In contrast to proteins recognizing small-molecule ligands, DNA-dependent enzymes cannot rely solely on interactions in the substrate-binding centre to achieve their exquisite specificity. It is widely believed that substrate recognition by such enzymes involves a series of conformational changes in the enzyme-DNA complex with sequential gates favoring cognate DNA and rejecting nonsubstrates. However, direct evidence for such mechanism is limited to a few systems. We report that discrimination between the oxidative DNA lesion, 8-oxoguanine (oxoG) and its normal counterpart, guanine, by the repair enzyme, formamidopyrimidine-DNA glycosylase (Fpg), likely involves multiple gates. Fpg uses an aromatic wedge to open the Watson-Crick base pair and everts the lesion into its active site. We used molecular dynamics simulations to explore the eversion free energy landscapes of oxoG and G by Fpg, focusing on structural and energetic details of oxoG recognition. The resulting energy profiles, supported by biochemical analysis of site-directed mutants disturbing the interactions along the proposed path, show that Fpg selectively facilitates eversion of oxoG by stabilizing several intermediate states, helping the rapidly sliding enzyme avoid full extrusion of every encountered base for interrogation. Lesion recognition through multiple gating intermediates may be a common theme in DNA repair enzymes.

DNA repair is essential to counteract damage to DNA induced by endo- and exogenous factors, to maintain genome stability. However, challenges to the faithful discrimination between damaged and non-damaged DNA strands do exist, such as mismatched pairs between two regular bases resulting from spontaneous deamination of 5-methylcytosine or DNA polymerase errors during replication. To counteract these mutagenic threats to genome stability, cells evolved the mismatch-specific DNA glycosylases that can recognize and remove regular DNA bases in the mismatched DNA duplexes. The Escherichia coli adenine-DNA glycosylase (MutY/MicA) protects cells against oxidative stress-induced mutagenesis by removing adenine which is mispaired with 7,8-dihydro-8-oxoguanine (8oxoG) in the base excision repair pathway. However, MutY does not discriminate between template and newly synthesized DNA strands. Therefore the ability to remove A from 8oxoG•A mispair, which is generated via misincorporation of an 8-oxo-2'-deoxyguanosine-5'-triphosphate precursor during DNA replication and in which A is the template base, can induce A•T→C•G transversions. Furthermore, it has been demonstrated that human MUTYH, homologous to the bacterial MutY, might be involved in the aberrant processing of ultraviolet (UV) induced DNA damage.

In protein evolution, functionally important intramolecular interactions, such as polar bridges or hydrophobic interfaces, tend to be conserved. We have analyzed coevolution of physicochemical properties in pairs of amino acid residues in the formamidopyrimidine-DNA glycosylase (Fpg) protein family, identified three conserved polar bridges (Arg54-Glu131, Gln234-Arg244, and Tyr170-Ser208 in the E. coli protein) located in known functional regions of the protein, and analyzed their roles by site-directed mutagenesis. The structure and molecular dynamic modeling showed that the coevolving pairs do not form isolated bridges but rather participate in tight local clusters of hydrogen bonds. The Arg54-Glu131 bridge, connecting the N- and C-terminal domains, was important for DNA binding, as its abolishment or even ion pair reversal inactivated Fpg and greatly decreased the enzyme's affinity for DNA. Mutations of the Gln234-Arg244 bridge, located at the base of the single Fpg β-hairpin zinc finger, did not affect the activity but sharply decreased the melting temperature of the protein, with the bridge reversal partially restoring the thermal stability. Finally, Tyr170 mutation to Phe decreased Fpg binding but did not fully inactivate the protein, whereas Ser208 replacement with Ala had no effect; molecular dynamics showed that in both wild-type and S208 A Fpg, Tyr170 quickly re-orients to form an alternative set of hydrogen bonds. Thus, the coevolution analysis approach, combined with biochemical and computational studies, provides a powerful tool for understanding intramolecular interactions important for the function of DNA repair enzymes.

