Base and Nucleotide Excision Repair of Oxidatively Generated Guanine Lesions in DNA*

The well known biomarker of oxidative stress, 8-oxo-7,8-dihydroguanine, is more susceptible to further oxidation than the parent guanine base and can be oxidatively transformed to the genotoxic spiroiminodihydantoin (Sp) and 5-guanidinohydantoin (Gh) lesions. Incubation of 135-mer duplexes with single Sp or Gh lesions in human cell extracts yields a characteristic nucleotide excision repair (NER)-induced ladder of short dual incision oligonucleotide fragments in addition to base excision repair (BER) incision products. The ladders were not observed when NER was inhibited either by mouse monoclonal antibody (5F12) to human XPA or in XPC−/− fibroblast cell extracts. However, normal NER activity appeared when the XPC−/− cell extracts were complemented with XPC-RAD23B proteins. The Sp and Gh lesions are excellent substrates of both BER and NER. In contrast, 5-guanidino-4-nitroimidazole, a product of the oxidation of guanine in DNA by peroxynitrite, is an excellent substrate of BER only. In the case of mouse embryonic fibroblasts, BER of the Sp lesion is strongly reduced in NEIL1−/− relative to NEIL1+/+ extracts. In summary, in human cell extracts, BER and NER activities co-exist and excise Gh and Sp DNA lesions, suggesting that the relative NER/BER product ratios may depend on competitive BER and NER protein binding to these lesions.

The overproduction of free radical intermediates and electrophiles by macrophages and neutrophils activated during the inflammatory response in chronically inflamed tissues induces persistent damage to cellular DNA (1). If not properly repaired, this DNA damage can increase levels of mutations and genomic instability and eventually can lead to the initiation of human cancers (2)(3)(4). The primary target of oxidatively generated DNA damage is guanine (5), the most easily oxidizable natural nucleic acid base (6). The best known oxidation product of gua-nine is 8-oxo-7,8-dihydroguanine (8-oxoG), 3 which is ubiquitous in cellular DNA and is used widely as a biomarker of oxidative stress (7). The 8-oxoG lesion is genotoxic, and failure to remove 8-oxoG before replication induces G:C 3 T:A transversion mutations (8). This lesion is even more easily oxidized than the parent base guanine (9), and its further oxidation by various oxidizing agents, including peroxynitrite, leads to the formation of stereoisomeric spiroiminodihydantoin (Sp) and 5-guanidinohydantoin (Gh) lesions (10 -15). In the case of guanine oxidation by peroxynitrite, the 5-guanidino-4-nitroimidazole (NIm) lesion (16 -18) is also produced and can serve as a biomarker of inflammation-related oxidation mechanisms (1). The NIm lesions together with the easily depurinated 8-nitroguanine, are typical products of guanine damage in calf thymus DNA induced by reactions with nitrosoperoxycarbonate (16,19) derived from the combination of carbon dioxide and peroxynitrite (20). The structures of these oxidatively generated guanine lesions are shown in Fig. 1A.
The chiral carbons in the Gh and Sp nucleobases give rise to a pair of R and S diastereomers. Oligonucleotides that contain single, site-specifically inserted Gh or Sp lesions can be separated and purified by anion-exchange HPLC methods (21). In aqueous solutions, the Gh diastereomers are easily interconvertible, and it is, therefore, not feasible to study their characteristics individually (22). In contrast, the Sp-S and Sp-R diastereomers are chemically stable and can be purified and studied individually (10,23). NMR structural studies in solution indicate that both stereoisomerically distinct Sp lesions (24) perturb adjacent base pairs and thus thermodynamically destabilize oligonucleotide duplexes (25). Molecular modeling studies indicate that the NIm lesions adopt flexible and multiple ringopened structures with the nitro and guanidino groups, providing multiple hydrogen bonding possibilities (26).
The Gh and Sp lesions are highly mutagenic, resulting in G 3 C and G 3 T transversion mutations (27). The accumulation of Sp lesions was detected in Nei-deficient Escherichia coli after treatment of these cells with chromate (28) and were also detected in both the liver and colon tissues of Rag2 Ϫ/Ϫ mice at levels ϳ100 times lower than those of 8-oxoG (29). Although the cellular levels of hydantoin lesions are low, these lesions can contribute to the malignant transformations of cells because they are at least 1 order of magnitude more mutagenic than 8-oxoG (27). Mangerich et al. (29) found a correlation between the cellular levels of Sp lesions and the progression of disease in a mouse model of inflammation-induced colon cancer.
