INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

BECAUSE OF THEIR LOCATION at the interface of the vascular system, endothelial cells in the blood vessel wall are from time to time exposed to peroxides, for instance, during local inflammatory reactions or contact with oxidized lipoproteins. In vitro experiments have indicated that endothelial and other cells, when subjected to oxidative stress with hydrogen peroxide or tert-butyl hydroperoxide (tBH), undergo remarkable changes in morphology and in the structure of the actin cytoskeleton, often resulting in membrane blebbing in a manner resembling that due to apoptosis (programmed cell death) (4, 14, 15, 19, 28). A whole variety of intracellular signals have been put forward as intermediates of these peroxide-evoked changes in cell function and structure, e.g., elevated levels of intracellular Ca²⁺ (9, 38); activation of phospholipase D (30), mitogen-activated protein kinases (5, 20), or calpain (28); heat shock protein 27 phosphorylation (12, 19); and changed concentrations of cGMP (23) or cAMP (16). However, the precise mechanism by which peroxides, and thus oxidative stress, trigger cells is still a matter of debate.

Another well-documented direct effect of peroxides, and especially of tBH, on cells is the oxidation of cytosolic glutathione, which is normally present in its reduced form (GSH) (8, 33). Although there are good indications of ties between oxidative stress, glutathione depletion, and the onset of apoptosis (reviewed in Ref. 6), these have not yet been studied in endothelial cells. In the present paper, we report on the effects of tBH on a number of structural and functional properties of human umbilical vein endothelial cells (HUVEC) and two HUVEC-derived cell lines, ECV304 (34) and ECRF24 (10). We note remarkable differences in the capacities of these cells to respond by actin reorganization, membrane blebbing, and nuclear condensation, which closely parallel differences in sensitivity to GSH oxidation. Because these tBH-evoked morphological changes suggested the commencement of apoptosis, we determined two markers of this process, phosphatidylserine (PS) externalization (21, 35) and DNA fragmentation (31). It appeared that even prolonged incubation with high doses of tBH failed to induce these apoptotic responses.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Materials. Cell culture media were purchased from Life Technologies (Breda, The Netherlands), human serum was provided by the Red Cross Blood Bank Zuid-Limburg (Maastricht, The Netherlands), bovine brain extract was from Boehringer (Mannheim, Germany), heparin was from ICN (Zoetermeer, The Netherlands), and FCS was from Integro (Zaandam, The Netherlands). Oregon green 488-labeled annexin V (Apoptest) was supplied by Nexins Research (Hoeven, The Netherlands), and rhodamine-labeled phalloidin was supplied by Molecular Probes (Leiden, The Netherlands). tBH was obtained from Aldrich (Milwaukee, WI). Other reagents, including 4',6-diamidino-2-phenylindole (DAPI) and L-buthionine-[S,R]-sulfoximine (BSO), were purchased from Sigma (St. Louis, MO); 1,3-bis(2-chloroethyl)-1-nitrosourea (BCNU) was a gift from Bristol-Myers Squibb (Evansville, IN).

Cell culturing. HUVEC were collected by trypsin digestion and cultured on gelatin-coated culture dishes in 1 vol medium 199 and 1 vol RPMI 1640, supplemented with 20% (vol/vol) human serum, bovine brain extract (5 µg/ml), and heparin (10 U/ml), as described previously (39). Repetitions of experiments were always performed with HUVEC derived from different umbilical cords (passage 4). ECRF24 cells, kindly provided by Dr. H. Pannekoek (Dept. of Biochemistry, Academic Medical Centre, Amsterdam, The Netherlands), were grown in HUVEC culture medium, but without bovine brain extract or heparin. ECV304 cells were obtained from the European Collection of Animal Cell Cultures (Salisbury, UK) and cultured in medium 199 plus 10% (vol/vol) FCS. All experiments were performed with endothelial cells in confluent monolayers.

Microscopic inspection and staining of tBH-treated cells. Confluent monolayers of endothelial cells grown in 12-well plates were treated with 0.1-1 mM tBH in serum-containing culture medium. Cells were monitored for membrane blebbing every 10 min by phase-contrast microscopy. All experiments were run in triplicate with cells in three different wells. Usually, three independent experiments were performed in which, for HUVEC, a different isolation was used for each experiment. The observer was blind to the experimental condition. For F-actin and nuclear staining, cells grown on gelatin-coated glass coverslips were washed with HEPES buffer (pH 7.45), fixed with 1% (wt/vol) paraformaldehyde for 10 min, washed with PBS, permeabilized with 0.02% (wt/vol) saponin, and finally stained with rhodamine-labeled phalloidin (2.5 U/ml) for 45 min in the dark. DAPI (62.5 µg/ml) was used for nuclear counterstaining. For quantification of actin rearrangements and nuclear condensation, two to three independent experiments were performed and for each experiment at least three microscopic fields were inspected. Confocal laser scanning microscopy was performed with a Bio-Rad MRC 600 system as described elsewhere (24).

