Impact of 1,N6-ethenoadenosine, a damaged ribonucleotide in DNA, on translesion synthesis and repair

Incorporation of ribonucleotides into DNA can severely diminish genome integrity. However, how ribonucleotides instigate DNA damage is poorly understood. In DNA, they can promote replication stress and genomic instability and have been implicated in several diseases. We report here the impact of the ribonucleotide rATP and of its naturally occurring damaged analog 1,N6-ethenoadenosine (1,N6-ϵrA) on translesion synthesis (TLS), mediated by human DNA polymerase η (hpol η), and on RNase H2–mediated incision. Mass spectral analysis revealed that 1,N6-ϵrA in DNA generates extensive frameshifts during TLS, which can lead to genomic instability. Moreover, steady-state kinetic analysis of the TLS process indicated that deoxypurines (i.e. dATP and dGTP) are inserted predominantly opposite 1,N6-ϵrA. We also show that hpol η acts as a reverse transcriptase in the presence of damaged ribonucleotide 1,N6-ϵrA but has poor RNA primer extension activities. Steady-state kinetic analysis of reverse transcription and RNA primer extension showed that hpol η favors the addition of dATP and dGTP opposite 1,N6-ϵrA. We also found that RNase H2 recognizes 1,N6-ϵrA but has limited incision activity across from this lesion, which can lead to the persistence of this detrimental DNA adduct. We conclude that the damaged and unrepaired ribonucleotide 1,N6-ϵrA in DNA exhibits mutagenic potential and can also alter the reading frame in an mRNA transcript because 1,N6-ϵrA is incompletely incised by RNase H2.

High-fidelity DNA polymerase-mediated replication is a central tenet of biology. It is important to ensure that the genome is copied completely and accurately during replication without any incorrect insertions to avoid further consequences. On the basis of several studies done during the past decade, ribonucleotide incorporation in DNA is considered a major threat to genome stability (1,2). Mitochondrial (pol ␥) 2 as well as nuclear DNA polymerases (pols ␣, ␤, and ⑀) (3,4) incorporate ribonucleotides into the genome because the cellular concentrations of ribonucleoside triphosphates (rNTPs) are much higher than the dNTPs (5)(6)(7). Discrimination between deoxyribo-and ribosugars varies among DNA polymerases, mainly due to the active site "steric gate" residue that plays an important role in limiting rNTP insertions (8). Other sources of ribonucleotide incorporation include HIV-1 reverse transcriptase (RT)-mediated reverse transcription and incomplete removal of Okazaki fragment primers (1,9). Embedded ribonucleotides in the genome present an enigma in the DNA replication process due to the presence of a 2Ј-OH group, which renders the backbone susceptible to cleavage and various sugar puckered conformations (10 -12). Cells repair such ribonucleotides (in DNA) primarily by the ribonucleotide excision repair (RER) process (13)(14)(15)) and a mechanism that involves RNase H2 enzyme-mediated recognition and incision of ribonucleotides (15)(16)(17). The loss of RNase H2 activity is linked to Aicardi-Goutieres syndrome in humans (18). If RNase H2 is unable to remove these ribonucleotides (19), they can have both favorable (20,21) and lethal biological consequences (22). Recent investigations have shown that ribonucleotides are deleterious lesions that can lead to mutations, large deletions, replication stress, strand breaks (topoisomerase-1-mediated repair), chromosomal rearrangements, loss of hereditary information, and disruption of transcription processes (1,(23)(24)(25)(26).
In cells, reactive oxygen species and lipid peroxidation can cause damage of the nucleotide pool (damaged rNTPs/dNTPs) (7,27) and can be a threat to the synthesis of genetic material (28). Thus, the coding potential of genetic material can be compromised if damaged rNTPs/dNTPs are incorporated into the genome during replication. Oxidation of a deoxyribo-to a ribosugar can also occur (29). In the cellular nucleotide pool, free dGTP is more vulnerable to oxidation than the G present in DNA and leads primarily to the formation of 8-oxo-dGTP (30), which can be inserted into the nascent DNA (31)(32)(33)(34)(35). Among the rNTPs, the cellular concentration of rATP is very high, and it may therefore be inserted into DNA (33). In this connection, Clausen and co-workers also (36) reported the levels of rATP in mitochondrial DNA. The possibility of insertion of rATP and damaged rATPs into the DNA during replication cannot be excluded (33). The biological consequences of rATP and damaged rATPs in DNA remain elusive. Our own work has shown that low-fidelity human DNA polymerase (hpol ) incorporates rATP opposite a cyclobutane pyrimidine dimer as well as 8-oxo-dG lesions (37,38).
DNA and RNA lesions are ubiquitous to the genome. If DNA damage is left unrepaired, cells utilize low-fidelity DNA polymerases to continue replication across damaged sites (to rescue the stalled replication forks) for cell survival, a process termed translesion synthesis (TLS) (39,40). The TLS process has garnered considerable attention due to the tolerance of damage in DNA in an error-free or error-prone manner (39,40). TLS also plays an important role in tolerance of ribonucleotides in DNA, and the main key player is hpol (especially in dealing with lagging strands derived from Okazaki fragments) (16,24,38,(41)(42)(43)(44)(45)(46). Studies with 8-oxo-rG in DNA indicate that hpol is required for the tolerance of this adduct (24). The versatile hpol is now also known for its reverse transcriptase and transcription-associated roles (38,(47)(48)(49)(50). Damage in RNA has not been considered to disturb the cellular processes, possibly due to general lack of replication of RNA and its innately lower stability as well as the more rapid turnover (51)(52)(53)(54). However, RNA is more vulnerable to oxidative damage due to its abundance and single-stranded nature (55,56), and this damage can be detrimental for the cell (57). RNA damage is also associated with the pathogenesis of some neurodegenerative diseases (58 -60). RNA is not exempt from damaging events from reactive oxygen species and lipid peroxidation.
2-Chloroethylene oxide, an oxidation product of the known human liver carcinogen vinyl chloride (VC), reacts with nucleophilic sites in DNA and RNA and leads to the formation of etheno (⑀) lesions (61). Exposure of animals to VC or vinyl bromide (VB) resulted in chemical modification of RNA, and a common product was 1,N 6 -ethenoadenosine (1,N 6 -⑀rA) (62). Several studies have shown that 1,N 6 -⑀rA can be formed in the presence of metabolites of VC or other compounds that generate reactive epoxides (63)(64)(65). Bolt and co-workers (66, 67) mea-sured in vivo alkylation of RNA by VC or VB and its effect on protein synthesis; they reported the synthesis of two unusual proteins from mRNA of VC-exposed animals as compared with control rats (67). Exposure of rats to VB resulted in the synthesis of one unusual protein (67). Schulman and Pelka (68) reported that a total of five adenosine structures were altered after reaction of 2-chloroacetaldehyde (a rearrangement product of the VC metabolite 2-chlorooxirane, also known to generate ⑀ adducts) with Escherichia coli fMet tRNA and also suggested that adenosine residues were modified more rapidly than the cytosine in tRNA (69). Biernat et al. (70) also reported 1,N 6 -⑀rA formation in tRNA after treatment with 2-chloroacetaldehyde. Transcription errors were also observed in the presence of 1,N 6 -⑀rA (71). ⑀-Lesions can also be generated indirectly from oxidative stress through lipid peroxidation (72). Leumann and co-workers (73) studied various damaged bases in RNA in the context of translation, but there is only limited knowledge of the biochemistry of damage in stable RNA. It is important to understand the possible consequences of 1,N 6 -⑀rA in RNA, especially for RT and translation processes.
We now report TLS, RT, and RNA primer extension events and incision of the most abundant nucleotide, rATP, and its naturally occurring damaged analog 1,N 6 -⑀rA. We utilized hpol for the TLS, RT, and RNA primer extension studies due to its recently discovered novel functions (37,38,44,47,49,74,75) and RNase H2 for the incision studies. Our laboratory has recently shown that hpol can act as a reverse transcriptase in human cells (38) and also that it utilizes RNA as a primer in in vitro replication assays, apart from its role in TLS (74). We examined the roles of hpol as a TLS polymerase and reverse transcriptase and in RNA primer extension events in the presence of a damaged ribonucleotide in DNA as well as in RNA under physiological conditions (Fig. 1).

