|Home | About | Journals | Submit | Contact Us | Français|
Human 8-oxoguanine-DNA-glycosylase (OGG1) efficiently removes mutagenic 8-oxoguanine (8-oxoGua) and 2,6-diamino-4-hydroxy-5-formamidopyrimidine (FapyGua) when paired with cytosine in oxidatively damaged DNA. Excision of 8-oxoGua mispaired with adenine may lead to G→T transversions. Posttranslational modifications such as phosphorylation could affect cellular distribution and enzymatic activity of OGG1. Mutations and polymorphisms of OGG1 may affect enzymatic activity and have been associated with increased risk of several cancers. In this study we use double-stranded oligodeoxynucleotides containing 8-oxoGua:Cyt or 8-oxoGua:Ade pairs, as well as γ-irradiated calf thymus DNA, to investigate the kinetics and substrate specificity of several known OGG1 polymorphic variants and phosphomimetic Ser→Glu mutants. Among the polymorphic variants, A288V and S326C displayed opposite-base specificity similar to that of wild type OGG1, while OGG1-D322N was 2.3-fold more specific for the correct opposite base than the wild type enzyme. All phosphomimetic mutants displayed ~1.5–3-fold lower ability to remove 8-oxoGua in both assays, whereas the substrate specificity of the phosphomimetic mutants was similar to that of the wild type enzyme. OGG1-S326C efficiently excised 8-oxoGua from oligodeoxynucleotides and FapyGua from γ-irradiated DNA but excised 8-oxoG rather inefficiently from γ-irradiated DNA. Otherwise, kcat values for 8-oxoGua excision obtained from both types of experiments were similar for all OGG1 variants studied. It is known that human AP endonuclease APEX1 can stimulate OGG1 activity by increasing its turnover rate. However, when wild type OGG1 was replaced with one of the phosphomimetic mutants, very little stimulation of 8-oxoGua removal was observed in the presence of APEX1.
8-Oxo-7,8-dihydroguanine (8-oxoGua) and 2,6-diamino-4-hydroxy-5-formamidopyrimidine (FapyGua) are pre-mutagenic DNA lesions that appear in DNA damaged by reactive oxygen species of endogenous and environmental origin . During replication, 8-oxoGua directs misincorporation of dAMP  and thereby induces G→T transversions, which in mammals can activate oncogenes or inactivate tumor suppressor genes [3, 4]. Likewise, FapyGua pairs with Ade and leads to G→T transversions in mammalian cells [5, 6]. A causal role of oxidative damage to DNA in human cancer development has not been demonstrated directly; nevertheless, oxidatively induced DNA lesions, including 8-oxoGua, are responsible for mutations that may play a role in carcinogenesis .
8-oxoGua and FapyGua are removed from DNA by base excision repair (BER) . As part of this process, all organisms possess an enzymatic system that averts mutagenic load caused by these two lesions. In humans, a system has been described that consists of three enzymes: 8-oxoGua-DNA glycosylase OGG1 (UniProt accession number O15527), mismatched adenine-DNA glycosylase MUTYH, and 8-oxo-7,8-dihydrodeoxyguanosine triphosphatase NUDT1 (MTH1) (reviewed in ). OGG1 excises 8-oxoGua paired with Cyt, the context in which this oxidized base is naturally formed, but not from 8-oxoGua:Ade pairs that appear following misincorporation of dAMP opposite 8-oxoGua or by insertion of 8-oxodGTP opposite Ade. MUTYH removes Ade from 8-oxoGua:Ade pairs followed by additional repair processes that convert this mispair into 8-oxoGua:Cyt, which is repaired by OGG1. In parallel, NUDT1 hydrolyzes 8-oxodGTP, preventing its misincorporation during DNA replication. In addition to 8-oxoGua, human and other OGG1 proteins efficiently remove FapyGua from DNA with similar excision kinetics to that of removal of 8-oxoG [10–13]. In agreement with this fact, FapyGua paired with Cyt is also efficiently removed by human OGG1 from synthetic oligodeoxynucleotides . Simultaneous inactivation of OGG1 and MUTYH in transgenic mice predisposes these animals to lymphomas, lung and ovarian tumors, associated with many G→T transversions in codon 12 of the K-ras protooncogene .
