Effect of Ribonucleotides Embedded in a DNA Template on HIV-1 Reverse Transcription Kinetics and Fidelity*

Background: Under limiting dNTP concentrations, HIV-1 RT incorporates rNTPs during DNA synthesis. Results: HIV-1 RT utilizes dNTP less efficiently around rNMPs, and mismatch extension fidelity is significantly reduced. Conclusion: Presence of an rNMP in DNA template slows HIV-1 RT-mediated DNA synthesis and reduces fidelity. Significance: This study provides insight into how rNMP incorporation during proviral DNA synthesis can affect HIV-1 replication kinetics and fidelity. HIV-1 reverse transcriptase (RT) frequently incorporates ribonucleoside triphosphates (rNTPs) during proviral DNA synthesis, particularly under the limited dNTP conditions found in macrophages. We investigated the mechanistic impacts of an rNMP embedded in DNA templates on HIV-1 RT-mediated DNA synthesis. We observed that the template-embedded rNMP induced pausing of RT and delayed DNA synthesis kinetics at low macrophage dNTP concentrations but not at high T cell dNTP concentrations. Although the binding affinity of RT to the rNMP-containing template-primer was not altered, the dNTP incorporation kinetics of RT were significantly reduced at one nucleotide upstream and downstream of the rNMP site, leading to pause sites. Finally, HIV-1 RT becomes more error-prone at rNMP sites with an elevated mismatch extension capability but not enhanced misinsertion capability. Together these data suggest that rNMPs embedded in DNA templates may influence reverse transcription kinetics and impact viral mutagenesis in macrophages.

HIV-1 reverse transcriptase (RT) frequently incorporates ribonucleoside triphosphates (rNTPs) during proviral DNA synthesis, particularly under the limited dNTP conditions found in macrophages. We investigated the mechanistic impacts of an rNMP embedded in DNA templates on HIV-1 RT-mediated DNA synthesis. We observed that the templateembedded rNMP induced pausing of RT and delayed DNA synthesis kinetics at low macrophage dNTP concentrations but not at high T cell dNTP concentrations. Although the binding affinity of RT to the rNMP-containing template-primer was not altered, the dNTP incorporation kinetics of RT were significantly reduced at one nucleotide upstream and downstream of the rNMP site, leading to pause sites. Finally, HIV-1 RT becomes more error-prone at rNMP sites with an elevated mismatch extension capability but not enhanced misinsertion capability. Together these data suggest that rNMPs embedded in DNA templates may influence reverse transcription kinetics and impact viral mutagenesis in macrophages.
Human immunodeficiency virus type 1 (HIV-1) uniquely infects both activated CD4 ϩ T cells and terminally differentiated/ non-dividing macrophages (1,2). These cells differ in the concentrations of their intracellular deoxyribonucleoside triphosphates (dNTPs). Activated CD4 ϩ T cells, which are dividing cells, contain high levels of dNTPs (1-16 M) as compared with terminally differentiated macrophages (20 -40 nM) (3,4). Recently, we and others reported that the low dNTP pool in macrophages is partially due to active hydrolysis of cellular dNTPs by SAMHD1, a cellular dNTP triphosphohydrolase (5,6). Although the dNTP pools differ considerably, both cell types have similarly high concentrations of ribonucleoside triphos-phates (rNTPs) 2 (4). In addition, it has been shown that this disparity exists in yeast (7). Because macrophages contain much lower dNTP levels, the disparity between dNTPs and rNTPs is far greater in these cells as compared with activated CD4 ϩ T cells. Given that the only difference between a dNTP and an rNTP is the absence of a 2Ј OH on the dNTP, it is important that DNA polymerases possess a mechanism to minimize rNTP incorporation into genomic DNA during synthesis. Indeed, most DNA polymerases, including HIV-1 RT, have evolved to discriminate against rNTPs using a bulky residue located in their active site as a steric gate (8,9). HIV-1 RT uses a tyrosine residue at position 115 (Tyr-115) as steric gate, and mutations of this residue have been shown to promote greater rNTP incorporation (10). Despite this discriminatory mechanism, it has been shown that cellular DNA polymerases do incorporate rNTPs during DNA synthesis (7,11,12). We have previously demonstrated that HIV-1 RT frequently incorporates rNTPs during DNA synthesis under macrophage dNTP/ rNTP conditions at ratio of 1 rNTP every 146 bases (4,13). HIV-1 RT synthesizes the first strand (negative) proviral DNA from the viral RNA genome and the second strand (positive) DNA from the newly synthesized first strand DNA. Thus, the incorporation of rNTPs during first strand synthesis generates an rNMP-containing DNA template for second strand DNA synthesis. However, until now the mechanistic and kinetics impacts of chimeric DNA templates containing rNMPs on HIV-1 RT have not been explored.
