Improved Prediction of HIV‐1 Coreceptor Usage with Sequence Information from the Second Hypervariable Loop of gp120 (2024)

V2 Loop Improves Tropism Prediction • JID 2010:202 (1 November) • 1435 MAJOR ARTICLE Improved Prediction of HIV-1 Coreceptor Usage with Sequence Information from the Second Hypervariable Loop of gp120 Alexander Thielen, 1 Nadine Sichtig, 2 Rolf Kaiser, 2 Jeffrey Lam, 3 P. Richard Harrigan, 3 and Thomas Lengauer 1 1 Max Planck Institute for Informatics, Saarbru¨cken, and 2 University of Cologne, Cologne, Germany; 3 British Columbia Centre for Excellence in HIV/AIDS, Vancouver, Canada Background. Human immunodeficiency virus type 1 (HIV-1) uses the CD4 receptor and a coreceptor to gain cell entry. Coreceptor usage is mainly determined by the V3 loop of gp120. Therefore, coreceptor usage is currently inferred from the genotype on the basis of V3 alone. However, several mutations outside V3 have been repeatedly reported to influence coreceptor usage. In this study, the impact of the V2 loop on coreceptor usage prediction was analyzed. Methods. Sequences were analyzed for differences at specific positions and position-independent features with the Fisher exact and Student t tests. Prediction models were trained with support vector machines and evaluated in cross-validation on clonal data. Models trained on the clonal data set were validated on 2 clinical data sets. Results. Several mutations and position-independent features within V2 were statistically significantly different between R5 and X4 viruses. Cross-validation on the clonal data set revealed a statistically significantly higher area under the receiver operating characteristic curve if features of both loops were used, compared with those using only V2 or V3 alone. Similar results were found with clinically derived data sets. Conclusions. The ability of the V2 loop to improve coreceptor usage prediction has been shown in a large data set. Utilization of this information can lead to considerable improvements in the prediction of coreceptor use both on clonal data sets and on clinically derived data sets. The infection of human cells by human immunodefi- ciency virus (HIV) is mediated through the CD4 re- ceptor and a second coreceptor. Several different co- receptors have been described from experiments per- formed in vitro. Only 2 of them have been shown to be relevant in vivo: CCR5 and CXCR4 [1–6]. A virus is called an R5 virus if it can only use CCR5, whereas it is called an X4 virus if it only uses CXCR4 or a dual- tropic (R5X4) virus if it can infect cells through both coreceptors. Viral quasispecies containing mixed pop- Received 1 December 2009; accepted 26 May 2010; electronically published 27 September 2010. Potential conflicts of interest: none reported. Presented in part: 17th International HIV Drug Resistance Workshop, Sitges, Spain, 10–14 June 2008 (abstract A100); 3rd International Workshop on Targeting HIV Entry, Washington, DC, 7–8 December 2007 (abstract A1). Financial support: German Ministry of Health (grant BMG 310/4476/02/3). Reprints or correspondence: Alexander Thielen, Max Planck Institute for Informatics, Stuhlsatzenhausweg E1.4, 66123 Saarbru¨cken, Germany (athielen@mpi-inf.mpg.de). The Journal of Infectious Diseases 2010; 202(9):1435–1443  2010 by the Infectious Diseases Society of America. All rights reserved. 0022-1899/2010/20209-0018$15.00 DOI: 10.1086/656600 ulations capable of using CCR5 as well as CXCR4 are called dual-mixed viruses. In vivo, CXCR4-using strains usually appear together with CCR5-using viruses and only rarely alone [7]. R5 and dual-mixed isolates are observed in different phases of infection. In general, R5 viruses predominate early in infection whereas dual-tropic or X4 viruses emerge only in the later stage of the disease in about half of patients [2, 7–9]. Several studies have shown that the emergence of CXCR4-using viruses is corre- lated with accelerated disease progression [10, 11] and reduced survival time in untreated individuals [11, 12]. Whether this is the cause or the result of immune ex- haustion is still unclear. Patients hom*ozygous for a 32-bp deletion in the CCR5 gene that prevents the expression of the receptor seem to be almost resistant against HIV [13–16]. This led to the assumption that only R5 viruses are trans- mitted during infection [17] and to the development of a new class of antiretroviral drugs that block the CCR5 receptors [18]. The first of these coreceptor an- tagonists—maraviroc [18, 19]—was approved in the Downloaded from https://academic.oup.com/jid/article-abstract/202/9/1435/847890 by guest on 11 March 2019

