Characterization of matrix metalloproteinases produced by rat alveolar macrophages.

Evidence presented in the accompanying article (Gibbs, D. F., T. P. Shanley, R. L. Warner, H. S. Murphy, J. Varani, and K. J. Johnson. 1999. Role of matrix metalloproteinases in models of macrophage-dependent acute lung injury: evidence for alveolar macrophage as source of proteinases. Am. J. Respir. Cell Mol. Biol. 20:1145-1154) implicates alveolar macrophage matrix metalloproteinases (MMPs) in two models of acute lung inflammation in the rat. As a prerequisite to understanding which specific MMPs might be involved in the injury and how they might function, it was necessary to know the spectrum of enzymes present. To this end, alveolar macrophages were obtained from normal rat lungs by bronchoalveolar lavage, placed in culture with and without various agonists, and assessed by a variety of techniques for MMPs. The identification process involved characterization by gelatin, beta-casein, and kappa-elastin zymography, with confirmation of identity by Western blot/immunoprecipitation. Message levels of detected MMPs were assessed by Northern blot. Rat alveolar macrophages were found to produce a low constitutive level of MMP-2 (72-kD gelatinase A) that was only modestly upregulated following stimulation with phorbol myristate acetate, bacterial lipopolysaccharide, or immunoglobulin A-containing immune complexes. Although control cells were found to produce little or no MMP-9 (92-kD gelatinase B) or MMP-12 (metalloelastase), both enzymes were markedly upregulated upon stimulation. In the same stimulated macrophages there was little activity against type I collagen (associated with MMP-13 [collagenase-3] on the basis of Western blotting), no activity suggestive of stromelysin or matrilysin, and no measurable secretion of the serine proteinases, elastase and cathepsin G. These data demonstrate the ability of rat alveolar macrophages to elaborate certain MMPs under proinflammatory conditions, consistent with their possible involvement in the progression of acute inflammation.

Role of matrix metalloproteinases in models of macrophage-dependent acute lung injury. Evidence for alveolar macrophage as source of proteinases.

Matrix metalloproteinases (MMPs) have been implicated in the tissue injury seen in neutrophil-dependent models of acute lung injury. However, the role of MMPs in macrophage-dependent models of lung injury is unknown. To address this issue, the macrophage-dependent immunoglobulin A immune complex-induced lung injury model and the macrophage-dependent portion of the lipopolysaccharide-induced acute lung injury model in the rat were assessed for MMP involvement and for the source of these activities. In both models, injury was inhibited by the recombinant human tissue inhibitor of metalloproteinases-2. Bronchoalveolar lavage fluids (BALFs) from injured animals in both models showed increased levels of MMPs. Characterization of MMP production by isolated lung fibroblasts, endothelial cells, type II epithelial cells, and alveolar macrophages revealed that only the macrophage had the same spectrum of MMP activity as seen in the BALF. Further, isolated alveolar macrophages from injured lungs showed evidence of in vivo activation with the release of the same spectrum of MMP activities. Together these studies show that MMPs are produced during macrophage-dependent lung injury, that these MMPs play a role in the development of the lung injury, and that the alveolar macrophage is the likely source of these MMPs.

Regulatory effects of endogenous protease inhibitors in acute lung inflammatory injury.

Inflammatory lung injury is probably regulated by the balance between proteases and protease inhibitors together with oxidants and antioxidants, and proinflammatory and anti-inflammatory cytokines. Rat tissue inhibitor of metalloprotease-2 (TIMP-2) and secreted leukoprotease inhibitor (SLPI) were cloned, expressed, and shown to be up-regulated at the levels of mRNA and protein during lung inflammation in rats induced by deposition of IgG immune complexes. Using immunoaffinity techniques, endogenous TIMP-2 in the inflamed lung was shown to exist as a complex with 72- and 92-kDa metalloproteinases (MMP-2 and MMP-9). In inflamed lung both TIMP-2 and SLPI appeared to exist as enzyme inhibitor complexes. Lung expression of both TIMP-2 and SLPI appeared to involve endothelial and epithelial cells as well as macrophages. To assess how these endogenous inhibitors might affect the lung inflammatory response, animals were treated with polyclonal rabbit Abs to rat TIMP-2 or SLPI. This intervention resulted in significant intensification of lung injury (as revealed by extravascular leak of albumin) and substantially increased neutrophil accumulation, as determined by cell content in bronchoalveolar lavage (BAL) fluids. These events were correlated with increased levels of C5a-related chemotactic activity in BAL fluids, while BAL levels of TNF-alpha and chemokines were not affected by treatment with anti-TIMP-2 or anti-SLPI. The data suggest that endogenous TIMP-2 and SLPI dynamically regulate the intensity of lung inflammatory injury, doing so at least in part by affecting the generation of the inflammatory mediator, C5a.

