ABSTRACT
The OGG1 gene of Saccharomyces cerevisiae codes for a DNAglycosylase that excises 7,8-dihydro-8- oxoguanine (8-OxoG) and 2,6-diamino-4-hydroxy-5-N-methylformamidopyrimidine (Fapy) from damaged DNA. In this paper, we have analysed the substrate specificity and the catalytic mechanism of the Ogg1 protein acting on DNA subtrates containing 8-OxoG residues or apurinic/apyrimidinic (AP) sites. The Ogg1 protein displays a marked preference for DNA duplexes containing 8-OxoG placed opposite a cytosine, the rank order for excision of 8-OxoG and cleavage efficiencies being 8-OxoG/C > 8-OxoG/T >> 8-OxoG/G and 8-OxoG/A. The cleavage of the DNA strand implies the excision of 8-OxoG followed by a [beta]-elimination reaction at the 3'-side of the resulting AP site. The Ogg1 protein efficiently cleaves a DNA duplex where a preformed AP site is placed opposite a cytosine (AP/C). In contrast, AP/T, AP/A or AP/G substrates are incised with a very low efficiency. Furthermore, cleavage of 8-OxoG/C or AP/C substrates implies the formation of a reaction intermediate that is converted into a stable covalent adduct in the presence of sodium borohydre (NaBH4). Therefore, the Ogg1 protein is a eukaryotic DNA glycosylase/AP lyase. Sequence homology searches reveal that Ogg1 probably shares a common ancestor gene with the endonuclease III of Escherichia coli. A consensus sequence indicates a highly conserved lysine residue, K120 of endonuclease III or K241 of Ogg1, respectively. Mutations of K241 to Gln (K241Q) and Arg (K241R) have been obtained after site directed mutagenesis of OGG1. Mutation K241Q completely abolishes DNA glycosylase activity and covalent complex formation in the presence of NaBH4. However, the K241Q mutant still binds DNA duplexes containing 8-OxoG/C. In contrast, K241R mutation results in a catalytically active form of Ogg1. These results strongly suggest that the free amino group of Lys241 is involved in the catalytic mechanism of the Ogg1 protein.
Free radicals and reactive oxygen species (ROS) can attack DNA causing base and sugar damage (1 ). Oxidative DNA damage has been suggested to play a role in the etiologies of degenerative pathologies in man such as cancer and aging (2 -4 ). Several lines of evidence suggest that an oxidatively damaged form of guanine, 7,8-dihydro-8-oxoguanine (8-OxoG), is a highly mutagenic DNA lesion (5 ,6 ). Escherichia coli possesses two DNA glycosylases that prevent mutagenesis by 8-OxoG: the Fpg protein which excises 8-OxoG in damaged DNA and the MutY protein which excises the adenine residues incorporated by DNA polymerases opposite 8-OxoG (5 -7 ). Inactivation of both the fpg (mutM) and mutY (micA) genes of E.coli results in a strong GC -> TA mutator phenotype (8 -10 ). In Saccharomyces cerevisiae, the OGG1 gene encodes a protein of 376 amino acids, the Ogg1 protein, which possesses a DNA glycosylase activity that catalyses the removal of 8-OxoG and Fapy residues from damaged DNA (11 ). Recently, we have demonstrated that Ogg1-deficient strains of S.cerevisiae exhibit a mutator phenotype and specifically accumulate GC -> TA transversions (12 ). These results show that base excision repair of 8-OxoG, by proteins like Fpg or Ogg1, protects genomes from the deleterious action of ROS in prokaryotes or in eukaryotes (13 ).
The Fpg protein of E.coli was initially isolated as a DNA glycosylase which excises Fapy residues in alkylated DNA (14 ,15 ). The Fpg protein also releases other damaged purines such as 2,6-diamino-4-hydroxy-5-formamidopyrimidine (Fapy-G), 4,6-diamino-5-formamidopyrimidine (Fapy-A) and 8-oxoG (16 -19 ). Besides its DNA glycosylase activity, the Fpg protein has a physically associated activity which incises DNA at apurinic/ apyrimidinic (AP) sites and another activity which removes 5'-terminal deoxyribose-phosphate from DNA (20 -22 ). The AP nicking mechanism implies successive [beta]- and [delta]-elimination reactions leaving a single nucleoside gap in DNA (20 ,23 ).
