Lys66 residue as a determinant of high mismatch extension and misinsertion rates of HIV-1 reverse transcriptase



V. R. Prasad, Albert Einstein College of Medicine, 1300 Morris Park Avenue, Golding 401, Bronx, NY 10461, USA

Fax: 718 430 8976

Tel: 718 430 2517



A major factor contributing to the high mutation rate of HIV-1 reverse transcriptase (RT) is its high propensity for misincorporation. Misincorporation requires both deoxyribonucleotide triphosphate (dNTP) misinsertion and the subsequent extension of the mismatched terminus thus formed. We hypothesized that Lys66 is a determinant of mismatch extension based on its position near the primer terminus. This hypothesis was tested by steady-state kinetic studies using wild-type HIV-1 RT and four Lys66 substitution mutants: Lys66Arg, Lys66Ala, Lys66Asn and Lys66Thr. The mismatch extension efficiency was reduced for all mutants, with Lys66Ala, Lys66Asn and Lys66Thr showing a four- to six-fold reduction compared with wild-type HIV-1 RT. Surprisingly, the nonconservative substitutions also led to large decreases in misinsertion efficiency, ranging from as low as three-fold to values much higher than 23-fold. Thus, the Lys66Arg mutant was akin to wild-type HIV-1 RT, whereas all nonconservative mutants displayed significantly decreased efficiency for both events. Our results suggest that Lys66, much like Lys65, is a determinant of both dNTP misinsertion and mismatch extension efficiency. While Lys65 is known to contact the γ-phosphate of incoming dNTP, the Lys66 side chain is in the vicinity of the primer terminus. However, our results suggest that both residues have a similar influence on dNTP misinsertion and mispair extension efficiencies of HIV-1 RT. When we tested the mutants for susceptibility to selected nucleoside analog and non-nucleoside analog drugs, similarly to Lys65Arg, the Lys66Ala and Lys66Asn mutants displayed mild resistance to the nucleoside analog drug 3′-azido-3′-deoxythymidine-5′-triphosphate (AZTTP).








2′-deoxyadenosine 5′-triphosphate


2′-deoxycytosine 5′-triphosphate


2′-deoxyguanosine 5′-triphosphate


deoxynucleotide triphosphate


2′-deoxythymidine 5′-triphosphate



f ext

mismatch extension efficiency

f ins

misinsertion efficiency


inhibitory concentration 50%


Integrated DNA Technologies


polynucleotide kinase


ribonuclease H


reverse transcriptase


template primer


tenofovir diphosphate


wild type


HIV-1 encodes reverse transcriptase (RT) (EC and EC, a multifunctional enzyme responsible for converting the single-stranded viral genomic RNA to a double-stranded DNA form, a process that is essential for viral replication and for establishing infection. Although it has many structural similarities to other high-fidelity replicative DNA polymerases, such as a structure resembling a right hand, with fingers, palm and thumb subdomains [1] and substrate-induced structural changes [2-4], HIV-1 RT is quite error prone with an error rate in the order of 3 × 10−5 mutations/nucleotide/cycle [5]. The relatively high error rate of HIV-1 RT, in conjunction with the rapid replication of the virus in vivo, leads to high genetic variation of the HIV-1 genome in patients, ultimately resulting in immune escape mutants and in drug-resistant variants [6, 7].

Several studies of HIV-1 RT have established a basic model for nucleotide incorporation [3, 8, 9]. The general mechanism of the polymerization process begins with the binding of the template-primer (t/p) to the enzyme (E) to form a binary complex (E: t/p). The incoming deoxyribonucleotide triphosphate (dNTP) is subsequently bound in a loose ‘open’ tertiary complex (E: t/p: dNTP), which undergoes a conformational change into a ‘closed’ tertiary complex ready to undergo the chemical step (E': t/p: dNTP) [3]. A conformational change has been identified as the rate-limiting step of dNTP incorporation [8, 9]. However, kinetic studies demonstrate that the extent to which the conformational change is rate limiting may differ [10], and it has been specifically suggested that in mismatch extension, conformational change may no longer be the rate-limiting step [11]. During the conformational change, the p66 fingers subdomain closes down on the incoming dNTP, helping to precisely align the 3′-OH of the primer, the α-phosphate of the dNTP and the polymerase active site for chemistry [12]. Upon formation of the new phosphodiester bond, pyrophosphate is released, leaving the t/p extended by one nucleotide and the RT bound such that the primer 3′-end occludes the nucleotide-binding site (Fig. 1) [13, 14]. This complex, referred to as a pretranslocation complex, can be stabilized in vitro by the presence of the pyrophosphate analog, foscarnet. In order to extend the primer with an additional nucleotide, the RT must translocate along the t/p and reposition itself such that the primer 3′-end resides in the priming site [13, 14], leaving the nucleotide-binding site [13, 14] free for nucleotide binding. This post-translocation complex (Fig. 1) can be stabilized in vitro using a 3′-chain-terminated primer in the presence of the next complementary nucleotide.

Figure 1.

Graphical representation of the positioning of the RT on matched and mismatched primer templates. Cleavage positions in the RNaseH active site are shown with stars, with the black star indicating the precise position of cleavage for each complex. The positioning of the priming (P) and nucleotide-binding (N) sites of the polymerase active site [13, 14], relative to the t/p are highlighted as green and pink rectangles respectively. The RT forms two complexes with matched primer termini: pretranslocation and post-translocation. Binding to a mismatched primer results in an additional complex (−2).

Discrimination to reduce error can occur either at nucleotide insertion or following insertion, by promoting removal of an incorrect nucleotide. It is well known that HIV-1 RT does not possess proofreading capability [15]; thus, error prevention must occur at the point of nucleotide incorporation. The incorporation of an incorrect nucleotide in a DNA strand is the result of two processes: misinsertion and mismatch extension. Misinsertion is the initial insertion of a noncomplementary nucleotide, whereas mismatch extension is the subsequent extension of the mismatched terminus thus formed. HIV-1 RT can efficiently perform both misinsertion and mismatch extension events [16, 17].