Growth factor-induced epidermal invasion of the dermis in human skin organ culture: expression and role of matrix metalloproteinases.

Matrix metalloproteinase activity was assessed in culture fluids of organ-cultured human skin by gelatin zymography. Both the 92-kD gelatinase/type IV collagenase and the 72-kD gelatinase/type IV collagenase were detected. Production of the 92-kD enzyme was substantially increased in the presence of epidermal growth factor (EGF) and hepatocyte growth factor (HGF) as compared to control but not in the presence of insulin-like growth factor-1 (IGF-1) or keratinocyte growth factor (KGF). This is of interest because our recent studies have shown that EGF and HGF induce the epithelial cells to invade the underlying stroma while normal architecture is maintained in the presence of IGF-1 and KGF. Addition of tissue inhibitor of metalloproteinase-2 to the organ culture fluids blocked expression of the active forms of both enzymes and concomitantly blocked invasion. Epidermal keratinocytes, dermal fibroblasts and dermal endothelial cells were grown in monolayer culture and examined for matrix metalloproteinase production. The 92-kD enzyme accounted for most of the gelatinase activity in keratinocyte culture fluids while the 72-kD enzyme accounted for most of the activity in the dermal fibroblast and endothelial cell culture fluids. Increased production of the 92-kD enzyme was seen in keratinocytes upon exposure to the growth factors that induced invasion (EGF and HGF) while the two factors that did not induce invasion (IGF-1 and KGF) were much less effective. Production of the 72-kD enzyme in fibroblasts and endothelial cells was not upregulated by any of the four growth factors. Taken together, these data indicate that matrix metalloproteinase activity is increased in the epithelium under the influence of invasion-inducing growth factors and contributes to invasion.

Expression of serine proteinases and metalloproteinases in organ-cultured human skin. Altered levels in the presence of retinoic acid and possible relationship to retinoid-induced loss of epidermal cohesion.

Neonatal human foreskin obtained at circumcision was cut into 2 x 2-mm pieces and placed in organ culture. Culture medium consisted of a serum-free, growth factor-free basal medium containing either 0.15 mmol/L Ca2+ or 1.4 mmol/L Ca2+. Some cultures were left as control, whereas others were treated with 3 mumol/L all-trans retinoic acid (RA). In the presence of RA, epidermal cohesion was disrupted and the upper layers separated from the viable epidermis beneath. This effect was observed under both low Ca2+ and high Ca2+ conditions. At 2-day intervals, culture fluids were collected and analyzed for serine and metalloproteinase activities. Serine proteinase activity was detected in the culture fluids and virtually all of the detected activity was dependent on the presence of plasminogen. Activity was elevated in the RA-treated tissues and this was due to increased amounts of both urokinase-type plasminogen activator (u-PA) and tissue-type plasminogen activator (t-PA). Elastase and cathepsin G were not detected in either control or RA-treated cultures. Increased plasminogen activator levels were also detected in RA-treated keratinocytes and fibroblasts in monolayer culture. Significant amounts of t-PA (though not u-PA) were found in fibroblast culture fluids, whereas both t-PA and u-PA were detected in culture fluids from keratinocytes. Metalloproteinase activity was also detected in the culture fluids of control and RA-treated tissues but in contrast to plasminogen activator, metalloproteinase activity decreased in the presence of RA. Casein and gelatin zymographic studies indicated the presence of both 92- and 72-kd gelatinases and stromelysin-1 and suggested that the decreased activity was primarily due to reduction in the 92- and 72-kd gelatinases. When serine proteinase inhibitors (aprotinin and soybean trypsin inhibitor) were included in the culture medium throughout the incubation period, epidermal discohesion was reduced. A metalloproteinase inhibitor, tissue inhibitor of metalloproteinase-2, did not have this effect. Taken together, these data show that a number of proteolytic enzymes are produced during organ culture of human skin. They suggest that these proteases may influence the structural integrity of the tissue.

