Pharmacophore Modeling And Related Database Building For GPCR Ligands

GPCR (G-protein coupled receptors) superfamily is one of the most important drug targets in human body. With the advances of computer technology, many new methods and techniques for medical study have employed in GPCR field. These methods and techniques provide visual representation of theories for pharmic experts and make medical research more convenient, intuitionistic and definite.In Chapter 1, the basic knowledge of GPCR and CADD was reviewed. Virtual screening and pharmacophore models were then introduced as two important approaches used in this thesis.Adenosine receptors are widespread throughout the body, they are involved in a variety of physiological processes and pathology including neurological, cardiovascular, inflammatory diseases and cancer. Wide expression and overlapping functions have been the major problems to designing specific adenosine agonists and antagonists with few associated negative side effects. Ligand-based methods were used to investigate the selective mechanism of antagonists of four adenosine receptor subtypes. Firstly 3D-pharmacophore models of A1 A2A A2B A3 were developed based on compounds carefully selected from a large number of adenosine receptor antagonists. Their correlation coefficient, root mean square deviation (RMSD), and cost difference values indicated the hypothesis we buit were reasonable and reliable. Then the crystal structure superimposition validation indicated A2A model well agreement with the crystal structure of A2A receptor. Finally, simulated virtual screening methods were implemented on A1 A2A A2B and A3 subtypes, Good enrichment factor of 6.51,11.0,7.1,6.90 were achieved in pharmacophore-based virtual screening, respectively These results demonstrated that our models could provide helpful informations and a robust tools for the discovery of potent adenosine receptor antagonists.To explore the selective mechanism of antagonists binding models of three opioid receptor subtypes, a serial molecular modeling methods including pharmacophore modeling, homology modeling and molecular docking were implemented together. At first three pharmacophore models were built for the antagonists of three opioid receptor subtypes and validated by the test set out of training compounds. The results of test set validation were 0.785 0.723 0.730, respectively. The best models for delta, kappa, mu antagonists contained three, foue, five features, respectively. Then homology models of three receptor subtypes were modeled on the basis of the crystal structure of beta2-adrenergic receptor, and molecular docking was conducted further. These findings may be crucial for the development of novel selective opioid receptor antagonists.To date, numerous efforts have been engaged into GPCR field and a significant number of diverse GPCR ligands have been discovered and developed for clinical use. These ligands give us valuable information for drug design. Managing and getting benefits from this useful information become urgent issues recently. In chapter four, a pharmacophore database called GPCR_PharmDB, which contains 122 records covering 11 GPCR families and 24 subset GPCR, was constructed. The methods to format and verify the collected pharmacophore models were also discussed. From the statistical data of the pharmacophore, we can see that 87% of these models contain hydrophobic feature,80% contain aromatic feature, and 65% contain Position ions. Then the implication of GPCR_PharmDB was discussed especially in pharmacophore based inverse virtual screening. GPCR_PharmDB will greatly facilitate the discovery of novel and selective GPCR ligands.The last chapter summarized the whole work.