Site-specific Incorporation of Nucleoside Analogs by HIV-1 Reverse Transcriptase and the Template Grip Mutant P157S

Studies of drug-resistant reverse transcriptases (RTs) reveal the roles of specific structural elements and amino acids in polymerase function. To characterize better the effects of RT/template interactions on dNTP substrate recognition, we examined the sensitivity of human immunodeficiency virus type 1 (HIV-1) RT containing a new mutation in a “template grip” residue (P157S) to the 5′-triphosphates of (−)-β-2′,3′-dideoxy-3′-thiacytidine (3TC), (−)-β-2′,3′-dideoxy-5-fluoro-3′-thiacytidine (FTC), and 3′-azido-3′-deoxythymidine (AZT). A primer extension assay was used to monitor quantitatively drug monophosphate incorporation opposite each of multiple target sites. Wild-type and P157S RTs had similar catalytic activities and processivities on heteropolymeric RNA and DNA templates. When averaged over multiple template sites, P157S RT was 2–7-fold resistant to the 5′-triphosphates of 3TC, FTC, and AZT. Each drug triphosphate inhibited polymerization more efficiently on the DNA template compared with an RNA template of identical sequence. Moreover, chain termination by 3TC and FTC was strongly influenced by template sequence context. Incorporation of FTC and 3TC monophosphate varied up to 10-fold opposite 7 different G residues in the DNA template, and the P157S mutation altered this site specificity. In summary, these data identify Pro157 as an important residue affecting nucleoside analog resistance and suggest that interactions between RT and the template strand influence dNTP substrate recognition at the RT active site. Our findings are discussed within the context of the HIV-1 RT structure.


Studies of drug-resistant reverse transcriptases (RTs) reveal the roles of specific structural elements and amino acids in polymerase function. To characterize better the effects of RT/template interactions on dNTP substrate recognition, we examined the sensitivity of human immunodeficiency virus type 1 (HIV-1) RT containing a new mutation in a "template grip" residue (P157S) to the 5-triphosphates of (؊)-␤-2,3-dideoxy-3thiacytidine (3TC), (؊)-␤-2,3-dideoxy-5-fluoro-3-thiacytidine (FTC), and 3-azido-3-deoxythymidine (AZT).
A primer extension assay was used to monitor quantitatively drug monophosphate incorporation opposite each of multiple target sites. Wild-type and P157S RTs had similar catalytic activities and processivities on heteropolymeric RNA and DNA templates. When averaged over multiple template sites, P157S RT was 2-7-fold resistant to the 5-triphosphates of 3TC, FTC, and AZT. Each drug triphosphate inhibited polymerization more efficiently on the DNA template compared with an RNA template of identical sequence. Moreover, chain termination by 3TC and FTC was strongly influenced by template sequence context. Incorporation of FTC and 3TC monophosphate varied up to 10-fold opposite 7 different G residues in the DNA template, and the P157S mutation altered this site specificity. In summary, these data identify Pro 157 as an important residue affecting nucleoside analog resistance and suggest that interactions between RT and the template strand influence dNTP substrate recognition at the RT active site. Our findings are discussed within the context of the HIV-1 RT structure.
Reverse transcriptase (RT) 1 converts the human immuno-deficiency virus type 1 (HIV-1) plus-stranded RNA genome into double-strand DNA through the complex process of reverse transcription (1). Common HIV-1 therapies employ nucleoside analogs that are metabolized to their active 5Ј-triphosphates in vivo and are incorporated into viral DNA by RT, terminating DNA synthesis (2,3). However, the efficacy of nucleoside-based chemotherapy is significantly reduced by the emergence of drug-resistant HIV-1 variants containing mutations in RT that confer reduced susceptibility to nucleoside analogs (2,3).
Feline immunodeficiency virus (FIV) has been developed as a model for studying HIV-1 pathogenesis (13) and drug resistance (14). FIV RT is similar to HIV-1 RT in amino acid sequence, physical properties, catalytic activities, and nucleoside analog susceptibility (15)(16)(17)(18). Moreover, a valine or threonine substitution at Met 183 of FIV RT, the residue analogous to HIV-1 Met 184 (19,20), confers resistance to 3TC (18). Recently, a new variant of FIV resistant to 3TC and 3Ј-azido-3Ј-deoxythymidine (AZT) was identified (21). Resistance was attributed to a novel proline to serine mutation at position 156 in FIV RT. The analogous position in HIV-1 RT, Pro 157 , is one of several residues that compose the template grip, a DNA polymerase structural motif that interacts with the template strand (7)(8)(9). Hence, the FIV P156S mutant identified a new region of RT that affects active site substrate discrimination. This is interesting because it implies that template interactions away from the active site influence dNTP substrate recognition.