Ultimately, fidelity of the 8-oxoGua repair system depends on discrimination between 8-oxoGua:Cyt and 8-oxoGua:Ade pairs by OGG1. This enzyme possesses two catalytic activities, a strong DNA glycosylase activity specific for 8-oxoGua and FapyGua, and a relatively weak apurinic/apyrimidinic (AP) lyase activity which, after base excision, cleaves the DNA backbone by elimination of the 3′-phosphate of the damaged deoxynucleotide (β-elimination) [11, 16, 17]. Due to the weak AP lyase activity and high affinity for the AP product, the turnover of OGG1 is low but the enzyme is stimulated by the major human apurine/apyrimidine endonuclease APEX1 (UniProt accession number P27695) [18–21]. OGG1 is highly selective for 8-oxoGua:Cyt substrates and discriminates against 8-oxoGua:Thy, 8-oxoGua:Gua, and especially 8-oxoGua:Ade, both at the level of its glycosylase and the AP lyase activity [22, 23]. The C/A specificity of OGG1 is influenced by several factors including ionic strength, presence of Mg2+ ions , and interactions with APEX1 .
Many single-nucleotide polymorphisms of OGG1 gene have been found in human populations and deposited in the NCBI dbSNP database  or reported individually [26–29]. Of these polymorphisms, 14 change the amino acid sequence of its major protein isoform OGG1-1a (A3P, P27T, A53T, A85S, R131Q, R154H, R229Q, E230Q, A288V, G308E, S320T, D322N, S326C). Two more variants, R46Q and S232T, have been reported only from human tumors [26, 30]. Few proteins encoded by these variants have been characterized with respect to their function, kinetics and substrate specificity. Most attention has been given to the OGG1-S326C variant, which is associated with an increased risk of lung, and possibly gastrointestinal, cancer, especially in patients exposed to environmental factors such as smoking or animal protein consumption [31, 32]. Still, functional characterization of this protein has been inconclusive. In E. coli mutator strain complementation tests, OGG1-S326C has been reported as either being less efficient than wild-type OGG1  or providing normal complementation . Cell extracts from lymphocytes from OGG1-S326 and OGG1-S326 homozygous individuals show similar abilities to excise 8-oxoGua . OGG1-S326C exhibits less efficient excision of 8-oxoGua and FapyGua from γ-irradiated DNA than the wild-type enzyme  and shows less proficiency in excising 8-oxoGua from oligodeoxynucleotides . Among other OGG1 polymorphic variants, limited kinetic information is available for R46Q, A53T, R154H, and A288V [29, 36].
Many BER proteins undergo posttranslational modification, including acetylation and phosphorylation . OGG1 interacts physically with protein kinases CDK4, c-ABL and PKC, with CDK4 and PKC able to modify OGG1 in vitro [38, 39]. Phosphorylation of OGG1 by CDK4 was reported to activate the enzyme , while phosphorylation by PKC had no effect on OGG1 activity , suggesting that several sites in OGG1 may be phosphorylated. In no case has the site of OGG1 phosphorylation been identified. Additionally, the OGG1-S326C variant, which shows aberrant intracellular sorting, can be rescued by mutating residue 326 to Glu, a substitution approximating the bulk and charge of phosphoserine .
In this report, we analyze the activity, substrate specificity and kinetics of two naturally occurring polymorphic variants of OGG1, A288V and D322N, comparing them with wild-type and S326C variants of the enzyme. We use a neural network trained on a large set of experimentally proven protein phosphorylation sites to predict additional sites of high phosphorylation probability in OGG1, then introduce phosphomimetic Ser→Glu substitutions at these positions, determining changes in the activity, substrate specificity, and interactions with AP endonuclease of the resulting enzyme variants.
Association of OGG1 polymorphisms with succeptibility to human cancer and other diseases is an area of active research [31, 41]. Among known polymorphic variants, S326C, associated with the increased risk of lung cancer, has been extensively studied, since the frequency of this allele in the general population is ~0.25. Several functional defects have been found in this form of the OGG1 protein, including abnormal cell cycle-dependent localization , protein dimerization, changes in opposite-base specificity and inability to be stimulated by APEX1 . Therefore, we have used OGG1-S326C as a “reference” variant, with which to compare other enzyme variants. Of other polymorphic OGG1 forms, we have chosen OGG1-A288V and OGG1-D322N for structural reasons. In the OGG1-DNA complex , Ala-288 forms direct contacts to DNA, while a highly conserved Asp-322 is involved in positioning the imidazole ring of an absolutely conserved His-270 residue, which in turn binds to the 5′-phosphate of the damaged nucleotide monophosphate (Fig. 1B). The A288V polymorphism in the germline has been found in Alzheimer’s disease patients and the activity of the OGG1-A288V has been reported to be lower than of the wild-type enzyme . The activity of OGG1-D322N has not previously been addressed.
