HiV Articles  
Characterization of the E138K Resistance Mutation in HIV-1 Reverse Transcriptase Conferring Susceptibility to Etravirine in B and Non-B HIV-1 Subtypes
  Download the PDF here
Antimicrob. Agents Chemother. February 2011
Eugene L. Asahchop, Maureen Oliveira, Mark A. Wainberg, Bluma G. Brenner, Daniela Moisi, Thomas d'Aquin Toni and Cecile L. Tremblay


We have selected for resistance to etravirine (ETR) and efavirenz (EFV) in tissue culture using three subtype B, three subtype C, and two CRF02_AG clinical isolates, grown in cord blood mononuclear cells. Genotypic analysis was performed at baseline and at various weeks of selection. Phenotypic resistance in regard to ETR, EFV, and nevirapine (NVP) was evaluated at weeks 25 to 30 for all ETR-selected viruses and in viral clones that contained specific resistance mutations that were inserted by site-directed mutagenesis into pNL-4.3 and AG plasmids. The results show that ETR selected mutations at positions V90I, K101Q, E138K, V179D/E/F, Y181C, V189I, G190E, H221H/Y, and M230L and that E138K was the first of these to emerge in most instances. The time to the emergence of resistance was longer in the case of ETR (18 weeks) compared to EFV (11 weeks), and no differences in the patterns of emergent mutations could be documented between the B and non-B subtypes. Viral clones containing E138K displayed low-level phenotypic resistance to ETR (3.8-fold) and modestly impaired replication capacity (2-fold) compared to wild-type virus. ETR-selected virus showed a high degree of cross-resistance to NVP but not to EFV. We identified K101Q, E138K, V179E, V189I, G190E, and H221Y as mutations not included among the 17 currently recognized resistance-associated mutations for ETR.

HIV-1 reverse transcriptase (RT) is responsible for the conversion of the single-stranded RNA genome into double-stranded DNA and is therefore an important target of anti-HIV-1 therapy (16). Non-nucleoside reverse transcriptase inhibitors (NNRTIs) are noncompetitive inhibitors of HIV-1 RT that bind to the polymerase active site and disrupt enzyme function. NNRTIs represent an important component of standard antiretroviral therapy in HIV-1-infected patients. NNRTIs can also exert additive effects when combined with nucleoside reverse transcriptase inhibitors (NRTIs) (5) and have potency comparable to that of protease inhibitors (PIs) (39, 41). However, the efficacy of narrow-spectrum NNRTIs such as nevirapine (NVP) and efavirenz (EFV) is often hindered by their low genetic barrier for resistance, since only a single amino acid substitution in the NNRTI binding pocket can often lead to high-level resistance against these drugs (3, 25). Furthermore, mutations associated with NNRTI resistance can be transmitted to others, and this has been documented in both developed and developing countries as occurring in between 10 and 20% of newly infected individuals (1, 19).

Etravirine (ETR) is an expanded-spectrum NNRTI with potent antiviral activity against both wild-type (wt) HIV-1 subtypes and against some viruses resistant to narrow-spectrum NNRTIs (2, 47). Clinical studies have shown that ETR plus optimized background regimens consisting of NRTIs, as well as integrase and protease inhibitors, significantly decreased viral loads in patients with resistance against narrow-spectrum NNRTIs and some PIs (9, 27, 33).

The structure of ETR allows it to bind to the RT enzyme in more than one distinct mode through conformational adaptations based on changes in the NNRTI-binding pocket. This allows ETR to reorient itself and provides alternative binding conformation when mutations in the binding pocket occur (12). In vitro selection studies and both the DUET-1 and the DUET-2 clinical trials identified 17 ETR resistance-associated mutations (RAMs) (V90I, A98G, L100I, K101E/H/P, V106I, E138A, V179D/F/T, Y181C/I/V, G190A/S, and M230L) and have permitted the assignment of a weighted score for each mutation (48). In general, three or more ETR RAMs are required for a diminished responsiveness to ETR to occur.

Non-subtype B infections represent >90% of the global HIV problem. Non-B viruses predominate in sub-Saharan Africa and are becoming more prevalent even in areas dominated by subtype B, such as North America and western Europe (18). Although some studies have established that subtype polymorphisms can play a role in the development of drug resistance, based on specific mutations that are selected by antiviral pressure (6, 7, 11, 21), most have focused on narrow-spectrum antiretrovirals. It is known, however, that naturally occurring polymorphisms at RT sites that correspond to ETR RAMs can vary among subtypes (24, 25, 28, 36). We sought to investigate the possible importance of such polymorphisms in the development of ETR resistance, even though ETR seems to possess similar antiviral activity against various HIV subtypes (A to H) and some circulating recombinant forms (CRFs).

