A series of cyclic, disulfide- or dithioether-containing tetrapeptides based on previously reported potent mu- and delta-selective analogs has been explored with the aim of improving their poor affinity to the kappa-opioid receptor. Specifically targeted were modifications of tetrapeptide residues 3 and 4, as they presumably interact with residues from transmembrane helices 6 and 7 and extracellular loop 3 that differ among the three receptors. Accordingly, tetrapeptides were synthesized with Phe(3) replaced by aliphatic (Gly, Ala, Aib, Cha), basic (Lys, Arg, homo-Arg), or aromatic sides chains (Trp, Tyr, p-NH(2)Phe), and with d-Pen(4) replaced by d-Cys(4), and binding affinities to stably expressed mu-, delta-, and kappa-receptors were determined. In general, the resulting analogs failed to exhibit appreciable affinity for the kappa-receptor, with the exception of the tetrapeptide Tyr-c[d-Cys-Phe-d-Cys]-NH(2), cyclized via a disulfide bond, which demonstrated high binding affinity toward all opioid receptors (Ki(mu) = 1.26 nm, Ki(delta) = 16.1 nm, Ki(kappa) = 38.7 nm). Modeling of the kappa-receptor/ligand complex in the active state reveals that the receptor-binding pocket for residues 3 and 4 of the tetrapeptide ligands is smaller than that in the mu-receptor and requires, for optimal fit, that the tripeptide cycle of the ligand assume a higher energy conformation. The magnitude of this energy penalty depends on the nature of the fourth residue of the peptide (d-Pen or d-Cys) and correlates well with the observed kappa-receptor binding affinity.