Efflux pump like P-glycoprotein (P-gp) is known to be a major barrier to drug delivery. Functional P-glycoprotein has been recently identified in cornea and corneal cell lines. Thus, it is probable that P-glycoprotein may restrict in vivo ocular drug absorption, resulting in low ocular bioavailability. Experiments were designed using New Zealand albino (New Zealand White) rabbits to assess inhibitors of P-gp efflux to increase drug absorption. Anesthetized rabbits were given constant topical infusions of [(14)C]erythromycin in the presence and absence of inhibitors. Testosterone, verapamil, quinidine, and cyclosporine A were selected as P-gp inhibitors. Transport experiments were conducted in Madin-Darby canine kidney cells transfected with the human mdr1 gene (MDCK-MDR1). Erythromycin exhibited significant efflux out of MDCK-MDR1 cells, suggesting that erythromycin is a good substrate for P-gp. Ocular pharmacokinetic studies were conducted using a topical single-dose infusion method. Maximum inhibition of P-gp mediated efflux was observed with 500 microM testosterone. Area under the curve (AUC)(0- infinity ) of erythromycin with 500 microM testosterone was almost 4 times higher than AUC(0- infinity ) without any inhibitor. Rate of elimination (k(10)) for erythromycin and those with inhibitors was found to be similar (141 +/- 23 min), suggesting that elimination pathways were not altered. All the inhibitors were found to be nontoxic. Verapamil also inhibited the efflux pump with moderate change in AUC(0- infinity ) and C(max) compared with control. Thus, P-gp is found to be active in vivo, and it restricts topical erythromycin absorption across the cornea, which can be inhibited by known P-gp inhibitors. Therefore, ocular bioavailability of P-gp substrates can be significantly enhanced by proper selection of P-gp inhibitors.