The Ogg1 protein of S.cerevisiae which excises Fapy and 8-OxoG is also endowed with an AP nicking activity (11 ). It was proposed that the strand cleavage at the 3' side of an AP site occurs via a [beta]-elimination reaction (11 ,24 ). Several studies have led to the emergence of a unified catalytic mechanism for DNA glycosylases/AP lyases such as Fpg protein or endonuclease III of E.coli (25 ,26 ). These enzymes proceed by a nucleophilic attack at the sugar of the damaged base nucleotide using an imino group as the attacking nucleophile. The consequence is the formation of an imino enzyme-DNA intermediate as the product of the glycosylase step. Sodium borohydride has been used to trap this intermediate in reactions catalysed by DNA glycosylases/AP lyases (24 -27 ).
In this study, we report an analysis of the substrate specificity and of the catalytic mechanism of the Ogg1 protein of S.cerevisiae. The results show that the Ogg1 protein has a marked preference for lesions, either 8-OxoG or AP sites, placed opposite a cytosine in DNA. The mechanisms of strand cleavage for both 8-OxoG/C and AP/C duplexes imply the formation of a Schiff base intermediate which can be converted into a stable adduct in the presence of NaBH4. The results also show that lysine 241 is a crucial residue for the catalytic activities of the Ogg1 protein but not for DNA binding.
Escherichia coli strains BH410 (fpg::Kanr) and PR195 (fpg::Kanr mutY::Kanr) and plasmid pYSB10 containing the OGG1 gene of S.cerevisiae were from our laboratory (11 ,28 ,29 ). Endonuclease III (Nth), endonuclease IV (Nfo), uracil DNA glycosylase and Fpg protein from E.coli were purified from overproducing strains. Restriction endonucleases, DNA polymerases, T4 polynucleotide kinase and T4 DNA ligase were from commercial sources.
Plasmid pYSB10 was used as a template in a PCRreaction to amplify OGG1. Primer 1 was used to engineer an EcoRI restriction site at the beginning of OGG1 (start codon is in bold). Primer 2 was used to introduce a HindIII restriction site at the end of OGG1 (stop codon is in bold).
Primer 1: 5'-CCGGAATTCATGTCTTATAAATTCGG-3'
Primer 2: 5'-GCCCAAGCTTCTAATCTATTTTTGCTTC-3'
The amplified fragment containing the OGG1 gene was cloned into plasmid pKK223-3 (Pharmacia) after digestion by EcoRI and HindIII restriction enzymes yielding the pYSB160 plasmid. The sequence of the OGG1 gene in pYSB160 does not reveal mutations when compared with the published sequence of the OGG1 gene (11 ).
The Altered site II Systems (Promega) was used to introduce single base pair mutations in the OGG1 gene. For this purpose, OGG1 from pYSB160 was cloned in the pALTER-1 vector between the EcoRI and HindIII sites. The following oligonucleotides were used to generate the mutant Ogg1 proteins where the base resulting in the Lys (K) -> Gln (Q) or Lys -> Arg (R) substitutions are indicated in bold type and underlined: K241Q protein, GGTGTAGGCCCC
The Ogg1 protein was purified from isopropyl [beta]-d-thiogalactopyranoside (IPTG)-induced E.coli BH410 (fpg-) hosting pYSB160. The cells (29 g) were resuspended in 200 ml of Buffer A (25 mM Tris-HCl pH 7.6, 2 mM Na2EDTA and 5% glycerol) containing 250 mM NaCl. The cell suspension was supplemented with lysozyme (1 mg/ml) and incubated at 0oC for 20 min, followed by 15 min at 37oC and 15 min at -80oC. The resulting lysate was centrifuged at 130 000 g for 45 min at 4oC. The supernatant fraction was taken as crude extract (fraction I: 200 ml). Fraction I was loaded onto a QMA-Acell anion exchange column (Waters) equilibrated with Buffer A containing 250 mM NaCl. The Fapy DNA glycosylase activity was not retained under these conditions (fraction II: 380 ml). Fraction II was dialysed against Buffer A containing 50 mM NaCl and applied to a Phospho-Ultrogel A6 column (IBF-LKB). Proteins were eluted with a linear salt gradient (50-600 mM NaCl). Fractions containing the activity eluted at 350 mM NaCl (fraction III: 130 ml). Fraction III was precipitated using ammonium sulfate, 500 g/l (w/v). The precipitate was collected by centrifugation and resuspended in Buffer A containing 1 M NaCl and loaded onto an AcA54 gel filtration column (IBF-LKB). Fractions containing the activity (fraction IV: 43 ml) were dialysed against Buffer A containing 100 mM NaCl and loaded onto a double-stranded DNA cellulose column (Sigma). Proteins were eluted with a linear salt gradient (100-800 mM NaCl). The active fractions eluted at 500 mM NaCl (fraction V: 12 ml). Fraction V was dialysed against Buffer A containing 100 mM NaCl and loaded onto a MonoS HR5/5 (FPLC, Pharmacia). Proteins were eluted with a salt gradient (100-800 mM NaCl). Fractions containing the activity eluted at 450 mM NaCl (fraction VI: 4 ml). Protein concentration was measured according to Bradford (30 ).
