De Novo Design of Proteins That Target Natural Protein Epitopes

The ability to design highly specific protein binders to disrupt protein-protein interactions would have immediate application to treatment of many diseases and for the dissection of protein-interaction networks. To date, no general computational method has been proposed for the generation of de novo protein binders. We present a new algorithm that centers on the incorporation of “hot-spot” residues onto protein scaffolds to generate protein binders to specific epitopes on proteins of interest. The focus on key interactions reduces the dimensionality of the sequence and conformational search problem considerably, opening the way to the generation of dozens of redesigned scaffolds for experimental testing. Our method focuses first on the generation of a small number of disembodied high affinity sidechain-target interactions which bias our search for suitable binding modes. Once a conformation compatible with the high-affinity interaction site has been identified, the scaffold is designed with general RosettaDesign to hold these key interactions in place. Experimentally, we examine designed binders via yeast surface display, which allows us, in subsequent steps, to optimize novel binders using directed evolution methodologies on.