Impact of Ribonucleotide Backbone on Translesion Synthesis and Repair of 7,8-Dihydro-8-oxoguanine*

Numerous ribonucleotides are incorporated into the genome during DNA replication. Oxidized ribonucleotides can also be erroneously incorporated into DNA. Embedded ribonucleotides destabilize the structure of DNA and retard DNA synthesis by DNA polymerases (pols), leading to genomic instability. Mammalian cells possess translesion DNA synthesis (TLS) pols that bypass DNA damage. The mechanism of TLS and repair of oxidized ribonucleotides remains to be elucidated. To address this, we analyzed the miscoding properties of the ribonucleotides riboguanosine (rG) and 7,8-dihydro-8-oxo-riboguanosine (8-oxo-rG) during TLS catalyzed by the human TLS pols κ and η in vitro. The primer extension reaction catalyzed by human replicative pol α was strongly blocked by 8-oxo-rG. pol κ inefficiently bypassed rG and 8-oxo-rG compared with dG and 7,8-dihydro-8-oxo-2′-deoxyguanosine (8-oxo-dG), whereas pol η easily bypassed the ribonucleotides. pol α exclusively inserted dAMP opposite 8-oxo-rG. Interestingly, pol κ preferentially inserted dCMP opposite 8-oxo-rG, whereas the insertion of dAMP was favored opposite 8-oxo-dG. In addition, pol η accurately bypassed 8-oxo-rG. Furthermore, we examined the activity of the base excision repair (BER) enzymes 8-oxoguanine DNA glycosylase (OGG1) and apurinic/apyrimidinic endonuclease 1 on the substrates, including rG and 8-oxo-rG. Both BER enzymes were completely inactive against 8-oxo-rG in DNA. However, OGG1 suppressed 8-oxo-rG excision by RNase H2, which is involved in the removal of ribonucleotides from DNA. These results suggest that the different sugar backbones between 8-oxo-rG and 8-oxo-dG alter the capacity of TLS and repair of 8-oxoguanine.

DNA replication is essential to maintain genomic integrity. Replicative DNA polymerases (pols) 3 synthesize a new DNA strand by incorporating deoxyribonucleotide triphosphates (dNTPs) with high fidelity. In the cellular nucleotide pool, the concentration of RNA precursors, i.e. ribonucleotide triphosphates (rNTPs), is 1 or 2 orders of magnitude higher than that of dNTPs (1,2). Although pols can discriminate between dNTPs and rNTPs during DNA replication, this selectivity is not perfect (1). More than 10 6 ribonucleotides can be incorporated into the genome per cell (3). Ribonucleotides embedded in the genome are repaired by RNase H2-dependent ribonucleotide excision repair (RER) (4). If ribonucleotides are not efficiently removed from DNA, nucleophilic attack of the 2Ј-oxygen on the ribonucleotide sugar backbone renders DNA chemically unstable, leading to genomic instability. The absence of RER results in S-phase checkpoint activation (5), slow cell growth (6), and deletion mutations in repetitive DNA sequences in yeast (7). RNase H2 defects are embryonically lethal in mice (3). In humans, mutations in the gene coding RNase H2 are associated with Aicardi-Goutières syndrome, which is a neurological disease with symptoms of systemic autoimmunity (8). The aberrant accumulation of ribonucleotides has been observed in fibroblast cells from RNase H2-defective Aicardi-Goutières syndrome patients (9). Furthermore, the accumulated ribonucleotides activate DNA damage signaling (9), suggesting that ribonucleotide incorporation in DNA could be detrimental to genome stability.
Nucleotides are subjected to oxidation by reactive oxygen species (ROS), which are generated by normal aerobic metabolism in cells (10,11). 7,8-Dihydro-8-oxo-2Ј-deoxyguanosine triphosphate (8-oxo-dGTP) is a major oxidized form of dGTP and can be incorporated into the nascent DNA strand during DNA replication and repair (12). 8-Oxo-dG adopts an anticonformation that forms Watson-Crick pairs with cytosine and a syn-conformation that forms Hoogsteen pairs with adenine. This causes G to T transversion mutations after DNA replication. Base excision repair (BER) is the primary * This work was supported by Grant-in-Aid for Young Scientists B (16K16195) and for Scientific Research B (25281022) from the Ministry of Education, Culture, Sports, Science and Technology in Japan and the Intramural Research Program of the National Institutes of Health, NIEHS Grants Z01 ES050158 and Z01 ES050159. The authors declare that they have no conflicts of interest with the contents of this article. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. 1 To whom correspondence may be addressed. Japan. Tel.: 81-3-3700-1141; Fax: 81-3-3700 -2348; E-mail: a-sassa@nihs.go.jp. 2 To whom correspondence may be addressed. Tel.: 81-3-3700-1141; Fax: 81-3-3700-2348; E-mail: m-yasui@nihs.go.jp.