Markers of apoptosis and necrosis. Endothelial cells were treated with tBH or staurosporine in serum-containing medium for 1 or 16 h in 12-well plates, after which the medium (possibly containing floating cells) was removed and stored. Cells were trypsinized, and then the corresponding medium was added again. Cells pelleted by centrifugation were resuspended in 500 µl HEPES buffer, pH 7.45, containing (in mM) 150 NaCl, 10 HEPES, 5 KCl, 1.8 CaCl₂, 1 MgCl₂,10 D-glucose, and 4 L-glutamic acid and 0.25% (wt/vol) BSA. The surface expression of negatively charged PS in these unfixed cells was determined by flow cytometry, by using as a probe Oregon green 488-labeled annexin V (0.25 µg/ml) (17, 21). The membrane-impermeable DNA stain propidium iodide (1 µg/ml) was included in these experiments to determine membrane integrity as a marker of cell vitality. Cells with a low level of propidium iodide staining were defined to be vital; cells with a high level of annexin V binding together with a low level of propidium iodide staining were defined as apoptotic. Finally, all cells with a high level of propidium iodide staining were defined as necrotic (36).

Determination of GSH and total glutathione levels. After incubation of the endothelial cells on six-well culture plates, cellular proteins were precipitated by adding directly to the monolayer 500 µl of an ice-cold solution of 4% (wt/vol) TCA and 2 mM EDTA. The culture plates were then incubated at 4°C for 15 min, after which cell lysates were collected and centrifuged at 9,000 g for 5 min (4°C). The resulting supernatants were used for the measurement of GSH and total glutathione levels, as previously described (37). Glutathione levels are expressed as percentages of the levels from control incubations. Unstimulated HUVEC and ECRF24 and ECV304 cells each contained ~120 nmol total glutathione/mg cellular protein.

Inhibition of glutathione reductase levels. Confluent monolayers of endothelial cells were incubated with specific glutathione reductase inhibitor BCNU (50 µM) (37) in culture medium. After 30 min the BCNU-containing medium was replaced by fresh culture medium and the cells were allowed to recover for 2 h, essentially as described previously (5). The BCNU-treated cells were then used for other experimentation.

Determination of NADPH in endothelial cell monolayers. Endothelial cells were cultured on gelatin-coated glass coverslips in 12-well culture plates. The coverslips were mounted in an open chamber that was placed on the stage of a Diaphot inverted fluorescence microscope (Nikon, Tokyo, Japan). Changes in autofluorescence, reflecting intracellular levels of NADPH (22a), in individual cells at 37°C were measured with a Quanticell fluorometric system (Applied Imaging, Sunderland, UK), basically following previously described procedures (17). Excitation and emission wavelengths were 340 and 510 nm, respectively.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Changes in actin cytoskeleton and cell morphology. The effects of tBH on confluent monolayers of endothelial cells in the continuous presence of serum-containing culture medium were investigated. HUVEC were treated with a low dose of 0.1 mM tBH, and a microscopic inspection showed that this resulted, within 30 min, in the development of plasma membrane blebs in, on average, 26% of the cells, although there was considerable variation between cells derived from different umbilical veins. After the cells were stained for F-actin to visualize peroxide-induced changes in the actin cytoskeleton, it appeared that, in the bleb-forming cells, the normal filamentous pattern of F-actin had been completely changed into a pattern with F-actin redistributed as a diffuse ring along the cellular border and blebs (Fig. 1). The interior of the cell body and the membrane blebs were practically devoid of F-actin. Counterstaining with DAPI indicated that most of the blebbing cells also had a condensed nucleus (Fig. 2). Quantitative analysis revealed that the number of cells with nuclear condensation was indeed similar to the number of bleb-forming cells (Table 1). When a higher dose of 0.4 mM tBH was used, a larger fraction of cells showed these morphological changes. The same effects on cellular structure were observed when HUVEC were incubated with hydrogen peroxide instead of tBH. For instance, 0.5 mM hydrogen peroxide induced membrane blebbing in 14.7 ± 4.6% of total cells (mean ± SE; n = 5). To study the dose dependency of this process, HUVEC were exposed to various concentrations of tBH in culture medium and visually checked for the appearance of membrane blebs. As shown in Fig. 3, the blebbing response was found to increase in the range of 0.1-1.5 mM tBH, with almost all cells showing blebs at the highest concentration. Typically, the number of peroxide-induced blebbing cells decreased considerably after longer incubation times with lower peroxide concentrations, and the majority of them regained normal morphology. For hydrogen peroxide stimulation (0.5 mM), 94.6 ± 3.8% (n = 4) of the blebs had disappeared after 2 h of incubation, whereas with tBH (0.1 mM) an incubation time of 4-8 h was needed to achieve this level of reduction. These results strongly suggest that the bleb-forming process is reversible.

