A comparative structural workflow for investigating structural features associated with immunoglobulin E cross-reactivity between cow’s milk and respiratory lipocalin allergens

Abstract

Background: Cow’s milk allergy and allergic asthma can occur in the same individuals, but it is not well understood at the molecular level. Bovine β-lactoglobulin (Bos d 5) is a major cow’s milk allergen and a member of the lipocalin protein family. Several clinically important respiratory allergens are lipocalins. Their shared fold provides a basis for comparing surfaces that may be relevant to IgE recognition. Objective: To develop a structural workflow for comparing Bos d 5 IgE-binding regions with corresponding regions in milk, respiratory and human lipocalins. Methods: Literature-reported linear IgE-binding regions of Bos d 5 and the conformational epitope defined from the Bos d 5 IgE–Fab complex were mapped onto homologous lipocalin structures using structural alignment. In parallel, comparative Fab–query allergen models were generated by positioning each query structure in the binding orientation defined by the Bos d 5 IgE–Fab complex. Structural and physicochemical descriptors were calculated after standardized preparation without additional energy minimization (RAW) and with minimization (MIN). Similarity to Bos d 5 was assessed using principal component analysis (PCA), Euclidean distance-based ranking, and group comparisons using permutation tests. Results: The dataset comprised the Bos d 5 reference structure (n = 1), respiratory allergens (n = 23), human lipocalins (n = 18), and β-lactoglobulin (βLG) homologs (n = 17 Across mapped linear IgE-binding regions, non-bovine ruminant βLGs most frequently ranked closest to Bos d 5, occupying the top rank in 8 of 12 analyses under the RAW protocol and 7 of 12 analyses under the MIN protocol. For the mapped conformational epitope region, non-bovine ruminant βLGs showed the smallest median distance to Bos d 5 under both protocols (RAW: 0.82; MIN: 1.10). The comparison with respiratory allergens was significant under RAW (p = 0.0311), but not under MIN (p = 0.0774). In the MIN composite analysis, median distances were 1.60 for non-bovine ruminant β-lactoglobulins, 4.46 for respiratory allergens and 4.51 for human lipocalins. Several respiratory lipocalins, including Fel d 4, Can f 4, Mus m 1, Fel d 7, and Cav p 1, repeatedly appeared among nearer neighbors in selected descriptor sets; however, this pattern was not consistent across all descriptor sets. Conclusion: This study developed a reproducible structural comparison workflow for evaluating IgE-binding surface similarity across homologous lipocalins. Application to Bos d 5 consistently recovered close similarity among non-bovine ruminant βLGs, as expected from their close evolutionary and structural relationship. A subset of respiratory lipocalins showed partial structural similarity in selected analyses, but this pattern was not consistent across all comparisons. The workflow therefore provides a way to prioritize candidate proteins and surface regions for experimental testing of IgE binding and possible cross-reactivity.