A theoretical model of the human thrombin receptor (PAR-1), the first known protease-activated g-protein-coupled receptor

Abstract A three-dimensional molecular model of the human thrombin receptor (TR; PAR-1), a member of the seven-transmembrane G-protein-coupled receptor (GPCR) superfamily of glycoproteins and the first GPCR to have a tethered-peptide ligand (ref. 7), was constructed by using de novo computer-based modeling techniques. Because of numerous short comings associated with the use of the bacteriorhodopsin template for building GPCR models, we employed an alternative modeling strategy involving GPCR alignment and topological properties, as first described by Moereels (ref. 10). To initiate assembly of the seven-helix bundle for PAR-1, we relied on specific, conserved residues in the transmembrane (TM) domains, which in this instance are Asn-120 in TM1, Asp-148 in TM2, and Asp-367 in TM7. The side chains of these residues then comprise a hydrogen-bonding network that orients the three helices so that they are properly positioned to interact with each other. The energy-minimized three-helix bundle (TM1-TM2-TM7) served as a cornerstone for assembly of the complete seven-helix bundle, through systematic addition of TM3 through TM6 involving suitable orientation followed by limited translation, rotation, and tilting. The three extracellular (EC) loops were added to the energy-minimized, seven-helix construct via loop-search routines, and the disulfide bridge between two GPCR-conserved cysteines on loops EC1 and EC2 was formed using the protein manipulation software program SCULPT. Finally, part of the extracellular N-terminus, from Tyr-95 (at the extracellular end of TM1) to Asn-75, was attached to the receptor. The known structure-activity relationship surrounding the thrombin receptor agonist peptide motif SFLLRN, coupled with the results from our receptor-based, site-directed mutagenesis, were used to define and enhance our computer-generated, energy-minimized models of the ligand-bound thrombin receptor. We examined the acidic residues in the three extracellular loops (seven total: D167 and E173 in EC1; E241, D256, E260, and E264 in EC2; and E347 in EC3) to identify electrostatic interactions that could account for the ammonium N-terminus and/or the Arg residue of SFLLRN. When both EC1 acidic residues were mutated to Ala, there was no change in receptor functional responses. However, when all four acid residues in EC2 were mutated to Ala, there was a significant reduction in receptor activation in the presence of thrombin or agonist peptide. When the single acidic residue in EC3 was mutated to Ala, reduced functional activity was also observed. Comtrary to published observations (refs. 20 and 24), we were unable to find an important role for Glu-260 (E260A mutant TR), whereas replacement of Asp-256 (D256A mutant TR) had a significant effect on receptor function, especially in response to the TR-agonist peptide SFLLRNP-NH 2 (TRAP-7). Our mutagenesis work also confirmed the importance of the peptide triad Asn-120/Asp-148/Asp-367 (viz. refs. 10 and 18) and the C-175/C-254 disulfide bridge (viz. ref. 25) in receptor function. On the basis of our studies, we suggest that the ammonium group at the N-terminus of the peptide ligand interacts with the carboxylic acid side chain of Glu-347, at the extracellular surface of TM7, and that the aromatic ring of the ligand's critical Phe residue binds within a hydrophobic pocket on the receptor defined by the side chains of Phe-182, Leu-340, Tyr-337, and Phe-339.