Doravirine and Islatravir Have Complementary Resistance Profiles and Create a Combination with a High Barrier to Resistance

ABSTRACT

Doravirine (DOR), a non-nucleoside reverse transcriptase inhibitor (NNRTI), was approved for treatment of HIV-1 infection in 2018. In the pivotal phase 3 trials, DRIVE-FORWARD and DRIVE-AHEAD, 7 out of 747 (0.9%) treatment-naive participants treated with DOR plus two nucleos(t)ide reverse transcriptase inhibitors (NRTIs) met protocol-defined virologic failure criteria and showed phenotypic resistance to DOR at week 48. The most common DOR resistance-associated mutation (RAM) detected in 5 of the 7 resistant isolates was F227C. Six isolates bearing NRTI RAMs (M184V and/or K65R) were resistant to lamivudine (3TC) and emtricitabine (FTC) but not to other approved NRTIs. All DOR-resistant isolates were susceptible or hypersusceptible (fold change of <0.25) to islatravir (ISL), a nucleoside reverse transcriptase translocation inhibitor (NRTTI). Isolate hypersusceptibility to ISL required F227C, in contrast to zidovudine, an NRTI, which required M184V. Based on the frequent emergence of F227C, we hypothesized that DOR and ISL would create a combination (DOR/ISL) with a high barrier to resistance. In de novo resistance selection studies in MT4-GFP cells (MT4 cells engineered to express green fluorescent protein), DOR/ISL synergistically prevented viral breakthrough at a threshold of 2× the half-maximal inhibitory concentration (IC₅₀). DOR/ISL exhibited a higher barrier to resistance than DOR/3TC and dolutegravir (DTG)/3TC. Resistance analysis showed no emergence of substitutions at F227, an observation consistent with its ability to confer hypersusceptibility to ISL. Overall, the data demonstrate that DOR/ISL creates a 2-drug combination with a higher barrier to resistance, consistent with the reported clinical activity.

INTRODUCTION

Human immunodeficiency virus type 1 (HIV-1) reverse transcriptase (RT) plays an essential role in the life cycle of the virus by converting the single-stranded RNA genome into double-stranded DNA that can be integrated into the host genome. HIV-1 RT accomplishes this conversion via a series of carefully orchestrated events that are made possible by the polymerase and RNase H catalytic activities of the enzyme (1, 2). RT is a heterodimeric enzyme composed of a p66 subunit that contains the catalytic activities and a structural p51 subunit that is missing the C-terminal portion of the larger p66 subunit. Given its critical role in reverse transcription, inhibition of RT has emerged as one of the primary therapeutic strategies for developing antiviral agents to block replication of HIV-1 (3, 4). Two classes of RT inhibitors are currently approved for HIV-1 infection: nucleoside or nucleotide reverse transcriptase inhibitors (NRTIs) and non-nucleoside reverse transcriptase inhibitors (NNRTIs). NRTIs are active-site-directed substrate analogues, for example, 3′-azido-2′-deoxythymidine (ZDV, zidovudine) and 2′,3′-dideoxy-3′-thiacytidine (3TC, lamivudine), that compete with the natural substrate for enzyme inhibition. NNRTIs, including nevirapine (NVP), efavirenz (EFV), etravirine (ETR), rilpivirine (RPV), and doravirine (DOR), have chemical structures that are unrelated to the natural substrate; they bind in an allosteric hydrophobic pocket within the p66 subunit 10 Å from the polymerase active site (5–7). NNRTI binding causes conformational changes in RT that repositions the active-site residues into an inactive conformation, resulting in enzyme inhibition (1, 8, 9).

In addition to RT inhibitors, other major classes of approved antiretroviral agents include integrase strand transfer inhibitors (InSTIs) and protease inhibitors (PIs). As the effectiveness of antiretroviral agents can be impacted by emergence of resistance-associated mutations (RAMs), regimens that include a combination of drugs that target different mechanisms are recommended for treatment. The current standard of treatment for people living with HIV-1 (PLWH) is highly active antiretroviral therapy (HAART), which is typically composed of two or three drugs with at least two different mechanisms of action (10). Patients undergoing HAART have experienced profound and continuous viral suppression, in many cases with substantial immune system recovery leading to a halt in the progression to clinical disease (11, 12). Consensus guidelines for the use of HAART in antiretroviral-naive subjects recommend the use of 2 NRTIs in combination with an InSTI, an NNRTI, or a boosted PI as well as 2-drug combinations in some settings (13, 14). Despite the effectiveness of HAART, the impact of RAMs always remains a concern, as a single RAM can often lead to significant reductions in susceptibility to inhibitors within a class (15–17). Hence, antiretrovirals with a high genetic barrier for the development of resistance are of the utmost importance to overcome treatment failure from emergence of resistance.

