SARS-CoV-2 Spike Protein Stabilized in the Closed State Induces Potent Neutralizing Responses

Abstract

The majority of SARS-CoV-2 vaccines in use or advanced development are based on the viral spike protein (S) as their immunogen. S is present on virions as prefusion trimers in which the receptor binding domain (RBD) is stochastically open or closed. Neutralizing antibodies have been described against both open and closed conformations. The long-term success of vaccination strategies depends upon inducing antibodies that provide long-lasting broad immunity against evolving SARS-CoV-2 strains. Here, we have assessed the results of immunization in a mouse model using an S protein trimer stabilized in the closed state to prevent full exposure of the receptor binding site and therefore interaction with the receptor. We compared this with other modified S protein constructs, including representatives used in current vaccines. We found that all trimeric S proteins induced a T cell response and long-lived, strongly neutralizing antibody responses against 2019 SARS-CoV-2 and variants of concern P.1 and B.1.351. Notably, the protein binding properties of sera induced by the closed spike differed from those induced by standard S protein constructs. Closed S proteins induced more potent neutralizing responses than expected based on the degree to which they inhibit interactions between the RBD and ACE2. These observations suggest that closed spikes recruit different, but equally potent, immune responses than open spikes and that this is likely to include neutralizing antibodies against conformational epitopes present in the closed conformation. We suggest that closed spikes, together with their improved stability and storage properties, may be a valuable component of refined, next-generation vaccines. IMPORTANCE Vaccines in use against SARS-CoV-2 induce immune responses against the spike protein. There is intense interest in whether the antibody response induced by vaccines will be robust against new variants, as well as in next-generation vaccines for use in previously infected or immunized individuals. We assessed the use as an immunogen of a spike protein engineered to be conformationally stabilized in the closed state where the receptor binding site is occluded. Despite occlusion of the receptor binding site, the spike induces potently neutralizing sera against multiple SARS-CoV-2 variants. Antibodies are raised against a different pattern of epitopes to those induced by other spike constructs, preferring conformational epitopes present in the closed conformation. Closed spikes, or mRNA vaccines based on their sequence, can be a valuable component of next-generation vaccines.

Keywords: SARS-CoV-2; glycoproteins; immunization; neutralizing antibodies.

FIG 1

Design of constructs. (A) Overview of constructs used in the study indicating the positions where residues have been mutated. (B) Overview of the trimeric spike structure indicating the positions of the mutated residues. The insertion of cysteine residues at positions 413 and 987 leads to the formation of a disulfide bond (dotted yellow line in inset) that stabilizes S in the closed prefusion form.

FIG 2

Characterization of fusogenicity and ACE2 binding. (A) Infection of HEK293T-hACE2 cells with SARS-CoV-2 S-pseudotyped HIV virions carrying a GFP reporter gene. The relative area of infected cells was quantified by GFP fluorescence 48 h postinfection using an Incucyte S3 live cell imager. Viruses were produced using either full-length S constructs or a C-terminal deletion of 19 amino acids (del 19) to increase infectivity. Infections were carried out using quantities of virus containing equivalent amounts of S. Virions pseudotyped with wild-type (WT) S are compared to those pseudotyped with S-R and S-R/x2. (B) Biolayer interferometry sensograms for binding kinetics of ACE2 to 600 nM RBD (magenta) and 1,000 nM S-R/PP (black), S-R (green), and S-R/x2 (red). The data are shown in gray with fits to the data in their respective colored lines. The dissociation constants (K_d) shown were calculated from panel C. (C) Concentration series sensograms with fits of the association (k_on) and dissociation (k_off) constants, where K_d = k_off/k_on. The data are summarized in Table 1.

FIG 3

Immunization strategy and proteins for immunization. (A) Overview of immunization strategy. Eight- to ten-week-old BALB/c females were immunized twice with a 4-week interval. Mice were bled 1 week prior to immunization (PB), and serial bleeds (SB) were taken at 2-week intervals after the first immunization, and again at 21 weeks, after which animals were sacrificed and spleens taken for T-cell assays. (B) Oligomeric state of S proteins before and after addition of adjuvant assessed by negative-stain EM. In all cases, the proteins are predominantly in the prefusion, trimeric state. Scale bar, 200 nm.

