Repair of mtDNA in Vertebrates

Abstract

The subject of this review is a process once thought by many to be nonexistent. Spurred by an early observation that pyrimidine dimers introduced into mtDNA were not repaired (Clayton et al. Clayton et al., 1974Clayton DA Doda JN Friedberg EC The absence of a pyrimidine dimer repair mechanism in mammalian mitochondria.Proc Natl Acad Sci USA. 1974; 71: 2777-2781Crossref PubMed Scopus (401) Google Scholar), many people stampeded to the conclusion that mammalian mitochondria were incapable of repairing their DNA. Proponents of this view held that damaged mtDNA molecules might be either targeted for destruction or diluted out in the cell's pool of thousands of undamaged molecules. Perhaps wastage of an occasional damaged 16-kb DNA molecule would be a small price to pay to avoid duplication of the DNA-repair tool chest inside mitochondria. This careless disregard of DNA maintenance would be expected to contribute to the incidence of human diseases involving mtDNA mutations that compromise mitochondrial function. Although these considerations may apply to some types of mtDNA damage, we now know that other types of damage occur too frequently to go unrepaired; these include spontaneous base loss, giving rise to abasic sites (also referred to as “AP sites” [for “apurinic/apyrimidinic sites”]), and certain key examples of endogenous damage to DNA bases. It is instructive to consider mtDNA damage and its potential repair in light of the repair pathways described for bacterial and eukaryotic nuclear DNA. In those genomes, most types of DNA damage are processed by one of three general mechanisms described in table 1. In each case, DNA repair consists of two coupled steps, involving (1) recognition of damage and (2) synthesis of a repair patch. Recognition and resection of a lesion from DNA are rate-limiting steps in repair. For the purposes of this review, it is appropriate to consider that the three repair processes highlighted in table 1 require distinct sets of proteins for lesion recognition—an oversimplification, since certain proteins may act in more than one repair pathway. This article will review the classes of DNA damage that are amenable to repair in mtDNA, the current understanding of the enzymatic machinery available for DNA repair in mitochondria, and the implications that human mtDNA-repair mechanisms have for aging and for mitochondrial disease. The emerging pattern is that mitochondria in higher organisms are reasonably well equipped to conduct base-excision repair (BER) but appear to be deficient in nucleotide-excision repair (NER) and mismatch repair (MMR).Table 1Simplified View of Nuclear DNA–Repair Pathways: Extent of Duplication in MitochondriaRepair ProcessLesions ProcessedProteins Acting in Nuclear DNA RepairGenetic CorrelationsBERSpontaneous AP sites; glycosylase products“Short patch” (AP endonuclease, DNA pol β, DNA ligase I/ III); “long patch” (AP endonuclease, DNA pol δ [or η], FEN1, proliferating-cell nuclear antigen, replication factor C, DNA ligase I/ III)Mouse knockouts of AP endonuclease, DNA pol β, and DNA ligase I embryonic lethalNERPyrimidine dimers; bulky chemical adducts16 proteins for excision (XPA, XPC [p125, p58], RPA [p70, p34, p11], TFIIH [XPB/ERCC3, XPD/ERCC2, p62, p44, cdk7/p41, cycH/p38, 34], XPF 9 [p112, p33], XPG); gap filling (DNA pol η, proliferating-cell nuclear antigen, replication factor C [6 subunits], DNA ligase I/III)Xeroderma pigmentosum (alleles XPA–XPG); Cockayne syndrome; trichothiodystrophyMMRBase mismatches; short insertions/deletionsMultiple mut-S and mut-H homologues; excision activities; proliferating-cell nuclear antigen; replication factor C; DNA pol δHereditary nonpolyposis colon cancer Open table in a new tab The classical experiment to document DNA repair is to