Distinct repair activities of human 7,8-dihydro-8-oxoguanine DNA glycosylase and formamidopyrimidine DNA glycosylase for formamidopyrimidine and 7,8-dihydro-8-oxoguanine.

7,8-dihydro-8-oxoguanine (8-oxoG) and 2,6-diamino-4-hydroxyformamidopyrimidine (Fapy) are major DNA lesions formed by reactive oxygen species and are involved in mutagenic and/or lethal events in cells. Both lesions are repaired by human 7, 8-dihydro-8-oxoguanine DNA glycosylase (hOGG1) and formamidopyrimidine DNA glycosylase (Fpg) in human and Escherichia coli cells, respectively. In the present study, the repair activities of hOGG1 and Fpg were compared using defined oligonucleotides containing 8-oxoG and a methylated analog of Fapy (me-Fapy) at the same site. The k(cat)/K(m) values of hOGG1 for 8-oxoG and me-Fapy were comparable, and this was also the case for Fpg. However, the k(cat)/K(m) values of hOGG1 for both lesions were approximately 80-fold lower than those of Fpg. Analysis of the Schiff base intermediate by NaBH(4) trapping implied that lower substrate affinity and slower hydrolysis of the intermediate for hOGG1 than Fpg accounted for the difference. hOGG1 and Fpg showed distinct preferences of the base opposite 8-oxoG, with the activity differences being 19.8- (hOGG1) and 12-fold (Fpg) between the most and least preferred bases. Surprisingly, such preferences were almost abolished and less than 2-fold for both enzymes when me-Fapy was a substrate, suggesting that, unlike 8-oxoG, me-Fapy is not subjected to paired base-dependent repair. The repair efficiency of me-Fapy randomly incorporated in M13 DNA varied at the sequence level, but orders of preferred and unpreferred repair sites were quite different for hOGG1 and Fpg. The distinctive activities of hOGG1 and Fpg including enzymatic parameters (k(cat)/K(m)), paired base, and sequence context effects may originate from the differences in the inherent architecture of the DNA binding domain and catalytic mechanism of the enzymes.

For introduction of Fapy as a unique lesion, the use of ionizing radiation, oxidizing chemical reagents, or photooxidation is impractical since these agents are known to produce not only Fapy but also other damage (1). Until now, most studies on the genotoxicity/cytotoxicity (13,14) and repair (15)(16)(17) of Fapy have been conducted using 2,6-diamino-4-hydroxy-5-N-methylformamidopyrimidine (me-Fapy), an N-methylated analog of Fapy, since me-Fapy can be introduced into DNA as a major damage (15). To introduce me-Fapy, DNA was first treated with methylating agents followed by base to open the imidazole ring. Although me-Fapy is a major product formed in this treatment, complete conversion of the target guanine to me-Fapy is practically impossible without introducing other unintended damage. The primary reason is that methylation of DNA generates not only 7-methylguanine (m 7 G, a precursor of me-Fapy) but also O-and N-alkylated bases and methyl phosphates simultaneously (18). In addition, certain methylated purines such as 3-methyladenine (m 3 A) and 3-methylguanine (m 3 G) readily undergo depurination, thereby generating abasic sites. Therefore, certain precaution is necessary to interpret the results obtained with DNA substrates prepared by the conventional method.
Repair enzymes that recognize 8-oxoG also act on Fapy and me-Fapy and excise them from DNA. Saccharomyces cerevisiae Ntg1/yOGG2 protein was originally reported to recognize me-Fapy but not 8-oxoG (19,20). However, the recent studies on Ntg1/yOGG2 have suggested that both lesions are substrates, and the activity for 8-oxoG varies depending on the paired base (21,22). Therefore, 8-oxoG DNA glycosylases from bacteria (Fpg), yeast (yOGG1 and Ntg1/yOGG2), and human (hOGG1/ hMMH) so far identified recognize me-Fapy as well. Among these enzymes, OGG1 and Fpg are functionally similar enzymes, and cellular repair activity for 8-oxoG and me-Fapy primarily relies on these enzymes. Despite such a functional similarity, their primary amino acid sequences are quite different. Moreover, OGG1, but not Fpg, is a member of endonuclease III superfamily that contains a hairpin-helix-hairpin (HhH)-GPD motif (reviewed in Ref. 8). For hOGG1, Lys-249 and Asp-268 in the HhH-GPD motif are likely to be involved in the glycosylase/AP (apurinic/apyrimidinic)-lyase activity (23), whereas for Fpg, a proline residue in the N-terminal region is responsible for it (24). Therefore, it is still equivocal how these structurally unrelated proteins can recognize and catalyze excision of both 8-oxoG and me-Fapy. Such a mechanistic question can be best addressed by comparing the intrinsic activities of OGG1 and Fpg for 8-oxoG and me-Fapy using common substrates.
