The MIT-T-COVID vaccine realizes an n-times coverage objective by encoding multiple epitopes for each target MHC class I and II diplotype to (1) expand diverse sets of T cell clonotypes to fight viral infection, (2) accommodate variations in epitope immunogenicity between vaccinees, and (3) reduce the chances that viral evolution will lead to immune system escape (12). The MIT-T-COVID vaccine consists of eight MHC class I epitopes and three MHC class II epitopes ( Figure 1A ; Supplementary Methods and Supplementary Table 1 ). The MHC class I and II vaccine peptides are encoded into a single mRNA construct for delivery with the same Acuitas LNP delivery platform that is used by the Pfizer-BioNTech Comirnaty^® vaccine ( Figure 1A ). The eight MIT-T-COVID MHC class I epitopes are the HLA-A*02:01 subset of the MHC class I de novo MIRA only vaccine design of Liu et al. (11) that used combinatorial optimization to select vaccine epitopes to maximize n-times population coverage over HLA haplotype frequencies (12). For inclusion in the assembled construct, the eight MHC class I vaccine peptides were randomly shuffled, and alternate peptides were flanked with five additional amino acids at each terminus as originally flanked in the SARS-CoV-2 proteome. Selected epitopes were flanked to test if flanking enhanced or impaired epitope presentation. The three MHC class II epitopes were selected by considering all SARS-CoV-2 proteome windows of length 13-25 and selecting conserved peptides that are predicted to be displayed by the mouse MHC class II allele H-2-IAb ( Supplementary Methods ). The mRNA construct encodes a secretion signal sequence at the N-terminus and an MHC class I trafficking signal (MITD) at the C-terminus (13). Peptide sequences are joined by non-immunogenic glycine/serine linkers (14). The construct also included control peptides for HLA-A*02:01 and H-2-IAb (CMV pp65: NLVPMVATV for HLA-A*02:01 and Human CD74: KPVSKMRMATPLLMQAL for H-2-IAb) (15).

We immunized with three test articles: a negative control injection (PBS), the Pfizer-BioNTech Comirnaty^® vaccine (wastage that was refrozen and then thawed), and the MIT-T-COVID vaccine ( Figure 1B ). We immunized HLA-A*02:01 human transgenic mice with these three vaccines, immunizing 11 age-matched male mice with each vaccine. Comirnaty^® or MIT-T-COVID immunizations contained of 10 µg of mRNA. The mice were immunized at Day 0 and boosted at Day 21 ( Supplementary Methods ). At Day 35 three mice from each group were sacrificed for immunogenicity studies, and at Day 37 the remaining eight mice were challenged intranasally (I.N.) with 5 x 10⁴ TCID₅₀ of the SARS-CoV-2 B.1.351 variant. Challenged mice were subjected to daily monitoring for the onset of morbidity (i.e., weight changes and other signs of illness) and any mortality. At Day 39 (2 days post infection) three mice were taken from each group and sacrificed to determine viral burdens and to perform lung histopathology. At Day 44 (7 days post infection) the remaining mice were sacrificed to determine viral burdens and to perform lung histopathology.

Additional methods are presented in the Supplementary Material .

Immunization with MIT-T-COVID vaccine expanded CD8⁺ and CD4⁺ T cells that expressed interferon gamma (IFN-γ) or tumor necrosis factor alpha (TNF-α) when queried by vaccine epitopes ( Figures 2A, B ). The observed variability of immunogenicity of MIT-T-COVID epitopes in convalescent COVID patients ( Supplementary Table 1 ) and in the present study supports our usage of multiple epitopes per MHC diplotype for n-times coverage. Immunization by Comirnaty^® produced no significant T cell responses to vaccine epitopes (including a Spike epitope) when compared to PBS. CD8⁺ T cells that are activated by the CD8-4 epitope (YLQPRTFLL, Spike 269-277) are expanded in animals immunized with the MIT-T-COVID vaccine (IFN-γ 1.32% ± 0.53% of CD8⁺ T cells, P = 0.0087 vs. PBS; TNF-α 0.38% ± 0.09% of CD8⁺ T cells, P = 0.0149 vs. PBS; Figure 2A ). Similarly, the CD8-8 epitope (TVYSHLLLV, ORF3a 89-97) activated CD8⁺ T cells that are expanded by the MIT-T-COVID vaccine (IFN-γ 0.60% ± 0.02% of CD8⁺ T cells, P = 0.001 vs. PBS; TNF-α 0.25% ± 0.12% of CD8⁺ T cells, P = 0.015 vs. PBS) and the CD4-2 epitope (RFYFYTSKTTVASLIN, ORF1ab 1421-1436) activated CD4⁺ T cells are expanded by the MIT-T-COVID vaccine (TNF-α 0.078% ± 0.027%, P = 0.058 vs. PBS, P = 0.010 vs. Comirnaty^®, IFN-γ not significant, Figure 2B ). We also measured interleukin-2 (IL-2) expression and found the MIT-T-COVID vaccine significantly expanded CD4⁺ T cells activated by the SARS-CoV-2 CD4⁺ pool (P = 0.0015 vs. PBS, P = 0.0013 vs. Comirnaty^®, Supplementary Figure 12 ). The lack of HLA-A*02:01 transgenic animal response to certain CD8⁺ epitopes that were immunogenic in patients ( Supplementary Table 1 ) is consistent with past studies of transgenic mouse models (16), the variability of immunogenicity of epitopes between in-bred mice (17), and the potential of immunodominance of the immunogenic epitopes.

