Translesion Synthesis by MmLV-, AMV-, and HIV-Reverse Transcriptases Using RNA Templates Containing Inosine, Guanosine, and Their 8-oxo-7,8-Dihydropurine Derivatives

Inosine is ubiquitous and essential in many biological processes, including RNA-editing. In addition, oxidative stress on RNA has been a topic of increasing interest due, in part, to its potential role in the development/progression of disease. In this work we probed the ability of three reverse transcriptases to catalyze the synthesis of cDNA in the presence of RNA templates containing inosine (I), 8-oxo-7,8-dihydroinosine (8oxo-I), guanosine (G), or 8-oxo-7,8-dihydroguanosine (8-oxoG), and explored the impact that these purine derivatives have as a function of position. To this end, we used 29-mers of RNA (as template) containing the modifications at position-18 and reverse transcribed DNA using 17-mers, 18-mers, or 19-mers (as primers). Generally reactivity of the viral RTs, MMLV / AMV / HIV, towards cDNA synthesis was similar for templates containing G or I, as well as for those with 8-oxoG or 8-oxoI. Notable differences are 1) that templates containing I enabled the incorporation of dT when using 17-mers (for exploring incorporation of dNTPs opposite the site of interest); 2) that the use of 18-mers of DNA (to explore cDNA synthesis past the lesion) led to DNA elongation inhibition in the case when a G:dA wobble pair was present, while the presence of I, 8-oxoI, or 8-oxoG led to full synthesis of the corresponding cDNA, with the latter two displaying a more efficient process; 3) that HIV-RT is more sensitive to modified base pairs in the vicinity of cDNA synthesis; and 4) that the presence of a modification two positions away from transcription initiation has an adverse impact on the overall process. Steady-state kinetics were established to determine substrate specificities towards canonical dNTPs (N = G, C, T, A). Overall we found evidence that RNA templates containing inosine are likely to incorporate dC > dT > > dA, where reactivity in the presence of dA was found to be pH dependent (process abolished at pH 7.3); and that the absence of the C2-exocyclic amine, as displayed with templates containing 8-oxoI, leads to increased selectivity towards incorporation of dA over dC. The data will be useful in assessing the impact that the presence of inosine and/or oxidatively generated lesions have on viral processes and adds to previous reports where I codes exclusively like G.


INTRODUCTION
There are two aspects that provided the motivation for this work, one regards to the importance of inosine (naturally occurring modification), and the other to the impact that oxidatively generated lesions (chemically formed from endogenous or exogenous sources) within RNA have on enzymatic processes, reverse transcription (RTn) in this case.
Another aspect of interest was to gain an understanding of how different modifications or lesions behave in various contexts, which increases their potential use as tools to unravel mechanistic aspects of biologically relevant pathways, e.g., the use of 8-bromopurines to explore conformational & H-bonding changes; and the use of 8-oxoinosine or inosine to explore the role of the C2-exocyclic amine in 8-oxoG or G (modifications used in this work).
Inosine (I) is the deamination product of adenosine (A) ( Figure 1) and its formation is catalyzed by deaminases that act on RNA (ADARs). 1, 2 Furthermore, IMP is the first nucleotide that is generated in the de novo purine synthesis pathway and is then enzymatically derivatized to yield the corresponding AMP or GMP (via XMP). 3 Examples that highlight the importance of this modification include that 1) it is commonly observed in a variety of functions that include editing (changes from A to I alter the H-bonding interactions and, as a consequence the coding properties of mRNA), e.g., in the maturation of tRNA; 4, 5 2) it has also been identified in short RNAs such as micro-RNAs, albeit at lower levels than their longer precursors (pri-miRNA); 6 or 3) its presence can cause ribosome stalling. 7 Furthermore, this modification has been associated with disease, along with xanthosine and 8-oxoG, 8 and various human pathologies [e.g., as profiled in the inosinome atlas 9 ]. It is no surprise then of the existence of enzymes that specifically cleave RNA containing inosine, 10 or that remove them from cellular nucleotide pools. 11 It is also important to note that, while the presence of I has been characterized as a marker of viral infection, using respiratory syncytial virus-RSV as model; 12 its incidence on other viral and cellular RNAs has been quantified at low levels on cells infected with Zika virus, Dengue virus, hepatitis C virus (HCV), poliovirus and human immunodeficiency virus type 1. (13) Thus bringing into question the relevance of this modification in distinct viral RNAs. However, we hypothesize that it is plausible that reverse transcriptases encounter inosine within viral or cellular RNA, thus understanding how they cope with its presence within templates of RNA is of importance to understand potential mechanisms, outcomes and/or strategies addressing the synthesis of viral DNA or other factors involving reverse transcription.
