Thermostability of Well-Ordered HIV Spikes Correlates with the Elicitation of Autologous Tier 2 Neutralizing Antibodies

Abstract

In the context of HIV vaccine design and development, HIV-1 spike mimetics displaying a range of stabilities were evaluated to determine whether more stable, well-ordered trimers would more efficiently elicit neutralizing antibodies. To begin, in vitro analysis of trimers derived from the cysteine-stabilized SOSIP platform or the uncleaved, covalently linked NFL platform were evaluated. These native-like trimers, derived from HIV subtypes A, B, and C, displayed a range of thermostabilities, and were "stress-tested" at varying temperatures as a prelude to in vivo immunogenicity. Analysis was performed both in the absence and in the presence of two different adjuvants. Since partial trimer degradation was detected at 37°C before or after formulation with adjuvant, we sought to remedy such an undesirable outcome. Cross-linking (fixing) of the well-ordered trimers with glutaraldehyde increased overall thermostability, maintenance of well-ordered trimer integrity without or with adjuvant, and increased resistance to solid phase-associated trimer unfolding. Immunization of unfixed and fixed well-ordered trimers into animals revealed that the elicited tier 2 autologous neutralizing activity correlated with overall trimer thermostability, or melting temperature (Tm). Glutaraldehyde fixation also led to higher tier 2 autologous neutralization titers. These results link retention of trimer quaternary packing with elicitation of tier 2 autologous neutralizing activity, providing important insights for HIV-1 vaccine design.

Fig 1. Schematic of stability test for well-ordered trimers.

Representative soluble trimers from clade A, B and C in the NFL or SOSIP format were cross-linked with glutaraldehyde and/or treated under selected conditions (i.e. 4°C or 37°C and/or in adjuvants) for up to 7 days. Trimer stability was assessed by antibody binding profile and BN-PAGE using samples taken over the time course. Guinea pigs were immunized with selected trimers to access potential correlates between trimer stability and immunogenicity.

Fig 2. Stability of trimers in selected conditions.

(A) Schematic showing how relative binding of the different mAbs to the trimers by BLI was determined. The binding response (nm) of the trimers from samples taken at day 1, 3, 5, and 7 was compared to the binding level at day 0 (100%) to calculate the relative binding (%) for each day. (B) Stability of BG505 NFL and JRFL NFL at 4°C and 37°C in PBS. Binding of VRC01, PGT145 and F105 was analyzed by BLI using anti-Fc sensors to capture the mAbs with the trimers as the analyte (top). Relative binding of VRC01 and PGT145 through the time course is plotted on left Y-axis. F105 binding level at each day was plotted on the right Y-axis (Response, nm). Stability of BG505 NFL and JRFL NFL at 37°C was also assessed by BN-PAGE shown below (trimer, red arrow; monomer, blue arrow). (C) Stability of BG505 NFL in ISCOMATRIX adjuvant and in Adjuplex at 37°C. The experiments were performed two times independently.

Fig 3. Characterization of well-ordered trimers before and after fixation with glutaraldehyde.

(A) Comparison of trimers before (Wt) and after fixation (X-link) by BN- and SDS-PAGE. Bands corresponding to the trimer (red arrow), dimer (orange arrow), and monomer (blue arrow) are indicated. (B) DSC chromatograph of BG505 NFL before (Wt, blue) and after fixation (X-link, black). (C) EM 3D reconstructions comparing Wt (left) and X-link (middle) BG505 NFL trimer in the unliganded form (upper panel; Wt, EMD-8270; X-link, EMD-8271) or bound to VRC01 (lower panel; Wt, EMD-8269; X-link, EMD-8268). On the right, the Wt BG505 NFL EM density (solid gray) is superimposed on the X-link BG505 NFL trimer EM density (mesh orange). Top and side views are presented with the high-resolution cryo-EM BG505 SOSIP structure fitted within (PDB ID 3J5M; gp120 in blue, V1V2 in magenta, V3 in green, and gp41 in brown; VRC01, PDB ID 4S1Q, in red). The gel and DSC experiments were performed at least two times independently.

Fig 4. Antigenicity of Wt and X-link trimers.

(A) Antigenic profile of Wt (left) and X-link (right) of JRFL NFL by ELISA using His-capture of the trimer (top) or with the trimer directly coated onto the plate (bottom). Binding of the trimer-preferring antibodies PGT145 (pink) and PG9 (orange), the broadly neutralizing CD4bs mAb, VRC01, and the non-neutralizing mAb, F105 are shown. The assay was performed two times independently. (B) Potential lysine residues (red) that can be cross-linked by glutaraldehyde are shown on the surface of BG505 SOSIP (PDB 4ZMJ) as an example.

