Glycan modifications to the gp120 immunogens used in the RV144 vaccine trial improve binding to broadly neutralizing antibodies

Abstract

To date, the RV144 HIV vaccine trial has been the only study to show that immunization can confer protection from HIV infection. While encouraging, the modest 31.2% (P = 0.04) efficacy achieved in this study left significant room for improvement, and created an incentive to optimize the AIDSVAX B/E vaccine immunogens to increase the level of vaccine efficacy. Since the completion of the RV144 trial, our understanding of the antigenic structure of the HIV envelope protein, gp120, and of the specificity of broadly neutralizing monoclonal antibodies (bN-mAbs) that bind to it, has significantly improved. In particular, we have learned that multiple families of bN-mAbs require specific oligomannose glycans for binding. Both of the monomeric gp120 immunogens (MN- and A244-rgp120) in the AIDSVAX B/E vaccine used in the RV144 trial were enriched for glycans containing high levels of sialic acid, and lacked critical N-linked glycosylation sites required for binding by several families of bN-mAbs. The absence of these epitopes may have contributed to the low level of efficacy achieved in this study. In this report, we describe our efforts to improve the antigenic structure of the rgp120 immunogens used in the vaccine by optimizing glycan-dependent epitopes recognized by multiple bN-mAbs. Our results demonstrated that by shifting the location of one PNGS in A244-rgp120, and by adding two PNGS to MN-rgp120, in conjunction with the production of both proteins in a cell line that favors the incorporation of oligomannose glycans, we could significantly improve the binding by three major families of bN-mAbs. The immunogens described here represent a second generation of gp120-based vaccine immunogens that exhibit potential for use in RV144 follow-up studies.

Fig 1. Modification of N-linked glycosylation sites in MN- and A244-rgp120.

(A) The A244-rgp120 or MN-rgp120 sequences were analyzed for the presence of highly conserved glycans known to be important for bN-mAb binding within the C2-C3 domains. Glycosylation sites are represented as either black (present in RV144 immunogen) or grey (absent in original RV144 immunogen) structures. (B) A ribbon diagram depicts the 3-dimensional arrangement of the N289, N301, N332, and N334 PNGS. The structure is based on crystal structure of the BG505 SOSIP.664 gp140 trimer (in gold) bound to the PGT122 bN-mAb (in grey) [41]. The N301 and N332 glycan structures immobilized by the PGT122 antibody are indicated in green, while the asparagine residues at the base of relevant PNGS are indicated in red. (C) Site directed mutagenesis was used to create MN- or A244-rgp120 variants introducing one or more of the indicated PNGS. A summary of the PNGS variant constructs assayed is shown. The constructs with identical number and location of PNGS to the RV144 rgp120 immunogens are marked with an asterisk.

Fig 2. Endo H digest and immunoblot of A244 gp120 glycan variants.

A244-rgp120s containing either N332 or N334 based PNGS were expressed in either CHO-S (lanes 1–4) or HEK 293 GnTI^- cells (lanes 5–8) via transient transfection. Purified protein was subjected to Endo H or mock digest (digest buffer alone), and analyzed for mobility on 4–12% reducing SDS-PAGE gels. Immunoblots were probed with the mouse monoclonal 34.1 that binds a conformation independent epitope in the N-terminal gD tag of all expressed proteins, and visualized with goat-anti-mouse HRP conjugated polyclonal sera.

Fig 3. Binding of A244-rgp120 glycan variants to bN-mAbs.

Purified A244-rgp120 glycan variants were compared by FIA for binding to a panel of bN-mAbs. Results are reported as the half maximal effective concentration (EC50) in μg/mL, or the concentration of antibody required for a half-maximal binding, measured in Relative Fluorescence Units (RFU) on a titration-binding curve. Values are reported as ≥2.5μg/mL if titration curves did not plateau, or mean EC50 was calculated as ≥2.5μg/mL. Each curve represents the average of four independent assays. The EC50 values of the rgp120 glycan variants that are significantly different (p<0.05) from the A244_GNE EC50 to the same bN-mAb are highlighted in bold. Human IgGK polyclonal antibody was used as a negative control, and binding curves to purified rabbit polyclonal antibody (PB94) raised against rgp120 were used as a coating control.

Fig 4. Endo H digest and immunoblot of MN gp120 glycan variants.

Wildtype MN rgp120 expressed in CHO GnTI^- cells, or MN glycoform mutants with single, double, or triple glycan additions were expressed in GnTI^- cells via transient transfection. Purified MN_GNE or GnTI^- cells supernatants containing expressed rgp120 were immunoprecipitated via a monoclonal antibody to an N-terminal gD tag bound to protein-G coated beads. Immunoprecipiated rgp120 variants were subjected to Endo H or mock digest (digest buffer alone), and analyzed for mobility on 4–12% reducing SDS-PAGE gels. Immunoblots were probed with the mouse monoclonal 34.1 and visualized with goat-anti-mouse HRP-conjugated polyclonal sera.

Fig 5. Screen of MN glycan mutant supernatants for improvements to bN-mAb binding.

A FIA was used to identify 293 GnTI^- expressed MN-rgp120 glycan variants exhibiting improved bN-mAb binding profiles as compared to the wildtype MN sequence expressed in GnTI^- (MN358). Recombinant gp120s were expressed in GnTI^- 293 cells via transient transfection, and transfection supernatants were normalized to contain ~2μg/mL. MN-rgp120 variants were captured onto 96 well black plates using a 1μg/mL concentration of mouse monoclonal antibody to an N-terminal gD tag. Binding curves to the VRCO1 bN-mAb, which binds a conformation dependent epitope in the CD4 binding site, were used to assay for maintenance of overall secondary and tertiary structure. All screening assays were performed in duplicate. MN-rgp120 glycan variants were assayed for improved antigenicity to a panel of glycan dependent bN-mAbs to be considered for further analysis.

Fig 6. Binding of MN glycan variants to extended panel of bN-mAbs.

The MN-rgp120 variants MN358 and MN1320 expressed in GnTI^- cells were compared to MN_GNE for improved binding to an array of bN-mAbs. Recombinant gp120s were expressed in GnTI^- 293 cells via transient transfection. Purified MN_GNE or transfection supernatants were normalized to contain 4μg/mL rgp120 and captured using 2μg/mL of mouse monoclonal antibody 34.1, then assayed by FIA for bN-mAb binding. Results are reported in μg/mL as (EC50), the concentration of antibody required for a half-maximal binding, measured in Relative Fluorescence Units (RFU) on a titration-binding curve. Values are reported as ≥2.5μg/mL if titration curves did not plateau or if mean EC50 was ≥2.5μg/mL. Binding curves to bN-mAbs were performed in quadruplicate. The rp120 constructs exhibiting statistically significant differences in EC50 values (p<0.05) from the MN_GNE are noted in bold. Human IgGK polyclonal antibody was used as a negative control and purified goat polyclonal antibody raised against rgp120 (PB94) was used as a coating control.