Protein-Protein Recognition: An Analysis by Docking Simulation

Abstract

In this paper, we make use of a docking algorithm developped by Wodak & Janin^1–3 and simulate protein-protein recognition by reconstituting complexes from component molecules. When applied to protease-inhibitor or antigen-antibody complexes of known X-ray structure, the procedure efficiently retrieves native modes of association. However, it also selects a number of non-native modes with structural and physical-chemical features that would be expected only from native complexes⁴. We find that these ‘false positives’ cannot be discriminated from the correct solution on simple criteria such as the number of H-bonds, buried polar groups or cavities at the interface, or more elaborate ones like conformational energy⁵. To predict a complex of unknown structure, additional information must be available. As an example, we model the complex between the hemagglutinin antigen from the flu virus and the Fab fragment of a monoclonal antibody raised against this antigen⁶, taking into account the known location of mutations which affect interaction of the antigen with the antibody.