Here, we investigated two novel antiretrovirals, DOR and ISL (previously known as MK-8591), as the components of a two-drug combination regimen. DOR is an approved NNRTI indicated in combination with other antiretrovirals for treatment of HIV-1 infection. It exhibits potent antiviral activities across subtypes and variants bearing common NNRTI RAMs that are largely responsible for transmitted drug resistance (TDR), including K103N, Y181C, G190A, and K103N/Y181C (18, 19). ISL is a novel nucleoside analogue that inhibits reverse transcription by blocking RT translocation, a mechanism that is better described as NRTTI (nucleoside reverse transcriptase translocation inhibitor) (20). Like DOR, ISL is highly active across HIV-1 subtypes and variants bearing common NRTI RAMs that are largely responsible for TDR, including thymidine analogue mutations (TAMs), K65R, and M184I/V (21). As combination partners, DOR and ISL were evaluated against viruses bearing substitutions in RT that emerged in DOR phase 3 clinical studies. The viruses were constructed based on the mutations identified in isolates from treatment-naive participants who met the criteria for protocol-defined virologic failure (viral rebound after suppressing to HIV-1 RNA of <50 copies/mL) after treatment with DOR plus 2 NRTIs and showed phenotypic resistance to DOR in the pivotal phase 3 DOR clinical trials, DRIVE-FORWARD and DRIVE-AHEAD. In addition, the barrier to resistance of the independent compounds and the combination were assessed in vitro. Characterization of emergent viruses from the resistance selection studies revealed complementary resistance profiles and a high barrier to resistance for the combination partners.

RESULTS

Susceptibility of clinical isolates from participants who met criteria for protocol-defined virologic failure and had phenotypic resistance to DOR in DRIVE-FORWARD and DRIVE-AHEAD.

The pivotal phase 3 clinical trial DRIVE-FORWARD evaluated the efficacy of DOR plus 2 NRTIs (either tenofovir disoproxil fumarate [TDF] with emtricitabine [FTC] or abacavir [ABC] with 3TC) compared with ritonavir-boosted darunavir (DRV/r) plus 2 NRTIs (as described above) in treatment-naive HIV-1 participants. The primary endpoint of the trial was the proportion of participants achieving HIV-1 RNA of <50 copies/mL after 48 weeks of treatment. The efficacy of DOR (84%) was noninferior to that of DRV/r (80%) (22). In this trial, 1 out of 383 participants in the DOR arm met the criteria for protocol-defined virologic failure and developed phenotypic resistance to DOR (Table 1). Mutant viruses containing each NNRTI substitution identified in the clinical trial were established, and their susceptibility to DOR was evaluated in vitro. Variants containing the V106I or H221Y substitution were susceptible to DOR, with 1.6- or 4.6-fold potency reduction compared to the wild type (WT), respectively; the F227C variant was resistant, conferring a 70-fold reduction in potency to DOR.

TABLE 1.

Genotypic and phenotypic resistance of clinical isolates from participants who met criteria for protocol-defined virologic failure and had phenotypic resistance to DOR in DRIVE-FORWARD and DRIVE-AHEAD^a

Genotype (amino acid substitutions)	Phenotype (fold change in potencyb) of NNRTI susceptibility
	DOR	EFV
DRIVE-FORWARD
V106I/H221Y/F227C/M184V	>96	1.7 (S)
DRIVE-AHEAD
A98G/F227C/M184V	>93	9.1
A98G/V106I/H221Y/F227C/M184V	>110	19
V106A/P225H/Y318F/K65R	>210	4.8
V106I/F227C	>105	2.5 (S)
V106M/F227C/M184V/K65R	>98	11
Y188L/M184V	>181	>120

^aData are from studies conducted at Monogram Bioscience. DOR, doravirine; EFV, efavirenz; NNRTI, non-nucleoside reverse transcriptase inhibitor; RAMs, resistance-associated mutations; NNRTI RAMs, regular font; NRTI RAMs, boldface font.

^bClinical or biological fold cutoffs for inhibitor (EFV) susceptibility (S), 3.

A second pivotal phase 3 clinical trial, DRIVE-AHEAD, evaluated the efficacy of DOR as a fixed-dose combination with 3TC and TDF (DOR/3TC/TDF) compared with the fixed-dose combination of EFV/FTC/TDF in treatment-naive HIV-1 participants. The primary endpoint of the trial was the proportion of participants achieving HIV-1 RNA of <50 copies/mL after 48 weeks of treatment. The efficacy of DOR (84.3%) was noninferior to that of EFV (80.8%) (23). In this trial, 6 out of 364 participants met the criteria for protocol-defined virologic failure in the DOR arm and developed phenotypic resistance to DOR (Table 1).

