Comparative analysis of 8-oxoG:C, 8-oxoG:A, A:C and C:C DNA repair in extracts from wild type or 8-oxoG DNA glycosylase deficient mammalian and bacterial cells

DNA Repair 2 (2003) 707–718

Comparative analysis of 8-oxoG:C, 8-oxoG:A, A:C and C:C DNA repair in extracts from wild type or 8-oxoG DNA glycosylase deficient mammalian and bacterial cells Francoise Dantzer1 , Magnar Bjørås, Luisa Luna, Arne Klungland, Erling Seeberg∗ Centre for Molecular Biology and Neuroscience, and Institute of Medical Microbiology, University of Oslo, Rikshospitalet, 0027 Oslo, Norway Received 22 May 2002; received in revised form 26 February 2003; accepted 26 February 2003

Abstract We have investigated repair of DNA containing 8-oxoguanine and certain mismatches in cell-free extracts from mouse embryonic fibroblasts (MEFs) using a plasmid substrate with a single lesion at a defined position. Repair synthesis was monitored in a small restriction fragment with different size single strands in order to follow the fate of repair reactions in both strands at the same time. An important part of the study was to assess the role of OGG1 in various repair reactions and the experiments were carried out with extracts from mouse embryonic fibroblasts diploid for a mogg1 deletion (Ogg1−/− ) as well as wild type. In wild type, DNA containing 8-oxoG:C was repaired in the expected fashion predominantly through short-patch repair. Overall repair was reduced to 20% in the Ogg1−/− extracts and to 40% if only long-patch repair was considered. The 8-oxoG:A pair was processed similarly in wild type and Ogg1−/− extracts and repair synthesis at A as well as at 8-oxoG could be demonstrated, however, to the same extent in Ogg1−/− and wild type for both strands. Extracts from Ogg1−/− behaved normally in the correction of A:C and C:C mismatches, with a strong bias for correction of A for A:C and no significant strand discrimination for C:C. Similar experiments with extracts from Escherichia coli showed a 50% reduction in the repair of 8-oxoG:C in fpg extracts and an increase to 50% above wild type in mutY. These results show that the mouse OGG1 is the major enzyme for 8-oxoG repair in the MEF cells and does not participate in mismatch repair of A:C or C:C. Furthermore, 8-oxoG opposite A appears to be repaired by a two-step repair pathway with sequential removal of A and 8-oxoG mediated by enzymes different from OGG1. © 2003 Elsevier Science B.V. All rights reserved. Keywords: 8-Oxoguanine; DNA glycosylase; Base excision repair; OGG1-deficient cell extracts; Fpg-deficient cell extracts

1. Introduction Reactive oxygen species (ROS) generated during normal cellular oxygen metabolism and by oxidative ∗ Corresponding author. Tel.: +47-2307-4059; fax: +47-2307-4061. E-mail address: [email protected] (E. Seeberg). 1 Present address: UPR9003/CNRS, ESBS, Bld S´ ebastien Brant, 67400 Illkirch, France.

stress conditions have been considered an important factor contributing to carcinogenesis and ageing in higher organisms [1,2]. ROS induce a large number of sugar and base lesions in DNA of which 7,8-dihydro8-oxoguanine (8-oxoG) is abundant and highly mutagenic because of its ability to pair with adenine during replication, yielding G:C → T:A transversions [3,4]. In Escherichia coli, the mutagenic effects of 8-oxoG are counteracted by the combined action of three enzyme functions: (i) Fpg (MutM), a DNA glycosylase

1568-7864/03/$ – see front matter © 2003 Elsevier Science B.V. All rights reserved. doi:10.1016/S1568-7864(03)00041-7

708

F. Dantzer et al. / DNA Repair 2 (2003) 707–718

that catalyses the excision of 8-oxoG opposite C [5,6], (ii) MutY, a DNA glycosylase that removes adenine inserted across 8-oxoG during replication [7], and (iii) MutT, a nucleotide phosphatase that catalyses the hydrolysis of 8-oxodGTP to 8-oxodGMP, thus preventing 8-oxodGTP from being incorporated opposite A during replication [8]. Human sequence homologues of MutY and MutT, termed hMYH and hMTH1, have been identified and characterised [9,10]. In contrast, the major 8-oxoG DNA glycosylase in mammalian cells, termed mOGG1 in mouse and hOGG1 in humans, is sequence-unrelated to bacterial Fpg/MutM and belongs to the hairpin–helix–hairpin family of DNA glycosylases [11,12]. However, human homologues of Fpg have also recently been identified but these enzymes appears to be minor or backup activities for removal of 8-oxoG [13–15]. In human cells, 8-oxoG:C is repaired mainly through the base excision repair (BER) process initiated by the DNA glycosylase/AP lyase activities of OGG1. There are two pathways for BER: a single nucleotide insertion pathway (short-patch repair (SPR)) and a long-patch repair pathway (LPR) that involves resynthesis of 2–10 nucleotides [16]. SPR uses DNA pol␤ for the resynthesis step and requires AP endonuclease 1 (HAP1/APE1), XRCC1, poly (ADP-ribose) polymerase 1 (PARP-1) and either DNA ligase I or III [17,18]. However, DNA Pol␦ and/or Polε might function as backup DNA polymerases for DNA Pol␤ in the short-patch repair pathway [19]. LPR is PCNA-dependent and involves HAP1, Replication factor C (RFC), Proliferating cell nuclear antigen (PCNA), Replication protein A (RPA), PARP-1, Flap endonuclease 1 (FEN1), DNA Pol␦/ε and DNA ligase I [18,20–23]. However, long-patch repair of reduced AP sites has been shown to require DNA Pol␤ for incorporation of the first nucleotide [24]. The type of DNA glycosylase initiating repair will to some extent determine the choice of pathway [25]. Bifunctional DNA glycosylases such as hOGG1 and hNTH1 induce mainly SPR, while monofunctional glycosylases such as uracil DNA glycosylase (UNG), or alkylpurine DNA glycosylase (ANPG), initiate both short- and long-patch repair. Characterisation of 8-oxoguanine repair in mammalian cells using cell extracts have indicated that repair of DNA with 8-oxoG is mediated through a Pol␤-dependent

