Computational studies of metalloproteins, quantum refinement and ligand binding
Abstract
We develop and apply theoretical methods to study the structure and function of metalloproteins with high scientific, medicinal and industrial interest. For example, we study the reaction mechanisms of nitrogenases, lytic polysaccharide monooxygenase and particulate methane monooxygenase. We will also calibrate methods to calculate redox potentials acid constants in proteins. We will compare different approaches to study enzymes with computational methods and we will develop methods to calculate charges for molecular dynamics simulations. The project builds on the unique methods developed in our group, viz. methods to combine quantum mechanical (QM) and molecular mechanics (MM) calculations, as well as method to calculate free energies at the QM/MM level. In particular, we use calculations with big QM systems (600–1200 atoms) to obtain stable energies.
X-ray crystallography is the main source of structural information for proteins. With the European Spallation Source (ESS), the hope is that neutron crystallography will become an important complement. We have long developed methods (called quantum refinement, QR) to use computational chemistry to interpret, complement and improve macromolecular crystal structures. We will continue this work by applying QR to structures of high scientific interest, e.g. hydrogenase and photosystem II. We will extend this approach to X-ray free-electron laser, electron diffraction and time-resolved serial crystallography. We will also develop methods to simplify and speed up the refinement of crystal structures, e.g. by identifying positions of deuterons in neutron structures or water molecules in X-ray structures. Moreover, we develop methods to improve the atomic scattering factors (Hirshfeld atom refinement) to be applicable for whole proteins.
We also develop and improve methods to predict the binding free energy of drug candidates to a macromolecular receptor. This is one of the greatest challenges in drug development: If the binding affinity could be accurately predicted, the synthesis of most of the drug candidates could be avoided, which could save an enormous amount of money and time. We will develop methods to perform FEP calculations for proteins where the ligand binds to a metal. We will also develop approaches based on QM or QM/MM geometry optimisations and we will develop machine-learning methods to improve the results of various computational ligand-binding approaches.
