JOURNAL OF CELLULAR PHYSIOLOGY 153:53-61 (1992)

Human Umbilical Vein Endothelial Cells Submitted to Hypoxia-Reoxygenation In Vitro: Implication of Free Radicals, Xanthine Oxidase, and Energy Deficiency CARINE MICHIELS,* THIERRY ARNOULD, ANDREE HOUBION, AND JOSEREMACLE Laboratoire de Biochimie Cellulaire, F ~ c L J / ~Notre-Dame & de la Paix, 8-5000 ,harnur, Belgium Ischemia-reperfusion is observed in various diseases such as myocardium infarct. Different theories have been proposed to explain the reperfusion injury, among them that the free radical generation plays a crucial role. To study the mechanisms of the reperfusion injury, a hypoxia (H)-reoxygenation (R) model upon human umbilical vein endothelial cells in culture was developed in order to mimic the in vivo situation. Different parameters were quantified and compared under H or H/R, and we found that oxygen readmission led to damage amplification after a short hypoxia period. To estimate the importance of various causes of toxicity, the effects of various protective molecules were compared. Different antioxidant molecules, iron-chelating agent, xanthine oxidase inhibitors, and energy-supplying molecules were very efficient protectors. Synergy could also be observed between the antioxidants and the energy-supplying molecules or the xanthine oxidase inhibitors. The toxic effect of 0, . (-) could be lowered by the presence of SOD or glutathione peroxidase in the culture medium, whereas glutathione peroxidase was the most efficient cnzyme when injected into the cells. The production of 0,. (-) and of H,O, by endothelial cell3 was directly estimated to be, respectively, of 0.17 and 0.(135 pmol/minlmg prot during the R period. 0, . (-) production was completely inhibited whcn allopurinol was added during H and R. In addition, a xanthine oxidase activity of 21.5 10 U/mg prot could he observed by a direct assay in cells after H but not in control cells, thus confirming the previous conclusions of xanthine oxidase as a potent source of free radicals in these conditions. Thanks to the use of cultured human endothelial cells, a clear picture was obtained of the overall process leading to cell degenerescence during the reoxygenation process. We particularly could stress the importance of thc low energetic state of these cells, which is a critical factor acting synergistically with the oxidant molecules to injure the cells. These results also open new possibilities for the development of new therapeutics for ischemia. o 1992 W ~ I PLM,~ Inc.

Myocardial infarction is often treated by thrombolysis to achieve early reperfusion since it is one effective way to reduce infarct size and to improve cardiac functions. However, reperfusion by itself may have deleterious effects on the ischemic myocardial area. Cells remaining viable after ischemia can be injured by readmission of blood and oxygen. This enhanced toxicity is called “reperfusion injury.” Different theories have been advanced to explain this injury but the most frequently accepted is the generation of free radicals at the time of reperfusion (McCord, 1985).Readmission of oxygen causes a burst of free radical production that can damage cells and lead to this injury. Differcnt sources of production of these oxygen species have been implicated in vivo, such as PMN infiltration, mitochondrial electron transport, or catecholamine autooxidation, but the most important seems t o be xanthine oxidase (Parks et al., 1982).This enzyme is synthetized a s xanthine dehydrogenase but can be transformed into 0 1992 WILEY-LISS. INC.

xanthine oxidase during ischemia. This conversion is probably performed by a protease called calpai’ne, which is activated by calcium (Bindoli et al., 1988). Additionally, the depletion of ATP during ischemia results in a n elevated concentration of AMP, which is catabolized to hypoxanthine (Saugstad, 1988). When oxygen is readmitted, xanthine oxidase utilises hypoxanthine and 0, to produce superoxide anion and hydrogen peroxide (for a review, see Granger et al., 1986; Simpson and Lucchesi, 1987). This production of oxygen free radicals at the time of reperfusion was actually observed in vivo (Baker et al., 1988; Zweier et al., 19881, and protection of different organs by superoxide

Heceived October 9,1991; accepted April 6,1992. *To whom reprint requests/correspondence should be addressed.

54

MICIIIELS ET AL.

dismutase and catalase (Zweier et al., 19871, by cx-tocopherol (Schimke et al., 19871, or by xanthine oxidase inhibitor (Werns et al., 1987) are strong arguments for the involvement of these species in the reperfusion injury. The fact that a similar mechanism is observed in different organs suggests that there may be a common source. Endothelial cells are ubiquitous and seem t o be the first cell type to be injured in ischemia-reperfusion (Sunnergen and Rovetto, 1987; Suva1 et al., 1987; Lindal et al., 1988). In addition, damages to the microvasculature can induce a cascade of other damages leading to necrosis of the subendothelial tissue such as the necrosis described in the reperfusion injury. Capillary endothelial cells are a rich source of xanthine oxidase and the only potential source of this enzyme in the human heart (Jarasch et al., 1986) even ifxanthine oxidase has never been demonstrated in this organ (Downey et al., 1987). Because of the importance of endothelial cells as the possible source of free radical production and its role as the first line of defense of the microvascular tissue, we decided t o develop a n in vitro model of hypoxia-reoxygenation on cultured human umbilical vein endothelial cells that mimics the ischemia-reperfusion process. This in vitro model allows a reproduction of the damages with a direct quantification of the protective effects of various molecules so that their relative importance in the reoxygenation injury could be estimated. We also performed a quantification of the free radical production during R and gave evidence of a xanthine oxidase activity in human endothelial cells exposed to hypoxia.

MATERIALS AND METHODS Endothelial cell cultures Human umbilical vein endothelial cells were isolated according to Jaffe et al. (1973b). Cords were stocked a t 4°C just after birth in stock buffer (4mM KC1, 140 mM NaC1, 10 mM Hepcs, 1mM glucose, 100 pgiml streptomycin, 100 Uiml penicillin, and 0.25 pgiml fungizone, pH 7.3). Before manipulation, they were rinsed with phosphate-buffered saline (PBS) containing antibiotics and fungizone a t the same concentration as in stock buffer. Cords were incubated 35 min at 37°C with 4 ml collagenase type I1 (Sigma, St. Louis, MO) 0.05% in PBS. Cells were then harvested in M199 + 20% fetal calf serum (FCS) containing 10 mM Hepes (Gibco, Paisley, Scotland), centrifugated 10 min a t 1,000 rpm, and seeded in 0.25% gelatine-coated culture dishes (25 cm', Falcon Plastics, Oxnard, CA). After 1 day, plated cells were washed with medium to eliminate erythrocytes. Cells were used for experimental protocols in the first passage. Confirmation of their identity as endothelial cells was obtained by detecting factor VIII antigen assessed by immunofluorescence staining (Jaffe e t al., 1973a).

