Protection against Methylation-induced Cytotoxicity by DNA Polymerase β-Dependent Long Patch Base Excision Repair*

Using a plasmid-based uracil-containing DNA substrate, we found that the long patch base excision repair (BER) activity of a wild-type mouse fibroblast extract was partially inhibited by an antibody to DNA polymerase β (β-pol). This suggests that β-pol participates in long patch BER, in addition to single-nucleotide BER. In single-nucleotide BER, the deoxyribose phosphate (dRP) in the abasic site is removed by the lyase activity of β-pol. Methoxyamine (MX) can react with the aldehyde of an abasic site, making it refractory to the β-elimination step of the dRP lyase mechanism, thus blocking single-nucleotide BER. MX exposure sensitizes wild-type, but not β-pol null mouse embryonic fibroblasts, to the cytotoxic effects of methyl methanesulfonate (MMS) and methylnitrosourea. Expression of β-pol in the null cells restores the ability of MX to modulate sensitivity to MMS. The β-pol null cells are known to be hypersensitive to MMS and methylnitrosourea, and in the presence of MX (i.e. under conditions where single-nucleotide BER is blocked) the null cells are still considerably more sensitive than wild-type. The data are consistent with a role of β-pol in long patch BER, which helps protect cells against methylation damage-induced cytotoxicity.

Using a plasmid-based uracil-containing DNA substrate, we found that the long patch base excision repair (BER) activity of a wild-type mouse fibroblast extract was partially inhibited by an antibody to DNA polymerase ␤ (␤-pol). This suggests that ␤-pol participates in long patch BER, in addition to single-nucleotide BER. In single-nucleotide BER, the deoxyribose phosphate (dRP) in the abasic site is removed by the lyase activity of ␤-pol. Methoxyamine (MX) can react with the aldehyde of an abasic site, making it refractory to the ␤-elimination step of the dRP lyase mechanism, thus blocking single-nucleotide BER. MX exposure sensitizes wildtype, but not ␤-pol null mouse embryonic fibroblasts, to the cytotoxic effects of methyl methanesulfonate (MMS) and methylnitrosourea. Expression of ␤-pol in the null cells restores the ability of MX to modulate sensitivity to MMS. The ␤-pol null cells are known to be hypersensitive to MMS and methylnitrosourea, and in the presence of MX (i.e. under conditions where single-nucleotide BER is blocked) the null cells are still considerably more sensitive than wild-type. The data are consistent with a role of ␤-pol in long patch BER, which helps protect cells against methylation damage-induced cytotoxicity.
Single base lesions in mammalian genomic DNA are repaired by the process known as base excision repair (BER). 1 Typically, such base lesions arise through endogenous events including spontaneous base loss, uracil incorporation, base deamination, and base oxidation, and also through base alkylations introduced by alkylating agents. Recent evidence has indicated that BER in mammalian cells is mediated through at least two subpathways that are differentiated by the repair patch sizes and the enzymes involved and are designated as "single-nucleotide BER" and "long patch BER" (2-10-nucleotide patch) (1)(2)(3). The most simple single-nucleotide BER pathway is an ordered sequential process initiated by base excision by a monofunctional glycosylase. Removal of the deoxyribose phosphate (dRP) in the abasic site is by the dRP lyase activity of DNA polymerase ␤ (␤-pol) (4). Single-nucleotide BER of a uracil-containing oligonucleotide substrate can be reconsti-tuted in vitro with four purified human proteins: uracil-DNA glycosylase (UDG), apurinic/apyrimidinic endonuclease (APE), ␤-pol, and DNA ligase I (4), and a complex containing these enzymes can be isolated from extracts of mammalian cells (5). In addition, genetic evidence with ␤-poland DNA ligase I-deficient mammalian cell lines indicates that these proteins are important for the repair of alkylating agent induced lesions (6,7).
The choice of subpathway in BER (single-nucleotide or long patch) depends on whether the dRP intermediate can be efficiently removed by the ␤-pol lyase activity to yield a 5Ј-phosphorylated DNA strand capable of serving as a substrate for DNA ligase (4,8). When such processing is not efficient, for example during repair of a reduced abasic site, long patch BER can occur (9). In addition, it has been suggested that the properties of the damage-specific DNA glycosylase (e.g., monofunctional, exhibiting only glycosylase activity, or bifunctional, having glycosylase and ␤-lyase activity) can determine the BER subpathway (10). For BER studied in vitro, the two subpathways can co-repair the same type of DNA lesion and operate in the same extract (10,11).
