Relationship between immunogenicity and protein structure at amino acid substitution sites of blood group antigens

Polypeptide blood group antigens, many of which are created by single exofacial amino acid substitutions, have varying immunogenicities. Why some amino acid substitutions are more immunogenic than others is little understood. Using AlphaFold2, an artificial intelligence system that predicts 3-dimensional protein structure, along with multiple other structure analysis programs, we investigated protein structure at sites of amino acid substitutions that create 9 clinically significant blood group antigens. Based on structure predictions, the amino acid substitutions that create the 4 most immunogenic of the 9 antigens (K, Jka, Lua, and E) were typically buried or partially buried in rigid, ordered protein regions, usually helices and β-strands. This was reflected by their lower mean relative solvent accessibility (RSA) than the 5 less immunogenic antigens (c, M, Fya, C, and S; 0.13 vs 0.81; P = .003) and higher mean AlphaFold2 confidence score (92.5 vs 48.3; P = .001; scores <50 predict protein disorder). Substitutions creating the 5 least immunogenic antigens (c, Fya, M, C, and S) were all predicted to be in flexible regions with high accessibility, either in surface-accessible loops (C, c) or disordered coils (Fya, M, and S). Scatter plots revealed a positive linear correlation of immunogenicity with confidence score (R2 = 0.826; P = .0007) and percent helix/β-strand in 15-mers centered around the substitution sites (R2 = 0.763; P = .0021) and a negative linear correlation with RSA (R2 = 0.688; P = .0057). Therefore, based on an informatics analysis, the protein secondary and tertiary structures at amino acid substitution sites that create blood group antigens are significant correlates and potential determinants of immunogenicity.