In vivo suppression of immune complex-induced alveolitis by secretory leukoproteinase inhibitor and tissue inhibitor of metalloproteinases 2.

The pulmonary tree is exposed to neutrophil-derived serine proteinases and matrix metalloproteinases in inflammatory lung diseases, but the degree to which these enzymes participate in tissue injury remains undefined, as does the therapeutic utility of antiproteinase-based interventions. To address these issues, an in vivo rat model was examined in which the intrapulmonary deposition of immune complexes initiates a neutrophil-mediated acute alveolitis. In vitro studies demonstrated that rat neutrophils can release neutrophil elastase and cathepsin G as well as a neutrophil progelatinase, which was subsequently activated by either chlorinated oxidants or serine proteinases. Based on structural homologies that exist between rat and human neutrophil proteinases, rat neutrophil elastase and cathepsin G activities could be specifically regulated in vitro by recombinant human secretory leukoproteinase inhibitor, and rat neutrophil gelatinase activity proved sensitive to inhibition by recombinant human tissue inhibitor of metalloproteinases 2. When either of the recombinant antiproteinases were instilled intratracheally, in vivo lung damage as assessed by increased permeability or hemorrhage was significantly reduced. Furthermore, the coadministration of the serine and matrix metalloproteinase inhibitors almost completely prevented pulmonary damage while effecting only a modest decrease in neutrophil influx. These data support a critical role for neutrophil-derived proteinases in acute lung damage in vivo and identify recombinant human secretory leukoproteinase and recombinant human tissue inhibitor of metalloproteinases 2 as potentially efficacious interventions in inflammatory disease states.

Chemiluminescence in activated human neutrophils: role of buffers and scavengers.

Human neutrophils (PMNs) suspended in Hanks' balanced salt solution (HBSS), which are stimulated either by polycation-opsonized streptococci or by phorbol myristate acetate (PMA), generate nonamplified (CL), luminol-dependent (LDCL), and lucigenin-dependent chemiluminescence (LUCDCL). Treatment of activated PMNs with azide yielded a very intense CL response, but only a small LDCL or LUCDCL responses, when horse radish peroxidase (HRP) was added. Both CL and LDCL depend on the generation of superoxide and on myeloperoxidase (MPO). Treatment of PMNs with azide followed either by dimethylthiourea (DMTU), deferoxamine, EDTA, or detapac generated very little CL upon addition of HRP, suggesting that CL is the result of the interaction among H2O2, a peroxidase, and trace metals. In a cell-free system practically no CL was generated when H2O2 was mixed with HRP in distilled water (DW). On the other hand significant CL was generated when either HBSS or RPMI media was employed. In both cases CL was markedly depressed either by deferoxamine or by EDTA, suggesting that these media might be contaminated by trace metals, which catalyzed a Fenton-driven reaction. Both HEPES and Tris buffers, when added to DW, failed to support significant HRP-induced CL. Nitrilotriacetate (NTA) chelates of Mn2+, Fe2+, Cu2+, and Co2+ very markedly enhanced CL induced by mixtures of H2O2 and HRP when distilled water was the supporting medium. Both HEPES and Tris buffer when added to DW strongly quenced NTA-metal-catalyzed CL. None of the NTA-metal chelates could boost CL generation by activated PMNs, because the salts in HBSS and RPMI interfered with the activity of the added metals. CL and LDCL of activated PMNs was enhanced by aminotriazole, but strongly inhibited by diphenylene iodonium (an inhibitor of NADPH oxidase) by azide, sodium cyanide (CN), cimetidine, histidine, benzoate, DMTU and moderately by superoxide dismutase (SOD) and by deferoxamine LUCDCL was markedly inhibited only by SOD but was boosted by CN. Taken together, it is suggested that CL generated by stimulated PMNs might be the result of the interactions among, NADPH oxidase, (inhibitable by diphenylene iodonium), MPO (inhibitable by sodium azide), H2O2 probably of intracellular origin (inhibitable by DMTU but not by catalase), and trace metals that contaminate salt solutions. The nature of the salt solutions employed to measure CL in activated PMNs is critical.