On a different note, oxidative damage of RNA is a topic that has increasingly captured attention due to its potential role in the development/progression of disease. [14][15][16][17] Oxidized RNA has been shown to occur in various types of RNA including rRNA, 18 miRNA, 19 or mRNA, 20 and intracellular mechanisms in charge of diminishing the impact of oxidation have been reported. [21][22][23][24] Furthermore, the role of oxidative stress on viral pathogenesis 25 is a factor that could lead to interactions between reverse transcriptases and oxidatively damaged RNA. The trend for the oxidation potential of purine nucleobases (I > A > G) 26,27 makes G the likeliest candidate to undergo transformation under oxidative stress. In fact 8-oxoG is one of the most abundant oxidatively generated lesions, a result of oxidation at the C8-position, with trends in their redox properties matching those of the precursor structures (8-oxoI > 8-oxoA > 8-oxoG). 28 Although this physical property decreases the probability that 8-oxoI or 8-oxoA are formed, rendering them not as biologically relevant as 8-oxoG, the lesions have attracted interest in other contexts. For example in their potential role in a prebiotic scenario, where it was established that none of the oxidized versions of G, A, or I (8-oxoG, 8-oxoA, 8-oxoI) are likely substrates for prebiotic RNA replication. Interestingly, inosine was found to be a possible candidate in this regard. 29 Other examples highlighting their use as tools to probe for biochemical mechanisms include, where 8-oxoinosine and other C8-subsituted purines have been employed to understand the base excision of 8-oxoG by the MutY glycosylases; 30   Therefore, considering the prominent role that inosine has on various biological processes and that oxidation of RNA is ubiquitous, with purines undergoing oxidation to their corresponding 8-oxo-7,8-dihydropurine derivatives, we decided to probe these modifications/lesions within RNA templates using reverse transtription (RTn) as framework.
Briefly, RTn begins by association of a reverse transcriptase (RT) with a nucleic acid substrate, typically composed by a primer and a template in a way that the 3'-end of the primer strand is bound at the priming site; followed by formation of a ternary complex with the incoming dNTP to form a phosphodiester bond; and nucleic acid translocation relative to the RT to continue with processive DNA synthesis. 32 Structural differences among all RTs used in this work are well established and have been reviewed for HIV-RT, 33 AMV-RT, 34 and MMLV-RT. 35 Notably both AMV and MMLV-RT have a sequence homology of 23 % and contain the same five domains needed for RTn, 36 while the HIV-RT has structural differences that lead to distinct fidelity.
Previous work on reverse transcription of RNA containing inosine has been reported using an engineered form of MMLV-RT, Superscript IV, to show that I codes like G in experiments aimed at developing deep sequencing technologies. (37) Furthermore, reverse transcription of RNA containing 8-oxoG in the presence of AMV-or MMLV-RTs showed insertion of A and C opposite this lesion. (38) In this work, we were interested in 1) probing the impact of the exocyclic amine on both G and 8-oxoG, thus we directly compared strands of RNA containing this nucleobases to RNA containing I or 8-oxoI; 2) extending the impact of reverse transcription to HIV-RT; and 3) establishing the reactivity of the reverse transcriptase within the vicinity of the reactive site containing G, I, 8-oxoG, or 8-oxoI (for which three DNA primers of varying length were used).
Interestingly some unexpected reactivity was observed on RNA containing I, where experiments carried out in the presence of dTTP or dATP allowed for the enzyme to facilitate their incorporation into DNA opposite this modification.
Steady-state kinetics showed that the order of selectivity was dC > dT >> dA and that there was a dependence on pH, where decreasing the pH led to a lack of recognition towards dA and decreased recognition towards dT. RNA templates containing 8-oxoG displayed the same reactivity as that previously reported, where both dC or dA are incorporated.
Furthermore, reverse transcription in the presence of RNA templates containing 8-oxoI displayed a higher efficiency towards the incorporation of dA and lower efficiency for dC, with respect to 8-oxoG, which points to the importance of the exocyclic amine in fidelity and efficiency. Furthermore the importance of using reverse transcription in sequencing and the effects of various modifications was highlighted recently, 39 and is therefore important to understand their specificity/reactivity in the presence of various modifications that are found/generated in biological processes.

