5 * -Nicked Apurinic/Apyrimidinic Sites Are Resistant to b -Elimination by b -Polymerase and Are Persistent in Human Cultured Cells after Oxidative Stress*

Genomic DNA is continuously exposed to oxidative stress. Whereas reactive oxygen species (ROS) preferentially react with bases in DNA, free radicals also ab-stract hydrogen atoms from deoxyribose, resulting in the formation of apurinic/apyrimidinic (AP) sites and strand breaks. We recently reported high steady-state levels of AP sites in rat tissues and human liver DNA (Nakamura, J., and Swenberg, J. A. (1999) Cancer Res. 59, 2522–2526). These AP sites were predominantly cleaved 5 * to the lesion. We hypothesized that these endogenous AP sites were derived from oxidative stress. In this investigation, AP sites induced by ROS were quantitated and characterized. A combination of H 2 O 2 and FeSO 4 induced significant numbers of AP sites in calf thymus DNA, which were predominantly cleaved 5 * to the AP sites (75% of total aldehydic AP sites). An increase in the number of 5 * -AP sites was also detected in human cultured cells exposed to H 2 O 2 , and these 5 * -AP sites were persistent during the post-exposure period. b -Elimination by DNA b -polymerase efficiently excised 5 * -regular AP sites, but not 5 * -AP sites, in DNA from cells exposed to H 2 O 2 . These results suggest that 5 * -oxidized AP sites induced by ROS are not efficiently repaired by the mammalian short patch base excision repair pathway. Reactive oxygen species (ROS) 1 are generated continuously in cells during normal metabolic processes and by a number of exogenous agents, including ionizing radiation. ROS can react with cellular components such as proteins, lipids, and

Reactive oxygen species (ROS) 1 are generated continuously in cells during normal metabolic processes and by a number of exogenous agents, including ionizing radiation. ROS can react with cellular components such as proteins, lipids, and nucleic acids to induce DNA adducts such as 8-hydroxy-2Ј-deoxyguanosine (8-OH-dG) (1,2). It is believed that these oxidized bases are predominantly repaired by a base excision repair pathway (3). In this process, a bifunctional 8-OH-dG-DNA glycosylase with apurinic/apyrimidinic (AP) lyase such as 8-hydroxy-2Ј-deoxyguanine-DNA glycosylase cleaves the N-glycosylic bond between 8-hydroxyguanine and deoxyribose and incises immediately 3Ј to AP sites, leaving 3Ј-nicked AP sites (4,5). The 3Ј-AP sites generated by the DNA glycosylase are subsequently excised by class II AP endonuclease (3), resulting in a 3Ј-hydroxyl group and a 5Ј-phosphate group. Repair is completed by polymerase and ligase activity. Recently, it has been reported that mammalian cell extracts repair 8-OH-dG preferentially via single nucleotide replacement reactions (6,7). The contribution of nucleotide excision repair to the removal of 8-OH-dG was not significant in experiments using human cell extracts (6,7). These results indicate that base excision repair plays a central role in counteracting oxidized base lesions.
In addition to base damage in DNA, ROS also induce lesions by hydrogen abstraction of the deoxyribose, frequently producing oxidized AP sites as well as DNA strand breaks (8). AP sites are also generated spontaneously by chemical depurination of labile oxidized bases and enzymatically by DNA glycosylases as mentioned above. Hydrogen abstraction has been examined extensively for model deoxyribose and polynucleotides (9). Although Ͻ10% of the hydroxyl radicals attack sugar residues in single-stranded polynucleotides, it has been proposed that oxidized AP sites induced by ROS may be one of the major oxidative lesions in double-stranded DNA (3,9). These studies demonstrated that all hydrogen atoms of deoxyribose and ribose are potential targets for direct attack by oxygen radicals. In B-form duplex DNA, however, hydrogen atoms at the C-4Ј and C-5Ј positions of deoxyribose are the most accessible to ROS (10). ROS-induced sugar lesions and strand cleavage in genomic DNA are difficult to examine, mainly due to the large variety of products as well as their instability even at mild temperatures and neutral pH (11). Many oxidized sugars are very labile, as terminal sugar lesions tend to be modified spontaneously during experimental procedures.
We recently developed a sensitive aldehyde reactive probe slot-blot (ASB) assay to detect aldehydic AP sites in DNA, which can quantitate Ͻ1 AP site/10 6 nucleotides (12). Using this assay, we detected 50,000 -200,000 AP sites in mammalian cells under normal physiological conditions (13). Large numbers of AP sites were detected in brain, heart, and colon DNAs, which appear to be continuously exposed to higher levels of oxidative stress. These endogenous AP sites were predominantly cleaved 5Ј to the AP sites. Therefore, we hypothesized that oxidative stress directly induces 5Ј-nicked oxidized AP sites, which may contribute to a high steady-state level of AP sites in mammalian cells and tissues. To test this hypothesis, we have quantitated and characterized AP sites induced by ROS. We also have examined the repair efficiency of these AP sites in human cultured cells.

