ABSTRACT
Vicriviroc (VCV) is a CCR5 antagonist that blocks the viral entry of CCR5-tropic (R5) virions by binding to and inducing a conformational change in the chemokine receptor. Clinical resistance to CCR5 antagonists occurs in two phases, competitive and noncompetitive stages. In this study, we analyzed two subjects, from a phase 2b VCV clinical trial, whose quasispecies contained R5 and dual-mixed virions at the earliest recorded time of virological failure (VF). Genotypic analysis of R5-tropic patient-derived envelope genes revealed significant changes in the V1/V2 coding domain and convergence toward a more homogenous sequence under VCV therapy. Additionally, a small population of baseline clones sharing similar V1/V2 and V3 domains with the predominant VF isolate was observed. These clones were denoted preresistant based on their genotype. Preresistant clones and chimeric clones containing V1/V2 regions isolated during VF displayed high 50% inhibitory concentration (IC50) values relative to those at baseline, consistent with early competitive resistance. Genotypic analysis of the dual-tropic clones also showed significant changes in the V1/V2 region, different from the resistant R5-tropic viruses. Our findings suggest that the V1/V2 domain plays a key role in the initial step of development of drug resistance.
IMPORTANCE It is believed that each CCR5 antagonist-resistant isolate will develop its own unique set of mutations, making it difficult to identify a signature mutation that can effectively predict CCR5 antagonist resistance. This may explain why we do not observe shared mutations among clinical studies. The present study examined the earliest events in the development of drug resistance with viral quasispecies that continued the use of CCR5 for entry. Genotypic and phenotypic assays demonstrated a distinct role of the variable domain V1/V2 in competitive resistance to CCR5 antagonist therapy. Thus, future studies analyzing the development of clinical resistance should focus on the relationship between the V1/V2 and V3 domains.
INTRODUCTION
Infection with HIV-1 is a multistep, dynamic process that requires three key events (reviewed in reference 1). First, the viral envelope (Env) trimer binds to CD4 on the host cell surface, resulting in a conformational change that exposes the coreceptor interaction site, which primarily consists of the third variable (V3) loop. Second, the V3 region binds to a chemokine receptor, commonly CCR5 or CXCR4. Third, the extracellular domain of the transmembrane glycoprotein gp41 undergoes a rearrangement to insert its fusion peptide domain into the target membrane. Depending on the coreceptor utilized, Env variants can be classified as CCR5 tropic (R5), CXCR4 tropic (X4), or dual tropic (R5/X4).
In recent years, several viral entry inhibitors have been developed to block CD4 binding, CCR5 binding, and gp41 rearrangements (reviewed in reference 2). Inhibitors of CCR5 binding include TAK-779, the FDA-approved drug maraviroc (MVC), and the investigational agent vicriviroc (VCV). These CCR5 antagonists bind to and induce a conformational change in the chemokine receptor, effectively blocking the entry of viruses that exclusively use CCR5. Therefore, screening for the presence of dual-mixed (DM) viruses prior to initiating CCR5 antagonist treatment is required. Tropism can be determined phenotypically through cell-based infection assays or genotypically through prediction algorithms that utilize the V3 coding domain. Current clinical guidelines recommend phenotypically screening the patient quasispecies (3).
There are many ways in which a patient may fail CCR5 antagonist therapy. Treatment failure can be broadly classified into three groups. The patient quasispecies can (i) shift to dual tropism, (ii) retain R5 tropism, or (iii) have a mixture of both. True resistance to CCR5 antagonists occurs when the virus retains its R5 phenotype but alters its interaction with the drug (4). Resistance has been documented both in vitro through the serial passaging of viruses with increasing drug concentrations (5, 6) and in vivo through clinical trials (7, 8).
The development of clinical resistance is believed to occur in two phases: competitive and noncompetitive stages (4, 6). The first step involves a competitive mechanism whereby the virus develops an increased affinity for the drug-free form of CCR5 and is characterized by an increase in 50% inhibitory concentration (IC50) values. The virus continues to replicate until it accumulates the mutations needed for the noncompetitive stage. During this phase, the virus gains the ability to use drug-bound CCR5 for viral entry and is characterized by a flattening of the dose-response curve and a decrease in maximal percent inhibition (MPI). The majority of studies (8–12) identifying escape mutations have focused on the later stages, typically weeks after virological failure (VF). Very few studies have examined early, competitive resistance (6, 13).
Genetic analysis of competitive and noncompetitive resistant HIV-1 variants has primarily focused on the V3 region, due to its key roles in viral entry and coreceptor usage. To date, no mutations that can definitively predict CCR5 antagonist resistance have been identified. This may suggest that each virus utilizes its own distinct genetic pathway to evade R5 viral entry inhibitors. More recently, some investigators have proposed that resistance may also depend on sites elsewhere in the Env glycoprotein (11, 12, 14).
Like the V3 domain, the V1/V2 region is thought to influence coreceptor usage (15). However, to our knowledge, no studies have analyzed the V1/V2 region in the context of clinical CCR5 antagonist resistance. Out of the five hypervariable regions, V1/V2 is the most diverse in sequence and length (16). The role of the V1/V2 region in HIV-1 infection is not as well defined as that of V3, but several studies have provided evidence showing the V1/V2 region’s involvement in viral entry (reviewed in reference 17) and immune evasion (18).
The present study aims to understand the retention of R5 tropism under CCR5 antagonist therapy and characterize regions in gp120 that may contribute to the development of drug resistance. We isolated individual clonal Envs from two subjects who were found to retain R5 tropism despite failure of VCV therapy. Genotypic analysis of the R5-tropic clones revealed two key findings. First, the V1/V2 domain shifted toward a relatively homogenous sequence under CCR5 antagonist treatment. Second, a small population of baseline clones was selected for by CCR5 antagonist treatment. These “preresistant” clones shared V1/V2 and V3 sequences similar to those found at VF. Drug sensitivity assays conducted on these clones showed increasing IC50 values, which is indicative of competitive resistance. Taken together, our findings suggest that the V1/V2 domain may play a key role in the early development of CCR5 antagonist resistance.
RESULTS
Panel of recombinant Env clones.Subjects were selected from a phase 2b clinical trial of VCV (AIDS Clinical Trials Group 5211 [ACTG 5211]) added to current combined antiretroviral therapy (7). Twenty-nine of the 118 (24.6%) enrolled participants showed drug resistance, as indicated by an increase in viral load. Thirteen of the 29 patients with VF exhibited dual-mixed (DM) virus at the time of treatment failure, between 2 and 28 weeks after initiation of VCV (7). Envelope amplicons were obtained from baseline (sample x.1) and treatment failure (sample x.2) time points for 7 of the 13 individuals. Next-generation sequence analysis was performed for the V3 region. Sequences encoding complete V3 domains were analyzed using the Geno2Pheno algorithm, which predicted CXCR4 coreceptor usage.
Five patients showed a complete shift from R5 to DM tropism at the time of VF (not shown). Two patients, subjects 2 and 5, retained a population of R5-tropic virions (subject designations as in reference 19). Both of these individuals exhibited VCV resistance by 8 weeks. The patient 2 Env quasispecies was predicted to have an even proportion of DM to R5 tropism: 451/887 (50.8%) V3 reads were consistent with R5 tropism, and 436/887 (49.2%) were consistent with X4-using strains (Fig. 1A and B). In contrast, patient 5 had a significantly larger proportion of predicted R5-tropic viruses: 949/1,084 (87.6%) V3 reads were consistent with R5 tropism, and 135/1,084 (12.4%) were consistent with X4-using strains (Fig. 1A and C).
Comparison of amplicon and functional libraries. The amplicon library was comprised of V3 Illumina sequencing reads extracted from patients with viremia. The functional library was comprised of V3 sequences from replication-competent clones, determined through a single-round infection assay. (A) Comparison between the predicted (amplicon library [AL]) and true (functional library [FL]) proportions of DM (pink) to R5 (blue) viruses at VF. Individual V3 Illumina reads were analyzed by the Geno2Pheno algorithm to predict CXCR4 usage. A summary of the patient 2 quasispecies tropism is shown on the left, and that for patient 5 is on the right. (B) Comparison between the predicted (AL) and true (FL) frequencies of clonal V3 sequences at baseline (sample 2.1) and VF (sample 2.2). Illumina reads spanning the entire V3 sequence were included in the AL analysis. Unique clonal sequences are represented by abundance and are color-coded across all libraries. (C) A similar analysis was conducted for patient 5 baseline (sample 5.1) and VF (sample 5.2) clones.
Recombinant HIV-1 plasmids were generated by cloning patient 2 and 5 Env amplicons into pNL4-3. We followed the cloning protocol established in our previous study to reduce the risk of PCR-induced mutations or artifacts (19). Individual clones expressing Env were isolated by single-clone analysis, and tropism was phenotypically determined through single-round infection assays. Positive coreceptor usage was defined as 2 standard deviations above the mean for negative-control wells (Fig. 2).
Panel of viral clones from patients 2 and 5. Tropism was determined through a single-round infection assay. Recombinant viral constructs were transfected into HEK 293T cells to produce replication-competent virions. Viral stocks were harvested from the supernatant and used to infect CCR5- and CXCR4-expressing U87.CD4 cell lines. Positive coreceptor usage was defined by an average luciferase readout of >2 standard deviations above the average value for negative-control wells. The results were found to be consistent with the tropism predicted by Geno2Pheno and further confirmed by repeating the infection assay. (A) Twenty-six baseline clones and 49 VF clones were isolated from patient 2. Of the 49 VF clones, 28/49 (57.1%) were dual tropic, and 21/49 (42.9%) were R5 tropic. (B) Thirty-five baseline clones and 40 VF clones were isolated from patient 5. Baseline clones were homogenous for R5 tropism. VF clones were heterogenous for R5 and dual tropisms. Of the 40 VF clones, 1/40 (2.5%) was dual tropic, and 39/40 (97.5%) were R5 tropic. No X4-tropic virus was identified at either time point.
The functional library for patient 2 comprised 26 baseline and 49 VF clones. For patient 5, these values were 35 baseline and 40 VF clones (Fig. 2). Baseline Envs from both patients were homogenous for CCR5 coreceptor usage. In contrast, VF Env sequences were heterogeneous for coreceptor use. In patient 2, 28/49 (57.1%) sequences were dual tropic, and 21/49 (42.9%) were R5 tropic (Fig. 1A and B, V3_FL for sample 2.2 obtained at VF). In patient 5, 1/40 (2.5%) sequences was dual tropic, and 39/40 (97.5%) were R5 tropic (Fig. 1A and C, V3_FL for sample 5.2 obtained at VF). No X4-tropic virus was isolated at either time point. The proportion of R5 to DM virions was compared to the expected proportion predicted from the V3 Illumina reads (Fig. 1A), and no statistically significant differences were seen (P = 0.46 and 0.10 for patients 2 and 5, respectively, by chi-squared analysis).
Genetic difference between R5-tropic baseline and VF clones.To understand the sequence signature of HIV-1 and its relationship to VCV resistance, a phylogenetic analysis was performed for the functional Env amino acid sequences for each patient (Fig. 3A and D). A wide spectrum of genetic diversity was apparent. The clones clustered into three main groups based on bootstrap analysis. In patient 2 (Fig. 3A), the top cluster contained 24/26 baseline R5 clones (red squares) and 3/21 VF R5 clones (blue squares), the middle cluster contained 18/21 VF R5 and 2/26 baseline R5 clones, and the bottom cluster contained 28/28 VF dual-tropic clones (yellow circles). In patient 5 (Fig. 3D), these numbers were 32/35 baseline R5 clones for the top cluster, 39/39 VF R5 clones and 3/35 baseline R5 clones for the middle cluster, and 1/1 VF dual-tropic clone for the bottom cluster. Both duration of VCV treatment and tropism were key distinguishing characteristics. The three clusters can be broadly classified as (i) baseline R5-tropic (red squares), (ii) VF R5-tropic (blue squares), and (iii) VF dual-tropic (yellow circles) clones.
Phylogenetic analysis of individual clones. Shown are data for hierarchical clustering analysis of Env (A and D), V3 (B and E), and V1/V2 (C and F) variants from functional libraries based on full-length amino acid sequences from patient 2 (A to C) and patient 5 (D to F). Tropism was phenotypically characterized through a single-round infection assay. Branch points with bootstrap values equal to 100% are labeled.
Some baseline R5 Env sequences were grouped with the VF R5 cluster, meaning that they were more genetically similar to clones from VF than baseline. In patients 2 and 5, we observed 2/26 (7.7%) and 3/35 (8.6%) such clones, respectively (Fig. 3A and D, red squares clustering with blue squares).
Relationship of V3 sequences to VCV sensitivity.Due to its well-established relationship to coreceptor usage, the V3 region has been studied extensively in understanding CCR5 antagonist susceptibility (20). Previous studies suggested that mutations in the V3 region are sufficient for competitive (13) and noncompetitive (8) resistance. To determine if the V3 region influences drug resistance in our samples, we performed a phylogenetic and genetic analysis of the clonal V3 amino acid sequences.
The V3 phylogenetic analysis showed no statistically significant difference between the R5-tropic baseline and VF V3 regions (Fig. 3B and E). However, it is worth noting that we observed an enrichment of minority, preexisting V3 sequences under VCV therapy (Table 1). In patient 2, the predominant R5-tropic V3 sequence at VF expanded from 3.8% to 20.4% (R5 weeks 0 and 8) (Table 1, boldface type, and Fig. 1B, orange). In patient 5, these numbers rose from 11.1% to 97.5% (R5 weeks 0 and 8) (Table 1, boldface type, and Fig. 1C, blue). These findings are consistent with previous studies which also observed an enrichment of V3 sequences under VCV treatment (8, 19).
V3 sequence analysis
Unlike the Env phylogenetic analysis, which separated sequences based on tropism and duration of treatment, the V3 phylogenetic analysis separated sequences based only on tropism. Genetic analysis revealed that the dual-tropic clones shared multiple V3 mutations with one another (Table 1), including the 11/25 rule (21). These findings are consistent with those of previous studies showing that the V3 region is a good predictor of coreceptor specificity and help confirm the validity of our genetic findings (22).
Selection pressure of V1/V2 sequences under VCV.To determine if any other region contributes to CCR5 antagonist resistance, we repeated the phylogenetic analysis for all the constant and variable regions in the Env glycoprotein (not shown). Our single-clone approach allowed us to analyze each region in relation to the full-length Env sequence and phenotype. Only the V1/V2 region produced statistically significant results (Fig. 3C and F). The V1/V2 sequences clustered into three groups: (i) baseline R5-tropic, (ii) VF R5-tropic, and (iii) VF dual-tropic clones. The V1/V2 phylogenetic analysis matched the Env phylogenetic analysis discussed above and indicates that the genetic differences observed in Env can be attributed to differences in the V1/V2 domain.
There was a statistically significant difference between the baseline and VF R5-tropic clusters (Fig. 3C and F, red versus blue squares). Alignments of the individual V1/V2 R5-tropic sequences revealed multiple, shared amino acid mutations at VF. In patient 2, most of the V1/V2 sequences at VF were highly similar to one another (Table 2). In patient 5, all the V1/V2 sequences at VF were identical (Table 2, boldface type). In both patients, the V1/V2 region shifted toward a homogenous sequence under VCV therapy, indicating that the V1/V2 region may contribute to the development of early CCR5 antagonist resistance.
V1/V2 sequence analysis
Notably, the V1/V2 phylogenetic analysis also showed a statistically significant difference between dual- and R5-tropic clusters (Fig. 3C and F, yellow squares). Alignments of dual-tropic clones from patient 2 revealed multiple common V1/V2 mutations. These mutations were different from those found in the VF R5-tropic clones, which suggests that the V1/V2 region may also play a role in coreceptor switching (Table 2). In patient 5, we isolated one dual-tropic clone with similar results. Genetic analysis of the full-length Env revealed that the dual-tropic clone(s) from both patients had significant V1/V2 and V3 mutations, which is consistent with a previous in vitro study highlighting the importance of V1/V2 mutations in coreceptor switching (22).
Selection of preresistant clones under VCV therapy.Previous studies observed the enrichment of V3 sequences under VCV therapy, leading some investigators to propose the existence of preresistant viruses (13). Our approach allows us to test this theory by genotypically and phenotypically analyzing recombinant Env isolated from clinical samples. In both patients, we identified clones that were more genetically similar to clones at VF than at baseline (Fig. 3A and D). These clones were characterized as “preresistant” based on the phylogenetic analysis described above (Fig. 3A and D, red squares clustering with blue squares). In patients 2 and 5, there were 2/26 (7.7%) and 3/35 (8.5%) preresistant clones observed, respectively.
Genetic analysis of the Env of the preresistant clones revealed that the enrichment of the V3 domain is linked with the selection of the V1/V2 domain. These clones shared V1/V2 and V3 sequences similar to the predominant R5-tropic sequence at VF (R5 week 0 sequences) (Tables 1 and 2, boldface type). Our genotypic results indicate that certain V1/V2 and V3 sequences are positively selected by CCR5 antagonists.
To test if the preresistant clones were less sensitive to CCR5 antagonists than baseline clones, we conducted VCV and MVC susceptibility assays on representative clones from both patients. VF clones were used as the positive control. Selection criteria were based on two characteristics: infectivity and genotype. Clones with high infectivity (>106 relative luciferase units [RLU]) based on the single-round infection assay and clones with the predominant V1/V2 sequences were given priority.
Dose-response curves with VCV (Fig. 4A and C) and MVC (Fig. 4B and D) are shown for representative clones for patient 2 (Fig. 4A and B) and patient 5 (Fig. 4C and D). All clones showed a sigmoidal dose-dependent curve for both VCV and MVC. The rightward shift in the dose-response curve is consistent with competitive drug resistance (8), and the differences were statistically significant (P ≤ 0.01 for week 0 versus week 8, and week 0 versus week 0, preresistant curves). We observed moderate increases in IC50 values from baseline to VF, with 4.2- and 5.4-fold increases in patients 2 and 5, respectively (Table 3). As expected, the preresistant clones were less sensitive to CCR5 antagonists and had IC50 values comparable to those for clones from VF, with 4.8- and 4.7-fold increases from the baseline in patients 2 and 5, respectively. Similar results were seen in both patients for MVC. The phenotypic results show that both patient quasispecies contained preresistant virions, which expanded under VCV therapy.
CCR5 antagonist susceptibilities of baseline, VF, and preresistant clones. Preresistant clones were defined as baseline clones that shared similar V1/V2 and V3 coding domains as VF clones. Clones were assayed in triplicates, and each assay was repeated in at least two independent experiments. Representative VF and baseline clones served as the positive and negative controls, respectively. Percent inhibition was defined as the RLU at various drug concentrations relative to the average RLU with no drug. (A and B) The dose-response curves for patient 2 recombinant Env (p02env0 and p02env8) and preresistant Env (p02env0 pre-R) are shown for VCV (n = 2) (A) and MVC (n = 2) (B). (C and D) Similar curves for patient 5 recombinant Env (p05env0 and p02env8) and preresistant Env (p05env0 pre-R) are also shown for VCV (n = 2) (C) and MVC (n = 3) (D).
IC50 values for VCV and MVC susceptibilitiesa
Introduction of resistant V1/V2 into baseline Env sequences increases resistance to VCV.To determine the role of V1/V2 in CCR5 antagonist susceptibility, we generated chimeric clones by incorporating a resistant V1/V2 region into a sensitive baseline backbone (Fig. 5). Preresistant and VF clones were used as the positive controls. The chimeric constructs were tested for functionality through a single-round infection assay before testing their sensitivity to VCV and MVC. Both constructs were found to be replication competent, with high infectivity (>106 RLU). The chimeric clones exhibited a rightward shift in the sigmoidal dose-response curve, which is consistent with competitive drug resistance, and the differences were statistically significant (P ≤ 0.001 for baseline versus p02env0χV1V2 and p05env0χV1V2 curves). The VCV and MVC susceptibilities of the chimeric clones were comparable to those of resistant VF clones (P ≥ 0.05 for p02env8 versus p02env0χV1V2 and p05env8 versus p05env0χV1V2 curves). We observed 2.5- and 6.8-fold increases in VCV IC50 values relative to the baseline for patients 2 and 5, respectively, and 4.3- and 4.1-fold increases in MVC IC50 values for patients 2 and 5, respectively.
CCR5 antagonist susceptibilities of V1/V2 chimeric clones. Chimeric clones were generated by introducing a resistant VF V1/V2 region into a sensitive baseline backbone. The reverse chimeric clone was also generated by introducing a sensitive baseline V1/V2 region into a resistant VF backbone. Clones were assayed in triplicates, and each assay was repeated in at least two independent experiments. Representative VF and baseline clones served as the positive and negative controls, respectively. (A and B) The dose-response curve for patient 2 recombinant Env (p02env0 and p02env8), chimeric Env (p02env0χV1V2), and reverse chimeric Env (p02env8χV1V2) are shown for VCV (n = 2) (A) and MVC (n = 2) (B). (C) Schematic representation of patient 2 chimeric Env. Numbering is based on the HXV2 reference. (D and E) Similar curves for patient 5 recombinant Env (p05env0 and p05env8) and chimeric Env (p05env0χV1V2) are shown for VCV (n = 2) (D) and MVC (n = 3) (E). (F) Schematic representation of patient 5 chimeric Env.
Reverse chimeric constructs were generated by incorporating the sensitive V1/V2 region into the resistant backbone. The patient 5 reverse chimeric construct was functional, with reduced infectivity (<104 RLU), which was too low for the drug assay. The patient 2 reverse chimeric construct was functional, with high infectivity (>106 RLU). We observed a leftward shift in the dose-response curve with IC50 values comparable to those of the baseline clones, with 1.1- and 1.6-fold increases for VCV and MVC, respectively. Our phenotypic data confirm that the V1/V2 region derived from the VCV-resistant virus can increase the resistance of the baseline viruses in both patients.
DISCUSSION
Here we analyzed patient samples from a phase 2b VCV clinical trial at two time points, baseline and VF. We used deep sequencing of the V3 region to guide our analysis. Five out of seven patients had significant mutations in the V3 region at VF. Two patients were notable: one showed almost 50%-50% significant and nonsignificant mutations (patient 2), and another did not show any significant mutations (patient 5). We used a genotypic algorithm to predict coreceptor usage from the V3 Illumina reads (23). Sequences containing a significant number of mutations were predicted to utilize CXCR4 for viral entry, and a lack of significant mutations was predicted for clones that utilize only CCR5. Since the purpose of this study was to understand the retention of R5 tropism under VCV treatment, we focused our efforts on two subjects: patients 2 and 5.
Previous studies suggested that the V3 region was responsible for competitive (13) and noncompetitive (8, 11) resistance to CCR5 antagonists. More recently, some investigators proposed that regions outside V3 can also influence resistance (11, 12, 14). Our study focuses on the competitive stage of resistance, because our clinical samples came from the earliest reported time at which the subjects had undergone protocol-defined VF. The V3 region of the R5-tropic clones at VF (week 8) did not share any common mutations. This is in contrast to noncompetitive resistance, where shared V3 mutations have been identified. Tsibris et al. identified a single V3 mutation, S306P, that correlated with the emergence of noncompetitive resistance 12 weeks after VF (8, 11). This led us to explore if other regions outside V3 could influence early competitive resistance.
We utilized deep sequencing and single-clone analysis to construct a comprehensive representation of the patient Env quasispecies. The Env phylogenetic analysis showed a clear genetic difference between R5-tropic clones from baseline to VF. Drug susceptibility assays confirmed that the VF clones were less sensitive than baseline clones. We observed a modest, but statistically significant, increase in IC50 values: 4.2- and 5.4-fold increases in patients 2 and 5, respectively. These values are consistent with data from in vitro studies involving competitive resistance, which showed 3-fold increases (6).
Our data suggest that the genetic differences between R5-tropic baseline and VF clones can be attributed to the V1/V2 domain. We provide several lines of evidence showing that the V1/V2 region may play a key role in competitive resistance to CCR5 antagonists. First, we observed the selection of preexisting, minority V1/V2 sequences as the viral load increased. Second, we observed a small population of baseline clones that were more genetically similar to VF clones, which we designated “preresistant.” Genetic analysis of the full-length Envs of preresistant clones showed that they shared similar V1/V2 and V3 regions with the predominant VF isolate. This suggests that the selection of the first three variable domains is linked, which is consistent with the enrichment of V3 sequences under VCV therapy as described in previous studies (8). Drug susceptibility assays confirmed that preresistant and VF clones were less sensitive to CCR5 antagonists and had comparable IC50 values (Fig. 4 and Table 3). Third, we show that incorporating a resistant V1/V2 domain into a sensitive baseline backbone can provide partial resistance to CCR5 antagonists (Fig. 5 and Table 3). Conversely, incorporating a sensitive V1/V2 region into a resistant Env backbone can fully sensitize a resistant virus (Fig. 5 and Table 3). To our knowledge, this is first paper directly linking the V1/V2 region to CCR5 antagonist resistance.
Why do we see the selection of a specific V1/V2 region in the presence of CCR5 inhibitors? The answer may lie in the mechanism for competitive resistance. During the early phase of resistance, it is believed that virions need to effectively compete for binding to CCR5 in order to continue replication. VCV may act as a selective pressure on the viral quasispecies, favoring virions that can effectively compete for binding of the drug-free chemokine receptor. While this is the first paper linking the V1/V2 region to CCR5 antagonist resistance, there have been previous studies suggesting that the first two variable domains can affect CCR5 binding affinity. One in vitro study found that a single residue in the V1 region was responsible for the ability of Env to bind to nearly undetectable levels of CCR5 (24, 25). More recently, Pastore et al. found that V1/V2 mutations in the absence of V3 mutations can increase the affinity for CCR5 (26). Future studies are needed to determine the mechanism by which the V1/V2 domain can contribute to competitive resistance.
Our findings provide new insights into viral entry inhibitor resistance that may help explain the patient outcomes of the phase 2b clinical trial (7). Out of 29 virologic failures, 13 were attributed to coreceptor switching, from R5 to DM. We obtained 7 of the 13 paired samples for further analysis. Two of the seven samples exhibited both R5 and dual tropisms, meaning that there were multiple mechanisms of resistance present. By analyzing these special cases, we can understand why one mechanism may prevail over the other. Our findings show that CCR5 antagonists can act as a selective pressure on the viral quasispecies, favoring the outgrowth of preresistant R5-tropic clones. The retention of R5 tropism may be favored over dual-tropic variants due to the high fitness cost of switching coreceptors (26). In the absence of such clones, VCV may act as a selective pressure favoring the sequential mutations needed for dual tropism. Due to our limited study size, future studies should screen more clinical samples for the presence of R5 VCV-resistant viruses to confirm if the pathway of drug evasion depends on the presence of preresistant V1/V2 sequences in the baseline quasispecies.
In the dual-tropic clones, we observed multiple, shared V1/V2 mutations in both patients (Table 2). These mutations were different from the ones observed in the R5-tropic clones, suggesting that the V1/V2 domain may also play a role in coreceptor switching. One in vitro study found that V1/V2 mutations could compensate for loss-of-fitness mutations in the V3 region during coreceptor switching (26). Future studies should analyze the remaining five patients, who shifted completely to DM virions at VF, to confirm if the V1/V2 region significantly changed. Taken together, the results of our study suggest that the V1/V2 domain not only plays a key role in treatment failure through resistance occurring with the retention of CCR5 usage but also may play a role in resistance occurring with the emergence of CXCR4 usage.
MATERIALS AND METHODS
Patient samples.HIV-1 Env amplicons were obtained from seven treatment-experienced participants in a VCV phase 2b clinical trial (7). Samples came from individuals who experienced a shift in coreceptor usage, from R5 to DM, based on the Monogram Biosciences tropism assay and had a confirmed HIV-1 RNA level of <1 log10 copies/ml below the baseline level at or after 16 weeks (7). The results from the Trofile assay are summarized in Table 4. For each patient, two time points were obtained: (i) baseline (week 0) and (ii) VF (week 8) (27).
Trofile assay results from patient samples
Single-clone analysis.The plasmid libraries were generated as previously described (19). Patient Env amplicons were cloned into a pNL4-3.Luc.ΔE backbone, which included an AfeI site introduced by site-directed mutagenesis. The vector was linearized using NotI and AfeI. PCR amplification conditions for the Env amplicon were previously described (19). The two products were ligated using NEBuilder master mix with an insert-to-vector molar ratio of 1:1 and transformed into ElectroMax Stbl4 competent cells. Colony PCR with Bullseye R-Taq DNA polymerase master mix (MidSci) was used to screen for recombinant vectors containing the 3-kb Env, using the following oligonucleotides: EnvF-ATG (5′-AGAGCAGAAGACAGTGGCAAT-3′) and Nef-EnvR (5′-CCACTTGCCACCCATCTTAT-3′).
Cell lines.HEK 293T cells were maintained in complete Dulbecco’s modified Eagle’s medium (DMEM) with 10% fetal bovine serum (FBS), 2 mM l-glutamine, 1 mM sodium pyruvate, and 1× antibacterial-antimycotic solution (containing 100 U/ml penicillin, 100 μg/ml streptomycin, and 250 ng/ml amphotericin). U87.CD4 cells were stably transfected with pBABE-CCR5-GFP or pBABE-CXCR4-GFP, constructed as previously described, and maintained in DMEM supplemented with 15% fetal bovine serum, 2 mM l-glutamine, 1 mM sodium pyruvate, 1× antibacterial-antimycotic solution, 0.2 mg/ml G418, and 1 μg/ml puromycin (27).
Viral stock preparation.Replication-competent HIV-1 was prepared by transfection of 3 μg of a viral recombinant plasmid in 1.0 × 106 HEK 293T cells using TransIT-KT1 transfection reagent (Mirus). The viral supernatant was harvested 60 h after transfection and passed through a 0.45-μm syringe filter to remove any cell debris. Titers in the stocks were determined using a luciferase-based assay. Triplicate serial dilutions of virus (2-fold; 100 to 12.5 μl) were used to infect U87.CD4.CCR5 cells, and cells were cultured for 48 h before luciferase activity was measured on a GloMax 96 microplate luminometer (Promega). Linear regression analysis was performed to obtain the RLU per unit volume of stock (11).
Phenotypic assay.Tropism and infectivity of the recombinant viruses were determined by a single-round-infection, U87.CD4.CCR5/U87.CD4.CXCR4 cell-based assay. One day prior to infection, 15,000 cells in 50 μl of 15% FBS–complete DMEM were plated in a 96-well plate. The day of infection, 50 μl of fresh DMEM containing 16 μg/ml DEAE-dextran was added to each well, and the mixture was incubated for 1 h. After 1 h, the cells were inoculated with 50 μl of viral stock and cultured for 48 h before luciferase activity was measured. The cells were lysed with 0.2% Triton X-100 (Sigma-Aldrich) in phosphate-buffered saline (PBS), and luciferase activity was determined on a GloMax 96 microplate luminometer (Promega). A positive result was defined as an RLU readout greater than 2 standard deviations above the mean for mock-infected control wells. Each recombinant virus was assayed in triplicate, and each assay was repeated in an independent experiment.
Phylogenetic and statistical analyses.The Env gene was sequenced for each functional isolate. Amino acid sequences were aligned using MUSCLE (28). Hierarchical clustering analysis was performed using SeaView (29). Phylogenetic trees were constructed by PhyML using the HIV-Wm substitution model, and the reliability of branching order was tested by bootstrap analysis using 1,000 replicates (30). Branches with bootstrap values above 95% were considered statistically significant.
Drug susceptibility assay.Susceptibilities to R5 entry inhibitors of the HIV-1 patient clones were determined by a U87.CD4.CCR5 cell-based assay (27). One day prior to infection, 15,000 cells in 50 μl of 15% FBS–complete DMEM were plated in a 96-well plate. On the day of infection, 50 μl of fresh DMEM containing 16 μg/ml DEAE-dextran and serially diluted VCV (5-fold, starting with 2 μM) was added, and the cells were incubated for 1 h. After 1 h, the cells were inoculated with the necessary amount of viral stock needed to produce 1.5 × 105 RLU on the 96-well plate microreader and cultured for 48 h before RLU were measured. The percent inhibition was calculated as the ratio of RLU in drug-containing wells to the RLU in the drug-free control wells. Each virus was assayed in triplicate, and the assay was repeated at least twice. Dose-response curves were generated using the best-fit curves of susceptibility using GraphPad Prism 7. P value calculations for dose-response curves used the extra sum-of-squares F test, with an α value of 0.01.
Generation of chimeric clones.All oligonucleotides were designed according to the In-Fusion HD protocol for PCR primer design (TaKaRa) (Table 5). For patient 2, recombinant pNL4-3.LucR containing the sensitive, baseline p02env0 Env was used as the template. Two large PCR fragments comprising the majority of the backbone were amplified using two pairs of primers. The first PCR product was ∼8,709 bp (pNL4-3 Amp forward/patient 2 C1 reverse) and contained the HIV 5′ genome from the long terminal repeat (LTR) 5′ to the C1 region. The second PCR product was ∼7,460 bp (patient 2 C2 forward/pNL4-3 Amp reverse) and contained the HIV 3′ genome from the C1 region to the 3′ LTR. The resistant V1/V2 domain was amplified using resistant p02env8 Env as the template. The third PCR product was ∼279 bp (patient 2 C1 forward/patient 2 C2 reverse). The PCR reactions were set up using Q5 Hot Start HiFi master mix (New England BioLabs) according to the manufacturer’s protocol. PCR fragments were purified from agarose gel using the Qiagen gel extraction kit. The products were ligated using an In-Fusion HD cloning kit with a molar ratio of 1:1:1 according to the manufacturer’s protocol (TaKaRa) and transformed into ElectroMax Stbl4 competent cells. We used the same approach for patient 5 with patient-specific primers. For patient 5, we used recombinant pNL4-3.LucR containing sensitive p05env0 Env as the template to amplify the two PCR products comprising the backbone, and the resistant p05env8 Env was used as the template to amplify the V1/V2 region. All chimeric proviral plasmids were confirmed by sequencing the full-length Env gene.
Primers used to create chimeric constructs
ACKNOWLEDGMENTS
This work was supported by National Institutes of Health grant R01 AI06361 to L.R.
We thank R. Gulick, D. Kurtizkes, T. Henrich, members of the ACTG A5211 protocol team, as well as the participating ACTG clinical research sites for providing the plasma samples used in this study. We thank G. Kyei and D. Rauch for critical reviews of the manuscript.
FOOTNOTES
- Received 11 January 2019.
- Accepted 11 February 2019.
- Accepted manuscript posted online 20 February 2019.
- Copyright © 2019 American Society for Microbiology.