treat cells with DNA-damaging agents and monitor the kinetics with which DNA is repaired. If the damage does not introduce DNA breaks, samples can be treated with a damage-specific nuclease to monitor the persistence of endonuclease-sensitive sites. In the early days of DNA-repair research, such experiments used alkaline elution and alkaline sucrose gradient sedimentation to monitor the size of DNA fragments after DNA damage. These techniques were not widely applied to the study of mtDNA repair, because of the small quantities of mtDNA in somatic cells. The application of Southern blot hybridization methods to the study of repair of specific DNA sequences (Bohr et al. Bohr et al., 1985Bohr VA Smith CA Okumoto DS Hanawalt PC DNA repair in an active gene: removal of pyrimidine dimers from the DHFR gene of CHO cells is much more efficient than in the genome overall.Cell. 1985; 40: 359-369Abstract Full Text PDF PubMed Scopus (975) Google Scholar) made it possible to study the repair of mtDNA in the presence of contaminating nuclear DNA. Work from several laboratories has shown that mtDNA damage caused by some oxidizing agents and alkylating agents can be repaired (Myers et al. Myers et al., 1988Myers KA Saffhill R O'Connor PJ Repair of alkylated purines in the hepatic DNA of mitochondria and nuclei in the rat.Carcinogenesis. 1988; 9: 285-292Crossref PubMed Scopus (86) Google Scholar; Satoh et al. Satoh et al., 1988Satoh MS Huh NH Rajewsky MF Kuroki T Enzymatic removal of O6-ethylguanine from mitochondrial DNA in rat tissues exposed to N-ethyl-N-nitrosourea in vivo.J Biol Chem. 1988; 263: 6854-6856Abstract Full Text PDF PubMed Google Scholar; Pettepher et al. Pettepher et al., 1991Pettepher CC LeDoux SP Bohr VA Wilson GL Repair of alkali-labile sites within the mitochondrial DNA of RINr 38 cells after exposure to the nitrosourea streptozotocin.J Biol Chem. 1991; 266: 3113-3117Abstract Full Text PDF PubMed Google Scholar; Driggers et al. Driggers et al., 1993Driggers WJ LeDoux SP Wilson GL Repair of oxidative damage within the mitochondrial DNA of RINr 38 cells.J Biol Chem. 1993; 268: 22042-22045Abstract Full Text PDF PubMed Google Scholar). In contrast, no repair has been reported for larger alkylating groups, such as N-7 butyl adducts or bulky adducts such as benz(a)pyrene diolepoxide, aflatoxin, or psoralen (Backer and Weinstein Backer and Weinstein, 1980Backer JM Weinstein IB Mitochondrial DNA is a major cellular target for dihydrodiol-epoxide derivative of benzo[a]pyrene.Science. 1980; 209: 297-299Crossref PubMed Scopus (214) Google Scholar; Niranjan et al. Niranjan et al., 1982Niranjan BG Bhat NK Avadhani NG Preferential attack of mitochondrial DNA by aflatoxin B1 during hepatocarcinogenesis.Science. 1982; 215: 73-75Crossref PubMed Scopus (168) Google Scholar; Cullinane and Bohr Cullinane and Bohr, 1998Cullinane C Bohr V DNA interstrand crosslinks induced by psoralen are not repaired in mammalian mitochondria.Cancer Res. 1998; 58: 1400-1404PubMed Google Scholar). The situation is more complex for some agents that cause chemically heterogeneous lesions. It has been reported that some damage introduced into mtDNA by bleomycin and 4-nitroquinoline oxide can be repaired, whereas some lesions may be refractory to repair (Snyderwine and Bohr Snyderwine and Bohr, 1992Snyderwine EG Bohr VA Gene- and strand-specific damage and repair in Chinese hamster ovary cells treated with 4-nitroquinoline 1-oxide.Cancer Res. 1992; 52: 4183-4189PubMed Google Scholar; Shen et al. Shen et al., 1995Shen CC Wertelecki W Driggers WJ LeDoux SP Wilson GL Repair of mitochondrial DNA damage induced by bleomycin in human cells.Mutat Res. 1995; 337: 12-23Crossref Scopus (52) Google Scholar). Similarly, only interstrand cross-links introduced by cis-platinum (cis-Pt) are repaired; but common intrastrand cross-links persist in mtDNA (Singh and Maniccia-Bozzo Singh and Maniccia-Bozzo, 1990Singh G Maniccia-Bozzo E Evidence for lack of mitochondrial DNA repair following cis-dichlorodiammineplatinum treatment.Cancer Chemother Pharmacol. 