Statistical correlation of nonconservative substitutions of HIV gp41 variable amino acid residues with the R5X4 HIV-1 phenotype

Important features of the HIV-1 induced disease are determined by the interaction of three main classes of viruses with different subsets of CD4+ cells, currently designated as R5, X4 and R5X4 viruses depending on the coreceptor they use to enter cells (CCR5, CXCR4, or both, respectively). CCR5 is expressed mainly by macrophages and the activated/memory T subset, whereas CXCR4 is predominantly expressed by the naïve, but also the memory, subsets of CD4+ T-lymphocytes and by CD4+ T-cell lines. R5 viruses are responsible for transmission and persist through the whole course of the disease in most of patients. The appearance of R5X4 and X4 viruses in blood associates with the onset of AIDS [1].

Entry of the HIV-1 genome into target cells depends on trimmeric complexes of the viral envelope glycoprotein (Env) heterodimer, which is composed of a hypervariable surface subunit (gp120), and a more conserved, though highly variable, transmembrane subunit (gp41) [2]. CD4 binding to gp120 induces the exposure/formation of the binding site for the coreceptor [3]. The gp120-CD4-coreceptor interaction then allows the extension of gp41 and the insertion of the fusion peptide into the target membrane. Current models indicate that packing of three gp41 C-terminal helices into the grooves of a coiled coil formed by the N-terminal helices forms a structure known as the six-helix bundle, enforcing virus-cell membrane fusion [4, 5].

Determinants of HIV-1 coreceptor tropism have been identified mainly in the hypervariable gp120 V3 loop, where a high positive net charge associates with X4 tropism [6, 7]. V1, V2 and V5 loops modulates the V3 effects [8‐12]. In addition, experimental evidence of the participation of gp41 in coreceptor recognition has been provided [13‐16]. Gp41 contains approximately 346 amino acids and is composed of an ectodomain, a membrane spanning domain, and a long C-terminal tail (CTT). The ectodomain is organized in an N-terminal fusion peptide, two helical regions known as HR1 and HR2, a central loop, and the membrane proximal external region (MPER). In the ectodomain, HR2 concentrates the highest variation rate [17], whereas the C-terminal tail display the higher average diversity in the protein [2]. Theoretical studies have shown the statistical association of gp41 with coreceptor tropism although there is no a consensus about the putative sites implicated [18‐21], and congruency with experimental investigations of coreceptor associated mutations [13, 14] is not clear. Given the high variability and adaptive nature of gp41, discordances may be caused by differences in the databases used, as well as to distinct analytical approaches. Thus, while it seems clear that different gp41 domains participate in determination of virus phenotype, the specific changes involved may develop in a complex, context-dependent manner, similarly to the different mutational pathways observed in studies of the correlates of the gp120 sequence with coreceptor tropism [10] or that obtained for resistance to maraviroc of R5-tropic viruses [22].

Unlike other studies, we focused our analysis on the relationship of the hydropathy index and charge of amino acid positions between coreceptor groups in order to determine if general physicochemical properties of gp41 residues correlate with different virus phenotypes. In addition, we focused on highly variable amino acid positions of gp41 since conserved positions are most probably engaged in maintaining the highly stringent structural properties required for membrane fusion. With this purpose, we retrieved amino acid sequences of a set of 2823 HIV-1 strains from all clades with known coreceptor tropism from Los Alamos HIV Database. Consensus sequences were constructed for homologous sequences (those obtained from the same patient and having the same tropism) in order to avoid bias due to sequence overrepresentation. Then, the variability (entropy) per site was determined and amino acid positions with high variability scores or with large differences in variability between coreceptor groups were selected. Next, we performed a statistical analysis for the association of the viral tropism (X4, R5 and R5X4) with the hydropathy index (HI) and charge (Q) of amino acid residues at those positions. Twelve positions were found linked to coreceptor usage in this analysis. We suggest that some of the most gp41 variable residues are involved in the coreceptor recognition process.

Our results indicate that the R5X4 phenotype associates with a hydrophobic tendency of positions at the C-terminal half of the loop (619) the HR2 (640, 641), the so called minor ectodomain (723), and the putative MSD3 (756), as well as with a hydrophilic/charged tendency in a residue at the disulfide bridge region of the loop (602), and the HR2 (636, 662). The location of the nine positions belonging to the ectodomain is shown in the structure of the six-helix bundle in Fig. 4. Since this study is correlative, it does not necessarily implicates that these residues establish contact with coreceptor molecules, but only that hydrophobic or hydrophilic residues at these positions are more frequently harbored by R5X4 than R5 and X4 viruses. However, it can be speculated that they may contribute to virus phenotype by several mechanisms. Position 602 is the most variable site in the disulfide bridge region of the loop (Figs. 1 and 4). It is known that hydrophobicity of the loop is important for the stability of the gp120-gp41 association [27], so a hydrophilic residue at position 602 may favor gp120 shedding and fusion. Position 619 is part of the LEQ – leucine-glutamate-glutamine in the HXB2 strain – highly variable triplet located at the loop-HR2 boundary (Fig. 4). To our knowledge, there are no experimental studies regarding the role of this position. However, a more conserved fragment comprising nearby residues 579–613 of the loop (which includes the 602 residue) and another fragment containing the 619 amino acid, interact with and perturb cellular and model membranes [28‐30]. It has been hypothesized that the loop may bind to and destabilize the host cell membrane, as well as stabilize the trimeric helical hairpin, then favoring the formation of the fusion pore [28]. Thus, a hydrophobic 619 residue in R5X4 strains may enhance the interaction of the loop with membranes. On the other hand, since the loop is part of a wide region composing the gp120-gp41 interface [27, 31], it may influence the efficiency of gp120 shedding. It has been demonstrated that gp120 shedding requires the presence of CXCR4 [5], although a similar analysis for CCR5 is still lacking.

