Modulation of the fibrinolytic response of cultured human vascular endothelium by extracellularly generated oxygen radicals.

Clinical and experimental data indicate that activated oxygen species interfere with vascular endothelial cell function. Here, the impact of extracellular oxidant injury on the fibrinolytic response of cultured human umbilical vein endothelial (HUVE) cells was investigated at the protein and mRNA levels. Xanthine (50 microM) and xanthine oxidase (100 milliunits), which produces the superoxide anion radical (O2-) and hydrogen peroxide (H2O2), was used to sublethally injure HUVE cells. Following a 15-min exposure, washed cells were incubated for up to 24 h in serum-free culture medium. Tissue-type plasminogen activator (t-PA) antigen, plasminogen activator inhibitor-1 (PAI-1) antigen, and PAI-1 activity were determined in 1.25 ml of conditioned medium and t-PA and PAI-1 mRNA in the cell extracts of 2 x 10(6) HUVE cells. Control cells secreted 3.9 +/- 1.3 ng/ml (mean +/- S.D., n = 12) within 24 h. Treatment with xanthine/xanthine oxidase for 15 min induced a 2.8 +/- 0.4-fold increase (n = 12, p less than 0.05) of t-PA antigen secretion after 24 h. The t-PA antigen was recovered predominantly in complex with PAI-1. The oxidant injury caused a 3.0 +/- 0.8-fold increase (n = 9, p less than 0.05) in t-PA mRNA within 2 h. Total protein synthesis was unaltered by xanthine/xanthine oxidase. The oxidant scavengers superoxide dismutase and catalase, in combination, abolished the effect of xanthine/xanthine oxidase on t-PA secretion and t-PA mRNA synthesis. Xanthine/xanthine oxidase treatment of HUVE cells did not affect the PAI-1 secretion in conditioned medium nor the PAI-1 mRNA levels in cell extracts. Thus extracellular oxidant injury induces t-PA but not PAI-1 synthesis in HUVE cells.

and urokinase-type plasminogen activator (u-PA), highly specific but structurally distinct serine proteases which convert plasminogen to plasmin (1). The PA activity is controlled by two potent specific plasminogen activator inhibitors (PAIs) designated PAI-1 and PAI-2 (2,3). The endothelial cell synthesizes and secretes t-PA, u-PA, PAI-1, and possibly  and contributes to the regulation of intravascular fibrinolysis. Diverse factors, including enzymes, hormones, and cytokines, modulate the fibrinolytic properties of endothelial cells thereby altering the balance between PAS and PAIs. Intracellularly generated oxidants during anoxia/reoxygenation markedly alter both t-PA and PAI-1 synthesis and secretion (7).
The vascular endothelium is a target of oxidant injury in a variety of inflammatory conditions, possibly mediated by activated phagocytes which release oxygen radicals (8). In the present study we have investigated the effect of exposure to sublethal extracellular oxidants, generated by a mixture of xanthine and xanthine oxidase, on the expression of components of the fibrinolytic system in primary cultures of human umbilical vein endothelial (HUVE) cells. The results show that the synthesis and secretion of t-PA antigen and mRNA but not of PAI-1 antigen and mRNA are increased following exposure to oxidants.
