Phenotypic characterization of drug resistance-associated mutations in HIV-1 RT connection and RNase H domains and their correlation with thymidine analogue mutations

Abstract

Objectives

HIV-1 reverse transcriptase (RT) mutations associated with antiviral drug resistance have been extensively characterized in the enzyme polymerase domain. Recent studies, however, have verified the involvement of the RT C-terminal domains (connection and RNase H) in drug resistance to RT inhibitors. In this work, we have characterized the correlation of recently described C-terminal domain mutations with thymidine analogue mutations (TAMs), as well as their phenotypic impact on susceptibility to zidovudine and nevirapine.

Methods

HIV-1 RT sequences from Brazilian patients and from public sequence databases for which the C-terminal RT domains and treatment status were also available were retrieved and analysed for the association of C-terminal mutations and the presence of TAMs and treatment status. Several C-terminal RT mutations previously characterized were introduced by site-directed mutagenesis into an HIV-1 subtype B molecular clone in a wild-type, TAM-1 or TAM-2 pathway context. Mutants were tested for drug susceptibility to the prototypic drugs zidovudine and nevirapine.

Results

Subtype B-infected patient database analysis showed that mutations N348I, A360V/T, T377M and D488E were found to be selected independently of TAMs, whereas mutations R358K, G359S, A371V, A400T, K451R and K512R increased in frequency with the number of TAMs in a dose-dependent fashion. Phenotypic analysis of C-terminal mutations showed that N348I, T369V and A371V conferred reduced susceptibility to zidovudine in the context of the TAM-1 and/or TAM-2 pathway, and also conferred dual resistance to nevirapine. Other mutations, such as D488E and Q547K, showed TAM-specific enhancement of resistance to zidovudine. Finally, mutation G359S displayed a zidovudine hypersusceptibility phenotype, both per se and when combined with A371V.

Conclusions

This study demonstrates that distinct RT C-terminal mutations can act as primary or secondary drug resistance mutations, and are associated in a complex array of phenotypes with RT polymerase domain mutations.

Introduction

Although highly active antiretroviral therapy (HAART) has efficiently improved the inhibition of HIV-1 replication in infected patients, the inevitable emergence of mutations that diminish virus drug susceptibility is still one of the major factors undermining effective response to antiretroviral therapy.¹ Among these, thymidine analogue mutations (TAMs) play an important role due to the common administration of zidovudine or stavudine as the nucleoside reverse transcriptase inhibitor (NRTI) component of HAART. In addition to primary drug resistance mutations that have been well characterized in reverse transcriptase (RT),² other secondary mutations that may further decrease drug susceptibility remain unknown.

Previous research studies that led to the characterization of drug resistance mutations in HIV-1 RT have been limited to its N-terminal domain (codons 1–300), comprising the RT catalytic (polymerization) activity. The catalytic domain is the target of the major RTI classes, NRTIs and non-NRTIs (NNRTIs). Consequently, little is known about the influence of the HIV-1 C-terminal RT domains, the connection (CN) and the RNase H (RNH) domains, on viral susceptibility to drugs. Recently, however, evidence has accumulated that different mutations in the HIV-1 C-terminal RT domains contribute to and are implicated in drug resistance,^3–14 and research focus in these regions has increased. RT CN and RNH domains have recently been implicated in both RNH-dependent and -independent mechanisms of resistance to both NRTIs and NNRTIs.^4,5,15–17 Moreover, a recent dual (NRTI and NNRTI) resistance mechanism has been proposed for some C-terminal mutations.¹⁶ The role of RT C-terminal domain mutations in the clinical outcome of HIV-infected subjects under antiretroviral therapy is just starting to emerge, and there are currently conflicting results. Whereas it has been suggested that mutations in the RT CN domain are associated with lack of virological response in the OPTIMA trial,¹⁸ a more recent study failed to find such association.^19,20 Therefore, the clinical implication of these mutations still warrants further assessment.

With the purpose of identifying the role of HIV-1 RT C-terminal domains in drug resistance, we have identified mutations in the CN and RNH domains of RT that are strongly related to treatment in a Brazilian cohort and in public databases.²¹ Previous work by Cane et al.²² identified C-terminal mutations associated with TAMs or NNRTI-associated mutations; however, the analysis was restricted to the first 400 codons of RT, excluding any associations with RNH mutations.

