The hypothesis was tested that M2-selective antagonists partially utilize the allosteric site of muscarinic M2 receptors. The interactions of the allosteric agent W84 (hexane-1, 6-bis[dimethyl-3'-phthalimidopropyl-ammonium bromide]) were studied with the M2/M4-selective AF-DX 384 [(+/-)-5, 11-dihydro-11-([(2-(2-[(dipropylamino)methyl]-1-piperidinyl)ethyl)amino]carbonyl)-6H-pyrido(2,3-b)(1,4)-benzodiazepine-6-one], the nonselective N-methylscopolamine (NMS), and a number of other muscarinic antagonists. In isolated paced guinea pig atria, the antagonistic effect of W84 against oxotremorine- and arecaidine propargyl ester-induced negative inotropic actions reached a limiting value at higher W84 concentrations, revealing negative cooperativity (factors of cooperativity alpha = 311 and alpha = 495, respectively). The antagonistic potency of W84 in this M2 receptor model (W84 binding constant KA approximately 160 nM) was higher than at M1/M4-like receptors of rabbit vas deferens (KB approximately 800 nM) and at M3 receptors of guinea pig ileum (KB approximately 4,000 nM). In paced atria, combinations of W84 with muscarinic antagonists yielded more-than-additive antagonistic effects against oxotremorine in case of conventional antagonists such as NMS (alpha = 18) but less-than-additive effects with the M2-preferring AF-DX 384 (alpha = 444). In guinea pig heart homogenates, the equilibrium binding of [3H]NMS was only partially inhibited by W84 (alpha = 2.4), whereas [3H]AF-DX 384 binding could be suppressed completely (alpha = 194). The difference in cooperativity reflects that W84 inhibits [3H]NMS dissociation with a approximately 40-fold higher potency (ECdiss = 900 nM) than [3H]AF-DX 384 dissociation (ECdiss = 33,300 nM). [3H]NMS dissociation also could be retarded by AF-DX 384 (ECdiss = 22,000 nM), probably via an interaction with the site used by W84. The results suggest that the binding domain of AF-DX 384 partially overlaps with the common allosteric site of the M2 receptor protein.