Endonuclease VIII (Nei) from Escherichia coli is a DNA repair enzyme that removes a wide range of oxidized pyrimidine bases from DNA. As inferred from the crystal structures and biochemical studies, recognition of DNA lesions by Nei involves several conformational changes in both protein and DNA, such as DNA kinking, damaged base eversion into the enzyme's active site, and insertion of a loop of the enzyme into the void formed by the eversion. Excision of the damaged base by Nei also proceeds through several chemical steps: N-glycosidic bond breakage, β-elimination and δ-elimination of the phosphates flanking the lesion. We have used stopped-flow kinetics with fluorescence detection to follow conformational changes in the Nei molecule when the enzyme binds normal DNA, damaged but uncleavable DNA, or several cleavable damaged DNA substrates. Binding normal or damaged uncleavable DNA proceeded in two fluorescently discernible reversible stages, while processing of cleavable substrates involved three reversible stages followed by and irreversible stage and equilibrium with the reaction product. Individual rate constants were calculated for each reaction step. Based on the stopped-flow data, crystal structure, and a comparison with the stopped-flow kinetics of E. coli formamidopyrimidine-DNA glycosylase, a homolog of Nei, we propose the nature of some of the steps that may be involved into the recognition and excision of damaged bases by Nei.

Isolated human ribosomal protein uS3 has extra-ribosomal functions including those related to base excision DNA repair, e.g. AP lyase activity that nicks double-stranded (ds) DNA 3΄ to the abasic (AP) site. However, the ability of uS3 residing within ribosome to recognize and cleave damaged DNA has never been addressed. Here, we compare interactions of single-stranded (ss) DNA and dsDNA bearing AP site with human ribosome-bound uS3 and with the isolated protein, whose interactions with ssDNA were not yet studied. The AP lyase activity of free uS3 was much higher with ssDNA than with dsDNA, whereas ribosome-bound uS3 was completely deprived of this activity. Nevertheless, an exposed peptide of ribosome-bound uS3 located far away from the putative catalytic center previously suggested for isolated uS3 cross-linked to full-length uncleaved ssDNA, but not to dsDNA. In contrast, free uS3 cross-linked mainly to the 5΄-part of the damaged DNA strand after its cleavage at the AP site. ChIP-seq analysis showed preferential uS3 binding to nucleolus-associated chromatin domains. We conclude that free and ribosome-bound uS3 proteins interact with AP sites differently, exhibiting their non-translational functions in DNA repair in and around the nucleolus and in regulation of DNA damage response in looped DNA structures, respectively.

Uracil-DNA glycosylases are enzymes that excise uracil bases appearing in DNA as a result of cytosine deamination or accidental dUMP incorporation from the dUTP pool. The activity of Family 1 uracil-DNA glycosylase (UNG) activity limits the efficiency of antimetabolite drugs and is essential for virulence in some bacterial and viral infections. Thus, UNG is regarded as a promising target for antitumor, antiviral, antibacterial, and antiprotozoal drugs. Most UNG inhibitors presently developed are based on the uracil base linked to various substituents, yet new pharmacophores are wanted to target a wide range of UNGs. We have conducted virtual screening of a 1,027,767-ligand library and biochemically screened the best hits for the inhibitory activity against human and vaccinia virus UNG enzymes. Although even the best inhibitors had IC50 ≥ 100 μM, they were highly enriched in a common fragment, tetrahydro-2,4,6-trioxopyrimidinylidene (PyO3). In silico, PyO3 preferably docked into the enzyme's active site, and in kinetic experiments, the inhibition was better consistent with the competitive mechanism. The toxicity of two best inhibitors for human cells was independent of the presence of methotrexate, which is consistent with the hypothesis that dUMP in genomic DNA is less toxic for the cell than strand breaks arising from the massive removal of uracil. We conclude that PyO3 may be a novel pharmacophore with the potential for development into UNG-targeting agents.

