Kinetic Conformational Analysis of Human 8-Oxoguanine-DNA Glycosylase*

7,8-Dihydro-8-oxoguanine (8-oxoG) is one of the major DNA lesions formed by reactive oxygen species that can result in transversion mutations following replication if left unrepaired. In human cells, the effects of 8-oxoG are counteracted by OGG1, a DNA glycosylase that catalyzes excision of 8-oxoguanine base followed by a much slower β-elimination reaction at the 3′-side of the resulting abasic site. Many features of OGG1 mechanism, including its low β-elimination activity and high specificity for a cytosine base opposite the lesion, remain poorly explained despite the availability of structural information. In this study, we analyzed the substrate specificity and the catalytic mechanism of OGG1 acting on various DNA substrates using stopped-flow kinetics with fluorescence detection. Combining data on intrinsic tryptophan fluorescence to detect conformational transitions in the enzyme molecule and 2-aminopurine reporter fluorescence to follow DNA dynamics, we defined three pre-excision steps and assigned them to the processes of (i) initial encounter with eversion of the damaged base, (ii) insertion of several enzyme residues into DNA, and (iii) enzyme isomerization to the catalytically competent form. The individual rate constants were derived for all reaction stages. Of all conformational changes, we identified the insertion step as mostly responsible for the opposite base specificity of OGG1 toward 8-oxoG:C as compared with 8-oxoG:T, 8-oxoG:G, and 8-oxoG:A. We also investigated the kinetic mechanism of OGG1 stimulation by 8-bromoguanine and showed that this compound affects the rate of β-elimination rather than pre-excision dynamics of DNA and the enzyme.

Living cells continuously experience a great number of insults from reactive oxygen species that are both produced during aerobic respiration and generated by environmental factors such as ionizing radiation (1). The adverse effects of oxidative damage to DNA include miscoding and mutagenesis, cytotoxicity, and disregulation of gene expression and may ultimately lead to cancers and aging (2). To counteract these effects, cells maintain an extensive system of antioxidant defense, one branch of which is formed by DNA repair mechanisms (3). Base excision repair is one subpathway that mostly deals with non-bulky base lesions and single-strand breaks, the lesions that are predominantly produced by oxidative damage (4,5). Base excision repair of damaged bases is initiated by DNA glycosylases, enzymes that recognize such lesions and hydrolyze their N-glycosidic bonds (3). In eukaryotes, a prominent role is played by 8-oxoguanine-DNA glycosylase OGG1, which excises 8-oxoguanine (8-oxoG) 3 (Fig. 1), a major damaged purine, from DNA (6 -8). Defects in OGG1 have been associated with human cancers (9) and enhanced mutagenesis (10,11), and accelerated senescence has been observed in a mouse strain with thermolabile OGG1 (12).
In addition to 8-oxoG, OGG1 has been shown to excise formamidopyrimidine derivatives of G (13,14). During the reaction, a covalent Schiff base intermediate is formed between an active site lysine residue (Lys-249 in human OGG1) and C-1Ј of the damaged nucleoside (15). Similar to other DNA glycosylases that form the Schiff base (16), OGG1 possesses an abasic (apurinic/apyrimidinic (AP)) lyase activity, which catalyzes ␤-elimination at the nascent AP site to break the damaged DNA strand. This activity is weak, being about 1 order of magnitude lower than the glycosylase activity (17,18), but can be enhanced by analogs of 8-oxoG such as 8-bromoguanine   (19,20). OGG1 demonstrates a profound specificity for the base opposite the lesion, especially if DNA strand cleavage is used as the assay end point, with C being preferred at this position and purines strongly discriminated against (17,18). If C is opposite the lesion, the enzyme may even excise bases not cleaved in their natural context; for example, it excises 8-oxoadenine from 8-oxoA:C mispairs (18,21). As with many other DNA glycosylases, the structure of OGG1 is known (22), but not all aspects of its substrate preference can be explained solely by structural considerations, suggesting that dynamic features of lesion recognition can contribute to the enzyme specificity (reviewed in Ref. 23).
