A computational approach for analysis of communication in proteins

Abstract

Allostery and communication in proteins are crucial for the regulation of protein functions. The binding of a ligand changes the affinity of the protein at a distal ligand binding site and leads to conformational changes which are important for its function. In other words, the functional motions of the proteins reveal the communication patterns that are inherent to native architecture. The discrete-time, discrete-state Markov process applied on the network of interacting residues determines the potential pathways of signal transduction. This novel method involves the evaluation of two basic quantities: hitting and commute times, which express the information flow between the residue pairs. Furthermore maximum likelihood communication pathways are computed based on the Markov transition probabilities. This methodology is applied to adenylate kinase, triosephosphate isomerase, PDZ signaling protein and Cdc25B. As a result, the key interactions and the communication ability of the residues in the selected networks are ascertained. It is shown that the catalytic residues are located at the minima of the mean commute time curves, which indicates their efficient signal transduction abilities. Mobile residues are found to be slow communicators and the residues in the protein core are illustrated to be efficient communicators of the network. Besides, the residues on the maximum likelihood pathways are, at large, evolutionarily conserved; this issue confirms the functional importance of the residues on the pathways. In summary, the new methodology provides insights into understanding of communication patterns in proteins.