HR2 amino acids 636, 640, and 641 may participate in coreceptor recognition by interacting with the gp120 coreceptor binding site. The HR2-based peptide T-20 interacts with peptides derived from the bridging sheet [32], and can block the interaction of gp120-CD4 complexes with the CXCR4 coreceptor through binding a region near the base of the gp120 V3 loop [33]. Recently, Moseri and cols. showed that T-20 binds to the conserved region 4 of R5 gp120 trough mostly hydrophobic interactions [34]. On the other hand, the direct interaction of the gp41 ectodomain with the coreceptor molecule has been suggested by the observation that T-20 and the related T22 peptide, inhibited the binding to CXCR4 of the anti-CXCR4 HIV-blocking antibody 12G5 [35]. CXCR4, but not CCR5, contains a highly hydrophobic groove in the region located between the second and third extracellular loops. Since the second extracellular loop is critical for coreceptor function [36, 37], this region represents a putative site for interaction with the hydrophobic residues of the gp41 ectodomain of R5X4 viruses. Finally, it is possible that residues 619, 640 and 641 of R5X4 gp41 proteins strengthen the interaction of this molecule with membrane lipids. HR1 and HR2 peptides interact with membrane vesicles and it has been proposed that they play an important role in the interaction of gp41 with the viral and cellular membranes during the opening of the fusion pore [38‐42]. Current structural models indicate that residues 636, 640, 641 are not part of the HR1-HR2 interface in the six-helix bundle [43], so they would be exposed on this structure and available for membrane interactions in late stages of the fusion process, contributing to fusogenicity and pathogenicity of R5X4 viruses (Fig. 4).

Importantly, correlation analysis revealed that the hydropathy index of pairs 602–640, 602–723, 619–640, 636–640, 640–662, and 640–756, covariate with higher correlation coefficients in the R5X4 group than in the R5 and X4 groups (Table 4), suggesting a complementary functionality of these residues for determination of the R5X4 phenotype. The positive covariation of the 619–640 and 640–756 pairs suggests a joint hydrophobic effect of these positions in R5X4 viruses for membrane lipid interactions (Fig. 2). On the other hand, the negative covariation observed for positions with opposed hydropathy tendencies (602–640, 602–723, 636–640 and 640–662) remarks the importance of the concurrence of hydrophilicity at positions 602, 636 and 662 (Fig. 2) for the R5X4 phenotype. In particular, the participation of position 640 in five of six covariations and the exposed position of this residue on the six-helix bundle structure (Fig. 4), suggest an important role of this residue for the R5X4 phenotype.

Residue 723 is part of a region in the C-terminal tail that may be transiently exposed on the surface virus and infected cells and is so called the minor ectodomain [23‐25], while position 756 locates in a region that may constitute a third membrane spanning domain (MSD3) during exposition of the minor ectodomain [25, 26]. A hydrophobic residue at this position may favor the exposure of the minor ectodomain, although with still unsuspected consequences.

A less restrictive analysis (Q_FD = 0.1) rendered additional positions located at different domains of gp41 and again, only for the R5X4 vs. R5 comparison. Thus, statistical analysis suggests a role for gp41 in the R5X4 virus phenotype.

Our analysis of the relationship of the gp41 sequence with virus phenotype did not yield differences between the X4 and R5 groups. It is well known that V3 gp120 residues influence the macrophage-tropic R5 (M-R5) and T-cell tropic (T-X4) viral phenotypes [44, 45], yet the role of V3 as a major determinant of phenotype is less clear in the case of dually tropic viruses [8]. Since our analysis was performed independently of the gp120 sequence, it is likely that we only observed residues influencing the R5-R5X4 shift in gp41, whereas residues in gp120 would be significant in determination of the R5 and X4 phenotypes.

R5 and R5X4 are the two main classes of viruses found in the circulation of patients with HIV-1 infection. Our analysis suggests that physicochemical properties of the variable amino acid residues at positions 602, 619, 636, 640, 641, 662, 723 and 756 of gp41 may contribute to enhanced virus-host membrane fusion of R5X4 viruses respective to R5 viruses.