Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Tumor necrosis factor (TNF)-, originally described for its antitumoral activity, is a polypeptide that exerts a pleiotropic action on multiple cell functions regulating the immune response, host defense reactions, and gene expression. The importance of the effects of TNF- has been clearly shown in conditions such as inflammation and sepsis, in which endothelial cells are the first target of this cytokine (1, 2). An increase in cellular oxidative stress is involved in the cytotoxic effects of TNF- on tumor cells (3) and in the regulation of gene expression (4). In tumor cells, hepatocytes or cardiomyocytes, reactive oxygen species (ROS) production induced by TNF- was accompanied by a cytotoxic or pro-apoptotic effect (3, 5). In these studies, mitochondria have been identified as a major source of TNF--induced ROS production (6). In endothelial cells, increased ROS production may interfere with crucial endothelial functions such as nitric oxide and cyclooxygenase pathways (7). Most studies have evaluated the effect of long exposure to TNF- (i.e., several hours), whereas several endothelium-mediated vascular effects of TNF-, as well as the activation of TNF- intracellular transduction pathways, have been shown to occur rapidly (3, 6). However, the specific involvement of ROS production in the early effects of TNF- on the endothelial cell has not been reported.

The cellular pathways activated by TNF- vary (8, 9). There is evidence to support the role of ceramide as a second messenger of TNF- in different cell types, including endothelial cells (10, 11), where it participates in the regulation of cell death via apoptosis, cell proliferation, and gene regulation through the activation of nuclear transcription factors such as nuclear factor-B (4). Exogenous cell-permeable ceramide analogs have been recently demonstrated to act on the mitochondrial electron transport chain leading to hydrogen peroxide and ROS production in rat liver hepatocytes (5). However, the specific involvement of ceramide in TNF--induced ROS production in endothelial cells is not known.

The aim of the present study was to assess the effect of TNF- on mitochondrial oxidative stress in endothelial cells and to identify the pathways involved in signal transduction. For this purpose, we first investigated the time course of ROS production in human umbilical vein endothelial cells (HUVEC) using the fluorescent probe dichlorodihydrofluorescein diacetate (DCFH-DA) after a short exposure to TNF-. Second, we aimed at identifying the site of ROS production in the absence or in the presence of different inhibitors of the mitochondrial respiratory chain. To investigate the potential contribution of other ROS-generating pathways such as the membrane-bound nicotinamide adenine dinucleotide reduced (NADH) nicotinamide adenine dinucleotide phosphate (NADPH)-dependent oxidase, xanthine oxidase (XO), and nitric oxide (NO) and cyclooxygenase pathways, we also studied the effects of TNF- in the presence of specific inhibitors of these pathways.

We finally studied the role of ceramide in intracellular signaling pathways of TNF--induced mitochondrial ROS production in endothelial cells. The impact of ceramide was first explored by means of the acidic sphingomyelinase (ASMase) inhibitor desipramine. Subsequently, the mechanism of ceramide-activated intracellular signaling pathways was examined using the ceramide-activated protein kinase (CAPK) inhibitor dimethylaminopurine (DMAP) (10).

We provide evidence that TNF--induced ROS production in endothelial cells primarily occurs in mitochondria and is supported by an intracellular signaling pathway initiated by ceramide.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Cells and Culture

HUVEC were obtained from PromoCell Laboratories (Heidelberg, Netherlands) as cryopreserved cells. After thawing, cells were plated (5,000 to 10,000 cells/cm²) in culture flasks precoated with rat-tail collagen I (5 µg/cm²; Boehringer Mannheim, Mannheim, Germany) and cultured to confluence in MCDB 131 medium containing: Hepes (28 mM), fetal calf serum (2%), human recombinant epidermal growth factor (0.1 ng/ml), human recombinant basic fibroblast growth factor (1.0 ng/ml), synthetic hydrocortisone (1.0 µg/ml), gentamycin (50 µg/ml), amphotericin B (50 ng/ml), and supplemented with Supplement Mix (PromoCell) containing endothelial cell growth factor and heparin.

