Delayed Chain Termination Protects the Anti-hepatitis B Virus Drug Entecavir from Excision by HIV-1 Reverse Transcriptase*

Entecavir (ETV) is a potent antiviral nucleoside analogue that is used to treat hepatitis B virus (HBV) infection. Recent clinical studies have demonstrated that ETV is also active against the human immunodeficiency virus type 1 (HIV-1). Unlike all approved nucleoside analogue reverse transcriptase RT) inhibitors (NRTIs), ETV contains a 3′-hydroxyl group that allows further nucleotide incorporation events to occur. Thus, the mechanism of inhibition probably differs from classic chain termination. Here, we show that the incorporated ETV-monophosphate (MP) can interfere with three distinct stages of DNA synthesis. First, incorporation of the next nucleotide at position n + 1 following ETV-MP is compromised, although DNA synthesis eventually continues. Second, strong pausing at position n + 3 suggests a long range effect, referred to as “delayed chain-termination.” Third, the incorporated ETV-MP can also act as a “base pair confounder” during synthesis of the second DNA strand, when the RT enzyme needs to pass the inhibitor in the template. Enzyme kinetics revealed that delayed chain termination is the dominant mechanism of action. High resolution foot-printing experiments suggest that the incorporated ETV-MP “repels” the 3′-end of the primer from the active site of HIV-1 RT, which, in turn, diminishes incorporation of the natural nucleotide substrate at position n + 4. Most importantly, delayed chain termination protects ETV-MP from phosphorolytic excision, which represents a major resistance mechanism for approved NRTIs. Collectively, these findings provide a rationale and important tools for the development of novel, more potent delayed chain terminators as anti-HIV agents.

Co-infection with the hepatitis B virus (HBV) 3 and the human immunodeficiency virus, type 1 (HIV-1) is common and complicates treatment (1). Although both viruses share a highly related target for pharmaceutical intervention, there are few drugs that are approved to treat HBV and HIV-1 infection simultaneously (1,2). HBV and HIV replicate through a reverse transcription step, carried out by the virally encoded HBV DNA polymerase and the HIV-1 reverse transcriptase (RT), respectively (3,4). Lamivudine (3TC) and emtricitabine are cytidine analogues that exert potent antiviral effects against HBV and HIV-1, targeting their DNA polymerase (5)(6)(7). Once intracellularly phosphorylated to their triphosphate forms, these drugs act as chain terminators. However, the development of resistance can limit their clinical utility (8 -10). In HIV-1, both 3TC and emtricitabine can select for the M184V mutation in the conserved YMDD motif at the active site of RT (11). A structurally and functionally equivalent mutation (i.e. M204V) is selected in HBV (10,(12)(13)(14)(15)(16). This mutation is often accompanied with other mutations, that appear to compensate for fitness deficits that are introduced by the primary mutation (17). The incidence of 3TC resistance is high in part because 3TC has been for a long time the only small molecule approved for the treatment of HBV, and 3TC has also been and remains an important component in recommended drug regimens for the treatment of HIV (2). The acyclic phosphonate tenofovir is active against 3TC-resistant HIV and HBV strains (18 -21), whereas the structurally related compound adefovir has been approved for HBV treatment only, although it has activity against HIV (22,23).
Entecavir (ETV) belongs to the few available drugs that retain potency against 3TC-resistant HBV (24,25). This compound does not appear to select for M204 and L180M in HBV, although the preexistence of the two mutations can decrease susceptibility to ETV (26,27). Other additional mutations can further amplify clinically relevant levels of resistance to this drug (28). In contrast, it has recently been demonstrated that ETV can select for the M184V mutation in HBV/HIV-co-infected individuals (29). These data showed unambiguously that ETV can exert antiretroviral effects, which led to the recommendation that ETV should not be administered in co-infected individuals unless these persons are simultaneously on highly active antiretroviral therapy (29,30). Subsequent case reports and clinical studies with smaller cohorts are consistent with the original study, and in vitro selection experiments as well as phenotypic susceptibility measurements with HIV-1 strains containing the M184V mutation confirmed the clinical data (31)(32)(33)(34)(35). Moreover, pre-steady-state kinetics with wild type HIV-1 RT demonstrated that the enzyme can incorporate ETV-monophosphate (MP) (36). The M184V mutant appears to be able to discriminate against the inhibitor, since the efficiency of incorporation of ETV-MP is severely diminished.
