Natural polymorphism of HIV-1 subtype G protease and cleavage sites.

Abstract

The main difficulty in HIV-1 therapy with protease inhibitors (PI) is the rapid emergence of resistant strains with primary mutations on the protease gene, whose functionality is compensated by the appearance of secondary mutations both at protease and its cleavage sites. Accumulating data from PI-naive patients have documented significantly more naturally occurring PI resistance-associated secondary substitutions in HIV-1 non-B subtypes than in subtype B strains [1]. The presence of pre-existing secondary mutations could reduce the effectiveness of PI treatment through a more rapid progression to a resistant phenotype. Moreover, in the presence of some PI, proteases from non-B viruses show a higher biochemical fitness [2]. A previous genetic analysis of HIV-1 strains circulating in Lisbon, Portugal, demonstrated a high prevalence of non-B subtypes, with a predominance of subtype G, CRF02_AG and CRF14_BG [3,4]. In order to assess natural resistance mutations to PI in HIV-1 subtype G, we analysed protease and cleavage site amino acid substitutions in 52 antiretroviral therapy-naive individuals, 27% of whom were of African origin, half of them having reported injecting drug use. Previous phylogenetic characterization of env sequences and gag subtyping by heteroduplex mobility assay [3,4] had assigned 38 of the infecting viruses to subtype G, eight to CRF14_BG, and six to CRF02_AG. A 600 base pair fragment of proviral DNA, covering the 3′ end of gag and the protease coding sequence, was amplified from peripheral blood mononuclear cell lysates by nested polymerase chain reaction, with primer sets P1F, 5′–GGCTGTTGGAAATGTGGAAAGG–3′ (nucleotides 2023–2044 on HXB2), and P2R, 5′–TGGAGTATT GTATGGATTTTCAGG–3′ (nucleotides 2703–2726 on HXB2), in the first round, and P3F, 5′–GGAAAG GAAGGACACCAAATGAAAG–3′ (nucleotides 2038– 2062 on HXB2), and P4R, 5′–CTGTCAATGGCC ATTGTTTAACT–3′ (nucleotides 2609–2631 on HXB2), in the second round. Polymerase chain reaction products were sequenced and phylogenetically analysed as previously described [3]. Sequences have been deposited in GenBank under accession numbers AJ334987 to 535016, AJ535018 to 535021, and AJ535023 to 535040. Putative amino acid sequences were compared with HIV-1 subtype B consensus (available at http://hiv-web.lanl.gov and http://hivdb.stanford.edu/) and resistance associated mutations were defined following the Stanford HIV Database (SHD) criteria. Protease amino acid consensus sequences of subtype G and CRF14_BG viruses analysed show a similar polymorphism pattern and differ from the subtype B consensus in nine positions (I13V, K14R, K20I, E35D, M36I, R41K, H69K, V82I, L89M) (Table 1). CRF02_AG shares only seven of these positions, E35D and V82I being absent. These results were anticipated by phylogenetic tree analysis (data not shown), because CRF14_BG protease sequences were interspersed in the subtype G radiation studied, whereas the CRF02_AG sequences formed a separate cluster in clade G. Analysis of protease sequences on databases confirmed this polymorphism profile. However, lower frequencies of V82I have been reported for subtype G protease [5] (SHD), mainly because of the inclusion of CRF02_AG sequences. V82I, present in approximately 1.7% of untreated patients infected with other group M subtypes (SHD), is found in 93.5% of the G and CRF14_BG viruses studied and should therefore be considered a subtype G protease signature.Table 1: Polymorphism of Portuguese HIV-1 subtype G proteases, comparatively to subtype B consensus.Only two major mutations (D30N and N88S) were observed in the protease region, but a high rate of secondary mutations was found, the baseline K20I/M36I/V82I combination being observed in 93.5% of the subtype G proteases analysed. Other less frequent secondary mutations, L63P (n = 4), G16E (n = 3), and I47V (n = 2), along with six mutations occurring only once, account for or 17.4 and 2.2% of G sequences with four and six minor substitutions, respectively. Amino acid sequences for p7/p1, p1/p6gag, p6pol/protease, and protease/reverse transcriptase cleavage sites were also analysed and compared with corresponding sequences on databases (http://hiv-web.lanl.gov/). A low variability was observed for p7/p1 and protease/reverse transcriptase in subtype G and CRF02_AG strains, in accordance with their high degree of conservation within group M viruses. Concerning p1/p6gag, 95.6% of the subtype G sequences studied are characterized by mutation S451N, whereas 83.3% of the CRF02_AG sequences show the L449P polymorphism typical of subtype A, CRF01_AE and CRF02_AG. Mutations A431V, L449F, and P453L, associated with PI resistance [6,7], were not detected. The 5′ sequence of p6pol/protease shows high variability both intra and inter-subtypes. The amino acid motif IP/SLS/NL, consistently found in the G viruses studied (82.6%), is uncommon among other HIV-1 clades. The data presented provide evidence that natural secondary mutations are frequent in HIV-1 subtype G protease. Polymorphisms not associated with PI resistance were also observed both at protease and cleavage sites. The effect of these genetic differences on HIV-1 subtype G drug susceptibility should be evaluated on phenotypic assays.