Mapping Binding Landscapes of Different Trypsin/BPTI Complexes by Experimental and Computational Methods

Molecular interactions in protein-protein complexes have been studied for a long time. Yet, it remains unclear how the interplay between these interactions produces such large differences in PPI binding affinities. This study aims at mapping binding landscapes of four homologous complexes between trypsin-like proteases and their inhibitor BPTI. While structurally very similar, the four complexes span more than nine orders of magnitude in binding affinity. To map the binding landscapes for these interactions, we constructed a library of BPTI mutants containing all possible single mutations in the BPTI binding interface. Using the yeast surface display technology, we sorted the BPTI single mutants into four affinity groups when interacting with each of the four target proteases. The BPTI populations having high-, WT-like-, lower- and the lowest-affinity for each protease were collected and sequenced with the next generation sequencing. The frequency of each variant in each population was used to create comprehensive binding landscapes of BPTI interacting with four proteases. The generated binding landscapes correlated well with our computational predictions and with already published binding energy values for some purified BPTI mutants. In each protease/BPTI complex, we saw a particular pattern of hot- spots (i. e. positions where all mutations lead to affinity decrease) and cold-spots (i. e. positions where most mutations lead to affinity increase) and were able to analyze the conservation pattern of these hot and cold spots between homologous complexes. Our results bring invaluable insights on evolution of protein-protein interactions and facilitate design of novel protein-based therapeutic molecules.