In the present work, we have prepared oligonucleotide and DNA substrates containing me-Fapy as unique lesions. The oligonucleotide substrate was tested for hOGG1 and Fpg, and the enzymatic parameters, influences of the paired base, and reaction mechanisms were compared with those for 8-oxoG embedded in the same site. Furthermore, the effects of surrounding sequence contexts on the recognition of me-Fapy by the two enzymes were compared using M13 DNA containing randomly distributed me-Fapy lesions.

EXPERIMENTAL PROCEDURES
Enzymes-E. coli DNA polymerase I Klenow fragment (Pol I Kf) and T4 polynucleotide kinase were purchased from Life Technologies, Inc., and New England Biolabs, respectively. Formamidopyrimidine DNA glycosylase (Fpg) and endonuclease IV (Endo IV) were overexpressed in E. coli cells harboring plasmids containing the fpg or nfo gene (gifts from S. S. Wallace and Z. Hatahet, University of Vermont) and purified as described (25). Human 8-oxoguanine glycosylase (hOGG1)/MutM homolog (hMMH) was purified as described (26 -28). Preliminary studies were performed with the protein prepared by the method of Roldan-Arjona et al. (26) or Aburatani et al. (27). The results presented in this paper were obtained with hOGG1/hMMH (type 1a isoform) prepared by the method of Monden et al. (28). The purified hOGG1/hMMH contained 5 additional N-terminal amino acid residues (GPLGS) derived from glutathione S-transferase. The repair enzymes used in this study (Fpg, Endo IV, and hOGG1) were apparently homogenous in SDS-PAGE analysis. Liquid chromatography-mass spectrometry (LC-MS) analysis showed that the purity of hOGG1/hMMH was over 99%.
Oligonucleotides and DNA-Oligonucleotides used in this study are listed in Table I. Oligonucleotides 15PRM, 20PRM, 25COM-N (N ϭ A, G, C, T), 30COM-C, 25G, and 25OX containing 8-oxoG were synthesized by the phosphoramidite method and purified by reversed phase high pressure liquid chromatography. The duplex substrate 25MG/ 25COM-C containing a single 7-methylguanine (m 7 G) lesion at the specific position was prepared by the DNA polymerase reaction using 7-methyl-2Ј-deoxyguanosine 5Ј-triphosphate (m 7 dGTP) as a substrate. m 7 dGTP was previously shown to serve as a substrate for Sequanase (T7 DNA polymerase deficient in 3Ј-5Ј exonuclease activity) (29). 15PRM was 5Ј-end-labeled with [␥-32 P]ATP (110 Tbq/mmol, Amersham Pharmacia Biotech) and T4 polynucleotide kinase and purified as described (30). 15PRM annealed to the template 25COM-C (25 pmol as template/primer) in buffer A (200 l) was extended by Pol I Kf (25 units) in the presence of dCTP, dTTP (both 20 M), and m 7 dGTP (200 M, Sigma) at 25°C for 40 min. The composition of buffer A was 66 mM Tris-HCl (pH 7.5), 1.5 mM mercaptoethanol, and 6.6 mM MgCl 2 . The reaction was terminated by the addition of EDTA (final concentration 50 mM). The control duplex substrate 25G/25COM-G containing G at the same position of m 7 G was also prepared by a similar DNA polymerase reaction, except that dGTP (20 M) was used instead of m 7 dGTP. The reaction mixture was extracted by phenol, and DNA was recovered by ethanol precipitation. DNA was resuspended in water, purified by a Sephadex G-25 column, and finally recovered by ethanol precipitation. The duplex substrate 25FP/25COM-C containing a me-Fapy lesion was prepared by the alkali treatment of 25MG/25COM-C. 25MG/25COM-C in a microdialysis cup was dialyzed against 20 mM phosphate buffer (pH 11.4) containing 2 mM EDTA at room temperature for 10 h, and then against 10 mM Tris-HCl (pH 7.5) and 1 mM EDTA at room temperature for 12 h, and finally against the same buffer at 4°C for 12 h.
