Role of 3-methyladenine-DNA glycosylase in host-cell reactivation of methylated T7 bacteriophage

Purified T7 phage, treated with methyl methanesulfonate, was assayed on four Escherichia coli K12 host cells: (1) AB1157, wild-type; (2) PK432-1, lacking 3-methyladenine-DNA glycosylase (tag); (3) NH5016, lacking apurinic endonuclease VI (xthA); (4) p3478, lacking DNA polymerase I (polA), the latter three strains being deficient in enzymes of the base excision repair pathway. For inactivation measured immediately after alkylation, phage survival was lowest on strains PK432-1 and p3478; for delayed inactivation, measured after partial depurination of alkylated phage, survival was much lower on strain p3478 than on PK432-1. These results demonstrate the important role played by 3-methyladenine-DNA glycosylase in the survival of methylated T7 phage. Quantitative analysis of the data, using the results of Verly et al. (Verly, W.G., Crine, P., Bannon, P. and Forget, A. (1974) Biochim. Biophys. Acta 349, 204–213) to correlate the dose with the number of methyl groups introduced into phage DNA, revealed that 5–10 3-methyladenine residues per T7 DNA constituted an inactivation hit for the tag mutant. Thus, 3-methyladenine may be as toxic a lesion as an apurinic site.     

Excision repair of adozelesin-N3 adenine adduct by 3-methyladenine-DNA glycosylases and UvrABC nuclease

Adozelesin is a synthetic analog of the antitumor antibiotic CC-1065, which alkylates the N3 of adenine in the minor groove in a sequence-selective manner. Since the cytotoxic potency of a DNA alkylating agent can be modulated by DNA excision repair system, we investigated whether nucleotide excision repair (NER) and base excision repair (BER) enzymes are able to excise the bulky DNA adduct induced by adozelesin. The UvrABC nuclease and 3-methyladenine-DNA glycosylase, that exhibit a broad spectrum of substrate specificity, were selected as typical NER and BER enzymes, respectively. The adozelesin-DNA adduct was first formed in the radiolabeled restriction DNA fragment and its excision by purified repair enzymes was monitored on a DNA sequencing gel. The treatment of the DNA adduct with a purified UvrABC nuclease and sequencing gel analysis of cleaved DNA showed that UvrABC nuclease was able to incise the adozelesin adduct. The incision site corresponded to the general nuclease incision site. Excision of this adduct by 3-methyladenine-DNA glycosylases was determined following the treatment of the DNA adduct with a homogeneous recombinant bacterial, rat and human 3-methyladenine-DNA glycosylases. Abasic sites generated by DNA glycosyalses were cleaved by the associated lyase activity of the E. coli formamidopyrimidine-DNA glycosylase (Fpg). Resolution of cleaved DNA on a sequencing gel showed that the DNA glycosylase from different sources could not release the N3-adenine adducts. A cytotoxicity assay using E. coli repair mutant strains showed that E. coli mutant strains defective in the uvrA gene were more sensitive to cell killing by adozelesin than E. coli mutant strain defective in the alkA gene or the wild type. These results suggest that the NER pathway seems to be the major excision repair system in protecting cells from the cytotoxicity of adozelesin.     

Pseudomonas putida AlkA and AlkB Proteins Comprise Different Defense Systems for the Repair of Alkylation Damage to DNA – In Vivo,In Vitro,and In Silico Studies

