The experimental determination of conformational allergen epitopes recognized by IgE is a difficult task because they often involve discontinuous amino acid residues, being separated in the primary allergen sequence, and require the correct allergen fold.

We sought to develop a computational tool for the localization of conformational IgE epitopes by using a structure-based comparison of allergen surfaces and IgE cross-reactivity data.

Our approach involves the quantitative analysis of geometric and physicochemical surface parameters and the subsequent correlation of surface similarity scores to immunologic data. The software tool Surface comparison–based Prediction of Allergenic Discontinuous Epitopes (SPADE) is able to predict the IgE epitopes of an allergen given the availability of at least 2 structural models and IgE reactivity data.

We report on the application of our tool to 3 allergen families: the lipocalins, the group 10 pathogenesis-related proteins, and the group 2/3 grass pollen allergens. First, we succeeded in the partial relocalization of IgE epitopes of bovine β-lactoglobulin and grass pollen Phl p 2 as known from the x-ray structures of their antibody complexes. Second, we measured the relative binding of anti–Bet v 1 IgE to 10 homologous proteins and correlated these data to surface comparison results involving Bet v 1, 5 of the homologs, and 2 hypoallergenic Bet v 1 isoforms. Thereby we predicted IgE-reactive surface portions in agreement with IgE epitope–mapping studies.

Our approach is the first for the prediction of IgE epitopes by combining structural and IgE cross-reactivity data. It should be useful for the development of point-mutated or structurally disrupted allergen derivatives for allergen-specific immunotherapy.