Recognition and excision of damaged bases by DNA glycosylases is accompanied by several conformational rearrangements that bring the base into the enzyme catalytic site. As a rule, DNA is severely kinked at the site of the lesion, the damaged nucleotide is everted from the double helix and inserted in a deep catalytic pocket, and several amino acid residues of the enzyme are inserted into the resulting void in DNA (a process commonly referred to as "plugging") to stabilize the whole structure (for a recent review, see Ref. 24). These processes occur in a rapid consequence and are not easily studied by conventional steady-state enzyme kinetic methods. Recently, stopped-flow techniques have been applied by several groups, including ours, to investigate the multiple conformational changes accompanying damage recognition (20,(25)(26)(27)(28)(29). In particular, we have used stopped-flow with detection of intrinsic tryptophan (Trp) fluorescence to study the dynamics of 8-oxoG recognition and removal by human OGG1 and its activation by . We were able to derive a minimal kinetic scheme (Scheme I) of this process and to suggest the conformational changes underlying each step. However, DNA conformational dynamics during OGG1 catalysis was not addressed, and neither was the amazing opposite-base specificity of this enzyme. In the present study, we have investigated changes in the conformation of DNA processed by OGG1 by using 2-aminopurine (2-aPu) as a fluorescent marker complementing the Trp fluorescent studies. This nucleobase analog has a high quantum yield in aqueous solution, but the fluorescence is highly quenched when it is incorporated into DNA or transferred into a nonpolar environment (30,31). These properties make 2-aPu attractive for stopped-flow analysis of protein-DNA interaction; indeed, the combination of Trp and 2-aPu fluorescence has been used to dissect the kinetic pathway of damage recognition by the repair enzyme uracil-DNA glycosylase (32). We have also analyzed the DNA dynamics associated with stimulation of the AP lyase activity of the enzyme by 8-BrG. Finally, we have addressed the origins of OGG1 specificity for the base opposite the lesion by stopped-flow kinetics.
Oligodeoxynucleotides and Enzymes-Human OGG1 was purified as described (20). The final enzyme stock of 23.4 M, as determined spectrophotometrically using the Gill-von Hippel algorithm (34) (⑀ 280 ϭ 6.84 ϫ 10 4 M Ϫ1 cm Ϫ1 ), was stored at Ϫ20°C. The purified OGG1 had 65% of the active enzyme form as determined by borohydride trapping with 8-oxoG substrate as follows. The reaction mixture (10 l) included 2 M enzyme, 25 mM potassium phosphate (pH 6.8), 100 mM NaCl, 100 mM NaBH 4 , and varying amounts of oligonucleotide duplex containing an 8-oxoG residue opposite C. The samples were incubated for 1 h at room temperature and analyzed by 12% SDS-PAGE. The gel was stained with Coomassie Blue and quantified using Gel-Pro Analyzer 4.0 software (Media Cybernetics, Silver Spring, MD). The concentration of active form of enzyme was taken into account in all experiments. The sequences of ODNs used in this work are listed in Table 1. The ODNs were synthesized on an ASM-700 synthesizer (Biosset, Novosibirsk, Russia) using phosphoramidites purchased from Glen Research (Sterling, VA) and purified by anion exchange HPLC on a Nucleosil 100-10 N(CH 3 ) 2 column followed by reverse-phase HPLC on a Nucleosil 100-10 C 18 column (both columns from Macherey-Nagel, Düren, Germany). The purity of ODNs exceeded 98% as estimated by electrophoresis in 20% denaturing PAGE after staining with the Stains-All dye (Sigma-Aldrich). The concentration of the ODNs was determined from their absorbance at 260 nm. ODN duplexes were prepared by annealing modified and complementary strands taken at 1:1 molar ratio in the reaction buffer described under "Materials and Buffers." Stopped-flow Fluorescence Measurements-Stopped-flow measurements with fluorescence detection were carried out essentially as described (20,35) using a model SX.18MV stopped-flow spectrometer (Applied Photophysics). To detect intrinsic Trp fluorescence only, ex ϭ 283 nm was used and em Ͼ 320 nm was followed as transmitted by a Schott filter WG 320 (Schott, Mainz, Germany). If 2-aPu was present in the ODNs, ex ϭ 310 nm was used to excite 2-aPu residues, and their emission was followed at em Ͼ 370 nm (Corion filter LG-370); Trp fluorescence of the enzyme ( ex ϭ 283 nm) was detected in this case using a Corion filter P10-340, which transmits a 10-nm-wide band at ϳ340 nm to avoid overlapping with 2-aPu emission. The dead time of the instrument was 1. Bleaching of Enzyme Fluorescence-For the correction of the measured data on bleaching effect, the fluorescence intensities were recalculated using Equation 1 (28), where F is the corrected fluorescence intensity, F obs is the observed fluorescence intensity, F b is the background fluorescence, and k b is the coefficient determined for each substrate concentration in experiments with noncleaved substrates. The difference between the observed and corrected values did not exceed 10%.