In light of the above-mentioned scenario, the present study aimed to elucidate the potential correlations between HIV-1 RT mutations located in the CN and RNH C-terminal domains, previously identified as related to NRTI exposure,²¹ and the accumulation of TAMs. We have further phenotypically characterized several of these mutations in the context of the two TAM pathways, TAM-1 (M41L, L210W and T215Y) and TAM-2 (D67N, K70R, T215F and K219Q),²³ for their impact on resistance to NRTIs and NNRTIs.

Materials and methods

Association of CN and RNH mutations with TAMs

Thirty sequences of HIV-1 subtype B RNH RT domain from Brazilian HIV-infected patients previously investigated by our group,²¹ and for which RT polymerase sequences were also available, were used in this analysis. We also evaluated 873 CN and RNH sequences from the Stanford HIV Drug Resistance Database (http://hivdb.stanford.edu) and from the Los Alamos HIV database (http://hiv-web.lanl.gov) of the HIV-1 subtype B pol gene. While sequences from drug-experienced subjects were retrieved from the former, those from drug-naive subjects were from the latter. Only sequences of  >326 codons of RT (the first position analysed herein) were included in this study. To avoid duplicated sequences, all sequence data were subjected to phylogenetic inference, as previously described.²¹

Sequences were grouped according to the total number of TAMs (0, 1–2 or ≥3); the group with no (0) TAMs was subdivided into treated or untreated patients. The following TAMs were considered in our analyses: M41L, D67any, K70R, L210W, T215any and K219any. C-terminal CN and RNH domains of each viral sequence were surveyed for the presence of the following mutations: I326V, G335D, N348I, R358K, G359S, A360V/T, K366R, T369V, A371V, I375V, T377M, S379G, K390R, I393M and A400T in the CN subdomain; and K451R, L452S, T470N, H483Q, D488E, I506L, K512R, K527N, K530R and Q547K in the RNH domain.

To estimate the correlation between each of these mutations and the accumulation of TAMs, analyses of frequency distribution were performed for the following two situations: (i) analysis between the groups of treated and untreated patients with no (0) TAMs, for which Fisher's exact test was used; and (ii) comparison between the groups 0 TAMs, 1–2 TAMs and ≥3 TAMs among sequences from treated subjects, by using a χ² test, and P values <0.05 were considered statistically significant.

Plasmids, clones and mutagenesis

For the phenotypic characterization of mutations, site-directed mutagenesis was conducted with the QuikChange XL site-directed mutagenesis kit (Stratagene). The following mutants were constructed: N348I, K358R, G359S, A360V, T369V, A371V, I375V, T377M, K451R, D488E, I506L, K512R, K530R and Q547K. The double mutants A371V + G359S and I375V + K530R were also constructed, in view of their high co-occurrence in sequences by visual inspection. In the case of R358K, 358K was already present in the parental pNL4-3 virus from which our molecular clone was derived; therefore we reverted codon 358 to an arginine (K358R) to test its impact on drug susceptibility. All mutants and relevant TAM backbone mutations were confirmed by DNA sequencing prior to phenotypic analysis. Mutations were analysed in three luciferase-expressing HIV-1 vectors: pHL-WT (where WT stands for wild-type); pHL-TAM-1 (containing TAMs M41L, L210W and T215Y); and pHL-TAM-2 (containing mutations D67N, K70R, T215F and K219Q), as previously described.⁵

Virus production and single replication cycle drug susceptibility assays

Cell transfections and single replication cycle susceptibility assays were conducted as previously described.⁶ Briefly, 293T cells were plated at a density of 5 × 10⁶ and on the following day were transfected with the different CN and RNH mutants in a pHL-WT or pHL-TAM backbone. All transfections were done in combination with pHCMV-G, a plasmid expressing the vesicular stomatitis virus envelope protein.²⁴ Forty-eight hours after transfection, culture supernatants were harvested and filtered through a 0.45 μm pore-size membrane (Nalgene). A 90 min ultracentrifugation at 4°C at 25 000 rpm was used to pellet and concentrate viruses, which were further stored at −80°C until use.