Here, we demonstrate that the incubation of site-specifically modified oligonucleotide duplexes containing single Sp or Gh lesions are excised by the human NER and BER systems when incubated in cell-free HeLa S3 cell extracts, whereas NIm is resistant to NER but is a substrate of BER.
The HeLa S3 cells were purchased from American Type Culture Collection (Manassas, VA). SV40-transformed XP4PA-SV-EB fibroblasts deficient in XPC (GM15983) and the fully XPC-competent (GM16248) fibroblasts also derived from XP4PA-SV-EB (GM15983) by stable transfection with XPC-cDNA using the plasmid pXPC3 were purchased from the Coriell Institute for Medical Research (Camden, NJ). The cells were maintained according to the manufacturer protocols. Mouse monoclonal antibody (5F12) against human XPA (ab65963, Abcam, Cambridge, MA) was used to inhibit the NER activities. The full-length (amino acid residues 1-940) and truncated forms of XPC-RAD23B (the latter missing residues 1-155 only but active in binding to DNA (45)) were expressed in SF-9 insect cells as described (46). Another sample of fulllength XPC-RAD23B was kindly provided by Dr. Orlando Schärer (Stony Brook University, Stony Brook, NY). Murine NEIL1 enzyme containing C-terminal His 6 tag was cloned, expressed, and purified as described previously (32,47). Active site concentrations of the NEIL1 samples were determined using the Sp-S duplex as a substrate (37,48).
The full details of the DNA repair experiments in extracts from HeLa S3 cells are described in detail elsewhere (50,51). Briefly, the 50-l reaction mixtures contained an ϳ1 nM concentration of the 32 P-internally labeled 135-bp duplexes with or without a lesion or adduct in a standard buffer solution (25 mM HEPES-KOH, pH 7.9, 0.1 M KCl, 12 mM MgCl 2 , 1 mM EDTA, 17% glycerol, 2 mM DTT, and 2 mM ATP) containing 20 -30 g of protein from the cell extracts. The reactions were stopped after the desired incubation time by the addition of SDS to a concentration of 0.5% proteinase K (0.4 gm/ml), and this mixture was incubated for 1 h at 45°C. The oligonucleotide excision products and intact 135-bp DNA duplexes were isolated by ethanol precipitation, subjected to denaturing 12% polyacrylamide gel electrophoresis, and analyzed by gel autoradiography using a Typhoon FLA 9000 laser scanner, and densitometric traces were generated from the autoradiographs using the ImageQuant software.

Results and Discussion
BER and NER Activities Are Observable in the Same Human Cell Extract DNA Repair Assays-The hallmark of successful NER activity in cell extracts is the appearance of 32 P-labeled oligonucleotides that are generally 24 -32 nucleotides in length that contain the lesions (52,53). In contrast, glycosylases cleave the N-glycosidic bond of the damaged base, thus resulting in the formation of an apurinic/apyrimidinic (AP) site in the damaged strand. The AP site is cleaved by lyase action, thus giving rise to a strand break at the site of the lesions (54) that results in the appearance of 67-mer incision products.
Typical results of the DNA repair assays in HeLa cell extracts obtained with 135-mer oligonucleotide duplexes containing single Sp-S, Gh, or NIm lesions are depicted in the autoradiograph of denaturing polyacrylamide gels shown in Fig. 2A. The electrophoretic migration distances of oligonucleotide size markers and their lengths (expressed in terms of numbers of nucleotides per strand) are shown in lane 1 ( Fig. 2A). The results shown in this gel were obtained from experiments that were conducted with aliquots from the same cell extract and at the same time; it is, therefore, possible to evaluate the relative incision activities of different substrates from such gels. After incubation of the 135-bp duplexes with the B[a]P-dG adducts for 15 and 30 min, the characteristic ladders of NER dual incision products of ϳ24 -30 nucleotides in lengths are clearly visible in the case of the bulky B[a]P-N 2 -dG adducts ( Fig. 2A, lanes 5 and 6). These observations demonstrate that this cell extract was NER-active (42,43). Using the same cell extract, a similar series but somewhat shorter oligonucleotide bands were observed in the case of the non-bulky Gh (lanes 9 -11) and Sp-S lesions (lanes 19 -21). These bands are also attributed to NER as shown in more detail below. However, analogous bands were not observed in the case of NIm (lanes [13][14][15][16], which thus appears to be NER-resistant. The repair efficiencies using duplexes containing the other Sp diastereomer, Sp-R, yielded qualitatively similar results (Fig. 3A).
