mRNA vaccines encoding membrane-anchored RBDs of SARS-CoV-2 mutants induce strong humoral responses and can overcome immune imprinting

Abstract

We investigated mRNA vaccines encoding a membrane-anchored receptor-binding domain (RBD), each a fusion of a variant RBD, the transmembrane (TM) and cytoplasmic tail fragments of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) spike protein. In naive mice, RBD-TM mRNA vaccines against SARS-CoV-2 variants induced strong humoral responses against the target RBD. Multiplex surrogate viral neutralization (sVNT) assays revealed broad neutralizing activity against a range of variant RBDs. In the setting of a heterologous boost, against the background of exposure to ancestral whole-spike vaccines, sVNT studies suggested that BA.1 and BA.5 RBD-TM vaccines had the potential to overcome the detrimental effects of immune imprinting. A subsequent heterologous boost study using XBB.1.5 booster vaccines was evaluated using both sVNT and authentic virus neutralization. Geometric mean XBB.1.5 neutralization values after third-dose RBD-TM or whole-spike XBB.1.5 booster vaccines were compared with those after a third dose of ancestral spike booster vaccine. Fold-improvement over ancestral vaccine was just 1.3 for the whole-spike XBB.1.5 vaccine, similar to data published using human serum samples. In contrast, the fold-improvement achieved by the RBD-TM XBB.1.5 vaccine was 16.3, indicating that the RBD-TM vaccine induced the production of antibodies that neutralize the XBB.1.5 variant despite previous exposure to ancestral spike protein.

Keywords: COVID vaccine; anchored RBD vaccine; immune imprinting; lipid nanoparticle; mRNA vaccine; virus neutralization.

Graphical abstract

Figure 1

Comparison of immunogenicity induced by RBD-TM mRNA and whole-spike mRNA vaccines (A) Schematic diagram comparing whole SARS-CoV-2 spike protein (1,273 amino acids) with the RBD-TM construct (328 amino acids). (B) Common features of the mRNA vaccines used in this study. We used TriLink CleanCap reagent to produce the Cap1 structure, and used a 125 nucleotide polyA tail. The UTRs were designed de novo with reference to known sequences. (C) General features of the LNP delivery system used. The identities and concentrations of ionizable and PEGylated lipids are described in the materials and methods section. (D) Cartoon representation of the proteins resulting from translation of the whole-spike and RBD-TM mRNAs. (E) Neutralization of infection of Vero cells by serum samples from mice vaccinated intramuscularly (IM) with WT SARS-CoV-2 RBD-TM vaccine. BALB/c mice were vaccinated on day 0 and 21 with either 1, 3, or 10 μg mRNA. Serum samples were collected on day 42. Viral strains used were the VIC01 isolate of WT SARS-CoV-2 or the Beta B.1.351 variant. The control serum was obtained by pooling serum collected on day 56 after vaccination of BALB/c mice with 30 μg whole-spike vaccine. (F) Viral titers in lungs of the mice from (E) 3 days after challenge with an N501Y mutant of SARS-CoV-2 (hCoV-19/Australia/VIC2089/2020) on day 65, 44 days after the second dose of vaccine. The titer of infectious virus (TCID₅₀) in the lungs of individual mice were determined by titrating lung homogenate supernatants on Vero cell monolayers and measuring viral cytopathic effect 5 days later. Control animals were unvaccinated aged-matched BALB/c mice. (G and H) RBD-specific Ab titers determined by ELISA in mouse serum samples after either 1- or 5-μg doses IM on days 0 and 21 of either WT whole-spike or WT RBD-TM vaccines. Ab titers were determined on day 21 (G) or on day 42 (H). (I) Neutralization of infection of Vero cells, by VIC01 or Beta strains of virus, by the day 42 serum samples used for ELISA studies in (H) (geometric mean ± geometric SD; n = 5). (J and K) sVNT studies using multiplexed variant RBD-beads indicating relative neutralization of SARS-CoV-2 variants by the serum samples as used in (I) after doses of either 5 μg of WT whole-spike (J) or 5 μg of WT RBD-TM (K) vaccines. The half-maximal inhibitory dilution (sVNT₅₀) is indicated for each serum sample. Horizontal lines show geometric mean; error bars show geometric SD (n = 5 mice); statistical analysis shown in (G and H) used the Kruskal-Wallis test with Dunn’s multiple comparisons test, ∗p < 0.05, ∗∗p < 0.01. For all panels, unless shown, multiple comparisons were not statistically significant.

