Dolutegravir Monotherapy of Simian Immunodeficiency Virus-Infected Macaques Selects for Several Patterns of Resistance Mutations with Variable Virological Outcomes

A growing number of anti-HIV drug combinations are effective in suppressing virus replication in HIV-infected persons. However, to reduce their cost and risk for toxicity, there is considerable interest in simplifying drug regimens. A major concern with single-drug regimens is the emergence of drug-resistant viral mutants. It has been speculated that DTG monotherapy may be a feasible option, because DTG may have a higher genetic barrier for the development of drug resistance than other commonly used antiretrovirals. To explore treatment initiation with DTG monotherapy, we started SIV-infected macaques on DTG during either acute or chronic infection. Although DTG initially reduced virus replication, continued treatment led to the emergence of a variety of viral mutations previously described to confer low-level resistance of HIV-1 to DTG, and this was associated with variable clinical outcomes. This unpredictability of mutational pathways and outcomes warns against using DTG monotherapy as initial treatment for HIV-infected people.

ABSTRACT

Drug resistance remains a major concern for human immunodeficiency virus (HIV) treatment. To date, very few resistance mutations have emerged in first-line combination therapy that includes the integrase strand transfer inhibitor (INSTI) dolutegravir (DTG). In vitro, DTG selects for several primary mutations that induce low-level DTG resistance; secondary mutations, while increasing the level of resistance, however, further impair replication fitness, which raised the idea that DTG monotherapy may be feasible. The simian immunodeficiency virus (SIV) rhesus macaque model of HIV infection can be useful to explore this concept. Nine macaques were infected with virulent SIVmac251 and started on DTG monotherapy during either acute (n = 2) or chronic infection (n = 7). Within 4 weeks of treatment, all animals demonstrated a reduction in viremia of 0.8 to 3.5 log RNA copies/ml plasma. Continued treatment led to overall sustained benefits, but the outcome after 10 to 50 weeks of treatment was highly variable and ranged from viral rebound to near pretreatment levels to sustained suppression, with viremia being 0.5 to 5 logs lower than expected based on pretreatment viremia. A variety of mutations previously described to confer low-level resistance of HIV-1 to DTG or other INSTI were detected, and these were sometimes followed by mutations believed to be compensatory. Some mutations, such as G118R, previously shown to severely impair the replication capacity in vitro, were associated with more sustained virological and immunological benefits of continued DTG therapy, while other mutations, such as E92Q and G140A/Q148K, were associated with more variable outcomes. The observed variability of the outcomes in macaques warrants avoidance of DTG monotherapy in HIV-infected people.

IMPORTANCE A growing number of anti-HIV drug combinations are effective in suppressing virus replication in HIV-infected persons. However, to reduce their cost and risk for toxicity, there is considerable interest in simplifying drug regimens. A major concern with single-drug regimens is the emergence of drug-resistant viral mutants. It has been speculated that DTG monotherapy may be a feasible option, because DTG may have a higher genetic barrier for the development of drug resistance than other commonly used antiretrovirals. To explore treatment initiation with DTG monotherapy, we started SIV-infected macaques on DTG during either acute or chronic infection. Although DTG initially reduced virus replication, continued treatment led to the emergence of a variety of viral mutations previously described to confer low-level resistance of HIV-1 to DTG, and this was associated with variable clinical outcomes. This unpredictability of mutational pathways and outcomes warns against using DTG monotherapy as initial treatment for HIV-infected people.

INTRODUCTION

Despite the increasing arsenal of potent antiretroviral drugs (ARV) and combinations to combat human immunodeficiency virus (HIV) infection, the emergence of viral mutants with reduced susceptibility (drug resistance) remains a major concern. This is especially the case for middle- and low-income countries, where factors such as late presentation, an inconsistent supply chain, poor adherence, suboptimal drug regimens, and a lack of routine viral load monitoring contribute to ongoing virus replication, which promotes the emergence and persistence of such drug-resistant mutants (1).

Although the correlation between the emergence of specific mutations and their impact on drug susceptibility and replication fitness in vitro is generally well understood, in vitro conditions do not fully capture the many complex and constantly changing events and interactions that occur in vivo during virus infection and antiretroviral treatment (ART). Thus, an important question pertains to the clinical implications of what to do when genotypic or phenotypic assays detect viral mutants with reduced in vitro drug susceptibility in a sample from an HIV-infected person receiving ART. At one end of the spectrum, drug-resistant mutants could abrogate the efficacy of an ART regimen for the patients in which they emerged, and as such variants can possibly be transmitted, they may reduce the efficacy of preexposure prophylaxis regimens or influence the outcome of first- and second-line ARV treatment regimens (2). At the other end, the detection of drug resistance mutations may be associated with sustained therapeutic benefits of drug therapy due to a variety of factors, such as a residual partially efficacious antiviral effect of the drug regimen or an impaired replication fitness of the viral mutants which cannot be fully restored by compensatory mutations. In this scenario, withdrawing the drug would be recommended only if alternative treatment strategies are feasible.