Aside from the clinical importance of these findings, the observation that ETV exhibits antiretroviral activity has potential implications for the development of novel drugs that may evade major resistance pathways in HIV. Unlike all approved nucleoside analogue RT inhibitors that lack the 3Ј-hydroxyl group of the sugar moiety and act as chain terminators, ETV contains this group, which can attack the next incoming nucleotide on its ␣-phosphate (Fig. 1). Thus, DNA synthesis may be inhibited at several steps after incorporation of ETV-MP. In vitro studies with HBV replication complexes (37) and purified HIV-1 RT (36), respectively, indicated pausing of DNA synthesis immediately following incorporation of ETV-MP and later following incorporation of up to three additional nucleotides. Such late pausing is referred to as "delayed chain termination." Very few nucleotide analogues have been described that show this type of inhibition of HIV-1 RT (38); however, these compounds are toxic and/or are not converted intracellularly into their triphosphate form (39). ETV may therefore be exploited as a model compound for the study of delayed chain termination and its implications in current drug development efforts. Of note, ETV is fully susceptible against a background of thymidine analogue-associated mutations (TAMs) (36), which reduce susceptibility to literally all approved nucleotide analogue RT inhibitors (NRTIs), albeit at different degrees (40). TAMs include changes at positions 41, 67, 70, 210, 215, and 219 that were shown to increase the phosphorolytic excision of incorporated nucleotide analogues (41,42). These mutations are able to recruit ATP as a PP i donor, which ultimately removes the inhibitor from the primer terminus and leads to the rescue of DNA synthesis. ETV may evade this resistance mechanism through delayed chain termination.
Here, we demonstrate that delayed chain termination at position n ϩ 3 (i.e. three nucleotides following incorporation of ETV-MP) is the major mechanism of inhibition. Although ETV-MP is efficiently excised with TAMs containing RT enzymes, the inhibitor is not excised when the primer was extended by three additional nucleotides. These proof-of-principle studies show that delayed chain terminators are protected from excision.
Synthesis of ETV-TP-2-Amino-9-[4-hydroxy-3-(hydroxymethyl)-2-methylidene-cyclopentyl]-3H-purin-6-one (4.71 mg, 1.7 mmol) was dissolved in 200 ml of dry 1,3-dimethyl-2oxohexahydropyrimidine N,NЈ-dimethylpropylene urea with 15 molecular sieves under nitrogen and stirred for 24 h. The mixture was chilled with an ice-water bath and stirred for 1 h, followed by the slow addition of phosphorus oxychloride (3 eq) and stirred for another 50 min. A solution of tributylammonium pyrophosphate (4 eq) in 200 l of 1,3-Dimethyl-3,4,5,6tetrahydro-2(1H)-pyrimidinone and tributyl amine (15 eq) were simultaneously added to the reaction. After 2 h, the reaction was quenched with ice-cold water and slowly brought to room temperature. The solution was washed with chloroform, and the aqueous layer was collected and co-evaporated with water three times. The product was resuspended in 100 ml of water and purified by high performance liquid chromatography (HPLC), followed by co-evaporation with water, giving a total yield of 9% with 98% purity. ETV-TP was finally purified by ion exchange HPLC.  Nucleotide incorporation was initiated by the addition of MgCl 2 to a final concentration of 10 mM, and the reactions were allowed to proceed for 3 min. The reaction conditions were optimized such that the rate of the product formation was within its linear range with respect to the enzyme concentration and time point, which is consistent with the steady-state approach. The reactions were stopped by the addition of 3 reaction volumes of formamide containing traces of bromphenol blue and xylene cyanol. The samples were then subjected to 18% denaturing PAGE followed by phosphorimaging. The incorporation of single nucleotides was quantified as the fraction of the DNA substrate (template-primer) converted to product (template-primer ϩ 1 nucleotide). The rate of the reaction was plotted versus the concentration of nucleotide substrate. The data were fitted to the Michaelis-Menten equation by use of GraphPad Prism (version 4.0) to determine K m and k cat values for dGTP and ETV-TP. Significant figures for the fitted data are as reported by the software. k cat was defined as the maximal rate of single nucleotide incorporation. K m was defined as the concentration of dNTP at which the rate of single nucleotide incorporation equals half of the maximal rate. In the experiments where both dGTP and ETV-TP were present for the incorporation at position n, the concentration of ETV-TP at which the incorporation of dGTP was reduced by 50% (IC 50 ) was determined by plotting the percentage incorporation of dGTP versus the concentration of ETV-TP. The data were fit to a sigmoidal curve (variable slope).