In vitro model of hypoxia-reoxygenation Ischemia-reperfusion was simulated by exposing cells to hypoxia (100% N,) followed by reoxygenation. Cells were seeded a t 2 x lo4 cells;cm2 on gelatinecoated glass coverslips (15 x 15 mm') set down in multidish (Sterilin, Feltham, UK). After l day, coverslips

were transferred in Petri dishes (0= 6 cm, Falcon) containing 1 ml of M199 buffered at pH 7.4 with 10 mM Hepes + 20% FCS. Medium was reduced to a n uniform thin layer to decrease the diffusion distance of the atmospheric gases. The hypoxia was produced with a n atmosphere of 100% N, in a n incubator gas chamber and oxygenation was restorcd with a 95%)air-5% CO, atmosphere; 1 ml of M199 + 20% FCS was added during reoxygenation. For experiments performed in Hank's solution, reoxygenation was performed in normal air in order to avoid pH changes due to the presence of CO,. PO, in the medium was 130 mmHg in normal conditions, about 15 mm Hg after 30 min H as described here, and it reached the air value (130 mm Hg) in 5 min during R. The concentration of dissolved oxygen was measured according to Winkler (Rodier, 1975). Chemicals Enzymes were purchased from Sigma (St. Louis); superoxide dismutase (SOD) and glutathione peroxidase (Gpx) were from bovine erythrocytes and catalase from bovine liver. To provide suitable control, SOD was inactivated 1h 30 a t 90°C in phosphate buffer 5 mM pH 7.2. This was checked by a SOD assay performed according to Corbisier et al. (1987). Allopurinol, a-tocopherol, folic acid, glutathione, and pyruvate also came from Sigma; mannitol and ascorbic acid were from Merck AG (Darmstadt, Germany); and desferrioxamine from Ciba-Geigy (Grand Bigard, Belgium). P-hydroxybutyrate was generously given by Solvay SA (Bruxelles, Belgium). 18-'H]-adenine was purchased from Amersham Corp. (Bruxelles, Belgium) (specific activity 23.5 Ciimmol). Cell viability Cell injury was assessed by following the erythrosine B uptake after incubation with 0.4% erythrosine in PBS for 5 rnin (Phillips, 1973).A minimum of 300 cells were counted for each determination. Cells unable to exclude the colorant were considered to be nonviable. [3H]-adenine release Cells seeded in Petri dishes were radiolabelled with 1 pCiiml in 2 ml M199 + 2 0 4 FCS for 20 h. After labelling, the cells were washed four times with 1ml Hank's buffer. Then, 1.4 ml of Hank's buffer was added and cells were incubated a t 37°C during different times under nitrogen (hypoxia) followed or not by 45 min of reoxygenation. After these incubations, the medium was collected and counted for 4 min in a liquid scintillation counter after the addition of Aqualuma (Lumac, Landgraaf: Nederlandsi. The radioactivity remaining in the cell monolayer was assayed on cells removed from the plate by 1 ml Hank's buffer containing 29% triton X-100. The percentage of isotope release was calculated as: 100 x (number of dpm in the extracellular fluididpm in the extracellular fluid + dpm remaining in the monolayer) (Andreoli et al., 1985). Microinjection Cells were first subcultivated in squared Petri dishes at a density of 100 cellsicm' and incubated a t 37°C under normal atmosphere. The squared dish and the low plating density allow the individual localization of

HYPOXIA-REOXYGENATION UPON ENDOTHELIAL CELLS

the injected cells. After 1 day, cells were injected and further exposed to 2 h H t 45 min. Microinjection by glass capillary was performed a s described by Graessmann et al. (1980). The injection buffer contained 5.45 mM K,HPO,, 4.55 mM KH2P0,, 70.5 KC1, 7.05 mM NaC1, pH 7.2. This technique allows injection of about lop1’ ml per cell. From this value we can calculate t h a t 0.13-, 3.2-, and 420-fold of the native cellular enzymatic activities are injected into each endothelial cell from solutions, respectively, of 360 Uiml Gpx, 11,000 Uiml catalase, and 240,000 Uiml SOD. Enzyme-injected cells and the corresponding control cells and buffer-injected cells were in the same dish to decrease the variability from dish to dish. Microinjection is not per se damaging for the cell growth o r the cell metabolism nor does it disturb the DNA or protein synthesis (Michiels et al., 1990b). The viability percentage was then determined a s previously described.

55

internal standard. The units of activity were calculated as described by Beckman et al. (1989). For the xanthine oxidase assay, endothelial cells were grown a t confluence in 75 cm2 flasks (Sterilin, Feltham, UK) that contain 3 x lo6 cells. They were washed twice with PBS, scraped, and homogenized in 50 mM K+-Hepesbuffer pH 7.4 containing 10 mM DTT (dithiothreitol), 0.1 mM EDTA, and 0.2 mM PMSF (phenyl methyl sulfonyl fluoride). The assay was performed immediately after the homogenization because of a rapid inactivation of the xanthine oxidase in these conditions.

Statistical analysis Results are expressed as means sd. Differences were evaluated for statistical significance using the Student’s t-test with a = 0.05.