In recent years, the integrase (IN) strand transfer inhibitor (INSTI) dolutegravir (DTG) has become a key component of many combination ART regimens. It has been speculated that DTG may have a higher genetic barrier for the development of drug resistance than other compounds commonly used in therapy (3). To date, very few mutations conferring resistance to DTG have emerged in first-line combination therapy regimens that include DTG. In vitro, DTG selects for primary mutations, such as R263K and G118R, that induce low-level DTG resistance, but secondary mutations, while increasing resistance, unexpectedly further impair replication fitness; no compensatory mutations that fully restore replication fitness have so far been observed to emerge in vitro or in vivo (4). Because of this apparent high barrier to the development of DTG-resistant virus with high replicative fitness, there has been interest to investigate DTG monotherapy. Because of ethical concerns, initiating prolonged DTG monotherapy in treatment-naive people has never been formally tested in the context of a randomized trial; in real-life settings, however, cases of treatment initiation with DTG monotherapy are likely to occur. In contrast, several studies have investigated DTG monotherapy as a maintenance therapy in HIV-infected patients who have previously achieved virological suppression with combination ART regimens. Despite promising results in several initial small studies, in subsequent larger studies, including the DOMONO study, occasional virological failures associated with the detection of DTG resistance mutations have been reported, raising ethical concerns whether this strategy deserves further consideration (5–15).

Because, as explained above, it is difficult to extrapolate from in vitro observations to in vivo virulence, relevant animal models can be useful to gain additional insights that supplement information from human studies. Simian immunodeficiency virus (SIV)-infected rhesus macaques have been a useful animal model to study the emergence and clinical implications of drug-resistant mutants, as demonstrated for drugs like zidovudine, tenofovir, emtricitabine, and lamivudine (reviewed in reference 16). These nonhuman primate studies demonstrated that, depending on the drug and viral and individual host factors, the clinical implications of drug-resistant virus can range widely, from an apparent loss of clinical benefits of continued therapy to sustained benefits of continued therapy (17–21). This has led to a model of viral dynamics during drug therapy in vivo that incorporates many factors, including altered replication fitness and the virulence of viral mutants, residual drug activity, and the role of the immune system, which can more readily attack and eliminate replication-attenuated variants (reviewed in reference 16).

Previous in vitro experiments demonstrated that under increasing DTG concentrations SIVmac239 develops mutations that have also been found for HIV-1 under selection pressure from DTG or other INSTI (22), suggesting the potential relevance of SIV-infected macaques as a model to study DTG resistance in vivo. Accordingly, the current study was designed to study the effects of DTG monotherapy initiated during either acute or chronic SIV infection. We used the highly virulent uncloned SIVmac251 isolate, which consistently replicates to high levels, as the many replication events would maximize the chances to select for mutations. Our results demonstrate that DTG treatment of SIVmac251-infected macaques selects for several mutational pathways previously recognized as conferring reduced in vitro susceptibility to DTG or other INSTI. Despite the emergence of these mutations, a fraction of the animals demonstrated sustained suppression of viremia during continued DTG therapy, while other animals showed a partial to nearly complete viral rebound and gradual disease progression, suggesting that a balance between drug resistance and the associated fitness costs determines where viral replication levels and plasma viremia will equilibrate for a given mutation or pattern of mutations. This variability and unpredictability in outcomes indicate that DTG monotherapy as initiation therapy in humans is definitely contraindicated and also raises concerns about the potential use of DTG monotherapy as maintenance therapy for patients previously suppressed on combination ART.

RESULTS

Experimental design and drug treatment.

As outlined in Fig. 1, nine SIVmac251-infected juvenile rhesus macaques were started on DTG monotherapy during either acute or chronic infection.

FIG 1.

Experimental design. Two groups of SIVmac251-infected macaques were started on early (n = 2) or late (n = 7) DTG monotherapy. Therapy in group 1 was stopped after 50 weeks, while the animals of group 2 were maintained on treatment until the time of euthanasia.

Group 1 (n = 2) started on DTG (2.5 mg/kg of body weight once daily subcutaneously) at 2 weeks of infection. In an effort to increase drug selection pressure, their daily DTG dose was increased to 5 mg/kg after 16 weeks of treatment. When both animals had been on daily treatment for 24 to 25 weeks, a single peripheral blood sample was collected right before their daily dose (i.e., ∼24 h after their previous dose) for analysis of DTG levels. Plasma trough levels of DTG were 382 to 866 ng/ml plasma, which is in the range of the drug levels observed in humans 24 h after an oral DTG dose (23) and which suggested an appropriate dosage regimen of DTG. DTG dosing was stopped at 50 weeks of treatment to monitor the impact of drug withdrawal.

Late treatment initiation has been associated with larger viral reservoirs and increased rates of drug resistance compared with those seen with early treatment initiation. Accordingly, in an attempt to more strongly select for drug-resistant mutants, the group 2 animals (n = 7) were started on DTG treatment during chronic infection, at 18 weeks postinoculation. The daily subcutaneous DTG regimen, which was 10 mg/kg, was maintained until the time of euthanasia (at 10 to 23 weeks of treatment).

The data were compared to those for historical control animals infected with SIVmac251. The SIVmac251 isolate used in our studies is highly virulent, as infected animals typically have peak viremia of >7 log RNA copies/ml and a viral set point of 6 to 8 log RNA copies/ml of plasma (Fig. 2A), with about half of them requiring euthanasia prior to 6 months of infection due to clinical deterioration. In our studies, no untreated SIVmac251-infected control animal ever gained spontaneous control of viremia to <2 log RNA copies/ml (21, 24–28).