The 135-bp duplexes with Sp, Gh, and NIm lesions are also incised at the sites of the lesions, thus generating 32 P-labeled 67-mer oligonucleotide fragments (Figs. 2A and 3A) that are attributed to BER activity (see below). However, there are no analogous 67-mer bands in the case of the bulky B[a]P-dGmodified duplexes that are not substrates of BER ( Fig. 2A). The zero incubation time points in lanes 2, 7, 12, and 17 ( Fig. 2A) represent control experiments (incubations in cell extracts were terminated within the first 20 -25 s after manual mixing, and the samples were subjected to the same post-incubation treatment as all of the other samples). It is evident that the extent of observable incisions is negligible. A full set of controls with unmodified 135-mer duplexes incubated in cell extracts for different time intervals up to 60 min is shown in Fig. 4, lanes [2][3][4][5]. Neither BER nor NER bands were observed in any of these control experiments.
Examples of densitometry traces derived from Fig. 2A at the 30-min incubation time points (lanes 11, 21, and 16 for the Gh, Sp-S, and NIm samples, respectively) are depicted in Figs. 2, B-D. The relative yields of the 67-mer BER and the shorter NER products can be determined separately as a percentage of the initial 135-mer DNA strands incised. Because the repair activities are known to vary in different cell extract preparations, the BER and NER yields were normalized with respect to the Sp-S yield that was assigned an arbitrary value of 100 in each of the three independent experiments (the actual fractions of duplexes incised by the NER mechanisms were in the range of 7-25%). The averaged relative NER and BER yields derived from these three independent experiments in different cell extracts are summarized in Fig. 2E. In the case of the NER yields, the standard deviations indicate that there are no significant differences between the Sp, Gh, and B[a]P-dG lesions. In each of the individual cell extract experiments ( Fig. 2A), the NER yields in the case of the B[a]P-dG adduct, Sp-S, and Gh lesions did not vary by ϾϮ20%. Similar variations were noted in the case of the BER yields observed with the Gh, Sp-S, and NIm lesions (Fig. 2, B-D 6). The difference in mobilities of the single-stranded oligonucleotides with Gh and Sp-S lesions relative to those of the B[a]P-dG-modified excision products were further investigated. We synthesized 23-mer oligonucleotide strands containing either Sp or B[a]P-dG lesions. The 23-bp duplexes constructed from the 32 P-labeled B[a]P-dG and Sp-S sequences and the corresponding complementary strands were mixed with an excess of the unlabeled 135-bp duplexes, incubated with HeLa cell extracts for fixed periods of time (0, 30, and 60 min), and analyzed by denaturing polyacrylamide gel electrophoresis (Fig. 4, lanes 7-15). The 23-mer oligonucleotides with Sp-S lesions (lanes 9, 12, and 15) migrate approximately like 21-mer unmodified oligonucleotides (lane 6). This difference in mobilities (Fig. 4) is too small to account for the observed Sp and Gh NER fragments that are as short as 18 -19 nucleotides (Fig. 2). However, Sugasawa et al. (55) showed earlier that the NER excision products of DNA with the UV light-induced cyclobutane thymine-thymine dimers and 6-4 pyrimidine-pyrimidone photoproducts exhibit similar patterns of dual incision products that are Ͻ25 nucleotides in lengths. Huang et al. (56,57) reported earlier that in cell extracts NER excision products are progressively degraded to shorter oligonucleotides as the incubation time is increased, thus shifting the group of NER dual incision products farther down the gel autoradiographs. Therefore, the faster-migrating ladder of bands in the case of the dual incision products ( Fig. 2A) observed in the case of the Gh and Sp-S lesions (lanes 10 and 11 and 20 and 21) relative to the slower mobility bands with B[a]P-dG adducts (lanes 5 and 6) can be attributed to (i) an intrinsic faster electrophoretic mobility of the Gh-and Sp-modified short oligonucleotides in comparison with the same length B[a]P-dG fragments, and (ii) the partial degradation of the excised oligonucleotides by nonspecific nucleases (56, 57) that gradually diminish the size of the NER dual incision products. The oligonucleotide dual incision products in the case of the bulky B[a]P-dG products appear to be more resistant to nuclease digestion. The resistance of B[a]P-dG adducts to digestion catalyzed by exonucleases was described earlier (58).