It is still unclear which enzymes are involved in vivo in the various BER subpathways. Biochemical and genetic evidence indicate that, in addition to those mentioned above, 3-methyladenine DNA glycosylase, XRCC1, DNA ligase III, flap endonuclease 1 (FEN1), and poly(ADP-ribose) polymerase are important proteins for BER and the repair of alkylating agentinduced lesions (9,(12)(13)(14)(15)(16)(17). For single-nucleotide repair, ␤-pol is the polymerase of choice for the resynthesis step (6); however, ␤-pol-deficient extracts are able to perform single-nucleotide BER (11). For long patch BER, repair has been shown to be stimulated by proliferating cell nuclear antigen which has led to the suggestion that DNA polymerases ␦ and/or ⑀ are involved (1,18). However, the role of proliferating cell nuclear antigen may be limited to stimulation of FEN1-dependent flap cleavage (19), and proliferating cell nuclear antigen can promote ␤-poldependent long patch repair by this mechanism (9). There is additional evidence pointing to a role of ␤-pol in long patch repair DNA synthesis. For example, ␤-pol antibody was found to inhibit long patch repair mediated by cell extracts (9), and ␤-pol-deficient extracts were unable to repair a reduced AP site on a linear DNA substrate that could be repaired by wild-type extracts (20). Additionally, recent data indicate that ␤-pol plays an essential role in the strand displacement synthesis of long patch BER (21,22).
In this paper, we describe further analysis of the role of ␤-pol in BER and dissect the contribution of ␤-pol in protecting cells against the cytotoxic effects of simple methylating agents. Using wild-type mouse embryonic fibroblast cell extracts, we find that repair of a plasmid-based uracil-containing BER substrate by long patch BER is partially blocked by an antibody to ␤-pol. Thus, this extract is capable of both ␤-pol-dependent and ␤-pol-independent long patch BER, in addition to the single-nucleotide BER we have described previously (6). We also examined the cellular phenotype of a ␤-pol-dependent long patch BER deficiency in mouse fibroblast cells. First, single-nucleotide BER was chemically blocked by methoxyamine (MX), and then the effect of ␤-pol gene deletion on cellular sensitivity to DNAalkylating agents was examined. The results demonstrate a role of ␤-pol-dependent long patch BER in protection of cells against the cytoxicity of the monofunctional DNA-alkylating agents methyl methanesulfonate (MMS) and methylnitrosourea (MNU).
Cell Lines-The cell line utilized for preparation of cell extracts was a clone of the wild-type mouse embryonic fibroblast cell line M␤16tsA described previously (6). The wild-type mouse fibroblasts, a clone of the isogenic ␤-pol null line M␤19tsA (6), and the wild-type ␤-pol minigenetransfected null cell line, 19/A5 (27), were routinely grown at 34°C in a 10% CO 2 incubator in Dulbecco's modified Eagle's medium supplemented with GlutaMAX-1, 10% fetal bovine serum, and hygromycin (80 g/ml). Primary Xpaϩ/Ϫ and XpaϪ/Ϫ isogenic mouse embryonic fibroblasts (28) were obtained from Dr. Harry van Steeg (National Institute of Public Health and Environmental Protection, Bilthoven, The Netherlands). These cells were transformed with SV40 T antigen as described for the wild-type and ␤-pol null fibroblasts (6). Transformed Xpa fibroblasts were routinely grown at 34°C in a 10% CO 2 incubator in Dulbecco's modified Eagle's medium supplemented with glutamine, 10% fetal bovine serum, minimal essential medium non-essential amino acids, and hygromycin (80 g/ml). All cells were routinely tested and found to be free of mycoplasma contamination.
Cell Extracts-For the long patch BER assay, wild-type mouse fibroblast cells were cultured in 150-mm dishes until near-confluent, washed three times with Dulbecco's phosphate-buffered saline (Life Technologies, Inc.), and harvested by scraping. Approximately 5 ϫ 10 7 cells were suspended in 1 ml of buffer (5 mM NaPO 4 , pH 7.1, 150 mM NaCl, 2.5 mM KCl) plus protease inhibitors, aprotinin (10 g/ml), leupeptin (0.5 g/ ml), and Pefabloc SC (10 g/ml) supplied by Roche Molecular Biochemicals. The suspension was subjected to five cycles of freeze/thaw and further disrupted by sequential passage through 18-and 22-gauge needles. The resulting homogenate was centrifuged at 20,000 ϫ g for 10 min at 4°C. The supernatant fraction was removed, aliquoted, and stored at Ϫ80°C. Protein concentration was measured using the Bio-Rad protein assay dye reagent. For the AP endonuclease assay, cell extracts were prepared from wild-type mouse fibroblasts as described previously (20).
Construction of DNA Substrate-A partially duplex oligonucleotide was formed by annealing a phosphorylated 12-mer (5Ј-ACCGG-TACUGGC-3Ј) containing a uracil residue at position 9 and a 20-mer (5Ј-ACGTGCCGGTACCGGTCTAG-3Ј). The 12-mer (6 nmol) and 20mer (6 nmol) were mixed in 30 l of buffer containing 10 mM Tris-HCl, pH 8.0, 1 mM EDTA, 250 mM NaCl, heated to 98°C for 3 min, and then slowly cooled to room temperature. The annealed oligonucleotide duplex was purified by precipitation, washed twice with 70% ethanol, and resuspended in 30 l of 10 mM Tris-HCl, pH 8.0, 1 mM EDTA. An aliquot of this DNA was analyzed by 20% non-denaturing polyacrylamide gel electrophoresis to confirm that annealing had occurred.