Global Nonlinear Simulation Fitting of Stopped-flow Data-Accurate modeling of all stopped-flow traces was obtained by numerical fitting using DynaFit software (BioKin, Pullman, WA) (36) as described (20,29,35). Differential equations were written for each species in the mechanisms described by Schemes I-VII (see "Results"), and the stopped-flow fluorescence traces were directly fit by expressing the corrected fluorescence intensity (F c ) at any reaction time t as the sum of the background fluorescence (F b ) and the fluorescence intensities of each protein species, where F i (t) ϭ f i (E i (t)), f i are the coefficients of specific fluorescence for each discernible OGG1 conformer, and (E i (t)) are the concentrations of the conformers at any given time t (i ϭ 0 relates to the free protein and i Ͼ 0, to the protein-DNA complexes). These specific fluorescence coefficients describe only the part of fluorescence that changes due to DNA binding. The minimal nature of the reaction schemes was confirmed by analyzing the dependence of the standard deviation of the residuals on the number of steps in the scheme using a scree plot (35).
Product Analysis-To analyze products formed by OGG1, the substrate oligonucleotides were 5Ј-32 P-labeled using T4 polynucleotide kinase and [␥-32 P]ATP. Reaction mixtures (20 l) contained reaction buffer, 1, 2, or 4 M 32 P-labeled substrate, and 2 M enzyme. The reaction was initiated by adding the enzyme and allowed to proceed at 25°C for 25-60 min. Aliquots (2 l) were withdrawn as required, mixed with 3 l of gel-loading dye containing 7 M urea, and analyzed by 20% denaturing PAGE. The gels were exposed to Agfa CP-BU x-ray film (Agfa-Geavert), and the autoradiograms were scanned and quantified using Gel-Pro Analyzer, version 4.0. Kinetic parameters were obtained by numerical fitting using Microcal Origin version 7.0 software (OriginLab, Northampton, MA).

Rationale
Recently, we studied the dynamics of fluorescence of OGG1 tryptophan residues in the course of enzymatic reaction using stopped-flow technique (20). When the substrate contained 8-oxoG opposite C, a series of changes in Trp fluorescence intensity were observed, attributed to binding and catalytic stages of enzyme process. The proposed kinetic scheme of OGG1 interaction with its DNA substrate included at least three fluorescently discernible consecutive steps accompanying damaged base recognition and its binding in the active site of the enzyme.
The rate of damaged base excision (DNA glycosylase activity) by OGG1 is ϳ10-fold higher than the rate of elimination of the 3Ј-phosphate of the damaged nucleotide (AP lyase activity) (18). This difference allowed us to propose the kinetic mechanism containing three equilibrium steps of the specific complex formation and two non-equilibrium catalytic steps. The nonequilibrium steps were attributed to glycosylase reaction and ␤-elimination. At the end of the process, formation of an enzyme-product complex was observed indicated by a decrease in final Trp fluorescence intensity in comparison with its initial values (Scheme I). However, only weak changes in the fluorescence intensity of the Trp residues of the enzyme were detected when OGG1 interacted with duplexes containing G, F, or AP residues opposite C (20).
In the present work, to gain a better understanding of the recognition of various substrates by OGG1, we studied a set of model duplexes (see Table 1 for the sequences) in which the target residue (8-oxoG, AP site, F, or G; Fig. 1) was flanked by 2-aPu, a fluorescent marker sensitive to the structure of DNA duplex. This approach allowed us to observe conformational changes in the oligonucleotide duplex in parallel with those in the enzyme and to assign conformational changes in the interacting molecules to the elementary steps in Scheme I. Furthermore, we determined the enzyme specificity for substrates with different bases opposite 8-oxoG and defined the stages of discrimination of these substrates by OGG1. We also analyzed the effect of 8-BrG on the processing of substrates containing 8-oxoG or AP.

Interactions of OGG1 with Substrates Containing 2-aPu
G-ligand-The process of OGG1 binding the nonspecific G-ligand was essentially complete by 0.02 s ( Fig. 2A). An increase in fluorescence intensity of 2-aPu was observed, indicating destabilization of Watson-Crick or stacking interactions in the primary nonspecific enzyme-DNA complex (30,31). Fitting the experimental data to the one-site binding model (Scheme II) gave the values for the forward and reverse rate constants, F-ligand-The kinetic curves for the F-ligand, a non-cleavable analog of the AP site, were characterized by a decrease in 2-aPu fluorescence intensity during 5 s, suggestive of transition of the 2-aPu residue to a more hydrophobic environment (Fig.  2B). This could be the result of filling the abasic void in the DNA duplex by amino acids of OGG1 (37). The rate constants of the forward and reverse reactions of this single-stage mechanism (Scheme II) are k 1 ϭ (0.48 Ϯ 0.05) ϫ 10 6 M Ϫ1 s Ϫ1 and k Ϫ1 ϭ 0.23 Ϯ 0.04 s Ϫ1 , respectively. SCHEME I. Kinetic mechanism of OGG1 processing of the 8-oxoG substrate. E, OGG1; OG, 8-oxoG substrate; (E⅐OG) n and E⅐AP, different enzyme-substrate complexes; P, reaction product; E⅐P, enzyme-product complex; k i and k Ϫi , individual rate constants.

