Electron-Microscopy-Based Epitope Mapping Defines Specificities of Polyclonal Antibodies Elicited during HIV-1 BG505 Envelope Trimer Immunization

Abstract

Characterizing polyclonal antibody responses via currently available methods is inherently complex and difficult. Mapping epitopes in an immune response is typically incomplete, which creates a barrier to fully understanding the humoral response to antigens and hinders rational vaccine design efforts. Here, we describe a method of characterizing polyclonal responses by using electron microscopy, and we applied this method to the immunization of rabbits with an HIV-1 envelope glycoprotein vaccine candidate, BG505 SOSIP.664. We detected known epitopes within the polyclonal sera and revealed how antibody responses evolved during the prime-boosting strategy to ultimately result in a neutralizing antibody response. We uncovered previously unidentified epitopes, including an epitope proximal to one recognized by human broadly neutralizing antibodies as well as potentially distracting non-neutralizing epitopes. Our method provides an efficient and semiquantitative map of epitopes that are targeted in a polyclonal antibody response and should be of widespread utility in vaccine and infection studies.

Keywords: BG505; Env; HIV; SOSIP; antibodies; antibody epitope mapping; cryo-EM; electron microscopy; negative-stain EM; polyclonal antibodies; vaccine.

Graphical abstract

Figure 1

Epitope Mapping of the Antibody Response to BG505 SOSIP.664 Trimers in Rabbits by nsEM Defines Different Antibody Classes (A) Immunization schedule used for the four rabbits analyzed in this study. (B) Representative reference-free 2D class averages obtained after each BG505 SOSIP.664 immunization of rabbit 3417. Fabs bound to BG505 SOSIP.664 Env trimers (white) are highlighted with false coloring: red, bottom of trimer (BOT); blue, glycan hole (GH); orange, cleft of trimer (COT). (C) 3D reconstructions of the four basic antibody classes elicited by BG505 SOSIP.664 immunization of rabbits. Refined 3D models were fitted onto a low-pass filtered Env trimer reference structure (PDB: 5I8H; displayed as ribbons with gp120 in bright blue and gp41 in dark gray). For display of the surface, the density map for fully glycosylated BG505 SOSIP.664 (EMD: 5782) was also fitted with Chimera and displayed in semitransparent gray. Side and top views are shown for representative 3D reconstructions and 2D class averages. Densities for mAbs 10A and PGT151 are added as semitransparent references in the GH2 and COT 3D reconstructions, respectively. BOT-, GH-, and COT-specific antibodies are highlighted with red, blue, and orange, respectively. Representative 2D class averages are shown below each view. A comparison of the two GH classes and variations of the BOT epitope recognition is depicted in a box on the right. See also Figures S1 and S2.

Figure 2

Epitope Mapping of the Antibody Responses in the Four Rabbits at Different Time Points during the Immunization Schedule Reveals that Neutralization Correlates with Appearance of GH1 Class Antibodies Refined 3D models were fitted onto a low-pass filtered Env trimer reference structure (PDB: 5I8H; displayed as ribbons with gp120 in bright blue and gp41 in dark gray). Densities corresponding to Fabs were separated and colored. For display of the surface, the density map for fully glycosylated BG505 SOSIP.664 (EMD: 5782) was aligned as described above and rendered in semitransparent gray. Side and top views are displayed. The bright-green wedge (between the side and top views) illustrates the development of autologous neutralizing titers, correlating with GH1 class antibodies, in two of the rabbits at different time points. See also Figure S1.

Figure 3

Generation of a Standard Curve for Measuring Fab Occupancy in Immune Complexes (A) Elution volumes of Fabs purified from the indicated mAbs. (B and C) Elution volumes of immune complexes saturated with a single Fab specificity (B) or combined Fab specificities (C) for determination of a standard curve for the calculation of Fab occupancy in immune complexes. Multiple dots represent different Fab-BG505 SOSIP.664 combinations with the same stoichiometry of binding. Note that complexes containing PGT151 and 35O22 were excluded from the right panel because of their aberrant elution behavior, which in the case of 35O22 only could be explained by a prolonged retention time of the Fabs themselves. See also Table S1.

Figure 4

Analysis of 241 and 289 GH Binding Shows the Evolution of the Immune Response and Higher Affinities of GH-Specific Antibodies Than of Non-GH Antibodies Fab occupancy of immune complexes formed with rabbit polyclonal Fabs and BG505 SOSIP.664 with or without the GH was compared. (A) Factors used for normalization to rabbit 3417 PB1 EC₅₀ were determined by ELISA. (B) Occupancy of GH-specific (white) and non-GH-specific (black) antibodies in immune complexes normalized to PB1-EC₅₀ of rabbit 3417. (C) Comparison of relative affinities of GH-specific (white) and non-GH-specific (black) antibodies determined by measurement of Fab occupancy in complexes formed with titrated amounts of rabbit 3417 Fabs. See also Figure S2.

Figure 5

Analysis of a Polyclonal Immune Complex Structure Obtained by cryoEM with Sub-nanometer Resolution (A) Side and bottom views of two representative 3D reconstructions for immune complexes with Fabs originating from PB1 of rabbit 3417. EM densities are depicted in beige (Env), blue (GH binding Fabs), and red (BOT binding Fabs). The crystal-structure coordinates of BG505 SOSIP (PDB: 5I8H) were fitted into the EM densities and are depicted as a backbone in beige for gp120 and in gray for gp41. The 10A light chains are colored cyan, and heavy chains are colored green (cf. B–D). (B) Crystal structure of rabbit mAb 10A (PDB: 6CJK). The long LCDR3 extends away from the surface of the paratope. (C) Close-up views of a high-resolution cryoEM map of BG505 SOSIP.664 in complex with polyclonal Fabs. The cryoEM BG505 SOSIP.664 structure (PDB: 5ACO) and crystal structure of Fab 10A were fitted onto the map. (D) Close-up view of the epitope-paratope. The long LCDR3 makes the majority of contacts with Env in a lysine-rich loop directly above the S241 GH residue. Position N289, whose glycosylation would interfere with antibody binding (but is not glycosylated in BG505 SOSIP.664), is indicated in the structure. The glycan present at N88 is represented by sticks and exhibits density in the cryoEM map that interacts with the light chain. See also Figure S3 and Table S2.

Figure 6

Semiquantitative Analysis of Epitope Occupancy in cryoEM 3D Reconstructions of Immune Complexes between BG505 SOSIP.664 and Fabs Originating from Rabbit 3417 PB1 (A) Examples of different occupancies for the indicated GH epitope. For normalizing thresholds of the individual 3D reconstructions, all density maps were overlaid with a reference structure for BG505 SOSIP.664 (EMD: 8312) and adjusted in volume threshold until the Env volumes were identical to those of the reference structure. Occupancy was then estimated, starting from the GH1 antibody present in all structures (referred to as GH₁), in a clockwise direction. (B) Number of particles displaying the indicated full occupancy. (C) Number of particles containing full (darker colors), medium, and partial (lighter colors) occupancies for each of the indicated antibody-binding sites. See also Figure S3.