The closed circular DNA substrate containing a unique G:U base pair was constructed according to a previously described procedure (29) with slight modifications. Briefly, pUC19 plasmid (500 g) was digested with PstI (750 units) overnight at 37°C, and the reaction was checked for FIG. 1. Properties of mammalian cell extract-mediated single-nucleotide and long patch BER on a plasmid-based uracil-containing DNA substrate. A, diagram illustrating plasmid substrate and restriction enzyme analysis of the [ 32 P]dCMP-labeled repair product. The repair reaction mixture contained [ 32 P]dCTP, 5 g of wild-type mouse embryonic fibroblast cell extract, plasmid substrate, and other components as described under "Experimental Procedures." After incubation, the product DNA was digested either with SphI ϩ NcoI to obtain a labeled 10-nucleotide fragment (lane 1) or HaeIII to obtain a 6-nucleotide fragment (lane 2) and then electrophoresed in a denaturing 20% polyacrylamide gel. An autoradiogram is shown. The positions of the first six nucleotides in the repair patch are indicated by numbers, and the positions for incorporation of dCMP into the respective restriction fragments are indicated by asterisks. B, effect of anti-␤-pol rabbit serum on dCMP and dAMP incorporation into the HaeIII fragment. The antiserum or buffer and cell extract (5 g) were pre-incubated as a 1:1 mixture for 30 -60 min at 0 -1°C prior to the BER incubation. dCMP incorporation is into positions 1 and 2 of the repair patch, and is primarily a measure of single-nucleotide BER, and dAMP incorporation is into position 3 and is a measure of long patch BER. Bars represent the activity in the absence (Ϫ) of antiserum (57 ϫ 10 Ϫ6 pmol of dCMP and 6 ϫ 10 Ϫ6 pmol of dAMP incorporated/min) and the relative incorporation with addition of anti-␤-pol antiserum (ϩ). The results shown represent the mean and standard error of three experiments. complete linearization by 1% agarose gel electrophoresis. The reaction mixture was diluted with five volumes of Qiagen binding buffer, and the plasmid was purified using a Qiagen-tip 500. The purified PstI-linearized pUC19 plasmid DNA (300 pmol) was ligated to the G:U base pair-containing oligonucleotide duplex (800 pmol) in a 2-ml reaction mixture containing 100 mM Tris-HCl, pH 7.5, 10 mM MgCl 2 , 100 mM NaCl, 1 mM dithiothreitol (DTT), 2 mM ATP, 800 units of PstI, and 250 units of T4 DNA ligase at 15°C overnight. The ligation product was purified using a Qiagen-tip 500 as described above and 5Ј-phosphorylated in a reaction mixture (500 l) containing 10 mM Tris-HCl, pH 8.0, 5 mM MgCl 2 , 100 mM NaCl, 1 mM 2-mercaptoethanol, 2.5 mM ATP, 600 units of BamHI, and 250 units of T4 polynucleotide kinase for 3 h at 37°C. After purification by Qiagen-tip 500, the DNA was circularized at 15°C overnight in a 5-ml reaction mixture containing 50 mM Tris-HCl, pH 7.5, 10 mM MgCl 2 , 100 mM NaCl, 1 mM DTT, 1 mM ATP, 750 units of BamHI, 750 units of BglII, and 300 units of T4 DNA ligase. The reaction product was precipitated with three volumes of ethanol, washed twice with 70% ethanol, and resuspended in 500 l of H 2 O. The circular DNA substrate was separated from linear DNA by 0.8% agarose gel electrophoresis. A small portion (1 l) of the sample was electrophoresed as a marker in a separate lane. This marker, but not the preparative sample, was visualized by UV light, to minimize UV/ ethidium bromide-induced nicking of the DNA substrate. The closed circular DNA substrate was isolated by electroelution and purified using a Qiagen polymerase chain reaction purification kit. The product was eluted in 10 mM Tris-HCl, pH 7.5, and DNA was quantified using Hoechst 33258 dye.
Long Patch BER Assay-BER assays were performed in a reaction mixture (11 l) that contained 5 g of cell extract, 1 nM closed circular DNA substrate, 50 mM Hepes, pH 7.5, 5 mM MgCl 2 , 1 mM DTT, 0.1 mM EDTA, 5 mM diTris-phosphocreatine, 10 units of creatine phosphokinase, 4 mM ATP, 0.5 mM NAD, 4 M [␣-32 P]dCTP (specific activity 1 ϫ 10 6 DPM/pmol), and 4 M each dATP, TTP, and dGTP. Alternatively, 4 M [␣-32 P]dATP (specific activity 1 ϫ 10 6 DPM/pmol) was included as the labeled nucleoside triphosphate with 4 M each of the remaining three unlabeled nucleoside triphosphates. After incubation at 37°C for 60 min, the reactions were stopped by adding 1.5 l of 0.5 M EDTA. The samples were extracted with phenol/chloroform, and the DNA was purified using a MicroSpin S-400 HR column and ethanol precipitation. The DNA samples were resuspended in 8 l of H 2 O, and repair was assessed by digestion with either HaeIII alone or with SphI ϩ NcoI (10 units each) at 37°C overnight. Samples were then electrophoresed on a 20% denaturing polyacrylamide gel. Radiolabeled products were quantified using a Molecular Dynamics PhosphorImager 450 and Image-Quant software.