JOURNAL OF BIOLOGICAL CHEMISTRY 1031
AP Substrate-The substrate containing the AP site is expected to interact with the enzyme in a more complicated way. In this case, DNA binding and void-filling by OGG1 should be followed by ␤-elimination and dissociation of the enzymeproduct complex. The last process resulted in an increase in the 2-aPu fluorescence intensity due to a transition of the 2-aminopurine to a more hydrophilic environment (Fig.  2C) at times Ͼ100 s. Scheme III describes the observed fluorescence changes in minimal terms. Its first step obviously reflects substrate binding and the transition to the catalytically active complex (E⅐AP). The irreversible step was attributed to the reaction of ␤-elimination. Therefore, the final step of the scheme most likely corresponds to the equilibrium between OGG1 and the reaction product. However, although we did see the beginning of this last reversible stage, the equilibrium could not be achieved, likely because of a very tight product binding. Therefore, numerical values were not calculated for K p ; Table 2 presents the rate constants obtained by fitting. The forward and reverse rate constants of the binding step (k 1 and k Ϫ1 , respectively) were close for the F-ligand and AP substrate, suggesting that this step reflects identical processes in both cases.
8-oxoG Substrate-The interaction of OGG1 with the 8-oxoG substrate included additional steps during formation of the catalytically active complex as can be seen from Fig. 2D. The formation of the primary nonspecific complex led to partial duplex melting as in the case of the G-ligand ( Fig.  2A). However, the increase in fluorescence intensity was more pronounced for 8-oxoG substrate than for the G-ligand. We suggest that, during this time interval, the 8-oxoG residue is flipped out from DNA helix and inserted into the active site of the enzyme, leaving a void in the DNA helix. This void is then filled with several amino acid residues of OGG1, resulting in a decrease in 2-aPu fluorescence, as was also observed with the AP substrate. Fluorescence traces in Fig. 2D indicate that at least two intermediates were involved in this process, which proceeded 5-10-fold more slowly than in the case of the AP substrate and was completed in 50 -100  SCHEME II. E is OGG1; S is DNA substrate; E⅐S is the enzyme-substrate complex; k 1 and k ؊1 are the rate constants of formation and dissociation of ES. SCHEME III. Kinetic mechanism of OGG1 processing of the AP substrate. E, OGG1; AP, AP substrate; E⅐AP, enzyme-substrate complex; P, reaction product; EP, enzyme-product complex; k 1 , k Ϫ1 , and k 2 , individual rate constants; K p , dissociation constant of the enzyme-product complex.
s. The minimal kinetic scheme describing the observed changes of 2-aPu fluorescence intensity was identical to that proposed for the description of Trp fluorescence changes (Scheme I) and contained three equilibrium steps that characterized substrate binding followed by two irreversible chemical steps and then an equilibrium step of product release.
Trp Fluorescence of the 2-aPu-containing 8-OxoG Substrate-In addition to 2-aPu fluorescence, we followed the dynamics of fluorescence intensity of OGG1 Trp residues during processing of 2-aPu-containing substrates. Fig. 2E shows that Trp fluorescence intensities remained essentially unchanged at the time scales (10 -20 ms) corresponding to nonspecific enzyme-substrate binding and 8-oxoG eversion. However, the Trp fluorescence intensity decreased afterward in concordance with void filling by the amino acid residues of the enzyme and the formation of the catalytic complex.
All fluorescently discernible steps of catalytic complex formation between OGG1 and the specific 8-oxoG substrate were completed by 50 -100 s. The following chemical steps and the dissociation of the enzyme-product complex led to an increase in the fluorescence intensity of both 2-aPu (because of transition of this residue to a more hydrophilic environment) and Trp (the return of the protein to its initial free conformation). Scheme I was also valid for the description of these Trp fluorescence intensity changes. The rate constants of the elementary steps and the total binding constant of OGG1 association with the 8-oxoG substrate (K bind ) estimated according to this kinetic scheme are listed in Table 2.
Chemical Quench Assay-The rates of formation of the APintermediate and the nicked product were directly measured by PAGE with 32 P-labeled 8-oxoG substrate (Fig. 3A). To observe formation of the AP-intermediate after base excision, the reaction mixture was treated with alkali. Scheme IV was used to describe the kinetic curves obtained. The value of equilibrium constant was taken from stopped-flow data ( Table 2) as the total binding constant, K bind . The constant of the irreversible step, k glyc , estimated by nonlinear fitting, was 0.03 s Ϫ1 , closely matching the value for k 4 , obtained from the stopped-flow experiments.