Inhibition of HIV-1 RT by Entecavir
DNA Synthesis at Position n ϩ 1-The reactions were conducted as described above, except that (i) T50A/P1 was used as the DNA/DNA hybrid, (ii) 1 M dGTP or 10 M ETV-TP was used for the incorporation at position n, and (iii) the time point was extended to 30 min to ensure maximum product formation at position n. Then increasing concentrations of dTTP from 0.02 to 2 M with n ϭ dGMP and 0.078 to 10 M were allowed to proceed for 4 min.
DNA Synthesis at Position n ϩ 4-The reactions were conducted as described, except that (i) T50A6/P1 was used as the DNA/DNA hybrid and (ii) increasing concentrations of dCTP from 0.03 to 256 M were added in a series of 2-fold dilutions. Reactions were allowed to proceed for 4 min. In the experiment where dGTP, ETV-TP, dTTP, and dCTP were present at the same time, only dGTP concentrations were varied from 0.03 to 2 M, whereas the concentrations of ETV-TP, dTTP, and dCTP were kept constant at 5 and 0.5 M, respectively. In addition, the concentrations of HIV-RT and DNA substrate were adjusted to 100 nM, and the reactions were stopped after 5 min to enhance the signal at the pausing sites.
Site-specific Footprinting-In preparation of the footprinting reaction, the 5Ј-end-labeled template T50A was heat-annealed with the primer P1. 50 nM DNA/DNA hybrid was incubated with 750 nM HIV-RT for 10 min in a reaction mixture containing sodium cacodylate, pH 7 (120 mM), NaCl (20 mM), DTT (0.5 mM), MgCl 2 (10 mM), and 25 M ddGTP or ETV-TP in a final volume of 15 l. At the 10 min time point, increasing concentrations of foscarnet (PFA) or dTTP from 0.8 to 500 M were added to the reactions in a series of 5-fold dilutions, followed by an incubation of 5 min at 37°C. The complex was treated with 0.1 mM ammonium Fe(II) sulfate hexahydrate (44). The reactions were allowed to proceed for 5 min and were processed and analyzed as described.
Ribonuclease H (RNase H) Activity Assays-Primer P1, annealed to T50A or T50A6 templates, was extended by 2 or 4 nucleotides, respectively, such that dGMP or ETV-MP were incorporated at position n, followed by 1 or 3 more nucleotides. These primers with n ϭ dGMP or ETV-MP were gel-purified and reannealed to T50A or T50A6 RNA templates, respectively. 3Ј-End-labeling with [ 32 P]pCp and T4 RNA ligase was carried out as described (45). The RNA/DNA hybrid (50 nM) was incubated with 750 nM HIV-RT in a reaction buffer containing 50 mM Tris-HCl, pH 7.8, 50 mM NaCl, and 0.3 mM EDTA at 37°C for 10 min. Heparin (4 mg/ml, final) was added to the reaction mixture and allowed to incubate for variable time ranging from 0 to 60 s, after which MgCl 2 (10 mM, final) was added to initiate the RNase H activity. The reactions were stopped by the addition of 3 reaction volumes of formamide containing traces of bromphenol blue and xylene cyanol. The samples were subjected to 18% denaturing PAGE, followed by phosphorimaging. RNase H activity was monitored and quantified based on the appearance of the 3Ј-end-labeled RNA template degradation products. RNase H-specific products were identified based on the experiments controlling for the efficiency of the heparin trap as well as for RNase H activity in the Containing dGMP, ddGMP, or ETV-MP at the 3Ј-end of the primer-The rate constant (k cat ) for dGTP incorporation using 150 nM T50A/P1 primertemplate was determined from the slope of the linear portion of product formation versus time. Reactions were carried out in the presence of 50 nM HIV-RT and 1 M dGTP. The effect of the addition of 300 nM primer-templates already containing dGMP, ddGMP, or ETV-MP at the 3Ј-end of the primer on the k cat was monitored to assess potential differences in relative affinities of dGMP-, ddGMP-, or ETV-containing primer-templates.