3Ј End Labeling-A 49-mer oligodeoxyribonucleotide containing a uracil residue at position 21 (5) was labeled at the 3Ј end by terminal deoxynucleotidyltransferase using [␣-32 P]ddATP and annealed to its complementary strand by heating the solution at 90°C for 3 min, followed by slow cooling to 25°C. 32 P-Labeled duplex oligodeoxynucleotide was separated from unincorporated [␣-32 P]ddATP using a NEN-SORB-20 column according to the manufacturer's suggested protocol. The radiolabeled oligodeoxynucleotide was lyophilized, resuspended in H 2 O, and stored at Ϫ30°C.
Preparation of DNA Substrates for dRP Lyase and Excision Assays-32 P-Labeled uracil containing duplex DNA (62.5 nM) was pretreated with 10 nM UDG for 20 min at 37°C in 100 l of buffer containing 70 mM Hepes, pH 7.4, 0.5 mM EDTA, and 0.2 mM DTT. The reaction mixture was then supplemented with 10 mM MgCl 2 and APE, either 5 nM (for normal AP-DNA) or 100 nM (for MX-adducted DNA) and the incubation continued for an additional 20 min at 37°C. To prepare MX-adducted substrate, the 32 P-labeled UDG-treated duplex oligonucleotide was mixed with 33 mM MX in buffer containing 50 mM KPO 4 , pH 7.1 and incubated at 37°C. After 30 min, the DNA was recovered by ethanol precipitation, lyophilized, resuspended in water, and stored at Ϫ30°C.
AP Endonuclease Assay-AP endonuclease activity was assayed in a reaction mixture (10 l) containing 50 mM Hepes, pH 7.4, 10 mM MgCl 2 , 2 mM DTT, and 20 nM 32 P-labeled MX-adducted DNA. The reaction was initiated by adding the indicated concentrations of purified APE or cell extract, and incubation was at 37°C. Samples were withdrawn at the indicated time periods. The reaction was terminated by transfer to 0 -1°C, and the DNA product was stabilized by addition of NaBH 4 to a final concentration of 340 mM and incubating 30 min on ice. The stabilized DNA products were recovered by ethanol precipitation in the presence of 0.1 g/ml tRNA and resuspended in 10 l of gel loading buffer (95% formamide, 20 mM EDTA, 0.02% bromphenol blue, and 0.02% xylene cyanol). Alternatively, the reactions were terminated without product stabilization by the addition of an equal volume of gel loading buffer. After incubation at 75°C for 2 min, the reaction products were separated by electrophoresis in a 20% polyacrylamide gel containing 8 M urea in 89 mM Tris-HCl, 89 mM boric acid, and 2 mM EDTA, pH 8.8, and visualized by autoradiography. To quantify the product, the gel was scanned on a PhosphorImager 450 and the data were analyzed using ImageQuant software.
dRP Lyase Assay-dRP lyase activity was performed in a reaction mixture (10 l) containing 50 mM Hepes, pH 7.4, 10 mM MgCl 2 , 2 mM DTT, and 20 nM pre-incised 32 P-labeled normal AP-DNA or MX-adducted DNA. The reaction was initiated by adding ␤-pol (0 -25 nM as indicated) and incubated at 37°C for 15 min. The reaction was terminated by transfer to 0 -1°C, and the DNA product was stabilized by addition of NaBH 4 as described above. The stabilized DNA products were processed as described above and visualized by autoradiography.
Excision Assay-The excision reaction was reconstituted in a reaction mixture (10 l) that contained 50 mM Hepes, pH 7.4, 2 mM DTT, 10 mM MgCl 2 , 0.5 mM EDTA, 2 mM ATP, 20 M each dATP, dGTP, dCTP, and TTP, and APE-preincised 32 P-labeled duplex oligonucleotide substrate (20 nM). The reaction was initiated by adding 20 nM FEN1 and ␤-pol, as indicated in the figure legend. Incubation was at 37°C for 30 min. The reaction was terminated by transfer to 0 -1°C, and the DNA product was stabilized by addition of NaBH 4 as described above. The stabilized DNA product was recovered by ethanol precipitation, separated by electrophoresis, and visualized by autoradiography.
