Skip to main navigationSkip to main content
The University of Southampton

Research project: Essex: The simulation of protein-ligand systems

Currently Active: 

Protein-ligand interactions are critical in determining the effectiveness of pharmaceutical compounds, and computer simulations are ideally suited to probing these interactions.

There are essentially two problems to be solved: First, is it possible to predict the structure of a protein-ligand complex given the protein structure, and second, can the binding free energy of the resulting complex be accurately calculated? If these objectives can be met, then the reliable use of structure to predict novel pharmaceutical compounds will become a reality. In the area of structure prediction, most existing methods invoke a rigid receptor hypothesis, whereas the algorithm we have developed incorporates receptor flexibility (J. Comput. Chem. 24, 2003, 1637-1656). This method is currently undergoing further validation and testing, and will ultimately be extended to allow explicit, structurally critical, water molecules to be incorporated as the calculation proceeds.

Predicted and experimental protein-ligand structures
Click to expand

Regarding the prediction of binding constants, there are two main limiting factors affecting existing methods. First, for the prediction to be reliable, the inter and intramolecular forces must be described accurately. Second, the simulation must sample sufficient configuration space to generate a representative ensemble of structures. The former problem is very significant, and perhaps ultimately only soluble through the use of quantum mechanical methods of energy evaluation. We are currently investigating these approaches. Regarding the sampling problem, we have developed a novel application of the Replica Exchange approach to free energy calculations (J. Phys. Chem. B 107, 2003, 13703-13710). In free energy calculations, the overall simulation is sub-divided into a series of isolated calculations, the results of which are then summed to yield the final free energy. The Replica Exchange methodology we have developed allows the configurations seen by each sub-simulation to be shared with all others, thereby markedly improving efficiency and precision. The application of the approach to handling the rare configurations arising from water-contaminated organic solvent is particularly impressive (J. Phys. Chem. B 107, 2003, 13711-13718). Through the careful development and parameterisation of a continuum solvent model (J. Comp.Chem. 25, 2004, 1760-1770), and its efficient implementation in our Monte Carlo code (J. Chem. Theory Comput. 2, 2006, 732-739), we are now able to calculate binding affinities using our Replica Exchange approach very rapidly, with stable rank orderings of the ligands being achieved within a few hours (J. Med. Chem. 49, 2006, 7427-7439). As a result, binding free energy calculations are now within the reach of the pharmaceutical industry. As another application of this technology, we have examined the role of tightly-bound water molecules in mediating protein-ligand interactions, by calculating their binding affinities, correlating these results with whether or not the water is known to be displaced, and developing a simple structural model to predict these calculated data (J. Am. Chem. Soc. 129, 2007, 2577-2587). The objective of this work is to predict which waters may be usefully displaced as part of a rational drug design process. Finally, to extend these methods to calculating the binding free energies of structurally diverse molecules, we have developed new approaches (J. Chem. Theory Comput. 3, 2007, 1645-1655) based on dual-topology methods, and have successfully applied these to binding affinity predictions in a de novo design context.

Related research groups

Computational Systems Chemistry
Share this research project Share this on Facebook Share this on Twitter Share this on Weibo
Privacy Settings