Structural studies with DNA duplexes have shown that rNMPs embedded in DNA induce a global structural change, shifting it from a B-form helix to an A-form helix, which is more typical of dsRNA (14 -16). However, another study suggested a more localized structural change around an rNMP embedded in a DNA duplex (17). More recent studies by McElhinny et al. (7,18) on cellular DNA polymerases have shown that an rNMP present in a DNA template causes pausing of a yeast replicative DNA polymerase and is mutagenic if not repaired. This is in agreement with other studies that implicated pause sites as mutation hot spots for DNA polymerases, including HIV-1 RT (19,20). Furthermore, it has been shown that rNMPs in dsDNA are targeted by RNase H2-initiated repair mechanism and could be mutagenic if not removed (21). However, we have shown that RNase H2-mediated repair for rNMPs embedded in DNA is significantly delayed in macrophages as compared with dividing cells (13).
Collectively, our previous findings indicate that HIV-1 RT frequently incorporates rNTPs particularly in macrophages. Thus, in this study we biochemically tested whether rNTPs incorporated during first strand proviral DNA synthesis affect polymerization kinetics and enzyme fidelity of HIV-1 RT during second strand DNA synthesis. This study provides invaluable insights as to how rNMP incorporation during proviral DNA synthesis, which is mechanistically promoted by extremely limited canonical dNTP levels in macrophages, affects HIV-1 RT kinetics and enzyme fidelity.

EXPERIMENTAL PROCEDURES
Protein Expression-SIV agm Sab-1 RT gene was previously cloned and purified (22), and avian myeloblastosis virus (AMV) protein was obtained from New England Biolabs. Hexahistidine-tagged HXB2 HIV-1 RT gene (23) was introduced into pET28a (Novagen) and overexpressed in BL21 Escherichia coli (Novagen). The RT protein was purified using Ni 2ϩ chelating chromatography as described previously (24,25). The concentration and purity of the protein was analyzed by 10% SDSpolyacrylamide gel using 1.5 g of bovine serum albumin (Sigma) as a control.
Primer Extension Assay-An HIV-1, SIV agm , and AMV RT primer extension assay was performed as previously described but with minor modifications (4). Briefly, a 17-mer primer was 5Ј end 32 P-labeled and annealed to 48-mer rNMP-containing or rNMP-free DNA templates in the presence of 100 mM NaCl, 10 mM Tris-HCl (pH 8.0), and 1 mM EDTA. Reactions with a final volume of 20 l contained equal amounts of RT, 10 nM template/primer (T/P), and macrophage or T cell dNTP concentrations in a 1ϫ reaction buffer (12.5 mM Tris-HCl (pH 7.5), 12.5 mM NaCl, and 2.5 mM MgCl 2 ). The reactions were incubated at 37°C for 5, 10, 20, 40, 80, or 120 min then quenched with 40 mM EDTA. The products were resolved on a 14% urea-PAGE gels under denaturing conditions and visualized by Personal Molecular Imager (Bio-Rad).
Circular Dichroism (CD)-T/P were diluted and annealed in a buffer containing 25 mM Tris HCl (pH 7.8) and 100 mM NaCl 2 . A 0.1-mm path length quartz cuvette (Starna) was used for all readings. Wavelength scans were recorded from 320 to 220 nm (1-nm increment, 1-nm bandwidth, 10-s averaging time) at 25°C. Data from three scans were averaged, and background from buffer was subtracted.
RNase H-mediated Cleavage Analysis-To examine RT RNase H-mediated cleavage of rNMP-containing templates during primer extension assays, a 48-mer template with or without rAMP was 5Ј end 32 P-labeled. The templates were then annealed to a 17-mer primer and incubated at 37°C with HIV-1 RT in reaction buffer described above for 1 h. As a positive control, the labeled templates were incubated at 37°C with 0.3 M KOH without RT for 1 h. The products were resolved on a 14% urea-PAGE gels under denaturing conditions and visualized by Personal Molecular Imager.
Surface Plasmon Resonance-To analyze HIV-1 RT T/P interaction, we utilized a surface plasmon resonance technology that was previously described (26,27). In this study we used a Biacore T200 (Biacore Inc., Piscataway, NJ) and a Series S streptavidin-coated (SA) sensor chip (GE Healthcare). 3Ј-Endbiotinylated 48-mer templates with or without an rAMP at position 23 relative to 3Ј end were synthesized by Integrated DNA Technologies. To analyze HIV-1 RT interaction with T/P when an rAMP is at the ϩ1, or Ϫ1 position in the active site, a 21-or 23-mer primer, respectively, was annealed to the biotinylated template (1:2 template to primer ratio). The chip was preconditioned with buffer (1 M NaCl, 50 mM NaOH), and the biotinylated T/P diluted in running buffer (100 mM NaCl, 10 mM HEPES 7.5, 0.005% Tween 20) was immobilized to obtain a signal of ϳ30 response units. The kinetic experiment was performed using 120-s injections of four different concentrations (2.5, 5, 10, and 20 nM) of HIV-1 RT at 75 l/min for the association phase and 300 s monitoring for dissociation phase. The surface was regenerated using 0.1% SDS injected for 120 s at 50 l/min. All experiments were done at 25°C. The data obtained were analyzed using double referencing (28) and were fit to a 1:1 binding model to obtain the k a , k d , and K D values using the Biacore T200 evaluation software (GE Healthcare).
