Evidence for an imino intermediate in the DNA polymerase beta deoxyribose phosphate excision reaction.

A recent study demonstrated that rat DNA polymerase β (β-pol) releases 5′-deoxyribose phosphate (dRP) termini from preincised apurinic/apyrimidinic DNA, a substrate generated during certain types of base excision repair. This catalytic activity resides within the amino-terminal, 8-kDa domain of β-pol and occurs via β-elimination as opposed to hydrolysis (Matsumoto, Y., and Kim, K. (1995) Science 269, 699-702). The latter finding suggested that the dRP excision reaction might proceed via an imine intermediate. In order to test this hypothesis, we attempted to trap β-pol on preincised apurinic/apyrimidinic DNA using NaBH4 as the reducing agent. Both 8-kDa domain-DNA and intact β-pol-DNA complexes were detected and identified by autoradiography coupled to immunoblotting. Our results indicate that the chemical mechanism of the β-pol dRpase reaction does proceed through an imine enzyme-DNA intermediate and that the active site residue responsible for dRP release must therefore contain a primary amine.

Vertebrate DNA polymerase ␤ (␤-pol), 1 a constitutively-expressed monomeric protein ranging from 39 to 45 kDa, has been implicated in DNA repair. Specifically, in vitro experiments indicate that ␤-pol can perform two of the five reactions involved in base excision repair, excision of a 5Ј-terminal dRP from a preincised AP site (1) and DNA synthesis to fill the single-nucleotide gap (2)(3)(4)(5). The overall pathway for base excision repair that has been initiated by simple N-glycosylases (glycosylases lacking a concomitant AP lyase activity) is believed to proceed according to the following scheme. (i) A damage-specific DNA N-glycosylase creates an AP site by cleaving an N-C glycosyl bond, thereby releasing a modified base; (ii) a class II AP endonuclease incises the DNA 5Ј to the AP site; (iii) ␤-pol or another dRpase excises the 5Ј dRP terminus to leave a single-nucleotide gap; (iv) ␤-pol or another polymerase fills in the single-nucleotide gap; and (v) DNA ligase seals the gap (6). 5Ј-Terminal dRP might eventually undergo an uncatalyzed ␤-elimination reaction to generate a single-nucleotide gap (half-life Ϸ 2 h) (7), but this reaction is probably too slow to be relied upon in vivo. Coordination of steps (iii) and (iv) by ␤-pol presumably would lead to more efficient DNA repair. Thus, reaction (iii) assures an adequate substrate for ␤-pol, an enzyme without an intrinsic 5Ј-exonuclease activity.

Materials-[␣-
Preparation of Preincised AP DNA Substrate-Uracil-containing 49mer oligonucleotide (5Ј-AGCTACCATGCCTGCACGAAUTAAGCAAT-TCGTAATCATGGTCATAGCT-3Ј) was 32 P-labeled on its 3Ј end with terminal deoxynucleotidyltransferase and then annealed to its complement. Free [␣-32 P]ddATP and transferase were separated from the duplexed DNA by passage of the mixture over a Nensorb-20 column as per the supplier's instructions. Column fractions containing the substrate DNA were evaporated to dryness in a Savant SC110 SpeedVac and resuspended in UDG reaction buffer (70 mM Hepes, pH 8.0, 1 mM EDTA, 1 mM ␤-ME). Recovery of DNA from the column was assessed by chromatography of pre-and post-column samples of the kinase mixture on Whatman DE-81 paper with 0.3 M NH 4 COOH as the solvent. The chromatograph was scanned on a PhosphorImager 450 machine (Molecular Dynamics), and the counts remaining at the origin (i.e. radioactivity associated with labeled substrate) were quantitated using Im-ageQuant software (Molecular Dynamics). Substrate was incubated for 20 min at 37°C with excesses of UDG (0.25 units/pmol uracil) and APE (10-fold molar excess) to sequentially remove uracil from the DNA backbone and then incise the DNA backbone 5Ј to the resultant AP site.