DNA Isolation from Calf Thymus
Thymus was harvested from a newborn Holstein calf and quickly frozen on dry ice. After thawing, the calf thymus was homogenized in lysis buffer (Gentra Systems, Inc.) with 10 mM 2,2,6,6-tetramethylpip-* This work was supported in part by NIEHS Superfund Basic Research Program Grant P42-ES05948 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

H 2 O 2 and FeSO 4 Treatment of Calf Thymus DNA
Calf thymus DNA isolated in our laboratory or commercially obtained calf thymus DNA (Sigma) pretreated with MX was incubated with H 2 O 2 and/or FeSO 4 in 10 mM Tris-HCl/KOH (pH 7.4) at 37°C for 10 min with or without TEMPO. The AP site assay was performed immediately after the Fenton reaction. For measurement of oxidative base lesions, the Fenton reaction was quenched by addition of 15 mM TEMPO, and the DNA was recovered by precipitation with cold ethanol. After washing the DNA pellet with 70% ethanol, DNA was resuspended in distilled water containing 1 mM TEMPO.

ASB Assay
The AP site assay was performed following a procedure slightly modified from that reported by Nakamura and Swenberg (13). Briefly, 8 g of DNA in 150 l of phosphate-buffered saline was incubated with 1 mM aldehyde reactive probe at 37°C for 10 min. After precipitation using cold ethanol, DNA was resuspended in TE buffer (10 mM Tris-HCl, pH 7.4, containing 1 mM EDTA). The DNA concentration was measured by a UV spectrophotometer, and the DNA solution was then prepared at 0.5 or 1 g/100 l of TE buffer. Heat-denatured DNA was immobilized on a nitrocellulose membrane (Hybond-C Super, Amersham Pharmacia Biotech). The nitrocellulose membrane was soaked with 5ϫ SSC and then baked in a vacuum oven for 30 min. The membrane was preincubated with 10 ml of Tris-HCl containing bovine serum albumin for 15 min and then incubated in the same solution containing streptavidin-conjugated horseradish peroxidase at room temperature for 45 min. After rinsing the nitrocellulose membrane, the enzymatic activity on the membrane was visualized by enhanced chemiluminescence reagents. The nitrocellulose filter was then exposed to x-ray film, and the developed film was analyzed using an Ultrascan XL scanning densitometer.

AP Site Cleavage Assay
The AP site cleavage assay was performed as described (13) with a slight modification.
Regular AP Site Assay-The number of total AP sites was measured by the ASB assay as described above.
5Ј-Cleavage Assay-Eight g of DNA and 145 units of Escherichia coli exonuclease III (Exo III) (New England Biolabs Inc.) were incubated in 135 l of 10 mM Tris-HCl/KOH (pH 7.5) containing 50 mM NaCl and 5 mM MgCl 2 at 37°C for 1 min and immediately analyzed by the ASB assay.
3Ј-Cleavage Assay-Eight g of DNA, 10 mM EDTA, and 100 mM putrescine were incubated in 135 l of 10 mM Tris-HCl/KOH at 37°C for 30 min and immediately analyzed by the ASB assay.
Detection of Residual AP Sites-Eight g of DNA and 145 units of exonuclease III in 110 1 of 10 mM Tris-HCl/KOH were incubated at 37°C for 1 min, immediately followed by addition of 0.1 volume of 100 mM EDTA. The sample was incubated with 100 mM putrescine in the reaction buffer at 37°C for 30 min, immediately followed by the ASB assay.