The rate constant of the ␤-elimination reaction, k elim , was determined in the same way using Scheme V to describe the time course curves obtained (Fig. 3B). The rate constants of the chemical steps (k glyc and k elim ) of enzymatic process obtained from the direct PAGE analysis of the reaction products were in a good agreement with the values of these constants (k 4 and k 5 , respectively) obtained from the fluorescence studies ( Table 2).

TABLE 2 The rate constants for interactions of OGG1 with G-and F-ligands and AP and 8-oxoG substrates
See Fig. 1 for a description of the abbreviations used in this table. Here and throughout the text, the superscript Trp indicates data obtained in and constants derived from the experiments with tryptophan fluorescence detection, and the superscript 2-aPu indicates data obtained in and constants derived from the experiments with 2-aminopurine fluorescence detection.

Effect of 8-Bromoguanine on the Rates of OGG1-catalyzed Reactions
The 8-oxoG base released by the glycosylase reaction accelerates ␤-elimination by OGG1 (19). It was shown earlier that the rate of the AP substrate cleavage increases at least 10-fold by 8-BrG, an 8-oxoG analog, present in the reaction mixture at 0.5 mM (19,20). Recently we have shown that 8-BrG increases the rate of ␤-elimination not only for the AP substrate but also for the 8-oxoG substrate (20).
We addressed the effect of 8-BrG on 2-aPu fluorescence of AP and 8-oxoG substrates (Fig. 4A). The rate constant of ␤-elimination of the AP substrate (k 2 in Scheme III) was increased 5.7-fold (from 0.0021 to 0.012 s Ϫ1 ) ( Table 3). When OGG1 processed the 8-oxoG substrate, the fluorescent traces showed a notable change at times exceeding 100 s (Fig. 4A). Instead of two irreversible steps describing base excision and ␤-elimination in Scheme I, the minimal kinetic scheme in the presence of 8-BrG (Scheme VI) contained only one irreversible step, characterized by the rate constant k 4 . The values of k 4 determined from 2-aPu and Trp fluorescence traces were 0.0070 and 0.0077 s Ϫ1 , respectively (Table 3, Scheme VI).
Direct PAGE analysis of the products of 8-oxoG substrate cleavage in the presence of 8-BrG was also performed (Fig. 4B). The data could be described by Scheme VII. The rate constant of the irreversible stage k elim was 0.012 s Ϫ1 . This value is higher than those obtained by fitting of the fluorescence data. In the latter case, significant bleaching at longer times can lower the accuracy of k 4 value determination. On the other hand, this value of k elim coincides with the k 2 value obtained for the AP substrate and is twice the rate constant of ␤-elimination without 8-BrG. Consequently, the rate of ␤-elimination of the 8-oxoG substrate was increased at least 2-fold in the presence of 8-BrG.
Therefore, we suggest that at high concentrations 8-BrG binds in the active site of OGG1. This complex binds the AP substrate just like the free enzyme, but the chemical step proceeds faster with 8-BrG present. When OGG1 interacts with the 8-oxoG substrate and cleaves its N-glycosidic bond, 8-oxoG can either be retained in the active site of the enzyme or quickly exchanged for the free 8-BrG base.

Interaction of OGG1 with 8-OxoG-containing Mispairs
Recognition by OGG1 of its natural substrate, a 8-oxoG:C pair, involves formation of a set of specific bonds with cytosine opposite the lesion (22). Removal of the oxidized base from mispairs 8-oxoG:A, 8-oxoG:G, and 8-oxoG:T would lead to a mutation in DNA. This may be the reason for pronounced opposite-base kinetic specificity of eukaryotic OGG1 proteins (17,18,38), their substrate order of reference being 8-oxoG: C Ͼ 8-oxoG:T Ͼ 8-oxoG:G Ͼ 8-oxoG:A. This preference is more pronounced at the level of strand nicking than at the level of base excision (17,18).