hpol tolerates the damaged ribonucleotide 1,N 6 -⑀rA
To gain detailed insights into the biological consequences of 1,N 6 -⑀rA and rATP, multiple modified oligonucleotides were used for the TLS, RT, and RNA primer extension studies (Table  S1). A primary concern was feasible side reactions related to the presence of the ribo-backbone in the DNA and RNA. Recent studies have shown that strand cleavage, as well as transesterification reactions (isomerization from 3Ј-5Ј to 2Ј-5Ј), at ribonucleotide sites are drastically suppressed in nucleosome core particles (76). In contradiction, it is also known that a ribobackbone in DNA or RNA is temperature-, pH-, and metal ion-sensitive and can lead to transesterification as well as strand cleavage (77). Accordingly, we excluded the possibility of side reactions during our experiments by maintaining physiological conditions, as well as a desired metal ion concentration (Mg 2ϩ ), low temperature, and pH.
To investigate TLS across a ribonucleotide embedded in DNA, full-length extension and single nucleotide incorporation reactions were carried out using physiological concentra-tions of dNTPs as well as rNTPs (78,79) and hpol ( Fig. 2 and Fig. S1 (A and B); see "Experimental procedures" for details). Full-length extension reactions were carried out using higher concentrations of hpol , whereas single-nucleotide incorporation reactions were carried out using lower concentrations of hpol . With an unmodified DNA template (DNA-dA), hpol rapidly extended the DNA primer using dNTPs as well as rNTPs (Fig. S1A, lanes 1-8). When an rA-embedded DNA template was used (DNA-rA), hpol fully extended the DNA primer using dNTPs ( Fig. 2A, lanes 1-4). The DNA primer was also elongated to the full-length product in the presence of rNTPs, but with less processivity ( Fig. 2A, lanes 5-8). These results show that generation of a ribo-ribo pair at an rA site in a DNA template is an unusual behavior of hpol . Changing the coding base from dA to rA had only a minor effect on rNTP incorporation. TLS activities were examined in the presence of the modified ribonucleotide 1,N 6 -⑀rA, and hpol was less efficient in bypassing 1,N 6 -⑀rA in a DNA template (Fig. 2C, lanes 1-4) as compared with an rA template. However, extension reactions were strongly blocked when rNTPs were used (   Translesion synthesis of 1,N 6 -ethenoadenosine 2C, lanes 5-8). As expected, the TLS process was attenuated in the presence of 1,N 6 -⑀rA ( Fig. 2 (A and C), compare lane 1).
In the single-nucleotide incorporation assays, hpol preferentially added dTTP opposite A in DNA-dA and DNA-rA templates ( Fig. S1B and Fig. 2B (lanes 1-3)). With the DNA-rA template, the incorporation preference was dTTP Ͼ dATP Ͼ dGTP Ͼ dCTP Ͼ rUTP Ͼ rATP Ͼ rGTP (Fig. 2B, lanes 1-24). These results showed that hpol can generate an rA-rU pair during TLS by inserting rUTP opposite rA in DNA. In the case of a 1,N 6 -⑀rA-modified DNA template, hpol preferably added dATP and dGTP, as compared with dTTP and dCTP (Fig. 2D, lanes 1-24). Additionally, no rNTP incorporation was observed opposite 1,N 6 -⑀rA in DNA (with a lower concentration of hpol ), indicating that hpol prefers to add deoxypurines instead of ribopurines opposite 1,N 6 -⑀rA.
Our laboratory has previously shown that hpol can act as a reverse transcriptase in human cell extracts (38), and here we examined this novel function of hpol using modified RNA as a template, employing physiological concentrations of dNTPs and rNTPs (Fig. 3) under in vitro conditions. With an unmodified RNA template, hpol elongated a DNA primer in the presence of a mixture of dNTPs (Fig. 3A, lanes 1-4), but only one or two rNTPs were added even after 3 h (Fig. 3A, lanes [5][6][7][8]. In the case of a 1,N 6 -⑀rA-modified RNA template, the DNA primer was elongated with a mixture of dNTPs with low processivity (Fig. 3C, lanes 1-4), whereas extension reactions were strongly blocked in the assays with rNTPs (Fig. 3C, lanes [5][6][7][8]. These results indicate that hpol acts as a specific reverse transcriptase in the presence of 1,N 6 -⑀rA and prefers dNTPs under physiological conditions. As a reverse transcriptase, hpol preferentially added dTTP opposite an unmodified RNA template (Fig. 3B, lanes 1-3), and the incorporation preference was dTTP Ͼ dATP Ͼ dGTP Ͼ dCTP Ͼ rUTP (Fig. 3B, lanes 1-24). These results also suggest that hpol can generate rA-rU pairs during reverse transcription. In the case of a 1,N 6 -⑀rA-modified template, hpol preferably added dATP and dGTP (Fig. 3D, lanes 1-24), following similar base selectivity as in the TLS process. Overall, hpol  Table S1 for the oligonucleotide sequences used. P, FAM-labeled DNA primer.