Examination of the HIV-1 RT–DNA–dNTP ternary structure implicates a number of amino-acid residues in the formation of the dNTP-binding pocket [12]. Many of the residues lining the dNTP-binding pocket are situated in the highly flexible β3–β4 loop of the fingers subdomain of the p66 subunit (Fig. 2). The β3–β4 loop plays a critical role in orienting and stabilizing the substrate for catalysis [18, 19]. This loop encompasses residues 60–75, which includes Lys70, Lys74, Lys65, Arg72 and Asp67. The three residues of the loop Arg72, Asp67 and Lys65 make contact with the α-, β- and γ-phosphates of the incoming nucleotide, respectively [12, 20-22].

Figure 2.

Positioning of Lys66Arg in the β3–β4 loop relative to the primer in the dNTP-binding site. The distance between the Lys66 and the phosphate backbone of the penultimate base pair is labeled with a red line and is measured using the distances tool under the structural analysis menu of Chimera [47]. The ultimate and penultimate nucleotides are indicated. The incoming dNTP is represented as a mesh surface, while the β3–β4 strand is shown in light gray. The relative positions of atoms were obtained from 1rtd [12].

Several residues in the β3–β4 loop are implicated in drug resistance as primary mutations, but only a few, including Leu74, Val75 and Lys65, have been identified as playing a role in misinsertion and mismatch extension. Leu74Val mutations have shown up to a six-fold reduction in misinsertion frequency, and a 15- to 26-fold reduction in mismatch extension frequency [23]. Matamoros et al. showed that the mutation Val75Ile increased the mismatch extension fidelity by three-fold while showing no apparent effect on misinsertion fidelity. The authors showed that the increased fidelity is caused by a decrease in nucleotide affinity in the presence of mismatched t/p [24]. Early work from our laboratory investigating the influence of Lys65Arg mutations on overall mutation rates and error specificity indicated a qualitative defect in both misinsertion and mismatch extension, as seen by a reduction in extension products on the gel-based single dNTP exclusion assay [18]. Further investigation into the influence of mutations at Lys65 showed that a Lys65Arg mutation conserves polymerase activity similar to that of the wild-type (WT) enzyme, whereas a nonconservative substitution (i.e. Lys65Ala) results in reduced RT catalytic efficiency (kcat/Km) [25] as well as decreased affinity for dNTP [26]. The increased fidelity of the Lys65Ala mutation appears to be directly caused by a cumulative effect of a reduction in kpol and increased dNTP Kd in both misinsertion and mismatch extension [26]. Although mutagenesis studies have investigated the role of a number of β3–β4 residues in fidelity [25, 27], a role for Lys66 in RT fidelity has not been investigated so far.

In the primary sequence of RT, Lys66 is in a cluster of three lysines in the fingers subdomain. Unlike Lys64, which points away from the polymerase active site in the 3D structure of a ternary complex, Lys66 lies adjacent to the phosphate of the penultimate primer base/nucleotide (Fig. 2) [12]. This conformation is strictly seen in the presence of t/p and dNTP, while in the absence of bound dNTP, the fingers subdomain is further from the active site [28, 29]. Owing to its location in the vicinity of the dNTP-binding pocket, we hypothesized that Lys66 could play a role in mismatch extension. The significance of this residue is further supported by its broad conservation among RTs of retroviruses and other retroelements [30]. Likewise, Lys66 is conserved in patient isolates. Mutations identified among patient isolates at this position include: Lys66Arg, Lys66Asn, Lys66Thr and Lys66Glu. These mutations are not known to confer drug resistance but rather occur at a low frequency in both drug-naïve and drug-resistant isolates [31, 32].

In the present study, we investigated the potential role of the Lys66 residue in RT fidelity by creating the substitutions Lys66Arg, Lys66Ala, Lys66Asn and Lys66Thr and testing them for both misinsertion efficiency (fins) and mismatch extension efficiency (fext). Our results showed that not only did the Lys66 substitutions display decreased fext, they were also poor at dNTP misinsertion. In addition, drug-susceptibility tests indicated that substitutions at position 66 confer low resistance to nucleotide RT inhibitors such as 3′-azido-3′-deoxythymidine-5′-triphosphate (AZTTP). Furthermore, we showed (using site-specific hydroxyl radical footprinting) that these substitutions do not alter the positioning of HIV-1 RT on the t/p, but rather, similarly to the WT enzyme, their positioning on the t/p substrate is affected in the presence of a mismatched primer terminus.


Mismatch extension fidelity

In order to examine the effect of the substitutions at Lys66 on the ability of RT to extend from a mismatched terminus, single-nucleotide incorporation assays were performed. Lys66Ala was used as a nonconservative substitution in addition to the three substitutions (Arg, Asn and Thr) observed in clinical isolates [31, 32]. WT and Lys66 substitution mutants of recombinant HIV-1 RT were used to extend either a matched (T:A) or a mismatched (G:A) terminus by incorporation of the next complementary nucleotide [2′-deoxyadenosine triphosphate (dATP)]. A broad range of nucleotide concentrations was tested (Fig. 3). Steady-state kinetic parameters were calculated using this single-nucleotide insertion assay, with reaction conditions adjusted such that < 30% primer extension was observed [33].

Figure 3.

Single-nucleotide mismatch extension assays for WT, Lys66Arg, Lys66Ala, Lys66Asn or Lys66Thr. The reactions were performed in the presence of 100 nm of γ-[32P] 5′-end-labeled mismatched primer annealed to a template, 10 nm enzyme and 0–800 μm dATP. The reactions were run for 2, 5 or 10 min and then stopped with a solution containing 95% formamide, 15 mm EDTA and 0.2 mg·mL−1 of Bromophenol Blue.