The [3H]Fapy-poly(dG-dC) was prepared as previously described (31 ). The assay mixture (100 [mu]l) was composed of 25 mM Tris-HCl pH 7.6, 100 mM KCl, [3H]Fapy-poly(dG-dC) and Ogg1 protein. The reaction was carried out at 37oC for 15 min. Ethanol-soluble radioactive material was quantitated and the chemical nature of this material was monitored by HPLC (31 ).
The assay mixture (50 [mu]l) contained 25 mM Tris-HCl pH 7.6, 100 mM KCl, 10 pmol 8-OxoG/N duplex (34mer DNA duplexes containing a single 8-OxoG placed opposite each of the four DNA bases in the complementary strand) and Ogg1 protein. The reaction was carried out at 37oC for 15 min. The products of the reaction were analysed by HPLC with electrochemical detection as previously described (11 ).
The 34mer oligonucleotide containing a single 8-oxoG was synthesized as previously described (32 ): 5'-GGCTTCATCGTTATT(8-OxoG)ATGACCTGGTGGATACCG-5'*. Complementary sequences with a C, T, G or A placed opposite 8-OxoG after hybridization were also synthesized. The nucleotide at the 3'-end was inverted yielding a 5'-(N)n-3'-P-3'-N-5'* sequence with two 5'-ends (33 ). Thus, the 34mer containing 8-OxoG was labelled at both ends using [[gamma]-32P]ATP (3000 Ci/mmol) and T4 polynucleotide kinase (32 ,33 ). The labeled strand was hybridized with a complementary sequence yielding 8-OxoG/N duplexes. The assay mixture (20 [mu]l) contained 25 mM Tris-HCl pH 7.6, 2 mM Na2EDTA, 100 mM KCl, 50 fmol of 32P-labeled 8-OxoG/N duplex and Ogg1 protein. The reactions were performed at 37oC for 15 min. The products of the reactions were separated by 20% PAGE containing 7 M urea. Cleavage was quantitated after scanning of the autoradiographs using the NIH V1.59 software.
The 30mer oligonucleotide containing a single uracil residue and the four complementary sequences with C, T, G or A placed opposite the uracil after hybridization were of commercial origin: 5'-TACGGATCGCAG(U)TGGGTTAGGGAAGTTGG-3'. This 30mer oligonucleotide was 32P-labeled at the 5'-end and annealed with one of the four complementary sequences yielding U/N duplexes. To generate AP sites, 500 fmol of each U/N duplex was incubated with 20 ng of purified uracil DNA glycosylase for 15 min at 37oC yielding the AP/N duplexes (30mer DNA duplexes containing a single AP site placed opposite each of the four DNA bases in the complementary strand). Assay conditions are as described for 8-OxoG/C nicking assay.
The trapping assays were performed using either 8-OxoG/N (Reaction 1) or AP/N (Reaction 2) duplexes as substrates. Reaction 1 (20 [mu]l) contained 25 mM Tris-HCl pH 7.6, 2 mM Na2EDTA, 100 fmol labelled 8-OxoG/N duplex and 50 mM of either NaBH4, NaCNBH3 or NaCl. Finally, 50 ng of either Ogg1 or Fpg protein was added to the mixture. The reaction was carried out at 37oC for 20 min. Reaction 2 (20 [mu]l) contained 25 mM Tris-HCl buffer pH 7.6, 2 mM Na2EDTA and 100 fmol labeled AP/N duplex, 50 mM NaBH4 and 50 ng of Ogg1 protein. Immediately before the reaction, the Ogg1 protein was premixed with NaBH4 and added to the rest of the assay mixture. The reaction was carried out at 4oC for 20 min followed by 5 min at 37oC. The reactions were stopped by addition of 10 [mu]l SDS-PAGE loading buffer/formamide blue dye and heating at 90oC for 3 min. The products of the reactions were separated onto 10-20% gradient SDS-PAGE and analysed as previously described.
Binding reactions were performed at 4oC for 20 min. The reaction mixture (20 [mu]l) contained 50 mM Tris-HCl pH 7.6, 100 mM KCl, 2 mM Na2EDTA, 5% glycerol, 0.2 mg/ml poly(dI-dC), 100 pM (8000 c.p.m.) of 32P-labelled 8-OxoG/C oligonucleotide duplex and 10 [mu]g of proteins from bacterial crude lysate. The samples were analysed using non-denaturing 10% PAGE as described by Castaing et al. (34 ).