To clarify the mechanism of discrimination of different 8-oxoG-containing mismatched substrates, we investigated the reaction of OGG1 with 8-oxoG:T, 8-oxoG:G, and 8-oxoG:A substrates by stopped-flow kinetics with Trp fluorescence detection. The fluorescence traces had the same overall shape for these three substrates and differed from that for 8-oxoG:C (Fig. 5A). At the same time, the absolute values of the fluorescence intensity changes differed between the mispaired substrates, indicating different depths of their conversion by the enzyme. The smallest change in the fluorescence intensity was observed for the 8-oxoG:G substrate. A direct PAGE analysis of the products of cleavage for 8-oxoG:N substrates more accurately defined the preference order under the conditions used (Fig. 5, B and C). The cleavage efficiency, both for base excision Scheme I was used to obtain the rate constants of the elementary steps (Table 4) and to define the stages contributing most in the discrimination of good versus poor substrates (23). The first step, nonspecific binding, proceeded with approximately similar rate constants for both the forward and the reverse reaction for all substrates. The equilibrium constant of the first step fell in the range of (0.6 -2.0) ϫ 10 6 M Ϫ1 for both 8-oxoG:C and the mispaired substrates, with a small preference for 8-oxoG:C (K 1 was 1.0 -3.5-fold higher than for other substrates). The second step of this scheme clearly distinguishes the 8-oxoG:C substrate from other mispairs. The rate constant of the forward reaction at this step was at least twice as high for 8-oxoG:C substrate as for other substrates, and the rate constant of the reverse reaction was at least twice as low. The preference for 8-oxoG:C at this step (in terms of K 2 ) was 4.5-fold over 8-oxoG:T, 17-fold over 8-oxoG:A, and 35-fold over 8-oxoG:G. In contrast, the equilibrium constant of the third step disfavored 8-oxoG:C as compared with all other substrates (K 3 was 5-22-fold lower for 8-oxoG:C), coinciding with our observations for Escherichia coli Fpg (35). Overall K bind values for 8-oxoG:C exceeded those for the mispaired substrates by 3-20fold in the order of opposite-base preference, C Ͼ T Ͼ A Ͼ G. The rate constants of the N-glycosylase and ␤-elimination reactions did not fluctuate much between different substrates (0.03-0.07 and 0.006 -0.010 s Ϫ1 , respectively), with no preference shown for 8-oxoG:C. The equilibrium constant of the product release stage was 3-13-fold higher for substrates containing T, G, or A opposite 8-oxoG, an evidence of less stable OGG1 complexes with products of mispaired substrates.

DISCUSSION
In our previous study (20) we had investigated the dynamics of Trp fluorescence of OGG1 and suggested the most likely assignments of different fluorescently discernible steps to individual conformational change events during damaged base recognition and removal. However, Trp fluorescence alone could not offer unambiguous identification of conformational changes in the DNA molecule, leaving room for uncertainty in this assignment. In the present work, we used stopped-flow with detection of 2-aPu fluorescence to address conformational dynamics of DNA ligands and substrates during their processing by OGG1, as reported earlier for uracil-DNA glycosylase (32). The quantum yield of 2-aPu fluorescence decreases in a less polar environment (31); therefore, increases in the 2-aPu signal indicate destabilization of the stacking interactions around this residue, and decreases in the signal correspond to their strengthening.
To ensure a proper comparison of 2-aPu and Trp dynamics, we also traced the Trp fluorescence of OGG1 processing 2-aPucontaining substrates and compared it with the fluorescence of OGG1 on substrates lacking this fluorescent base (20). The minimal kinetic scheme of this reaction on the 8-oxoG substrate with 2-aPu adjacent to the lesion (Scheme I) was identical to that determined earlier for 8-oxoG-containing substrates without 2-aPu (20). The values of the individual rate constants were generally similar; however, in some cases more significant changes were observed (compare the last column in Table 2 and the first column in Table 4). Incorporation of 2-aPu had a pronounced effect on the third equilibrium in Scheme I (which probably corresponds to an isomerization of the E⅐S complex to a catalytically competent state; see below) because of an increase in k 3 and a decrease in k Ϫ3 . In addition, product release was impeded with 2-aPu-containing ODNs, with K P decreasing up to 1 order of magnitude. These differences are evident from the shapes of the respective fluorescent traces and thus are unlikely to be calculation artifacts. They may be due to the sequence effect of the 2-aPu replacing C in the immediate vicinity of the 8-oxoG residue, with the accompanying changes in the DNA structure around the lesion. Most importantly, the introduction of 2-aPu affected neither the overall reaction scheme nor the direction of any equilibrium step except the third one in Scheme I.
The combination of tryptophan and 2-aminopurine fluorescence changes (Fig. 6) permits us to make more precise conclusions about the nature of each stage in the kinetic mechanisms of interaction of OGG1 with G-and F-ligands and AP and SCHEME VI. Kinetic mechanism of OGG1 processing of the 8-oxoG substrate in the presence of 8-BrG. E, OGG1; OG, 8-oxoG substrate; (E⅐OG) n and E⅐AP, different enzyme-substrate complexes; P, reaction product; E⅐P, enzyme-product complex; k i and k Ϫi , individual rate constants. SCHEME VII. Accumulation of the ␤-elimination product in OGG1-catalyzed reaction in the presence of 8-BrG. E, OGG1; OG, 8-oxoG substrate; E⅐OG, enzyme-substrate complex; P, product of ␤-elimination; K bind , equilibrium constant of E⅐OG formation; k elim , rate constant of ␤-elimination.