Figure 2

Immunogenicity of Beta RBD-TM mRNA vaccine as a function of dose (A–D) RBD-specific Ab titers in mouse serum after doses of Beta RBD-TM vaccine administered IM on days 0 and 21. Titers on day 21 against Beta RBD (A) and titers against WT RBD on day 21 (B), day 42 (C), or day 56 (D). NMS, normal mouse serum; 49C9, control antiserum; control, serum from mice treated with 30 μg native mRNA vaccine. Horizontal lines show the geometric mean; error bars show the geometric SD (n = 5 mice). Statistical analysis Kruskal-Wallis test with Dunn’s multiple comparisons test, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. (E) Neutralization of infection of Vero cells by serum samples from mice vaccinated IM with various doses of Beta RBD-TM vaccine. Neutralization of WT VIC01 or a Beta variant are shown. (F and G) RBD-specific Ab titers (F) and VNT (G) in mouse serum at day 56 after two doses of Beta RBD-TM vaccine (day 0 and day 21) in an experiment to extend the dose range to 10 μg. (H) sVNT study using multiplexed variant RBD-beads on day 56 after 3-μg doses of Beta RBD-TM vaccine. Half-maximal inhibitory dilution (sVNT₅₀) is indicated for each serum sample. Horizontal lines show geometric mean; error bars show geometric SD (F–H) (n = 5 mice). For all panels, unless shown, multiple comparisons were not statistically significant.

Figure 3

Immunogenicity and protective efficacy of the Beta RBD-TM mRNA vaccine administered in different LNP formulations (A–D) WT (A, C) or Beta (B, D) RBD-specific Ab titers in Balb/c mouse serum 21 days after a single dose (A, B) or on day 42 (C and D), 21 days after a second dose of 3 μg Beta RBD-TM mRNA formulated in each of four LNPs. The LNPs were formulated with either of two ionizable lipids (ALC-0315 at 46.3 mol% or DLin-MC3-DMA at 50 mol%) with either 1.5 mol% or 0.15 mol% DMG-PEG2000. Horizontal bars show the geometric mean; error bars show the geometric SD (n = 5 mice). Statistical analysis Kruskal-Wallis test with Dunn’s multiple comparisons test, ∗p < 0.05. (E) Half-maximal VNT titers determined against VIC01 or Beta SARS-CoV-2 using the day 42 serum samples used in (C and D). Horizontal bars show the geometric mean; error bars show the geometric SD. (F and G) Viral titers in lungs (F) or nasal turbinates (G) of the mice from (A–E) 3 days after aerosol challenge with a Beta variant (B.1.351) of SARS-CoV-2 on day 65, 44 days after the second dose of vaccine. Control mice were untreated aged-matched Balb/c mice. Horizontal lines show geometric mean titers in control mice. Titers in all four groups of vaccinated mice were below the limit of detection. Results reported in Figures 1, 2, 4, 5, and 6 were obtained after the administration of vaccines in a single LNP formulation using 50 mol% DLin-MC3-DMA and 0.15% DMG-PEG2000. For all panels, unless shown, multiple comparisons were not statistically significant.

Figure 4

Immunogenicity of Kappa and Omicron BA.1 RBD-TM mRNA vaccines (A and B) sVNT study of BALB/c mouse serum using multiplexed variant RBD-beads on day 56, after two 3-μg doses (days 0 and 21) of either Kappa (A) or Omicron BA.1 (B) RBD-TM vaccines. (C) Half-maximal VNT titers determined against naturally occurring Omicron BA.1, BA.2 and BA.4 variants of SARS-CoV-2 in serum collected on day 56 after two doses (day 0 and 21) of Omicron BA.1 RBD-TM mRNA vaccine administered at doses of 1, 3, or 10 μg mRNA. BA.4 VNT was below the effective limit of detection of the assay for all three doses. (D) sVNT study of BALB/c mouse serum on day 56 after 3- or 10-μg doses of Omicron BA.1 RBD-TM mRNA using WT SARS-CoV-2, BA.1, BA.2 or BA.5-RBD-coated beads. Half-maximal inhibitory dilution (sVNT₅₀ or VNT(ID₅₀)) is indicated for each serum sample. Horizontal lines or bars show the geometric mean; error bars show the geometric SD (in A–D) (n = 5 mice). Statistical analysis Kruskal-Wallis test with Dunn’s multiple comparisons test, ∗p < 0.05, ∗∗p < 0.01. For all panels, unless shown, multiple comparisons were not statistically significant.

