performed for the plasmid pALTER:MTH-ss, using the mutagenesis primers listed in Table II, according to the Promega technical manual of altered sites II in vitro mutagenesis system but with the modifications described (28). Mutagenized DNA were digested with a set of restriction enzymes SacI and NcoI and subcloned into pTT100:hMTH1. A mixture of plasmids pTT100:hMTH1(G36X), for example, carrying mutations at the codon for Gly 36 thus obtained was applied to E. coli CC101T and cultured on agar medium containing minimal A salts, 0.2% glucose, 0.05% phenyl-␤-D-galactopyranoside, and 40 g/ml of 5-bromo-4-chloro-3-indolyl-␤-Dgalactopyranoside. After incubation for 5 days at 37°C, papillae formation was examined, and appropriate samples were subjected to further analyses. Fluctuation Test-Bacteria were grown overnight in LB medium in the presence of 50 g/ml of ampicillin, and aliquots containing about 100 cells were inoculated into 5 ml of LB medium containing 50 g/ml of ampicillin at 37°C for 48 h. Mutant frequencies were determined by plating aliquots of cultures on normal and 0.4% lactose-containing minimal A medium plates with 50 g/ml of ampicillin. Three independent experiments were done, and mutation rates were calculated (30,31).
Western Blotting-Bacterial cell crude extracts, prepared as described (27), were subjected to 15% SDS-polyacrylamide gel electrophoresis and then to Western blotting. Purified 18-kDa Val 83 -MTH1d polypeptide (27) was used as a standard for quantification of mutant MTH1 protein in the bacterial extracts. Western blotting was done as described (28), using anti-MTH1 or anti-M77 antibodies.
8-Oxo-dGTPase Activity-8-oxo-dGTPase activity was measured as described (11,28). One unit of 8-oxo-dGTPase was defined as the amount of enzyme that produced 1 pmol of 8-oxo-dGMP/min at 30°C. Specific activity of mutant MTH1 protein was calculated based on the content of each mutant MTH1 protein in the extract, determined by Western blotting.

The 23-Residue Sequences of Human MTH1 and of E. coli
MutT Are Functionally Equivalent-To determine whether the 23-residue sequence of MTH1 is functionally equivalent to that of MutT, we constructed plasmid pTT100:MTH1-Ec, encoding a chimeric protein MTH1-Ec, in which the sequence of MTH1 (Gly 36 -Gly 58 ) was replaced with that of MutT (Gly 37 -Gly 59 ), as shown in Fig. 1. Plasmids, pTT100, pTT100:hMTH1, and pTT100:MTH1-Ec were introduced into E. coli strain CC101T (mutT Ϫ , lacZ Ϫ ), which exhibits an elevated spontaneous mutation rate of A:T to C:G transversion, which can be detected by LacZ ϩ reversion (Table III). At 30°C, expression of MTH1 reduced the elevated mutation rate (1.13 ϫ 10 Ϫ6 ) seen in CC101T, to 1 ⁄73 of the level, and expression of MTH1-Ec in CC101T also reduced the spontaneous mutation rate to 1.99 ϫ 10 Ϫ7 , which reaches 1 ⁄13 of the level attained by expression of MTH1 itself. This result clearly indicates that the chimeric protein MTH1-Ec has a significant capacity to suppress the increased spontaneous mutation rate in mutT Ϫ -deficient cells, albeit the level being lower than that seen with the wild type MTH1. Furthermore, we found that the potential of MTH1-Ec to suppress the occurrence of spontaneous mutations in CC101T cells is thermolabile, because expression of MTH1-Ec at 37°C only suppressed the elevated mutation rate (2.47 ϫ 10 Ϫ6 ) seen in CC101T to 1 ⁄3 of the level (Table III).
To obtain biochemical parameters that may explain the temperature-sensitive nature of MTH1-Ec, levels of expression and catalytic activity were compared with those for MTH1. As shown in Fig. 2A, amounts of each protein were determined by Western blotting, using purified MTH1 protein as standards with anti-M77. At 37°C, the cellular content of MTH1 protein was 170 ng/mg of total protein, whereas that of MTH1-Ec protein was 52 ng/mg. At 30°C, though the content of wild type MTH1 decreased to 53 ng/mg, that of MTH1-Ec increased to 90 ng/mg (Table III), indicating that MTH1-Ec is more stable at a lower temperature.