DNA Synthesis across a Template Containing ETV-MP-The T45/P1 RNA/DNA hybrid was used to generate a template containing ETV-MP. The RT-associated RNase H activity degraded the original T45 RNA template. The newly synthesized DNA template containing ETV-MP was annealed to 5Ј-end-labeled primer P2 and incubated with 500 nM HIV-1 RT in the presence of an MgCl 2 (10 mM) and dNTP mix (0.5 M). Aliquots were taken at time points 1, 2, 5, 20 min, and data were processed and analyzed as described.
ATP-dependent Phosphorolysis-The radiolabeled T50A6/P1 DNA/ DNA hybrid was extended by a single nucleotide to generate an oligonucleotide with ETV-MP or dGMP at the 3Ј-end. Primers were also further extended by three residues, gel-purified, and annealed with T50A6. The 50 nM DNA/ DNA template-primer hybrid with ETV-MP at the 3Ј-end of the primer or ETV-MP followed by an additional 3 nucleotides was incubated with 750 nM HIV-1 RT in a  DECEMBER 5, 2008 • VOLUME 283 • NUMBER 49 buffer containing 50 mM Tris-HCl, pH 7.8, 50 mM NaCl, 10 mM MgCl 2 , and 3.5 mM ATP (pyrophosphatase-treated).

RESULTS
ETV-TP Is Able to Compete with dGTP-In order to assess the efficiency with which ETV-MP is incorporated in comparison with its natural counterpart dGMP, we determined the concentration of ETV-TP required for 50% inhibition (IC 50 ) of incorporation of dGMP (Fig. S1). High resolution polyacrylamide gels allowed us to distinguish between primers extended by dGMP or ETV-MP, respectively (Fig. S1A). As expected for a competitive inhibitor, we found that the IC 50 (ETV-TP) value increased with increasing concentrations of dGTP (Fig. S1B). Concentrations of dGTP as low as 0.5 M, which is within the physiologically relevant range, require relatively high concentrations of 19 M ETV-TP to obtain 50% inhibition. These data are in good agreement with steady-state kinetics that show 14 -18-fold differences in efficiencies of incorporation ( Table 1).

Incorporation of ETV-MP Causes Pausing at Positions n and n ϩ 3-
The effects of incorporation of ETV-MP at position n were analyzed by studying the efficiency of subsequent nucleotide incorporation events (Fig. 2). Enzyme pausing is evident at positions n and n ϩ 3 and to a lesser extent also at position n ϩ 2 (Fig. 2B). Pausing at positions n and n ϩ 2 can be overcome, whereas pausing at position n ϩ 3 represents the final product when DNA synthesis   was limited to four nucleotide incorporation events (Fig. 2C).
Increasing the concentrations of dCTP reduced pausing and resulted in increased product formation at position n ϩ 4 (Fig. 2D). Thus, incorporation of ETV-MP appears to exert an immediate effect on the next nucleotide at position n ϩ 1, and pausing at position n ϩ 3 points to long range effects of the inhibitor that can affect nucleotide incorporation at position n ϩ 4.
