Protein structure elucidation from computational techniques and sparse EPR data
Alexander, Nathan Scott
Computational methods that allow application of electron paramagnetic resonance (EPR) spectroscopy data for protein structure prediction were developed. An implicit cone-model for the spin label enabled EPR distance information to be incorporated into structure prediction methods. The small, soluble proteins T4-lysozyme and αA-crystallin were used to demonstrate the ability of EPR measurements to guide atomic-detail protein structure prediction. In addition, a spin label rotamer library was developed and incorporated into Rosetta, allowing structural interpretation of distances and dynamics observed through EPR. These methods were applied to investigate the overall structure of the GPCR rhodopsin in complex with the conjugate G-protein transducin, as well as de novo protein structure prediction of membrane proteins. The results show the ability to utilize data from EPR to aid in the prediction of membrane protein structures approaching atomic-detail accuracy. In addition, significant conformational changes were predicted to occur as transducin binds to rhodopsin, and the formation and disruption of stabilizing residue interactions were mapped according to Rosetta energy changes.