In this work, the sensitivities of purified HIV-1 RTs (wildtype (WT), P157S, and M184V) to nucleoside analogs were examined as a means to address the effects of RT/primertemplate interactions on substrate selection. Primer extension assays were used to detect quantitatively drug monophosphate incorporation opposite each of multiple sites on heteropolymeric DNA and RNA templates. We found that P157S confers moderate resistance to 3TC-5Ј-triphosphate (3TCTP) and FTC-5Ј-triphosphate (FTCTP) and low resistance to AZT-5Ј-triphosphate (AZTTP). We also found that the levels of 3TC and FTC monophosphate incorporation by HIV-1 RT vary at different template sites and that this site specificity is altered by mutation of Pro 157 in the template grip. These findings are discussed in the context of the recently published structure of an HIV-1 RT catalytic complex (9) and suggest that interactions between the RT template grip and the template affect dNTP substrate recognition at the polymerase active site.
Cloning, Mutagenesis, and Purification of Wild-type and Mutant HIV-1 RT Heterodimers-The coding region of each WT RT subunit (nucleotides (nt) 2551-4229 for p66 and nt 2551-3869 for p51) was amplified from the infectious HIV-1 clone pNL4-3 (Ref. 26; a kind gift of Dr. Arnold Rabson, New Jersey Center for Biotechnology and Medicine, Piscataway, NJ) using polymerase chain reaction (PCR; Ref. 27). The following oligonucleotide primers were used: 5Ј-end of p66 and p51, 5Ј-ACTAGTGAATTCATGCCCATTAGTCCTATTGAGAC-3Ј; 3Ј-end of p66, 5Ј-CTGGAGAAGCTTTCACTATAGTACTTTCCTGATTCCAG-3Ј; 3Ј-end of p51, 5Ј-CTGGAGAAGCTTTCACTAGAAAGTTTCTGCTCCT-ATTA-3Ј. Recognition sites for EcoRI and HindIII are underlined; bold nucleotides are the start codon (5Ј-end oligonucleotide) and stop anticodons (3Ј-end oligonucleotides). PCR conditions were 30 cycles at 94°C, 1 min; 60°C, 1.5 min; and 72°C, 1.5 min and were carried out in 10 mM Tris-HCl, pH 8.0, 50 mM KCl, 2 mM MgCl 2 , 200 M each dNTP, 200 pmol each primer, 75 ng of double-strand template, and 2.5 units of Taq DNA polymerase. PCR products were purified through 2% low melting temperature agarose gels, digested with EcoRI and HindIII, and ligated into the corresponding sites in pKK223-3 using standard protocols (27). The entire coding region for each subunit was sequenced at the University of Utah Sequencing Core Facility to ensure no errors were introduced during PCR. The p66 coding region DNA was then excised at the EcoRI and HindIII sites and subcloned into M13mp19, which was used to generate uracilated DNA for site-directed mutagenesis (28). The following mutagenic oligonucleotides were used: 5Јphosphate-ATCAATACGTGGATGATTTG-3Ј, to change methionine 184 to valine (M184V) and 5Ј-phosphate-AAGGATCATCAGCAATATT-C-3Ј, to change proline 157 to serine (P157S; mutagenic nts are italicized). After mutagenesis, a 934-nt EcoRI/AgeI fragment was removed and used to replace the corresponding WT fragment in the pKK223-3 p66 and p51 expression clones to generate RT expression clones of WT, P157S, and M184V for each RT subunit.
