Development of Human Immunodeficiency Virus Type 1 Resistance to 4′-Ethynyl-2-Fluoro-2′-Deoxyadenosine Starting with Wild-Type or Nucleoside Reverse Transcriptase Inhibitor-Resistant Strains

ABSTRACT

4′-Ethynyl-2-fluoro-2′-deoxyadenosine (EFdA, MK-8591, islatravir) is a nucleoside reverse transcriptase translocation inhibitor (NRTTI) with exceptional potency against wild-type (WT) and drug-resistant HIV-1 in phase III clinical trials. EFdA resistance is not well characterized. To study EFdA resistance patterns that may emerge in naive or tenofovir (TFV)-, emtricitabine/lamivudine (FTC/3TC)-, or zidovudine (AZT)-treated patients, we performed viral passaging experiments starting with WT, K65R, M184V, or D67N/K70R/T215F/K219Q HIV-1. Regardless of the starting viral sequence, all selected EFdA-resistant variants included the M184V reverse transcriptase (RT) mutation. Using recombinant viruses, we validated the role for M184V as the primary determinant of EFdA resistance; none of the observed connection subdomain (R358K and E399K) or RNase H domain (A502V) mutations significantly contributed to EFdA resistance. A novel EFdA resistance mutational pattern that included A114S was identified in the background of M184V. A114S/M184V exhibited higher EFdA resistance (∼24-fold) than either M184V (∼8-fold) or A114S alone (∼2-fold). Remarkably, A114S/M184V and A114S/M184V/A502V resistance mutations were up to 50-fold more sensitive to tenofovir than was WT HIV-1. These mutants also had significantly lower specific infectivities than did WT. Biochemical experiments confirmed decreases in the enzymatic efficiency (k_cat/K_m) of WT versus A114S (2.1-fold) and A114S/M184V/A502V (6.5-fold) RTs, with no effect of A502V on enzymatic efficiency or specific infectivity. The rather modest EFdA resistance of M184V or A114S/M184V (8- and 24-fold), their hypersusceptibility to tenofovir, and strong published in vitro and in vivo data suggest that EFdA is an excellent therapeutic candidate for naive, AZT-, FTC/3TC-, and especially tenofovir-treated patients.

INTRODUCTION

As of 2019, 38 million people worldwide are living with HIV/AIDS, with 1.7 million new HIV infections and 690,000 AIDS-related deaths annually (1). However, AIDS-related morbidity and mortality rates have declined in recent years, largely due to the widespread use of highly active antiretroviral therapy (HAART) (2). HAART is typically based on a combination of two nucleoside reverse transcriptase inhibitors (NRTIs) in combination with a nonnucleoside reverse transcriptase inhibitor (NNRTI), a protease inhibitor, or an integrase inhibitor (3). Azidothymidine (AZT), didanosine (ddI), lamivudine (3TC), emtricitabine (FTC), abacavir (ABC), and tenofovir (TFV) are the six NRTIs included in HAART regimens (3–5). However, the prevalence of HIV strains resistant to these compounds is rapidly increasing, both in treatment-experienced and newly infected patients (5–10). High-level resistance to AZT generally requires multiple mutations, including D67N, K70R, T215F, and K219Q (4, 11–13), while an M184I/V mutation grants resistance to both 3TC and FTC (4, 14–16). Meanwhile, the K65R mutation imparts resistance to TFV (17–20). A great concern is the emergence of virus strains with cross-resistance to multiple NRTIs, which can limit treatment options following viral escape from first-line HAART (5). M184V strains show resistance to ddI and abacavir, while mutations selected by AZT are resistant to FTC; also, K65R has reduced sensitivity to ddI, 3TC, and FTC (4, 17, 19, 21, 22). Use of the currently available NRTIs can lead to toxicity and side effects (5, 23–26). In addition, the success of antiretroviral regimens at preventing HIV-1 infection has moved attention to the question of compliance and increased the interest in therapeutic agents that may be suitable for long interval dosing (27, 28). All of the above are reasons for the development of novel therapeutic agents that combine high antiviral efficacy with improved quality-of-life characteristics.

