We have previously demonstrated that the ROMK channel maintains the property of arachidonic acid (AA) sensitivity observed originally in the native ATP-sensitive K⁺ channel of the rat cortical collecting duct (16). We used the patch-clamp technique to extend these studies to other NH₂-terminal splice variants of the ROMK channel family, ROMK2 and ROMK3, expressed in Xenopus oocytes to determine the mechanism by which AA inhibits channel activity. Although the conductance, channel open probability, and open/closed times of the three homologs were determined to be similar, addition of 5-10 µM AA caused only a moderate inhibition of ROMK2 (15 ± 8%) and ROMK3 (13 ± 9%) activity, indicating that differences in the NH₂ termini of ROMK channels strongly influence the AA action. We consequently examined the effect of AA on a ROMK1 variant, R1ND37, in which the NH₂ terminal amino acids 2-37 were deleted, and on a mutant ROMK1, R1S4A, in which the serine-4 residue was mutated to alanine. Like ROMK2 and ROMK3, AA had a diminished effect on these variants. Addition of 1 nM exogenous protein kinase C (PKC) inhibited ROMK1 but not the mutant, R1S4A. However, the effect of AA is not a result of stimulation of a membrane bound PKC, since PKC inhibitors, calphostin C and chelerythrine, failed to abolish the AA-induced inhibition. In contrast, application of 5 µM staurosporine, a nonspecific protein kinase inhibitor at high concentration, abolished the effect of AA. We conclude that phosphorylation of serine-4 residue in the NH₂ terminus plays a key role in determination of AA effect on ROMK channels.