Structure-based prediction of Ras-effector binding affinities and design of 'branchegetic' interface mutations

Abstract

Ras is a central cellular hub protein controlling multiple cell fates. How Ras interacts with a variety of potential effector proteins is relatively unexplored, with only some key effectors characterized in great detail. Here, we have used homology modelling based on X-ray and AlphaFold2 templates to build structural models for 54 Ras-effectors complexes. These models were used to estimate binding affinities using a supervised learning regressor. Furthermore, we systematically introduced Ras branch-pruning (or branchegetic) mutations to identify 200 interface mutations that affect the binding energy with at least one of the model structures. The impacts of these branchegetic mutants were integrated into a mathematical model to assess the potential for rewiring interactions at the Ras hub on a systems level. These findings have provided a quantitative understanding of Ras-effector interfaces and their impact on systems properties of a key cellular hub.