DNA damages that escape from repair can block replication, possibly leading to cell death. To counteract the deleterious effects of DNA damage, cells possess specialized pols that bypass DNA damage during replication. This process is called translesion DNA synthesis (TLS) and contributes to cell survival by incorporating dNMPs opposite the DNA damage in the template DNA. For example, pol incorporates dCMP opposite N 2 -guanine adducts of benzo[a]pyrene diolepoxide (21)(22)(23). pol accurately and efficiently incorporates nucleotides opposite UV-induced cyclobutane pyrimidine dimers (24,25). In addition, pols and are involved in error-prone and errorfree TLS, respectively, of 8-oxo-dG in vivo; pol knockdown reduces G to T transversion mutations caused by 8-oxo-dG in human cells (26), and the absence of pol decreases the accuracy of TLS across 8-oxo-dG in yeast (27).
As with various DNA damages, ribonucleotides in DNA templates retard DNA synthesis in vitro by replicative pols ␦ and ⑀ but not by TLS pol (1,28,29). Thus, TLS pols are important for the damage tolerance of embedded ribonucleotides. However, it is still unclear whether TLS pols other than pol bypass ribonucleotides. In addition, the miscoding properties of damaged ribonucleotides such as 8-oxo-rG during TLS remain unclear. In this study, we analyzed TLS of rG or 8-oxo-rG by human replicative pol ␣ and TLS pols and as undamaged or damaged ribonucleotide in the template DNA. We observed that pol ␣ efficiently bypassed rG but not 8-oxo-rG. The primer extension reactions catalyzed by pol were strongly retarded by rG and 8-oxo-rG. On the other hand, pol rapidly bypassed rG and 8-oxo-rG as efficiently as dG and 8-oxo-dG. pol ␣ exclusively inserted dAMP opposite 8-oxo-rG and 8-oxo-dG. Both TLS pols and preferentially inserted the correct dCMP opposite 8-oxo-rG, whereas pols were more prone to incorporate dAMP opposite 8-oxo-dG. We also examined the repair of 8-oxo-rG in DNA by BER in vitro; OGG1-mediated BER was completely inhibited by the ribonucleotide sugar of 8-oxo-rG. Furthermore, OGG1 interfered with the excision of 8-oxo-rG by RNase H2. Therefore, the ribonucleotide sugar backbone can alter the specificity of TLS and BER activity.
Influence of the 8-Oxoguanine Sugar on the Miscoding Specificities Catalyzed by Translesion DNA Polymerases-To examine base substitutions and deletions during TLS, primer extension reactions were performed in the presence of all four dNTPs using pol and pol . The extended products (ϳ28-to 32-mer) past dG, rG, 8-oxo-rG, and 8-oxo-dG were recovered, digested with EcoRI, and subjected to two-phase polyacrylamide gel electrophoresis as described in Fig. 2A and under "Experimental Procedures." A standard mixture of six Alexa Fluor 546 (Alexa 546)-labeled oligonucleotides containing dC, dA, dG, or dT opposite the modified nucleotide or one-and two-base deletions can be resolved by this method. The percentage of 2Јdeoxynucleotide monophosphate (dNMP) incorporation was normalized to the amount of the starting primer.
Steady-state Kinetic Studies on rG-, 8-Oxo-dG-, and 8-Oxo-rG-modified DNA Templates-To more accurately measure the frequency of dNMP incorporation (F ins ) opposite rG, 8-oxo-dG, or 8-oxo-rG and chain extension (F ext ) from the primer terminus, steady-state kinetic analysis was performed using pols ␣, , and .