Alkylating agents introduce cytotoxic and/or mutagenic lesions to DNA bases leading to induction of adaptive (Ada) response, a mechanism protecting cells against deleterious effects of environmental chemicals. In Escherichia coli, the Ada response involves expression of four genes: ada, alkA, alkB, and aidB. In Pseudomonas putida, the organization of Ada regulon is different, raising questions regarding regulation of Ada gene expression. The aim of the presented studies was to analyze the role of AlkA glycosylase and AlkB dioxygenase in protecting P. putida cells against damage to DNA caused by alkylating agents. The results of bioinformatic analysis, of survival and mutagenesis of methyl methanesulfonate (MMS) or N-methyl-N’-nitro-N-nitrosoguanidine (MNNG) treated P. putida mutants in ada, alkA and alkB genes as well as assay of promoter activity revealed diverse roles of Ada, AlkA and AlkB proteins in protecting cellular DNA against alkylating agents. We found AlkA protein crucial to abolish the cytotoxic but not the mutagenic effects of alkylans since: (i) the mutation in the alkA gene was the most deleterious for MMS/MNNG treated P. putida cells, (ii) the activity of the alkA promoter was Ada-dependent and the highest among the tested genes. P. putida AlkB (PpAlkB), characterized by optimal conditions for in vitro repair of specific substrates, complementation assay, and M13/MS2 survival test, allowed to establish conservation of enzymatic function of P. putida and E. coli AlkB protein. We found that the organization of P. putida Ada regulon differs from that of E. coli. AlkA protein induced within the Ada response is crucial for protecting P. putida against cytotoxicity, whereas Ada prevents the mutagenic action of alkylating agents. In contrast to E. coli AlkB (EcAlkB), PpAlkB remains beyond the Ada regulon and is expressed constitutively. It probably creates a backup system that protects P. putida strains defective in other DNA repair systems against alkylating agents of exo- and endogenous origin.     

Excision Repair of 2,5-Diaziridinyl-1,4-Benzoquinone (DZQ)-DNA Adduct by Bacterial and Mammalian 3-Methyladenine-DNA Glycosylases

The mechanisms of anticancer activity of 2,5-diaziridinyl-1,4-benzoquinone (DZQ) are believed to involve the alkylation of guanine and adenine bases. In this study, it has been investigated whether bacterial and mammalian 3-methyladenine-DNA glycosylases are able to excise DZQ-DNA adduct with a differential substrate specificity. DZQ-induced DNA adduct was first formed in the radiolabeled restriction enzyme DNA fragment, and excision of the DNA adduct was analyzed following treatment with homogeneous 3-methyladenine-DNA glycosylase from E. coli, rat, and human, respectively. Abasic sites generated by DNA glycosylases were cleaved by the associated lyase activity of the E. coli formami-dopyrimidine-DNA glycosylase. Resolution of cleaved DNA on a sequencing gel with Maxam-Gilbert sequencing reactions showed that DZQ-induced adenine and guanine adducts were very good substrates for bacterial and mammalian enzymes. The E. coli enzyme excises DZQ-induced adenine and guanine adducts with similar efficiency. The rat and human enzymes, however, excise the adenine adduct more efficiently than the guanine adduct. These results suggest that the 3-methyladenine-DNA glycosylases from different origins have differential substrate specificity to release DZQ-DNA lesions. The use of 3-methyladenine-DNA glycosylase incision analysis could possibly be applied to quantify a variety of DNA adducts at the nucleotide level.     

In vitro host cell reactivation of alkylated bacteriophage T7 deoxyribonucleic acid by repair-deficient strains of Escherichia coli.

An in vitro system capable of packaging bacteriophage T7 deoxyribonucleic acid (DNA) into phage heads to form viable phage particles has been used to monitor the biological consequences of DNA dam aged by alkylating agents, and an in vitro DNA replication system has been used to examine the ability of alkylated T7 DNA to serve as template for DNA synthesis. The survival of phage resulting from in vitro packaging of DNA preexposed to various concentrations of methyl methane sulfonate or ethyl methane sulfonate closely paralleled the in vivo situation, in which intact phage were exposed to the alkylating agents. Host factors responsible for survival of alkylated T7 have been examined by using wild-type strains of EScherichia coli and mutants deficient in DNA polymerase I (polA) or 3-methyladenine-DNA glycosylase (tag). For both in vivo and in vitro situations, a deficiency in 3-methyladenine-DNA glycosylase dramatically reduced phage survival relative to that in the wild type, whereas a deficiency in DNA polymerase I had an intermediate effect. Furthermore, when the tag mutant was used as an indicator strain, phage survival was enhanced when alkylated DNA was packaged with extracts prepared from a wild-type strain in place of the tag mutant or by complementing a tag extract with an uninfected tag+ extract, indicating in vitro repair during packaging.     

