Contribution of Gag and Protease to HIV-1 Phenotypic Drug Resistance in Pediatric Patients Failing Protease Inhibitor-Based Therapy

Abstract

Protease inhibitors (PIs) are used as a first-line regimen in HIV-1-infected children. Here we investigated the phenotypic consequences of amino acid changes in Gag and protease on lopinavir (LPV) and ritonavir (RTV) susceptibility among pediatric patients failing PI therapy. The Gag-protease from isolates from 20 HIV-1 subtype C-infected pediatric patients failing an LPV and/or RTV-based regimen was phenotyped using a nonreplicativein vitroassay. Changes in sensitivity to LPV and RTV relative to that of the matched baseline (pretherapy) sample were calculated. Gag and protease amino acid substitutions associated with PI failure were created in a reference clone by site-directed mutagenesis and assessed. Predicted phenotypes were determined using the Stanford drug resistance algorithm. Phenotypic resistance or reduced susceptibility to RTV and/or LPV was observed in isolates from 10 (50%) patients, all of whom had been treated with RTV. In most cases, this was associated with protease resistance mutations, but substitutions at Gag cleavage and noncleavage sites were also detected. Gag amino acid substitutions were also found in isolates from three patients with reduced drug susceptibilities who had wild-type protease. Site-directed mutagenesis confirmed that some amino acid changes in Gag contributed to PI resistance but only in the presence of major protease resistance-associated substitutions. The isolates from all patients who received LPV exclusively were phenotypically susceptible. Baseline isolates from the 20 patients showed a large (47-fold) range in the 50% effective concentration of LPV, which accounted for most of the discordance seen between the experimentally determined and the predicted phenotypes. Overall, the inclusion of thegaggene and the use of matched baseline samples provided a more comprehensive assessment of the effect of PI-induced amino acid changes on PI resistance. The lack of phenotypic resistance to LPV supports the continued use of this drug in pediatric patients.

FIG 1

Genotypic and phenotypic resistance of pediatric patient samples to LPV and RTV at treatment failure. The treatment histories of the patients are shown, where RTV + LPV/r indicates the receipt of both RTV as a single PI and LPV/r at some point, RTV indicates the receipt of RTV as a single PI only, and LPV/r refers to the receipt of LPV/r only. The Stanford HIV drug resistance database was used to predict the phenotype. Asterisks, the six patients for whom discordance between the predicted and the actual phenotype was found. Levels of drug resistance were classified as follows: sensitive (S; no shading), reduced susceptibility (RS; light gray shading), and resistant (R; dark gray shading). The assay cutoffs used are as follows: for sensitive, a <3.6-FC in the EC₅₀ for LPV and a <3.8-FC in the EC₅₀ for RTV; for reduced susceptibility, a 3.6- to 55-FC in the EC₅₀ for LPV and a >3.8-FC in the EC₅₀ for RTV; and for resistant, a >55-FC in the EC₅₀ for LPV. Protease amino acid changes in bold represent major PI mutations. Gag amino acid changes in italics represent those at non-CSs that were previously shown to be associated with PI exposure (28).

FIG 2

(A) Phenotypic susceptibility to LPV and RTV of Gag-protease from the baseline samples. Dose-response curves of baseline (pre-ART) samples from 20 pediatric patients (gray) relative to the dose-response curve for control plasmid p8.MJ4GP (black) for both LPV and RTV. Dotted line, 50% inhibition. (B) Dose-response curves of the baseline sample and a sample obtained at the time of treatment failure from patient 1. The baseline sample from patient 1 showed increased susceptibility to LPV and RTV relative to the assay control, p8.MJ4GP. The fold change in the EC₅₀ relative to that for the sample obtained at the time of treatment failure is therefore higher than that for the baseline sample rather than the control plasmid. (C) Dose-response curves of samples obtained at the baseline and the time of treatment failure from patient 10. The baseline sample from patient 10 showed reduced susceptibility to LPV and RTV relative to that of the assay control, p8.MJ4GP. The fold change in the EC₅₀ relative to that for the sample obtained at the time of treatment failure is therefore lower than that for the baseline sample rather than the control plasmid. For actual values, see Fig. 1 and Table S1 in the supplemental material.

FIG 3

Impact of Gag amino acid changes on phenotypic resistance to LPV and RTV. The results for Gag CS mutations A431V, E428D, and N374S in various combinations with major protease amino acid changes (M46I, I54V, and V82A) are shown. Black, p8.MJ4 (MJ4), the positive control for the assay; vertical dotted lines, lower assay cutoff values; error bars, standard deviations of the means.