Adenine release is fast in MutY-catalyzed hydrolysis of G:A and 8-Oxo-G:A DNA mismatches.

MutY, a DNA repair enzyme, is unusual in that it binds exceedingly tightly to its products after the chemical steps of catalysis. Until now it was not known whether the product being released in the rate-limiting step was DNA, adenine, or both. MutY hydrolyzes adenine from 8-oxo-G:A (OG:A) base pair mismatches as the first step in the base excision repair pathway, as well as from G:A mismatches. The products are adenine and DNA containing an apurinic (AP) site. Tight product binding may have a physiological role in preventing further damage at the OG:AP site. We developed a rate assay using [8-14C]adenine in OG:A or G:A mismatches that distinguishes between adenine hydrolysis and adenine release. [8-14C]Adenine was released quickly from the MutY.AP-DNA.[8-14C]adenine complex, with a rate constant greater than 5 min-1. This was much faster than the rate-limiting step, at 0.006-0.015 min-1. Gel retardation experiments showed that AP-DNA release was very slow, consistent with it being the rate-limiting step. Thus, the kinetic mechanism involves fast adenine release after hydrolysis followed by rate-limiting AP-DNA release. Adenine appears to be buried deep in the protein.DNA interface, but there is enough flexibility or open space for it to dissociate from the MutY.APDNA.adenine complex. These results have implications for the catalytic mechanism of MutY.

Unlike "normal" enzymes, MutY releases its products exceedingly slowly after the chemical steps of catalysis are complete. MutY is a DNA repair enzyme that recognizes 8-oxo-G:A (OG:A) 1 base pair mismatches in the first step of the base excision repair pathway (1, 2) ( Fig. 1). OG residues are one of the most common forms of oxidative DNA damage (3). If OGcontaining DNA is replicated, A is incorporated with a 200-fold preference over C. If this OG:A mismatch is not repaired, a further round of replication results in a G3 T mutation (4). MutY hydrolyzes the adenine base from OG:A, forming an apurinic site (AP), OG:AP. OG:AP sites are susceptible to spon-taneous strand breakage and are substrates for MutM, a DNA glycosylase/␤-lyase (5) that excises OG from OG:C mismatches. With OG:AP, MutM causes a strand break opposite AP, removing all genetic information from that site. Tight binding of MutY to (OG:AP)-DNA may help protect it from strand breakage or inappropriate enzymatic reactions.

MutY reactions display biphasic kinetics when [MutY] Ͻ [DNA]
, with a pre-steady state burst followed by a slow steady state rate (6). The pre-steady state burst reflects adenine hydrolysis and has a rate constant of 15 min Ϫ1 with OG:A mismatches and 2 min Ϫ1 with G:A (6 -10). The steady state rate constant is much lower, 0.003 min Ϫ1 with OG:A and 0.02 min Ϫ1 for G:A, which has been attributed to rate-limiting AP-DNA release (9). The steady state rates parallel the higher affinity of MutY for OG:AP sites than for G:AP sites, with dissociation constants (K d ) of 50 -120 pM and 2-21 nM, respectively (11)(12)(13). It was not known whether this reflects slow adenine release followed by rapid AP-DNA release, or vice versa, or coordinated release of both. The x-ray crystal structure of MutY⅐adenine reveals a deep binding pocket, where it appears that adenine release would be blocked in a MutY⅐AP-DNA⅐adenine complex (14).
Tight product binding also has implications for catalysis. Glycosylase reactions proceed through oxocarbenium ion(-like) transition states or intermediates (1) with a positive charge on the sugar ring (15,16), and they are often susceptible to inhibition by imino sugars such as 2. However, (OG:2)-DNA binding to MutY is no better than (OG:AP)-DNA, whereas an adenine ring mimic in (OG:3)-DNA improved binding by more than 65-fold (17,18). This raises the possibility of a catalytic mechanism similar to purine nucleoside phosphorylase (PNP), which is believed to bind tightly with the leaving group and nucleophile but has little interaction with the ribose ring (19). Similarly, MutY could form tight interactions with adenine during catalysis and/or in the product complex. An activity assay was developed to detect free [8-14 C]adenine, allowing us to distinguish between the kinetics of hydrolysis and adenine release.