Preparation of the Oxidant Generating System-A mixture of xanthine (50 p M ) and xanthine oxidase (100 milliunits) in HEPES-Tyrode buffer which generates superoxide anion radicals and/or hydrogen peroxide (11,12) was used. Xanthine oxidase was dialyzed overnight at 4 "C against 10 mM sodium phosphate buffer containing 1 mM EDTA, pH 7.8, sterilized by filtration through a 0.2-pm filter and stored at 4 "C. Dialyzed xanthine oxidase was monitored for possible contaminating trypsin-like activity with N-benzoyl-L-arginine ethyl ester (13). Before each experiment, the activity of xanthine oxidase was measured using the ferricytochrome c assay (11). The amount of superoxide produced was calculated using an absorbancy at A,,,,, = 2.1 X lo4 M" cm" for reduced cytochrome c. Xanthine (50 p~) and xanthine oxidase (100 milliunits) in our system produced 2.3 to 2.9 nm of O;/min for the first 20 min. The rate of 0; production was constant throughout the cells' exposure to the generating system. Determination of Sublethal Oxidant Injury-The sublethal effects of xanthine (50 p~) and xanthine oxidase (100 milliunits) as compared to control cultures were determined by morphological observation using light microscopy and by viability studies consisting of trypan blue dye exclusion (14) and of retention of fluorescein diacetate (15). The release of 61Chromium (16) and lactate dehydrogenase (17) into the medium was also quantitated. Control and treated cultures were assessed for injury immediately following the initial 15min treatment and after a subsequent 24-h incubation in HEPES-Tyrode buffer.
Experimental Protocol-Primary cultures were grown to confluence as described above. Medium was removed from the cells and monolayers washed twice with Hanks' balanced salt solution. Xanthine (50 p~) and xanthine oxidase (100 milliunits) were then added in HEPES-Tyrode buffer or, in control experiments, buffer alone for 15 min at 37 "C. Immediately following treatment, medium was removed, cells were washed twice with Tyrode buffer, and incubated with fresh HEPES-Tyrode buffer for up to 32 hours. Conditioned medium then was collected, centrifuged at 10,000 X g to remove cellular debris, and stored following the addition of PBS-Tween 80 at -20 "C for assay of antigen levels of t-PA, u-PA, PAI-1, and of PAI-1 activity. Cells were lysed with 1 ml/well of 5 M guanidine isothiocyanate and stored at -80 "C until RNA isolation. In other experiments, the remaining cells were treated with lysing buffer for 1 h at 4 'C, scraped, and centrifuged at 10,000 X g. The supernatant was collected and frozen at -20 "C for antigen and activity assays.
Additional Protocols-In order to evaluate the protective effect of antioxidant scavengers, 200 pg/ml superoxide dismutase and 50 pg/ ml catalase were added to HUVE cultures. Inactivated superoxide dismutase and catalase were used in order to validate the specificity of the scavengers. Superoxide dismutase was inactivated by incubation with H202 at pH 10 (18) and catalase by incubation with 3amino-l,2,4-triazole and H202 (19). Inhibition of protein synthesis was accomplished with 1 pg/ml of cycloheximide in serum-free medium following the initial 15-min treatment.
Assay Techniques-The concentration of t-PA antigen was measured with an ELISA as described (20) and expressed in nanograms/ milliliter by comparison with the International Reference Preparation for t-PA (National Institute for Biological Standards and Control, London) which contains approximately 500,000 IU/mg. u-PA antigen was measured by ELISA as described (21). PAI-1 antigen was measured in conditioned medium, cell lysates, and extracellular matrix using monoclonal antibody-based specific ELISAs described elsewhere (22). PAI-1 activity was determined by the method of Verheijen et al. (23) and expressed in International Units of t-PA neutralized. PAI-1, which is converted to an inactive form during routine cell culture conditions, was reactivated by the method of Levin and Santell(24) in order to quantitate latent PA1 activity. Briefly, samples were made 0.2% in SDS, incubated at 37 "C for 10 min, and then made 2% in Triton X-100, by dialysis prior to assay performance.
Molecular characterization of secretion products was undertaken in the following way. Conditioned medium was harvested and concentrated from 10 to 1 ml by ultrafiltration on a Centricon-30 membrane (Amicon Corp., Lexington, MA). The concentrate was then subjected to immunoadsorption with rabbit anti-t-PA IgG. The antibodies were isolated by chromatography of rabbit anti-human t-PA serum on Protein A-Sepharose and were coupled to cyanogen bromide activated-Sepharose 4B. For immunoadsorption, 20 pl of a 50% substituted Sepharose slurry was added to 1 ml of media concentrate. Immunoadsorbed t-PA antigen was then eluted by incubation in SDS sample buffer prior to 10% SDS-polyacrylamide gel electrophoresis (25). Fractionated proteins were electrophoretically transferred to nitrocellulose and immunoblotted to the same affinity purified rabbit anti-t-PA IgG (20 pg/ml) by the method of Towbin et al. (26).  (22). The protein concentration of cell-free conditioned medium was determined by the method of Bradford (28) using bovine serum albumin for calibration.
