Cut or bind? Antigen-specific processing mechanisms define CD4⁺ T cell immunodominant epitopes for SARS-CoV-2 S and N...

CD4⁺ T cell responses are key to adaptive immunity, yet the mechanisms underlying peptide selection and immunodominance across MHC class II variants in humans remain poorly defined. Two non-mutually exclusive models - First Bind-then cut (FBtc) and First Cut-then bind (FCtb) - have been proposed to explain immunodominant peptide selection, but experimental evidence in humans is mostly limited to a single allotype (HLA-DRB1*01:01).

To generalize processing mechanisms across DRB1 alleles we developed an integrative strategy combining in silico prediction and a reconstituted antigen processing system. The independent and combined outcome of both approaches was validated on curated SARS-CoV-2 epitope data (IEDB) for responses to the Spike and Nucleocapsid proteins across a panel of 11 DRB1 allotypes, covering over 90% of European Caucasian populations. Potential immunogenic regions identified by the combination of both methods enabled the design of minimalistic peptide pools whose performance was validated via flow cytometry and ELISpot assays in post-Covid19 and pre-pandemic donors. Mechanistic insights for the selection of immunodominant peptides were derived analyzing biophysical parameters and proteolysis of the model antigens.

Three prediction tools used showed limited concordance for some allotypes (< 5%), but their combined output for all allotypes considered revealed potential immunogenic hotspots in the model antigens. Complementary, the reconstituted in vitro system identified allotype-dependent and promiscuous peptide candidates. Minimal peptide pools designed from the overlap of both methods featured improved performance to identify IEDB entries and induced robust CD4⁺ T cell activation in post-COVID-19 donors. Mechanistic modeling classified most immunodominant peptides from the Spike protein as arising via FCtb while FBtc predominated for Nucleocapsid. Epitope selection pathways are therefore antigen-dependent defined by proteolytic resistance and solvent accessibility.

We establish a scalable, genomics-informed framework for decoding CD4⁺ T cell immunodominance across diverse HLA contexts. Our findings reveal that antigen-intrinsic features govern the preferential processing pathway - FCtb for Spike and FBtc for Nucleocapsid - and validate the utility of minimal peptide pools for population-level immune-monitoring. These insights inform the design of personalized immunotherapies and broadly effective vaccines.