When studying extension from the matched terminus, the catalytic efficiency (kcat/Km) of extension was comparable for the WT and the Lys66Arg enzymes (Table 1A). While the nonconservative mutants (Lys66Ala, Lys66Asn and Lys66Thr) exhibited slightly lower extension efficiencies (up to two-fold), the data suggested that substitutions at position 66 have a minimal effect on the efficiency of extension from a matched terminus (Table 1A). However, the presence of a mismatch (G:A) at the primer terminus resulted in a greater decrease in kcat/Km, especially for the nonconservative mutants, which showed reductions of up to 12-fold when compared with WT RT (Table 1B). Thus, Lys66 substitutions hamper the innate ability of the enzyme to extend from a mismatched terminus. This effect appears to be caused both by an increase in substrate Km and a decrease in the overall kcat of the enzyme (Table 1B).

Table 1. Single nucleotide extension assay. ƒext is calculated as (kcat /Km)incorrect /(kcat /Km)correct. ∆ƒext represents the fold change compared with the WT enzyme. Data shown are mean and SD values from at least three independent experiments. Kinetic constants were calculated using graphpad prism
dNTPRTkcat (min−1) ± SEKm (nm) ± SEkcat /Km (nm−1·min−1) ± SE ƒ ext ƒext
A. Steady-state extension kinetics from a matched template primer
dATPWT6.88 ± 0.2934.51 ± 6.120.20 ± 0.03  
K66R14.21 ± 1.2154.80 ± 18.310.26 ± 0.07  
K66A6.95 ± 0.4050.61 ± 11.580.14 ± 0.03  
K66N6.33 ± 0.8485.35 ± 22.270.08 ± 0.02  
K66T14.51 ± 0.6486.56 ± 13.780.17 ± 0.02  
B. Steady-state extension kinetics from a mismatched template primer
dATPWT2.74 ± 0.43116.6 ± 60.10.024 ± 0.0090.12 
K66R2.99 ± 0.75124.5 ± 100.80.024 ± 0.0150.091
K66A0.87 ± 0.18326.4 ± 163.60.003 ± 0.00090.026
K66N0.80 ± 0.23412.8 ± 266.90.002 ± 0.00070.034
K66T1.37 ± 0.39414.6 ± 263.00.003 ± 0.0010.026

In order to elucidate the overall effect on mismatch extension fidelity in more detail, these data were expressed as fext, defined as [(kcat/Km)incorrect/(kcat/Km)correct]. The Lys66Arg substitution displayed an fext comparable with that of the WT enzyme, whereas the reduction in fext for the nonconservative mutants (Lys66Ala, Lys66Asn and Lys66Thr) was four- to six-fold (Table 1B).

Misinsertion fidelity

The efficiency of insertion of each nucleotide was measured using single-nucleotide insertion assays. Figure 4 illustrates the results from an assay in which primer was extended in the presence of the correct (dATP) or each incorrect nucleotide by the WT and mutant RTs.

Figure 4.

Gel illustrating single nucleotide misinsertion assays in the presence of 10 nm enzyme (WT, Lys66Arg, Lys66Ala, Lys66Asn or Lys66Thr) and 100 nm of matched t/p. The term ‘Enz’ is used to designate which of the enzymes are used in each reaction, ‘dNTP’ is used as a label for the four nucleotides, ‘P’ represents the unextended primer and ‘p + 1’ designates the primer extended by one nucleotide, and so forth. The reactions were performed in the presence of 10 nm enzyme and 0–1 mm dATP, dGTP, dCTP and dTTP. The reactions were run for 10 min and then stopped with a solution containing EDTA.

All enzymes tested (WT and Lys66 substitution mutants) efficiently inserted the correct nucleotide with similar efficiencies, with the most impaired RT being Lys66Asn, leading to a decrease of approximately three-fold in kcat/Km (Table 1A). Under conditions where 2′-deoxyguanosine triphosphate (dGTP) is inserted opposite a template T, resulting in a G:T mispair, all of the enzymes, except for Lys66Ala, showed similar efficiencies of misinsertion. The Lys66Ala effect was dramatic in this instance, resulting in a 23-fold decrease in kcat/Km. Interestingly, WT, Lys66Arg and Lys66Asn enzymes displayed increased G:T mispair catalytic efficiency (kcat/Km) compared with A:T. These experiments were repeated using a template with a different sequence, in which an additional nucleotide was inserted immediately 5′ to the repetitive CCCC (5′-CCCC-3′ to 5′-ACCCC-3′) sequence. The addition of the insulating nucleotide allowed for the same percentage of extension forming a G:T mismatch, suggesting that the mismatch was formed by misincorporation and not by template slippage. Additionally, insertion of the nucleotide abolished +2, +3 and +4 extension products, which are a result of canonical G:C extension following G:T mismatch formation in the original template, further suggesting that the observed multiple extension events were probably caused by +1 product elongation and were unlikely to be caused by template slippage events (data not shown). However, it is difficult to rule out the contribution of template slippage entirely. Under conditions where a T:T mispair was formed, WT and Lys66Arg behaved similarly, whereas the nonconservative mutants were severely impaired with a considerable increase seen in Km. Furthermore, there was a dramatic effect of the substitutions on the ability of the RT to form a T:T mismatch. While WT and Lys66Arg retain the ability to misinsert 2′-deoxycytosine triphosphate (dCTP) opposite a template thymidine base, the nonconservative substitution mutants are inept, resulting in a six- to eight-fold reduction in fins compared with the WT RT. Although the mutants showed an effect on C:T and T:T mispairs, this reaction was already a lower efficiency reaction, as demonstrated by the low kcat/Km values. For example, the kcat/Km values for incorporation of A opposite T and C opposite T were 0.200 ± 0.031 and 0.033 ± 0.013 for WT RT and 0.137 ± 0.030 and 0.005 ± 0.0043 for the Lys66Ala mutant, respectively. As previously demonstrated for RT and many other polymerases, there was a preference for certain mispairs, with G:T mispairs being more favorable than C:T or T:T mispairs. This preference was unaffected by the Lys66 substitutions. Representation of these data in terms of fins demonstrated that WT and Lys66Arg had a similar misinsertion fidelity, whereas the nonconservative substitutions at position 66 caused enhanced fidelity with a reduction of up to 23-fold in fins (Table 2).