Processivity Assay-A trap assay was used to examine the processivity of HIV-1 RT as previously described (29). Briefly, a 17-mer primer was 5Ј end 32 P-radiolabeled and annealed to 48-mer rNMP-containing or rNMP-free template (2:1 primer to template ratio). The trap was prepared by annealing poly-rA to oligo-dT at a 1:2 ratio. RT was incubated with the annealed template (8 nM)/primer (16 nM) complex for 3 min at 37°C in RT reaction buffer. The reaction was initiated with the addition of MgCl 2 (6 mM), 200-fold excess trap, and macrophage or T cell dNTP concentrations. After incubation of reactions at 37°C for 10, 20, or 60 min, the reactions were terminated with EDTA. For a trap control, RT was added to a mixture of T/P and trap, and then extension was initiated after 3 min. In the positive control reaction, RT was incubated with T/P, and the reaction was initiated in the absence of trap. The products were resolved on a 14% urea-PAGE gel under denaturing conditions and visualized by Personal Molecular Imager. To determine the product formed, the bands corresponding to extended primer were quantified. The density in the entire lane was quantified and divided by two to account for the excess primers in the reaction. The density of the fully extended product and the density in the entire lane were then used to determine the percent of primers extended. This was then multiplied by the total template concentration in the reaction (8 nM) to obtain the amount of product formed.
Steady-state Kinetic Analysis-The reaction condition for the steady-state kinetic analysis is the same as the primer extension reaction with minor modifications (30,31). Briefly, the 17-, 18-, and 19-mer primers with matched 3Ј ends were 5Ј end 32 P-radiolabeled and annealed to rNMP-containing or rNMPfree templates. For each T/P pair and dNTP to be examined, RT and dNTP concentrations that extend 50% of 20 nM primer were established. Reactions to determine the K m and V max were performed with the predetermined protein amount and different dNTP concentrations. To assess a mismatch extension at rNMP site using a steady-state kinetics assay, a 19-mer primer was annealed to an rNMP-containing or rNMP-free template, and the incorporation of the correct dNTP after a mismatched 3Ј end of the primer was investigated. The reactions were resolved on 14% urea-PAGE under denaturing conditions and visualized by Personal Molecular Imager. The product was then quantified to determine the product formation rate, which was plotted against dNTP concentrations. The K m and V max were determined by fitting data obtained with the Michaelis-Menten equation ). K cat was calculated using the obtained values and RT concentration used in the reaction. The incorporation efficiency was determined by the expression K cat /K m .
Mismatched Primer Extension Assay-The assay was performed as described before but with minor modifications (32). The amount of protein that gave equal activity on matched rNMP-containing and rNMP-free T/P complex was determined by primer extension assay described above. This RT concentration was set as 1ϫ. 19-mer primers with varying 3Ј end nucleotide, either matched or mismatched, were 5Ј-32 P-radiolableled and annealed to a 38-mer DNA template containing rNMP or dNMP at position 19 from 3Ј end. These primers were then extended at 37°C with varying concentrations of RT. The protein concentrations are 1ϫ, 0.5ϫ, or 0.25ϫ for matched primers and 16ϫ, 8ϫ, and 4ϫ for mismatched primers. The reactions were terminated after 5 min with the addition of 40 mM EDTA. The products were analyzed as described above. To determine the product formed, the band corresponding to extended primer and the entire lane were quantified, and percent of primers extended was calculated. This was then multiplied by the total primer concentration in the reaction (20 nM) to obtain the amount of product formed.