Each of the 6 clones was expressed in Escherichia coli DH5␣I Q (Life Technologies, Inc.), and RT p66/p51 heterodimers were purified as described previously (22) with several modifications. The concentration of isopropyl-␤-D-thiogalactopyranoside used to promote RT expression was raised to 30 M, but the conditions for cell growth and harvesting cells were unchanged. A total of 10 g of E. coli cell paste consisting of 3 g of p66 paste and 7 g of p51 paste was used for purification. The resulting lysate had a 2-fold molar excess of p51 relative to p66 to facilitate the preferential association of p66/p51 heterodimers rather than p66/p66 homodimers (29). The lysate was centrifuged as described (22), and the supernatant was desalted by dialysis against buffer M (50 mM Tris-HCl, pH 7.0, 1 mM EDTA, 1 mM dithiothreitol, 0.1% Nonidet P-40, and 10% glycerol) at 4°C (22). The desalted, cleared lysate was loaded onto a 400-ml DEAE-cellulose column (Whatman) as described (22). RT activity eluted in the void volume in buffer M, and peak protein-containing fractions were pooled and loaded onto a buffer M-equilibrated heparin column (POROS 20 HE1, 10 ϫ 0.46 cm) at 5 ml/min using a BioCAD SPRINT perfusion chromatography system (Perseptive Biosystems, Framingham, MA). The column was washed with 4 column volumes of buffer M, and RT was eluted with a 12-column volume linear gradient of 0 -500 mM NaCl in buffer M. Excess p51 subunit eluted first (ϳ210 mM NaCl), followed by p66/p51 heterodimers at approximately 275 mM NaCl. The peak fractions were combined, dialyzed at 4°C against buffer M, and then loaded onto a buffer M-equilibrated strong cation exchange column (POROS 20 S, 10 ϫ 0.46 cm) at 5 ml/min using the BioCAD SPRINT. This column was washed with 5-column volumes of buffer M, and RT was eluted with a 10-column volume linear gradient of 0 -1000 mM NaCl in buffer M. RT consisting of equal molar amounts of p66 and p51 eluted near the start of the gradient at approximately 35 mM NaCl. Purity was approximately 95% as determined by Coomassie Bluestained SDS-polyacrylamide gel electrophoresis (PAGE; data not shown). The expression levels, chromatographic behavior, and yields of WT and mutant RTs were very similar, suggesting that the mutant RTs were properly folded. All three RT preparations were free of detectable 3Ј-5Ј exonuclease activity (data not shown). RT active sites were determined as described previously (23); each p66/p51 preparation was nearly 100% active (data not shown).
Processivity Assay-HIV-1 oligonucleotide primers (primer 4737 and primer 4670 for the DNA and RNA template, respectively), minusstrand pHIV-pol DNA and plus-strand pol RNA transcripts were created as described previously (22). 5Ј-32 P-End labeling of the oligonucleotide primer, annealing of primer-templates, and processivity reactions were carried out essentially as described (22,23,30) in 15-l volumes except the primer-template concentration was 10 nM, and concentrations of each RT were varied between 0 and 30 nM. After product resolution by 7 M urea, 8% PAGE, products were visualized with a Molecular Dynamics PhosphorImager and quantitated using IMAGEQUANT software. Values of k cat were calculated from steadystate reactions where product formation was linearly proportional to RT concentration.
Drug Susceptibility Assay-For DNA-templated polymerization, a synthetic DNA oligonucleotide 20 nt long (20-mer; Ref. 31) was 5Ј-endlabeled with [␥-32 P]ATP and hybridized to a 46-nt DNA oligonucleotide (46-mer; Ref. 31) as described previously (22). Primer extension reactions (15-l volume) contained 10 nM primer-template and 50 nM RT in buffer with 25 mM Tris-HCl, pH 8.0, 30 mM KCl, and 2 mM dithiothreitol. After a 5-min preincubation at 25°C, reactions were started by the addition of 10 mM MgCl 2 , 2 or 4 dNTPs, and 0 -200 M FTCTP, 3TCTP, or AZTTP (see figure legends for specific concentrations). After 10 min at 37°C, the reactions were stopped by the addition of EDTA to 50 mM final concentration. A portion of each reaction was mixed with formamide loading dye (27), resolved by 7 M urea, 16% PAGE, and visualized and quantitated as described for the processivity reactions.
For RNA-templated polymerizations, a 49-nt RNA was generated by in vitro transcription. Duplex DNA was first prepared by hybridization of complementary (ϩ) and (Ϫ) strand synthetic DNA oligonucleotides 66 nt long containing a T7 RNA polymerase promoter. The sequence of the ϩ strand oligonucleotide was 5Ј-TAATACGACTCACTATAGGGTA-TGGTACGCTGGACTTTGTGGGATACCCTCGCTTTCCTGCTCCTG-3Ј. The oligonucleotides were hybridized at a concentration of 0.1 mg/ml in 40 mM Tris-HCl, pH 8.0, and 100 mM KCl at 95°C for 5 min followed by cooling to room temperature. The resultant duplex DNA was precipitated with sodium acetate and ethanol (27), dissolved in RNase-free water to a concentration of 1.5 mg/ml, and used in in vitro transcription reactions according to instructions from Promega. This generated a 49-nt RNA with the same sequence as the 46-nt DNA template described above plus 3 additional nts at the 5Ј-end (5Ј-GGG . . . -3Ј). The RNA was annealed to the 20-mer primer, and drug sensitivity reactions were carried out as described above except that RNasin was included in the reactions (1 unit/l).