Translesion synthesis of 1,N 6 -ethenoadenosine
catalyzes reverse transcription with preference for incorporation of dNTPs over rNTPs. We previously reported that hpol can extend an RNA primer (74), and we next examined the ability of hpol to extend an RNA primer in the presence of an unmodified as well as a modified ribonucleotide in DNA using physiological concentrations of dNTPs and rNTPs. With the unmodified DNA template (DNA-dA), hpol readily extended the RNA primer using dNTPs, but the process was slow with rNTPs ( Fig. S1C, lanes [1][2][3][4][5][6][7][8]. Surprisingly, fully extended as well as stalled products were observed for a DNA-rA template (Fig. 4A, lanes 1-8).

Miscoding potential of the damaged ribonucleotide 1,N 6 -⑀rA
To further measure the efficiency and fidelity of hpol during TLS, RT, and RNA primer extension processes, steady-state kinetic analysis (80, 81) was done ( Table 1, Figs. 5 and 6, and Figs. S2-S16; see "Experimental procedures" for details). In the TLS assays, steady-state kinetic analysis showed that incorporation of dTTP opposite 1,N 6 -⑀rA was highly unfavorable. The specificity constant (k sp ; i.e. k cat /K m ) for dTTP insertion (0.001 M Ϫ1 min Ϫ1 , Table 1) was dramatically reduced, 4200-fold compared with DNA-dA (4.2 M Ϫ1 min Ϫ1 ; Table 1) and 1600fold compared with DNA-rA templates (1.7 M Ϫ1 min Ϫ1 , Table   Table S1 for the oligonucleotide sequences used. P, FAM-labeled RNA primer. Translesion synthesis of 1,N 6 -ethenoadenosine 1 and Fig. S2), respectively. For insertion of dATP opposite 1,N 6 -⑀rA, the specificity constant decreased 4.5-fold compared with DNA-dA and 6.5-fold compared with a DNA-rA template ( Table 1 and Fig. 5). The insertion of dGTP opposite 1,N 6 -⑀rA was also decreased 5-fold compared with a DNA-dA and 8-fold compared with a DNA-rA template (Table 1 and Fig. S3). In the case of dCTP insertion opposite to 1,N 6 -⑀rA, the specificity constant was decreased 50-fold compared with a DNA-dA and 80-fold compared with a DNA-rA template (Table 1 and Fig.  S4). As discussed earlier, rNTP insertion was only observed for the DNA-dA and DNA-rA templates. Surprisingly, the specificity constant for rUTP insertion opposite DNA-rA was only 1.5-fold lower compared with the DNA-dA template (Table 1 and Fig. S5). For rATP insertion, the same specificity constants were observed for both templates (Table 1 and Fig. S6), and rGTP incorporation was observed only with the DNA-rA template (Table 1 and Fig. S7).
In comparing the specificity constants for insertion of individual dNTPs opposite 1,N 6 -⑀rA, hpol incorporated dATP 200-fold and dGTP 40-fold more efficiently than dTTP and dCTP, a result driven by the K m values ( Table 1). Incorporation of dCTP showed the same low-specificity constant as dTTP (0.001 M Ϫ1 min Ϫ1 ; Table 1). We conclude that hpol catalyzes translesion synthesis across the damaged ribonucleotide 1,N 6 -⑀rA in a highly error-prone manner, with preference for insertion of deoxypurines, which can affect the reading frame of mRNA. This observation is in line with the previous bypass studies with the 1,N 6 -⑀dA adduct (82).
We next focused on the efficiency of hpol as a reverse transcriptase (Table 1, Fig. 6, and Figs. S8 -S11). Steady-state kinetic analysis revealed that the incorporation of dTTP across 1,N 6 -⑀rA was quite unfavorable, as seen in the TLS process. The specificity constant for insertion of dTTP (0.002 M Ϫ1 min Ϫ1 ; Table 1 and Fig. S8) decreased 2100-fold compared with the RNA-rA template (4.3 M Ϫ1 min Ϫ1 ; Table 1), driven by the much larger K m value for 1,N 6 -⑀rA template (Table 1) as compared with the RNA-rA template. For insertion of dATP opposite 1,N 6 -⑀rA, the specificity constant was 10-fold higher than with the RNA-rA template ( Table 1 and Fig. 6), again driven by the K m difference (Table 1). Thus, hpol prefers to insert dATP opposite 1,N 6 -⑀rA in an RNA template, similar to HIV-1 RT (83). In the case of dGTP insertion across 1,N 6 -⑀rA, the specificity constant was 2.5-fold higher compared with the RNA-rA template (Table 1 and Fig. S9). For dCTP insertion, the specificity constant was 1.25-fold lower compared with the RNA-rA template (Table 1 and Fig. S10). rUTP incorporation was only observed opposite the RNA-rA template (Table 1 and Fig. S11).
With 1,N 6 -⑀rA in the RNA template, hpol incorporated dATP 200-fold, dGTP 25-fold, and dCTP 4-fold more efficiently than dTTP (Table 1), indicating that hpol catalyzes reverse transcription across the damaged ribonucleotide 1,N 6 -⑀rA by the preferential addition of dNTPs instead of rNTPs. With the use of an RT strategy, the position of 1,N 6 -⑀rA in the RNA could be mapped concerning distinctive mutations in the DNA (84).
The efficiency of hpol in RNA primer extension was examined (Table 1 and Figs. S12-S16). Steady-state kinetic analysis showed that insertion of dTTP opposite 1,N 6 -⑀rA was highly unfavorable, as observed for both TLS and RT processes ( Table  1). The specificity constant (0.0004 M Ϫ1 min Ϫ1 ) was 1100-fold lower compared with DNA-dA (0.45 M Ϫ1 min Ϫ1 ) and 1200fold lower than for a DNA-rA template (0.50 M Ϫ1 min Ϫ1 ; Table 1 and Fig. S12). For dATP insertion opposite 1,N 6 -⑀rA, the specificity constant was 10-fold lower compared with a DNA-dA template and 15-fold lower compared with a DNA-rA template (Table 1 and Fig. S13). These results indicate a preference for dATP insertion opposite to the DNA-rA template (1.5fold higher specificity constant) compared with a DNA-dA template. For dGTP insertion opposite 1,N 6 -⑀rA, the specificity constant was 1.7-fold lower compared with a DNA-dA template and 20-fold lower compared with a DNA-rA template ( Table 1 and Fig. S14). These results indicate that dGTP is preferentially added opposite a DNA-rA template compared with DNA-dA, as judged by an 11-fold higher specificity constant. The dCTP incorporation was observed only with the DNA-dA and DNA-rA templates, not with 1,N 6 -⑀rA. The dCTP was preferentially added opposite the DNA-rA (7.5-fold higher specificity constant; Table 1 and Fig. S15). Among the four rNTPs, the insertion of rUTP was observed for DNA-dA and DNA-rA templates but not for the 1,N 6 -⑀rA template. The specificity constant for rUTP insertion opposite DNA-rA was Table 1 Steady-state kinetic analysis of hpol -catalyzed translesion synthesis, reverse transcription, and RNA primer extension ⑀rA, 1, N 6 -ethenoadenosine; rA, adenosine; dA, 2Ј-deoxyadenosine. The complete data set is presented in Figs. 5 and 6 and Figs. S2-S16.
The specificity constants for insertion of three dNTPs across 1,N 6 -⑀rA were compared. Interestingly, hpol incorporated dGTP 100-fold and dATP 25-fold more efficiently compared with dTTP ( Table 1), suggesting that hpol extends an RNA primer opposite the damaged ribonucleotide 1,N 6 -⑀rA-containing DNA template by the preferential addition of dGTP.