GO system is part of base excision DNA repair and is required for the correct repair of 8-oxoguanine (8-oxoG), one of the most abundant oxidative lesions. Due to the ability of 8-oxoG to mispair with A, this base is highly mutagenic, and its repair requires two enzymes: Fpg that removes 8-oxoG from 8-oxoG:C pairs, and MutY that excises the normal A from 8-oxoG:A mispairs. Here we characterize the properties of putative GO system DNA glycosylases from Staphylococcus aureus, an important human opportunistic pathogen that causes hospital infections and presents a serious health concern due to quick spread of antibiotic-resistant strains. In addition to Fpg and MutY from the reference NCTC 8325 strain (SauFpg1 and SauMutY), we have also studied an Fpg homolog from a multidrug-resistant C0673 isolate (SauFpg2), which is different from SauFpg1 in its sequence. Both SauFpg enzymes showed the highest activity at pH 7.0-9.0 and NaCl concentrations 25-75 mM (SauFpg1) or 50-100 mM (SauFpg2), whereas SauMutY was active at a broad pH range and had a salt optimum at ∼75 mM NaCl. Both SauFpg1 and SauFpg2 bound and cleaved duplexes containing 8-oxoG, 5-hydroxyuracil, 5,6-dihydrouracil or apurinic/apyrimidinic site paired with C, T, or G, but not with A. For SauFpg1 and SauFpg2, 8-oxoG was the best substrate tested, and 5,6-dihydrouracil was the worst one. SauMutY efficiently excised adenine from duplex substrates containing A:8-oxoG or A:G pairs. SauFpg enzymes were readily trapped on DNA by NaBH4 treatment, indicating formation of a Schiff base reaction intermediate. Surprisingly, SauMutY was also trapped significantly better than its E. coli homolog. All three S. aureus GO glycosylases drastically reduced spontaneous mutagenesis when expressed in an fpg mutY E. coli double mutant. Overall, we conclude that S. aureus possesses an active GO system, which could possibly be targeted for sensitization of this pathogen to oxidative stress.

Uracil-DNA glycosylase (Ung) is a DNA repair enzyme that excises uracil bases from DNA, where they appear through deamination of cytosine or incorporation from a cellular dUTP pool. DNA repair enzymes often use one-dimensional diffusion along DNA to accelerate target search; however, this mechanism remains poorly investigated mechanistically. We used oligonucleotide substrates containing two uracil residues in defined positions to characterize one-dimensional search of DNA by Escherichia coli Ung. Mg(2+) ions suppressed the search in double-stranded DNA to a higher extent than K(+) likely due to tight binding of Mg(2+) to DNA phosphates. Ung was able to efficiently overcome short single-stranded gaps within double-stranded DNA. Varying the distance between the lesions and fitting the data to a theoretical model of DNA random walk, we estimated the characteristic one-dimensional search distance of ~100 nucleotides and translocation rate constant of ~2×10(6) s(-1).

Apurinic/apyrimidinic (AP) endonucleases are important DNA repair enzymes involved in two overlapping pathways: DNA glycosylase-initiated base excision (BER) and AP endonuclease-initiated nucleotide incision repair (NIR). In the BER pathway, AP endonucleases cleave DNA at AP sites and 3'-blocking moieties generated by DNA glycosylases, whereas in NIR, the same AP endonucleases incise DNA 5' to a wide variety of oxidized bases. The flowering plant Arabidopsis thaliana contains three genes encoding homologues of major human AP endonuclease 1 (APE1): Arp, Ape1L and Ape2. It has been shown that all three proteins contain AP site cleavage and 3'-repair phosphodiesterase activities; however, it was not known whether the plant AP endonucleases contain the NIR activity. Here, we report that ARP proteins from Arabidopsis and common wheat (Triticum aestivum) contain NIR and 3'→5' exonuclease activities in addition to their AP endonuclease and 3'-repair phosphodiesterase functions. The steady-state kinetic parameters of reactions indicate that Arabidopsis ARP cleaves oligonucleotide duplexes containing α-anomeric 2'-deoxyadenosine (αdA) and 5,6-dihydrouridine (DHU) with efficiencies (kcat/KM=134 and 7.3 μM-1·min-1, respectively) comparable to those of the human counterpart. However, the ARP-catalyzed 3'-repair phosphodiesterase and 3'→5' exonuclease activities (kcat/KM=314 and 34 μM-1·min-1, respectively) were about 10-fold less efficient as compared to those of APE1. Interestingly, homozygous A. thaliana arp-/- mutant exhibited high sensitivity to methyl methanesulfonate and tert-butyl hydroperoxide, but not to H2O2, suggesting that ARP is a major plant AP endonuclease that removes abasic sites and specific types of oxidative DNA base damage. Taken together, these data establish the presence of the NIR pathway in plants and suggest its possible role in the repair of DNA damage generated by oxidative stress.

