Wnt signalling pathways are of considerable interest as targets for cancer treatment.1 The first step of Wnt signalling generally involves the binding of a Wnt protein to the cysteine-rich domain (CRD) of a Frizzled (Fzd) receptor; secreted Frizzled-related proteins (sFRPs), which contain a Fzd-type CRD, can antagonise Wnt signalling by preventing this interaction.2 19 Wnts, 10 Fzds and 5 sFRPs are known in humans, making systematic experimental investigation challenging. Additionally, lipidation of Wnt directly facilitates their binding to Fzd; computational approaches typically used to assess protein-protein binding affinity are generally incapable of considering residues other than standard amino acids. These challenges to both experimental and computational approaches have delayed a comprehensive understanding of these interactions and limited opportunities for rational drug design targeting these pathways.
In this study, we have developed a computational method for accurate prediction of Wnt-Fzd binding affinities.3 Homology models of the entire set of human and mouse Wnt-Fzd interactions were built, based on the crystal structure of a representative Wnt-Fzd interaction.4 We have then developed and validated a model for predicting Wnt-Fzd binding affinities, based on a set of Wnt-Fzd binding affinities determined by biolayer interferometry.5 The model incorporates terms from several protein-protein docking scoring functions, as well as a term to consider the contribution to binding made by the Wnt lipid. Based on the performance of the model in our training and test cases, we estimate that the model can predict binding affinities accurately for 75-80% of complexes, with an error similar to experiment. The model was used to predict the binding affinities of all mouse and human Wnt-Fzd CRD and Wnt-sFRP CRD interactions, and revealed trends in the binding promiscuity of specific Wnts, Fzds and sFRPs. The data further revealed specific Wnt and Fzd/sFRP residue positions that afford the greatest contributions to the binding energy.
The comprehensive predictions made in this study provide the basis for laboratory-based studies of previously unexplored Wnt-Fzd and Wnt-sFRP interactions, which may reveal further Wnt signalling pathways. Furthermore, the generated model may be valuable for the discovery of Wnt signaling antagonists for the treatment of hard-to-treat cancers.