Designed mosaic nanoparticles enhance cross-reactive immune responses in mice

Abstract

Nanoparticle vaccines displaying combinations of SARS-like betacoronavirus (sarbecovirus) receptor-binding domains (RBDs) could protect against SARS-CoV-2 variants and spillover of zoonotic sarbecoviruses into humans. Using a computational approach, we designed variants of SARS-CoV-2 RBDs and selected 7 natural sarbecovirus RBDs, each predicted to fold properly and abrogate antibody responses to variable epitopes. RBDs were attached to 60-mer nanoparticles to make immunogens displaying two (mosaic-2_COMs), five (mosaic-5_COM), or seven (mosaic-7_COM) different RBDs for comparisons with mosaic-8b, which elicited cross-reactive antibodies and protected animals from sarbecovirus challenges. Naive and COVID-19 pre-vaccinated mice immunized with mosaic-7_COM elicited antibodies targeting conserved RBD epitopes, and their sera exhibited higher binding and neutralization titers against sarbecoviruses than mosaic-8b. Mosaic-2_COMs and mosaic-5_COM elicited higher antibody potencies against some SARS-CoV-2 variants than mosaic-7_COM. However, mosaic-7_COM elicited more potent responses against zoonotic sarbecoviruses and highly mutated Omicrons, supporting its use to protect against SARS-CoV-2 variants and zoonotic sarbecoviruses.

Keywords: RBD; SARS-CoV-2; antibody; computational methods; nanoparticle; protein design; sarbecovirus; vaccination.

Graphical abstract

Figure 1

Overview of the design process (A) Structures of representative class 1 (C102, PDB: 7K8M), class 2 (C144, PDB: 7K90), class 3 (S309, PDB: 7JMX), and class 4 (CR3022, PDB: 6W41) antibodies bound to the WA1 SARS-CoV-2 RBD, and the structure of the WA1 RBD (PDB: 6W41) colored based on conservation scores calculated using the ConSurf database. (B) Overview of mosaic-2_COM and mosaic-5_COM RBD-NP designs. Starting from the WA1 RBD, computational analysis and machine learning models were used to calculate properties of potential RBD immunogens based on expression, antibody binding, and solubility. A set of selected RBDs was further filtered based on expression and binding measurements and used to construct the mosaic-2_COM and mosaic-5_COM RBD NPs. (C) Overview of designing mosaic-7_COM. A set of 8 RBDs was selected from naturally occurring zoonotic sarbecovirus RBDs to maximize (1) sequence diversity and (2) binding to class 3 and 4 but not class 1 and 2 RBD epitopes (RBD epitopes defined as described). The 8 selected RBDs were further filtered based on experimentally determined properties (see text), and the 7 remaining RBDs were used for mosaic-7_COM.

Figure S1

Illustration of how sequences for class 1 escape mutations were generated, related to Figure 2 From the 20 RBD positions with the highest escapes against class 1 anti-RBD antibodies, we generated all 38,760 possible combinations of 6 positions. For each combination, we further generated all possible ways to divide the 6 positions into 2 groups of 3, of which there were 10 possible divisions, resulting in 387,600 sets of positions. For each set, 1 group of 3 positions was assigned to RBD1 (colored as light blue or pink), and the other group of 3 positions was assigned to RBD2 (colored as dark blue or red). RBD1 and RBD2 would be mutated in their assigned positions to the amino acids listed in Table S2.

Figure 2

Overview of computational methods (A) Architecture of the neural network used to predict RBD expression. The input is an expression matrix, which is the element-wise product (multiplication of entries at the same positions) of the one-hot encoded sequence (each residue is represented as a 20D vector with entries of 1 for the matching amino acid and 0 for other amino acids) and the matrix of single-mutation expression changes. This is processed through a convolutional neural network to produce the predicted change in expression as an output. (B) ∼800,000 possible RBD sequences are screened for predicted expression relative to the WA1 RBD using a threshold value of −0.2 logMFI. Rejected RBD pairs are in blue, and selected pairs are in red. (C) ∼100,000 RBD sequences that passed predicted expression screening and were further screened for solubility based on a change in aggregation score relative to WA1 calculated using Aggrescan. Rejected RBD pairs are in blue, and selected pairs are in red. (D) The distribution of total mutational entropy over sets of 10 RBDs, and the set selected for experimental testing is the one with maximum entropy indicated by the red line. (E) Mean escape against class 1 and 2 anti-RBD antibodies and the mean escape against class 3 and 4 anti-RBD antibodies for naturally occurring sarbecoviruses. Rejected RBDs are in blue, and selected RBDs are in red. See also Figures S1 and S2.