E. coli Endonuclease III-sensitive Site Assay
Oxidative pyrimidine bases are repaired by E. coli endonuclease III (End III), leaving AP sites on the DNA backbone (3). End III was kindly provided by Dr. Y. W. Kow (Emory University). The End III-sensitive site assay was performed as described (13).

8-OH-dG Assay
Quantitation of 8-OH-dG was based on an HPLC/electrochemical detection approach that was modified from a method previously described by Richter et al. (14). DNA was hydrolyzed enzymatically to deoxyribonucleosides using deoxyribonuclease I, spleen phosphodiesterase, snake venom phosphodiesterase, and alkaline phosphatase. The digest was separated by reversed-phase HPLC, and 8-OH-dG was quantitated using an electrochemical array detector (ESA). Electrochemical oxidation was monitored at 200, 300, 375, 450, 525, 600, 700, and 800 mV. The concentration of 8-OH-dG was normalized to the amount of DNA analyzed, as determined by UV absorbance.

Cell Culture
HeLa S3 cells were obtained as suspension cells from the Lineberger Comprehensive Cancer Center at the University of North Carolina at Chapel Hill. After centrifugation, cells were resuspended in 25 ml of Dulbecco's modified Eagle's medium/nutrient mixture F-12 (Life Technologies, Inc.) without serum (4 ϫ 10 5 cells/ml). The cultured cells were exposed to H 2 O 2 (Sigma) at 37°C for 15 min, immediately followed by centrifugation. After washing twice with cold phosphate-buffered saline, cell pellets were frozen and stored at Ϫ80°C until use. To test the repair efficiency of oxidative DNA lesions, cells washed in phosphatebuffered saline were further resuspended in 20 ml of Dulbecco's modified Eagle's medium/nutrient mixture F-12 with 10% bovine serum (Hyclone Laboratories) and cultured at 37°C for up to 6 h.

DNA Isolation from Cultured Cells
DNA isolation from cultured cells was performed using the PureGene DNA extraction kit (Gentra Systems, Inc.). Briefly, cell pellets were thawed and lysed in lysis buffer supplemented with 20 mM TEMPO. After protein precipitation with a protein precipitation solution, the DNA/RNA mixture in the supernatant was precipitated with isopropyl alcohol. The DNA/RNA pellet was resuspended in lysis buffer with 10 mM TEMPO and incubated with RNases T1 (50 units/ml) and A (100 mg/ml) at 37°C for 30 min, followed by protein and DNA precipitation. The DNA pellet was resuspended in sterilized distilled water with 1 mM TEMPO. The DNA solution was stored at Ϫ80°C for assays.

AP Site Repair Assay with Human ␤-pol
The AP site repair assay was performed by a procedure slightly modified from the AP site cleavage assay.
5Ј-Regular AP Sites-Eight g of DNA pretreated with heat/acid buffer (12)  Calf Thymus DNA Pre-exposed to the Fenton Reaction and DNA Isolated from Cells Exposed to H 2 O 2 -Eight g of DNA pre-exposed to the Fenton reaction was incubated with human ␤-pol or putrescine at different concentrations as described above and analyzed by the AP site assay.
Repair Efficiency-The efficiency of AP site repair was calculated by the reduction of AP sites by ␤-pol divided by the reduction of AP sites by putrescine.

