Differential cleavage of oligonucleotides containing the benzene ...

Carcinogenesis vol.19 no.8 pp.1339–1343, 1998

Differential cleavage of oligonucleotides containing the benzenederived adduct, 1,N6-benzetheno-dA, by the major human AP endonuclease HAP1 and Escherichia coli exonuclease III and endonuclease IV B.Hang, A.Chenna, J.Sa´gi1 and B.Singer2 Donner Laboratory, Lawrence Berkeley National Laboratory, University of California, Berkeley, CA 94720, USA 1Present

address: Central Research Institute for Chemistry, Hungarian Academy of Sciences, H-1525 Budapest, PO Box 17, Hungary

2To

whom correspondence should be addressed

We report here that the newly synthesized DNA adduct, 1,N6-benzetheno-dA (pBQ-dA), in defined oligonucleotides [Chenna and Singer, Chem. Res. Toxicol., 8, 865–874], is a substrate for the major human AP endonuclease, HAP1, and the Escherichia coli AP endonucleases, exonuclease III and endonuclease IV. The mechanism of cleavage is identical to that reported previously for 3,N4-benzethenodC (pBQ-dC) and leads to a phosphodiester bond cleavage 59 to the adduct. There are, however, significant differences in the rate of cleavage of this adduct by these enzymes. The two bacterial AP endonucleases are both much more efficient than the human repair enzyme. In addition, using two random oligodeoxynucleotide sequences containing a single pBQ-dA, exonuclease III and endonuclease IV are similarly active, while HAP1 shows a distinct sequence preference of ~10-fold in efficiency of cleavage. The repair of this adduct by the three recombinant enzymes is further confirmed by using both active site mutant HAP1 proteins and by E.coli mutant strains lacking exonuclease III and/ or endonuclease IV. This sequence-dependent repair of pBQ-dA by HAP1 may play an important role in modulating benzene-induced carcinogenesis.

recognition of this new substrate led to a comparative study of recognition of the AP site and pBQ-dC by both HAP1 and mutants of its active site. Both wild-type HAP1 and mutant proteins had parallel activities toward these two structurally diverse lesions, indicating that they shared the same active site (16). The present work extends the study of enzyme recognition to another benzene-derived adduct, 1,N6-benzetheno-dA (pBQdA), which was synthesized and converted to the phosphoramidite (17). This new adduct was site-specifically incorporated into two random oligomer sequences for biochemical and biological studies. Repair experiments utilized HAP1, exonuclease III and endonuclease IV, as these three enzymes had been shown to be involved in the repair of the analogous derivative, pBQ-dC (16). HAP1 and exonuclease III belong to the same family of ‘class II’ AP endonucleases and share 27% homology, while endonuclease IV is identified as a member of another family of the same class, which is homologous to the Apn1 protein of the yeast Saccharomyces cerevisiae (1,2). Endonuclease IV and exonuclease III share no significant sequence homology and they both show some activities distinct from each other, such as the RNaseH and 39→59 exonuclease activities in exonuclease III, and α-dA activity in endonuclease IV (2,13). Nevertheless, all three repair enzymes were found to cleave the phosphodiester bond 59 to the pBQ-dA adduct. However, cleavage by both exonuclease III and endonuclease IV occurs with much higher efficiency than by HAP1. Moreover, HAP1 recognizes pBQdA with a .10-fold difference in the two sequences used, strongly suggesting a sequence-dependent enzymatic action by the human repair enzyme.

Introduction Class II or 59-AP endonucleases, which are considered multifunctional (1,2), have been primarily studied for their ability to cleave abasic (AP*) sites, either those arising from natural processes, or by specific DNA glycosylase activity (3), or with synthetic AP site analogues (4–6). There are other reported substrates for these repair enzymes, including 39 blocking lesions (7–10), urea (11), hydroxylamines (12) or α-deoxyadenosine (α-dA) (13), in which the base is in the α-configuration [reviewed by Singer and Hang (14)]. Recently, Hang et al. (15) found that a bulky exocyclic adduct, synthesized by reaction of p-benzoquinone (pBQ), a metabolite of the human carcinogen, benzene, with deoxycytidine to form 3,N4-benzetheno-dC (pBQ-dC), was a good substrate for both the major human AP endonuclease, HAP1, and the bacterial AP endonucleases, exonuclease III and endonuclease IV. Furthermore, incision 59 by these 59-AP endonucleases to the adduct in a defined 25mer oligonucleotide left the pBQ-dC on the 59 side of the 39 cleavage fragment as a ‘dangling base’. One approach to understanding the enzyme *Abbreviations: AP, abasic; DTT, dithiothreitol; pBQ, p-benzoquinone; pBQdC, 3,N4-benzetheno-dC. © Oxford University Press