hpol generates frameshifts opposite the damaged ribonucleotide 1,N 6 -⑀rA
Steady-state kinetic analysis of hpol -mediated TLS showed a preference for dATP and dGTP insertion opposite 1,N 6 -⑀rA. Other possible miscoding events during the lesion bypass were also considered, and semiquantitative LC-ESI-MS/MS analysis was performed using a previously developed protocol (85,86) for TLS processing (see "Experimental procedures" and Table S1 for the oligonucleotide sequences used) because both the RT and RNA primer extension processes were slower. A 2Ј-deoxyuridine (dU)-containing primer and physiological concentrations of dNTPs were used for the full-length extension reactions (see "Experimental procedures" for details). The fully extended products were treated sequentially with uracil-DNA glycosylase (UDG) and piperidine to obtain short pieces of extended products (Fig. S17) (85). Sequences of fulllength extended products and their relative yields were determined by LC-ESI-MS/MS analysis and are summarized in Table 2 (see Figs. S18 -S35 and Tables S2-S12 for fragmentation patterns).
As expected, hpol yielded error-free products for the DNA/ DNA-dA complex, with blunt end additions of G and A, respectively (m/z 938.3 (Ϫ3) and 933.0 (Ϫ3); Figs. S18 -S20 and Tables S2 and S3). hpol is well-known to catalyze the blunt  Table  1 for k cat and K m values (estimated using fits to a hyperbolic equation in Prism (GraphPad)); see "Experimental procedures" and Table S1 for the oligonucleotide sequences. P, FAM-labeled DNA primer.

Translesion synthesis of 1,N 6 -ethenoadenosine
end addition of A and G to oligonucleotides (86 -88). For the DNA/DNA-rA complex, hpol yielded ϳ89% extended products having T incorporated opposite rA (Table 2) Tables S4 and S5 for fragmentation patterns). Apart from the insertion of T, incorporation of G was also observed across rA, accounting for 11% of the products (m/z 946.7 (Ϫ3); Table 2 Fig. S24, and Table S6).
hpol -mediated bypass of the DNA/DNA-1,N 6 -⑀rA complex was analyzed (Figs. S25-S35 and Tables S7-S12), yielding multiple extended products of the same m/z (837.0 (Ϫ3) and 831.9 (Ϫ3)) and eluted together (Figs. S26, S30, S31, and S34). Mass spectral analysis revealed that frameshift products (one base deletion) contributed 44% of the total, with blunt end addi-tions of G and A, respectively (Table 2, m/z 837.0 (Ϫ3) and 831.9 (Ϫ3); Figs. S27 and S32 and Tables S7 and S10). Extended products with frameshift (one base deletion) followed by misinsertion of G were also observed (5%), with a blunt end addition of A (Table 2, m/z 837.0 (Ϫ3); Fig. S28 and Table S8). Overall, the frameshift products accounted for 49% of the total products derived using physiological concentrations of dNTPs. Apart from frameshift products, small amounts of products with insertion of G (12%) and A (8%) opposite 1,N 6 -⑀rA were also observed in the MS/MS analysis (Table 2, m/z 837.0 (Ϫ3) and 831.9 (Ϫ3); Figs. S29 and S33 and Tables S9 and S11). Interestingly, we did not detect insertion of T opposite 1,N 6 -⑀rA, indicating that hpol is not reading this lesion as a deoxyadenosine residue. Surprisingly, LC-ESI-MS/MS analysis also revealed an unexpected extended product containing a G opposite 1,N 6 -⑀rA and the subsequent misinsertion of another G (insertion of two Gs; Table 2, m/z 842.0 (Ϫ3); Fig. S35 and Table  S12). We included this product in a different category (Table 2) of base substitutions. Such base substitutions have been observed for 8-oxo-rG in human cells (24). LC-ESI-MS/MS provides insight regarding the frameshift products, and the steady-state kinetic and LC-ESI-MS/MS analyses both showed the insertion of A and G opposite 1,N 6 -⑀rA during TLS.

Human RNase H2-mediated recognition but partial incision of the damaged ribonucleotide 1,N 6 -⑀rA
RNase H2-mediated RER plays an important role in removing rNTPs in DNA (89), and we analyzed the endoribonuclease activity of RNase H2 enzyme opposite rA and 1,N 6 -⑀rA in the DNA template. On the basis of our steady-state kinetic and  Table 1 for k cat and K m values (estimated using fits to a hyperbolic equation in Prism (GraphPad)); see "Experimental procedures" and Table S1 for the oligonucleotide sequences. P, FAM-labeled DNA primer. Table 2 LC-ESI-MS/MS sequencing: Summary of products of extension of template-primer complexes by hpol X, 1, N 6 -ethenoadenosine (1,N 6 -⑀rA), adenosine (rA), or 2Ј-deoxyadenosine (dA). LC-ESI-MS/MS data used to summarize these products are presented in Figs. S18 -S35 and Tables S2-S12. ⌬ 1 , a one-base deletion. T, DNA template; P, FAM-labeled dU-containing DNA primer.