Polyclonal antibodies against Ogg1 were obtained after immunisation of female New Zealand rabbits. Antibodies were purified from the antiserum and the resulting preparation was referred to as BCE-1691.Total proteins from bacterial extracts were separated using 12.5% SDS-PAGE and transferred to Hybond-C membranes (Amersham). Western blots were performed using BCE-1691 at a dilution of 1:10 000 in phosphate buffered saline-0.1% Tween 20 (pH 7.4). Western blots were then developed using an anti-rabbit horseradish peroxidase-conjugated (Amersham) and detected with the BM chemiluminescence blotting substrate (POD) system (Boehringer).
PR195 cells harbouring either the pKK223-3, pYSB160, pYSB161 or pYSB162 vectors were grown in LB broth containing 100 [mu]g/ml ampicillin and 0.5 mM IPTG. To determine the spontaneous frequencies of rifampicin-resistant mutants and lactose revertants, appropriate dilutions of at least 10 overnight cultures, originally inoculated with 103 cells, were plated on either LB agar, LB agar with rifampicin (100 [mu]g/ml), or minimal lactose (0.2%) agar. Colonies were counted the following day in the case of the LB plates and 48 h after plating for the minimal lactose plates.
To overproduce the Ogg1 protein, the coding sequence of the OGG1 gene of S.cerevisiae was PCR-amplified and cloned in the expression vector pKK223-3 yielding the pYSB160 plasmid. The Ogg1 protein was purified from IPTG-induced E.coli BH410 (fpg::Kanr) cells harbouring pYSB160 (Table 1 ). The release of Fapy residues from [3H]Fapy-poly(dG-dC) was used as an activity assay during the course of the purification. The purity of the final fraction VI was assessed by the observation of a single protein band on a SDS-PAGE with a molecular weight of 43 kDa (Fig. 1 ). This 43 kDa polypeptide was transfered to a membrane and the N-terminal sequence was determined. The unique N-terminal sequence for the 10 first amino acids was (Ser-Tyr-Lys-Phe-Gly-Lys-Leu-Ala-Ile-Asn). This sequence is identical to that deduced from the nucleotide sequence of OGG1 which is confirmed as the structural gene coding for the Ogg1 protein.
Our study shows that the Ogg1 protein is a DNA glycosylase/AP lyase that cleaves DNA after the removal of a damaged base. However, we have not demonstrated that the Ogg1 protein possesses an activity that nicks DNA at preformed AP sites. Therefore, we have prepared 30mer DNA substrates, AP/N, in which a single AP site was placed opposite each of the four DNA bases. Figure 5 shows that the Ogg1 protein very efficiently cleaves the AP/C duplex where a cytosine was placed opposite the AP site. Indeed the cleavage efficiency of the AP/C substrate is similar to that of the 8-OxoG/C in a parallel reaction (Fig. 5 , insert). In contrast, the AP/T duplex is a very poor substrate for the Ogg1 protein and AP/G and AP/A duplexes are not incised at all (Fig. 5 ). Furthermore, the reduction of the AP site with NaBH4 before the reaction with Ogg1 makes the AP/C duplex resistant to cleavage by the Ogg1 protein (data not shown). These results strongly suggest that the Ogg1 protein possesses an intrinsic AP lyase activity.