and an alternative Pol␤-independent single nucleotide repair-patch pathway [25,27]. In this report, we have analysed repair in cell-free extracts from normal and mOGG1 deficient mouse embryonic fibroblasts (MEFs) to determine the role of OGG1 in the repair of both 8-oxoG:C and 8-oxoG:A pairs and to search for a functional interaction between OGG1 and MYH. A similar assay with bacterial cell extracts has confirmed the existence of a cooperative interaction between Fpg and MutY as was recently indicated from experiments with purified enzymes [28,29]. Errors in DNA replication, spontaneous base oxidation and homologous recombination also induce base pair mismatches. Most of them are repaired by the long-patch post-replicative mismatch repair system, which involves resynthesis of up to 1000 nucleotides. In addition, there are more specialised repair pathways that remove specific mismatched bases. In mammalian cells, G:T mismatches were shown to be processed by a thymine-DNA glycosylase (TDG) initiating base excision repair [30]. A:C mismatches were recently shown to be repaired by an Msh2 and Mlh1-independent repair mechanism involving replacement of the adenine associated with a 25 nt synthesis patch [31]. In E. coli, the mutHLS mismatch repair pathway was shown to correct efficiently G:G but not C:C base pairs [32]. E. coli possesses two C:C mismatch binding proteins of which one is the 8-oxoguanine DNA glycosylase Fpg, thus suggesting the possibility that Fpg is involved in the repair of these mismatches [33]. In order to examine a possible role of OGG1 in the repair of A:C and/or C:C mispairs in mammalian cells, we compared the repair of such mismatches in extracts from wild type and Ogg−/− cells without finding any evidence for a role of OGG1 in mismatch repair. A:C repair was specific for the A-containing strand whereas C:C correction occurred with similar efficiency in both strands.

2. Materials and methods 2.1. Cell culture Mouse embryonic fibroblasts were isolated by microdissection of 13.5-day-old embryos as described

F. Dantzer et al. / DNA Repair 2 (2003) 707–718

previously [34]. Spontaneously immortalised embryonic fibroblasts were maintained in DMEM/Ham’s F-12 (3:1) supplemented with 10% fetal calf serum (Integro), 100 units/ml penicillin and 100 ␮g/ml streptomycin (BioWhittaker), at 37 ◦ C in a 5% CO2 atmosphere. 2.2. DNA plasmid substrates Plasmids carrying a single lesion at a defined position were produced as described [16]. Briefly, pGEM3Zf(+) single-stranded DNA was annealed with the 22 bp oligonucleotides: (i) GATCCTCTAGA 8-oxoGTCGACCTGCA (generating a single 8-oxoG/ C base pair, pGEM-8-oxoG:C), (ii) GATCCTCTAGAG8-oxoGCGACCTGCA (generating a single 8-oxoG/ A base pair, pGEM-8-oxoG:A), (iii) GATCCTCTAGAATCGACCTGCA (generating a single A/C base pair, pGEM-A:C), (iv) GATCCTCTAGACTCGACCTGCA (generating a single C/C base pair, pGEM-C:C), (v) GATCCTCTAGAGUCGACCTGCA (generating a single U/A base pair, pGEM-U:A). Control pGEM-G:C plasmids were prepared with an oligonucleotide carrying the normal base G. Closed circular double-stranded DNA was obtained by incubation with T4 DNA polymerase, T4 gene 32 protein and T4 DNA ligase (Boehringer, Mannheim). 2.3. Escherichia coli extracts and Fpg enzyme E. coli strains AB1157 (wt), CSH117 (mutY) and BH20 (fpg) were obtained from the E. coli Genetic Stock Center (Yale University). Bacteria were grown in LB medium supplemented with corresponding antibiotics and extracts were prepared by lysozyme/EDTA treatment according to standard procedure. Fpg protein was kindly provided by S. Boiteux (CEA, Fontenay-aux-Roses, France). 2.4. In vitro repair assay The in vitro base excision repair reaction was carried out essentially as described [16]. Briefly, 300 ng of the indicated plasmid substrate were incubated with 50 ␮g of whole cell extracts of MEFs prepared as described [35], for 3 h at 30 ◦ C in 50 ␮l reaction buffer (Hepes/KOH 45 mM, pH 7.8; KCl 60 mM;

709

MgCl2 7.5 mM; DTT 0.9 mM; ATP 2 mM; phosphocreatine 40 mM; creatine phosphokinase (type I, Sigma) 2.5 ␮g; BSA 18 ␮g) containing 20 ␮M of each dNTP and 5 ␮Ci (3000 Ci/mmol, Amersham) of either ␣-[32 P]dGTP (for pGEM-8-oxoG:C, pGEM-8-oxoG:A, pGEM-A:C, pGEM-C:C) or ␣-[32 P]dTTP (for pGEM-U:A) to monitor the overall BER, and ␣-[32 P]dCTP to investigate the long-patch repair pathway or the repair process involved on the removal of the opposite base. After repair reaction, the plasmid DNA was recovered, purified and digested with SmaI and HindIII restriction endonucleases. The resulting 33 and 37 bp fragments were resolved on denaturating 20% polyacrylamide gel electrophoresis, and visualised by autoradiography. The amount of radioactivity was quantified on a PhosphoImager (Molecular Dynamics). In vitro repair assays with E. coli extracts were performed as described above except that the plasmid substrate was incubated in 50 ␮l of reaction buffer together with 80 ␮g of E. coli extract for 20 min at 37 ◦ C.

2.5. Expression, purification and biochemical characterisation of wild type and K249 Q human OGG1 proteins The expression and purification of the mutant hOGG1 protein K249 Q was perfomed as previously described for the wild type hOGG1 [12]. 8-oxoG cleavage assays were carried out essentially as described [12]. Briefly, a 24-mer oligonucleotide containing a single 8-oxoG residue at position 14 was 32 P-labeled at the 5 end using T4 polynucleotide kinase (New England Biolabs) ␥[32 P]ATP (3000 Ci/mmol; Amersham), and annealed to the complementary strand containing a C residue opposite 8-oxoG. Standard reaction mixtures contained 80 fmol double-stranded oligonucleotide substrates and increasing amounts of enzymes as indicated in a total volume of 10 ␮l. Samples were incubated for 30 min at 37 ◦ C, and the cleavage products were analysed by denaturating 20% polyacrylamide gels and PhosphorImaging. The electrophoretic mobility-shift assays were performed identically to the cleavage assays except that the reactions were performed at 4 ◦ C for 30 min and loaded onto a non-denaturating 10% polyacrylamide gel.