8-oxo-dGTPase activity in crude extracts was measured as shown in Fig. 2B, and the specific activity of MTH1 or MTH1-Ec was calculated based on their contents in crude extracts (Table III). Specific activity of MTH1 was determined to be 240.8 ϫ 10 2 units/g of MTH1 protein at 37°C and 248.8 ϫ 10 2 units/g at 30°C, respectively; values are close to that of purified MTH1 protein (300 ϫ 10 2 units/g protein) (27). On the other hand, the specific activity of 8-oxo-dGTPase for MTH1-Ec protein was 62.5 ϫ 10 2 units/g at 37°C, that is about 1 ⁄4 that for MTH1, indicating that total 8-oxo-dGTPase activity in cells expressing MTH1-Ec at 37°C is less than 1 ⁄12 that in cells expressing MTH1 itself. At 30°C, the specific activity increased 2-fold to 135.0 ϫ 10 2 units/g, comparable with the 54% level of MTH1 protein itself.
Thus, we concluded that the chimeric protein MTH1-Ec retains 8-oxo-dGTPase activity comparable with that for MTH1 itself and thus has a significant capability to suppress the increased occurrence of A:T to C:G transversion in mutT-deficient cells, but it is thermolabile in terms of structure and function. Based on these findings, we propose that the 23residue sequence constitutes an independent functional module for MTH1 and MutT family proteins.
Saturation Mutagenesis of the 23-Residue Module of MTH1 Protein-Residues in the 23-residue module of MTH1 are classified into three groups: conserved, semi-conserved, and nonconserved residues, as based on conservation among six MutT and MTH1 homologs, as shown in Fig. 3. We earlier reported that four conserved residues (Lys 38 , Glu 43 , Arg 51 , and Glu 52 ) in human MTH1 protein cannot be replaced with any other amino acid residue without losing function, the approach used was negative and positive mutant screening, and we concluded that such are essential for MTH1 protein (28). For a complete understanding of roles of each amino acid residue in the 23residue module, we carried out saturation mutagenesis of the other 19 residues in the module of human MTH1 protein, in combination with negative and positive mutant screening, using mutagenesis primers that contain a completely degenerated codon for each residue (Table II).
First, we isolated negative mutants that lost the capability to suppress the elevated level of LacZ ϩ reversion in E. coli CC101T cells after introduction of the mutagenized plasmid pTT100:hMTH1 into the cells (Tables IV-VI). Next, we used one of the negative mutants for each residue as a template for the second round of saturation mutagenesis to isolate positive mutants that acquired the capability to suppress the LacZ ϩ reversion in CC101T cells. As shown in Table VII, mutation rates of CC101T cells harboring negative mutants used for the second mutagenesis as templates were as high as that of the cells with vector pTT100. Expression levels of the negative mutant proteins were lower than that of wild type MTH1, and their 8-oxo-dGTPase activity was detectable in only a few cases   (21,32). Secondary structures of MTH1 (Val 83 ) were predicted using PHD methods (27,36). In the chimeric protein, MTH1-Ec, the 23-residue sequence, which consists of loop I and ␣-helix I in MTH1 protein, is replaced with that of E. coli MutT protein (shaded), as described under "Experimental Procedures." The numbers represent numbers of amino acid residues from the Nterminal end. Ņ Ņ, ␤-strand (A, B, C, D, E, and F); () ʛ, ␣-helix (I and II); , loop (I-IV). Thin lines in MTH1 and MTH1-Ec indicate no prediction made for the residues. and was very low.