To translate these findings into quantitative terms, we devised primer-template substrates that allowed us to determine the efficiency of nucleotide incorporation at positions n ϩ 1 and n ϩ 4. The primer strands contained either ETV-MP or dGMP at position n. Steady-state kinetics revealed that the incorporation of the next complementary nucleotide was ϳ7-fold reduced when the primer was terminated with ETV-MP (Table 2). Nucleotide incorporation at position n ϩ 4 was measured with primers that were further extended by 3 nucleotides to yield the n ϩ 3 substrate. The measurements revealed substantially higher reductions (i.e. Ͼ1000-fold) in efficiency of nucleotide incorporation with ETV-MP at position n. Thus, inhibition at position n ϩ 4 was more than 2 orders of magnitude more efficient when compared with inhibition at position n ϩ 1.
Effects of ETV-MP When Present in the Template-Given that increasing the concentrations of natural dNTP substrates can overcome RT pausing (Fig. 2D), it is conceivable that the incorporated ETV-MP may also exert inhibitory effects when located in the template strand. In order to identify positions that could be affected, we devised a primer-template substrate that contained the inhibitor two nucleotides away from the primer terminus in the template (Fig. 3A). A primer extension assay revealed specific pausing opposite the incorporated ETV-MP, also referred to as position n, and at position n ϩ 1 (Fig.  3B). Together, the data point to three distinct mechanisms of inhibition: (i) inhibition of nucleotide incorporation at position n ϩ 1 immediately following the inhibitor; (ii) inhibition at later stages, predominantly at position n ϩ 4; and (iii) inhibition during synthesis of the second strand, opposite the templated ETV-MP.
Inhibition at Position n ϩ 1 Correlates with Diminished Translocation of RT-We considered two possible scenarios that could lead to inhibition of DNA synthesis by the incorporated ETV-MP. First, the inhibitor may increase dissociation of the RT primer-template complex. Second, the incorporated ETV-MP may affect the positioning of RT on its primer-template. Binding and/or incorporation of the next nucleotide would be affected in either case, which ultimately results in enzyme pausing. To address this issue, we focused on inhibition at position n ϩ 1 (i.e. the equivalent of classic chain termination) and position n ϩ 4 (i.e. delayed chain termination).
We have recently developed site-specific footprinting tools that allowed us to determine the positioning of HIV-1 RT on DNA/DNA templates at single nucleotide resolution (44, 46 -48). One of these methods takes advantage of the RT-associated RNase H domain that trails the polymerase active site by ϳ18 base pairs (45). The RNase H active site accommodates Fe 2ϩ ions that cleave the DNA template in oxidative fashion. Specific cuts were seen at positions Ϫ17 and Ϫ18. The cut at Ϫ17 was indicative of a pretranslocation complex that exists immediately following nucleotide incorporation. In this configuration, the nucleotide binding site is still occupied with the 3Ј-end of the primer, and the polymerase needs to translocate to allow binding of the next incoming nucleotide. The cut at Ϫ18 was indicative for such a post-translocation complex. The PP i analogue PFA was shown to trap the pretranslocational state (44,47), whereas the nucleotide substrate traps the posttranslocational state (Fig. 4A). Termination caused by ETV-MP incorporation resulted in different patterns. Although the pretranslocated complex can still be identified in the presence of PFA, the post-translocated complex was only seen as a faint band even in the presence of the next nucleotide (Fig. 4B). These findings indicated that ETV-MP reduces the population of post-translocated complexes that are available for nucleotide incorporation, which may help to explain the subtle inhibitory effects at position n ϩ 1 ( Table 2). Increasing concentrations of the next nucleotide will eventually result in its incorporation; however, the bands remain faint, because this complex is not sufficiently stabilized due to the lack of the following dNTP substrate. In contrast, reactions with the ddGMP-terminated primer contain the correct next nucleotide (dTTP) that can stabilize the complex.
