A 78 kDa gastrin-binding protein (GBP) has previously been identified as the target of the anti-proliferative effects of non-selective gastrin/cholecystokinin receptor antagonists on colorectal carcinoma cell lines. The GBP was related in sequence to a family of fatty acid oxidation enzymes possessing enoyl CoA hydratase and 3-hydroxyacyl CoA dehydrogenase activity. This study aims to define the binding site for gastrin and gastrin antagonists in greater detail. The N- and C-terminal halves of the porcine GBP were expressed independently as glutathione S-transferase fusion proteins in E. coli. Affinities of gastrin and gastrin antagonists for the fusion proteins were measured by competition for 125I-[Nle15]-gastrin binding in a covalent cross-linking assay. The N- and C-terminal fusion proteins bound gastrin with affinities of 9.9 +/- 6.1 and 71 +/- 48 microM, respectively (n = 3). These values were 40-fold and 300-fold lower than the affinity of the full-length GBP for gastrin (0.23 +/- 0.15 microM). In contrast, the affinities of the N- and C-terminal halves for the antagonists proglumide (22 +/- 13 and 10 +/- 4 mM, respectively) and benzotript (350 +/- 90 and 400 +/- 160 micro M, respectively) were similar to each other and to the affinities of proglumide and benzotript for the full-length GBP (5.1 +/- 3.6 mM and 200 +/- 120 microM, respectively). It is concluded that proglumide and benzotript bind independently to both the hydratase and dehydrogenase active sites of the GBP, while a single molecule of gastrin may bind simultaneously to both active sites. A model is proposed which is consistent with these data, and which will assist in the development of more potent and selective GBP antagonists.