Extended Data Fig. 1: Study design and binding ELISA titres.

a, Structural model of the RBD-16GS-I53-50 (RBD–NP) immunogen. The genetic linker connecting the RBD antigen to the I53-50A trimer is expected to be flexible and thus the RBD may adopt alternate orientations to that shown. b, Negative-stain electron microscopy of RBD–NP. Scale bar, 100 nm. c, Dynamic light scattering (DLS) of RBD–NP and unmodified I53-50 lacking displayed antigen. The data indicate the presence of monodisperse nanoparticles with size distributions centred around 36 nm for RBD–NP and 30 nm for I53-50. In b and c, the samples were analysed following a single freeze/thaw cycle. d, Antigenic characterization by biolayer interferometry (BLI). RBD–NP was bound to immobilized CR3022 monoclonal antibody and maACE2-Fc receptor, both before and after one freeze/thaw cycle. Monomeric SARS-CoV-2 RBD was used as a reference antigen. e, Schematic representation of the study design. f, Serum concentrations of anti-spike IgG and anti-I53-50 NP IgG (anti-I53-50) in individual non-human primates detected by ELISA at day 42. Boxes show median, 25th and 75th percentiles and the whiskers show the range. The statistical difference between anti-spike and anti-I53-50 IgG response was determined using two-sided Wilcoxon matched-pairs signed-rank test. g, Spearman’s correlation between anti-spike IgG (described in Fig. 1) and anti-NP IgG responses at day 42. The error bands represent 95% confidence limits. Each symbol represents an animal. N = 4 for O/W, 10 for AS03 and 5 for all other groups.