Injury to Endothelial Cells by Phagocytosing Polymorphonuclear ...

2 downloads 0 Views 2MB Size Report
remained unaffected by scavengers of toxic oxygen species. During ... products of the lipoxygenase pathway of PMN play a modulatory role in this injury.
Vol. 55, No. 6

INFECTION AND IMMUNITY, June 1987, p. 1447-1454 0019-9567/87/061447-08$02.00/0 Copyright C) 1987, American Society for Microbiology

Injury to Endothelial Cells by Phagocytosing Polymorphonuclear Leukocytes and Modulatory Role of Lipoxygenase Products CHRISTINA M. J. E. VANDENBROUCKE-GRAULS,l2* HENRICUS M. W. M. THIJSSEN,2 KOK P. M. VAN KESSEL,2 B. SWEDER VAN ASBECK,3 AND JAN VERHOEF"2 Department of Clinical Microbiology and Laboratory of Infectious Diseases,' and Department of Internal Medicine,3 University Hospital Utrecht, and Laboratory of Microbiology, State University of Utrecht,2 Utrecht, The Netherlands Received 26 September 1986/Accepted 6 March 1987

Phagocytosis of microorganisms by polymorphonuclear leukocytes (PMN) is accompanied by inadvertent extracelHular release of microbicidal products; this could result in tissue damage. We investigated whether PMN damages endothelial cells when phagocytosis of Staphylococcus aureus occurs on the endothelial surface and how this damage might be modulated. Damage was assayed by the measurement of cell detachment or cell lysis of cultured endothelial cells that were radiolabeled with 51Cr. Uptake of bacteria was accompanied by nonlytic detachment of endothelial cells from the monolayer. This effect was inhibited by a-1-antitrypsin but remained unaffected by scavengers of toxic oxygen species. During phagocytosis, PMN adhered to the endothelial cells. Adherence could be prevented by inhibition of the lipoxygenase pathway of arachidonic acid metabolism of the PMN with nordihydroguaiaretic acid. This inhibition also resulted in a marked decrease of the detaching activity of the PMN. The addition of exogenous leukotriene B4 during phagocytosis greatly enhanced the damage to the endothelial monolayer. These results indicate that phagocytosis of staphylococci by PMN is accompanied by injury to endothelial cell monolayers due to released lysosomal proteases and that products of the lipoxygenase pathway of PMN play a modulatory role in this injury.

cells from cultured monolayers, an effect due to released granular enizymes (16). To our knowledge, no studies have shown a direct deleterious effect of the phagocytosis of bacteria by PMN on normal host tissue cells. In the present study, we investigated whether PMN phagocytosing S. aureus on the surface of endothelial cell monolayers caused injury to the cells. The choice of S. aureus was prompted by the well-known pyogenic properties of staphylococci and their ability to spread to distal sites by the hematogenous route with the formation of metastatic abscesses. Also, in contrast to several other microorganisms, S. aureus is known to be able to cause endocarditis on healthy cardiac valves. As endothelial damage did occur during phagocytosis of S. aureus, we studied which products from the PMN were responsible for this damaging effect and how this effect might be modulated. Products of the arachidonic acid metabolism, such as leukotrienes and prostaglandins, affect in several ways the interaction between leukocytes and endothelial cells (5, 12, 15, 30). Therefore, we investigated a possible modulatory role of arachidonic acid metabolites in the injury to the endothelial cells caused by PMN during phagocytosis. (Part of this research was presented at the 25th Interscience Conference on Antimicrobial Agents and Chemotherapy, Minneapolis, Minn., 29 September to 2 October 1985.)

The main function of polymorphonuclear leukocytes (PMN) is to ingest and kill microorganisms that invade host tissue. Killing of the ingested bacteria requires degranulation of lysosomes (18) and activation of the oxidative metabolism (1) of the PMN. Most often PMN function is studied in vitro by incubation of PMN with bacteria in suspension in medium containing serum. In vivo, however, at the infectious site, bacteria are phagocytosed in the presence of tissue cells. Thus, tissue cells may well be involved in the phagocytic process. In a recent study (35), we demonstrated that endothelial cells do not behave as passive bystanders during phagocytosis of Staphylococcus aureus by PMN but that they are actively involved in this process; that is, phagocytosis is enhanced when it occurs on the surface of endothelial cells. The fate of the tissue cells, however, is uncertain, since microbicidal products released by phagocytosing PMN can leak extracellularly through a mechanism called regurgitation during feeding (46); tissue injury as a result seems likely. The phenomenon of regurgitation during phagocytosis of microorganisms is well known. Yet, in most studies on the cytotoxic effects of PMN, the phagocytes are stimulated with soluble products, such as phorbol myristate acetate (PMA), N-formylmethionyl-leucyl-phenylalanine, or complement component C5a (27, 29, 42, 44). Activation by these compounds leads to a direct extracellular release of toxic PMN products and not to inadvertent leakage, as might occur during phagocytosis of microorganisms. Clark and Klebanoff (9) showed that tumor cells are lysed in the presence of PMN that are stimulated with opsonized zymosan and that for this effect the presence of halides is required. It is thought to be dependent on the myeloperoxidase system of the PMN. Phagocytosis of opsonized zymosan by PMN also induces detachment of endothelial *

MATERIALS AND METHODS Reagents. PMA, superoxide dismutase, catalase, a-iantitrypsin, phenylmethylsulfonyl fluoride, L-1-tosylamideserum bovine ketone, 2-phenylethyl-chloromethyl albumin, 1-chloro-2,4-dinitrobenzene (CDNB), and nordihydroguaiaretic acid (NDGA) were obtained from Sigma Chemical Co., St. Louis, Mo. Phenylmethylsulfonyl fluoride and L-1-tosylamide-2-phenylethyl-chloromethyl ketone were dissolved in dimethyl sulfoxide (J. T. Baker Chemicals, Deventer, The Netherlands); CDNB and NDGA were dissolved in ethanol, and all reagents were further diluted to the

Corresponding author. 1447

1448

VANDENBROUCKE-GRAULS ET AL.