Induction of SOS and adaptive responses by alkylating agents in Escherichia coli mutants deficient in 3-methyladenine-DNA glycosylase activities

The induction of SOS and adaptive responses by alkylating agents was studied in Escherichia coli mutants tagA and alkA deficient in 3-methyladenine-DNA glycosylase activities. The SOS response was measured using an sfiA::lacZ operon fusion. The sfiA operon, in the double mutant tagA alkA, is induced at 5-50-fold lower concentrations of all tested methylating and ethylating compounds, as compared to the wild-type strain. In all cases, the tagA mutation, which inactivates the constitutive and specific 3-alkyladenine-DNA glycosylase I (TagI), sensitizes the strain to the SOS response. The sensitization effect of alkA mutation, which inactivates the inducible 3-alkyladenine-DNA glycosylase II (TagII), is observed under conditions which allow the induction of the adaptive response. We conclude that the persistence of 3-methyladenine and 3-ethyladenine residues in DNA most likely leads to the induction of the SOS functions. In contrast, the adaptive response, evaluated by O6-methylguanine-DNA methyltransferase activity in cell extracts, was not affected by either tagA or alkA mutations. The results suggest that the SOS and adaptive responses use different alkylation products as an inducing "signal". However, adaptation protein TagII inhibits the induction of the SOS response to some extent, due to its action at the level of signal production. Finally, we provide conditions to improve short-term bacterial tests for the detection of genotoxic alkylating agents.     

Effect ofmutY andmutM/fpg-1 mutations on starvation-associated mutation inEscherichia coli: implications for the role of 7,8-dihydro-8-oxoguanine

MutY specifies a DNA glycosylase that removes adenines unnaturally paired with various bases including oxidized derivatives of guanine, such as 7,8-dihydro-8-oxoguanine (8-oxoG). The rate of mutation in starvedEscherichia coli cells is markedly raised inmutY mutants defective in this glycosylase. As predicted, the mutations produced include G to T transversions. Bacteria carryingmutM orfpg-1 mutations (defective in Fapy glycosylase, which removes oxidized guanine residues such as 8-oxoG) show little or no enhancement of mutation under starvation conditions. When present together withmutY, however,mutM clearly further enhances the rate of mutation in starved cells. Plasmids resulting in overproduction of MutY or Fapy glycosylases reduce the rate of mutation in starved cells. We conclude that, in non-growing bacteria, oxidized guanine residues, including 8-oxoG, constitute an important component of spontaneous mutation. Addition of catalase to the plates did not reduce the mutant yield, indicating that extracellular hydrogen peroxide is not involved in the production of the premutational damage. Singlet oxygen, known to give rise to 8-oxoG, may be the ultimate oxidative species.     

Characterization of Staphylococcus epidermidis mutants sensitive to ultraviolet radiation

Five UV-sensitive mutants obtained by N-methyl-N′-nitro-N-nitrosoguanidine (MNNG) treatment of the Staphyloccocus epidermidis W5 strain were characterized phenotypically by assaying their UV- and MNNG-sensitivities, lysogenic inducibility, host-cell reactivation and Weigle reactivation capacities. The results were compared with those of well-characterized Escherichia coli strains, permitting the identification of: 2 mutants that behave as Uvr⁻ Umu⁻; 1 mutant that appears analogous to Uvr⁻; 1 mutant that resembles LexA⁻ and 1 mutant that exhibits a RecA⁻ phenotype. The study of these mutants can contribute to the understanding of the repair mechanisms in S. epidermidis.     

Induction of transversion mutations in Escherichia coli by N-methyl-N''-nitro-N-nitrosoguanidine is SOS dependent.

Escherichia coli alkA mutants, which are deficient for an inducible DNA glycosylase, 3-methyladenine-DNA glycosylase II, are sensitive to mutagenesis by low doses of the alkylating agent N-methyl-N'-nitro-N-nitrosoguanidine (MNNG). As many as 90% of the alkA-dependent mutations induced by MNNG are also umuC+ dependent and thus are due to DNA lesions that are substrates for the mutagenic functions of the SOS response. A great number of these mutations are base substitutions at A . T sites, particularly A . T transversions. We discuss which DNA lesions may be responsible for these mutations. Our results show that the induction of 3-methyladenine-DNA glycosylase II, which occurs as part of the adaptive response to alkylating agents such as MNNG, significantly reduces the mutagenicity as well as the lethality of alkylation damage.     