EXPERIMENTAL PROCEDURES
MutY-MutY was prepared (7) and the active site concentration was determined (6) using an overexpressing Escherichia coli strain supplied by Prof. Sheila David (University of Utah). Protein concentration was determined from A 280 , using ⑀ 280 ϭ 77,400 M Ϫ1 cm Ϫ1 as determined by the method of Edelhoch (20). * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Adenine Hydrolysis/Release Assay-All reactions were at 37°C, and chromatography or PAGE separations were at 6°C. Substrate DNA (200 nM, 1000 cpm 14 C, with 7.5 mM KCl) was equilibrated with reaction buffer (20 mM Tris⅐HCl, pH 7.6, 12.5 mM KCl, 10 mM EDTA, 0.1 mg/ml bovine serum albumin) in 45 l. MutY (110 nM) was added to start the reaction. For acid-quenched reactions, 5 l of 3 M HCl was added, and the acidified reaction was placed in liquid nitrogen. It was then thawed and neutralized with 3 M KOH before chromatography.
For unquenched reactions, a 30 -50 l aliquot was quickly drawn into an ice-cold 50-l glass syringe and immediately injected onto a G-25 Sephadex Superfine (Amersham Biosciences) gel filtration column (16 ϫ 490 mm, 9.9 ml) with reaction buffer as eluant and a 2.5 ml/min flow rate. Fractions (2.5 ml) were collected, mixed with 16 ml of scintillation fluid (Liquiscint, National Diagnostics), and 14 where [P] t is product concentration at time ϭ t, k ϩ2 is the burst phase rate constant, and k ss is the steady state rate constant, equal to k ϩ3 [MutY], where k ϩ3 is the rate of product release ( Fig. 3

) (6).
Chromatographic Resolution-The adenine release rate calculation was based in part on the time needed for high molecular weight species (DNA and protein) to be separated from adenine on the gel filtration column. Chromatographic resolution, R s , is given by Equation 3, where ⌬t r is the difference in retention time between two peaks, and w av is the average width of the two peaks (23). Two peaks are fully resolved when R s Ն 1.5, with Ͻ0.1% overlap.
The intention was to separate adenine from DNA based on size, but it adsorbed on the gel resin, leading to an elution volume of 19 ml for a 9.9-ml bed volume (Fig. 4) and better than expected resolution. A 50-l test injection containing 2.5 g of denatured herring sperm DNA and 3.5 nmol of adenine was monitored by A 254 and gave R s ϭ 3.8 (Fig. 4a). The UV chromatogram matched the peak position and shape of 14 C elution from an acid-quenched reaction with 15-s fractions. The rate of peak travel and widening was simulated numerically based on that chromatogram. These values were used to calculate peak travel, peak widths, and R s within the gel filtration column at increasing times throughout separation. R s was also measured for test runs with injection volumes from 50 to 2000 l. The numerical simulations agreed well with experimental values of R s , lending support for their accuracy.
DNA Product Release Assay-Pulse-chase experiments were used to detect transient dissociation of MutY⅐AP-DNA complexes during gel filtration. 33 A-DNAs were reacted with MutY and then chased with unlabeled AP-DNAs followed by gel retardation PAGE (24). Reactions were as described above, with 200 nM (OG: 33 A)-DNA incubated with 250 nM MutY for 30 s. A 20 -50 l aliquot was then added to 400 nM (OG:AP)-DNA. The mixture was chromatographed as described above but on a 5-ml HiTrap Desalting column (G-25 Sephadex Superfine, Amersham Biosciences) to reduce sample dilution. Loading buffer (20 l, 200 mM Tris, pH 7.6, 200 mM KCl, 100 mM EDTA, 0.25% bromphenol blue) was added to the fraction with the highest [DNA] (the 10th 200-l fraction), and 50 l was applied to a 145 ϫ 165 ϫ 1.5-mm nondenaturing 10% PAGE run in 0.5ϫ TBE buffer. The polyacrylamide gel was run at 300 V until all the dye had entered the gel (6 -8 min) and then was reduced to 70 V. Bands were visualized and quantitated by storage phosphor-autoradiography.
Reactions with (G: 33 A) substrates were the same as (OG: 33 A) but with 3-min incubations, and the samples were electrophoresed without loading buffer because (G:AP)-DNA release was accelerated by some component of the loading buffer, presumably bromphenol blue.
In a second set of experiments, unlabeled substrates were used with a (OG: 33 AP)-DNA chase. The experiments were as described above but with 150 nM MutY and 50 nM (OG: 33 AP)-DNA.

RESULTS
Adenine Hydrolysis/Release Assays-Previous MutY assays took advantage of facile ␤-elimination at the AP site to cleave the DNA backbone at 90°C in 0.2 M NaOH. This allowed separation and quantitation of the resulting fragments by PAGE. An assay was developed in this study to report directly on the adenine product by using gel filtration chromatography to separate [8-14 C]adenine from 14 A-DNA substrate (Fig. 4). There was no detectable difference between the acidquenched and unquenched reactions (Fig. 5, Table I). That is, the free [8-14 C]adenine concentration in solution was equal to total [8-14 C]adenine, indicating that adenine was released quickly from the MutY active site after the chemical steps of catalysis.
