Structural determinants for specific recognition by T4 endonuclease V.

DNA glycosylases catalyze the scission of the N-glycosyl bond linking either a damaged or mismatched base to the DNA sugar phosphate backbone. T4 endonuclease V is a glycosylase/apurinic (AP) lyase that is specific for UV light-induced cis-syn pyrimidine dimers. As a proposed transition state analog/inhibitor for glycosylases, a phosphoramidite derivative containing a pyrrolidine residue has been synthesized. The binding of endonuclease V to this duplex was analyzed by gel mobility shift assays and resulted in a single stable complex of reduced mobility and an apparent Kd of 17 nM. To assess the importance of the positive charge for specific binding, studies using other non-cleavable substrate analogs were performed. Wild type T4 endonuclease V shows an 8-fold decreased affinity for a tetrahydrofuran as compared with the pyrrolidine residue, demonstrating the significance of the positive charge for recognition. A 2-fold increase in binding affinity for a reduced AP site was observed. Similar assays using catalytically compromised mutants (E23Q and E23D) of endonuclease V demonstrate altered affinities for the pyrrolidine as well as tetrahydrofuran and reduced AP sites. This approach has provided insight into the structural mechanism by which specific lesions are targeted by the protein as well as the structural determinants of the DNA required for specific recognition by T4 endonuclease V.

DNA glycosylases catalyze the scission of the N-glycosyl bond linking either a damaged or mismatched base to the DNA sugar phosphate backbone. T4 endonuclease V is a glycosylase/apurinic (AP) lyase that is specific for UV light-induced cis-syn pyrimidine dimers. As a proposed transition state analog/inhibitor for glycosylases, a phosphoramidite derivative containing a pyrrolidine residue has been synthesized. The binding of endonuclease V to this duplex was analyzed by gel mobility shift assays and resulted in a single stable complex of reduced mobility and an apparent K d of 17 nM. To assess the importance of the positive charge for specific binding, studies using other non-cleavable substrate analogs were performed. Wild type T4 endonuclease V shows an 8-fold decreased affinity for a tetrahydrofuran as compared with the pyrrolidine residue, demonstrating the significance of the positive charge for recognition. A 2-fold increase in binding affinity for a reduced AP site was observed. Similar assays using catalytically compromised mutants (E23Q and E23D) of endonuclease V demonstrate altered affinities for the pyrrolidine as well as tetrahydrofuran and reduced AP sites. This approach has provided insight into the structural mechanism by which specific lesions are targeted by the protein as well as the structural determinants of the DNA required for specific recognition by T4 endonuclease V.
Base excision repair provides organisms a major line of defense against a multitude of base damage including UV-induced cyclobutane pyrimidine dimers, alkylation, and misincorporation (1). The initiating cascade of events in base excision repair includes specific site location, damage recognition, and excision of the improper or modified base by a DNA glycosylase. All DNA glycosylases catalyze the breakage of the N-C 1 Ј glycosyl bond linking the damaged base to the sugar phosphate backbone. A subset of these glycosylases has a concomitant apurinic (AP) 1 lyase activity, which catalyzes a ␤-elimination reaction leaving a 3Ј ␣,␤-unsaturated aldehyde and a 5Ј phosphate. This product is then further processed by an endonuclease to create a 3Ј-OH for polymerization and ligation.