Determination of radical formation 0,. (-j production was measured by cytochrome c reduction. Endothelial cells at confluence in Petri dishes (0 = 3.5 cm, Falcon) containing 0.33 10” cells were exposed to 2 h H. After this incubation, the cells were washed twice and 0.7 ml of cytochrome c a t 50 pM diluted in Hank’s balanced salt solution was added during 45 min R a t 37°C. At the end of the incubation, the test solutions were removed and their absorbance was measured in a cuvette at 550 nm (McCord and Fridovich, 1969). 0, . (-) concentration was calculated with E = 21 mM-’ cm-l. H202was determined fluorimetrically using scopoletin (Meier et al., 1989). Endothelial cells at confluence in Petri dishes (0 = 3.5 cm, Falcon) containing 0.33 lo6 cells were exposed to 2 h H. After this incubation, the cells were washed twice and 2 ml of PBS (phosphatebuffered saline) containing 2 pl of scopoletin (Fluka, Switzerland) a t 40 pM in 0.5 M phosphate buffer pH 7.0 and 10 pl of horse radish peroxidase solution (type IV, Sigma) (4 mg/ml in 0.05 M phosphate buffer pH 7.0) were added during the reoxygenation (45 min a t 37°C). At the end of incubation, the test solutions were removed and their fluorescence measured in a quartz cuvette. The excitation wavelength was 380 nm and the emission wavelength was 436 nm. Calibration was done with authentic H202.

RESULTS Hypoxia-reoxygenationtoxicity Cells were initially subjected to variable periods of hypoxia (H) followed or not by a 45 min period of reoxygenation (R) in order to determine when the toxicity due to the reoxygenation was observed (Fig. 1). The viability was measured by dye exclusion (Fig. 1A). Hypoxia alone led to cell death after 2 h 30 (11%mortality). However, reoxygenation increased significantly the mortality after 1h 30,2 h, and 2 h 30 of hypoxia. As a n example, there was 4% mortality after 2 h hypoxia but 13%after H + R. After 3 h, hypoxia alone was too toxic to study reoxygenation-induced damages. A longer period of reoxygenation (2 h) did not further increase the mortality (data not shown); 2 h H + 45 min R was selected as optimal for the experiments because cells died mainly due to reoxygenation and not to hypoxia. We calculated a mean on all the experiments performed for the viability of cells incubated 2 h under hypoxia and found 98.4 * 1.47% for 33 groups of 300 cells. We obtained 90.8 I 3.26% for 46 groups of 300 cells for the cells incubated 2 h under hypoxia followed by 45 min reoxygenation. These data show the high reproducibility of this experimental model. These results were confirmed when the cytotoxicity was measured by release of [€b3H1-adenine (Fig. 1B). There was a n increase of 14.5% of the adenine release during the 45-min reoxygenation, which followed the 2 h hypoxia.

Xanthine oxidase assay The oxidation of pterin to the fluorescent product isoxanthopterin was used to assay xanthine dehydrogenase and oxidase (Beckman e t al., 1989). The excitation wavelength was 345 nm and the emission wavelength was 390 nm. For the assay, 40 p1 of cell homogenate were diluted in 2 ml of 50 mM K+-phosphate, 0.1 ml EDTA, pH 7.4, warmed to 37°C in the quartz cuvette and any baseline drift measured. The rate of pterin oxidation was determined after adding 20 ~1 of 1 mM pterin, reflecting xanthine oxidase activity; 20 p1 of 1 mM methylene blue were then added a s a n electron acceptor to measure combined activities of xanthine dehydrogenase plus oxidase. The reaction was then inhibited by the addition of 20 p1 of 1 mM allopurinol. Finally, 20 pl of 0.1 pM isoxanthopterin were added as

Protection experiments Different mechanisms have been described in order to explain cell death during reoxygenation. The most important seems to be free radical generation (McCord, 1985). Therefore, the protection of the reoxygenationinduced cell death was first tested with antioxidant molecules. These molecules were added in the culture medium at different concentrations during the whole incubation. The viability percentage was then determined after H or H + R and compared t o control cells exposed to the same incubation without any supplementation. An example of such a n experiment is illustrated in Figure 2 for a-tocopherol, a lipophilic antioxidant molecule. The effect of hypoxia alone was very low, between 1 and 3% mortality, and there is no effect of

MICHIELS ET AI,.

56

80' 0

50

100

150

1

I

control

-3

-3,3

-4

-5

1

-6

200

Vitamin E concentration (log M)

Hypoxia time (min)

Fig. 2. Effect of different concentrations of or-tocopherol on human endothelial cells submitted to 2 h of hypoxia ( 0 ) or 2 h of hypoxia followed by 45 min of reoxygenation (m). Results are expressed as mean i_ 1 sd for 2 groups of 300-350 cells in two different dishes (n = 4).* significantly different from control cells incubated without u-tocopherol added with OL = 0.05.

I 0

50

100

150

200

Hypoxia time (rnin)

Fig. 1. A, Evolution of the cellular viability measured by dye exclusion of human endothelial cells incubated during increasing times under hypoxia ( 0 )or hypoxia followed by 45 min of reoxygenation (=I. Results are expressed as means i 1 sd for 1 group of 300-350 cells in two different dishes (n = 2). B, Evolution of the cytotoxicity measured by release of [8-'H]-adenine from human endothelial cells incubated during increasing times under hypoxia ( 0 )or hypoxia followed by 45 minutes of reoxygenation (M). Results are expressed as means ? 1 sd for two dishes. * significantly different from cells incubated under hypoxia alone with N = 0.05.

a-tocopherol on this process. The reoxygenation step increased by 12%the cell mortality and the toxicity was well counteracted in a dose-dependent manner by a-tocopherol with a nearly complete protection at M compared to the H alone. a-tocopherol was also tested a t 5 x l o p 4M when added only during reoxygenation. The protection was the same as when added during H R (data not shown). This resull shows that damages occurred only when oxygen was readmitted and not during hypoxia. This confirms the free radical hypothesis. Other molecules were also tested: ascorbic acid (hydrophilic antioxidant), mannitol (OH . scavenger), glutathione and cysteine (0, . ( -) scavengers, and thiol containing molecules), desferrioxamine (DFO, iron chelator), and three antioxidant enzymes (SOD, Gpx, and catalase). We used different concentrations for the

+

chemicals and for the antioxidant enzymes a s proposed by Ratych et al. (1987). The results from all these experiments are presented in Table 1A for the chemicals and Table 1B for the enzymes with the statistical analysis. As for a-tocopherol, ascorbic acid, mannitol, and DFO afforded a complete protection that decreased with dilution. These molecules had no influence on the viability after hypoxia alone. Glutathione presented a n optimal concentration for protection M). At lower concentrations, it did not give any protection. This protection a t 10 M was significant, highly reproducible, and also observed with cysteine, another thiol-containing molecule, a t the same concentration. Different concentrations of extracellular antioxidant enzymes were also studied: catalase even a t 1,000 Uiml was not efficient but SOD at 150 Uiml but not a t 15 Uiml and Gpx at 5.1 Uiml were reproducibly very good protectors. This protection was actually due to enzymatic activity since heat-inactivated SOD (150 Uiml) did not give any protection. We also tested these different antioxidant enzymes when injected directly into the cells by the microinjection technique just before the hypoxia incubation. The concentrations chosen were those used in a previous work on endothelial cells submitted to hyperoxia (Michiels et al., 1990a). The results shown on Table 2 show a highly significant protection by Gpx and a weak protection by catalase. SOD was inefficient. Similar results of protection with catalase and Gpx but not with SOD were obtained for endothelial cells submitted t o hyperoxia (Michiels et al., 1990a). This similarity as well a s the results of Table 1 confirm the importance of the oxidative stress during the reoxygenation period. They suggest that free radicals are generated during reoxygenation and are responsible for cell death. Different sources of free radicals have been described during H + R but the most frequently advanced is xanthine oxidase (Granger, 1988). In order to test this hypothesis, we used two inhibitors of this enzyme, allopurinol and folic acid (Granger et al.,

57

HYPOXIA-REOXYGENATION UPON ENDOTHELIAL CELLS TABLE 1. Viability 1%) of cndothelial cells exposed to 2 h H A. Concentrations

lo-,''

Vitamin C

Gluthath ione

Mannitol

M M

510 4 1 0 ~M 10-5 M lo-' M 10-7 M

+ 45 min R measured by dyc exclusion' Desferrinxam ine

97.7 i 0.35" 96.5 ? 1.48"

M

98.1 i 1.6* 94.2 i- 1.34 95.5 t 2.3 88.5 i 4.9 90.9 i 6.08

96.1 ? 2.1" 96 t 2.82 94.5 2.12

98.2 ? 1.06" 94 2 0 93.6 t 0.14 93.8 L 1.41

+

99.5 ? 0.58 (4)" 99 2 0.82 (4Y 99.5 2 0.58 (4Y 96.67 ? 4.9 (3) 98.88 I 0.85 (41"

M

Cysteine lo-' M Control cells

97.9 t 0.63' 87 i 4.2

90

?

lt3)

93.3 i 1.52 (3)

B. Enzymes

92.25

-t

3.1 (4)

Enzymes

Catalase (1000 Uimll Catalase (100 U/ml) SOD (150 Uiml) SOD (15 Uiml) Gpx (5.1Uimlj Control cells

SOD (150 U/ml) inactivated SOD

88.5 ? 1.7 90.1 t 0.7 97.7 t 2 . 4 89.8 t 4.03 97.3 t 0.91"

98.2 & 0.35" 94.15 t 1.62

90.3 t 3.5

94.1 i 0.49

'Different molecules were diluted in the culture medium Values for conlrol cells are noted for each experiment. Results are expressed as means ? 1sd on onc group of300 cells in two different dishes n - 2) except in A. ifnoted hetween parentheses (n = 3 or 41. *Significantly different from control cclls with N = 0.05.

TABLE 2. Viability

(%)

of endothelial cells exposed to 2 h H

+ 45 min R measured by dye exclusion'

Enzyme

Enzyme injection

Buffer injection

No injection (Control cells)

Catalase (11,000 1Jiml) SOD (40.000 U/mlj Gpx (360 Uiml)

97.5 k 1.73 (4)" 94.2 ? 1.92 (5) 99.5 i- 0.58 14)","

95.67 2 2.31 (31b 95.75 f 2.22 (416 94.2 2.28 l5N

95.25 f 0.50 (4) 96.0 i 1.22 (5) 95.4 L 1.14 (5)

+

'DilTerent enzymes were injected directly into the cells before these incubations. Values for control cells ar e noted for each experiment. Results are expressed as means 1 sd on three. four, or five groups of 80 cells (specifiedhetween parenthcscs). *Significantly different from buffer-injected cells with LI - 0.05, significantly different from noninjecled cells with a = 0.05: S nonsignificantly different from noninjected cells. +

TABLE 3. Viahilily (36)of endothelral cells exposcd to 2 b I1 Concentrations

5 10-4 M 10-4 M M 10-6 M 10-7 M M 10-9 M Control cells

Allopurinol

+ 45 min R measured

Folic acid

by dye exclusion'

BOBA

Pyruvate

*

98.2 0.28 (3)" 96.8 i- 0.07 (3)" 95.5 i- 0.91* 97.3 i- 0.91" 96.8 2 1.18" 92.9 i- 0.63% 91.7 ? 0.7" 87.4 i 2.19 (4)

98.4 f 0.8" 98.6 t 0" 94.9 f 1.56 94.8 0*

*

91.8 I 1.56 (3)

97.25 ? 1.5 (41% 97 1 0* 95 t 2 (3)* 96.5 i 0.7" 94.2 i- 1.41* 93.1 f 0"

91.3

2

1.15 (3)

97.6 +95.8 94.3 i93.2 -t

0.14 0.75

92.6

0.85 (3)

1.13"

* 1.4

+

'Diffcrcnt molecules wcrc diluted 111 the culture medium during lhe whole incubation. Values for control cells alp noted for trach experiment. Results are expressed as means ? 1 sd on 1 group of 300 cells in two different dishes ( n = 2) except if noted hetween parentheses In = 3 or 41. ^Significantly different from control cells with a - 0 05