Experimental Procedures
General. The detailed synthesis for the phosphoramidites of 8-oxoI and 8-BrI is included in the supporting materials (pp. S3-S18). It includes spectroscopic information for all known and novel intermediates. 1 H NMR, 13 C NMR, and 31 P NMR spectra, recorded at 300, 75, and 121.5 MHz respectively (using a standard broadband multinuclear probe on a 300MHz Avance III platform from Bruker); IR spectra were recorded on a diamond ATR sampler using powders of pure materials; high-resolution mass spectrometry was carried out via ESI/APCI; and UV-vis spectroscopy of all small molecules was carried out on a Perkin Elmer λ-650 UV/vis spectrometer using quartz cuvettes (1 cm pathlength). The synthesis corresponding to the phosphoramidite for 8-oxoG was carried out according to a previous report. 40 All experiments described herein were carried out in triplicate, unless otherwise noted.
C18-Sep-Pak cartridges were obtained from Waters and used to desalt the purified oligomers using 5 mM NH 4 OAc as the elution buffer. Oligonucleotides were dissolved in H 2 O and used as obtained for subsequent experiments. Unmodified ONs were purchased from IDT-DNA or ChemGenes and, following quantification via UV-vis, used without further purification.

UV-vis Spectroscopy.
Concentrations of all oligonucleotides (no secondary structure was detected, via CD, for any oligonucleotide used/measured herein) were obtained via UV-vis using a 1 mm path-length with 1 μL volumes (Thermo Scientific Nano Drop Nd-1000 UV-vis spectrometer). Origin 9.1 was used to plot the spectra of monomers and oligonucleotides for comparison. Oligonucleotide radiolabeling. T4 polynucleotide kinase (PNK) and γ-32 P-ATP-5′-triphosphate were obtained from Perkin Elmer. Oligonucleotides were labeled by mixing PNK, PNK buffer, ATP, DNA, and water (final volume = 50 μL) according to manufacturer's procedure followed by incubation at 37° C for 45 min. Radiolabeled materials were passed through a G-25 sephadex column followed by purification via electrophoresis (20 % denaturing PAGE). The bands of interest (slowest) were extruded and eluted over a saline buffer solution (0.1 M NaCl) for 12 h at 37° C. The remaining solution was filtered and concentrated to dryness under reduced pressure followed by precipitation over NaOAc and ethanol. Supernatant was removed and the remaining oligonucleotide was concentrated under reduced pressure and dissolved in water. Activity was assessed using a Beckmann LS 6500 scintillation counter. Electrophoresis was not necessary for DNA strands previously purified via HPLC (purchased from manufacturer).
Electrophoretic mobility shift assays. Radiolabeled oligonucleotides were mixed in buffers under the desired conditions and all samples were heated to 90° C with slow cooling to room temperature before loading. All samples were electrophoresed using 20 % non-denaturing PAGE (10 × 8 cm). Samples were typically mixed in a 1:1 mixture with 75 % glycerol loading buffer. Quantification of radiolabeled oligonucleotides was carried out using a Molecular Dynamics Phosphorimager 840 equipped with ImageQuant Version 5.1 software.
They were obtained from commercial sources and details are delineated in Table 1  diluted to 1x to yield the concentrations described above), and water to a final volume of 20 μL. For example, a cocktail solution was prepared as follows: 5'-32 P*-DNA (1 μL, < 1 pmol), RNA (2 μL, 2pmol), 10x buffer (2 μL), and water (16 μL); and annealed by placing in a heat block at 90 °C followed by slow cooling to room temperature (over app. 1-1.5 h).
RTn. The annealed solution (3 μL) was transferred to a new tube followed by addition of dNTP (or water for control experiments, 1 μL), and RT (2 μL). The mixture was then incubated at 37 °C (or room temperature) for the desired amount of time. Addition of loading buffer (6 M urea) followed, along with mixing and heating to 90 °C for 5-10 min.
The tube(s) are then allowed to cool down to rt and centrifuged before loading onto a 20 % denaturing PAGE (43 × 35 cm). A voltage was applied to the gels until the xylene cyanol dye passed ¾ the length of the gel, followed by exposure using an autoradiography cassette (Amersham Biosciences) overnight. Quantification of radiolabeled oligonucleotides was carried out using a Molecular Dynamics Phosphorimager 840 equipped with ImageQuant Version 5.1 software.

Steady-State Kinetics.
