Abstract
There is currently limited understanding of the severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) adaptation to the human leukocyte antigen (HLA) proteins which mediate CD8 (HLA-I) and CD4 (HLA-II) T cell immune responses. We investigated population-level T cell immune escape in SARS-CoV-2 Spike protein at amino acid binding positions (the anchor motifs) preferred by the highly restrictive peptide binding grooves of the HLA. SARS-CoV-2 Spike protein sequences isolated in South Africa from January 2020 until June 2022, were used. All possible 9-mer and 15-mer peptides in the sequence alignment were scanned for matches to HLA-I and HLA-II anchor motifs, respectively. Peptide positions with matched anchor motifs and ≥1% mismatched sequences were investigated for immune escape using immunoinformatic prediction methods and directional evolution along the phylogenetic tree. Toggling of short-lived immune escape mutations at HLA-I anchor motifs was observed in 17 peptides across Spike. Eight of these overlapped with HLA-II escape mutations. Six mutations were related to zoonotic adaptation. All 17 sites were under significant directional evolution along the phylogenetic tree, and 16/17 are within published confirmed or inferred T cell epitopes. Immune escape predictions for HLA- A*66:01/A*68:01 were common (n = 7/17). HLA- A*02:05, A*03:01, B*07:02, B*08:01, B*58:01, DRB1*04:01 and DQA1*01:02-DQB1*06:02 were each associated with at least two escape mutations. This immunoinformatic prediction of T cell immune escape at HLA anchor motifs: (i)shortlisted potentially understudied population-specific HLA and immune escape (ii)revealed a footprint of underlying toggling of short-lived immune escape mutations, and (iii)has potential to cost-effectively guide pre-clinical research questions on the inclusion of partially conserved but dominant epitopes in vaccine immunogens.