desired concentration in the appropriate medium. Lysostaphin was from Schwarz/Mann, Orangeburg, N.Y.; Triton X-100 (TX-100) was from BDH, (Poole, United Kingdom), thiourea was from Cooperatieve Apothekersvereniging OPG, Utrecht, The Netherlands; methionine was from E. Merck AG, Darmstadt, Federal Republic of Germany; and indomethacin was from Merck Sharp & Dohme, Haarlem, The Netherlands. U-60,257B [6,9 deepoxy 6,9-(phenylimino)-6,8 PGI1 potassium] and leukotriene B4 (LTB4) were kindly provided by The Upjohn Co., Kalamazoo, Mich. Isolation of human PMN. PMN were isolated from venous blood from healthy adult donors by dextran sedimentation of erythrocytes, differential density centrifugation on FicollPaque (Pharmacia Fine Chemicals, Uppsala, Sweden), and NH4Cl lysis of contaminating erythrocytes (6, 38). PMN were suspended in Hanks balanced salt solution with 0.1% gelatin (Gel-HBSS) at a concentration of 5 x 106 cells per ml. S. aureus strain. S. aureus Ev (38) was grown in MuellerHinton broth (Difco Laboratories, Detroit, Mich.) and supplemented when needed with 0.02 mCi of [3H]thymidine (specific activity, 5 Ci/mmol; Amersham Corp., Radiochemical Centre, Amersham, United Kingdom) for radiolabeling (40). The bacteria were washed with sterile phosphatebuffered saline (PBS; pH 7.4), killed by Formalin treatment, washed, and suspended in Gel-HBSS to a final concentration of 109 CFU/ml (39). When necessary, bacteria were opsonized in 5% human serum (pooled from 10 healthy donors) in Gel-HBSS (38). Culture and radiolabeling of human venous endothelial cells. Endothelial cells were isolated from human umbilical cord veins by a method modified from Jaffe et al. (21) as previously described (35). Cells were cultured in RPMI 1640 (GIBCO Biocult Ltd., Paisley, United Kingdom) supplemented with 20% human serum (pooled from 15 healthy donors), 2 mM glutamine, 100 U of penicillin per ml, and 100 ,ug of streptomycin per ml. For experiments, cells from individual cords in first or second passage were harvested with 0.05% trypsin-0.02% EDTA (GIBCO) and cultured in 48-well microtiter plates (no. 3548; Costar, Cambridge, Mass.) or 96-well culture plates (Nunc, Roskilde, Denmark). The cells were identified as endothelial cells by cell morphology and by immunofluorescent staining with rabbit antibodies against factor VIII (20, 21). Endothelial cells were radiolabeled with radioactive chromium (51Cr) as sodium chromate (specific activity, 350 to 600 mCi/mg of chromium; Amersham). To this purpose, 96-well microtiter plates containing confluent monolayers were washed once with culture medium and then incubated overnight with culture medium containing 1 ,uCi of 51Cr per well. The monolayers were washed free of radioactive medium by five washes with Gel-HBSS and used for measurement of cell detachment or cell lysis. Phagocytosis of S. aureus by PMN on endothelial monolayers. Phagocytosis was performed as previously described (35). Briefly, radiolabeled bacteria (0.25 ml; 5 x 108 CFU/ml) were added to the wells of 48-well microtiter plates containing a confluent endothelial cell monolayer which was previously washed twice with Gel-HBSS. The culture plates were centrifuged for 5 min at 1,600 x g, and PMN (0.25 ml; 5 x 106/ml) were then added. This resulted in a ratio of PMN to bacteria of 1:100. The effector-to-target ratio for PMN and endothelial cells was estimated as approximately 20:1. The plates were incubated in a stationary position at 37°C in a 95% air and 5% CO2 atmosphere. After 30, 60, or 90 min, ice-cold PBS was added to the wells. After careful resuspension, the content of each well was transferred to a

INFECT. IMMUN.

vial, and the wells were rinsed vigorously four times with PBS. The rinses were added to the respective vials and the non-leukocyte-associated bacteria were removed by two cycles of centrifugation at 160 x g. The pellet was resuspended in 2.5 ml of PBS with 1 ,ug of lysostaphin per ml and incubated for 30 min at 37°C to lyse extracellular, adherent S. aureus (40). Two more centrifugation cycles were then performed. To determine the total radioactivity (a measure of the total number of bacteria) added to each well, 0.25 ml of bacterial suspension was added directly to a separate biovial and centrifuged at 1,600 x g for 15 min. Leukocyte pellets and bacterial pellets were suspended in scintillation liquid and counted. Phagocytosis was expressed as a percentage of the total amount of added radioactivity. Measurement of detachment and lysis of endothelial cells. Endothelial cell detachment and lysis were determined by a method modified from Harlan et al. (16). 51Cr-labeled endothelial monolayers grown in 96-well microtiter plates were used. After the wells were washed with Gel-HBSS, 100 ,ul of bacterial suspension (2.5 x 108 CFU/ml) was added to each well and the plates were centrifuged for 5 min at 1,600 x g. PMN (100 RI; 2.5 x 106/ml) were then added (PMN-tobacteria ratio, 1:100; PMN-to-endothelial cell ratio, approximately 15:1), and the plates were incubated as described above for phagocytosis. For the determination of detachment at the different times, the reaction was stopped by the addition of Gel-HBSS. The wells were gently rinsed twice to remove detached endothelial cells. Adherent cells were lysed with 200 RI of 0. 1% TX-100, and 100 ,ul was transferred to vials for counting. Total radioactivity was measured by lysis of the cells in separate reference wells that had not been rinsed. Spontaneous detachment was determined in wells incubated with Gel-HBSS only. Detachment was calculated by the following formula: (counts per minute in experimental wells/counts per minute in wells incubated with Gel-HBSS only) x 100. Spontaneous detachment varied with the endothelial cell isolate; only results from experiments with a spontaneous detachment of less than 20% were considered. For the determination of cell lysis, the release of 51Cr was measured. For this purpose, after incubation of the monolayers with S. aureus and PMN as described above, cells were pelleted in the plates by centrifugation at 160 x g for 5 min. Supernatant fluid (100 RI) was then carefully removed for counting. Total radioactivity was determined in separate wells incubated with TX-100, and spontaneous 51Cr release was determined in wells incubated with Gel-HBSS only. Specific 51Cr release was calculated by the following formula: [(51Cr release in experimental wells - 51Cr spontaneous release)/(51Cr total release - 51Cr spontaneous release)] x 100.