(This study was performed by E.L.A. in partial fulfillment of the requirements of a Ph.D. degree for the Dˇpartement de Microbiologie et d'Immunologie, Universitˇ de Montrˇal, Montrˇal, Quebec, Canada.)


Although ETR is currently approved for NNRTI-experienced patients, a number of studies have suggested that its use in patients with NNRTI mutations at baseline might lead to poor virological response (VR) (22, 29, 30, 37). In most of these studies, the presence in the background regimen of newer drugs such as enfuvirtide, darunavir, or raltegravir was positively associated with VR, whereas previous exposure to NVP and the presence of baseline mutations such as E138A, V179I, and Y181C were negatively associated with VR. We provide here a comprehensive in vitro analysis of the resistance pattern of ETR in wild-type clinical isolates of subtype B, subtype C, and CRF02_AG grown over 25 to 30 weeks in CBMCs in increasing concentrations of ETR compared to EFV. Our data show that ETR RAMs emerged after 18 weeks of drug pressure except for one isolate (BK132) that selected for E138E/K at weeks 11 to 13. The E138K mutation was selected in all isolates and was almost always the first mutation to emerge.

Although E138K has been reported in a few cases of ETR treatment failure (44, 49), its clinical significance has not yet been elucidated. Site-directed mutagenesis of E138K revealed low-level resistance to ETR in both pNL-4.3 and AG plasmids, with a fold change slightly above the clinical cutoff (CCO). Others have reported a median ETR FC of 2.6-fold for E138K with first- and third-quartile values of 2.0- and 4.1-fold, respectively, data that are consistent with the site-directed mutagenesis results reported here (4). Although E138K cannot on its own exclude ETR as a potentially useful drug, its presence might facilitate the accumulation of other resistance mutations or may result in an increased FC for resistance for ETR. In an analysis of the DUET studies, two different groups showed that mutations at position 138 were among the most frequent with E138K emerging in three patients (44, 49). Furthermore, all patients with E138 mutations demonstrated an increased FC for ETR over the upper CCO for resistance for ETR (>13), documenting the role of mutations at position 138 in ETR resistance. Of course, the emergence of resistance in the treatment-experienced DUET patients may not be comparable to in vitro selections beginning with wt viruses, especially because several of the NRTI mutations that patients possessed in the DUET studies are known to hypersensitize them to ETR. Clinical studies with ETR in NNRTI-naive patients will be required to validate the present results.

In contrast, an FC for EFV below the CCO was observed with the site-directed mutants of E138K. This is consistent with the fact that E138K is not selected in vitro by EFV and was not observed in EFV-treated patients. One study reported the emergence of E138K after in vitro passage of viruses containing amino acid substitutions at position 135 at baseline (15). These authors demonstrated FCs for EFV of 2 and 7 for E138K and I135T/E138K, respectively.

The ETR RAMs observed here include V90I, K101Q, E138K, V179D/E/F, Y181C, V189I, G190E, H221Y, and M230L, a pattern similar to that observed by others who also observed a high frequency of L100I if viruses contained some NNRTI mutations at baseline (47). These authors also selected E138G if they began with a virus containing the K103N substitution.

We did not observe any difference in the pathway of ETR resistance in B and non-B subtypes. The Y181C mutation was associated with either E138K or V179D/E/F. Both Y181C and V179F were also observed to co-emerge in the DUET studies (44, 48), a finding that is possibly due to the fact that the combination of either E138K and Y181C or of V179D/E/F and Y181C improves viral replication capacity compared to the presence of only a single mutation. In a recent analysis of genotypic and phenotypic changes of rebounders in the DUET studies, it was shown that V179 (n = 35), E138 (n = 17), Y181 (n = 15), and K101 (n = 13) (44) were the positions at which new mutations were most frequently observed. In contrast, EFV selected for classic mutations, including V106I and V106M in subtypes B and C, respectively, at weeks 11 to 13, as previously described (6). The specific selection here of G190A and G190E by EFV and ETR, respectively, confirms the nonassociation of G190A with ETR (50). However, G190E has been shown to be selected by both ETR and another novel NNRTI, rilpivirine (RIL) (4, 47).

Our phenotypic data show that accumulation of several ETR RAMs especially Y181C and M230L, resulted in high level ETR resistance. Viruses possessing high-level resistance to ETR also displayed moderate and high-level resistance to EFV and NVP, respectively. This confirms cross-resistance between ETR and NVP and is also consistent with studies showing that the prior use of NVP is associated with resistance to ETR (23, 31, 42). The low and moderate resistance to ETR displayed by isolates 6399 (E138K) and BK132 (E138K, G190E, and H221H/Y) are probably due to the low weight of these mutations on resistance to ETR.