710

F. Dantzer et al. / DNA Repair 2 (2003) 707–718

3. Results 3.1. Ogg−/− mouse embryonic fibroblast extracts are impaired in the repair of 8-oxoguanine in DNA Plasmid DNA carrying a single 8-oxoG:C base pair at a defined position (pGEM-8oxoG:C) was in-

cubated with whole cell extracts of Ogg+/+ (+/+, wt) and Ogg−/− (−/−) MEF cells in the presence of either ␣-[32 P]dGTP to monitor overall repair synthesis (SPR + LPR) or ␣-[32 P]dCTP to measure LPR exclusively. After the repair reaction, the plasmid DNA was purified and digested with SmaI and HindIII restriction enzymes to excise the 33 and 37 bp fragment

Fig. 1. Repair of a DNA containing a single 8-oxoG in wt and Ogg1−/− MEF extracts. (A) pGEM-8-oxoG:C (lanes 3–6, 9–12) or pGEMcontrol plasmids (lanes 1, 2, 7, 8) were incubated with a cell-free extract of wt (+/+) or Ogg−/− (−/−) MEFs under standard repair conditions. When indicated, cell extracts were preincubated with 300 ng of either wild type purified human OGG1 (lanes 5 and 11) or mutant K249 Q purified hOGG1 (lanes 6 and 12) before repair assay. To measure the overall repair (SPR + LPR), repair replication was performed in the presence of ␣-[32 P]dGTP (lanes 1–6) and to measure the long-patch repair (LPR), ␣-[32 P]dCTP was used (lanes 7–12). After the repair reaction, plasmids were purified and digested with SmaI and HindIII restriction endonucleases to release a 33/37 bp DNA fragment. Repair products were analysed by autoradiography after electrophoresis on a 20% denaturating polyacrylamide gel. The exposure time for the autoradiography showing LPR (right gel) was increased compared to the autoradiography showing overall repair (left gel). For each repair pathway, radioactivity in fragment bands of four independent experiments were quantified on a PhosphorImager and average values were plotted in the histogram. Error bars represent standard deviation. (B) 8-oxoG DNA glycosylase activity of hOGG1 and K249 Q. Increasing amounts of either wt purified hOGG1 or mutant K249 Q purified hOGG1 were incubated with a [32 P] 5 end-labelled 24 bp duplex oligodeoxyribonucleotide containing a single 8-oxoG residue opposite C. The cleavage products were analysed by 20% denaturating polyacrylamide gel electrophoresis and PhosphorImaging. (C) Protein–DNA complexes formed upon incubation of increasing amounts of either hOGG1 or K249 Q with the [32 P]-labelled 8-oxoG:C containing duplex, as visualized by a band-shift assay after 10% polyacrylamide gel electrophoresis under native conditions.

F. Dantzer et al. / DNA Repair 2 (2003) 707–718

in which the lesion was originally inserted (Fig. 1). After denaturating polyacrylamide gel electrophoresis of the restricted DNA, efficient ␣-[32 P]dGTP and ␣-[32 P]dCTP incorporation was detected in the 33 bp fragment that contained the 8-oxoG residue (Fig. 1A; lanes 3 and 9). Quantification by PhosphorImaging of the incorporation of both labelled nucleotides normalized to the number of potential sites revealed that approximately 65% (±17) of the 8-oxoG was repaired via SPR and 35% (±9) was repaired via LPR in mouse cells. This estimate is close to the 72% of SPR and 28% of LPR reported previously for human cells [27]. The repair was damage-specific since no repair synthesis was observed in the undamaged pGEM-G:C plasmid (lanes 1, 2, 7, 8). Ogg−/− cell extracts showed a major deficiency in the overall repair synthesis of 8-oxoG:C (20% remaining, compare lane 4 with lane 3) and to a lesser extend for LPR (40% remaining, compare lane 10 with lane 9). The persistence of some repair synthesis of 8-oxoG:C in the Ogg−/− cell extracts probably reflects some backup repair activity possibly mediated by the recently discovered hFpg1/Neil1 DNA glycosylase [15]. It seems unlikely that this will represent nucleotide excision repair as there is now evidence for any repair synthesis 3 to the lesion with these extracts ([34], Klungland, unpublished). The repair deficiency observed in the Ogg−/− cell extracts was restored by the addition of purified wild

711

type human OGG1 (wt) but not by an active-site mutant protein (K249 Q). Whereas the wild type hOGG1 used efficiently cleaved a 24 bp DNA duplex containing an 8-oxoG:C base pair, no cleavage was observed for the K249 Q mutant protein (Fig. 1B). In contrast, the mutant K249 Q showed essentially normal DNA binding properties (Fig. 1C). When purified wild type human OGG1 was added to the Ogg−/− cell extracts, both overall and long-patch repair activities were slightly above the level measured with wild type cell extracts (Fig. 1A, compare lanes 5 and 11 with lanes 3 and 9, respectively). This could indicate that endogeneous mouse OGG1 is limiting in the repair reaction and that the supplied human enzyme alleviates this shortage. In contrast neither overall repair, nor the LPR defect in Ogg−/− cell extracts could be restored by the addition of the catalytically inactive mutant K249 Q, thus further emphasising the catalytic requirement of OGG1 in the base excision repair of 8-oxoG:C lesions (Fig. 1A, compare lanes 6 and 12 with lanes 3 and 9, respectively). 3.2. Ogg−/− cell extracts are proficient in the repair of U:A base pairs For comparison, wild type cell extracts and Ogg−/− cell extracts were also examined for their ability to repair a single U:A base pair (Fig. 2). ␣-[32 P]dTTP was used to monitor the overall base excision repair in

Fig. 2. Repair of a single uracil residue by wt and Ogg1−/− MEF extracts. pGEM-U:A plasmid was incubated with a cell-free extract of wt (+/+) or Ogg−/− (−/−) MEF under standard repair conditions and in the presence of ␣-[32 P]dTTP. Repaired DNA was digested with SmaI and HindIII to release the 33 bp fragment. Reaction products were analysed by autoradiography after separation by 20% denaturating polyacrylamide gel electrophoresis.

