Cellular membranes from a cell line expressing P-glycoprotein (Pgp(+)) and from a cell line that does not express Pgp (Pgp(-)) were immobilized on the surface of glass capillaries (25 cm x 100 microm i.d.) by non-covalent interactions using the avidin-biotin coupling system to create two open tubular columns, Pgp(+)-OT and Pgp(-)-OT. Frontal displacement chromatography on the Pgp(+)-OT demonstrated that the immobilized Pgp retained its ability to specifically bind the known Pgp substrates vinblastin and ketoconazole. The calculated affinities, expressed as K(d), for vinblastin and ketoconazole were 97 nM and 12.1 microM, which were comparable with previously reported K(d) values of 37 nM and 8.6 microM, respectively. The results confirm that the Pgp(+)-OT can be used to quantitatively estimate binding affinities for the Pgp. Frontal displacement chromatography on the Pgp(-)-OT demonstrated that the immobilized membranes retained the ability to bind some Pgp substrates, but that the binding was not due to specific binding to Pgp. A cohort of compounds containing high affinity Pgp substrates (vinblastin, prazosin) and moderate-low affinity Pgp substrates (doxorubicin, verapamil, ketoconazole) and a non-substrate (nicotine) were chromatographed on the Pgp(+)-OT and Pgp(-)-OT using fast frontal analysis and mass spectrometric detection. The results demonstrated that when the retention on the Pgp(+)-OT was corrected by subtraction of the retention on the Pgp(-)-OT, the test compounds could be accurately sorted into high, moderate-low and non-substrate categories. The data from the study indicates that a single 30-min parallel chromatographic experiment can be used to rank a compound based upon its relative affinity for the immobilized Pgp.