1990; 26: 97-100Crossref PubMed Scopus (34) Google Scholar; LeDoux et al. LeDoux et al., 1992LeDoux SP Wilson GL Beecham EJ Stevnsner T Wasserman K Bohr VA Repair of mitochondrial DNA after various types of DNA damage in Chinese hamster ovary cells.Carcinogenesis. 1992; 13: 1967-1973Crossref PubMed Scopus (218) Google Scholar). Some of the DNA-damaging agents used in these studies, such as bleomycin and cis-Pt, are administered to patients as cytotoxic chemotherapeutic agents, but the extent to which these treatments affect the structure of mtDNA in cancer survivors is largely unexplored. Virtually all cases cited above in which mtDNA repair has been documented involve oxidative damage or alkylation, classes of DNA damage that also occur by endogenous mechanisms (Lindahl Lindahl, 1993Lindahl T Instability and decay of the primary structure of DNA.Nature. 1993; 362: 709-715Crossref PubMed Scopus (4018) Google Scholar). Oxidative damage is of particular relevance to mtDNA, since a large fraction of reactive oxygen species (ROS) in cells is thought to originate by leakage of electrons from the electron-transport system. The proximity of mtDNA to the ROS generator suggests that there may be a high rate of endogenous oxidative damage to mtDNA (Shigenaga et al. Shigenaga et al., 1994Shigenaga MD Hagen TM Ames BN Oxidative damage and mitochondrial decay in aging.Proc Natl Acad Sci USA. 1994; 91: 10771-10778Crossref PubMed Scopus (1749) Google Scholar). Some of the immediate products of ROS damage would include 8-oxo-guanine and thymine glycol as examples of base damage, as well as malondialdehyde and other products of oxidation of the deoxyribose backbone. However, it is difficult to measure these lesions with confidence, given the small amount of mtDNA available in cells. The incidence of base damage has generally been measured by high-performance liquid-chromatography analysis of hydrolyzed DNA; DNA-backbone damage can be assessed by PCR. An early study reported a significantly higher incidence of 8-oxo-dG in mtDNA than in nuclear DNA (Richter et al. Richter et al., 1988Richter C Park JW Ames BN Normal oxidative damage to mitochondrial and nuclear DNA is extensive.Proc Natl Acad Sci USA. 1988; 85: 6465-6467Crossref PubMed Scopus (1432) Google Scholar). 8-oxo-dG is of particular interest, since it has been well documented to lead to G→T transversions in DNA (Grollman and Moriya Grollman and Moriya, 1993Grollman A Moriya M Mutagenesis by 8-oxoguanine: an enemy within.Trends Genet. 1993; 9: 246-249Abstract Full Text PDF PubMed Scopus (710) Google Scholar). An initial report of high steady-state levels of 8-oxo-dG in the absence of oxidative stress has not been confirmed by other studies. The authors of the original report now acknowledge that contaminants may have caused them to overestimate the prevalence of this adduct (Helbock et al. Helbock et al., 1998Helbock H Beckman K Shigenaga M Walters P Woodhall A Yeo H Ames B DNA oxidation matters: the HPLC-electrochemical detection assay of 8-oxo-deoxyguanosine and 8-oxo-guanine.Proc Natl Acad Sci USA. 1998; 95: 288-293Crossref PubMed Scopus (602) Google Scholar). If it is assumed that the lower steady-state incidence of these lesions is correct, then mitochondrial repair mechanisms would appear to be quite efficient at repairing the damage caused by ROS. As noted in table 1, the BER pathway processes abasic sites that have been generated either by spontaneous base loss or by enzymatic removal of bases by damage-specific glycosylases. The spectrum of base damage that can be repaired