Two of the seven DOR-resistant clinical isolates from both trials were susceptible to EFV. Based on the mutations identified from this trial, additional mutant viruses were established: A98G, P225H, Y318F, and Y188L. The susceptibility of the variants to DOR was evaluated in a multiple-cycle assay in MT4-GFP cells (MT4 cells engineered to express green fluorescent protein). The variants conferred 5.8-, 4.3-, 9.9- and >100-fold reductions in susceptibility to DOR, respectively. Previous susceptibility studies (PhenoSense data) with a V106M variant showed the substitution conferred a 3.4-fold potency reduction to DOR.

Susceptibility of DOR-resistant clinical isolates to approved NRTIs and islatravir.

As DOR was administered in both pivotal phase 3 trials with 2 NRTIs (ABC/3TC, FTC/TDF, or 3TC/TDF), susceptibility of all seven DOR-resistant clinical isolates from the virologic failure participants (1 from DRIVE-FORWARD and 6 from DRIVE-AHEAD) to a panel of NRTIs was assessed (Table 2). The five clinical isolates containing M184V were substantially resistant to 3TC and FTC with potency reductions of >100-fold. All seven isolates were susceptible to stavudine (d4T), didanosine (ddI), ABC, and TFV. The impacts of the DOR-resistant isolates on the antiviral activity of ZDV and TFV were notable. Five of the variants showed hypersusceptibility (defined by a fold shift of <0.25) to ZDV, while the other two remained susceptible. Isolates bearing M184V showed enhanced susceptibility to TFV (Table 2). All isolates were susceptible to ISL, with one (Y188L/M184V) conferring a moderate 6.9-fold reduction in IC₅₀. Interestingly, two of the DOR-resistant isolates were hypersusceptible to ISL (Table 2).

TABLE 2.

Susceptibility of DOR-resistant clinical isolates to approved NRTIs and islatravir^a

Virus	Fold change in potencyb
	ZDV	d4T	ddI	ABC	FTC	3TC	TFV	ISL
WT	1.0	1.0	1.0	1.0	1.0	1.0	1.0	1
A98G/F227C/M184V	0.1	0.7	1.2	2.7	>100	>100	0.5	2.46
A98G/V106I/H221Y/F227C/M184V	0.1	0.6	1.2	3.2	>100	>100	0.6	2.50
V106A/P225H/Y318F/K65R	1.0	1.4	1.8	2.4	7.7	12	1.6	0.23
V106I/F227C	0.2	0.7	1.0	0.7	2.8	3.1	0.3	0.24
V106I/H221Y/F227C/M184V	0.2	0.8	1.1	3.9	>100	>100	0.4	2.67
V106M/F227C/M184V/K65R	0.1	0.5	1.5	2.8	>100	>100	0.4	1.4
Y188L/M184V	0.5	0.8	1.6	2.9	>100	>100	0.8	6.93

^aData are from studies conducted at Monogram Bioscience. ZDV, zidovudine; d4T, stavudine; ddI, didanosine; ABC, abacavir; FTC, emtricitabine; 3TC, lamivudine; TFV, tenofovir. NNRTI RAMs, regular font; NRTI RAMs, boldface font.

^bFold change in potency relative to activity against wild-type HIV-1. Clinical or biological fold cutoffs for inhibitor susceptibility were the following: ZDV, 1.9; d4T, 1.7; ddI, 1.3 to 2.2; ABC, 4.5 to 6.5; FTC, 3.5; 3TC, 3.5; TFV, 1.4 to 4.

To assess if the detected NNRTI RAMs contributed to the observed inhibitor susceptibility, viruses bearing the relevant substitutions were generated in the absence of NRTI RAMs by site-directed mutagenesis (SDM) and evaluated against the NRTIs and ISL. In the absence of NRTI RAMs, the NNRTI RAM-containing viruses were susceptible to all of the NRTIs except one (A98G/V106I/H221Y/F227C), which conferred a moderate 3.9-fold reduction in susceptibility to 3TC (Table 3). No hypersusceptibility (fold shift of <0.25) was evident with ZDV; however, two (A98G/F227C and V106I/F227C) of the five original isolates demonstrated enhanced susceptibility (fold-shift of <1) to ZDV in the absence of M184V.

TABLE 3.

