Isolated islets were either studied immediately after isolation (fresh; F), or were cultured for 6 days at 11 mM glucose (desensitized; D), or were incubated for 2 h at 5.5 mM glucose following D (recovered; R). Glucose-stimulated insulin secretion in D islets was reduced compared with F and R islets. In the presence of 3-isobutyl-1-methylxanthine, glucose also increased cyclic adenosine monophosphate (cAMP) levels in F islets, but failed to affect cAMP generation in R or D islets. Glucagon alone or in the presence of glucose stimulated insulin release in F and R islets, but the response was blunted in D islets. Glucagon-like peptide 1 (GLP) potentiated insulin secretion in R islets, but not in D islets. Glucagon (0.01-0.1 microM) did not increase cAMP levels in D islets, whereas GLP (0.1 microM) increased cAMP as much as 4.5-fold. R islets recovered adenylyl cyclase responsivity to glucagon, and GLP increased cAMP levels as much as 9-fold. In F islets pretreated with forskolin for 2 h, the cAMP responses to glucose and GLP were inhibited. The cAMP response to forskolin stimulation was similarly inhibited in D islets and in islets pretreated for 2 h with forskolin. Forskolin pretreatment significantly attenuated the islet insulin release response to glucose, although the combined stimulus of glucose and GLP restored insulin release to control values. Insulin secretion in response to glucose and cAMP analogue (Sp)5,6-dichloro-1-beta-D-ribofuranosylbenzimidazole-3'-5'-cyclic monophosphorothioate was lower than that observed in F islets. In conclusion, beta-cell cAMP accumulation in response to several stimuli acting through different mechanisms is impaired following continuous glucose stimulation. However, cAMP levels are not the definitive second messenger in the recovery of glucose-sensitive insulin secretion in glucose desensitized islets.