The SARS-CoV-2 spike reversibly samples an open-trimer conformation exposing novel epitopes

Abstract

Current COVID-19 vaccines and many clinical diagnostics are based on the structure and function of the SARS-CoV-2 spike ectodomain. Using hydrogen-deuterium exchange monitored by mass spectrometry, we have uncovered that, in addition to the prefusion structure determined by cryo-electron microscopy, this protein adopts an alternative conformation that interconverts slowly with the canonical prefusion structure. This new conformation-an open trimer-contains easily accessible receptor-binding domains. It exposes the conserved trimer interface buried in the prefusion conformation, thus exposing potential epitopes for pan-coronavirus antibody and ligand recognition. The population of this state and kinetics of interconversion are modulated by temperature, receptor binding, antibody binding, and sequence variants observed in the natural population. Knowledge of the structure and populations of this conformation will help improve existing diagnostics, therapeutics, and vaccines.

Figure 1.. SARS-CoV-2 spike ectodomain and Hydrogen-Deuterium Exchange monitored by mass spectrometry (HDX-MS) experimental overview.

(A) Schematic of the prefusion-stabilized SARS-CoV-2 spike protein and a model of the trimeric prefusion conformation (24). (B) Schematic of HDX-MS experiment and the resulting mass distributions for a peptide that exists in either one (left) or two (right) separable conformations. In order for the two conformations to result in a bimodal mass distribution, they must not interconvert during the timescale of the HDX experiment (hours). Rapid interconversion would result in a single mass distribution with the ensemble averaged mass profile.

Figure 2.. Peptide-level HDX is consistent with the known prefusion conformation.

Percent deuteration after one minute of deuterium labeling for every peptide in the S-2P continuous exchange dataset for the S1 (top) and S2 (bottom) domains. Each line represents an individual peptide spanning the residues indicated on the x-axis, with percent deuteration after one minute of exchange indicated on the y-axis (for bimodal peptides, only the less-exchanged centroid shown). Secondary structures in the prefusion conformation are shaded in blue (alpha helices) and green (beta strands). A measure of solvent accessibility is shown above in Ų (calculated as a three-residue sliding average using the default get_area function in pymol) using a model of the full-length prefusion trimer with all three RBDs in a down position (24). These data are consistent with the SARS-CoV-2 spike trimer secondary structures, notably regions buried in the trimer interface, such as the central helix, show increased protection relative to more exposed regions lacking secondary structure. Important sequence features are indicated above the plot including the N-terminal domain (NTD - green), receptor binding domain (RBD - blue), fusion peptide (FP - cyan), heptad repeat 1 (HR1 - yellow), central helix (CH - orange), core domain (CD - purple) and heptad repeat 2 (HR2 - yellow). Locations of glycans are noted with stars with three categories - glycans detected in at least one peptide of our data set (black, 9/22), glycans known to be on the spike protein where we lack coverage (white, 12/22), and glycans known to be on the spike protein but for which non-glycosylated peptides are observed (pink, 1/22).

Figure 3.. The Spike ectodomain reversibly samples two conformations.

(A) Left: SARS-CoV-2 spike monomer with all regions that have peptides showing bimodal mass distributions colored in blue. Right: Example mass spectra from two peptides after one minute of deuteration (top: residues 982–1001, bottom: residues 878–903) with overlaid fitted gaussian distributions that describe each protein conformation in blue (light blue: less exchanged A state, dark blue: the more exchanged B state). (B) Conformational preference for the S-2P spike construct at 25 °C, 4 °C and 37 °C monitored by pulsed-labeling. At 25 °C S-2P converts from primarily state A to ~50:50 A:B after 4 days. At 4 °C, S-2P prefers state B while at 37 °C, S-2P prefers state A. (C) The kinetics of interconversion between the A and B states for different of spike variants. Starting from an initial prefusion conformation (state A, 37 °C), samples were rapidly transferred to 4 °C and assayed for conversion to state B over time using pulsed-labeling HDX-MS. To estimate fraction state A, peptides from two different regions (residues 982–1001 (circles) and residues 878–903 (triangles)) were fit to two gaussians. Data from both regions were used to determine the rate of interconversion.