Several studies have focused on a group of novel 4′-substituted NRTIs, the most promising of which is 4′-ethynyl-2-fluoro-2′-deoxyadenosine (EFdA, MK-8591, or islatravir) (reviewed in reference 29). EFdA is a deoxyadenosine analog with an ethynyl group at the 4′ carbon of the ribose and a fluorine at the 2 position of the adenine base (13, 30–32). EFdA also retains a 3′-OH, unlike all other NRTIs currently used in HIV therapeutic regimens. The presence of the 3′-OH improves the recognition of EFdA as a substrate of the cellular deoxycytidine kinase (13, 33, 34) and of the viral HIV reverse transcriptase (RT) at the deoxynucleoside triphosphate (dNTP) binding site (35–37). This may contribute to the efficient production of EFdA-triphosphate (EFdA-TP) and the efficient incorporation of EFdA-monophosphate (EFdA-MP) into nascent HIV DNA during reverse transcription. Deamination of adenosine analogs can reduce intracellular concentrations of the active metabolite, resulting in a decreased efficacy of a drug; notably, deamination of EFdA by cellular enzymes is greatly reduced by the 2-fluorine substitution, thus increasing the intracellular half-life of the molecule compared to similar nonhalogenated compounds (30, 34). The 4′-ethynyl group of EFdA interacts with a hydrophobic pocket in the polymerase active site of RT, affecting translocation and blocking efficient extension of the DNA primer, despite the available 3′-OH (35, 38, 39). EFdA is therefore termed a nucleoside RT translocation inhibitor (NRTTI). EFdA is highly potent against both wild-type (WT) HIV-1, HIV-2, and NRTI-resistant strains (34, 40–42). The NRTI-resistant virus, K65R, shows hypersensitivity to EFdA (41, 43). EFdA has shown in vivo potency against both HIV-1 in humanized mouse (44–46) and SIV in nonhuman primate (47–49) models. EFdA imparts minimal toxicity in the animal models, as well as in multiple in vitro assays, due to minimal inhibition of human DNA polymerases, leading to a high selectivity index (44, 46, 50). Importantly, EFdA is being tested for once-weekly, once-monthly, and once-yearly dosing regimens (48, 49, 51–53). Recently, Merck Co. and Gilead Sciences announced a collaboration that will focus on the development of oral and injectable formulations of two long-acting anti-HIV therapeutics: islatravir and lenacapavir (GS-6207) (54). Taken together, these results suggest that EFdA has great promise as a therapeutic agent.

To our knowledge, there is no information about the capacity of HIV to develop high-level resistance to EFdA. The M184V RT mutation imparts a low decrease in sensitivity to EFdA (13, 34, 41, 42, 55, 56). M184V, along with I142V and T165R, was also selected during passage of WT virus in the presence of EdA, a related compound that has the 4′-ethynyl, but not the 2-fluoro modification (34). The I142V/T165R/M184V mutant virus exhibited a 22-fold increase in EFdA resistance relative to the WT; however, it is uncertain whether these mutations would arise during passage of virus in the clinically relevant EFdA itself or whether novel mutations or combinations of mutations conferring even greater resistance are possible. It is also not well understood how resistance to EFdA develops in virus strains with specific preexisting resistance mutations to other NRTIs, which would be relevant to its potential use as salvage therapy for patients failing HAART. Hence, there is incomplete information on EFdA-resistant strains and their cross-resistance to NRTIs currently used in the clinic.

Here, we explored how preexisting NRTI resistance mutations affect the development of EFdA resistance. We selected viruses resistant to EFdA by serial passage of HIV-1 that was either WT or resistant to TFV (K65R), 3TC/FTC (M184V), or AZT (D67N/K70R/T215F/K219Q). We found that M184V/I was included in all selected EFdA-resistant strains. In some cases, mutations also appeared in the RT connection domain (R358K and E399K) and the RNase H domain (A502V); the latter three changes did not significantly affect resistance when alone, and they also did not appear to compensate for the loss of specific infectivity imparted by primary mutations. We identified a double-mutant virus with a moderate level of resistance, A114S/M184V (∼24-fold), with greater resistance to EFdA than that observed with M184V alone. In addition, these EFdA resistance mutations were found to be remarkably hypersensitive to tenofovir (∼50-fold increase in susceptibility). However, A114S/M184V RT was found to have reduced catalytic efficiency (∼6-fold) and impart reduced virus-specific infectivity.

RESULTS

Virus breakthrough during serial passage of viruses in increasing EFdA concentrations.

Serial passages were initiated by infecting MT-2 cells with WT (xxLAI), K65R, M184V, or D67N/K70R/T215F/K219Q stock virus in the presence of EFdA and monitored as described in Materials and Methods. At the end of the passaging, all EFdA-selected viruses, regardless the initial strain or number of passages, induced ≥75% syncytium formation in untreated MT-2 cells within 7 days of infection, indicating that they had retained replication fitness.

The ability of the viruses to grow in the presence of EFdA was assessed by time to viral breakthrough (defined as 75% syncytium formation) at each passage. This time to breakthrough and the concentration of EFdA in each passage is shown in Fig. 1. For WT, M184V, and D67N/K70R/T215F/K219Q strains, virus breakthrough typically occurred after approximately 7 days. As EFdA concentrations were raised in passages 8 and 9, the time to virus breakthrough increased, until virus replication was no longer observed in passage 10. Passage 10 corresponded to an EFdA concentration of 550 to 800 nM (Fig. 1A, C, and D).

FIG 1.

Selection of resistance to EFdA by serial passage. Time to breakthrough was determined (in MT-2 cells) as the number of days required for ≥75% syncytium formation, at which point the supernatants were harvested, assayed for p24 content, and used to infect the subsequent passage. EFdA concentration was doubled every second passage until P₆ and then every passage thereafter. Results represent a single trial for each passage. Passages were initiated with WT (xxLAI) (A), K65R (B), M184V (C), and D67N/K70R/T215F/K219Q (D).