RESULTS
An rNMP Embedded within a DNA Template Induces Pausing of HIV-1 RT-We recently reported that HIV-1 RT incorporates rNMPs frequently at macrophage dNTP/rNTP concentrations but not at activated CD4 ϩ T cell concentrations. This is due to the higher discrepancy between dNTP and rNTP levels in macrophages (4). Also, previously published studies have shown that the presence of rNMPs in a DNA template causes DNA polymerases to pause during DNA synthesis (7,23). Thus, we tested whether the presence of rNMPs in DNA templates also has the same pausing impact on HIV-1 RT-mediated DNA synthesis. For this test we performed a time course primer extension assay of HIV-1 RT on 48-nucleotide DNA templates containing a single rNMP (rAMP, a purine, or rUMP, a pyrimidine) at position 23 from 3Ј end of the template (see N in Fig. 1, A and B, and supplemental Table 1 showing sequences of all templates and primers used) under macrophage (20 -40 nM) or T cell (2-5 M) dNTP conditions. As shown in Fig. 1A, at the macrophage dNTP concentration, HIV-1 RT generated two paused products near the rNMP site (N site: arrows 1 and 2), whereas no evident RT pausing was observed in the reaction with the template containing normal dAMP at the correspond-ing positions, even at the macrophage dNTP concentrations (Fig. 1A). We also observed RT pausing in the reactions with the same DNA template but containing dUMP at this position (supplemental Fig. 1), although the pausing was less pronounced as compared with that observed on the rAMP-containing template. In contrast, when we conducted the same primer extension reaction at the high dNTP concentrations found in activated T cells (Fig. 1B), the rAMP-induced RT pausing was not observed. This supports that the abundant dNTP substrate in T cells could allow HIV-1 RT to bypass an rNMP in the DNA template without pausing, whereas pausing readily occurs when the dNTP pools are limited. We also tested the impact of an rNMP in a different sequence and at a different position in the DNA template (Fig. 1C). A similar pausing pattern of HIV-1 RT was observed only with the T/P containing an rNMP at macrophage dNTP concentrations (see two arrows in Fig. 1C). These data suggest that the rAMP-induced HIV-1 RT pausing is sequence-and position-independent. Although the RNase H activity of HIV-1 RT is not known to cleave a single rNMP in DNA, it is possible that the pausing may be induced by the premature cleavage of the rNMP-containing DNA template at the rNMP sites by the RNase H activity of HIV-1 RT during the reaction. To test this possibility, we employed the 5Ј endlabeled template used in Fig. 1A and incubated with or without HIV-1 RT. As shown in Fig. 1D, there was no cleavage of the rAMP-containing template in reactions with HIV-1 RT. As a control, the same template was treated with 0.3 M KOH, which will hydrolyze 3Ј of the rNMP in the DNA template. Under this condition it is evident that the rNMP-containing template is hydrolyzed, generating a smaller sized product (cleaved product (CP) in Fig. 1D), excluding the possibility that the pausing of HIV-1 RT shown in Fig. 1 resulted from the RT mediated cleavage of the rNMP containing template. This further supports that the RNase H activity of HIV-1 RT does not cleave DNA templates at the single rNMP site.
Upon examining the template sequence at the pause sites observed in Fig. 1, A and C, we found that with the rAMP template and at macrophage dNTP pools, HIV-1 RT generated two paused products. The first paused product is the primer terminated at two nucleotides upstream (arrow 1) of rNMP site (N site), which is generated by the delay at the one nucleotide upstream (Ϫ1 site) of the N site. The second paused product is the primer terminated at the N site (arrow 2), which is generated by the delay at the one nucleotide downstream (ϩ1 site) of the N site. However, on a template containing rUMP, in addition to less pronounced pausing overall, there was no pausing observed at N site. This indicates a possible difference in the effect of purines and pyrimidine on RT pausing. Basically, the data presented in Fig. 1, A and C, and supplemental Fig. 1 suggest that the DNA polymerase activity of HIV-1 RT is significantly reduced both at one nucleotide downstream and upstream of the rAMP site but only at one nucleotide downstream of an rUMP site.
Several studies have reported that an rNMP present in dsDNA does change the structure of the duplex DNA by promoting a shift from a B-form helix to an A-form (14,15,17). To investigate the effect of the rNMP present in our DNA tem-

Impact of rNMPs on HIV-1 RT Kinetics and Fidelity
plate, we utilized CD, which has been employed to monitor changes in the structure of DNA (16,33). Because one of the pause sites was observed at the position corresponding to rNMP present in the DNA template, we analyzed the structure of the same DNA template while annealed to a 22-mer primer (see the diagram in Fig. 1E). In this configuration the rNMP will be at the n position of RT active site and will be in the singlestranded region of the T/P duplex. Although rAMP is present in a single strand region of the T/P, the CD spectra did show an elevated ellipticity near wave length 280 nm (see box in Fig. 1E), that is indicative of a structural shift toward an A-form helix, and this is a similar effect of rNMP as previously observed when in duplex DNA (16). In addition, we observed a similar shift when rAMP is within the duplex region of our T/P (supplemental Fig. 2, A and B). Our CD data suggest that the observed two pause sites could occur due to DNA structural change by the embedded rNMP, which may affect the interaction of RT with template and dNTP incorporation kinetics. Collectively, the data shown in Fig. 1 support that rNMPs embedded in DNA template induces HIV-1 RT pausing more frequently in limited dNTP pools, possibly due to local structural changes in the T/P, which may slow down the overall HIV-1 RT-mediated DNA synthesis. Importantly, it also suggests that HIV-1 RT could bypass the rNMP-induced pausing when RT kinetics is optimal due to abundant dNTPs.
