Accumulation of adenine DNA glycosylase-sensitive sites in human mitochondrial DNA.

The mitochondrial respiratory chain inevitably produces reactive oxygen species as byproducts of aerobic ATP synthesis. Mitochondrial DNA (mtDNA), which is located close to the respiratory chain, is reported to contain much more 8-oxoguanine (8-oxoG), an oxidatively modified guanine base, than nuclear DNA. Despite such a high amount of 8-oxoG in mtDNA (1-2 8-oxoG/10(4) G), mtDNA is barely cleaved by an 8-oxoG DNA glycosylase or MutM, which specifically excises 8-oxoG from a C:8-oxoG pair. We find here that about half of human mtDNA molecules are cleaved by another 8-oxoG-recognizing enzyme, an adenine DNA glycosylase or MutY, which excises adenine from an A:8-oxoG pair. The cleavage sites are mapped to adenines. The calculated number of MutY-sensitive sites in mtDNA is approximately 1.4/10(4) G. This value roughly corresponds with the electrochemically measured amount of 8-oxoG in mtDNA (2.2/10(4) G), raising the possibility that 8-oxoG mainly accumulates as an A:8-oxoG pair.

Mitochondria are power plants that are responsible for more than 80% of energy in eukariotic cells. Mitochondria have extranuclear genes, which is common for all vertebrates. One cell contains 10 3 to 10 4 molecules of mitochondrial DNA (mtDNA). mtDNA codes for 13 subunits of the mitochondrial respiratory chain, 22 tRNAs, and 2 rRNAs, all of which are essential for proper function of the respiratory chain. Maintaining the integrity of mtDNA is, therefore, essential for aerobic ATP synthesis (19).
More than 1% of the oxygen consumed by cells is converted to reactive oxygen species under physiological conditions (20). Mitochondrial respiration accounts for about 90% of cellular oxygen consumption, and thus the respiratory chain in mitochondria is principally responsible for the production of reactive oxygen species. Accordingly, mtDNA may suffer from more oxidative damage than nuclear DNA. In fact, there are a number of reports that mtDNA contains much more 8-oxoG than does nuclear DNA (see Ref. 21 for review). The amount of 8-oxoG in mtDNA is reported to further increase with age (2,22). Recently, the mitochondrial dysfunction caused by the oxidative damage of mtDNA has been implicated in aging, neurodegenerative diseases, and cardiovascular disorders.
Despite such a high amount of 8-oxoG in mtDNA, mtDNA is barely cleaved with MutM protein (23,24). This discrepancy between the insensitivity of mtDNA to MutM protein and the high amount of 8-oxoG in mtDNA has remained to be resolved. At present, little is known about actually pairing partners with 8-oxoG in mtDNA in vivo. We report here that human mtDNA is cleaved at A:T sites on the mtDNA sequence by another 8-oxoG-recognizing enzyme MutY protein, suggesting that 8-oxoG accumulates as an A:8-oxoG pair but not as a C:8-oxoG pair.

EXPERIMENTAL PROCEDURES
Materials-Restriction enzymes EcoRI, BamHI, and PvuII were purchased from Takara (Seta, Japan). Phosphocellulose P11 was from Whatman. DNAzol reagent was from Life Technologies, Inc. Mono S column (1 ml) was from Amersham Pharmacia Biotech. Other reagents were of analytical grade.
