Consistent patterns of change during the divergence of human immunodeficiency virus type 1 envelope from that of the inoculated virus in simian/human immunodeficiency virus-infected macaques

Abstract

We have analyzed changes to proviral Env gp120 sequences and the development of neutralizing antibodies (NAbs) during 1 year of simian/human immunodeficiency virus SHIV-89.6P infection in 11 Macaca nemestrina macaques. Seven macaques had significant env divergence from that of the inoculum, and macaques with greater divergence had higher titers of homologous NAbs. Substitutions in sequons encoding potential N-linked glycosylation sites (PNGs) were among the first to be established, although overall the total number of sequons did not increase significantly. The majority (19 of 23) of PNGs present in the inoculum were conserved in the sequences from all macaques. Statistically significant variations in PNGs occurred in multiple macaques within constrained regions we term "hot spots," resulting in the selection of sequences more similar to the B consensus. These included additions on V1, the N-terminal side of V4, and the outer region of C2. Complex mutational patterns resulted in convergent PNG shifts in V2 and V5. Charge changes in Env V1V2, resulting in a net acidic charge, and a proline addition in V5 occurred in several macaques. Molecular modeling of the 89.6P sequence showed that the conserved glycans lie on the silent face of Env and that many are proximal to disulfide bonds, while PNG additions and shifts are proximal to the CD4 binding site. Nonsynonymous-to-synonymous substitution ratios suggest that these changes result from selective pressure. This longitudinal and cross-sectional study of mutations in human immunodeficiency virus (HIV) env in the SHIV background provides evidence that there are more constraints on the configuration of the glycan shield than were previously appreciated.

FIG. 1.

Phylogenetic analysis of individual HIV-1 envelope gp120 proviral clones from macaques infected with SHIV-89.6P at 32 weeks postinfection. The maximum likelihood tree is rooted on the consensus of the inoculum. Sequences from each macaque are clustered together and uniformly colored. Branches were collapsed if the bootstrap value was below 50; bootstrap values over 50 are labeled on the tree.

FIG. 2.

Graphical representation of potential N-linked glycosylation sites in SHIV-89.6P Env and HIV-1 clade B Env. N-Glycosite was used to quantify the locations and prevalence of PNGs within the SHIV-89.6P inoculum (A), pooled SHIV-89.6P variants from all macaques (B), and an HIV-1 clade B reference alignment (C). The amino acid number of each PNG is labeled on the x axis. The prevalence of each PNG is graphed as a fraction of the total on the y axis. Variable regions are denoted as a bar at the bottom of each figure; shaded boxes within the variable regions represent hot spots of PNG variation. PNGs that are conserved (>90% prevalent) are shown in blue, PNGs that are heterologous within the inoculum are shown in orange, convergent PNG additions are shown in green, and convergent PNG shifts are shown in purple. The PNGs that are conserved in both SHIV-89.6P and HIV-1 clade B are denoted with asterisks in panels B and C.

FIG. 3.

PNG combinations within the Env quasispecies members of each macaque at 32 weeks postinfection. Each horizontal bar corresponds to a combination of PNG changes that exists within the quasispecies of each animal. The number of clones with a given combination is shown to the left of the bars. The position of each PNG is shown along the top. If the PNG is present, it is shown as a vertical line. Identical combinations found within different animals are denoted with symbols on the right.

FIG. 4.

Three-dimensional structural illustration of locations of N-glycosylation sites. Locations of PNGs from the SHIV-89.6P clones were mapped onto a model of gp120 based on the X-ray structure of the CD4-bound YU2 gp120 core (35) (PDB code 1RZK). The V1, V2, and V3 loops (pink) of YU2 are modeled onto the core (red) for completeness and identification of N-glycosylation sites. In the gp120 orientation shown, the viral and target membranes would be located above and below the protein, respectively. The yellow lines show the disulfide bridges. The N-glycosylation sites that are stable (blue), added (green), shifted (purple), and heterologous within the inoculum (orange) are shown as balls and labeled with the corresponding amino acid numbers. The predicted CD4bs is shown in gray; residues were included based on the contacts identified within 5 Å of CD4 in the CD4-bound YU2 crystal structure (35). (A) Cartoon diagram showing the locations of all PNGs in 89.6P. The inset highlights a subset of PNGs proximal to the disulfide bridges. (B) Space-filling model of the inner face. The predicted CD4bs (gray) is partially occluded by the modeled V1V2 loop (pink). All variant PNGs (underlined) lie on the inner face of Env. (C) Space-filling model of the outer face. The PNGs on the outer domain of gp120 are predominantly conserved.

FIG. 5.

Accumulation of Env sequence changes over time in two macaques. The variable regions of representative longitudinal sequences from two macaques with significant divergence are shown. The consensus sequence of Env in the SHIV-89.6P inoculum is shown along the top. A charged 40-residue region from positions 152 to 192 is highlighted in V1/V2.

FIG. 5.

Accumulation of Env sequence changes over time in two macaques. The variable regions of representative longitudinal sequences from two macaques with significant divergence are shown. The consensus sequence of Env in the SHIV-89.6P inoculum is shown along the top. A charged 40-residue region from positions 152 to 192 is highlighted in V1/V2.

FIG. 6.

Net charge of the V2 region becomes more acidic at week 32 in animals with diversified quasispecies. A 40-residue region from positions 152 to 192 had a median charge of +4 in the inoculum. The four macaques with minimal divergence (K97107, K98099, L97191, and J97168) maintained this charge. The seven animals with viruses that diverged from the inoculum (A98069, T98098, J97156, K97246, L98152, T98108, and J98071) had mutations in this region that resulted in a more acidic (more negative) net charge. The vertical line divides animals that do not have any significant PNG changes from those that do. Macaque numbers on the x axis are arrayed in order of increasing levels of divergence.

FIG. 7.

Development of homologous NAbs as a function of time postinfection. Sera from the SHIV-89.6P-infected macaques were tested against pseudotyped SHIV-89.6P in a Tzm-bl cell assay. Reciprocal titers for 50% neutralization are shown. The y axis for animal K98099 is different (higher values) from the rest. Each sample was run in triplicate, and most data points represent the geometric means of three independent assays. Macaques K97107, L97191, J97168, and L98152 had no detectable NAbs at any time point tested (not shown).