The Appearance of 67-Mer Incision Products in Mouse Embryonic Fibroblast Is Consistent with BER Activities-To further support the conclusion that the 67-mers are indeed BER excision fragments, the 135-bp Sp-S and Sp-R duplexes were incubated with extracts from MEF NEIL1 ϩ/ϩ and NEIL1 Ϫ/Ϫ cells (Fig. 5). These wild type and NEIL1 knock-out cells were selected because it is well established that the Sp lesions are substrates of NEIL1 in vitro (32). In Fig. 5 it is shown that the excision of Sp lesions is significantly reduced in NEIL1 Ϫ/Ϫ cells (lanes 3, 6, and 9) relative to wild type cells (lanes 4, 7, and 10) in three different sequence contexts. Densitometric analysis shows that the reduction in activity is ϳ18-fold in these MEF NEIL1 Ϫ/Ϫ cells. The Sp and Gh are good substrates of the Nei family glycosylases NEIL2 and NEIL3 in vitro (32,39). It is, therefore, likely that the residual BER activity observed in extracts from NEIL1 Ϫ/Ϫ cells (lanes 3, 6, and 9, Fig. 5) might be associated with the presence of DNA glycosylases other than NEIIL1. We note that these particular MEF cell extracts did not show any NER activity, although very weak activities were observed in some samples. By contrast, extracts from human fibroblasts do exhibit NER activities (see below), and thus studies of NER in MEF extracts were not further pursued.
Time Dependence of NER and BER Excision Activities-Examples of the BER and NER kinetics at different 135-bp Sp-S substrate concentrations are depicted in Fig. 6, A and C. In these figures the yields of excision products calculated by integration of the histograms derived from the denaturing gel autoradiographs (Fig. 2, B-D) are plotted as a function of incubation time in HeLa cell extracts. The NER kinetics are linear up to at least 45 min (Fig. 6C) but level off at higher incubation times (data not shown). In contrast, the BER kinetics shown in Fig. 6A are nonlinear in HeLa cell extracts. The yield of 67-mer excision products rises rapidly within the first 1-2 min followed by a much slower linear phase. These biphasic kinetics resemble classical base excision reaction kinetics that exhibit a rapid  burst phase due to single turnover reactions followed by a linear steady-state phase that is determined by the rate of release of DNA BER incision products (37,48,59,60).
It has been shown that BER activities in cell-free extracts from HeLa cells are associated with endonuclease VIII-like proteins (NEIL1, NEIL2, and NEIL3) (61-63) that cleave the N-glycosyl bonds of Sp/Gh lesions, thus generating abasic sites (32, 37, 39, 64 -67). NEIL1 and NEIL2 are bifunctional DNA glycosylases that further cleave the abasic sites by the intrinsic lyase activity that results in the formation of 5Ј-side fragments with 3Ј-phosphate groups at the ends (via ␤,␦ elimination) (54,63). Therefore, the faster-migrating 67-mer oligonucleotide BER incision products (lanes 8 -11, 13-16, and 18 -21, Fig. 2A) can be assigned to the 5Ј-side cleavage fragments with 3Ј-phosphate ends (32,37,38). In HeLa cell extracts, the 3Ј-phosphate groups are removed by the phosphatase activity of polynucleotide kinases to form the slower migrating fragments with OH groups at the 3Ј-ends (68 -70). Indeed, these slower migrating 67-mer fragments with 3Ј-OH ends (lanes 9 -11, 14 -16, and 19 -21, Fig. 2A) were observed at longer incubation times (4 -30 min), which was attributed to the pronounced phosphatase activity of these particular HeLa cell extracts (68). These 5Ј-side cleavage fragments with 3Ј-OH groups are not detected in BER experiments with Sp and Gh lesions conducted with purified NEIL1 and NEIL2 in vitro (32,37,38). However, NEIL3 does not seem to play a role in the BER activity shown in Fig. 2A. NEIL3 is a monofunctional glycosylase without intrinsic lyase activity and an AP endonuclease (APE1) is typically recruited to cleave the abasic site (63). The latter mechanism generates a 5Ј-side fragment formed that contains an ␣,␤-unsaturated aldehyde group (␤-elimination) at the 3Ј-ends. Such a 5Ј-fragment is not evident in our cell extract experiments ( Fig. 2A) because its mobility would be even slower than that of a fragment with a 3Ј-OH group (70).