We immunized an additional cohort of female HLA-A*02:01 transgenic mice ( Supplementary Methods ), and results are shown in Supplementary Figure 8 . We also immunized female mice with synthetic peptides mixed with poly IC adjuvant and found no significant SARS-CoV-2 specific T cell responses compared to PBS controls ( Supplementary Figures 9A, B and Supplementary Methods ). We found a significant increase in the number of effector and memory CD8⁺ CD44⁺ T cells in MIT-T-COVID and Comirnaty^®-immunized mice, compared to those immunized with Peptide/poly IC or PBS ( Supplementary Figure 9C ).

T cells that lack IFN-γ responses can produce effective immune mediators (18). We found that the fraction of CD4⁺ T cells that are Foxp3⁺, designated regulatory T cells (Treg), were not expanded in Comirnaty^® or MIT-T-COVID-immunized animals, suggesting that Treg cells that could induce tolerance were not expanded by immunization ( Figure 2C ).

Upon viral challenge, both PBS and MIT-T-COVID-immunized animals exhibited a more than 10% weight loss by Day 3 (PBS mean weight reduction 11.386% ± 1.688%, MIT mean weight reduction 10.851% ± 0.641%), with the MIT-T-COVID-immunized animals beginning to recover from Day 4 onward ( Figure 3A ). The weight phenotype was mirrored by the clinical score phenotype, with PBS animals not recovering and MIT-T-COVID-immunized animals improving from Day 5 onward ( Figure 3B and Supplementary Methods ). The Comirnaty^® vaccine protected animals from significant weight loss and poor clinical scores.

When the yields of infectious progeny virus were measured at 2 days post infection (dpi), we noted that mice immunized with MIT-T-COVID vaccine had 6.706 ± 0.076 log10 TCID₅₀/g, compared to 7.258 ± 0.367 log10 TCID₅₀/g in the PBS control, representing a moderate 3.6-fold reduction in viral replication (P = 0.046; Figure 3D ). In contrast, mice immunized with Comirnaty^® had infectious viral titers that were below the detection limits at 2 dpi. Infectious viral titer was not significant for all test articles at 7 dpi ( Figure 3D ). Lung viral mRNA levels measured by qPCR followed the same trends as infectious virus progeny in the lungs, with Comirnaty^® showing no significant levels. Despite a reduced content of total and sub-genomic viral RNAs associated with MIT-T-COVID vaccine samples as assessed by qPCR, the reduction was insignificant, compared to those of PBS control ( Supplementary Figure 1 ).

All the Comirnaty^® and MIT-T-COVID-immunized mice survived to 7 dpi, when the study was terminated, with the difference in clinical scores of these two vaccine groups becoming insignificant ( Figure 3C ). In contrast, four of five PBS control mice had been euthanized because of weight loss > 20%. The survival of all five animals immunized with Comirnaty^® and the MIT-T-COVID, respectively, was significant, compared to that of PBS-immunized control (P = 0.0053, logrank test).

We noted that mice immunized with the Comirnaty^® vaccine elicited substantial specific antibodies capable of neutralizing Beta and WA-1 variants of SARS-CoV-2 with 100% neutralizing titers (NT₁₀₀) of 896 ± 350 and fold 2048 ± 1887, respectively ( Figure 3E ). The higher neutralizing titer against WA-1 is expected as it is the strain matched to Comirnaty^®. As expected for a peptide-based vaccine, no neutralizing titer could be readily detected in the serum from mice immunized with MIT-T-COVID vaccine or PBS. Total specific IgG/IgM antibodies were also measured by ELISA against a cell lysate prepared from WA-1-infected Vero E6 cells. Comirnaty^® has a log10 IgG/IgM titer of 4.915 ± 0.213, while the log10 titer produced by PBS was 2.452 ± 0.321 and the MIT-T-COVID vaccine produced a log10 titer of 2.266 ± 0.291 ( Figure 3F ). The low but detectable titers in PBS and MIT-T-COVID vaccine-immunized mice may represent an early IgM response to viral infection.

