Endothelial Cells Inhibit Receptor-mediated Superoxide Anion Production by Human Polymorphonuclear Leukocytes via a Soluble Inhibitor R. E. Basford, Robert L. Clark, Ronald A. Stiller, Sandra S. Kaplan, Douglas B. Kuhns, and Jean E. Rinaldo Departments of Microbiology, Biochemistry and Molecular Biology, Medicine, and Pathology, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania and the Center for Lung Research, Vanderbilt University, Nashville, Tennessee

Confluent monolayers of bovine pulmonary artery endothelial cells (BPAE) or human umbilical vein endothelial cells (HUVE) inhibited by 80 to 90% the production of O2- by added human neutrophils (PMNs) stimulated by plasma membrane receptor-mediated activators (formylmethionylleucylphenylalanine [fMLP], opsonized zymosan, heat-killed Staphylococci), but not by non-plasma membrane receptor-mediated activators (phorbol myristate acetate and delta-hexachlorocyclohexane). Degranulation induced by fMLP was also inhibited by BPAE. Inhibition was not affected by eicosotetraynoic acid (ETYA) or indomethacin. To assess the role of cell-cell contact, 0.45-p.m-pore culture plate inserts were employed to prevent PMN-endothelial cell contact during incubation. A similar amount of inhibition of stimulated PMNs superoxide production was seen as compared to PMN-endothelial incubations where contact occurred. A soluble component released by BPAE monolayers, when added to PMNs, duplicated the inhibition seen by BPAE-PMN co-incubation. Incubation of BPAE with adenosine deaminase did not reduce inhibition of O 2- production compared to controls without adenosine deaminase. There was no evidence of endothelial scavenging of O2- generated by hypoxanthine-xanthine oxidase, and inhibition of endothelial superoxide dismutase did not diminish the inhibitory effort. We conclude that cell contact is not required for BPAE inhibition of fMLP-stimulated O2- production by PMN, and that scavenging of superoxide anion is not the mechanism. The inhibitor appears to be a polypeptide with an apparent molecular weight between 1,000and 10,000 D and does not appear to be adenosine, an arachidonate metabolite, or superoxide dismutase. The mechanism may involve down-regulation of plasma membrane receptor-mediated activation of PMNs.

When neutrophils encounter appropriate mediators emanating from areas of acute extravascular inflammation, receptor-coupled events cause them to adhere to the walls of contiguous vessels and then to migrate through intraendothelial junctions to extravascular sites of inflammation. Concur(Received in original form April 10, 1989 and in revised form October 26, 1989) Address correspondence to: Jean E. Rinaldo, M.D., Center for Lung Research, B-1308 Medical Center North, Vanderbilt University Medical Center, Nashville, TN 37232. Dr. Kuhns' present address is: Neutrophil Monitoring Laboratory, Clinical Immunology Services, N.C.L, Frederick Cancer Research Facility, Frederick, MD 21701. Dr. Clark's present address is: Wayne State University, Department of Surgery, Section of Emergency Medicine, Detroit, MI 48201. Abbreviations: angiotensin-converting enzyme, ACE; bovine pulmonary artery endothelial cells, BPAE; diethyldithiocarbarnate, DEDTC; endothelial cell growth supplement, EGCS; eicosotetraynoic acid, ETYA; N-formylmethionylleucylphenylalanine, fMLP; delta-hexachlorocyclohexane, o-HCCH; human umbilical vein endothelial cells, HUVE; interleukin I,lL-I; Krebs Ringer phosphate buffer containing glucose, KRPG; Medium 199, M199; Ryan growth supplement, RGS. Am. J. Respir. Cell Mol. BioI. Vol. 2. pp. 235-243, 1990

rently, neutrophils may release a variety of cytotoxic substances that have the potential to injure endothelial cells. Studies in vitro have implicated reduced oxygen species (O,, HzOz, and/or OH') (1, 2), or neutral proteases, particularly elastase (3-5), as potentially damaging agents. When this sequence occurs in the lung, there is speculation that it may result in alveolar flooding, gas exchange dysfunction, and death, as in the sepsis-induced adult respiratory distress syndrome (ARDS) (6). However, this putative pathogenic mechanism remains controversial. Recent clinical studies do not fully support the hypothesis (7), and results using animal models are inconsistent (8, 9). Furthermore, neutrophil-mediated lung injury in a well-established animal model induced by the activation of complement is evanescent (10), suggesting that protective mechanisms in vivo may prevent the severe pulmonary endothelial injury mediated by the activation of intravascular neutrophils. A variety of protective mechanisms may play a role in limiting neutrophil-mediated endothelial injury. For example, serum (11) and erythrocytes (12, 13) contain free radical scavengers that substantially ameliorate neutrophil-mediated endothelial cell injury in vitro and may afford substantial pro-



