HIV-1 Envelope and MPER Antibody Structures in Lipid Assemblies

Abstract

Structural and functional studies of HIV envelope glycoprotein (Env) as a transmembrane protein have long been complicated by challenges associated with inherent flexibility of the molecule and the membrane-embedded hydrophobic regions. Here, we present approaches for incorporating full-length, wild-type HIV-1 Env, as well as C-terminally truncated and stabilized versions, into lipid assemblies, providing a modular platform for Env structural studies by single particle electron microscopy. We reconstitute a full-length Env clone into a nanodisc, complex it with a membrane-proximal external region (MPER) targeting antibody 10E8, and structurally define the full quaternary epitope of 10E8 consisting of lipid, MPER, and ectodomain contacts. By aligning this and other Env-MPER antibody complex reconstructions with the lipid bilayer, we observe evidence of Env tilting as part of the neutralization mechanism for MPER-targeting antibodies. We also adapt the platform toward vaccine design purposes by introducing stabilizing mutations that allow purification of unliganded Env with a peptidisc scaffold.

Keywords: HIV-1 Env; MPER broadly neutralizing antibody; bicelle; nanodisc; peptidisc.

Graphical abstract

Figure 1

Preparation of Env Lipid Assemblies (A) Overview of the workflow used to generate and analyze different lipid assemblies. (B) Typical size-exclusion chromatograms of different Env assemblies. 3D reconstructions below the chromatogram peaks illustrate the corresponding forms of the assembly. The lipid bilayer is highlighted in yellow. Incorporation of lipids was confirmed by mass spectrometry from a pool of AMC011 FL nanodiscs and bicelles. See also Figures S1, S2, and S6.

Figure 2

Comparison of Different Sample Preparation Approaches and Major Components of Sample Heterogeneity Stabilizing PGT151 Fab highlighted in blue. Different assembly types are presented with cartoons next to negative stain and cryo-EM 3D reconstructions. The Env ectodomain is highlighted in gray, PGT151 in blue, and the micelle and lipid bilayer in yellow. (A) The ectodomain could be reconstructed to 3–5 Å resolution from all lipid-detergent micelle samples. (B and C) When the whole nanodisc assembly was reconstructed from cryo-EM data, global resolution remained at ∼9 Å, but now the height from the bilayer surface could be measured and is indicated for PC64FL (B) and BG505ΔCT (C). Height was estimated by fitting an Env model to the ectodomain density and a modeled lipid bilayer patch in the EM density corresponding to the disc. The distance between the stable ectodomain ending at Asp664 and the closest atom of the bilayer is reported as the ectodomain height from the bilayer surface. Density that passes through the disc of PC64FL is highlighted in purple. (D) Discs larger than the 10 nm in diameter defined by the MSP1D1 scaffold were considered as bicelles. In these, additional compositional heterogeneity limits the analysis to negative-stain EM reconstructions. (E) Illustrative summary of the cumulative rotational, positional, and compositional heterogeneity indicated by the arrows and shading of Env in different assembly types. See also Figures S2, S3, and S6.

Figure 3

Cryo-EM Reconstructions of FL Env-MPER Fab Complexes in Detergent-Lipid Micelles (A) PC64FL and AMC011FL Envs with a panel in of MPER-targeting Fabs showing variable Fab positions and occupancies. The estimated membrane position is indicated as well as the height of the structurally stable part of ectodomain (ending at Asp664) from the membrane surface. Membrane surface position in the absence of the bilayer is estimated based on MPER Fab position. Particle classes with one, two, and three Fabs could be classified from most of the datasets with all showing similar flexibility and heterogenous density in the micelle. When the second or third MPER Fab is not visible, it is bound behind the micelle, pointing away from the viewer. (B) In the complex between PC64FL and VRC42.01 Fab, the MPER density could be followed through to the TMD as continuous density allowing docking of crystal structure of the Fab (PDB: 6MTO) and NMR structure of TMD helices (PDB: 6B3U). The position of R696 as the crossing point of the helices is indicated as well as residues R683 and R707 that are commonly positioned at the membrane boundaries. See also Figures S2, S4, and S5 and Table S1.

