UCSF

Virtual screening is a powerful tool in drug discovery, with the potential to find novel ligands for therapeutically relevant target structures. However, the field is plagued by both false positive and false negative predictions. This is due to approximations within the scoring functions, leading to the failure to distinguish between true ligands and high-ranking nonbinders (decoys). To compound the problem, in a typical target the complexity of the ligand-receptor interactions prevents us from unraveling the many components of the binding energy that lead to the incorrect predictions. Model binding sites provide simpler systems in which individual terms can be isolated and studied.

In Chapter 1, cytochrome c peroxidase (CCP) W191G, an anionic, wet, and buried cavity is introduced. This cavity primarily binds aromatic monocations; dications and most neutral molecules do not bind detectably.

In Chapter 2, CCP W191G is included in a series of model systems (the T4 lysozyme L99A hydrophobic and L99A/M102Q polar cavities) to evaluate MM-GBSA rescoring of docking hit lists; both chapters consider the case for CCP W191G where the scoring function must balance the cost of ligand desolvation with the favorable electrostatic interaction energy between ligand and protein.

Chapter 3 returns to the T4 lysozyme L99A/M102Q polar models system for absolute and relative binding free energy predictions. This system proved to be difficult for the free energy methods, but not due to the additional polarity, as we had initially predicted. Instead, protein conformational change, sampling of reasonable ligand orientations and methodological failures proved a challenge to accurate predictions.

In the final chapter, a new open cavity in cytochrome c peroxidase, created by the W191GP190G deltaG192-A193 deletion mutant is introduced. The cavity contains multiple ordered waters and an interface to bulk solvent. This more complicated cavity presents an opportunity to investigate displacing individual ordered waters, and the potential for neutral ligands in a charged cavity. Implications for this new charged, open cavity and preliminary results are discussed in Chapter 4: Future Directions.