Duplex substrates containing me-Fapy paired with four different bases (A, G, C, and T) were prepared as follows. 25MG/30COM-C was prepared by the DNA polymerase reaction using 32 P-labeled 15PRM (primer) and 30COM-C (template) as described for 25MG/25COM-C and converted to 25FP/30COM-C by the alkali treatment. Complete conversion of m 7 G to me-Fapy in 25MG was essential, otherwise remaining m 7 G underwent depurination during the subsequent purification step. 25FP/30COM-C was heat-denatured at 50°C in gel loading buffer and separated by 16% PAGE. The use of 30COM-C in place of 25COM-C as a template facilitated separation of 25FP and the template. The band corresponding to 25FP was detected by autoradiography and excised from the gel. 25FP in the crashed gel was extracted by 500 mM ammonium acetate and 10 mM magnesium acetate, purified by a Sep-Pak cartridge, and finally annealed to appropriate complementary strands (25COM-A, -G, -C, -T). These substrates were resistant to Endo IV, indicating that contamination of oligonucleotides containing abasic sites was negligible (data not shown). The substrates containing 8-oxoG paired with four different bases were prepared by annealing 25OX to 25COM-A, -G, -C, -T.
Duplex M13 DNA containing randomly distributed m 7 G was prepared using the method similar to oligonucleotides with slight modifications. Single-stranded M13mp18 DNA (5 g, ϳ2 pmol) was primed with 20PRM (5Ј-end-labeled, 4 pmol) and replicated by Pol I Kf (5 units) in buffer A (400 l, MgCl 2 was replaced by 0.5 mM MnCl 2 ) in the presence of dATP, dCTP, dTTP (all 50 M), dGTP (10 M), and m 7 dGTP (50 M). m 7 dGTP was incorporated into DNA more uniformly in the presence of Mn 2ϩ than Mg 2ϩ . 2 The reaction was continued at 37°C for 30 min. Purification of DNA and the alkali treatment to convert m 7 G to me-Fapy was performed as described for 25FP.
Reaction with Repair Enzymes-To follow the conversion of m 7 G to me-Fapy, untreated and alkali-treated 25MG/25COM-C (5 nM) were incubated with Fpg (100 ng) in buffer B (10 l) at 37°C for 30 min. The composition of buffer B was 10 mM Tris-HCl (pH 7.5), 1 mM EDTA, 100 mM NaCl, and 0.1 mg/ml BSA. To reveal abasic sites potentially formed during substrate preparation, untreated and alkali-treated 25MG/ 25COM were also treated with Endo IV. Untreated and alkali-treated 25MG/25COM-C (5 nM) in buffer C (10 l) was incubated with Endo IV (8 ng) at 37°C for 30 min. The composition of buffer C was 10 mM Tris-HCl (pH 7.5), 1 mM EDTA, and 50 mM NaCl.
Enzymatic parameters of Fpg and hOGG1 for 8-oxoG and me-Fapy were determined as follows. 25OX/25COM-C and 25FP/25COM-C were incubated with Fpg (0.5 ng) in buffer B (10 l) at 37°C for 3 min. For hOGG1, the substrates were incubated with hOGG1 (6 ng) in buffer D (10 l) at 37°C for 15 min. The composition of buffer D was 50 mM Tris-HCl (pH 7.5), 2 mM EDTA, 50 mM KCl, and 0.1 mg/ml BSA. The concentration ranges of the substrate used in the experiments were indicated in the figures. Parameters (V max and K m ) were evaluated by a hyperbolic curve-fitting program. The activity of Fpg and hOGG1 to me-Fapy and 8-oxoG paired with four different bases was determined by an essentially similar manner. The substrates (25FP/25COM-N or 25OX/25COM-N (N ϭ A, G, C, T), 5 nM) were incubated with Fpg (2 ng) for 5 min or hOGG1 (6 ng) for 15 min at 37°C.
Heat and Piperidine Treatments-For chemical characterization of m 7 G introduced into substrates, 25MG/25COM-C was heated in water at 90°C for 1 h to induce depurination of m 7 G and strand scission. Alternatively, substrates containing m 7 G or me-Fapy were treated with 10% piperidine at 90°C for 1 h. Piperidine was removed by repeated evaporation with water, and the sample was resuspended in gel loading buffer.