Materials and methods AP endonucleases and bacterial strains The recombinant major human AP endonuclease, HAP1, and its mutant proteins, D219A (aspartate → alanine), N212A (asparagine → alanine) and D210N (aspartate → asparagine) (18,19), were a gift from Dr I.D.Hickson (University of Oxford, UK). Exonuclease III was purchased from Gibco BRL. Endonuclease IV was a gift from Dr D.M. Wilson III (Lawrence Livermore National Laboratory). All the above recombinant proteins were ù95% pure based on SDS–PAGE analysis. Escherichia coli strains BW32(AB1157), BW9109[AB1157 1 ∆(xthpncA)90], BW554[AB1157 1 ∆(xth-pncA)90 nth-1::kan 1 pBR322-nfo] and BW528[AB1157 1 ∆(xth-pncA)90 nth-1::kan] were all gifts from Dr B.Weiss (University of Michigan). All strains were grown in 2 l of LB broth at 37°C to the late log phase. For the strain BW554, paraquat (methyl viologen; Sigma), which was shown to induce the endonuclease IV (20), was added at 0.25 mg/ml to the culture and further incubated for 1 h. The cells were harvested by centrifugation for 10 min at 10 000 r.p.m. in a SS-34 rotor (Sorvall), washed with a buffer containing 50 mM Tris–HCl, pH 7.8, then resuspended in twice the pellet volume using 50 mM Tris–HCl, pH 7.8, 5 mM EDTA and 1 mM phenylmethylsulfonyl fluoride (PMSF), and stored at –75°C. The preparation of crude extracts from these strains were carried out essentially as described by Singer et al. (21). Oligonucleotide substrates and 32P-end labeling Two oligonucleotides containing pBQ-dA, and one containing pBQ-dC for comparative purposes were used as substrates for examining AP endonuclease activities. These had the following sequences: 59-CCGCT-pBQ-A-GCGGGT-

1339

B.Hang et al.

Fig. 2. Plot of pBQ-dA-containing oligomer (Sequence I) cleaved in 10 min as a function of enzyme concentration. The abscissa show the fmol cleaved for each AP endonuclease, in nanograms; Endo IV (j), Exo III (u), and HAP1 (n).

Fig. 1. Autoradiogram of cleavage of pBQ-dA-containing oligomer (Sequence I) using three AP endonucleases. Left, HAP1; center, E.coli exonuclease (Exo III); right, endonuclease IV (Endo IV). The amount of each enzyme and time are given above each group. The 5mer size marker for the cleaved product is on the left (bottom). Note that the amounts of each enzyme were chosen to reflect differences in cleavage.

ACCGAGCTCGAAT-39 (sequence I); 59-GCGTCGCG-pBQ-A-CTGGGAAAACCCC-39 (sequence II) and 59-CCGCTAG-pBQ-C-GGGTACCGAGCTCGAAT-39. The synthesis of these modified nucleosides, their phosphoramidites and site-specific incorporation into 25mer oligonucleotides were previously described by Chenna and Singer (17). These modified oligomers were all purified by reverse-phase HPLC. The oligomer size markers used were also synthesized and purified by OPC cartridges (Applied Biosystems). The oligonucleotides, both modified and the size markers, were 59end labeled with [γ-32P]ATP (sp. act. 6000 Ci/mmol, Amersham) and T4 polynucleotide kinase (USB) in a modified kinase buffer (low Mg21) containing 50 mM HEPES–KOH pH 7.5, 10 mM β-mercaptoethanol, 2 mM MgCl2 at 37°C for 1 h. The modified oligomers were then annealed to complementary oligonucleotides (1.5-fold excess) in a buffer containing 10 mM HEPES– KOH pH 7.5 and 100 mM NaCl, by slowly cooling from 80°C to room temperature. Enzyme assay The incision assay used to examine AP endonuclease activities toward oligonucleotides containing pBQ adducts was carried out essentially as described by Rydberg et al. (22,23) with minor modifications. The standard reaction was performed in a total volume of 10 µl in 25 mM HEPES–KOH pH 7.0, 0.5 mM dithiothreitol (DTT), 0.5 mM spermidine, 200 µM MgCl2, 400 µg of BSA, 10% glycerol and various amounts of enzyme at 37°C. Routinely, 79 fmol of the 32P-labeled oligomers were used in the reaction mixture with varying amounts of repair enzymes. The reactions were stopped by addition of a formamide/EDTA gel loading buffer and eletrophoresed on an 8 M urea 12% polyacrylamide gel. The bands corresponding to the cleavage products, as determined by the co-migrating size markers, and the remaining uncut substrates were quantitated using a Molecular Dynamics PhosphorImager. Thermal melting experiments The experimental method is the same as given by Sa´gi et al. (24). The pBQdA-containing oligonucleotide (sequence I) was compared with the analogous pBQ-dC-containing oligomer in terms of Tm, ∆Tm, H260, and ∆T as described.