View larger version (80K): [in this window] [in a new window]   Fig. 1.   Confocal laser scanning micrograph of stained F-actin in human umbilical vein endothelial cell (HUVEC) treated with tert-butyl hydroperoxide (tBH). HUVEC grown to confluence were incubated with 0.1 mM tBH at 37°C. After 60 min of treatment, cells were fixed and stained for F-actin with rhodamine-labeled phalloidin. A representative image of a blebbing cell is shown. Note that blebbing cell has rounded up to a higher plateau and is observed on top of surrounding cells. Bar, 8 µm.

View larger version (74K): [in this window] [in a new window]   Fig. 2.   Nuclear condensation in blebbing endothelial cells. Photographs are from 1 microscopic field observed for F-actin (A) or DNA (B) staining. HUVEC grown to confluence were incubated with 0.1 mM tBH for 60 min, fixed, and stained for F-actin with rhodamine-labeled phalloidin and then counterstained for DNA with 4',6-diamidino-2-phenylindole (DAPI). Note condensed DNA of blebbing cell. Bars, 11 µm.

View larger version (12K): [in this window] [in a new window]   Fig. 3.   Dose dependency of tBH-induced membrane blebbing in HUVEC. Confluent monolayers of HUVEC were incubated with 0.1-1.5 mM tBH in culture medium (37°C). Cells were inspected by microscopy for membrane blebbing for 10-60 min to determine maximal no. of bleb-forming cells. Data are means ± SE, n = 3.

View larger version (140K): [in this window] [in a new window]   Fig. 4.   Confocal laser scanning micrograph of stained F-actin in ECRF24 cells treated with tBH. Confluent ECRF24 cells were incubated 5 times with 0.1 mM tBH, with 10-min intervals between each incubation. After 60 min of treatment, cells were fixed and stained for F-actin with rhodamine-labeled phalloidin. Representative image of a blebbing cell is shown. Bar, 5 µm.

Bleb formation in the absence of apoptosis. The appearance of circular F-actin, membrane blebbing, and nuclear condensation in the peroxide-treated HUVEC and ECRF24 cells closely resembles the alterations in cell morphology in an early stage of apoptosis (26, 27, 40). We therefore investigated whether tBH was capable of initiating this process in these endothelial cells by measuring two characteristic apoptotic responses: appearance of PS at the plasma membrane outer surface as an early marker (21, 35) and fragmentation of DNA to mark more-advanced stages (31). For comparison, the cells were triggered with the protein kinase inhibitor staurosporine, which is a well-known inducer of apoptosis in endothelial cells (3).

View larger version (35K): [in this window] [in a new window]   Fig. 5.   Short-term effects of tBH on phosphatidylserine (PS) exposure and DNA fragmentation in endothelial cells. Confluent monolayers of HUVEC (left) or ECRF24 cells (right) remained untreated (A, B) or were incubated with tBH (0.1 mM; 1 treatment for HUVEC and 5 treatments for ECRF24 cells) (C, D) or 4 µM staurosporine (E, F). After 60 min of incubation, cells were harvested and were either stained with Oregon green (OG) 488-labeled annexin V or, alternatively, fixed and stained with propidium iodide. Fluorescence of collected cells was analyzed with a flow cytometer. Lower channel nos. represent autofluorescence of unlabeled cells (dotted lines); nos. in italics are percentages of annexin V-positive cells. Results are from typical experiment representative of 4-6 others.

View larger version (36K): [in this window] [in a new window]   Fig. 6.   Long-term effects of tBH on PS exposure and DNA fragmentation in endothelial cells. HUVEC (left) or ECRF24 cells (right) remained untreated (A, B) or were treated with tBH (C, D) or staurosporine (E, F) during a period of 16 h. Further conditions and analysis are as described for Fig. 5. Numbers in italics are percentages of annexin V-positive cells. Results are representative of those from 4-6 experiments.