Preparation and Isolation of HUVE Cell Extracellular Matrix (ECM)-HUVE ECM were prepared from confluent primary monolayers according to the method of Knudsen et al. (29). The culture medium was first removed and centrifuged at 12,000 X g for 2 min to remove floating cells and debris. The resulting supernate was collected and made 0.01% in Tween 20, and then frozen at -20 "C until analysis. The monolayer was then washed three times with PBS. Cells were lysed by treatment with 0.5% Triton X-100 in PBS for 10 min at room temperature, and cell lysates were collected and stored at -20 "C. The wells were washed with PBS and treated with 25 mM NH,OH (approximately 10 min) until light microscopic examination showed the removal of residual cell debris from the surface of the well. The material remaining on the surface after this treatment is defined as ECM. Wells were then washed with PBS and ECM was extracted by scraping into 0.2% SDS in PBS. ECM samples were then made 2% in Triton X-100 and stored at -20 "C.
Qunntitation of t-PA and PAI-1 mRNA Levels by Slot Blot and Northern Blot Analysis-Total cellular RNA was extracted from cells by lysis with 1 ml/well of 5 M guanidinium isothiocyanate, 25 mM sodium citrate, pH 7.0, containing 0.5% Sarkosyl and 8% 8-mercaptoethanol, as described by Chomczynski and Sacchi (30), followed by cold phenol extraction (performed only once). The final RNA pellet was resuspended in 50 pl of HzO and the concentration determined by absorbance at 260 nm.
RNA was denatured with 6 M gloxal for 1 h at 50 "C and 2.0, 1.0, or 0.5 pg was applied to a nylon membrane (Zetaprobe, Bio-Rad) according to the manufacturer's instructions, using a slot blot filtration apparatus (Schleicher & Schuell). A standard curve of known concentrations of t-PA or PAI-1 mRNA was applied to each membrane. The t-PA and PAI-1 mRNA used for the construction of these calibration curves were prepared using an in vitro SP6 polymerase system (Riboprobe; Promega, Leiden, the Netherlands) and quantitated by absorbance at 260 nm (31). A 2000 base pair long PAI-1 cDNA, containing most of the coding sequence, and 700 base pairs of the 3' untranslated sequence (32) was cloned into pSP65. Alternatively, the BglII cDNA fragment of t-PA (nucleotides 230-2205) according to the published t-PA cDNA sequence (33) was cloned into pSP65. Human spleen RNA was used as a control to determine nonspecific hybridization. After application, the membrane was baked for 1 h at 80 "C under vacuum and prehybridized for at least 4 h at 65 "C in the following hybridization mixture: 50% formamide, 5 X SSC (saline-sodium citrate buffer), 10 X Denhardt's solution, 50 mM phosphate buffer, pH 7.6, 5% dextran sulfate, 1 mM EDTA, 1% SDS containing heat-denaturing transfer RNA (500 rg/ml) and sonicated, heat-denatured salmon sperm DNA (200 pglml).
