Ab initio Prediction of the Conformation of Solvated and Adsorbed Proteins

Abstract

Proteins are among the most important groups of biomolecules, with their biological functions ranging from structural elements to signal transducers between cells. Apart from their biological role, phenomena related to protein behaviour in solutions and at solid interfaces can find a broad range of engineering applications such as in biomedical implants, scaffolds for artificial tissues, bioseparations, biomineralization and biosensors. For both biological and engineering applications, the functionality of a protein is directly related to its three-dimensional structure (i.e. conformation). Methods such as homology and threading that depend on a large database of existing experimental knowledge are the most popular means of predicting the conformation of proteins in their native environment. Lack of sufficient experimentally-derived information for non-native environments such as general solutions and solid interfaces prevents these knowledge-based methods being used for such environments. Resort must, instead, be made to so-called ab initio methods that rely upon knowledge of the primary sequence of the protein, its environment, and the physics of the interatomic interactions. The development of such methods for non-native environments is in its infancy – this thesis reports on the development of such a method and its application to proteins in water and at gas/solid and water/solid interfaces. After introducing the approach used – which is based on evolutionary algorithms (EAs) – we first report a study of polyalanine adsorbed at a gas/solid interface in which a switching behaviour is observed that, to our knowledge, has never been reported before. The next section reports work that shows the combination of the Langevin dipole (LD) solvent method with the Amber potential energy (PE) model is able to yield solvation energies comparable to those of more sophisticated methods at a fraction of the cost, and that the LD method is able to capture effects that arise from inhomogenities in the water structure such as H-bond bridges. The third section reports a study that shows that EA performance and optimal control parameters vary substantially with the PE model. The first three parts form the basis of the last part of the thesis, which reports pioneering work on predicting ab initio the conformation of proteins in solutions and at water/solid interfaces.