Modelling protein unfolding: a solvent insertion protocol

Abstract

A novel protocol for modelling the unfolding pathway has been developed which focuses on the protein–solvent interface. The method is based on a cavity search algorithm, PRO-ACT, which allows us to locate and define cavities within native protein structures. By comparison with experimental data from X-ray crystallography on buried solvent molecules in proteins, we have found that the cavities which are likely to be occupied by solvent are larger and have a more polar surface. By placing solvent molecules into cavities which may not normally be fully hydrated and then relaxing the structure with short molecular-dynamics simulations, we are able to force the protein to gently unfold. We present our method and results on the unfolding of a small protein, barnase, which has been particularly well characterised experimentally. We find that the partially unfolded structure has properties that are consistent with models for ‘molten globules’ which have been used to describe the partially unfolded states of many proteins and that it is consistent with the experimental data on the unfolding pathway of barnase. Our method shows (i) that there is a clear role for solvent in the unfolding of a small globular protein and (ii) that modelling methods can be devised which allow unfolding using relatively small computational resources.