Phosphorylation of OGG1 can affect its biological functions at several level, including the intrinsic activity and intracellular localization [39, 40]. The sites of phosphorylation in this enzyme are presently unknown. Thus, to select residues for phosphomimetic Ser/Thr modifications, we used the NetPhos 2.0 server (www.cbs.dtu.dk/services/NetPhos/), a neural network that predicts the probability of phosphorylation at a given site using a constantly updated learning set based on the sequences of experimentally proven phosphorylation sites . In Table 1, we summarize the results of an analysis of overall phosphorylation probability within the OGG1 sequence. It should be noted that the NetPhos score is not the exact probability, but rather a function of the probability of a site being phosphorylated. A NetPhos score > 0.5 is generally considered a threshold for prediction of a Ser/Thr residue as a possible phosphorylation site, and the higher the score, the higher the probability of the site being phosphorylated . As an additional criterion of possible phosphorylation, we used the surface accessibility of the Ser/Thr residues in the structure of OGG1, limiting the range of mutagenesis targets to the residues not buried in the protein globule according to their surface exposure ratio (Table 1). Therefore, we have chosen S231, S232, S280, and S326, the residues with the highest overall scores (> 0.99), for biochemical characterization of the phosphomimetic Ser to Glu substitution. Additionally, a double mutant S231E/S232E, mimicking double phosphorylation at two adjacent sites, was studied. All of these residues are located at the surface of the OGG1 protein globule far away from the protein–DNA interface (Table 1, Fig. 1A) and thus are accessible for phosphorylation; Ser-326 is missing from the OGG1–DNA crystal structure  but is inferred to be surficial and distant from DNA.
OGG1 is part of an enzymatic system responsible for prevention of mutations generated by 8-oxoGua and FapyGua . Since 8-oxoGua directs pre-mutagenic misincorporation of dAMP during replication, a distinguishing feature of OGG1 is its preference for removal of 8-oxoGua from 8-oxoGua:Cyt pairs compared with that from 8-oxoGua:Ade pairs [22, 23, 45]. To study the effect of amino acid substitutions on the activity and opposite-base specificity of OGG1, we determined the kinetic constants kcat and KM for reaction of cleavage of 8-oxoGua:Cyt and 8-oxoGua:Ade substrates by wild-type and mutant OGG1 enzymes. Fig. 2 shows a typical dependence of the reaction velocity on the substrate concentration in double reciprocal coordinates for the wild-type enzyme. The specificity constant, ksp = kcat/KM, was calculated for each enzyme and substrate, and the ratio of the ksp for 8-oxoGua:Cyt to the ksp for 8-oxoGua:Ade was used as a measure of the biologically relevant opposite-base specificity (C/A specificity) . In the wild-type enzyme, the C/A specificity of 4.9 was due mostly to the lower value of KM for the 8-oxoGua:Cyt substrate (Table 2, Table 3), similar to what was reported in the literature [23, 45]. The KM values for cleavage of 8-oxoGua:Cyt by OGG1-A288V and OGG1-D322N were higher than that for wild-type OGG1. Due to a concomitant increase in kcat for OGG1-A288V, no significant difference in ksp and C/A specificity was observed for this form of the enzyme (Table 2, Table 3). Interestingly, the activity of OGG1-D322N towards the 8-oxoGua:Cyt substrate was the lowest of all polymorphic variants studied, but this variant showed an even lower activity on the 8-oxoGua:Ade substrate. As a result, the overall C/A specificity of OGG1-D322N was 12, 2.4-fold higher than the C/A specificity of wild-type OGG1 (Table 2, Table 3). In the OGG1-S326C variant, the KM value for the cleavage of 8-oxoGua:Cyt substrate was nearly the same as for the wild-type OGG1, and decreased for the 8-oxoGua:Ade substrate in the mutant, but, as the ksp value decreased for both 8-oxoGua:Cyt and 8-oxoGua:Ade, the C/A specificities of wild-type OGG1 and OGG1-S326C were similar (Table 2, Table 3). Thus, of all studied natural variants of the enzyme, OGG1-D322N demonstrated the highest C/A specificity. The values of kinetic constants found for cleavage of 8-oxoGua:Cyt by OGG1-A288V and OGG1-S326C were in an overall agreement with published data [29, 35].
In the reaction of 8-oxoGua:Cyt cleavage by phosphomimetic mutants of OGG1, we observed an increase in both KM and kcat for OGG1-S231E, OGG1-S232E and OGG1-S231S/S232E, and a decrease in kcat for OGG1-S280E and OGG1-S326E, as compared with wild-type OGG1 (Table 2). Overall, the decrease in ksp for all phosphomimetic mutants of OGG1 but OGG1-S231E reveals that these enzymes are ~2-fold less active than wild-type OGG1. For OGG1-S231E, the increase in KM was compensated by an increase in kcat, leading to only a marginal decrease in the activity of the mutant enzyme. For the 8-oxoGua:Ade substrate, the KM value in the phosphomimetic mutants either decreased in comparison with that for wild-type OGG1 (OGG1-S231E and OGG1-S280E), or did not change (OGG1-S232E, OGG1-S231S/S232E, OGG1-S326E). The kcat value decreased in all cases; as a result, all phosphomimetic mutants excised 8-oxoGua from 8-oxoGua:Ade pairs less efficiently than did the wild-type enzyme (Table 3). The C/A specificity for all phosphomimetic mutants of OGG1 resembled closely that of the wild-type enzyme (Table 3).
