Sensitive cell-based assay for determination of human immunodeficiency virus type 1 coreceptor tropism

Abstract

CCR5 antagonists are a powerful new class of antiretroviral drugs that require a companion assay to evaluate the presence of CXCR4-tropic (non-R5) viruses prior to use in human immunodeficiency virus (HIV)-infected individuals. In this study, we have developed, characterized, verified, and prevalidated a novel phenotypic test to determine HIV-1 coreceptor tropism (VERITROP) based on a sensitive cell-to-cell fusion assay. A proprietary vector was constructed containing a near-full-length HIV-1 genome with the yeast uracil biosynthesis (URA3) gene replacing the HIV-1 env coding sequence. Patient-derived HIV-1 PCR products were introduced by homologous recombination using an innovative yeast-based cloning strategy. The env-expressing vectors were then used in a cell-to-cell fusion assay to determine the presence of R5 and/or non-R5 HIV-1 variants within the viral population. Results were compared with (i) the original version of Trofile (Monogram Biosciences, San Francisco, CA), (ii) population sequencing, and (iii) 454 pyrosequencing, with the genotypic data analyzed using several bioinformatics tools, i.e., the 11/24/25 rule, Geno2Pheno (2% to 5.75%, 3.5%, or 10% false-positive rate [FPR]), and webPSSM. VERITROP consistently detected minority non-R5 variants from clinical specimens, with an analytical sensitivity of 0.3%, with viral loads of ≥1,000 copies/ml, and from B and non-B subtypes. In a pilot study, a 73.7% (56/76) concordance was observed with the original Trofile assay, with 19 of the 20 discordant results corresponding to non-R5 variants detected using VERITROP and not by the original Trofile assay. The degree of concordance of VERITROP and Trofile with population and deep sequencing results depended on the algorithm used to determine HIV-1 coreceptor tropism. Overall, VERITROP showed better concordance with deep sequencing/Geno2Pheno at a 0.3% detection threshold (67%), whereas Trofile matched better with population sequencing (79%). However, 454 sequencing using Geno2Pheno at a 10% FPR and 0.3% threshold and VERITROP more accurately predicted the success of a maraviroc-based regimen. In conclusion, VERITROP may promote the development of new HIV coreceptor antagonists and aid in the treatment and management of HIV-infected individuals prior to and/or during treatment with this class of drugs.

Fig 1

(A) Strategy to introduce patient-derived env PCR fragments into a proprietary vector via yeast homologous recombination as described previously (34, 35). (B) Overview of the novel HIV-1 coreceptor tropism assay (VERITROP). Patient-derived viral amplicons were introduced into a vector lacking the corresponding HIV-1 env sequence and cotransfected into HEK293T cells with a plasmid encoding the α fragment of the β-galactosidase gene (pCMVα). Cell-to-cell fusion events were quantified by transfecting the target cells (U87.CD4.CCR5 or U87.CD4.CXCR4) with a vector expressing the ω fragment (pCMVω) of the β-galactosidase gene and mixing them with the HIV-1 env- and pCMVα-expressing cells.

Fig 2

HIV-1 coreceptor tropism test: proof of principle. The new assay was tested by mixing different amounts of env amplicons from R5 (HIV-1_BaL) and X4 (HIV-1_NL4-3) prior to cloning into the HIV expression vector. Cell-to-cell fusion events were quantified, including controls to take into account potential nonspecific luminescence expression, i.e., (i) mock transfection, (ii) transfecting HEK293T cells with the α fragment but no env amplicon and target (U87) cells with no ω fragment, and (iii) HEK293T cells expressing HIV-1 env (R5 BaL or X4 NL4-3) with the α fragment in the total absence of target (U87) cells. Technical cutoffs for the quantification of env-mediated cell fusion events (green and blue dashed lines) were calculated as the means plus two standard deviations (SD) of the β-galactosidase activity detected after HEK293T cells transfected with 100% R5 HIV-1_BaL or 100% X4 HIV-1_NL4-3 were incubated in cell-to-cell fusion experiments with U87.CD4.CXCR4 or U87.CD4.CCR5 cells, respectively. The bar denoting the minimal amount of non-R5 (X4) env detected with statistical significance (0.3%) is enclosed by a red dashed line. RLU, relative light units.

Fig 3

Analytical sensitivity. The ability of the novel HIV-1 coreceptor tropism assay to detect minority non-R5 viruses within the HIV-1 population was evaluated by mixing at different proportions: plasmid DNA (HIV-1_BaL and HIV-1_NL4-3) prior to transfection of the HEK293T cells (A), yeast colonies prior to plasmid DNA isolation (used to transfect HEK293T cells) (B), and HIV-1_BaL and HIV-1_NL4-3 DNA prior to PCR amplification (C). R5 and X4 technical cutoffs were calculated as described in the legend to Fig. 1 and Materials and Methods. Bars denoting the minimal amount of non-R5 (X4) env detected with statistical significance (0.3%) are enclosed by a red dashed line. RLU, relative light units.

Fig 4

Reproducibility of the HIV-1 coreceptor tropism assay. (A) env PCR products from three viruses with different and known HIV-1 coreceptor usage (X4, HIV-1_NL4-3; R5, HIV-1_BaL; and dual/mixed, HIV-1_10-172) were analyzed 10 times (10× replicates), by three operators, with distinct lots of critical reagents over a 4-week period. The coefficient of variation (%) is indicated above each bar. R5 and X4 technical cutoffs were calculated as described in the legend to Fig. 1 and Materials and Methods. D/M, dual/mixed. (B) Reproducibility of the entire assay, starting from RNA extraction from 15 plasma samples from HIV-infected individuals, as determined by two different operators with different lots of critical reagents over a 14-day period. Pearson correlation coefficient was used to determine the strength of association between the two determinations; r, correlation coefficient; p, two-tailed P value.

Fig 5

Comparison of the novel HIV-1 coreceptor tropism assay with another phenotypic (Trofile) and two genotypic (population and 454 sequencing) assays. Plasma samples from 76 HIV-infected individuals prior to treatment with the CCR5 antagonist maraviroc were analyzed. HIV-1 coreceptor tropism was determined using two phenotypic assays (Trofile and VERITROP) and inferred from population and 454 sequences using a series of algorithms as described in Materials and Methods. (A) Hierarchical clustering analysis was used to group the HIV-1 coreceptor tropism determinations by similarity. Dendrograms were calculated using the Euclidean distance and Complete cluster methods with 100 bootstrap iterations as described (http://www.hiv.lanl.gov/content/sequence/HEATMAP/heatmap.html). Green and red blocks indicate the absence or presence of non-R5 (X4) viruses, respectively, as determined by each assay. G2P, Geno2Pheno; PSSM, position-specific scoring matrices. Bootstrap values above 55% are indicated. (B) The ability of the different HIV-1 coreceptor tropism tests to predict the success of a maraviroc-based therapy was determined as the percentage of patients where non-R5 viruses were detected at baseline (no. of non-R5 viruses) from the total number of patients who were not able to enter (due to the presence of non-R5 viruses at baseline as determined by Trofile) or failed to respond to the maraviroc-based regimen (n = 36), i.e., % treatment success prediction = (no. of patients with non-R5 viruses/no. of patients not entering or failing the maraviroc-based regimen [n = 36]) × 100. Values for the tests with the three highest percentages (best capability to predict the success of maraviroc-based therapy) are indicated.