Tumor necrosis factor-alpha (TNF-alpha) may increase vascular endothelial permeability through alteration of the extracellular matrix (ECM). Incubation of bovine pulmonary microvascular endothelial (BPMVE) cells grown to confluence on microporous filters with 10(4) U/ml TNF-alpha for 24 h increased monolayer permeability to 125I-labeled albumin two- to threefold. TNF-alpha treatment also induced expression of a 96-kDa gelatinolytic metalloproteinase that was present in the medium and bound to the ECM. The induced 96-kDa metalloproteinase was purified from conditioned medium and found to cleave fibronectin, laminin, types IV and V collagens, and gelatins from types I and III collagens, suggesting identity as a type IV collagenase-gelatinase. Incubation of BPMVE cells with the 96-kDa gelatinase increased monolayer permeability, an effect prevented by inclusion of either tissue inhibitor of metalloproteinase (TIMP) or 1,10-phenanthroline. When BPMVE cells were incubated with the 96-kDa gelatinase or 10(4) U/ml TNF-alpha and then stripped from the filters, the remaining ECM displayed increased permeability to 125I-albumin compared with matrix from untreated BPMVE. The ECM extracts from both TNF-alpha- and enzyme-treated cells were found to contain less fibronectin, whereas their total protein contents were similar to those of untreated controls. These results suggest that the 96-kDa metalloproteinase induced by TNF-alpha contributes to increased vascular endothelial permeability through the degradation of specific extracellular matrix components.

IntroductionWe examined the effects of tumor necrosis factor-a (TNFa) stimulation of endothelial cells on the increase in endothelial permeability induced by H202. Bovine pulmonary microvascular endothelial cells (BPMVEC) were grown to confluence on a microporous filter and the 1251-albumin clearance rate across the monolayer was determined. Pretreatment with TNFa (100 U/ml) for 6 h had no direct effect on transendothelial 1251-albumin permeability. However, TNFa pretreatment enhanced the susceptibility of BPMVEC to H202; that is, H202 (10 MM) alone had no direct effect, whereas H202 increased 1251-albumin permeability more than threefold when added to monolayers pretreated for 6 h with TNFa. Determination of lactate dehydrogenase release indicated that increased permeability was not due to cytolysis. We measured the intracellular contents of GSH and catalase to determine their possible role in mediating the increased susceptibility to H202. TNFa treatment (100 U/ ml for 6 h) decreased total GSH content and concomitantly increased the oxidized GSH content, but did not alter the cellular catalase activity.

The cell-surface localization and site of activation of type IV collagenases/gelatinases (matrix metalloproteinases, MMP) in bovine pulmonary microvascular endothelial (BPMVE) cells was examined. Sucrose density centrifugation of plasma membranes and immunofluorescent staining of whole cells indicated association of 72 kDa (MMP-2) and 96 kDa (MMP-9) type IV collagenase/gelatinases with the plasma membrane. Incubation of the BPMVE cells with rhodaminated MMP-9 demonstrated colocalization with beta 1-integrin, indicating incorporation into the focal contacts. The focal contacts were extracted with saponin, and associated proteolytic activity was examined by zymography. The focal contacts contained latent MMP-2, and stimulation of the cells with cytochalasin D or with 8-bromoadenosine 3',5'-cyclic monophosphate with 3-isobutyl-1-methylxanthine increased both latent and activated MMP-9 in the focal contacts. Addition of these stimuli in unconditioned culture medium did not produce this effect, indicating that the MMP-9 in focal contact extracts was derived from previously secreted enzyme. The activated metalloproteinase degraded extracellular matrix collagens and was inhibited by 1,10-phenanthroline. These findings indicate that endothelial cells release MMP into the extracellular milieu and then concentrate and activate MMP-9 from medium at the focal contacts.

Anionic glutathione S-transferases were purified from human lung and placenta. Chemical and immunochemical characterization, including polyacrylamide-gel electrophoresis, gave strong evidence that the anionic lung and placental enzymes are chemically similar, if not identical, proteins. The electrophoretic mobilities of both proteins were identical in conventional alkaline gels as well as in gels containing sodium dodecyl sulphate. Gel filtration of the intact active enzyme established an Mr value of 45000; however, with sodium dodecyl sulphate/polyacrylamide-gel electrophoresis under dissociating conditions a subunit Mr of 22500 was obtained. Amino acid sequence analysis of the N-terminal region of the placental enzyme revealed a single polypeptide sequence identical with that of lung. Results obtained from immunoelectrophoresis, immunotitration, double immunodiffusion and rocket immunoelectrophoresis also indicated the anionic lung and placental enzymes to be closely similar. The chemical similarity of these two proteins was further supported by protein compositional analysis and fragment analysis after chemical hydrolysis. Immunochemical comparison of the anionic lung and placental enzymes with human liver glutathione S-transferases revealed cross-reactivity with the anionic omega enzyme, but no cross-reactivity was detectable with the cationic enzymes. Comparison of the N-terminal region of the human anionic enzyme with reported sequences of rat liver glutathione S-transferases gave strong evidence of chemical similarity, indicating that these enzymes are evolutionarily related. However, computer analysis of the 30-residue N-terminal sequence did not show any significant chemical similarity to any other reported protein sequence, pointing to the fact that the glutathione S-transferases represent a unique class of proteins.