Discovering the Correlation between Theoretical and Experimental Binding Data for the HLA-A*0201 Molecule Using an Inverse Folding Approach

Background: The estimation of peptides binding to the Human Leukocyte Antigen (HLA) has significant implications for rational vaccine design and other immunotherapies. The present work employs an Inverse folding (IF) approach to estimate HLA-binding peptides and to establish a correlation between experimental binding affinity and the estimated interaction potential score of HLAA* 0201-peptide interactions. Methods: The interaction potential score of a peptide is computed as the total energy of interaction with the contact residues of the HLA and peptide using statistical pairwise contact potentials, namely Miyazawa-Jernigan (MJ), and Betancourt-Thirumalai (BT). Results: The results showed a negative correlation between computed MJ and BT interaction potential scores and experimental binding affinity of peptides to the HLA-A*0201 molecule. BT scores correlate better with the experimental binding affinities of peptides, as indicated by Pearson's correlation coefficient (-0.568, R² =0.322, p <0.001), compared to MJ scores (-0.457, R2 =0.208, p <0.001). Discussion: Since the HLA class I molecules have specificity for certain amino acid side chains in distinct binding pockets within the binding cleft, the structural templates of peptide-bound HLA molecules may provide a means to interrogate the interactions between HLA and peptide. IF indirectly explores the underlying fitness landscape by focusing exploration on regions where the protein's backbone fold is preserved. Given the involvement of multiple template structures for combinations of peptides and HLA alleles, the IF approach is expected to be more robust. Conclusion: The IF approach demonstrates potential for evaluating the intricate assembly of HLA molecules with any novel peptide, which holds significant relevance for vaccine development.