Table 2. Steady-state kinetics of nucleotide misinsertion by WT and Lys66 substitution mutants. ƒins is calculated as (kcat/Km)incorrect/(kcat/Km)correct. ∆ƒins represents the fold change compared with the WT enzyme (Table 1A). Data shown are mean and SD values from at least three independent experiments
dNTPRTkcat (min−1)  ± SEKmm)  ± SEkcat/Kmm−1·min−1)  ± SE ƒ ins ƒins
dGTPWT5.52 ± 0.5712.65 ± 7.800.46 ± 0.252.32 
K66R8.30 ± 0.7714.20  ± 8.070.58 ± 0.302.931
K66A1.52 ± 0.2274.38 ± 43.640.02 ± 0.010.1023
K66N1.52 ± 0.084.31 ± 1.950.35 ± 0.151.771
K66T2.83 ± 0.2015.41 ± 6.690.18 ± 0.070.923
dCTPWT3.30 ± 0.49100.7 ± 51.890.03 ± 0.010.161
K66R6.52 ± 1.29212.7 ± 116.700.03 ± 0.010.151
K66A1.79 ± 0.65351.1 ± 326.50.005 ± 0.0030.036
K66N6.92 ± 2.321529 ± 10910.007 ± 0.0020.028
K66T8.01 ± 1.531618 ± 656.40.005 ± 0.0010.028
dTTPWT1.59 ± 0.3571.89 ± 63.300.010 ± 0.0030.051
K66R12.60 ± 1.78888.3 ± 234.70.014 ± 0.0020.071
K66A2.17 ± 1.22603.5 ± 729.40.004 ± 0.0030.023
K66N11.36 ± 2.221623 ± 659.70.007 ± 0.0020.041
K66T11.48 ± 2.41615 ± 704.00.007 ± 0.0020.041

Site-specific hydroxyl radical footprinting

Differential positioning of the RT on the t/p can be a potential explanation for the different fidelity exhibited by the Lys66 substituted enzymes on matched and mismatched primers. We have previously shown that a mismatch at the primer 3′-terminus is sufficient to cause the repositioning of the RT on a t/p [26]. To investigate whether Lys66 substitutions also cause such repositioning upon encountering a mismatched primer terminus, site-specific footprinting was carried out with all five enzymes. Previously, Gotte et al. showed that the addition of Fe2+ to an extension reaction results in the generation of hydroxyl radicals in the vicinity of the ribonuclease H (RNaseH; EC domain of HIV-1 RT. These radicals result in cleavage of the template in a site-specific manner and thus allow one to deduce the positioning of the polymerase active site with respect to the primer terminus.

In order to determine the relative positions of the RT on the template, we compared the positions obtained following site-specific cleavage with five different primers (A, B, C, D, E) with a ladder generated by potassium permanganate cleavage (Fig. 5A). Primers B and D are used to determine the position of the RT on a matched terminus. In the presence of foscarnet, cleavage occured at position −18, corresponding to pretranslocation (Fig. 5A, arrow a). In the presence of the next complementary nucleotide, and utilizing a primer terminated with a dideoxynucleoside monophosphate to prevent further extension, cleavage occured at position −17, corresponding to post-translocation (Fig. 5A, arrow b). Primers C and E were used to determine the position of RT on a mismatched terminus. In the presence of the next complementary nucleotide and a dideoxynucleotide monophosphate-terminated primer, cleavage occured at position −19 (Fig. 5A, arrow c). This corresponds to a position where cleavage occurred on a matched t/p shortened by a single nucleotide at the 3′-end, as shown in primer pair A. To characterize the positioning of the WT and the Lys66 substitution mutants on the matched and mismatched t/p, we performed our reactions in the presence of primer B or primer C, respectively. Experimental control reactions showed that template cleavage in these reactions was contingent on the presence of both the enzyme and Fe2+ (Fig. 5B) as no cleavage products were observed in their absence. These data were quantified by densitometry and are represented as the percentage of enzyme present at positions −17, −18 or −19 (Fig. 5C). Under conditions where the enzyme encounters a matched primer terminus, ~ 60% of the enzyme is in the post-translocation (−17) position, with ~ 30% at the −18 position and ~ 10% at the −19 position. These results were similar for the WT RT and for the substitution mutants. However, the situation was quite different when the enzymes encountered a mismatched terminus. Whereas previously the majority of the enzyme was trapped at the post-translocation (−17) position, now an equal amount is positioned at an additional pretranslocation state, which we termed the −2 state (corresponding to the polymerase active site) (Figs 1 and 5).

Figure 5.

Site-specific RNaseH-directed hydroxyl radical footprinting. (A) Various primers (A–E) were separately annealed to the template to help distinguish cleavage positions. Pretranslocation and post-translocation positions are designated by arrows a and b, respectively. (B) Representative footprinting gel showing cleavage positions for each enzyme bound to a matched (T:A) or a mismatched (G:A) t/p. No cleavage products were observed in the absence of enzyme or Fe2+. (C) Graphical representation of the positioning of WT and mutant RTs on either a matched (T:A) or a mismatched (G:A) template, shown as positions relative to the 3′ primer terminus. P < 0.025 represents statistical significance between the percentage of WT and Lys66Asn complexes at position −19. P < 0.004 represents statistical significance between Lys66Arg and Lys66Ala complexes at position 19. P < 0.02 represents statistical significance between the percentage of WT and Lys66Ala complexes at position 17.

