a The three mRNA vaccine candidates for COVID-19. RQ3011-RBD contains an mRNA encoding the receptor binding domain of S with a C-terminal membrane-anchoring helix. RQ3012-Spike contains an mRNA encoding the full-length wild-type (WT) S. RQ3013-VLP contains three mRNAs encoding S, M, and E proteins that can assemble into VLPs. b Electron microscopy images of VLPs produced by RQ3013-VLP. VLPs in the cell culture supernatant were purified and concentrated by ultracentrifugation and subjected to negative staining for electron microscopy. Scale bar, 500 nm. c The expression of S with various codons. GC-content for each sequence is indicated in the parenthesis. All S mRNAs have uridine replaced by Ψ. d Screening for optimal modified nucleotides in M and E mRNAs. A codon-optimized sequence for each gene was tested for protein expression with six different kinds of modified nucleotides. They are uridine (lane 1), m5C/Ψ (lane 2), Ψ (lane 3), mo5U (lane 4), m1Ψ (lane 5), and m5C (lane 6). All mRNAs were used to transfect HEK 293A cells, and protein expression was detected by western blotting. EGFP mRNA was used as a negative control. The mRNAs used in vaccines are indicated by stars. e The scheme of mice immunization. Mice (n = 10 per group) were either mock-immunized with Placebo (empty LNP, blue circle) or vaccinated with RQ3011-RBD (red square), RQ3012-Spike (green triangle), and RQ3013-VLP (purple diamond) intramuscularly. Time points of vaccination and bleeding are indicated by arrows. f The antibody response was analyzed by ELISA using the S antigen. The black dashed line indicates the titer of pre-immune mice (n = 10). g Neutralizing antibodies were analyzed by the VSV-based pseudovirus assay. Lines represent average titers of all animals in each vaccine group. The black dashed line indicates the limit of detection of the assay (reciprocal titer of 100). Any measurement below the limit of detection was assigned a value of half the limit of detection for plotting and statistical purposes. h Frequency changes of CD4+ and CD8+ T cells 8 weeks after vaccination. The RQ3011-RBD group was omitted for analysis. The significance of the ratio difference was indicated. i Frequency of CD44high/CD4+ and CD44high/CD8+ T effector cells 8 weeks after vaccination. j, k Antigen-specific responses were assessed by in vitro antigen restimulation. PBMCs isolated from the blood were stimulated with either SARS-CoV-2 VLPs or S protein and analyzed via flow cytometry for the frequency of VLP or Spike-specific CD4+ and CD8+ T cells expressing IFN-γ. Statistical analyses were carried out by Student’s t test when two groups were analyzed, and by ANOVA when more than two groups were analyzed (**P < 0.005; ***P < 0.001; ****P < 0.0001).