To gain further insights into the BER and NER processes, we analyzed the reaction kinetics at different concentrations of the Sp-S duplexes (Fig. 6). The biphasic kinetics (Fig. 6A) can be represented by the standard scheme used to describe kinetics of the base excision repair of Sp and Gh lesions in vitro by hNEIL1 (37), 8-oxoG by MutY (48), and hOGG1 (59), and thymine glycol by hNTH1 (60):

Reaction1
Under multiple turnover conditions the concentration of the DNA substrate exceeds that of the enzyme, ([DNA S ] Ͼ [E BER ] 0 ), and the yield of BER products is characterized by a fast  burst of product formation (amplitude A 0 ) with the observed rate constant (k obs ) followed by a slower, steady-state linear phase with the slope (k ss ) (Fig. 6A). The kinetics of the appearance of DNA cleavage products ([DNA P ] t ) as a function of time (t) is described by the classic burst equation, An increase in the concentration of the DNA substrate leads to increases of the values of A 0 and k ss , and in the high concentration limit A 0 ϭ [E BER ] 0 , k obs ϭ k BER , and k ss ϭ k P [E BER ] 0 , where k BER is the rate constant of the chemical reaction in the enzyme-(DNA substrate) complex (DNA S ⅐E BER ), whereas k P is the rate constant of product release from the enzyme-(DNA product) complex, (DNA P ⅐E BER ) (48). The overall BER product yield as a function of DNA concentration in HeLa cell extract is consistent with these predictions (Fig. 6, A and B). The results shown in Fig. 6B indicate that the burst amplitude does not significantly increase at DNA substrate concentrations above [DNA S ] Ͼ 10 nM, which provides an approximate value of [E BER ] 0 ϭ 0.15-0.19 nM BER enzyme concentration in this example. In terms of this scheme, the active concentrations of BER enzymes in HeLa cell extract are almost the same for all three BER substrates studied (Sp-S, Gh, and NIm) (data not shown). The values of the burst phase rate constant, k BER , are too high to be kinetically resolvable under our experimental conditions that are based on manual mixing procedures; using rapid quench flow techniques, values of k BER ϳ 100 -180 min Ϫ1 were determined for the excision of Gh and Sp lesions by hNEIL1 in vitro (37).
The NER dual incision kinetics are linear up to at least 45 min (Fig. 6C) as shown earlier in the case of the cyclobutane thymine-thymine dimers and 6 -4 pyrimidine-pyrimidone photoproducts as well as other DNA lesions in cell extracts (50) and in cells (71). Increasing the concentration of the DNA substrate enhances the observed rate of NER product formation, V obs ϭ ⌬͓DNA p ͔/⌬t (Eq. 2) calculated from the linear portions of the product concentrations (Fig. 6C). Under these conditions, the value of V obs became independent of the DNA substrate concentration and attained the maximum rate V max ϭ k NER [E NER ] 0 , which is ϳ18 pM⅐min Ϫ1 at [DNA S ] Ͼ 10 nM) in the particular HeLa cell extract described in this example (Fig. 6D).
XPC-RAD23B Complementation Experiments and Impact of a Monoclonal XPA Antibody on NER Activity-The eukaryotic and heterodimeric protein XPC-RAD23B binds stably to a variety of DNA lesions and is a critical factor in NER (72,73). The XPC-RAD23B binding step initiates the cascade of NER steps that ultimately result in the appearance of the dual incision fragments (74,75). In some cases DDB is required for the recruitment of XPC. This effect is well established in whole cells (for review, see Ref. 74). However, in vitro, a role of DDB has been proposed for specific lesions like cyclobutane pyrimidine dimers under some conditions but not in others (74). The importance of XPC-RAD23B is supported by the observation that the 20 -27-mer products generated by NER activity are absent when the 135-bp duplexes containing Sp-S (lane 5, Fig.  7) or Sp-R (Fig. 3) are incubated in whole cell extracts from XPC Ϫ/Ϫ fibroblasts that are not expressing XPC. In contrast, 67-mer fragments produced by BER activity are still observed, because XPC-deficiency affects only the NER activity. The extracts from HeLa cells and XPC ϩ/ϩ fibroblasts used as a positive control (lanes 3 and 4, Fig. 7) demonstrate a normal NER activity (as in Figs. 2 and 3). The NER activity of cell extracts from XPC Ϫ/Ϫ fibroblasts can be restored by the addition of XPC-RAD23B. Indeed, dual incision products are clearly observed after the addition of the XPC-RAD23B heterodimer assembled from full-length XPC (lanes 7 and 8, Fig. 7) but not from the truncated form of XPC that is missing amino acids 1-155 in the N-terminal region (lane 6, Fig. 7).