Calculation of Chain Termination Probabilities-Incorporation of a nucleotide analog lacking a 3Ј OH terminates primer extension. Therefore, the proportion of primers that terminate at a given site relative to the total amount of primer elongated up to and beyond that site defines the probability of incorporating a chain terminating nucleotide. We call this chain termination probability. A similar parameter was previously used to describe pauses in DNA synthesis in steady-state reactions (22,32). Chain termination probabilities were calculated from band intensities quantified using IMAGEQUANT software after visualization by a Molecular Dynamics PhosphorImager.

RESULTS
Recent studies of the model AIDS virus, FIV, identified P156S as a novel RT mutation that confers low level resistance to both 3TC and AZT (21). Alignment of FIV Pro 156 with the sequence and crystal structure of HIV-1 RT indicates that this residue is part of the template grip region of the enzyme (7)(8)(9)21). This suggests that dNTP analog discrimination in FIV is influenced by residues involved in template binding that do not directly contribute to the RT active site.
To determine whether the analogous residue (Pro 157 ) plays a similar role in HIV-1 RT and to characterize better the effect of template interactions on active site substrate discrimination, we examined the catalytic activity, processivity, and drug triphosphate susceptibility of purified recombinant P157S HIV-1 RT. A primer extension assay was used to measure polymerization and drug monophosphate incorporation at multiple sites within DNA and RNA templates. The biochemical properties of this template grip mutant were compared with those of WT HIV-1 RT and M184V RT, an active site mutant that exhibits both resistance to 3TCTP and increased dNTP incorporation fidelity.
Activity and Processivity of P157S HIV-1 RT-To determine if the P157S mutation affects RT activity or processivity, primer extension reactions on HIV-1 ϩ strand RNA and Ϫ strand DNA templates were performed (Fig. 1). Product analysis by urea-PAGE revealed that polymerization terminated at distinct sites (termed "pause sites," indicated as R4 -R9 and D1-D3 (22) on the HIV-1 RNA and DNA templates, respectively). At low RT concentrations (0.5-5 nM; Fig. 1, lanes 2-4), the pausing patterns in each reaction set remained essentially unchanged as RT concentrations were increased, demonstrating that the reactions were in steady state and polymerization products resulted from processive synthesis initiated at the 20-mer primer (22).
WT, P157S, and M184V RTs had similar processivities on the HIV-1 templates. The weighted average processivity of each RT was approximately 25 nt on the RNA template (Fig.   1A) and 66 nts on the DNA template (Fig. 1B). With the exception of M184V RT, the k cat values calculated from these steady-state data were approximately equal for each RT on the DNA and RNA templates (data not shown). The RNA-directed DNA synthesis activity of M184V RT was 30% lower than the other RTs. Taken together, these data show that the P157S mutation minimally affects HIV-1 RT activity and processivity on both RNA and DNA templates.
Primer Extension Assay for Incorporation of Chain Terminating Nucleotide Analogs-Purified FIV P156S RT is resistant to the 5Ј-triphosphates of FTC, 3TC, and AZT when assayed on homopolymeric templates (21). Although assays on homopolymers provide a useful measure of overall drug sensitivity, these templates are somewhat artificial and do not yield information about the effects of template sequence context on drug incorporation.
To address these issues in our study of HIV-1 RT mutants, we employed a primer extension assay that detects the incorporation of chain terminating nucleotide analogs opposite multiple sites on heteropolymeric RNA and DNA templates (Fig.  2). In this assay, preformed 5Ј-32 P-labeled primer-templates are incubated with excess HIV-1 RT in the presence of fixed low concentrations of normal dNTPs and increasing concentrations of drug triphosphate. The resulting primer extension products are quantified by PhosphorImaging after resolution on urea-PAGE sequencing gels. In the absence of drug triphosphate, the hybridized 5Ј-32 P-labeled primers are extended nearly quantitatively to full-length products (or to defined shorter products when one or more normal dNTPs are omitted). In contrast, when reactions contain chain terminating drug triphosphates (e.g. AZTTP or FTCTP), template-directed incorporation of the drug causes termination of primer growth and accumulation of product bands with lengths determined by the sites of drug incorporation. As drug triphosphate concentration is increased, there is a concomitant increase in chain terminated products until saturation conditions are reached. The amount of drug triphosphate required to cause chain termination is a measure of the ability of RT to polymerize drug monophosphate into the growing primer strand. Wild-type RT catalyzes chain termination at relatively low drug triphosphate concentrations, whereas RT mutants resistant to drug triphosphates require higher concentrations of the inhibitors to achieve equivalent levels of chain termination (see below).
