Design and crystal structure of a native-like HIV-1 envelope trimer that engages multiple broadly neutralizing antibody precursors in vivo

Abstract

Induction of broadly neutralizing antibodies (bNAbs) by HIV-1 envelope glycoprotein immunogens would be a major advance toward an effective vaccine. A critical step in this process is the activation of naive B cells expressing germline (gl) antibody precursors that have the potential to evolve into bNAbs. Here, we reengineered the BG505 SOSIP.664 glycoprotein to engage gl precursors of bNAbs that target either the trimer apex or the CD4-binding site. The resulting BG505 SOSIP.v4.1-GT1 trimer binds multiple bNAb gl precursors in vitro. Immunization experiments in knock-in mice expressing gl-VRC01 or gl-PGT121 show that this trimer activates B cells in vivo, resulting in the secretion of specific antibodies into the sera. A crystal structure of the gl-targeting trimer at 3.2-Å resolution in complex with neutralizing antibodies 35O22 and 9H+109L reveals a native-like conformation and the successful incorporation of design features associated with binding of multiple gl-bNAb precursors.

Figure 1.

Design and biophysical properties of a germline-targeting SOSIP trimer. (A) Schematic of the BG505 SOSIP.v.4.1-GT1 construct (also referred to as GT1 trimer). The constant (C1–C5) and variable (V1–V5) regions in gp120 and the HR1 and HR2 regions in gp41 are indicated. The SOSIP mutations as well as the added N332 PNGS are shown in red. The E64K and A316W stabilizing mutations introduced to the SOSIP.664 construct to create SOSIP.v4.1 are indicated in blue. The mutations then introduced to SOSIP.v4.1 to induce gl-bNAb binding are indicated in green. The approximate position of a seven amino acid deletion is indicated with magenta arrows and a white dashed line. The glycan composition is adapted from Behrens et al. (2016). (B) Overview of the 18 changes introduced to BG505 SOSIP.664 to obtain SOSIP.v4.1-GT1. (C) NS-EM analyses of the GT1 trimer purified by PGT145. The 2D class averages are shown. On the basis of loop movement, compactness, and angles between individual protomers, the trimers are classified as closed native-like, partially open native-like, or nonnative (Pugach et al., 2015). The proportion of each class is indicated. (D) DSC analysis of the GT1 trimer purified with PGT145. The T_m value is indicated. (E) Glycan profiles of PGT145-purified trimer variants as determined by hydrophilic interaction liquid chromatography–ultraperformance liquid chromatography. The percentages of Man_5-9GlcNAc₂ glycans (M5-M9; shown in green), as a proportion of the total glycan population, are listed in Table S6.

Figure 2.

Antigenicity of the BG505 SOSIP.v4.1-GT1 trimer with a panel of bNAbs and gl-bNAbs. (A) Binding of bNAbs and gl-bNAbs to different SOSIP trimers was assessed by capture ELISA. Half maximal binding concentrations (EC₅₀ [μg/ml]) are shown, and ranges in nanograms per milliliter are color coded. (B) Representative binding SPR curves of the binding of PG16 and VRC01 mature and germline versions to SOSIP.v4.1 and SOSIP.v4.1-GT1. The sensorgrams show the response (RU) over time (seconds). The association phase is 300 s, and the dissociation is followed over 600 s. Curves for concentration ranges (see inset) are shown in color with the modeled fits in black overlaid with the corresponding dissociation constant (K_d = K_d1 for the monovalent initial interaction; see Table S7). SPR experiments were performed at least three times independently.

Figure 3.

Overall architecture of BG505 SOSIP.v4.1-GT1-N137A at 3.2 Å. (A) Left: Side view of the trimer in complex with 9H+109L and 35O22 Fabs. The three protomers of the trimer complex are depicted as a surface representation (front left), spheres (back), and ribbon (front right), the latter with gp41 (in cyan) and gp120 (in blue). Each Env protomer (blue) is associated with one Fab 9H+109L (brown) and one Fab 35O22 (yellow). Glycans are shown in red sticks. Right: Structural alignment of one protomer (ribbon) from BG505 SOSIP.v4.1-GT1-N137A at 3.2 Å (gp120 in blue and gp41 in cyan) superimposed on one protomer of BG505 SOSIP.664-N137A (gray; PDB accession no. 5CEZ). The root-mean-square deviation (RMSD) is indicated, and N-linked glycans are shown and numbered by their respective Asn residues. (B) Ribbon representation of one protomer illustrating the mutations introduced to SOSIP.v4.1-GT1 to improve stability and enhance gl-bNAb-binding. (C) Zoomed-in 180° views of the apex region. Mutations are indicated with arrows, and the side chains are represented as sticks. The asterisk indicates the location of the truncated V2 loop after the seven amino acid deletion. (D) Detailed view of the V3 showing the A316W substitution with a 2Fo-Fc electron density map contoured at 1.0σ.

