AI-aided design of multi-epitope peptide vaccine elicits cellular and humoral immunity to broad Omicron subvariants

Peptide vaccines offer a flexible, rapidly updatable platform for responding to SARS-CoV-2 antigenic drift. However, many candidates target a single epitope class, predominantly CD4+/CD8+ T-cell or short linear B-cell epitopes, limiting their capacity to induce both high-titer neutralizing antibodies and robust T cell immunity. Here, to elicit potent humoral and cellular responses, we applied AI-aided epitope prediction tools to analyze 18 Omicron subvariants and designed two candidate peptides from the RBD of XBB.1.5, named LY54-XBB (L455-Y508) and P67-XBB (Y351-K378). In silico TCR-pMHC binding analysis and structural modeling validated that both peptides engage T-cell epitopes. The immunogenicity of the two peptide nanoemulsions was validated in murine and NHP models. Mixed vaccination elicited RBD-binding, ACE2-blocking, and pseudovirus-neutralizing antibody responses, together with a Th1-biased cross-reactive cellular immune response and no observable adverse reactions. Crucially, mixed vaccination protected the lungs of HLA-A2/DR1-hACE2 transgenic mice from the SARS-CoV-2 BA.5 variant challenge. Meanwhile, ex vivo stimulation of hPBMCs from COVID-19 convalescent plasma donors confirmed that both peptides elicit antigen-specific CD4+ and CD8+ T cell responses, validating the inclusion of effective T-cell epitopes. These complementary peptides, supported by both experimental and computational validation, represent promising, rapidly updatable booster candidates. Our epitope-based pipeline offers a generalizable framework for vaccine design that may contribute to sustaining population immunity against SARS-CoV-2 and other antigenically diverse pathogens.