Structural basis of transmembrane coupling of the HIV-1 envelope glycoprotein

Abstract

The prefusion conformation of HIV-1 envelope protein (Env) is recognized by most broadly neutralizing antibodies (bnAbs). Studies showed that alterations of its membrane-related components, including the transmembrane domain (TMD) and cytoplasmic tail (CT), can reshape the antigenic structure of the Env ectodomain. Using nuclear magnetic resonance (NMR) spectroscopy, we determine the structure of an Env segment encompassing the TMD and a large portion of the CT in bicelles. The structure reveals that the CT folds into amphipathic helices that wrap around the C-terminal end of the TMD, thereby forming a support baseplate for the rest of Env. NMR dynamics measurements provide evidences of dynamic coupling across the TMD between the ectodomain and CT. Pseudovirus-based neutralization assays suggest that CT-TMD interaction preferentially affects antigenic structure near the apex of the Env trimer. These results explain why the CT can modulate the Env antigenic properties and may facilitate HIV-1 Env-based vaccine design.

Fig. 1. Structures of TMD–CT^LLP2 and MPER–TMD–CT^LLP2 trimers in bicelles.

a Ribbon representation of the TMD–CT^LLP2 average structure from the calculated ensemble. The unstructured KS (residues 711–736) is omitted for clarity. The H1 and H2 helices forming the inner and outer rings of the CT baseplate, respectively, are indicated in the bottom view (right). b Close-up view of the residues establishing the CT–TMD interactions. Hydrophobic and hydrophilic interactions are shaded in yellow and light blue, respectively. The TMD and the CT are shown as green and pink ribbons, respectively. c Same as (b) but for the CT–CT interactions. The palmitoylation site C764 (S764 in our construct) faces the lipid bilayer interior. d Ribbon representation of the merged MPER–TMD–CT^LLP2 model showing the MPER, TMD, and CT^LLP2 in cyan, green, and pink, respectively. The placement of the structure in the lipid bilayer was determined experimentally using the PPT method. The conserved intramembrane R696 is represented as spheres. e Fit of the MPER–TMD–CT^LLP2 model and the structure of the SOSIP Env trimer (yellow; pdb ID: 5T3Z) into the low-resolution EM density (gray) of the HIV-1 Env trimer on the virion surface by cryo-electron tomography (Env trimer EMDB ID: EMD-5019; viral membrane EMDB ID: EMD-5020).

Fig. 2. Independent evidence of CT^LLP2–TMD interaction.

a Schematic illustration of the sample preparation strategy used for intermolecular PRE analysis. TMD–KS and CT^LLP2 (C789) are expressed separately: TMD–KS is isotopically-enriched for NMR readout while CT^LLP2 carries the spin-label (at C789). After purification, the two segments are mixed at ~1:1 molar ratio and co-reconstituted in bicelles. Upon interaction, the TMD–KS is expected to experience PRE generated by the CT^LLP2 spin-label. b Residue-specific PRE (I/I₀) of (¹⁵N, 85% ²H)-labeled TMD–KS mixed with MTSL-labeled CT^LLP2 (left). Error bars represent the uncertainty derived from cross-peaks signal to noise. Missing bars are due to prolines (indicated by gray triangles) or overlapping residues. The horizontal dash lines mark the four PRE regimes used to map the PREs onto the protein structure (right). The TMD–KS and the CT^LLP2 are shown as white ribbons and blue cylinders, respectively. c Labeling scheme for probing intermolecular CT^LLP2–CT^LLP2 interaction. CT^LLP2 (white) is isotopically-enriched for NMR readout while CT^LLP2 (blue) carries the spin-label at C789, and TMD–KS (green) serves as scaffold. After purification, the three proteins are mixed at ~1:2:3 molar ratio, respectively, and co-reconstituted in bicelles. d Residue-specific PRE (I/I₀) of (¹⁵N, 85% ²H)-labeled CT^LLP2 mixed with MTSL-labeled CT^LLP2 and scaffold TMD–KS (left). Error bars represent the uncertainty derived from cross-peaks signal to noise. Missing bars are due to overlapping residues. The horizontal dash lines mark the four PRE regimes used to map the PREs onto the protein structure (right). Source data are provided as a Source data file.

Fig. 3. Destabilization of the TMD–CT interactions loosens the CT baseplate.

a Comparison of the 2D ¹H–¹⁵N TROSY-HSQC spectra of the TMD–CT^LLP2 (left) and CT2-tmd (right), showing strong agreement among the two constructs of the chemical shift of residues 679–710 (TMD), 711–725 (KS), and 770–788 (H2) highlighted in green, orange, and pink, respectively. Bearing the mutation site, H1 exhibited strong chemical shift changes (black cross-peaks) and thus was not further analyzed. b Comparison of residue-specific PRE (I/I₀) of TMD–CT^LLP2 (red) versus those of CT2-tmd (blue). Each sample was prepared mixing ~1:1 (¹⁵N, 85% ²H)-labeled protein (NMR readout) with unlabeled (and thus “NMR-invisible”) protein carrying the MTSL spin-label at position S764C, following the strategy summarized in Fig. 2a (see the Methods section for details). Error bars represent the uncertainty derived from cross-peaks signal to noise. Missing bars are due to prolines (indicated by gray triangles) or overlapping residues. The position of the paramagnetic tag is marked by a star. Source data are provided as a Source data file.

Fig. 4. Effect of mutations in the CT on Env antibody sensitivity.

a Antibody neutralization of pseudovirus containing the 92UG037.8 Env determined for non-neutralizing antibodies, including b6 (CD4 binding site; blue), 3791 (V3; cyan), and 17b (CD4-induced; purple), and trimer-specific bnAbs, including PG9 (orange), PG16 (red), and PGT145 (magenta). The CD4 binding site bnAb VRC01, used as a control antibody, is shown in green. b Antibody neutralization of pseudovirus containing the L702S-A756N-D759R mutant. Right panel shows the relative decrease of sensitivity to the trimer-specific bnAbs of double and triple mutants bearing the D759R. c Antibody neutralization of pseudovirus (upper panel) containing a larger number of mutations in the CT (lower panel). d Same as (c) but for the mutant with mutations on the opposite side of the CT–TMD interface. e Same as (c) but for the mutant containing TMD mutations that break the hydrophobic and hydrophilic TMD–CT interactions. Source data are provided as a Source Data file.

Fig. 5. Conformational coupling between the Env CT and ectodomain.

a Backbone conformational dynamics of the unlocked MPER–TMD (red), locked MPER–TMD (green) and TMD–CT^LLP2 (black) probed by the product of ¹⁵N R₁ and R₂ relaxation rates, which reflects ms–μs motion. Error bars represent the uncertainty derived from R₁ and R₂ fitting errors. The regions characterized by higher R₁R₂, i.e., the MPER, the C-terminal part of the TMD, and its interacting CT region (H1), are highlighted in cyan, green, and purple, respectively. The position of the hydrophobic core (or the TM hinge) is shown in gray. The yellow box highlights the decrease of R₁R₂ in the C-terminal part of the TMD due to locking the MPER or CT presence. b Proposed mechanism of CT-ectodomain coupling mediated by the TMD. For simplicity, only two chains of the trimer are illustrated. Left: the CT baseplate confines the TMD motion, important for stabilizing the Env trimer in the prefusion state. Right: deletion of the CT baseplate or disruption of CT–TMD interactions allows greater TMD hinge motion, destabilizing the MPER, which in turn modulates the antigenic structure of the ectodomain. Source data are provided as a Source data file.