The only source of 14 C in these experiments was [8-14 C]A residues in the substrate, which were converted to [8-14 C]adenine upon hydrolysis. No detectable 33 P co-eluted with adenine ( Fig. 4a), demonstrating that the 14 C detected with 14 A-DNA substrates was truly from free [8-14 C]adenine and not a chromatographic artifact. The small 33 P peak at t ϭ 3.5 min corresponded to free [ 33 P]phosphate, which co-eluted with other nonadsorbed low molecular weight species.
For the (OG: 14 A)-DNA reactions, the burst phase was essentially complete in the first time points, and thus the fitted values of k ϩ2 , 9 and 18 min Ϫ1 (Table I,  MutY⅐AP-DNA Release-With adenine release being fast, an important question became whether it is released directly from the MutY⅐AP-DNA⅐adenine complex or whether transient release and rebinding of AP-DNA allows adenine release. Pulsechase experiments and gel retardation PAGE (24) were used to examine the high molecular weight peak from gel filtration. Gel retardation exploits the change in DNA mobility on nondenaturing PAGE when complexed with protein and has been used with MutY to probe DNA binding (9,(11)(12)(13).
(OG: 33 A)-or (G: 33 A)-DNA was reacted with MutY, and then excess unlabeled (OG:AP)-DNA was added immediately before gel filtration chromatography. If 33 AP-DNA dissociated from MutY during chromatography, unlabeled (OG:AP)-DNA would out-compete it to rebind MutY, resulting in displacement of 33 AP-DNA. The small proportion of free 33 AP-DNA (Fig. 6a, Table II) demonstrated that most of the product DNA did not dissociate even once from MutY during the 9 min between the start of the reaction and separation on gel retardation PAGE. The fraction of free 33 AP-DNA was in excellent agreement with that calculated from k ϩ3 and was far from the equilibrium value expected if there was dissociation/rebinding of AP-DNA. k ϩ2 is the rate of the chemical step, adenine hydrolysis. Previously, k ϩ3 was shown to be slow, but whether the upper or lower pathway was followed was not known. Our results show that the lower kinetic pathway is followed, with fast adenine release and slow (OG:AP)-DNA release.
The complementary experiment with unlabeled substrate DNA and an (OG: 33 AP)-DNA chase gave the same result, with only a small fraction of labeled 33 P appearing in the MutY⅐AP-DNA complex (Fig. 6b, Table II).
Adenine Release Rate-The adenine release rate was estimated from the chromatographic data. The acid-quenched and unquenched reactions were essentially identical for both (OG: A)-and (G:A)-DNA, arguing that adenine release was complete on the time scale of the assay. For short reaction times, the time on the gel filtration column was longer than the reaction time, and adenine release could have occurred during chromatography. If this were the case, the high molecular weight peak would be normal, but the adenine peak would have a long leading edge due to adenine being transported partway down the column in the MutY⅐AP-DNA⅐adenine complex before release. Also, the peak would be smaller for the unquenched reaction. There was no detectable 14 C eluted between the high molecular weight peak and adenine (Fig. 4b), and [[8-14 C]adenine] free was the same for acid-quenched and unquenched reactions (Fig. 5). Thus, most or all adenine release occurred before chromatographic separation.
The shortest elapsed time between the start of reaction and separation on the gel filtration column was ϳ23 s, comprising 7 s reaction time, 5-7 s to withdraw an aliquot into an ice-cold syringe and inject it, and 7.8 s for the sample to reach the top of the resin, based on the internal volume of the tubing and fittings. Once on the resin, numerical simulations showed that 1.0 s was required to separate [8-14 C]adenine from 14 A-DNA and MutY⅐AP-DNA⅐[8-14 C]adenine. Assuming that 23 s represented at least 3 half-lives of adenine release (87.5% released), then t1 ⁄2 Յ 8 s and k ϩ32 Ն 0.09 s Ϫ1 or 5 min Ϫ1 (Fig. 3). DISCUSSION Reaction Rates and Adenine Release-The rate constants k ϩ2 and k ϩ3 determined in this study were very similar to those  reported previously, where k ϩ2 ϭ 14 -15 min Ϫ1 (OG:A) and 1.3-2.5 min Ϫ1 (G:A), and k ϩ3 ϭ 0.002-0.005 min Ϫ1 (OG:A) and 0.02-0.03 min Ϫ1 (G:A) (6 -10). There was no sequence similarity between the DNAs used here and in previous studies, showing that MutY is generally insensitive to the sequence context of the mismatch.