FIG 4

Neutralization by sera and T-cell response. (A) SARS-CoV-2 pseudovirus neutralization IC₅₀ values for individual mice at five bleed points postimmunization with one of four S protein variants. The bleed time points are 2, 4, 6, 8, and 21 weeks postprime. (B) SARS-CoV-2 pseudovirus neutralization IC₅₀ values for individual mice at 21 weeks postprime using pseudoviruses expressing original S (from 2019 SARS-CoV-2 lineage B) or from VOCs. (C) Box-and-whisker plots for endpoint antibody titers representing the serum dilution point at which the cytopathic effect in Vero cells caused by SARS-CoV-2 BetaCoV/Australia/VIC01/2020 infection was no longer inhibited, compared to controls. Endpoint titer means ranged between 12,800 and 25,600 for vaccine sera. (D) Comparison of persisting T-cell responses 21 weeks postvaccination with stabilized S constructs. T-cell responses were generated by all constructs with some heterogeneity: the SFU range from 0 to 88 for S-R/PP.13, to 79 for S-GSAS/PP, and to 31 to 424 for S-R/x2.

FIG 5

Inhibition by sera of RBD-ACE2 interaction. (A) Boxplot showing IC₅₀ values for the surrogate virus neutralization assay that measures the inhibition of RBD-ACE2 interactions by individual mouse sera. (B) Sera from S-R/x2-immunized mice were significantly more potent neutralizers of SARS-CoV-2 pseudovirus than expected based on their direct inhibition of RBD-ACE2 interaction. Predicted neutralizations are shown as a solid line; 95% confidence intervals are shown as dashed lines. S-R/x2 sera and predictions are shown in red, and all other sera are shown in black.

FIG 6

Quantification of binding of sets of three sera immunized with S-R/PP or S-R/x2 to an array of overlapping 15mer linear peptides covering S (see Table S1). (A) The binding strengths of different sera are shown as blue, green, and orange lines. The x axis is the residue number at the center of the peptide, and the y axis is the intensity of binding shown on a log scale. Different sera bind different patterns of peptides. The positions of the most common linear epitopes identified in human sera by Shrock et al. (38) are shown as pale green rectangles; the positions of two linear peptides that induce a neutralizing antibody response, as identified by Poh et al. (40), are shown as purple rectangles. Structural features are indicated: RBD, S1/S2 cleavage site, fusion peptide (RP), and heptad repeat 2 region (HR2). (B) The positions of the bound epitopes are illustrated on the structure of closed, trimeric S, showing the maximum binding value of the three sera on a log color scale at the center of each peptide (as in panel A) for S-R/PP or S-R/x2. The bound epitopes are almost entirely surface exposed.

FIG 7

Binding of sera to S protein antigens. (A) IC₅₀ values of sera from individual mice immunized with one of four trimeric S constructs binding to five separate antigens. The antigens are SARS-CoV-2 RBD and the four trimeric S constructs. Binding of sera from mice immunized with different S proteins, to different S antigens, was measured using the Luminex assay. All trimeric S constructs induce antibodies that bind all constructs and RBD. Sera from mice immunized with S-R/x2 bound the S-R/x2 antigen more than sera from other mice. (B) Visualization of correlation in IC₅₀ values between different antigens. Narrower, darker ellipses indicate stronger correlations. The three scatterplots on the right-hand side are illustrative examples showing the raw relationship for high, low, and intermediate correlations. There is generally a good correlation between the strengths of individual sera binding to different S antigens. This correlation is less strong for RBD and very different for S-R/x2, suggesting that S-R/x2 is displaying different epitopes in this assay than the other S proteins. Heating S-R/x2 leads to it displaying epitopes that are more similar to those of the other S proteins. (C) Sera from S-R/x2-immunized mice were significantly more potent neutralizers of SARS-CoV-2 pseudovirus than expected based on their spike binding IC₅₀. Predicted neutralization is shown as a solid line; 95% confidence intervals are shown as dashed lines. S-R/x2 sera and predictions are shown in red, all other sera are shown in black.