FIG 2.

DTG treatment of acute infection. Two male juvenile macaques were inoculated intravenously with SIVmac251 at time zero and started on chronic DTG therapy (yellow box) 2 weeks later. Plasma viremia (number of RNA copies per milliliter) (A) and the CD4⁺/CD8⁺ T cell ratio (B) were monitored regularly. The limit of detection of plasma viral RNA was 15 copies/ml. The dotted lines represent, for comparison, data on recent SIVmac251-infected historical control animals (juveniles and adolescents). In panel A, the main mutations observed in integrase (Tables 1 and 2) are listed after the animal numbers. Both animals were Mamu-A*01 negative but Mamu-B*01 positive. Animal 43921, which had additional testing, was Mamu-B*17 positive and Mamu-B*08 negative.

Virological and immunological outcome of DTG treatment.

The 2 animals of group 1 (early treatment) were started on DTG at week 2 of infection, when they had high peak viremia (>7 log RNA copies/ml) and a decrease in the CD4⁺/CD8⁺ T cell ratio to below 1, similar to the values for historical controls infected with SIVmac251 (Fig. 2). Within 4 weeks of starting DTG treatment, both animals had a decrease (200- to 3,000-fold) in plasma viremia. Subsequently, their respective courses diverged. Despite similar DTG plasma trough levels, one animal (animal 43549) had a partial rebound and developed a viral set point of ∼5 × 10⁶ log RNA copies/ml, which was still ∼1 log below the average for untreated animals infected with this virus (Fig. 2). The animal was initially clinically healthy, with stable CD4⁺ T cell counts and CD4⁺/CD8⁺ T cell ratios of >1 for almost a year, but then deteriorated with a gradual decrease in CD4⁺ T cell percentages and the CD4⁺/CD8⁺ T cell ratio (Fig. 2) and had to be euthanized at week 56 due to progressive anemia, a sign of late-stage SIV infection.

In contrast, the viremia of the 2nd animal (animal 43921) gradually declined to <100 RNA copies/ml with intermittent viral blips (Fig. 2A). This pattern of viremia is ∼5 logs below what was typically seen in 32 untreated historical control animals infected with SIVmac251 (Fig. 2A). When DTG was withdrawn at week 52, a small viral rebound (to a peak of 400 copies/ml 5 weeks after DTG withdrawal) was observed but was then controlled again. The animal was euthanized while it was still healthy at 66 weeks of infection, when viremia was 25 copies/ml plasma and CD4⁺/CD8⁺ T cell ratios (>1.6) and CD4⁺ T cell counts (>900/μl) in peripheral blood were all within the normal range.

The 7 animals of group 2 (late treatment) were started on DTG at 18 weeks of infection. Within 4 weeks after the start of DTG monotherapy, there was, on average, a 220-fold reduction of plasma viremia (i.e., the lowest viremia relative to that at the start of treatment; range, 6- to 2,500-fold; P = 0.001, two-tailed paired t test) (Fig. 3A). Similar to the 2 animals of group 1, continued treatment led to variable responses, as 3 out of the 7 animals (Fig. 3B to D) demonstrated a partial to substantial rebound of viremia (a <2-log reduction relative to that before DTG), while the remaining 4 animals demonstrated more sustained suppression (a >2-log reduction relative to that before DTG) (Fig. 3E to H). No pretherapy marker in peripheral blood (plasma viremia, CD4⁺ and CD8⁺ T cell counts) nor their major histocompatibility complex (MHC) class I genotype (Mamu-A*01 and -B*01 alleles) was predictive of the posttherapy virological outcome. Overall, for all 7 of these animals, the viral set point during prolonged daily DTG monotherapy (average viral load from week 26 to euthanasia) was 1.5 logs lower than the viral pretherapy viral set point (defined as the average viral load from weeks 10 to 18 of infection) (P = 0.007, two-tailed paired t test; Fig. 3A). A mixed-effect regression analysis also indicated a significant decrease of viremia due to DTG treatment (P = 6.293e−08). The changes in viral load were also reflected in an improvement in the average CD4⁺ T cell values and CD4⁺/CD8⁺ T cell ratio (Fig. 3A); however, this increase was not significant in a mixed-effect regression (P = 0.381). Of these 7 animals, 1 animal that had the highest viral rebound (Fig. 3B, animal 44837) had to be euthanized at 28 weeks of infection due to recurring rectal prolapse, not directly related to SIV infection. The other animals were monitored for 39 to 41 weeks of infection and then, prior to meeting the clinical criteria of simian AIDS, were euthanized.

FIG 3.

DTG monotherapy of chronic SIVmac251 infection. Seven SIVmac251-infected juvenile macaques with a high viral set point (>6 log RNA copies/ml plasma) were started on DTG monotherapy (yellow). Viral RNA levels in plasma and CD4⁺/CD8⁺ T cell ratios were monitored regularly. (A) Means for both markers; (B to H) individual values for all 7 animals. For each individual animal, the earliest detection of integrase mutations is indicated (arrow); the main mutations observed at the end of the study are summarized on the top right of each graph (Tables 1 and 2). The sex and presence of the Mamu-A*01 and Mamu-B*01 alleles are indicated after each animal number. F, female; M, male; A01+, Mamu-A*01 positive; A01−, Mamu-A*01 negative; B01+, Mamu-B*01 positive; B01−, Mamu-B*01 negative.

