Background: Co-receptor antagonists constitute a promising new class of anti-HIV-1 drugs, with several lead compounds currently in full clinical development. Testing for co-receptor utilization (or tropism) prior to initiating therapy with a CCR5 antagonist will be critical to avoid the use of these compounds in patients that are infected with X4 or R5X4 strains. At least two tropism prediction algorithms (PSSM, http://ubik.microbiol.washington.edu/computing/pssm/; and Geno2Pheno (G2P), http://coreceptor.bioinf.mpi-sb.mpg.de/cgi-bin/coreceptor.pl) are publicly available, both based on the analysis of specific amino acid (aac.) characteristics of the V3-loop of HIV-1 env. In the current study we compared the predictions yielded by the PSSM and G2P on clonal V3 env sequences.

Methods: The HIV-1 env V3-V4 region (174 aac. in JR-CSF) was amplified from 389 clinical plasma samples in a single round RT-PCR reaction using the SuperScript III One-Step RT-PCR with Platinum Taq High Fidelity (Invitrogen, Belgium). Amplification products were pooled and cloned into the pCR4-TOPO vector (Invitrogen, Belgium). Individual clones were selected for insert amplification and sequencing.

Results: The env V3-V4 region was successfully amplified in 345 clinical samples (89%). A clonal library of 256 entrees was created. Ten clones contained a premature stop codon and were excluded from further analysis. The env-encoded domain amplified between primers varied from 165 to 190 amino acids (24 different lengths, with the most frequent (n = 39) being 174 aac. long). The tropism predictions are summarized in Table 1. Most of the G2P unpredictable clones were scored as R5 by PSSM, however the scores were significantly less ‘R5-like’ (p<0.0001). The aac. length of the V3 regions were: 34 (in 28 out of 32 clones, aac. 24 was missing), 35, and 37 (35 aac. in JR-CSF; 36 aac. in HXBII) (Table 1).

Table 1: Tropism prediction of 246 clones sorted by V3-loop length.

D = dual tropism, ? = tropism unpredicted.

Conclusion: A wide diversity in length and composition of clonal sequences was observed. Tropism predictions using genotyping should be interpreted carefully since these may vary depending on the program used.