Molecular docking: Application of current methodologies to peroxisome proliferator activated receptors

Computational molecular docking methods seek to determine structures of molecular complexes from the structures of their components. This thesis examines aspects of docking methodology in the context of applications to specific systems. It presents improvements in docking methodology as well as new approaches to scoring docked complexes of putative ligands with various receptors.; Peroxisome proliferator activated receptors (PPARgamma) constitute a therapeutically important family of nuclear receptors. We investigated a number of known PPARgamma structures using computational solvent mapping and identified conformational changes in the ligand-binding domain upon ligand binding. Molecular docking of specific ligands to PPARgamma crystal structures, determined with different ligands, demonstrated the typical difficulties associated with "induced-fit" conformation changes in a receptor and the need for adequate conformational sampling.; Phthalate esters constitute potential xenobiotic ligands of PPARgamma, inducing transactivation upon binding---an important feature of their environmental toxicity. Using molecular docking and free energy calculations we developed a method to identify novel phthalate ligands of PPARgamma and validated it using several agonists of PPARgamma, whose binding orientations were correctly reproduced. Rescoring, using an improved scoring function (CHARMM/ACE), yielded a better correlation than initial docking results between experimentally determined EC50 values and the calculated lowest binding energies. Several new putative PPARgamma-binding phthalates were identified by applying this method to screen 73 mono-ortho-phthalate esters listed in the Available Chemicals Directory. These constitute important subjects for future toxicity studies. A molecular-dynamics analysis of the PPARgamma binding pocket revealed conformational changes in sidechain orientations due to induced-fit binding. Re-docking using the predominant orientations, however, did not improve docking accuracy.; We have applied several scoring functions to a diverse set of other protein-ligand complexes to assess the accuracy of molecular docking and scoring as a means to correctly reproduce the crystal structures of receptor-ligand complexes. A new approach based on clustering docked ligand orientations was shown to perform at least as accurately as traditional scoring functions. The decoy-set produced as part of this study will be further utilized in the development of a new improved scoring function, DARS (Decoys As the Reference State) potential. This work demonstrates the scale of applicability of recent molecular docking methods and potential pitfalls that need to be addressed.