Protein Flexibility in Drug Discovery: From Theory to Computation

Abstract

Nowadays it is widely accepted that the mechanisms of biomolecular recognition are strongly coupled to the intrinsic dynamic of proteins. In past years, this evidence has prompted the development of theoretical models of recognition able to describe ligand binding assisted by protein conformational changes. On a different perspective, the need to take into account protein flexibility in structure-based drug discovery has stimulated the development of several and extremely diversified computational methods. Herein, on the basis of a parallel between the major recognition models and the simulation strategies used to account for protein flexibility in ligand binding, we sort out and describe the most innovative and promising implementations for structure-based drug discovery.

Understanding the challenges of protein flexibility in drug design

“…According to this model, the ligand selectively stabilizes specific unbound-receptor conformations, causing a shift in the population toward a particular bound-receptor conformation [27]. In other words, changes in the population are dictated by the laws of statistical mechanics, and the underlying energy landscape is reshaped upon ligand binding [25], [30].…”

“…These two biomolecular recognition models have often been used to provide protein-ligand docking methods with some theoretical justification, and to classify these methods [9], [25]. In fact, direct analogies can be drawn between these theoretical models and some of the docking approaches accounting for protein flexibility.…”

“…It does not allow for larger conformational changes, such as large side-chain rotations or backbone motions [34], [35]. Another drawback of soft docking is its tendency to increase the rate of false positives [25], [36]. Nowadays, soft docking is used only rarely on its own, but sometimes as a first docking step within more complex methods (see Section II-D) [37], [38].…”

“…Considering target flexibility has a direct impact on docking experiments, potentially leading to different results, and consequently having major implications for drug discovery [22], [23]. Therefore, protein flexibility has become a major goal in molecular docking, as acknowledged by several recent reviews [5], [24], [25]. Because of the diversity of docking methods accounting for protein flexibility, general classifications are useful to provide overviews of shared concepts and applications.…”

“…Because of the diversity of docking methods accounting for protein flexibility, general classifications are useful to provide overviews of shared concepts and applications. In this context, analogies with the major theories underlying biomolecular recognition (i.e., induced fit [26] and conformational selection [27]) are sometimes used to classify these docking methods [9], [25]. Although convenient, this classification can become arbitrary, as several methods are clearly blurring the boundary between both categories [25].…”