Genetic changes in HIV-1 gag and pol genes associated with protease inhibitor-based therapy failure in paediatric patients

Abstract

HIV-1 infection in paediatric patients is a major public health issue with an estimated 740 000 children living with HIV-1 globally, and South Africa accounting for almost half of these infections. In South Africa, children <3 years of age are initiated on a protease inhibitor (PI)-based regimen on diagnosis. For those who fail therapy, there are limited second-line treatment options available. While virological failure is often associated with HIV-1 drug resistance, a number of studies have shown infrequent selection of protease mutations in adults and children. The protease enzyme recognises and cleaves the gag substrate at specific sites to release structural components and enzymes essential for viral replication. Thus, changes occurring at the gag cleavage sites (CS) represent an alternate route of PI resistance. We therefore investigated the genotypic changes in the gag substrate in paediatric patients failing PI-based therapy to determine if changes in gag could contribute to PI failure. The phenotypic effects of these changes on PI susceptibility both in the presence and absence of PI mutations were also examined. In addition, the impact of key protease and gag mutations were assessed by site-directed mutagenesis on phenotypic resistance. Furthermore, cross-resistance to other PIs were tested to identify PIs suitable for subsequent use. Plasma samples from 20 HIV-1 subtype C-infected children <2 years of age at pre-treatment and failing PI-based therapy were genotyped to identify changes in gag and protease associated with regimen failure. Fourteen patients received RTV as a single PI of which 7 were later switched to LPV/r. Six patients received LPV/r only. Following laboratory confirmed virologic failure, genotyping was performed on the first available sample (mean time to failure of 285 days). Major PI mutations (M46I, I54V and V82A) were found in 8 (40%) patients, all of whom had received RTV as a single PI at some point. Gag CS mutations were observed in 5 of these 8 patients (at codons 374, 378, 428, 431 and 451). In addition, amino acid changes at gag non-CS were noted, some of which were predicted to be under HLA/KIR immune-mediated pressure and/or drug selection pressure. These changes in gag following PI therapy failure were further explored by phenotypic analysis to assess whether they contributed to resistance. The gag-protease genes from pre-treatment and post-failure specimens from all 20 patients were cloned into an HIV-1 expression vector and tested for their PI susceptibility using an in vitro single-cycle phenotypic assay. Phenotypic resistance to LPV and RTV (up to >100-fold) was seen in 7 of the 8 patients with major PI mutations. Of the 12 patients with genotypically wild-type protease, 3 showed reduced susceptibility to RTV, 1 of which also had reduced susceptibility to LPV. These patients had changes in gag CS (L449P and P453L) or non-CS (codons 62 and 69 also under positive selection pressure) causing up to a 7-fold increase in resistance. Investigation of cross-resistance to other PIs, revealed that of the 8 patients with PI major mutations, 6 showed reduced susceptibility to atazanavir (ATV), 4 to amprenavir (APV), 1 to darunavir (DRV) and 4 to saquinavir (SQV). In addition, 3 patients with genotypically wild-type protease who showed reduced susceptibility to ATV (up to 11-fold) and in one case also to DRV (5-fold), had gag CS (S373A, L449P and P453L) and/or gag non-CS mutations (T62A and K69R). To investigate the effects of individual gag and protease mutations on PI susceptibility, single and multiple mutants were created in a subtype C reference vector by site-directed mutagenesis and tested in the single-cycle phenotypic assay. Analysis of single and multiple gag and protease site-directed mutants showed that individual changes were not sufficient to cause resistance in this assay system. However, in combination with major protease mutations, gag mutations were shown to contribute to PI resistance by reducing susceptibility to PIs by up to 4-fold. This suggests a minimal but significant impact and is consistent with findings using clinical samples. Overall, these results show that RTV therapy selects for genotypic and phenotypic resistance that caused cross-resistance to other PIs. This rarely occurred following LPV/r-based therapy which is the standard first-line treatment for infants. PI resistance was predominantly explained by the presence of major protease mutations. In some patients, gag mutations reduced PI susceptibility but their effect was more pronounced when they were present in combination with major protease mutations. Thus, incorporating gag in drug resistance algorithms could improve interpretation of genotyping data and drug resistance prediction. Furthermore, these results support the use of genotypic testing of children failing first-line treatment to guide future PI therapy options. However, eight children failed PI therapy in the absence of gag and protease mutations, highlighting the importance of further investigation into the reasons for paediatric failure, including reliable measures of adherence.