To study potential differences in complex dissociation with ETV-MP versus dGMP-terminated primers, we measured RNase H activity directly in the presence of increasing concentrations of substrate (see "Experimental Procedures"). However, the dissociation rate constants k off (ETV-MP) ϭ 0.27 s Ϫ1 and k off (dGMP) ϭ 0.22 s Ϫ1 were almost identical (Fig. 4C), which argues against such a contribution to inhibition. Moreover, ETV-MP and dGMP-terminated primer and template substrates show the same inhibitory effects when added to reaction mixtures containing preformed ternary complexes (Fig. 4D).
Inhibition at Position n ϩ 4 Correlates with a Misplacement of RT-We were unable to obtain specific cleavage in our footprinting experiments when using the ETV-MP-containing primer that was further extended by 3 nucleotides. Since the intrinsic RNase H activity provides a more robust pattern, we used this activity again directly to identify potential differences between the substrates (Fig. 5). The pattern with the natural primer with dGMP at position n and an additional 3 nucleotides up to position ϩ3 shows three distinct cuts (Fig. 5B). The central RNase H cut was seen 18 nucleotides upstream of the 3Ј-end of the primer, which is indicative of the post-translocated configuration (Fig. 5A). The two cuts upstream and downstream did not depend on a specific position of the primer terminus and are therefore referred to as polymerase-independent. These two cuts are likewise observed with the primer that contains ETV-MP at position n (Fig. 5B). However, specific polymerase-dependent cuts were missing. The controls in the absence of inhibitor with and without PFA identified cuts that represent pre-and post-translocated complexes, and neither of these bands were seen with the ETV-modified primer. These findings suggested that ETV-MP at position n misplaces the 3Ј-end of the primer at position n ϩ 3. The sum of RNase H activity at the three cleavage events was identical with both primers, which resulted in very similar k off values that argued again against a contribution of complex dissociation to inhibition (Fig. 5C).

ETV-MP Is Efficiently Excised at Position n, and Delayed Chain Termination Provides Protection from Excision-
Whereas binding and incorporation of nucleotides can only FIGURE 5. RNase H activity on RNA/DNA template-primer substrates containing dGMP or ETV-MP 4 nucleotides upstream the 3-end of the primer. A, RNase H activity monitored with 3Ј-end-labeled RNA/DNA template-primer following incubation with heparin for variable time. Lane c1, a control experiment where heparin was omitted. Lane c2, a control experiment where MgCl 2 was omitted. Lane c3, a control experiment where trap was added prior to the addition of HIV-1 RT in order to assess the efficiency of the trap. The arrows point to the specific polymerase-dependent RNase H cleavage products that are indicative for pre-and post-translocated conformations. Lane c4, a control experiment where RNA template was subjected to alkaline hydrolysis to produce a latter. Asterisks show polymerase-independent RNase H cleavage. B, RNA/DNA substrate used in the reaction. Position Ϫ19 illustrates the distance in nucleotides between the polymerase and RNase H active sites of HIV-1 RT with respect to the 3Ј-end of the primer. ETV-MP or dGMP is incorporated at position n. The arrows and asterisks assign the RNase H cuts to the sequence of the template. C, rate of the dissociation of the RNA/DNA template-primer was determined as described in the legend to Fig. 4. occur in the post-translocated state, excision of incorporated nucleotides can only occur in the pretranslocated state that allows productive binding of PP i or the PP i donor ATP (49). The site-specific footprinting data suggested that ETV-MP may not affect the formation and/or stability of the pretranslocated complex (Fig. 4), which is a prerequisite for the excision reaction. To address this question directly, we compared efficiencies of excision of ETV-MP and dGMP with wild type RT and a mutant enzyme that contained major TAMs: M41L, D67N, L210W, and T215Y (Fig. 6). ATP-dependent excision of dGMP and ETV-MP with wild type RT was generally low (Fig. 6B). In contrast, TAMs showed significant increases in the efficiency of the removal of ETV-MP (Fig. 6, A and B). Based on these data, one would predict that TAMs decrease susceptibility to ETV; however, in vitro susceptibility experiments have shown that TAM-containing viruses are fully sensitive to this compound (36).