To estimate virus viability, 293T cells were plated at a density of 4 × 10⁵ in 96-well plates and infected using four 10-fold dilutions of virus stocks. Luciferase activity was assessed 48 h after infection using the Britelite Plus kit (PerkinElmer) in a 96-well luminometer (LUMIstar Galaxy; BMG Labtech). The virus amount corresponding to 5 × 10⁴ luciferase units was used in a single replication cycle drug susceptibility assay⁶ for zidovudine and nevirapine, the prototypic antiretroviral drugs of the two major classes of RT inhibitor. Inhibition curves, based on percentage inhibition of luciferase activity, were constructed using SigmaPlot v.11.0 to calculate the drug concentration required to inhibit virus replication by 50% (IC₅₀). Replicates ranging from 2 to 11, from an average of three independent experiments, were performed for each virus. The IC₅₀s of zidovudine and nevirapine for each mutant clone were compared with those for their WT counterparts (in either a WT or a TAM-1/2 context) using Student's t-test.

Results

Correlation between C-terminal RT mutations and TAMs

A total of 920 sequences of subtype B viruses were analysed, 47 of which were selected from previous data generated by our group.²¹ The majority of the sequences from treated subjects (64.4%) were derived from patients who took more than one regimen. Thirty-four percent of patients had been subject to two sequential drug regimens, 16% had three and another 14% had over three regimens. Even for those patients who took a single regimen, there were 19 different antiretrovial drug combinations (not shown). With regard to the geographical distribution of the sequences, approximately two-thirds (65%) came from the USA, 16% were from Europe, 15% were from Latin America and the Caribbean, and the remaining 4% from the rest of the world. Nineteen percent of the sequences were derived from samples collected in the pre-HAART era (before 1996), while the majority (81%) were from the HAART era.

Four hundred and eighty-eight sequences were found with none of the evaluated TAMs (331 from untreated and 157 from treatment-experienced patients), while the group with 1–2 TAMs was composed of 157 sequences, and the group with ≥3 TAMs harboured 288 sequences.

Statistical analyses of frequency distribution were performed to estimate the correlation between the C-terminal mutations I326V, G335D, N348I, R358K, G359S, A360V/T, K366R, T369V, A371V, I375V, T377M, S379G, K390R, I393M, A400T, K451R, L452S, T470N, H483Q, D488E, I506L, K512R, K527N, K530R and Q547K and the accumulation of TAMs. The comparison between treated and untreated patients carrying viruses with no TAMs revealed that 4 out of the 25 analysed mutations had significantly higher prevalence among treated patients (Figure 1a): N348I (P = 0.0464), A360V/T (P = 0.0219), T377M (P = 0.0192) and D488E (P = 0.0057). Interestingly the mutation D488E, with the highest significance, was absent among 106 sequences from viruses of untreated patients, but had a prevalence of 8% (6/80) in treatment-experienced patients in the absence of TAMs.

Figure 1.

(a) Frequency of HIV-1 RT C-terminal mutations among viral sequences of antiretroviral-naive and -treated subjects in the absence of TAMs in the RT polymerase domain. (b) The same as in (a) except that viral sequences from treated subjects containing 0, 1–2 or ≥3 TAMs are compared. Only mutations for which significant P values were achieved are shown. P values of Fisher's exact tests (a) and χ² tests for trend (b) are depicted above each comparison group.

When the comparison of groups consisting of treated patients with 0, 1–2 or ≥3 TAMs was conducted, six of the analysed mutations were found to be significantly related to the number of TAMs in a dose-dependent fashion (Figure 1b). The lowest P values were found for R358K, G359S (both with P values of 0.001) and A371V (P < 0.0001). The latter showed the largest difference of ∼12-fold in prevalence when the extreme groups (0 and ≥3 TAMs) were compared. The mutation A400T showed the highest prevalence in the three groups of treated patients, ranging in frequency from 54% in the 0 TAM group to 66% in the ≥3 TAMs group. Mutations K451R and K512R also showed highly significant P values of 0.0046 and 0.0004, respectively.

Mutations in the CN and RNH domains reduce zidovudine drug susceptibility in the presence of TAMs

We assessed how mutations previously identified as being increased with the accumulation of TAMs in HIV-1 subtype B sequences contribute to zidovudine drug resistance in the absence and presence of primary mutations in the context of viral infections. IC₅₀ data showed that none of the analysed mutations significantly reduced susceptibility to zidovudine in a WT RT polymerase backbone, with the exception of N348I (fold change of 1.6; Table 1). These results are consistent with previous observations that most CN or RNH mutations do not increase zidovudine resistance in the absence of TAMs.^3–5

Table 1.

