Structure-Based Design of a Soluble Prefusion-Closed HIV-1 Env Trimer with Reduced CD4 Affinity and Improved Immunogenicity

Abstract

The HIV-1 envelope (Env) trimer is a target for vaccine design as well as a conformational machine that facilitates virus entry by transitioning between prefusion-closed, CD4-bound, and coreceptor-bound conformations by transitioning into a postfusion state. Vaccine designers have sought to restrict the conformation of the HIV-1 Env trimer to its prefusion-closed state as this state is recognized by most broadly neutralizing, but not nonneutralizing, antibodies. We previously identified a disulfide bond, I201C-A433C (DS), which stabilizes Env in the vaccine-desired prefusion-closed state. When placed into the context of BG505 SOSIP.664, a soluble Env trimer mimic developed by Sanders, Moore, and colleagues, the engineered DS-SOSIP trimer showed reduced conformational triggering by CD4. Here, we further stabilize DS-SOSIP through a combination of structure-based design and 96-well-based expression and antigenic assessment. From 103 designs, we identified one, named DS-SOSIP.4mut, with four additional mutations at the interface of potentially mobile domains of the prefusion-closed structure. We also determined the crystal structures of DS-SOSIP.4mut at 4.1-Å resolution and of an additional DS-SOSIP.6mut variant at 4.3-Å resolution, and these confirmed the formation of engineered disulfide bonds. Notably, DS-SOSIP.4mut elicited a higher ratio of tier 2 autologous titers versus tier 1 V3-sensitive titers than BG505 SOSIP.664. DS-SOSIP.4mut also showed reduced recognition of CD4 and increased thermostability. The improved antigenicity, thermostability, and immunogenicity of DS-SOSIP.4mut suggest utility as an immunogen or a serologic probe; moreover, the specific four alterations identified here, M154, M300, M302, and L320 (4mut), can also be transferred to other HIV-1 Env trimers of interest to improve their properties.IMPORTANCE One approach to elicit broadly neutralizing antibodies against HIV-1 is to stabilize the structurally flexible HIV-1 envelope (Env) trimer in a conformation that displays predominantly broadly neutralizing epitopes and few to no nonneutralizing epitopes. The prefusion-closed conformation of HIV-1 Env has been identified as one such preferred conformation, and a current leading vaccine candidate is the BG505 DS-SOSIP variant, comprising two disulfides and an Ile-to-Pro mutation of Env from strain BG505. Here, we introduced additional mutations to further stabilize BG505 DS-SOSIP in the vaccine-preferred prefusion-closed conformation. In guinea pigs, our best mutant, DS-SOSIP.4mut, elicited a significantly higher ratio of autologous versus V3-directed neutralizing antibody responses than the SOSIP-stabilized form. We also observed an improvement in thermostability and a reduction in CD4 affinity. With improved antigenicity, stability, and immunogenicity, DS-SOSIP.4mut-stabilized trimers may have utility as HIV-1 immunogens or in other antigen-specific contexts, such as with B-cell probes.

Keywords: HIV-1; immunogen design; protein stabilization.

Copyright © 2017 American Society for Microbiology.

Figures

FIG 1

Structure-based stabilization of DS-SOSIP into its prefusion-closed conformation. (A) Amino acid mutational positions from 103 DS-SOSIP variants evaluated antigenically and mapped onto the structure of BG505 SOSIP.664 (PDB accession number 4TVP), with one protomer shown in rainbow colors (residues colored coded as defined in the legend) and the other two in gray ribbon representation. (B) Crystal structure of BG505 SOSIP.664 (PDB accession number 4TVP) shown in gray, with one gp120 protomer shown in green and another shown in salmon (central image). The gp120 domains are color coded as defined in the legend on the figure. Structural models of DS-SOSIP variants with the most favorable antigenic profiles (Table S3) are shown in the insets; for each variant, stabilizing mutations relative to DS-SOSIP are modeled in stick representation and labeled.

FIG 2

Properties of stabilized HIV-1 Env trimers. (A) Size exclusion chromatography (SEC) profiles of stabilized DS-SOSIP variants prior to V3 negative selection. (B) SDS-PAGE of stabilized DS-SOSIP variants prior to V3 negative selection. R, reducing conditions; NR, nonreducing conditions. (C) Expression yield of stabilized DS-SOSIP variants after SEC or SEC with V3 negative selection.

FIG 3

Antigenicity of stabilized Env trimer determined by Meso Scale Discovery (MSD). (A) Antigenicity of stabilized DS-SOSIP prior to V3 negative selection (top) or after V3 negative selection (bottom), assessed with broadly neutralizing antibodies (VRC01 and PGT145), weakly or nonneutralizing antibodies (F105, 447-52D, and 3074), and a negative control antibody (CR9114, an influenza virus antibody with no recognition of HIV-1 Env). AUC, area under the concentration-time curve. (B) Antigenicity of stabilized DS-SOSIP variants after V3 negative selection assessed on a panel of CD4-induced antibodies (17b and 48d, with and without soluble CD4), CD4-binding site antibodies (VRC01, VRC13, and b12), V2 apex-directed antibodies (PGT145 and CAP256-VRC26.25), glycan-V3 antibodies (PGT121, PGT128, and 2G12), weakly neutralizing V3-directed antibodies (447-52D, 3074, and 2557, with and without soluble CD4), and gp41-gp120 interface antibodies (PGT151, 35O22, and 8ANC195).