712

F. Dantzer et al. / DNA Repair 2 (2003) 707–718

the 33 bp SmaI and HindIII restriction fragment. No difference in the overall BER of a single uracil residue was observed between the wt and Ogg−/− cell extracts implying that the Ogg defect does not influence other steps of the BER pathway, as would be expected. 3.3. 8-oxoG opposite A is removed by a long-patch repair pathway independent of OGG1 Previous studies with purified enzymes have suggested a cooperative interaction between MutY and

Fpg to prevent mutations induced by 8-oxoG [28,29]. To analyse for a possible functional interaction between the corresponding mouse OGG1 and MYH, we analysed repair synthesis at 8-oxoG:A mispairs in wild type (wt) and Ogg−/− cell extracts (Fig. 3). Repair synthesis with ␣-[32 P]dGTP at 8-oxoG was 3.4-fold higher opposite C than A in the wild type (Fig. 3B; 33 bp, lanes 1 and 3). In contrast, incorporation of ␣-[32 P]dGTP in the A containing strand was minimal in both wt and Ogg1−/− (37 bp, lanes 3 and 4). Furthermore, there was no reduction in the repair

Fig. 3. Repair of 8-oxoG:C and 8-oxoG:A by Ogg1−/− MEF extracts, pGEM-8-oxoG:C (lanes 1, 2, 5, 6) or pGEM-8-oxoG:A plasmid (lanes 3, 4, 7, 8) was incubated with a cell-free extract of wt (+/+) or Ogg−/− (−/−) MEFs under standard repair conditions and in the presence of either ␣-[32 P]dGTP or ␣-[32 P]dCTP as indicated. Repaired DNA was digested with SmaI and HindIII to release the 33/37 bp fragment. Repair products were analysed by autoradiography after separation by 20% denaturating polyacrylamide gel electrophoresis. One out of three independent experiment is shown. Spots of all three experiments were quantified on a PhosphorImager and average numbers are plotted in the histogram. Error bars represent standard deviation.

F. Dantzer et al. / DNA Repair 2 (2003) 707–718

of 8-oxoG opposite A in Ogg1−/− (33 bp, lanes 3 and 4) consistent with the inability of OGG1 to act at 8-oxoG:A [11]. When we measured the incorporation of ␣-[32 P]dCTP in the 8-oxoG-containing strand, we observed a reduced capacity of the Ogg−/− cell extract to perform long-patch repair of 8-oxoG:C (33 bp, compare lane 6 with lane 5) as already shown in Fig. 1. No difference in the LPR at 8-oxoG across A was observed (33 bp, lanes 7 and 8) and the LPR of 8-oxoG opposite C was only 1.4-fold higher than opposite A (33 pb, lanes 5 and 7). In addition, quantification of the incorporation of ␣-[32 P]dCTP and ␣-[32 P]dGTP in the 33 bp of 8-oxoG:A revealed that the ratio of long- to short-patch repair was about 1.5. Altogether, these results suggest a two-step repair pathway where A is replaced by C, presumably mediated by MYH, and the 8-oxoG is then removed from the resulting 8-oxoG:C basepair through a long-patch repair pathway mediated by an 8-oxoguanine DNA glycosylase different from OGG1. A by C replacement was efficiently detected in our assay for both wt and Ogg−/− cell extracts (37 bp, lanes 7 and 8). In contrast, no incorporation of ␣-[32 P]dGTP (37 bp, lanes 3 and 4) or ␣-[32 P]dTMP (data not shown) could be observed. These findings suggest that the repair synthesis in the A-containing strand does not exceed two nucleotides. This repair is specific to 8-oxoG:A since no radioactive incorporation was detected in the C-containing strand of 8-oxoG:C (37 bp, lanes 5 and 6). 3.4. A:C and C:C mismatches are efficiently corrected by Ogg−/− cell extracts Oda et al. [31] have recently reported efficient repair of A:C mismatches in both Msh2−/− and Mlh1−/− mouse cell lines suggesting the presence of a repair pathway different from ordinary long-patch mismatch repair. Furthermore, we and others have identified bacterial Fpg and yeast and human OGG1 as C:C and A:C mismatch binding proteins, suggesting a possible connection between base excision and mismatch repair pathways [33,36]. To investigate whether OGG1 could be involved, repair synthesis at 8-oxoG:C, A:C and C:C base pairs in DNA were analysed by ␣-[32 P]dGTP incorporation mediated by extracts from wt and Ogg1−/− cells (Fig. 4). The wild type cell extracts repaired a single 8-oxoG:C

713

Fig. 4. Repair of A:C and C:C mismatches by Ogg1−/− MEF extracts, pGEM-8-oxoG:C (lanes 1 and 2), pGEM-A:C (lanes 3 and 4) or pGEM-C:C (lanes 5 and 6) plasmid was incubated with a cell-free extract of wt (+/+) or Ogg−/− (−/−) MEFs under standard repair conditions and in the presence of ␣-[32 P]dGTP. Repaired DNA was digested with SmaI and HindIII to release the 33/37 bp fragment. Repair products were analysed by autoradiography after separation by 20% denaturating polyacrylamide gel electrophoresis. One out of three independent experiment is shown. Fragment bands of all three experiments were quantified on a PhosphorImager and average results plotted in the histogram.

on average 2 and 1.4-fold better than single A:C and C:C mismatches, respectively (33 bp, lanes 1, 3 and 5). For A:C mispairs, the repair was largely specific to the A-containing strand (lanes 3 and 4, 33 bp versus 37 bp). In contrast, for C:C, incorporation was observed to nearly the same extent in both strands (lanes 5 and 6). As shown in Fig. 1, Ogg−/− cell extracts were deficient in the overall base excision repair of 8-oxoG:C base pair (33 bp, compare lane 2 with lane 1), while no significant repair defect was detected in the repair of either A:C or C:C mismatches (lanes 3 and 4, and lanes 5 and 6, respectively). The bias for A repair could be ascribed to the use of radiolabelled dGTP, however, similar results were obtained when repair synthesis was investigated using ␣-[32 P]dCTP as labelled nucleotide (data not shown). Overall, these results indicate that OGG1 is not involved in the repair of A:C or C:C mismatches.