tection in vivo. There is also evidence that endothelial cells manifest functional properties that tend to mitigate neutrophil-induced injury during acute inflammation. Hoover and coworkers (14, 15) reported that co-incubation of neutrophils with endothelial cells markedly inhibited the extracellular release of superoxide anion in vitro after activation of the neutrophils. Other studies have shown that endothelial cells can inactivate substantial quantities of hydrogen peroxide (16, 17). Recent preliminary reports have suggested that culture supernatants from endothelial cells attenuate the release of oxygen free radicals from activated neutrophils (18, 19). The present study was designed to test the hypothesis that soluble mediators produced by endothelial cells directly attenuate neutrophil activation sequences, and to investigate the mechanisms of such modulation. In this study, we confirm the finding of Hoover and coworkers (14, 15) that bovine pulmonary artery endothelial monolayers inhibit O2- production by PMNs. We extend these studies by showing that degranulation is similarly inhibited. We further show that chemotactic peptide, opsonized zymosan, and opsonized bacteria but not phorbol ester (PMA) or delta-hexachlorocyclohexane (a-HCCH)-stimulated O2 - production is inhibited by endothelial cell monolayers, suggesting that only plasma membrane receptor-coupled activation may be down-regulated by endothelial cells. We provide evidence that the apparent inhibition of superoxide release is not caused by endothelial scavenging of superoxide anion. We show that PMN adherence to endothelial cells is not required for inhibition, but rather that cultured endothelial cells produce and release into the medium a soluble inhibitor of PMN O2- production. These studies suggest a potential for pulmonary artery endothelial cells to down-regulate neutrophil activation by elaborating soluble products that modulate receptor-coupled activation sequences in contiguous blood neutrophils. If this mechanism is operative in vivo, it could limit neutrophil cytotoxic responses in intravascular loci, while permitting a more complete expression of activation responses at extravascular sites of inflammation.

Materials and Methods Materials and Reagents Tissue culture materials. Medium 199 (MI99), FCS, and Hepes buffer were obtained from M.A. Bioproducts (Cabin John, MD). Penicillin, streptomycin, and L-glutamine were purchased from GIBCO (Grand Island, NY). Ryan growth supplement (RGS) was obtained from the laboratory of Dr. Una S. Ryan (Department of Medicine, University of Miami). Type V collagenase and trypsin (0.1 %) were obtained from Sigma Chemical Co. (St. Louis, MO). Bovine endothelial cell growth supplement (ECGS) was purchased from Collaborative Research (Lexington, MA). For endothelial cell identification, the angiotensin-converting enzyme substrate pH]benzoyl-Phe-Ala-Pro (PH]BPAP) was obtained from the Ventrex Co. (Portland, MA). Rabbit anti-factor VIII antibody and FITC-goat-anti-rabbit antibodies were purchased from Calbiochem (San Diego, CA). Tissue culture plasticware included 25-cm~ Falcon Primeria flasks for primary culture and subculture and Falcon 2-cm~, 24-well tissue culture plates for most co-incubation assays. Millicell-HA, 0.45-/-tm-pore filter culture plates inserts, 12