Killing of endothelial cells and release of arachidonic acid. Synergistic effects among hydrogen peroxide, membrane-damaging agents, cationic substances, and proteinases and their modulation by inhibitors.

51Chromium-labeled rat pulmonary artery endothelial cells (EC) cultivated in MEM medium were killed, in a synergistic manner, by mixtures of subtoxic amounts of glucose oxidase-generated H2O2 and subtoxic amounts of the following agents: the cationic substances, nuclear histone, defensins, lysozyme, poly-L-arginine, spermine, pancreatic ribonuclease, polymyxin B, chlorhexidine, cetyltrimethyl ammonium bromide, as well as by the membrane-damaging agents phospholipases A2 (PLA2) and C (PLC), lysolecithin (LL), and by streptolysin S (SLS) of group A streptococci. Cytotoxicity induced by such mixtures was further enhanced by subtoxic amounts either of trypsin or of elastase. Glucose-oxidase cationized by complexing to poly-L-histidine proved an excellent deliverer of membrane-directed H2O2 capable of enhancing EC killing by other agonists. EC treated with rabbit anti-streptococcal IgG were also killed, in a synergistic manner, by H2O2, suggesting the presence in the IgG preparation of cross-reactive antibodies. Killing of EC by the various mixtures of agonists was strongly inhibited by scavengers of hydrogen peroxide (catalase, dimethylthiourea, MnCl2), by soybean trypsin inhibitor, by polyanions, as well as by putative inhibitors of phospholipases. Strong inhibition of cell killing was also observed with tannic acid and by extracts of tea, but less so by serum. On the other hand, neither deferoxamine, HClO, TNF, nor GTP gamma S had any modulating effects on the synergistic cell killing. EC exposed either to 6-deoxyglucose, puromycin, or triflupromazin became highly susceptible to killing by mixtures of hydrogen peroxide with several of the membrane-damaging agents. While maximal synergistic EC killing was achieved by mixtures of H2O2 with either PLA2, PLC, LL, or with SLS, a very substantial release of [3H]arachidonic acid (AA), PGE2, and 6-keto-PGF occurred only if a proteinase was also added to the mixture of agonists. The release of AA from EC was markedly inhibited either by scavengers of H2O2, by proteinase inhibitors, by cationic agents, by HClO, by tannic acid, and by quinacrin. We suggest that cellular injury induced in inflammatory and infectious sites might be the result of synergistic effects among leukocyte-derived oxidants, lysosomal hydrolases, cytotoxic cationic polypeptides, proteinases, and microbial toxins, which might be present in exudates. These "cocktails" not only kill cells, but also solubilize AA and several of its metabolites. However, AA release by the various agonists can be also achieved following attack by leukocyte-derived agonists on dead cells. It is proposed that treatment by "cocktails" of adequate antagonists might be beneficial to protect against cellular injury in vivo.

Mesangial cell killing by leukocytes: role of leukocyte oxidants and proteolytic enzymes.

Mesangial cells from human and rat kidney were examined for sensitivity to killing by neutrophils. Cells from both species were sensitive to killing by phorbol myristate acetate-stimulated neutrophils. Catalase was highly protective while superoxide dismutase was less protective and a number of protease inhibitors were not protective. Strong protection was also observed with the iron chelators, deferoxamine and phenanthroline, and with the hydroxyl radical scavengers, dimethylthiourea and 5,5-dimethyl-1-pyrroline N-oxide. Pretreatment of the mesangial cells with deferoxamine followed by washing also provided protection. Mesangial cells were also killed by reagent hydrogen peroxide (H2O2) but were much less sensitive to injury by direct application of proteolytic enzymes. The ability of H2O2 to injure mesangial cells was prevented by pre-incubation of the H2O2 with human leukocyte myeloperoxidase. These data suggest that killing is due primarily to the generation of H2O2 by the stimulated neutrophils and its further reduction in an iron-catalyzed reaction. The hydroxyl radical may be the reduction product that actually mediates lethal injury but lack of scavenger specificity prevents definitively concluding this. Mesangial cell killing by activated neutrophils could be significantly inhibited by monoclonal antibodies to CD11/CD18 molecules, suggesting that close contact between the target and effector cells is required for cytotoxicity. Although qualitatively similar to endothelial cells, the mesangial cells appeared to be quantitatively more oxidant sensitive than previously examined human and rat endothelial cells. Taken together, these data show that mesangial cells from rat and human are sensitive to leukocyte-induced injury and that injury results via an oxidant pathway.(ABSTRACT TRUNCATED AT 250 WORDS)