T4 endonuclease V has served as the prototype for mechanistic studies of the molecular basis of recognition and catalysis for the glycosylase/AP lyase enzymes, including determination of the active site (2,3), the residues necessary for DNA binding (4), and the chemical basis of catalysis (5). T4 endonuclease V is a 16-kDa cyclobutane pyrimidine dimer-specific glycosylase with an associated AP lyase activity. It has been shown that T4 endonuclease V binds DNA through electrostatic interactions and then scans along the helix in a salt-dependent one-dimensional search until a pyrimidine dimer is encountered (6,7). The enzyme then cleaves the N-glycosyl bond between the base and the sugar on the 5Ј side of the dimer and subsequently cleaves the phosphodiester bond between the pyrimidines. The x-ray crystal structure of T4 endonuclease V has been solved and has revealed the presence of a glutamic acid residue (Glu-23) near the active site nucleophile (Thr-2), thus implicating this acidic residue in the reaction chemistry (8,9). Recently, building on the vast amount of information obtained on T4 endonuclease V, a unified catalytic mechanism for DNA glycosylases and glycosylase/AP lyases, has been proposed (10). In this model, the initial catalytic process proceeds by a nucleophilic attack at the sugar C 1 Ј of the damaged base ( Fig. 1, structure 1). It is proposed that the glycosylase/AP lyase enzymes use an amino group as the attacking nucleophile, resulting in an imino intermediate that can be trapped experimentally by reduction with NaBH 4 (Fig. 1, structure 3). Enzymes with only the glycosylase activity use a nucleophile from the medium such as an activated water molecule. Recently, the co-crystal structure of a catalytically inactive T4 endonuclease V mutant (E23Q) with dimer containing DNA has been solved (11). Interestingly, the crystal structure revealed a sharp kink in the DNA helix at the thymine dimer, and the adenine base opposite the 5Ј-thymine of the dimer is flipped out of the DNA duplex and into a pocket on the protein surface (11).
Although the biochemistry and structural information on T4 endonuclease V have led to the proposed catalytic mechanism and the identification of two active site groups (the N-terminal amino acid and Glu-23), there are still many gaps in our understanding of glycosylase mechanisms. One of the barriers to obtaining a detailed picture of the interactions of glycosylases with DNA is the relatively fleeting nature of the complexes that are formed. Thus, the production of stable long-lived complexes between DNA repair enzymes and DNA would facilitate an understanding of the interactions that occur prior to catalysis, including site-specific recognition and base flipping. In an effort to design molecules that bind DNA repair proteins in stable long-lived complexes suitable for structural analysis, Schä rer et al. (12) presented data on a pyrrolidine-based inhibitor for AlkA, an Escherichia coli DNA glycosylase. The pyrrolidine residue contains a positively charged nitrogen in place of the endocyclic oxygen mimicking a proposed transition state for glycosylases, where a positive charge is accumulated at the O 1 Ј/C 1 Ј (Fig. 1, structure 2). Recently, Schä rer et al. (12) have shown that AlkA binds very tightly to this duplex with an apparent K d of 16 pM and that this transition state analog also serves as a potent inhibitor for the AlkA-catalyzed reaction.
This study utilizes the pyrrolidine inhibitor to make comparative measurements of binding affinities for T4 endonuclease V on a series of related DNA structures to gain insight into the mechanism of specific site recognition. This approach may prove useful to 1) examine the catalytic distinction between glycosylases and glycosylase/AP lyases, 2) determine the structural features of the DNA that mimic the transition state for UV-specific glycosylases, 3) determine the molecular architecture of the reaction intermediates, and 4) determine the structural features of the protein required for differential recognition and catalysis.

EXPERIMENTAL PROCEDURES
T4 Endonuclease V-T4 endonuclease V (wild type) was purified from E. coli AB2480 (recA Ϫ , uvrA Ϫ ) cells transformed with a denV expression vector as described previously (13). E23Q and E23D endonuclease V mutants were created by site-directed mutagenesis and purified as described (14).
Oligonucleotide Substrates-DNA oligonucleotides containing a sitespecific pyrrolidine residue, reduced abasic site residue, or a propanediol residue were synthesized as described (15). DNA containing a site-specific tetrahydrofuran residue was a generous gift from Francis Johnson (SUNY, Stonybrook, NY). Complementary sequences were synthesized using standard procedures, and the deprotected oligonucleotides were electrophoretically purified on 20% denaturing polyacrylamide gels. A cis-syn cyclobutane thymine dimer-containing 10-base oligonucleotide was kindly provided by John-Stephen Taylor (Washington University, St. Louis). A 49-base oligonucleotide containing a sitespecific cyclobutane thymine dimer (underlined) with the sequence 5Ј-AGCTACCATGCCTGCACGAATTAAGCAATTCGTAATCATGGTC-ATAGCT-3Ј was constructed using the 10-mer as described previously (16). All oligonucleotide strands containing the base analogs, thymine dimer, and a control sequence were 32 P labeled on the 5Ј-end with T4 polynucleotide kinase (New England Biolabs) and annealed to their complementary strands to form the duplex DNA.