Discussion
The DNA replication process is continuously challenged by various obstacles, such as non-B-form structures of DNA (e.g. G-quartets), fragile sites, transcription factors, and numerous DNA lesions (92). In addition to this, rNTP insertion during DNA replication is a major concern because more than one million rNTPs are incorporated per cell cycle (in human cells) due to their cellular abundance (93). rNTPs and dNTPs are the main precursors of RNAs and DNAs and can be damaged endogenously in the nucleotide pool (e.g. formation 8-oxo-dGTP). Recent studies have shown that 8-oxo-rG, an oxidized form of rGTP, is incorporated into DNAs and that nucleotide excision repair and RER mechanisms are not able to remove  Table S1 for the DNA sequences used (FAM-labeled templates and biotin-labeled complementary strands (CS) were used).

Translesion synthesis of 1,N 6 -ethenoadenosine
this adduct efficiently (24). 1,N 6 -⑀rA, a well-known oxidized adduct derived from rATP, is generated from VC (and its reactive metabolites) as well as from lipid peroxidation. 1,N 6 -⑀rA is known to affect protein translation, as discussed earlier. In the present study, we investigated the impact of rATP and the damaged ribonucleotide 1,N 6 -⑀rA on TLS, reverse transcription, RNA primer extension, and ribonucleotide incision processes catalyzed by hpol and human RNase H2 employing steadystate kinetics, MS, and incision assays. Interestingly, the TLS process was strongly affected due to the presence of 1,N 6 -⑀rA in DNA as compared with rA (Fig. 2). It has previously been reported that hpol bypasses the major groove DNA adduct 1,N 6 -⑀dA poorly (82,94). Steady-state kinetic analysis revealed the error-prone bypass of 1,N 6 -⑀rA with the preferable addition of deoxypurines (Table 1). These outcomes are congruent with the results of our previous studies on hpolcatalyzed TLS past a 1,N 6 -⑀dA DNA lesion (i.e. more efficient insertion of dATP (3.7-fold) and dGTP (2.5-fold) than dTTP opposite 1,N 6 -⑀dA) (82). Among the dNTPs, the dGTP insertion opposite DNA-dA and DNA-rA templates suggests that hpol preferably adds dGTP opposite DNA-rA, with a specificity constant 1.6-fold higher than for the DNA-dA template (Table 1). In comparing the specificity constants for DNA-dA and DNA-rA templates, the insertion of dATP was 1.4-fold more efficient for DNA-rA compared with the DNA-dA template. dCTP insertion opposite a DNA-rA template yielded a higher specificity constant (1.6-fold) than for the DNA-dA template, with the basis being in the K m values (Table 1).
When we compare 1,N 6 -⑀dA and 1,N 6 -⑀rA lesions, the impact of ribo-backbone on the TLS process can be seen. These studies led to a new direction in understanding whether slow TLS across 1,N 6 -⑀rA is due to the ribo-backbone or the damaged ribonucleotide. Recent investigations with 8-oxo-rG in DNA indicate that the ribonucleotide backbone has a strong influence on biological processes (24,35).
Further, the specificity constants for insertion of dNTP versus rNTP opposite the DNA-dA and DNA-rA templates were compared. Interestingly, the incorporation of dTTP was 700fold more efficient compared with rUTP for the DNA-dA template, and incorporation of dTTP was 410-fold more efficient compared with rUTP for the DNA-rA template ( Table 1). The specificity constants for dATP and rATP insertion opposite the DNA-dA and DNA-rA templates indicated that incorporation of dATP was favored 900-fold compared with rATP for the DNA-dA template, and incorporation of dATP was 1200-fold more efficient than rATP for the DNA-rA template. It is important to note that the sugar selectivity was well-maintained during the TLS as reflected in the kinetic parameters (Table 1).
LC-ESI-MS/MS analysis showed that the presence of 1,N 6 -⑀rA in DNA produces extensive frameshift (one-base deletion) products (Table 2). On the basis of earlier results, we hypothesize that TLS may occur in two different ways. First, the incoming A or G can be stacked across from 1,N 6 -⑀rA in cross-strand fashion and result in frameshifts (one base deletion), with hpol skipping the lesion (82,95). This process could contribute to the misinsertion opposite the next base T, positioned 5Ј to the lesion on the template strand. These results are consonant with our previous LC-ESI-MS/MS analysis of a 1,N 6 -⑀dA DNA lesion, which yielded 37% frameshifts (82). Structural studies revealed that incoming dATP and dGTP molecules are present in a staggered configuration across the anti-oriented 1,N 6 -⑀dA lesion (82). Subsequent analyses of these studies for both of the lesions indicates that insertion of A and G was reduced drastically across 1,N 6 -⑀rA as compared with 1,N 6 -⑀dA. We assume that the ribo-backbone of 1,N 6 -⑀rA may play an important role in this case. The addition of two Gs by hpol , possibly due to the tight binding of hpol with dGTP during the extension process, leads to the apparent misincorporation of G opposite the T positioned 5Ј to the lesion (base substitution) or due to the misalignment of the template strand (95,96). Overall, we have observed extensive frameshifts (one deletion) followed by misinsertion of G as well as another unusual product that indicates the insertion of two Gs ( Table 2). The preference for insertion of purines over pyrimidines is presumably due to betterstacking interactions to achieve a proper orientation