Hybridization was performed overnight at 65 "C in the same solution containing specific "P-RNA probes for t-PA mRNA, PAI-1 mRNA, or p-actin mRNA. Antisense RNA probes were generated by either SP6 polymerase transcription of linearized pSP65 plasmids containing nucleotides 278-885 of t-PA or nucleotides 1045-1481 of PAI-1 or by T7 polymerase transcription of a linearized pGemZ plasmid containing nucleotides 616-1015 of @-actin according to the published cDNA sequences (32-34). The probes were prepared freshly using a Promega Transcription Kit (Promega, Leiden, the Netherlands) and routinely had a specific activity of 0.5-2.0 X log cpmlpg RNA. The probes were heat denatured and added to the hybridization solution at a concentration of 1 0 ' cprn/ml. Membrane washing conditions at 65 "C were as described by the manufacturer of Zetaprobe with a final wash of 0.05 X SSC at 65 "C. Autoradiography was carried out using Hyperfilm (Amersham, Brussels, Belgium) at -72 "C. Quantitation was achieved by measuring the radioactivity of each slot by liquid scintillation counting. For Northern blot analysis, glyoxal-treated RNA (5 or 10 pg) was electrophoresed in a 1% agarose gel followed by capillary transfer to a Zetaprobe membrane. Hybridizations were performed as described for slot blots.
Statistics-Statistical significance of experimental results was determined using the Student's t test for nonpaired comparisons.

Effect of Xanthine/Xanthine Oxidase on the Synthesis and
Secretion of Fibrinolytic Components t-PA-A 15-min exposure of HUVE cells to 50 p~ xanthine and 100 milliunits of xanthine oxidase had no significant effect upon HUVE cell viability/cytotoxicity when assayed immediately following oxidant exposure or following a subsequent 24-h period of incubation, as judged from the morphology, trypan blue dye exclusion, fluorescein diacetate inclusion, and 'lChromium or lactate dehydrogenase release. Xanthine/xanthine oxidase treatment resulted in a significant increase in t-PA antigen within 16-32 h, with a 2.8 f 0.4-fold increase at 24 h (Table I). Histamine, a known agonist of t-PA antigen (35) secretion resulted in a 4-7-fold stimulation (Table 11).
Addition of xanthine or xanthine oxidase alone did not alter t-PA antigen secretion, higher concentrations or pro-  Incubation of HUVE cells with 1.0 pg/ml of cycloheximide, a protein synthesis inhibitor, following 15 min of incubation with buffer, histamine, or xanthine/xanthine oxidase, resulted in a marked reduction of t-PA antigen secretion within 24 h (Table 11), but did not affect cell viability as assessed by trypan blue dye exclusion.
Western blotting with anti-t-PA antiserum of 24-h conditioned medium (Fig. 1) revealed a main band with M, = 110,000, representing t-PA presumably in a 1:l complex with PAI-1, but no free t-PA at M, 70,000. Conditioned medium from cells stimulated with histamine (lane B) or xanthine/ xanthine oxidase ( l u n e C) showed an increase in this band, corresponding to the increase observed in t-PA antigen synthesis and secretion when assayed by ELISA. PAI-1-Xanthine/xanthine oxidase or histamine treatment of HUVE had no effect on PAI-1 antigen nor upon t-PA inhibitor activity before or after reactivation (Table I).

Effects of Oxygen Radical Scavengers on the Xanthine/ Xanthine Oxidase-induced Synthesis and Secretion of Fibrinolytic Components
Simultaneous addition of superoxide dismutase (200 pg/ml) and catalase (50 pg/ml) abolished the effect of xanthine/ xanthine oxidase on t-PA secretion by HUVE cells (Table   200- 111). When superoxide dismutase (200 pglml) and catalase (50 pglml) were added individually to the test system, neither was as effective as the combination of superoxide dismutasel catalase in preventing the xanthinelxanthine oxidase induced increase in t-PA antigen secretion by HUVE cells. Chemically or enzymatically inactivated superoxide dismutase and catalase at identical concentrations had no effect on the oxidantinduced stimulation of t-PA antigen secretion (Table 111).