Figure 4

Analysis of FL Env in Lipid Bicelles and Nanodiscs in Complex with MPER-Targeting Fabs (A) Representative 2D class averages from negative-stain EM data from lipid assemblies of PC64FL and AMC011FL in complex with MPER-targeting antibodies. In all bilayer-assembled Env-MPER antibody complexes, Env was tilted at varying degrees and displaced to the edge of the bilayer. The degree of the Fab binding angle is estimated in relation to the bilayer from the given 2D class average. The lipid bilayer is highlighted in yellow, and MPER Fab is shown in red. (B) Comparison of the VRC42.01 Fab-binding angle in the nanodisc and micelle. The angle between the Fab and the ectodomain was identical, while in the nanodisc, an additional angle can be measured between the bilayer and the Fab. PGT151 Fab is highlighted in blue throughout the figure. (C) Low-pass filtered cryo-EM reconstructions of the AMC011FL nanodisc in complex with 10E8 Fab with different Fab occupancies. The highest resolution and particle count were obtained with the complex containing one copy of PGT151 Fab and three copies of 10E8 Fab, which is used for (E)–(I). (D) Low-pass-filtered reconstructions with one, two, or three copies of 10E8 Fab showing degree of tilting and distance of Env from bilayer in different 10E8 Fab occupancy states. (E) Highest resolution reconstruction and epitopes of the three Fabs with 10E8 Fab crystal structure (PDB: 5T80) docked in. Epitope components are highlighted as indicated in the panel below. The distance from the bilayer surface is also indicated. (F) Local resolution estimation showing up to 5 Å resolution in the ectodomain and stabilized 10E8 Fab epitope. (G) Fab A and Fab C stabilized TMD orientations are highlighted in purple. Residues marking the outer (R683) and inner (R707) bilayer surfaces are highlighted in orange and R696 in blue, which marks the crossing point of TMDs in micelle samples and is now separated. The angles of the two TMDs are estimated in relation to the bilayer surface. (H) Two Fab orientations and dependency of Fab C on the PGT151 position are highlighted in red in low-pass-filtered maps. (I) Comparison of 10E8 Fab docking into the AMC011FL nanodisc and JRFLΔCT micelle reconstructions (EMD-3312) showing an ∼10 Å change in the distance between the MPER peptides (residue Q135 in PDB: 5T80). See also Figures S2, S4, and S6 and Table S1.

Figure 5

Assembly of Vaccine Design Adapted BG505-ST-710 into Peptidiscs (A) Examples of BG505-ST-710 peptidiscs with and without the addition of MPER-targeting Fabs as determined from negative-stain EM 2D class averages. (B) Antigenic profiling of BG505-ST-710 peptidiscs compared to equivalent soluble Env (BG505-ST-664) by lectin capture-based biolayer interferometry. The top graph shows binding signals of the indicated Fabs after normalization to the respective PGT121 signal. Exemplary traces of the indicated Fabs are shown below. (C) Antigenicity comparison of AMC011FL in detergent micelles and lipid nanodiscs with protein A-IgG capture mode. See also Table S1.

Figure 6

Model for MPER-Targeting Neutralization Mechanism Steps Based on Lipid Bilayer Assemblies Suggested HR1 (blue), HR2 (orange), MPER (red), and TMD (purple) orientations are presented. Based on the lack of density corresponding to the TMD in our EM reconstructions, we hypothesize that in their ground state, TMDs are fluctuating in a loosely folded scissoring motion. Similarly, native Env spontaneously samples different functional states. (A) In the absence of an MPER-targeting Fab, (1) CD4 triggers stabilization and conformational changes in HR2, MPER and TMD in addition to opening of the ectodomain and exposure of the CCR5 coreceptor binding site. (2) This leads to coreceptor binding and formation of a lower energy state post-fusion conformation of fusion peptide (arrows), HR1, HR2, MPER, and TMD. Further downstream (not represented in this schematic), this extended three-helix bundle undergoes a further condensation into a six-helix bundle and, in concert with adjacent Env molecules, facilitates membrane fusion and viral entry into the host cell. (B) In the MPER-targeting neutralization path (1), approaching antibody forms a wedge between the ectodomain (red arrow) and bilayer surface (blue arrow), tilting the ectodomain, increasing the exposure of MPER peptide, and stabilizing the scissoring of MPER-TMD. While on a planar bilayer, the tilting may restrict access to additional MPER epitopes within the trimer, we observe at least a second MPER-binding event in all tested assemblies. (2) Thus, subsequent Fab arms binding to other protomers may eventually lead to displacement or increased shedding of gp120 in the absence of stabilizing mutations or PGT151 Fab. MPER-TMDs are now separated and locked by the MPER antibody.