Binding Assays-The pyrrolidine, reduced abasic site, tetrahydrofuran, and propanediol-containing oligonucleotides were 5Ј-end labeled with [␥-32 P]ATP and annealed to a complementary strand with the indicated base opposite the modification ( Table I). Binding of T4 endonuclease V to this duplex was assayed using gel mobility shift analysis. The reactions contained 25 mM sodium phosphate (pH 6.8), 100 mM KCl, 5% glycerol, 100 g/ml BSA, 20 pM DNA duplex, and the indicated amounts of T4 endonuclease V in a total volume of 20 l. Following a 30-min incubation at 25°C, the free DNA and enzyme-bound DNA was separated through a 7.5% native polyacrylamide gel in 0.5 ϫ TBE (45 mM Tris borate, 1 mM EDTA) buffer for 2 h at 120 volts. Bands were visualized by autoradiography of the wet gels using Hyperfilm-MP x-ray film (Amersham).
The binding data were quantitated using a Molecular Dynamics PhosphorImager and ImageQuant software (Sunnyvale, CA). The data were plotted and fit to a simple hyperbolic curve function using Kaleidagraph (Synergy Software, Reading, PA) as described: where x ϭ the amount of free substrate, a ϭ maximum capacity of binding, and b ϭ K d . The K d is calculated from the regression analysis based on the best fit curve through the mean values of n ϭ 3 independent experiments. The fraction of active molecules of wild type endonuclease V was determined by active site titration analysis on dimer-containing DNA by sodium borohydride trapping the covalent intermediate as described previously (5). The 5Ј-end-labeled 49-mer, containing a cis-syn thymine dimer (CS-49), was annealed to the complementary strand and incubated (1 nM) with increasing amounts of T4 endonuclease V (0 -10 nM). The standard binding buffer was used with the addition of 100 mM NaBH 4 immediately following addition of T4 endonuclease V. Reactions were incubated at 25°C for 4 h, and the covalent enzyme-DNA complexes were separated on a 7.5% polyacrylamide gel. Bands were visualized and quantitated as described above.
Determination of Nonspecific Binding Affinity-Standard gel mobility shift binding assays were performed in the presence of increasing amounts of nontarget DNA. A control 25-base oligonucleotide containing a cytosine at position 13 was 5Ј-end labeled and annealed to its complementary strand. In a separate experiment, a pBKS plasmid (3 kilobases) was used as a nontarget DNA competitor. T4 endonuclease V (40 nM) and pyrrolidine containing duplex (100 pM) were incubated in the presence of various amounts of nonspecific DNA as indicated. Following 30 min at 25°C, the samples were loaded onto a 7.5% native polyacrylamide gel. The bound and free substrate bands were visualized and quantitated as described above.
Thymine Dimer-specific Nicking Activity and Inhibition of T4 Endonuclease V-The CS-49 duplex substrate (250 pM) was incubated with T4 endonuclease V in the standard reaction buffer (25 mM sodium phosphate (pH 6.8), 100 mM KCl, 100 g/ml BSA) in a total volume of 20 l. The potential inhibitors were added simultaneously with the substrate where indicated, prior to the addition of limiting enzyme (25-50 pM). Reactions were incubated at 25°C for 5 min and terminated by placing the reaction mixture on a dry ice-ethanol bath. An equal volume of loading buffer (95% (v/v) formamide, 20 mM EDTA, 0.02% (w/v) bromphenol blue, 0.02% (w/v) xylene cyanol) was added, and the samples were heated to 90°C for 5 min prior to loading on 15% denaturing polyacrylamide gels (8 M urea) in a 1 ϫ TBE buffer (90 mM Tris borate, 2 mM EDTA, pH 8.0). The DNAs were separated by electrophoresis for 3 h at 800 volts. Bands were visualized by autoradiography of the wet gels using Hyperfilm-MP x-ray film (Amersham). The appearance of the 20-base product band was quantitated using a Molecular Dynamics PhosphorImager and ImageQuant software (Sunnyvale, CA). The data were plotted using Kaleidagraph (Synergy Software, Reading, PA).