714

F. Dantzer et al. / DNA Repair 2 (2003) 707–718

Fig. 5. Repair of a single 8-oxoG lesion in fpg extracts from E. coli. (A) pGEM-8-oxoG:C (lanes 3–5, 8–10) or pGEMcontrol plasmid (lanes 1, 2, 6, 7) was incubated with an AB1157 (wt) or BH20 (fpg) extract in a repair reaction buffer containing either ␣-[32 P]dGTP to investigate overall repair (SPR + LPR) (lanes 1–5) or ␣-[32 P]dCTP to measure long-patch repair (LPR) (lanes 6–10) and subsequently digested with SmaI and HindIII restriction enzymes. The 33/37 bp fragment containing the repair product was analysed by autoradiography after separation by 20% denaturating polyacrylamide gel electrophoresis. One out of three independent experiment is shown. Spots of all three experiments were quantified on a PhosphorImager and average numbers plotted in the histogram. When indicated, E. coli extracts were preincubated with 50 ng of purified wild type Fpg protein before repair assay (lanes 5 and 10). (B) pGEM-8-oxoG:C plasmid was incubated with an AB1157 (wt) or BH20 (fpg) extract in a repair reaction buffer containing ␣-[32 P]dGTP. Short-patch repair (SPR) was investigated in the 17 bp SmaI and HincII restriction fragment (lanes 1 and 2) and long-patch repair (LPR) was measured in the 16 bp HincII and HindIII restriction fragment (lanes 3 and 4). Repair products were analysed by autoradiography after separation by 20% denaturating polyacrylamide gel electrophoresis. One out of three experiments giving the essentially the same result is shown.

3.5. Repair of 8-oxoG:C is more efficient in extracts from E. coli mutY than wild type For comparison, we have also performed similar experiments with extracts from E coli. (Fig. 5A). Compared to wild type, fpg extracts showed a 40% decrease (lanes 3 and 4) in the overall DNA repair of a single 8-oxoG:C mispair and a 75% decrease in LPR (lanes 8 and 9). Repair synthesis was restored to wild type levels by the addition of purified Fpg (lanes 5 and 10). Repair was damage specific as no incorporation was observed with the control pGEM-G:C plasmid (lanes 1 and 2, and lanes 6 and 7). The relative contribution of short- and long-patch repair pathways was also exam-

ined by measuring the incorporation of ␣-[32 P]dGTP into two smaller fragments, 17 and 16 bp, generated by an additional HincII cleavage (Fig. 5B. Similar relative extent of incorporation was observed in both fragments indicating that LPR is the predominant pathway in the wild type (lanes 1 and 3). However, the remaining activity in the fpg extracts was primarily SPR (compare lanes 4 and 3, and lanes 2 and 1). In vitro, MutY has been shown to bind efficiently to 8-oxoG:C base pairs and thereby inhibit the Fpg DNA glycosylase activity suggesting a functional interaction between the two proteins [28]. To further study Fpg and MutY interactions in a system with other components of repair, we compared the

F. Dantzer et al. / DNA Repair 2 (2003) 707–718

715

3.6. Repair of 8-oxoG across A requires MutY in E. coli

Fig. 6. Repair of 8-oxoG:C and 8-oxoG:A in extracts from E. coli fpg and mutY. (A) pGEM-8-oxoG:C plasmid was incubated with an AB1157 (wt) or CSH117 (mutY) extract in a repair reaction buffer containing ␣-[32 P]dGTP to investigate overall repair or ␣-[32 P]dCTP to investigate long-patch repair. Repaired plasmid were digested with SmaI and HindIII restriction enzymes and the 33 bp fragment was analysed by autoradiography after separation on a 20% denaturating polyacrylamide gel. Bands of four independent experiments were quantified on a PhosphorImager and mean data are plotted in the histogram. (B) Western blot analysis of Fpg in wt and mutY extracts. Forty microgram of total protein from AB1157 (wt, lane 1) or CSH117 (mutY, lane 2) and 10 ng of purified Fpg (lane 3) were analysed by Western blot according to standard procedure using a polyclonal ␣-Fpg antibody kindly provided by S. Boiteux (CEA, Fontenay-aux-Roses). (C) pGEM-8-oxoG:C (lanes 1–3) or pGEM-8-oxoG:A (lanes 4–6) plasmids were incubated with a an AB1157 (wt), BH20 (fpg) or CSH117 (mutY) extract under standard repair conditions in the presence of ␣-[32 P]dCTP. Repaired DNA was digested with SmaI and HindIII to release a 33 bp fragment originally containing 8-oxoG and a 37 bp fragment with C on the opposite strand. Repair products were analysed by autoradiography after separation on a 20% denaturating polyacrylamide gel.

base excision repair efficiency of wt, fpg and mutY extracts on both 8-oxoG:C and 8-oxoG:A DNA (Fig. 6). MutY deficient extracts were on the average 1.8 and 1.6-fold more efficient than wt extracts in the overall and long-patch base excision repair of 8-oxoG:C, respectively Fig. 6A and C, 33 bp, lanes 1 and 3), although no overexpression of Fpg protein was detected in the mutY extracts compared to the wt extracts (Fig. 6B). This extends the previous finding of interaction/interference between Fpg and MutY.

We also compared the long-patch base excision repair of 8-oxoG:C and 8-oxoG:A using ␣-[32 P]dCTP as labelled nucleotide (Fig. 6C). Extracts from wild type were more efficient (approximately 1.7-fold) in repair of 8-oxoG opposite C than A (lanes 1 and 4). Repair of 8-oxoG opposite A was not affected by the absence of Fpg but completely abolished in mutY (lane 6). The extent of repair in the A-containing strand (37 bp fragment) was close to the limit of detection in our assay but was comparable in the wt and fpg extracts and abolished in mutY as expected (lanes 4–6). These results are parallel to those obtained with the mammalian extracts showing that Fpg in E. coli like OGG1 in mammalian cells is not required for the repair of 8-oxoG opposite A. In addition, they show that MutY is required for the 8-oxoG repair.