We also measured the F ins and F ext for the substrates using pol ( Table 3). The F ins for dCMP incorporation opposite rG (4.90 ϫ 10 Ϫ1 ) was 2-fold lower than that for dG. Because the F ext for dC:rG (1.21) was comparable with that for dC:dG, the F ins ϫ F ext past dC:rG was only 1.7-fold lower than that for dC:dG. The values of F ins for dCMP and dAMP incorporation opposite 8-oxo-dG were similar (2.82 ϫ 10 Ϫ1 and 3.19 ϫ 10 Ϫ1 , respectively). The F ins ϫ F ext past dC:8-oxo-dG and dA:8oxo-dG were comparable because the F ext values were similar between dC:8-oxo-dG and dA:8-oxo-dG (1.37 and 1.53, respectively). Next, the F ins for the incorporation of the correct nucleotide (dCMP) opposite 8-oxo-rG (2.63 ϫ 10 Ϫ1 ) was 2.7-fold higher than that for dAMP (9.87 ϫ 10 Ϫ2 ) and was 4.9-and 67-fold higher than those for dGMP and dTMP, respectively. The F ins ϫ F ext for dC:8-oxo-rG was 2.5-fold higher than that for dA:8-oxo-rG and was 53-and 510-fold higher than that for dG:8-oxo-rG and dT:8-oxo-rG, respectively.
Suppression of RNase H2 Activity by 8-Oxoguanine and OGG1-Our working model suggests that 8-oxo-rG could not be efficiently excised by RNase H2 due to the presence of the abnormal base, i.e. 8-oxoguanine. Furthermore, specific bind-  ing of OGG1 to 8-oxoguanine-containing DNA (34) may influence the RNase-mediated cleavage of 8-oxo-rG, because DNA glycosylases bound to the damaged base could inhibit damage processing by other enzymes (20,35,36). To investigate this model, the activity of RNase H2 was compared in the absence or presence of OGG1 for different substrates including rG or 8-oxo-rG (Fig. 4, A and B). We observed complete cleavage of rG by RNase H2 (Fig. 4A, lanes 2-4), which was not affected by the addition of OGG1 (Fig. 4A, lanes 5-7). With the 8-oxo-rG substrate, the activity of RNase H2 was decreased (Fig. 4A, lanes  9 -11), and 8-oxo-rG excision was further suppressed in the presence of OGG1 (Fig. 4A, lanes 12-14). Quantification of the results indicate that RNase H2 was inhibited ϳ4-fold by OGG1 (Fig. 4B). Furthermore, the binding capacity of OGG1 was analyzed and compared for the substrates with 8-oxo-dG and 8-oxo-rG. The results showed that OGG1 was capable of bind-ing to 8-oxo-rG (Fig. 4C), which could interfere with RNase H2 (Fig. 4A).

Discussion
The concentration of rGTP (ϳ500 M) is 2 orders of magnitude higher than that of dGTP (ϳ5 M) in cells (2), implying that a substantial amount of 8-oxo-rGTP can be produced in the presence of ROS. Because rNTPs (including 8-oxo-rGTP) can be incorporated into DNA during replication, ribonucleotides, if not repaired, are problematic DNA lesions due to its potential to retard DNA replication (1,28). However, the ability of pols ␣, , and to bypass the ribonucleotides is still unclear. Furthermore, little is known about the repair mechanisms of oxidized ribonucleotides such as 8-oxo-rG. In this study, we analyzed the activities and specificities of TLS across rG or 8-oxo-rG catalyzed by pols ␣, , and . In addition, the  influence of the 8-oxo-rG sugar backbone on the activity of BER enzymes was examined.

TABLE 2 Kinetic parameters for nucleotide insertion and chain extension reactions catalyzed by human DNA polymerase
We found that the replicative pol ␣ efficiently bypassed rG but not 8-oxo-rG. Furthermore, pol ␣ exclusively inserted dAMP opposite 8-oxo-rG. These results suggest that 8-oxo-rG in DNA can be highly cytotoxic and pro-mutagenic. This result highlights the importance of TLS pols for the damage tolerance against the ribonucleotide. Regarding TLS pols, primer extension reactions catalyzed by pol were retarded by rG or 8-oxo-rG one base before the ribonucleotide. In contrast, pol easily bypassed rG and 8-oxo-rG as efficiently as dG and 8-oxo-dG, respectively. This was also supported by steady-state kinetic analyses. The different TLS efficiencies between pol and may reflect the evolution of their active sites to bypass specific DNA template lesions. pol , but not pol , constitutes a molecular splint that stabilizes the structure of damaged DNA (37,38). This might enable pol to efficiently bypass the ribonucleotide. Both pols exclusively inserted the correct nucleotide (dCMP) opposite rG, indicating that rG itself does not have a miscoding potential during TLS.
