Abstract

P-glycoprotein is an ATP Binding Cassette (ABC) transporter, expressed at numerous sanctuary sites protecting the body against xenobiotics. Overexpression of P-glycoprotein in cancer cells is associated with a multidrug resistance phenotype. P-glycoprotein is one of the most promiscuous efflux pumps, expelling more than 300 structurally and functionally unrelated compounds. To understand this poly-specificity, numerous research teams across the globe have attempted to elucidate the pharmacophore of the drug. Pharmacological data demonstrated the presence of at least four distinct binding sites, but their locations and the mechanism of binding have not been fully elucidated. 

My PhD project contained two parts; the first one was to develop and test the polymer-based protein purification and reconstitution system. The need to devise new reconstitution systems is vital as the traditional lipid vesicles are prone to biochemical changes (oxygenation/hydrolysis) over time making them leaky. More recently, block copolymers have gained attention, offering vesicles with low permeability and high stability. The di-block copolymer PBd-PEO (Poly-buta-diene polyethylene oxide) has been used for the reconstitution of proteins previously. However, since various membrane proteins require lipids to function, hybrid vesicles with both polymer and lipids would be more appropriate. I tested the different ratios of a hybrid polymer-lipid (PBd-PEO and POPC) reconstitution system and found that a 1:1 molar ratio of polymer and lipid seemed best to begin with. However, P-glycoprotein was inactive after successful reconstitution. Changing the lipids to the total E.coli lipids resulted not only in successful reconstitution but also retained the function. 

The second and main aim of my PhD study was to define the binding sites for two transported substrates of P-glycoprotein; the anticancer drug vinblastine and the fluorescent probe rhodamine 123. I took a site-directed mutagenesis approach to mutate at least four residues in positions proximal to previously defined drug-interacting residues on P-glycoprotein for these substrates. The protein was purified and reconstituted into styrene-maleic acid lipid particles (SMALPs) to measure the apparent drug binding constant, or into liposomes for the assessment of drug-stimulated ATP hydrolysis. The biochemical data were reconciled with structural models of P-glycoprotein using molecular docking. The data indicated that the binding of rhodamine 123 occurred predominantly within the central cavity of P-glycoprotein. In contrast, the significantly more hydrophobic vinblastine bound to both the lipid-protein interface and within the central cavity. This suggests that the binding of vinca alkaloids occurs in a processive manner, starting at the lipid-protein interface before being internalised to the central cavity for export.