Sequence analysis for detection of viral integrase mutations.

The integrase region of the pol gene from plasma, axillary lymph nodes (AXLN), or jejunum (JEJ) RNA samples (collected at the time of euthanasia, after 10 to 50 weeks of DTG treatment) was genotyped to identify potential integrase resistance mutations. This analysis showed that after 4 weeks of treatment, 4 animals already had a known drug resistance mutation; the timing of detection coincided with early viral rebound (Fig. 3). After 10 to 13 weeks of treatment, RNA samples from all animals contained at least one drug resistance substitution (Table 1). Insertions were also commonly found. Tissue RNA samples were found to harbor substitutions and insertions, which typically corresponded to the circulating viruses (Table 2). In some cases, there was apparent genetic evolution in the viral RNA between two time points, including in virologically suppressed animals. For example, viral RNA analysis of animal 44853 showed at week 13 the development of viruses with an E92G pathway that were subsequently replaced by viruses bearing the Q148K substitution at week 23. Week 23 analysis of the AXLN and JEJ tissues revealed that the E92G/E mutation was present as a mixture, whereas the Q148K mutation was present only in the JEJ. Overall, among these 9 animals, the most common resistance substitutions to emerge were E92Q, Q148K, and G118R, with the last one being found independently in 3 of the 9 animals.

TABLE 1.

Integrase substitutions acquired in plasma of DTG-treated macaques

Treatment group and macaque identifier	Baseline polymorphisms (at treatment initiation)	Substitution(s) at the following wk after start of DTG treatmenta:
		4	10	12	18–23	54–55
Early treatment group
43549	V73I	None		V54I, E92Q, G106S, R231K		Q91R, G106S, A122T, V141I, V172L, insertion after R234 (SREGK), E255D
43921	V73I	None		G118R, A119T		A54V, G106S, G118R, A122T, D232N
Late treatment group
44837	G106S, I110V	Q148K	G140A, V141I, Q148K		NA
44853	G106S, I110V	None	I31L, E92G, D232N		I31L, V141I, Q148K
44854	V73I, V88A	G118R	G118R, A122T, R263K		G118R, A122T, R263K
44869	G106S, I110V	Insertion after position 228 (EGRGIK)	Insertion after position 228 (EGRGIK)		Q91R, E92Q, insertion after 235 (GIKGR), L251I
45092	G106S, I110V	E92Q	Q91R, D163N, insertion after Y226 (YREGR), insertion after R228 (EGRYK)		Q91R, D163N, insertion after R228 (EGRYK)
45250	G106S, I110V	None	V141I, Q148K, V151M		G140A, V141I, Q148K
45274	G106S, I110V	G118R	G118R, A122T, R263K		G118R, A119T, A122T, V172L, R263K

^aBold indicates mutations at positions previously associated with reduced susceptibility to DTG. NA, not available (the animal was euthanized earlier).

TABLE 2.

Integrase substitutions acquired in tissue viral RNA of DTG-treated macaques

Group and macaque identifier	Time (wk)a	IN substitutions acquired in viral RNA of different tissue compartmentsb
		Axillary lymph node	Jejunum
Acute infection group
43549	54	Q91R, A122T, V141I, V172L, insertion after R234 (SREGK), E255D	Q91R, A122T, V141I, V172L, insertion after R234 (SREGK), E255D
43921	61	G118R, A122T, R263K	G118R, A122T, D232N
Chronic infection group
44837	10	G140A, Q148K	G140A, Q148K
44853	23	E92G/E, V141I, Q148K	E92G/E
44854	22	V88A, G118R, A122T, R263K	V88A, G118R, A122T, R263K
44869	22	Q91R, E92Q, insertion after 235 (GIKGR), L251I	Q91R, E92Q, insertion after 235 (GIKGR), L251I
45092	21	Q91R, D163N, insertion after R228 (EGRYK)	Q91R, D163N, insertion after R228 (EGRYK)
45250	23	G140A, V141I, Q148K	E92G/E, G140A, V141I, Q148K
45274	18	E92Q/E, G118R, A119T, A122T, V172L, R263K	E92Q/E, G118R, A119T, A122T, V172L, R263K

^aTime (in weeks) after the start of DTG treatment.

^bBold indicates mutations previously associated with reduced susceptibility to DTG.

We have reported previously on the effects of E92Q, Q148K, G118R, and R263K on the antiretroviral drug susceptibility of SIV, as these mutations confer low-level (2- to 12-fold) resistance to DTG in vitro (22, 29). Here, to explore the importance of viral fitness on treatment failure versus viral suppression, we documented the replicative capacities of wild-type (WT), E92Q mutant, and G118R mutant viruses in monkey peripheral blood mononuclear cells (PBMCs) over 12 days (Fig. 4A). This experiment showed that the G118R mutant, but not the E92Q mutant, was severely impaired in its ability to replicate in culture. In addition, we report the detection of a new G118R plus R263K combination of substitutions that raised the possibility that these substitutions may have compensatory effects on each other. Infectivity assays (Fig. 4B to E) did not confirm this hypothesis since the infectiousness of the G118R/R263K SIV mutant was lower than that of the single mutants.

FIG 4.

Effects of integrase mutations on SIV replicative capacity in PBMCs sand infectivity in TZM-bl cells. (A) Macaque PBMC cultures were infected with equivalent amounts (normalized on RT activity) of wild-type SIVmac239 or SIVmac239 engineered to have the indicated integrase mutations. Replicative capacity was measured over time using RT activity quantification. (B to D) TZM-bl reporter cells were infected with wild-type SIVmac239 or SIVmac239 engineered to have the indicated integrase mutations, and infectivity relative to that of the wild-type virus was measured. (E) Calculated half-effective infectious concentrations. Higher EC₅₀ values indicate lower infectiousness. RLU, relative light units; CI, confidence interval.

DISCUSSION

The current study underscores the relevance of nonhuman primate models to study the emergence and clinical implications of drug-resistant viral mutations in vivo. For antiretroviral drugs to which HIV-1 and HIV-2/SIV share similar susceptibility, the SIVmac251 infection model is very relevant. The high replication levels of SIVmac251 make it more likely that rigorous drug selection pressure will select for the emergence of viral mutants with reduced drug susceptibility. As primary drug resistance mutations generally reduce viral replication fitness, continued treatment pushes the virus into mutational pathways with secondary and compensatory mutations that reflect attempts to restore viral replication fitness in the harsh in vivo environment where any replicative fitness impairment renders the virus extra vulnerable to the synergistic attacks by antiviral drugs and the immune system (reviewed in reference 16).

In the current study, the initial virological response to DTG was quite variable. This variability was not unexpected, considering that SIVmac251 is very virulent and that at the onset of dolutegravir treatment the animals likely had various degrees of immunosuppression. Our previous studies demonstrated that the strength of antiviral immune responses plays a significant role in determining the virological response to antiviral drug treatment (reviewed in reference 16). What was, however, unexpected was the variability in mutational pathways during prolonged DTG treatment. In previous studies with reverse transcriptase (RT) inhibitors in SIV-infected macaques, the emergence of primary drug resistance mutations was generally very consistent (e.g., tenofovir selects for the mutation K65R; lamivudine and emtricitabine select for the mutation M184V) (25, 30). In the current study with DTG, although all 9 animals were infected with the same virus stock and within each study group DTG treatment was initiated at the same time, there was a surprising variability in the virological outcomes and mutational pathways of the virus. With the caveat of the limited animal numbers, no pretherapy markers in peripheral blood (such as viremia, viral sequence, or general immune status) were predictive of the mutational pathway that a particular animal would enter, suggesting a stochastic process and/or unidentified viral or host factors that we did not study here (e.g., the extent of viral reservoirs in lymphoid tissues prior to therapy, the extent of viral reservoirs in sanctuary sites with poor drug penetration, and the magnitude and breadth of the antiviral immune responses). The virological outcomes following the emergence of these resistance mutations also covered a broad spectrum, from sustained suppression (5 animals) to the partial and nearly complete rebound of viremia (4 animals).

While the limited number of animals in these studies precludes definite conclusions, several interesting associations between mutations and virological outcomes can be made. Interestingly, 3 of the 5 animals that demonstrated sustained viral suppression (a >2-log reduction of viremia relative to pre-DTG levels) developed viral mutants bearing the G118R plus R263K combination of substitutions; this combination was not seen in any of the 4 animals that had on-treatment viral rebound (Fig. 2 and 3). G118R and R263K were both identified during preclinical in vitro selection experiments with HIV under DTG pressure and later found in patients failing DTG-based ART (6, 31–33). Notably, both G118R and R263K were found to emerge from HIV sequences in the context of dolutegravir monotherapy (6, 13). We previously showed that introducing G118R in SIVmac239 integrase caused an 80% decrease in enzymatic efficiency (22, 34). We also confirmed here that the SIV replicative capacity was negatively impaired by introducing the G118R substitution, whereas E92Q was mostly innocuous in this regard (Fig. 4). Infectivity assays (Fig. 4) support the hypothesis that the G118R-R263K combination of resistance substitutions may protect against uncontrolled viral replication for reasons of decreased replicative capacity.

The G118R mutation was also detected in animal 43921, 1 of the 2 macaques that were started on DTG monotherapy during acute infection. After continuous treatment, this animal gradually achieved remarkable virological suppression (Fig. 2). This animal did not have the MHC class I allele Mamu-A*01 or B*08 but had B*01 and B*17. These alleles are sometimes associated with an increased frequency of spontaneous control of viremia (A*01, B*08, and B*17) or of higher viremia (B*01) of certain SIVs, but not for the SIVmac251 isolate used in this study (20, 21, 28, 35–37). When after 50 weeks of treatment DTG was interrupted, a transient small virological rebound in this animal was followed by renewed control of virus replication for at least 11 weeks, when the animal, while still healthy, was euthanized, having reached an experimental endpoint. During this drug-free period, the G118R virus did not revert back to the wild-type sequence. Although we were not able to evaluate any longer the durability of the sustained suppression, the current observation is very unusual in the SIVmac251 model; SIVmac251-infected macaques that are started on ART at peak viremia or during the chronic stage rarely achieve sustained control of wild-type virus replication after ARV withdrawal (16, 38–40). We speculate that this sustained virological suppression in this animal after DTG withdrawal was multifactorial, namely, a result of a combination of immunological control (aided by the early start of drug treatment) and the impaired replication fitness of the G118R viral mutants.

