Evolution of protein interactions: From interactomes to interfaces

•Interface stability is the most useful proxy for the conservation of an interaction.•Sequence conservation is a good proxy except in protein families with many paralogs.•Despite global interface conservation, epistasis causes high residue-level plasticity.•Many protein complexes can be modelled using template-based docking.•Conservation and coevolution can improve discrimination in template-free docking.

Protein–protein interactions lie at the heart of most cellular processes. Many experimental and computational studies aim to deepen our understanding of these interactions and improve our capacity to predict them. In this respect, the evolutionary perspective is most interesting, since the preservation of structure and function puts constraints on the evolution of proteins and their interactions. However, uncovering these constraints remains a challenge, and the description and detection of evolutionary signals in protein–protein interactions is currently a very active field of research. Here, we review recent works dissecting the mechanisms of protein–protein interaction evolution and exploring how to use evolutionary information to predict interactions, both at the global level of the interactome and at the detailed level of protein–protein interfaces. We first present to what extent protein–protein interactions are found to be conserved within interactomes and which properties can influence their conservation. We then discuss the evolutionary and co-evolutionary pressures applied on protein–protein interfaces. Finally, we describe how the computational prediction of interfaces can benefit from evolutionary inputs.