We studied contractile effects in isolated electrically driven (1 Hz) atrial preparations from patients undergoing cardiac bypass surgery. ATP concentration dependently (10, 30, and 100 microM) and rapidly decreased force of contraction (negative inotropic effect, NIE) and thereafter more slowly increased force of contraction. The maximum positive inotropic effect (PIE) at 100 microM ATP amounted to 152% of the predrug value (n = 9) and was stable and could be washed out fast and completely. The PIE did not affect time parameters of contraction (time to peak tension and time of relaxation). Moreover, a similar NIE and PIE were noted with adenosine 5'-O-(2-thiotriphosphate) (100 microM). In contrast 2-methyl-thio-ATP did not exert a NIE but only a PIE. In a second set of experiments, preparations were first incubated for 30 min with purinoreceptor antagonists and, in their continuous presence, 100 microM ATP was applied. However, the PIE and NIE of ATP could neither be blocked with suramin (100 and 500 microM), pyridoxalphosphate-6-azophenyl-2',4'-disulfonic acid (50 microM), nor reactive blue 2 (30, 100, and 500 microM), which are known blockers for subtypes of P(2) receptors, or 1,3-dipropyl-cyclopentvl-xanthine (1 and 10 microM), a subtype (A(1) adenosine) P(1) receptor blocker. Likewise, the inhibitor of phospholipase C (PLC) activity (U-73122) and the inhibitor of adenylate cyclase activity (SQ-022563) (10 microM each) failed to affect the NIE and the PIE of ATP. We tentatively suggest that the PIE of ATP might be mediated via P(2X4)-like receptors. In summary, we describe a novel biphasic effect of ATP on force contraction in the isolated human atrium. It is conceivable that ATP plays a physiological role in the human heart, for instance, after cardiac injury to sustain contractility.