in mtDNA is sure to be limited by the collection of DNA glycosylases found in the organelle. Unfortunately, our present knowledge of the DNA glycosylase genes encoded in the human nuclear genome is still very incomplete, so it is uncertain how many genes encode either dedicated mitochondrial glycosylases or mitochondrially targeted isoforms of such enzymes. DNA glycosylases are relatively well understood in bacteria, where the recent sequencing of whole bacterial genomes is providing a wealth of information on these gene families. Table 2 provides a list of several well-characterized DNA glycosylases in Escherichia coli and the current understanding of their human homologues.Table 2mtDNA Glycosylases: Relation to Nuclear Homologues of Bacterial EnzymesE. coli EnzymeaSeveral glycosylases have been shown to act on more than one type of modified base.Human HomologuePotential MLS?b“(t)” denotes that the MLS has been tested for its ability to direct a reporter protein to mitochondria; “(c)” denotes that the signal has been identified only by computerized sequence analysis.Protein Documented in Mitochondria?UDG, ungUDGYes (t)Yes, by Slupphaug et al. (Slupphaug et al., 1993Slupphaug G Markussen FH Olsen LC Aasland R Aarsaether N Bakke O Krokan HE et al.Nuclear and mitochondrial forms of human uracil-DNA glycosylase are encoded by the same gene.Nucleic Acids Res. 1993; 21: 2579-2584Crossref PubMed Scopus (117) Google Scholar)Thymine glycol glycosylase, endonuclease III, nthhNTHYes (t)Not determinedFormamidopyrimidine glycosylase, FAPY glycosylase, mut-mOGGYes (t)Yes, by Nishioka et al. (Nishioka et al., in pressNishioka K, Ohtsubo T, Oda H, Fujiwara T, Kang D, Sugimachi K, Nakkabeppu Y. Expression and differential localization of two major forms of human 8-oxoguanine glycosylase encoded by alternatively spliced OGG1 mRNAs. Mol Biol Cell (in press)Google Scholar)AG glycosylase, mut-YhMYHYes (t)Not determinedAlkyl adenine glycosylase, aagMAG, AAGYes (c)Not determineda Several glycosylases have been shown to act on more than one type of modified base.b “(t)” denotes that the MLS has been tested for its ability to direct a reporter protein to mitochondria; “(c)” denotes that the signal has been identified only by computerized sequence analysis. Open table in a new tab The prototypical DNA glycosylase, uracil DNA glycosylase (UDG), is directed to either nuclei or mitochondria by differential splicing (Slupphaug et al. Slupphaug et al., 1993Slupphaug G Markussen FH Olsen LC Aasland R Aarsaether N Bakke O Krokan HE et al.Nuclear and mitochondrial forms of human uracil-DNA glycosylase are encoded by the same gene.Nucleic Acids Res. 1993; 21: 2579-2584Crossref PubMed Scopus (117) Google Scholar). Only during the past few years have other glycosylases have come under study in higher eukaryotes. Some progress has been made in following the UDG model by analysis of candidate genes that appear to encode DNA glycosylases with potential N-terminal mitochondrial localization signals (MLSs). This is done first by computer sequence analysis and subsequently by testing the ability of these N-terminal sequences to direct a reporter gene (such as green fluorescent protein [GFP]) to mitochondria. As shown in table 2, there is strong evidence to suggest that mitochondria contain 8-oxo-guanine glycosylase (OGG) (Nishioka et al., Nishioka et al., in pressNishioka K, Ohtsubo T, Oda H, Fujiwara T, Kang D, Sugimachi K, Nakkabeppu Y. Expression and differential localization of two major forms of human 8-oxoguanine glycosylase encoded by alternatively spliced OGG1 mRNAs. Mol Biol Cell (in press)Google Scholar). The N-termini of other glycosylase genes—including hNTH1, hMYH, and AAG—appear to encode potential MLSs, and the first two of these have been shown to function to direct GFP to mitochondria (Takao et al. Takao et al., 1998Takao M Aburatani H Kobayashi K Yasui A Mitochondrial targeting of human DNA glycosylases for repair of oxidative DNA damage.Nucleic Acids Res. 1998; 26: 2917-2922Crossref PubMed Scopus (228) Google Scholar). It is interesting to note that the human homologue of another enzyme (albeit not a glycosylase) involved in the response to oxidative damage—that is, human 8-oxo-dGTPase (a homologue of the bacterial mutT)—has also been identified in mitochondria (Kang et al. Kang et al., 1995Kang D Nishida J Iyama A Nakabeppu Y Furuichi M Fujiwara T Sekiguchi M et al.Intracellular localization of 8-oxo-dGTPase in human cells, with special reference to the role of the enzyme in mitochondria.J Biol Chem. 1995; 270: 14659-14665Crossref PubMed Scopus (157) Google Scholar). Thus, it seems likely that mitochondria contain all of the components of the so-called GO system of E. coli, which responds to oxidative damage to guanosine either as a free nucleotide or within DNA (Grollman and Moriya Grollman and Moriya, 1993Grollman A Moriya M Mutagenesis by 8-oxoguanine: an enemy within.Trends Genet. 1993; 9: 246-249Abstract Full Text PDF PubMed Scopus (710) Google Scholar). The data in table 2 are almost certainly incomplete; other enzyme activities are only now being investigated (Croteau et al. Croteau et al., 1997Croteau DL ap Rhys CM Hudson EK Dianov GL Hansford RG Bohr VA An oxidative damage-specific endonuclease from rat liver mitochondria.J Biol Chem. 1997; 272: 27338-27344Crossref PubMed Scopus (132) Google Scholar), and many others remain to be explored. The existence of DNA glycosylases and the unavoidable spontaneous formation of abasic sites in mtDNA suggests that the mitochondria should maintain a complete system for repair of abasic sites. Since my laboratory has been characterizing the mitochondrial replicative DNA pol γ, my colleagues and I undertook to purify other enzymes that might cooperate with this enzyme in the repair of abasic sites in mtDNA. By analogy to BER in bacterial and nuclear systems, this repair was expected to require at least three other enzymatic activities—those of an AP endonuclease, an AP lyase, and a DNA ligase (fig. 1). Although we succeeded in reconstituting BER with mitochondrial enzymes (Pinz and Bogenhagen Pinz and Bogenhagen, 1998Pinz KG Bogenhagen DF Efficient repair of abasic sites in DNA by mitochondrial enzymes.Mol Cell Biol. 1998; 18: 1257-1265Crossref PubMed Scopus (152) Google Scholar), a great deal remains to be learned concerning the enzymes that can act in this pathway, as discussed below. Despite the frequency and importance of abasic sites in mtDNA, we are still largely ignorant of AP endonucleases in mitochondria. From our work with Xenopus enzymes and from a previous study using mouse cells (Tomkinson et al. Tomkinson et al., 1988Tomkinson AE Bonk RT Linn S Mitochondrial endonuclease activities specific for apurinic/apyrimidinic sites in DNA from mouse cells.J Biol Chem. 1988; 263: 12532-12537Abstract Full Text PDF PubMed Google Scholar), it appears that the mitochondrial AP endonuclease cleaves on the 5′ side of the AP site and behaves as a single chromatographic species. The mitochondrial enzymes have not yet been cloned, but Tomkinson et al. (Tomkinson et al., 1988Tomkinson AE Bonk RT Linn S Mitochondrial endonuclease activities specific for apurinic/apyrimidinic sites in DNA from mouse cells.J Biol Chem. 1988; 263: 12532-12537Abstract Full Text PDF PubMed Google Scholar) and Pinz and Bogenhagen (Pinz and Bogenhagen, 1998Pinz KG Bogenhagen DF Efficient repair of abasic sites in DNA by mitochondrial enzymes.Mol Cell Biol. 1998; 18: 1257-1265Crossref PubMed Scopus (152) Google Scholar, and unpublished data) have found that antisera raised against the nuclear enzyme cross-react with proteins