1.6 ϫ 10 7 7.1 ϫ 10 6 5.1 ϫ 10 7 8-oxoG substrates. Nonspecific binding of the enzyme to DNA (G-ligand) quickly led to destabilization of the double helix, with the characteristic time of the process being Ͻ10 ms. This destabilization may be because of the distortion introduced by sharp kinking of nonspecific DNA by OGG1, demonstrated by atomic force microscopy (39). A recently published structure of OGG1 in a covalent disulfide complex with nonspecific DNA suggests that the enzyme is capable of everting the undamaged G into an "exo-site," a base-accommodating site different from the catalytic pocket (40). Therefore, the increase in 2-aPu fluorescence we observed could be due to the disruption of the stacking interactions between 2-aPu and G after the eversion of the latter. Caution is prudent when analyzing the structures of undamaged DNA cross-linked to DNA repair proteins by the disulfide method, which by its nature favors conformations most resembling the respective specific complexes (41). For instance, the OGG1/undamaged DNA structure features the void-filling residues inserted into the helix (40). However, this may well be an artifact, because the cross-linking was done through one of these residues to the base opposite the everted G and thus the procedure would automatically select for voidfilled conformations. 2-aPu fluorescence traces with the G-ligand do not show any step that would correspond to voidfilling. Hence, binding of OGG1 to nonspecific DNA may lead to DNA kinking and the eversion of the normal base into the exo-site, but it is unlikely to be stabilized there by void-filling, allowing the search for the lesion to be resumed after fast sampling of the normal base (23,42). F-ligand and AP substrate revealed one-step binding by 2-aPu fluorescence. Notably, the accumulation of the fluorescently discernible enzyme-DNA complex was much slower in these cases, with k 1 being ϳ2-3 orders of magnitude lower than for the G-ligand, albeit a compensatory decrease in k Ϫ1 produced similar overall K d values for the G-ligand, F-ligand, and AP substrate. In addition, this single binding step was accompanied with the decrease in 2-aPu fluorescence rather than its increase, as was the case with the G-ligand. It is clear that the processes registered in one-step binding of the F-ligand and AP substrate are physically distinct from those seen during onestep binding of the G-ligand. As both the F-ligand and AP substrate lack a base adjacent to a 2-aPu residue, we suggest that the initial eversion of the damaged nucleoside is not detected in these cases and that an observed slower change in fluorescence reflects a process of void-filling in which the environment of 2-aPu is made more hydrophobic by the insertion of Asn-149/ Asn-150 into the void (37), with a characteristic time of ϳ1 s.
In our earlier experiments with Trp fluorescence detection of OGG1 interactions with the AP substrate (20), we failed to detect the irreversible catalytic stage in the absence of 8-BrG, attributing this to very slow ␤-elimination and/or tight product binding. In the present study, the introduction of 2-aPu label into DNA helped us to detect the catalytic process and calculate its rate constant (k 2 for the AP substrate). Clearly, the elimination of the C-3Ј-O bond causes a conformational rearrangement in the enzyme-DNA complex (43) that is accompanied by no Trp movement but some changes in the 2-aPu environment. However, k 2 for the AP substrate is still slower than the corresponding rate constant, k 5 , for the 8-oxoG substrate, most likely because 8-oxoG, which acts as a general base in ␤-elimination (19), is absent from the active site in the former case.
When a damaged base is present, as in the 8-oxoG substrate, three equilibrium steps preceding the irreversible chemical steps are detected by both 2-aPu and Trp fluorescence (Fig. 2, D and E; see also Fig. 6). The first one proceeds with the same characteristic time as the only step observed with the G-ligand and is also accompanied with an increase in 2-aPu fluorescence. Therefore, if one assumes that this step in G-ligand corre- sponds to the bona fide eversion of G into the exo-site, one can expect that 8-oxoG initially also falls into the exo-site. Quantum mechanical/molecular mechanical (QM/MM) free energy simulations suggest that 8-oxoG can indeed be stabilized in the exo-site and that it is energetically preferred over G there, albeit by a small margin (40). The second step has the same characteristic time scale as the only equilibrium step in the case of F-ligand and AP substrate, and it is accompanied by a decrease in 2-aPu fluorescence, Moreover, this step had the highest discriminatory power favoring C, the natural and preferred opposite base, over A, G, or T. Accordingly, the second step likely corresponds to the void-filling process, during which contacts with the base opposite the lesion are formed. The third step then should reflect isomerization of the everted and plugged enzyme-DNA complex into the pre-catalytic conformation; this may include transfer of 8-oxoG from the exosite to the catalytic site as well as the conformational adjustment of the protein globule. This most plausible interpretation of the observed fluorescent traces suggests that damaged nucleotide eversion and plugging are separated in time rather than concurrent (as proposed earlier based solely on the analysis of structures of free OGG1 and OGG1 bound to damaged DNA (44)). The sequential mechanism of damaged base recognition, eversion, and plugging is remarkably similar to the multistep mechanism previously observed for uracil-DNA glycosylase (25,26,32) and E. coli Fpg (29,35), perhaps underlying the general commonality of the kinetic mechanisms of lesion recognition (23).