Additional evidence that the appearance of the 20 -27-mer oligonucleotide ladder is due to NER activity, which causes the observed dual incision activities of the Sp-and Gh-containing 135-mer duplexes, are the observations that the latter disappears in the presence of low concentrations of monoclonal antibody against XPA. This protein is a key component of the NER machinery (76), and inhibition of XPA activity by anti-XPA antibodies has been widely used for validating the existence of NER mechanisms (77)(78)(79)(80). For example, the mouse monoclonal anti-XPA (5F12) antibody strongly inhibits NER of oxidatively generated 5Ј,8-cyclopurine lesions (78) and the 1,3-intrastrand d(GpTpG)-cisplatin cross-linked lesions (79). Indeed, the addition of anti-XPA (5F12) to the 135-bp Sp-S duplexes selectively inhibits the appearance of the dual incision products, whereas the 67-mer incision fragments produced by the BER activity remain at the same level (lanes 10 -12, Fig. 8A). The B[a]P-dG adducts embedded in 135-mer duplexes are well established NER substrates (42,43) and provide a positive control for validating the existence of NER mechanism in these experiments (lanes 4 -6, Fig. 8A). The yields of dual incision products elicited either by the B[a]P-dG adducts (Fig. 8B) or by Sp-S lesions (Fig. 8C) decrease as the concentration of the XPA antibody is increased. The complete inhibition of the dual incision products at anti-XPA antibody concentrations of ϳ 0.2-0.3 nM suggests that the XPA concentration in these HeLa cell extracts is in that range as well (Fig. 8, B and C).
It is remarkable that human NER and BER proteins can function in parallel in repairing Sp and Gh DNA lesions in human HeLa cell extract experiments. These observations suggest that these two pathways complement one another to maximize the efficiency of removal of these DNA lesions and may operate in whole cells as well.
Competition between BER and NER Pathways-Our working hypothesis is that the observation of parallel BER and NER activities in cell extracts is due to competitive BER and NER protein binding to the Gh and Sp substrates. According to this  hypothesis, the addition of increasing amounts of an exogenous BER glycosylase should enhance the yield of BER incisions, whereas the NER activity should decrease. This hypothesis is consistent with the competition experiments depicted in Fig. 9. Increasing the concentrations of the bifunctional BER glycosylase NEILI to cell extracts containing 135-bp duplexes with single Sp-S lesions decreased the NER product yields; the levels of the NER dual incision products became negligible at the highest concentrations of the BER glycosylase employed in these experiments (Fig. 9E). By contrast, the relative yield of the 67-nucleotide BER cleavage products increased markedly with increasing concentrations of NEIL1 (Fig. 9D). These results are attributed to a competition between NEIL1 and the initial NER DNA lesion recognition factor XPC-RAD23B for binding to the same DNA lesion that increasingly favors NEIL1 as its concentration is increased.
Mechanistic Considerations-An interesting question is why the NIm lesion is resistant to NER, whereas the structurally similar Gh lesion is an excellent NER substrate. The two lesions are distinguished from one another only by the replacement of the keto group in Gh by an -NO 2 group in NIm (Fig. 1). Molecular modeling studies indicate that the NIm lesions adopt flexible and multiple ring-opened structures with the nitro and guanidino groups providing multiple hydrogen bonding possibilities (26). By contrast, Gh can exist in the forms of two slowly interconverting R and S stereoisomers (81,82). On the other hand, the stereoisomeric Sp structures exist as stable and identifiable R and S stereoisomers (24,83). All three lesions, NIm, Gh, and Sp, cause significant thermodynamic destabilization of DNA duplexes (21,25,26,(82)(83)(84)(85)(86)(87), which is a common feature of DNA substrates that are recognized by the NER machinery. Thus, the Sp and Gh lesions are recognized by the prokaryotic UvrABC nuclease (40) as well as by the eukaryotic NER machinery in human cell extracts as demonstrated here. However, the NER resistance of NIm appears to be correlated not with a destabilization but with its conformational flexibility in double-stranded DNA (26). It is noteworthy that the behavior of NIm lesions relative to Sp and Gh lesions is also quite different in translesion bypass experiments catalyzed by the human polymerase ␤ (pol ␤); the NIm lesions are efficiently bypassed by pol ␤ (16), whereas pol ␤ is completely stalled by Sp and Gh lesions (88).