Phenotypic susceptibility (IC₅₀ in μM ± SD) of HIV-1 RT C-terminal mutants to zidovudine (CCO1 = 1.5; CCO2 = 11.4^a) in different resistance contexts

Mutation(s)	pol resistance context
	WT	TAM-1	TAM-2
WT	0.070 ± 0.02 (1.0)	0.746 ± 0.32 (10.6)	0.713 ± 0.17 (10.1)
N348I	0.110 ± 0.02 (1.6)	ND	1.397 ± 0.14 (19.8)
K358R	0.066 ± 0.01 (0.9)	0.515 ± 0.07 (7.3)	0.539 ± 0.14 (7.6)
G359S	0.052 ± 0.02 (0.7)	0.196 ± 0.04 (2.8)	0.482 ± 0.05 (6.8)
A360V	0.085 ± 0.01 (1.2)	0.882 ± 0.27 (12.5)	0.909 ± 0.17 (12.9)
T369V	0.085 ± 0.01 (1.2)	4.320 ± 0.38 (61.2)	11.946 ± 2.51 (169.5)
A371V	0.048 ± 0.01 (0.7)	1.811 ± 0.59 (25.7)	1.970 ± 0.29 (27.9)
I375V	0.060 ± 0.00 (0.9)	0.761 ± 0.28 (10.8)	0.880 ± 0.15 (12.5)
T377M	0.052 ± 0.01 (0.7)	0.872 ± 0.23 (12.4)	0.607 ± 0.22 (8.6)
K451R	0.039 ± 0.02 (0.6)	ND	0.943 ± 0.23 (13.4)
D488E	0.057 ± 0.02 (0.8)	ND	1.048 ± 0.15 (14.9)
I506L	0.061 ± 0.00 (0.9)	1.153 ± 0.04 (16.4)	0.890 ± 0.17 (12.6)
K512R	0.036 ± 0.02 (0.5)	0.784 ± 0.06 (11.1)	0.678 ± 0.24 (9.6)
K530R	0.052 ± 0.02 (0.7)	0.953 ± 0.54 (13.5)	0.841 ± 0.43 (11.9)
Q547K	0.057 ± 0.03 (0.8)	2.070 ± 1.29 (29.4)	0.622 ± 0.14 (8.8)
G359S + A371V	0.061 ± 0.01 (0.9)	1.199 ± 0.50 (15.9)	1.991 ± 0.42 (28.2)
K530R + I375V	0.051 ± 0.01 (0.7)	0.683 ± 0.11 (9.7)	0.619 ± 0.03 (8.8)

ND, not done.

The fold change in IC₅₀ versus the respective WT context is shown in parentheses.

Bold values refer to IC₅₀s significantly different from the respective resistance (WT, TAM-1 or TAM-2) context at the 0.05 level.

^aCCO1 and CCO2, clinical cut-offs 1 (20% reduced response) and 2 (80% reduced response) as last updated by VIRCO.³¹

In the context of TAMs, however, a significant decrease in zidovudine susceptibility was observed for 5 of the 16 individual mutants (Table 1). For both TAM backbones, zidovudine susceptibility was reduced in the presence of mutations A371V (fold change of 25.7 in TAM-1 and 27.9 in TAM-2) and T369V, which in fact depicted the highest resistance levels found in this study (fold change of 61.2 in TAM-1 and 169.5 in TAM-2, respective to WT; or 5.8 in TAM-1 and 16.8 in TAM-2, compared with the respective TAM backbones). The mutations N348I and D488E also showed reduced susceptibility in a TAM-2 backbone. Interestingly, the RNH mutation Q547K showed reduced susceptibility to zidovudine exclusively in a TAM-1 context (2.8-fold less susceptible compared with the TAM-2 backbone). The A371V + G359S double mutant showed a higher fold change in the TAM-2 backbone (28.2 versus 15.9 in TAM-1). Interestingly, the mutation G359S significantly increased virus susceptibility to zidovudine in both TAM backgrounds (Table 1). When the double mutant G359S + A371V was compared with A371V alone in a TAM-1 context, a neutralization effect of A371V was observed, corroborating the sensitization effect of G359S.