Cytotoxicity Studies-Cytotoxicity was determined by growth inhibition assays. It has been shown previously that SV40-transformed mouse embryonic fibroblasts grow logarithmically and at similar rates under the conditions of the assay (6). Cells were seeded at a density of 40,000 cells/well in six-well dishes. The following day, they were exposed for 1 h to a range of concentrations of MMS or MNU in growth medium without hygromycin. MMS was dissolved directly in the medium. A stock solution of MNU was freshly-prepared in dimethyl sulfoxide and dissolved in medium at the time of the experiment. After 1 h, the cells were washed with Hanks' balanced salt solution and fresh medium was added. For UV irradiation (254 nm), cell monolayers were washed twice with Hanks' balanced salt solution before UV exposure (0 -20 J/m 2 ) in a Stratalinker model 1800 (Stratagene, La Jolla, CA), followed by addition of growth medium. Dishes were incubated for 4 -5 days at 34°C in a 10% CO 2 incubator until untreated control cells were approximately 80% confluent. Cells (triplicate wells for each drug concentration) were counted by a cell lysis procedure (30), and results were expressed as the number of cells in drug-treated wells relative to cells in control wells (% control growth). The IC 50 and IC 90 values (defined as the concentration of agent required for 50% or 90% growth inhibition compared with untreated controls) were determined from concentration-percentage of growth inhibition curves.
Cytotoxicity studies were also conducted in the presence of MX. A stock solution of MX (1-5 M in phosphate-buffered saline) was prepared immediately before use and NaOH added to achieve neutral pH. MX stock solution was added to the volume of medium required for the experiment, and the pH was re-adjusted to 7.2 by further addition of NaOH. Dilutions of MMS or MNU were prepared in the MX-containing medium, and cells were dosed as described above. In certain experiments, cells were further incubated in MX-medium for up to 7 h following the 1-h alkylating agent exposure.

RESULTS
Long Patch BER on a Plasmid-based Substrate-We used a plasmid-based uracil-containing DNA substrate (see "Experimental Procedures") that allows simultaneous quantification of repair by the single-nucleotide and long patch subpathways of BER. Conditions were standardized to measure the efficiency (k cat /K m,DNA ) of BER and involved using a limiting concentration of DNA and saturating or near-saturating concentrations of dNTP for both single-nucleotide and long patch BER subpathways. Restriction enzyme and electrophoretic analysis of BER products formed with [␣-32 P]dCTP as the labeled nucleoside triphosphate indicated that the level of dCMP incorporation by wild-type cell extract into the first nucleotide of the repair patch (SphI ϩ NcoI fragment) was similar to dCMP incorporation into the first two nucleotides of the repair patch (HaeIII fragment; Fig. 1A). In the HaeIII fragment, the ratio of dCMP incorporation (first and second nucleotide):dAMP incorporation (third nucleotide) was approximately 10:1 (data not shown). Thus, as expected from earlier work (11), single-nucle-otide BER of uracil-DNA was predominant over long patch BER in a wild-type cell extract.
In the experiment shown in Fig. 1B, it was found that the combined short and long patch BER activity of the wild-type extract (incorporation of labeled dCTP into position 1 and 2 of the repair patch) as well as long patch BER activity (incorporation of labeled dATP into position 3) were both partially inhibited by a neutralizing antibody to ␤-pol. Non-immune serum from the same rabbit was used in control incubations and did not inhibit single-nucleotide or long patch BER activity (data not shown). These results indicate clearly that a portion of long patch BER activity measured in the wild-type cell extract is dependent upon ␤-pol (Ϸ75%). Taken together, these results indicate that in addition to a role in single-nucleotide BER, ␤-pol also participates in long patch BER in mouse fibroblast extracts.
Use of MX to Block Single-nucleotide BER-The primary amine of MX is capable of reacting with the aldehydic C1Ј atom of the abasic site, as illustrated in Scheme 1.
The Schiff base intermediate produced after MX attack on the C1Ј atom spontaneously resolves into the stably adducted SCHEME 1   FIG. 3. Measurements of AP endonuclease activity using MX-adducted and normal AP site DNA substrates. The reaction conditions and product analyses are as described under "Experimental Procedures." A, diagram of a 3Ј-[ 32 P]ddAMP-labeled DNA substrate containing either the MX-adducted or normal AP site. Enzymatic cleavage of the DNA substrate by AP endonuclease results in (29-mer ϩ MX-dRP or 29-mer ϩ dRP/29-mer) product, respectively. B, 32 P-labeled MX-adducted duplex oligonucleotide (20 nM) was incubated with 100 nM APE at 37°C, and aliquots were taken at the indicated time intervals. The incision product was stabilized and analyzed by polyacrylamide-urea gel electrophoresis. The k cat for APE acting on a normal AP site is 10 1/s (24). C, an autoradiogram illustrating the AP endonuclease activity of purified AP endonuclease or wild-type cell extract on MX-adducted AP site DNA (lanes 1, 2, 4, and 5) and normal AP site DNA (lanes 3 and 6). Incubation was at 37°C for 15 min. The APE and cell extract concentrations and the positions of the substrate and products are indicated. Note, under the reaction conditions in C where the incision product was not stabilized, cleavage of the normal AP site results in 29-mer final product (panel C, lanes 3 and 6). sugar molecule. Therefore, MX treatment will render the abasic site refractory to the ␤-elimination step involved in the dRP lyase activity of ␤-pol. This property of the MX-adducted abasic site DNA is confirmed by the experiments illustrated in Fig. 2. Using the duplex oligonucleotide substrates shown in Fig. 2A, the dRP lyase activity of ␤-pol could be detected on a APEcleaved normal AP site, resulting in a 29-mer product (Fig. 2B,  lanes 5 and 6), but not on a pre-incised MX-adducted AP site (lanes 2 and 3). We conclude that MX is capable of chemical deletion of single-nucleotide BER that depends upon elimination of the dRP moiety in the single-nucleotide gap.
