Climbing Up and Down Binding Landscapes through Deep Mutational Scanning of Three Homologous Protein-Protein Complexes.

Michael Heyne,Itay Cohen,Yoav Peleg,Julia M Shifman,Evette S Radisky,Niv Papo,Jason Shirian

doi:10.1021/jacs.1c08707

Abstract

Protein–protein interactions (PPIs) have evolved to display binding affinities that can support their function. As such, cognate and noncognate PPIs could be highly similar structurally but exhibit huge differences in binding affinities. To understand this phenomenon, we study three homologous protease–inhibitor PPIs that span 9 orders of magnitude in binding affinity. Using state-of-the-art methodology that combines protein randomization, affinity sorting, deep sequencing, and data normalization, we report quantitative binding landscapes consisting of ΔΔGbind values for the three PPIs, gleaned from tens of thousands of single and double mutations. We show that binding landscapes of the three complexes are strikingly different and depend on the PPI evolutionary optimality. We observe different patterns of couplings between mutations for the three PPIs with negative and positive epistasis appearing most frequently at hot-spot and cold-spot positions, respectively. The evolutionary trends observed here are likely to be universal to other biological complexes in the cell.

Highlights

Protein function is determined by the protein amino acid sequence, which has undergone billions of years of evolution while subjected to various selection pressures
BPTIWT was expressed on the surface of a yeast cell with a C-terminal myc-tag for monitoring protein expression through binding of an antimyc antibody and a secondary antibody conjugated to phycoerythrin (PE) (Figure 2A)
We explored the robustness and evolvability of the bovine pancreatic trypsin inhibitor (BPTI) sequence toward its main function, high-affinity binding to bovine trypsin (BT)

Summary

Introduction

Protein function is determined by the protein amino acid sequence, which has undergone billions of years of evolution while subjected to various selection pressures. Fitness landscapes explore the effects of all possible mutations on the ability of proteins to perform their main function Such landscapes reveal how far a particular protein is from its functional maximum, what fraction of mutations leads up and down the “fitness hill”, how large the mutational steps are, and which residues are the most critical to protein function.[12] Mapping of fitness landscapes is an attractive strategy for approaching various protein engineering projects with the goal to improve or modify protein function since the best mutations could be identified from the fitness landscape.[13,14] Development of new strategies for protein randomization and advances in next-generation sequencing (NGS) enabled several exciting studies that report fitness landscapes for a number of biological systems.[1,15−27] In these studies, the effects of mutations on enzyme catalysis, fluorescence, thermostability, and other functions have been reported, giving invaluable insights on how different biological functions have evolved

Objectives

Methods

Results

Conclusion