MutY was purified to homogeneity essentially according to the methods of Lu and co-workers (27). Briefly, the cells overproducing MutY (5 g) were suspended in 15 ml of Buffer T (50 mM Tris-HCl, pH 7.6, 0.1 mM EDTA, 0.5 mM dithiothreitol, and 0.1 mM phenylmethylsulfonyl fluoride) and sonicated. The cell debris was removed by centrifugation, and 16 ml of the supernatant was treated with 4 ml of 25% streptomycin sulfate. After stirring for 45 min on ice, the solution was centrifuged, and 18 ml of the supernatant was collected. Ammonium sulfate (3.6 g) was added to the supernatant, and the proteins were precipitated for 1 h on ice. After centrifugation, the protein pellets were resuspended in 3 ml of Buffer T and dialyzed against 500 ml of Buffer T at 4°C for 3 h. The dialyzed sample (5 ml) was diluted 4-fold with Buffer A (20 mM potassium phosphate, pH 7.4, 0.5 mM dithiothreitol, 0.1 mM EDTA, and 0.1 mM phenylmethylsulfonyl fluoride) containing 50 mM KCl and applied to a 2-ml phosphocellulose column equilibrated with Buffer A containing 50 mM KCl. After washing with 20 ml of equilibration buffer at 0.2 ml/min, proteins were eluted with a linear gradient of KCl (50 -500 mM) at 0.2 ml/min for 300 min. Fractions containing the A:G-specific nicking activity were pooled. The pool fractions (20 ml) were dialyzed two times against 1 liter of Buffer A containing 50 mM KCl and 10% glycerol at 4°C for 1 h. The dialyzed sample was applied to a 1-ml Mono S column equilibrated with Buffer S (50 mM potassium phosphate, pH 7.5, 0.5 mM dithiothreitol, 0.1 mM EDTA, 0.1 mM phenylmethylsulfonyl fluoride, and 10% glycerol) containing 100 mM NaCl. After washing with 10 ml of equilibration buffer at 1.0 ml/min, proteins were eluted with a linear gradient of NaCl (100 -500 mM) for 20 min in Buffer S. The fractions containing the activity were pooled and stored at Ϫ80°C. All chromatography was performed with an Amersham Pharmacia Biotech FPLC System. MutM was purified to homogeneity essentially the same as described above except for omitting streptomycin sulfate precipitation.
Cleavage Reaction by MutM or MutY-The cells in a half-confluent state were harvested, and the total DNA was extracted with DNAzol reagent according to the manufacturer's instructions. The extracted DNA was digested with RNase A and BamHI (or PvuII), phenol-extracted, ethanol-precipitated, and solubilized in 0.1ϫ TE (1 mM Tris-HCl, pH 7.5, and 0.1 mM EDTA). The concentration of DNA was determined by measuring absorbance at 260 nm.
The standard reaction was performed in 10 l of the mixture containing 25 mM HEPES-NaOH, pH 7.6, 0.5 mM EDTA, 10 mM NaCl, 0.5 mM dithiothreitol, 50 g/ml bovine serum albumin, 30 ng of MutM or 300 ng of MutY, and 0.5 g of the total DNA at 37°C for 3 h. In some cases, the reaction was performed in the presence of the 35-mer oligonucleotide substrates listed in Table I. For detection of the fluorescein isothiocyanate-labeled oligonucleotides, DNA was denatured in 80% formamide by boiling, resolved on a 15% polyacrylamide/6 M urea gel, and analyzed with a FluorImager 595 (Molecular Dynamics). The mtDNA was analyzed by Southern blotting as described below.
Southern Blotting and Ligation-mediated Polymerase Chain Reaction-Southern blotting of mtDNA and the nuclearly encoded 18 S rRNA gene was performed as described previously (28). Total DNA (0.5 g) was treated with the purified MutM or MutY protein. After glyoxaldimethylsulfate denaturation (29), the DNA was electrophoresed on a 0.6% agarose gel in 10 mM sodium phosphate, pH 7.0, and transferred onto a nylon membrane (Hybond-N ϩ , Amersham Pharmacia Biotech). mtDNA (16 kbp) was probed with a 1.9-kbp fragment of mtDNA (nt 2573-4431) for the BamHI-digested DNA. For the PvuII-digested DNA, mtDNA (ϳ16 kbp) and the 18 S rRNA gene (ϳ12 kbp) were probed with a 0.8-kbp XbaI fragment of mtDNA (nt 7441-8286) and a 1.5-kbp XbaI fragment of the 18 S rRNA gene (nt 435-1951), respectively. Each probe was labeled with [␣-32 P]dCTP using a Megaprime DNA labeling kit (Amersham Pharmacia Biotech) or alkaline phosphatase using an Alkphos Direct labeling kit (Amersham Pharmacia Biotech). In the former case, the signals were detected and quantified with a radio image scanner, STORM (Molecular Dynamics). In the latter case, the signals were visualized using CDP-Star chemiluminescent reagent (Amersham Pharmacia Biotect) and quantified with LAS1000 (Fuji Film, Tokyo, Japan). The MutM-or MutY-cleaved strands of mtDNA were amplified by ligation-mediated polymerase chain reaction (LMPCR) using the primer sets D6, D7, and D8 as described previously (30).