A solution containing the duplex of interest, in the corresponding buffer, was placed on a heat block at 90 C followed by slow cooling to room temperature (over app. 1-1.5 h). A solution containing the dNTP at concentrations that varied with the nature of the RNA template (2 mM -850 nM) was mixed with the anneal solution, followed by addition of the reverse transcriptase. These reaction mixtures were then incubated at 37 C for 5, 10, 15, or 45 minutes depending on the process being measured. The mixtures were quenched with loading buffer (6M urea), followed by heating of the sample at 90 C for 5-10 minutes. The samples, each containing various dilutions of dNTP, were loaded into a 20% denaturing PAGE, and developed as described above. Kinetics (V max , K m , and V max /K m ) were determined using a Hanes-Woolf plot, and assessed using multiple experiments per graph (3-5 gels/plot). Reaction velocities were measured at various substrate concentrations and the values for V max and K m were obtained by plotting [S]/V vs [S].
Different enzyme batches, from same manufacturer, were used throughout.

Single-point Kinetics -Relative Rates at constant [dNTP] and [RT] as a function of time (see supporting info). A solution
containing the duplex of interest in the appropriate buffers was placed on a heat block at 90 °C followed by slow cooling to room temperature (over app. 1-1.5 h). A solution containing the reverse transcriptase was mixed into the reaction tube followed by addition of the dNTP. A determined volume is then withdrawn and added to a tube containing loading buffer (6M urea), followed by heating of the sample to 90 °C for 5-10 min. The last step is repeated at various time intervals followed by loading into a 20 % PAGE, and developed as described before. Importantly for these experiments, the incubations were carried out at room temperature.

RESULTS
Reverse Transcription -cDNA Synthesis. We studied cDNA synthesis via reverse transcription using RNA (29-nt long) templates containing guanosine, inosine, 8-oxoinosine, or 8-oxoguanosine at position-18; which were prepared via solid phase synthesis, purified and characterized prior to their use. The selectivity and efficiency of each RT for enabling the incorporation of canonical dNTPs was explored using DNA primers of three lengths (17-, 18-, or 19-nt long) to explore RTn 1) opposite the modification/lesion; 2) past a modified/canonical base pair generated at the end; or 3) on an RNA:DNA duplex in which a modified base pair is present two positions away from the start of cDNA synthesis. The RTs used were Avian Myeloblastosis Virus (AMV); Moloney Murine Leukemia Virus (M-MLV); human immunodeficiency virus (HIV); and a genetically modified version of M-MLV (Superscript-II). The overall process was explored with canonical dNTPs (dG, dC, dT, dA) individually and as a mixture, to explore transcriptase selectivity and efficiency. In addition, the kinetic parameters of some reactions were derived to establish relative rates. We also compared the results to RNA templates containing 8-bromoinosine at the site of interest, which was used to probe for potential relationships between H-bonding and anti/syn conformational variations. This allowed us to establish potential differences among these lesions/modifications and gain a better understanding of their corresponding incorporation ratios.
Since the only structural aspect differentiating I from G, and 8-oxoI from 8-oxoG, is the presence of an exocyclic amine at position-2, we initially reasoned/expected that these pairs would exhibit similar outcomes. Experiments were carried out at two concentrations of RT while keeping the concentration of the corresponding dNTPs in excess (1.5 -0.5 mM) with respect to the RNA:DNA duplexes (50 -100 nM). In this manner, the effect of the RT was explored under higher RT concentrations that allowed us to establish reactivity and selectivity, and under lower RT concentrations that aided in identifying efficiency and specificity/selectivity (see Table 1).