of the lesion and incoming nucleotide in the active site of hpol for further replication. Previous TLS studies with 1,N 6 -⑀dA and an abasic site also showed that hpol predominantly follows the purine rule (82,97). The in vitro ribonucleotide incision assays indicate that recognition of unmodified rA and 1,N 6 -⑀rA by human RNase H2 is completely different, apparently due to the presence of an etheno group on the adenosine (Fig. 7). It is important to note that ribonucleotide excision repair involves RNase H2-mediated incision of a 5Ј-ribonucleotide followed by strand displacement, flap cleavage by the flap endonuclease FEN1, and ligation (13)(14)(15). Our results indicate that E. coli RNase HII (91) recognizes and incises 1,N 6 -⑀rA slowly. Overall, RNase H2-mediated incomplete incision of 1,N 6 -⑀rA from DNA poses a threat to genome stability. Alternative repair pathways such as base excision repair may have potential roles in removing this adduct from DNA employing the alkyladenine DNA glycosylase AAG. If this adduct persists in the DNA, the versatile human TLS pol may tolerate this adduct but in an error-prone way. These studies provide new insight into the consequences of 1,N 6 -⑀rA when TLS is the only means of coping with damaged ribonucleotides in DNA (outlined in Fig. 8).
Damage to DNA has been studied extensively, as compared with RNA. RNA is an innately reactive biomolecule, reacting with both endogenous and exogenous genotoxic agents to yield a plethora of lesions and changes in structural integrity. Leumann and co-workers (98) studied various damaged bases in RNA (8-oxo-rA, 8-oxo-rG, ribo-abasic site, 5-hydroxycytidine, and uridine) in the context of protein translation, as discussed earlier. The ⑀-adducts in RNA were also detected in various tissues of mice after exposure to [ 14 C]VC as discussed earlier (63). In liver, RNA accumulated 1,N 6 -⑀rA, and in vivo levels of ⑀-adducts in RNA were higher than in DNA (63). 1,N 6 -⑀rA was also detected in mouse liver RNA after treatment with ethyl carbamate (64).
To this end, we explored the 1,N 6 -⑀rA modification to understand the RT employing hpol . As a result, hpol was able to synthesize DNA utilizing 1,N 6 -⑀rA-modified RNA as a template with the preferred addition of dNTPs (Fig. 3). These studies suggested that hpol may utilize RNA as a template for double-strand break repair (99, 100). However, hpol showed Translesion synthesis of 1,N 6 -ethenoadenosine reduced reverse transcriptase activities as compared with HIV-1 RT. The preferable addition of deoxypurines across 1,N 6 -⑀rA by hpol (Table 1) was observed. Previous RT studies showed a similar trend with HIV-1 RT, which preferentially added dATP and dGTP across 1,N 6 -⑀rA (83). Overall, 1,N 6 -⑀rA was bypassed by hpol , and yielded full-length DNA molecules in this study may be due to the lack of a more effective polar filter, as proposed by Nair and co-workers (75).
RNA primer extension activities indicate that hpol may play a role in the extension of RNA primer across the templates containing unmodified ribonucleotides using only dNTPs (Fig.  4). The incorporation pattern showed that the hpol prefers to add deoxypurines (preferably dGTP) across 1,N 6 -⑀rA. This result differs from the TLS and RT processes but the basis is unclear for the G preference over A with the RNA primer.
RNase H2 activity toward an oligonucleotide duplex was strongly attenuated when rA (opposite the DNA strand) was replaced by 1,N 6 -⑀rA (Fig. S36), as seen in both the steady-state and pre-steady-state kinetics. The steady-state kinetic rate was reduced 6-fold. RNase H2 shows burst kinetics, indicative of a rate-limiting step following cleavage (38,101). The amplitude of the burst (Fig. S36) was reduced from 38 to 1.8 nM (i.e. 20-fold). Although the calculated pre-steady-state burst rate was still 78 s Ϫ1 , the error in this measurement is high due to the very low amplitude. We previously reported that the presence of a cyclobutane dimer (T^T) opposite the DNA strand only reduced the rate of RNase H2 cleavage ϳ2-fold relative to two Ts (T-T) (38). The effect of the presence of 1,N 6 -⑀rA was much more dramatic than a cyclobutane dimer. There is precedent for the presence of a DNA adduct to change the burst amplitude of an enzyme, specifically a DNA polymerase (102,103). One kinetic explanation, which has been supported over the years, is that the presence of an adduct facilitates the equilibrium of the enzyme with nonproductive forms (103,104). This may well be the case here, although other kinetic possibilities could be involved and cannot be excluded without more extensive experiments. The main point is that the adduct 1,N 6 -⑀rA is highly disruptive of RNase H2 catalysis, as judged by both steady-state and pre-steady-state kinetics (Fig. S36).
In summary, our findings indicate the unique roles of hpol in dealing with an oxidatively damaged ribonucleotide. hpol tolerates the damaged ribonucleotide 1,N 6 -⑀rA during the TLS process with lower efficiency and also generates extensive frameshifts. RNase H2-mediated partial incision of 1,N 6 -⑀rA in DNA was supported by the observation of a (low) burst amplitude. hpol acts as specific reverse transcriptase in the pres-ence of damaged ribonucleotide 1,N 6 -⑀rA in RNA, showing preference for dNTPs. In addition, poor RNA primer extension activities were observed by hpol .