Apurinic/apyrimidinic (AP) endonucleases are key enzymes involved in the repair of abasic sites and DNA strand breaks. Pathogenic bacteria Mycobacterium tuberculosis contains two AP endonucleases: MtbXthA and MtbNfo members of the exonuclease III and endonuclease IV families, which are exemplified by Escherichia coli Xth and Nfo, respectively. It has been shown that both MtbXthA and MtbNfo contain AP endonuclease and 3'→5' exonuclease activities. However, it remains unclear whether these enzymes hold 3'-repair phosphodiesterase and nucleotide incision repair (NIR) activities. Here, we report that both mycobacterial enzymes have 3'-repair phosphodiesterase and 3'-phosphatase, and MtbNfo contains in addition a very weak NIR activity. Interestingly, depending on pH, both enzymes require different concentrations of divalent cations: 0.5mM MnCl2 at pH 7.6 and 10 mM at pH 6.5. MtbXthA requires a low ionic strength and 37 °C, while MtbNfo requires high ionic strength (200 mM KCl) and has a temperature optimum at 60 °C. Point mutation analysis showed that D180 and N182 in MtbXthA and H206 and E129 in MtbNfo are critical for enzymes activities. The steady-state kinetic parameters indicate that MtbXthA removes 3'-blocking sugar-phosphate and 3'-phosphate moieties at DNA strand breaks with an extremely high efficiency (kcat/KM=440 and 1280 μM(-1)∙min(-1), respectively), while MtbNfo exhibits much lower 3'-repair activities (kcat/KM=0.26 and 0.65 μM(-1)∙min(-1), respectively). Surprisingly, both MtbXthA and MtbNfo exhibited very weak AP site cleavage activities, with kinetic parameters 100- and 300-fold lower, respectively, as compared with the results reported previously. Expression of MtbXthA and MtbNfo reduced the sensitivity of AP endonuclease-deficient E. coli xth nfo strain to methylmethanesulfonate and H2O2 to various degrees. Taken together, these data establish the DNA substrate specificity of M. tuberculosis AP endonucleases and suggest their possible role in the repair of oxidative DNA damage generated by endogenous and host- imposed factors.

Azide-alkyne cycloaddition ("click chemistry") has found wide use in the analysis of molecular interactions in living cells. 5-ethynyl-2-(hydroxymethyl)tetrahydrofuran-3-ol (EAP) is a recently developed apurinic/apyrimidinic (AP) site analog functionalized with an ethynyl moiety, which can be introduced into cells in DNA constructs to perform labeling or cross-linking in situ. However, as a non-natural nucleoside, EAP could be subject to removal by DNA repair and misreading by DNA polymerases. Here, we investigate the interaction of this clickable AP site analog with DNA polymerases and base excision repair enzymes. Similarly to the natural AP site, EAP was non-instructive and followed the "A-rule", directing residual but easily detectable incorporation of dAMP by E. coli DNA polymerase I Klenow fragment, bacteriophage RB69 DNA polymerase and human DNA polymerase β. On the contrary, EAP was blocking for DNA polymerases κ and λ. EAP was an excellent substrate for the major human AP endonuclease APEX1 and E. coli AP exonucleases Xth and Nfo but was resistant to the AP lyase activity of DNA glycosylases. Overall, our data indicate that EAP, once within a cell, would represent a replication block and would be removed through an AP endonuclease-initiated long-patch base excision repair pathway.

The major human apurinic/apyrimidinic (AP) site endonuclease, APEX1, is a central player in the base excision DNA repair (BER) pathway and has a role in the regulation of DNA binding by transcription factors. In vertebrates, APEX1 knockouts are embryonic lethal, and only a handful of knockout cell lines are known. To facilitate studies of multiple functions of this protein in human cells, we have used the CRISPR/Cas9 system to knock out the APEX1 gene in a widely used non-cancer hypotriploid HEK 293FT cell line. Two stable knockout lines were obtained, one carrying two single-base deletion alleles and one single-base insertion allele in exon 3, another homozygous in the single-base insertion allele. Both mutations cause a frameshift that leads to premature translation termination before the start of the protein's catalytic domain. Both cell lines totally lacked the APEX1 protein and AP site-cleaving activity, and showed significantly lower levels of the APEX1 transcript. The APEX1-null cells were unable to support BER on uracil- or AP site-containing substrates. Phenotypically, they showed a moderately increased sensitivity to methyl methanesulfonate (MMS; ~2-fold lower EC50 compared with wild-type cells), and their background level of natural AP sites detected by the aldehyde-reactive probe was elevated ~1.5-2-fold. However, the knockout lines retained a nearly wild-type sensitivity to oxidizing agents hydrogen peroxide and potassium bromate. Interestingly, despite the increased MMS cytotoxicity, we observed no additional increase in AP sites in knockout cells upon MMS treatment, which could indicate their conversion into more toxic products in the absence of repair. Overall, the relatively mild cell phenotype in the absence of APEX1-dependent BER suggests that mammalian cells possess mechanisms of tolerance or alternative repair of AP sites. The knockout derivatives of the extensively characterized HEK 293FT cell line may provide a valuable tool for studies of APEX1 in DNA repair and beyond.