mm in diameter, were obtained from the Millipore Corp. (Bedford, MA). Neutrophi l activators. N-formy ImethionylleucyIphenylalanine (fMLP) , phorbol myristate acetate (PMA), a-hexachlorocyclohexane (a-HCCH), and zymosan were purchased from Sigma. PMA and fMLP were dissolved in DMSO and a-HCCH in dimethylformanide at 10-3 M, stored at 0° C, and diluted to the indicated concentrations on the day of the experiment. Staphylococcus aureus were obtained from American Type Culture Collection (Rockville, MD). Zymosan (1 mg/ml) and heat-killed S. aureus were opsonized by incubation at 37° C for 30 min in Krebs Ringer phosphate buffer containing glucose (KRPG) (145 mM NaCl, 5 mM KCl, 0.5 mM csci; 0.5 mM MgS04, 16 mM Na~HP04, and 4.4 mM glucose) containing 50% human plasma, sedimented, washed once, and resuspended in KRPG. Other reagents. Bovine erythrocyte SOD, horse heart cytochrome c, grade 1 xanthine oxidase, hypoxanthine, 4methylumbelliferone, diethyl-dithiocarbamic acid, eicosotetraynoic acid (ETYA), 4-methylumbelliferyl-{3-D-glucuronide, and indomethacin were obtained from Sigma. NaCl, sodium acetate, KCl, CaCb, MgS04, and glucose were purchased from Fisher Scientific (Fairlawn, NJ). Isolation and Culture of Cells Bovine pulmonary artery endothelial cells (BPAE). BPAE were isolated and subcultured according to the methods of Ryan and Maxwell (20) without exposure to enzymes during primary culture or passage. Briefly, primary cultures were obtained by scraping the luminal surface of the main pulmonary artery of calves within 6 h of killing. Primary cultures were incubated in M 199 containing 10% heat-inactivated FCS, 10% RGS, penicillin (100 U/ml), streptomycin (l00 /-tg/ml), supplemental L-glutamine, and 20 mM Hepes buffer. Endothelial cell colonies were identified by morphologic criteria, harvested, and subcultured by scraping gently with a rubber policeman. After reaching confluence, endothelial cell cultures were maintained in MI99 containing 5 % FCS + 5 % RGS, fed twice weekly, and passaged weekly by scraping. The identity of the endothelial cell cultures was confirmed by determining the angiotensin-converting enzyme (ACE) activity with PH]BPAP as substrate, and an immunofluorescent assay for factor VIII using FITC-labeled goat antibody. The cultures were used from passages 5 through 15, with the ACE assay being repeated every 4 passages. For the co-incubation experiments with human PMNs, BPAE were grown to confluence in 24-well, 2-cm~ tissue culture plates and studied within 48 h of confluence. Incubation assays with PMN reported here were repeated on at least 3 different BPAE cell lines to confirm qualitative reproducibility. Human umbilical vein endothelial cells (HUVE). HUVE were obtained according to the method of Jaffee and associates (21). Briefly, human umbilical cords were obtained within several hours after delivery. The umbilical vein was cannulated with a #19 butterfly needle, and approximately 30 ml of HBSS washed through the vein. The distal end of the cord was clamped, and 10 ml of Type V collagenase 0.2 % (Sigma) infused and incubated at 37° C for 15 min. The cord was sectioned just proximal to the clamp and the collagenase, and another 30-ml wash of HBSS collected and

Basford, Clark, Stiller et al.: Endothelial Cell-derived Inhibitor of Neutrophil Activation