The transfer RNA processing defect in Escherichia coli strains BN and CAN is not due to a mutation in RNAase D or RNAase II

Escherichia coli strains BN and CAN are unable to support the growth of bacteriophage T4 psu₁⁺-amber double mutants. For strain BN, this phenotype has been attributed to a defect in 3′ processing of the precursor to psu₁⁺ tRNA^Ser. Since RNAase D and RNAase II are the only well-characterized 3′ exoribonucleases to be implicated in tRNA processing, the status of these activities and their genes in the mutant strains was investigated. Although extracts of strains BN and CAN were defective for hydrolysis of the artificial tRNA precursor, tRNA-C-U, these strains contained normal levels of RNAase D and RNAase II, and purified RNAase D or RNAase II could only partially complement the mutant extracts. Introduction of the wild-type RNAase D gene into strains BN and CAN did not correct the mutant phenotype. Likewise, strains defective in RNAase D and/or RNAase II plated T4psu₁⁺-amber phage normally. These results indicate that the tRNA processing defect in strains BN and CAN is not due to a mutation in either RNAase U or RNAase II. The possibility that the mutation in these strains affects another exoribonuclease or a factor influencing the activity and specificity of RNAase D or RNAase II is discussed.     

3-Methyladenine residues in DNA induce the SOS function sfiA in Escherichia coli.

The induction by methylating agents of the SOS function sfiA was measured by means of a sfiA::lac operon fusion in Escherichia coli mutants defective in alkylation repair. The sfiA operon was turned on at a 10-fold lower concentration of methylmethane sulfonate or dimethyl sulfate in tagA strains, lacking specific 3-methyladenine-DNA glycosylase, than in wild-type strains. In contrast, the induction of sfiA by u.v. light was not affected by a tagA mutation. We confirm that tagA strains specifically accumulate 3-methyladenine in their DNA. We conclude that the persistence of 3-methyladenine in E. coli DNA most likely induces the SOS functions. Results on in vitro DNA synthesis further suggest that this induction is due to an unscheduled arrest of DNA synthesis at this lesion.     

Isolation of a formamidopyrimidine-DNA glycosylase (fpg) mutant of Escherichia coli K12

Summary The fpg ⁺ gene of Escherichia coli coding for formamidopyrimidine-DNA glycosylase was previously cloned on a multicopy plasmid. The plasmid copy of the fpg ⁺ gene was inactivated by cloning a kanamycin resistance gene into the open reading frame, yielding the fpg-1:: Kn^r mutation. This mutation was transferred to the chromosome in the following steps: (i) linearization of the plasmid bearing the fpg-1::Kn^r mutation and transformation of competent bacteria (recB recC sbcB); (ii) selection for chromosomal integration of the fpg-1::Kn^r mutation; (iii) phage P1 mediated transduction of the fpg-1::Kn^r mutation in the AB1157 background. The resulting fpg ^- mutant exhibited no detectable Fapy-DNA glycosylase activity in crude lysates. The insertion mutation was localized by means of genetic crosses between mtl and pyrE, at 81.7 min on the E. coli linkage map. Sequence analysis confirmed this mapping and further showed that fpg is adjacent to rpmBG in the order fpg, rpmGB, pyrE. The formamidopyrimidine-DNA glycosylase defective strain does not show unusual sensitivity to the following DNA damaging treatments: (i) methylmethanesulfonate, (ii) N-methyl-N-nitro-N-nitrosoguanidine, (iii) ultraviolet light, (iv) -radiation. The fpg gene is neither part of the SOS regulon nor the adaptive response to alkylating agents.     

Effects of mutations in genes for proteins involved in disulphide bond formation in the periplasm on the activities of anaerobically induced electron transfer chains in Escherichia coli K12

The assembly of anaerobically induced electron transfer chains in Escherichia coli strains defective in periplasmic disulphide bond formation was investigated. Strains deficient in DsbA, DsbB or DipZ (DsbD) were unable to catalyse formate-dependent nitrite reduction (Nrf activity) or synthesize any of the known c-type cytochromes. The Nrf⁺ activity and cytochrome c content of mutants defective in DsbC, DsbE or DsbF were similar to those of the parental, wild-type strain. Neither DsbC expressed from a multicopy plasmid nor a second mutation in dipZ (dsbD) was able to compensate for a dsbA mutation by restoring nitrite reductase activity and cytochrome c synthesis. In contrast, only the dsbB and dipZ (dsbD) strains were defective in periplasmic nitrate reductase activity, suggesting that DsbB might fulfil an additional role in anaerobic electron transport. Mutants defective in dipZ (dsbD) were only slightly more sensitive to Cu⁺⁺ ions at concentrations above 5?mM than the parental strain, but strains defective in DsbA, DsbB, DsbC, DsbE or DsbF were unaffected. These results are consistent with our earlier proposals that DsbA, DsbB and DipZ (DsbD) are part of the same pathway for ensuring that haem groups are attached to the correct pairs of cysteine residues of apocytochromes c in the E. coli periplasm. However, neither DsbE nor DsbF are essential for the reduction of DipZ (DsbD).     

