| Summary: | Membrane proteins are an important class of drug targets, making up at least 25% of proteins in the human genome. In this thesis I investigated two aspects of alpha-helical membrane protein structures. Firstly, I investigated kinks in alpha-helices, many of which are thought to have functional roles. Kinks are changes of direction in helices, often defined in a binary fashion, but here I move towards defining them on a continuum. I found that kink angles are not generally a conserved property of homologues, pointing either to their not being functionally critical or to their function being related to conformational flexibility. I found correlation in kink angles and conformational change upon activation in GPCRs, reinforcing the belief that helix kinks are key, functional, flexible points in structures. Secondly, I turned to the biogenesis of alpha-helical membrane proteins, and how this might be used to improve structure prediction. These proteins are inserted into the membrane during the process of translation by the ribosome, therefore the N-terminus may be able to adopt its tertiary fold before the C-terminus is translated. I found a weak signal in a non-redundant set of structures that membrane proteins exhibit asymmetry between the N- and C-termini. This might be expected if they are folding cotranslationally, as had been seen in soluble proteins. Motivated by this, I predicted the structures of membrane proteins using SAINT2, a cotranslational structure prediction program, and achieved promising results. I developed SAINT2-ScafFold, which folds proteins around a rigid N-terminus, but the accuracy of prediction of the remaining protein was no better than when the entire chain was sampled. A membrane potential was implemented in SAINT2, which slightly improves the accuracy of models generated. Finally, the SAINT2-ScafFold method was applied to the completion of homology models that do not cover the entire target. An RMSD of less than 5 Å was achieved in more than half of the cases where a terminal transmembrane helix of membrane protein structures was predicted. This was an encouraging result for the prediction of membrane proteins from partial templates, and could easily be extended to soluble proteins.
|
|---|