DNA damage can affect various regulatory elements of the genome, with the consequences for DNA structure, dynamics, and interaction with proteins remaining largely unexplored. We used solution NMR spectroscopy, restrained and free molecular dynamics to obtain the structures and investigate dominant motions for a set of DNA duplexes containing CpG sites permuted with combinations of 5-methylcytosine (mC), the primary epigenetic base, and 8-oxoguanine (oxoG), an abundant DNA lesion. Guanine oxidation significantly changed the motion in both hemimethylated and fully methylated DNA, increased base pair breathing, induced BI→BII transition in the backbone 3' to the oxoG and reduced the variability of shift and tilt helical parameters. UV melting experiments corroborated the NMR and molecular dynamics results, showing significant destabilization of all methylated contexts by oxoG. Notably, some dynamic and thermodynamic effects were not additive in the fully methylated oxidized CpG, indicating that the introduced modifications interact with each other. Finally, we show that the presence of oxoG biases the recognition of methylated CpG dinucleotides by ROS1, a plant enzyme involved in epigenetic DNA demethylation, in favor of the oxidized DNA strand. Thus, the conformational and dynamic effects of spurious DNA oxidation in the regulatory CpG dinucleotide can have far-reaching biological consequences.

Endonuclease VIII-like protein 1 (NEIL1) is a DNA repair enzyme found in higher eukaryotes, including humans. It belongs to the helix-two turn-helix (H2TH) structural superfamily together with Escherichia coli formamidopyrimidine-DNA glycosylase (Fpg) and endonuclease VIII (Nei), and removes a variety of oxidized purine and pyrimidine bases from DNA. Structural, modeling and kinetic studies have established that the bacterial H2TH superfamily enzymes proceed through several conformational intermediates while recognizing and removing their cognate lesions. Here we apply stopped-flow kinetics with detection of intrinsic Trp fluorescence and Förster resonance energy transfer fluorescence to follow the conformational dynamics of human NEIL1 and DNA when the enzyme interacts with undamaged DNA, or DNA containing cleavable or non-cleavable abasic sites, or dihydrouracil lesions. NEIL1 processed a natural abasic site and a damaged base in DNA equally well but showed an additional fluorescently discernible step when DHU was present, likely reflecting additional rearrangements during base eversion into the enzyme's active site. With undamaged DNA and DNA containing a non-cleavable abasic site analog, (3-hydroxytetrahydrofuran-2-yl)methyl phosphate, NEIL1 was diverted to a non-productive DNA conformation early in the reaction. Our results support the view of NEIL1 as an enzyme that actively destabilizes damaged DNA and uses multiple checkpoints along the reaction coordinate to drive substrate lesions into the active site while rejecting normal bases and non-substrate lesions.

Apurinic/apyrimidinic (AP) endonucleases play critical roles in the repair of abasic sites and strand breaks in DNA. Complete genome sequences of Helicobacter pylori reveal that this bacterial specie has a single AP endonuclease. An H. pylori homolog of Xth (HpXth) is a member of exonuclease III family, which is represented by Escherichia coli Xth. Currently, it remains unknown whether this single AP endonuclease has DNA repair activities similar to those of its counterpart in E. coli and other bacteria. We report that HpXth possesses efficient AP site cleavage, 3'-repair phosphodiesterase, and 3'-phosphatase activities but not the nucleotide incision repair function. Optimal reaction conditions for HpXth's AP endonuclease activity are low ionic strength, high Mg2+ concentration, pH in the range 7-8, and temperature 30 °C. The kinetic parameters measured under steady-state conditions showed that HpXth removes the AP site, 3'-blocking sugar-phosphate, and 3'-terminal phosphate in DNA strand breaks with good efficiency (kcat/KM = 1240, 44, and 5,4 μM-1·min-1, respectively), similar to that of E. coli Xth. As expected, the presence of HpXth protein in AP endonuclease-deficient E. coli xth nfo strain significantly reduced the sensitivity to an alkylating agent and H2O2. Mutation of active site residue D144 in HpXth predicted to be essential for catalysis resulted in a complete loss of enzyme activities. Several important structural features of HpXth were uncovered by homology modeling and phylogenetic analysis. Our data show the DNA substrate specificity of H. pylori AP endonuclease and suggest that HpXth counteracts the genotoxic effects of DNA damage generated by endogenous and host-imposed factors.