Acid-quenched and unquenched reactions were identical within experimental error for both (OG: 14 A)-and (G: 14 A)-DNA, indicating that adenine release was complete on the time scale of these experiments. If adenine release could only occur after rate-limiting AP-DNA release, there would have been a lag in its appearance followed by release at the steady state rate (Fig.  5, dotted line, insets). The estimated lower limit for the adenine release rate constant, k ϩ32 Ն 5 min Ϫ1 , was faster than the chemical step, k ϩ2 Ϸ 2 min Ϫ1 , for (G: 14 A)-DNA. For (OG:A)-DNA k ϩ2 Ϸ 15 min Ϫ1 , and so it was not possible to determine which was faster in this case, the chemical step or adenine release. Adenine release was 1000-and 300-fold faster than k ϩ3 for (OG:AP)-or (G:AP)-DNA release, respectively.
Consequences of Fast Adenine Release-Fast adenine release has several important consequences for understanding MutY. First, in the kinetic mechanism (Fig. 3), the lower pathway predominates with k ϩ32 Ն 5 min Ϫ1 , and therefore, presumably, k ϩ34 is equal to the experimental values of k ϩ3 , i.e. 0.015 min Ϫ1 for G:A and 0.006 min Ϫ1 for OG:A.
Second, contrary to appearances in the MutY⅐adenine cocrystal structure (Fig. 7), adenine is not trapped in a deep pocket by bound AP-DNA. This implies either structural flexibility in the complex and/or that a solvent-accessible route remains open in the MutY⅐AP-DNA⅐adenine complex.
Third, adenine does not contribute to the observed affinity of MutY for AP-DNA. The previously determined values of K d for (OG:AP)-DNA (50 to 120 pM) and (G:AP)-DNA (2 to 21 nM) were measured under conditions in which adenine would have been released and therefore reflect the equilibrium dissociation constants for AP-DNA only (11)(12)(13).
The fact that K d reflects AP-DNA binding means that the presence of an oxocarbenium ion mimic in the MutY inhibitor (OG:2)-DNA, K d ϭ 65 pM, does not contribute to binding. Imino sugars such as 2 and deoxynojirimycin are oxocarbenium ion (1) mimics that are glycosylase inhibitors (25,26). The lack of effect with MutY could reflect a catalytic mechanism similar to PNP. PNP catalysis is believed to involve tight binding of the leaving group and phosphate, the nucleophile, but relatively little interaction with the ribose ring (19). PNP can transiently bind the hypoxanthine product very tightly, K d ϭ 1 pM (27). Iminoribitol (4), an analog of 2, is a poor inhibitor of M. tuberculosis PNP, K d ϭ 2.9 mM, whereas immucillin-H (5), containing an analog of the leaving group, is potent, with K d ϭ 28 pM (28,29). The fact that (OG:3)-DNA binds to MutY Ͼ 65-fold more tightly than (OG:2)-DNA indicates a role for adenine ring binding. Two other nucleic acid N-glycosylases, ricin (30) and uracil DNA glycosylase (31), show improved binding of oxocarbenium ion-mimic inhibitors in the presence of their base products, adenine or uracil.
Rapid adenine release from MutY⅐AP-DNA does not, of course, preclude tight binding in the transition state. Tight transition state binding could be catalytically important, with those interactions lost in the product complex. Also, rapid adenine release does not rule out significant affinity of MutY for adenine. With a rate constant of Ն5 min Ϫ1 for adenine release, a K d as low as 1 nM is possible, 2 assuming diffusion rate-limited association (10 8 M Ϫ1 s Ϫ1 ). Transition state analysis of MutY will elucidate adenine binding at the transition state.
Conclusions-A new activity assay for MutY-based separation of free [8-14 C]adenine from 14 C-labeled DNA and MutY⅐AP-DNA⅐ [8-14 C]adenine was used to examine adenine release. Adenine release was fast for both (OG:A)-and (G:A)-DNA substrates and was faster than hydrolysis in the case of G:A. Gel retardation PAGE demonstrated that AP-DNA release was much slower, consistent with its being the ratelimiting step. The high affinity of MutY for (OG:AP)-and (G:   From Fig. 6, lanes 2, 4, 6, 8. c Calculated from k ϩ3 for (G:AP)-DNA and (OG:AP)-DNA (Table I)