in preparations of mouse and Xenopus mitochondrial AP endonuclease. DNA pol γ is capable of binding the 3′-OH group produced by AP endonuclease to fill a one-nucleotide gap. The Xenopus enzyme was found to contain an additional activity as an AP lyase (Pinz and Bogenhagen Pinz and Bogenhagen, 1998Pinz KG Bogenhagen DF Efficient repair of abasic sites in DNA by mitochondrial enzymes.Mol Cell Biol. 1998; 18: 1257-1265Crossref PubMed Scopus (152) Google Scholar), which was subsequently confirmed for the recombinant human DNA pol γ (Longley et al. Longley et al., 1998Longley M Prasad R Srivastava D Wilson S Copeland WC Identification of 5′ deoxyribose phosphate lyase activity in human DNA polymerase γ and its role in mitochondrial base excision repair in vitro.Proc Natl Acad Sci USA. 1998; 95: 12244-12248Crossref PubMed Scopus (163) Google Scholar). Although this represents the second example of a DNA polymerase that can act as an AP lyase activity, the corresponding activity previously identified in DNA pol β has a much higher catalytic activity (Matsumoto and Kim Matsumoto and Kim, 1995Matsumoto Y Kim K Excision of deoxyribose phosphate residues by DNA polymerase β during DNA repair.Science. 1995; 269: 699-702Crossref PubMed Scopus (625) Google Scholar). DNA polymerases related to E. coli DNA polymerase I and T7 DNA polymerase also contain an AP lyase activity with a low turnover capacity (K. G. Pinz and D. F. Bogenhagen, unpublished data). DNA pol γ is a heterodimeric protein whose larger subunit contains the catalytic activities for the polymerase, 3′–5′ proofreading exonuclease, as well as for the newly discovered AP lyase activity. The smaller subunit acts as a processivity factor (Carrodeguas et al. Carrodeguas et al., in pressCarrodeguas JA, Kobayshi R, Lim SE, Copeland WC, Bogenhagen DF (1999) The accessory subunit of X. laevis mitochondrial DNA polymerase γ increases processivity of the catalytic subunit of human DNA polymerase γ and is related to class II amino acyl tRNA synthetases. Mol Cell Biol (in press)Google Scholar) but is not required for gap-filling synthesis in vitro (Longley et al. Longley et al., 1998Longley M Prasad R Srivastava D Wilson S Copeland WC Identification of 5′ deoxyribose phosphate lyase activity in human DNA polymerase γ and its role in mitochondrial base excision repair in vitro.Proc Natl Acad Sci USA. 1998; 95: 12244-12248Crossref PubMed Scopus (163) Google Scholar; K. G. Pinz and D. F. Bogenhagen, unpublished data). The final step in the BER pathway is the resealing of the DNA by DNA ligase. To date, all DNA ligases characterized in eukaryotes are ATP-dependent enzymes that act through an enzyme-adenylate intermediate to reseal single-strand breaks. mtDNA ligase has been purified from Xenopus oocyte mitochondria as an ∼100-kD protein active in an adenylation assay (Pinz and Bogenhagen Pinz and Bogenhagen, 1998Pinz KG Bogenhagen DF Efficient repair of abasic sites in DNA by mitochondrial enzymes.Mol Cell Biol. 1998; 18: 1257-1265Crossref PubMed Scopus (152) Google Scholar). The enzyme has been shown to be clearly distinct from DNA ligase I, which is the predominant source of ligase activity in Xenopus oocytes. My colleagues and I have noted that, in published cDNA sequences of human DNA ligase III and IV (Wei et al. Wei et al., 1995Wei YF Robins P Carter K Caldecott K Pappin DJC Yu GL Wang R et al.Molecular cloning and expression of human cDNAs encoding a novel DNA ligase IV and DNA ligase III, an enzyme active in DNA repair and recombination.Mol Cell Biol. 1995; 15: 3206-3216Crossref PubMed Scopus (223) Google Scholar), each of these gene products could be translated from in-frame initiation codons that were upstream of those indicated in the initial reports. Remarkably, in each case, the upstream sequences revealed a