Fig 5. Stability of trimers after fixation in selected conditions.

(A) Stability of JRFL NFL Wt (hollow symbol) and after glutaraldehyde fixation (X-link, solid symbol) in PBS, ISCOMATRIX adjuvant and Adjuplex at 37°C was detected by BLI (top) and BN-PAGE (bottom). The red arrow indicates the trimer and the blue arrow marks the monomer band on the gels. The experiments were performed two times independently. (B) Table summarizing the stability of NFL and SOSIP trimers at 37°C in selected conditions at day 7. Stable (+++) is defined as 80–100% relative binding of PGT145 and VRC01 with no increase in F105 binding, no monomer band and retains 80–100% trimer by BN-PAGE. Partially stable (++) is defined as 30–80% relative binding of PGT145 and VRC01 with an increase of F105 binding along with a visible monomer band and decreased trimer (30–80% remaining) by BN-PAGE. Unstable (–) is defined as < 30% relative binding of PGT145 and VRC01 with increased F105 binding coupled with a marked decrease (<30% remaining) of the trimer band by BN-PAGE.

Fig 6. Immunogenicity of well-ordered trimers with or without fixation.

Guinea pigs (n = 6) were immunized at week 0, 4, 12, and 24 with Wt JRFL, 16055, or BG505 SOSIP compared to JRFL or BG505 NFL with or without fixation (X-link). (A) Geometric mean IgG titers (GMT, ± SD; n = 6) are shown, as measured by His-capture ELISA to the autologous trimer immunogen (listed on the right by increasing T_m). Sera were collected prior to inoculation (gray dash line) and at 2 weeks after. Assays were performed on two independent occasions with similar results. (B) Neutralizing ID₅₀ titers (serum fold dilution factor) of sera collected two weeks after the fourth Env inoculation (wk 26) in guinea pigs (n = 6) against a panel of tier 1 viruses using the TZM-bl pseudovirus assay. The analysis was performed at multiple dilutions of each serum sample, and the data are representative of two independent experiments for most viruses.

Fig 7. Autologous tier 2 neutralization correlates with trimer stability.

(A) Autologous neutralization (ID₅₀ titers, serum fold dilution factor) against tier 2 viruses elicited by JRFL, 16055, and BG505 well-ordered trimers, which are listed on top from left to right in increasing stability, as measured by T_m. Neutralizing activity was measured in sera collected two weeks after the fourth Env (wk 26) inoculation in guinea pigs (n = 6) using the TZM-bl pseudovirus assay with multiple dilutions of each serum sample. The data are representative of at least three independent experiments. (B) Correlation analysis of stability and immunogenicity. Trimer T_m values were plotted against the serum neutralization titer (geometric mean) of tier 1 viruses and tier 2 autologous viruses. The P value of T_m vs tier 2 autologous neutralization (right) is 0.0238 by Spearman’s rank correlation analysis. No significant correlation was found between T_m and neutralization against tier 1 viruses.

Fig 8. Increased trimer stability relates to improved elicitation of autologous neutralization.

(A) Comparison of longitudinal tier 2 autologous neutralization elicited by Wt or X-link NFL trimers in guinea pigs (n = 6), as shown by box plots of neutralizing ID₅₀ titers determined by the TZM-bl pseudovirus assay; horizontal line, median; box, interquartile range; whiskers, min/max. Left panel, JRFL autologous neutralization, *, P = 0.0152 after the third immunization, **, P = 0.0022 after the fourth immunization. Right panel, BG505 autologous neutralization, *, P = 0.0152 after the fourth immunization. Significance was evaluated by Mann-Whitney. The experiment was performed once using multiple dilutions of each serum sample. (B) Model depicting how native-like trimer stability, in particular when cross-linked (lower panel; black horizontal lines), presents limited conformations to the immune system during affinity maturation (block arrows) to generate a greater frequency of neutralizing antibodies (blue) that cross react with the functional spike on the virus. In contrast, less stable trimers (top panel) with lower T_m values (i.e. more flexible) may unfold more rapidly in adjuvant or in vivo, leading to the elicitation of more diverse B cell responses during the GC reaction, resulting in many different antibody specificities (red and yellow) that do not recognize the functional spike.