Human umbilical vein endothelial cell killing by activated neutrophils. Loss of sensitivity to injury is accompanied by decreased iron content during in vitro culture and is restored with exogenous iron.

First passage human umbilical vein endothelial cells (HUVECs) were sensitive to killing by activated neutrophils and reagent hydrogen peroxide (H2O2). Catalase and deferoxamine prevented killing whereas soybean trypsin inhibitor and superoxide dismutase did not. In these regards, HUVECs are similar to previously characterized endothelial cells from bovine and rat. Although first passage HUVECs were killed by activated neutrophils, sensitivity fell off rapidly as the cells were maintained in culture. At passage 2 (four population doublings), and beyond, HUVECs were highly resistant. The cells also became resistant to killing by reagent H2O2. The acquisition of resistance to killing was not accompanied by a failure to up-regulate neutrophil adhesion molecules or to support neutrophil adhesion. Levels of intracellular anti-oxidants (total thiols, though not glutathione, glutathione peroxidase or catalase activity) increased as a function of passage in culture. However, levels of glutathione and total thiols in late passage (resistant) HUVECs were similar to levels in late passage rat pulmonary artery endothelial cells, that were sensitive to killing by activated neutrophils. Cell-associated iron in HUVECs fell as a function of time in culture. By passage 2, the amount of total iron measurable with the Ferrozine reagent was only about 30% of the amount recovered from first passage HUVECs. The loss of iron from the cells may underlie much of the concomitant resistance to killing because when the cells were pretreated with iron under conditions in which it could be taken up, sensitivity to killing by activated neutrophils and by H2O2 was restored.

Modulation of Ca2+ levels in keratinocytes by all-trans retinoic acid.

Human epidermal keratinocytes, that have been growth-arrested by removal of epidermal growth factor from the culture medium, are stimulated to proliferate by all-trans retinoic acid (RA). The same treatment inhibits the onset of differentiated features and reduces cell-substrate adhesion. In the present study we show that the same treatment results in a decrease in total cell-associated Ca2+ as measured by changes in the amount of 45Ca2+ bound to cells at equilibrium following RA treatment and by a decrease in intracellular free Ca2+ levels as measured with the Ca(2+)-sensitive dye, Indo-1. The alterations in Ca2+ levels were evident within an hour after RA treatment, were in the range of 30-35% and occurred over the same RA concentration range that stimulated proliferation (i.e., 0.25-1.0 micrograms/ml). When the extracellular Ca2+ concentration was elevated from the normal level of 0.15-1.4 mM, intracellular free Ca2+ increased by a factor of 2 while total cell-associated Ca2+ increased approximately 6-fold. Even under conditions of high extracellular Ca2+, RA was able to reduce cell-associated and intracellular free Ca2+. These data indicate that RA has the capacity to lower Ca2+ levels in keratinocytes concomitantly with its effects on biological behavior.

Effects of all-trans retinoic acid on neutrophil-mediated endothelial cell injury in vitro and immune complex injury in rats.

All-trans retinoic acid (RA) has beneficial effects when used in a variety of inflammatory skin conditions. In this study, the authors found that RA inhibited superoxide anion production and proteolytic enzyme release by human and rat neutrophils. Concomitantly, the authors found that RA-treated neutrophils were less able than untreated neutrophils to injure endothelial cells in culture even though the adhesion of the RA-treated neutrophils to endothelial cell monolayers was not diminished. Inhibition of cytotoxicity occurred over the same range of concentrations that inhibited oxygen radical formation and protease release. In additional studies, it was observed that pretreatment of endothelial cells with RA-induced resistance to subsequent injury by activated neutrophils. Finally, in vivo studies showed that pretreatment of rats for 3 days with RA (1-10 mg/day, IP) reduced the degree of injury in the lungs and skin sites after treatment with bovine serum albumin and antibodies to bovine serum albumin in the reverse-passive Arthus reaction. Thus, RA can modulate neutrophil-mediated endothelial cell injury by an effect on both the neutrophils and their target cells. Together, these effects may underlie the reduction in immune complex-mediated injury seen in experimental animals. The beneficial effects that retinoids have in a variety of inflammatory skin diseases may likewise be a reflection of their effects on the physiology of both neutrophils and endothelial cells.