Reagents
The catalytic core of human DNA pol (amino acids 1-432) was expressed in E. coli and purified as reported previously (105,106). Human RNase H2 was expressed in E. coli and purified as reported previously (107,108). Unlabeled dNTPs, rNTPs, UDG, and E. coli RNase HII were purchased from New England Biolabs (Ipswich, MA). C18 Sep-Pak columns were purchased from Waters Corp. (Milford, MA). Micro Biospin-6 columns were purchased from Bio-Rad. Piperidine was from Sigma-Aldrich. Unmodified oligonucleotides and FAM-labeled DNA and RNA primers (twice HPLC-purified) were purchased from Integrated DNA Technologies (Coralville, IA). 1,N 6 -⑀rA phosphoramidite and FAM-labeled modified oligonucleotides were purchased from ChemGenes Corp. (Wilmington, MA). The tert-butyldimethylsilyl-protected ribo-phosphoramidites were purchased from Glen Research (Sterling, VA).

Oligonucleotide synthesis, purification, and characterization
The oligonucleotide sequences were designed on the basis of our previous in vitro and cell-based studies (38,74). Solid-phase synthesis of the 1,N 6 -⑀rA-modified DNAs and RNAs was done on a PerSeptive Biosystems model 8909 DNA synthesizer (see Table S1 for the in-house oligonucleotide sequences). tert-Butyldimethylsilyl-protected ribo-phosphoramidites were used, and the modified ribo-and deoxyribooligonucleotides were synthesized on a 1-mol scale using the appropriate controlled pore glass as solid support. 5-Ethylthio-1H-tetrazole was used as an activator, and the coupling time for the modified phosphoramidites was 10 min. The deprotection of modified ribo-and deoxyribooligonucleotides was carried out using a reported procedure (83). The crude modified ribo-and deoxyribooligonucleotides were purified by denaturing PAGE (20% PAGE, 7 M urea) at 55 W for 2.5 h using 1ϫ TBE buffer (89 mM Tris-HCl buffer containing boric acid (89 mM) and 2 mM EDTA, pH 8.0). The gel was visualized under a UV lamp at 260 nm, and the desired DNA/RNA bands were isolated from the gel and extracted using TEN buffer (10 mM Tris-HCl buffer containing 1 mM EDTA and 300 mM NaCl, pH 8.0) overnight at room temperature. The modified ribo-and deoxyribooligonucleotides were desalted using C18 Sep-Pak columns, and their

Physiological concentrations of dNTPs and rNTPs (78, 79)
All polymerase reactions were carried out using physiological concentrations of dNTPs/rNTPs. Physiological concentrations of dNTPs used in these studies were 25 M dATP, 30 M dCTP, 90 M dGTP, and 40 M dTTP. Physiological concentrations of rNTPs used in these studies were 2 mM rATP, 0.25 mM rCTP, 0.5 mM rGTP, and 0.5 mM rUTP. The desired stocks of dNTPs/rNTPs were prepared in RNase-free water. Fig. S1 (A and C)).

Full-length extension reactions for translesion synthesis, reverse transcription, and RNA primer extension studies (Figs. 2 (A and C), 3 (A and C), and 4 (A and C) and
A primer-template complex containing a FAM-labeled DNA or RNA primer and an unmodified or modified template were annealed (1:1 molar ratio) at 75°C for 5 min followed by slow cooling overnight. Full-length extension reactions were carried out using 40 mM Tris-HCl buffer (pH 7.5) containing 100 mM KCl, 5 mM MgCl 2 , 5% glycerol (v/v), 10 mM DTT, and 0.1 mg/ml BSA at 37°C. The final concentration of primer-template complex was 0.5 M, and 200 nM hpol was used to obtain fully extended primers. Reactions were initiated by adding a 1-l mixture of dNTPs/rNTPs (physiological concentrations) to a total volume of 10 l. Aliquots (1.5 l) of reaction mixtures were taken at different time points (0, 5, 30, 60, and 180 min) and quenched with 8.5 l of 10 mM EDTA (pH 8.0) in 95% deionized formamide (v/v). Products were separated using 20% PAGE (7 M urea), and results were visualized using a Typhoon scanner (GE Healthcare) and analyzed by ImageJ software (80).

Single nucleotide incorporation reactions (Figs. 2 (B and D), 3 (B and D), and 4 (B and D) and Fig. S1 (B and D))
A primer-template DNA complex containing a FAM-labeled DNA or RNA primer and an unmodified or modified template were annealed (1:1 molar ratio) at 75°C for 5 min, followed by slow cooling overnight. All single-nucleotide insertion reactions were performed using 40 mM Tris-HCl buffer (pH 7.5) containing 100 mM KCl, 5 mM MgCl 2 , 5% glycerol (v/v), 10 mM DTT, and 0.1 mg/ml BSA at 37°C. The final concentration of primer-template complex was 0.5 M, and 50 nM hpol was used, followed by the addition of 1 l of an individual dNTP/ rNTP (physiological concentrations) to a total volume of 10 l. Aliquots (1.5 l) of reaction mixtures were taken at each time point (0, 5, 30, and 60 min) and quenched with 8.5 l of 10 mM EDTA (pH 8.0) in 95% deionized formamide (v/v). Products were separated using 20% PAGE (7 M urea), and results were visualized using a Typhoon scanner (GE Healthcare) and analyzed by ImageJ software. (Table 1, Figs. 5 and 6, and Figs. S2-S16) A primer-template DNA complex containing a FAM-labeled DNA or RNA primer and an unmodified or modified template were annealed (1:1 molar ratio) at 75°C for 5 min, followed by slow cooling overnight. All steady-state kinetic reactions were performed using 40 mM Tris-HCl buffer (pH 7.5) containing 100 mM KCl, 5 mM MgCl 2 , 5% glycerol (v/v), 10 mM DTT, and 0.1 mg/ml BSA at 37°C for 5 min. The final concentration of primer-template complex was 0.5 M, and hpol was added, ranging from 0.25 to 38 nM to achieve steady-state kinetics of dNTP/rNTP insertion. Reactions were initiated by adding 1 l of an individual dNTP/rNTP stock solution, at each of 10 different concentrations, to a total volume of 5 l. A general rule of Յ20% (P ϩ 1) product formation was maintained (80). Aliquots (1.5 l) of reaction mixtures were taken at 5-min time points and quenched with 8.5 l of 10 mM EDTA (pH 8.0) in 95% deionized formamide (v/v). Products were separated using 20% PAGE (7 M urea), and results were visualized using a Typhoon scanner (GE Healthcare) and analyzed by ImageJ software. Data points are shown as means Ϯ S.D. from two independent experiments (see Table 1 for k cat and K m values) and estimated using fit to a hyperbolic equation in Prism software (GraphPad, San Diego, CA) to obtain k cat /K m (k sp ) and k cat directly (81).

LC-ESI-MS/MS analysis: Full-length extension followed by UDG and piperidine treatment (Fig. S17)
A 2Ј-deoxyuridine (dU)-containing FAM-labeled 18-mer primer and an unmodified or modified template were annealed (1:1 molar ratio) at 75°C for 5 min followed by slow cooling overnight. The full-length extension reactions were carried out under the same reaction conditions as described for the fulllength extension assays, except that the final concentrations were as follows: primer-template complex was 2.5 M; hpol was 1.2 M for DNA/DNA-dA and DNA/DNA-rA complexes and 2 M for DNA/DNA-1,N 6 -⑀rA complex, in a total reaction volume of 85 l. Reactions were carried out in the presence of mixture of dNTPs (physiological concentrations) at 37°C for 4 h for DNA/DNA-dA and DNA/DNA-rA complexes and overnight for DNA/DNA-1,N 6 -⑀rA complex. Reactions were terminated using Micro Biospin-6 column separations to extract Mg 2ϩ and dNTPs. The resulting products were treated with 25 units of UDG at 37°C for 4 h, followed by 0.25 M piperidine at 95°C for 1 h (85). The reaction mixture was evaporated to dryness by lyophilization. The dried pellet was resuspended in 500 l of nuclease-free water and taken to dryness by lyophilization. Finally, this dried pellet was resuspended in 25 l of nucleasefree water for LC-ESI-MS/MS analysis.

