The base excision repair pathway is initiated by the action of a class of enzymes known as DNA glycosylases, which recognize and release the damaged base, and thus give specificity to the repair process. Mammalian cells carry two major DNA glycosylases for the repair of oxidized bases, oxoguanine DNA glycosylase (OGG1) and Endonuclease III homologue (NTH1). We found that OGG1 plays a crucial role in the repair of oxidized lesions in mitochondria and is probably the only DNA glycosylase for 8-oxoG removal in these organelles. We have also shown that NTH localizes to mitochondria, where it is involved in removing oxidized pyrimidines. One strong point of our studies is that we assay for DNA repair activities and measure the actual occurrence of the lesions in DNA using chromatographic techniques. We analyzed the levels of 8-oxoG and other oxidized bases in mouse liver DNA and found that the levels of the ring-opened oxidative lesion fapyguanine (FapyG) is higher than that of 8-oxoG. Using mouse models deficient for these glycosylases we find that 8-oxoG and Fapy-G accumulate in DNA from OGG1-/- mouse and that FapyG and fapyadenine (FapyA) accumulate in DNA from NTH1-/- mice. We also show that FapyG and FapyA are repaired by the same set of DNA glycosylases that remove 8-oxoG and thymine glycols from DNA, both in the nucleus and in mitochondria. These results indicate that the accumulation of these lesions may have important biological consequences, at least as relevant as those of 8-oxoG. Moreover, we established the mitochondrial localization of the newly identified DNA glycosylase NEIL1, which has higher specificity for the ring-opened substrates. In human cells two distinct OGG1 isoforms are expressed, alpha and beta. Beta-OGG1 localizes exclusively to mitochondria and was believed to provide the 8-oxoG glcycosylase activity. We purified recombinant b-OGG1 and found that the protein lacks glycosylase activity. Site-directed mutagenesis studies identified two aminoacids that are found in the b-isoform that render the a-isoform inactive. We also found that approximately 10% of a-OGG1 localizes to mitochondria and may account for the mitochondrial 8-oxoG glycosylase activity. Because of the high abundance of the b-OGG1 protein in human mitochondria we are now investigating whether it has any biological function. For this we are establishing cell lines with isoform-specific stable knockdown of b-OGG1, in order to identify possible biological endpoints altered in the absence of this protein. All BER enzymes are encoded in the nucleus and transported to mitochondria; however there is very limited information on the regulation of mitochondrial BER. We measured BER activities in mitochondria that lack mtDNA (rho-). Despite the absence of mtDNA, a complete set of BER enzymes was present in mitochondria, and most activities were only slightly decreased compared to wild type mitochondria. Interestingly, nuclear BER activities were also affected by the absence of mtDNA, suggesting an interesting cross-talk between BER in both compartments. Mitochondria are comprised of two membranes (outer and inner) enclosing an aqueous matrix compartment. We studied the spatial organization of BER in mitochondria and find that most BER activities are not freely soluble in the matrix, but rather associated with the membrane fraction. This association is likely electrostatic in nature, as it can be disrupted by high salt concentration. The existence of this higher order DNA repair complex has profound implications for mtDNA repair, as it suggests a mechanism in which the DNA flows through this stationary complex. In mammalian mitochondria the mtDNA is found in a large protein-DNA complex known as the nucleoid. One of the most abundant protein components of mammalian nucleoids is the transcription factor TFAM, which has been postulated to have a structural function in compacting the mtDNA in the nucleoid. Using recombinant human TFAM we are now investigating whether TFAM modulates mtDNA repair. We find that TFAM binds with higher affinity to DNA containing oxidized bases, and that when TFAM is bound the catalytic activity of BER enzymes is decreased, most likely because of poor accessibility to the damaged base. These results indicate that TFAM may function to modulate BER through post-translational modifications that change its DNA binding affinity. We are now investigating whether mammalian mitochondria have any of the other repair pathways that operate in the nucleus, such as mismatch repair (MMR). Our results show that human mitochondria can catalyze mismatch repair in vitro and contain a mismatch binding activity. Using affinity purification with a mismatch-containing DNA substrate, and mass spectrometry-peptide analyses we identified 3 proteins in the mismatch-bound complex, the transcription factor YB-1, the Citochrome oxidase-assembly factor LRP130 and an UV-resistance associated gene of unknown activity. We showed mitochondrial localization of YB-1 using both the endogenous as well as ectopic expressed protein. Interestingly, abrogation of YB1 levels by RNA interference significantly decreased mitochondrial-catalysed mismatch repair activity in an in vitro assay, indicating that this protein is involved in mitochondiral MMR. These observations, along with results from others clearly establish that mammalian mitochondria have a functional mismatch repair pathway. Another important set of proteins involved in mitochondrial DNA metabolism are the helicases SUV3 and PIF1. We have investigated the biochemical functions of SUV3, and it appears to interact with some mitochondrial and telomere proteins, making it possible that it functions both in telomeres and in mitochondria. This is under further investigation.