Effect of an rNMP Embedded within a DNA Template on T/P Binding of HIV-1 RT-Although previously it had been shown that HIV-1 RT binding affinity to DNA-primed DNA and RNA templates are similar (34), HIV-1 RT may interact with an rNMP-containing DNA template differently, particularly when the 3Ј end of the primer lies near the rNMP site (N site). In addition, the structural change observed previously and, with our CD data combined with the pause site observed during primer extension, prompted us to test whether the embedded rNMP can alter the T/P binding kinetics of HIV-1 RT. The overall T/P binding affinity of polymerases is determined by two opposing events. Initial binding was measured by the association rate or on-rate (k a or k on ), and release from T/P, was measured by the dissociation rate or off-rate (k d or k off ). Together this results in the binding affinity (K D ), which is determined by k d /k a (35,36). To examine these parameters, we employed a surface plasmon resonance-based biosensor technology, which has been used previously for the same purpose (35). We determined the K D values of HIV-1 RT with primers aligned at the two paused products, marked by arrows 1 and 2 in Fig. 1, which were generated by the delayed incorporation at the Ϫ1 and ϩ1 sites relative to the N site (see diagram in Table 1 and supplemental Fig. 3). As shown in Table 1 and supplemental Fig. 3, when the 3Ј end of the primer is aligned at the positions corresponding to the site of pause products, the experimentally determined k a , k d and K D values of HIV-1 RT to T/P were not significantly altered by presence of rNMP. Therefore, the data presented in Table 1 and supplemental Fig. 3 together with the CD data indicate that although an rNMP embedded in DNA template alters the T/P structure, this alteration does not significantly affect the T/P interaction of HIV-1 RT. Furthermore, because the T/P binding kinetics of HIV-1 RT at the rNMP-mediated paused products were not altered by the rNMP, this supports that the T/P binding step of HIV-1 RT is not a part of the mechanistic cause for the rNMP-mediated pausing of HIV-1 RT.
Effect of an rNMP on DNA Polymerase Kinetics of HIV-1 RT-Because the T/P binding kinetics of HIV-1 RT is not mechanistically involved in the rNMP-mediated pausing, next we tested whether an rNMP embedded in DNA template directly affects the overall dNTP incorporation kinetics of RT near the rNMP site. To test this, we performed a single nucleotide extension steady-state kinetics assay using a 38-mer template containing dA, rA, dT, or rU, which was annealed to the radioactively labeled primer for the incorporation of correct dNTPs at the Ϫ1, N, or ϩ1 position relative to the rNMP site (see supplemental Table 1 for the sequence of template and primers used). These T/Ps positions the rNMP at ϩ1, N, or Ϫ1, relative to HIV-1 RT active site (N indicates active site as shown in Table 2). The steady-state Michaelis-Menten constant (K m ) and catalytic constant (k cat ) of HIV-1 RT were experimentally determined, and the dNTP incorporation efficiency (k cat /K m ) was calculated (Table 2). Interestingly, HIV-1 RT showed similar k cat /K m values during the incorporation of dTTP at dAMP

The K a , K d , and K D values of HIV-1 RT on rAMP-containing and rAMP-free templates
Fold changes of each parameter as compared on the dAMP to rAMP site are given in the parentheses. a The dashed arrow indicates the position of rNMP relative to RT active site. The solid arrow indicates the position of RT active site. b The K a , K d , and K D values were derived by fitting the SPR data with a 1:1 binding model.