Results Comparative cleavage of pBQ-dA by three AP endonucleases In this work, we extended the study of enzyme recognition of pBQ-dC to the purine pBQ adduct, pBQ-dA, using three recombinant 59-AP endonucleases from human (HAP1) and E.coli (exonuclease III and endonuclease IV; Figure 1). All of 1340

these enzymes have previously been shown to recognize and cleave the oligonucleotide containing pBQ-dC (15). The general methodology used was to 32P-end label the oligonucleotides containing site-specific pBQ-dA or pBQ-dC, incubate with appropriate amounts of HAP1, exonuclease III or endonuclease IV, and separate the 59 cleavage fragments by denaturing gel electrophoresis. Figure 1 is a representative autoradiogram showing increasing cleavage of pBQ-dA-containing 25mer oligonucleotide with all three enzymes as a function of time. The products co-migrating with the 5mer size marker (left side), and uncleaved substrates could be scanned and quantitated with a PhosphorImager. Figure 2 shows fmol cleaved/ng enzyme for the pBQ-dAcontaining oligonucleotide reacted with varying amounts of three individual repair enzymes. It should be noted that pBQdA is a very poor substrate for the human repair enzyme HAP1 compared with the two E.coli enzymes. Nevertheless, cleavage of pBQ-dA-containing oligomer by HAP1 can be demonstrated using a high level of enzyme (Figure 1, left panel). In contrast, both exonuclease III and endonuclease IV act in parallel with much higher efficiency. Since HAP1 was so inefficient on the oligomer, which we name sequence I, the effect of sequence and/or oligonucleotide structure was considered. Therefore, another pBQ-dA-containing oligomer (sequence II) was tested for the same enzymes (Figure 3). The results for HAP1 were clear in that the new sequence led to a 10-fold higher cleavage by HAP1 (Figure 3, Table I), while much less differential effect on the rate of cleavage was found with exonuclease III and endonuclease IV (Figure 3, Table I). Comparative cleavage of pBQ-dA and pBQ-dC by wild-type HAP1 and its active site mutants To confirm the HAP1 activity toward pBQ-dA, three HAP1 mutant proteins (D219A, N212A and D210N) were tested for their activities toward the pBQ-dA adduct, compared with their pBQ-dC activity which has been previously described (16). Figure 4 shows the relative cleavage by HAP1 proteins, as a function of time, with differing wild-type HAP1 and mutant concentrations. Note that for pBQ-dC (right) the enzyme level used is 100-fold less than for pBQ-dA (left). Under these conditions (Figure 4), in which similar levels of repair with wild-type HAP1 can be demonstrated for the two adducts, pBQ-dA and pBQ-dC (j), the relative repair by three mutant proteins, which are either inactive (N212A, u, D210N, e) or markedly reduced in activity (D219A, n), is

Repair of pBQ-dA by 59-AP endonucleases

Fig. 3. Comparison of cleavage of sequences I and II by HAP1 (left), Exo III (center), and Endo IV (right) expressed as fmol/ng enzyme over a 15 min period. The scale of cleavage by HAP1 is from 0 to 3.5, while the E.coli AP endonucleases are from 0 to 50. The change in scale is necessary to be able to visualize the difference in cleavage, since HAP1 is very inefficient. The symbols, as shown, are for the two oligonucleotide sequences used.

