Macromolecular Modeling Blog ™

Incorporation of flexibility into docking simulations and accounting for backbone conformational changes during association is probably the toughest problem protein-protein dockers are facing today. Normal Mode Analysis is lately the answer of many groups to this problem. We hereby review two approaches of incorporating NMA into docking simulations, reflecting the induced fit vs. conformational sampling association mechanisms. 

Two weeks ago I attended the EMBO practical course on Docking Predictions of Protein-Protein Interactions which was conducted at the nicely situated Barcelona Supercomputing Center. The course, organized by Prof. Joel Janin (Universite Paris-Sud, Orsay, France) and Dr. Juan Fernandez-Recio (BSC, Barcelona, Spain), gathered together some of the leading figures in the computational protein-protein docking field to give a very insightful and interesting review of the accomplishments, theory and challenges remaining in protein-protein docking. 

One of the hottest topics that seemed to re-emerge during the 4 days workshop was the incorporation of Normal Mode Analysis (NMA) in order to account for backbone conformational changes between the bound and unbound monomers during the association process. This topic was covered during talks by Prof. Martin Zacharias (Jacobs University, Bremen, Germany) and Prof. Michael Sternberg (Imperial College, London, England) as well as during table-side conversations. I will try to briefly review the subject based on these talks and references provided by the speakers.

Upon protein-protein association, a large variety of conformational changes can occuer. Ranging from side chains motion at the time scales of 0.1ps to 0.1ns and up to domain motions that can take even 1ms. Two mechanisms have been suggested to account for these motions: 1. During the association process the interacting molecules can adapt to each other, inducing conformational changes during the binding process (induced fit binding mechanism. see Koshland, D. E.). 2. The unbound monomers might sample the bound conformation even in the absence of their binding partner, each protein adopts a range of conformations and the recognition step primarily involves an appropriate pairwise matching between the ensembles of conformations (conformational sampling/selection binding mechanism. see Ma, B. et al.).

Ideally, computational approaches to protein-protein docking should include such conformational changes during the docking simulations. In principle, it is possible to perform docking searches allowing for conformational changes of both protein partners in all degrees of freedom, However this is not yet computationally feasible (for generic systems) and may also present a problem of creating too much false positive solutions.

NMA: Normal modes of vibration are simple harmonic oscillations characterizing the dynamics of the system around an energy minimum. Each mode describes a state of the system where all particles are oscillating with the same characteristic frequency. The dynamic behavior of most systems of physical interest may be approximated by a linear combination of normal modes. The vectors calculated in NMA describe the direction and relative magnitude of each atom’s motion. The overall scale of motion is not defined and depends on the conditions of the system, e.g., temperature. The atomic fluctuations will be greatest for low-energy modes (soft modes).

NMA has been used for more than three decades to study protein flexibility,it has many limitations such as use of the harmonic approximation, neglect of solvent damping, and an inability to model energy barriers and multiple minima (Elber R. & Karplus M.,Frauenfelder H et al.,Kitao A. et al.). Furthermore, the trajectory of protein motion will rarely be in a straight line. Despite these limitations, many studies have found agreement between features of the motion predicted by NMA and the observed or simulated conformational change of one or a small number of proteins. 

The first approach implemented by the Zacharias group (ATTRACT docking software) working under the assumption that the deformation directions may overlap with “soft collective degrees of freedom” of the protein partners, is incorporating possible movements in the modes’ directions during the docking process, in a way simulating an induced fit process. They employ: 

• Pre-calculation of the soft modes of the protein partners: either by principal component analysis of an MD simulation or by an approximate normal mode calculation. 
• Structure relaxation in soft degrees of freedom during docking by energy minimization. 

This have worked nicely in both for protein-ligand docking and protein-protein docking (see Zacharias & Sklenar, Zacharias, May & Zacharias, May & Zacharias). Amongst others to incorporate NMA to model flexibility in docking: Lindahl E. et al., Cavasotto CN et al.,Schneidman-Duhovny D et al.).

The second approach investigated by the Sternberg group and presented in a paper by Dobbins, S. E. et al. uses coarse-grained normal mode simulations (also used by: Hinsen, K.,Tobi D. & Bahar I.) which takes into account C-alpha atoms only. The normal modes of a benchmark set of 20 proteins observed to undergo substantial conformational change on docking were calculated. The study evaluated the extent to which normal mode simulations can reproduce the location and the direction of observed conformational change.

In 35% of the proteins studied, a single low-frequency normal mode was found that describes well the direction of the observed conformational change. Furthermore, they found that for a set of 134 proteins from a docking benchmark, the characteristic frequencies of normal modes can be used to predict reliably the extent of observed conformational change, which provides guidelines as to whether a protein is likely to undergo substantial conformation change on docking.

This research bodes strongly of the important role conformational selection plays in the recognition process. (as well as Gunasekaran K. & Nussinov R., Henzler-Wildman K. A. et al.,Lange O. F. et al.) and is suggestive that running a standard docking procedure with a conformation selected by NMA before the docking is ran, might do the job.

It is worth mentioning that other approaches exists for modeling flexibility of the complex such as the use of detailed molecular dynamics simulations to probe flexibility (see Grunberg, R., et al., Smith, G. R. et al.). Systematic conformational generator approaches, based on peptide backbone segments/systematic dihedral angle sampling/stable side chain rotamerstates (see Bruccoleri & Karplus Biopolymers 1987) Monte Carlo simulations (see Wang C. et al.) and of course soft docking approaches. (A good review on incorporating flexibility into docking was written by Andrusier N. et al)

To conclude, NMA is rapidly gaining popularity in the modeling world and specifically in the flexible docking domain, new algorithms and implementations for including NMA results into simulations are likely to be of need in the near future.  By Nir London.