Development of Prediction Method for GPCR­G-protein Coupling Selectivity Using Amino Acid Properties

ABSTRACT

We describe a novel method for predicting G-protein coupled receptor (GPCR) - G-protein coupling selectivity using amino acid properties of specific residues in GPCR sequences. We have evaluated various amino acid properties obtained with experimental or theoretical studies. The GPCRs having reliable G-protein binding information were collected from Guide to Receptors and Channels (GRAC) and gpDB databases, and these sequences were aligned with the amino acid sequence of bovine rhodopsin, whose structure is known, to identify positions of each amino acid residue and its secondary structure. The collected properties were used as feature values to calculate Fisher’s ratio (FR) for each residue in GPCRs related with rhodopsin residue numbers. Some amino acid properties with high FR value were picked up as the effective characteristics for selecting G-protein type, and they were used as feature vectors in support vector machine (SVM) to predict GPCR - G-protein coupling selectivity. Applying this method to known GPCR sequences, each binding G-protein is predicted with high sensitivity and specificity of more than 96%. This result strongly suggests that the amino acid properties of specific residues are appreciably important for GPCR - G-protein coupling to determine Gprotein binding selectivity and our method could be an effective tool to investigate the mechanism of GPCR - G-protein coupling through site directed mutagenesis experiments.

Keywords: