Characterizing the Diverse Mutational Pathways Associated with R5-Tropic Maraviroc Resistance: HIV-1 That Uses the Drug-Bound CCR5 Coreceptor

Abstract

Entry inhibitors represent a potent class of antiretroviral drugs that target a host cell protein, CCR5, an HIV-1 entry coreceptor, and not viral protein. Lack of sensitivity can occur due to preexisting virus that uses the CXCR4 coreceptor, while true resistance occurs through viral adaptation to use a drug-bound CCR5 coreceptor. To understand this R5 resistance pathway, we analyzed >500 envelope protein sequences and phenotypes from viruses of 20 patients from the clinical trials MOTIVATE 1 and 2, in which treatment-experienced patients received maraviroc plus optimized background therapy. The resistant viral population was phylogenetically distinct and associated with a genetic bottleneck in each patient, consistent with de novo emergence of resistance. Recombination analysis showed that the C2-V3-C3 region tends to genotypically correspond to the recombinant's phenotype, indicating its primary importance in conferring resistance. Between patients, there was a notable lack of commonality in the specific sites conferring resistance, confirming the unusual nature of R5-tropic resistance. We used coevolutionary and positive-selection analyses to characterize the genotypic determinants of resistance and found that (i) there are complicated covariation networks, indicating frequent coevolutionary/compensatory changes in the context of protein structure; (ii) covarying sites under positive selection are enriched in resistant viruses; (iii) CD4 binding sites form part of a unique covariation network independent of the V3 loop; and (iv) the covariation network formed between the V3 loop and other regions of gp120 and gp41 intersects sites involved in glycosylation and protein secretion. These results demonstrate that while envelope sequence mutations are the key to conferring maraviroc resistance, the specific changes involved are context dependent and thus inherently unpredictable.IMPORTANCE The entry inhibitor drug maraviroc makes the cell coreceptor CCR5 unavailable for use by HIV-1 and is now used in combination antiretroviral therapy. Treatment failure with drug-resistant virus is particularly interesting because it tends to be rare, with lack of sensitivity usually associated with the presence of CXCR4-using virus (CXCR4 is the main alternative coreceptor HIV-1 uses, in addition to CD4). We analyzed envelope sequences from HIV-1, obtained from 20 patients who enrolled in maraviroc clinical trials and experienced treatment failure, without detection of CXCR4-using virus. Evolutionary analysis was employed to identify molecular changes that confer maraviroc resistance. We found that in these individuals, resistant viruses form a distinct population that evolved once and was successful as a result of drug pressure. Further evolutionary analysis placed the complex network of interdependent mutational changes into functional groups that help explain the impediments to the emergence of maraviroc-associated R5 drug resistance.

Deep Sequencing of the HIV-1 env Gene Reveals Discrete X4 Lineages and Linkage Disequilibrium between X4 and R5 Viruses in the V1/V2 and V3 Variable Regions

“…We also note that there were three env gene clones (from subjects C05 and C06) whose encoded proteins retained some infectivity in the presence of both the CCR5 and CXCR4 inhibitors. It was reported previously that CCR5-tropic but maraviroc-resistant variants can use inhibitor-bound CCR5 (44). The three env clones from this study may encode Env proteins that are examples of a naturally occurring CCR5-tropic maraviroc-resistant virus, or they may use coreceptors other than CCR5 or CXCR4.…”

HIV-1 drug resistance and resistance testing

“…HIV-1 can also develop maraviroc resistance mutations that allow HIV-1 gp120 to bind to an inhibitor-bound R5 receptor. However, reports of such resistance have been documented primarily in vitro and in only a small number of clinical viral isolates (Jiang et al, 2015; Westby et al, 2007). …”

Accurate predictions of population-level changes in sequence and structural properties of HIV-1 Env using a volatility-controlled diffusion model

“…Similar to the antigenicity features, volatility indices measured using plasma samples from Iowa City and Seattle correlated well ( p -value < 10 −6 in a Spearman correlation test, Fig 3E). Volatility indices were also measured using a third panel of plasma samples from 20 clade B–infected individuals who enrolled in the MOTIVATE trial, which examined efficacy of the CCR5 inhibitor Maraviroc [89, 90] (see alignment of these Envs in S4 Data). Only plasma samples collected prior to initiation of Maraviroc treatment were studied.…”

“…We emphasize that the approach does not take into account the likelihood of each amino acid substitution; incorporation of substitution likelihoods further improves the predictive capacity of this basic model. We also note that the model treats all amino acid residues as independent variables and does not acknowledge the well-defined networks of association that exist within the V3 loop [89, 96, 97]. Inclusion of such considerations is expected to improve performance of the model still further.…”