Another mutational pathway that was detected in 3 animals was G140A plus Q148K (2 animals) or Q148K (without G140A; 1 animal). Genotyping results show that the development of the G140A plus Q148K combination of substitutions was associated in one animal with virological control and in another with early virological failure (Fig. 3). This finding suggests that the development of this specific resistance pathway is not sufficient by itself to predict treatment failure. As described earlier (reviewed in reference 16), other host and viral factors may contribute to these different outcomes, including immunological factors (such as the strength of the antiviral immune responses) and also the reservoir size, given that this parameter was recently reported to correlate with virological failure in people using dolutegravir monotherapy as maintenance (13). These G140A and Q148K mutations were not unexpected, considering that HIV-1 mutants with changes at integrase positions 140 and 148 were recently described to emerge in an HIV-infected humanized mouse that was treated with DTG and led to a partial viral rebound (41). Codon 92 was another position at which mutations (E92Q, E92G) associated with variable outcomes were detected in 3 animals; these codon 92 mutations have previously been associated with resistance to another INSTI, elvitegravir (42–44).

Surprisingly, insertions within the integrase-coding sequences were identified in RNA samples from 3 animals; in each of these 3 animals, these insertions were associated with virological failure. These insertions varied in location (after positions 226, 228, and 234; Table 1) and length (5 to 6 amino acids). In all cases, at least one G and two positively charged (R or K) amino acids were found to be inserted, with the latter often being in the form of an (R/K)(X)_2-3(R/K) motif. A previous independent study reported the development of a 5-amino-acid insertion at position 232 in SIVmac251 integrase in macaques treated with suboptimal doses of cabotegravir (45). Altogether, these results are indicative of an evolutionary convergence toward resistance against integrase inhibitors through the development of such insertions. Given the importance of the C-terminal region of HIV integrase for interaction with the viral DNA, it is possible that the above-mentioned insertions, by altering this activity, render SIVmac251 partially resistant to inhibition by DTG.

In addition to affecting the response to DTG treatment, albeit to a variable degree, it is important to acknowledge that many of the mutations described here also confer cross-resistance to other commonly used INSTI (42, 43). Thus, the possible emergence of these mutations during DTG monotherapy would also impact future treatment options with other INSTI.

In conclusion, the emergence of drug resistance mutations, the unpredictability of the mutational pathways, and the associated clinical and virological outcomes observed in the current animal study strongly suggest that DTG monotherapy is not indicated as initial therapy for HIV-infected patients. As recent reports have also demonstrated the occasional breakthrough of DTG-resistant virus when people on a suppressive ARV combination regimen were switched to DTG monotherapy, our data cast further doubt on the feasibility and ethicality of this approach. Instead, the data suggest that DTG and, most likely, also other INSTI provide maximal benefits only when used as part of combination regimens, which have a higher barrier to the emergence of resistance. Consistent with observations in humans, treatment of SIVmac251-infected macaques with DTG as part of potent combination regimens has so far resulted in sustained viral suppression and the animals have not shown the emergence of DTG-resistant SIV (38, 46).

The current study also emphasizes the usefulness of the robust SIVmac251-infected macaque model to identify atypical changes in integrase in response to suboptimal IN-based treatment. This model is also amenable to future studies investigating the impact of regimens consisting of 2 or more drugs.

MATERIALS AND METHODS

Animals.

All rhesus macaques (Macaca mulatta) in the study were from the SIV-, HIV-2-, simian retrovirus type D-, and simian T cell lymphotropic virus type 1-free breeding colony of the California National Primate Research Center (CNPRC). The animals were healthy juveniles. Group 1 and group 2 animals were 24 to 28 months and 9 to 11 months of age, respectively, at the time of SIV infection. They were housed in accordance with American Association for Assessment and Accreditation of Laboratory Animal Care standards. We strictly adhered to the Guide for the Care and Use of Laboratory Animals, prepared by the Committee on Care and Use of Laboratory Animals of the Institute of Laboratory Animal Resources, National Research Council (47). The study was approved by the Institutional Animal Care and Use Committee of the University of California, Davis.

Genetic assessment of MHC class I alleles.

DNA extracted from lymphoid cells was used to screen for the presence of the major histocompatibility complex (MHC) class I alleles Mamu-A*01 and Mamu-B*01, using a PCR-based technique (48, 49). One animal of group 1 (animal 43921) was tested for additional alleles (B*08 and B*17) as described previously (50).

SIV inoculations.

All animals were inoculated with highly virulent uncloned SIVmac251. The SIVmac251 stock (reference number 8/12), which was propagated on rhesus PBMCs, contains approximately 5 × 10⁴ 50% tissue culture infective doses (TCID₅₀) per ml. The 2 animals of group 1 (acute infection study) were inoculated with 10⁴ TCID₅₀ by the intravenous route. The 7 animals of group 2 (chronic infection study) were inoculated with 10² TCID₅₀ of this stock by the intravenous route and were, before being enrolled in this DTG study, part of another study in which, reflecting the virulence of the SIVmac251 stock and inoculation dose, 5 of 12 infected animals had already met the clinical criteria for euthanasia by 18 weeks postinfection.