dNTP Incorporation Opposite the Modification on Duplexes 1:5 -4:5. To begin our studies, templates of RNA (29-nt long) 1-4 were set for hybridization with the corresponding DNA primer, 17-mer (5) (Figure 2A). The sequence of the template was chosen based on a report by Alenko et al., 38 with the exception that the length of the RNA strand was extended by nine nucleotides to explore continuation of cDNA synthesis in more detail. All solutions were prepared in the buffer provided by the manufacturer (see experimental details). Formation of the corresponding duplexes was confirmed in two ways: 1) via CD using phosphate buffered solutions (10 mM sodium phosphate at pH 7.2, 1 mM NaCl, 5 mM MgCl 2 , Figure S11-S13); and 2) via electrophoretic analyses (in buffer provided by manufacturer) where a slower band was observed on native PAGE gels, that can be assigned to hybridization of the corresponding RNA and DNA to their duplexes ( Figure S19). It is important to note that the use of buffers recommended for RTn were not compatible with CD spectroscopy, given that strong absorption was observed at all wavelengths where features of an A-form duplex are expected (see Figure S10 for comparison between duplex and ss-RNA), thus the use of the phosphate buffered mixture mentioned above was employed for all experiments that required CD (including T m analyses). Thermal denaturation transitions corresponding to duplexes 1:5 -4:5 led to values that were equivalent (Figure 2A -70 °C). We initiated experiments using AMV-RT ( Figure 2B) to observe: 1) that the duplex where the template contains an I (2:5) enables the incorporation of dA while the analogous G-containing duplex (2:5) does not catalyze this process under the conditions described herein (lanes 15 and 5); 2) that the I-template exhibits lower incorporation selectivity between dC and dT than that observed on the G-template, which preferentially adds dC ( Figure 2C); 3) that the G-or I-containing templates enabled the incorporation of dT, while the 8-oxopurine analogues do not (Lanes 4, 14,9,19); and 4) that duplexes  Figure 2D, lanes 26, 32, 38, 44). The results indicate that in the case of the canonical template strand (1), dCTP keeps adding up to the +4N-position (opposite one additional A and two Cs) stopping at a site where the enzyme encounters a U (lane 23). This is consistent with previous reports where formation of a C:dT base pair is likely to form, while a U:dT base pair occurs with very low efficiency. 43 The same trend was observed in the case of template strand containing inosine (2), albeit with less efficient addition at +3N, and +4N; confirmed by plotting % conversion as a  To corroborate the dependence of reverse transcription with the base pairing interactions at this position, duplexes where the last nucleotide of the RNA template is base pairing to dC (duplexes 1:7 -4:7) were probed for RTn ( Figure   3C). As expected, formation of a Watson-Crick base pair in duplex 1:7 restored the ability of the RT to carry out cDNA synthesis in the presence of a dNTP mix; where the other three duplexes also displayed cDNA synthesis under these conditions (lanes 26,32,38,44). Steady-state kinetics showed that all duplexes enabled the incorporation of dT with similar efficiencies (Figure 3C, table). In particular, we initially expected dT incorporation to be most efficient upon use of the canonical G:dC-containing duplex (1:7), given that this is the canonical analogue of the reaction. It is unclear at the moment on why the templates containing the oxidative modifications, or I, enable the incorporation of dT in this context However, it is possible that this process is facilitated by additional interactions between the RNA:DNA duplex and the RT-enzyme.
To further explore this unexpected reactivity, RTn experiments were carried out using duplexes 1:8 -4:8, where the site of interest is base pairing to dT ( Figure 3D). In agreement with previous results, the duplex containing an I:dT base pair (2:8) displayed the most efficient incorporation of dTTP, over duplexes containing G:dT (7.5 × slower) and 8-oxoI:dT (37.5 × slower), with inefficient dTTP incorporation on duplex 3:8 (containing 8-oxoG). The fact that dTTP incorporation is inefficient in the presence of an 8-oxoG:dT or 8-oxoI:dT base pair is reflected on experiments carried out in the presence of a mix of all dNTPs, where cDNA synthesis is halted or affected (lanes 50, 56, 62, 68).  (Figure 4, left). To this end, the RNA template containing 8-BrI (9) was prepared and cDNA synthesis was explored in the presence of DNA primers 5-8 using AMV-RT. As indicated within Figure 4 (upper-right insert), we were surprised to find that this chemical modification induces thermal destabilization on duplex 9:5, compared to analogue 2:5 (Figure 2A), suggesting that it is somehow interacting to destabilize this duplex in the absence of a direct base pair opposite 8-BrI. On the other hand, a duplex containing an 8-BrI:dT base pair (9:8) resulted in the more stable species, with duplexes 9:6 and 9:7 displaying equivalent T m values; which suggests that this chemical modification base pairs with similar strengths to dC or dA (at the end of the duplex). RTn using AMV-RT and duplex 9:5 showed that the incorporation of dNTPs opposite 8-BrI followed the trend: C > T ≈ A ≈ G, with inefficient cDNA synthesis in the presence of dNTP mix (lanes 1-6). Steady-state kinetics showed that the incorporation of dCTP opposite 8-BrI is about 5 orders of magnitude less efficient than its analogous process for an RNA template containing I ( Figure 2C). This result suggests that 8-BrI is in its anti-conformation, as the pattern is similar to that observed when using RNA template 2, containing I (Figure 2A), and that the bromine at the C8-position adversely affects cDNA extension. In addition, RTn showed that duplexes formed using templates 7, and 8 led to incorporation of dT as well as cDNA synthesis in the presence of dNTP mix (lanes 16,22,18,24), while hybridization with DNA primer 6 (containing an 8-BrI:dA base pair) did not show a band corresponding to this N+11 product (lane 12).