Template–primer binding kinetics

The catalytic and binding properties of the purified WT enzyme and the mutant enzymes were characterized. To measure the kinetics of binding between the enzyme and the t/p substrate, we performed surface plasmon resonance experiments using the WT and four mutant RTs (Lys66Arg, Lys66Ala, Lys66Asn and Lys66Thr) with matched or mismatched t/p. The purified enzymes were allowed to interact with the immobilized t/p at concentrations ranging from 0 to 200 nm to generate binding sensorgrams such as those shown in Fig. S1.

The WT enzyme exhibited a high affinity for both matched and mismatched t/p, with equilibrium dissociation constants (KD) of 1 and 1.94 nm, respectively (Table S1). The dissociation constant on matched t/p was well within the range of previously published KD values [9, 16, 34]. Interestingly, a conservative substitution at position 66 (Lys66Arg) affected the affinity of the enzyme for the matched t/p (6 nm) and mismatched t/p (8 nm) by six- and eight-fold, respectively, whereas a nonconservative Lys66Ala alteration had minimal effects. Under conditions where the enzyme contained a Lys66Asn or Lys66Thr substitution at position 66, the affinity was decreased by five- to 49-fold for a matched t/p and by 10- to 32-fold for a mismatched t/p.

Drug-resistance profiles

In an attempt to study the role of Lys66 in drug susceptibility, qualitative gels and quantitative filter-based drug-resistance assays were performed using the various enzymes (WT, Lys66Arg, Lys66Ala, Lys66Asn, Lys66Thr, Lys65Ala and Lys65Arg) in the presence of three nucleoside RT inhibitors and one non-nucleoside RT inhibitor. The resistance profiles of the mutants to the nucleotide reverse transcriptase inhibitor, AZTTP, and to the non-nucleoside reverse transcriptase inhibitor, efavirenz (EFV), were specifically examined because the small percentage of mutations seen at position 66 occur in patients on regimens that include the aforementioned inhibitors [31, 32].

Recombinant WT HIV-1 RT, and Lys66 and Lys65 substitution mutants, were used to extend a long t/p in the presence of all four dNTPs and at a range of inhibitor concentrations (Figs S2 and S3). In the presence of AZTTP, Lys66Arg and WT RT appeared to have a similar resistance profile to the WT enzyme, as less full-length product was seen, even at 1 μm AZTTP (Fig. S2). In contrast, the nonconservative substitution mutants, specifically Lys66Ala and Lys66Asn, allowed for more full-length product formation, even at the highest concentration of AZTTP (4 μm) (Fig. S2). The qualitative data were supported by drug-susceptibility measurements (inhibitory concentration 50% [IC 50] values) obtained from our quantitative filter-based assays (Table 3). The data from these experiments showed no difference in susceptibility between the WT and Lys66Arg RTs (185 nm versus 250 nm), whereas Lys66Ala and Lys66Asn exhibited IC50 values that were twice as high (559 nM and 563nM respectively), resulting in a three-fold decrease in susceptibility (Table 3).

Table 3. IC50 values of recombinant RTs for three nucleotide reverse transcriptase inhibitors (AZTTP, d4TTP and TFV-DP) and one non-nucleoside reverse transcriptase inhibitor (EFV), determined using a filter-based assay. Data shown are mean and SD values from three independent experiments. Values in parentheses represent the fold resistance
RTDrug susceptibility (IC50)
AZTTP (nm)d4TTP (μm)TFV-DP (nm)EFV (nm)
WT185 ± 1.114.52 ± 1.16950.3 ± 1.492067 ± 1.07
K66R250.4 ± 1.22 (1)4.87 ± 1.23 (1)1056 ± 1.22 (1.1)2005 ± 1.14 (1)
K66A559.3 ± 1.10 (3)7.91 ± 1.22 (1.7)1635 ± 1.16 (1.7)1669 ± 1.08 (1.2)
K66N563.1 ± 1.08 (3)8.64 ± 1.23 (1.9)1065 ± 1.21 (1.1)1870 ± 1.08 (1.1)
K66T218.9 ± 1.28 (1.2)6.84 ± 1.18 (1.5)1222 ± 1.16 (1.3)1255 ± 1.07 (1.6)
K65A372.5 ± 1.17 (2)15.70 ± 1.17 (3.5)5663 ± 1.23 (5.9)960.3 ± 1.05 (2)
K65R453.9 ± 1.44 (2.4)16.68 ± 1.34 (3.7)2531 ± 1.16 (2.6)1281 ± 1.05 (1.6)

Extension in the presence of 2′,3′-didehydro-TTP (d4TTP) showed, once again, that Lys66Arg allows for full-length product extension similar to the WT enzyme (Fig. S2). However, in this instance the nonconservative mutants showed an intermediate resistance profile compared with WT and Lys65Arg (Fig. S2). This was confirmed by the quantitative assays, where the IC50 values of the nonconservation mutants (7.91, 8.64 and 6.84 μm) showed a decrease in susceptibility of approximately two-fold, whereas the Lys65 mutants showed a three-fold decrease compared with the WT enzyme (Table 3).

Interestingly, in the presence of tenofovir diphosphate (TFV-DP), even though the three nonconservative Lys66 substitutions exhibited slightly higher levels of full-length product at the highest inhibitor concentration compared with WT or Lys66Arg (Fig. S3), there did not appear to be a large difference in the IC50 values obtained from the filter-based assays (Table 3). A larger difference was seen between the Lys65 substitution mutants and the rest of the enzymes, with the Lys65 mutants showing two- to six-fold more resistance to TFV-DP (Table 3).

Lastly, no apparent differences were observed for the WT, the Lys66 substitution mutants or the two Lys65 mutants in the presence of EFV, with the exception of Lys65Ala. Analysis of the IC50 values showed that the Lys65Ala mutant was the most susceptible to the inhibitor with an IC50 of 960 nm (a two-fold increase in susceptibility), while the Lys66 substitution mutants were more similar to the Lys65Arg mutant (Table 3).