Trapping of ␤-pol-DNA Covalent Complexes by NaBH 4 -Due to the inherent instability of preincised AP DNA, trapping experiments were started immediately following preparation of the substrate. Therefore, when called for, pretreatments of ␤-pol were carried out during the same 20-min interval in which the 49-mer duplex was being incubated with UDG and APE. Pretreatments included: (i) inactivation of intact ␤-pol or 8-kDa domain by heating at 90°C for 20 min, (ii) exposure of the 8-kDa domain to 150 mM NaBH 4 for 20 min on ice, and (iii) preincubation of the 8-kDa domain with pre-or postimmune antiserum against a 32-amino acid peptide for 20 min on ice. For (ii), a highly concentrated stock of 8-kDa domain was used so that the enzyme plus NaBH 4 could be diluted 100X into enzyme dilution buffer (20 mM Hepes, pH 7.5, 50 mM KCl, 5 mM ␤-ME, 100 g/ml acetylated BSA) before addition into the trapping reaction ([residual NaBH 4 from pretreatment] f ϭ 1.5 mM). Dilutions of untreated ␤-pol and 8-kDa domain were prepared using the same dilution buffer. Trapping reactions were initiated by the almost simultaneous addition of varying amounts of either intact or 8-kDa ␤-pol enzyme and either NaBH 4 or NaCl into a reaction mixture containing 20 mM Hepes, pH 7.5, 50 mM KCl, 5 mM ␤-ME, 10 mM MgCl 2 , preincised AP substrate (1.5 ng ϭ 95 fmol of damaged strand DNA), and 125 ng of poly(dI-dC) (80-fold excess (w/w) over damaged strand DNA). In one inhibition experiment, 32-amino acid peptide was present in the reaction mixture at a 4000-fold (1.1 g ϭ 370 pmol) or a 40,000-fold (11 g ϭ 3.7 nmol) molar excess over substrate. In another experiment, trapping of the 32-amino acid peptide was attempted in the absence of any intact ␤-pol or 8-kDa domain. All reactions were allowed to proceed for 60 min at 37°C before they were terminated by the addition of an equal volume of formamide loading buffer (95% (v/v) formamide, 20 mM EDTA, 0.02% (w/v) bromphenol blue, 0.02% (w/v) xylene cyanol). The samples were heated for 3 min at 80°C, loaded onto 10% polyacrylamide gels containing 8 M urea, and subjected to electrophoresis at 800 V for 3 h. Alternatively, the reactions were terminated by the addition of an equal volume of SDS-PAGE loading buffer (125 mM Tris-HCl, pH 6.8, 4% (w/v) SDS, 10% (v/v) ␤-ME, 20% glycerol (v/v), 0.01% (w/v) bromphenol blue). The samples were then heated for 3 min at 95-100°C, loaded onto either 10% (intact ␤-pol samples) or 15% (8-kDa domain samples) SDS-PAGE gels containing 10% glycerol, and subjected to electrophoresis at 175 V for 5 h.
Detection of ␤-pol-DNA Covalent Complexes-Labeled substrate and substrate-␤-pol complexes were visualized by autoradiography of the wet gels with Hyperfilm-MP film (Amersham Corp.) at Ϫ70°C, typically for several hours with two DuPont Quanta III intensifying screens enclosed in the cassette. Following development of the films, autoradiographs were overlaid on the gels, and, for SDS-PAGE gels, the positions of the prestained molecular size markers were recorded. In order to obtain quantitative results, the wet gels also were scanned on a Phos-phorImager 450 machine and the data analyzed using ImageQuant software. Whether the ␤-pol protein was intact, truncated in the form of the 8-kDa domain, or complexed to substrate, it was detected by immunoblotting of the SDS-PAGE gels. Specifically, proteins were transferred onto a nitrocellulose membrane in a Transblot apparatus for 4 h at 290 mA. The membrane was then blocked overnight with 5% milk in Tris-buffered saline before being probed with affinity-purified anti-␤pol (1:10,000 dilution). Goat anti-rabbit IgG/horseradish peroxidase conjugate (1:10,000 dilution) served as the secondary antibody and was detected with the ECL chemiluminescence system (DuPont). The nitrocellulose membrane was stained with Coomassie Blue, and the positions of the prestained molecular size markers and the uncomplexed ␤-pol proteins were noted.