All lung samples were subjected to immunohistochemistry (IHC) staining for the SARS-CoV-2 spike protein ( Supplementary Figure 2 ). We found specimens immunized with PBS exhibited extensive staining indicative of viral infection throughout the epithelium of both the bronchioles and the alveolar sacs, with the viral infection appearing more intense at 2 dpi. Although viral infection is significantly reduced by 7 dpi, viral antigen was still readily detectable throughout alveoli. In comparison, specimens immunized with the MIT-T-COVID vaccine exhibited similarly extensive viral infection at 2 dpi throughout the bronchiolar and alveolar epithelia, albeit somewhat reduced in intensity. However, by 7 dpi, viral infection was significantly reduced in both extent and intensity, with brown puncta being detected only in a few alveoli scattered throughout the tissue. Contrasted with both PBS and MIT-T-COVID-immunized specimens, the Comirnaty^®-immunized specimens exhibited significantly reduced viral infections at both 2 and 7 dpi. Apart from a single area at 7 dpi (see Supplementary Figure 3 ), viral antigen was undetected at both timepoints in Comirnaty^®-immunized animals.

Paraffin-embedded and H&E-stained lung specimens of differentially immunized mice, harvested at 2 and 7 dpi, were subjected to histopathological examination ( Supplementary Figure 4 ). We found that at 7 dpi mice immunized with MIT-T-COVID vaccine exhibited extensive lymphocytic infiltrations in perivascular regions and spaces from around bronchi, bronchioles, to alveoli. Fewer infiltrations were found in mice immunized with either Comirnaty^® or PBS. Additionally, these infiltrations only localized at perivascular regions around bronchi and large bronchioles. Despite the less intensive and localized inflammatory infiltrates, we also noted widespread congestion, hemorrhage, and few foci of thromboembolism were exclusively observed within the lungs of Comirnaty^®-immunized mice but not others ( Supplementary Figure 4 ). We also noted the lung histopathology was milder but the same pattern at 2 dpi than those of 7 dpi (data not shown).

Lung specimens were subjected to IHC staining for CD8⁺ and CD4⁺ cells at both 2 and 7 dpi ( Figure 4 and Supplementary Figures 5–7 ). At 2 dpi, we found a significant increase in CD8⁺ T cells infiltrating the lungs in mice immunized with MIT-T-COVID (12.6% ± 5.91% of all nucleated cells were CD8⁺) or Comirnaty^® (12.4% ± 2.35%) compared to mice immunized with PBS (PBS mean 4.48% ± 1.99%; P = 0.044 vs. MIT-T-COVID, P = 0.050 vs. Comirnaty^®; Figure 4A ). At 7 dpi, we found a significant increase in CD8⁺ T cells infiltrating the lungs in mice immunized with MIT-T-COVID (24.0% ± 5.21% of all nucleated cells were CD8⁺) compared to mice immunized with Comirnaty^® (Comirnaty^® mean 4.35% ± 2.20%, P = 0.001) or PBS (PBS mean 7.32% ± 4.88%, P = 0.001; Figure 4A ). We observed a significant increase in CD8⁺ T cells infiltrating the lungs in mice immunized with MIT-T-COVID between 2 and 7 dpi (P = 0.0039), and a significant decrease in CD8⁺ T cells infiltrating the lungs in mice immunized with Comirnaty^® between 2 and 7 dpi (P = 0.044; Figure 4A ). At7 dpi, we also found a significant increase in CD4⁺ T cells infiltrating the lungs in mice immunized with MIT-T-COVID (7.09% ± 4.05%) compared to mice immunized with PBS (PBS mean 0.41% ± 0.39%; P = 0.0062; Figure 4B ). In the unchallenged cohort of female mice, we found an increase in CD8⁺ T cells infiltrating the lungs in mice immunized with MIT-T-COVID (1.12% ± 0.59% of all nucleated cells were CD8⁺) or Comirnaty^® (0.87% ± 0.56%) compared to mice immunized with PBS (0.09% ± 0.10%; P = 0.001 vs. MIT-T-COVID, P = 0.003 vs. Comirnaty^®; Supplementary Figures 10, 11 ).