A novel computational method for predicting the transmembrane structure of G-protein coupled receptors: application to human C5aR and C3aR.

A novel algorithm was applied to the sequences of bacteriorhodopsin (BRh), of rhodopsin (Rh), and of the two human anaphylatoxin receptors, C5a-receptor (hC5aR) and C3a-receptor (hC3aR), that predicts their transmembrane domains (TMD) according to energy criteria alone, on the basis of their sequences and a template structure for each. Two consecutive criteria were applied for the predictions: the first is hydrophobicity of a sequence of residues, which determines the candidate stretches of residues that form one of the transmembrane helices. The second criterion is an energy function composed of inter residue contact energies, of hydrophobic contributions due to membrane exposure and of the interactions of a few residues with the phospholipid head groups. The sequence of candidate residues for each helix is longer than that of the template, and is finally determined by threading each of the candidate stretches on each of the template helices and evaluating the energy for all possible configurations. Contact energies between residues were taken from a database (Miyazawa S and Jernigan RL (1996) J Mol Biol 256 623-44). The algorithm predicts well the TMD structure of BRh based on its own template, and the TMD structure of Rh conforms well with the model of Baldwin et al (Baldwin JM Schertler GFX and Unger VM (1997) J Biol Chem 272 144-64). Results for the construction of the TMD of hC5aR and hC3aR were compared, employing the template structure of Rh. Most of the results for these receptors are in accord with alignments and with mutation experiments on hC5aR and hC3aR. The predictions may serve as a basis for future mutagenesis experiments of these receptors.