In some experiments, PMN were stimulated with PMA which was dissolved in dimethyl sulfoxide and further diluted in Gel-HBSS. The final concentration of dimethyl sulfoxide in reaction mixtures never exceeded 0.1%. In the experiments in which LTB4 was added to the reaction mixture, Gel-HBSS was replaced by Tyrode buffer with 0.2% BSA (Sigma). Adherence of PMN to endothelial monolayers. Adherence of PMN to the monolayers was measured as described previously by van Kessel et al. (37). Briefly, PMN (107/ml) were labeled with Na51Cr (specific activity, 350 to 600 mCi/mg of chromium) for 60 min at 37°C, washed five times with Gel-HBSS, and resuspended to a concentration of 5 x 106/ml of Gel-HBSS. Bacteria and radiolabeled PMN were added to endothelial monolayers in 96-well microtiter plates as for the detachment experiments, but the plates were

VOL. 55, 1987

ENDOTHELIAL DAMAGE BY PHAGOCYTOSING NEUTROPHILS

subjected to an additional centrifugation step after the addition of PMN (1 min at 160 x g) for rapid sedimentation of the leukocytes. Previous experiments (37), which supported the results of others (8), showed that adherence was not influenced by centrifugation; when PMN are allowed to settle by gravity, comparable amounts of PMN adhere to the monolayer. The plates were incubated for 10 min at 37°C and then gently rinsed twice with Gel-HBSS. A 200-pul sample of 0.1% TX-100 was added to each well to disrupt adherent PMN, and 100 ,ll was aspirated for counting. To determine the total radioactivity (a measure of the total number of PMN) added to each well, TX-100 was added to reference wells that were not rinsed. The adherence of PMN was expressed as a percentage of total radioactivity. Measurement of GSH and depletion of GSH stores in endothelial cells. For the measurement of cellular glutathion (GSH; 17), endothelial cells were harvested from tissue culture flasks with 0.05% trypsin-0.02% EDTA, washed three times with PBS, and counted. Cells (106) were pelleted at 160 x g for 10 min and lysed with 140 ,ul of 0.2% TX-100 and 2.5% sulfosalicylic acid in 0.2% EDTA-PBS buffer. After centrifugation at 11,000 x g for 5 min, 50 ,ul of the acid-soluble extract was mixed with 100 ,ul of 0.3 M Na2HPO4. GSH was measured in these samples by spectrophotometric determination of the reduction of 5,5'-dithiobis(2-nitrobenzoic) acid at 412 nm. A standard curve with known concentrations of GSH (1 to 100 nmol) was measured at the same time. Results are expressed as nanomoles of GSH per 106 endothelial cells. To deplete GSH stores in endothelial cells, cell cultures were incubated with CDNB, which was dissolved in ethanol (final concentration of ethanol in reaction mixtures, 0.05%) and further diluted in culture medium to 10 ,uM, for 60 min. The cell cultures were then washed three times with GelHBSS. Measurement of lysozyme. Lysozyme was measured as a marker for lysosomal enzyme release. After incubation of S. aureus, PMN, and endothelial cells for 30 min as described above for phagocytosis experiments, the microtiter plates were centrifuged for 5 min at 160 x g. Supernatant fluids from duplicate wells were collected for lysozyme determination. The pellets were suspended in ice-cold PBS, transferred to appropriate vials, and sonicated three times for 20 s (Sonifier B12; Branson Sonic Power Co., Danbury, Conn.). The amounts of lysozyme in the supernatant fluids and in the cell lysates were determined by the rate of lysis of Micrococcus lysodeikticus (dried cells; Sigma) measured by the decrease in absorption at 450 nm (10). The total amount of enzyme in the PMN was determined in PMN incubated with endothelial cells in the absence of bacteria. Enzyme release was calculated by subtracting enzyme activity in the cell lysates (residual activity after phagocytosis) from total activity and was then expressed as a percentage of the total enzyme activity. Statistical analysis. All assays, except the GSH determination, were run in triplicate. Results are expressed as the mean of three or more independent observations plus or minus the standard error of the mean (SEM). For statistical analysis, the Student t test was performed. P values exceeding 0.05 were considered not significant. RESULTS Damage to endothelial cells by PMN during phagocytosis of S. aureus. Phagocytosis of both preopsonized and unopsonized S. aureus occurs on the surface of endothelial monolay-

1449

100.

E u a)(U 80.. 4)

I U

60.

401

z,4i

LL

201l

30

1,

~

1_ 60 90

time (min)

FIG. 1. Damage to endothelial cell monolayers by PMN during phagocytosis of S. aureus or on activation by PMA. O, Control PMN (2.5 x 106/ml); 0, PMN stimulated with unopsonized S. aureus (2.5 x 1081/ml); *, PMN stimulated with PMA (10 ng/ml). Each value is the mean + SEM of at least four separate experiments performed in triplicate.

ers (35). In the experiments described in the present study, a PMN-to-bacteria ratio of 1:100 was used. Under these conditions, uptake of S. aureus reached 59 + 2% after 60 min of incubation. The uptake of bacteria by the PMN injured the endothelial cells. After 1 h of incubation, half of the cells had lifted off from the monolayer (Fig. 1). Resting PMN or S. aureus by itself without PMN did not cause detachment. The endothelial monolayers were also exposed to PMN stimulated with the soluble agent PMA. Under these conditions, detachment was only slightly higher than when phagocytosing PMN were used (Fig. 1). To evaluate whether detachment is dependent on the phagocytic process, the effect of increasing amounts of phagocytosed bacteria or of phagocytosing cells was determined by the use of higher numbers of PMN or by the addition of more bacteria per PMN. These experiments showed that the detachment increased with the number of PMN and the number of bacteria

(Fig. 2).

We also measured the ability of phagocytosing PMN to lyse endothelial cells. No lysis, as determined by measurement of 51Cr release, was observed during the incubation with PMN and S. aureus or with PMA-stimulated PMN. Lysis of endothelial cells is usually described as a relatively late event, which occurs after 5 to 6 h of incubation with activated PMN (17, 44). Therefore, lysis was also determined after an 18-h incubation period. These experiments showed little lysis by phagocytosing PMN (always less than 10%), whereas PMA-activated PMN did cause extensive 51Cr release (Fig. 3). Critical role of lysosomal enzymes in endothelial cell damage. During phagocytosis, PMN produce several toxic oxygen products, such as superoxide anion (O2-), hydrogen peroxide (H202), and hydroxyl radical (-OH). In combination with halides and by the action of myeloperoxidase, H202 forms highly toxic hypohalides. The role of these

1450

VANDENBROUCKE-GRAULS ET AL.

INFECT. IMMUN.

I

c

E

£

-I-

0

la

r

1:20

number of PMN x 2.5

1:100 1:200 PMN: bacteria

FIG. 2. Effect of number of PMN and number of bacteria on detachment of endothelial cells. (A) Percent detachment of endothelial cells after incubation with S. aureus and various PMN concentrations. FO1, Control PMN; *, PMN stimulated with unopsonized S. aureus (2.5 x 108/ml). (B) Percent detachment after incubation with PMN (2.5 x 106/ml) and various S. aureus concentrations. Each value is the mean + SEM of three separate experiments performed in triplicate.

different oxygen-derived products was evaluated by the addition of various enzymes and scavengers to the assay system. Superoxide dismutase (100 ,ug/ml) was used to inactivate superoxide anion, and catalase (100 [Lg/ml) was used to degrade H202. Detachment of endothelial cells by phagocytosing PMN was 72 + 4%; superoxide dismutase and catalase had no significant effect (66 ± 5 and 68 ± 6%, respectively). Thiourea has been shown to be an effective hydroxyl-radical scavenger (28) and was used for this purpose. Detachment of endothelial cells in the presence of 10 mmol of this agent amounted to 80 ± 6%. Methionine is known to react with hypohalides and acts as a scavenger for these products (41). Methionine had no protective effect on the damage caused by the phagocytosing PMN to the

80_

80_

-W

0

I

0

in,-64I

c

0

E 0

601 *

0

h.

0 ._

endothelial monolayer (detachment was 80 + 1% in the presence of 10 mM methionine). Control experiments showed that none of the agents by itself caused detachment of the endothelial cells. Results are given for unopsonized S. aureus; with opsonized S. aureus the same findings were recorded. The lack of effect of these agents could be owing to the protective role of the GSH redox cycle in the endothelial cells (17). If the oxygen products generated by the phagocytosing PMN are rapidly destroyed by oxidation of GSH then no effect can be expected from the exogenous added enzymes or scavengers. Therefore, endothelial stores of GSH were depleted with CDNB (17). CDNB treatment of endothelial monolayers effectively diminished their GSH content; untreated cells contained 22.6 nmol of GSH per 106 cells, and treated cells contained 8.2 nmol of GSH per 106 cells, a reduction of 63%. No difference was found, however, either in detachment (67 + 9 versus 67 + 7%) or cell lysis (1.2 + 0.2 versus 1.7 + 0.3%) of normal or GSHdepleted endothelial cells by phagocytosing PMN. To exclude convincingly a role for toxic oxygen species in the disruption of the integrity of endothelial cell monolayers by PMN during phagocytosis of bacteria, detachment was also determined in an experiment in which PMN from a patient with chronic granulomatous disease were used. These PMN were as effective as normal PMN in detaching endothelial cells from the monolayers (results not shown). These results suggest a critical role for lysosomal proteases in the injurious effect of phagocytosing PMN, with primary granule constituents and especially elastase (22) as obvious candidates. Indeed, when a-1-antitrypsin (100 ,ug/ml), a well-known inhibitor of proteolytic activity of granular enzymes, was added to the incubation wells, detachment of endothelial cells was reduced from 74 to 32% (Fig. 4). Total inhibition was never achieved, even in the presence of higher concentrations of cx-1-antitrypsin. The effect of (x-1-antitrypsin was specific, since bovine serum albumin at the same concentration (100 ,ug/ml) did not affect endothelial cell detachment. In ancillary studies, we exam-

40.

40.

(L

0

e

20.

201

90

TrL18

min hrs S. aureus

90 18 min hrs PMA

FIG. 3. Lysis of endothelial cells after incubation with PMN and S. aureus or PMA. PMN (2.5 x 106/ml) were stimulated with S. aureus (2.5 x 108/ml) (O) or with PMA (30 ng/ml) (0). Each value is the mean + SEM of three separate experiments performed in

triplicate.

100 75 50 control

a-I-at

pg/mI

FIG. 4. Protective effect of a-1-antitrypsin on detachment of endothelial cells by PMN during phagocytosis of S. aureus. PMN (2.5 x 106/ml) were stimulated with S. aureus (2.5 x 108/ml) in the absence (LI) and presence (0) of a-1-antitrypsin (a-l-at). Each value is the mean ± SEM of at least four separate experiments performed in triplicate. *, P < 0.05.

ENDOTHELIAL DAMAGE BY PHAGOCYTOSING NEUTROPHILS

VOL. 55, 1987

1451

60_ A

0 0

B

0 la

T 40.

T I

20.1.