The results indicate that the 8-BrI:A base pair does not form a stable interaction, which differs from the result obtained using an I:A base pair and suggests that bromine induces an adverse interaction in this case. On the other hand, the observation that duplex 9:7 enables the incorporation of dT and restores cDNA synthesis, suggests that the 8-BrI:C base pair is stable enough to allow for processivity with dNTP addition (in this case dT). Importantly, there is no direct relationship between thermal stability of the RNA:DNA duplex and RTn efficiency in these cases, thus suggesting that enzyme:duplex:dNTP interactions may differ upon cDNA synthesis.  Incorporation of dATP opposite I is pH dependent. The unexpected reverse transcriptase activity on templates containing I, i.e., that RNA template 2:5 facilitates the incorporation of dATP on its DNA primer in the presence of AMV-RT; or that contrary to duplex 1:6, duplex 2:6 enables cDNA synthesis; motivated us to explore this process in more detail. We hypothesized that this reactivity was due to the formation of seemingly stable interactions between I and dA, and that varying the conditions could lead to changes in selectivity. Since all of the buffers provided by the manufacturer are at a pH value of 8.3, we decided to prepare buffered solutions with the same salt concentrations and different pH values of 5.5, 7.3, 8.4 (as control), or 9.5. Control reactions using duplexes 1:5 -4:5 at pH 8.4 displayed results that were equal to those described on Figure 2A; and increasing the pH to 9.5 led to a drastic decrease in selectivity in the case of every RNA template. On the other hand, decreasing the pH closer to physiological conditions of 7.3 resulted in inhibited addition of dATP opposite I, while displaying the same reactivity on templates containing G, 8-oxoG, or 8-oxoI. Experiments carried out at different incubation times, as low as 5 min, displayed the same trend. Thus suggesting that the formation of I:dA interactions may not be biologically relevant, or dependent on H-bonding interactions around the vicinity of this potential Wobble base pair. Furthermore, lowering the pH to 5.5 decreased the activity of the reverse transcriptase in a significant manner. The set of data for these experiments is shown in figure S24.
Reverse Transcription using MMLV-RT. To explore potential differences/similarities of reactivity amongst reverse transcriptases, we probed for differences in fidelity and reactivity using MMLV-RT. As illustrated in Figures S25-S26, the selectivity and reactivity was similar to that described for AMV-RT, with the only measurable difference in the incorporation ratios of dC or dA opposite 8-oxoI or 8-oxoG. However these were only observed at high enzyme concentrations, thus both RTs have overall similar selectivity against the canonical dNTPs. In addition, experiments carried out using DNAs 6 or 7 as primers led to similar results where the use of DNA 6 led to cDNA synthesis inhibition in the case where a G:A base pair is present at the end while the rest of the modifications (I/8-oxoG/8-oxoI) allowed for full cDNA synthesis. As expected, cDNA synthesis is restored upon using DNA 7 with a canonical G:C base pair at this position (full cDNA synthesis is also observed in all other cases). Furthermore there was no difference in the trends of dNTP addition upon using Superscript-II as RT ( Figure S27), which is not surprising given that this RT is obtained from a MMLV source.

Reverse Transcription using HIV-RT.
Taking into consideration that the HIV-RT is a less selective/specific reverse transcriptase, we set out to probe its reactivity towards the modifications in this work. As shown in figure 6, the use of  (lanes 58, 60). Lastly to explore the impact of a modified base pair two positions away from the transcription site, experiments were carried out using duplexes 1:10 -4:10 and 1:11 -4:11. Interestingly, the presence of a modified base pair had a big impact and only efficient cDNA elongation was observed in the cases where a G:dC, I:dC, or G:dA (1:11, 2:11, 1:10) base pair was present. Suggesting that reverse transcription stalls upon encountering an oxidative lesion in this context. Supporting this observation is the fact that duplexes containing 8-BrI (9:10 / 9:11) also resulted in an inefficient process (see Figure S29 for this experiment series).  Reverse transcription using RNA templates containing inosine.
I:dC. The reactivity where RNA templates containing G or I can efficiently incorporate dC was expected, 42 and can be explained via formation of canonical WC-type interactions. The incorporation of dC opposite RNA templates containing G was measured as a more efficient process (app. 1.6 ×) than the corresponding I-containing analogue. An observation that can be explained by the lack of a H-bond interaction in I, compared to that expected from the exocyclic amine in G.
