To test the hypothesis that the direct thrombin inhibitor, melagatran is able to inhibit local pro-carboxypeptidase U (proCPU) activation that occurs during thrombolytic treatment, t-PA alone, or in combination with melagatran, was given to dogs with a coronary artery thrombosis. Blood samples from the great cardiac vein and aorta were collected at baseline, during thrombus formation, throughout the t-PA+/-melagatran infusion and during the patency period, for analysis of CPU activity using a novel assay. A higher CPU activity in venous compared to arterial blood (V-A difference) indicates CPU activation in coronary vessels. Efficacy was assessed by determination of time to lysis, duration of patency and blood flow during patency. Dogs (n = 26) were randomized to receive either 1) t-PA, 1 mg/kg as an intravenous 20-min infusion; 2) t-PA as in group 1, +melagatran bolus, 0.3 mg/kg, followed by a 3-h infusion (0.15 mg/kg per h); 3) sham-operated but no coronary thrombus, and administered t-PA as for Group 1. All groups had similar baseline characteristics. Significant increases in CPU activity were observed in Groups 1 and 2 during thrombus formation, with V-A differences of 5.5 and 4.5 U/L, respectively. No significant V-A difference was observed in the sham-operated group. CPU activity increased in Group 1 during the t-PA infusion (V-A difference 15.9 U/L), whereas the V-A difference in Group 2 decreased to 2.6 U/L following melagatran treatment. These results demonstrate that melagatran attenuates generation of CPU in the coronary circulation. The mechanism is probably indirect, via inhibition of thrombin-mediated activation of proCPU.