I

-

PMN

ITT T

T 1 10 NDGA

1 10 100 500pM indomethacin

PMN + S.aureus

E1I -

0.0010.01 0.1 LTB4

PMN

1

-

0.01 0.1 1 PM LTB4

PMN + S.aureus

FIG. 5. Adherence of PMN to endothelial cells during phagocytosis and effects of lipoxygenase products. (A) Effect of inhibition of arachidonic acid metabolic pathways. D1, Control PMN (2.5 x 106/ml); 0, PMN stimulated with S. aureus (2.5 x 108/ml); E, NDGA-treated PMN stimulated with S. aureus; 0, indomethacin-treated PMN stimulated with S. aureus. Each value is the mean SEM of at least four separate experiments performed in triplicate. (B) Effect of exogenous LTB4. O, Control PMN (2.5 x 106/ml); 0, PMN stimulated with S. aureus (2.5 x 1081/ml); *, PMN with LTB4 (1 to 0.01 ,ug/ml); EI1, PMN with LTB4 and stimulated with S. aureus. Each value is the mean SEM of three separate experiments performed in triplicate. *, P < 0.05 as compared with base-line adherence; *, P < 0.05 as compared with adherence of control PMN stimulated by S. aureus.

ined the role of methionine, which is located at the reactive center of the a-1-antitrypsin molecule. Oxidation of the methionyl residue leads to loss of the inhibitory activity of a-1-antitrypsin toward several serine proteases (23). Oxidative inactivation of a-1-antitrypsin could occur during incubation with activated PMN. In an attempt to protect a-iantitrypsin, an excess of methionine (1 to 10 mM) was added to the reaction mixture. Despite this increase in oxidizable substrate, no improvement of the effect of a-1-antitrypsin was observed. Further evidence for the role of lysosomal proteases was provided by the inhibitory effect of serine protease inhibitors phenylethylsulfonyl fluoride (1 mM) and L-1-tosylamide-2phenylethyl-chloromethyl ketone (1 ,uM), which inhibited endothelial cell detachment by 45 + 6 and 44 7%, respectively. Modulation of endothelial cell detachment by products of the arachidonic acid metabolism. Adherence of PMN to endothelial cells during phagocytosis of S. aureus could play a role in the damage caused to the endothelial monolayer. Leukotrienes and prostaglandins are known to affect the adhesion of PMN to endothelial cells (15). As both PMN and endothelial cells synthesize several products of the arachidonic acid metabolism (13, 19, 24, 30), we studied a possible modulatory role for these products. NDGA and U-60,247B were used to inhibit the lipoxygenase pathway of arachidonic acid metabolism (2), and indomethacin was used to inhibit the cyclooxygenase pathway (36). Unstimulated PMN hardly adhered to the endothelial surface (about 10% of the total added PMN), whereas the ±

presence of S. aureus on the monolayer caused a threefold increase in the number of adherent PMN (Fig. 5A). Preincubation of endothelial cells with NDGA or indomethacin had no effect on the adherence of the stimulated PMN to the endothelial cells or on the detachment of endothelial cells during phagocytosis of S. aureus by the PMN. The adhesion induced by the S. aureus, however, could be prevented by preincubation of the PMN with NDGA (Fig. 5A). Preincubation of PMN with NDGA also markedly reduced the injurious effect of the phagocytosing PMN on the monolayer (Fig. 6A). Adherence and the detaching activity of PMN were inhibited at higher concentrations of indomethacin (100 to 500 ,uM) (Fig. 5A and 6A). At these concentrations, however, indomethacin affects the arachidonic acid metabolism via non-cyclooxygenase-dependent mechanisms (31). Since NDGA may have effects on PMN function other than inhibition of lipoxygenase activity (27), a second inhibitor, U-60,247B, was used. Preincubation of PMN with this agent (10 ,uM) confirmed the results obtained with NDGA; detachment of endothelial cells was inhibited by 73 + 10%. To further explore the role of lipoxygenase products, we added purified LTB4, a major product of the lipoxygenase pathway of arachidonic acid metabolism in PMN, to the test system. LTB4 by itself caused significant adhesion of PMN, to an even greater extent than the adhesion induced by S. aureus. The effect of LTB4 and S. aureus, however, was not additive; incubation of PMN with bacteria in combination with LTB4 did not cause a further increase in the leukocyte adhesion to the endothelial cells (Fig. SB). LTB4 by itself did not stimulate the PMN to detaching activity, since in the

1452

VANDENBROUCKE-GRAULS ET AL.

INFECT. IMMUN.

80B

B

A

-

T-

E 60.

T

T

0

T

T

40.

1.

20_

Ni PMN

-

1 10 NDGA

10 100 500 pM indomethacin

1

PMN + S. aureus

-

-

0.1 1 LTB4

001 0.1 1 PM LTB4 PMN+ S.aureus

PMN

FIG. 6. Modulation of endothelial cell detachment by lipoxygenase products. (A) Effect of inhibition of arachidonic acid metabolic pathways. 2l, Control PMN (2.5 x 106/ml); 0, PMN stimulated with S. aureus (2.5 x 108/ml); P, NDGA-treated PMN stimulated with S. aureus; E, indomethacin-treated PMN stimulated with S. aureus. Each value is the mean SEM of four separate experiments performed in triplicate. (B) Effect of exogenous LTB4. 2l, Control PMN (2.5 x 106/ml); 0, PMN stimulated with S. aureus (2.5 x 108/ml); *, PMN with LTB4 (1 to 0.01 ,ug/ml); lID, PMN with LTB4 stimulated with S. aureus. Each value is the mean SEM of three separate experiments performed in triplicate. *, P < 0.05. ±

±

absence of bacteria, no significant detachment above baseline values was recorded. LTB4 added to phagocytosing PMN, however, greatly enhanced the damage to the monolayers (Fig. 6B). This large increase of detachment probably explains the lack of an additive effect of S. aureus and LTB4 on PMN adhesion; a correct measurement of adherence is hampered when the endothelial cells start to detach from the monolayer. To exclude the possibility that the effect of LTB4 on endothelial cell detachment is nonspecific and is the result of the addition of a fatty acid which may serve as a precursor of toxic lipid peroxides, control experiments were performed with arachidonic acid at the same concentrations as LTB4 and with an inactive analog of LTB4, 12epi,6t-LTB4 (kindly provided by P. L. Bruynzeel, Department of Pulmonary Disease, University Hospital Utrecht, Utrecht, The Netherlands). Neither compound had an effect on the detaching activity of the PMN. The effects of NDGA and LTB4 could be mediated via an effect on the release of lysosomal enzymes during phagocytosis (3). Therefore, release of lysozyme was determined. NDGA and LTB4 had no significant effect on the release of this enzyme compared with the control PMN (38 + 10 and 44 + 6%, respectively, versus 49 + 6%). Also, the amounts of enzyme activity recovered in the extracellular medium were not different for treated or control PMN. The effects of NDGA and LTB4 were not due to diminished or enhanced phagocytosis; uptake of S. aureus was not altered by these agents. DISCUSSION In response to phagocytosis, human PMN release lysosomal enzymes, generate oxygen-derived products, and

produce inflammatory mediators (45). All of these products have been implicated in tissue injury. Lysosomal enzymes have been implicated for their effect on extracellular matrices and for their nonlytic damage to cultured cellular monolayers (7, 16, 33, 43); oxygen-derived products have been cited for their cytotoxic effects (9, 26, 27, 29, 39, 42, 44); and inflammatory mediators, such as leukotrienes and prostaglandins, have been implicated for their effects on vascular permeability and for their vasomotor and chemotactic effects (for a review, see reference 32). Evidence for a role for these products from PMN in tissue damage, however, has been obtained in studies in which PMN were stimulated with soluble agents or with opsonized zymosan, a large phagocytosable particle, or in studies in which more or less purified products were used. We studied the question of whether phagocytosis of bacteria per se was accompanied by damage to surrounding cells. PMN were allowed to phagocytose S. aureus on the surface of endothelial cells in an experimental setting which permitted the measurement of nonlytic detachment of endothelial cells from the monolayer and of lysis of the endothelial cells. In previous studies, we showed that PMN are able to phagocytose opsonized as well as unopsonized S. aureus on an endothelial surface and that in both cases, the microbicidal mechanisms of the PMN are triggered to the same extent (34, 35). Uptake of S. aureus by PMN on the endothelial surface resulted in nonlytic damage to the monolayer; cells were detached from the culture surface but not lysed. This effect appeared to be dependent on the amount of phagocytosis that occurred on the endothelial surface, since it was dependent on the number of PMN as well as on the ratio of PMN

VOL. 55, 1987

ENDOTHELIAL DAMAGE BY PHAGOCYTOSING NEUTROPHILS

to bacteria. It was also evident that cell detachment was solely mediated by lysosomal proteases, whereas there was no role for toxic oxygen species. The question arises as to whether detachment of endothelial cells from cultured monolayers may be called endothelial cell injury. Detachment of endothelial cells from cultured monolayers certainly is a sign of endothelial damage, since healthy endothelial cells readily attach to the culture surface and do not detach unless a noxious agent is added. Moreover, endothelial cell detachment was always preceded by endothelial cell contraction, a sign of endothelial cell reaction (14). It is difficult to judge whether the release of lysosomal enzymes from PMN activated during phagocytosis of bacteria would fully detach endothelial cells in vivo, but it is conceivable that even slight alterations in the endothelial lining of the vessel wall would change its permeability and seriously affect normal vessel wall physiology. Damage to endothelial cells by activated PMN has been the subject of extensive investigations. Several authors (25, 29, 44) have shown that PMN can cause endothelial cell lysis on stimulation with soluble stimuli. In these studies, hydrogen peroxide appears to be the main cytotoxic agent. Still, the role and the mechanisms of activated PMN in endothelial cell lysis are controversial. Harlan et al. (16) found no cell lysis when PMN were stimulated with opsonized zymosan. This could mean that endothelial lysis can only be achieved by PMN activated by soluble agents. Harlan and co-workers (17), however, showed that GSH effectively protects endothelial cells from oxidative damage; they were able to detect endothelial cell lysis by PMA-activated PMN only after depletion of GSH stores in the endothelial cells. In our experimental model, when bacteria were used as a stimulus, even the inhibition of endothelial antioxidant defense (by depletion of GSH stores with CDNB) did not result in an increase in either endothelial cell detachment or endothelial cell lysis. Also, after prolonged incubation of endothelial cells with S. aureus and PMN, endothelial cell lysis was minimal, in contrast to the extensive lysis caused by PMAactivated PMN. Therefore, we conclude that the leakage of microbicidal products from PMN in the course of phagocytosis mainly results in cell detachment, due to release of lysosomal proteases. Interestingly, although significant protection of the endothelial monolayer was achieved by ca-1antitrypsin, a total inhibition of the detaching activity of the PMN was never reached, even with high concentrations of a-1-antitrypsin. It is possible that a-1-antitrypsin cannot reach the site of lysosomal enzyme activity because of close contact between PMN and endothelial cells (7). Close approximation between PMN and endothelial cells has been shown to be an important prerequisite for lytic damage to the endothelium by activated PMN (4, 29). Therefore, the question arises as to whether such proximity is also important in the nonlytic damage caused by the phagocytosing PMN. Adherence of PMN to endothelial cells is largely modulated by products of the arachidonic acid metabolism. Several leukotrienes, but especially LTB4, stimulate adherence of PMN to the endothelium (12), whereas prostaglandins inhibit this adherence (5). Many anti-inflammatory agents interact with the metabolism of arachidonic acid. This prompted us to investigate the possible role of leukotrienes and prostaglandins in the damage to the endothelial monolayer by the phagocytosing PMN. The presence of bacteria on the endothelial surface, whether opsonized or not, induced a large increase in the adherence of PMN to the monolayer. This is in contrast to the finding of Harlan (mentioned in reference 11), who saw no stimulation