Mutations N348I and T369V exhibit dual resistance to zidovudine and nevirapine

To determine whether any of the mutations in the CN or RNH domains tested conferred dual resistance to NRTIs and NNRTIs, resistance to nevirapine was first tested for all mutants in the background of a WT RT. As shown in Table 2, N348I and T369V displayed reduced susceptibility to nevirapine even in the background of a WT pol, with fold change values of 3.3 and 8.6, respectively, compared with the WT control. We further tested N348I and T369V, in addition to G359S, A371V and G359S + A371V, for nevirapine susceptibility in the presence of TAM-1- and TAM-2-containing viruses (Table 3). Interestingly, both T369V and A371V mutants still exhibited reduced nevirapine susceptibility in the TAM-1 context (fold change of 3.4 and 2.3, respectively versus the TAM-1 control). T369V also showed reduced susceptibility in the context of TAM-2 (fold change of 3.2; Table 3). Interestingly, in the TAM background, the observed nevirapine resistance conferred by N348I and T369V was diminished (compare Tables 2 and 3). Conversely, in a TAM-1 context, the mutation A371V was able to confer reduced susceptibility to nevirapine, a phenomenon not displayed by that mutation by itself. We have also studied the sensitizing effect of G359S to nevirapine in different TAM backbones, which was confirmed on its own and together with A371V, particularly in a TAM-1 backbone (Table 3).

Table 2.

Phenotypic susceptibility (IC₅₀ in μM ± SD) of HIV-1 RT C-terminal mutants to nevirapine (BCO = 6.0)^a in the absence of other resistance mutations

Mutation(s)	IC50 ± SD
WT	0.041 ± 0.02 (1.0)
N348I	0.135 ± 0.01 (3.3)
K358R	0.022 ± 0.01 (0.5)
G359S	0.030 ± 0.01 (0.7)
A360V	0.035 ± 0.01 (0.8)
T369V	0.353 ± 0.01 (8.6)
A371V	0.051 ± 0.01 (1.2)
I375V	0.038 ± 0.00 (0.9)
T377M	0.048 ± 0.01 (1.2)
K451R	0.028 ± 0.01 (0.7)
D488E	0.024 ± 0.00 (0.6)
I506L	0.042 ± 0.01 (1.0)
K512R	0.032 ± 0.00 (0.8)
K530R	0.037 ± 0.01 (0.9)
Q547K	0.041 ± 0.00 (1.0)
G359S + A371V	0.041 ± 0.02 (1.0)
K530R + I375V	0.038 ± 0.00 (0.9)

The fold change in IC₅₀ versus the WT is shown in parentheses.

Bold values refer to IC₅₀s significantly different from the respective resistance context at the 0.05 level.

^aBCO, latest biological cut-off by VIRCO (http://vircolab.com).

Table 3.

Phenotypic susceptibility (IC₅₀ in μM ± SD) of selected HIV-1 RT C-terminal mutants to nevirapine (BCO = 6.0^a) in different resistance contexts

Mutation(s)	pol resistance context
	WT	TAM-1	TAM-2
WT	0.042 ± 0.01 (1.0)	0.042 ± 0.01 (1.0)	0.026 ± 0.00 (0.6)
N348I	0.135 ± 0.01 (3.3)	ND	0.072 ± 0.02 (1.8)
G359S	0.030 ± 0.01 (0.7)	0.042 ± 0.01 (1.0)	0.027 ± 0.00 (0.7)
T369V	0.353 ± 0.01 (8.6)	0.140 ± 0.01 (3.4)	0.133 ± 0.00 (3.2)
A371V	0.051 ± 0.01 (1.2)	0.093 ± 0.01 (2.3)	0.042 ± 0.00 (1.0)
G359S + A371V	0.041 ± 0.02 (1.0)	0.067 ± 0.01 (1.6)	0.053 ± 0.00 (1.3)

ND, not done.

The fold change in IC₅₀ versus the respective WT context is shown in parentheses.

Bold values refer to IC₅₀s significantly different from the respective resistance context at the 0.05 level.

^aBCO, latest biological cut-off by VIRCO (http://vircolab.com).

Discussion

Increasing efforts have been made to better understand the mechanisms governing HIV-1 drug resistance conferred by RT C-terminal mutations.²⁵ It has been postulated that mutations in RT RNH³ or CN⁵ domain impair RNH cleavage activity, providing additional time for nucleoside analogue excision and therefore enhancing resistance. The current study extended and corroborated this hypothesis by establishing a correlation between RT C-terminal mutations and accumulation of TAMs in HIV-1 subtype B, by assessing their phenotypic role in HIV drug resistance, and by improving our knowledge of the contribution of RT C-terminal domains to HIV-1 drug resistance.

A total of 25 mutations were evaluated to assess their correlation with TAMs. Although we were unable to associate specific RT C-terminal mutations with distinct polymerase domain mutations (due to the complexity of drug scheme regimens taken by the majority of the patients studied), we did find relevant associations with therapy. Four of these mutations were found to be significantly correlated with therapy when comparing sequences from naive and treatment-experienced isolates in the absence of TAMs, namely N348I, A360V/T, T377M and D488E. The first two corroborate previous reports, confirming that N348I and A360V both confer dual drug resistance to zidovudine and nevirapine.^4,5,10,11,15 Mutations R358K, G359S, A371V, A400T, K451R and K512R were found to be correlated with TAMs in a dose-dependent manner, being more frequent as the number of TAMs increased in the RT polymerase domain. Some of these mutations were previously shown to be associated with TAMs, as was the case for G359S, A371V and K512R.^22,26 Mutation K451R has been recently confirmed as being correlated with treatment.²⁷ Mutation A400T has been recently associated with increased zidovudine resistance through zidovudine monophosphate excision and RNH cleavage reduction in both subtypes B and CRF01_AE.²⁸ A400T has been shown to vary in frequency among drug-naive and experienced patients infected with different HIV-1 subtypes, and further characterization of the role of this polymorphism is warranted.

Some C-terminal mutations failed to provide statistical significance in co-occurrence with TAMs (T369V, I375V, I506L, K530R and Q547K). The small number of available sequences in this HIV genomic region, particularly at the RNH domain, might explain at least partially the observed lack of statistical robustness. Another potential caveat of our analysis is the heterogeneous proportion of subtype B sequences from different geographical locations. Approximately two-thirds of our sequences are from the USA, a fact that could bias WT and mutant codon compositions analysed. However, we failed to detect any obvious differences in codon prevalence between subtype B sequences from different geographical locations, or between the sequences generated from our cohort and those retrieved from the databases. Nonetheless, previous reports showing the correlation of those C-terminal mutations with treatment experience²¹ prompted us to assess their phenotypic impact on resistance to NRTIs and NNRTIs.

The majority of the mutations analysed in this work failed to show significant reductions in susceptibility to zidovudine or to nevirapine in the absence of TAMs, as expected (Tables 1 and 2), with the exception of N348I, for which modest increases in fold change values relative to WT were observed for zidovudine. For nevirapine, N348I (fold change of 3.3) and T369V (fold change of 8.6) showed modest to high increases in fold change values respective to WT. Interestingly, N348I was among the mutations associated with therapy independently of TAMs. One may speculate that these CN substitutions act as primary resistance mutations, selected by treatment in the absence of other mutations, and display a reduction in drug susceptibility per se. Consistent with this hypothesis is the observation that N348I may arise early in therapy, even before TAMs.¹⁰ Moreover, it has been recently suggested that both N348I and T369I/V reduce viral replication capacity,²⁹ a feature commonly displayed by primary resistance mutations.

The phenotypic characterization of the single C-terminal mutants in the context of two different TAM backbones showed diverse effect patterns. While some mutations decreased susceptibility to zidovudine of viruses containing either TAM-1 or TAM-2 mutations, as was the case for T369V and A371V, others showed more pronounced effects for a single TAM backbone pathway. That was the case for RNH mutation Q547K in a TAM-1 backbone, and the mutant D488E in TAM-2. We can think of these C-terminal mutations as secondary or accessory, as they act by further decreasing the susceptibility of viruses containing primary TAMs, but do not have a direct effect on zidovudine susceptibility. For those mutations that did not affect drug susceptibility, they could be acting by increasing the viral fitness of drug-resistant viruses, and additional investigation of this scenario is warranted. Although T369V showed decreased susceptibility with both TAM backbones, the effect was even more pronounced in TAM-2, also providing evidence of distinct TAM-specific phenotypes. In an in vitro analysis of virus passage in increasing zidovudine concentrations, mutation A371V was selected in the presence of TAMs, with a 26-fold increse in resistance over WT,¹² corroborating our phenotypic results. A371V has also been evaluated by Delviks-Frankenberry et al.⁴ in the context of mutation A360V in a TAM-2 background, providing a 4-fold increase in resistance over TAMs. Finally, it has been reported that A371V was the only baseline CN domain mutation associated with antiretroviral virological failure in the OPTIMA trial.¹⁸ Although we found a modest enhancement of the resistant phenotype for N348I and A371V, other reports that tested N348I showed fold change values of higher magnitudes when compared with WT viruses, ranging from 2.0 to 6.9.^10,11,20 Differences in the backbone virus used, as well the methodology used in the zidovudine susceptibility assays, might account for the disparities in the absolute fold change values observed in all three studies. Nonetheless, all reports agree on the reduction in susceptibility conferred by N348I when in combination with TAMs.

