Coupling of the three alpha 2-adrenergic receptor (alpha 2AR) subtypes to Gi and Gs was studied in membranes from transfected CHO cells. We observed that in the presence of low concentrations of the alpha 2AR agonist UK-14304, alpha 2C10 mediated inhibition of adenylyl cyclase activity, whereas at high concentrations of agonist, alpha 2C10 mediated stimulation of adenylyl cyclase activity. We considered that this biphasic response was due to the coupling of alpha 2C10 to both Gi and Gs. To isolate functional Gs and Gi coupling, cells were treated with pertussis toxin or cholera toxin in doses sufficient to fully ADP-ribosylate the respective G-proteins. Following treatment with cholera toxin, agonists elicited only alpha 2C10-mediated inhibition (approximately 50%) of adenylyl cyclase while after pertussis toxin treatment, agonists elicited only alpha 2C10-mediated stimulation (approximately 60%) of adenylyl cyclase. Incubation of membranes with antisera directed against the carboxyl-terminal portion of Gs alpha blocked this functional alpha 2AR.Gs coupling to the same extent as that found for beta 2AR.Gs coupling. In addition to functional Gs coupling, we also verified direct, agonist-dependent, physical coupling of alpha 2AR to Gs alpha. In agonist-treated membranes, an agonist-receptor-Gs alpha complex was immunoprecipitated with a specific alpha 2C10 antibody, and the Gs component identified by both western blots using Gs alpha antibody, and cholera toxin mediated ADP-ribosylation. Due to the differences in primary amino acid structure in a number of regions of the alpha 2AR subtypes, we investigated whether G-protein coupling was subtype-selective, using UK-14304 and cells with the same alpha 2AR expression levels (approximately 5 pmol/mg). Coupling to Gi was equivalent for alpha 2C10, alpha 2C4, and alpha 2C2: 53.4 +/- 8.8% versus 54.9 +/- 1.0% versus 47.6 +/- 3.5% inhibition of adenylyl cyclase, respectively. In marked contrast, distinct differences in coupling to Gs were found between the three alpha 2AR subtypes: stimulation of adenylyl cyclase was 57.9 +/- 6.3% versus 30.7 +/- 1.1% versus 21.8 +/- 1.7% for alpha 2C10, alpha 2C4, and alpha 2C2, respectively. Thus, alpha 2AR have the potential to couple physically and functionally to both Gi and Gs; for Gi coupling we found a rank order of alpha 2C10 = alpha 2C4 = alpha 2C2, while for Gs coupling, alpha 2C10 greater than alpha 2C4 greater than alpha 2C2.