In addition to measuring kinetic constants of DNA glycosylases on oligodeoxynucleotide substrates containing 8-oxoGua, the substrate specificity of these enzymes may be analyzed using high molecular weight DNA damaged by γ-irradiation or other treatment, with the following analysis of excised bases by gas chromatography/mass spectrometry (GC/MS) with isotope dilution . This assay reveals the spectrum of damaged bases released by a given enzyme, including those not easily introduced into oligodeoxynucleotides, such as formamidopyrimidines. When applied to wild-type human OGG1 and its forms R46Q, R154H, and S326C, this approach has shown that OGG1 excises only 8-oxoGua and FapyGua out of more than 20 oxidized bases detected in this system [11, 36]. Both OGG1 and OGG1-S326C excise 8-oxoGua and FapyGua, with kcat and ksp for OGG1-S326C reported about twofold lower than for wild-type OGG1 .
To determine the full spectrum of substrate bases excised from their naturally occurring base pairs by OGG1 and its variants, we used γ-irradiated calf thymus DNA and employed E. coli Fpg protein, a functional counterpart but not a structural homolog of OGG1, with well established specificity for 8-oxoGua, FapyGua, and 4,6-diamino-5-formamidopyrimidine (FapyAde) [48, 49], as an additional control. All studied OGG1 variants were able to excise FapyGua and 8-oxoGua from DNA, with OGG1-S326C being the least active for excision of 8-oxoGua (Table 4). Figures 3A and 3B illustrate excision of 8-oxoGua, FapyGua and FapyAde by OGG1 and Fpg, respectively. In agreement with previous results, OGG1 excised 8-oxoGua and FapyGua, but not FapyAde, whereas all three products were removed by Fpg from DNA. Other modified bases monitored by GC/MS were not excised, indicating that mutant OGG1 forms do not acquire broader substrate specificity compared with the wild-type enzyme.
The values of kinetic constants for excision of FapyGua and 8-oxoGua by various forms of OGG1 are summarized in Table 4. Excision of 8-oxoGua by OGG1-A288V was characterized by a somewhat lower kcat than that for the wild-type enzyme but, due to a concomitant decrease in KM, the value of ksp for OGG1 and OGG1-A288V were very similar. The values of kcat and KM for FapyGua excision were higher than for 8-oxoGua excision by both OGG1 and OGG1-A288V, making these two forms of the enzyme equally well suited for excision of both lesions. The polymorphic variant OGG1-D322N showed notably lower kcat and ksp for excision of both lesions, with a more pronounced effect on 8-oxoGua excision. In this case, the ksp(WT)/ksp(mutant) ratio was 4.3 for 8-oxoGua excision and 1.6 for FapyGua excision, consistent with a decrease in OGG1-D322N activity observed with oligodeoxynucleotide substrates. Interestingly, OGG1-S326C was the least active variant in excising 8-oxoGua while retaining appreciable activity towards FapyGua. For the latter substrate, the value of kcat decreased 3.8-fold in comparison with that for the wild-type enzyme, but due to a concomitant decrease in KM for OGG1-S326C, the ksp value for FapyGua excision by this variant was only twofold lower than the ksp for FapyGua excision by OGG1. In contrast, ksp for 8-oxoG excision by OGG1-S326C was 6.2-fold lower than that of wild-type OGG1.
All phosphomimetic mutants of OGG1 demonstrated reduced abilities to excise FapyGua and especially 8-oxoGua when compared to the wild-type enzyme. Both kcat and KM for 8-oxoGua excision by OGG1-S231E, OGG1-S232E, OGG1-S231E/S232E, OGG1-S280E, and OGG1-S326E were elevated in comparison with the kinetic constants for wild-type OGG1; as a result, ksp was 2.2–3.6-fold lower for all phosphomimetic mutants than for wild-type OGG1. The reduction in ksp for FapyGua excision also was evident although not as pronounced (1.1–1.7-fold) as in the case of 8-oxoGua (Table 4). For OGG1-S326E, ksp characterizing the excision of both 8-oxoGua and FapyGua was lowered in comparison with the wild-type OGG1 due to an increase in KM with a much less effect on kcat. Overall, kcat values of 8-oxoGua excision from irradiated DNA are in a good agreement with data for the cleavage of 8-oxoGua:Cyt oligodeoxynucleotide substrates (compare Tables 2 and and4).4). Much higher values obtained for apparent KM in the irradiated DNA assay are due to a much lower concentration of damaged bases in this substrate, which causes KM to increase due to longer lesion search time and a correspondingly lower association rate constant in the Michaelis–Menten equation as discussed previously .