LC-ESI-MS/MS analysis (Figs. S18 -S35 and Tables S2-S12)
LC-ESI-MS/MS analyses were performed on a Finnigan LTQ mass spectrometer (Thermo Scientific Corp., San Jose, CA) connected to an Acquity ultraperformance LC (UPLC) system (Waters). The data were acquired in the ESI negative ion mode using an C18 column (Acquity UPLC BEH, 1.7 m, 2.1 ϫ 100 mm). UPLC conditions were as follows. Buffer A contained 10 mM NH 4 CH 3 CO 2 , 2% CH 3 CN, and 98% H 2 O (v/v), pH 7.0, and buffer B contained 10 mM NH 4 CH 3 CO 2 , 95% CH 3 CN, 5% H 2 O (v/v), pH 7.0. The column temperature was 50°C. The following gradient program was used with a flow rate of 0.3 ml/min: 0 -30% B over 8 min, 30 -90% B over 1 min, held at 90% B for 1 min, 90 -0% B over 1 min, held at 0% B for 4 min (all v/v). A 15-l sample aliquot was injected with an autosampler system. ESI settings were as follows: source voltage 4.5 kV, sweep gas Translesion synthesis of 1,N 6 -ethenoadenosine flow rate 5, auxiliary gas flow rate 30, capillary voltage Ϫ48 V, sheath gas flow 60, capillary temperature 270°C, and tube lens voltage Ϫ143 V. MS/MS conditions were as follows: activation Q 0.25, activation time 30 ms. Product spectra were acquired over the range m/z 300 -2000. The most abundant species were fragmented in the ion trap with a normalized collision energy of 35% (data-dependent mode). The extended product sequences were identified (Figs. S18 -S35 and Tables S2-S12) by comparing the observed CID fragments with the theoretical values using the Mongo Oligo Mass Calculator version 2.08 (State University of New York at Albany). The relative yields of extended product sequences were calculated based on the peak areas of extracted ion chromatograms. For the co-eluted products, relative peak areas were calculated on the basis of the intensity of representative fragments (as shown in Fig. S30 for a 6 -B(Ϫ2) and Fig. S34 for W 3 (Ϫ1) fragments). The fragmentation patterns of the observed products were confirmed by comparing with the fragmentation patterns of commercial standards (IDT, Coralville, IA). (Fig. 7 and Fig. S37) The complementary sequences (CS_1 to CS_3; Table S1) were annealed with FAM-labeled unmodified and modified template (1:1 molar ratio) at 75°C for 5 min followed by slow cooling overnight (see Table S1 for the oligonucleotide sequences). The final concentration of dsDNA was 0.5 M. The assays were done at 37°C in the incision buffer: 40 mM Tris-HCl buffer (pH 7.5) containing 50 mM NaCl, 10 mM MgCl 2 , 1 mM DTT, 5% glycerol (v/v), and 0.1 mg/ml BSA. The reactions were initiated by adding 10 nM RNase H2. Aliquots (1.5 l) of reaction mixtures were taken at different time points (0, 5, 15, 30, 45, and 60 min) and quenched with 8.5 l of slow-moving dye (10 mM EDTA (pH 8.0) in 95% deionized formamide (v/v) and xylene cyanol FF, only slow moving dye). Products were separated using 20% PAGE (7 M urea), and the results were visualized using a Typhoon scanner (GE Healthcare) and analyzed by ImageJ software. The E. coli RNase HII-mediated incision assays were carried out in 1ϫ ThermoPol reaction buffer (New England Biolabs) using 1 unit of enzyme at 37°C. The same time points were taken as described above.

RNase H2-mediated incision assays
Pre-steady-state kinetics (Fig. S36) The complementary sequence (CS_1 ; Table S1) was annealed with a FAM-labeled unmodified as well as a modified ribonucleotide-containing DNA template (1:1 molar ratio) at 75°C for 5 min, followed by slow cooling overnight (see Table  S1 for oligonucleotide sequences). The final concentration of dsDNA was 0.25 M, and concentration of RNase H2 was 203 nM (for DNA-rA) and 263 nM (for DNA-1,N 6 -⑀rA), respectively. Pre-steady-state kinetic experiments were carried out at 37°C in the incision buffer (40 mM Tris-HCl buffer (pH 7.5) containing 50 mM NaCl, 10 mM MgCl 2 , 1 mM DTT, 5% glycerol (v/v), and 0.1 mg/ml BSA. The dsDNA and RNase H2 were mixed in an RQF-3 rapid quench-flow instrument (KinTek Corp., Snow Shoe, PA), allowing the reaction to proceed for time ranges from 0.005 to 2 s. Aliquots (6 l) of each reaction mixture were taken with 24 l of slow-moving dye (0.5 mM EDTA (pH 8.0) in 90% deionized formamide (v/v) and xylene cyanol FF (0.01%, w/v)). Part of each reaction mixture (7 l) was loaded on the 20% PAGE (7 M urea), and the results were visualized using a Typhoon scanner (GE Healthcare) and analyzed by ImageJ software, and the data were fit to a burst equation, y ϭ A(1 Ϫ e Ϫkpt ) ϩ k ss E 0 t, using GraphPad Prism software (GraphPad), where A denotes the apparent concentration of the active form of the enzyme, k p denotes the burst rate, k ss represents the steady-state rate, E 0 is the total enzyme concentration, and t is the time. The initial time points (0 -0.08 s) were analyzed by non-linear regression of an exponential decay equation fits in Prism software and presented in Fig. S36 (inset).

Supporting information
The supporting information contains a list of oligonucleotide sequences used for the studies; PAGE analyses of full-length extension and single-nucleotide incorporation using physiological concentrations of dNTPs and rNTPs; PAGE analyses of steady-state kinetic analysis and the respective Prism graphs; PAGE analyses of full-length extension assays for LC-ESI-MS/MS analysis using physiological concentrations of dNTPs; LC-ESI-MS/MS sequencing analysis of full-length extension reactions for DNA/DNA-dA, DNA/DNA-rA, and DNA/DNA-1,N 6 -⑀rA complexes; respective extracted ion chromatograms and CID fragmentation spectrums; Prism graphs of pre-steadystate kinetic analysis; PAGE of E. coli RNase HII-mediated incision assays; and the MALDI spectrum for DNA-1,N 6 -⑀rA and RNA-1,N 6 -⑀rA.

Data availability
All data are included in the article and supporting information.