Characterisation of lipid-protein interactions through computational modelling

Thesis

Parallel advances in lipidomics and the structural biology of membrane proteins over the past decade have revealed the highly diverse and complex nature of cell membrane composition. These compositional complexities influence the behaviour of embedded proteins, and a number of biomedically important membrane proteins are now known to be modulated via interactions with their lipid environment. This project aims to apply an array of computational tools to probe and characterize the molecular level detail of these interactions, and obtain information to complement experimental studies. In particular the interactions of modulatory lipid species with the transmembrane domain of the epidermal growth factor (EGF) receptor were addressed using coarse-grained potential of mean force (PMF) calculations. The results suggest this approach, widely applied in other areas of computational biology, may be successfully adapted to probe the strength and selectivity of lipid-protein interactions, and the effects of protein mutation. Subsequently these calculations were applied in tandem with equilibrium simulations to explore cardiolipin interactions with a key mitochondrial transporter, the ADP/ATP Translocase. The data show the coarse-grained model employed is capable of accurately identifying three specific cardiolipin binding sites on the protein, in agreement with prior crystallographic and NMR data. These sites were shown to be specific to cardiolipin, rather than general phospholipid interaction sites. In the third project, prospective predictions were made as to the location of PIP2 lipid binding sites on the membrane-exposed surface of a class A GPCR, the Neurotensin Receptor 1 (NTS1). A protein known to specifically bind these lipids from native mass spectrometry (MS) measurements at hitherto unknown regions. The results suggest PIP2 binds to defined loop regions on the intracellular portion of the protein. Finally, atomistic simulations are applied to examine the effect of a crystallographically resolved cholesterol molecule on the dynamics of a class F GPCR, Smoothened. The data suggest a marked influence of bound cholesterol on intra-domain dynamics of the extracellular region of Smoothened.

Authors: Hedger, G

• DOI: 10.5287/ora-nrkkm4jk4
• Type of award: DPhil
• Level of award: Doctoral
• Awarding institution: University of Oxford