Structure prediction of small transmembrane helix bundles.

In this work, we will introduce a novel computational approach to predict the structures of small helical hetero-oligomeric transmembrane bundles. The approach is based on the generation and evaluation of a large library of randomly generated helix bundle conformations. This library will be evaluated by energy-dependent distributions of the structural parameters of the conformations. The approach enables us to model a subunit of cytochrome c oxidase (occ), consisting of four TM helices, to an accuracy of 1.7A and the transducer protein of the sensory Rhodopsin II-transducer complex to an accuracy of 2.3A when including two transducer-contacting Rhodopsin helices. As the approach does not afford a unique solution for each protein, experimental data would be needed to discriminate the possible models. In addition to predicting the structure of helix bundles, one can also gain insight into possible higher-energy conformations or flexible regions of the protein.