A cell suspension in nutritive medium was plated onto 25-mm-diameter plastic Thermanox coverslips (Poly Labo, Strasbourg, France) to obtain a seeding density of approximately 5 × 10³ cells/ cm². Medium was replaced every 48 h and cells were maintained in a humidified air/5% CO₂ atmosphere at 37°C until confluence was reached. All cells were studied at the third passage. To avoid variability related to differences between cultures, replicate experiments were carried out using cells from separate cultures.

Perfusion System

Endothelial cell monolayers on plastic coverslips were placed in a stainless steel flow-through chamber (1 ml volume; Penn Century Co., Philadelphia, PA). The chamber was sealed using thin wafer gaskets and mounted in a heated (37°C) water serpentine on an inverted microscope (Leica SA, Rueil-Malmaison, France). A water-jacketed glass equilibration column (37°C) (Radnotti Glass Technology Inc., Monrovia, CA), mounted above the microscope stage, was used to equilibrate the perfusate with 25% O₂/5% CO₂/70% N₂ air mixture. The perfusate consisted of Krebs-Henseleit bicarbonate buffer (pH 7.4, 37°C) containing in mM: NaCl 117.3, KCl 4.7, NaHCO₃ 25, MgSO₄ 1.3, KH₂PO₃ 1.2, CaCl₂ 1.23, lactate 1, and glucose 11.1. Gas tension in the chamber was as follows: PO₂, 151.2 mm Hg; PCO₂, 38.6 mm Hg; and pH of the perfusate was 7.38. The perfusion system was a nonrecirculating system with a constant flow rate of 1.3 ml/min.

Fluorescence and Light Microscopy

An inverted microscope (model DM-IRB; Leica) was equipped for epifluorescent illumination and included a mercury light source (50 W), a 12-bit digital cooled camera (model RTE/CCD-1317-K/1; Princeton Instruments Inc., Trenton, NJ), and a shutter (model D-122; Uniblitz, Rochester, NY) under computer control (Metamorph Imaging System 3.5; Universal Imaging Corporation, West Chester, PA), with appropriate excitation and emission cubes. Fluorescent cell images were obtained using a 20× objective for fluorescence (Leica). Data were acquired and analyzed using Metamorph software (Universal Imaging Corp.).

Measurement of ROS and Assessment of Cell Death

ROS generation in cells was assessed using the probe 2,7-dichlorofluorescein (DCF) (Molecular Probes Europe BV, Leiden, The Netherlands). The membrane-permeable diacetate form of the dye (reduced DCF: DCFH-DA) was added to the perfusate at a final concentration of 5 µM. Within the cell, esterases cleave the acetate groups on DCFH-DA, thus trapping the reduced form of the probe (DCFH) intracellularly. ROS in the cells oxidize DCFH, yielding the fluorescent product DCF (12). After 1 h, fluorescence was measured using an excitation wavelength of 480 ± 20 nm, a dichroic 505-nm long pass, and an emitter bandpass of 527 ± 15 nm. Previous studies of the behavior of DCFH revealed that the probe is readily oxidized by H₂O₂ or hydroxyl radical but is relatively insensitive to superoxide (13, 14).

To evaluate cell viability, we used the fluorescent probe propidium iodide (PI) (Molecular Probes Europe BV), which was added to the perfusate at a final concentration of 5 µM. This probe is a red fluorescent, cell-impermeant dye that is widely used to detect dead or dying cells and only penetrates in the cell when the membrane is injured, binding to the DNA. Once this dye is bound to nucleic acids, its fluorescence is enhanced 20- to 30-fold, its excitation maximum is shifted ~ 30 to 40 nm to the red, and its emission maximum is shifted ~ 15 nm to the blue. After 1 h, fluorescence was measured using an excitation wavelength of 515 to 560 nm, a dichroic 580-nm long pass, and an emitter bandpass of 590 nm.