Hydrogen peroxide-induced cell and tissue injury: protective effects of Mn2+.

Recent evidence indicates that under in vitro conditions, superoxide anion and hydrogen peroxide (H2O2) are unstable in the presence of manganese ion (Mn2+). The current studies show that in the presence of Mn2+, H2O2-mediated injury of endothelial cells is greatly attenuated. A source of bicarbonate ion and amino acid is required for Mn2+ to exert its protective effects. Injury by phorbol ester-activated neutrophils is also attenuated under the same conditions. EDTA reverses the protective effects. Acute lung injury produced in vivo in rats by intratracheal instillation of glucose-glucose oxidase is almost completely blocked in rats treated with Mn2+ and glycine. Conversely, treatment of rats with EDTA, a chelator of Mn2+, markedly accentuates lung injury caused by glucose-glucose oxidase. These data are consistent with the findings of others that Mn2+ can facilitate direct oxidation of amino acids with concomitant H2O2 disproportionation. This could form the basis of a new therapeutic approach against oxygen radical-mediated tissue injury.

Inhibition of epithelial cell adhesion by retinoic acid. Relationship to reduced extracellular matrix production and alterations in Ca2+ levels.

Human squamous epithelial cells maintained in growth factor-deficient medium were examined for sensitivity to all-trans retinoic acid (retinoic acid). Under conditions of low external Ca2+ (0.15 mmol/l [millimolar]), or high external Ca2+ (1.4 mmol/l), retinoic acid stimulated proliferation. Concomitantly, cell-substrate adhesion was decreased. Enzyme-linked immunosorbent assays were used to assess production of two extracellular matrix components, ie, fibronectin and thrombospondin. In the presence of retinoic acid, production of both was decreased. Because both fibronectin and thrombospondin serve as epithelial cell adhesion factors, the decreased production of these moieties could contribute to reduced adhesion. Using 45Ca2+ to measure total cell-associated Ca2+ and the Ca2(+)-sensitive dye Indo-1 to measure intracellular free Ca2+, it was found that concentrations of retinoic acid that altered cell-substrate adhesion in the squamous epithelial cells had no effect on total, cell-associated Ca2+, but reduced intracellular free Ca2+ by 50% to 60%. Because Ca2+ is a regulator of adhesion, the ability of retinoic acid to modulate Ca2+ levels in the squamous epithelial cells may explain, in part, how retinoic acid influences their adhesiveness.

H2O2-mediated cytotoxicity of rat pulmonary endothelial cells. Changes in adenosine triphosphate and purine products and effects of protective interventions.

H2O2-mediated cytotoxicity (as measured by 51Cr-release) of rat pulmonary artery endothelial cells was time-dependent and related to the concentration of H2O2 employed. The cytotoxic effects of H2O2 were, as expected, prevented by catalase and the degree of protection was directly related to its time of addition. Endothelial cells were incubated with [14C]adenosine to achieve intracellular labeling of ATP, after which the cells were exposed to H2O2. Based on analysis of cell extracts by high-performance liquid chromatography, there was a time-dependent loss of intracellular radioactivity and ATP with the simultaneous appearance of purine degradation products including xanthine/hypoxanthine. Approximately 50% of the intracellular ATP was lost after 15 minutes of exposure and up to 80% was lost by 30 minutes. The extracellular fluid of cells exposed to H2O2 contained significant amounts of xanthine/hypoxanthine. The ferric iron chelator deferoxamine provided almost complete protection against H2O2-mediated cytotoxicity. Two inhibitors of xanthine oxidase, allopurinol and oxypurinol, were also protective as was deoxycoformycin, an inhibitor of adenosine deaminase. Remarkably, cells protected by these agents showed the same loss of intracellular ATP as unprotected, H2O2-treated cells. These findings demonstrate the dissociation between ATP loss per se and oxidant injury of endothelial cells. ATP breakdown may be an important event leading to cellular injury in that this results in the formation of substrate for xanthine oxidase.