I:dT. Interestingly both of these templates enabled the incorporation of dT, albeit with less efficient rates of app. 930× and 390× for G or I respectively (comparison amongst duplexes 1:5 or 2:5, Figure 2). This is in agreement with results obtained on the duplex family 1:8 -4:8 where the DNA:RNA duplexes containing an I:dT or G:dT base pair enabled cDNA synthesis, with 8-oxoI:dT following in efficiency ( Figure 3D). Where, steady-state kinetics showed that sample 2:8 (containing I:dT) allowed for dTTP incorporation app. 7.5× faster than the canonical analogue 1:8, and 37.5× more efficiently than duplex 4:8 (containing 8-oxoI:dT). Stable base pair interactions have been previously reported on G:U wobble pairs and can be rationalized, in part, from the form0ation of stable G:T / I:T H-bonds that are independent of the exocyclic amine ( Figure 8A); thus presumably facilitating dTTP incorporation and/or cDNA elongation. Another aspect to consider, particularly in cases where the modification is already in place (duplexes 1:8 -4:8) involves the role of electrostatics in processivity, as determined from crystal structures bound to DNA duplexes. 44,45 Taking into consideration that the presence of G:U wobble pairs induce structural differences to form an electronegative environment in the major groove, 46 it is reasonable to expect that this will also have an impact on RT─RNAbinding and/or processivity. While this relationship was initially surprising, G:U and I:U wobble pairs have also shown similar reactivity in other enzymatic contexts, such as in translation elongation. 47 I:dA. Contrary to the reactivity observed with the use of RNA templates containing G at the site of interest, templates containing I also enabled the incorporation of dA in the presence of the RTs used in this work, albeit in a much less efficient manner ( Figure 2B, lane 15). In this regard, G:A base pairs have been reported 48 and base pairing involving I and dA has been characterized, via crystallography, in the context of the ribosomal decoding center. 49 An interaction that may be enhanced due to the lack of the C2-exocyclic amine ( Figure 8B). The same behavior was corroborated in experiments where duplexes 1:6 -4:6 were used, and contrary to 8-oxoG/I/8-oxoI (which allowed cDNA synthesis), duplex 1:6 (containing a G:A base pair at the start site) did not display incorporation of any dNTP efficiently and resulted in cDNA synthesis inhibition ( Figure 3B, lanes 1-6). Furthermore, steady state kinetics on this family of duplexes dis played that the trend in efficiency for reverse transcription to occur when encountering inosine base pairing with a deoxynucleotide at the start site was I:dT > I:dC > I:dA (Figure 3-D, C, B respectively). Although this is an inefficient process we were interested in probing its prevalence as a function of pH, where lowering the pH to 7.3 abolished the addition of dA opposite I. Since it is unlikely that the protonation states of the nucleobases change in this range, the most probable cause for this observation may be due to varying interactions between the enzyme:duplex:dNTP complex.

Reverse transcription using templates containing 8-oxoG or 8-oxoI.
In agreement with previous reports, 38 the RNA template containing 8-oxoG (3) allowed for incorporation of dA with a slight preference over dC (app. 1.5× more efficient). It can be assumed that the expected H-bonding interactions between 8-oxopurines, in its syn-conformation, and A play a role in this behavior ( Figure 8C). This is also in agreement with 8-oxoG exhibiting H-bonding interactions as Uridine, arising from a conformational change around the glycosidic bond, in other enzymatic contexts. 50 Interestingly, the corresponding 8-oxoI analogue also allowed for incorporation of dC and dA, exhibiting a more efficient process in the presence of dATP (app. 20×, Figure 2B). The fact that the RNA template containing I enabled the incorporation of dA more efficiently than dC (4.3× -2:5 compared to 1:5) suggests that the C2exocyclic amine may be playing a significant role within the active site. This pattern was also observed in duplexes that already contained an 8-oxopurine base pairing to dC or dA (1 :6 -4:6, 1:7 -4:7); where the use of DNA templates containing an additional dA did not allow for efficient cDNA synthesis on the canonical duplex 1:6. The results highlight the ability of both 8-oxopurines, as well as inosine, to form catalytically active base pairs in this context. Unexpectedly, the use of duplexes 1:7 -4:7 (containing an additional dC) led to similar rates of dT incorporation, with the trend: 8-oxoG suggests that 8-oxoI may be behaving conformationally, like inosine in its anti-conformation.

Reactivity on templates containing 8-bromoinosine.