4. Discussion Targeted disruption of the mogg1 gene in mice has permitted the assessment of the in vivo function of OGG1; the major DNA glycosylase for removal of 8-oxoG in mammalian cells [34,37]. The animals are essentially without phenotypic abnormalities but show a slightly elevated rate of spontaneous mutations in non-proliferative tissues and a 5–10-fold higher level of endogenously induced 8-oxoG in the genome. The mild phenotype indicates the presence of alternative pathways for the repair of 8-oxoG in mammalian cells. Here, we have used extracts from embryonic fibroblasts from the mutated mice to investigate the function of OGG1 in short and long-patch base excision repair measuring repair synthesis in a plasmid DNA containing a single 8-oxoguanine opposite C or A. Parallel studies were also performed with E. coli and mutant extracts lacking the major 8-oxoG DNA glycosylase in bacterial cells; Fpg. Both biochemical and structural analysis of the mammalian OGG1 have indicated a strong preference for 8-oxoG opposite C and a concomitant repulsion of 8-oxoG paired with A [11,38–40]. Similarly, bacterial Fpg also rejects 8-oxoG paired with A [11]. This is consistent with the need for these enzymes in removing 8-oxoG prior to replication since 8-oxoG in

716

F. Dantzer et al. / DNA Repair 2 (2003) 707–718

the template will generate 8-oxoG:A mispairs that should not be corrected to T:A. Instead, homologous DNA glycosylase for removal of A across 8-oxoG have evolved; termed MutY in bacteria and MYH in mammals. Our results confirm that 8-oxoG:C is predominantly processed by the short-patch BER pathway as was recently described [25–27]. We also show that Ogg−/− cells are affected in both short- and long-patch base excision repair of 8-oxoG opposite C. Residual long-patch repair could still be detected in the Ogg−/− cells, which is consistent with the presence of an alternative OGG1-independent repair pathway for 8-oxoG:C [15,41]. The BER defect of the Ogg−/− mutant depended on the enzymatic functions of OGG1 since the repair defect could be restored by the addition of wild type hOGG1 but not when a catalytically inactive mutant of hOGG1 was added. In contrast to mammalian cell extracts, E. coli cell extracts repaired 8-oxoguanine in DNA efficiently via LPR only. Although the overall repair was reduced in the fpg extracts, residual repair synthesis could still be detected. This alternative pathway most likely can be attributed to Endonuclease VIII (Nei) [42,43]. The recently reported interference by MutY with the activity of Fpg in E. coli prompted us to examine if a functional interaction also would exist between the corresponding mammalian enzymes, MYH and OGG1. Despite that MYH in vitro is shown to be a DNA glycosylase little is known about the in vivo function of this enzyme. Using our in vitro cell-free repair assay, we observed BER of A across 8-oxoG and with a polymerisation step limited to two-nucleotides, since we do not find incorporation by dTTP needed for repair synthesis at the third nucleotide in the 5 –3 direction. However, human MYH has recently been reported to be associated with long-patch repair proteins HAP1, PCNA and RPA in HeLa cells and therefore most likely follows the LPR pathway consistent with a replacement of two nucleotides [44]. Most strikingly, incorporation in the A-containing strand is always associated with a similar incorporation in the 8-oxoG containing strand. This can be explained by a co-ordinated two-step model for BER of 8-oxoG:A in mammalian cells. After removal of adenine by the MYH glycosylase activity, APE1 induces cleavage of the resulting AP-DNA [45]. DNA polymerase ␤ incorporates a cy-

tosine opposite 8-oxoG and the resulting 8-oxoG:C is processed by an OGG1-independent long-patch base excision repair process. In human cells, it is supposed that MYH is targeted to the daughter strand through interactions with hMutS␣ and replication proteins, and that PCNA might coordinate DNA replication with repair [46]. The same proteins could recruit the proposed DNA glycosylase involved in the repair of 8-oxoG:C formed upon MYH-dependent 8-oxoG:A repair. A reasonable candidate DNA glycosylase for this action is hFPG1/NEIL1, which by presumed analogy to its bacterial homologue Fpg is expected to carry out LPR predominantly, as is observed for replacement of 8-oxoG opposite A in our system. Similarly as for Ogg−/− cell extracts, Fpg-deficient cell extracts are not affected in the repair of 8-oxoG opposite A, consistent with the low activity at 8-oxoG:A by Fpg. In contrast, E. coli Nei has affinity for 8-oxoG opposite A [42,43] and may be recruited to such a site together with MutY. Interestingly, we also found that the mutY extracts were more efficient in repairing 8-oxoG:C than wt extracts. This is not associated with overexpression of Fpg in the mutant but is most likely due to lack of inhibition of 8-oxoG:C repair by MutY as described by Li et al. [28] based on studies on the inhibiting effect of purified MutY on Fpg DNA glycosylase activity in vitro. Our results indicate that this inhibition is relevant for the reconstituted pathway in a whole cell extract system. The interaction between hMYH and hMutS␣ represents a functional connection between base excision repair and mismatch repair in mammalian cells [46]. Several other studies are consistent with a redundancy among various DNA repair pathways. The repair of both T:G and A:C mismatches have been shown to occur both by ordinary mismatch repair and by separate pathways independent of the ordinary mismatch repair proteins Msh2 and Mlh1 [30,31]. Although no significant A:C DNA glycosylase activity could be detected in the extracts used in their study, we examined a potential role of OGG1 in the processing of this lesion because we have observed that human OGG1 binds weekly to A:C mismatches. In the extract system, removal of A opposite C, was considerably less efficient than repair of 8-oxoG:C and was independent of OGG1. We also tested for C:C mismatch repair in the extract system without finding any evidence for an involvement of OGG1.

F. Dantzer et al. / DNA Repair 2 (2003) 707–718

In conclusion, we propose that post-replicate repair of 8-oxoG:A, arising from insertion of A across 8-oxoG in the template, is a co-ordinated process that requires MYH or alternative for removal of A and a glycosylase removing 8-oxoG to regenerate G:C. Support for this model is obtained with the bacterial system, which shows that the action of MutY is required for removal of 8-oxoG paired with A and not only the A. Results with both the bacterial and the mammalian system says that the proposed 8-oxoG DNA glycosylase involved is different from Fpg and OGG1, respectively.

Acknowledgements We thank Serge Boiteux (CEA, Fontenay-auxRoses, France) for providing us with Fpg enzyme and the polyclonal ␣-Fpg antibody. This work was supported by the International Agency of Cancer Research (IACR, Lyon, France), Norwegian Cancer Society, the Research Council of Norway and en EU Contract (QRLT 1999-02002).