RESULTS AND DISCUSSION
Experimental Rationale-As outlined in the Introduction, the dRpase reaction of ␤-pol is hypothesized to proceed via an imine intermediate. Accordingly, once the ␣or the ⑀-amino group of the active site residue has engaged in a nucleophilic attack on the sugar C-1Ј of an AP nucleoside, it should be possible to trap the ␤-pol-DNA intermediate by reduction with NaBH 4 . The covalent nature of the resultant complex can be tested by subjecting it to denaturing polyurea gel electrophoresis or SDS-PAGE. Although ␤-pol can release dRP from both single-stranded and double-stranded substrate, duplexed substrate was chosen as a matter of convenience; human APE incises double-stranded DNA more efficiently than singlestranded DNA. Magnesium was included in the reaction buffer to ensure that the 8-kDa domain would be able to bind to double-stranded DNA with a gap of only one nucleotide (24,1). NaBH 4 -mediated Trapping of the Intact ␤-pol-DNA and 8-kDa Domain-DNA Covalent Complexes-Upon exposure to NaBH 4 but not NaCl, ␤-pol (M r ϭ 39,000) was expected to be bound covalently to the 3Ј-labeled 28-mer (M r ϭ 9,000) produced by sequential UDG and APE treatments of 3Ј-labeled 49-mer containing uracil at position 21. Indeed, polyurea gel electrophoresis revealed a slow-migrating band (Fig. 1A, lane  4), presumably the covalent ␤-pol-DNA complex. In order to determine the molecular mass of the presumed intact ␤-pol-DNA complex and to identify it via immunoblotting, a more extensive version of the same experiment was carried out using SDS-PAGE. Fig. 1 (B and C) shows the autoradiograph and the corresponding immunoblot from SDS-PAGE trapping experiments on intact ␤-pol, respectively. As predicted, the most prominent product band on the autoradiograph migrated with an apparent molecular mass of 48 kDa (39,000 ϩ 9,000 ϭ 48,000). This complex, highlighted by asterisks, was easily identified by its immunoreactive and radioactive properties. Minor bands at 62 and 67 kDa on the autoradiograph also had counterparts on the immunoblot. Since SDS-PAGE does not necessarily denature DNA, these bands probably corresponded to complexes of (␤-pol ϩ 28-mer ϩ 49-mer) (calculated M r ϭ 64,000) and (␤-pol ϩ 28-mer ϩ 49-mer ϩ 20-mer) (calculated M r ϭ 71,000). They would not have been observed in Fig. 1A because polyurea electrophoresis completely denatures DNA. Unlabeled 20-mer would have been produced along with the labeled 28-mer during the APE preincision reaction. Secondary enzyme-DNA complexes and ␤-pol breakdown products were not observed when less ␤-pol was introduced into the reactions (data not shown), and heat inactivation of ␤-pol completely eliminated the formation of covalent complexes (lane 5). Finally, there was no evidence to indicate that any of the protein-DNA complexes were derived in whole or in part from APE (M r ϭ 30,000), UDG (M r ϭ 26,000), or contaminating bacterial dRpases (RecJ M r ϭ 60,000, Fpg M r ϭ 30,000).
Analogous to intact ␤-pol, 8-kDa domain was expected to be bound covalently to the 3Ј-labeled 28-mer (M r ϭ 9,000) by NaBH 4 reduction of an imine intermediate. A single product did survive polyurea gel electrophoresis (Fig. 2A, lane 4) and possessed a mobility in between that of the 28-mer substrate and the intact ␤-pol product seen in Fig. 1A (Figs. 1A and 2A were derived from the same gel). Fig. 2 (B and C) shows the autoradiograph and the corresponding immunoblot from trapping experiments on the 8-kDa domain of ␤-pol, respectively. The primary product band on the autoradiograph migrated with an apparent molecular mass of 17 kDa (8,000 ϩ 9,000 ϭ 17,000). This product, highlighted by asterisks, was both immunoreactive and radioactive. Again, a protein doublet appeared on the autoradiograph, but this time it accounted for 50% of the complexed protein. The protein pair was not detected on the immunoblot shown in panel C but was seen on other occasions. With observed molecular masses of 30 and 37 kDa, the secondary bands probably corresponded to complexes of (8-kDa domain ϩ 28-mer ϩ 49-mer) (calculated M r ϭ 33,000) and (8-kDa domain ϩ 28-mer ϩ 49-mer ϩ 20-mer) (calculated M r ϭ 40,000). Importantly, as demonstrated in lanes 7 and 8, NaBH 4 trapping occurred in a substrate-dependent manner. Pre-exposure of the 8-kDa domain to NaBH 4 failed to signifi-cantly reduce the enzyme's ability to become trapped to substrate upon re-exposure to NaBH 4 .