Product Analysis-After enzymatic or chemical reactions, the samples were mixed with gel loading buffer (0.05% xylene cyanol, 0.05% bromphenol blue, 20 mM EDTA, and 98% formamide), heated at 50°C for 5 min, and separated by 8% (M13 DNA) or 16% (oligonucleotides) denaturing polyacrylamide gel electrophoresis. Harsh heat denaturation at high temperatures (i.e. boiling) was avoided to prevent break down of DNA containing the heat-labile lesions. The gel was autoradiographed at Ϫ80°C overnight. Alternatively, the radioactivity of the separated bands was analyzed by Fuji BAS 2000.
Cross-link Reaction with NaBH 4 -Cross-link reactions between repair enzymes (Fpg or hOGG1) and substrates (25G/25COM-C, 25OX/ 25COM-C, 25FP/25COM-C) were performed as follows. An aqueous solution of NaBH 4 (500 mM, 1 l) was added to 8 l of buffer B (Fpg) or D (hOGG1) containing the substrate (final concentration 2-40 nM). In NaBH 4 reactions, the salts (NaCl and KCl) and BSA were omitted in buffer B and D. Immediately after, the solution was mixed with Fpg (1.5 ng, 1 l) or hOGG1 (19.5 ng, 1 l), and incubation was continued at 37°C for 30 min. The reaction solution was mixed with an equal volume of SDS-loading buffer (100 mM Tris-HCl (pH 6.8), 8% SDS, 24% (v/v) glycerol, 4% 2-mercaptoethanol, 0.02% SERVA Blue G) and heat-denatured. The sample was electrophoresed on a 10% SDS-polyacrylamide gel. Autoradiography and quantitation of the radioactivity were performed as described above. To analyze Schiff base intermediates without NaBH 4 reduction, 25OX/25COM-C (5 nM) was incubated with Fpg and hOGG1 (both 4, 15, 45 ng) in buffer B (Fpg) or D (hOGG1) at 37°C for 5 min. The sample was subjected to SDS-PAGE as described above.

Preparation and Characterization of Substrates-
The presence of m 7 G in 25MG prepared by the DNA polymerase reaction was confirmed by the heat or piperidine treatment. In the heat treatment, 25MG/25COM-C and control 25G/25COM-C (prepared by the similar polymerase reaction) was heated at 90°C for 1 h, and products were analyzed by PAGE (Fig. 1). The m 7 G with a heat-labile N-glycosidic bond underwent depurination, and the resulting abasic site was totally cleaved (lane 6). By comparison with the gel mobility of 15PRM (lane 1), the products were identified as ␤-(minor bands) and ␦-elimination (major band) products. In the hot piperidine treatment, 25MG was totally converted to the ␦-elimination product migrating slightly faster than 15PRM (lane 7). The complete conversion of 25MG to the ␤and ␦-elimination products indicated that m 7 G was quantitatively incorporated in the designed position of 25MG. 25G similarly prepared by Pol I Kf was stable under these conditions (lanes 3 and 4).
25FP/25COM-C containing me-Fapy was prepared by alkali treatment (pH 11.4) of 25MG/25COM-C. To follow the alkalicatalyzed ring opening reaction of m 7 G, oligonucleotides before and after the alkali treatment were analyzed with Fpg that specifically recognizes me-Fapy. Before the treatment, 25MG (paired with 25COM-C) was not cleaved by Fpg (Fig. 2A, lane 3) but was almost quantitatively converted to the ␦-elimination product after the alkali treatment for 10 h (lane 6). These results indicate quantitative conversion of m 7 G to me-Fapy by the alkali treatment. Partial depurination of m 7 G (yielding abasic sites) or fission of the imidazole ring (yielding me-Fapy) in untreated 25MG was ruled out by the resistance to Endo IV (lane 4) and Fpg (lane 3). 25MG treated by alkali for 10 h was also resistant to Endo IV (lane 7), showing the absence of abasic sites in this substrate. Fig. 2B shows the time course of the conversion of m 7 G in 25MG to me-Fapy during the alkali treatment. The percent of me-Fapy in the oligonucleotide was determined by quantifying the nicked and unnicked products by Fpg. The proportion of me-Fapy increased with the incubation time with alkali and reached a plateau (ϳ96%) around 10 h. Thus, me-Fapy lesion was specifically introduced into DNA by the present method.