The weak AP lyase activity of OGG1 on AP substrates can be activated by 8-oxoG base and its analogs, such as 8-BrG, 8-aminoguanine, or even guanine (19). 8-BrG is an especially useful reagent in this assay because, just as 8-oxoG, it has an electronegative substitute at C-8 of the purine moiety and is much more soluble that 8-oxoG. The proposed mechanism of activation involves a capture of the base in the active site of the enzyme, and its use as a general base to strip C-2Ј of its pro-Sproton, which becomes increasingly acidic after the Schiff base formation (19). In our previous work, we had also found that 8-BrG enhances the AP lyase activity of OGG1 on 8-oxoGcontaining substrates too (20), likely indicating fast exchange of the excised base for the base analog free in solution. Although barely influencing individual rate constants at pre-excision equilibrium stages, 8-BrG accelerates the ␤-elimination step to the point that it is no longer rate-limiting in the reaction as detected by Trp fluorescence (20). In the present study, we confirmed that the same conclusions hold if the DNA dynamics is followed by 2-aPu fluorescence. The reaction scheme for the 8-oxoG substrate (Scheme VI) also included three pre-excision and one post-excision equilibrium steps with parameters similar to those in the absence of 8-BrG, but both irreversible steps coalesced into a single fluorescently discernible stage. Similarly, 8-BrG did not significantly influence the reaction scheme or parameters for the AP substrate except for the acceleration of the lyase step. Overall, the present results are consistent with the kinetic model of stimulation by 8-BrG proposed earlier from our previous data on Trp fluorescence (20).
Kinetic mechanisms of discrimination of good and poor substrates are of special interest for substrates that do not principally diverge in their basic chemical structure, i.e. they can in principle undergo the same chemical transformation (23). The same damaged base, paired with different opposite bases, constitutes a notable example of such substrates for DNA glycosylases. In human OGG1, the cytosine base forms multiple hydrogen bonds with Asn-149, Arg-154, and Arg-204 and is engaged in a van der Waals interaction with Tyr-203 (22). All of these interactions require insertion of the amino acid residues into DNA after base eversion. No structure of OGG1 complexed with any DNA containing G, T, or A opposite the lesion is presently available. In a bacterial 8-oxoG-DNA glycosylase Fpg, which is not a structural homologue of OGG1, conformational adjustments required to accommodate T or G opposite the lesion are minimal, are confined to the region immediately  surrounding the opposite base, and do not extend into the 8-oxoG-binding pocket (45). Thus, although we cannot state it in full confidence, the preference of OGG1 for C is probably not due to some disorientation of the catalytic machinery when the base other than C is placed opposite 8-oxoG. An indirect endorsement for this suggestion comes from steady-state kinetics of mouse OGG1 showing that the contribution of k cat into the overall opposite-base specificity is ϳ10-fold less than the contribution of K m (18). We also found little difference here in k 4 and k 5 values between different opposite bases ( Table 4).
The strongest discrimination of good versus poor substrates with different bases opposite 8-oxoG was found to occur at the second pre-excision equilibrium step, which we attribute to the insertion of the plugging residues into DNA (see above). This makes perfect sense from the structural point of view, because the specific bonds with the base opposite the lesion should indeed be formed at the plugging step. In similar experiments with Fpg we observed that the strongest discrimination also occurred at the second step, which, in that case, was attributed to the initial destabilization of the DNA at the site of the lesion (35). As Fpg and OGG1 belong to different structural classes and recognize 8-oxoG by different means (23,24), the differences in the key discriminatory step is not too surprising. More importantly, selection of the correct substrate in both enzymes follows a common kinetic theme, i.e. the step following the primary encounter with the lesion heavily favors the correct substrate and disfavors poor substrates. Glycosylases most likely locate their cognate lesions by fast sliding along DNA ("scanning"), and therefore full sampling of all bases would be kinetically very costly. A good strategy in this case would be to reject non-substrates at some early step in recognition, without having the base brought into the active site after all of the conformational changes required to bring the substrate into the active site. This, however, carries a risk of incorrectly rejecting a good substrate, because when the enzyme is just beginning to sample a base, sliding kinetically competes with entering the sampling branch of the kinetic scheme. Both OGG1 and Fpg seem to pull good substrates quickly into the sampling process, whereas poor substrates allow the enzyme to easily resume scanning the DNA for its cognate lesions.