Computational modelling of enzymes: predicting and understanding the selectivity of an epoxide hydrolase

Abstract

Enzymes provide improved catalytic efficiency, renewability and higher selectivity compared to traditional chemical synthesis methods. Stereoselective enzymes in particular are highly desired within industry for the production of fine chemicals and drugs. Epoxide hydrolases (EH) are a family of enzymes whose selective properties facilitate the production of valuable intermediates used for the synthesis of many pharmaceuticals. There is a wealth of structural and biochemical data available for EHs, however the origin of their selectivity remains uncertain. Computational and statistical methods can aid in the design and discovery of selective enzymes and interpretation of the experimental data produced during their characterisation. This thesis focuses on the EH from the fungus Aspergillus niger (AnEH) and its aim is two-fold. The first aim is to explore and develop machine learning methods to predict selectivity enhancing mutations for AnEH. The second aim is to get insight into the mechanistic determinants of AnEH selectivity through molecular modelling methods. Publications during candidatureBook chapters • Zaugg J., Gumulya Y., Gillam E. M. J. and Bodén M. (2014) Computational tools for directed evolution: a comparison of prospective and retrospective strategies. Methods in Molecular Biology 1179:315-333 Contributions by others to the thesis Chapter 2 Zaugg J (candidate) and Bodén M researched and wrote the manuscript. Zaugg J, Bodén M, Gumulya Y and Gillam E edited the manuscript. Chapter 3 Zaugg J (candidate) carried out all computational experiments. Gumulya Y provided experimental data. Bodén M and Malde AK provided guidance in experimental design and analysis of results. Chapter 4 Zaugg J (candidate) carried out all computational experiments. Gumulya Y provided experimental data. Bodén M provided guidance in experimental design and analysis of results. Bodén M wrote much of the software (Bnkit) used for creating and training Bayesian networks. Wang Y developed code to allow specification of Dirichlet priors for Bayesian networks. Essebier A provided code for testing Bayesian networks. Chapter 6 Zaugg J (candidate) designed and carried out all computational experiments. Gumulya Y provided experimental data. Chapter 7 Zaugg J (candidate) carried out all computational experiments and designed the protocol used for docking of ligands. Malde AK and Mark AE provided guidance in the design and implementation of molecular dynamics simulations and free energy calculations and the analysis of results. Gumulya Y provided experimental data. Zaugg J, Gumulya Y, Bodén M, Mark AE and Malde AK wrote and edited the manuscript. IX Statement of parts of the thesis submitted to qualify for the award of another degree None. X Research Involving Human or Animal Subjects No animal or human subjects were involved in this research. XI Abstract I List of Figures XVI List of Tables XIX