Quantification of 8-Oxo-2Ј-deoxyguanosine (8-oxo-dG)-Mitochondria were prepared by differential centrifugation from approximately 1 ϫ 10 9 cells of HeLa MRV11. mtDNA was purified as supercoiled DNA from the mitochondrial fraction by using a Qiagen plasmid mini kit (Hilden, Germany) according to the manufacturer's instructions. The amount of 8-oxo-dG in mtDNA was measured by an electrochemical method as described previously (25).  (Table I), respectively, in the presence of 0.5 g of total DNA (Fig. 1B). The amount of 8-oxoG in total DNA is reported to be, at most, 1/10 5 G or roughly 4 fmol of 8-oxoG in 0.5 g of DNA (21,31). Therefore, the MutM or MutY is sufficient for digesting the endogenous 8-oxoG sites in the following experiments.

Cleavage of Oligonucleotides
DNA Digestion by MutM and MutY-After treatment of total DNA with the purified MutM or MutY protein, the amount of uncleaved mtDNA was quantified by Southern blot analysis. As reported by other groups (23,24), mtDNA remained uncleaved by the MutM protein both in the nondenaturing and denaturing conditions ( Fig. 2A, lanes 2 and 5). In contrast, the amount of uncleaved mtDNA was decreased to approximately one-half by the MutY protein in the glyoxal denaturing conditions ( Fig. 2A, lane 4). The MutY-treated mtDNA was apparently uncleaved in the nondenaturing conditions ( Fig. 2A, lane  1), indicating that single-strand cleavage occurs by the MutY protein treatment, which is consistent with the nature of adenine DNA glycosylase. In Fig. 2B, changes in the amount of 16-kbp mtDNA on the denaturing gels are shown. The amount of full-length mtDNA of HeLa MRV11 cells was barely decreased by the MutM protein (91.9 Ϯ 26.9, n ϭ 12, mean Ϯ S.D.), whereas the amount was decreased to nearly one-half that of the control by the MutY protein (46.2 Ϯ 11.0, n ϭ 12). In the case of HeLa MR51 cells that over-produce human MTH1 approximately 100-fold (15), the amounts of full-length mtDNA after treatment of the MutM and MutY proteins were 99.6 Ϯ 26.2 and 59.1 Ϯ 17.3 that of the control (mean Ϯ S.D., n ϭ 6), respectively (Fig. 2B).
Identification of the Cleavage Sites by MutY-To determine the cleavage sites, we amplified the cleaved strands by LMPCR. Consistent with the results in the Southern blot experiments, there were few signals detected in the control and MutM protein-treated DNAs (Fig. 3, lanes 5 and 6), whereas signals were observed in the MutY protein-treated DNA (Fig. 3,  lane 7). The same pattern was also observed in other regions of mtDNA when using other primer sets, D6 and D7 (results not shown). Essentially all of the MutY cleavage sites were mapped to adenine sites without sequence specificity around the adenines. This cleavage pattern is completely compatible with an adenine DNA glycosylase activity of MutY protein. All adenine sites were not evenly cleaved by the MutY protein (Fig. 3, lane  7). The cleavage was preferentially observed at sites where an A:T pair runs consecutively.