AP Sites
Induced by the Fenton Reaction-One of the most significant oxygen radicals is the hydroxyl radical, which is generated by the reaction of reduced transition metals with H 2 O 2 via the Fenton reaction (15). To address whether oxygen radicals induced by the Fenton reaction directly generate AP sites in DNA, calf thymus DNA pretreated with MX was incubated with 10 M FeSO 4 with or without H 2 O 2 at 37°C for 10 min under neutral pH conditions. The number of AP sites in MX-pretreated calf thymus DNA increased following treatment with 10 M FeSO 4 and was further enhanced by H 2 O 2 (Fig. 1A). TEMPO, a radical-trapping reagent containing a nitrone group (9), is known to reduce the number of 8-OH-dGs in mammalian tissues at 1 mM (16). To investigate whether TEMPO inhibits AP site formation by the Fenton reaction, MX-pretreated calf thymus DNA was reacted with 10 M H 2 O 2 and FeSO 4 with or without TEMPO. TEMPO prevented AP site formation in a dose-dependent manner and completely protected DNA from AP site formation at concentrations of 10 mM (Fig. 1B).
Base Lesions Induced by the Fenton Reaction-To test whether AP sites are major oxidative lesions induced by the Fenton reaction, we compared the number of AP sites, End

5Ј-Nicked AP Sites Induced by Oxidative Stress
III-sensitive sites, and 8-OH-dGs in calf thymus DNA following the Fenton reaction. End III cleaves the N-glycosylic bond between deoxyribose and most oxidized pyrimidines, leaving 3Ј-cleaved AP sites (3). The number of End III-sensitive sites was calculated from the number of AP sites with End III treatment minus the number of AP sites with putrescine treatment. Since commercially available calf thymus DNA contains relatively large amounts of oxidative base lesions even without any treatment, we isolated DNA from fresh calf thymus with 10 mM TEMPO in this experiment. Whereas the steady-state level of AP sites was detected at 8 lesions/10 6 nucleotides in isolated calf thymus DNA, endogenous 8-OH-dG (detection limit: 1 lesion/10 7 dGs) was not detectable, and End III-sensitive sites (detection limit: 2 lesions/10 6 nucleotides) were around the detection limit. Using calf thymus DNA isolated in this laboratory, a combination of 10 M H 2 O 2 and 10 M FeSO 4 generated End III-sensitive sites and 8-OH-dG at 96 and 424 lesions/ 10 6 nucleotides, respectively (Fig. 1, C and D). These results indicated that the Fenton reaction induced by 10 M H 2 O 2 and 10 M FeSO 4 produced predominantly 8-OH-dG, followed by pyrimidine base lesions and AP sites (the ratio of 8-OH-dGs, End III-sensitive sites, and AP sites was ϳ9.7:2.2:1). In addition to AP sites, the formation of these oxidative base lesions by the Fenton reaction was almost completely protected by TEMPO at concentrations ranging from 10 to 20 mM (Fig. 1, C  and D). In the subsequent experiments, we isolated DNA from cultured cells with lysis buffer supplemented with 20 mM TEMPO to avoid artifactual formation of oxidative base lesions as well as AP sites.
AP Site Cleavage Assay for AP Sites Induced by the Fenton Reaction-ROS can induce sugar lesions directly by hydrogen abstraction of deoxyribose, resulting in AP sites as well as DNA strand breaks. AP sites are also generated spontaneously by chemical depurination of labile oxidized bases and unmodified bases and enzymatically by cleavage of the N-glycosylic bond between the sugar and modified bases. We recently developed an AP site cleavage assay to examine the site of cleavage at AP sites (13). To test whether the AP sites induced by ROS were 5Јor 3Ј-nicked or intact, the AP site cleavage assay was performed for MX-pretreated calf thymus DNA exposed to the Fenton reaction. MX-pretreated calf thymus DNA containing 5.9 Ϯ 0.9 (mean Ϯ S.D.) AP sites/10 6 nucleotides was incubated with 10 M H 2 O 2 and 10 M FeSO 4 for 10 min. The number of AP sites increased to 37 AP sites/10 6 nucleotides ( Fig. 2A). In this assay, we used Exo III as the class II AP endonuclease to identify 3Ј-cleavage of AP sites and putrescine to detect 5Јnicks. Immediately after the Fenton reaction, DNA was incubated with Exo III and/or putrescine, followed by the ASB assay. A single treatment of Exo III reduced the number of AP