Activity of NRTIs in viruses engineered to exclude NRTI but maintain NNRTI resistance-associated mutations

Virus	Fold changea
	ZDV	d4T	ddI	ABC	FTC	3TC	TFV	ISL
WT	1.0	1.0	1.0	1.0	1.0	1.0	1.0	1.0
A98G/F227C	0.3	1.3	1.1	1.7	2.7	0.4	1.5	0.2
A98G/V106I/H221Y/F227C	1.9	0.9	1.2	0.9	0.5	3.9	0.7	1.0
V106A/P225H/Y318F	1.7	0.9	0.9	0.8	0.6	1.2	1.2	0.8
V106I/F227C	0.3	0.6	0.7	0.7	0.7	0.4	0.3	0.4
V106I/H221Y/F227C	1.4	1.0	0.9	0.8	0.9	1.0	0.9	0.5
V106M/F227C	0.8	1.0	0.1	0.7	0.9	0.4	0.8	0.4
Y188L	1.5	1.4	1.6	1.5	1.4	0.9	1.8	1.5

^aFold change in potency relative to activity against wild-type HIV-1. ZDV, zidovudine; d4T, stavudine; ddI, didanosine; ABC, abacavir; FTC, emtricitabine; 3TC, lamivudine; TFV, tenofovir; ISL, islatravir.

Unlike ZDV, exclusion of NRTI RAMs from the DOR-resistant variants did not abolish the enhanced susceptibility to ISL (Table 3). Notably, the sensitizing effect was extended to variants that originally harbored M184V. As the variants that exhibited enhanced susceptibility to ISL contained F227C with or without A98G or V106I, viruses bearing these single substitutions were also generated and tested, among others (see Table S1 in the supplemental material). Only the F227C-containing variant was hypersusceptible to ISL, demonstrating that the substitution was necessary and sufficient (in the absence of other substitutions) to elicit the enhanced susceptibility. As M184V negatively impacts the potency of ISL and 3TC, the variants lacking NRTI RAMs (particularly M184V) were evaluated against 3TC as well. Unlike ISL, the enhancing effect of the F227C substitution was not observed with 3TC (Table S1).

In vitro resistance selection with two-drug combinations.

We inferred from the enhanced antiviral activity of ISL in the presence of DOR-resistant mutations that the two inhibitors would create a combination with a higher barrier to resistance. To test this hypothesis, in vitro resistance selection studies were conducted with DOR/ISL and compared to DOR/3TC (as control) and the marketed dual combination of dolutegravir (DTG) and 3TC. The combinations were tested at comparable potencies by evaluating them at equivalent multiples of the individual compound’s in vitro IC₅₀ against wild-type HIV-1 in MT4-GFP cells as described in Materials and Methods. All the combinations were evaluated over a concentration range of 0.25× to 8× IC₅₀ of each compound in the combination. Each combination was monitored for viral breakthrough, and RNA was isolated from wells with viral breakthrough and sequenced to assess the presence of RAMs.

The combination of DOR/ISL efficiently prevented viral breakthrough, with only 3/16 (14%) wells showing evidence of viral replication at a concentration of 2× IC₅₀ and none above that concentration (Fig. 1). The other inhibitor combinations required higher concentrations (4× to 8× IC₅₀) to suppress viral breakthrough. At concentrations of 4× IC₅₀, the combinations of DOR/3TC and DTG/3TC had 8/16 (50%) and 15/16 (94%) wells, respectively, showing evidence of viral replication.

FIG 1.

De novo resistance selection studies in MT4-GFP cells with inhibitor combination pairs of DOR/ISL (A), DOR/3TC (B), and DTG/3TC (C). Expression of GFP under the regulation of HIV-1 tat and rev turns MT4-GFP green and signifies viral breakthrough; the results shown are from passage 13. Rows A through H represent 8 replicates of the concentrations tested in columns 1 through 12 in the study. Inhibitor concentrations (as fold multiples of IC₅₀) tested in columns 1 through 11 were 8×, 4×, 4×, 2×, 2×, 1×, 1×, 0.5×, 0.5×, 0.25×, and 0.25× (as outlined in Table 1). Column 12 had no inhibitor added and served as a control. Abbreviations for inhibitors studied: 3TC, lamivudine; DOR, doravirine; DTG, dolutegravir; ISL, islatravir.