Figure 4.

Structural mechanism of germline engagement. (A) Model of the interaction between PG9 HCDR3 (red) and V1V2 epitope of BG505 SOSIP.v4.1-GT1-N137A (blue). Relevant amino acid positions are indicated, and side chains are shown as blue sticks for Env and red sticks for HCDR3. Predicted interactions (<4 Å) between side chains are indicated with black dashed lines. (B) Model of interaction between the light chain of gl-VRC01 (green) and the loop D of SOSIP.v4.1-GT1-N137A (blue). Predicted interactions (<4 Å) between side chains are indicated with black dashed lines. The D462 residue, a substitution made to delete a possible obstructing glycan, is shown in the background (surface red). An intraprotomer H-bond (<4 Å) between T455 and S471 is indicated with black dashed lines. (C) Interaction between gl-VRC01 and the CD4bs, modeled in two different views. The positions of three PNGS in BG505 are indicated with spheres and their likely clashes with gl-VRCO1 light chain by red explosion shapes.

Figure 5.

BG505 SOSIP.v4.1-GT1 initiates antibody responses in knock-in mice expressing the predicted germlines of VRC01 and PGT121 bNAbs. (A) Calcium flux in B cells expressing either gl-VRC01 (left) or VRC01 (right) as a B cell receptor, stimulated with the indicated trimers at a 1 µM final concentration. In this and subsequent panels, all of the trimers used as immunogens and ELISA antigens were of the BG505 genotype. (B) Endpoint antibody binding titers in sera from gl_H-VRC01 mice immunized thrice with SOSIP.v4.1-GT1 (left), SOSIP.v4.1 (center), or SOSIP.664 (right) trimers, measured against the indicated His-tagged BG505 trimer variants by ELISA. The median titers are indicated by the black lines. Statistically significant differences are indicated by asterisks (*, P ≤ 0.05; **, P ≤ 0.01; Wilcoxon matched-pairs signed rank test). The ELISA curves can be found in Fig. S5 A. (C) Antibody specificity determinations. For each mouse, the area under the curve for a given gl-VRC01 epitope knockout trimer (SOSIP.v4.1-D368R or SOSIP.v4.1-GT1-D368R) was subtracted from the area under the curve obtained with the corresponding unmodified trimer (i.e., SOSIP.v4.1 or SOSIP.v4.1-GT1). The resulting “area above knockout (KO)” values are plotted as bars. The ELISA curves used for the area-under-the-curve analyses can be found in Fig. S5 A. The method has been described before. The mean and SEM are indicated. Statistically significant differences are indicated by asterisks (*, P ≤ 0.05; Wilcoxon matched-pairs signed rank test). (D) Endpoint antibody binding titers in sera from gl_H-PGT121 mice immunized thrice with either SOSIP.v4.1-GT1 (left) or SOSIP.v4.1 (right) trimers, measured against the indicated His-tagged BG505 trimer variants by ELISA. The median titers are indicated by the black lines. Statistically significant differences are indicated by asterisks (*, P ≤ 0.05; **, P ≤ 0.01; Wilcoxon matched-pairs signed rank test). The ELISA curves can be found in Fig. S5 B. (E) Antibody specificity determinations. For each mouse, the area under the curve for the PGT121 epitope knockout trimer (SOSIP.v4.1-GT1-N137A/N332A/N301A/H330A) was subtracted from the area under the curve obtained with the unmodified trimer (i.e., SOSIP.v4.1-GT1). The resulting area above KO values are plotted as bars. The ELISA curves used for the area-under-the-curve analyses can be found in Fig. S5 B. The mean and SEM are indicated. Statistically significant differences are indicated by asterisks (*, P ≤ 0.05; Wilcoxon matched-pairs signed rank test).