The behavior of K65R deviated from the other mutants studied in two ways. First, there was a reduction in the time to breakthrough from passages 1 to 4, likely because K65R (Fig. 1B), which confers hypersensitivity to EFdA (43), began to be lost as early as P₂ and was not detected at passage 6 (Table 1 and Fig. 1B). The time to breakthrough then increased as EFdA concentrations were raised, as observed with other viruses.

TABLE 1.

Amino acid mutations in WT-, K65R-, M184V-, and D67N/K70R/T215F/K219Q-derived viruses during serial passage in progressively increasing concentrations of EFdA

Virus	Passagea	Sequencingb
		Amino acid mutation(s)c	Proportion of sequence population (%)
WT (xxLAI)	0	None	100
	6	M184I	100
	9	M184I	36.4
		M184I, E399K	31.8
		M184V	18.2
		None	13.6
K65R	0*	None	100
	6	R65K, M184I	90.9
		R65K, M184V	9.1
	9	R65K, M184V, A502V	43.5
		R65K, M184I	34.8
		R65K, M184V	17.4
		R65K, M184I, A502V	4.3
	10	R65K, A114S, M184V, A502V	90.9
		R65K, A114S, M184V	4.5
		R65K, A114S, A502V	4.5
M184V	0*	None	100
	6	R358K	45
		None	55
	9	R358K	90.5
		None	9.5
	0*	None	100
D67N/K70R/T215F/K219Q	6	M184V	66.7
		None	33.3
	9	M184V	18.2
		M184V, E399K	81.8

^a*, unpassaged stock virus.

^bPopulation and clonal sequencing of viral RNA from passage supernatants were performed as described in Materials and Methods.

^cRelative to unpassaged stock virus of the appropriate strain.

In every case, the passaged viruses were able to replicate in the presence of greater EFdA concentrations than the parental virus, indicating that some degree of EFdA resistance had developed. The most significant resistance appeared in the last and longest passage initiated with K65R virus (Table 1).

Dose response of EFdA-passaged viruses to EFdA.

To evaluate the degree of resistance selected in the passaging experiments, sensitivity to EFdA was determined for virus supernatants from passage 6 (P₆) onward. Data are presented for the viruses obtained from P₆, P₉, and, in one case, P₁₀. P₆ was the first passage for which extensive sequencing analysis was performed, P₉ was the effective endpoint for the WT-, M184V-, and D67N/K70R/T215F/K219Q-derived viruses, and P₁₀ was chosen as the endpoint for the K65R-derived strain. For each of the viruses, the later passages demonstrated increased EC₅₀ values compared to WT and earlier passage viruses, demonstrating that resistance had been acquired as expected (Fig. 2).

FIG 2.

EFdA dose response for viruses selected during serial passage in EFdA. P4-R5 MAGI cells were treated with various concentrations of EFdA and infected with supernatants from the P₀, P₆, P₉, or P₁₀ (where applicable) passages of experiments using WT or mutant-HIV. Untreated control is P₀ of WT virus with no EFdA treatment. After incubation for 48 h, the cells were lysed, and the β-galactosidase activity was assessed. (A to D) Inhibition of virus replication relative to infected untreated cells. Dashed line represents 50% inhibition of viral replication. The results represent the means ± the standard deviations (SD) from one experiment with three replicates.

Development of amino acid mutations in reverse transcriptase during serial passage of viruses with EFdA.

Clonal sequencing confirmed the identity of the P₀ stock viruses. Similar sequencing was carried out on clones from various passages to identify any changes in RT that may have arisen during replication in the presence of EFdA. Table 1 summarizes the data for P₆ and P₉ isolates for all parental viruses, as well as viruses from P₆, P₉, and P₁₀ for K65R.

The WT P₆ population sequencing revealed only M184I (Table 1); by P₉, approximately 30% of sequences were M184I, 30% were M184I/E399K, and the remainder were divided between M184V and WT. Consistent with the initial passage data, sequencing data showed that all clones of the K65R P₆ virus had reverted the K65R mutation back to WT. The reversion occurred rapidly, since K65R was not seen in the P₂ population. About 90% of clones in P6 contained M184I, with the remainder containing M184V. Sequencing data also showed that the proportion of M184V increased to approximately one-quarter of the population by P₈. The virus harvested following P₉ had a more heterogeneous population, with most sequences containing M184V/A502V or M184I. By P₁₀, the diversity decreased dramatically, and A114S/M184V/A502V became the dominant sequence. A114S/M184V and A114S/A502V were found at a much lower frequency.

During the passages that started with M184V, the connection domain mutation R358K appeared and became increasingly dominant as passaging progressed. Passage of the D67N/K70R/T215F/K219Q virus led to the rapid emergence of M184V (dominant in passage 6). This was joined by E399K in late passages (M184V/E399K was dominant in passage 9). None of the starting AZT-resistance-associated mutations were lost during passaging.

