SUMMARY1. In voltage-clamped whole cells dialysed with GTP, extracellular application of ACh elicits an inwardly rectifying K+ current which subsequently decreases to a steady-state level well below the maximally induced current (desensitization). The mechanism of desensitization of the acetylcholine(ACh)-activated K+ channel current was studied in rat neonatal atrial cells at the single-channel level using the patch-clamp technique.2. In cell-attached patches with ACh in the pipette, a similar pattern of K+ channel current desensitization was present. Single-channel analyses revealed that the initial rapid decrease in channel activity was associated with progressive shortening of the mean open time (To) and prolongation of the mean closed time (rc) of the K+ channel.3. In excised, inside-out patches with ACh in the pipette, GTP activated K+ channels with a ro of 1.0 ms. Addition of ATP to the cytosolic surface resulted in progressive increases in To (from 1 to 5 ms) and channel activity. These changes are similar but opposite in direction to those observed during the early phase of AChinduced channel desensitization in cell-attached patches. 4. The effect of ATP on the channel kinetics was abolished in Mg2+-free solution AMP-PNP (adenylyl-imidodiphosphate, a non-hydrolysable analogue of ATP), ADP, CTP (cytidine triphosphate), ITP (inosine triphosphate) or UTP (uridine triphosphate) did not alter the channel kinetics, suggesting that the ATP effect on channel gating probably occurs via phosphorylation by a membrane-bound kinase. H-8 (an isoquinolinesulphonamide derivative which inhibits protein kinases A and C) failed to prevent the action of ATP on the channel.5. The increases in ro and channel activity produced by ATP could be completely reversed by an elevation of cytosolic [Ca21] to 3 x 10-5 M or above. 6. The effect of Ca21 on the ATP-induced changes in channel kinetics was blocked by sodium vanadate, a general phosphatase inhibitor. Okadaic acid, an inhibitor of protein phosphatase 1 and 2A, did not block the Ca2' effect. Calmodulin antagonists, N-(6-aminohexyl)-5-chloro-1 -naphthalenesulphonamide (W-7), trifluoroperazine, and calmidazolium, partially blocked the effect of Ca2 .7. Alkaline phosphatase (20 units/ml) reversed the ATP-induced increases in T.MS 8700 and channel activitv. These results suggest that the ACh-activated K+ channel can be modulated by phosphorvlation and dephosphorylation.8. The early phase of desensitization of the K+ current activated by ACh can largely be explained by time-dependent changes in the open-and closed-time durations of the channel, possibly by activation of a Ca2+-calmodulin-dependent phosphatase. The secondary slower phase of desensitization was due to a progressive time-dependent decrease in the frequency of opening, and may involve effects on the receptor---G protein coupling.