Mesothelin Structural Modeling

Lee, Byungkook

Abstract

Using various existing protein structure prediction tools, we built a plausible model of the 327-residue mesothelin molecule. (Sathyanarayana et al, BMC Structural Biology, 9:1, 2009) The predicted model turns out to be an α/αsuperhelix, a fold type that is mostly involved in protein-protein interaction. In the mean time, Dr. Mitchell Ho of the Antibody Therapy Unit of LMB determined that only the N-terminal 64-residue portion of mesothelin binds MUC16. (Kaneko et al., J. Biol. Chem. 284:3739-3749, 2009) We built a separate model for the 64-residue unit using existing structure prediction servers. This model is related to, but not the same as, that in the model of the full 327-residue mesothelin molecule. We then built a new model of the full mesothelin molecule by replacing the 64-residue unit with the new model and adjusting the joining region. We believe that the new model is a refinement of the old model and is likely to be closer to the real structure of mesothelin. We are currently collecting and examining the structures of carbohydrate-protein complexes and of protein complexes that involve α/α superhelical structures. If we obtain a carbohydrate-protein complex structure from the database in which the protein portion is similar to one of the model structures, we already have a good trial mesothelin-carbohydrate complex structure. Otherwise, we will use available programs to predict carbohydrate binding sites and build a model of the complex. Such a model will not be highly accurate, particularly because we do not know the structure of the MUC16 carbohydrate that binds mesothelin. But we hope that they will be good enough to suggest the residues that interact with the carbohydrate. Dr. Ho can verify these residues by measuring the changes in binding to MUC16 upon mutating them. We will refine our model by incorporating these experimental data. Once we have such a model, we can start designing the inhibitor molecules.