References [1] L.J. Marnett, Oxyradicals and DNA damage, Carcinogenesis 21 (2000) 361–370. [2] S. Raha, B.H. Robinson, Mitochondria, oxygen free radicals diseases and ageing, Trends Biochem. Sci. 25 (2000) 502– 508. [3] K.C. Cheng, D.S. Cahill, H. Kasai, S. Nishimura, L.A. Loeb, 8-hydroxyguanine, an abundant form of oxidative DNA damage, causes G:T and A:C substitutions, J. Biol. Chem. 267 (1992) 166–172. [4] M. Moriya, Single-stranded shuttle phagemid for mutagenesis studies in mammalian cells: 8-oxoguanine in DNA induces targeted G:C → T:A transversions in simian kidney cells, Proc. Natl. Acad. Sci. U.S.A. 90 (1993) 1122–1126. [5] S. Boiteux, T.R. O’Connor, J. Laval, FormamidopyrimidineDNA glycosylase of Escherichia coli: cloning and sequencing of the fpg structural gene and overproduction of the protein, EMBO J. 6 (1987) 3177–3183. [6] J. Tchou, H. Kasai, S. Shibutani, M.H. Chung, J. Laval, A.P. Grollman, S. Nishimura, 8-oxoguanine (8-hydroxyguanine) DNA glycosylase and its substrate specificity, Proc. Natl. Acad. Sci. U.S.A. 88 (1991) 4690–4694. [7] M.L. Michaels, L. Pham, Y. Nghiem, C. Cruz, J.H. Miller, The GO system protects organisms from the mutagenic effect of the spontaneous lesion 8-oxoguanine (7,8-dihydro8-oxoguanine), Nucleic Acids Res. 18 (1990) 3841–3845.

717

[8] H. Maki, M. Segikuchi, MutT protein specifically hydrolyses a potent mutagenic substrate for DNA synthesis, Nature 355 (1992) 273–275. [9] M.M. Slupska, C. Baikalov, W.M. Luther, J.H. Chiang, Y.F. Wei, J.H. Miller, Cloning and sequencing of a human homolog (hMYH) of the Escherichia coli mutY gene whose function is required for the repair of oxydative DNA damage, J. Bacteriol. 178 (1996) 3885–3892. [10] K. Sakumi, M. Furuichi, T. Tsuzuki, T. Kakuma, S.I. Kawabata, H. Maka, M. Sekiguchi, Cloning and expression of cDNA for a human enzyme that hydrolyses 8-oxo-dGTP, a mutagenic substrate for DNA synthesis, J. Biol. Chem. 268 (1993) 23524–23530. [11] T.A. Rosenquist, O.D. Zharkov, A.P. Grollman, Cloning and characterization of a mammalian 8-oxoguanine DNA glycosylase, Proc. Natl. Acad. Sci. U.S.A. 94 (1997) 7429– 7434. [12] M. Bjørås, L. Luna, B. Johnsen, E. Hoff, T. Haug, T. Rognes, E. Seeberg, Opposite base-dependent reactions of a human base excision repair enzyme on DNA containing 7,8-dihydro-8-oxoguanine and abasic sites, EMBO J. 16 (1997) 6314–6322. [13] T.K. Hazra, T. Izumi, I. Boldogh, B. Imhoff, Y.W. Kow, P. Jaruga, M. Dizdaroglu, S. Mitra, Identification and characterization of a human DNA glycosylase for repair of modified bases in oxidatively damaged DNA, Proc. Natl. Acad. Sci. U.S.A. 99 (2002) 3523–3528. [14] V. Bandaru, S. Sunkara, S.S. Wallace, J.P. Bond, A novel human DNA glycosylase that removes oxidative DNA damage and is homologous to Escherichia coli endonuclease VIII, DNA Repair 1 (2002) 517–529. [15] I. Morland, V. Rolseth, L. Luna, T. Rognes, M. Bjoras, E. Seeberg, Human DNA glycosylases of the bacterial Fpg/MutM superfamily: an alternative pathway for the repair of 8-oxoguanine and other oxidation products in DNA, Nucleic Acids Res. 30 (2002) 4926–4936. [16] G. Frosina, P. Fortini, O. Rossi, F. Carrozzino, G. Raspaglio, L.S. Cox, D.P. Lane, A. Abbondandolo, E. Dogliotti, Two pathways for base excision repair in mammalian cells, J. Biol. Chem. 271 (1996) 9573–9578. [17] Y. Kubota, R.A. Nash, A. Klungland, P. Schär, D. Barnes, T. Lindhal, Reconstitution of DNA base excision-repair with purified human proteins: interaction between DNA polymerase ␤ and the XRCC1 protein, EMBO J. 15 (1996) 6662–6670. [18] F. Dantzer, G. de La Rubia, J. Menissier-De Murcia, Z. Hostomsky, G. de Murcia, V. Schreiber, Base excision repair is impaired in mammalian cells lacking poly (ADP-ribose) polymerase-1, Biochemistry 39 (2000) 7559–7569. [19] M. Stucki, B. Pascucci, E. Parlanti, P. Fortini, S.H. Wilson, U. Hubscher, E. Dogliotti, Mammalian base excision repair by DNA polymerase ␦ and ⑀, Oncogene 17 (1998) 835–843. [20] G.L. Dianov, B.R. Jensen, M.K. Kenny, V.A. Bohr, Repair pathways for processing of 8-oxoguanine in DNA by mammalian cell extracts, Biochemistry 38 (1999) 11021– 11025.