5Ј-Nicked AP Sites Induced by Oxidative Stress
sites to 34 AP sites/10 6 nucleotides. This reduction was comparable to the data we published earlier (13) and may be due to the combination of enzymatic incision on the 5Ј-side by Exo III and nonspecific 3Ј-cleavage of AP sites during incubation with Exo III. In contrast, putrescine treatment resulted in significant reduction of the original number of AP sites. After incubation with Exo III followed by putrescine, the number of AP sites was reduced by 86% from the original number of AP sites in MX-pretreated calf thymus DNA exposed to the Fenton reaction. The summarized fractions of intact and cleaved AP sites and residual aldehydic lesions are shown in Fig. 2B (left). A major finding was that the AP site cleavage fractions induced by the Fenton reactions were different from those induced by heat/acid depurination (Fig. 2B, right) (13).
Oxidative DNA Lesions in Cells Exposed to H 2 O 2 -To evaluate AP site formation in cellular DNA by oxygen radicals, we exposed HeLa S3 cells to H 2 O 2 at 3-20 mM without serum at 37°C for 15 min. Toxicity of H 2 O 2 to cells was determined by the trypan blue exclusion assay. The viability of cells was Ͼ95% when the cultured cells were harvested. HeLa cells showed a slight increase in the number of AP sites following exposure to H 2 O 2 in a dose-dependent manner (Fig. 3A). The number of End III-sensitive sites and 8-OH-dGs was also increased by treatment with H 2 O 2 (the ratio of induction of 8-OH-dGs, End III-sensitive sites, and AP sites at 10 mM H 2 O 2 was ϳ0.6:1.4:1) (Fig. 3B and Table I). If we assume that H 2 O 2 exposure induces the Fenton reaction in cellular DNA, AP sites become among one of the major oxidative DNA lesions in cells. Furthermore, these data suggest that the repair of 8-OH-dG may be more efficient compared with the repair of AP sites and oxidized pyrimidine base lesions.
Repair Efficiency of Oxidative DNA Lesions in Cells Exposed to H 2 O 2 -To further investigate the repair efficiency of these oxidative DNA lesions, the cultured cells were post-incubated in fresh medium with 10% serum for up to 6 h after the exposure to 10 mM H 2 O 2 . 8-OH-dG was repaired ϳ83% within 6 h, and oxidized pyrimidines were repaired ϳ40% (Fig. 3C). In contrast, we detected no reduction in the number of AP sites after the 6-h repair period. The data further confirmed that AP sites induced by H 2 O 2 are more resistant to cellular excision repair pathways compared with oxidized bases.
Characterization of AP Sites in Cells Exposed to H 2 O 2 -The AP sites in cells exposed to H 2 O 2 were characterized using the AP site cleavage assay. The number of 5Ј-AP sites and residual aldehydic lesions increased 2-3 times compared with controls after exposure to 10 mM H 2 O 2 (Fig. 4). These lesions tended to accumulate during the repair period. In contrast, the combined fraction of 3Ј-nicked and intact AP sites did not increase in cells exposed to H 2 O 2 . To better understand the persistence of 5Ј-AP sites in cells after exposure to H 2 O 2 , we tested whether ␤-pol could excise 5Ј-AP sites introduced by oxidative stress. MXpretreated calf thymus DNA exposed either to the Fenton reaction or to heat/acid buffer followed by the incision 5Ј to AP sites by Exo III was incubated with ␤-pol or putrescine. The efficiency of AP site repair was calculated by the reduction of AP sites through ␤-elimination by ␤-pol divided by the reduction of AP sites through ␤-elimination by putrescine. ␤-pol efficiently excised 5Ј-regular AP sites at a concentration of 60 ng/67.2 l (Fig. 5). In contrast, 5Ј-AP sites directly introduced by ROS were less efficiently excised from the DNA backbone by ␤-pol. To address whether 5Ј-AP sites in cells exposed to H 2 O 2 are repaired like 5Ј-regular AP sites or 5Ј-AP sites/ROS, the DNA from HeLa cells exposed to 20 mM H 2 O 2 was incubated with ␤-pol at 60 ng/67.2 l, followed by the ASB assay. These 5Ј-AP sites were also excised less efficiently by ␤-pol compared with 5Ј-regular AP sites. We also detected a ␤-pol-resistant AP site fraction after a combined treatment of Exo III and ␤-pol (data not shown). DISCUSSION A large number of AP sites are produced continuously by spontaneous depurination in mammalian cells (12), leaving intact AP sites. Oxidative stress also induces labile ring-saturated pyrimidine adducts that result in intact AP sites by chemical depyrimidination. These intact AP sites are subsequently incised 5Ј to AP sites by class II AP endonuclease. Most   0.1, 0.1, Ͻ0.1, and Ͻ0.1).