Fluorescence intensity was assessed by the imaging system Metamorph 3.5 on a region of interest that included the whole observational field. Intensity values are reported as percentage changes of initial values.

Experimental Protocol

Cells plated onto plastic coverslips were placed in the perfusion chamber and flow was started. After 1 h of stabilization, cells were either exposed to a control perfusion or to TNF-. Four groups of at least five monolayers each were studied: three groups were exposed to different doses of TNF- in the Krebs-Henseleit perfusate (0.1, 1, or 10 ng/ml, respectively) for 1 h. The control group was exposed to the same perfusate containing the diluent for TNF-, phosphate-buffered saline (PBS), with 1% bovine serum albumin (BSA), prepared as for the TNF- experiments. After 1 h exposure to control or TNF-, perfusion was returned to a Krebs-Henseleit buffer without any additives and the observation was conducted for another hour. During the 2 h of the experiment, one fluorescent image was acquired every 10 min and analyzed to assess ROS production. To exclude that ROS production was due to nonspecific stimulation, three control experiments were carried out using TNF- (1 ng/ml) in association with neutralizing anti-TNF- antibodies; the absence of effect of anti-TNF- antibodies alone was also checked.

To identify the site of intracellular production of ROS in response to TNF-, additional experiments were performed using the same design, with 1 ng/ml of TNF-, because this dose did not alter cell viability in preliminary experiments.

Within the mitochondria, ROS production occurs primarily at two sites: complex I (NADH dehydrogenase) and complex III. To identify the site of ROS production in mitochondria, we used inhibitors of mitochondrial electron transport. Cells were exposed to 1 ng/ml of TNF- in the presence of both rotenone (10 µM) + thenoyltrifluoroacetone (TTFA) (10 µM). Rotenone is an inhibitor of complex I (NADH dehydrogenase), inhibiting electron entry from complex I to ubiquinone. The electron transport from complex II (succinate dehydrogenase) to ubiquinone is blocked by the addition of TTFA. In another set of experiments, antimycin A (AA), which inhibits the electron flow at complex III, was added at the concentration of 10 µM and perfused 20 min after TNF- to characterize the precise role of mitochondrial complex III in ROS generation once TNF--induced production had started.

To study the contribution of the membrane-bound NADPH-dependent oxidase, we ran a set of experiments in the presence of the NADPH oxidase inhibitor diphenylene iodonium (DPI) (10 µM) (15). To test the contribution of XO to ROS production, experiments were conducted with a Krebs-Henseleit buffer supplemented with the XO inhibitor allopurinol (10 µM) (16).

Finally, the contribution of both cyclooxygenase and endothelial nitric oxide synthase (NOS) to ROS generation was explored by means of simultaneous administration of the cyclooxygenase competitive inhibitor mefenamic acid (20 µM) (17) and the inhibitor of the L-arginine pathway N-nitro-L-arginine (L-NNA) (1 mM) (18). All the previously mentioned inhibitors were administered together with TNF- after uptake of DCFH and equilibration of fluorescence, except for AA, as mentioned previously.

TNF- signal transduction pathways involve sphingomyelin hydrolysis to produce ceramide as a lipid second messenger in many cell types (19). We therefore explored the potential involvement of the ceramide-activated intracellular pathway in TNF--induced ROS production in HUVEC. For this purpose, we studied the effect of 1 ng/ml TNF-, as described previously, in two different conditions. In one group, cells were incubated in the presence of the ASMase inhibitor desipramine (5 µM) for 2 h before the beginning of the experiment (11). In another group, cells were studied in the presence of the CAPK inhibitor DMAP (1 mM). DMAP administration began 1 h before TNF- administration and was maintained throughout the whole experiment. The concentration of DMAP was chosen according to the work of Marino and coworkers (10) in bovine aorta endothelial cells.

Presentation of the Results and Statistical Analysis

Differences among groups were analyzed using a one-way or two-way analysis of variance (within factor, time; between factor, treatment group) (20). Statistical significance was determined at the 0.05 level. Values are reported as mean ± standard error of the mean (SEM).