Sequencing of viral RNA isolated from wells with breakthrough viruses revealed emergent mutations (Table 4). All combinations selected for resistant variants that harbored M184I/V substitutions in HIV-1 RT. Substitutions detected in integrase from breakthrough viruses from the selection studies with DTG included S39N, S153S/T, and M154M/I; each of these was detected in only one well, and none of these are clinically associated with integrase inhibitor resistance; therefore, they were not studied further. In the DOR/ISL and DOR/3TC selection studies, RAMs associated with both inhibitor classes were detected in breakthrough viruses. M184I was the most common NRTI RAM, and V106A, V108I, L234I, and Y318F substitutions were NNRTI RAMs frequently detected in breakthrough viruses. A comparison of the emergent RAMs from the DOR-containing selection studies showed, overall, a less divergent set of NNRTI RAMs in the combination with ISL than with 3TC. This was particularly evident at concentrations above 1× IC₅₀, where a higher selective pressure is applied. F227C/V substitutions were detected in viruses that emerged at the highest level, 4× IC₅₀, for the DOR/3TC combination, as the F227C/V variants are highly resistant to DOR and have a lower replicative capacity. Other RAMs detected at the various concentration levels for DOR/3TC and other combinations are summarized in Table 4. The emergent NNRTI substitutions identified in the resistance selection study were evaluated in a multiple-cycle replication assay as described in Materials and Methods. Variants containing V108I, Y318F, or V106A showed fold changes of 6.8, 9.9, and 15, respectively, versus DOR; P236L was previously evaluated against DOR with a fold change of 2.8. All the NNRTI variants were susceptible to ISL with a fold change of <2-fold.

TABLE 4.

Mutations detected from in vitro two-drug resistance selection studies

Concn (IC50 multiples)	Mutation(s) detected witha:
	DOR/ISL	DOR/3TC	DTG/3TC
0.5×	V108I, M184I	V108I	WT
1×	V106A, V108I, M184I, P236L, Y318F	V106A, V108I, M184I, H221Y, L234I, Y318F	M184I
2×	V106A, V108I, M184I	V90I, V106A, V108I, M184I, M230I, L234I, Y318F	M184I, M184V
4×	No breakthrough	D67N, V106A, M184I, F227C, F227V, M230I, L234I	M184I
8×	No breakthrough	No breakthrough	M184I

^a3TC, lamivudine; BIC, bictegravir; DOR, doravirine; DTG, dolutegravir.

To assess possible synergy between DOR and ISL, the ability of the combination to prevent viral breakthrough was compared to that of the individual components over 12 passages in MT4-GFP cells (Fig. S1). While concentrations above 24× and 8× IC₅₀ for DOR and ISL, respectively, were required to prevent viral breakthrough when the compounds were present alone, the combination synergized with no breakthrough evident at concentrations greater than 3× and 1× IC₅₀ for DOR and ISL, respectively (Fig. S1). The data for DOR/ISL were consistent with that from Fig. 1. Sequencing of RNA isolated from breakthrough viruses detected the following NNRTI RAMs in RT from multiple wells in the studies (with all having been detected and characterized in previous studies [24]): DOR only, V106A, V108I, F227L/Y, and L234I. RT M184I was the only mutation identified in the ISL only studies; it confers <5-fold potency reduction to ISL. V108I and M184I/V were the only mutations detected in RT for the DOR/ISL combination, and only M184I was seen in combination with V108I. The emergent substitutions were generated and tested in a multicycle replication assay in MT4-GFP cells (see Materials and Methods). V108I, M184I, and M184V variants showed 6.8-, 1.1-, and 1.1-fold reduction in susceptibility, respectively, to DOR and 0.8-, 6.2-, and 6.8-fold reduction in susceptibility, respectively, to ISL (compared to WT). The dual V108I/M184I variant conferred a reduction in susceptibility of 4.2- and 7.6-fold to DOR and ISL, respectively.

DISCUSSION

In the pivotal phase 3 clinical studies, DRIVE-FORWARD and DRIVE-AHEAD, 7 out of 747 participants (0.9%) from the combined DOR arms met the criteria for protocol-defined virologic failure and emerged with viruses that exhibited phenotypic resistance to DOR. The resistance development pathways were comparable to those originally identified in in vitro selection studies (24) and provide support for translatability of the in vitro results into the clinic. The emergent viruses from the clinical trials conferred substantial potency losses to DOR (Table 1). Sequencing of the variants revealed that a combination of 2 or more NNRTI RAMs is required to overcome the suppressive effect of DOR, except for one variant that harbored a single NNRTI RAM, Y188L. Interestingly, 2 nucleotide changes are required in the codon to convert tyrosine (Y) to leucine (L). Thus, overall, a minimum of 2 nucleotide changes was required for the emergence of resistance to DOR. As DOR was administered with 2 NRTIs (3TC/ABC, 3TC/TDF, or FTC/TDF) in the pivotal phase 3 clinical trials, the DOR-resistant variants were tested against several NRTIs. Among the components of the regimens used, the variants were mostly susceptible to ABC and TFV but not to 3TC or FTC. This is likely due to selection of M184V, which confers high levels of resistance to 3TC and FTC but exerts a sensitizing effect on ZDV, d4T, or TFV (25, 26). Indeed, when viruses lacking M184V were generated, the negative (greater resistance) and positive (higher susceptibility) effects exerted by the mutation on 3TC/FTC and ZDV/TFV, respectively, were abolished.