Regulation of protein–protein interactions by post-translational modification, including phosphorylation, is widely encountered in nature. We and others have shown that human AP endonuclease APEX1 stimulates the activity of wild-type OGG1, most likely through DNA-mediated protein–protein interactions [18–21]. Therefore, we asked whether putative phosphorylation of OGG1 at sites of high phosphorylation probability could influence the ability of APEX1 to stimulate OGG1. To address this question, we investigated the activity of phosphomimetic mutants of OGG1 in the presence and in the absence of APEX1. All forms showed a significantly lower ability to be stimulated by APEX1 than the wild-type enzyme (Fig. 4). APEX1 elicited only a moderate stimulation of OGG1-S326E, OGG1-S231E and OGG1-S232E, whereas the activity of OGG1-S280E and OGG1-S321E/S232E in the presence and in the absence of the AP endonuclease was nearly indistinguishable. Also, OGG1-S280E, OGG1-S326E, and possibly OGG1-S231E, lacked a pronounced burst phase characteristic of wild-type OGG1 (compare panel A with panels B–D in Fig. 4). This result may indicate that reaction rates are limited by chemical steps of the reaction rather than by the product release step, as had been suggested for cleavage of suboptimal substrates, including 8-oxoGua:Ade, by wild-type OGG1 .
Relatively few polymorphisms affecting the protein sequence of OGG1 have been characterized with respect to their function. Population data are available for only five polymorphisms that deviate from the reference sequence . By far, the most widely encountered variant is OGG1 326C allele (refSNP ID rs1052133), the frequency of which varies from ~0.1 in African Americans to > 0.5 in some Japanese populations . The other alleles are much less common: the reported frequency of the OGG1 85S allele (refSNP ID rs17050550) is ~0.04 (Centre d’Etude du Polymorphisme Human population sample, Caucasian origin), and of the 229Q allele (refSNP ID rs1805373), 0.008 (NIEHS HSP_GENO_PANEL population sample, ethnic origin not specified) to 0.1 (NIEHS YRI_GENO_PANEL population sample, Sub-Saharan African). The OGG1 288V and 322N alleles also are rare; in the NIH PDR90 population sample, the global frequency of OGG1 288V allele (refSNP ID rs1805373) is 0.011, and the global frequency of OGG1 322N allele is 0.006 . Given the functional defects reported for OGG1-S326C and OGG1-R229Q proteins [33, 35, 40, 51–53], it was interesting to analyze various aspects of activity of other variants of OGG1. We have selected OGG1-A288V and OGG1-D322N as the variants in which, as deduced from the structural data , the DNA-binding interface of the protein could be affected.
The OGG1-A288V variant has been observed in patients with Alzheimer’s disease . A very limited kinetic analysis of this variant has been reported, suggesting that KM of OGG1-A288V is moderately higher than that of the wild-type enzyme . In our experiments, A288V was ~30% less efficient (in terms of ksp) than wild-type OGG1 in the oligodeoxynucleotide cleavage assay (8-oxoGua:Cyt substrate) but virtually indistinguishable from wild-type enzyme in the irradiated DNA assay. Little difference was observed in cleavage of 8-oxoGua:Ade substrate between wild-type OGG1 and OGG1-A288V, making the latter the least specific form of all OGG1 variants studied. In the OGG1–DNA complex , the Ala-288 backbone amide forms a hydrogen bond with an internucleoside phosphate p(5) residing in the non-damaged strand and remote from the active site. Additionally, the side chain methyl group of Ala-288 makes van der Waals contacts with non-bridging oxygens of the same phosphate. While the hydrogen bond may be lost in the lesion search complex  and in some late complexes , the van der Waals contacts are present in all reported OGG1-DNA complexes [42, 54–60]. The bulkier isopropyl side chain of Val may induce local distortion in the region of p(5), partly destabilizing the OGG1–DNA complex. However, it is not clear whether the moderate decrease in the activity and C/A-specificity of OGG1-A288V, as measured on oligodeoxynucleotide substrates, may impair the activity of this variant in vivo and contribute to the pathogenesis of Alzheimer’s disease.