To reconcile these findings, we further analyzed the excision reaction at the level of delayed chain termination that may provide a certain degree of protection (Fig. 6, C and D). The natural primer without the inhibitor is readily cleaved with the TAMcontaining enzyme. Cleavage products are seen down to position n. In contrast, excision of the ETV-MP-modified primer was considerably less efficient, which is consistent with our RNase H mapping studies that detected neither pre-nor posttranslocated complexes (Fig. 5). Thus, the combined data provide strong evidence that delayed chain termination provides protection from excision and is therefore identified as a major mechanism for inhibition of HIV-1 RT by ETV.

DISCUSSION
The structure of the anti-HBV drug ETV differs from all approved anti-HIV NRTIs in that the sugar moiety contains a 3Ј-hydroxyl group. The lack of the 3Ј-hydroxyl group of estab-lished NRTIs causes chain termination, whereas the presence of this group allows in principle the nucleophilic attack on the ␣-phosphate of the next dNTP substrate. Thus, the mechanism of inhibition associated with this type of nucleoside analogues can differ from classic chain terminators. In this study, we identified three distinct mechanisms of inhibition of HIV-1 RT with ETV: (i) inhibition of nucleotide incorporation at position n ϩ 1 immediately following the incorporated ETV-MP at position n, (ii) inhibition of nucleotide incorporation at position n ϩ 4, and (iii) inhibition of nucleotide incorporation opposite the templated ETV-MP.
Inhibition at position n ϩ 1 has been described before for several compounds that target the hepatitis C virus RNA-dependent RNA polymerase (50). 2Ј-C-methylated ribonucleotides are readily incorporated by hepatitis C virus RNA polymerase and prevent binding of the next complementary nucleotide. As a consequence, this nucleotide is not incorporated, and RNA synthesis is literally terminated. Thus, these compounds act as chain terminators, despite the presence of the 3Ј-hydroxyl group. It has been suggested that some 2Ј-Cmethylated ribonucleotides may block RT translocation, which provides a mechanism for the inability to accommodate the next incoming nucleotide (51). Modeling studies with ETV-TP bound to an HIV-1 RT-derived structure of HBV polymerase point to a similar scenario (52). A putative steric clash between Tyr 203 in the vicinity of the active site of the HBV enzyme and the exocyclic double bond of the inhibitor can disfavor the posttranslocated complex. This model is consistent with our sitespecific footprinting experiments with HIV-1 RT that showed marked reductions in the population of post-translocated complexes, whereas the stability of the pretranslocated complex was not affected. However, kinetic measurements revealed that the inhibition of nucleotide incorporation at position n ϩ 1 was inefficient and classic chain-termination was, therefore, unlikely to be a major mechanism of action for ETV.
Inhibition of nucleotide incorporation at position n ϩ 4 was 2-3 orders of magnitude more efficient than inhibition at n ϩ 1, suggesting that delayed chain termination represented an important mechanism of action. Mapping the RT-associated RNase H activity on the RNA template strand provided snapshots of complexes that exist at this advanced stage of DNA synthesis. The location of primary RNase H cuts revealed the existence of pre-and post-translocated configurations solely in the absence of ETV-MP. With primers containing ETV-MP at position n, the RNase H pattern lacked these polymerase-dependent cuts, whereas polymerase-independent cleavage increased concomitantly. The incorporated inhibitor does not appear to increase enzyme dissociation, the data rather show that the 3Ј-end of the primer was not properly positioned at the polymerase active site to allow binding and incorporation of the next nucleotide. Based on these findings, we propose a model in which the various complexes co-exist in equilibrium (Fig. 7). The incorporated ETV-MP appears to "repel" the 3Ј-end of the primer from the active site, which causes a bias toward polymerase-independent conformations.
Crystal structures of HIV-1 RT bound to primer-template substrates show important contacts with the first six 3Ј-terminal residues of the primer (53)(54)(55). ETV-MP, as part of the extended primer, interacts with residues that form the "minor grove binding track" or "translocation track" of HIV-1 RT.