potential N-terminal MLS, suggesting that alternative translational initiation could generate both a mitochondrial and a nuclear isoform of each of these enzymes. Indeed, when we cloned cDNAs encoding Xenopus DNA ligase III and IV, we found that the putative MLS regions were conserved in these genes and that the putative MLS regions of DNA ligase III can direct GFP to mitochondria (R. M. Perez-Jannotti, S. M. Klein, and D. F. Bogenhagen, unpublished data). Moreover, an antiserum raised against a DNA ligase III–specific epitope was found to cross-react with the mtDNA ligase. Although protein sequence data or other definitive proof that mtDNA ligase is a product of the DNA ligase III gene is lacking, it is likely that the DNA ligase III gene will join the collection of DNA glycosylase genes described above, as a dual-function gene capable of providing enzyme to both the mitochondrion and the nucleus. When DNA ligase III resides in the nucleus, it acts in association with the XRCC1 gene product that has not yet been shown to enter mitochondria. It is unknown whether the mitochondrial isoform acts on its own or has a molecular partner in place of XRCC1. It is also uncertain whether DNA ligase IV may enter mitochondria in some cells. Since mtDNA ligase was purified by use of an adenylation assay, and since DNA ligase IV has been reported to purify in an adenylated form that is not efficiently labeled in vitro, it is possible that a mitochondrial form of DNA ligase IV may have been overlooked because of the use of the adenylation assay. In the mouse, DNA ligase IV is an essential gene (Frank et al. Frank et al., 1998Frank K Sekiguchi J Seidl K Swat W Rathbun G Cheng HL Davidson L et al.Late embryonic lethality and impaired V(DJ) recombination in mice lacking DNA ligase IV.Nature. 1998; 396: 173-177Crossref PubMed Scopus (444) Google Scholar); whether mitochondrial abnormalities contribute to the demise of DNA ligase IV knockout-mouse embryos is unknown. Although many proteins involved in repair and maintenance of mtDNA remain to be discovered, several examples of repair proteins that are produced from dual-function genes that contribute products to both the nucleus and mitochondria have already been identified. This type of organization is not unprecedented, since a number of proteins involved in tRNA processing have been shown to be products of dual-function genes (Martin and Hopper Martin and Hopper, 1994Martin NC Hopper AK How single genes provide tRNA processing enzymes to mitochondria, nuclei and the cytosol.Biochimie. 1994; 76: 1161-1167Crossref PubMed Scopus (63) Google Scholar). It is appropriate to ask both how the expression of proteins such as UDG, OGG1, and DNA ligase III is controlled to suit the needs of both organelles and how this regulation varies in different cell types. One can envision circumstances in which the supply of repair proteins to mitochondria may be either excessive or insufficient. In the case of UDG, ∼25% of the total activity in a cell is directed to mitochondria, which contain only 0.1% of the cell's DNA. Thus, relative to the nucleus, mitochondria contain a large excess of this enzyme scanning DNA for misincorporated U residues. Because some glycosylases have a propensity to remove normal bases at a measurable efficiency, and because the overexpression of at least one member of these enzymes is mutagenic (Berdal et al. Berdal et al., 1998Berdal K Johansen R Seeberg E Release of normal bases from intact DNA by a native DNA repair enzyme.EMBO J. 1998; 17: 363-367Crossref PubMed Scopus (143) Google Scholar), it is possible that delivering too much of a repair glycosylase to mitochondria may be detrimental. Conversely, it is possible that, in postreplicative cells, which may be expected to reduce their expression of repair and replication