Cleavage of mtDNA and Nuclear DNA-To examine whether the MutY-sensitive sites accumulate in mtDNA more than in nuclear DNA, we detected the cleavage of nuclear DNA and mtDNA simultaneously. The cleavage of a nuclear 18 S rRNA gene (about 12 kbp) was marginal compared with that of mtDNA (Fig. 4A), suggesting that mtDNA contains more MutY-sensitive sites than does the nuclear 18 S rRNA gene. It is known that damage-specific glycosylases can attack normal nucleotide pairs as a part of the damage-scanning process (32)  protein-treated sample is reduced to approximately 50% of that in the control (Fig. 4B). The 18 S rRNA gene was barely cleaved by either the MutM or MutY protein. Similar results were also obtained when using Jurkat cells (results not shown).

DISCUSSION
Preparation of mtDNA from purified mitochondria raises a possibility that intact mtDNA is preferentially collected from intact mitochondria. This is considered to be one of the reasons why mtDNA is resistant to MutM (21,33). Although we used total cellular DNA to avoid any possible bias resulting from the mtDNA preparation, we could hardly detect MutM-dependent cleavage of mtDNA by Southern blot analysis. The collective number of MutM-cleavable lesions such as C:8-oxoG, C:formamidepyrimidine (Fapy), and an abasic site is below 1 per 10 molecules of mtDNA as long as we observe the amount of 16-kbp mtDNA. It has been recently reported that the oxidized bases exist predominantly in fragmented mtDNA (33). We, however, did not observe such fragmented bands of mtDNA by Southern blot analysis ( Fig. 2A).
In contrast, approximately one-half of mtDNA molecules were sensitive to the MutY protein, i.e. there is one MutYsensitive site per two mtDNA molecules. The cleavage of mtDNA is not due to nuclease(s) remaining in the purified MutY sample for the following reasons: 1) the MutY protein was highly purified (Fig. 1A); 2) the cleavage was a singlestrand break in nature (Fig. 2); 3) essentially all of the cleavage sites were mapped to adenines (Fig. 3); 4) a nuclearly encoded 18 S rRNA gene was barely cleaved (Fig. 4); and 5) the oligonucleotide FY was specifically cleaved at an A:G site (Fig. 1B). Accordingly, we conclude that mtDNA is cleaved by an adenine DNA glycosylase activity of the MutY.
MutY removes adenine residues from A:G and A:8-oxoG pairs with similar efficiency and also removes the residue from an A:C pair, but with 20-fold lower efficiency (27). In general, MutY-sensitive sites are generated the following three ways: misincorporation of normal nucleotides (A:G and A:C pairs); misincorporation of dAMP opposite 8-oxo-dG in DNA (A:8-oxoG pair); and misincorporation of 8-oxo-dGMP opposite template adenine (A:8-oxoG pair). The MutY cleavage sites in mtDNA were adenines of A:T on the mtDNA sequence (Fig. 3), suggesting that dGMP, dCMP, or 8-oxo-GMP is misincorporated opposite template adenine. Chick embryo DNA polymerase ␥ exhibits a fidelity of less than one error for every 260,000 bases polymerized when normal dNTPs are used as substrates (34). It is therefore unlikely that human DNA polymerase ␥ misincorporates dGMP or dCMP opposite template adenine as frequently as observed here. On the other hand, an A:T to C:G transversion rate for the chick embryo DNA polymerase ␥ is increased more than 1000-fold for synthesis in the presence of 8-oxo-dGTP at a concentration equal to that of normal dGTP (35). We electrochemically measured 8-oxo-dG in mtDNA that is prepared from mitochondria by an alkaline SDS method for extraction of closed circular DNA. The amount of 8-oxo-dG was approximately 2.2/10 4 G. There are about 3,500 G per single strand of mtDNA. One MutY-sensitive site per two mtDNA molecules means that human mtDNA contains approximately one MutY-sensitive site per 7,000 G or 1.4 per 10 4 G, roughly corresponding to electrochemically measured 8-oxo-dG of 2.2/  (16 kbp) and an 18 S rRNA gene (12 kbp) were simultaneously detected and quantified by Southern blotting using alkaline phosphatase-labeled probes for mtDNA and the 18 S rRNA gene as described under "Experimental Procedures." B, the PvuII-digested total DNA of U937 cells was treated with MutM or MutY. mtDNA was first probed. After deprobing, the 18 S rRNA gene was reprobed. The detection and quantification were performed as in A. 10 4 G. Taken together, although the nature of the MutY-sensitive sites is not completely understood at present, it is plausible that a certain part of the MutY-sensitive sites are A:8-oxoG. The over-expression of human MTH1, however, did not much reduce the MutY-sensitive sites (Fig. 2B). K m of human MTH1 for 8-oxo-dGTP is 12.5 M (36), the value of which is level with intracellular concentration of normal dGTP (ϳ30 M) (37). The inability of human MTH1 in reducing the MutY-sensitive sites might rather suggest that additional factor(s) may be required to eliminate 8-oxo-dGTP.