Figure S2

The total escape against class 1 and 2 antibodies for all ∼90,000 RBD pairs that pass screening, related to Figure 2 The total escape is obtained from DMS experiments^,,,, and antibody RBD-epitope classes are defined in Barnes et al.

Figure 3

Designed SARS-CoV-2 RBDs and sarbecovirus RBDs exhibit desired properties (A and B) HiLoad 16/600 Superdex 200 SEC profiles of designed RBDs (A) and sarbecovirus RBDs (B). RBD3 and RBD8 exhibited suboptimal expression, indicated by no signal for an RBD monomer (RBD3) or a peak in the void volume (RBD8). (C and D) Fold reduction of selected monoclonal anti-RBD antibodies (mAbs) or a human ACE2-Fc construct (hACE2) to designed SARS-CoV-2 RBDs (C) and sarbecovirus RBDs (D) compared with binding to WA1 RBD. (E) Superose 6 increase 10/300 SEC profiles after SpyTagged RBDs were conjugated to SpyCatcher-mi3 showing peaks for RBD NPs and free RBDs. (F) SDS-PAGE for each RBD-NP after pooling appropriate SEC fractions. See also Figure S3.

Figure 4

Computationally designed mosaic RBD NPs elicit cross-reactive antibody binding and pseudovirus neutralization responses in immunized mice The mean of mean titers is compared in (B) and (C) by Tukey’s multiple comparison test with the Geisser-Greenhouse correction calculated using GraphPad Prism, with pairings by viral strain. Error bars associated with each data point are standard deviations. Significant differences between immunized groups linked by horizontal lines are indicated by asterisks: ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001, ^∗∗∗∗p < 0.0001. (A) Left: schematic of immunization regimen. Middle: numbers and colors used for sarbecovirus strains within clades throughout the figure. Right: colors and symbols (squares) used to identify immunizations (colors) and matched (filled in) versus mismatched (not filled in) viral strains. (B) Left: ELISA binding titers at day 56 for serum IgG binding to RBDs, represented as mean ED₅₀ values. Middle left: means of ELISA binding titers for each immunization. Middle right: means of ELISA binding titers for each immunization against only SARS-CoV-2 variant RBDs. Right: means of ELISA binding titers for each immunization against zoonotic sarbecovirus RBDs. Each circle represents the mean serum IgG binding titer against matched (solid circles) and mismatched (open circles) RBDs. (C) Left: ELISA binding titers at day 84 for serum IgG binding to RBDs, represented as mean ED₅₀ values. Middle left: means of ELISA binding titers for each immunization. Middle right: means of ELISA binding titers for each immunization against only SARS-CoV-2 variant RBDs. Right: means of ELISA binding titers for each immunization against zoonotic sarbecovirus RBDs. Each circle represents the mean serum IgG binding titer against matched (solid circles) and mismatched (open circles) RBDs. (D) Left: neutralization titers at day 84 for serum IgG neutralization of pseudoviruses derived from the virus strains in (A), represented as mean ID₅₀ values. Middle left: means of all neutralization titers for each immunization. Each circle represents the mean neutralization titer against matched (Khosta-2 for mosaic-7_COM; solid circle) and mismatched (open circles) pseudoviruses. Middle right and right: neutralization titers against XBB.1.5 and Khosta-2. Each circle represents a neutralization titer from an individual mouse serum sample.

Figure S3

RBD amino acid sequence identities for computationally designed mosaic RBD NPs and mosaic-8b, related to Figure 3 Asterisks indicate strains used for mosaic RBD NPs.

Figure 5

Differences in epitope targeting of antibodies elicited in mice immunized with mosaic and homotypic RBD NPs (A) DMS line plots for analyses of sera from mice that were immunized as shown in Figure 4A. DMS was conducted using a SARS-CoV-2 Beta RBD library. The x axis shows RBD residue positions, and the y axis shows the total sum of Ab escape for all mutations at a given site, with larger values indicating greater Ab escape. Each faint line represents a single antiserum with heavy lines indicating the average of n = 4 sera for each group. Lines are colored differently based on RBD epitopes from the 4 major classes (color definitions are shown in the legend below this panel and gray for residues not assigned to an epitope). (B) Mean site-total antibody escape for a SARS-CoV-2 Beta RBD library determined using sera from mice immunized with the indicated immunogens mapped to the surface of the WA1 RBD (PDB: 6M0J). White indicates no escape, and dark pink indicates sites with the most escape (residue numbers are denoted with epitope-specific colors as denoted by the legend between A and B).

