Bioscience seminar series

  • Date: –09:30
  • Location: Biomedicinskt centrum, BMC C4:2 Seminar room
  • Lecturer: Katarina Roos, Department of Cell and Molecular Biology, Uppsala University
  • Contact person: Anna Nilsson
  • Seminarium

Molecular modeling in drug design by free energy and quantum mechanical methods

An important role of computer aided drug design is to guide modifications of ligands to improve binding to a protein target, to aid in prioritizing which out of all possible virtual molecules should be selected for synthesis. A successful method must be able to describe the non-covalent interactions between the ligand and protein with high accuracy and at a reasonable time scale. I will describe our work on computational methods for predicting protein-ligand binding affinities with a focus on heterocycle variations, such as e.g. diverse core structures of drug-like molecules. 

Free energy perturbation (FEP) is one of the most rigorous methods to calculate protein-ligand free energy of binding, including dynamical effects. The interactions between the ligand and protein is computed using a classical force field, with parameters describing the properties of different atom types and how they interact. Although FEP has been shown to reproduce and predict the relative affinity for a wide range of systems, there are cases where polarization effects are important and the current standard classical force fields are inadequate. With a high level quantum mechanical (QM) method electronic effects can be described, however, extensive conformational sampling is at present computationally too expensive. 

We have showed that a QM approach, in combination with a representation of the binding pocket of a protein using only key residues, could reproduce the experimental relative binding affinity of core fragments with diverse variations extracted from inhibitors of drug targets such as e.g. beta-secretase, matrix metalloproteinase and mineralocorticoid receptor. However, in the general case sampling of multiple protein configurations is required. We have further explored and improved FEP methods for calculating protein-ligand free energy of binding. We identified deficiencies in the FEP classical force field description and developed a new force field, OPLS3e in collaboration with Schrodinger, with extended coverage for drug-like small molecules.

I will talk about different methods for molecular modeling and the application on predicting protein-ligand binding for a large number of targets and ligands, and with a special focus on challenging heterocycles.