Reaction Products and Enzymatic Parameters of Fpg and hOGG1-The activities of Fpg and hOGG1 for 8-oxoG and me-Fapy were compared using 25OX/25COM-C and 25FP/ 25COM-C. The substrates were treated with Fpg or hOGG1, and products were analyzed by PAGE (Fig. 3). The treatment of 8-oxoG and me-Fapy with Fpg exclusively resulted in ␦-elimination products (Fig. 3, lanes 6 and 9) migrating slightly faster than the 15-mer marker (Fig. 3, 15PRM, lane 1). The treatment with hOGG1 resulted in putative ␤-elimination products with their gel mobilities slower than the marker (Fig. 3, lanes 7 and  10). However, the ␤-elimination products migrated as several separated bands. The ␤-elimination products generated by AP lyases, e.g. endonuclease III, sometimes give rise to multiple bands in PAGE analysis, presumably due to Tris-adduct formation (31) and/or isomerization of the 3Ј-hydroxypentenal terminus (32,33). To clarify the nature of the putative ␤-elimination products, hOGG1-treated samples were further treated with Endo IV having 3Ј-phosphodiesterase activity. By this treatment, the ␤-elimination products were converted to 3Ј-OH products (Fig. 3, lanes 12 and 14) co-migrating with the standard maker 15PRM (Fig. 3, lane 1). The same treatment of the ␦-elimination products formed by Fpg also resulted in the 3Ј-OH products (Fig. 3, lanes 11 and 13). Therefore, the products generated by hOGG1 are likely to be a mixture of ␤-elim-  ination products bearing different 3Ј-terminal deoxyribose modifications.
To obtain kinetic parameters, the initial velocities of the reaction (V) were determined at varying substrate concentrations (S). The S-V plots for Fpg and hOGG1 are shown in Fig.  4. The enzymatic parameters are summarized in Table II.
NaBH 4 Trapping of the Reaction Intermediates-hOGG1 and Fpg are known to form transient imine intermediates (Schiff base) in the reaction with substrates containing 8-oxoG (reviewed in Ref. 8). Reduction of the imine bond by NaBH 4 leads to formation of irreversibly cross-linked complexes between the enzyme and substrate. To verify the proposed mechanism further, NaBH 4 -trapping experiments were performed using the substrate containing me-Fapy. 25FP/25COM-C and 25OX/ 25COM-C were incubated with Fpg or hOGG1 in the presence of NaBH 4 , and products were analyzed by SDS-PAGE (Fig. 5). The upper shifted bands clearly indicated formation of the cross-linked complexes between 25FP and Fpg (Fig. 5, lane 9) or hOGG1 (Fig. 5, lane 10). The complex for Fpg (molecular mass ϭ 30 kDa) migrated faster than that for hOGG1 (39 kDa) due to the molecular mass difference of the cross-linked enzymes. Similar upper shifted bands were also observed for 25OX containing 8-oxoG (Fig. 5, lanes 6 and 7) but not for intact 25G (Fig. 5, lanes 3 and 4), indicating that the Schiff base formation was specific to the lesioned substrates (25FP and 25OX).
The amount of cross-linked products was determined at varying substrate concentrations to obtain parameters for the Schiff base formation. The correlation between the concentrations of substrate ([S]) and trapped Schiff base ([SB]) is shown in Fig. 6. For both 8-oxoG and me-Fapy, the amount of crosslinked products increased with the substrate concentration and reached plateaus. By applying simple Michaelis-Menten analysis to the data, approximate dissociation constants (K ES ) were evaluated (Table III). For more detailed analysis of the data, see under "Discussion." Paired Base Effects on the Repair Activity of Fpg and hOGG1-25OX/25COM-N and 25FP/25COM-N (N ϭ A, G, C, T) were treated with Fpg and hOGG1, and products were analyzed by PAGE. The activity was compared based on the amount of nicked products (Fig. 7). Consistent with the previous studies (34, 35), 8-oxoG paired with A was poorly recognized by Fpg, whereas 8-oxoG paired with other bases (G, C, T) was a good substrate and excised 12-fold more efficiently than the 8-oxoG:A pair (Fig. 7A). Surprisingly, such a substrate preference was almost abolished for me-Fapy. The activity of Fpg for me-Fapy was virtually independent of the paired base and comparable to those for 8-oxoG paired with G, C, T. The activity difference was at most 1.7-fold between the most (C) and least (A) preferred bases. hOGG1 recognized 8-oxoG paired with C most efficiently and then T (Fig. 7B). Those paired with purines (A and G) were poor substrates. The relative activities for 8-oxoG paired with C, T, G, A were 19.8:4.3:1.5:1, respectively. These results are consistent with those reported for human (26,27,36) and yeast (37) OGG1. Although hOGG1 showed some preference for me-Fapy:C, the paired base effect was much less obvious than that for 8-oxoG. The activity difference was only ϳ2-fold between the most (C) and least (A) preferred bases.