For four conserved residues, Gly 36 , Gly 37 , Glu 55 , and Glu 56 , no positive mutant, except for each true revertant, was isolated, indicating that these residues cannot be replaced with any other residue without losing function. For Thr 44 and Gly 58 , two other conserved residues, various positive mutants were obtained (Table IV). Among four semi-conserved residues, for three of them (Gly 42 , Glu 46 , and Gln 54 ), one or two positive mutants besides true revertants were obtained. However, only the true revertants were obtained as positive mutants for Leu 53 from 1105 mutagenized clones examined (Table V). For five nonconserved amino acid residues (Val 39 , Glu 41 , Ile 45 , Gly 48 , and Ser 57 ), no positive mutant except the true revertants was isolated from at least 170 clones examined. Positive mutants were obtained for four nonconserved residues (Gln 40 , Asp 47 , Ala 49 , and Arg 50 ) are shown in Table VI. In conclusion, 14 residues of the 23-residue module of human MTH1 are essential for function to suppress the elevated A:T to C:G transversion in mutT Ϫ cells, determined when combining the results obtained in the present study and previous studies (28).
Positive Mutants for the Conserved Residues Thr 44 and Gly 58 -Seven different positive mutants were obtained from the negative mutant T44amber for the conserved Thr 44 residue as well as the true revertant (Tables IV and VIII). Among the  1 and 4), pTT100:hMTH1 (lanes 2 and 5), and pTT100:MTH1-Ec (lanes 3 and 6) were grown at 37 or 30°C, and 20 g of crude extracts were subjected to SDS-polyacrylamide gel electrophoresis and to Western blotting using anti-M77. Arrow indicates 18-kDa MTH1 and MTH1-Ec polypeptides. B, 8-oxo-dGTPase assay for MTH1 and MTH1-Ec proteins. Crude extracts (1.0 g of protein) prepared from CC101T cells harboring plasmids, pTT100 (lanes 1 and 4), pTT100: hMTH1 (lanes 2 and 5), and pTT100:MTH1-Ec (lanes 3 and 6) were grown at 37 or 30°C. The crude extracts (1.0 g of protein) prepared from these cells were incubated with ␣-32 P-labeled 8-oxo-dGTP, and the reaction products were subjected to thin layer chromatography on PEI cellulose, as described (11). C represents a control reaction without extract. coli MutT, residues 37-45 constitute loop I, and residues 46 -59 constitute ␣-helix I. Conserved amino acid residues are boxed in black with white letters, and residues preserved at least between MTH1 and E. coli MutT protein, namely semi-conservative residues, are shaded with gray. The asterisks indicate residues that could not be replaced with any other residue without losing function. Residues shown in bold type with a thick arrow indicate those that can replace the corresponding residue in wild type MTH1 protein with the full function of MTH1. Residues with a thin arrow indicate those that can replace the corresponding residue in wild type MTH1 protein with a partial function of MTH1. positive mutants, the lowest mutation rate of LacZ ϩ reversion in CC101T was attained by expression of a positive mutant T44Q, and the mutation rate, 1.89 ϫ 10 Ϫ8 , was 2.1-fold higher than that attained by wild type MTH1. A positive mutant T44K exhibited the highest mutation rate, 3.32 ϫ 10 Ϫ8 , that is, 3.7-fold higher compared with wild type MTH1. Mutation rates for the other positive mutants were in the following order: Gln Ͼ Leu Ͼ Glu Ͼ Ser Ͼ Tyr Ͼ Met Ͼ Lys.
Amounts of T44E and T44S mutant proteins expressed in CC101T cells equalled amounts of the wild type, whereas amounts of others were 51-65% of the level of the wild type, except that the expression level of T44Y was significantly low: 22% of the wild type. All of the positive mutants for the residue Thr 44 possess a specific activity of 8-oxo-dGTPase comparable to that of wild type.
Five positive mutants were obtained for the conserved residue Gly 58 from the negative mutant G58amber in addition to the true revertant (Tables IV and VIII). Mutation rates attained by expression of these positive mutants in CC101T cells were 4.1-fold or higher than that by wild type MTH1, indicating that suppression of the elevated A:T to C:G transversion by these mutant MTH1 is partial. Amounts of these mutant MTH1 proteins in CC101T cells were very low, mostly less than 10% that of wild type MTH1. The amount of G58R, the highest among these five positive mutants, was 15.3% that of wild type MTH1. Three positive mutants possess specific activities of the 8-oxo-dGTPase, corresponding to the 60 -94% level of wild type, whereas that of a mutant G58R was twice that of the wild type. A mutant G58Y had much lower specific activity, 20% level of wild type. These results indicate that positive mutants for the residue Gly 58 possess substantial activity of 8-oxo-dGTPase comparable to wild type MTH1, but expression levels are significantly low, suggesting that the Gly 58 residue is important for maintaining the stable structure of MTH1 protein.