718

F. Dantzer et al. / DNA Repair 2 (2003) 707–718

[21] G.L. Dianov, R. Prasad, S.H. Wilson, V.A. Bohr, Role of DNA polymerase ␤ in the excision step of long-patch mammalian base excision repair, J. Biol. Chem. 274 (1999) 13741–13743. [22] Y. Matsumoto, K. Kim, J. Hurwitz, R. Gary, D.S. Levin, A.E. Tomkinson, M.S. Park, Reconstitution of proliferating cell nuclear antigen-dependent repair of apurinic/apyrimidinic sites with purified human proteins, J. Biol. Chem. 274 (1999) 33703–33708. [23] B. Pascucci, M. Stucki, Z.O. Jonsson, E. Dogliotti, U. Hubscher, Long-patch base excision repair with purified human proteins. DNA ligase I as patch size mediator for DNA polymerase ␦ and ⑀, J. Biol. Chem. 274 (1999) 33696–33702. [24] A.J. Podlutsky, I.I. Dianova, V.N. Podust, V.A. Bohr, G.L. Dianov, Human DNA polymerase ␤ initiates DNA synthesis during long-patch repair of reduced AP-sites in DNA, EMBO J. 20 (2001) 1477–1482. [25] P. Fortini, E. Parlanti, O.M. Sidorkina, J. Laval, E. Dogliotti, The type of DNA glycosylase determines the base excision repair pathway in mammalian cells, J. Biol. Chem. 274 (1999) 15230–15236. [26] E. Cappelli, P. Degan, G. Frosina, Comparative repair of the endogeneous lesions 8-oxo-7,8-dihydroguanine (8-oxoG), uracil and abasic site by mammalian cell extracts: 8-oxoG is poorly repaired by human cell extracts, Carcinogenesis 21 (2000) 1135–1141. [27] G. Dianov, C. Bischoff, J. Piotrowski, V.A. Bohr, Repair pathways for processing of 8-oxoguanine in DNA by mammalian cell extracts, J. Biol. Chem. 273 (1998) 33811– 33816. [28] X. Li, P.M. Wright, A.L. Lu, The C-terminal domain of MutY glycosylase determines the 7,8-dihydro-8-oxoguanine specificity and is crucial for mutation avoidance, J. Biol. Chem. 275 (2000) 8448–8455. [29] T.K. Hazra, J.W. Hill, T. Izumi, S. Mitra, Multiple DNA glycosylases for repair of 8-oxoguanine and their potential in vivo functions, Prog. Nucleic Acid. Res. Mol. Biol. 68 (2001) 193–205. [30] P. Schär, J. Jiricny, Eukaryotic mismatch repair, Nucleic Acids Mol. Biol. 12 (1998) 199–247. [31] S. Oda, O. Humbert, S. Fiumicino, M. Bignami, P. Karran, Efficient repair of A:C mismatches in mouse cells deficient in long-patch mismatch repair, EMBO J. 19 (2000) 1711–1718. [32] S.S. Su, R.S. Lahue, K.G. Au, P. Modrich, Mispair specificity of methyl-directed DNA mismatch correction in vitro, J. Biol. Chem. 263 (1988) 6829–6835. [33] T. Nakahara, Q.M. Zhang, K. Hashiguchi, S. Yonei, Identification of proteins of Escherichia coli and Saccharomyces cerevisiae that specifically bind to C/C mismatches in DNA, Nucleic Acids Res. 28 (2000) 2551–2556. [34] A. Klungland, I. Rosewell, S. Hollenbach, E. Larsen, G. Daly, B. Epe, E. Seeberg, T. Lindahl, D.E. Barnes, Accumulation of premutagenic DNA lesions in mice defective in removal of oxidative base damage, Proc. Natl. Acad. Sci. U.S.A. 96 (1999) 13300–13305.

[35] G. Frosina, P. Fortini, O. Rossi, F. Carrozzino, A. Abbondandolo, E. Dogliotti, Repair of abasic sites by mammalian cell extracts, Biochem. J. 304 (1994) 699–705. [36] E. Seeberg, L. Luna, I. Morland, L. Eide, B. Johnsen, E. Larsen, I. Alseth, F. Dantzer, K. Baynton, R. Aamodt, K.I. Kristiansen, T. Rognes, A. Klungland, M. Bjørås, Base removers and sytrand scissors: different strategies employed in base excision and strand incision at modified base residues in DNA, Cold Spring Harbor Symp. Quant. Biol. 65 (2000) 135–142. [37] O. Minowa, T. Arai, M. Hirano, Y. Monden, S. Nakai, M. Fukuda, M. Itoh, H. Takano, Y. Hippou, H. Aburatani, K. Masumura, T. Nohmi, S. Nishimura, T. Noda, Mmh/Ogg1 gene inactivation results in accumulation of 8-hydroxyguanine in mice, Proc. Natl. Acad. Sci. U.S.A. 97 (2000) 4156–4161. [38] D.O. Zharkov, T.A. Rosenquist, S.E. Gerchman, A.P. Grollman, Substrate specificity and reaction mechanism of murine 8-oxoguanine-DNA glycosylase, J. Biol. Chem. 275 (2000) 28607–28617. [39] M. Bjørås, E. Seeberg, L. Luna, L. Pearl, T. Barrett, Reciprocal “flipping” underlies substrate recognition and catalytic activation by the human 8-oxoguanine DNA glycosylase, J. Mol. Biol. 317 (2002) 171–177. [40] S.D. Bruner, D.P. Norman, G.L. Verdine, Structural basis for recognition and repair of the endogenous mutagen 8-oxoguanine in DNA, Nature 403 (2000) 859–866. [41] F. Le Page, A. Klungland, D.E. Barnes, A. Sarasin, S. Boiteux, Transcription coupled repair of 8-oxoguanine in murine cells: the ogg1 protein is required for repair in non-transcribed sequences but not in transcribed sequences, Proc. Natl. Acad. Sci. U.S.A. 97 (2000) 8397–8402. [42] T.K. Hazra, T. Izumi, R. Venkataraman, Y.W. Kow, M. Dizdaroglu, S. Mitra, Characterization of a novel 8-oxoguanine DNA glycosylase activity in Escherichia coli and identification of the enzyme as endonuclease VIII, J. Biol. Chem. 275 (2000) 27762–27767. [43] J.O. Blaisdell, Z. Hatahet, S.S. Wallace, A novel role for Escherichia coli endonuclease VIII in prevention of spontaneous G:T transversions, J. Bacteriol. 181 (1999) 6396– 6402. [44] A. Parker, Y. Gu, W. Mahoney, S.H. Lee, K.K. Singh, A.L. Lu, Human homolog of the MutY repair protein (hMYH) physically interacts with proteins involved in long-patch DNA base excision repair, J. Biol. Chem. 276 (2001) 5547–5555. [45] H. Yang, W.M. Clendenin, D. Wong, B. Demple, M.M. Slupska, J.H. Chiang, J.H. Miller, Enhanced activity of adenine-DNA glycosylase (Myh) by apurinic/apyrimidinic endonuclease (Ape1) in mammalian base excision repair of an A/GO mismatch, Nucleic Acids Res. 29 (2001) 743–752. [46] Y. Gu, A. Parker, T.M. Wilson, H. Bai, D.-Y. Chang, A.L. Lu, Human MutY homolog (hMYH), a DNA glycosylase involved in base excision repair, physically and functionally interacts with mismatch repair proteins hMSH2/hMSH6, J. Biol. Chem. 277 (2002) 11135–11142.