Reagents

All products, reagents, and inhibitors were purchased from Sigma Chemical Co. (St. Louis, MO). They are detailed in the previous paragraph. Recombinant human TNF- (bioactivity: EC₅₀ = 0.01 to 0.1 ng/ml using L929 cells for inhibition) and antihuman TNF- antibodies (bioactivity: 0.4 µg/ml concentration of antibody neutralizes 50% of the cytotoxicity of 1 U of recombinant human TNF- in the mouse cell line A-9) were purchased from Sigma. Lyophilized powder of TNF- was dissolved in 0.2-µm-filtered PBS + 1% BSA. To avoid repeated freezing and thawing, working aliquots were prepared and stored at 20°C.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

TNF- Induces a Rapid Production of ROS in HUVEC

After 1 h of stabilization, endothelial cell monolayers were exposed to either 0.1, 1, or 10 ng/ml of TNF- or to a control perfusion containing the vehicle of the TNF-, for 1 h. As shown in Figure 1, a significant dose effect of TNF- was observed (P < 0.01). At 0.1 ng/ml of TNF-, DCF fluorescence did not differ from control values, whereas at 1 ng/ml TNF-, it increased within 1 h after beginning the perfusion, indicating a rise in ROS production by the cells. At 10 ng/ml of TNF-, a dramatic and more rapid increase in DCF fluorescence was observed; however, the results were more variable and PI fluorescence increased, indicating a loss of cell viability. Anti-TNF- immunoglobulin G antibodies had no effect per se (data not shown) but they totally abolished the increase in DCF fluorescence induced by 1 ng/ml TNF- (P < 0.01) (Figure 1). PI fluorescence did not vary in the 1 ng/ml-treated group (4.9 ± 1.7%), whereas it increased dramatically in those monolayers exposed to 10 ng/ml TNF- (70.5 ± 19.5%; n = 3, P < 0.01).

View larger version (20K): [in this window] [in a new window]   Figure 1.   Effects of TNF- on DCF fluorescence in HUVEC. Data are presented as percentage of the initial value of DCF fluorescence after exposure to three different doses of TNF- (0.1, 1, or 10 ng/ml), control perfusion (Krebs buffer), or 1 ng/ml TNF- + anti-TNF- antibodies for 1 h. TNF- induces a rapid production of ROS in HUVEC. After 60 min of stabilization, images were taken every 10 min for two consecutive hours. A significant dose-dependent effect was observed (n = 5 experiments for each group, except for the dose 1 ng/ml, n = 10, P < 0.01; n = 3 for anti-TNF- antibodies, P < 0.01). *Cell death occurred after this time for the highest dose of TNF-.

TNF--Induced ROS Production Is Located within the Mitochondrial Electron Transport Chain

To determine the source of endothelial ROS generation in response to TNF-, the mitochondrial respiratory chain inhibitors rotenone and TTFA were used to block the electron flow between mitochondrial complexes I and III, and between mitochondrial complexes II and III, respectively. Their site of action is schematically represented in Figure 2A. When rotenone (10 µM) and TTFA (10 µM) were added together with TNF-, TNF--induced DCFH oxidation was significantly inhibited (P < 0.05) (Figure 2B). Thus, the inhibition of both potential electron entry sites to ubiquinone with rotenone + TTFA resulted in an inhibition of mitochondria respiratory chain ROS signaling. In a separate set of experiments, AA was used to block electron flow at complex III; a very rapid increase in DCFH oxidation was observed (118% in 3 min; n = 4, P < 0.01) (Figure 2C). Thus, inhibition of complex III, which blocks the oxidation of ubisemiquinone to cytochrome c1, potentiated ROS production at the semiquinone site.