DNA intermediates at the Escherichia coli replication fork. II. Studies using dut and ung mutants in vitro.

The incorporation of uracil into and excision from DNA were studied in vitro using lysates on cellophane discs made from Escherichia coli strains with defects in the enzymes dUTPase (dut) and uracil-DNA glycosylase (ung).Results with dut ung lysates indicate that dUTP is competitively incorporated with dTTP at the replication fork. Such incorporation is not due to DNA polymerase I. There is a mild discrimination (2.5-fold) against incorporation of dUTP versus dTTP. These data, together with in vivo uracil incorporation data (Tye et al., 1978) permit a rough estimate of the pool of dUTP in vivo (~0.5% of the dTTP pool).These in vitro data indicate that uracil-DNA glycosylase is the initial step in at least 90% of uracil excision events. However, in a strain defective in uracil-DNA glycosylase (ung-1), uracil-containing DNA is still more subject to single-strand scission than non-uracil-containing DNA, albeit at a rate at least tenfold less than in an ung⁺ strain.A number of qualitative statements may also be made about different steps in uracil incorporation and subsequent excision and repair events. When high levels of dUTP are added in vitro, a dut ung⁺ strain has a higher steady-state level of uracil in newly synthesized DNA than does an isogenic dut⁺ ung strain. Thus the dUTPase in these lysates has a higher capacity to be overloaded than does the excision system (i.e. uracil DNA glycosylase). However, the DNA sealing system (presumably DNA polymerase I and DNA ligase) apparently can handle all single-strand interruptions being introduced by uracil excision at the maximal rate, at least so that DNA synthesis can continue.     

Induction of the SOS response in <Emphasis Type="Italic">Escherichia coli</Emphasis> cells under osmotic stress and in the presence of N-methyl-N′-nitro-N-nitrosoguanidine

We investigated the dynamics of the SOS response induction and the frequency of reversions induced by the monofunctional alkylating compound N-methyl-N′-nitro-N-nitrosoguanidine in Escherichia coli cells exposed to osmotic stress for 1 h. During the stress treatment of the wild-type cultures adapted and not adapted to the alkylating agent, the maximum SOS response values and induced reversion frequencies were recorded twice. The SOS response values and induced reversion frequencies remained unchanged during the whole period after attaining the maximum values in adapted and nonadapted cells carrying a mutation in the excision repair gene. Presumably, the SOS mutagenesis mechanisms are turned on in the cells with an inactivated excision repair system earlier than in wild-type cells.     

Mutagenic DNA repair in Escherichia coli XIII Proofreading exonuclease of DNA polymerase III holoenzyme is not operational during UV mutagenesis

We have introduced a mutD5 mutation (which results in defective 3′–5′-exonuclease activity of the ϵ proofreading subunit of DNA polymerase III holoenzyme) into excision-defective Escherichia coli strains with varying SOS responses to UV light. MutD5 increased the spontaneous mutation frequency in all strains tested, including recA430, umuC122::Tn5, and umuC36 derivatives. It had no effect of UV mutability or immutability in any strain or on misincorporation revealed by delayed photoreversal in UV-irradiated umuC36, umuC122::Tn5, or recA430 bacteria. It is concluded that the ϵ proofreading subunit of DNA polymerase III holoenzyme is excluded, inhibited, or inoperative during misincorporation and mutagenesis after UV.     

Mutagen sensitivity of Drosophila melanogaster. I. Isolation and preliminary characterization of a methyl methanesulphonate-sensitive strain

Sensitivity to the monofunctional alkylating agent methyl methanesulfonate (MMS) has been tested as a selection technique to isolate mutant strains which can provide insights into the genetic control of DNA replication, DNA repair and recombination in the complex eucaryote, Drosophila melanogaster. The successful isolation of an X-linked MMS-sensitive strain, mut^s, has suggested that mutagen sensitivity is a feasible methodology for the selection of mutant strains of Drosophila which will be useful in the genetic and biochemical analysis of these cellular functions. Preliminary characterization of this mutant strain indicates that: (A) it is extremely sensitive to killing by MMS; (B) it is more mutable by MMS than the parent wildtype strain; and (C) it appears to possess mutator gene activity.     