TABLE 2 Steady-state kinetics values of HIV-1 RT for incorporation of correct dNTP on rNMP-containing and rNMP-free DNA templates
Fold changes of dNTP incorporation efficiency as compared on the dAMP to rAMP site are given in the parentheses. and rAMP sites (9973.9 Ϯ 578 and 7534.0 Ϯ 443 M Ϫ1 min Ϫ1 , respectively), suggesting that the rAMP in the template does not affect the dNTP incorporation efficiency of HIV-1 RT at the actual site complementary to the rAMP. In contrast, when we measured the dNTP incorporation efficiency of HIV-1 RT when the rNMP is present at the ϩ1 or Ϫ1 sites relative to the rAMP (N) site, where Fig. 1 showed the delayed kinetics and consequent generation of the two paused products, HIV-1 RT showed a 4-and 6-fold reduction in the dNTP incorporation efficiency ( Table 2). In addition, HIV-1 RT displays a 3-fold reduction in dNTP incorporation efficiency at the Ϫ1 position with respect to rUMP (supplemental Table 2), a position corresponding to where delayed kinetics and consequent generation of paused product were observed (supplemental Fig. 2). Interestingly, although the K cat values of HIV-1 RT at the Ϫ1 and ϩ1 sites were not affected by the rAMP, the K m values were increased at those sites on rAMP containing template as com-pare with rAMP-free template. Thus, this kinetic finding supports that the rNMP-mediated pausing of HIV-1 RT at the Ϫ1 and ϩ1 positions was generated by inefficient dNTP incorporation, leading to a severe kinetic block at those sites when the dNTP availability become limited (below the K m values). Next, we tested if the rAMP embedded in DNA template directly affects the DNA synthesis rate at low macrophage and high T cell dNTP concentrations during single round primer extension. To establish the single round primer extension, we employed a molar excess of unlabeled T/P trap, which captures the HIV-1 RT molecules released from the template after the initial primer extension and prevents the rebinding of RT to the 5Ј end 32 P-labeled T/P thus limiting the extension to a single round (37)(38)(39). As shown in Fig. 2, at both low macrophage (Fig.  2, A and C) and high T cell (Fig. 2, B and D) dNTP concentrations, without the trap (ϪT), a large amount of fully extended products was observed due to the multiple rounds of primer . The reactions were terminated at 10, 20, and 60 min after the reaction initiation, T1, T2, and T3, respectively. Triplicate reactions were performed, and the products were quantified for macrophages (B) and activated T cells (D) dNTP pools. TC, HIV-1 RT was preincubated with a trap and labeled T/P mixture and the reaction was initiated by the addition of dNTPs. The reaction was terminated after 60 min for macrophage dNTP condition and after 20 min for T cell dNTP condition. No primer extension should be detected in this control. ϪT control, HIV-1 RT was preincubated only with the 32 P-labled T/P in the absence of the trap, and the reaction was initiated by dNTPs and terminated at 20 min. In this condition, RT undergoes multiple rounds of primer extension and generates a greater amount of the fully extended product. C, control. As a negative control, no RT reaction was initiated by dNTP and terminated at 20 min. C, the reaction products were analyzed by 14% denaturing PAGE. F indicates fully extended product, P indicates unextended primer, and N indicates the location of rAMP and corresponding dAMP. The two pause sites are indicated by arrows. The template sequence from the 3Ј end of the annealed primer was marked at the side of the gel. extension. However, when HIV-1 RT was preincubated with the trap before the primer extension (TC), no product was observed, showing that the trap was successfully able to capture and prevent HIV-1 RT from initiating the DNA synthesis. For the single round of the primer extension, the primer extension reactions were initiated by mixing the HIV-1 RT, which was prebound to the 5Ј end 32 P-labeled T/P, with a mixture of dNTPs (macrophage or T cell concentration), Mg 2ϩ , and the unlabeled T/P trap and terminated at three different time points (T1-T3). The reactions with the rAMP containing T/P produced strong pauses (see the arrows) but only at the macrophage/low dNTP concentrations ( Fig. 2A), which is similar with what we observed during the multiple rounds of the primer extension. However, no pausing was observed with the dAMPcontaining template regardless of the dNTP concentrations used. More importantly, as seen in the level of the fully extended products (F in Fig. 2A) and quantified product (Fig.  2C), the reactions with the rAMP-containing T/P at the macrophage dNTP concentrations generated reduced fully extended product compared with the reactions with the dAMP-containing T/P, supporting that the rAMP embedded in the DNA template restricts the processive DNA synthesis of HIV-1 RT at limiting dNTP concentrations. In contrast, under T cell dNTP concentrations no pause site was observed, and the fully extended product was similar on both rNMP-containing and rNMP-free templates (Fig. 2, B and D). Interestingly, the paused products in the rAMP containing T/P ( Fig. 2A) gradually decreased during incubation under the single round primer extension condition. In addition, fully extended product on the rAMP-containing template continued to increase in a time-dependent manner (Fig. 2C), supporting that some of the HIV-1 RT molecules, which paused at the Ϫ1 and ϩ1 sites relative to rNMP, remained bound to the template and were able to continue extending the prematurely terminated primers at the pause sites. Basically, this kinetic study supports that the rNMPs in DNA template can further slow down the processive DNA polymerization kinetics of HIV-1 RT at low dNTP concentrations.