Figure 6

Mosaic-7_COM immunization in pre-vaccinated mice elicits superior cross-reactive antibody responses The mean of mean titers is compared in (C) and (E) by Tukey’s multiple comparison test with the Geisser-Greenhouse correction calculated using GraphPad Prism, with pairings by viral strain. Error bars associated with each data point are standard deviations. Significant differences between immunized groups linked by horizontal lines are indicated by asterisks: ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001, ^∗∗∗∗p < 0.0001. Binding responses at day 0 (before NP or other vaccine immunizations) showed significant differences across cohorts in titers elicited by the pre-vaccinations. To account for different mean responses at day 0 between cohorts, we applied baseline corrections (see STAR Methods). Uncorrected binding data for (B) and (C) are shown in Figures S4B and S4C. (A) Left: schematic of vaccination regimen. Mice were pre-vaccinated with mRNA-LNP encoding WA1 spike and bivalent WA1/BA.5 prior to prime and boost immunizations with RBD NPs at day 0 and day 28 or an additional WA1/BA.5 mRNA-LNP immunization at day 0. Middle: colors and symbols (squares) used to identify immunizations (colors) and matched (filled in), mismatched (not filled in), or matched to pre-vaccination (half-filled in) viral strains (squares). Right: numbers and colors used for sarbecovirus strains within clades throughout the figure. (B) Log₁₀ mean fold change in ELISA ED₅₀ binding titers from day 0 at the indicated days after priming with the indicated immunogens against spike or RBD proteins from the indicated sarbecovirus strains (numbers and color coding as in A). (C) Log₁₀ means of fold change in ELISA titers for each type of immunization at the indicated days. Each circle represents the log₁₀ mean fold change in ED₅₀ titers from mice against a single viral strain of sera from mice that were immunized with the specified immunogen (solid circles = matched; open circles = mismatched; colors for different strains defined in A). (D) Mean change in neutralization ID₅₀ titers from day 0 at the indicated days against the indicated sarbecovirus strains (numbers and color coding as in A). (E) Means of all neutralization titers for each type of immunization at the indicated days. Each circle represents the mean neutralization IC₅₀ titer against a single viral strain of sera from mice that were immunized with the specified immunogen (solid circles = matched; open circles = mismatched; colors for different strains defined in A). See also Figure S4.

Figure S4

Mosaic-7_COM immunization in pre-vaccinated mice elicited superior cross-reactive antibody responses, related to Figure 6 The mean of mean titers is compared in (C) and (E) by Tukey’s multiple comparison test with the Geisser-Greenhouse correction calculated using GraphPad Prism, with pairings by viral strain. Error bars associated with each data point are standard deviations. Significant differences between immunized groups linked by horizontal lines are indicated by asterisks: ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001, ^∗∗∗∗p < 0.0001. Binding responses at day 0 (before NP or other vaccine immunizations) showed significant differences across cohorts in titers elicited by the pre-vaccinations. To account for different mean responses at day 0 between cohorts, we applied baseline corrections in Figure 6 (see STAR Methods). Here, binding data are shown as both baseline corrected (B and C) and not baseline corrected (D and E). (A) Left: schematic of vaccination regimen. Mice were pre-vaccinated with mRNA-LNP encoding WA1 spike and bivalent WA1/BA.5 prior to prime and boost immunizations with RBD NPs at day 0 and day 28 or an additional WA1/BA.5 mRNA-LNP immunization at day 0. Middle: colors and symbols (squares) used to identify immunizations (colors) and matched (filled in), mismatched (not filled in), or matched to pre-vaccination (half-filled in) viral strains (squares). Right: numbers and colors used for sarbecovirus strains within clades throughout the figure. (B) ELISA ED₅₀ binding titers in serum samples from mice immunized with the indicated immunogens measured at days 0, 28, and 56 against spike or RBD proteins from the indicated sarbecovirus strains (numbers and color coding as in A). (C) Mean ELISA titers for each type of immunization at the indicated days. Each circle represents the mean ED₅₀ titers from mice against a single viral strain of sera from mice that were immunized with the specified immunogen (solid circles = matched; open circles = mismatched; colors for different strains defined in A). (D) Mean fold change in ELISA ED₅₀ binding titers from day 0 in serum samples from mice immunized with the indicated immunogens measured at days 0, 28, and 56 against spike or RBD proteins from the indicated sarbecovirus strains (numbers and color coding as in A). (E) Means of fold changes in ELISA titers for each type of immunization at the indicated days. Each circle represents the mean fold change in ED₅₀ titers from mice against a single viral strain of sera from mice that were immunized with the specified immunogen (solid circles = matched; open circles = mismatched; colors for different strains defined in A).