Analysis and refinement of criteria for predicting the structure and relative orientations of transmembranal helical domains.

We are interested in modeling the membrane-spanning domain of the serotonin 5-HT1A G-protein coupled receptor. This superfamily of proteins is predicted to share the topology of the seven transmembrane helices of bacteriorhodopsin (BR), even though no significant sequence homology had been identified. We found significant homologies by allowing for helix shuffling corresponding to minimal exon shuffling during evolution. Consequently, our strategy for building the model for the 5-HT1A receptor has been to construct hypotheses concerning helix-helix interactions, their orientations, and arrangement in bundles surrounded by lipid, based on the 3.5 A resolution structure of BR. Inferences resulting from such models were tested against the 2.3 A resolution structure of the photosynthetic reaction center (PRC) from Rhodobacter Viridis. These comparisons led us to a reevaluation of current methods for the identification and topological orientation of membrane-embedded alpha-helices. We find that methods used currently in the construction of helical transmembrane domains could be misleading if used indiscriminately. These methods include the hydrophobicity profile, the hydrophobic moment, helix amphiphilicity, and charge neutralization. A refinement is proposed here, based on empirical observations, molecular modeling, and physicochemical considerations designed to overcome some of the shortcomings inherent in the use of the above mentioned methods. Here we present the analysis of two of the motifs identified in our study that led to the proposed refinements: the distribution of acidic and basic residues in the transmembranal domains, and the kink induced by a Pro residue in an alpha-helix.