Computationally-guided design and affinity improvement of a protein binder targeting a specific site on HER2

A protein binder with a desired epitope and binding affinity is critical to the development of therapeutic agents. Here we present computationally-guided design improvement recognizing specific site on domain IV human epidermal growth factor receptor 2 (HER2). As model, scaffold composed Leucine-rich repeat (LRR) modules was used. We designed binders which appear bind target using computational method. Top 10 designs were expressed tested assays, lead low micro-molar selected. Binding selected further increased by two-orders magnitude through mutual feedback between experimental methods. The utility potential our approach demonstrated determining interface developed its crystal structure in complex HER2 IV.