In this report, we have shown that Lys66, located in the β3–β4 loop of HIV-1 RT, contributes to mismatch extension fidelity of HIV-1 RT. Perturbing this residue by mutation decreases fext. Interestingly, in addition to mismatch extension fidelity, Lys66 also appears to be important for dNTP insertion fidelity. In addition, we showed that substitutions at this position have little effect on the positioning of the enzyme on the substrate t/p.

In the event that a mispair is created, the enzyme may extend past the mispair, dissociate from the substrate or remain bound as an inactive complex, in which case the enzyme would no longer contribute to the rate of the reaction. Analysis of the t/p-binding kinetic parameters ka (association rate) and kd (dissociation rate) shows that t/p binding does not correlate with mismatch discrimination. A ‘bulkier’ (Arg) or polar uncharged (Asn or Thr) residue shows decreased ka and increased kd for a net slight decrease in affinity indicated by a five- to 50-fold increase in KD. Furthermore, the effect on the KD for matched and mismatched t/p follows similar trends such that no significant differences are seen in affinity of the RT for a matched or a mismatched t/p. As seen previously, a mismatch has little effect on WT t/p binding [27, 32]. Thus, WT RT does not discriminate against a mismatch based on t/p binding and the Lys66 mutations that do increase discrimination do not do so through binding. Although substitutions at position 66 do alter t/p binding, they do not discriminate against the mismatched terminus; rather, they lead to an overall reduction in affinity for the DNA substrate. This was unexpected, as given the proximity of the Lys66 residue to the primer strand, we had hypothesized that discrimination would probably be through altered interaction with the t/p. Furthermore, substituting Lys66 with Arg, Ala, Asn and Thr residues in the structure of a ternary complex of HIV-1 RT showed that nonconservative substitutions caused a greater distortion in the active-site geometry than did the conservative substitutions, and particularly changed the relative position and proximity of the side-chain of residue 66 and the phosphate on the penultimate nucleotide (Fig. 6).

Figure 6.

Mapping of the relative positions of Lys66 substitution mutants with respect to the phosphate of the penultimate base pair. Molecular models of residue 66 in comparison with the conservative and nonconservative substitution mutants using the 3° crystal structure of Huang et al. [12]. The β3–β4 strands are shown in yellow, and the incoming dNTP is shown as a mesh surface. The primer strand and the template strand are both labeled, whereas the nucleotides in both strands are color-coded: dA is blue; dG is green; and dC is pink. Each of the nonconservative mutations was introduced into position 66 of model 1rtd using the command line swapaa and the Dunbrack rotamer library [48]. Distances (represented by red lines) were measured from the indicated atom of the residue at position 66 to the phosphate on the primer strand using the distances tool under the structural analysis menu of Chimera [47].

Data from our single-nucleotide extension assays showed that while WT and Lys66Arg are efficient in extending from a mismatched primer terminus, the nonconservative mutants showed a five- to six-fold reduction in fext. Analysis of catalytic efficiency (fext) allows one to delineate the influence of the substitution specifically on the mismatch extension, as the role of the mutation in normal catalysis is compensated for. Hence, this reduction in efficiency implicates Lys66 explicitly in mismatch extension fidelity.

The Lys66Arg substitution did not alter the misinsertion fidelity, whereas the nonconservative mutants showed a reduced ability to misinsert. The G:T mispairs are purine:pyrimidine mispairs and are more efficiently formed compared with C:T or T:T mispairs because they are similar to A:T base pairs. All the enzymes were relatively efficient in G:T mispair formation whereas the C:T and T:T mispairs were less tolerated.

Taken together, the data suggest that the Lys66 residue not only facilitates the rather high WT levels of mismatch extensions but also misinsertions. It is likely that the effects on misinsertion and mispair extension are caused by local, active-site distortions, demonstrated by the increases in the distance of the residue side chain to the primer (Fig. 6) for the substitution mutants.

Although there are subtle, and statistically significant, differences between the positions of the WT and nonconservative Lys66 substituted enzymes on the mismatched primer terminus, these differences do not account for the differences in fidelity that were observed. The extent to which the enzyme is found in a −19 (−2) additional pretranslocation position differs significantly for WT RT and the Lys66Ala, Asn and Thr mutants (Fig. 5C, right panel). In the context of a mismatched t/p, more of the WT enzyme shifts backwards to a −19 (−2) position compared with the nonconservative mutants. This suggests that upon encountering an incorrect base pair at the primer 3′-terminus the enzyme could potentially realign its active site by sliding backwards on the t/p. Likewise, there is bias in the percentage of complexes found at position −17 on the mismatched t/p, with more of the nonconservative mutant complexes found at this position. Interestingly, although the enzymes in complexes at position −17 are positioned such that they can readily accept an incoming nucleotide, this increase in post-translocation complexes does not correlate with an increase in extension from a mismatched t/p for Lys66Ala, Asn and Thr. This discrepancy between t/p positioning and mismatch extension efficiency was also described previously for a Lys65Ala mutant [26].

HIV-1 RT has evolved the ability to extend mismatches and misinsertions efficiently. None of the amino-acid residues that are responsible for this property have been delineated. For dNTP misinsertion, a number of residues that line the dNTP-binding pocket, including Lys65, Thr115 and Met184, have been shown to impact RT fidelity [26, 35, 36]. Residues affecting the fidelity of RT are not limited to those situated in the polymerase active site. Several substitutions at residues that do not directly line the active site, but yet impinge upon it via interaction with template or primer, such as Met230Ile (in the primer grip) [37] and Glu89Gly (at position −2 on the template) [23, 38] have also been shown to bring about increases in fidelity. Similarly, as seen from our data, substitutions at the Lys66 residue, which appears poised to contact the penultimate nucleotide at the primer 3′-terminus, led to enhanced fidelity of misinsertion. In addition to the impact on mispair extension fidelity that we predicted based on the proximity of Lys66 to the primer terminus, the Lys66 substitutions also enhanced the fidelity of misinsertion. Previously, through substitution of Lys65, which contacts the gamma-phosphate of the incoming nucleotide, it was demonstrated that this residue also has a role in both aspects of fidelity. Although the Lys65 and Lys66 residues are adjacent to each other in the β3–β4 loop, the similarity of the effect of substitutions of these residues on fidelity was unexpected, as their orientation, as shown in the crystal structure, is quite different from each other.