parallel. The data on pBQ-dC cleavage (right panel) for all three proteins are in agreement with those described previously (16). Identification of exonuclease III and endonuclease IV as repair enzymes for pBQ-dA To rule out the possibility that the bacterial AP endonuclease activity toward pBQ-dA could be due to an unknown contaminant in the enzyme preparations, cell-free crude extracts from the wild-type E.coli strain, and mutants in exonuclease III and/ or endonuclease IV were tested against an oligomer containing pBQ-dA (sequence II). A nicking assay, using the 32P-labeled oligomer (Figure 5), clearly showed that BW32 (wild-type) cleaved the oligonucleotide 59 to pBQ-dA (lane 2), while with BW9109, the exonuclease III mutant, almost all of the activity toward pBQ-dA adduct was lost (lane 3), which is in agreement with the fact that exonuclease III accounts for .80% of the cellular AP endonuclease activity in E.coli. With BW554, a strain carrying multicopy nfo1 plasmid, which encodes the minor 59 AP activity in E.coli, endonuclease IV, the cleavage was enhanced as expected (lane 4). Note that BW554 also lacked exonuclease III activity. Finally, the exonuclease III and endonuclease IV double mutant, BW 528, showed virtually no activity toward the adduct (lane 5), thus confirming exonuclease III and endonuclease IV as the responsible repair enzymes for pBQ-dA. The same results were also obtained for the repair of the pBQ-dC adduct (Figure 5, right panel), in which this genetic approach confirms the previous finding that exonuclease III and endonuclease IV are specific enzymes for this adduct. Effect of pBQ-dA on the apparent thermal stability of the annealed 25-double strand There is a significant destabilization of the pBQ-dA-containing 25mer oligomer (sequence I) as indicated by a decrease in Tm of 9.2° compared with the control 25mer. The hyperchromicity is 18.4% compared with the control value of 20%, and width of transition is 14.7° compared with the control value, 11.2°. These small differences do not show a pattern indicative of a major structural effect or biphasic melting which could occur from preferential melting resulting from placing the adduct 6 bases from the 59 end. All parameters measured are, within experimental error, the same as found for the comparable pBQ-dC containing oligomer (24). Discussion We had previously established that pBQ-dC, site-specifically incorporated into a defined 25mer, was efficiently cleaved by

Table I. Reaction rates of HAP1, exonuclease III and endonuclease IV toward two pBQ-dA-containing oligonucleotidesa Oligomer

pBQ-dA (sequence I) pBQ-dA (sequence II)

Reaction rate (fmol/ng/min)b HAP1

Exo III

Endo IV

0.023 0.24

3.9 6.4

2.6 3.2

aThe

enzymatic reactions were carried out under the standard assay conditions using two pBQ-dA-containing sequences as described in Materials and methods. For details see the legend to Figure 3. bThese numbers were obtained from the rate of reaction at a 5 min point in the time course curves shown in Figure 3.

Fig. 4. Comparison of cleavage of oligomers containing pBQ-dA (left) or pBQ-dC (right) by HAP1 and three mutant proteins. The scales are identical, but wt HAP1 for pBQ-dA is 100 ng while for pBQ-dC, 1 ng is used. In all cases, the mutant proteins use twice these levels, i.e. 200 ng for pBQ-dA and 2 ng for pBQ-dC.

the major human AP endonuclease HAP1, as well as the bacterial functional counterparts, E.coli exonuclease III and endonuclease IV (15). The primary purpose of the present work was to identify and explore the recognition of these bacterial and human AP endonucleases toward the purine exocyclic adduct, pBQ-dA, which is now under investigation for possible mutagenicity. In this study, it has been shown that the repair of pBQ-dA with exonuclease III and endonuclease IV is much more efficient than with HAP1 (Figure 2). Interspecies differences with different repair enzyme sources have been reported 1341

B.Hang et al.

Fig. 5. Cleavage of oligomers containing pBQ-dA (left, lanes 1–5) and pBQ-dC (right, lanes 6–10) using cell-free extracts from wild-type E.coli strain BW32 and mutant strains for 1 h incubation at 37°C. Note the same discrimination pattern for each derivative. All proteins used in the left panel were at the same amount, 3.2 µg, while in the right panel, the protein amount for each lane was 1.6 µg. The relevant phenotypes of the strains used are as follows: BW32(AB1157), BW9109[AB1157 1 ∆(xth-pncA)90], BW554[AB1157 1 ∆(xth-pncA)90 nth-1::kan 1 pBR322-nfo] and BW528[AB1157 1 ∆(xth-pncA)90 nth-1::kan].