In conclusion, different positive mutants for the conserved residues Thr 44 and Gly 58 were obtained; however, they retain a limited level of capability to suppress the elevated A:T to C:G transversion in mutT Ϫ cells, probably because structural stability is poor.
Positive Mutants for the Semi-conserved Residues Gly 42 , Glu 46 , and Gln 54 -Positive mutants for the four semi-conserved residues (Gly 42 , Glu 46 , Leu 53 , and Gln 54 ) were screened from at least 1000 clones. Positive mutants G42P and G42T from a negative mutant G42I, positive mutants E46P and E46S from a negative mutant E46amber, and a single positive mu-  tant Q54R from a negative mutant Q54P were obtained, respectively, but no positive mutant for the residue Leu 53 was isolated (Tables V and VIII).
Mutation rates attained by expression of these mutant proteins in CC101T cells were significantly lower than that for the wild type, the range being from 0.4 to 0.76 ϫ 10 Ϫ8 . Amounts of mutant proteins for residues Gly 42 and Glu 46 , which exhibited 28 -46% of the specific activity of 8-oxo-dGTPase for wild type MTH1, exceeded the 60% level of the wild type. In particular the level of E46S protein was almost twice that of wild type MTH1. A single positive mutant for the residue Gln 54 , Q54R, had lower expression level (27%) with a higher 8-oxo-dGTPase activity (185%) compared with wild type, suggesting that the higher 8-oxo-dGTPase activity compensates for the lower expression level. Thus CC101T cells expressing Q54R protein attain significantly lower mutation rates than those expressing wild type MTH1 protein.
A positive mutant Q40H had much higher capability to suppress the elevated spontaneous mutation rate in CC101T cells than did wild type MTH1, reaching the rate of 0.27 ϫ 10 Ϫ8. Another positive mutant Q40R had a capability almost equiv-alent to that of wild type MTH1 protein. Amounts of these two mutant proteins expressed in CC101T cells were about twice that of the wild type and retained more than a 50% level of 8-oxo-dGTPase activity for wild type protein. Thus, replacement of the residue Gln 40 with histidine or arginine led to significant stabilization of the MTH1 protein.
D47R, the single positive mutant obtained for residue Asp 47 , had about 60% the level of expression for the wild type protein and a slightly higher specific activity of 8-oxo-dGTPase than did the wild type. Again D47R had a much higher capability to suppress the elevated mutation rate in CC101T cells than did the wild type MTH1, reaching the rate of 0.42 ϫ 10 Ϫ8 , probably because of increased catalytic activity⅐ For residue Ala 49 , A49S was obtained as a single positive mutant; however, the mutation rate in the CC101T cells expressing the mutant protein was 3.4-fold higher than that of wild type, thus reflecting lower levels of both its expression and catalytic activity (Table VIII). These results suggest that residue Ala 49 is important both for stability and catalytic functions of MTH1.
Eleven positive mutants were obtained for residue Arg 50 from a negative mutant (R50ocher). The mutation rates in CC101T cells expressing two positive mutants, R50A and R50I, were lower than in cells expressing the wild type MTH1, and cells expressing R50K, R50L, and R50V had slightly higher mutation rates (Ͻ1.38 ϫ 10 Ϫ8 ) than cells expressing the wild a Number of mutants isolated for each codon is shown in parentheses. Codons in bold type were used as negative mutant templates for positive mutant screening.
b Number of clones whose DNA sequence were determined is shown in parentheses.  Most of the mutant proteins retained higher or equivalent levels of specific activity of 8-oxo-dGTPase, compared with the wild type, the order being Phe Ͼ Gln Ͼ Ile Ͼ Tyr Ͼ Val Ͼ His Ͼ Leu Ͼ Asn Ͼ Lys Ͼ Ala Ͼ Ser. The most active mutant protein R50F had a 2-fold higher specific activity; however, expression level was low.