1986). Table 3 illustrates the effects of both molecules a t different concentrations. The results show that both inhibitors can completely protect cells against reoxygenation-induced cell death and that this protection decreased with dilution. Allopurinol seemed to be more efficient than folic acid. No effect was detectable on viability after hypoxia alone. These results suggest that xanthine oxidase is a major source of free radicals when human endothelial cells are incubated under H + R. Since reoxygenation occurs after hypoxia, a deficiency in energy was also proposed a s one main factor

for the toxic action of the reoxygenation (McDonough and Spitzer, 1983). ATP is indeed metabolized during hypoxia due to the arrest of oxidative phosphorylation. Even if this decrease does not lead to viability loss after 2 h of hypoxia, it can weaken cells or decrease their defenses against free radicals for example. This hypothesis can be tested in the model by addition to the culture medium of molecules that can supply for energy: P-hydroxybutyrate (BOBA) or pyruvate. The results show that both chemicals can also completely protect cells against H + R (Table 3). This protection decreased with dilution, but BOBA was more

58

MICHIELS ET AL

TARLE 4. Recapitulation of protectivc effects of the different molecules added to the culture medium when endothelial cells are exposed t o 2 h H + 45 min R’ Concentrations that gave ahout 50% protection

Molecules Vitamin F Wit E! Vitamin C Wit C) Mannitol Glutathione (GSH) Dcsferrioxamine (DFOJ Allopurinol (allo) Fohc acid 8-hydroxyhutyrate (BORA) Pyruvate Catalase

M 5 M 10-‘-M 5 lo-” M 10-7

M

510 “ M 10 f i M 510 ‘M

-

SOD

M

100 U/ml 3.6 U/ml

Gpx

’ The “‘ir protection” was calculated a6 ~ollows: % viability of protected cells after 2 h I1 t 46 rnin R

c( viahility of control cells after 2 h H alone

~

Q viability of’control cells after 2 b H t 45 min K % viability of control cells after 2 h H + 45 min R

efficient than pyruvate. BOBA was also tested when added only during reoxygenation and found a s efficient as when added during the whole incubation (data not shown). Since it is used immediately by the Kreb’s cycle, it can speed up the ATP recovery (McDonough and Spitzer, 1983). In order to have a comparative view of all these effects, we calculated for each protective molecule the concentration that gives 50% protection (Table 4).I n this way, the efficiency of the nine molecules can be compared in the same conditions in one experimental model for their protection of endothelial cells against the reoxygenation process. If the protection observed with all these molecules confirmed the involvement of their respective mechanism in the cell toxicity, the table stresses the importance of the xanthine oxidase inhibition by allopurinol (5 x lope M), the requirement of energy supplementation by P-hydroxybutyrate (5 x lop8M)and the Fcnton reaction, which is counteracted by desfcrrioxamine ( l o p 7M).

a xanthine oxidase inhibitor (allopurinol) or to DFO. However, ascorbic acid and a-tocopherol were not synergistic.

Free radical production O2 . (-1 and H,O, were found released during reoxygenation of endothelial cells after 2 h H. However, we could not detect a significant production of these radicals when cells were incubated in normal atmosphere. Quantification of 0, . (-j by cytochrome c reduction allowed to detect the production of 0.169 r 0.014 FM; minimg prot (n = 3) during the 45 min R. This production was completely inhibited by 1,200 Uiml SOD in the test solution during R or by allopurinol 5 x M when added in the test solution during H and R since the obtained values were 0.027 k 0.027 yMiminimg prot ( n = 4) in the presence of allopurinol and 0.025 * 0.024 pMimidmg prot (n = 4) for the control air-incubated cells. Five x lop4M allopurinol was the optimal protective concentration against R injury. H,O, generation was measured fluorimetrically with scopoletin and horse radish peroxidase. We observed a production of 0.035 i 0.005 pM/midmg prot (n = 3) during the 45 min R. This production was completely inhibited by 1,000 U/ml catalase. These results demonstrate that free radicals were actually generated by endothelial cells during reoxygenation, that they could escape in the extracellular fluid, that 0, . (-) seemed to be more abundant than H,02, and that xanthine oxidase could be the source of the 0, . (-1 produced during R. Xanthine dehydrogenase and oxidase assay In order to identify the suspected source of free radicals produced by endothelial cells during R, a xanthine dehydrogenase and oxidase assay were performed according to Beckman et al. (1989) on human endothelial cells. The activity of both enzymes were measured in control cells and in cells after 2 h H or 2 h H + 45 min R. We were able to detect xanthine oxidase only in cells submitted to 2 h H: the activity was 21.53 2 14.15 lop6 Uimg prot (n = 6). The xanthine oxidase activity in cells exposed to 2 h H + 45 min R seemed to be just below the detection level of the assay. Xanthine dehydrogenase activity was never detected. In addition, the xanthine oxidase activity was very labile and disappeared in 1h even at 4°C. This can reflect the possibility of the presence of a specific inhibitor of xanthine oxidase and dehydrogenase in human endothelial cells (B.A. Freeman, personal communication).

Combinations of different protective molecules Since the toxic process seems to be a complex phenomenon involving various mechanisms a s exemplified in Table 4, we decided to test the possible synergistic effect when acting on various levels of the process. The mortality of cells after 2 h H + 45 min R is relaDISCUSSION tively low (10%)so that low concentrations were used in Reperfusion injury is a complex process that has multhese experiments in order to observe eventual additive or synergistic effects. Table 5 illustrates the results of tiple clinical implications and where free radicals have three experiments with various combinations of mole- been proposed to be one of the factors responsible for the cules. Comparison of the observed effect with the theo- cell toxicity (McCord, 1985). The readmission of oxygen retical protection calculated from the addition of the after a period of ischemia leads to a burst of reactive separated molecules gives a n estimation of the syner- oxygen-derived molecules, which can damage the cells gistic effect of the molecules. We actually obtained syn- weakened by the ischemia. The production of O2 . (-1 ergistic effects when a n energy supplier (BOBA) was and O H . has been shown by Zweier et al. (1988) and combined with a n antioxidant molecule (a-tocopherol, Zweier (1988), whereas Brown et al. (1988) could demDFO) or with a n inhibitor of the free radical-generating onstrate the presence of H20,. The increase of GSSG system (allopurinol). Synergistic effects were also ob- concentration is also a clear indication of a n oxidative served when a n antioxidant (a-tocopherol) was added t o stress (Jenkinson et al., 1988). In spite of these evi-