RESULTS
Experimental Rationale-To assess the relative binding affinities of T4 endonuclease V for specifically modified oligonucleotides representing a series of proposed transition state analogs or substrate analogs, gel mobility shift analysis was utilized. This analysis has provided a detailed picture of the apparent equilibrium binding affinities (K d ) and the architecture of the DNA necessary for optimal discrimination between nontarget and target sites.
As shown in Fig. 2, the DNAs chosen include a proposed transition state analog (pyrrolidine) and DNA containing abasic site analogs either incapable of sugar ring opening (tetrahydrofuran) or incapable of ring closure (reduced apurinic site). A structure completely lacking the sugar ring but retaining the phosphodiester backbone (propanediol) was also included in the study.
Determination of Active Molecules of Endonuclease V-The fraction of active molecules of wild type endonuclease V was determined by active site titration analysis on dimer-containing DNA. The 5Ј-end-labeled CS-49 was annealed to the complementary strand, and 1 nM was incubated with increasing amounts of endonuclease V (0 -10 nM) in the standard reaction buffer in the presence of NaBH 4 , a reducing agent previously shown to covalently trap endonuclease V on pyrimidine dimercontaining DNA (5). Following incubation at 25°C, the reactions were analyzed for the number of endonuclease V molecules covalently trapped on the DNA, indicating the number of active sites available (data not shown). Under stoichiometric conditions, 50% of the T4 endonuclease V molecules were active, and thus all reported concentration values reflect this activity. As it was not possible to determine the number of active sites for catalytically compromised mutants, the binding constants are reported assuming 100% active protein, and thus may be underestimating the true affinity of the mutants for these DNAs.
Wild Type Endonuclease V Binding to Substrate Analogs-Binding of T4 endonuclease V to the pyrrolidine-containing duplex was monitored by gel mobility shift analysis. The interaction of T4 endonuclease V with this DNA results in the formation of a single stable complex in a 7.5% native polyacrylamide gel. This complex is specific for the pyrrolidine-containing duplex DNA, since it was resistant to competition with 100ϫ excess nonspecific competitor, and no binding to the single-stranded pyrrolidine oligonucleotide was observed (data not shown). A representative autoradiogram of the binding assays and the corresponding binding curve is shown in Fig. 3. Titration of the duplex with increasing amounts of wild type T4 endonuclease V demonstrates a simple hyperbolic binding isotherm with an apparent K d of 17 nM (Table II). To assess the binding of T4 endonuclease V to this proposed transition state analog as compared with other structurally related analogs, similar assays were performed with wild type endonuclease V binding to DNA duplexes containing a site-specific tetrahydrofuran, reduced apurinic, or propanediol residue (Fig. 2, Table  I). As with the pyrrolidine-containing duplex, T4 endonuclease V forms a single stable complex on both DNAs containing the reduced AP site and the tetrahydrofuran; however, the affinities for the two are quite different (Table II). Binding to the tetrahydrofuran-containing DNA was 16-fold less than binding to the reduced AP DNA, which bound with a 2-fold higher affinity than the pyrrolidine DNA. The binding to the propanediol-containing duplex was less specific (K d ϭ 220 nM) and was seen as a less pronounced band on the gel mobility shift assay.