Dolutegravir treatment.

Dolutegravir (DTG) powder (purchased from Cedarlane and WuXi Apptec) was dissolved in polyethylene glycol 400, mixed with an equal volume of phosphate-buffered saline, and then filter sterilized to obtain a final concentration of 5 to 10 mg/ml. It was stored frozen at −20°C and thawed to make up syringes for 1 week at a time. Once daily, DTG was administered subcutaneously in the back of the animal at a regimen of 2.5 to 10 mg/kg, as a previous study had demonstrated that an injectable triple-drug regimen containing DTG at 2.5 mg/kg gave plasma drug levels similar to those observed in humans receiving a therapeutic dose (23, 46).

Collection and processing of blood and tissue specimens.

When necessary, animals were immobilized with ketamine HCl (Parke-Davis, Morris Plains, NJ) at 10 mg/kg, injected intramuscularly. Blood samples were collected regularly for monitoring viral and immunological parameters as described previously (21). Complete blood counts (CBC) were performed on EDTA-anticoagulated blood samples. Samples were analyzed using a Pentra 60C+ analyzer (ABX Diagnostics); differential cell counts were determined manually. Tissue specimens collected at euthanasia were snap frozen for sequence analysis.

Measurement of viral RNA.

SIV RNA levels in plasma samples were tested by a real-time reverse transcription-PCR assay for SIV gag, as described in detail previously (51); this assay has a cutoff value of 15 copies/ml for a 0.5-ml plasma sample. The viral set point was defined as the average viral load over a certain time period; the average viral load was calculated by dividing the area under the curve of the log-transformed viral RNA level-versus-time curve by the time window. For animals that started DTG therapy at 18 weeks of infection, the pretherapy viral set point was defined as the average viral load from weeks 10 to 18 of infection; the posttherapy viral set point was defined as the average viral load from week 26 to the time of euthanasia.

Viral integrase sequence analysis.

The integrase region of the pol gene from plasma, axillary lymph nodes (AXLN), or jejunum (JEJ) RNA samples was sequenced using the single-genome-amplification approach. Here the viral nucleic acid template was diluted so that most PCR mixtures did not contain a viral template and the majority of wells that contained a template originated only from a single copy. For plasma samples, viral RNA was extracted using a QIAamp viral RNA minikit (Qiagen). For tissue samples, viral RNA was extracted using a phenol extraction protocol. Tissue pieces were homogenized in 1 ml of TRI Reagent (Molecular Research Center) in a 2-ml extraction tube containing Lysing Matrix D (MP Biomedicals) using a Precellys 24 homogenizer (Bertin Instruments). RNA was immediately subjected to cDNA synthesis using SuperScript III reverse transcriptase according to the manufacturer’s recommendations (Invitrogen). In brief, a cDNA reaction mixture consisting of 1× RT buffer, 0.5 mM each deoxynucleoside triphosphate, 5 mM dithiothreitol, 2 U/ml RNase inhibitor (RNaseOUT), 10 U/ml of SuperScript III reverse transcriptase, and 0.25 mM antisense primer SIVEnvR1 (5′-TGT AAT AAA TCC CTT CCA GTC CCC CC-3′) was incubated at 50°C for 60 min and 55°C for 60 min and then heat inactivated at 70°C for 15 min, followed by treatment with 1 U of RNase H at 37°C for 20 min. PCR amplification was performed on cDNA in a 1× PCR buffer containing 2 mM MgCl₂, 0.2 mM each deoxynucleoside triphosphate, 0.2 μM each primer, and 0.025 U/μl Platinum Taq polymerase (Invitrogen). A first-round PCR was performed with sense primer SIVmacPolF1 (5′-GGC AAT GCA GAG CCC CAA GAA GAC AGGG-3′) and antisense primer SIVIntR1 (5′-AAG CAA GGG AAA TAA GTG CTA TGC AGT AA-3′) under the following conditions: 1 cycle of 94°C for 2 min and 35 cycles at 94°C for 15 s, 55°C for 30 s, and 72°C for 4 min, followed by a final extension of 72°C for 10 min. Next, 1 μl from the first-round PCR product was added to a second-round PCR mixture that included the sense primer SIVmacPolF2 (5′- GGG ATG CTG GAA ATG TGG AAA AAT GGA CC-3′) and antisense primer SIVIntR3 (5′-CAC CTC TCT AGC CTC TCC GGT ATCC-3′). The second-round PCR was performed under the same conditions used for the first-round PCR, but with a total of 45 cycles. Correctly sized amplicons were identified by agarose gel electrophoresis and directly sequenced with the second-round PCR primers and SIV-specific primers using the BigDye Terminator technology. To confirm PCR amplification from a single template, chromatograms were manually examined for multiple peaks, indicative of the presence of amplicons resulting from PCR-generated recombination events, Taq polymerase errors, or multiple variant templates.

Determination of dolutegravir concentrations in plasma.