As illustrated by the amounts of free versus complexed ␤-pol or 8-kDa domain that were transferred to the Western blots (compare arrows versus asterisks on panel C in Figs. 1 and 2), only a small fraction of either protein was trapped. Although autoradiography was adequate to detect ␤-pol-DNA complexes when substrate was incubated with as little as a 20-fold molar excess of ␤-pol, immunodetection became extremely difficult (data not shown). PhosphorImager/ImageQuant analysis of the polyurea gel shown in Fig. 1A revealed that NaBH 4 mediated the incorporation of 0.9 ng of the intact ␤-pol (23 fmol, approximately 0.04% of the protein) into enzyme-DNA complexes. In  1 and 2) or a 600X molar excess (2200 ng ϭ 56 pmol) of intact ␤-pol (lanes 3 and 4). Seconds after control buffer or intact ␤-pol was pipetted into the reaction mixtures, trapping was initiated by the addition of NaCl (lanes 1 and 3) or NaBH 4 (lanes 2 and 4) to a final concentration of 150 mM. Incubations were terminated after 60 min at 37°C by the addition of an equal volume of formamide loading buffer. The samples were denatured, loaded onto a 10% polyacrylamide gel containing 8 M urea, and subjected to electrophoresis. Oligonucleotide sizing markers ranging from 8 to 32 bases were run in the outer lanes of the gel, and their migration distances are marked to the left of panel A. The asterisk to the left of panel A denotes the position of the intact ␤-pol-DNA complex, and the arrow to the left of panel A points at 28-mer substrate. Panels B and C, preincised AP DNA (1.6 ng ϭ 95 fmol) was incubated with enzyme dilution buffer (lanes 1 and 2), a 150X molar excess (560 ng ϭ 14 pmol) of active (lanes 3 and 4) or heat-inactivated (lane 5) intact ␤-pol, or a 600X molar excess (2200 ng ϭ 56 pmol) of active (lanes 6 and 7) intact ␤-pol. Seconds after control buffer or intact ␤-pol was pipetted into the reaction mixtures, trapping was initiated by the addition of NaCl (lanes 1, 3, and 6) or NaBH 4 (lanes 2, 4, 5, and 7) to a final concentration of 150 mM. Incubations were terminated after 60 min at 37°C by the addition of an equal volume of 4% SDS loading buffer. The samples were denatured, loaded onto a 10% SDS-PAGE gel, and subjected to electrophoresis. Prestained low and high molecular size markers were run in the outer lanes of the gel, and their migration distances are marked in between panels B and 37-kDa complex. A comprehensive comparison of the quantitative results from Figs. 1 and 2 suggests that intact ␤-pol was trapped somewhat more readily than the 8-kDa domain.
Experimental Design Considerations-Reproducible trapping of the ␤-pol-DNA and 8-kDa domain-DNA complexes was accomplished by rigorous adherence to the protocols outlined under "Experimental Procedures." Pilot studies established that our substrate preparation achieved complete conversion of the duplex 49-mer to the intended product. When preincised AP DNA was run on 10% polyacrylamide gels containing 8 M urea, it co-migrated as expected with a 28-mer oligonucleotide sizing marker (see Figs. 1 and 2, panel A, lanes 1 and 2). On SDS-PAGE gels, the uracil-containing 49-mer, AP-containing 49-mer, and preincised AP 28-mer DNAs could be distinguished from each other because each ran as a distinct series of bands. The complex banding patterns varied slightly from gel to gel and were influenced by several factors: the polyacrylamide and glycerol contents of the gels, the type (NaCl versus NaBH 4 ) and concentration of salt in the samples, and the presence or absence of poly(dI-dC).
