B-cell epitope prediction facilitates the design of antibody-binding constructs for the development of novel vaccines and immunodiagnostics. This work aimed to gain insights into the problem of B-cell epitope prediction using structural energetics. Methods: Structural-energetic analysis was applied to peptide and protein antigens. A possible rate-limiting process of local epitope unfolding was considered for the cross-reaction of antipeptide antibody with protein antigen. Immunodominance was treated as a thermodynamically determined hierarchical steric-exclusion phenomenon. The algorithm thus developed was implemented as the computer program SAPPHIRE (Structural-energetic Analysis Program for Predicting Humoral Immune Response Epitopes), with the estimated affinity for antibody as the main criterion for epitope prediction. Predictions were rendered on the cross-reactivities of polyclonal antibodies to 38 peptides with 15 globular proteins of known structure and evaluated against published experimental data comprising 18 positive and 20 negative binding interactions. Results: Structural-energetic parameters could not be unambiguously assigned to cysteine in view of its capacity for disulfide bond formation. The energetic contribution of histidine could not be determined in view of the uncertainty of its protonation state at physiologic pH. The binding contexts defined by the types of participating antibodies (antipeptide or antiprotein) and antigens (peptide or protein) were all fundamentally different from one another. Predictions on genuine antibody-antigen cross-reactivity could be evaluated against empirical data only with regard to interaction between antipeptide antibody and protein antigen. Maximum areas under the receiver operator characteristic curves were approximately 0.71 either with or without consideration of immunodominance. Conclusions: 1) B-cell epitope prediction is potentially complicated by the presence of cysteine and histidine. 2) The evaluation of B-cell epitope predictions against empirical data is meaningful only if the two pertain to exactly the same types of antibody (antipeptide or antiprotein) and antigen (peptide or protein). 3) Predictions on genuine antibody-antigen cross-reactivity can be evaluated against empirical data in the case of interaction between antipeptide antibody and protein antigen. Molecular

Username
Password