It had been reported previously that MX-adducted AP sites are resistant to cleavage by mammalian APE (31,32). Using the normal AP site or MX-adducted AP site substrate shown in Fig. 3A, we found that APE is able to cleave the MX-adducted AP site, yet the turnover number of APE acting on the MXadducted AP site is substantially lower (Ϸ330-fold) than that for the normal AP site (Fig. 3B). This is further illustrated in Fig. 3C. A MX-adducted AP site is only partially cleaved by 100 nM APE after 15 min (lane 2), whereas under the same conditions, 5 nM APE efficiently cleaves a normal AP site (lane 3). Next, using wild-type cell extracts, partial cleavage of a MXadducted AP site is observed with 8 and 16 g of cellular protein (lanes 4 and 5), yet the cleavage of a normal AP site is complete with 8 g of protein (lane 6). In these experiments, reaction products were not stabilized by addition of NaBH 4 ; therefore, the product obtained after APE-dependent cleavage of the normal AP site substrate is a 29-mer, rather that a 29-merϩdRP (Fig. 3C). Thus, incision of a MX-adducted AP site is readily observed in cell extracts and suggests that such cleavage can occur in the intact cell as part of the BER pathway.
A MX-adducted AP site, pre-incised with APE, is an active substrate for the FEN1/␤-pol-mediated flap cleavage reaction of long patch BER. As shown in Fig. 4A, neither ␤-pol alone (lane 2) nor FEN1 alone (lane 3) was able to cleave the MXadducted AP site. The combination of ␤-pol and FEN1 cleaved the characteristic flap oligomer (dRP-N 3 ) (21) from the 5Ј end of the substrate molecule, resulting in predominant formation of a 26-mer product and leaving a four-nucleotide repair patch (Fig. 4A, lane 4). With the normal pre-incised AP site as substrate, ␤-pol alone was partially able to release the dRP moiety, resulting in a 29-mer product (Fig. 4B, lane 2), and FEN1 alone was able to release the dRP moiety plus 1 dNMP residue, resulting in a 28-mer product, and leaving a two-nucleotide gap in the substrate (Fig. 4B, lane 3). In agreement with previous results (33), FEN1 could not release the dRP moiety without an additional dNMP. The combination of ␤-pol and FEN1 produced the four-nucleotide gap characteristic of long patch BER and release of the dRP-N 3 oligomer and resulted in formation of a 26-mer product (Fig. 4B, lane 4). The combination of ␤-pol and FEN1 exhibited a similar level of activity toward the normal dRP-containing substrate and the MX-adducted substrate. We conclude from these results that MX adduction of the dRP moiety does not interfere with long patch BER involving excision of the dRP-N 3 oligomer.
Effect of MX on Methylation-induced Cytotoxicity-To examine any cellular role of ␤-pol-dependent long patch BER in sensitivity to DNA alkylating agents, we used the approach of treatment of cells with MX to eliminate single-nucleotide BER at the time of exposure to the methylating agents. Modification of abasic sites in cells by MX has been demonstrated previously to influence the cytotoxic effects of simple alkylating agents (34). High concentration and/or lengthy exposure to MX is equally cytotoxic to wild-type and ␤-pol null cell lines, but under all of the conditions reported here, MX was found to result in 10% or less growth inhibition (data not shown). Coexposure to MX was able to sensitize wild-type cells to the cytotoxic effects of both MMS and MNU. For MMS, the effect was both time-and concentration-dependent (Fig. 5). At a concentration of 10 mM MX, near-maximal sensitization was achieved at 4 h, with only a slightly enhanced effect at 8 h (Fig.  5A). The requirement for prolonged exposure to MX is consistent with the known slow repair of cytotoxic methyl adducts and therefore the appearance of AP sites as intermediates of BER in cells (35). For a 4-h exposure to MX, near maximal sensitization was achieved at a concentration of 20 mM, with an only slightly enhanced effect at 30 mM (Fig. 5B). For both MMS (Fig.  5) and MNU (data not shown), a similar maximal sensitization was seen for exposures to 10 mM MX for 8 h or 30 mM for 4 h, resulting in a 25% decrease in IC 50 values. We propose that this modulation of sensitivity is a result of MX-induced blockage of the predominant single-nucleotide BER pathway in the cells. In contrast, in the ␤-pol null cells, in which single-nucleotide BER is inefficient, there was no significant sensitization to MMS (Fig. 5C) or MNU (data not shown) by MX under the experimental conditions (10 mM MX for 8 h and 30 mM MX for 4 h) that were most effective in the wild-type cells. In ␤-pol null cells partially complemented by stable transfection of wild-type ␤-pol, co-exposure to MX (30 mM for a total of 4 h) resulted in sensitization to MMS (Fig. 5D). This result confirms that there is a requirement for ␤-pol, and presumably ␤-pol-dependent single-nucleotide repair, for the MX-induced modulation of sensitivity.