For sample preparation, 150 μl of EDTA-treated monkey plasma was mixed with 300 μl of methanol, and the mixture was kept on ice for 10 min and then centrifuged at 4°C for 10 min at 13,000 rpm (maximum speed) in a microcentrifuge. Two-hundred-microliter aliquots of the supernatants were then injected onto a high-performance liquid chromatograph (HPLC). The plasma samples included samples from both naive and DTG-treated macaques to determine which peaks in the complex mixture derived from DTG. Reverse-phase HPLC separation was done on a Jupiter C₁₈ 4.6- by 150-mm column (catalogue number 00F-4053-E0; Phenomenex) at a flow rate of 1.0 ml/min using a Dionex Ultimate HPLC system equipped with an Ultimate 3000 pump, an Ultimate 3000 column compartment, an Ultimate 3000 photodiode array detector, and a Foxy 200 fraction collector. Buffer A was 0.1% trifluoroacetic acid (TFA) in water. The column was equilibrated with 2% buffer B (0.1% TFA in acetonitrile). The gradient of buffer B was 2% buffer B for 10 min, 2% to 60% buffer B for 60 min, and 60% isocratic buffer B for 10 min. Peaks were detected by measurement of the UV absorption at 206, 260, and 280 nm.

Various amounts DTG were then added to the plasma of untreated monkeys, and the DTG peaks were measured in order to determine the yield of DTG extraction. DTG peaks could be quantitated based on integrated peak areas of the experimentally determined extinction coefficient for DTG. DTG concentrations were calculated based on the measured DTG peaks and normalization for extracted plasma volumes.

Phenotyping of lymphocyte populations.

Multiparameter flow cytometric analysis was performed to characterize lymphocyte populations in PBMC and tissue cell suspensions. Cell suspensions were stained with peridinin chlorophyll protein (PerCP)-conjugated anti-human CD8 (clone SK1; Becton, Dickinson Immunocytometry Inc., San Jose, CA), fluorescein isothiocyanate-conjugated anti-human CD3 (clone SP34; Pharmingen), phycoerythrin-conjugated anti-human CD4 (clone M-T477; Pharmingen), and allophycocyanin-conjugated anti-human CD20 (clone L27; Becton, Dickinson) for 4-color flow cytometry and analyzed on a FACSCalibur flow cytometer, as described previously (21). CD4⁺ T lymphocytes, CD8⁺ T lymphocytes, NK cells, and B cells were defined as CD4⁺ CD3⁺, CD8⁺ CD3⁺, CD8⁺ CD3⁻, and CD20⁺ CD3⁻ lymphocytes, respectively.

Criteria for euthanasia and necropsy.

Euthanasia of animals with simian AIDS was determined by previously described established criteria, indicative of a severe life-threatening situation (28). A complete necropsy with collection of formalin-fixed tissues was performed.

Replicative capacity and infectiousness of integrase mutant viruses.

Whole blood (obtained from Bio-Resources, Inc., Vaudreuil-Dorion, QC, Canada) was collected from uninfected cynomolgus macaques and delivered in BD Vacutainer heparin tubes (Becton, Dickinson). Peripheral blood mononuclear cells (PBMCs) were isolated from the blood by Ficoll-Hypaque (GE Healthcare) gradient centrifugation. Stimulation of PBMCs was facilitated with RPMI 1640 medium (Gibco/Invitrogen) supplemented with 10% heat-inactivated fetal bovine serum (FBS; Invitrogen), 2 mM l-glutamine, 50 U of penicillin/ml, 50 μg of streptomycin/ml, 500 U interleukin-2 (IL-2; Roche), and 10 μg of phytohemagglutinin (Invitrogen)/ml and incubation at 37°C under 5% CO₂. Prior to infection, PBMC cultures were stimulated for 48 to 72 h. Full-length infectious proviral DNA of SIVmac239 SpX was obtained from the National Institutes of Health (NIH) AIDS Research and Reference Reagent Program (catalogue number 12249). Genetically homogeneous viruses were produced by transfecting 12.5 μg wild-type (WT) or mutated SIVmac239 proviral DNA into 293T cells using the Lipofectamine 2000 reagent (Invitrogen) as recommended by the manufacturer. Fresh medium was added at 4 h posttransfection. After 48 h, culture supernatants were harvested, centrifuged, passed through a 0.45-μm-pore-size filter to remove cellular debris, treated with Benzonase (EMD) to degrade the transfected plasmids, and stored at −80°C for future use. Reverse transcriptase (RT) activity quantification was performed using a previously described in vitro HIV RT assay (11) and used to normalize viral infections of PBMCs to 30,000 RT units of either WT or mutant viruses. The same RT quantification assay was used to measure the SIV replicative capacity over time (RT). Infectivity assays were performed as previously described (22). Briefly, TZM-bl reporter cells were infected with serial dilutions of wild-type or mutated SIVmac239, and luciferase production was used to measure infectivity. Infectivity curves were plotted against RT activity measurements and used to calculate half-effective infectious concentrations (EC₅₀) using Prism (v5) software (GraphPad Software, Inc.).

Statistical analysis.

Statistical analyses were performed with Prism (v7) software (GraphPad Software, Inc.) for Mac and in the R programming environment. A value of P of <0.05 was considered statistically significant.

To assess the possible relationship of the treatment on the viral load and the CD4⁺/CD8⁺ T cell ratio, we performed linear mixed-effect regression using the lmer command in the lme4 package. The significance of random effects was assessed on the basis of likelihood ratio test between nested models.

This article is dedicated to Mark Wainberg (1945–2017).