We used the patch-clamp technique to study the effects of ATP on the small-conductance potassium channel in the apical membrane of rat cortical collecting duct (CCD). This channel has a high open probability (0.96) in the cell-attached mode but activity frequently disappeared progressively within 1-10 min after channel excision (channel "run-down"). Two effects of ATP were observed. Using inside-out patches, low concentrations of ATP (0.05-0.1 mM) restored channel activity in the presence of cAMP-dependent protein kinase A (PKA). In contrast, high concentrations (1 mM) of adenosine triphosphate (ATP) reduced the open probability (Po) of the channel in inside-out patches from 0.96 to 0. 1.2 mM adenosine diphosphate (ADP) also blocked channel activity completely, but 2 mM adenosine 5'-[beta,gamma-imido]triphosphate (AMP-PNP), a nonhydrolyzable ATP analogue, reduced Po only from 0.96 to 0.87. The half-maximal inhibition (Ki) of ATP and ADP was 0.5 and 0.6 mM, respectively, and the Hill coefficient of both ATP and ADP was close to 3. Addition of 0.2 or 0.4 mM ADP shifted the Ki of ATP to 1.0 and 2.0 mM, respectively. ADP did not alter the Hill coefficient. Reduction of the bath pH from 7.4 to 7.2 reduced the Ki of ATP to 0.3 mM. In contrast, a decrease of the free Mg2+ concentration from 1.6 mM to 20 microM increased the Ki of ATP to 1.6 mM without changing the Hill coefficient; ADP was still able to relieve the ATP-induced inhibition of channel activity over this low range of free Mg2+ concentrations. The blocking effect of ATP on channel activity in inside-out patches could be attenuated by adding exogenous PKA catalytic subunit to the bath. The dual effects of ATP on the potassium channel can be explained by assuming that (a) ATP is a substrate for PKA that phosphorylates the potassium channel to maintain normal function. (b) High concentrations of ATP inhibit the channel activity; we propose that the ATP-induced blockade results from inhibition of PKA-induced channel phosphorylation.