Figure 5

Heterologous boost studies using Beta or Omicron BA.1 RBD-TM vaccines after exposure to WT whole-spike mRNA vaccine (A and B) Vaccines administered to each group of mice in heterologous boost tests of Beta (A) or Omicron BA.1 (B) RBD-TM vaccines. (A) Relates to data shown in (C–E). (B) Relates to the data shown in panels (F–K) and lists the alternative vaccines administered on day 56 after two 8-μg doses of WT spike vaccine administered on days 0 and 21. (C) Beta RBD-specific Ab titers in BALB/c mouse serum samples at day 42 and day 70 after two 5-μg doses (days 0 and 21) of WT whole-spike mRNA vaccine, and on day 91 after one of three alternative booster doses administered on day 70. (D and E) Beta RBD-specific (D) and Omicron BA.1 RBD-specific Ab titers in mouse serum samples from the three groups on day 125. Horizontal bars show geometric mean; error bars show geometric SD (n = 5 mice); statistical analysis Kruskal-Wallis test with Dunn’s multiple comparisons test, ∗p < 0.05, ∗∗p < 0.01. (F–H) Half-maximal inhibitory dilution sVNT₅₀ values determined in a multiplex RBD bead assay indicating the relative neutralization of binding of variant RBDs to ACE2 by mouse serum sampled on day 90, following day 56 boost with either 8 μg WT whole-spike (F), 8 μg Omicron BA.1 RBD-TM (G) or 2 μg Omicron BA.1 RBD-TM (H). sVNT₅₀s for early variants, Omicron variants, or SARS-CoV-1 are shown in green, orange, or blue, respectively. (I–K) Relative immunogenicity of five alternative boost vaccines administered on day 56, compared by determining sVNT₅₀ of day 90 mouse serum samples against WT SARS-CoV-2 RBD (I), Omicron BA.1 RBD (J), and Omicron XBB.1.5 RBD (K). Horizontal bars show the geometric mean; error bars show the geometric SD (in F–K) (n = 5 mice). For all panels, unless shown, multiple comparisons were not statistically significant.

Figure 6

Heterologous boost study with Omicron BA.5 RBD-TM vaccine after exposure to WT whole-spike mRNA vaccine (A) Vaccines administered to each group of mice in the heterologous boost study. (B–D) Half-maximal inhibitory dilution sVNT₅₀ values determined in a multiplex RBD bead assay indicating the relative neutralization of binding of variant RBDs to ACE2 by mouse serum sampled on day 90, following day 56 boost with either 5 μg WT whole-spike (B), 5 μg Omicron BA.5 whole-spike (C), or 5 μg Omicron BA.5 RBD-TM (D). (E–G) Relative immunogenicity of three alternative boost vaccines administered on day 56, compared by determining sVNT₅₀ of day 90 mouse serum samples against WT SARS-CoV-2 RBD (E), Omicron BA.5 RBD (F), and Omicron XBB.1.5 RBD (G). Horizontal lines or bars show the geometric mean; error bars show the geometric SD (in B–G) (n = 5 mice). Statistical analysis Kruskal-Wallis test with Dunn’s multiple comparisons test. There were no significant differences in (E–G). The p values for multiple comparisons are shown in (F and G).

Figure 7

Heterologous boost study with Omicron XBB.1.5 RBD-TM vaccine after exposure to WT whole-spike mRNA vaccine (A) Vaccines administered to each group of mice in the heterologous boost study. (B–D) Half-maximal inhibitory dilution sVNT₅₀ values determined in a multiplex RBD bead assay indicating the relative neutralization of binding of variant RBDs to ACE2 by mouse serum sampled on day 90, following day 56 boost with either 5 μg WT whole-spike (B), 5 μg Omicron XBB.1.5 whole-spike (C), or 5 μg Omicron XBB.1.5 RBD-TM (D). (E and F) Relative immunogenicity of three alternative boost vaccines administered on day 56, compared by determining sVNT₅₀ of day 90 mouse serum samples against WT SARS-CoV-2 RBD (E) or Omicron XBB.1.5 RBD (F). (G) Neutralization of XBB.1.5 virus infection of Vero cells by serum samples collected on day 90. Where neutralization was not detected, the individual samples are at the limit of detection of 40. In (B–G), the horizontal lines or bars show geometric mean; error bars show geometric SD (n = 8 mice). Statistical analysis Kruskal-Wallis test with Dunn’s multiple comparisons test ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.