View larger version (11K): [in this window] [in a new window]   View larger version (15K): [in this window] [in a new window]   View larger version (11K): [in this window] [in a new window]   Figure 2.   Mitochondrial electron transport and ROS production in endothelial cells. (A) Schematic representation of the mitochondrial electron transport showing the site of ROS generation and the action of the pharmacologic inhibitors used. (B) Effects of the mitochondrial electron transport inhibitors (rotenone 10 µM + TTFA 10 µM) on DCF fluorescence in HUVEC exposed to 1 ng/ml TNF- compared with 1 ng/ml TNF- alone (n = 7). Both inhibitors and TNF- were administered simultaneously and perfused for 1 h. Data are presented as mean ± SEM. TNF--induced increase in DCFH oxidation was significantly attenuated by these inhibitors (P < 0.05), suggesting that ROS generation from mitochondria in endothelial cells is distal to complex I (see DISCUSSION for details). (C) AA (10 µM), which blocks mitochondrial electron transport downstream from the ubiquinone site, increased DCF fluorescence, suggesting that ROS generation occurs at this site. Data were acquired every minute after AA addition, because cell death occurred rapidly (n = 4, P < 0.01). *Cell death after this time.

We then explored several other potential sources of ROS generation in endothelial cells. The NADPH-dependent oxidase inhibitor DPI did not significantly inhibit the rise in ROS production induced by 1 ng/ml TNF-, nor did the XO inhibitor allopurinol or the NO and cyclooxygenase pathway inhibitors L-NNA and mefenamic acid. Data obtained after 1 h exposure to TNF- in the presence of these inhibitors are shown in Figure 3 and compared with the inhibition observed in the presence of rotenone + TTFA at the same time.

View larger version (16K): [in this window] [in a new window]   Figure 3.   Effects of the NAD(P)H oxidase inhibitor DPI (10 µM, n = 4), the XO inhibitor allopurinol (10 µM, n = 6), the NOS pathway inhibitor L-NNA (1 mM), and the cyclo-oxygenase inhibitor mefenamic acid (20 µM, n = 4) on DCF fluorescence in HUVEC exposed to TNF- (1 ng/ml). Data are expressed as mean ± SEM at 60 min and compared with values obtained with either TNF- (1 ng/ml) alone or rotenone/TTFA + TNF-. Except for rotenone/TTFA, none of these inhibitors prevented the increase in DCF fluorescence induced by TNF-.

Role of Ceramide Signaling in TNF--Induced ROS Production

As shown in Figure 4, a 2-h preincubation of HUVEC with the ASMase inhibitor desipramine (5 µM) inhibited ROS production upon exposure to TNF- in endothelial cells by ~ 80% (P < 0.005). This suggests that ceramide generated by ASMase accounts for TNF--induced mitochondrial DCFH oxidation.

View larger version (20K): [in this window] [in a new window]   Figure 4.   Effects of desipramine on DCF fluorescence in HUVEC exposed to TNF- (1 ng/ml). Cells were incubated in the presence of the ASMase inhibitor desipramine (5 µM; solid circles) for 2 h before the beginning of exposure to TNF-. DCF fluorescence in HUVEC treated with desipramine decreased compared with untreated cells (mean ± SEM; n = 4, P < 0.0001). Solid squares, cells incubated without desipramine.

Continuous administration of DMAP (1 mM), a CAPK inhibitor, almost totally inhibited TNF--induced ROS production in HUVEC (P < 0.01; Figure 5), thus suggesting that TNF--induced mitochondrial ROS generation requires protein kinase activity associated with ceramide.

View larger version (20K): [in this window] [in a new window]   Figure 5.   Effects of the CAPK inhibitor DMAP (1 mM) on DCF fluorescence in HUVEC exposed to TNF- (1 ng/ml). When DMAP was administered 1 h before TNF- and kept throughout the whole experiment, DCF fluorescence decreased (solid circles) (mean ± SEM; n = 6, P < 0.01). In the experiment without DMAP (solid squares), DCF fluorescence increased.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

The present study demonstrates that TNF- induces a rapid increase in mitochondrial ROS production in human endothelial cells. The major source of ROS signaling appears to be the mitochondrial electron transport system. The increase in ROS signaling is mediated by ceramide, based on the observation that blocking either ceramide production or CAPKs inhibited intracellular ROS production.