Effects on glutathione oxidation. To determine how the cell line-dependent variation in membrane blebbing was related to the redox or sulfhydryl state of the cells, we compared the effects of tBH on the intracellular glutathione levels in HUVEC and ECRF24 and ECV304 cells. For HUVEC, a single incubation with 0.1 mM tBH immediately decreased the level of GSH to ~30% of the resting value, after which it was slowly restored. After 60 min, a considerable part of the oxidized glutathione still had not been regenerated as GSH (Table 3). Higher tBH concentrations resulted in more severe GSH oxidation (data not shown). When ECRF24 cells were used, treatment with 0.1 mM tBH also resulted in a pronounced decrease in GSH level but the decrease was followed by a more rapid recovery, so that all glutathione had regained the reduced form within 60 min (Table 3). The repeated addition of 0.1 mM tBH to ECRF24 cells considerably delayed this recovery process. At 60 min after five consecutive tBH incubations, the GSH level still reached only 52.9 ± 11.8% (mean ± SE; n = 4) of the total glutathione concentration. These data suggest that HUVEC differ from ECRF24 cells, not so much in their GSH peroxidase reactivities (i.e., the enzymatic reaction inactivating peroxides at the expense of GSH) but more in their glutathione reductase capacities (by which oxidized glutathione is reformed to GSH). It should be noted that, in both HUVEC and ECRF24 cells, the total glutathione concentration decreased gradually during the incubation period (Table 3), possibly because of the efflux of oxidized glutathione out of the cells, as has been reported for hepatocytes (8).

Relation between glutathione oxidation and bleb formation. To further investigate the role of GSH oxidation in tBH-induced membrane blebbing, cells were pretreated with the specific glutathione reductase inhibitor BCNU (8, 37). When applied to monolayers of HUVEC or the cell lines, this treatment was without effect on the total glutathione concentration. For HUVEC and ECRF24 cells, the subsequent addition of tBH caused a considerable reduction in the GSH redox state, as expected. At the same time, the BCNU treatment more than doubled the peroxide-induced blebbing response (Table 4). However, in ECV304 cells treated with BCNU, tBH still did not lead to notable GSH oxidation and membrane blebbing remained absent. Significant GSH oxidation and bleb formation were also not detected in ECV304 cells that were pretreated with the glutathione synthesis inhibitor BSO, although the level of total glutathione was considerably reduced in the cells (see above). In another series of experiments, HUVEC and other cells were treated with the membrane-permeable sulfhydryl agent N-ethylmaleimide (30 µM), an agent that oxidizes GSH and protein thiols in endothelial cells (29). When applied to HUVEC, this treatment resulted in massive F-actin rearrangement and in membrane blebbing in 10.0 ± 1.0% of the cells (mean ± SE; n = 3). Similar blebbing morphology was observed when these cells were incubated with another sulfhydryl reagent, diamide.

View larger version (21K): [in this window] [in a new window]   Fig. 7.   NADPH-oxidizing effect of tBH on endothelial cells in monolayer. HUVEC on glass coverslips were incubated with 0.1 mM tBH at 37°C. NADPH autofluorescence was determined by fluorescence microscopic imaging at excitation and emission wavelengths of 340 and 510 nm, respectively. Shown are traces of autofluorescence of 2 individual cells within 1 microscopic field. Later visual inspection showed blebbing in ~15% of cells in this particular experiment. Data are representative of those from 4 independent measurements.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

In this paper, we contrast the responses of HUVEC and two HUVEC-derived cell lines (ECRF24 and ECV304) to the glutathione-oxidizing compounds tBH and hydrogen peroxide with regard to membrane blebbing, actin cytoskeleton rearrangement, nuclear condensation, and GSH oxidation. For HUVEC, the peroxides increased in a dose-dependent way the number of cells undergoing bleb formation and the other morphological changes. As far as we could detect, many or all of the blebbing cells contained an actin cytoskeleton located on the surface and also a condensed nucleus (Table 1, Fig. 2). The time scale of these changes was quite similar to that of the transient, peroxide-induced oxidation of GSH (Table 3). Compared with HUVEC, the cell lines, ECV304 more so than ECRF24, were more resistant to both the morphological changes and GSH oxidation. Further confirmation of this relation between peroxide-induced glutathione-oxidizing capacity and membrane blebbing could be obtained by treating the cells with the glutathione reductase inhibitor BCNU, which increased the bleb formation and the degree of GSH oxidation in HUVEC and ECRF24 cells but was unable to induce either response in ECV304 cells (Table 4).