To learn more about the potential H-bonding interactions, and understand more on the role of a modification at the C8position, the phosphoramidite of 8-BrI was used to prepare RNA templates with this moiety. Thermal denaturation transitions showed that 8-BrI can form stable base pairs with the trend dT > dA > dC (Figure 4), and RTn experiments carried out on duplex 9:5 led to incorporation of dC preferentially. Indicating that the formation of a stable base pair with the incoming dNTP is not an accurate indicator of what will occur when the modification is already in place (duplexes 9:6 -9:8). Furthermore, it was found that RNA modified with an 8-BrI at the start site facilitates the insertion of dC or dT opposite to it and not dA, suggesting that the I:dA base pair likely forms with I in the anti-conformation. Thus 8-BrI may potentially base pair with both dC and dA as proposed on Figure 8D. The fact that templates containing 8-BrI do not lead to cDNA synthesis, while those containing I do, implies that the bromine may be posing adverse interactions that may prevent protein:RNA contacts that affect transcription efficiency. Pointing out to the importance of having these interactions intact and that the conformation and H-bonding between base pairs is not sufficient to dictate enzymatic reactivity, since other aspects and changes in structure must be taken in consideration. 52

Reactivity using HIV-RT.
The HIV-RT was the only RT that enabled incorporation of dA opposite an RNA template containing G but less efficiently than the corresponding I-analogue (duplexes 1:5 -4:5). Base pairing involving I:A can be rationalized from the formation of anti-I:anti-A or anti-I:syn-A ( Figure 8B), with the latter previously reported, 53 thus making this a more likely candidate. In addition the thermal stability trend for base pairing for oligonucleotides of DNA containing dI is I:C > I:A . I:T ≈ I:G, 54 also supporting the formation of a relatively stable base pair. Another difference was that templates containing 8-oxoI had no selectivity between the incorporation of dA or dC, contrary to experiments in the presence of AMV-RT, where duplex 4:5 exhibited more efficient incorporation of dATP. It is important to note that kinetic data is still needed to confirm these trends. On the other hand, there was no selectivity towards insertion of dA or dC in the presence of templates containing 8-oxoG (3:5). This result is in contrast with previous reports that used DNA duplexes containing 8-oxoG, where HIV-RT incorporated dCTP more efficiently than dATP. 55 This may be due to differences in substrate (RNA vs DNA) as well as experimental conditions. Another difference can be observed upon comparison of duplexes containing 8-oxoI and 8-oxoG, where experiments using the former led to incorporation of dT in higher conversion ratios. Which confirms the importance, and impact, that the exocyclic amine may have in this and other enzymatic contexts. Overall, the use of duplex 2:5 (containing inosine) suggests that this modification can code like G, A, and U to a lesser extent and that this process is pH dependent. The use of DNA 18-mers 1:6 -4:6 and 1:7 -4:7 shows that the HIV-RT is effective at continuing cDNA synthesis if there is a canonical base pair at the 3'-end of the DNA, while both AMV and MMLV are less affected by the presence of a modification at this position. The same behavior was also observed when an oxidative lesion is two positions away from the start of RTn.
Overall, the results obtained using 8-oxoG and AMV-or MMLV-RT are in good agreement with a previous report, 38 where A and C added efficiently opposite 8-oxoG compared to addition of C and T opposite G. In a separate report, C and T were reported to extend DNA synthesis opposite 8-oxoG using RAV2-RT and exclusively C using HIV-RT. 56 This report contrasts with the results obtained herein, possibly due to the use of different experimental conditions.

CONCLUSION
The reactivity described provides important information on the behavior of oxidatively generated lesions, or inosine, within RNA and their role on reverse transcription. A reliable method for the synthesis for the phosphoramidites of 8-oxoinosine and 8-bromoinosine along with their subsequent incorporation into RNA is reported. Importantly, the findings described herein shed light into this process in cases where 1) the process starts at the site of modification; 2) the modification/lesion is already involved in a base pair interaction; or 3) the modification/lesion is two bases away from the start of RTn. It is important to highlight that when analyzing some of the processes reported herein, e.g., I:dT or I:dA interactions, one must take into consideration the kinetics data to assess on the likelihood that they may be relevant in vivo (or within the active site under varying local H-bonding networks). Furthermore, the dependence on pH that was observed suggests that while I seems to be decoded as A or U (in addition to the more commonly accepted G), 7 this process may not be relevant in vivo. The use of inosine and 8-oxoinosine pointed to the impact that the exocyclic amine has on reverse transcription recognition and efficiency. Since a link between the presence of inosine, and/or oxidative stress, on viral pathogenesis exists, the information provided herein will be useful. It is plausible that reverse transcriptases encounter the lesions/modification explored in this work, thus a good understanding on how RNA templates containing them will be useful in assessing their impact. Funding from NIGMS (1R15GM132816) is also acknowledged.