To determine the affinity of endonuclease V to these structures as compared with nontarget DNA, a nonspecific binding affinity was determined by competition gel mobility shift analysis (Fig. 4). Standard binding assays were performed as described above in the presence of increasing amounts of nonspecific competitor DNA. Either nontarget 25-base duplex DNA (0 -8 M, squares) or nondamaged plasmid DNA (0 -4 M, circles) were examined. Dixon plot analysis was performed for the determination of nonspecific binding affinities for both the linear 25-bp duplex and the plasmid DNAs (Fig. 4). Using the K d for T4 endonuclease V binding to the pyrrolidine-containing 25-bp DNA (determined in Fig. 3), the following equation (17) was utilized to determine the nonspecific binding affinity (K ns ): where S ϭ substrate DNA concentration and K d is the binding affinity for substrate DNA as determined by gel mobility shift assays. Using this analysis, the apparent equilibrium binding affinity of T4 endonuclease V for the 25-bp duplex and the plasmid DNA are 1.5 ϫ 10 Ϫ6 M and 1.2 ϫ 10 Ϫ6 M, respectively. Catalytically Compromised T4 Endonuclease V Mutants Binding to Substrate Analogs-Equilibrium binding studies were performed using gel mobility shift assays as described above. Two previously described catalytically compromised en- donuclease V mutants (E23Q and E23D) were examined for relative binding affinities to the series of substrate analogs. Both mutants are devoid of glycosylase activity, while E23D retains approximately 60% of wild type AP lyase activity (14). Representative autoradiographs of the binding assays are shown in Fig. 5, and the corresponding K d values are shown in Table II. E23Q binds with wild type affinity to the pyrrolidinecontaining DNA while E23D has a 7-fold decreased affinity for this duplex (Fig. 5, A and B). Both mutants bind with higher affinity to the reduced AP site as compared with the pyrrolidine and with decreased affinity to the tetrahydrofuran, exhibiting the same trend as wild type enzyme (Table II). Since these mutants do not exhibit any pyrimidine dimer glycosylase activity, it was possible to determine apparent K d values for the thymine dimer substrate DNA. As previously reported, both mutants qualitatively retain the ability to specifically bind dimer-containing DNA (14). However, this study shows that their affinities differ by 7-fold, with E23Q binding tighter than E23D. Both mutants bind to the dimer substrate with approximately the same affinity as they bind to the pyrrolidine DNA, suggesting that the presence of the thymine dimer does not affect the specificity of binding for these mutants. Interestingly, the gel mobility shift assays with E23Q demonstrate the appearance of a higher molecular weight complex on all the substrate analogs tested at enzyme concentrations well above the K d value (Fig. 5). These complexes have been shown previously to correspond to a molecular weight consistent with two molecules of mutant T4 endonuclease V bound to the DNA (14). These complexes also appear for E23D on the reduced AP substrate (Fig. 5D). The higher molecular weight complexes are not evident even at exceedingly high concentrations (1 M) of wild type endonuclease V on any of the DNAs investigated.
Determination of binding constants for the wild type enzyme on the dimer-containing substrate DNA are complicated due to the efficient catalysis that occurs on this substrate. Therefore, to demonstrate that endonuclease V shows no nicking activity on the substrate analogs, standard assays were carried out as described. Under these conditions, T4 endonuclease V shows no catalytic cleavage activity on any of the substrate analogs investigated in this study (data not shown).
Inhibition of Dimer-specific Glycosylase/AP Lyase Activity-To examine the ability of these analogs to inhibit the glycosidic bond cleavage at a cis-syn thymine dimer by T4 endonuclease V, the enzyme (50 pM) was incubated with CS-49 duplex in the presence of increasing amounts of pyrrolidine or reduced AP DNA. As shown in Fig. 6, both DNAs inhibit the dimer-specific nicking activity of T4 endonuclease V when present in molar excess over the enzyme. When the pyrrolidine DNA is in an approximate equimolar ratio to the dimer substrate, 50% inhibition is achieved. As expected the reduced AP DNA inhibits the reaction more effectively, achieving 50% at concentrations below that of the dimer substrate.

DNA Structural Determinants of Specific Recognition by T4
Endonuclease V-This study has demonstrated specific binding of T4 endonuclease V to a pyrrolidine-containing analog/inhibitor as well as other noncleavable abasic site analogs. This glycosylase/AP lyase shows relatively tight binding in the nanomolar range for these DNAs. In the studies using the pyrrolidine-containing DNAs, the presence of a positive charge enhanced the binding of T4 endonuclease V 8-fold as compared with a neutral sugar ring-closed structure (tetrahydrofuran), while the stable ring-opened form (reduced AP) enhanced the binding 16-fold. Thus, both the positive charge and the ringopened structure facilitate tighter T4 endonuclease V binding. As seen in Fig. 1, the proposed chemistry would predict that both of these structures resemble intermediates along the reaction pathway. However, it has yet to be determined whether the formation of the covalent intermediate occurs before or after ring opening of the sugar. As reported previously, AlkA, an E. coli DNA glycosylase, also bound to the pyrrolidine DNA with high affinity; however, unlike T4 endonuclease V, it showed a 3 ϫ 10 3 -fold enhancement of binding for the posi- substrate analogs The apparent equilibrium dissociation constants for T4 endonuclease V were determined as described in Figs. 3 and 5. All data represent the mean of three independent determinations. The T4 endonuclease V mutant binding isotherms did not fit a simple hyperbolic function since either the binding did not saturate at the highest endonuclease V concentration tested or some of the plots were sigmoidal in shape. Therefore, the K d values for the mutants were estimated based on the concentration of the mutant enzyme, which gave 50% binding of input DNA. ND, not determined.  4. Dixon plot analysis of nonspecific binding of T4 endonuclease V. Endonuclease V (40 nM) was incubated with the pyrrolidine DNA (100 pM) in the standard binding buffer (25 mM sodium phosphate (pH 6.8), 100 mM KCl, 5% glycerol, and 100 g/ml BSA) in the presence of increasing amounts of nonspecific competitor plasmid DNA (closed circles) or 25-bp DNA (closed squares) as indicated. Following a 30-min incubation at 25°C, the reactions were loaded onto a 7.5% native polyacrylamide gel, and the bound and free substrate bands were quantitated as described.