Of all variants studied, OGG1-D322N possessed the highest C/A specificity. In the crystal structure of the complex of DNA with catalytically inactive OGG1 , and in several other structures of OGG1, either free or bound to DNA [54–62], the side chain carboxyl group of Asp-322 forms a hydrogen bond with the Nδ1 atom of His-270. The Nε2 atom of the His-270 imidazole ring in turn hydrogen bonds to a non-bridging oxygen of the phosphodiester bond immediately 5′ to the damaged deoxynucleoside (Fig. 1B). Substitutions of Ala or Leu for His-270 drastically decrease OGGl activity . The structures of OGG1/DNA complexes approximating other intermediates of the catalytic cycle suggest considerable dynamics of His-270, which stacks with undamaged G in the lesion search complex , disengages from this interaction in the early and advanced lesion detection complexes [59, 62], and stacks with Phe-319 in the late abasic product complex  and in the free enzyme . In all these cases, however, the bond between Asp-322 and either Nδ1 or Nε2 of His-270 is maintained. Donating two hydrogen bonds to acidic moieties requires the imidazole ring of His-270 to be in the doubly protonated, positively charged state, which may be important in interactions of His-270 with the negatively charged DNA backbone or transient stacking of His-270 with DNA bases during lesion search and recognition. Substitution of Asn for Asp-322 would likely maintain the hydrogen bonding with His-270 but eliminate the positive charge. This change appears to destabilize modestly the Michaelis complex with the 8-oxoGua:Cyt substrate while not affecting the catalytic constant (Table 2), suggesting that correct adjustment of catalytic residues in the OGG1-D322N Michaelis complex is preserved. In contrast, with the incorrect 8-oxoGua:Ade substrate, the KM value is nearly the same in both wild-type OGG1 and OGG1-D322N while kcat is reduced, possibly reflecting disorganization of the active site when the incorrect substrate binds to the D322N enzyme. On the other hand, in the irradiated DNA assay, kcat rather than KM was affected for OGG1-D322N, most probably because the reaction pathway leading to the Michaelis complex is different for short oligodeoxynucleotides carrying a single lesion and long DNA with interspersed lesions, In the latter case, the k1 association constant in the equation for KM is dominated by one-dimensional sliding to the lesion rather than by direct binding of the lesion . As substrate recognition by OGG1 proceeds through at least three kinetically stable intermediate complexes [45, 64], it is also possible that the D322N mutation may have an impact on selected steps of this process and/or on the sliding of the enzyme along DNA.
The OGG1 326C allele has been associated with an increased cancer risk in a number of epidemiological studies (reviewed in Refs [31, 32]). The activity of OGG1-S326C variant has been studied; however, the precise nature of the functional defects in this enzyme has not been established. The comparison of the ability of wild-type and S326C enzymes to counteract spontaneous or induced mutagenesis in E. coli, Salmonella, and cultured human cells showed either the functional equivalence of these two variants [11, 65] or a functional deficiency in OGG1-S326C [33, 52]. Extracts of lymphocytes from individuals homozygous for either form of OGG1 have the same ability to excise 8-oxoGua from DNA . No significant differences in the kinetic parameters of wild-type OGG1 and OGG1-S326C as glutathione-S-transferase fusion proteins has been found using the oligodeoxynucleotide cleavage assay while both kcat and ksp were reported ~2-fold lower than those for wild-type OGG1 in the γ-irradiated DNA cleavage assay . Unlike wild-type OGG1, OGG1-S326C is prone to dimerization, potentially producing a non-functional enzyme that is inefficiently stimulated by AP endonuclease . On the other hand, the functional impairment in OGG1-S326C may be due not to lower enzyme activity but to incorrect cell localization during the cell cycle .
In this study, we found that OGG1-S326C has a ~30% lower activity (in terms of ksp) than the wild-type OGG1 acting on 8-oxoGua:Cyt and 8-oxoGua:Ade oligodeoxynucleotide substrates. A different picture was observed in the irradiated DNA assay. While the removal of FapyGua lesions by OGG1-S326C was only ~2-fold lower than by wild-type OGG1, the S326C variant was much less efficient (~6-fold lower) than the wild-type in its ability to remove 8-oxoGua from high-molecular-weight DNA. Thus, our findings are in general agreement with an earlier study of the activity and substrate specificity of OGG1-S326C , confirming the usefulness of this variant as a reference point for kinetics of other OGG1 mutants. Differences in the relative efficiency of excision of certain damaged bases from oligodeoxynucleotide substrates and from high-molecular-weight DNA by the same enzyme is rather common for DNA glycosylases. In particular, such differences have been observed before for Fpg, a bacterial enzyme overlapping with OGG1 in its substrate specificity except for excision of FapyAde, which is not removed by OGG1 from DNA or oligodeoxynucleotides [14, 49, 50, 66]. It is possible that the S326C substitution more significantly affects the ability of OGG1 to participate in the repair of 8-oxoGua and thus represents a risk factor in carcinogenesis.