Although the oxidative stress imposed by tBH and hydrogen peroxide may evoke apoptosis in various cells (4, 14, 32), the morphological changes observed in the present study are unlikely to result from an apoptotic process, because they were not accompanied by early markers of apoptosis such as surface expression of PS or fragmentation of DNA. This holds for HUVEC as well as for ECRF24 cells (Figs. 5 and 6). More precisely, at lower peroxide concentrations, the number of blebbing cells was found to decrease in time (suggesting reversibility of the process), without indications of apoptosis (Table 2). On the other hand, at higher peroxide doses, at which many of the cells showed blebs, a considerable fraction of the cells became necrotic but not apoptotic. Apparently, depending on the peroxide concentration, blebbing endothelial cells either recover or die a nonapoptotic death. We therefore conclude that, under the present experimental conditions, 1) the peroxide-induced morphological changes correlate with the degree of GSH oxidation but are not early indications of apoptosis and 2) HUVEC and the HUVEC-derived cell lines are fairly resistant to apoptosis.

There is only limited literature directly comparing the properties of ECRF24 and ECV304 cells with those of the parental cells. ECRF24 was produced by transfection of HUVEC with the E6/E7 open reading frame of human papilloma virus 16 (10), whereas ECV304 is the result of a spontaneous HUVEC transformation (34). The ECRF24 line closely resembles HUVEC with respect to cell morphology, growth characteristics, and expression of von Willebrand factor and selective surface adhesion molecules (10). Similarly, the ECV304 line possesses many HUVEC-like characteristics in terms of morphology and signal transduction mechanisms, although a certain degree of dedifferentiation may have occurred (1, 11, 18). From the present results, it is likely that these cell lines differ from HUVEC in their capabilities to detoxify peroxides by using the redox equivalents from intracellular GSH. We hypothesize that this difference is responsible for the decreased blebbing response. However, the possibility that the cell lines lack one or more of the other intracellular signaling modules that are unrelated to glutathione metabolism and that participate in peroxide-induced cytoskeleton rearrangement and the blebbing response cannot be excluded. Indeed, a wide variety of peroxide-induced signaling pathways that may or may not be influenced by the glutathione state of the cells have been described (see INTRODUCTION). It is thus still possible that the processes of bleb formation and GSH oxidation are epiphenomena without a causal relationship. This might also explain why not all cells start to bleb in response to low peroxide concentrations, whereas most (or all) of them will undergo a change in their GSH levels. However, it is also possible that both the degree and time period of GSH oxidation may determine the blebbing phenomenon. In agreement with this is the observation that within a single endothelial monolayer, considerable cell-to-cell heterogeneity in the degree of peroxide-induced NADPH oxidation was observed (Fig. 7). Because the peroxide-induced oxidation of GSH and NADPH points to a changed intracellular redox state, it might well be that such an altered state, rather than a direct effect of oxidized GSH, is affecting the cell morphology.

The present results differ from those of others, who report that mild oxidative stress such as that evoked by peroxide triggers the apoptotic process in endothelial cells (7, 22, 32). It should be noted that our experiments were performed with cells that were treated with tBH or hydrogen peroxide in the continuous presence of serum with growth factors; most of the other reported studies have been performed with cells activated in serum (albumin)-free media. There are good indications that serum deprivation induces apoptosis in endothelial cells (41), whereas growth factor supplementation protects the cells from entering this process (13). At the molecular level, it has been proposed that the antiapoptotic effect of serum (albumin) and growth factors results from a cross talk between stress-stimulated protein kinase pathways (proapoptotic) and parallel growth factor-activated survival pathways (antiapoptotic) (2, 20). In this model, the final effect of stress activators such as peroxide would be influenced by the activity of the survival pathways. Thus peroxide application to endothelial cells that are maintained on serum might cause a mild stress response such as membrane blebbing (possibly by a changed GSH oxidation state) without trapping the cells into an apoptotic program.

Although the endothelial cells in the inner walls of blood vessels can easily become exposed to peroxides under stress conditions, such as those produced by activated neutrophils (25), the present data suggest that apoptosis is not a likely result under such in vivo conditions. On the other hand, the peroxide-evoked bleb formation may seriously impede the interface function of endothelial cells in the vessel wall, because it is necessarily accompanied by cell retraction and thus by the increased permeability of the endothelium.