tively charged pyrrolidine as compared with the tetrahydrofuran-containing DNA (12). Other glycosylases and glycosylase/AP lyases have also been shown to bind specifically to the pyrrolidine DNA. Interestingly, of those examined, only interactions of AlkA and endonuclease III with the analog/inhibitor were highly specific for the presence of the positive charge (15) while 2,6-dihydroxy-5N-formamidopyrimidine DNA glycosylase, adenine DNA glycosylase, and alkyl-N-purine DNA glycosylase behaved similarly to T4 endonuclease V, exhibiting only 10-fold or less discrimination between the pyrrolidine and the tetrahydrofuran. It appears from those studied to date that there is no correlation between the enzyme's substrate specificity and the relative discrimination between a positively charged analog as compared with a neutral abasic site analog.
To determine the specificity of T4 endonuclease V, binding affinities for linear and circular duplex nontarget DNA were determined and found to be approximately 1 ϫ 10 Ϫ6 M. Consequently, T4 endonuclease V discriminates between undamaged DNA and a pyrrolidine-containing DNA by a factor of 100, while for the tetrahydrofuran or propanediol the discrimination decreases to only 10-and 5-fold, respectively. This high affinity for nontarget DNA is expected, since T4 endonuclease V uses a nontarget scanning mechanism for specific site location.
During the course of these studies, it was noted that the base opposite the pyrrolidine residue does not affect T4 endonuclease V affinity for binding (data not shown). Using the 25-base oligonucleotide containing the pyrrolidine, complementary strands containing a cytosine, adenine, thymine, or guanine at the position opposite the pyrrolidine were synthesized. No difference in equilibrium binding affinities was observed as measured by gel mobility shift assays. This was unexpected as the co-crystal structure has shown that the adenine opposite the 5Ј-thymine of the thymine dimer is flipped out of the DNA helix and into a "pocket" in the enzyme (11). In the case of uracil DNA glycosylase, the pocket is highly specific for uracil. Preliminary fluorescence experiments utilizing the pyrrolidine residue with a 2-aminopurine in the complementary strand suggest that the base opposite the pyrrolidine is flipped extrahelical. 2 The binding results suggest that the T4 endonuclease V protein "pocket" may not be very specific for the flipped 2 A. McCullough and R. S. Lloyd, unpublished observations. FIG. 5. Binding of catalytically compromised endonuclease V mutants to a proposed transition state analog and substrate analogs as compared with substrate. 20 pM 5Ј-end-labeled pyrrolidine-containing duplex was incubated with increasing amounts of mutant T4 endonuclease V in 25 mM sodium phosphate (pH 6.8), 100 mM KCl, 5% glycerol, and 100 g/ml BSA. Following a 30-min incubation at 25°C, the complexes were separated by electrophoresis through a 7.5% native polyacrylamide gel electrophoresed for 2 h at 120 volts. Representative autoradiograms for each mutant are shown. A, E23Q (0, 2, 10, 14, 18, 20, 40, 60, and 80 nM; lanes 1-9, respectively) binding to pyrrolidinecontaining DNA. B, E23D (0, 40, 60, 80, 100, 200, and 400 nM; lanes 1-7, respectively) binding to pyrrolidine-containing DNA. C, E23Q (0, 1, 5, 10, 20, 40, 60, 80, and 100 nM; lanes 1-9, respectively) binding to reduced AP-containing DNA. D, E23D (0, 40, 60, 80, 100, 200, 400, and 800 nM; lanes 1-8, respectively) binding to reduced AP-containing DNA.
