Structural differences between the putative carbohydrate-recognition domains of human IL-1 alpha, IL-1 beta and IL-1 receptor antagonist obtained by in silico modeling

In a previous report (Cebo et al. J Biol Chem 276 (2001) 5685–5691), it was established that biologically active recombinant human IL-1α and IL-1β had different carbohydrate-binding properties. IL-1α recognized a di-antennary N-glycan with two α2-3-linked sialic acid residues, whereas IL-1β recognized the GM4, a α2-3-linked sialylated glycosphingolipid. These different carbohydrate-binding properties of two interleukins binding to the same receptor (IL-1R) could explain why these molecules had different biological effects and cell specificities. Molecular modeling of the ligands and in silico docking experiments defined putative carbohydrate-recognition domains localized in the same area of the two molecules, a domain different from that defined as the type I IL-1R binding domain. The calculated pattern of hydrogen bonding and of van der Waals interactions fulfilled the essential features observed for calcium-independent lectins (mammalian, viral or bacterial). The analysis of the same domain of the third members of this family of molecules, the IL-1R-antagonist, indicated it did not fulfill the criteria for carbohydrate-recognition domains. It is proposed that its role as a pure antagonist is due to the absence of lectin activity and consequently explained its inability to associate IL-1R with other surface molecular complexes necessary for signaling.