centrifuged at 200 x g for IS min: The cell pellet was suspended in 3 ml of MI99 containing 24 Ilmollml Hepes, 10 V/ml heparin, 100 Ilg/ml ECGS, 100 V/ml penicillin, and 100 Ilg/ml streptomycin. Cells were subcultured in T-25 tissue culture plates for experimental use. Endothelial cells were identified by their cobblestone morphology, by their ACE activity, and by their ability to express factor VIII as determined by direct immunofluorescence. Cells were passaged using 0.1% trypsin and used within the first 6 passages for these experiments. Human PMNs. Human PMNs were isolated from fresh citrated blood from patients with secondary polycythemia (Central Blood Bank of Pittsburgh) as described earlier (22). The condition of secondary polycythemia does not affect the responsiveness of PMNs to a variety of agonists (23). Assay Procedures O, - production by PMNs in the presence and absence of BPA£. Monolayers of BPAE grown on 24-well tissue culture plates in MI99 with phenol red were washed free of growth medium by 2 washes with KRPG at 37° C. BPAE in 24-well plates were incubated for 30 min at 37° C in KRPG (l mIl well), the conditioned medium aspirated off and saved, and immediately replaced by I ml of 100 11M cytochrome c in KRPG. For each condition (resting or control, stimulated or inhibited), 2 or 3 replicate wells were used plus I well containing 20 III of 2 mg/ml SOD. For most conditions tested, wells contained BPAE alone, BPAE plus I x IQ6 PMNs, or I X 106 PMNs alone. At time zero, activator (fMLP, opsonized zymosan, etc.) was added and the plates were incubated for 30 min in a Napco 5200 incubator with 95 % air/5 % COz as the gas phase. Although KRPG contains no HC0 3- , the pH of the KRPG under the described incubation conditions remained stable at pH 7.0 after 10 and 20 min of incubation and dropped to 6.95 by 30 min. At the end of the incubation period, the plates were cooled on ice and the supernatant fluid from each well was aspirated into microfuge tubes and centrifuged for 5 min in a Beckman microfuge. The supernatant fluid after centrifugation was transferred to I-ml cuvettes, and the absorbence at 550 nm measured in a Gilford spectrophotometer. Superoxide anion produced was calculated using a cytochrome c millimolar extinction coefficient of 21.1 after subtraction of the absorbance of the SOD blank. In some cases, when large numbers of samples were involved, a 0.2-ml aliquot of the final cytochrome c-containing supernatants was transferred to 96well flat-bottom plates, and the absorbance measured with a microplate reader (Titertek Multiskan MC, Type 340, ICN, McLean, VA) with a 550-nm filter. The cytochrome c millimolar extinction coefficient was reduced to 13.4 due to the reduction in light path in the 96-well plates. This figure was determined experimentally by reading the same sample of reduced cytochrome c in both the spectrophotometer and microplate reader and by calculating the change in extinction coefficient from the measured reduction in light path. When it was desired to separate PMNs from BPAE during the incubation period, BPAE were grown in M199 without phenol red on Millicell-HA, 0.45-llm-pore culture plate filter inserts, 12 mm in diameter. These were inserted into wells containing I x 1(}> PMNs in 400 III of 100 11M cytochrome c in KRPG and 400 III of 100 11M cytochrome c was


added into the insert. For each experiment using BPAE grown on Millicell-HA inserts, BPAE were also grown to confluence on 24-well plates in the usual way as a control for both cell growth and inhibition of PMN Oz- production. To check for BPAE confluence on the Millicell-HA inserts, after the incubation with cytochrome c, the inserts were rinsed in KRPG, fixed with 3 % glutaraldehyde in PBS, stained with hematoxylin, and examined microscopically. O,- generation by xanthine oxidase and hypoxanthine. Xanthine oxidase (6 to 30 mV) and 50 III of catalase, 3.6 mg/ml in 200 mM phosphate buffer, pH 7.4, were added to 1 ml of 100 11M cytochrome c in KRPG in empty 24-well tissue culture plates and to 24-well plates containing confluent BPAE, washed free of M199 with KRPG. Four concentrations of hypoxanthine were added (32 to 128 11M concentration) in quadruplicate, with 1 well per concentration containing 20 III of SOD, 2 mg/ml, incubated at 37° C for 15 min, and the O,: produced was measured as described above. HzOz production by PMNs in the presence and absence of BPAE. Monolayers of BPAE grown in 96-well tissue culture plates. in M199 with phenol red as described above for 24-well plates were washed free of growth medium by 2 washes with KRPG. HzOz production was measured using a Molecular Devices Vrna, plate reader housed in a VWR 1530 constant-temperature incubator at 37° C, essentially as described by Pick and Keisari (24). To each well was added ISO III of a phenol red solution containing 0.56 mM phenol red and 100 Ilg/ml horseradish peroxidase in KRPG. PMNs (1 X 106 ) were added to half of the wells, and the reaction was started by the addition of 0.1 11M fMLP or 50 Ilg/ml PMA. At various time points between 0 and 3 h, 20 III of I N NaOH was added (final volume, 200 Ill) and the plates were read at 610 nm. Controls, with PMNs alone, were run in 96-well plates coated with FCS or without any coating and were treated the same as BPAE alone or BPAE plus PMN s. Each condition was run in triplicate, with I well containing 100 Ilg of catalase. Degranulation. Todetermine the influence of BPAEmonolayers on fMLP-stimulated degranulation of PMN, 24-well, 2-cm Z tissue culture plates were employed as described for the measurement of Oz- production. BPAE alone, BPAE plus 2.5 X 106 PMNs, and PMNs alone were incubated 30 min at 37° C in KRPG without a stimulus or with 1 X 10-7 M fMLP ± 5 Ilg/ml cytochalasin B (final volume, 1 ml). To determine the total content of enzyme ({3-g1ucuronidase or lactic dehydrogenase), 0.2 % Triton X 100 was added to separate wells. For each condition tested, 3 replicate wells were used. After the 30-min incubation, the tissue culture plates were placed on ice for 15 min, and the contents of each well transferred to microfuge tubes and centrifuged in a Beckman Microfuge for 5 min. The supernatant fluids were transferred to 4-ml disposable plastic tubes and either assayed immediately for enzyme activity or stored overnight at 4°C. {3Glucuronidase activity was determined by measuring fluorometrically the release of 4-methy lumbelliferone from 4methylumbelliferyl-{3-D-glucuronide as described by Glew and colleagues (25). A 50-Ill aliquot of each sample was incubated with 5 mM substrate in 100 mM acetate buffer, pH 4.8, for 30 min at 37° C (final volume, 0.1 ml). The reaction was terminated by the addition of 2.9 ml of termination