Effect of an rNMP in the Template DNA on Enzyme Fidelity of HIV-1 RT-It has been reported that the rNMPs contained in DNAs are mutagenic, and cells harbor a repair system with RNase H2 that specifically removes rNMPs molecules in DNA duplex incorporated by cellular DNA polymerases during DNA replication (18,40,41). Thus, we tested whether rNMPs embedded in DNA templates could become mutational hotspots by affecting the enzyme fidelity of HIV-1 RT at the rNMP site by using a steady-state single nucleotide fidelity assay. The complete synthesis of a single mutation during DNA polymerization consists of two sequential mechanistic steps; 1) misinsertion (incorporation of incorrect dNTPs) and 2) extension of the mismatches generated by the misinsertion (32). Therefore, first we measured the misinsertion fidelity of HIV-1 RT during dNTP incorporation opposite a dAMP or rAMP. For the dAMP site, the misinsertion fidelity for the incorrect dGTP was 6.6 ϫ 10 4 . When the misinsertion fidelity was measured with the rAMP-containing template, similar misinsertion fidelity was observed (Table 3). This suggests that an rNMP embed-ded in the DNA template does not alter the misinsertion fidelity of HIV-1 RT at the N site.
Next, we determined the mismatch extension fidelity of HIV-1 RT with the primers containing dGMP at the 3Ј end, which is mispaired to either dAMP or rAMP in the template. Indeed, as shown in Table 4, HIV-1 RT showed nine times more efficient mismatch extension or decreased mismatch extension fidelity with the rAMP-containing template as compared with the dAMP-containing template (Table 4).
Next, we compared the capability of HIV-1 to extend matched and mismatched primers at the N site using the multiple nucleotide incorporation assay. In this assay the 5Ј endmatched and -mismatched primers annealed to the rAMP-or dATP-containing DNA template at the N site were extended by HIV-1 RT. As shown in Fig. 3, A and B, HIV-1 RT showed similar matched primer extension capability, which was assessed by the amount of the full-length product (F), with both rAMP-containing and rAMP-free templates. However, when provided with mismatched primers on templates containing rAMP at the N site, HIV-1 RT showed enhanced primer extension on rAMP-containing template as compared with rAMPfree templates (Fig. 3, A and B). Thus the data presented in Tables 2 and 3 combined with Fig. 3 suggest that the rNMPs  embedded in the DNA template alter the enzyme fidelity of HIV-1 RT by decreasing the mismatch extension fidelity.
Effect of rNMPs in DNA Template on Pausing of Other Reverse Transcriptase-Next, we tested whether the rNMP-induced pausing, which was observed with HIV-1 RT, is also observed with other retroviral RTs. For this test we first employed RT of the lentiviral simian immunodeficiency virus (SIVagm Sab-1 (SIV)). As shown in Fig. 4A, SIV RT also generated the two paused products near the N site of the rNMPcontaining template but only at macrophage dNTP concentrations, which is the same pattern as HIV-1 RT (Fig. 1). Finally we tested the rNMP-mediated pausing of AMV RT (oncoretrovirus). Importantly, unlike lentiviral RTs, which are functional in both high dividing cell and low non-dividing cell dNTP concentrations, oncoretroviral RTs are not functional at the low nondividing cell dNTP concentration because oncoretorviruses do not replicate in non-dividing cells such as macrophages (42,43). In addition we have shown biochemically that oncoretrovirus RTs, including AMV RT, function poorly in low dNTP concentration (30). Previously, it has been shown that a yeast replicative DNA polymerase, which normally functions with abundant dNTPs, pauses under high dNTP concentrations (7). This prompted us to investigate whether AMV RT may pause on the rNMP-containing template under the dNTP conditions in which it normally functions. Thus we tested the impact of rNMP on AMV RT at only high dNTP concentration found in T cells. Indeed, as shown in Fig. 4B, AMV RT did not display any pausing near the N site. Thus, the data shown in Fig. 4 demonstrate that the rNMP-mediated pausing, which can induce kinetic delay of the DNA synthesis, is specific for the lentiviral RTs such as HIV-1 and SIV RT proteins.

