Computer-aided Binding Mode Prediction and Affinity Maturation of LRR Protein 1 Binder without Structural Determination

27 Binding affinity maturation without structure determination remains a difficult challenge in the 28 computer-aided protein engineering. Precise binding mode identification is a vital prerequisite for 29 the affinity maturation. However, pure computational methods have been unreliable in practice so 30 far and experimental structural biology techniques are generally too costly. Herein, we show that 31 computational epitope localization followed by the full-atom energy minimization with 32 intermediate experimental validation can yield precisely bound complex model structure, which 33 ultimately enables effective affinity maturation and redesign of binding specificity. As a proof-of-34 concept, we targeted a leucine-rich repeat (LRR) protein binder which specifically binds to the 35 human IgG 1 (hIgG 1 ). Based on the computationally predicted binding mode of the LRR protein 36 binder to hIgG 1 , the binding affinity of the protein binder was significantly increased and its 37 specificity was redesigned toward multiple IgGs from other species. Experimental determination 38 of the complex structure showed that the predicted model closely matched the X-ray crystal 39 structure. Through the benchmark of therapeutically relevant existing LRR protein complexes, we 40 demonstrate that the present approach can be broadly applicable to other proteins which undergo 41 small conformational changes upon binding. 42