Sequence Context Effects on the Repair Efficiency-Duplex M13 DNA containing randomly distributed me-Fapy was treated with Fpg and hOGG1, and the repair efficiency of individual sites was analyzed by PAGE (Fig. 8). The ␦-elimination products formed by Fpg (lanes 3-5) migrated slightly faster than the corresponding ␤-elimination products by   hOGG1 (lanes 6 -9) in the electrophoresis. me-Fapy lesions in M13 DNA were not equally removed so that the band intensity varied significantly depending on the site. The variation of the band intensity was also observed with the enzymes. For quantitative analysis, the gel was run for a longer time than shown in Fig. 8 to achieve better band separation (particularly for G doublets). The intensity of the individual band formed by 10 ng of Fpg (Fig. 8, lane 4) or 200 ng of hOGG1 (Fig. 8, lane 8) was divided by that of the corresponding piperidine-generated band (Fig. 8, lane 2). Since me-Fapy sites are quantitatively cleaved by the piperidine treatment, the intensity of the piperidinegenerated band represents the actual lesion frequency of the individual site. This calculation corrects the site-dependent variation of the lesion frequency in the substrate DNA. It is also noted that the amount of nicked products increased with the increasing amount of Fpg (Fig. 8, lanes 3-5) and hOGG1 (Fig. 8, lanes 6 -9), indicating that the reaction conditions were within a dynamic (i.e. not saturating) range. The corrected repair efficiency for up to the 20th position of G is summarized in Table IV together with the surrounding sequences. To compare the repair efficiency of the individual site by Fpg and hOGG1, a normalized repair efficiency (nicked %/sum of nicked % for positions [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20] was calculated from the data in Table IV and plotted against the sequence (Fig. 9). The normalized repair efficiency represents the distribution of the repair event if the two enzymes remove the same amount of me-Fapy as a whole from the region of interest (i.e. positions [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20]. According to this analysis, me-Fapy at positions 5, 8, 10, 11, 12, 15, and 19 was excised by Fpg and hOGG1 with comparable efficiencies, and that at 1, 2, 6, 14, 17, 18, and 20 was preferred by Fpg, and that at 3, 4, 7, 9, 13, and 16 was preferred by hOGG1.

DISCUSSION
Reaction Kinetics of hOGG1 and Fpg-The activities of E. coli Fpg (34,35,38,39) and eukaryotic OGG1 (17,26,27,36) for 8-oxoG and me-Fapy have been studied previously. To our knowledge, two comparative studies have been performed using oligonucleotide substrates containing a single 8-oxoG or me-Fapy (34,38). However, minor portions of the target G (11 and 17%) were converted to me-Fapy in the substrates. In addition, some conflicting data were obtained in these studies with respect to the reaction efficiency (k cat /K m ) of Fpg to 8-oxoG and me-Fapy. One study indicates the ratio of the reaction efficiency for 8-oxoG versus me-Fapy was 1.4:1 and the other 15.8:1. The discrepancy mainly originated from the difference in k cat for these lesions. The present results (Table II) have shown that Fpg has a roughly 3-fold higher affinity for 8-oxoG (K m ϭ 13 nM) than me-Fapy (38 nM). However, with respect to k cat , me-Fapy (k cat ϭ 5.1 min Ϫ1 ) is preferred to 8-oxoG (k cat ϭ 1.8 min Ϫ1 ). Consequently, the opposite effects of the two parameters resulted in comparable k cat /K m values, indicating similar reaction efficiencies to 8-oxoG and me-Fapy. In contrast to Fpg, the activities of eukaryotic OGG1 to me-Fapy and 8-oxoG have not been compared on the basis of the kinetic parameters using defined substrates. The parameters determined in the present study (Table II) indicate that hOGG1 has comparable affinities and reaction efficiencies for me-Fapy than 8-oxoG (both paired with C). The substrate preference (k cat /K m ) of hOGG1 (8-oxoG versus me-Fapy ϭ 0.9:1, Table II) is notably different from that obtained for yOGG1 (8-oxoG versus me-Fapy ϭ 12.2:1) (17). The comparison for yOGG1 was made by measuring 8-oxoG and me-Fapy glycosylase activities (not kinetic parameters but the amount released products). In addition, the substrates were methylene blue/light-treated calf thymus DNA (8-oxoG) and conventionally prepared me-Fapycontaining poly(dG-dC), which may have contained other adducts interfering the analysis.
