A peptide-based structure-activity study is reported leading to the discovery of novel potent thrombin receptor antagonists. Systematic substitution of nonproteogenic amino acids for the second and third residues of the human thrombin receptor "tethered ligand" sequence (SFLLR) led to a series of agonists with enhanced potency. The most potent pentapeptide agonist identified was Ser-p-fluoroPhe-p-guanidinoPhe-Leu-Arg-NH2, 9 (EC50 approximately 0.04 microM for stimulation of human platelet aggregation, approximately 10-fold more potent than the natural pentapeptide). Systematic substitution of the NH2-terminal Ser in 9 with neutral hydrophobic NH2-acyl groups led to partial agonists and eventually antagonists with unprecedented potency (greater than 1000-fold increase over the previously reported antagonist 3-mercaptopropionyl-Phe-Cha-Cha-Arg-Lys-Pro-Asn-Asp-Lys-NH2). In the series of NH2-acyl tetrapeptide antagonists, N-transcinnamoyl-p-fluoroPhe-p-guanidinoPhe-Leu-Arg-NH 2, 41 (BMS-197525), was identified as the tightest binding (IC50 approximately 8 nM) and most potent with an IC50 approximately 0.20 microM for inhibition of SFLLRNP-NH2-stimulated platelet aggregation. Systematic single substitutions in 41 indicated that, in addition to the NH2-terminal acyl group, the side chains at the second and third positions were also responsible for important and specific receptor interactions. The p-fluoroPhe and p-guanidinoPhe residues in the second and third positions of 41 were observed to be optimal in both the agonist and antagonist series. In the case of antagonists, however, an appropriately positioned positively charged group (i.e., protonated base) at the third residue was required. In contrast, such a substitution was not required for potent agonist activity. An even more potent antagonist resulted when 41 was extended at the C-terminus by a single Arg residue giving rise to analog 90 (BMS-200261) which had an IC50 approximately 20 nM for inhibition of SFLLRNP-NH2-stimulated platelet aggregation. When the C-terminal Arg of 90 was replaced by an Orn-(Ndelta-propionyl) residue, the resulting antagonist 91 (BMS-200661) was suitable for use in radioligand binding assays (Kd = 10-30 nM). Antagonist activity observed for selected compounds was verified through secondary assays in that these analogs prevented SFLLRNP-NH2-stimulated GTPase activity in platelet membranes and Ca2+ mobilization in cultured human smooth muscle cells and mouse fibroblasts. Furthermore, this inhibition occurred at concentrations that had no effect on thrombin catalytic activity indicating a specific activity attributable to receptor binding and not enzyme inhibition.