[des-His1, des-Phe6,Glu9]Glucagon-NH2 is a newly designed glucagon antagonist. This analog has a binding IC50 of 48 nM (compared to glucagon IC50 of 1.5 nM) and demonstrates pure antagonism in an adenylate cyclase assay. Although the number of glucagon antagonists has grown rapidly recently, closer examination suggested that many of these antagonists retained very low, almost imperceptible levels of cAMP accumulation that were sufficient to elicit an in vivo biological response. To investigate more carefully this secondary biological signal, we measured cAMP accumulation in a revised assay using isolated hepatocytes in the presence of the phosphodiesterase (PDE) inhibitor Rolipram. The PDE inhibitors Rolipram and isobutyl-1-methylxanthine (IBMX) increased the sensitivity of the cAMP accumulation assay from approximately 10-fold for the native hormone to 35-fold above basal levels. On the other hand, amrinone, another PDE inhibitor, did not affect the cAMP accumulation caused by glucagon. The use of PDE inhibitors indicated that three glucagon analogs that had previously been reported to have strong antagonist properties in classical adenylate cyclase assays were actually weak partial agonists in this new assay system. [N alpha-Trinitrophenyl-His1, homo-Arg12]glucagon, [des-amino-His1,D-Phe4,Tyr5, Arg12, Lys17,18,Glu21]glucagon, and [des-His1,Glu9]glucagon-NH2 demonstrated 233%, 21%, and 5.5% cAMP accumulation relative to the native hormone in the presence of 25 microM Rolipram. On the other hand, [des-His1,des-Phe6,Glu9]glucagon-NH2, a newly designed glucagon antagonist, did not activate adenylate cyclase in the presence of Rolipram up to a maximal physiological concentration of 1 microM, indicating that it was a pure antagonist of glucagon-induced adenylate cyclase activity and also the first one in this class. This compound and others were tested in a glycogen phosphorylase assay. As [des-His1,des- Phe6,Glu9]glucagon-NH2 did not activate phosphorylase activity, it was chosen as our candidate for in vivo testing in streptozotocin-induced diabetic rats. An initial dose of 0.75 mg/kg was found to cause the greatest lowering of blood glucose levels (to 63% of the initial levels in 15 min) when the bolus was followed by continuous infusion of 25 micrograms/kgxmin for 1 h.