Validation of EFdA-resistance associated mutations using molecular clones.

To confirm that the identified mutations conferred EFdA resistance, we constructed relevant molecular clones in a WT (NL4-3) backbone and characterized their infectivity relative to WT (Fig. 3). M184V was the only individual mutation that conferred ∼8-fold EFdA resistance, consistent with previous reports (42, 57). Individual mutations A114S, R358K, E399K, and A502V had no effect on EFdA resistance (Fig. 3). The addition of A114S to M184V further increased the EFdA resistance, as seen in A114S/M184V (24-fold), A114S/M184V/R358K (28-fold), and A114S/M184V/A502V (25-fold). Hence, the highest levels of EFdA resistance were found in the combined presence of A114S and M184V.

FIG 3.

Resistance to EFdA of HIV-1 mutants selected during passaging. Mutations were introduced into NL4-3. TZM-GFP cells were pretreated with EFdA and infected after 24 h. GFP-positive cells (infected cells) were counted in various concentrations of EFdA. Mutants were normalized to WT infection to produce the fold change. EFdA dose-response curves were produced for each mutant and the EC₅₀s calculated. Statistical significance was determined using one-way ANOVA with Tukey’s posttest (***, P < 0.0001). The results represent the means and SD for four independent experiments performed in duplicate.

Susceptibility of select mutants to TDF and FTC.

To determine whether the various EFdA resistance mutations are susceptible to approved drugs such as tenofovir disoproxil fumarate (TDF) and emtricitabine (FTC), we prepared dose-response curves (Table 2). Remarkably, we found that all mutants with the EFdA resistance phenotype were hypersensitized to TDF. Most notably, the A114S mutation by itself and in the presence of M184V or A502V imparted an up to 50-fold decrease in 50% effective dose (EC₅₀; i.e., an increase in EFdA susceptibility). As expected, all M184V mutants were significantly resistant to FTC (>100-fold), which is known resistance mutation and causes therapy failure. There was a 2- to 3-fold decrease in EC₅₀s for FTC with A114S, A502V, and A114S/A502V strains.

TABLE 2.

EC₅₀ fold change compared to WT (NL4-3)

Parameter	Mean EC50 fold change ± the SD
	TDF	FTC
WT	1	1
A114S	0.04 ± 0.0042	0.54 ± 0.043
M184V	0.57 ± 0.0572	>100
A502V	0.12 ± 0.0544	0.44 ± 0.071
A114S/A502V	0.02 ± 0.0014	0.39 ± 0.027
M184V/A502V	0.15 ± 0.054	>100
A114S/M184V	0.04 ± 0.0027	>100
A114S/M184V/A502V	0.03 ± 0.0052	>100

Replication characteristics of molecular clone viruses.

To determine the effect of mutations on viral fitness, we tested viruses in single-cycle replication assays known as specific infectivity. There were no statistically significant differences between the specific infectivities of WT, R358K, or A502V. M184V had a reduced specific infectivity compared to WT; it did not appear to be statistically significant under these experimental conditions. In addition, M184V, in combination with all other mutants besides R358K, showed a significant decrease in specific infectivity. However, there was a decrease in specific infectivity of the A114S-containing mutants, A114S/M184V and A114S/M184V/A502V, compared to WT (Fig. 4). To further validate this result, we examined the biochemical properties of A114S RT, as described below.

FIG 4.

Single-round replication assays using TZM-GFP cells infected with individual mutants. An ELISA was also performed on the virus to determine amount of p24 (total virus in medium). The ratio of infected cells per p24 was then calculated and normalized to the WT. Statistical significance was determined using a one-way ANOVA with Dunnett’s multiple-comparison test (*, P < 0.05; **, P < 0.01). The results represent the means and SD for four independent experiments performed in triplicate.

Steady-state kinetics and EFdA susceptibility of mutant reverse transcriptases.

To better understand the effect of the mutations on the DNA polymerase activity of RT, we performed steady-state kinetics experiments to determine the catalytic efficiency, k_cat/K_m, for various RT mutants. We cloned, expressed, and purified the mutant RTs listed in Table 3. We found that in the presence of the A114S mutation there was a consistent decrease in the catalytic efficiency k_cat/K_m compared to WT RT: a 2.1-fold decrease for A114S and a 6.5-fold decrease for A114S/M184V/A502V. The changes in catalytic efficiencies of these mutants appeared to be primarily due to the change in the K_m. The K_m for A502V was slightly decreased but comparable to that for the WT. Of note, the catalytic efficiency (k_cat/K_m) for the A502V and M184V single mutants was comparable to that of the WT enzyme (Table 3).

TABLE 3.