We have reported previously that the ␤-pol null cells are not hypersensitive to the cytotoxic effects of UV irradiation (6), suggesting that ␤-pol-dependent BER is not involved in the repair of cytotoxic UV-induced DNA adducts. We now demonstrate that neither wild-type or ␤-pol null cell lines were sensitized to UV irradiation in the presence of MX (Fig. 6A). In addition, MX was not able to sensitize cells to the DNA crosslinking nitrogen mustard derivative, chlorambucil (data not shown). These results suggest that the MX-induced sensitization may be specific for agents that produce cytotoxic lesions that can be repaired by BER. XpaϪ/Ϫ mouse fibroblasts, deficient in nucleotide excision repair (NER) and extremely sensitive to killing by UV irradiation (data not shown), were found to be only slightly hypersensitive to MMS when compared with isogenic Xpaϩ/Ϫ fibroblasts (Fig. 6B). These data indicate that MMS cytotoxicity is a result of DNA adducts which are repaired by a pathway other than NER, most likely BER. Coexposure to MX is able to sensitize equally both Xpaϩ/Ϫ and XpaϪ/Ϫ cell lines to MMS, resulting in a 40% decrease in IC 50 values (Fig. 6B). This result suggests that NER is not involved in repair of a MX-adducted AP site in mouse fibroblasts.
We have shown previously that the ␤-pol null cells are hypersensitive to the cytotoxic effects of both MMS and MNU, 2.4-and 3.7-fold, respectively, at IC 50 (6). When MMS and MNU sensitivity are compared in the presence of MX (i.e. under conditions where there is elimination of single-nucleotide BER), the ␤-pol null cells are still considerably more sensitive than wild-type cells (1.5-and 2.3-fold, respectively, for MMS and MNU at IC 50 ; Fig. 7). These data are consistent with the hypothesis that, in addition to its protective function in single-nucleotide BER, ␤-pol also plays a role in long patch BER and protects cells against methylation damageinduced cytotoxicity. DISCUSSION In this study, we present evidence for a ␤-pol dependence in the long patch BER of both a normal and a MX-adducted AP site. In addition, the results demonstrate a role for ␤-pol in the long patch BER of cytotoxic lesions produced as a result of exposure of cells to methylating agents. It has been proposed previously that ␤-pol functions specifically in wild-type cells in single-nucleotide BER (6). However, there is increasing evidence pointing to an additional role of ␤-pol in a long patch BER subpathway. The long patch BER activity of a wild-type mouse fibroblast extract was partially inhibited by a neutralizing antibody to ␤-pol (Fig. 1B), clearly demonstrating that a portion of the repair of uracil-containing DNA by long patch BER, measured on a plasmid substrate, is dependent on ␤-pol.
In these experiments, ␤-pol-dependent long patch repair of a natural AP site is seen under conditions where ␤-pol-mediated single-nucleotide repair is also operative.
Previously, ␤-pol-dependent long patch repair has been demonstrated under conditions where single-nucleotide repair is not operative. For example, Biade et al. (see Fig. 1C in Ref. 20) showed that a linear DNA substrate bearing a synthetic sugar phosphate that is refractory to the ␤-pol dRP lyase activity is repaired by long patch BER by a wild-type mouse embryonic fibroblast cell extract, but is not repaired by an extract from an isogenic cell line lacking ␤pol. Similarly, Klungland and Lindahl (9) demonstrated that long patch BER of a reduced AP site on an oligonucleotide DNA substrate by a HeLa cell extract is strongly inhibited by a specific neutralizing antibody to ␤-pol.
Recently, Dianov et al. (21) have shown that it is the excision product release step during long patch BER of an internally labeled uracil-DNA circular substrate by a mouse fibroblast cell extract that is dependent upon ␤-pol. The results suggest that ␤-pol conducts the strand displacement synthesis of long patch BER. Further evidence for this role of ␤-pol in long patch repair was provided in the present study, by the demonstration of a requirement for ␤-pol, in addition to FEN1, for the flap cleavage, and therefore subsequent long patch repair, of a MX-adducted AP site (Fig. 4). In the case of cleavage of the MX-adducted AP site as described here, it is not known whether other polymerases could fulfill this same role in the cell. However, in the study by Dianov and co-workers (21), excision product release was deficient in extracts from ␤-pol null cells and was inhibited in wild-type cell extracts by a ␤-pol neutralizing antibody. It is likely that there is a similar requirement for ␤-pol for the processing and subsequent long patch BER of a MX-adducted site.
The quantitative reaction of AP sites in DNA with MX in vitro and its use in studies of BER have been reported previously (31,36). MX-adducted AP sites have been found to be refractory to ␤-elimination (31), which would include the dRP lyase activity of ␤-pol as demonstrated here (Fig. 2B). Therefore, MX can specifically block ␤-pol-dependent single-nucleotide repair, but does not interfere with long patch BER (Fig.  4A). From the present and previous data, it can be predicted that the preferred subpathway of repair of a MX-adducted AP site is long patch BER. Indeed, long patch BER (average of 10 nucleotides per MX-AP site) by Chinese hamster ovary cell extracts has been detected on a multiply MX-adducted plasmid substrate (37).