Lys 263 , Gly 262 , and Lys 259 are specific amino acids that may collide with the inhibitor when the primer is extended by an additional 3 residues. Modeling studies suggest that ETV-MP may induce structural distortions within the helix that could likewise affect productive interaction with the enzyme (36). Moreover, nucleotide analogues that are conformationally locked in the A-form show similar pausing patterns at positions n ϩ 2 and n ϩ 3 (38). Thus, this region appears to be sensitive to subtle structural differences of the nucleic acid substrate, although the underlying mechanisms may not necessarily be identical. The structural reasons that help to explain the inhibitory effects of ETV-MP when present in the template strand are likewise unknown. RT pausing opposite ETV-MP suggests that the inhibitor may affect proper base pairing. However, structural studies are required to address these issues.
The resistance profile of ETV provided independent support for our conclusion that delayed chain termination was a dominant mechanism of inhibition. Resistance to NRTIs is based on two major mechanisms: substrate discrimination and ATP-dependent excision, respectively (40,41). M184V discriminates against NRTIs, including ETV (36), (Table 3). M184V can cause significant decreases in susceptibility to most approved NRTIs, with few exceptions, including AZT and tenofovir (18 -21). In contrast, TAMs were shown to decrease susceptibility to all approved NRTIs, albeit at different degrees (40,41). Increases in resistance often correlate with increases in excision. AZT is efficiently excised and TAMs confer high level resistance to this drug, whereas 3TC and emtricitabine are to some degree protected from excision, and multiple TAMs are required to cause relevant levels of resistance in these cases (56). However, ETV appears to be fully susceptible in the context of HIV-1 variants containing four TAMs (36). Conversely, we show that ETV-MP was effectively excised at the point of incorporation with TAMs-containing HIV-1 RT. A possible explanation is that excision of ETV-MP does not play a significant role in biological settings.
We identified several factors that provide protection from excision of the incorporated ETV-MP. In contrast to classic chain termination, the subtle 7-fold reduction in the efficiency of nucleotide incorporation at position n ϩ 1 does not appear to give sufficient time for excision to occur. ATP-dependent excision of terminal nucleotides is at least 3 orders of magnitude slower than rates of nucleotide incorporation (57). Second, delayed chain termination at position n ϩ 3 diminishes the forward reaction as well as the reverse. Nucleotide incorporation can only occur post-translocation, whereas excision can only occur pretranslocation; however, neither of the two com-  Fig. 5 suggest that the RT enzyme can bind its nucleic acid substrate at various positions. Nucleotide binding can only occur when the complex exists in its post-translocated configuration, whereas the pyrophosphate analogue PFA stabilizes the pretranslocated complex. The data show that these polymerase-dependent conformations are in equilibrium with various polymeraseindependent conformations. A primer containing ETV-MP at position n followed by three natural nucleotides affects this equilibrium and favors polymerase-or excision-independent binding, which is indicated by the larger arrows.

TABLE 3
Summary of the single nucleotide incorporation kinetic constants for HIV-RT wild type and mutants at position n K m is in M dGTP or ETV-TP. k cat is in min Ϫ1 . SEL, selectivity, defined as a ratio of k cat /K m for dGTP over k cat /K m for ETV-TP. RES, resistance, defined as a ratio of selectivity of wild type (WT) over the selectivity of a mutant. S.D. was determined on the basis of at least three independent experiments. plexes can be detected at the point of DNA synthesis (Fig. 7). Nucleotide incorporation and also excision eventually occur at the terminal two residues, but reincorporation is again much more efficient than subsequent excision events that are required to remove ETV-MP at position n.

Enzyme
Together, these results demonstrate that efficient delayed chain-termination paired with insignificant inhibition at the point of incorporation can provide protection from a major resistance pathway (i.e. excision). Several studies have shown that ETV selects for M184V (29,(31)(32)(33), whereas TAMs do not appear to play a role in resistance (36). Thus, the combined biochemical and clinical findings warrant further investigation into the development of novel delayed chain terminators, related and unrelated to ETV, that show antiretroviral activity against a background of TAMs.