Steady-state enzyme kinetics for reverse transcriptase mutants

Parameter	WT RT	A114S	M184V	R358K	E399K	A502V	184V/358K/502V	114S/184V/502V
Km (nM)	1,448	2,758	1,791	2,008	1,795	1,110	2,327	6,659
Fold change, Km	1	1.9	1.2	1.4	1.2	0.8	1.6	4.6
Fold change, kcat/Km	1	2.1	1.2	1.4	1.2	1	1.3	6.5

DISCUSSION

NRTIs are the most widely used therapeutics to treat HIV infection. As such, NRTI-resistant HIV variants are becoming increasingly prevalent in the HIV-1-infected population. EFdA has a strong potential for long-acting first-line therapies for naive HIV-infected individuals or as an agent for preexposure prophylaxis, or salvage therapies. As such, we were interested in identifying and characterizing EFdA resistance mutations that might arise during exposure of WT HIV, and especially NRTI-resistant virus variants, to EFdA. Moreover, there is high interest in discovering novel mutations or combinations of mutations that impart levels of EFdA resistance that may have potential clinical implications.

Regardless the genotype of the starting virus, M184I/V consistently arose during passages through EFdA. M184V confers high-level resistance to both 3TC and FTC (14–16, 58) but only low-level resistance to EFdA (34, 41). Our data confirm the latter, with M184V conferring only an ∼8-fold resistance to EFdA. M184V is also the primary resistance mutation selected during serial passage of WT HIV with the related compounds EdA and Ed4T, suggesting that M184 is critical to the activity of 4′-ethynyl modified nucleoside analogues (34, 59).

D67N/K70R/T215F/K219Q HIV is highly resistant to AZT (11, 12). Similar to another AZT-resistant mutant, M41L/T215Y, which was previously shown to display marginal resistance to EFdA (34), we found that D67N/K70R/T215F/K219Q demonstrated a 1.8-fold increase in EFdA resistance compared to WT. As we have previously shown, the excision unblocking mechanism of resistance is not a major challenge for EFdA; although EFdA can indeed be excised, the efficiency of reincorporation is so high that the net result is not significant overall excision (38). The K65R mutation confers resistance to TFV and is the mutation responsible for virological failure in TFV-based therapies. K65R is also cross-resistant or selected during therapy with ABC, ddI, and 3TC/FTC (5, 17, 19–22). We previously showed that K65R is hypersensitive to EFdA, up to 5-fold lower EC₅₀ compared to the WT (34, 41, 43). The present data are consistent with this finding, since K65R displayed a longer time to breakthrough during initial passage in EFdA, and this breakthrough was associated with the loss of K65R. The rapid reversion of this mutation not only leads to a reduction in sensitivity to EFdA but also confers increased replication fitness to the virus. A114S and A502V were found after the reversion of K65R. Although there is no clear explanation for this observation, it may reflect a stochastic aspect of the drug resistance selection process. In addition, it is possible that the additional passaging steps may have contributed to conditions that led to high concentrations of EFdA, thus increasing the selective pressure on RT and allowing A114S to prevail.

Our results highlight the challenges of selecting for EFdA resistance. Even when starting the passages using a variety of clinically relevant genotypic backgrounds, mutations at residue 184 were consistently the first to emerge. Though in previous studies it was shown that M184V exhibited reduced viral fitness, we observed here a decrease in specific infectivity that was not statistically significant (60–62). This may be related to assay-condition-specific differences. In addition, EFdA was fully effective in maintaining suppression of mutant virus throughout the drug treatment period in WT SIV-infected rhesus macaques (47). EFdA-treated humanized mice infected with HIV remained fully suppressed while on treatment even when M184I emerged after therapy cessation (44). In addition, M184V is known to fail current therapies: 3TC and FTC (reviewed in references 63–67). Here, we have identified a novel EFdA resistance mutation, A114S, which appeared after ∼3 months of EFdA treatment (Fig. 1). The addition of A114S to the M184V background (A114S/M184V, A114S/M184V/R358K, and A114S/M184V/A502V) enhanced EFdA resistance up to ∼28-fold, or 350% higher than M184V by itself. Importantly, both virological and biochemical experiments demonstrated that this mutation, especially in the background of M184V, has a negative impact on specific infectivity (Fig. 4 and Table 3). Similar results were recently presented at a conference reporting that A114S with M184V conferred resistance but had minimal effect on its own (68). Specifically, it seems that the decrease in fitness is likely due to decreased binding of incoming deoxynucleoside triphosphates (dNTPs), reflected by the increased K_m, in all cases where A114S mutation was present. The A114S mutation has a longer side chain, which may extend closer to the 3′-OH of an incoming dNTP, thus reducing the available space in the binding pocket, leading to increased selectivity for the canonical dNTP. Consistent with previous data, we found that the A114S mutation decreased the DNA polymerase activity of RT (69, 70). Residues 114 and 184 are located at opposing sides of a hydrophobic pocket, along with Y115, F160, and the aliphatic chain of D185, which accommodates the 4′-ethynyl group of EFdA (35, 71). We speculate that the bulkier side chains of V184 and S114 impinge into the substrate envelope of the 4′-ethynyl binding pocket, thus causing a decrease in dNTP binding that may contribute to the observed decrease in specific infectivity. Nevertheless, A114S-containing mutants may not be a major problem in patients; A114 mutations are rarely seen in clinical samples. According to the Stanford HIV Drug Resistance database, A114S mutants have appeared a total of 7 times in 172,714 sequenced samples (72, 73).