In conclusion, four of the nonconserved residues could be replaced with other amino acid residues without losing the potential to suppress the elevated mutation rate in mutT Ϫ cells. However, only residue Arg 50 could be replaced with other residues retaining both its stability and catalytic activity. DISCUSSION Our main conclusion based on findings in the present study is that the 23-residue sequence highly conserved among MTH1 and MutT family proteins constitutes a functional module that is interchangeable between MTH1 and MutT proteins and that most of the residues in the functional module are essential to exert catalytic function as 8-oxo-dGTPase. As shown in Fig. 3, the six proteins identified biochemically are enzymes that hydrolyze dGTP or a mutagenic nucleotide, 8-oxo-dGTP (6,18). Over 10 putative homologs for bacterial MutT proteins, which also possess the 23-residue sequence and may hydrolyze 8-oxo-dGTP, can be retrieved by data base searching with the BLASTP alogrithm. In addition to the MTH1 or MutT family proteins, many other proteins with the 23-residue sequence that hydrolyze various nucleotide derivatives and diphosphoinositol derivatives other than 8-oxo-dGTP have been identified (23)(24)(25)(26). Here, we propose that the 23-residue module should be designated as the phosphohydrolase module.
For E. coli MutT, it has been established that the 23-residue sequence constitutes the active center for dGTPase activity (21,32), probably also for the 8-oxo-dGTPase activity, and that the module consists of loop I (Gly 37 -Thr 45 ) and ␣-helix I (Pro 46 -Gly 59 ) (33). The ␣-helix I in MutT is apparently amphipathic, and it was shown that MutT has two hydrophobic cores, one of them consisting of Val 50 , Val 51 , Leu 54 from the ␣-helix I clustering with Phe 65 and Leu 67 from the loop II, Val 87 from ␤-strand D, and Trp 95 from loop III (33). It may be that interactions between the phosphohydrolase module and other secondary structures are important to support stability or to provide functional moieties for the active center. Thus, residues Gly 38 , Glu 56 , and Glu 57 in the module of MutT are involved in coordination of enzymebound metal (Mg 2ϩ ), and residues Glu 53 , Arg 52 , and Lys 39 may be involved in the catalytic reaction (21,22). MTH1 and MutT proteins share only 30 residues, 14 of which constitute the 23-residue module, whereas the other 16 residues are scattered throughout the molecules (Fig. 1). Based on both physico-chemical and computational analysis of MTH1 protein, we earlier proposed that human MTH1 protein has secondary structures similar to those of MutT (27). We also have data supporting this model, as based on NMR analysis of MTH1 protein. 2 Thus, MutT or MTH1 protein consists of similar secondary structures with different primary sequences (Fig. 1B), suggesting that each of the 23-residue module is surrounded by a different microenvironment. The chimeric protein MTH1-Ec, in which the 23-residue sequence from MutT was placed in the corresponding region of MTH1, indeed retains catalytic function as 8-oxo-dGTPase. However, the chimeric protein is thermolabile in terms of suppression of mutagenesis as well as 8-oxo-dGTPase. There are 9 amino acid FIG. 4. Hydrophobic interaction between the 23-residue module and the ␤-sheet A in MTH1 and MutT protein. ␤-sheets (A, D, and C) are shown as boxes with arrowheads, in which amino acid residues in ␤-sheet A are shown. The 23-residue module consists of loop I and ␣-helix I, and each residue is shown as a circle with a diameter corresponding to its relative mass and with a grayscale corresponding to its hydrophobicity, determined by contact energy derived from three-dimensional data (37). residues that differ between the modules of two proteins; this difference maybe essential for each of the modules to fit into its own microenvironment.
