Long-term opioid agonist treatment results in adenylyl cyclase superactivation. A recent "RAVE" theory implicates a direct correlation between the ability of agonist to induce receptor internalization and the magnitude of adenylyl cyclase superactivation. We decided to test such a theory by examining the adenylyl cyclase superactivation after long-term activation of mu-opioid receptor (MOR) in an EcR293 cell model. We examined the magnitudes of adenylyl cyclase superactivation in the presence of naloxone after long-term treatment with morphine, etorphine, and methadone, three agonists reported to have differential activities in promoting MOR internalization. It can be shown that the magnitudes of adenylyl cyclase superactivation after treating with these three agonists, although different, were dependent on MOR density. Blunting MOR internalization with the dominant-negative mutant of dynamin, K44E, did not alter the magnitude of either morphine- or etorphine-induced adenylyl cyclase superactivation. In the presence of diprenorphine, the magnitude of adenylyl cyclase superactivation after etorphine treatment was identical to that observed with morphine. It could be demonstrated further that adenylyl cyclase superactivation is dependent on the cell surface-located MOR. Sucrose gradient fractionation demonstrated the colocalization of MOR and adenylyl cyclase V/VI with caveolin-1, a marker for lipid rafts. After long-term agonist treatment, the majority of MOR remained at the lipid rafts. Methyl-beta-cyclodextrin (MbetaCD) completely blunted the adenylyl cyclase superactivation and agonist-induced receptor internalization. These MbetaCD actions were reversed by incubating the cells with cholesterol. Thus, the adenylyl cyclase superactivation is not dependent on agonist-induced receptor internalization. Rather, the location of MOR at lipid rafts is an absolute requirement for the observed adenylyl cyclase superactivation.