Phosphorylation represents an established mechanism for regulating the function of certain proteins, including enzymatic activity, protein–protein interactions, cell sorting, etc. . As obtaining pure proteins phoshorylated at a defined site is hard to achieve, replacement of Ser or Thr with acidic residues, Asp or Glu, is often used as a convenient tool to study potential effects of phosphorylation in a diverse set of proteins. Such phosphomimetic mutations reproduce accurately both the structural and the functional consequences of phosphorylation [68–70]. OGG1 contains several Ser and Thr residues located in sequences with a high probability of phosphorylation (Table 1), and has been shown to be phosphorylated, although the modified residues have not been specifically identified [38, 39]. In fact, one of the putative phosphorylation residues is Ser-326, and the inability of OGG1-S326C to be phosphorylated at this site has been proposed as a possible cause of the functional deficiency of this OGG1 form . The phosphomimetic strategy was employed to explore the consequences of Ser-326 phosphorylation for cell sorting of OGG1 . However, data on the activity or substrate specificity of this phosphomimetic mutant other than confirmation that the OGG1-like activity is present in nuclear extracts of transfected HeLa cells, are unavailable. In this paper we have constructed and analyzed a series of phosphomimetic mutants at sites with the highest probability of phosphorylation (Table 1). All mutants were ~2-fold lower in activity than the wild-type protein in the oligodeoxynucleotide assay and 1.1–3.6-fold lower in the irradiated DNA assay, indicating that phosphorylation of OGG1 is not likely to be involved in regulating its activity. This result contrasts to the moderate activation of OGG1 by another posttranslational modification, acetylation at Lys-338/Lys-341 in the C-terminal tail of the protein . In other human DNA glycosylases, phosphorylation have been shown to increase the activity of MUTYH [72, 73] and uracil-DNA glycosylase (UNG) [74, 75].
Protein–protein interactions are important in the coordination of sequential BER steps; also as a potential target for regulation by phosphorylation. The ability of OGG1 to be stimulated by APEX1 is abrogated by the S326C substitution . We have shown that the same is true for phosphomimetic mutants of OGG1 (Fig. 4). Since Ser-231, Ser-232, Ser-280, and Ser-326 residues are located a significant distance apart on the surface of OGG1 globule, it is unlikely that all these mutations disrupt the OGG1–APEX1 interaction interface. However, the phosphomimetic mutations could alter the structure of some transient intermediate protein–DNA complexes that occurs during displacement of OGG1 by APEX1. The nature of such complexes is currently under investigation in our laboratory using stopped-flow enzyme kinetics. If regulation of functional interactions with APEX1 is indeed affected by phosphorylation of OGG1, this reaction may be involved in switching between APEX1-assisted and NEIL1-assisted subpathways of OGG1-initiated BER .
Other processes involving DNA glycosylases may be affected by protein phosphorylation. For instance, phosphorylation regulates proteasomal degradation of UNG [75, 77]. In the case of OGG1, phosphorylation may be required for association with chromatin  and localization in the nucleolus . It remains to be seen whether phosphomimetic mutants of OGG1 differ from wild-type protein in these aspects or in other properties such as intracellular trafficking, interactions with other BER components, etc.
The C/A specificity of OGG1 is important in preventing 8-oxoGua-induced mutagenesis. We have shown that the C/A specificity of OGG1 and Fpg is highest under nearly-physiological conditions, due to a sharp decrease in the enzyme’s activity on 8-oxoGua:Ade substrates with increasing ionic strength and Mg2+ concentration , and that APEX1 stimulates OGG1 to a higher degree on 8-oxoGua:Cyt than on 8-oxoGua:Ade substrates . Compared with these factors, the natural variations and phosphomimetic mutations in OGG1 had a lower impact on the C/A specificity, which varied between 70–240% of the specificity of the wild-type enzyme. Therefore, it is unlikely that the erroneous repair of 8-oxoGua:Ade mispairs by the studied forms of OGG1 would contribute significantly to the mutagenic load, or that phosphorylation of OGG1 could be used by the cell to regulate the enzyme’s opposite-base specificity.
The 8-oxoGua-containing oligodeoxyribonucleotide 5′-d(CTCTCCCTTCXCTCCTTTCCTCT)-3′ (X = 8-oxoGua) and its complementary strand 5′-d(AGAGGAAAGGAGNGAAGGGAGAG)-3′ (N = Ade or Cyt) were synthesized by Operon Biotechnologies (Huntsville, AL). The 8-oxoGua-containing strand was 32P-labeled using γ[32P]ATP and phage T4 polynucleotide kinase (New England Biolabs, Beverly, MA) according to the manufacturer’s protocol, then annealed to a complementary strand to produce duplexes containing an 8-oxoGua:Cyt or 8-oxoGua:Ade pair. His6-tagged human AP endonuclease APEX1 was purified as described .