Saturation mutagenesis of the 23-residue sequence of MTH1 with negative and positive mutant screening revealed that 14 residues are essential and cannot be substituted for other residue without losing function and that nine are conserved residues as expected, but five of them (Val 39 , Glu 41 , Ile 45 , Gly 48 , and Ser 57 ) are nonconserved residues between MutT and MTH1. These results clearly demonstrate that most of the conserved residues among the MutT homologs are essential to exert 8-oxo-dGTPase activity. The requirement of the five nonconserved residues, all of which are completely conserved among mammalian MTH1 homologs (Fig. 3), indicates that MTH1 homologs have a structural basis somehow different from that for the bacterial MutT homologs. Concerning the chimeric protein, our results suggest that five residues (Ile 38 , Met 40 , Pro 44 , Ala 47 , and Val 56 ) in MutT protein corresponding to the five nonconserved residues in MTH1 are not suitable for the microenvironment of MTH1, thus render the chimeric protein thermolabile.
By careful examination of the tertiary structure of MutT (ID 4271 and 5943) available from the Molecular Modeling Data Base at NCBI, we found that Val 50 in the 23-residue module locates close to the residue Ala 7 in the ␤-strand A, which mostly consists of hydrophobic residues (Fig. 4). As proposed in Fig. 4, Ala 49 and Leu 53 in the ␣-helix I of MTH1 may form a hydrophobic core with other hydrophobic residues, including the residue Leu 9 in the ␤-strand A that corresponds to the residue Ala 7 in MutT. However, the interaction is not likely to be as strong as in MutT. The residue Leu 53 in MTH1 reflects its conservation among MutT homologs, whereas Ala 49 can be substituted only with serine, but the expression level of the positive mutant A49S is much lower compared with that of the wild type MTH1. We considered that the residue Ala 49 is also essential for MTH1 and that the serine residue only partially supports the ␣-helix I structure. Ile 45 , which is a nonconserved residue but is essential for MTH1, may also be involved in formation of the hydrophobic core.
Many positive mutants obtained for residue Arg 50 have hydrophobic amino acid residues, such as Ala, Ile, Leu, and Val, in addition to Lys, and most are as stable and active as is the wild type MTH1. Rather R50I MTH1 protein is more stable and active than is the wild type, thereby indicating that replacement of the residue Arg 50 with a hydrophobic residue renders the MTH1 protein more stable and more active, probably providing an additional hydrophobic interaction between the essential residues Ala 49 , Leu 53 , and/or Ile 45 . Considering these positive mutants for Arg 50 , we emphasize that an amphipathic property of the ␣-helix I in MTH1 or MutT may determine stability of the catalytic center.
In wild type MTH1 protein, the positively charged Arg 50 may interact with a negatively charged residue Glu 46 instead of the hydrophobic residues, and thus contribute to stabilization of the ␣-helix I (Fig. 4). This notion can explain why Arg 50 can be replaced by hydrophobic residues as well as a positively charged lysine residue, with almost the same level of stability and activity. In the chimeric protein MTH1-Ec, Val 50 , Val 51 , and Leu 54 from the ␣-helix I of MutT may not interact properly with the cognate hydrophobic residues in MTH1, because of poor conservation of the primary sequence outside the 23residue sequence. Most likely Val 50 in the ␣-helix I of MutT cannot properly fit into the hydrophobic core opposite Leu 9 in the ␤-strand of MTH1, because the side chain of valine residue is larger than that of the alanine residue, which should be opposite to Leu 9 in MTH1 (Fig. 4). The hydrophobic interaction around residues Val 50 , Val 51 , and Leu 54 must be disturbed in the chimeric protein, and the chimeric protein becomes thermolabile in structure and function.
We reported that MTH1 but not MutT efficiently hydrolyzes two forms of oxidized dATP, 2-hydroxy-dATP and 8-oxo-dATP, as well as 8-oxo-dGTP (34), and that both proteins hydrolyze 8-oxo-GTP, to which MTH1 has much less affinity compared with findings with MutT protein (10,35). MTH1 has a much wider substrate specificity than does MutT, perhaps because of the diverged sequence located outside the 23-residue sequence. The possibility that the 23-residue module itself may determine the substrate specificities and that the positive mutants we isolated may shed a light on a role of each residue for the broad substrate specificity of MTH1 protein is worthy of consideration.