According to our steady-state kinetic analyses, dCMP was preferably inserted opposite 8-oxo-rG when bypassed by pols and ; the ratio of dCMP/dAMP insertion for 8-oxo-rG during TLS was 2.5 in the reactions catalyzed by both pol and pol , whereas the ratio for 8-oxo-dG was 0.095 and 0.79 in the reactions by pol and pol , respectively (Tables 2 and 3). Albeit the structural mechanism of 8-oxoguanine bypass is different between human pol and pol (31,39), the specificities of TLS catalyzed by both TLS pols were influenced by the sugar identity of 8-oxoguanine. Thus, it is plausible that the ribose sugar affects the conformation of the 8-oxoguanine base itself in the active sites of pols and that are more open and flexible than those of replicative pols. At physiological pH, 8-oxoguanine has a carbonyl group at C8 and a proton at N7. Therefore, when 8-oxo-dG is paired with cytosine in the anti-conformation, a steric clash could occur between the C8-oxygen of 8-oxoguanine and O4Ј of the deoxyribose sugar (40,41). Comparing the structure of the normal duplex DNA and the DNA containing a single ribonucleotide, the sugar pucker conformation of an embedded ribonucleotide is locally changed to C3Ј-endo, whereas that of deoxyribonucleotides in the normal DNA is C2Ј-endo (42). Because the sugar pucker conformation also affects the position of the base (42), sugar puckering of the ribose in 8-oxo-rG might change the position of 8-oxoguanine to avoid a potential steric clash between the C8-oxygen of 8-oxoguanine and the O4Ј of the sugar moiety, thereby promoting Watson-Crick pairing with cytosine. Because the active site of pol is adapted to accommodate 8-oxoguanine in the syn-conformation (39), the structural change of 8-oxoguanine in the anti-conformation by the ribose sugar might distort the active  site and result in stronger retardation of TLS compared with rG ( Table 2).
The ribonucleotide sugar backbone completely inhibited the excision of 8-oxo-rG by OGG1 and APE1, suggesting that the oxidized ribonucleotide was not repaired by BER. These find-ings suggest the following two possibilities: (i) OGG1 cannot remove 8-oxoguanine from 8-oxo-rG, or (ii) even if OGG1 removes 8-oxoguanine from 8-oxo-rG, the resulting abasic ribonucleotide is resistant to the lyase and AP-endonuclease activities of OGG1 and APE1, respectively. In the first case, the ribonucleotide backbone inhibits the flipping of 8-oxoguanine into the active site of OGG1, which is an essential step in the glycosylase reaction (43). Alternatively, even if base-flipping occurs, nucleophilic attack by the active site residue at the C1Ј position of ribose may not occur because of the steric hindrance with the 2Ј-OH of the ribose. The second scenario implies that the abasic ribonucleotide in DNA might be chemically and enzymatically resistant. Consistent with this idea, it has previously been demonstrated that an abasic RNA is more chemically stable than an abasic DNA (44). Further studies are needed to clarify the biochemical and biological relevance of abasic ribonucleotides in the genome.
Although RNase H2 was able to cleave 8-oxo-rG, it did so with lower efficiency than rG. In addition, OGG1 significantly suppressed 8-oxo-rG cleavage by RNase H2. Similarly, it has previously been shown that MYH cannot remove rA paired with 8-oxo-dG and interferes with the excision of rA by RNase H2 (20). Taken together, BER components have detrimental effects on ribonucleotide repair by RER when ribonucleotides "mimic" BER substrates. Thus, 8-oxo-rG is a poor substrate relative to undamaged ribonucleotides for RNase H2, supporting the idea that TLS pols are crucial for a damaged ribonucleotide tolerance pathway. The repair of 8-oxo-rG in the genome may rely on other repair pathways. The nucleotide excision repair pathway removes helix-distorting adducts and can play significant roles in the repair of ribonucleotides and oxidized DNA damages (45)(46)(47)(48). Thus, nucleotide excision repair may be involved in the repair of 8-oxo-rG to stabilize the genome.
In summary, our results suggest that 8-oxo-rG has the strong miscoding potential and acts as a replication blocking lesion for pol ␣. In contrast, 8-oxo-rG is bypassed by pols and , which preferentially inserts dCMP opposite 8-oxo-rG. BER cannot remove 8-oxo-rG and suppresses its excision by RNase H2. Based on these findings, we concluded that the sugar of the damaged nucleotide alters the capacity and specificity of TLS and repair in vitro. It would be interesting to examine the mutagenic potential and repair mechanisms of 8-oxo-rG in cells. Further investigation is required to completely understand the cellular impact of damaged ribonucleotides.