previously. For example, human repair (also in a 25mer) can be more efficient than that by lower species, as Saparbaev and Laval (25) and Saparbaev et al. (26) showed for repair of 1,N6-etheno-A and hypoxanthine. In these cases, the human m3A-DNA glycosylase was ~1000-fold more active toward the two substrates than the homologous E.coli enzyme, Alk A. In our case, repair of the pBQ adduct-containing oligomers differed significantly depending on the source of enzymes. pBQ-dA was a poor substrate for HAP1, but was cleaved by exonuclease III and endonuclease IV .13- to 170-fold more than by HAP1, depending on which of two random sequences was used (Table I). Another example of differences in substrate specificity by human and bacterial repair enzymes was summarized by Mattes et al. (27) who compared repair by m3ADNA glycosylase using a wide array of chemically modified substrates, but found no ‘rule’ for this interspecies difference. There also seems to be no clear rationale for the relatively good repair of pBQ-dA in sequence II by HAP1 compared with a 10-fold decrease in repair in sequence I of pBQ-dA (Figure 3, Table I) other than what appears to be the sequence context. However, the same sequence effect is much less for cleavage by exonuclease III (1.6-fold) and endonuclease IV (1.2-fold) (Figure 3, Table I). It would appear that different AP endonucleases, as studied here, exhibit sequence preference, to varying degrees. Thus, we believe that caution should be exercised in generalizing the effect of sequence on biochemical properties. It is likely that the major effect of sequence on efficiency of repair is the interaction of the substrate and neighboring bases in contact with the active site of the enzyme, as suggested by the crystal structure of HAP1 (28). Acknowledgements This work was supported by NIH Grants CA 47723 and CA 72079 (to B.S.) and was administered by the Lawrence Berkeley National Laboratory under Department of Energy contract DE-AC03-76SF00098. We are grateful to Dr I.D.Hickson for providing HAP1 and its mutant proteins, to Dr D.M.Wilson III for E.coli endonuclease IV and to Dr B.Weiss for E.coli mutant strains. The authors also thank M.Medina and S.Rao for expert technical assistance.

References 1. Demple,B. and Harrison,L. (1994) Repair of oxidative damage to DNA: Enzymology and biology. Annu. Rev. Biochem., 63, 915–948.