Susceptibility to FTCTP and 3TCTP-To evaluate whether P157S RT confers FTCTP resistance, primer extension reactions on DNA and RNA templates were carried out in the presence of fixed amounts of dATP and dCTP and increasing concentrations of FTCTP (Fig. 2, B and C). In the reactions lacking FTCTP, the primer was extended the expected 6 nts plus an additional 1 to 2 nts (Fig. 2, B and C, lanes 1; Note: addition of these extra nts is likely due to nucleotide misincorporations and/or trace contaminating dNTPs which are not expected to affect drug incorporation at the G positions in the template). As the concentration of FTCTP was increased in each reaction with WT RT, chain termination also increased 1 and 3 nucleotides from the primer, presumably due to incorporation of FTC monophosphate (FTCMP; Fig. 2B, WT lanes 2-8, bands labeled G1 and G2). As expected for a chain terminator, FTCMP incorporation was also accompanied by a corresponding decrease in the amount of full-length product formed. P157S RT appeared moderately resistant to FTCTP as evidenced by a shift in the distribution of products (lower yields of termination products at G 1 and G 2 and concomitant higher yields of full-length products; Fig. 2B, P157S lanes 2-8). This shift is most evident at the lower FTCTP concentrations (lanes 2-5). Similar results were seen on the RNA template (Fig. 2C) and in reactions containing 3TCTP in place of FTCTP, indicating that P157S RT may be broadly resistant to oxathiolane nucleoside analogs (data not shown). M184V RT formed almost no detectable chain terminated products at all FTCTP concentrations tested (Fig. 2, B and C, M184V lanes 2-8).
To quantitate the degree of resistance conferred by the P157S mutation, the fraction of full-length product formed in each reaction was determined and plotted as a function of drug concentration (Fig. 2D). This dose-response curve clearly shows the resistance of P157S RT to FTCTP. For example, at 1 M FTCTP on the DNA template (Fig. 2D), only 55% of the extended primer was polymerized to full-length product by WT RT (relative to control reactions lacking FTCTP). At this same FTCTP concentration, P157S RT incorporated less FTCMP than WT RT; as a result, nearly 85% of the extended primer was polymerized to full length. M184V RT was strongly resistant to FTCTP and polymerized 100% of the extended primers to full length at this FTCTP concentration. All data sets on the DNA and RNA templates demonstrated that the relative sensitivities of the RTs to FTCTP and 3TCTP were WT Ͼ P157S Ͼ Ͼ M184V RT (Fig. 2D and data not shown).
IC 50 values calculated from Fig. 2D and other dose-response curves not shown are summarized in Table I (IC 50 is defined as the concentration of drug that inhibits full-length product formation by 50%). The IC 50 values of FTCTP for P157S RT were 6.7 and 5.5 M on the RNA and DNA templates, respectively. Thus P157S RT was 2-3-fold resistant to FTCTP compared with WT RT. M184V RT, on the other hand, was approximately 50-fold resistant to FTCTP on these templates (data not shown). All three RTs were more sensitive to FTCTP than 3TCTP. This is consistent with a previous report (33) and shows that HIV-1 RT can distinguish between dNTPs differing only in an electronegative fluorine substituent at position 5 of the pyrimidine ring. Interestingly, this discrimination was slightly exaggerated by the P157S mutation. With WT RT the IC 50 values for 3TCTP (5-8 M) were about 3 times higher than those for FTCTP (2-3 M). Introduction of the P157S mutation increased this difference making the IC 50 values for 3TCTP (30 -50 M) about 7 times higher than for FTCTP (6 -7 M). The IC 50 values of FTCTP and 3TCTP were consistently higher for both WT and P157S RTs on the RNA template compared with DNA.
Susceptibility to AZTTP-P156S RT from FIV is cross-resistant to AZTTP (21). To determine if the corresponding HIV-1 RT mutant is also cross-resistant to AZTTP, primer extension reactions containing fixed amounts of all 4 normal dNTPs and increasing concentrations of AZTTP were carried out on the RNA and DNA templates (Fig. 3). In the absence of AZTTP, each RT extended the primer to the template end plus one additional nt (Fig. 3, lanes 1). This extra dNMP was likely added through a non-templated mechanism (34) that is not expected to affect drug incorporation. As increasing concentrations of AZTTP were added to the reactions, increasing amounts of chain termination occurred due to AZT monophosphate (AZTMP) incorporation opposite the template A residues 8, 15, and 20 nts from the primer (Fig. 3, lanes 2-8). Quantitation of the full-length products revealed that P157S RT synthesized slightly more full-length DNA relative to WT RT at all AZTTP concentrations below 5 M on the DNA template but not the RNA template (data not shown). M184V and WT RTs were  equally sensitive to AZTTP in these reactions, in agreement with the results of others (35,36). IC 50 values from these data are shown in Table I. For each RT, the AZTTP IC 50 values were 10 -100 times lower than those for FTCTP and 3TCTP. P157S RT was 2-fold resistant to AZTTP on the DNA template but not the RNA template. Thus it appears that, like FIV RT containing a serine at position 156, HIV-1 P157S RT has a low level of cross-resistance to AZTTP.