ISL is a novel nucleoside analogue that blocks RT translocation. When evaluated against the DOR-resistant variants, several showed enhanced susceptibility to ISL, which was differentiated from effects observed with marketed NRTIs such as ZDV with respect to M184V-containing variants (Tables 2 and 3). Given the sensitizing effect of M184V on ZDV, viruses bearing only NNRTI RAMs were generated by site-directed mutagenesis to assess their independent contribution to the antiviral activities of the compounds. Viruses bearing only the treatment-emergent NNRTI RAMs (in the absence of either M184V or K65R) were all susceptible to the NRTIs tested with no hypersusceptibility observed. Thus, the sensitizing effect of M184V on ZDV was confirmed. However, in contrast to ZDV, the absence of M184V did not abolish the enhanced susceptibility of the NNRTI RAM-containing viruses to ISL. The sensitizing effect of viruses lacking M184V on ISL was either attributable to the absence of the substitution or from an enhancing effect of NNRTI RAM(s). All the susceptibility-enhancing combinations of NNRTI mutations contained F227C; hence, a virus bearing that single NNRTI RAM was also generated. The F227C-containing virus, by itself, conferred hypersensitivity to ISL, demonstrating the substitution was necessary and sufficient (in the absence of other mutations) to confer the observed effect. As M184V also negatively impacts 3TC, the inhibitor was similarly evaluated against the F227C-containing virus as a control; no enhanced effect was observed. The data indicate that the F227C-enhancing effect is specific to ISL, which has a differentiated mechanism of action from 3TC and other marketed NRTIs. The structural basis for the positive impact of the F227C substitution in RT to ISL susceptibility remains to be fully characterized, but it is likely a result of an induced conformational change within the allosteric NNRTI binding pocket that is cascaded into the RT active site where ISL binds. Additional studies should provide a deeper understanding of the basis for the enhancing effect of F227C on ISL, which may include enhanced binding at the active site, faster incorporation into the viral DNA, or other kinetic or structural effects. Nonetheless, the data show a resistance profile of ISL complementary to that of DOR.

The observation that mutations elicited by DOR can enhance the antiviral activity of ISL suggested that the two inhibitors combine to exert a higher barrier to resistance. To test this hypothesis, in vitro resistance selection studies were conducted with DOR/ISL and compared to DOR/3TC (as a control given that 3TC also selects for M184V but is not similarly hypersusceptible to DOR RAMs) and DTG/3TC (a marketed two-drug combination). Based on the inhibitor concentrations at which emergence of breakthrough viruses occurred, the DOR/ISL combination had the highest barrier to the development of resistance. The pathways for the development of resistance were comparable to those previously reported for DOR and ISL in subtype B virus, except F227C was nonemergent. In previous in vitro selection studies (24), substitutions at position F227 were reported as a key mutation for development of resistance to DOR following emergence of V106A. In the pivotal DOR phase 3 clinical trials, DRIVE-FORWARD and DRIVE-AHEAD, five of the seven clinical isolates with phenotypic resistance to DOR detected by week 48 had an F227C substitution. As mentioned above, it is notable that in the 2-drug resistance selection studies with DOR/ISL, no substitution at position F227 was observed among breakthrough viruses at any concentration level (Table 4). This observation is consistent with hypersusceptibility of viruses bearing F227C to ISL (Table 3). In contrast, F227 substitutions emerged with the DOR/3TC combination, as a similar enhancing susceptibility effect is not observed with 3TC. A recent report also showed that escalating concentrations of the DOR/ISL combination resulted in a higher barrier to resistance over 8 to 24 weeks in cell culture compared to DOR alone (27). Unlike the DOR/ISL combination, there was no delay in the emergence of resistance with a rilpivirine (RPV) and ISL combination compared to RPV alone. Notably, no substitutions at RT position 227 were reported in the 4 subtypes (B, C, CRF01_AE, and CRF01_AG) investigated, a result consistent with our findings. The limited or lack of an impact of ISL on the barrier to resistance of the RPV/ISL combination (27) coupled with the nonemergence of F227 substitution(s) (this study) suggest that the higher barrier to resistance of DOR/ISL results from their complementary resistance profiles. The barrier to resistance observed for the DTG/3TC combinations was lower than that of DOR/ISL. However, no primary mutations in integrase typically associated with InSTI resistance were detected in breakthrough viruses. M184I/V, which confers substantial potency losses to 3TC, accounted for the lower barrier to resistance.