1342

2. Barzilay,G. and Hickson,I.D. (1995) Structure and function of apurinic/ apyrimidinic endonucleases. Bioessays, 17, 713–719. 3. Friedberg,E., Walker,G.C. and Seide,W. (1995) DNA Repair and Mutagenesis. ASM Press, Washington, DC. 4. Takeshita,M., Chang,C.N., Johnson,F., Will,S. and Grollman,A.P. (1987) Oligodeoxynucleotides containing synthetic abasic sites. Model substrates for DNA polymerases and apurinic/apyrimidinic endonucleases. J. Biol. Chem., 262, 10171–10179. 5. Sanderson,B.J., Chang,C.N., Grollman,A.P. and Henner,W.D. (1989) Mechanism of DNA cleavage and substrate recognition by a bovine apurinic endonuclease. Biochemistry, 28, 3894–3901. 6. Wilson,D.M. III, Takeshita,M., Grollman,A.P. and Demple,B. (1995) Incision activity of human apurinic endonuclease (Ape) at abasic site analogs in DNA. J. Biol. Chem., 270, 16002–16007. 7. Doetsch,P.W. and Cunningham,R.P. (1990) The enzymology of apurinic/ apyrimidinic endonucleases. Mutat. Res., 2–3, 173–201. 8. Chen,D.S., Herman,T. and Demple,B. (1991) Two distinct human DNA diesterases that hydrolyze 39-blocking deoxyribose fragments from oxidized DNA. Nucleic Acids Res., 21, 5907–5914. 9. Robson,C.N., Milne,A.M., Pappin,D.J. and Hickson,I.D. (1991) Isolation of cDNA clones encoding an enzyme from bovine cells that repairs oxidative DNA damage in vitro: homology with bacterial repair enzymes. Nucleic Acids Res., 5, 1087–1092. 10. Winters,T.A., Weinfeld,M. and Jorgensen,T.J. (1992) Human HeLa cell enzymes that remove phosphoglycolate 39-end groups from DNA. Nucleic Acids Res., 10, 2573–2580. 11. Kow,Y.W. and Wallace,S.S. (1985) Exonuclease III recognizes urea residues in oxidized DNA. Proc. Natl Acad. Sci. USA, 24, 8354–8358. 12. Kow,Y.W. (1989) Mechanism of action of Escherichia coli exonuclease III. Biochemistry, 8, 3280–3287. 13. Ide,H., Tedzuka,K., Shimzu,H., Kimura,Y., Purmal,A.A., Wallace,S.S. and Kow,Y.W. (1994) α-Deoxyadenosine, a major anoxic radiolysis product of adenine in DNA, is a substrate for Escherichia coli endonuclease IV. Biochemistry, 33, 7842–7847. 14. Singer,B. and Hang,B. (1997) What structural features determine repair enzyme specificity and mechanism in chemically modified DNA? Chem. Res. Toxicol., 7, 713–732. 15. Hang,B., Chenna,A., Fraenkel-Conrat,H. and Singer,B. (1996) An unusual mechanism for the major human apurinic apyrimidinic (AP) endonuclease involving 59 cleavage of DNA containing a benzene-derived exocyclic adduct in the absence of an AP site. Proc. Natl Acad. Sci. USA, 93, 13737–13741. 16. Hang,B., Rothwell,D.G., Sa´gi,J., Hickson,I.D. and Singer,B. (1997) Evidence for a common active site for cleavage of an AP site and the benzene-derived exocyclic adduct, 3,N4-benzetheno-dC, in the major human AP endonuclease. Biochemistry, 36, 15411–15418. 17. Chenna,A. and Singer,B. (1995) Large scale synthesis of p-benzoquinone29-deoxycytidine and p-benzoquinone-29-deoxyadenosine adducts and their site-specific incorporation into DNA oligonucleotides. Chem. Res. Toxicol., 8, 865–874. 18. Barzilay,G., Walker,L.J., Robson,C.N. and Hickson,I.D. (1995) Sitedirected mutagenesis of the human DNA repair enzyme HAP1: Identification of residues important for AP endonuclease and RNase H activity. Nucleic Acids Res., 23, 1544–1550. 19. Rothwell,D.G. and Hickson,I.D. (1996) Asparagine 212 is essential for abasic site recognition by the human DNA repair endonuclease HAP1. Nucleic Acids Res., 24, 4217–4221. 20. Chan,E. and Weiss,B. (1987) Endonuclease IV of Escherichia coli is induced by paraquat. Proc. Natl Acad. Sci. USA, 84, 3189–3193. 21. Singer,B., Antoccia,A., Basu,A.K., Dosanjh,M.K., Fraenkel-Conrat,H., Gallagher,P.E., Kusmierek,J.T., Qiu,Z.-H. and Rydberg,B. (1992) Both purified human 1,N6-ethenoadenine-binding protein and purified human 3-methyladenine-DNA glycosylase act on 1,N6-ethenoadenine and 3methyladenine. Proc. Natl Acad. Sci. USA, 89, 9386–9390. 22. Rydberg,B., Dosanjh,M.K. and Singer,B (1991) Human cells contain protein specifically binding to a single 1,N6-ethenoadenine in a DNA fragment. Proc. Natl Acad. Sci. USA, 86, 6839–6842. 23. Rydberg,B., Qiu,Z.-H., Dosanjh,M.K. and Singer,B. (1992) Partial purification of a human DNA glycosylase acting on the cyclic adduct, 1,N6-ethenodeoxyadenosine. Cancer Res., 52, 1377–1379. 24. Sa´gi,J., Chenna,A., Hang,B. and Singer,B. (1988) A single cyclic pbenzoquinone adduct can destabilize a DNA oligonucleotide duplex. Chem. Res. Toxicol., 11, 329–334. 25. Saparbaev,M. and Laval,J. (1994) Excision of hypoxanthine from DNA containing dIMP residues by the Escherichia coli, yeast, rat, and human alkylpurine DNA glycosylases. Proc. Natl Acad. Sci. USA, 91, 5873–5877.

Repair of pBQ-dA by 59-AP endonucleases 26. Saparbaev,M., Kleibl,K. and Laval,J. (1995) Escherichia coli, Saccharomyces cerevisiae, rat and human 3-methyladenine DNA glycosylases repair 1,N6-ethenoadenine when present in DNA. Nucleic Acids Res., 23, 3750–3755. 27. Mattes,W.B., Lee,C.S., Laval,J. and O’Connor,T.R. (1996) Excision of DNA adducts of nitrogen mustards by bacterial and mammalian 3methyladenine-DNA glycosylases. Carcinogenesis, 17, 643–648. 28. Gorman,M.A., Morera,S., Rothwell,D.G., de La Fortelle,E., Mol,C.D., Tainer,J.A., Hickson,I.D. and Freemont,P.S. (1997) The crystal structure of the human DNA repair endonuclease HAP1 suggests the recognition of extra-helical deoxyribose at DNA abasic sites. EMBO J., 16, 6548–6558. Received on March 13, 1998; revised on April 30, 1998; accepted on May 1, 1998

1343