1436 • JID 2010:202 (1 November) • Thielen et al Figure 1. Sequence profile of the V2 loop of human immunodeficiency virus type 1 (HIV-1). Positions are numbered according to HXB2. United States and Europe in 2007, and another—vicriviroc (Schering Plough)—is undergoing phase 3 clinical trials. Be- cause these drugs are only effective against R5 viruses and do not inhibit CXCR4-using viruses, a tropism test is mandatory before administration. The most widely used test measuring viral tropism is the Monogram Trofile tropism assay (Monogram Biosciences) [20]. It is a recombinant phenotypic assay that uses replication-de- fective viruses. It has been used in all clinical trials and has therefore become the de facto gold standard for measuring coreceptor tropism. Several other phenotypic assays have been developed [21, 22]. Phenotypic approaches for tropism determination are precise and reproducible, but they share substantial drawbacks with their counterparts in resistance testing: they are expensive, have slow turnaround times, and can only be performed by a small number of laboratories [23, 24]. As for resistance testing, ge- notypic approaches provide an alternative. These are faster and cheaper and yield good results on the basis of clonal data [25]. When used to determine tropism on clinically derived isolates, however, they have previously appeared to perform much worse than phenotypic approaches [25, 26]. A noteworthy limitation is that common prediction methods are based on only the third hypervariable loop of gp120 (V3). Although this short region of about 35 residues is known to be the major determinant of coreceptor usage [27–30], mutations in the bridging sheet [31– 37] and gp41 [38] have been correlated with tropism. So far, however, except for a study by Prosperi et al [35] with addi- tional patient features and mutations outside V3, whether the incorporation of these mutations can improve prediction meth- ods has not been evaluated. In this study, we address this ques- tion by analyzing the V2 loop, which has been reported re- peatedly to be responsible for tropism switches [32–34] in conjunction with V3 sequences. MATERIALS AND METHODS Data sets. Three distinct data sets of different character were used. The first (clonal data set) included all samples of the Los Alamos HIV Sequence Database (http://www.hiv.lanl.gov/ content/sequence/HIV/mainpage.html; accessed July 2007) that had sequence information from the V2 and V3 loops of the envelope protein gp120 as well as an experimentally determined phenotype. The data set consists of 916 sequences from 312 patients. In fact, the data set is mostly clonal; we could confirm that ∼85% of the samples are indeed clonal. However, restrict- ing the analysis to the subset of clonal samples showed only little effect. The second data set (therapy-naive data set) contained 268 samples from a subset of therapy-naive patients in the well- described British Columbia HAART Observational Medical Eval- uation and Research (HOMER) cohort [39] (HAART, highly active antiretroviral therapy). The cohort consists of HIV-posi- tive, antiretroviral-naive adults who started triple antiretroviral therapy through the British Columbia Drug Treatment Program between August 1996 and September 1999. Samples used in this study were the latest ones collected in the 180 days prior to therapy initiation [39]. Coreceptor usage was determined by the original Trofile assay. The third data set (therapy-experienced data set) consisted of isolates from 64 therapy-experienced patients that were screened with the original Trofile assay [20] for maraviroc ad- ministration. This data set was collected at the Institute for Virology of the University of Cologne (Cologne, Germany). Sequences are available in GenBank under the follow- ing accession numbers: HM176666–HM176793, HM211408– HM211675, and EF637088–EF638008. For specific subsets, see the Appendix, which appears only in the online version of the Journal. Determination of envelope sequences. Sequence variation of V2 and V3 was determined by population sequencing. Iso- lates from patients in the therapy-naive data set were genotyped as described elsewhere [10]. Viral RNA of samples collected for the therapy-experienced data set was isolated from plasma by using the MagNA Pure compact nucleic acid isolation kit (Roche Applied Science) according to manufacturer’s protocol. Reverse transcription and polymerase chain reaction (PCR) were performed using the OneStep reverse-transcription PCR Downloaded from https://academic.oup.com/jid/article-abstract/202/9/1435/847890 by guest on 11 March 2019

Download

Download

Download

Download

Download

Download

Download

Download

Download

Download