In addition, we have found that TDF inhibits the A114S/M184V EFdA-resistant HIV, significantly more efficiently than the WT (Table 2). We have previously reported that the K65R TDF resistance mutation in RT also hypersensitizes HIV to EFdA (43). Thus, a combination of tenofovir and EFdA could be a very efficient therapy since they complement each other’s resistance profile.

Selected residues R358K and E399K are proximal to G359-A360 and K395-E396, respectively, in the RNase H primer grip region that interacts with the DNA primer strand (74–77). R358K is selected both in NRTI-treated patients and during passage with NRTIs in tissue culture (78–82). E399 is located within a cluster of several conserved tryptophan residues involved in RT dimerization (83, 84). We found no significant variation in EFdA resistance between M184V/E399K or M184V/R358K and M184V alone. Hence, neither R358K nor E399K contribute directly to EFdA resistance. We initially considered that these mutations might provide a fitness benefit. However, we did not observe improved specific infectivity in our single-round replication assays. The A502V late-appearing mutation is located in the primer grip region of the RNase H domain, near residues Y501 and I505 (75). Under the conditions tested, we did not see changes in the biochemical fitness of A502V, but we did see minor decreases in K_m. M184V/A502V HIV subsequently acquired the A114S mutation, which coincided with a significant increase in EFdA resistance.

A major concern when introducing any new NRTI to clinical use is the potential for cross-resistance with current NRTIs. Our results confirm that EFdA drives the selection of M184I/V, producing virus with high-level cross-resistance to 3TC and FTC. However, EFdA is a potent inhibitor, even against M184I/V; it retains the ability to inhibit replication of these mutants at therapeutic doses (85). Conversely, M184I/V increases AZT sensitivity in both WT and AZT-resistant backgrounds (10, 16, 86, 87). Similarly, K65R confers resistance to TFV but is hypersensitive to EFdA (43); these data suggest that EFdA may be well suited for combination therapies. We plan to study these mutations in primary cells, as well as various subtypes, in future studies to further confirm our findings and understand EFdA inhibition for implementation on a global scale.

In summary, our data demonstrate that there is a strong barrier to HIV developing high-level EFdA resistance. The development of >10-fold resistance to EFdA required the combination of multiple mutations, including both M184V and A114S, that appear to negatively affect virus-specific infectivity. These RT mutants were also found to be dramatically hypersensitized to tenofovir (∼0.04-fold change compared to WT). Hence, EFdA has the potential to become a highly effective therapeutic and in combination with tenofovir.

MATERIALS AND METHODS

Reagents.

EFdA was synthesized by Life Chemicals (Burlington, Ontario, Canada). Stock solutions (10 mM) of EFdA was prepared in dimethyl sulfoxide and stored in aliquots at −20°C. MT-2 cells (88–90) were cultured in RPMI 1640 medium (Mediatech, Inc., Manassas, VA), supplemented with 10% fetal bovine serum (FBS; HyClone, Logan, UT) and 2 mM l-glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin (all from Mediatech, Inc.). P4-R5 MAGI cells were cultured in Dulbecco modified Eagle medium (DMEM; Mediatech, Inc.), supplemented with 10% FBS, 2 mM l-glutamine, 100 U/ml penicillin, 100 μg/ml streptomycin, and 1 μg/ml puromycin. HEK-293 (91) and HEK-293/17 (92) cells were cultured in DMEM supplemented with 10% FBS, l-glutamine, and penicillin/streptomycin. TZM-GFP cells from Massimo Pizzato laboratory (Trento University) (93–95) were cultured in DMEM (Corning) supplemented with 10% Serum Plus, 2 mM l-glutamine (Thermo Fisher, Waltham, MA), and 100 U/ml penicillin/streptomycin (Thermo Fisher). Jurkat LTR-GFP CCR5+ cells (JLTRG-R5) (96, 97) were cultured in RPMI 1640 medium (Cytiva Life Sciences, Marlborough, MA) supplemented with 10% FBS and 100 U/ml penicillin/streptomycin.

Generation of virus stocks and molecular clones.

