Structure-based design of native-like HIV-1 envelope trimers to silence non-neutralizing epitopes and eliminate CD4 binding

Abstract

Elicitation of broadly neutralizing antibodies (bnAbs) is a primary HIV vaccine goal. Native-like trimers mimicking virion-associated spikes present nearly all bnAb epitopes and are therefore promising vaccine antigens. However, first generation native-like trimers expose epitopes for non-neutralizing antibodies (non-nAbs), which may hinder bnAb induction. We here employ computational and structure-guided design to develop improved native-like trimers that reduce exposure of non-nAb epitopes in the V3-loop and trimer base, minimize both CD4 reactivity and CD4-induced non-nAb epitope exposure, and increase thermal stability while maintaining bnAb antigenicity. In rabbit immunizations with native-like trimers of the 327c isolate, improved trimers suppress elicitation of V3-directed and tier-1 neutralizing antibodies and induce robust autologous tier-2 neutralization, unlike a first-generation trimer. The improved native-like trimers from diverse HIV isolates, and the design methods, have promise to assist in the development of a HIV vaccine.

Fig. 1

Structure-based design of BG505 Olio6 trimer. a Structural model of BG505 SOSIP.664 trimer (PDBID: 4TVP), with two subunits shown as a grey surface, one subunit shown as green trace. The V3 loop is shown as a thick red trace and the six substituted positions are shown as slate spheres. The enlarged frames show the six substituted positions in sticks, in green are the BG505 amino acids and in cyan are the corresponding Olio6 amino acids. b Binding affinities of bnAbs as measured by SPR are shown for BG505 SOSIP.664, BG505 MD39, and BG505 Olio6. The K _Ds for V1V2 apex bnAbs were measured with trimer analyte and IgG ligand, but for other bnAbs Fab analyte and trimer ligand were used. c bnAb EC₅₀ values were measured by ELISA for stabilized trimers. When an EC₅₀ could not be determined, the symbol was placed at the highest concentration tested (20 μg/mL). d Binding of non-nAbs to trimers measured by SPR. Trimers were analytes at 1 μM. The SPR signal ratio is the observed response normalized to the expected response of one gp120 per antigen-binding site. e Binding of V3 non-nAbs to trimers by ELISA. BG505 SOSIP.664 v4.1 is reported in de Taeye et al. 2015 . The Area Under the Curve has units of Abs 450 nm×log(dilution). Measurements are done in duplicate and represent two independent experiments

Fig. 2

Design and evaluation of the CD4-KO mutation. a The difference in computed Rosetta energy between trimers with glycine at position 473 and all 20 amino acids except cysteine at position 473 in three states: unbound, VRC01-bound, and CD4-bound. Energies are colored from white to red, with energies similar to glycine in white and unfavorable energies in red. b SPR binding of trimers to VRC01-IgG and CD4-IgG. IgGs were ligands and trimers were analytes. c Binding of different trimers to CD4⁺ T cells. Fluorescently labeled Env trimer probes were mixed with human PBMC. All panels gated on CD4⁺ T cells. Blue oval gate indicated probe positive population. A ‘no probe’ staining was used as a negative control to set the lower bound of the positive gate. d Frequency of Env trimer probe positive CD4⁺ T cells. N = 3 individual human donor samples for each probe. The data are representative of two experiments with a total of four independent samples

Fig. 3

Circular permutation and glycan-masking of BG505 MD39 trimer. a gp140 trimer gene layout for standard native-like trimers (labeled gp140) and for circular permuted trimers (labeled gp140-cp). The trimer is shown in surface representation, subunits are colored green, blue and light pink. A model of the designed loop is shown in orange spheres. b A structural model of BG505 MD39 GRSF6. The trimer is colored as in a. Native glycans are shown in red spheres, and designed glycans are shown in yellow spheres. Each designed glycan is labeled in HXB2 numbering c Negative-stain EM 3D reconstruction of the base-binding monoclonal 12N Fab (blue) in complex with BG505 SOSIPv4.1 (grey). d The ELISA data for monoclonal rabbit antibody, 12N, binding to trimer variants. e ELISA assay format for testing VRC01 gH mouse sera. f ELISA AUC data for sera from six VRC01 gH mice binding to various trimers. ELISA antigens labeled on the x-axis and mouse identifiers in the legend. The Area Under the Curve has units of absorbance×log(dilution). g ELISA assay format for testing rhesus macaque sera. h The ELISA AUC data for sera from six BG505 Olio6-immunized rhesus macaques binding to various trimers. ELISA antigen is BG505 MD39, the competitors are listed on the x-axis, and the monkey identifiers are given in the legend. All trimers were made in 293F cells, unless otherwise noted. Measurements are done in duplicate and represent two independent experiments

Fig. 4

Thermostability of BG505 variants SOSIP, MD37 and MD64. a Differential Scanning Calorimetry (DSC) profile for BG505 SOSIP, the raw data are shown as a solid line and fit is shown as a dashed line. Melting Temperature (T _m) values from the fit are shown. b DSC profile for BG505 MD37 c DSC profile for BG505 MD64. The data are from one experiment

Fig. 5

Structural and antigenic assessment of AD8 trimer with different stabilization designs. a 2D class averages from negative-stain EM analysis of AD8 SOSIP trimer and AD8 MD64 trimer, classified as closed native-like, open native-like, or non-native. Red boxes indicate examples of open native-like trimers. b ELISA-binding curves for bnAbs and non-nAbs against AD8 SOSIP and the AD8 MD64 trimer. ELISA antigens are labeled on top of the plots, and the antibodies are labeled in the legend. c SPR signal ratio for non-nAbs binding to AD8 MD64 trimers. The data are from one experiment and measurements are done in duplicate

Fig. 6

Structural and antigenic assessment of 327c trimer with different stabilization designs. a 2D class averages from negative-stain EM analysis of 327c SOSIP, 327c MD37 and 327c MD39 trimers, which are classified as closed-native-like, open native-like or non-native. Red boxes indicate examples of open native-like trimers. b ELISA-binding curves for bnAbs and non-nAbs against 327c SOSIP, 327c MD37 and 327c MD39 trimers c SPR signal ratio for the V3 non-nAbs binding to 327c SOSIP, 327c MD37 and 327c MD39 trimers. The data are from one experiment and measurements are done in duplicate

Fig. 7

Rabbit immunization with 327c native-like trimers. a 327c trimer binding titers at week 0, week 8, week 10, week 24, week 26, week 32, and week 34 time points for rabbits immunized with 327c SOSIP, 327c MD39, and 327c MD37. b V3 binding titers at week 0, week 10, week 26, and week 34 time points for rabbits immunized with 327c SOSIP, 327c MD39, and 327c MD37. The V3 peptides correspond to the V3 peptide sequence found in the matching trimer immunogen. Stars indicate the significance of differences between WT and respective stabilized designs. c Tier-1 SF162 neutralization titers at week 26 and week 34. d Autologous 327c Tier-2 neutralization titers at week 26 and week 34. Group size N = 5. The data are from one experiment. Statistical comparisons were calculated using two-tailed Mann–Whitney U tests. * indicates p < 0.05, ** indicates p < 0.01