We and others have shown that MX-adducted AP sites are resistant to cleavage by APE ( Fig. 3B; Refs. 31 and 32). However, incision of a MX-adducted AP site is readily observed in cell extracts (Fig. 3C). This suggests that the APE-mediated cleavage of a MX-adducted AP site required for BER will be functional in the intact cell. It should be noted that approximately two-thirds or more of endogenous AP sites, and presumably therefore AP sites produced as a result of BER in cells, are already 5Ј-cleaved (38). Under these circumstances, and presuming that MX reacts with pre-incised AP sites as does the aldehyde reactive probe (38), the relatively inefficient cleavage of MX-adducted AP sites by APE would not be required for long patch BER in vivo.
In the present study, we found that MX treatment results in a concentration-and time-dependent sensitization of wild-type but not ␤-pol null cells to MMS (Fig. 5). With high concentrations of MX (30 mM), there was an apparent small sensitization in the ␤-pol null cells in certain experiments, perhaps consistent with the low levels of single-nucleotide repair that have been detected in extracts from these cells (11), but this sensitization was not statistically significant. Expression of ␤-pol in the null cells restores the ability of MX to modulate sensitivity to MMS (Fig. 5D). We suggest that the modulation in wild-type cells is a result of inhibition of DNA repair, specifically ␤-poldependent single-nucleotide BER. BER of N-methyl DNA adducts produced as a result of exposure to MMS and MNU are initiated by the monofunctional glycosylase, N-methylpurine- DNA glycosylase (39). Base excision, as well as spontaneous depurination, will result in formation of cytotoxic AP sites, which, after cleavage by APE, require removal of the dRP through the dRP lyase activity of ␤-pol. We show here that this lyase activity can be chemically blocked by MX adduction of the AP site (Fig. 2B).
A slot-blot assay using an aldehyde reactive probe has been utilized to detect AP site induction in cells following treatment with MMS (40). Using the same 1-h MMS treatment protocol as in the cytotoxicity studies, we were not able to detect a significant increase in AP sites in the wild-type fibroblast cells. 2 Therefore, it has not been possible to use this procedure to demonstrate directly that MX can react with AP sites formed in our cells. MX has been demonstrated previously to react with AP sites produced as a result of cell exposure to monofunctional DNA-alkylating agents (41). In these experiments, the presence of MX during cell treatment with diethyl sulfate has been shown to cause a concentration-and time-dependent inhibition of repair as assessed by alkaline elution (41).
We propose that MX-induced modulation of sensitivity is specific for agents that produce cytotoxic lesions that can be repaired by BER and where ␤-pol deficiency results in hypersensitivity. Whereas MX was able to specifically sensitize wildtype cells to monofunctional methylating agents that produce cytotoxic AP sites in cells (42), it was not able to sensitize cells to chlorambucil (data not shown) or UV irradiation (Fig. 6A). Chlorambucil initially produces DNA lesions that could be repaired by BER (43) (␤-pol null cells show low level hypersensitivity to this agent); however, cytotoxicity is thought to result from DNA cross-linking rather than AP site formation (43). UV irradiation produces primarily "bulky" cytotoxic DNA adducts that are repaired by NER rather than ␤-pol-dependent BER.
It has been reported that NER may be involved in the repair of a MX-adducted AP site (44). In agreement with data demonstrating that NER-deficient human fibroblasts, defective in repair of lesions formed by UV irradiation, possess a normal capacity for repair following MMS treatment (45), we show that the NER-deficient XpaϪ/Ϫcells are only slightly hypersensitive to MMS (Fig. 6B). In addition, we find that MX is able to sensitize NER-efficient Xpaϩ/Ϫ and NER-deficient XpaϪ/Ϫ cells equally to MMS (Fig. 6B), suggesting that NER is not involved in repair of a MX-adducted AP site in mouse fibroblasts.
It has been proposed that the hypersensitivity of ␤-pol null cells to monofunctional DNA-alkylating agents is a result of the deficiency in ␤-pol-dependent single-nucleotide repair of DNA damage (6). We describe here that, under conditions where there is elimination of single-nucleotide BER by MX, the ␤-pol null cells are still considerably more sensitive than wild-type cells (Fig. 7). These data are consistent with a role of ␤-pol-dependent long patch BER, in addition to ␤-pol-dependent singlenucleotide repair, in the protection of cells against methylation-induced cytotoxicity. In these experiments in wild-type cells, appreciable protection against cytotoxicity by ␤-pol-dependent long patch repair was seen in the absence of ␤-pol-dependent single-nucleotide repair. In contrast, experiments to quantitate repair in vitro demonstrate that single-nucleotide BER predominates over long patch repair in wild-type cell extracts. These results suggest that ␤-pol-dependent long patch repair can be up-regulated to compensate for a cellular deficiency in ␤-pol-dependent single-nucleotide repair.