Taken together, our findings suggest that the resistance profile of DOR greatly complements that of ISL, and together they create a combination with a high barrier to resistance in vitro. The predicted high barrier to resistance is consistent with the available clinical data recently reported for the DOR/ISL combination in a phase 2b dose-ranging trial (28). The combination of 100 mg DOR, the approved dose for HIV-1 treatment, with a dose of 0.25, 0.75, or 2.25 mg ISL maintained HIV-1 suppression for 48 weeks with no participant meeting the criteria for resistance testing (defined as confirmed HIV-1 RNA of ≥200 copies/mL). The in vitro data reported in this study provide additional support for the high barrier to resistance of the DOR/ISL combination.

MATERIALS AND METHODS

The 293T cells used in these studies were from the ATCC (Manassas, VA). Fugene HD transfection reagent was bought from Promega (Madison, WI). Dulbecco’s modified Eagle medium (DMEM) was purchased from Life Technology (Grand Island, NY). Fetal bovine serum (FBS) was obtained from HyClone (Logan, UT). Normal human serum (NHS) was obtained from Biological Specialty Corporation (Colmar, PA). G418 and hygromycin were from Life Technology (Grand Island, NY). pQBILTRGagGEPNeo was purchased from MP Biomedicals (Santa Ana, CA). The Bravo liquid-handling station used was obtained from Agilent Technologies (Santa Clara, CA). The Acumen eX3 cytometer was bought from TTP Labtech (Cambridge, MA). MagMAX Express 96, the Mag-MAX 96 viral RNA isolation kit, RPMI medium, the Superscript III one-step RT-PCR system, and the TOPO TA cloning system for sequencing were obtained from Life Technologies (Grand Island, NY). Black and clear 96-well poly-d-lysine-coated or noncoated plates were purchased from BD Biosciences (San Jose, CA). The SeqScape software V3 and ABI 3100 analyzer used were from Applied Biosystems (Foster City, CA). The wild-type HIV-1 was custom-made at Advanced Biotechnologies, Inc. (Columbia, MD), and mutant HIV-1 variants were produced at MSD Research Laboratories (MRL). RPMI 1640 medium was purchased from Invitrogen (Carlsbad, CA).

Generation of mutant viruses.

Each mutant was generated by the site-directed mutagenesis method via gene synthesis and subcloned into plasmid RT112, which contained full-length R8 provirus DNA. The resulting clones were sequenced to verify the presence of the desired mutation(s) and the absence of extraneous mutations.

Cell culture.

The 293T cells were grown in DMEM containing 10% FBS. MT4-GFP cells were grown in RPMI 1640 medium containing 10% FBS and 0.4 mg/mL G418; MT4-GFP/CCR5 cells were grown in RPMI 1640 medium containing 10% FBS, 0.4 mg/mL G418, and 0.4 mg/mL hygromycin. For virus production, 293T cells were seeded at 2.5 × 10⁶ per well of a 10-cm-diameter dish. After incubation for 24 h, the cells were transfected with 18 g of provirus plasmid DNA using Fugene HD transfection reagent. The supernatant was harvested at 48 h posttransfection. Each mutant virus was evaluated for infectivity in MT4-GFP reporter cells.

Resistance selection with inhibitor combinations.

In vitro selection experiments were performed in 96-well plates using R8 (subtype B) virus with the combinations of DOR/ISL, DOR/3TC, and DTG/3TC at concentrations defined by fixed multiples over each compound’s IC₅₀, as previously determined in the VIKING assay in MT4-GFP cells with 10% NHS (description of the VIKING replication assay is provided below). Compound plates were prepared with 2 μL of each compound combination at 200× the final concentration and stored at −80°C. One 96-well plate was used for each combination; the inhibitor concentrations (indicated as fold multiples of IC₅₀) studied across the columns are indicated in Table S2 in the supplemental material. Rows A through H represent 8 independent replicates of the concentrations studied in columns 1 through 12. Hence, 16 replicate wells were studied at final concentrations of 4-fold, 2-fold, 1-fold, 0.5-fold, and 0.25-fold of each compound’s IC₅₀ value. The IC₅₀ values determined for each compound and used in the selection studies were the following: 3TC, 372 nM; DOR, 3.6 nM; DTG, 8.1 nM; ISL, 0.79 nM.