In general, NNRTI mutations that emerge early in treatment-experienced patients, e.g., K103N, do not impair viral replication capacity to significant extend (10). We found that E138K in both pNL-4.3 and AG exerted only modest impact on replication capacity, a finding that helps to explain the frequency of this mutation in our selection, as well as the fact that E138K was usually the first mutation to emerge. In contrast, M230L impairs viral replication capacity by ~8-fold (51). This is consistent with the selection of E138K in all eight isolates, whereas M230L was selected in only two of eight isolates in the present study.

E138K has previously been selected in vitro by other broad-spectrum NNRTIs (4, 8, 13, 17, 20). However, it has low prevalence (0.5%) in clinical samples (43) and only little impact on susceptibility to narrow-spectrum NNRTIs. E138K (GAG->AAG) is found in the p51 subunit of RT as part of the NNRTI binding pocket. The K101 side chain can normally form a salt bridge with E138, but the presence of E138K causes residues to move away from the NNRTI pocket due to the juxtaposition of similar charges that result from the mutation (38), potentially preventing the specific binding of drug. In addition, interactions of Y318 and E138 in the presence of ETR have been described as unusual for NNRTI-RT complexes (45). The emergence of mutations at residues 138 and 318 could disrupt drug binding ability (4, 47). The identification of a new E138R mutation selected by RIL reinforces the importance of this position (4). E138K and other mutations (K101Q, V179E, V189I, G190E, and H221Y) selected in our study are not yet included among the list of ETR RAMs. Although G190E was previously selected by ETR and RIL (4, 47), its role, alone or in combination with other mutations, remains unknown. Both V179E and H221Y have also been previously shown to confer a degree of resistance to ETR (28, 31, 32). Additional studies in our laboratory will evaluate whether the mutations K101Q, V179E, V189I, G190E, and H221Y either alone or in combination with E138K might affect susceptibility to ETR and other NNRIs.

A recent study analyzed the weights of 17 ETR RAMs, based on the DUET-1 and -2 studies (48). Others have also assigned weighted scores for ETR mutations based on different work (32); the assigned scores were 2, 3, 1, and 1 for E138K, V179E, G190E, and H221Y, respectively. As new results of ETR resistance are released, there will be a need to improve algorithms to also take into account the partial resistance that is associated with certain mutations and mutational patterns.

Our in vitro results confirm that ETR is a potent antiviral drug with a higher genetic barrier for resistance than narrow-spectrum NNRTIs. Although the data also suggest that virological response to ETR in NNRTI-naive patients might be superior than in experienced individuals, clinical studies with patient samples will be needed to provide further information on this topic, as well as on the clinical significance of the E138K substitution in cases of ETR treatment failure.


Baseline genotyping of clinical isolates.The first 240 amino acids in the RTs of all of the isolates (BK132, 5326, 5331, 6383, 6399, Mole 03, BG 05, and 10680) were sequenced at baseline and compared to the HIV-1 subtype B reference strain HXB2 (Fig. 1). Amino acid changes were identified in 37 of 240 positions in RT and were more frequent in non-subtype B than in subtype B isolates. Only one of these substitutions, i.e., V90I, present at baseline in the 6383 isolate, is known to be implicated in HIV-1 resistance to NNRTIs with a weighted factor of 1.0 for ETR resistance (48).

Time to development of resistance to ETR.

ETR resistance was selected in culture using CBMCs as described above (Table 1). Despite attainment of high ETR concentrations at weeks 11 to 13, only the BK132 isolate selected the E138E/K mutation, a finding consistent with previous results on the difficulty of selecting resistance to this compound (47). Subsequently, all isolates showed the accumulation of other mutations at weeks 18 to 20 (V90I, K101Q, E138K, V179D/E/F, Y181C, V189I, G190E, H221Y, and M230L). E138K emerged as the first mutation in all cases except for isolate 5331, in which it was detected between weeks 25 and 30. All selected mutations were maintained in culture under drug pressure as other mutations accumulated. A strong association was observed between the presence of E138K and Y181C and between Y181C and substitutions at position 179 for all isolates. Isolate 6399 selected for E138K alone. No differences in the patterns of mutations for ETR were observed between subtype B and non-B viruses. Novel amino acid substitutions not included in lists of published RAMs for ETR (48) were K101Q, E138K, V179E, V189I, G190E, and H221Y.

Time to development of resistance to EFV.