Site Specificity of FTCMP/3TCMP Incorporation-In the experiments measuring sensitivity to FTCTP (Fig. 2) and 3TCTP (data not shown), drug monophosphate incorporation by WT RT at DNA template sites G 1 and G 2 did not appear equal (Fig. 2B, WT lanes 2-4). To quantify this difference, the probability of chain termination at each of these sites was determined as a function of FTCTP concentration (Fig. 4). This revealed a strong preference for drug monophosphate incorporation at template site G 2 . While ϳ2 M FTCTP was sufficient to cause 50% chain termination at G 2 , as much as 25 M was required to achieve a similar level of termination at G 1 . Based on these EC 50 values (concentration of drug triphosphate that results in 50% chain termination probability), we estimate that FTCMP incorporation by WT RT occurred approximately 10 times more readily at G 2 than at G 1 .
To determine whether FTCMP or 3TC monophosphate (3TCMP) incorporation varies at other G residues, we conducted primer extension assays on the same 46-mer DNA template ( Fig. 2A) but now in the presence of all four normal dNTPs. (Note: only dATP and dCTP were included in the assays summarized in Fig. 2 and Table I.) As expected, all three RTs efficiently extended the primer to the end of the template in the absence of drug triphosphate (Fig. 5, lanes 1). In reactions containing FTCTP or 3TCTP, the amount of chain termination at the seven template G sites increased in proportion to the amount of drug added, and the yields of full-length product correspondingly decreased (Fig. 5, lanes 2-8; termination sites labeled G1, G2, G3, etc.). A comparison of chain termination probabilities at each of the G residues in the DNA template showed that the levels of FTCMP and 3TCMP incorporation were site-specific. WT RT preferentially incorporated FTCMP opposite G 2 , G 3 , and G 6 with lower incorporation occurring at the other sites (Fig. 6A). P157S RT also exhibited site-specific preferences for drug monophosphate incorporation; however, these preferences differed somewhat from those of the WT RT (compare Fig. 6, A and B). Thus, the resistance of P157S RT to FTCTP is due largely to reduced incorporation at sites G 2 and G 3 ; EC 50 values at these sites were ϳ2 M for WT RT and ϳ14 M for P157S RT (data not shown). Resistance at sites G 4 and  1-8, respectively). Full-length products and the sites of AZTMP incorporation are indicated on the left and right, respectively. Product yields from these and other data not shown were determined by phosphorimaging (data not shown).

FIG. 4. Unequal incorporation of FTCMP by WT RT.
Chain termination probability was calculated as described under "Experimental Procedures" and is plotted as a function of FTCTP concentration at DNA template sites G 1 (circles) and G 2 (squares). Data are from reactions described in Fig. 2B. Each point is the average Ϯ S.D. of 2-5 independent determinations. Error bars less than 0.02 are too small to be visible.  lanes 1-8). 3TCTP concentrations were 0, 1, 3, 6, 10, 50, 100, and 200 M (B, lanes 1-8). Reaction products were resolved by 7 M urea, 16% PAGE and visualized by phosphorimagery. A representative image is shown. Full-length products and FTCMP/3TCMP incorporation sites are indicated. IC 50 values calculated from these data showed that P157S RT exhibited FTCTP/3TCTP resistance comparable to the values reported in Table I and that M184V RT was approximately 50and 80-fold resistant to FTCTP and 3TCTP, respectively. G 5 was relatively modest (1.5-2-fold), whereas at G 1 , G 6 , and G 7 both the P157S and WT RTs incorporated FTCMP almost equally (Fig. 6).
This experiment leads to two important conclusions. First, the efficiency of drug monophosphate incorporation by HIV-1 RT is dependent on template site. Second, a single amino acid change in the RT template grip (P157S) changes this site dependence. Taken together, these data indicate that interactions between RT and the template strand influence substrate recognition at the RT active site. DISCUSSION FIV RT containing a serine substitution for proline at position 156 is resistant to 3TCTP and AZTTP (21). The analogous residue in HIV-1 RT, Pro 157 , is part of the template grip (7-9), suggesting that RT/template interactions may influence substrate recognition at the RT active site. To determine if HIV-1 P 157 plays a role in dNTP analog discrimination, we monitored primer extension reaction products in the absence or presence of drug triphosphates (Fig. 2). We found that P157S RT is resistant to FTCTP, 3TCTP, and AZTTP (Figs. 2 and 3 and Table I). Moreover, our data show that the amount of FTCMP and 3TCMP incorporated varies up to 10-fold at different template sites and that this site specificity is altered by the P157S mutation (Figs. 4 -6).