5Ј-Nicked AP Sites Induced by Oxidative Stress
oxidative base lesions are also excised by bifunctional DNA glycosylases with AP lyase activity, which introduce 3Ј-AP sites. In addition, hydrogen abstraction directly induces both 5Ј-and 3Ј-nicked AP sites (8 -10). Therefore, a significant number of intact, 5Ј-and 3Ј-cleaved AP sites may be induced in cells under oxidative stress conditions. The present study demonstrated that oxidative stress predominantly induced 5Ј-cleaved AP sites in DNA in vitro and in vivo. Furthermore, 5Ј-nicked AP sites directly induced by ROS were efficiently released from the DNA backbone through ␤-elimination by putrescine, but not by ␤-pol (Fig. 5). In contrast, 5Ј-cleaved regular AP sites induced by heat/acid treatment followed by Exo III were efficiently excised by either putrescine or ␤-pol. These results indicate that the 5Ј-AP sites induced in vivo and in vitro by ROS are repaired differently than 5Ј-regular AP sites. In Bform duplex DNA, ROS most likely induce sugar lesions directly by abstraction of hydrogen atoms at the C-4Ј or C-5Ј position of deoxyribose (9 -11). Under aerobic conditions, hydrogen abstraction at C-4Ј results in DNA cleavage to produce the 3Ј-phosphoglycolate terminus, the base propenal, and the 5Ј-monoester phosphate terminus. In contrast, under anaerobic conditions, hydroxyl radicals induce C-4Ј-hydroxylated abasic sites with an equilibrium between C-4Ј-oxidized aldehydic AP sites. Hydrogen abstraction at C-5Ј has also been proposed to produce 5Ј-cleaved aldehydic AP sites under aerobic conditions (8,11). These AP sites with an aldehydic moiety should be a substrate for ␤-elimination and are detectable by the ASB assay. Therefore, we hypothesize that the 5Ј-AP sites induced by ROS represent oxidized AP sites such as 5Ј-cleaved C-4Ј-or C-5Ј-oxidized AP sites.
Whereas putrescine excised 5Ј-regular AP sites at 100 mM, ␤-pol efficiently cleaved 3Ј to the 5Ј-AP sites at 80 nM. These results are in good agreement with the differences in the efficiency of cleavage of intact and 5Ј-nicked AP sites by putrescine (17) and ␤-pol (18). A 49-base pair oligonucleotide duplex DNA (20 nM) with a single intact or 5Ј-incised AP site (10,000 lesions/ 10 6 nucleotides) was cleaved ϳ50% by treatment with 200 and 5 nM ␤-pol, respectively, for 15 min at 37°C (18). In contrast, we utilized long genomic DNA containing a much lower frequency of AP sites (20 lesions/10 6 nucleotides), which appears to be a more biologically relevant frequency based on the number of endogenous AP sites (13). Interestingly, ␤-pol excised the 5Ј-dRp (deoxyribose phosphate) moiety in long genomic DNA as efficiently as those in oligonucleotides at similar concentrations. These data suggest that ␤-pol efficiently recognizes and excises 5Ј-cleaved regular AP sites under physiologically relevant conditions. Aldehydic AP sites were relatively minor oxidative DNA lesions generated by the Fenton reaction in in vitro experiments, whereas these AP sites became one of the major oxidative lesions in genomic DNA from cells exposed to H 2 O 2 . Furthermore, 5Ј-cleaved AP sites were more persistent compared with oxidative base lesions in cultured cells after exposure to oxidative stress. As described above, putrescine, but not ␤-pol, efficiently excised 5Ј-AP sites induced by ROS. These data indicate that 5Ј-AP sites induced by oxidative stress are not repaired efficiently by cellular excision repair pathways. However, the results regarding the efficiency of repair by putrescine indicate that the lesions are potentially