The permeation characteristics of a model opioid peptide, H-Tyr-D-Ala-Gly-Phe-D-Leu-OH (DADLE), and its cyclic prodrugs [acyloxyalkoxy-based cyclic prodrug of DADLE (AOA-DADLE), coumarinic acid-based cyclic prodrug of DADLE (CA-DALE), and oxymethyl-modified coumarinic acid-based cyclic prodrug of DADLE (OMCA-DADLE)] across the blood-brain barrier (BBB) were determined using an in situ perfused rat brain model. The rat brains were perfused with Krebs-bicarbonate buffer containing test compounds in the absence or presence of a specific P-glycoprotein inhibitor (GF-120918). Brain samples were collected after perfusion and processed by a capillary depletion method. After liquid phase extraction with acetonitrile, samples were analyzed using high-performance liquid chromatography with tandem mass spectrometric detection. Linear uptake kinetics of DADLE and its cyclic prodrugs was observed within the range of 60 to 240 s of perfusion. The apparent permeability coefficient (P(app)) of DADLE across the BBB was very low (<10(-7) cm/s), probably due to its unfavorable physicochemical properties (e.g., charge, hydrophilicity, and high hydrogen-bonding potential). All three cyclic prodrugs, however, also exhibited low membrane permeation (P(app) <10(-7) cm/s) in spite of their more favorable physicochemical properties (e.g., no charge, high hydrophobicity, and low hydrogen-bonding potential). Inclusion of GF-120918 (10 microM) in the perfusates fully inhibited the P-gp activity in the BBB and dramatically increased the P(app) values of AOA-DADLE, CA-DADLE, and OMCA-DADLE by approximately 50-, 460-, and 170-fold, respectively. In contrast, GF-120918 had no effect on the P(app) value of DADLE. In addition, the observed bioconversions of the prodrugs to DADLE in the rat brains after 240-s perfusion were very low (5.1% from AOA-DADLE, 0.6% from CA-DADLE, and 0.2% from OMCA-DADLE), which was consistent with the in vitro bioconversion rates determined previously in rat brain homogenates.