Free energy simulations of biological systems

Thermochemistry of solvation plays important roles in numerous areas of biomedical, chemical, and industrial research.  The computation of ligand affinity to aid in drug design comprises one such area.  Another example is in the area of pharmacokinetics, where partitioning between various solvents and acid-base equilibria determine absorption properties.  Tautomeric equilibria of nucleic acid bases in solution are thought to affect the kinetics of DNA mutations by playing a part in abnormal base-pairing.  The evaluation of conformational free energies in solution can contribute to our understanding of the protein folding process, as well as to our ability to design molecules with useful structures, such as catalytic peptides or antigenic vaccines.  The problem of calculating relative stabilities of reactants versus products and their temperature dependence for reactions taking place in solution is also crucial to industrial process design.

Despite its importance, the calculation of solvation free energy presents a challenge to computational chemists.  Accurate solvation free energy is particularly difficult to evaluate for large systems, such as aqueous biosystems.   An accurate approach that has been applied to calculating solvation free energies is the use of either molecular dynamics (MD) or Monte Carlo (MC) simulations, for which the solute is surrounded with explicit solvent molecules.  Using this explicit solvent approach, solvation free energy differences between similar compounds can be obtained.  The displacement of water molecules surrounding the solute can be accounted for by integrating over a coupling parameter that varies as one solute molecule is gradually “mutated” into the other using the thermodynamic integration (TI) or free energy perturbation (FEP) technique. Although simulations have been demonstrated to produce acceptable solvation free energy differences for small organic molecules and peptides, these simulation methods are tedious and time consuming.  It is difficult to set up the path transforming one compound into the other, in which the two sets of parameters of the two solutes must be linked.  Also a large amount of time is required to converge all of the necessary simulations.

We have been developing a new methodology for simulations of solvation free energies from a single MD or MC calculations, without the need for performing free-energy simulations.  In the new method called Coupled Reference Interaction Site Model (RISM)/Simulation method, the thermochemistry of solvation can be determined from the RISM formalism, with the radial distribution functions obtained from either an MD or MC simulation, instead of by simultaneous solution of the RISM and closure relations.  This new methodology requires only one MD or MC simulation, and has shown to have a great potential in modeling large solvated systems such as biological systems, for which the high computational cost incurred makes free-energy simulation impractical.  Systematic applications of this method to several well-studied systems, for which different computational methods have been used and thus can provide an excellent basis for comparisons, shown that the coupled RISM/MD method has a comparable accuracy to the free energy simulation methods.  Applications to study biological systems such as RNA-small molecule complexes and others are being performed.