repairable through ␤-elimination by an amine moiety. It has been proposed that the amine residue Lys 72 in ␤-pol forms a Schiff base intermediate with the AP site and cleaves 3Ј to the AP site (19). The difference in dRp lyase activity between putrescine and ␤-pol for 5Ј-AP sites induced by ROS suggests that the amine moiety of Lys 72 in ␤-pol may not efficiently reach the aldehydic moiety of 5Ј-nicked oxidized AP sites. This inefficiency might be explained as follows: 1) ␤-pol inefficiently recognizes these 5Јaldehydic AP sites induced by ROS; or 2) after ␤-pol recognizes 5Ј-oxidized AP sites, Lys 72 in ␤-pol does not reach the aldehydic moiety of these AP sites due to structural difference of oxidized AP sites. Although 5Ј-nicked C-4Ј-oxidized AP sites induced by bleomycin followed by human AP endonuclease are excised by ␤-pol (20), the excision efficiencies of ␤-pol for 5Ј-nicked C-4Ј-or C-5Ј-oxidized AP sites directly induced by oxidative stress are still unknown. We hypothesize that 5Ј-oxidized AP sites directly induced by ROS may be repaired by the Flap endonuclease-1-dependent long patch base excision pathway. In our previous study, we found large numbers of endogenous 5Јnicked AP sites in rat tissues and human liver (13). Interestingly, the cleavage fractions of AP sites induced by the Fenton reaction are similar to those of endogenous AP sites in rat and human tissues. Although it has been believed that AP sites are efficiently repaired, oxidized AP sites are not excised as efficiently as regular AP sites in cells. Therefore, we believe that endogenous AP sites arise primarily from oxidized AP sites rather than from regular AP sites. We suggest that the high steady-state level of AP sites might be due to an inefficient short patch base excision repair pathway by ␤-pol.
It was originally demonstrated that ␤-pol required MgCl 2 for dRp lyase activity (21). However, Prasad et al. (18) proposed that ␤-elimination by ␤-pol is Mg 2ϩ -independent based on inhibition of dRp lyase activity by EDTA and restoration of dRp lyase function by supplementing with NaCl. Our results also showed that Ca 2ϩ , instead of Mg 2ϩ , quite efficiently excised dRp moieties from the DNA backbone. These data indicate that Mg 2ϩ in not an essential cofactor for the dRp lyase activity of ␤-pol. The human enzymes counteracting most oxidative base lesions are bifunctional DNA glycosylases such as human 8-hydroxy-2Ј-deoxyguanine-DNA glycosylase and human endonuclease III (22), leaving 3Ј-AP sites after releasing modified bases. Subsequently, class II AP endonuclease removes the 3Ј-blocked termini by 3Ј-phosphoesterase activity to create a 3Ј-OH group for DNA repair synthesis (3). Although ROS induce significant numbers of oxidized base adducts, there was no accumulation of the combined fraction of intact and 3Јnicked AP sites in cellular DNA after exposure to H 2 O 2 . These data suggest that class II AP endonuclease efficiently excises a large number of 3Ј-cleaved AP sites. In in vitro repair assays, 8-OH-dG and oxidized pyrimidines were repaired mainly by a short patch base excision repair pathway (6,7,23). However, the DNA repair synthesis at 8-OH-dG was less efficient than that at regular AP sites (7). Therefore, it has been proposed that the first three processes from base release to excision of 3Ј-AP sites may be rate-limiting steps. Our data suggest that 3Ј-phosphoesterase activity to repair 3Ј-AP sites is not ratelimiting in base excision repair. Based on these results, the excision of modified bases may be one of the