We investigated the effects of the thrombin inhibitor, argatroban ((2R,4R)-4-methyl-1-[N2-(3-methyl-1,2,3,4-tetrahydro-8- quinolinesulfonyl)-L-arginyl]-2-piperidinecarboxylic acid) on the endothelium-derived relaxing factor-nitric oxide (EDRF-NO)-dependent relaxant, and the endothelial cell-independent constrictor actions of thrombin. Experiments were performed in isolated rings of canine coronary arteries. Argatroban inhibited thrombin-induced relaxation (range of thrombin activity 0.003-0.3 U/ml), with an ED50 of 0.3 microM. The ED50 value was not different from inhibition of thrombin amidolytic cleavage of the chromogenic substrate N-p-tosylgly-pro-arg-p-nitroanilide acetate (TOGSPAN 0.28 microM), but inhibition was highly selective. Argatroban did not block EDRF-NO-dependent relaxations to trypsin (0.003-0.3 U/ml; Emax -88.7 + 2.0% without vs. -88.1 +/- 2.7% with argatroban), acetylcholine (ACh 1 nM to 1 microM; Emax -90.5 +/- 4.7% and -88.6 +/- 3.1%, with and without argatroban, respectively), or the calcium ionophore A23187 (1 nM to 1 microM; Emax -98.5 +/- 1.2 vs. -99.4 +/- 0.6%). The inhibitory effects of argatroban on thrombin-induced constriction were then compared with those of the irreversible thrombin inhibitor D-phenylalanyl-L-prolyl L-arginine chloromethyl ketone (PPACK). The highest concentration of argatroban (10 microM) inhibited the vasoconstrictor effects of thrombin but did not completely block the effects (Emax 21.4 +/- 8.1% of KCl constriction without argatroban and Emax 14.0 +/- 5.2% of KCl-induced constriction with argatroban). In contrast, both a 10- and a 100-fold lower concentration of PPACK (0.1-1 microM) prevented the thrombin-induced increase in tension. Thrombin-induced constriction therefore appeared to disclose mechanistic differences between the two thrombin inhibitors. Thrombin vasomotor actions were inhibited by argatroban, however, and this may contribute significantly to the therapeutic effect of argatroban.