As expected, all isolates had selected for mutations K103N/Q, V106I/M, V179D/V, Y188C/Y, and/or M230L/M that contribute to high-level EFV resistance by weeks 11 to 13 (Table 2). Additional mutations (L100I, V108I/V, K101E/K, and N348N/T) were observed at subsequent passages. This demonstrates the low genetic barrier of EFV. V106I and V106M were selected in subtypes B and C, respectively, due to a well-described polymorphism at this position (6). Other mutations selected by EFV included L100I, K103N/Q, V108V/I, V179D/V, Y188C/L, G190A, H221Y, and N348I.

Resistance profile of ETR-selected variants.

Table 3 shows the phenotypic susceptibility of viruses containing ETR RAMs. At baseline, the mean EC50s for ETR ranged between 1 and 4.5 nM, while those for EFV ranged between 1.2 and 3.5 nM. Isolate 6383 that harbored V90I at baseline had the highest EC50 of 4.5 nM for ETR. Isolate 6399 that selected for E138K alone after 25 to 30 weeks in culture displayed low-level resistance to ETR, EFV, and NVP, with fold resistances of 5.1, 3.7, and 6.7, respectively. Isolate BK132 (subtype B) variants selected at weeks 25 to 30 possessed E138K, G190E, and H221H/Y and displayed moderate and low-level resistance to ETR and EFV, i.e., 8.7- and 3.8-fold, respectively (Table 3). ETR-resistant variants selected by three other non-B isolates at weeks 25 to 30 contained mutations as follows: 6383 (V90I, E138K, V179E Y181C, and M230L), Mole 03 (V90I, E138K, V179D, and Y181Y/C), and BG 05 (E138K, V179F, Y181C, and M230L). These viruses all displayed very high-level resistance to ETR and NVP (>95-fold) and low-to-moderate resistance (<11-fold) to EFV. The presence of Y181C in combination with other RAMs may have contributed to high-level ETR resistance in these situations. As a control, we used the isolate BG05 (E138K, V179F, Y181C, and M230L) and the protease inhibitor LPV. The EC50s of BG05 wt and the BG05-selected variant (E138K, V179F, Y181C, and M230L) were 5.5 ± 2.1 and 6.0 ± 2.0 nM, respectively (fold change [FC] = 1.1).

Impact of E138K in the pNL-4.3 and AG plasmids.

We introduced E138K by site-directed mutagenesis into the pNL-4.3 and AG plasmids to determine its impact on resistance to ETR, EFV, and NVP in CBMCs. The EC50s of pNL-4.3wt and pNL-4.3E138K were 1.6 ± 0.25 nM and 6.2 ± 1.2 nM for ETR (FC = 3.8), 1.73 ± 0.25 nM and 4.8 ± 0.69 nM for EFV (FC = 2.7), and 26 ± 6 nM and 38 ± 10 nM for NVP (FC = 1.4), respectively. Similarly, the EC50s of AGwt and AGE138K were 1.3 ± 0.15 nM and 5.1 ± 0.17 nM for ETR (FC = 3.9), 1.46 ± 0.15 nM and 3.2 ± 0.42 nM for EFV (FC = 2.2), and 40.0 ± 7 nM and 75.0 ± 2 nM for NVP (FC = 1.8), respectively. Decreases in susceptibility were modest for ETR, i.e., 3.8- and 3.9-fold for pNL-4.3E138K and AGE138K, respectively (Fig. 2). EFV and NVP both retained relative sensitivity to the pNL-4.3E138K and AGE138K viruses containing E138K, i.e., FCs of 2.7 and 2.2 for EFV and FCs of 1.4 and 1.8 for NVP, respectively. The EC50s of a control drug, LPV, for pNL-4.3wt and pNL-4.3E138K were 9.5 ± 0.7 nM and 10.0 ± 0.02 nM (FC = 1.05), respectively.

Viral replication capacity.We also studied the relative replication capacity of wt virus and viruses carrying the E138K mutation by infecting TZM-bl cells with serially diluted viral stocks normalized for p24 production. The infectiousness of the wt and E138K viruses was determined by measuring the luciferase activity at 48 h postinfection. At a p24 input of 10,000 pg/ml, the relative replication of E138K compared to wt virus was diminished after 48 h by 2- and 3-fold for pNL-4.3E138K and AGE138K, respectively (P < 0.01 and P < 0.001) (Fig. 3 A and B). In a multiple round infection in CBMCs, viral growth experiments in the absence of drug confirmed that the E138K mutation conferred a 2-fold drop in replication rates for pNL-4.3E138K compared to pNL-4.3wt by both 6 and 8 days after infection (Fig. 3C) (P < 0.001).

  iconpaperstack view older Articles   Back to Top