Abstract

The ABC transporter P-glycoprotein (P-gp) is responsible for the translocation of a broad variety of different substances across the lipid bilayer. As a result, P-gp on the one hand exhibits important barrier and detoxifying functions, on the other hand it affects the ADMET properties of drugs, triggers unwanted drug-drug interactions and its overexpression is responsible for the development of multidrug-resistance (MDR), one major reason for the failure of antibiotic or anticancer therapies. Thus, reliable in silico methods for the early identification of P-gp modulators or virtual screening approaches for discovering new inhibitors is of high interest in drug discovery. This thesis outlines in three independent studies how structure-based methods can be used for tackling the problems triggered by P-gp. Although structure-based design has to be performed with precaution when dealing with membrane proteins, it is highly necessary for understanding intermolecular interactions between the ligand and its target protein. Two studies are presented, that describe the use of docking for the identification of binding modes of two classes of P-gp inhibitors. The workflows applied show how the implemention of external information is able to reduce the debatable use of scoring functions to a minimum. The results obtained provided useful information for screening for new P-gp inhibitors. Furthermore, a third study that compares the classification performance of ligand-based and structure-based in silico methods is also included in this thesis. Although the former models tend to be more accurate, the implementation of some ligand information could improve the structure-based models, emphasizing the importance of combined scoring functions (merging scoring and descriptor values) for future studies.