Of interest, we found that some of the RT C-terminal mutations tested act antagonistically on susceptibility to zidovudine, despite being selected together. That was the case for G359S, which showed a hypersusceptibility effect to zidovudine both per se and in the context of either TAM backbone (fold change ranging from 0.2 to 0.7). This mutation, despite being covariant with A371V, has a counteracting effect on the latter, increasing susceptibility to zidovudine when compared with viruses containing solely A371V. A recent survey in the Stanford HIV Drug Resistance Database (http://hivdb.stanford.edu; accessed on 8 August 2010) showed that ∼4% of drug-naive subtype B viruses have the 359S mutation. The impact of this polymorphism should be further evaluated in clinical assessments.

The mutation R358K was found to be correlated with the accumulation of TAMs in our study (Figure 1b). This agrees with a recent report by another group, where this mutation was found to be enriched in antiretroviral therapy-exposed subjects.¹⁹ As R358K was already present in the luciferase reporter vector used in our analysis (pNL4-3 derived), we reverted the lysine residue to an arginine through site-directed mutagenesis to assess its role in drug resistance. K358R showed an increase of 10% in susceptibility of the WT virus to zidovudine, and an increase of 30% and 20% in TAM-1 and TAM-2 backbones, respectively. For nevirapine, a 50% increase in susceptibility over WT was seen. In another study, R358K failed to show decreased susceptibility in a TAM-2 background; however, in combination with F416Y, the fold change was 4-fold higher in that context.¹² It is possible that 358R, as well as the previously mentioned 359S, are related to restoration of virus replicative capacity (RC) in association with other C-terminal and primary polymerase domain mutations. The antagonistic effect of these mutations is worth further assessment.

The susceptibility patterns of the single C-terminal mutants showed that N348I and T396V also conferred reduced susceptibility to nevirapine. A profile of dual resistance to NRTIs and NNRTIs has been recently proposed for N348I and T369I/V,^10,11,20,29 and our results are in agreement with those previous reports. It is noteworthy that the CN mutation G359S again showed a re-sensitization effect to nevirapine, in WT and TAM-containing viruses, and also an antagonistic effect of A371V for this drug. Taken together, our results highlight the complexity of the interactions between RT C-terminal and polymerase mutations in conferring drug resistance to nucleoside and non-nucleoside analogues. Additional enzymatic and molecular dynamic studies will be necessary to fully understand the molecular steps and atomic interactions governing these effects.

Overall, many of the RT C-terminal mutations herein characterized seem to be secondary or accessory in HIV drug resistance, mostly displaying a phenotypic effect of further reducing susceptibility of viruses containing TAMs. A role of these mutations in restoring RC of resistant viruses constitutes a likely scenario. In fact, preliminary evidence in that direction has already been shown for one of these mutations, A371V.^12,18 Further investigation of this issue for the remaining mutations characterized in our study is therefore merited. Data are just emerging on the impact of RT C-terminal mutations in virological failure of patients undergoing antiretroviral therapy,^18,19 as well as those carried by viruses from drug-naive subjects,³⁰ which will underscore the importance of evaluating these mutations in clinical practice. Our work has succeeded in establishing primary or secondary roles for many of the recently characterized mutations in the C-terminal domains of HIV-1 RT, adding to current knowledge on the contribution of this newly studied region of the RT enzyme to HIV drug resistance.

Funding

This work was supported by the Rio de Janeiro State Foundation (FAPERJ) grant no. E-26/102.858-2008, the Brazilian Science Council (CNPq) grant no. 134969/2007-3, SAIC–Frederick, Inc. and the Intramural Research Program of the National Institutes of Health, National Cancer Institute, Center for Cancer Research and the Intramural AIDS Targeted Antiviral Program, USA. The funders had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript.

This work is part of the Master of Sciences Dissertation by R. B. L. of the Graduate Program in Genetics, Universidade Federal do Rio de Janeiro, Brazil.

Transparency declarations

None to declare.

Disclaimer

The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, and mention of trade names, commercial products or organizations does not imply endorsement by the US Government.