A steady-state processivity assay showed that the polymerase activities of WT, P157S, and M184V RTs were comparable on both RNA and DNA templates (Fig. 1). Measures on homopolymeric primer-templates also showed that the three RTs had comparable polymerase activity (37). The only exception was M184V RT that had approximately 30% lower activity on the RNA template, a value that agrees well with other reports (33,35,38,39). The processivities of each RT were also virtually identical and similar to our previous observations using recombinant WT RT from the HXB2 strain of HIV-1 (22). Thus, the P157S mutation does not greatly affect overall polymerase activity. This is consistent with our recent observation that HIV-1 containing the P157S mutation in RT replicates in cultured HeLa cells at near wild-type levels (37). Our finding that M184V RT has a processivity similar to that of WT RT agrees with the data of Pandey et al. (11) and Wilson et al. (33) but contrasts with those of Back et al. (38) who reported a reduced processivity for M184V RT. This apparent discrepancy may be due to differences in primer-template sequences and dNTP concentrations used in these processivity assays (38).
To determine the susceptibility of the RTs to drug triphosphates, we developed a primer extension assay that quantitatively monitors drug monophosphate incorporation opposite each of multiple target sites within the template (Fig. 2). IC 50 values obtained from these experiments revealed that P157S RT is 2-7-fold resistant to FTCTP and 3TCTP and 2-fold resistant to AZTTP (Table I). This is very similar to data obtained with purified FIV P156S RT (21). Interestingly, HIV-1 containing the P157S mutation in RT is slightly hypersensitive to AZT in culture (37). The basis for this discrepancy is not known, although discrepancies between viral susceptibility to AZT and susceptibilities of purified RTs to AZTTP are well documented (40,41). The sensitivity of purified P157S RT to 3TCTP/FTCTP closely agrees with phenotypic data from the viral clone containing the P157S mutation (37). Taken together, these results demonstrate that the serine substitution at position 157 in HIV-1 RT confers resistance to FTC/3TC as predicted from work with FIV (21). Moreover, our results show that FIV variants resistant to nucleoside analogs are useful for predicting the contribution of analogous HIV-1 RT residues to drug resistance.
The proline at position 157 in HIV-1 RT is highly conserved in retroviruses, retrotransposable elements, retrons, and hepatitis B virus (5,6), suggesting that it is structurally and/or functionally important. In HIV-1 RT crystal structures, Pro 157 lies near the N terminus of ␣ helix E (residues 155-174) in a region of the RT template grip that is proximal to but not directly part of the catalytic active site (7)(8)(9). Pro 157 is directly involved in template binding through the minor groove and makes van der Waals contacts with the sugar and base of the template strand two base pairs "behind" the incoming dNTP (Fig. 7). Pro 157 does not appear to contact directly the incoming dNTP, indicating that the resistance of P157S RT to drug triphosphates does not result from direct interactions between the 157 position and the incoming dNTP substrate analog. Therefore, the effects of P157S on FTC/3TC monophosphate incorporation are likely indirect and may involve one or more alternative mechanisms. One possibility is that the P157S mutation mediates subtle structural rearrangements of other important amino acid residues due to the increased conformational flexibility of the peptide backbone imparted by the Ser replacement (42). Pro 157 is within 3-4 Å of Met 184 and Tyr 115 , residues that contact the 3Ј nucleotide of the primer and the incoming dNTP, respectively (9). Positional changes in these residues induced by the Pro to Ser substitution at 157 could influence the susceptibility of RT to nucleoside analogs. Alternatively, the serine substitution may alter the nature of the contact between the 157 position of RT and the template strand, changing the relative positioning of protein and nucleic acid components at the polymerase active site (12). Changes in the positions of catalytically relevant amino acids and/or the template strand may alter the active site geometry, resulting in decreased utilization of drug triphosphates as substrates.
Evidence for the involvement of RT/template interactions in dNTP substrate selectivity is provided by comparing the inhibition of RNA-directed versus DNA-directed polymerization by dNTP analogs. Each drug triphosphate inhibited polymerization more efficiently on the DNA template relative to an RNA template of identical sequence ( Fig. 2 and Table I). This is consistent with the observation of Wilson et al. (33) that K i values for the inhibition of WT and M184V HIV-1 RTs by 3TCTP and FTCTP are 30 -50% higher on RNA templates. Thus, the nature of the template (RNA versus DNA) influences the ability of HIV-1 RT to incorporate nucleoside analogs. It is not known if this is due to global structural differences between RNA and DNA templates and/or subtle differences imparted by the ribose 2Ј-OH at the polymerase active site, nor is it known if replicating virus shows a similar templatebiased drug sensitivity.