1453

of adherence of PMN by opsonized zymosan. Both the adherence of the PMN stimulated by the S. aureus and the detachment of endothelial cells during phagocytosis were markedly decreased when the lipoxygenase pathway of arachidonic acid metabolism of the PMN was inhibited with NDGA (2). Preincubation of endothelial cells with NDGA and of PMN and endothelial cells with low concentrations of indomethacin (1 to 10 p,M) remained without effect. Therefore, a lipoxygenase product from the PMN appears to play a modulatory role in the adherence to endothelial cells and in the detaching activity of the PMN. A major product of the lipoxygenase pathway in activated human PMN is LTB4. This is the most effective lipoxygenase product for stimulation of adherence of PMN to endothelial cells (13). The addition of exogenous LTB4 to PMN indeed resulted in a large increase of PMN adherence to endothelial cells; detachment of endothelial cells by phagocytosing PMN also showed an almost twofold increase. NDGA and LTB4 had no effect on the uptake of S. aureus by the PMN or on enzyme release by the activated PMN. It is tempting to postulate that the effect of inhibition of lipoxygenase activity or of exogenous LTB4 on detachment was mediated through the effect on adherence. It should be noted that adherence by itself (induced by LTB4 in the absence of S. aureus) was not sufficient to stimulate the detaching activity of the PMN. Adherence during phagocytosis, however, appears to play a major role in the injury caused by the PMN to the endothelium. Further studies are required to determine the relative importance of adherence and enzyme release in endothelial

injury. In conclusion, phagocytosis of S. aureus by PMN appears to cause nonlytic detachment of endothelial cells, thereby providing a model of tissue injury at infectious sites and possibly a hypothesis of the means by which hematogenous spread of staphylococcal infections may occur. The role of leukotrienes in the modulation of this damage deserves further investigation, especially because the reduction of damage that followed the inhibition of leukotriene formation was not the result of a diminished uptake of S. aureus. LITERATURE CITED 1978. Oxygen-dependent microbial killing by Engl. J. Med. 298:659-668. and P. V. Reed. 1981. Evidence for inhibition of synthesis by 5,8,11,14-eicosatetraynoic acid in guinea pig polymorphonuclear leukocytes. J. Biol. Chem. 256: 4156-4159. Bokoch, G. M., and P. W. Reed. 1981. Effect of various lipoxygenase metabolites of arachidonic acid on degranulation of polymorphonuclear leukocytes. J. Biol. Chem. 256:53175320. Boogaerts, M. A., 0. Yamada, and H. S. Jacob. 1982. Enhancement of granulocyte-endothelial cell adherence and granulocyte induced cytotoxicity by platelet release products. Proc. Natl. Acad. Sci. USA 79:7019-7023. Boxer, L. A., J. M. Allen, M. Schmidt, M. Yoder, and R. L. Baehner. 1980. Inhibition of polymorphonuclear leukocyte adherence by prostacyclin. J. Lab. Clin. Med. 95:672-678. Boyum, A. 1968. Isolation of mononuclear cells and granulocytes from human blood. Scand. J. Clin. Lab. Invest. 97: S77-S89. Campbell, E. J., R. M. Senior, J. A. McDonald, D. Cox, J. M. Greco, and J. A. Landis. 1982. Proteolysis by neutrophils: relative importance of cell-substrate contact and oxidative inactivation of proteinase inhibitors in vitro. J. Clin. Invest. 70: 845-852. Charo, I. F., C. Yuen, and I. M. Goldstein. 1985. Adherence of human polymorphonuclear leukocytes to endothelial monolayers: effects of temperature, divalent cations, and chemotactic

1. Babior, B. M. phagocytes. N. 2. Bokoch, G. M., leukotriene A4

3.

4.

5. 6. 7.

8.

1454

VANDENBROUCKE-GRAULS ET AL.

factors on the strength of adherence measured with a new centrifugation assay. Blood 65:473-479. 9. Clark, R. A., and S. J. Klebanoff. 1975. Neutrophil-mediated tumor cell cytotoxicity: role of the peroxidase system. J. Exp. Med. 141:1442-1447. 10. Decker, L. A. 1977. Lysozyme, p. 185-186. In L. A. Decker (ed.), Worthington enzyme manual. Worthington Biochemical Corp. Freehold, N.J. 11. Diener, A. M., P. G. Beatty, H. D. Ochs, and J. M. Harlan. 1985. The role of neutrophil membrane glycoprotein 150 (GP-150) in neutrophil-mediated endothelial cell injury in vitro. J. Immunol. 135:537-543. 12. Gimbrone, M. A., Jr., A. F. Brock, and A. I. Schafer. 1984. Leukotriene B4 stimutlates polymorphonuclear leukocyte adhesion to cultured vascular endothelial cells. J. Clin. Invest. 74: 1552-1555. 13. Goldstein, I. M., C. L. Malmsten, H. Kindahl, H. B. Kaplan, 0. Radmark, B. Samuelsson, and G. Weissmann. 1978. Thromboxane generation by human peripheral blood polymorphonuclear leukocytes. J. Exp. Med. 148:787-792. 14. Gordon, J. L., and J. D. Pearson. 1982. Responses of endothelial cells to injury, p. 433-469. In H. L. Nossel and H. J. Vogel (ed.), Pathobiology of the endothelial cell. Academic Press, Inc., New

York. 15. Harlan, J. M. 1985. Leukocyte-endothelial interactions. Blood 65:513-525. 16. Harlan, J. M., P. D. Kilien, L. A. Harker, G. E. Striker, and D. G. Wright. 1981. Neutrophil-mediated endothelial injury in vitro: mechanisms of cell detachment. J. Clin. Invest. 68:

1394-1403. 17. Harlan, J. M., J. D. Levine, K. S. Callahan, B. R. Schwartz, and L. A. Harker. 1984. Glutathion redox cycle protects cultured endothelial cells against lysis by extracellularly generated hydrogen peroxide. J. Clin. Invest. 73:706-713. 18. Hirsch, J. G., and Z. A. Cohn. 1960. Degranulation of polymorphonuclear phagocytes following phagocytosis of microorganisms. J. Exp. Med. 112:1005-1014. 19. Ingerman-Wojenski, C., M. J. Silver, J. B. Smith, and E. Macarak. 1981. Bovine endothelial cells in culture produce thromboxane as well as prostacyclin. J. Clin. Invest. 67:1292-

1296. 20. Jaffe, E. A., W. Hoyer, and R. L. Nachman. 1973. Synthesis of antihemophilic factor antigen by cultured human endothelial cells. J. Clin. Invest. 52:2757-2764. 21. Jaffe, E. A., R. L. Nachman, C. G. Becker, and C. R. Minick. 1973. Culture of human endothelial cells derived from umbilical veins: identification by morphologic and immunologic criteria. J. Clin. Invest. 52:2745-2756. 22. Janoff, A., and J. Scherer. 1968. Mediators of inflammation in leukocyte lysosomes. IX. Elastinolytic activity in granules of human polymorphonuclear leukocytes. J. Exp. Med. 128:11371151. 23. Johnson, D., and J. Travis. 1979. The oxidative inactivation of human a-1-proteinase inhibitor: further evidence for methionine at the reactive center. J. Biol. Chem. 254:4022-4026, 24. Kuhn, H., K. Ponicke, W. Halle, T. Schewe, and W. Forster. 1983. Evidence for the presence of lipoxygenase pathway in cultured endothelial cells. Biomed. Biochim. Acta 42(5):K1-K4. 25. Martin, W. J., II. 1984. Neutrophils kill pulmonary endothelial cells by a hydrogen-peroxide-dependent pathway: an in vitro model of neutrophil-mediated lung injury. Am. Rev. Respir. Dis. 130:209-213. 26. Nathan, C. F., L. H. Brukner, S. C. Silverstein, and Z. A. Cohn. 1979. Extracellular cytolysis by activated macrophages and granulocytes. I. Pharmacologic triggering of effector cells and the release of hydrogen peroxide. J. Exp. Med. 149:84-99. 27. Ozaki, Y., T. Ohashi, and Y. Niwa. 1986. A comparative study on the effects of inhibitors of the lipoxygenase pathway on neutrophil function: inhibitory effects on neutrophil function may not be attributed to inhibition of the lipoxygenase pathway. Biochem. Pharmacol. 35:3481-3488.

INFECT. IMMUN. 28. Richmond, R., B. Halliwell, J. Chauhan, and A. Dabre. 1981. Superoxide-dependent formation of hydroxyl radicals: detection of hydroxyl radicals by the hydroxylation of aromatic compounds. Anal. Biochem. 118:328-335; 29. Sacks, T., C. F. Moldow, P. R. Craddock, T. K. Bowers, and H. S. Jacob. 1978. Oxygen radicals mediate endothelial cell damage by complement-stimulated granulocytes: an in vitro model of immune vascular damage. J. Clin. Invest. 6i:11611167. 30. Samuelsson, B. 1983. Leukotrienes: mediators of immediate hypersensitivity reactions and inflammation. Science 220:568575. 31. Siegel, M. I., R. T. McConnell, and P. Cuatrecasas. 1979. Aspirin-like drugs interfere with arachidonate metabolism by inhibition of the 12-hydroperoxy-5,8,10,14-eicosatetraenoic acid peroxidase activity of the lipoxygenase pathway. Proc. Natl. Acad. Sci. USA 76:3774-3777. 32. Stjernschantz, J. 1984. The leukotrienes. Med. Biol. (Helsinki) 62:215-230. 33. Taubman, S. B., and R. B. Cogen. 1975. Cell-detaching activity mediated by an enzyme(s) obtained from human leukocyte granules. Lab. Invest. 32:555-560. 34. Vandenbroucke-Grauls, C. M. J. E., H. M. W. M. Thissen, and J. Verhoef. 1984. Interaction between human polymorphonuclear leukocytes and Staphylococcus aureus in the presence and absence of opsonins. Immunology 52:427-435. 35. Vandenbroucke-Grauls, C. M. J. E., H. M. W. M. Thissen, and J. Verhoef. 1985. Phagocytosis of staphylococci by human polymorphonuclear leukocytes is enhanced in the presence of endothelial cells. Infect. Immun. 50:250-254. 36. Vane, J. R. 1971. Inhibition of prostaglandin synthesis as a mechanism of action for aspirin-like drugs. Nature (London) 231:232-233. 37. van Kessel, K. P. M., H. J. van Kats-Renaud, J. A. G. van StriJp, M. R. Visser, and J. Verhoef. 1986. Measurement of antibodymediated binding of human polymorphonuclear leukocytes to HSV-1 infected anchorage fibroblasts. J. Immunol. Methods 88:101-107. 38. Verbrugh, H. A., R. Peters, P. K. Peterson, and J. Verhoef. 1978. Phagocytosis and killing of staphylococci by human polymorphonuclear and mononuclear leukocytes. J. Clin. Pathol. 31:539-545. 39. Vercellotti, G. M., B. S. van Asbeck, and H. S. Jacob. 1985. Oxygen radical-induced erythrocyte hemolysis by neutrophils: critical role of iron and lactoferrin. J. Clin. Invest. 76:956-962. 40. Verhoef, J., P. K. Peterson, and P. G. Quie. 1977. Kinetics of staphylococcal opsonization, attachment, ingestion and killing by human polymorphonuclear leukocytes: a quantitative assay using [3H]-thymidine labelled bacteria. J. Immunol. Methods 14:303-311. 41. Weiss, S. J., M. B. Lampert, and S. T. Test. 1983. Long-lived oxidants generated by human neutrophils: characterization and bioactivity. Science 222:625-628. 42. Weiss, S. J., and A. F. LoBuglio. 1980. An oxygen-dependent mechanism of neutrophil-mediated cytotoxicity. Blood 55:10201024. 43. Weiss, S. J., and S. Regiani. 1984. Neutrophils degrade subendothelial matrices in the presence of alpha-1-proteinase inhibitor: cooperative use of lysosomal proteinases and oxygen metabolites. J. Clin. Invest. 73:1297-1303. 44. Weiss, S. J., J. Young, A. F. LoBuglio, A. Slivka, and N. F. Nimeh. 1981. Role of hydrogen peroxide in neutrophil-mediated destruction of cultured endothelial cells. J. Clin. Invest. 68: 714-721. 45. Weissman, G., H. M. Korchak, H. D. Perez, J. E. Smolen, I. M. Goldstein, and S. T. Hoffstein. 1979. The secretory code of the neutrophil. RES J. Reticuloendothel. Soc. 26:687-700. 46. Zucker-Franklin, D., and J. G. Hirsch. 1964. Electron microscope study of the degranulation of polymorphonuclear leukocytes following treatment with streptolysin. Am. J. Pathol. 47:419-433.