Mobility-based prediction of hydration structures of protein surfaces

Hydration largely determines solubility, aggregation of proteins and influences interactions between proteins and drug molecules. Despite the importance of hydration, structural determination of hydration structure of protein surfaces is still challenging from both experimental and theoretical viewpoints. The precision of experimental measurements is often affected by fluctuations and mobility of water molecules resulting in uncertain assignment of water positions.

Our method can utilize mobility as an information source for the prediction of hydration structure. The necessary information can be produced by molecular dynamics simulations accounting for all atomic interactions including water-water contacts. The predictions were validated and tested by comparison to more than 1500 crystallographic water positions in 20 hydrated protein molecules including enzymes of biomedical importance such as cyclin-dependent kinase 2. The agreement with experimental water positions was larger than 80% on average. The predictions can be particularly useful in situations where no or limited experimental knowledge is available on hydration structures of molecular surfaces.

The method is implemented in a standalone C program MobyWat released under the GNU General Public License, freely accessible with full documentation at http://www.mobywat.com.