Figure 4:. ACE2-binding effects are similar on isolated RBD and RBD in the context of the full-length ectodomain.

(A) Diagram of the spike structure with regions of interest highlighted. (B) Left: Heatmap showing the difference in RBD peptide deuteration (from continuous-exchange HDX-MS) in the presence and absence of ACE2 on the isolated RBD. Middle: The deuterium uptake plots are illustrated for three peptides of interest with error bars representing the standard deviation of three replicates (some are smaller than marker size). The uptake plot for the RBM (residues 487– 510) is shown for both the isolated RBD and HexaPro. Right: Schematic representation of the heatmap data on the structure of the RBD•ACE2 complex (PDB 6M0J). The structure of the RBD is colored based on the maximum change shown in the heatmap for that residue in any peptide. (C) Changes to peptides from HexaPro upon binding of ACE2 outside of the RBD during continuous-exchange HDX-MS. When ACE2 binds the canonical prefusion structure, state A, peptide 982–1001 (Region II) loses inter-subunit contacts with the RBD and thus exchanges faster, but when ACE2 binds the open trimer (state B) it does not, presumably because it is already maximally exposed. For peptide 878–903 (Region III), there is no change in the exchange rate to either state A or state B indicating this region is not affected by ACE2 binding. In the schematic for region II, one NTD has been removed to visualize the peptide of interest. (D). Time course of interconversion in the presence of ACE2. Top: Pulsed-labeling HDX-MS example spectra of S-2P peptide 878–902 with and without ACE2 before and after 24 hours of incubation at 25 °C. Bottom: time vs fraction state A for peptide 878–902 in S-2P with and without ACE2 over 24 hours monitored by pulsed-labeling. After 24 hours ACE2 bound S-2P prefers state B.

Figure 5:. Comparison of HDX on RBD in isolation versus in S-2P

Left: Heat map showing the difference in peptide deuteration for isolated RBD compared to the RBD in S-2P. Bottom: Selected uptake plots of isolated RBD and S-2P RBD. Right: Structure of the RBD (model of a single RBD taken from a full-length spike trimer model from (24)) colored based on the maximum change shown in the heat map for that residue in any peptide. For reference, spheres are shown denoting the beginning and end of the peptides displayed in the uptake plots.

Figure 6:. The antibody 3A3 binds selectively to state B in the 978–1001 region.

(A) Example mass spectra for HexaPro with and without 3A3 for two different peptides that have bimodal mass distributions. The bottom peptide (878–902) shows no change in the presence of 3A3, which indicates that the amount of state A and state B has not significantly changed 13 minutes after adding 3A3. The top peptide (978–1001), however, shows significant protection in the presence of 3A3, shifting the distribution belonging to state B to a deuteration amount indistinguishable from state A. These data are the three-minute time point from a continuous-exchange HDX-MS time course. (B) The kinetics of interconversion of S-2P in the presence of 3A3 monitored by pulsed-labeling HDX-MS. The addition of 3A3 accelerated the rate of conversion to state B at 4 °C. The binding of 3A3 prevents the return to state A at 37 °C. Dotted lines indicate the conversion in the absence of 3A3.

Figure 7.. Schematic of the energy landscape for the SARS-CoV-2 Spike ectodomain.

Reaction coordinate illustrating the different conformations accessible by Spike. Three different conformational states are depicted: the canonical prefusion ensemble, the expanded open trimer, and the postfusion conformation. The prefusion conformation contains all four RBD states (0,1,2, or 3 up). The relative energies and barrier heights as well as the placement of the open trimer along the reaction coordinate are drawn for illustration only.