Combining site-directed mutagenesis with information obtained from molecular modelling of the bradykinin (BK) human B2 receptor (hB2R) as derived from the bovine rhodopsin crystal structure [Science 289 (2000) 739], we previously defined a putative binding mode for the non-peptide B2 receptor antagonists, FR173657 and LF16-0687 [Can J Physiol Pharmacol 80 (2002) 303]. The present work is aimed to define the specific role of the quinoline moiety in the pharmacophore of these non-peptide antagonists. The effect of the mutations I110A, L114A (TM, transmembrane 3), W256A (TM6), F292A, Y295A and Y295F (TM7) was evaluated. None of the mutations affected the binding interaction of peptide ligands: the agonist BK and the peptide antagonist MEN 11270. The affinities in competing for [3H]-BK binding and in blocking the BK-induced IP production by the non-peptide antagonists LF16-0687 and FR173657 at the wild type and mutant receptors were analysed. While the affinities of LF16-0687 and FR173657 were crucially decreased at the I110A, Y295A, and Y295F mutants, the W256A mutation affected the affinity of the LF16-0687 only. The important contribution of the quinoline moiety was shown by the inability of an analogue of LF16-0687, lacking this moiety, to affect BK binding at the wild type receptor. On the other hand, the benzamidine group did not interact with mutated residues, since LF16-0687 analogues without this group or with an oxidated benzamidine displayed pairwise loss of affinity on wild type and mutated receptors. Further differences between FR173657 and LF16-0687 were highlighted at the I110 and Y295 mutants when comparing binding (pK(i)) and functional antagonist (pKB) affinity. First, the I110A mutation similarly impaired their binding affinity (250-fold), but at a less extent the antagonist potency of FR173657 only. Second, both the hydroxyl and the phenyl moieties of the Y295 residue had a specific role in the LF16-0687 interaction with the receptor, as demonstrated at the Y295F and Y295A mutants, respectively, but not in that of FR173657. Present data identify a receptor binding pocket comprised among TM3, 6, and 7, which concerns the interaction of the non-peptide antagonists FR173657 and LF16-0687, but not that of the peptide agonist or antagonist. Results indicate the quinoline group as the involved pharmacophoric element, and that the studied residues are differently involved in the interaction. The analysis performed by means of the GRID software led us to propose different spatial orientations of the quinoline moieties and partially overlapping binding pockets for the two ligands: that of LF16-0687 is located in the lipophilic environment amongst I110 (TM3), W256 (TM6), and Y295 (TM7) residues, whereas that of FR173657 lies essentially between I110 and Y295.