DISCUSSION
DNA polymerases have evolved to maintain an effective "gate" mechanism that prevents the entry of cellular rNTPs, which tends to be at 100 -1000 times higher concentrations than cellular canonical dNTPs, to the active site of DNA polymerases. This mechanism is engineered by the molecular clash between the conserved residues of DNA polymerases near the dNTP binding site and the 2ЈOH of the rNTP (8,12). However, this mechanism appears not to be completely successful, leading to the incorporation of noncanonical cellular rNTPs during chromosomal DNA replication. This is apparent because almost all organisms maintain a specific cellular repair mechanism that removes single rNMPs embedded in dsDNA initiated by RNase H2. The ubiquitous nature of the RNase H2 repair system strongly supports that the rNMP molecules embedded in DNA significantly impacts cellular DNA metabolisms (41,44). It has been reported that host DNA polymerases also pause near the sites of DNA-containing rNMPs (7). Two possible consequences that can be induced by rNMP-induced pausing are 1) the kinetic delay of chromosomal replication and 2) mutation synthesis. A 5Ј end 32 P-labeled matched or mismatched primer at its 3Ј end was annealed to rAMP-containing or rAMP-free DNA template. This aligns rNMP at Ϫ1 position in HIV-1 RT active site. RT concentration that extends about 50% of the primers was determined and set as 1ϫ. The primers were then extended with 1ϫ, 0.5ϫ, and 0.25ϫ proteins for matched and 16ϫ, 8ϫ, or 4ϫ for mismatched primers because the mismatch primer extension requires a higher RT amount to produce the detectable extended products. The reactions were then terminated after 5 min. A, the reaction products were analyzed by 14% denaturing PAGE. F indicates fully extended product, and P indicates unextended primer. The N in the diagram shows the location of embedded rNMP (rAMP) and corresponding dAMP (dAMP). B, triplicate reactions were performed, and products with 1ϫ for matched and 16ϫ for mismatched primers were quantified. The numbers above the graphs indicate -fold change as product on rNMP-containing template compared with that of dAMP-containing template.
Our previous study reported that HIV-1 frequently incorporates rNTPs during proviral DNA synthesis with a rate of 1/146 in macrophages due to the a greater discrepancy between cellular dNTP and rNTP concentration but not in dividing cells with much smaller rNTP/dNTP discrepancy (13). The incorporation of rNMPs during the first (Ϫ) strand synthesis produces DNA templates containing rNMPs, which will be used for the second (ϩ) strand synthesis. Our followup study here clearly demonstrates that HIV-1 RT also pauses near the rNMP sites of the DNA template, which can delay viral replication kinetics, and the pauses occur due to the delayed dNTP incorporation kinetics at one nucleotide before and after the rNMP site instead of the interrupted T/P interaction of HIV-1 RT. Furthermore, HIV-1 RT becomes more error-prone specifically at the rNMP sites specifically by facilitating the mismatch extension capability. Our data also suggest that it is plausible that the rNMP-mediated kinetic delay and fidelity change may be mechanistically related with the local structural change of the T/P near the rNMP site. The rNMP-induced pausing was also observed for SIV RT at the limited dNTP pools, suggesting that consequences of the rNMP incorporation are common for lentiviral RTs. The oncoretroviral AMV RT, which is active only with high dNTP concentrations, failed to generate the rNMPinitiated pausing under these conditions.
We recently reported that the RNase H2-mediated repair function is significantly less in macrophages as compared with the activated CD4 ϩ T cells and other dividing cells (13). Thus, we envision the postulated kinetic delay and potential mutagenic consequence of the rNMP incorporation could be mechanistic bottlenecks that lentiviruses encounter in macrophages. Our biochemical simulation in Fig. 1 clearly demonstrates that the rNMP-mediated pausing disappears when HIV-1 RT synthesizes DNA at a fast rate with elevated dNTP concentrations. In fact, we and others recently reported that the limited dNTP pool found in macrophages is due to the expression of host restriction factor, SAMHD1, which is a dNTP triphosphohydrolase (5,45). Indeed, we also reported that HIV-2 and some SIVs encode an accessory protein, Vpx, that antagonizes the anti-viral function of SAMHD1 by proteasomal degradation. This led to the elevation of cellular dNTPs in macrophages (5,46), dendritic cells (47), and resting T cells (48,49), which ultimately rescued the delayed reverse transcription process. Previously, we have shown that dNTP levels in macrophage are lower than the K m values of RT (4,50). However, Vpx-mediated degradation of SAMHD1 elevates dNTP levels above the K m values of RT (5,50), accelerating proviral DNA synthesis (46). Therefore, we further envision that Vpx-induced dNTP elevation in the non-dividing target cell types enables RTs to effectively overcome the rNMP-induced pausing as shown in Fig. 1B and consequently alleviating the kinetic delay in proviral DNA synthesis.
In summary, we have demonstrated that DNA templates containing rNMPs slow down HIV-1 RT-mediated DNA synthesis under limiting dNTP concentrations, and this impact is reversed under elevated dNTP concentrations. In addition, we show that rNMPs embedded in DNA templates decreases RT fidelity. Overall, this study demonstrates the effect of chimeric DNAs, which are generated during the first strand of proviral DNA synthesis, on HIV-1 replication kinetics and mutagenesis, particularly in macrophages.  Fig. 1 was extended with an equal activity of SIV agm RT protein for indicated time points. The reactions were conducted under macrophage dNTP concentration for SIV RT (A) and T cells dNTP concentration for AMV RT (B). The products were resolved by 14% denaturing PAGE and analyzed. F indicates fully extended product, and P indicates unextended primer. The N in the diagram shows the location of embedded rAMP or dAMP.