FIG 4

Negative-stain electron microscopy of stabilized Env trimers. Reference-free classification and averaging produced symmetrical propeller-like classes typical of the prefusion-closed conformation of the HIV-1 Env trimer, indicative of high homogeneity and correct folding/assembly of the proteins. Scale bar, 10 nm.

FIG 5

Binding of soluble CD4 to stabilized HIV-1 Env trimers as measured by SPR with single-cycle kinetics. (A) SPR curves. For BG505 SOSIP.664 and DS-SOSIP, the concentrations of CD4 injected were 180, 90, 45, 22.5, and 11.25 nM. For the DS-SOSIP mutants, the concentrations of CD4 injected were 500, 250, 125, 62.5, and 31.25 nM. (B) Dissociation constant (Kd), on-rate constant (ka), and off-rate constant (kd) values. Values in parentheses report standard errors from fitting data to a 1:1 Langmuir binding model.

FIG 6

Thermostability of stabilized HIV-1 Env trimers assessed by differential scanning calorimetry (DSC). At left are raw data from DSC shown in solid lines for the Env trimers identified by color according to the scheme shown on the right. Values for Tm, ΔT1/2 (width at half peak height), and enthalpy of unfolding from DSC (ΔH) are also shown. Cp, heat capacity at constant pressure.

FIG 7

Crystal structure of DS-SOSIP.4mut. (A) Crystal structure of DS-SOSIP.4mut Env trimer, as extracted from the crystallized ternary complex with antibodies 35O22 and PGT122 (11), with amino acids displayed in ribbon representation and glycans in stick representation. The 2Fo − Fc (observed and calculated structure factor amplitudes, respectively) electron density is displayed as a gray mesh at 1σ. (B) Structural details of the engineered disulfide in DS-SOSIP (I201C/A433C, shown in sticks). The crystal structure of BG505 SOSIP.664 (PDB accession number 4TVP) is shown in gray. DS-SOSIP.4mut residues with Cα deviations greater than 2 Å from BG505 SOSIP.664 are shown in magenta. Simulated annealing (SA)-omit and 2Fo − Fc electron density map around the disulfide bond region are also shown as gray mesh in the middle and the right panels, respectively. (C) Structural details of designed hydrophobic side chains (shown in sticks) for DS-SOSIP.4mut; in the left panel, the crystal structure of BG505 SOSIP.664 (4TVP) is provided in gray for comparison. The 2Fo − Fc electron density map of the four designed residues in the DS-SOSIP.4mut and BG505 SOSIP.664 (4TVP) structures are shown in the middle and right panels, respectively, as gray mesh.

FIG 8

Crystal structure of DS-SOSIP.6mut. (A) Crystal structure of DS-SOSIP.6mut Env trimer, as extracted from the crystallized ternary complex with antibodies 35O22 and PGT122 (11), with amino acids displayed in ribbon representation and glycans in stick representation. The 2Fo − Fc electron density at 1σ is displayed as a gray mesh. (B) Structural details of designed hydrophobic side chains (shown in sticks) for DS-SOSIP.6mut; in the left panel, the crystal structure of BG505 SOSIP.664 (PDB accession number 4TVP) is provided in gray for comparison. The 2Fo − Fc electron density map of the four designed residues in the DS-SOSIP.6mut and BG505 SOSIP.664 (4TVP) structures are shown in the middle and right panels, respectively, as gray mesh.

FIG 9

Immunogenicity of stabilized HIV-1 Env trimers. (A) Longitudinal V3 peptide-specific (left) and native trimer-specific (right) serum immunogenicity. Recip Dil, reciprocal dilution. (B) Week 30 serum neutralization (ID80) of autologous (left) and V3-sensitive tier-1 (right) pseudoviruses. (C) DS-SOSIP and its variants elicited reduced V3-sensitive tier-1 neutralization (left) while maintaining autologous neutralization compared to neutralizing activity of BG505 SOSIP.664 (*, P < 0.05, by a two-tailed Mann-Whitney test). (D) Autologous- to V3-sensitive tier-1 neutralization ratio of DS-SOSIP.4mut is greater than that of BG505 SOSIP.664 (*, P < 0.05, by Kruskal-Wallis test with a post hoc Dunn's multiple-comparison test).

FIG 10

Correlation between immunogenicity and thermostability. (A) Spearman correlation between autologous neutralization (ID80 from week 30, group geometric means) and enthalpy of unfolding. (B) Spearman correlation between V3-sensitive tier-1 neutralization and enthalpy of unfolding. (C) Spearman correlation between autologous-to-V3-sensitive tier-1 neutralization ratio and enthalpy of unfolding.