We studied the effects of the CuZn superoxide dismutase (SOD), catalase (CAT), and glutathione (GSH) on endothelial permeability to 'L51-albumin after activation of neutrophils (PMN) with phorbol 12-myristate-13-acetate (PMA; 1 O-8M). PMN were either in direct contact with the endothelial cell monolayer grown on a porous gelatin-coated microporous 10-pm-thick polycarbonate filter (upright system) or separated from the endothelium by a similar filter (inverted system). Transendothelial '251-albumin clearance rates were measured as an index of endothelial permeability. In the absence of antioxidants, activation of PMN increased transendothelial '251-albumin clearance rates in both systems from 0.041 2 0.006 pl/min (baseline) to 0.262 2 0.18 plimin (upright system) and from 0.063 ? 0.02 pl/min to 0.244 ? 0.06 pl/min (inverted system). PMA induced 80-90% of PMN to adhere to either gelatin-coated filters or to endothelial cells, from the basal PMN adhesion value of 5.3 * 2.2% and 4.3 -t 1.1%, respectively. SOD, which dismutates superoxide anion to hydrogen peroxide (H,O,), did not alter the transendothelial "'I-albumin clearance rates in either system at any concentration from 10-300 Uiml. CAT (1 00-1,000 U/ml) and GSH (0.5-10 mM), which remove the H,O, generated during PMN activation, did not alter the increase in transendothelial 'LSI-clearance rates after PMN activation in the upright system, but both agents prvented the increase in transendothelial '"I-clearance rates in the inverted system. We conclude that PMN activation with PMA causes endothelial injury irrespective of PMN contact to the endothelial monolayer. Moreover, H, O, , a release product of PMN activation, is a criticdl mediator of PMN-dependent endothelial injury. Finally, the results indicate that CAT and GSH prevent endothelial injury only in the absence of direct PMN contact with endothelial cells, suggesting that antioxidants such as GSH and CAT are excluded from sites of PMN-endothelial contact and thus are ineffective antioxidants.