OGG1 mutants were produced using a QuikChange Multi site-directed mutagenesis kit (Stratagene, Cedar Creek, TX) with pET-15b-hOGG1-1a plasmid  as a template. The presence of target mutation and the absence of other mutations were confirmed by sequence analysis. Plasmids carrying the mutant OGG1 coding sequence were used to transform E. coli BL21(DE3)RIL. Wild-type and mutant His6-tagged OGG1 were purified as described , except that pre-charged Ni-NTA chelating resin (Qiagen, Venlo, the Netherlands) was used for affinity chromatography. The concentration of the active wild-type enzyme was determined from burst phase kinetic experiments as described .
The standard reaction mixture (20μl) included 20 mM HEPES-NaOH (pH 7.5), 50 mM KCl, 1 mM DTT, 1 mM EDTA, and radioactively labeled 8-oxoGua:Cyt substrate (2–400 nM) or 8-oxoGua:Ade substrate (5–1500 nM). The cleavage reaction was initiated by adding wild-type or mutant OGG1 (10–20 nM for 8-oxoGua:Cyt; 20–50 nM for 8-oxoGua:Ade), allowed to proceed for 20 min (8-oxoGua:Cyt) or 30 min (8-oxoGua:Ade) and terminated by addition of putrescine-HCl (pH 8.0) to a final concentration of 125 mM and heating at 95°C for 5 min to fully cleave the AP site product of the OGG1 glycosylase reaction. An equal volume of formamide dye was added, the mixture was heated for 3 min at 95°C, and the reaction products were separated by electrophoresis in a denaturing polyacrylamide gel. The bands were quantified using a Storm 840 system and ImageQuant v5.2 software (GE Healthcare Life Sciences, Uppsala, Sweden). The kinetic constants, KM and kcat, were determined by non-linear least-square fitting using SigmaPlot v8.0 software (SPSS Inc., Chicago, IL).
Calf thymus DNA (Sigma-Aldrich, St. Louis, MO) was dissolved in phosphate buffer (pH 7.4) at a concentration of 0.3 mg/ml. Aliquots of this solution were bubbled with N2O and γ-irradiated at doses 5, 10, 20, 40 and 60 Gy. Irradiated samples were dialyzed against water for 18 h at 4°C. Aliquots of dialyzed samples (50μg) were dried under vacuum. For enzymatic hydrolysis, DNA samples were dissolved in 50μl of an incubation buffer consisting of 50 mM phosphate buffer (pH 7.4), 100 mM KCl, 1 mM EDTA and 0.1 mM dithiotreitol. Aliquots of 8-oxoGua-13C,15N3, FapyGua-13C,15N2 and FapyAde-13C,15N2 were added as internal standards. Samples were incubated with 1μg of an enzyme for 30 min at 37°C in a water bath, and then processed and analyzed by GC/MS as described [78, 79]. Time dependence of excision was measured by incubation of DNA samples, which were irradiated at 20 Gy, with 1μg of the enzyme for 0, 10, 20 and 30 min. For the measurement of excision kinetics, two sets of DNA samples γ-irradiated at 5, 10, 20, 40 and 60 Gy were prepared with three replicates for each data point. One set of these samples was used to determine the levels of modified bases in DNA samples. The samples were hydrolyzed with formic acid, derivatized and analyzed by GC/MS as described elsewhere . The measured levels were expresses as nmol of a lesion/mg of DNA. The concentration of each lesion in the incubation buffer was calculated on the basis of the DNA concentration (50μg/50μl of incubation buffer). The concentration ranges of FapyGua and 8-OH-Gua were 1.25–7.67μM 0.56–3.58μM, respectively. The other set was incubated with an enzyme and analyzed by GC/MS as described above. The kinetic constants and standard deviations were calculated by non-linear least square fitting.
The standard reaction mixture (20μl) included 20 mM HEPES-NaOH (pH 7.5), 50 mM KCl, 1 mM DTT, 5 mM MgCl2, 50 nM substrate and 10 nM wild-type or mutant OGG1. The reaction was allowed to proceed for 20 min, aliquots were withdrawn at the required time and quenched by heating with putrescine-HCl as described above. Reaction products were analyzed by electrophoresis in denaturing polyacrylamide gel followed by phosphorimaging.
We thank Dr. Masaaki Moriya (Stony Brook University) for his advice with mutagenesis procedures. Support from Russian Foundation for Basic Research (08-04-00596-a), the Presidium of the Russian Academy of Sciences (22.7, 22.14), and the Integration Project No. 98 from Siberian Division of the Russian Academy of Sciences is acknowledged. The project was supported in part by Grants R01 CA017395 and P01 CA047995 from the National Cancer Institute. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Cancer Institute or the National Institutes of Health. Certain commercial equipment or materials are identified in this paper in order to specify adequately the experimental procedure. Such identification does not imply recommendation or endorsement by the National Institute of Standards and Technology, nor does it imply that the materials or equipment identified are necessarily the best available for the purpose.