Using fluorescence imaging techniques, we reported here a dose-dependent, TNF--induced ROS production. At an intermediate dose of TNF- (1 ng/ml), a detectable ROS production was found within the first hour of exposure, without any evidence of cytotoxicity. This dose of TNF- is consistent with TNF- concentrations previously used on endothelial cells in vitro (21) and with the range of TNF- levels detected in patients during bacteremia and septic shock (22, 23). In the present study, the delay in detecting ROS production in endothelial cells was shorter than that reported by Goosens and colleagues (3) in tumor cells. These investigators described an increase in ROS production at least 120 min after exposure to TNF-, but not within the first hour. We observed that at 10 ng/ml TNF-, ROS production increased faster than at 1 ng/ml, but this was followed by the appearance of PI fluorescence, indicating irreversible membrane injury. These results are consistent with previous studies, suggesting that oxidative stress mediates TNF- cytotoxicity in hepatocytes (5, 24) or fibrosarcoma cells (3, 6, 25).

The involvement of the mitochondrial electron transport in TNF--induced ROS production has been demonstrated in several cell types (3, 6, 25) and in isolated mitochondria from hepatocytes (26), but not in endothelial cells. In hepatocytes, two sites of the respiratory chain produce ROS: one is dependent on the auto-oxidation of the flavin mononucleotide (FMN) from the NADH-dehydrogenase (complex I), whereas the other, probably the most important, depends on the auto-oxidation of the unstable ubisemiquinone (complex III), which is an intermediate of the Q-cycle reaction (27, 28). In our study, a potential ROS generation from complex I would have been increased by rotenone because rotenone inhibits electron transfer distal to the FMN. Yet, rotenone + TTFA attenuated DCFH oxidation, indicating that complex I is not the primary site of ROS generation during TNF- stimulation. By inhibiting the oxidation of ubisemiquinone in the Q cycle, AA increases the lifetime of the semiquinone (29). AA augmented DCFH oxidation, suggesting that complex III was the site of ROS generation in endothelial cells. This result is consistent with those obtained in mouse fibrosarcoma cells (3) and in liver mitochondria (5), showing that either TNF-- or ceramide-induced cytotoxicity was decreased by the inhibition of the mitochondrial electron transport at complex I and complex II and potentiated by AA.

Potential alternative sources of endothelial ROS production include NAD(P)H-dependent oxidases (30), XO (16), cyclooxygenases (17), and NOS (31). The NAD(P)H oxidase, originally characterized in neutrophils, is also present in endothelial cells (30) and is the principal source of superoxide production in some animal and human models of vascular disease (32). Interestingly, a recent study has reported two alternative ROS production pathways in endothelial cells: a mitochondrial pathway that is suppressed by rotenone and appears to be directly involved in TNF--induced apoptosis (35) and a membrane-dependent pathway that is associated with the NAD(P)H oxidase, regulated by the guanosine triphosphatase Rac-1, and protects against TNF--induced cell death (35).

In contrast to what we obtained with mitochondrial inhibitors, the NAD(P)H oxidase inhibitor DPI failed to inhibit significantly ROS generation in endothelial cells. However, it should be noted that a trend to decrease ROS production was observed with DPI (Figure 3). This could possibly be due to the known inhibitory effect of DPI on flavoproteins, including complex I (36). Nevertheless, the possibility of a small contribution of the membrane-bound NADPH oxidase cannot be formally excluded.

The role of XO in ROS production by endothelial cells has been reported in ischemia-reperfusion injury (16). We specifically assessed whether the inhibition of XO reduced TNF--induced ROS production. The absence of any inhibitory effect of allopurinol, contrasting with the inhibition of TNF-induced ROS production by the inhibitors of the mitochondrial respiratory chain rotenone and TTFA, allowed us to exclude the role of XO in the observed phenomenon.