Formamidopyrimidine-DNA glycosylase (Fpg) removes abundant pre-mutagenic 8-oxoguanine (oxoG) bases from DNA through nucleophilic attack of its N-terminal proline at C1' of the damaged nucleotide. Since oxoG efficiently pairs with both C and A, Fpg must excise oxoG from pairs with C but not with A, otherwise a mutation occurs. The crystal structures of several Fpg-DNA complexes have been solved, yet no structure with A opposite the lesion is available.

Apurinic/apyrimidinic (AP) endonucleases are key DNA repair enzymes involved in the base excision repair (BER) pathway. In BER, an AP endonuclease cleaves DNA at AP sites and 3'-blocking moieties generated by DNA glycosylases and/or oxidative damage. A Triticum aestivum cDNA encoding for a putative homologue of ExoIII family AP endonucleases which includes E. coli Xth, human APE1 and Arabidopsis thaliana AtApe1L has been isolated and its protein product purified and characterized.

Formamidopyrimidine-DNA glycosylase (Fpg) excises 8-oxoguanine (oxoG) from DNA but ignores normal guanine. We combined molecular dynamics simulation and stopped-flow kinetics with fluorescence detection to track the events in the recognition of oxoG by Fpg and its mutants with a key phenylalanine residue, which intercalates next to the damaged base, changed to either alanine (F110A) or fluorescent reporter tryptophan (F110W). Guanine was sampled by Fpg, as evident from the F110W stopped-flow traces, but less extensively than oxoG. The wedgeless F110A enzyme could bend DNA but failed to proceed further in oxoG recognition. Modeling of the base eversion with energy decomposition suggested that the wedge destabilizes the intrahelical base primarily through buckling both surrounding base pairs. Replacement of oxoG with abasic (AP) site rescued the activity, and calculations suggested that wedge insertion is not required for AP site destabilization and eversion. Our results suggest that Fpg, and possibly other DNA glycosylases, convert part of the binding energy into active destabilization of their substrates, using the energy differences between normal and damaged bases for fast substrate discrimination.

8-oxoguanine DNA glycosylase (OGG1), a major DNA repair enzyme in mammalian cells and a component of the base excision repair (BER) pathway, was recently shown to be associated with the microtubule network and the centriole at interphase and the spindle assembly at mitosis. In this study, we determined whether other participants in the BER pathway also bind microtubules in situ and in vitro. Purified recombinant human DNA polymerase beta (DNA Pol beta) and purified recombinant mNEIL2 were chemically conjugated to fluorochromes and photosensitive dyes and used in in situ localization and binding experiments. Results from in situ localization, microtubule co-precipitation and site-directed photochemical experiments showed that recombinant human DNA Pol beta and recombinant mNEIL2 associated with microtubules in situ and in vitro in a manner similar to that shown earlier for another BER pathway component, OGG1. Observations reported in this study suggest that these BER pathway components are microtubule-associated proteins (MAPs) themselves or utilize yet to be identified MAPs to bind microtubules in order to regulate their intracellular trafficking and activities during the cell cycle.

Human 8-oxoguanine-DNA glycosylase OGG1 is an enzyme that removes abundant oxidative lesion 8-oxoguanine (8-oxoG) from DNA. Excision of 8-oxoG by OGG1 is inhibited by the abasic DNA reaction product and is stimulated by AP endonuclease APEX1. Besides 8-oxoG, OGG1 shows activity towards several other base lesions. Here we report that APEX1 efficiently stimulates OGG1 on good substrates (8-oxoadenine, 8-oxoinosine, or 6-methoxy-8-oxoguanine opposite to cytosine) but the stimulation is low or absent with poor OGG1 substrates (8-oxoadenine or 8-oxoinosine opposite to thymine; 8-oxoG or 8-aminoguanine opposite to adenine; 8-oxonebularine, 8-metoxyguanine, inosine or guanine opposite to cytosine). APEX1 significantly improves the ability of OGG1 to excise 8-aminoguanine from its naturally occurring pair with cytosine, making it possible that OGG1 repairs this lesion. Overall, APEX1 serves to improve specificity of OGG1 for its biologically relevant substrates.