FIG. 6. Inhibition of T4 endonuclease V thymine dimer-specific activity. CS-49 duplex substrate (250 pM) was incubated with T4 endonuclease V in the standard reaction buffer (25 mM sodium phosphate (pH 6.8), 100 mM KCl, 5% glycerol, and 100 g/ml BSA) in the presence of increasing amounts of pyrrolidine-containing DNA (closed circles) or reduced AP-containing DNA (closed squares) as indicated. Following a 5-min incubation at 25°C, the reactions were terminated and loaded onto a 15% denaturing polyacrylamide gel. The appearance of the 20-base product was quantitated as described.
out base as both purines and pyrimidines may be accommodated. This is supported by the co-crystal structure in which the flipped base is arranged between two amino acids, not forming any hydrogen bonds with protein residues, and unpublished data cited by Vassylyev et al. (11), demonstrating that the base opposite the 5Ј-thymine does not affect T4 endonuclease V glycosylase activity (11).
Examination of the Structural Features of T4 Endonuclease V Required for Differential Recognition and Catalysis-The x-ray crystal structure of T4 endonuclease V revealed the presence of Glu-23 near the active site (8,9,11), and site-directed mutagenesis of this residue has demonstrated its involvement in the catalytic mechanism of the enzyme (3,14). E23Q demonstrates an increased affinity for both the reduced AP and the tetrahydrofuran as compared with the wild type endonuclease V but has the same affinity for the pyrrolidine DNA. In addition, E23Q retains relatively tight binding to a cis-syn thymine dimer substrate. Thus, the presence of the carboxyl group, though critical for catalysis, is not necessary for specific binding of the enzyme. In fact, the presence of a neutral residue at this position actually increases the DNA binding affinity of the enzyme 3.5ϫ for the tetrahydrofuran and 4.5ϫ for the reduced AP. The E23D mutant exhibited the same trend in binding affinities as did the E23Q mutant; however, the values were much higher indicating a lower affinity for all the DNAs investigated. These data are consistent with the E23D mutant being devoid of any glycosylase activity and having a diminished AP lyase activity despite it being a relatively conservative mutation. Thus, the position of a side chain in the active site is important for proper damage recognition by the enzyme, most probably due to a structural fit that may stabilize the complex. The decrease in specific binding by E23D may contribute to its diminished AP lyase activity previously reported (14). As with the E23Q, E23D showed the same affinity for the dimer-containing DNA as the pyrrolidine DNA; however, it bound with a 3-fold higher affinity to the reduced AP DNA and with only slightly less affinity to the tetrahydrofuran as compared with the pyrrolidine. Thus, for both mutants, the presence of the positive charge makes only a slight difference in the binding affinity (1.5-2-fold).
Use of Transition State and Substrate Analogs for Structural and Mechanistic Studies on DNA Glycosylase/AP Lyases-Base excision repair is initiated by a complex cascade of events leading to damage recognition and catalysis. These events include target site location, specific binding, active site positioning (which may involve a conformational change in the enzyme and base flipping), and catalytic chemistry. This study has demonstrated that stable enzyme-DNA complexes can be formed between a DNA glycosylase/AP lyase and a proposed transition state analog as well as other structurally related substrate analogs. These DNAs, as well as the catalytically compromised mutants, may provide a detailed examination of the steps preceding catalysis by allowing numerous structural and mechanistic studies on the stable enzyme-DNA complex, including NMR and x-ray crystallography. The structural characteristics of the analogs investigated in this study have revealed potential structures for a series of second generation transition state analogs, such as a positively charged sugar ring-opened structure that may form an even more stable complex with T4 endonuclease V. This approach will not only make structural studies more feasible but will provide a means of dissecting the pre-catalysis mechanism for T4 endonuclease V as well as other glycosylases (this study and Ref. 15).