rate-determining processes in the 8-OH-dG base excision repair pathway. Furthermore, in human cultured cells, oxidative stress induced AP endonuclease and rendered cells resistant to oxidative stress (26). These results also raised the possibility that 3Ј-AP sites generated by bifunctional DNA glycosylases may induce AP endonuclease. In addition to 3Ј-AP sites, ROS also induced other 3Ј-phosphate lesions, including 3Ј-phosphoglycolate. These 3Ј-blocked termini might be one of the reasons for AP endonuclease induction in cells under oxidative stress conditions. Although the Fenton reaction directly induced a significant number of intact AP sites in the in vitro system, the number of intact AP sites was not increased in cells exposed to H 2 O 2 . Both regular AP sites and C-4Ј-oxidized AP sites without strand breaks directly induced by bleomycin are repaired by an interaction of AP endonuclease and ␤-pol in vitro using oligonucleotides (20,27). Based on these data and our experiments, the regular and oxidized aldehydic AP sites with no cleavage on either side induced by ROS appear to be efficiently repaired in cells through a base excision repair pathway.
A high concentration of TEMPO almost completely protected the formation of AP sites, End III-sensitive sites, and 8-OH-dG induced by a high level of oxidative stress. Furthermore, DNA extracted from fresh calf thymus also showed very low amounts of 8-OH-dG (Ͻ1 lesion/10 7 nucleotides). In contrast, the range of steady-state levels of 8-OH-dG measured by HPLC/electrochemical detection has varied from 4 to 800 lesions/10 7 nucleotides in mammalian cells and tissues (28). There are many factors that artifactually induce oxidative DNA lesions during DNA extraction (24). The trapping of free radicals by TEMPO appears to be quite efficient for preventing artifactual DNA damage from oxidative stress. Therefore, the current DNA extraction method using a high concentration of TEMPO minimizes the artifactual induction of oxidative lesions during DNA extraction. In the present experiment, 10 mM H 2 O 2 increased the number of 8-OH-dGs by a factor of Ͼ30 over the control. These data indicate that reduction of artifactual oxidative DNA lesions will also allow us to more accurately determine dose-response relationships as well as the repair kinetics of these lesions after oxidative stress.
Further studies are needed to understand the biological consequences of 5Ј-AP sites persisting in cells under normal physiological conditions as well as after oxidative stress. Although H 2 O 2 killed HeLa cells within 24 h at 20 mM, a limited number of cells survived after exposure to 10 mM H 2 O 2 and started growing within 1-2 days (data not shown). These results suggest that 5Ј-oxidized AP sites are repairable by cellular DNA repair pathways. Recently, Jackson et al. (25) demonstrated that oxidative stress, but not UV radiation or methylating agent, induces frameshift mutations in microsatellite DNA. They proposed that a common lesion such as a strand break is more likely to contribute to genomic instability than the alteration of a specific nucleotide. It is possible that 5Ј-oxidized AP sites might be involved in the frameshift mutation in microsatellite DNA. To date, it has generally been believed that AP sites are repaired very efficiently in genomic DNA; however, the high steady-state level of 5Ј-nicked AP sites as well as persistent 5Ј-cleaved AP sites after oxidative stress suggest that some fraction of AP sites may not be efficiently repaired by the mammalian excision repair pathway.