IIYPOXIA-REOXYGENATION UPON ENDOTHELIAL CELLS

59

TABLE 5 . Combinations of different protective molecules upon endothelial cells incubated under 2 h H 45 min R1 Protection of molecules when added alone I%) Vit E 5

M

’M

BOBA 10

DFO M Vit E 5 10 M

31 18 47

Protection calculated for the addition of the separated molecules ‘7,)

Protection ohtained with combined molecules (F)

49

77

/ 65

85

---

’

allo

M

BOBA lo-’ M Vit C 5 10 Vit E 5

32 0

+

91

1 5 -

60

32

‘M

28

M 97

DFO 5 1 0 P M IFor each of th? three enpenments, the % protection (calculated a s In Table 41 IS given for each niolccule alone and for t h c combinations. The values calculated for a lheoretical addition of the separated proleclion are also noted. All data come from a mean oblained on one group of 300 cells In two different dishes ( n - 2).

dences, the importance of free radicals in the injury is questioned mostly because of the absence of direct evidence for a xanthine oxidase activity in human endothelial cells (Downey et al., 1987) and because of negative or very low level of protection by antioxidants of myocardium after ischemia-reperfusion found in some cases (for a review see Engler and Gilpin, 1989). In this work, a model of human endothelial cells in culture was developed in order to test and quantify different parameters affecting the reoxygenation toxicity. Only a few such in vitro experiments have been described in the literature using rat endothelial cells (Ratych et al., 1987) or bovine endothelial cells (Inauen et al., 1990a,b). All the other works are performed on perfused organs or in vivo. The model described here first showed that the reoxygenation-induced injury was mostly limited between 1 and 3 h of hypoxia with a maximal effect after 2 h. Thereafter, hypoxia alone is the main deleterious factor. This period could be compared to the “therapeutic window,” which allows some protection by antioxidants as demonstrated by Zweier et al. (1989) on perfused rabbit heart. A rational approach for testing the importance of the various molecules which could be involved in the reoxygenation process was based on a scheme modified from Granger et al. (1981) and presented in Figure 3. Antioxidant molecules and enzymes were tested directly by addition in the culture medium or after microinjection inside the cells. Other metabolic pathways could be blocked by inhibitors like allopurinol, folic acid, desferrioxamine or activated like the ATP or NADPH production by P-hydroxybutyrate or pyruvate. Three antioxidant molecules, a-tocopherol, ascorbic acid, and mannitol were protective with similar efficiency since micromolar concentrations were able to give 50% protection (Table 4). Cysteine and glutathione were also protective but were less efficient. This protection by antioxidant molecules indicated a clear involvement of free radicals in the reoxygenation-induced injury since the protection was the same if a-tocopherol, for example, was added only during the reoxygenation period.

@ J hTPL I

-

.

deleterous enecb

0 MDP

GDPH GSSO

2

QSH

J.

vitc

REOXYGENATDN

Fig. 3. Schematic representation of various reactions occurring during the hypoxia-reoxygenation process, which indicates the action of antioxidant molecules and enzymes as well as of the various inhibitors (modified from Granger et al., 1981)

The very efficient protection given by desferrioxamine, a n iron chelating agent, confirmed these conclusions and indicates that, as for the other oxidative stress states, the hydroxyl radical generation catalyzed by transition metals is a main intermediate responsible for the toxic effects. All the three antioxidant enzymes were effective but with different behaviour according to the tested system. In the culture medium, SOD and Gpx were protective but not catalase. When injected into the cells, Gpx and, to a less degree, catalase were protective but not SOD. In the two situations, Gpx was the most effective so that the importance of lipid peroxidation also appears as a main cause of cell death during the reoxygenation. Gpx is only effective on free fatty acid hydroperoxides that are released by phospholipase A, from peroxidized phospholipids (Sevanian et al., 1985). The

60

MICHIELS ET AL.

different behaviours of catalase and SOD according t o their localization are interesting since they indicate a different importance for H,02 and O2 (-1 inside and outside the cells. Quantitatively, the amount of 0,. (-1 in the medium was found fivefold higher than the H,O, production during the 45 min R. This difference in quantity as well as the more toxic effect of 02.(-) outside the cells might explain the extracellular protection by SOD compared to catalase. H,02 and 0,. (-) are produced intracellularly during the reoxygenation incubation. As we could measure them in the culture medium during this period, we suppose that they escape from the cells. O H . is then generated both intraand extracellularly and can initiate lipid peroxidation within the cells. The intracellular protection of catalase, however, confirms the higher toxicity of H,O, inside the cells submitted to oxidative stress (Michiels et al., 1990b). In vivo also, SOD (Werns et al., 1988; Zweier et al., 1987) and a n heme peroxidase (Menasche et al., 1986) are beneficial but few positive results have been obtained for catalase. The involvement of free radicals in hypoxia-reoxygenation was further investigated. First, free radical production was assayed during R: H,O, and more particularly O , - ( - ) were indeed produced, which is in agreement with the data of Zweier et al. (1988) on bovine endothelial cells. Second, we examined to what extent xanthine oxidase was responsible for this free radical production, as it has been claimed by Zweier et al. (1988) in their model. In our hands, the powerful protection observed with allopurinol at very low concentrations as well as the total inhibition of 0,. (-) production during R when allopurinol was added to the cells during H and R confirms the crucial role of xanthine oxidase in this production. This was also confirmed by Simpson et al. (1987), Miura et al. (1988), and Ratych et al. (1987). In addition, we detected some xanthine oxidase activity in human endothelial cells incubated under hypoxia. This is in disagreement with Downey et al. (1987) who failed to show any activity on human heart. But the low levels and the lability of the enzyme observed in our experiments could perhaps explain why the presence of xanthine oxidase in human endothelial cells is still controversial. If the normal oxygen concentration is not toxic for the cells, whereas i t is after a n hypoxia period, it is also probably because of the lowered energy content due to the period of hypoxia. This deficit could impair many cellular functions so that cells are not able to resist to the oxidative stress during reoxygenation. In fact, both p-hydroxybutyrate and pyruvate, which can be quickly incorporated into the Kreb’s cycle, were very efficient protectors. The protective effect of pyruvate on myocytes has already been described (McDonough and Spitzer, 1983) but p-hydroxybutyrate was far more efficient with a 50% protection at 5 x lo-’ M (Table 4). P-hydroxybutyrate is also capable of reducing NADPt thus providing reduction equivalents for the GSH cycle. Comparison of the efficiency of the molecules as calculated in Table 4 clearly shows that among important reactions to be controlled for lowering cell mortality are the production of 0,. (-) by xanthine oxidase with a possible inhibition by allopurinol, the formation of OH . in the Fenton reaction with a possible inhibition