FIG
Effects of XanthinelXanthine Oxidase on mRNA Levels of Fibrinolytic Components Total cellular RNA was obtained from HUVE cells which had been exposed to xanthinelxanthine oxidase with or without the simultaneous addition of 200 pg/ml superoxide dismutase and 50 pg/ml catalase. Slot blot analysis revealed an increase in t-PA mRNA which reached a peak at 2-h post treatment (3 f 0.8-fold increase) declining to 1.8 f 0.1-fold at 8 h and 2.4 f 0.1-fold at 24 h (Table IV). superoxide dismutase/catalase markedly reduced the effect of xanthine/ xanthine oxidase on t-PA mRNA at 2,8, and 24 h (Table IV). Fig. 2 shows results of Northern blot analysis of total RNA obtained from HUVE cells harvested 2 h after treatment with xanthinelxanthine oxidase. t-PA mRNA (2.8 kilobases) is barely detectable in control cells ( l a n e A ) , it is markedly increased in treated HUVE cells ( l a n e B ) , and this increase  is virtually abolished with superoxide dismutase and catalase. Hybridization with the @-actin probe showed that equivalent amounts of RNA had been applied to the gel and that 8-actin mRNA (1.8 kilobases) levels were not changed by oxidants or scavengers.

DISCUSSION
Regulation of intravascular fibrinolysis is partly obtained at the level of the endothelial cell, a significant site of synthesis of circulating t-PA and PAI-1 (37). Derangements leading to endothelial cell injury can cause cellular dysfunction resulting in abnormal hemostasis and in thrombotic and hemorrhagic disorders in man (38-40). A number of factors have been reported to affect both t-PA and PAI-1 activity in cultured endothelium.
The present study demonstrates that cultured human vascular endothelial cells respond to a brief, sublethal extracellular oxidant perturbation. A 15-min exposure of HUVE cells to the xanthinelxanthine oxidase mixture caused a profibrinolytic response consisting of increased t-PA synthesis and secretion without affecting the inhibitor PAI-1. In addition, an increase of t-PA mRNA was observed that was suggestive of enhanced transcription.
The exogenous addition of the oxygen radical scavengers superoxide dismutase/catalase in combination ablated the increase in t-PA synthesis and secretion as well as mRNA, confirming that the endothelial cell profibrinolytic response is specifically mediated via extracellular oxidant generation.
Oxygen-free radicals are generated in cells and tissues under a variety of conditons and thus mediate many pathologic processes. They induce injury to cells by oxidizing proteins, initiating lipid peroxidation, inactivating enzymes, etc. (41). Yet, neither the magnitude nor mechanism(s) of target cell injury is well-defined.
Endothelial cells located on the luminal surface of the vascular wall are vulnerable to attack by reactive oxygen species released by activated phagocytes as they are recruited to sites of injury. It is unclear to what extent nonlethal oxidant exposure generated by cell or cell-free systems may interfere with overall endothelial cell function(s). Cell lysis and death occur following long term exposure (42), whereas shorter treatments have led to disruption of several functions: plasma membrane organization (43), shape and transport (44), hydrolysis of inositol phospholipids (45), and impaired low density lipoprotein receptor-mediated endocytosis (46). We have earlier described alterations in platelet adherence and prostacyclin release from xanthinelxanthine oxidase-treated endothelium (47).
Although the xanthinelxanthine oxidase system has been extensively used to monitor the biologic impact of oxidant metabolites generated in the extracellular space, this model may be somewhat limited in providing direct information on  (48). This is in contrast to the xanthine/xanthine oxidase system which permits the accumulation of generated HzOz enabling more complex interactions between 05, transition metals, and H202 to proceed. Thus caution is mandatory when extrapolating extracellular effects mediated by cell-free systems to more physiologic sources of oxidants in uiuo. The use of a well-defined soluble generating system for oxygen radicals has several advantages which include the quantitation of dose response, neutralization, and/or amelioration with appropriate enzymes and scavangers as well as defined kinetics of onset and duration of effect.
Whether the observed profibrinolytic response of human endothelium following extracellular oxidant targeting actually occurs in uiuo remains speculative. This phenomenon may be an acute physiological response to external stimulation rather than a result of pathologic injury and remains to be further investigated.