Initial K65R, M184V, and D67N/K70R/T215F/K219Q viruses were generated by site-directed mutagenesis on an xxLAI HIV-1 backbone using the QuikChange XL site-directed mutagenesis kit (Agilent Technologies, Inc., Santa Clara, CA) according to the manufacturer’s protocols. Subsequently, 6 × 10⁵ 293-T cells were transfected with 10 μg of viral DNA using the PrimeFectimine mammalian transfection reagent (PrimGen, Oak Park, IL). After 72 h of incubation, HEK-293/T cell supernatants were harvested, filtered, and used to infect 1.8 × 10⁶ MT-2 cells. Infected MT-2 cells were incubated at 37°C (5% CO₂), inspected daily, and infectious virus was harvested at ≥50% syncytium formation. Emory’s Cloning Core was used to make individual mutants in the backbone of NL4.3. Virus stocks were made using HEK-293/17 cells that were transfected with 6 μg of viral DNA using Xtreme-GENE HP (Roche, Basel, Switzerland) transfection reagent on a 10-cm dish of 60% confluence. After 48 h of incubation, HEK-293/17 cell supernatants were harvested and concentrated overnight with a Lenti-X concentrator (Clontech, Mountain View, CA) according to the manufacturer’s protocol.

Determination of TCID₅₀ values and p24 content.

We determined 50% tissue culture infective doses (TCID₅₀) by infecting 5 × 10⁴ MT-2 cells per well in 96-well flat-bottom plates with 4-fold serial dilutions of virus stock. Three replicates were performed for each virus. Infected plates were inspected daily for syncytium formation; every 3 days, half of the supernatant was replaced with fresh media. The assay was terminated when no additional syncytium formation was noted for 2 days. The TCID₅₀ was then calculated using the Reed-Muench method (98). The p24 content of each virus stock was determined using an HIV-1 p24^CA Antigen Capture assay kit (SAIC-Frederick, Frederick, MA).

Serial passage for selection of resistant virus.

MT-2 cells were suspended at 2.5 × 10⁵ cells per ml in 10 ml of medium containing EFdA. Initial EFdA concentrations were chosen based on the EC₅₀ of WT stock virus (8.6 nM), with a similar amount chosen for the K65R virus and slightly higher concentrations for the D67N/K70R/T215F/K219Q (10 nM) and M184V (12 nM) strains. Passages were initiated by immediately adding 200 TCID₅₀ of the appropriate unpassaged (P₀) virus stock to the cells, followed by gentle mixing. Untreated cultures were initiated by infecting 2.5 × 10⁵ MT-2 cells per ml with 200 TCID₅₀ of P₀ virus stock in 10 ml of drug-free medium. All passages and untreated cultures were grown in T-25 tissue culture flasks. Every 2 to 3 days, the cells were split 1:3 with fresh media containing the appropriate concentration of EFdA. Cultures were visually inspected every 1 to 2 days for the presence of syncytia. At ≥75% syncytium formation, culture supernatants were harvested, concentrated using Amicon Ultra Ultracel 100K centrifugal filters (Millipore, Carrigtwohill, Co. Cork, Ireland) and syringe-filtered through 0.22-μm filters (Millipore). The p24 content of the resulting first passage (P₁) supernatant was determined as described above. This procedure was followed for all subsequent passages, with P_N initiated by infecting 2.5 × 10⁶ MT-2 cells with P_N-1 virus supernatant in media containing the appropriate concentration of EFdA. Untreated cultures were also initiated by infecting 2.5 × 10⁶ MT-2 cells, in drug-free media, with P_N-1 virus. Since infectivity differences were expected between the P₀ virus and the output strains from each passage, P₂ and all subsequent passages were initiated by infecting cells with a p24 amount of P_N-1 virus equivalent to the P₀ virus p24 content used to initiate P₁. The concentration of EFdA was doubled every two passages up to P₆, and then the amount of drug was doubled with every passage. Passages lasting more than 60 days without syncytium formation were terminated and repeated, along with the previous passage. If no syncytia were noted in a repeat passage after 30 days, the passage was terminated, and no further attempts were performed.

Sequencing of passaged viruses.

Viral RNA was purified from supernatants using the QIAamp Viral RNA minikit (Qiagen, Valencia, CA), the concentration was determined with a Spectronic BioMate*3 UV spectrophotometer (Thermo Scientific), and 500 ng was used as the template for cDNA synthesis. First-strand PCR was performed using random hexamer primers and the SuperScript III first-strand synthesis system for RT-PCR (Invitrogen, Carlsbad, CA). The resulting cDNA was PCR amplified using the HIV-1 LAI-specific primers ABR-RT-OF (1763 5′-GGAGCCGATAGACAAGGAACTG-3′) and ABR-RT-OR2 (3863 5′-GGCTACTATTTCTTTTGCTACTACAGG-3′). These primers anneal to the 3′ end of gag and the 5′ end of integrase, respectively, and generate a 2,127-bp product spanning the full length of the reverse transcriptase gene. PCR was performed using the Expand High-Fidelity PCR System dNTPack (Roche Diagnostics GmbH, Mannheim, Germany), with 4.5 mM MgCl₂ and 3.5 U of enzyme mix used for each reaction. Reactions were run in a PCR Sprint bench-top PCR cycler (Thermo Electron) with an initial denaturation of 3 min at 94°C, followed by 30 amplification cycles, each consisting of 30 s at 94°C, 45 s at 58°C, and 150 s at 72°C. After a 7-min final extension at 72°C, the samples were used immediately or stored at −20°C.