by desferrioxamine, a n inhibition of lipid peroxidation with good protection by a-tocopherol and Gpx, and a rapid increase in energy as given by p-hydroxybutyrate. Combinations of such protections were found very effective with a very pronounced synergistic effect when using a-tocopherol and the three other protectors. The combination of a-tocopherol and ascorbic acid was, however, inefficient . Even if these observations were obtained in a model with cells in culture, they show that human endothelial cells can be effectively protected by acting on various parameters involved in the production of free radicals produced during the reoxygenation. In vivo, the destruction of such cells could be amplified and leads to a reperfusion injury for example by recruiting and activating neutrophils (Wach e t al., 1987). The ischemiareperfusion is more complex, but we believe that the optimization of the protection of endothelial cells, as observed in this work, could be tested in many pathological situations linked to the ischemia-reperfusion toxicity like the myocardial infarct, the renal ischemia, and organ transplantation (Fuller et al., 1988).

ACKNOWLEDGMENTS The umbilical cords were graciously given by Civil Hospital and Notre Dame Hospital at Charleroi. C. M. is a Senior Research Assistant a t the F.N.R.S. (Fonds National de la Recherche Scientifique, Brussels), and T. A. is a fellow of I.R.S.I.A. (Institut pour 1’Encouragement de la Recherche Scientifique dans YIndustrie et 1’Agriculture). LITERATURE CITED Andreoli, S.P., Baehner, P.L., and Bergstein, J.M. (1985) In vitro detection of endothelial cell damage using 2-deo~y-D-~H-glucose: Comparison with chromium .il, ‘H-leucine, ‘H-adenine and lactate dehydrogenase. J. Lab. Clin. Med., 106:253-261. Baker, J.E., Felix, C.C., Olinger, G.N., and Kalyanaraman, B. (1988) Myocardial ischemia and reperfusion: Direct evidence for free radical generation by electron spin resonance spectroscopy. Proc. Natl. Acad. Sci., 85:278&2789. Beckman, J.S., Parks, D.A., Pearson, J.D., Marshall, P.A., and Freeman, B.A. (1989)A sensitive fluorometric assay for measuring xanthine dehydrogenase and oxidase in tissues. Free Rad. Biol. Med., 6.507-615. Bindoli, A., Cavallini, L., Rigobello, M.P., Coassin, M., and Dilisa, F. ( 1 988) Modification of the xanthine-converting enzyme of perfused rat heart during ischemia and oxidative stress. Free Rad. Biol. Med., 4r163-167. Brown, J.M., Terada, L.S., Grosso, M.A., Whitman, G.J., Velasco, S.E., Patt, A,, Harken, A.H., and Repine, J.E. (1988) Hydrogen peroxide mediates reperfusion injury in the isolated rat heart. Mol. Cell. Riochem., 84:173-175. Corbisier, P., Houbion, A., and Remade, J . (1987) A new technique for highly sensitive detection of superoxide dismutase activity by chemoluminescence. Anal. Biochem., I64 :240-247. Downey, J.M., Miura, T., Eddy, L.J., Chambers, D.E., Mellert, T., Hearse, D.J., and Yellon, D.M. (1987) Xanthine oxidase is not a source of free radicals in the ischemic rahbit heart. J. Mol. Cell. Cardiol., 19:1053-1060. Engler, R., and Gilpin, E. (1989) Can superoxide dismutase alter myocardial infarct size? Circul., 79:1135-1142. Fuller, B.J., Gower, J.D., and Green, C.J. (1988) Free radical damage and organ preservation: Fact or fiction’?Cryobiol., 25r77-393. Graessmann, A,, Graessmann, M., and Mueller, C. (1980) Microinjection of early SV40 DNA fragments and T antigen. Meth. Enzymol., 65:s 16-825. Granger, D.N. (1988) Role of xanthine oxidase and granulocytes in ischemia-reperfusion injury. Am. J . Physiol., 255:H1269-H1275. Granger, D.N., Rutili, G., and McCord, J.M.C. (1981) Superoxide radicals in feline intestinal ischemia. Gastroenterology, S1.22-29.