Additional evidence that RT/template interactions affect nucleotide selection comes from the experiment showing that both WT and P157S RTs exhibit site specificity in the levels of FTCMP/3TCMP incorporation (Figs. 4 -6). Site-specific differences in the incorporation levels of other nucleoside analogs by RT have also been reported (43,44). Interestingly, changing Pro 157 to Ser significantly altered the ability of RT to incorporate FTCMP and 3TCMP at two of the seven template G sites, thus changing the site specificity of the enzyme. Therefore, the susceptibility of HIV-1 RT to drug triphosphates is strongly influenced both by template sequence context and the nature of the amino acid residue at position 157 in the template grip. There is no obvious correlation between the identities of the 5Ј or 3Ј nucleotide immediately adjacent to any template G site and the amount of FTCMP/3TCMP incorporated. This suggests that larger portions of flanking sequence affect drug monophosphate incorporation by RT. This is reminiscent of the well established observation that RT fidelity is also highly dependent on template sequence context (45,46). It will be interesting to determine whether fidelity parallels site-specific drug incor-poration at these same sites.
Previous reports have alluded to the connection between the HIV-1 template grip and polymerase active site. For example, residue Glu 89 contacts the same template sugar moiety as Pro 157 (9), and mutation of Glu 89 to Gly confers resistance to nucleoside analogs and the pyrophosphate analog phosphonoformic acid (47) and increases dNTP insertion fidelity (48). Mutations in several other template grip residues also confer resistance to phosphonoformic acid ( Fig. 7; summarized in Refs. 3 and 4). The observation that template grip mutations impart resistance to both nucleoside and pyrophosphate analogs suggests that the template grip influences the organization and selectivity of the active site/dNTP binding pocket. Additional experiments are required to understand fully how the HIV-1 template grip contributes to active site discrimination.
Studies of other DNA polymerases suggest that their template grips also influence dNTP substrate recognition. Protein structure alignments of HIV-1 RT, Pol I family polymerases (Klenow, Taq, T7, and Bst), and a Pol-␣ family polymerase (RB69) show remarkable conservation of structure in the palm subdomains of these proteins including their template grips (49 -55). Moreover, amino acid residues known to affect fidelity and/or nucleoside drug susceptibility in HIV-1, E. coli, T4 phage, herpes simplex virus, and hepatitis B virus polymerases map to this region (56 -63). Hence, the template grip contributes to dNTP discrimination in evolutionarily diverse polymerases.
In summary, our studies identify Pro 157 as an important HIV-1 RT template grip residue that influences 3TC/FTCTP recognition. Three lines of evidence show that RT/template interactions influence active site discrimination as follows: 1) drug monophosphate incorporation is not equal on RNA and DNA templates of identical sequence; 2) drug monophosphate . The RT peptide backbone is represented as a "worm" diagram, and the DNA primer (light green) and template (dark blue) strands are in wire frame format. DNA position "zero" (0) corresponds to the template A residue that forms a base pair with the incoming dTTP (dark green) at the RT active site. DNA residues labeled Ϫ1, Ϫ2, Ϫ3, and Ϫ4 are 1-4 base pairs, respectively, "behind" the active site (i.e. 3Ј on the template strand), while template residues labeled ϩ1 and ϩ2 are 1 and 2 bases in "front" of the active site (i.e. 5Ј on the template strand). For clarity, only p66 RT residues Thr 58 -His 96 and Gln 145 -Ser 191 and DNA residues Ϫ4 through ϩ2 are shown. Pro 157 (red) contacts the Ϫ2 residue of the template strand. Other residues that interact with the template are indicated in gray (Phe 61 , Leu 74 , Asp 76 , Arg 78 , Asn 81 , Gln 91 , Leu 92 , Gly 93 , Ile 94 , Gln 151 , Gly 152 , and Lys 154 ; see Refs. 8 and 9). Met 184 (cyan), the active site carboxylate residue Asp 185 , and two bound Mg 2ϩ atoms (black) are also shown. This image was created by RasMol 2.7.1 for the Macintosh.
incorporation is template sequence-dependent; and 3) mutation of a residue known to interact with the template (Pro 157 ) changes the site specificity of drug monophosphate incorporation. These findings, together with recent polymerase fidelity studies (64), imply that the geometry of the RT active site responds to differences in template sequence and/or structure. The underlying mechanisms for this are not known. Specific interactions among amino acid side groups and template atoms may contribute. Changes in active site geometry propagated through subtle structural changes of the template grip may also be involved. Additional biochemical and structural studies are required to address these and other possible mechanisms.