factors, mitochondria could become depleted of these factors. This could contribute to problems in mtDNA maintenance during aging, since muscle- and nerve-cell mitochondria continue to require nuclear-gene products to repair and replace their DNA. This review has concentrated on BER of mtDNA as the one example of a repair pathway that does seem to function in mitochondria. As summarized above, there is little evidence that NER functions in mitochondria to remove bulky DNA adducts. NER of nuclear DNA requires a large collection of proteins (including the alphabet soup of XP and ERCC genes; see table 1), most of which probably have no access to mtDNA. This deficiency in NER is striking, in the face of extensive evidence that many chemical carcinogens form adducts with mtDNA, at frequencies often 10–50-fold higher than they do with nuclear DNA. In the case of chemical damage to mtDNA, the rate of accumulation of damaged bases and the fate of damaged DNA molecules have not been studied in detail. If these lesions are not removed, they will be encountered by DNA pol γ and mtRNA polymerase in the course of replication and transcription and will likely block both processes. Deletions or rearrangements may result when a DNA pol γ molecule is stalled at a bulky lesion. This model represents a plausible, if speculative, mechanism whereby unrepaired bulky damage to mtDNA may contribute to mitochondrial mutagenesis. However, damage to mtDNA by exogenous chemicals does not appear to cause a high rate of fixation of mutations (Mita et al. Mita et al., 1988Mita S Monnat RJ Loeb L Resistance of HeLa cell mitochondrial DNA to mutagenesis by chemical carcinogens.Cancer Res. 1988; 48: 4578-4583PubMed Google Scholar). It may be that most mtDNA molecules bearing bulky damage are indeed targeted for destruction, perhaps by a turnover mechanism that uses an arrested polymerase as a signal to promote degradation. More research on the fate of damaged mtDNA molecules is required for resolution of these issues. The suggestion in this review—that is, that mitochondria in higher organisms are deficient in mismatch repair—is also largely speculative. Saccharomyces cerevisiae maintains a mut-S homologue, msh-1, and mutations in msh-1 induce a high rate of mtDNA mutations (Reenan and Kolodner Reenan and Kolodner, 1992Reenan RA Kolodner RD Characterization of insertion mutations in the Saccharomyces cerevisiae MSH1 and MSH2 genes: evidence for separate mitochondrial and nuclear functions.Genetics. 1992; 132: 975-985Crossref PubMed Google Scholar; Vanderstraeten et al. Vanderstraeten et al., 1998Vanderstraeten S VandenBrule S Hu J Foury F The role of 3′-5′ exonucleolytic proofreading and mismatch repair in yeast mitochondrial DNA error avoidance.J Biol Chem. 1998; 273: 23690-23697Crossref PubMed Scopus (74) Google Scholar). However, a complete MMR pathway in yeast mitochondria has not yet been described. The presence of an active MMR system in yeast may help explain the lower rate of evolutionary change in yeast mtDNA genomes as compared with those in higher eukaryotes. A mitochondrial mut-S gene is retained in a coral mtDNA genome (Pont-Kingdon et al. Pont-Kingdon et al., 1995Pont-Kingdon GA Okada NA Macfarlane JL Beagley CT Wolstenholme DR Cavalier-Smith T Clark-Walker GD A coral mitochondrial mutS gene.Nature. 1995; 375: 109-111Crossref PubMed Scopus (92) Google Scholar) but not in other metazoans. No available evidence indicates that the mut-S or mut-L homologues of higher eukaryotes have access to mitochondria, but this lack of data may be seen as a challenge—to nucleic-acid biochemists, cell biologists, and geneticists—to show that MMR does indeed occur in mitochondria in higher organisms. The author's research is supported by National Institutes of Health grants GM29681 and ES04068.