Interaction of mammalian cells with polymorphonuclear leukocytes: relative sensitivity to monolayer disruption and killing.

Monolayers of murine fibrosarcoma cells that had been treated either with histone-opsonized streptococci, histone-opsonized Candida globerata, or lipoteichoic acid-anti-lipoteichoic acid complexes underwent disruption when incubated with human polymorphonuclear leukocytes (PMNs). Although the architecture of the monolayers was destroyed, the target cells were not killed. The destruction of the monolayers was totally inhibited by proteinase inhibitors, suggesting that the detachment of the cells from the monolayers and aggregation in suspension were induced by proteinases releases from the activated PMNs. Monolayers of normal endothelial cells and fibroblasts were much resistant to the monolayer-disrupting effects of the PMNs than were the fibrosarcoma cells. Although the fibrosarcoma cells were resistant to killing by PMNs, killing was promoted by the addition of sodium azide (a catalase inhibitor). This suggests that the failure of the PMNs to kill the target cells was due to catalase inhibition of the hydrogen peroxide produced by the activated PMNs. Target cell killing that occurred in the presence of sodium azide was reduced by the addition of a "cocktail" containing methionine, histidine, and deferoxamine mesylate, suggesting that hydroxyl radicals but not myeloperoxidase-catalyzed products were responsible for cell killing. The relative ease with which the murine fibrosarcoma cells can be released from their substratum by the action of PMNs, coupled with their insensitivity to PMN-mediated killing, may explain why the presence of large numbers of PMNs at the site of tumors produced in experimental animals by the fibrosarcoma cells is associated with an unfavorable outcome.

Endothelial cell killing by neutrophils. Synergistic interaction of oxygen products and proteases.

Killing of rat pulmonary artery endothelial cells by activated polymorphonuclear leukocytes (PMNs), as measured at 4 hours, is catalase sensitive, iron dependent, and unaffected by addition of protease inhibitors. If the time course for exposure of endothelial cells to activated PMNs is extended to 18 hours, progressive injury occurs. Endothelial cell injury resulting at 18 hours is partially inhibited by catalase and partially inhibited by soybean trypsin inhibitor. Together, these two inhibitors function synergistically to protect the cells from injury. Exposure of endothelial cells to reagent H2O2 and purified proteolytic enzymes (trypsin, chymotrypsin, elastase, and cathepsin G) mimics the effects of activated PMNs: H2O2 alone is cytotoxic with maximal killing achieved by 4 hours; proteolytic enzymes produce cytotoxicity only at high concentrations and only after prolonged incubation (longer than 8 hours); and, in combination, H2O2 and proteolytic enzymes act synergistically. These data provide compelling evidence that PMN-mediated injury of endothelial cells involves interaction between oxygen products and proteases.

Vascular endothelial cell killing by combinations of membrane-active agents and hydrogen peroxide.

Previous studies have demonstrated that a number of membrane-active agents are capable of binding to the surface of polymorphonuclear leukocytes (PMN) resulting in an augmentation of superoxide anion and hydrogen peroxide (H2O2) production in response to soluble stimuli. It is now demonstrated that these same membrane-active agents can bind to the surface of endothelial cells and enhance their susceptibility to killing by H2O2. Membrane-active agents which are capable of synergizing with H2O2 include cationic proteins, cationic poly-amino acids, lysophosphatides and enzymes which are capable of degrading membrane phospholipids (e.g., phospholipase C, phospholipase A2 and streptolysin S). In each case, treatment of the target cells with the membrane-active agent and H2O2 produces greater damage than the sum of the damage produced by either agent separately. Since inflammatory lesions, particularly sites of bacterial infection, may contain a rich mixture of cationic substances, phospholipases and phospholipid breakdown products, these substances may contribute to the tissue damage observed at sites of inflammation by enhancing endothelial cell sensitivity to PMN-generated H2O2 as well as by augmenting the generation of H2O2 by PMNs.