PCR products were separated by electrophoresis on a 1% agarose gel, and the 2,127-bp band was harvested using a QIAquick gel extraction kit (Qiagen). The concentration of DNA was determined by spectrophotometry, and 500 ng of sample was submitted to ACGT, Inc. (Wheeling, IL), for full-length, double-stranded population sequencing. The P₀ consensus sequences from each virus were aligned in Clustal X-2 (99) with an independent HIV-1 LAI nucleotide sequence (accession no. NC_001802) to ensure the stock viruses did not contain unexpected mutations. Nucleotide consensus sequences from each passage were aligned to the appropriate P₀ consensus. Chromatograms were also inspected visually using Chromatogram Explorer and DNABaser (Heracle BioSoft S.R.L., Pitesti, Romania) for the presence of heterogeneous peaks and minority sequence populations not detectable in the consensus sequence.

Clonal sequencing of full-length reverse transcriptase gene PCR products.

Approximately 105 ng of full-length PCR product was ligated into the pGEM-T vector system (Promega, Madison, WI) at a 3:1 insert/vector molar ratio and incubated overnight at 4°C. Ligations were transformed into MAX Efficiency DH5α competent cells (Invitrogen) by heat shock. Blue-white screening was used to select clones with successful ligations, and plasmids containing the full-length reverse transcriptase gene were isolated with a QIAprep Spin Miniprep kit (Qiagen). A minimum of 20 clones from each sample were sequenced. Primers ABR-RT-OF and ABR-RT-OR2 were used to sequence the 5′ and 3′ ends of reverse transcriptase, while an internal portion of the gene was sequenced with primer ABR-RT-IF (2211 5′-CAGAGATGGAAAAGGAAGGG-3′). Clones were aligned with the appropriate P₀ stock virus consensus in Clustal X-2, and the proportion of sequences with the novel substitutions was determined.

Dose response to individual mutants.

All individual mutations were cloned into pNL4.3 using Emory’s Cloning Core. After each virus was generated as described above, TZM-GFP cells were plated at 10,000 cells/well in a 96-well plate and with serially diluted EFdA, TDF, or FTC starting at 100 nM (for EFdA) or 10 μM (TDF and FTC). The cells and drug were incubated for 24 h. Next, the cells were infected with virus and a 1-μg/ml final concentration of DEAE-dextran, followed by incubation for 48 h. The green fluorescent protein (GFP)-positive cells were then counted using Cytation 5 (BioTek, Winooski, VT) with Gen5 v3.03 software. EC₅₀ curves were then determined using Prism 5 (GraphPad) software.

Specific infectivity.

TZM-GFP cells were plated at 10,000 cells/well in a 96-well plate and incubated for 24 h. Then, cells were infected with the various concentrations of the virus and a 1-μg/ml final concentration of DEAE-dextran, followed by incubation for 48 h. The GFP-positive cells were then counted as described above. After 48 h of incubation, the cells were imaged using Cytation 5 to quantify the numbers of GFP-positive cells, which were then split, and the medium was replaced. The cells were subsequently imaged. The p24 content of each virus was also determined using an enzyme-linked immunosorbent assay (ELISA).

Steady-state kinetics and in vitro IC₅₀s.

HIV-1 RT and mutants were expressed and purified as described previously (36, 77, 100–104). RT was expressed in JM109 cells (Invitrogen) and purified by nickel affinity chromatography and Mono-Q anion-exchange chromatography. Steady-state kinetic parameters, K_m, for the incorporation of dNTPs were determined using plate-based assays measuring an 18-nucleotide primer annealed to a 100-nucleotide DNA template (primer, 5′-GTCCCTGTTCGGGCGCCA; template, 5′-TAGTGTGTGCCCGTCTGTTGTGTGACTCTGGTAACTAGAGATCCCTCAGACCCTTTTAGTCAGTGTGGAAAATCTCTAGCAGTGGCGCCCGAACAGGGAC-3′) (105–110). The reactions were carried out in RT buffer with 6 mM MgCl₂, 40 nM Td100/Pd18, and 10 nM RT in a final volume of 20 μl for 30 min at 37°C and then arrested by 100 mM EDTA. The QuantiFluor dsDNA System (Promega) was used to quantify the amount of formed double-stranded DNA. Reactions were read at ex/em 504/531 nm in a Perkin-Elmer EnSpire Multilabel plate reader. K_m values were determined graphically using the Michaelis-Menten equation and Prism 5.

Statistics.

The statistical significance was determined using Prism 5. A one-way analysis of variance (ANOVA) was performed, as well as a Dunnett’s or Tukey’s multiple-comparison test (as mentioned in the figure legends.

Data availability.

Mutation sequences for BankIt2501100 Seq1 to Seq14 have been deposited in GenBank under accession no. OK263149 to OK263162.