In an attempt to investigate whether the similarity between Lys66 and Lys65 would extend to drug susceptibilities, we performed both gel-based and filter-based RT assays. In addition, Lys65Arg has also been identified as the primary mutation selected for by TFV [39, 40] and has been shown to confer cross-resistance to other inhibitors, including 2′-3′-didehydro-2′-3′-dideoxythymidine (d4T) in cell-based assays [41] and its triphosphate form, d4TTP in cell-free assays [42]. Here we showed that the nonconservative Lys66 mutants showed phenotypes that are quite different from the WT but not as profound as the resistance profiles seen with the Lys65 mutants. While the Lys65Arg mutation has been widely studied in drug resistance, the drug-resistance effects of mutations at position 66 have not been tested. However, it is of interest that mutations at position 66 (which rarely occur in the virus) lead to decreases in drug susceptibility, suggesting that the high conservation of this residue is important for viral fitness. The differences in mutation frequencies observed when substitutions of WT residues are created at positions 65 and 66 may reflect the degree of resistance conferred by these substitutions; thus, although the Lys65 substitutions have a greater effect on enzymatic activity than the Lys66 substitutions, they also lead to higher levels of drug resistance.


The Lys66 substitutions increased the fidelity of misinsertion and mispair extension events by increasing the sensitivity to other factors that disturb active-site geometry, such as mismatches, rather than by affecting t/p binding. When a mismatched primer terminus or a mispaired dNTP is encountered at the active site, the WT or Lys66Arg mutant enzymes would efficiently proceed to incorporate these errors, while Ala/Asn/Thr mutants would be inefficient. The data presented show that the Lys66 mutations do not grossly alter the kinetic constants, except when the active site is disrupted either by an incorrect nucleotide in the N-site or a mismatched primer terminus. Therefore, our results suggest that the HIV-1 RT has evolved the Lys66 residue and maintained it as it continues to evolve, to maintain a high mutation rate as multiple Lys66 substitutions all appear to decrease the efficiency of biochemical events that are necessary for mutation. Our results also suggest that, despite the fact that Lys66 residue orientation at the active site is distinct from that of Lys65, its impact on polymerase fidelity is similar to that of Lys65.

Experimental procedures


Isopropyl thio-β-d-galactoside, restriction enzymes and DNA-modification enzymes [T4 polynucleotide kinase (PNK) and BsmBI] were obtained from New England Biolabs (Ipswich, MA, USA). Deoxynucleotides [dATP, dGTP, dCTP and dTTP] were purchased from CLP (Mercury Ultra-pure Nucleotides, San Diego, CA, USA). All other chemicals were purchased from Sigma Aldrich Chemical Co. (St Louis, MO, USA). FPLC columns were obtained from GE Healthcare (Waukesha, WI, USA). Integrated DNA Technologies (IDT) (Coralville, IA, USA) synthesized all oligonucleotides.

Generating site-directed Lys66 mutants

The Lys66 mutants were constructed using cassette mutagenesis. BsmBI restriction sites flanking codons 49 to 69 were introduced into the pRT expression plasmid by PCR amplification, as described previously [43, 44]. Following cleavage of this intermediate construct with BsmBI, a double-stranded DNA adapter, designed to restore the deleted sequence as well as to include a change in the triplet codon at position 66 (AAA) to CGC (Lys66Arg), GCG (Lys66Ala), AAC (Lys66Asn) or ACC (Lys66Thr), was ligated into the BsmBI-digested plasmid. The mutations were verified by performing PCR and by sequencing in both directions to confirm that only the desired substitutions had been introduced.

Protein expression and purification

Mutations were made in the p66 subunit. The p66 and p51 subunits were overexpressed separately in DH5α, F'IQ. pRIL (codon plus) Escherichia coli, subsequently combined and the heterodimers purified as published previously [19]. The specific RT activity was tested using a Poly(rA) Oligo(dT) activity assay, as previously described [45]. All purified enzymes were additionally tested for nuclease contamination.

Single-nucleotide mismatch extension assay

Gel-purified oligonucleotides (IDT DNA) 5′-CAGGGTCTCCCGATCCCGGACGAGCCCCCGGT-3′ (matched) and 5′-CAGGGTCTCCCGATCCCGGACGAGCCCCCGGG-3′ (mismatched) were 5′-end-labeled with γ-[32P] using T4 PNK and annealed to the template 3′-ACGTCCCAGAGGGCTAGGGCCTGCTCGGGGGCCATCCCC-5′. Extension from either t/p pair was performed in the presence of 10 nm RT, 100 nm primer-template, 50 mm Tris/HCl (pH 8.0), 50 mm KCl, 6 mm MgCl2, 1 mm dithiothreitol, 0.1 mg·mL−1 of BSA and varying concentrations of dATP (the next complementary nucleotide). The concentrations used were 10, 20, 100, 200 and 1000 nm for the matched t/p and 2, 10, 50, 200 and 800 μm for the mismatched t/p. The reactions were allowed to proceed for 10 minutes (allowing for no more than 30% substrate utilization) and were stopped with a mixture of 95% formamide, 15 mm EDTA and 0.2 mg·mL−1 of Bromophenol Blue. The products were analyzed on a 12% denaturing polyacrylamide gel. The data were fitted to a Michaelis–Menten equation in graphpad prism to generate kinetic constants. Only data points that did not exceed 30% substrate utilization were used for determination of the kinetic constants.