With respect to the reaction kinetics of Fpg and hOGG1, two sets of parameters were obtained from steady state reactions (Table II) and NaBH 4 cross-link reactions (Table III), although the latter data were approximate values due to the limited number of experimental points (Fig. 6). The parameters in Table II are for the overall repair reaction, which includes (i) association/dissociation of the enzyme (E) and substrate (S); (ii) formation of the Schiff base (ES SB ) between the enzyme and substrate resulting in release of the damaged base and strand cleavage; (iii) hydrolysis of the imine bond between the enzyme and DNA; and (iv) dissociation of the enzyme and product (P) (Fig. 10). The data in Table III represent dissected parameters for the events before hydrolysis of the imine bond, i.e. steps i and ii. The parameters in Table III were calculated by applying simple Michaelis-Menten analysis to steps i and ii. In the calculation, it was assumed that approximately the same proportion of Fpg and hOGG1 was inactivated during the trapping reaction and that the Schiff base intermediate was quantitatively converted to the stable cross-linked product by NaBH 4 . The latter was confirmed by the absence of nicked products in the NaBH 4 -trapping reactions (see Fig. 5). According to the dissociation constant (K ES ) in Table III, the intrinsic affinity of Fpg for the substrates was severalfold greater than that of hOGG1, which was in contrast to the steady state parameters showing less obvious preferences between Fpg and hOGG1 (K m in Table II). Fpg has much higher k cat and k cat /K m than hOGG1 for the entire reaction (Table II). However, if the reaction is dissected, the relative rate constants for the Schiff base formation (k SB ) were essentially similar for Fpg and hOGG1 (note that k SB should be read as relative values since the effective concentration of the enzymes in the trapping reaction was not known). Accordingly, the reactions before hydrolysis of the Schiff base (k SB /K ES for Fpg versus hOGG1 ϭ 10:1 (8-oxoG) or 7:1 (me-Fapy)) account for only a part of the difference in the overall reaction efficiency between Fpg and hOGG1 (k cat /K m for Fpg versus hOGG1 ϭ 80:1). This suggests that hydrolysis of the imine bond and/or subsequent dissociation of hOGG1 proceeds more slowly (ϳ10-fold) than that of Fpg. The slow hydrolysis of    Table IV and plotted against the sequence. The sequence and damage positions (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20) are shown below the graph. Open bar, Fpg; closed bar, hOGG1. the imine bond of hOGG1 relative to Fpg was further supported by direct SDS-PAGE analysis of the reaction mixture without the NaBH 4 treatment (Fig. 11). In the SDS-PAGE analysis, a faint band migrating slower than the free substrate (25OX) was present with 45 ng of hOGG1 (Fig. 11, lane 8). The mobility of the band was comparable to the cross-linked product formed by NaBH 4 (Fig. 11, lane 9). Such a band indicates the Schiff base intermediate was not present when the same amount of Fpg was incubated in the absence of NaBH 4 (Fig. 11, lane 4). Therefore, these results are consistent with the prediction based on the kinetic parameter analysis. Consequently, the low reaction efficiency of hOGG1 relative to Fpg is likely due to the combined outcome of the low affinity to the substrate (K ES ) and slow hydrolysis of the Schiff base intermediate.
