The stimulatory guanine nucleotide-binding protein (Gs), which links cell-surface receptors to second-messenger effector systems, is assumed to be confined to plasma membranes. In the current studies we tested whether Gs redistributes within cells by treating S49 lymphoma cells with the beta-adrenergic agonist isoproterenol, then separating cytosol and crude membrane fractions (defined as pellet and supernatant, respectively, after centrifugation for 1 hr at 150,000 x g), and assaying fractions for the alpha subunit of Gs (alpha s) using a competitive ELISA and reconstitution techniques. Under basal conditions, a small (10%) pool of alpha s was identified in supernatant fractions of S49 cells. The size of this pool decreased in the first 15 min after agonist treatment of cells. This decrease was blocked by a beta-adrenergic receptor antagonist and did not occur in an S49 variant, UNC, which lacks functional interaction between receptors and Gs. The size of the alpha s pool in supernatant fractions increased to almost 50% of total cellular alpha s during a 1-hr incubation of cells with isoproterenol. Before isoproterenol treatment only the competitive ELISA was sensitive enough to detect cytosolic alpha s, whereas at later time points (greater than or equal to 30 min) the presence of alpha s in the cytosol was confirmed by both immunoblotting and by reconstitution of adenylyl cyclase [ATP pyrophosphate-lyase (cyclizing), EC 4.6.1.1] in Gs-deficient membranes derived from cyc-S49 cells. In contrast to membrane alpha s, cytosolic alpha s did not require activation (e.g., by AlF4-) in the reconstitution assay to stimulate adenylyl cyclase. Use of an antibody that selectively recognizes monomeric dissociated alpha s, but not heterotrimeric alpha s, indicated that cytosolic alpha s is monomeric. These data indicate that alpha s is not exclusively localized to the plasma membrane and that agonist treatment redistributes this protein within target cells.