IIY POXIA-REOXYGENATION UPON ENDOTHELIAL CELLS

Granger, D.N., Hollwarth, M.E., and Parks, D.A. (1986) Ischemiareperfusion in jury: Role of oxygen-derived free radicals. Acta Physiol. Scand. Supp., 548:47-63. Inauen, W., Granger, D.N., Meininger, C.J., Schelling, M.E., Granger, H.J., and Kvietys, P.R. (1990a) Anoxia-reoxygenation-induced neutrophil-mediated endothelial cell injury: Role of elastase. Am. J. Physiol., 259:H925-H931. Inauen, W., Granger, D.N., Meininger, C.J., Schelling, M.E., Granger, H.J., and Kvietys, P.R. (1990b) An in vitro model of ischemia/ reperfusion-induced microvascular injury. Am. cJ. Physiol., 259:G134-G139. Jaffe, E.A., Hoger, L.W., and Nachman, R.L. (1973a3 Synthesis of antihemophilic factor antigcn by human endothelial cells. J . Clin. Invest., 529757-2764. Jaffe, E.A., Nachman, R.L., Becker, C.G., and Minich, C.R. (197313) Culture of human endothelial cells derived from umbilical cord. J. Clin. Invest., 522745-2756. Jarasch, E.D., Bruder, G., and Heid, H.W. (1986) Significance of xanthine oxidase in capillary endothelial cells. Acta Physiol. Scand. Supp., 548t3946. Jenkinson, S.G., Marcum, R.F., Pickard, J.S.? Onzechowski, Z., Lawrence, R.A., and Jordan, J.M. (1988) Glutathione disulfide formation occuring during hypoxia and reoxygenation of rat lung. J. Lab. Clin. Med., 112:471480. Lindal, S., Sorlie, D., and Jorgensen, L. (1988) Endothelial cells of the cardiac microvasculature during and after cold cardioplegic ischemia. Scand. J . Thor. Cardiovasc. Surg., 22,257-265. McCord, J.M. (19853 Oxygen-derived free radicals in postischemic tissue injury. N. Engl. J . Med., 312:159-163. McCord, J.M., and Fridovich, I. (1969) Superoxide dismutase. J. Biol. Chem., 244r60494055. McDonough, K.H., and Spitzer, J . J . (1983) Effect ofhypoxia and reoxygenation on adult rat heart cell metabolism. Proc. SOC. Exp. Biol. Med., 173519-526. Meier, B., Radeke, M. H., Sellc, S., Youncs, M., Sies, H., Resch, K., and Habermehl, G.G. (1989) Human fibroblasts release reactive oxygen species in response to interleukin-1 or tumor necrosis factor-a. Biochem. J.,263t539-545. Menasche, P., Grousset, C., Gandel, Y., and Piwnica, A. (1986) A comparative study of free radical scavengers in cardioplegic solutions. J . Thorac. Cardiovasc. Surg., 92:264-271. Michiels, C., Toussaint, O., and Remacle, J. (1990a) Comparative study of oxygen toxicity in human fibroblasts and endothelial cells. J. Cell. Physiol., 144r295302. Michiels, C., Raes, M., Pigeolet, E., Corbisier, P., Lamhert, D., and Remacle, J. (1990b) Importance of a threshold for error accumulation in cell degenerative processes. I. Modulation of the threshold in a model of free radical-induced cell degeneration. Mech. Ageing Dev., 51:41-54. Miura, T., Yellon, D.M., Kinyma, J . , and Downey, J.M. (19881 Protection afforded by allopurinol in the first 24 hours of coronary occlusion is diminished after 48 hours. Free Rad. Biol. Med.. 425-30.

61

Parks, D.A., Bulkley, G.B., Granger, D.N., Hamilton, S.K., and McCord, J.M. (1982) Ischemic injury in the cat small intestine: Role of superoxide radicals. Gastroenterol., 82:9- 15. Phillips, H.J. (1973) Dye exclusion tests for cell viability in: Tissue Culture: Methods and Application. P.F. Kurse and M.K. Petterson, eds. Academic Press, New York, p. 406. Ratych, R.E., Chukugisha, R.S., and Bulkley, G.B. (1987)The primary localization of free radical generation after anoxia-reoxygenation in isolated endothelial cells. Surgery, 102:122-13 1. Rodier, J. 11975)In: L'analyse de l'eau, Dunod, Paris, pp. 384388. Saugstad, O.D. (1988) Hypoxanthine as an indicator of hypoxia: Its role in health and disease through free radical production. Ped. Kes., 23:143-150. Schimke, I., Schimke, E., Papies, B., and Moritz, V. (1987) Importance of the antioxidant potential for free radical induced heart damage. Riomed. Biochim. Acta, 46:S576-S579. Sevanian, A,, and Kim, E. (1985) Phospholipase A2 dependent release of fatty acid from peroxidized membranes. J. Free Rad. Biol., 1263271. Simpson, P.J.,and Lucchesi, B.R. (1987)Free radicals and myocardial ischemia and reperfusion injury. J . Lab. Clin. Med., 110t13-30. Simpson, P.J., Michelson, J.K., and Lucchesi, B.R. (1987) Free radical scavengers in myocardium ischemia. Fed. Proc., 46:2413-2421. Sunnergen, K.P., and Rovetto, M.J. (1987) Myocyte and endothelial injury with ischemia reperfusion in isolated rat hearts. Am. J. Physiol., 252:H1211-H1217. Suval, W.D., Duran, W.N., Boric, M.P., Hobson, R.W., Berendsen, P.B., and Ritter, A.B. (1987) Microvascular transport and endothelial cell alterations preceding skeletal muscle damage in ischemic and reperfusion injury. Am. J . Surg., 154r211-218. Wach, F., Hein, R.. Adelman-Grill, B.C., and Krieg, T. (1987) Inhibition of fibroblasts chemotaxis by superoxide dismutase. Eur. J . Cell. Biol., 44r124-127. Werns, S.W., Simpson, P.J., Michelson, J.K., Shea, M.J., Pitt, B., and Lucchesi, B.R. (1988) Sustained limitation by superoxide dismutase of canine myocardial injury due to regional ischemia followed by reperfusion. J. Cardiovasc. Pharmacol., 11:36-44. Zweier, J.L. (1988) Measurement of superoxide-derived free radicals in the reperfused heart. J. Biol. Chem., 263:1353-1357. Zweier, J.L., Rayburn, B.K., Flaherty, J.T., and Weisfeld, M.L. (1987) Recombinant superoxide dismutase reduces oxygen free radical concentrations in reperfused myocardium. J. Clin. Invest., 80:17281734. Zweier, J.L., Kuppusamy, P., and Lutty, G.A. (1988)Measurement of endothelial cell free radical generation: Evidence for a central mechanisms or free radical injury in post ischemic tissues. Proc. Natl. Acad. Sci., 855t4046-4050. Zweier, J.L., Kuppusamy, P., Williams, R., Rayburn, B.K., Smith, D., Weisfeldt, M.L., and Flaherty, J.T. (1989)Measurement and characterization of postischemic free radical generation in the isolated perfused heart. J. Biol. Chem., 264t18890-18895.