Finally, we studied the role of both cyclooxygenase and NOS pathways by means of the competitive cyclooxygenase inhibitor mefenamic acid (17) and the NOS pathway inhibitor L-NNA (18). We chose a combined pharmacologic approach because cyclooxygenase and NOS share many functional similarities in endothelial cells exposed to inflammatory stimuli, and both enzymes are potential sources of ROS (7). In conjunction, these data show that mitochondria are the main site of ROS production in endothelial cells.

The mechanisms that transduce binding of TNF- to membrane receptors into ROS generation in endothelial cells are not yet fully characterized. In hepatocytes, ceramide and membrane-permeant ceramide analogues, such as C2-ceramide, have been shown to exert a direct effect on mitochondrial electron transport, leading to an increase in ROS generation in DCF-stained rat liver mitochondria (5, 26). This led us to test whether the increase in ROS induced by TNF- in HUVEC also occurs via the ceramide-dependent signaling pathway.

Previous work in human leukemic T Jurkat cells, in U397 monocytes, and in a mouse pre-B cell line has shown that binding of TNF- to its 55-kD receptor rapidly activates two distinct types of sphingomyelinase, the membrane-bound neutral sphingomyelinase (NSMase) and the endosomal ASMase, which are present in independent subcellular sites (37), each coupled to a specific pathway of intracellular TNF- signaling. Our findings demonstrate the predominant role of ASMase in ceramide generation in human endothelial cells because mitochondrial ROS generation driven by TNF- is inhibited by the ASMase inhibitor desipramine. This is in line with the observation of Schutze and coworkers (9), who demonstrated that in Jurkat T lymphocytes, binding of TNF- to its receptor is followed by a rapid activation of an ASMase, with subsequent sphingomyelin hydrolysis and ceramide production within 3 min (4). Furthermore, Jensen and associates (40) have stressed that precocious ASMase activation, with subsequent ceramide generation, plays a pivotal role in the early phase of repair of the cutaneous permeability barrier, as demonstrated by the observation that skin repair is delayed when ASMase is inhibited by desipramine and imipramine.

It should be noted that ceramide accumulation has been reported to occur without any decrease in sphingomyelin content in cerebral endothelial cells after long-term exposure to TNF- + cycloheximide (CHX) (11). This mechanism of ceramide generation has been attributed to de novo ceramide biosynthesis via ceramide synthase rather than to the activation of sphingomyelinase pathways, which are normally responsive to TNF-, Fas, and -irradiation (41). One may speculate that a de novo synthesis pathway is likely to be involved in long-term effects, whereas short-term exposure to TNF induces acute ceramide generation by ASMases.

It has been proposed that CAPK activation depends on the membrane-bound NSMase rather than on the ASMase pathway (38, 39). In contrast, we show that ceramide, likely generated by the ASMase pathway, signals to mitochondria through CAPK. In fact, TNF--induced ROS production was inhibited by the CAPK inhibitor DMAP (Figure 5). This is in accordance with previous observations by Marino and coworkers (10), who showed that administration of DMAP inhibits TNF--induced phosphorylation and activation of CAPK and Jun amino-terminal kinase, and attenuates the cytotoxicity promoted by TNF/ CHX in bovine endothelial cells.

In conclusion, we demonstrated that TNF- induces a rapid mitochondrial ROS production in human endothelial cells at the ubisemiquinone site. This effect occurs via a ceramide-dependent signaling pathway initiated by ASMase and is abrogated by either pretreatment of HUVEC with the ASMase inhibitor desipramine or treatment with the CAPK inhibitor DMAP (Figure 6). However, the specific mechanism by which ceramide signaling affects the increase in ROS generation in endothelial cells remains to be elucidated.

View larger version (15K): [in this window] [in a new window]   Figure 6.   Proposed sequence of events initiated by TNF- binding to the membrane and putative signaling pathways involved in ROS generation.