Single-nucleotide misinsertion assays

Gel-purified oligonucleotides (IDT DNA) 5′-CAGGGTCTCCCGATCCCGGACGAGCCCCCGGT-3′ (matched) were 5′-end-labeled with γ-[32P] using PNK and annealed to the template (3′-ACGTCCCAGAGGGCTAGGGCCTGCTCGGGGGCCATCCCC-5′) by heating to 80 °C followed by slow cooling. Extension was performed in the presence of 10 nm RT, 100 nm matched t/p, 50 mm Tris/HCl (pH 8.0), 50 mm KCl, 6 mm MgCl2, 1 mm dithiothreitol and 0.1 mg·mL−1 of BSA. The reactions were run in the presence of either the correct (dATP) or the incorrect (dGTP, dCTP or dTTP) nucleotide. The concentrations used were 10, 100, 200, 750, 1000, 2000 and 3000 μm nucleotide. The reactions were allowed to proceed for 2, 5 or 10 min (allowing no more than 30% substrate utilization). The products were treated under the same conditions as mentioned earlier for mispair extension assays. Extension products were quantitated by including +1, +2 and +3 bands as a percentage of the total in the lane using Image Quant. The data were fitted to a Michaelis–Menten equation in graphpad prism to generate kinetic constants. Only data points that did not exceed 30% substrate utilization were used for determination of the kinetic constants.

Site-specific hydroxyl radical footprinting

The positioning of the RT on the t/p was detected by a technique initially described by Gotte et al. [46], and performed as previously described [26]. Briefly, 2.5 nm t/p was incubated for 5 min in a reaction mix containing 40 nm enzyme, 50 mm Tris/HCl (pH 8.0), 50 mm KCl, 6 mm MgCl2, 5 mm dithiothreitol and 0.1 mg·mL−1 of BSA. The reactions were subsequently incubated in the presence of foscarnet or dATP for 5 min, after which 33 μm Fe(NH4)2 was added to initiate the cleavage reaction. All incubations were performed at 37 °C. The cleavage reactions were stopped after 2 min using a stop mix of 95% formamide, 15 mm EDTA and 0.2 mg·mL−1 of Bromophenol Blue. The products were separated on a 15% denaturing polyacrylamide gel.

Template – primer binding (dissociation constant experiments)

Binding experiments for each of the enzymes with the matched and mismatched t/p were performed on a Biacore 3000 (GE Healthcare). Briefly, 500 nm biotinylated template was annealed to 2.5 μm oligonucleotide. The annealing reactions were allowed to proceed for 3 min at 80 °C and then cooled to below 30 °C. The streptavidin-pretreated carboxymethylated dextran sensor surface (Biacore SA chip; GE Healthcare) was preconditioned with three consecutive 1-min injections of a solution containing 1 m NaCl. Different concentrations (0, 0.06, 0.08, 0.2, 0.4, 0.6, 0.8, 1, 2, 4, 6, 8, 10, 12, 20, 50, 100, 120 and 200 nm) of RT were injected onto the matched (ggt) or mismatched (ggg) t/p. The binding of RT was detected as changes in the response units. A blank of running buffer alone was included to control for background readings. Analysis was performed using the BIAevaluation program. The data were fitted to a 1 : 1 Langmuir model where analyte A binds to analyte B to form complex AB, with the rate of AB formation measured as an association rate constant (ka) and the disassembly of AB measured as the off-rate constant (kd).

Gel-based resistance assays

Gel-purified oligonucleotide (Invitrogen/Life Technologies, Grand Island, NY, USA) VP229: 5′-CGC TTT CAG GTC CCT GTT CGG GCG CCC AC-3′ was 5′-end labeled with γ-[32P] using T4 PNK and annealed to the template PBSA (5′-TTT AGT CAG TGT GGA AAA TCT CTA GCA GTG GGC GCC CGA ACA GGG ACC TGA AAG CG-3′) by heating to 80 °C followed by slow cooling. Extension reactions were performed in the presence of 1 nm γ-[32P] 5′-end-labeled t/p, 10 nm enzyme, 50 mm Tris/HCl (pH 8.0), 50 mm KCl, 6 mm MgCl2, 1 mm dithiothreitol, 0.1 mg·mL−1 of BSA, 50 μm of each nucleotide not analogous to the inhibitor, 20 μm analogous nucleotide, 0, 1 and 4 μm AZTTP, d4TTP and TFV-DP, and 32 nm and 20 μm EFV. The reactions were run for 10 min at 37 °C and stopped with a solution containing 95% formamide, 15 mm EDTA and 0.2 mg·mL−1 Bromophenol Blue. The products were separated on an 8% denaturing polyacrylamide gel.

Filter-based RT assay

Gel-purified oligonucleotide (Invitrogen/Life Technologies) VP229 (8 nmol) was annealed to template PBSA (16 nmol) by heating to 80 °C. The RT assays were performed in the presence of annealed t/p (PBSa.VP229), 6.25 nm enzyme, 50 mm Tris/HCl (pH 8.0), 50 mm KCl, 6 mm MgCl2, 1 mm dithiothreitol, 0.1 mg·mL−1 of BSA and various concentrations of inhibitor. The reactions were allowed to proceed for 5 min and stopped by spotting onto diethylaminoethyl paper followed by phosphoimaging. Spot intensities were determined using imagequant software (GE Healthcare). The IC50 values were determined from plots of percentage activity versus inhibitor concentration using graphpad prism 4.0 software.


The authors thank Dr Stuart LeGrice for the kind gift of pRT6H51. This work was supported in part by the Center for AIDS Research at the Albert Einstein College of Medicine and Montefiore Medical Center funded by the National Institutes of Health (NIH AI-51519). This work was supported by NIH grant R37 AI030861 to VRP.