buffer (0.2 M NH 40H, 0.05 M glycine, pH 10.5). The fluorescence for each sample was determined using a TurnerSequoia spectrofluormeter (Model 450; Turner-Sequoia, Mountain View, CA), using an excitation wavelength of 360 nm (bandwidth, 15 nm) and an emission wavelength of 450 nm (bandwidth, 15 nm). The fluorescence was compared to that of standard 4-methylumbelliferone for quantitation. Lactate dehydrogenase activity was measured by determining the rate of NADH oxidation at 340 nm. A 50-pJ sample of supernatant fluid was added to the assay mixture (50 nM potassium phosphate buffer, pH 7.0, 2 mM pyruvate [lithium salt], and 1.3 mM NADH in a final volume of 1.0 ml) at 37° C, and the absorbance at 340 nm monitored for 5 min with a Gilford Model 252 recording spectrophotometer. A millimolar extinction coefficient for NADH of 6.22 was used to calculate activity, which was expressed as J..tmol of NADH oxidized/min. Assay of soluble inhibitor produced by BPAE. Conditioned media (KRPG, M199 without phenol red, or HBSS after incubation with BPAE for 30 min) was assayed directly or after dialysis against deionized water, lyophilization to dryness, and reconstitution with a small volume of 200 mM phosphate buffer, pH 7.4. Assay of inhibition of O2- production was carried out in polypropylene tubes as described by Kuhns and coworkers (22) for total O2- production. The soluble inhibitor was concentrated by ultrafiltration using a magnetically stirred Amicon ultrafiltration cell. The inhibitor passed through Diaflo membrane YM 10, which retains molecules of 10 kD and larger, and was retained by Diaflo YM 2, which retains molecules 1 kD and larger. All of the inhibitory activity that passed through YM 10 was retained by the YM 2 membrane. Statistics Significance of differences between means ± SE or SD in all tables and figures, except Figure 4, were determined by Wilcoxon rank sum test for unpaired samples and Wilcoxon signed rank test for paired samples. For Figure 4, repeated measure analysis of variance was used to test overall differences, followed by multiple comparisons techniques to test individual differences.

Results Reduction of Superoxide Production by Human Neutrophils in Co-culture with Endothelial Cells Co-incubating neutrophils with BPAE markedly reduced superoxide production by human neutrophils stimulated by a variety of receptor-mediated activators. Figure 1 shows that O 2- production by resting PMN as well as PMN stimulated with 10-9 to 10-6 M fMLP was inhibited by 67 to 80% in the presence of monolayers of BPAE. Figure 2 shows that a similar degree of inhibition occurred when the stimulant was serum-opsonized zymosan or serum-opsonized heat-killed S. aureus. O2- production by BPAE alone with fMLP, serum-opsonized zymosan, or serum-opsonized heat-killed S. aureus was ~ 1 nmol/well/30 min. Most of the studies we performed utilized BPAE isolated and cultured without enzymes. To verify that a qualitatively similar inhibitory effect could be demonstrated using human






28 Z


a. 21



ON 14

Endothelial cells inhibit receptor-mediated superoxide anion production by human polymorphonuclear leukocytes via a soluble inhibitor.

Confluent monolayers of bovine pulmonary artery endothelial cells (BPAE) or human umbilical vein endothelial cells (HUVE) inhibited by 80 to 90% the p...
1MB Sizes 0 Downloads 0 Views