In a recent Journal of Physics: Condensed Matter Special issue on Molecular scale theories and simulations of fluid systems, Professor Tyler Luchko and his team propose a new methods for calculating small molecule hydration energy and entropy, with huge potential for in silico drug design. Read the authors perspective below.
Why water is important.
Just as water is essential for life it is essential in biomolecule computer simulations. Water is the medium in which proteins, DNA and other biomolecules function. It forms complex interactions with these other molecules through discrete hydrogen bonds (see figure 1), stabilizing specific structures. It also mediates interactions between biomolecules, acting as a bridge between the molecules for example. Furthermore, in order for molecules to make direct contact, water has to be removed from the interface.
For application to biochemical processes, such as binding affinity, reaction rates, or solubility, accurate treatment of molecular water, and especially the hydration free energy, is essential. Because of the cost involved in experimentally determining these properties, computer simulation is an essential tool for cost-effective modern drug development.
Challenges in simulating water at the molecular scale.
Because both the discrete molecular and bulk properties of water matter at this size scale, water is difficult to model. Every oxygen and hydrogen atom can be explicitly represented, providing the same level of description as the solute molecules, but this is computationally expensive. Simulating water will typically require 95% of the simulation time, and different properties require different simulation methods. To reduce this cost, water may be represented as a dielectric continuum. However, the molecular nature of water is lost. Some properties can be recovered with additional modeling and parameterization but this is a difficult and labour-intensive task.
3D-RISM as an approach.
An alternative is to calculate the equilibrium properties of an all-atom water model using statistical mechanics. The 3D reference interaction site model (3D-RISM) is a specific example of this approach, which has been the focus of much of our research. 3D-RISM calculates an approximate 3D density distribution of the water oxygen and hydrogen surrounding the biomolecules (see figure 1). From this distribution, all of the thermodynamic properties of the water-biomolecule interaction can be calculated.
For small, drug-like compounds, all hydration properties can be calculated in a single calculation. For larger biomolecules, 3D-RISM can be used to provide mean solvation forces as part of a molecular dynamics simulation. Our implementation of 3D-RISM is available as part of the open source AmberTools molecular modeling suite.
3D-RISM extended to efficiently calculate energies and entropies.
In our paper, we extend 3D-RISM to calculate the energetic and entropic components of the hydration free energy. Energies and entropies are extracted from the temperature derivative of the hydration free energy and provide information about what drives different interactions. We test our method calculating hydration free energies, energies, and entropies of small molecules, comparing them against enthalpies and entropies from experiment (the difference between enthalpies and entropies is small here). By including, and in some cases adapting, various existing corrections for 3D-RISM’s hydration free energy, we found good agreement between theory and experiment (see figure 2).
While this new method will allow the rapid calculation of these fundamental hydration properties, the free energy corrections employed are not transferable to other thermodynamic properties. To provide accurate predictions for all properties of solvation, the underlying approximations in 3D-RISM theory need to be addressed.
Jesse Johnson recently received his Ph.D. from Rutgers University where he studied implicit solvent models and ligand-protein interactions. He is currently working for Valve in Seattle, Washington.
David A. Case is in the Department of Chemistry and Chemical Biology at Rutgers University. He leads the development team for the Amber suite of biomolecular simulation programs, and has made contributions to understanding the connections between biomolecular structure and NMR and X-ray observables, to the development of implicit solvent models, and to the understanding of the electronic structure of active sites in metalloenzymes.
Takeshi Yamazaki is a Research Associate at Vancouver Prostate Centre, Vancouver, Canada. His research focuses on the application and development of computational chemistry methods to predict the structure and the thermodynamic properties of macromolecule in solution.
Sergey Gusarov is a Research Council Officer at the National Research Council of Canada’s National Institute for Nanotechnology in Edmonton, Canada. He has a strong theoretical background in material design, nano-electronics, and biochemistry. Dr. Gusarov’s research interests are focused on development and application of new computational methods of theoretical physics and chemistry, including statistical mechanical molecular theory of solvation and multi-configurational density functional theory, for rational design of complex materials, catalysts and devices.
Andriy Kovalenko is a Senior Research Officer at the National Institute for Nanotechnology and Adjunct Professor, Department of Mechanical Engineering, University of Alberta in Edmonton, Canada. His focus is the development of theoretical methods capable of predicting the behavior of nanosystems. He proposed the statistical-mechanical 3D molecular theory of solvation (a.k.a. 3D-RISM-KH), which bridges the gap between electronic structure, atomistic simulations, and system functioning.
This work is licensed under a Creative Commons Attribution 3.0 Unported License