For cell culture, MT4-GFP cells were infected in bulk with R8 virus at a high multiplicity of infection (MOI) at which most cells were dead 4 days after infection in wells without compound. Seventy thousand infected cells were added per well to a room temperature compound plate for a final volume of 200 μL per well. Plates were incubated for 4 days at 37°C and 5% carbon dioxide prior to initiation of the next selection passage. The medium was exchanged at each passage every 3 to 4 days.

Analysis of RT mutation(s) in the breakthrough viruses from the resistance selection studies.

Viral RNA was extracted with the MagMAX 96 viral RNA isolation kit from culture supernatant of breakthrough virus stock from the resistance studies described above. The RT-encoding region was amplified by the one-step RT-PCR method. PCR products were genotyped by an automated population-based full-length sequencing method (covering amino acids 1 to 440 of the RT region). The primers used for PCR amplification of subtype B virus were 5′AAGCAGGAGCCGATAGACAA3′ (forward) and 5′TAATCCCGAATCCTGCAAAGCTAGA3′ (reverse). The sequencing primers were 5′-CCCTGTGGAAGCAC3′ and 5′-GGATGTGGGCGATGC3′. The amplification primers used for subtype A sequencing were 5′AGGCTATAGGTACAGTATTAGTAGGACCTAC3′ (forward) and 5′TGTTCAGCTTGATCCCTTACCTG3′ (reverse). Primers 5′GACCTACACCTGTCAACATAATT3′, 5′TGTCTTCCTCTGTCAGTAACATAC3′, 5′GGAAAGGATCACCGGCAATA3′, and 5′TTGCTCTATGCTGCCCTATTT3′ were used for sequencing analysis. The primers used in PCR amplification for subtype C sequencing were 5′GACACAGGAGCAGATGATACAG3′ (forward) and 5′AGCACTTTCCTGATTCCACTAC-3′ (reverse). Primers 5′ACCTGTCAACATAATTGGAAGAAAT3′, 5′TTCTGCCTTCCTTTGTCAGTAA3′, 5′GATGGAAAGGATCACCAGCA3′, and 5′GCTCTATGTTGCCCTATTTCTAAG3′ were used for sequence analysis. Sequencing results were reported as amino acid changes compared with WT HIV-1 92RW026 (subtype A) (GenBank accession no. AY669702.1), R8 (subtype B) (accession no. KT200357.1), and 93MW959 (subtype C) (accession no. AY669739.1) reference sequences.

HIV replication assay.

HIV-1 replication was monitored using MT4-gag-GFP clone D3 (here designated MT4-GFP). Infection of MT4-GFP cells with HIV-1 results in GFP expression approximately 24 h postinfection. Details of the assay, subsequently referred to as VIral KINetics in Green cells (VIKING) assay, have been reported (19). For infection, MT4-GFP cells were cultured in growth medium lacking G418 and infected overnight with wild-type HIV-1 (R8) or a resistant mutant at an approximate MOI of 0.01 under the growth conditions. Cells were then washed and resuspended in 10% or 100% NHS at 2 × 10⁵ cells/mL. Compound plates were prepared by dispensing compounds dissolved in dimethyl sulfoxide (DMSO) into 384-well poly-d-lysine-coated plates (0.2 μL/well) using an ECHO acoustic dispenser. Each compound was tested in a 10-point serial 3-fold dilution (typical final concentrations, 8,400 nM to 0.42 nM). Controls included no inhibitor (0.4% DMSO only) and a combination of 3 antiviral agents (EFV, indinavir, and an integrase strand transfer inhibitor, L-002254051, at final concentrations of 4 μM each). Cells were added (50 μL/well) to compound plates, and the infected cells were maintained at 37°C, 5% carbon dioxide, and 90% humidity. Infected cells were quantified at 2 time points, approximately 48 h and 72 h postinfection, by counting the number of green cells in each well using an Acumen eX3 scanner. The reproductive ratio, R, is calculated by dividing the number of infected cells at 72 h by those at 48 h postinfection. The percent inhibition caused by a test compound is calculated by the following formula: % inhibition = [1 − (R_{test compound} – R_{positive control})]/(R_DMSO only – R_{positive control}) × 100%. The dose-response curves for each test compound were plotted as percent inhibition against test concentration, and the IC₅₀ values were determined by nonlinear 4-parameter curve fitting (ActivityBase). The inhibitory potency (IC₅₀) for DOR against HIV-1 variants harboring mutations that confer resistance to NNRTIs was assessed in 100% NHS in the VIKING assay. The relative fold change in potency was calculated by dividing the IC₅₀ for each resistant variant by the IC₅₀ for the wild-type HIV-1 control.