The most critical element of the experimental design proved to be the order in which the reaction components were combined. Since ␤-pol cannot liberate dRP from reduced AP sites (1), a form of AP site that is resistant to ␤-elimination, NaBH 4 was added into the reaction mixture last. Preliminary experiments had indicated that 150 mM NaBH 4 could reduce virtually all of the reaction substrate in under 1 min. Pre-NaBH 4 exposure of the substrate to even a 20-fold molar excess of enzyme was also problematic. During even a 1-or 2-min lag period between the introductions of ␤-pol enzyme and NaBH 4 , enough substrate was consumed to prevent the formation of any ␤-pol-DNA complexes. Thus, there exists a very short time frame in which trapping of the ␤-pol-DNA complexes can occur.
Probing the Role of Helix-2 in the ␤-pol dRP Excision Reaction-A 32-amino acid peptide that corresponds to ␤-pol residues 27-58 served as a tool to probe the potential role of helix-2 in catalysis. Historically, this region of the protein (originally generated by proteolysis of the 8-kDa domain with Staphylococcus aureus V8 protease) has been studied for its singlestranded DNA-binding properties (22). Three variations on the NaBH 4 trapping experiment were attempted: (i) inhibition by the peptide of 8-kDa domain-DNA complex formation, (ii) inhibition by antipeptide antisera of 8-kDa domain-DNA complex formation, and (iii) trapping of a peptide-DNA complex (data not shown). In (i), peptide was allowed to prebind to the substrate prior to the addition of 8-kDa domain and NaBH 4 to the reaction mixture. However, this treatment did not interfere with trapping. In (ii), 8-kDa domain was preincubated with pre-or postimmune antiserum before it was added into the trapping reaction. Preincubation of the enzyme with post-versus preimmune antiserum reduced the level of 8-kDa domain-DNA trapping by 1%. In (iii), trapping reactions were started by the addition of varying amounts of peptide in place of ␤-pol and either NaBH 4 or NaCl. A covalent complex was not seen at the predicted 12-kDa position. Presented with these generally negative results, it appears unlikely that the active site residue for the ␤-pol dRP excision reaction resides within helix-2 or its adjacent loop segments.
Speculation on the Identity of the Active Site Residue Responsible for dRP Excision-In summary, this report has shown that the dRpase reaction of rat ␤-pol proceeds via an imine intermediate. This intermediate arises when the nitrogen atom of a primary amine forms a bond with the C-1Ј atom of an AP nucleoside. Our results would tend to support the recent finding of Matsumoto and Kim (1) that the 8-kDa domain of ␤-pol is necessary and sufficient to release 5Ј-deoxyribose phosphate termini from preincised AP DNA. The 8-kDa domain contains 15 lysine residues at the following amino acid positions: 3, 5, 27, 35, 41, 48, 52, 54, 60, 61, 68, 72, 81, 84, and 87. Six of these lysines (at positions 27, 35, 41, 48, 52, and 54) are located within helix-2 or its adjacent loops and can probably be eliminated as serious candidates for the active site role. A new crystal structure of human DNA ␤-pol complexed to gapped DNA shows Lys-35, Lys-68, Lys-72, and Lys-84 clustered around the 5Ј-phosphate at the upstream end of the gap (25). Given this finding and the knowledge that pyridoxal 5Ј-phosphate modification of K72 inactivates the DNA synthesis function of ␤-pol (26), preliminary work was carried out on the K72A mutant of ␤-pol. As assayed by a shift in mobility of the labeled substrate, the mutant seemed unable to release dRP from preincised AP DNA. 3 K72A could be trapped on the substrate, however, albeit to a lesser extent (Ϸ30%) than wild-type ␤-pol (data not shown). These results were not definitive because if K72 truly acts as the catalytic nucleophile, substitution of lysine with an alanine would be predicted to completely inactive the enzyme. Perhaps another residue substituted for K72, or the conformation of the mutant was altered to such an extent that the dRpase activity had been compromised. Alternatively, instead of the ⑀-amino group of a lysine residue, the ␣-amino group of the amino terminus may catalyze the dRpase reaction. Precedent for this mechanism exists; two N-glycosylase/AP lyase enzymes, T4 bacteriophage endonuclease V (17,21,27), and E. coli Fpg (19), 4 employ their amino-terminal ␣-NH 2 to form isolatable, enzyme-DNA intermediates. Clearly, further study will be required to identify the active site residue that catalyzes the dRP excision reaction of ␤-pol.