Distinct Paired-base Effects on the Repair of 8-oxoG and me-Fapy-Although Fpg and hOGG1 recognize 8-oxoG and me-Fapy, they have distinct preferences for paired bases when acting on 8-oxoG. Fpg preferentially acts on 8-oxoG paired with G, C, T, but not A (34,35), whereas human (27,36) and yeast (37) OGG1 prefers 8-oxoG paired with pyrimidines (particularly C) over purines with an order of C Ͼ T Ͼ G, A. These preferences were also confirmed in the present study (Fig. 7). Interestingly, the stringent paired base preferences of Fpg and hOGG1 were markedly relieved for me-Fapy (Fig. 7). me-Fapy in DNA constitutes a strong block to DNA replication but not a premutagenic lesion (13,14). These data suggest that template me-Fapy does not direct incorporation of any nucleotides opposite the lesion, hence arresting DNA replication. Therefore, me-Fapy (and probably its analog Fapy as well) is likely to exist exclusively as a me-Fapy (Fapy):C pair derived from a G:C pair in cells. Thus, unlike 8-oxoG potentially existing either as an 8-oxoG:C or an 8-oxoG:A pair in cells (6,7), Fpg and hOGG1 may not need to discern the base opposite me-Fapy (and Fapy) strictly since only me-Fapy (Fapy):C is actually subjected to repair by these enzymes. Granting that this is the case, mechanistic questions still remain as to how the base opposite 8-oxoG and me-Fapy differentially affects the damage recognition by Fpg and so does by hOGG1. The situation is further complicated by their distinct activities for abasic sites. With respect to the paired base effect on abasic sites, human (36) and yeast (37) OGG1 have a preference similar to that for 8-oxoG (C Ͼ T Ͼ G, A), whereas Fpg shows little preference (37). Combining the present and previous results, it follows that hOGG1 shows a stringent paired base preference for 8-oxoG and abasic sites (C Ͼ T Ͼ G, A) but not for me-Fapy, whereas Fpg shows the preference only for 8-oxoG (G, C, T Ͼ A) but not for me-Fapy and abasic sites. Despite their functional similarity, primary amino acid sequences of Fpg and hOGG1 are totally different. Fpg is a metalloenzyme with a zinc finger motif located at the C terminus and is suggested to use a proline residue in the N-terminal region for catalysis (24,40). In contrast, hOGG1 is a member of endonuclease III superfamily containing the HhH-GPD motif and is assumed to use Lys-249 and Asp-268 in the HhH-GPD motif for catalysis (23). Thus, the architecture of the active site accommodating lesioned DNA and amino acid residues involved in catalysis are expected to be quite different between Fpg and hOGG1. However, it remains to be seen how these differences lead to the distinct damage recognition by hOGG1 and Fpg.
Distinct Sequence Context Effects on Damage Recognition-There is heterogeneity in DNA repair, and the heterogeneity in base excision repair has been mostly related to the sequence context (Refs. 41 and 42 and references cited therein). In the present study, the repair efficiency of individual me-Fapy varied 9.7-63.8% for Fpg and 5.3-91.9% for hOGG1 (Table IV), showing sequence-dependent variations of 6.6-fold for Fpg and 17.2-fold for hOGG1. According to the reported data for Fpg, the sequence-dependent variations of the in vitro repair rate of me-Fapy (43) and 8-oxoG (44) are 2-and 33-fold, respectively. The repair rate of me-Fapy by Fpg was previously compared using conventionally prepared 12-mer substrates with only four different sequence contexts. A larger variation of the repair rate of me-Fapy observed in this study (6.6-fold for Fpg) is probably attributable to the more diverse sequence samples. Fpg has been also suggested to excise efficiently me-Fapy in G-rich regions (43). According to the present result (Table IV), me-Fapy flanked by 5Ј-G (i.e. 5Ј-GF-3Ј, F ϭ me-Fapy) was excised more efficiently than that flanked by 3Ј-G (i.e. 5Ј-FG-3Ј) when the relative repair efficiencies of me-Fapy in the individual G doublet were compared (positions 3/4, 7/8, 10/11, 12/13, 19/20). The influences of 5Ј-and 3Ј-flanking G were common for Fpg and hOGG1. Thus, G flanking at 5Ј and 3Ј sides exerted opposite influences on the repair of me-Fapy. Efficiently and inefficiently repaired consensus sequences for 8-oxoG have been also reported (44). In the four consensus sequences poorly recognized by Fpg, three of them contain 8-oxoG flanked by 3Ј-A. We compared the present data and reported sequence data. However, the me-Fapy lesions flanked by 3Ј-A were not necessarily poor substrates in this study (Table IV), implying the subtle interplay of the lesion structure (8-oxoG versus me-Fapy) and surrounding sequences in the Fpg-DNA interaction. This was also the case for efficiently excised sequences when the present and reported (44) data were compared. The sequence context effect on the repair rate of me-Fapy by hOGG1 was quite different from that by Fpg (Table IV). For instances, me-Fapy at position 16 was poorly removed by Fpg but efficiently by hOGG1. Conversely, me-Fapy at positions 17 and 18 was efficiently removed by Fpg but poorly by hOGG1. Thus, no correlation was observed between the orders of preferred to unpreferred sequences for Fpg and hOGG1 (correlation coefficient (r) ϭ 0.074). Although Fpg did not show any obvious consensus sequences for efficient or inefficient repair except for the G doublets mentioned above, hOGG1 showed a preference for 5Ј-(C/G)FC-3Ј (positions 8, 9, 11, and 13 in Table IV). No exception was observed for this rule. As mentioned in the previous section, Fpg and hOGG1 are functionally similar with respect the substrate specificity but are structurally unrelated