Influence of variant-specific mutations, temperature and pH on conformations of a large set of SARS-CoV-2 spike trimer vaccine antigen candidates

Abstract

SARS-CoV-2 subunit vaccines continue to be the focus of intense clinical development worldwide. Protein antigens in these vaccines most commonly consist of the spike ectodomain fused to a heterologous trimerization sequence, designed to mimic the compact, prefusion conformation of the spike on the virus surface. Since 2020, we have produced dozens of such constructs in CHO cells, consisting of spike variants with different mutations fused to different trimerization sequences. This set of constructs displayed notable conformational heterogeneity, with two distinct trimer species consistently detected by analytical size exclusion chromatography. A recent report showed that spike ectodomain fusion constructs can adopt an alternative trimer conformation consisting of loosely associated ectodomain protomers. Here, we applied multiple biophysical and immunological techniques to demonstrate that this alternative conformation is formed to a significant extent by several SARS-CoV-2 variant spike proteins. We have also examined the influence of temperature and pH, which can induce inter-conversion of the two forms. The substantial structural differences between these trimer types may impact their performance as vaccine antigens.

Figure 1

SARS-CoV-2 spike protein variants display variable levels of two trimer peaks by analytical size exclusion chromatography (SEC). (a) Example of an A₂₈₀ SEC profile of Alpha variant spike ectodomain fused to resistin (S(Alpha)-R) showing peaks with molar masses consistent with hexamers and two distinct trimer species. (b) Distribution of trimer and hexamer SEC peaks for a selection of spike variants in DPBS at pH 7.8.

Figure 2

A high-temperature melting transition by DSC correlates with the presence of trimer 2 species for different SARS-CoV-2 spike constructs. (a) Compared to the S(Ref)-F construct that exhibits a single, low-temperature melting transition, a high-temperature transition is detectable for S(Ref)-R and much more prominent for S(Delta)-R. (b) S(Delta)-noTD, expressed without a heterologous trimerization domain, is purified as a mix of monomeric and trimeric species and exhibits a high-temperature melting transition similar to trimeric species containing trimer 2. S(Ref)-noTD produced in the same way is monomeric and exhibits only the low-temperature transition.

Figure 3

HDX-MS Woods plots. Differential normalized HDX relative to S(Ref)-F (trimer 1) at a single representative HDX time point (60 min). Δ%D is plotted as a function of peptide ID (see Table HDX-S2 for a full list of peptides). S(Beta)-R and S(Delta)-R (green and red, respectively) are plotted in (a), while S(Hexa)-F and S(Ref)-noTD (blue and red, respectively) are plotted in (b). S(Ref)-R (black) is included in both plots as a reference. Data in (a) and (b) were collected in triplicate in two separate batches. Dashed lines represent ± 3× pooled SD for each unique state. Δ %D measurements outside the dashed lines demonstrate either significantly reduced exchange (− Δ %D) or increased exchange (+ Δ %D) based on a 1-p value of 0.98. Key structural domains are indicated above the Woods plots.

Figure 4

Projection of HDX onto 3-D structure. Differential normalized HDX of S(Hexa)-R (trimer 2) relative to S(Ref)-F (trimer 1) is projected onto a single protomer, where residues with significantly reduced or increased exchange in at least one labelling time point are shown in blue and red, respectively. The full spike trimer is inset as a reference (PDB 6VXX). Significance is based on a 1-p value of 0.98 and 3× pooled standard deviation.

Figure 5

Conformational variability of spike-resistin fusion constructs is reduced at low pH. UPLC-SEC (a), SV-AUC (b) and near-UV CD (c) all demonstrate that conformational variability observed at pH 7.8 is greatly reduced at pH 5.5. All data shown are for resistin fusion constructs.

Figure 6

Anti-spike S2 subunit V_HHs S2G4 and MRed22 bind preferentially to spike constructs containing trimer 2. (a) Single-concentration screening of V_HHs for binding to different spike preparations by ELISA. Microwell plates were coated with the indicated spike antigens binding to a single concentrations of different V_HH-Fcs was assessed. S2A3, S2G3, S2G4, MRed11, and MRed22 were characterized previously as S2-specific binders, while 11, 12 and SR01 bind different epitopes on the S1 subunit. MRed09 binds the resistin trimerization sequence. Shades of red indicate the absorbance measured by ELISA for the different conditions. (b) Dose–response analysis of V_HH 11 and S2G4 binding to spike preparations with varying proportions of trimer 1 and trimer 2. S2G4 binds with lower EC₅₀ to spike preparations consisting mostly of trimer 2 compared to preparations consisting mostly of trimer 1. In contrast, V_HH 11 binds similarly to both forms. ELISAs were performed with both dimeric, Fc-fused V_HHs (top panels) and monomeric, biotinylated V_HHs (bottom panels). EC₅₀s are averages of three technical replicates and error bars correspond to standard deviation.

Figure 7

Spike trimer models. Trimer conformations adopted by SARS-CoV-2 spike ectodomain-foldon and resistin fusions supported by the current study.