Excision repair of uracil incorporated in DNA as a result of a defect in dUTPase.

Mutants of Escherichia coli that are severely defective in the enzyme dUTPase (dut) accumulate short (4 to 5 S) Okazaki fragments following brief pulses with [³H]thymidine. The transient appearance of DNA fragments in these mutants is plausibly explained by the misincorporation of uracil in DNA as a result of an increase in available dUTP, followed by its rapid excision and repair. The evidence in support of this interpretation is the following: (1) accumulation of short DNA fragments can be partially suppressed by a mutation in dCTP deaminase, presumably by decreasing the intracellular level of dUTP relative to dTTP; (2) accumulation of the short DNA fragments can be almost completely suppressed by a mutation in uracil N-glycosidase, probably by preventing the introduction of nicks at the sites of uracil incorporation; (3) introduction of DNA polymerase I or DNA ligase mutations into dUTPase-defective strains results in the persistence of the 4 to 5 S fragments and rapid cessation of DNA synthesis. Uracil N-glycosidase, DNA polymerase I and DNA ligase must therefore be involved in the excision repair of uracil-containing DNA.     

A general role of the DNA glycosylase Nth1 in the abasic sites cleavage step of base excision repair in Schizosaccharomyces pombe

One of the most frequent lesions formed in cellular DNA are abasic (apurinic/apyrimidinic, AP) sites that are both cytotoxic and mutagenic, and must be removed efficiently to maintain genetic stability. It is generally believed that the repair of AP sites is initiated by the AP endonucleases; however, an alternative pathway seems to prevail in Schizosaccharomyces pombe. A mutant lacking the DNA glycosylase/AP lyase Nth1 is very sensitive to the alkylating agent methyl methanesulfonate (MMS), suggesting a role for Nth1 in base excision repair (BER) of alkylation damage. Here, we have further evaluated the role of Nth1 and the second putative S.pombe AP endonuclease Apn2, in abasic site repair. The deletion of the apn2 open reading frame dramatically increased the sensitivity of the yeast cells to MMS, also demonstrating that the Apn2 has an important function in the BER pathway. The deletion of nth1 in the apn2 mutant strain partially relieves the MMS sensitivity of the apn2 single mutant, indicating that the Apn2 and Nth1 act in the same pathway for the repair of abasic sites. Analysis of the AP site cleavage in whole cell extracts of wild-type and mutant strains showed that the AP lyase activity of Nth1 represents the major AP site incision activity in vitro. Assays with DNA substrates containing base lesions removed by monofunctional DNA glycosylases Udg and MutY showed that Nth1 will also cleave the abasic sites formed by these enzymes and thus act downstream of these enzymes in the BER pathway. We suggest that the main function of Apn2 in BER is to remove the resulting 3′-blocking termini following AP lyase cleavage by Nth1.     

Purification and structure of 3-methyladenine-DNA glycosylase I of Escherichia coli

We constructed a recombinant plasmid carrying a gene that suppresses tag mutation. To overproduce its gene product, a 0.8-kilobase DNA fragment which carries the gene was placed under the control of the lac promoter in pUC8. 3-Methyladenine-DNA glycosylase activity in cells carrying such plasmids (pCY5) was 450-fold higher than that of wild type strain, on exposure to isopropyl-beta-D-thiogalactopyranoside. From an extract of such cells, the enzyme was purified to apparent physical homogeneity, and the amino acid composition and the amino-terminal amino acid sequence of the enzyme were determined. The data were in accord with nucleotide sequence of the gene, determined by the dideoxy method. It was deduced that 3-methyladenine-DNA glycosylase I comprises 187 amino acids and its molecular weight is 21,100, consistent with the value estimated from the sodium dodecyl sulfate-polyacrylamide gel electrophoresis of the purified protein. Only 3-methyladenine was excised from methylated DNA by the purified glycosylase. These results show that the tag is the structural gene for 3-methyladenine-DNA glycosylase I.