The Ogg1 protein belongs to a family of base-excision repair proteins related to the endonuclease III of E.coli (24 ). The sequence similarity between these two proteins is limited to a region spanning amino acid residues 232-265 and 111-143 of Ogg1 and endonuclease III, respectively. This region corresponds to the helix-hairpin-helix (HhH) structural motif and to the catalytic active site of the E.coli endonuclease III (38 ). Sequence alignment suggests that lysine 241 of Ogg1 corresponds to lysine 120 of endonuclease III (24 ). On these basis, lysine 241 of Ogg1 was used as a target for substitution to yield the K241Q (lysine -> glutamine) and K241R (lysine -> arginine) mutants. The various Ogg1 proteins were expressed in E.coli BH410 from pYSB160 (K241W.T.), pYSB161(K241R) or pYSB162 (K241Q) plasmids, respectively. Figure 7 shows that cell free extracts containing the wild-type Ogg1 or the K241R mutant cleave the 8-OxoG/C duplex and form a high molecular weight covalent complex in the presence of NaBH4 (Fig. 7 a and b). In contrast, cell free extracts containing K241Q mutants do not incise nor form retardation complexes with 8-OxoG/C (Fig. 7 a and b). Western-blot analysis using BCE-1691 antibodies shows that the wild-type and the two mutant Ogg1 proteins are present in similar amounts in cell free extracts (Fig. 7 c). These results strongly suggest that the replacement of Lys241 by Arg241 generates a catalytically functional protein, perhaps less active. In contrast, replacement of Lys241 by Gln241 results in a protein without catalytic activity. The expression vectors pKK223-3, pYSB160 (K241W.T.), pYSB161 (K241Q) and pYSB162 (K241R) were transformed in E.coli PR195 which displays a strong spontaneous mutator phenotype. The frequencies of either mutation to rifampicine resistance or reversion to lac+ were determined. The results shown in Table 2 demonstrate that expression of wild-type Ogg1 protein (K241W.T.) and K241R mutant in PR95 E.coli strain reduces the mutation frequencies for both markers analysed whereas expression of K241Q mutant does not. Thus, the cleavage and trapping efficiency of K241W.T., K241R and K241Q proteins parallels the antimutator effect of these proteins. Nevertheless, the lack of activity of the K241Q mutant could be due either to a defect in the catalytic mechanism or to an incapacity to bind modified DNA. Figure 8 shows that the wild-type as well as the mutant K241R or K241Q Ogg1 proteins bind the 34mer 8-OxoG/C duplex. The shifted bands were found to migrate at the same position either with purified Ogg1 protein or with the various cell extracts containing wild-type or mutant Ogg1 proteins (Fig. 8 ). These latter observations suggest that lysine 241 has a functional role rather than a structural role in the catalytic activities of Ogg1 protein.
Table 2
In this study, we have purified theOgg1 protein of S.cerevisiae from E.coli (fpg-) cells harboring the overproducing plasmid pYSB160. The Ogg1 protein is a monomer of 43 kDa which is endowed with two enzymatic activities, (i) a DNA glycosylase activity which excises Fapy and 8-OxoG, and (ii) an AP lyase activity which incises DNA at AP sites either preformed or resulting from the excision of a damaged base. The catalytic mechanism of the Ogg1 protein for the cleavage of DNA substrates containing either 8-OxoG or an AP site implies the formation of a transient enzyme-DNA intermediate which can be converted into a stable covalent adduct in the presence of NaBH4. This is the first demonstration of a covalent intermediate between a DNA glycosylase and a preformed AP site. Therefore, these results suggest a unique mechanism for cleavage of DNA by the Ogg1 protein, (i) after excision of a damaged base and (ii) at a preformed AP site. Our data also show that the Ogg1 protein reacts preferentially with 8-OxoG and AP sites placed opposite a cytosine. This strong bias in favor of lesions placed opposite a cytosine is a hallmark of the Ogg1 protein and could reflect the biological properties of the Ogg1 protein that is required to repair endogenous mutagenic DNA damage thus preventing G.C -> T.A transversions (12 ). It is worth noting that the Fpg protein, which is the functional homologue of the Ogg1 protein in bacteria, incises AP/N duplexes at approximatively the same rate (unpublished results). On the other hand, the rank order for cleavage efficiency by the Fpg protein is as follows, 8-Oxo/C = 8-OxoG/T > 8-OxoG/G >> 8-OxoG/A (32 ). Furthermore, the mechanism of strand cleavage at AP sites is also different since Ogg1 proceeds via a [beta]-elimination reaction whereas Fpg proceeds via [beta]- and [delta]-elimination reactions, respectively. To conclude, the substrate specificities and the catalytic mechanism of Fpg and Ogg1 are significantly different which, in turn, may suggest that these two proteins are not closely related.
The authors thank the Commissariat l'Energie Atomique and the Centre National de la Recherche Scientifique UMR217. This work was supported by the ACC-SV8 (S.B.) and the Comit de Radioprotection of EDF (RB96-09). The authors thank Drs J.Laval and T.R.O'Connor (Villejuif, France) for the kind gift of uracil DNA glycosylase and endonuclease IV of E.coli and Dr P.Nehls (Essen, Germany) for the kind gift of the oligonucleotide containing 8-OxoG. We thank P.Auffret van der Kemp and C.Dherin for their excellent technical assistance.
*To whom correspondence should be addressed. Tel: +33 1 46 54 88 58; Fax: +33 1 46 54 88 59; Email: boiteux@dsvidf.cea.fr
Vector
Cloned gene
Rifr/108
Lac+/108
pKK223-3
none
118 +- 18
2027 +- 263
pYSB160
OGG1
13 +- 1
83 +- 7
pYSB161
OGG1K241Q
185 +- 61
1698 +- 300
pYSB162
OGG1K241R
12 +- 3
620 +- 132
REFERENCES