8-Oxo-dGMP is incorporated opposite C and A with essentially the same efficiency (4). Given that a certain part of the MutY-sensitive sites are A:8-oxoG, the fewer MutM-sensitive sites than the MutY-sensitive sites suggest that C:8-oxoG is more efficiently repaired in mitochondria than is A:8-oxoG that is formed by misincorporation of 8-oxo-dGMP opposite adenine. C and A are incorporated opposite 8-oxoG with almost the same efficiency after removal of A from A:8-oxoG, whereas G is predominantly incorporated opposite C after removal of 8-oxoG from C:8-oxoG. Thus a MutY-like activity is intrinsically less efficient in eliminating A:8-oxoG than is a MutM-like activity in eliminating C:8-oxoG. In addition, a MutY-like activity itself might be insufficient for repair of A:8-oxoG in mitochondria. The activity of 8-oxoG DNA glycosylase is actually detected in human mitochondria (16). On the other hand, the activity of adenine DNA glycosylase has not been detected in human mitochondria to date, although recombinant human MYH is targeted into mitochondria (18). This might reflect the fact that the human MYH activity is low relative to the human OGG1 activity in mitochondria. Another possibility is that human MYH might be less effective in repair of A:8-oxoG formed by misincorporation of 8-oxo-dGMP than that formed by misincorporation of dAMP. Although hydrolysis of 8-oxo-dGTP is proposed for prevention of 8-oxo-dGMP misincorporation in the current MutM, Y, and T systems (6,7), the repair of A:8-oxoG once formed by 8-oxo-dGMP misincorporation is not well elucidated. In the case that 8-oxo-dGMP is misincorporated opposite adenine, a strong MutY-like activity would enhance A:T to C:G transversion (6,7). In this light, it is conceivable that human MYH is regulated in mitochondria to act mainly on A:8-oxoG formed by misincorporation of dAMP. If so, A:8-oxoG formed by misincorporation of 8-oxo-dGMP might be repaired at later stages by other unknown mechanism(s), e.g. in a transcription-coupled way. We might also need to consider mechanism(s) other than base excision repair, such as selective degradation of mtDNA containing 8-oxoG.
Despite presumed strong oxidative damage of mtDNA, A:T to C:G or C:G to A:T transversion does not frequently occur as is evident from reported mutations (38). It is yet unclear to what extent 8-oxoG contributes to oxidative damage-induced mutations in mitochondria. The mutual roles of MutT-, MutM-, and MutY-like activities are not completely understood in mitochondria, either. Examination of pairing partners with 8-oxoG in a steady state is important both for understanding of mu-tagenesis by 8-oxoG and for estimation of relative activities among the three repair enzymes in vivo. To this purpose, the combination of DNA cleavage by DNA glycosylases and LMPCR, as used here, appears to be useful.