Ceramide Induces Cytochrome c Release from Isolated Mitochondria

In the present study we show thatN-acetylsphingosine (C2-ceramide),N-hexanoylsphingosine (C6-ceramide), and, to a much lesser extent, C2-dihydroceramide induce cytochromec (cyto c) release from isolated rat liver mitochondria. Ceramide-induced cyto c release is prevented by preincubation of mitochondria with a low concentration (40 nm) of Bcl-2. The release takes place when cytoc is oxidized but not when it is reduced. Upon cytoc loss, mitochondrial oxygen consumption, mitochondrial transmembrane potential (ΔΨ), and Ca2+ retention are diminished. Incubation with Bcl-2 prevents, and addition of cytoc reverses the alteration of these mitochondrial functions. In ATP-energized mitochondria, ceramides do not alter ΔΨ, neither when cyto c is oxidized nor when it is reduced, ruling out a nonspecific disturbance by ceramides of mitochondrial membrane integrity. Furthermore, ceramides decrease the reducibility of cytoc. We conclude that the apoptogenic properties of ceramides are in part mediated via their interaction with mitochondrial cytoc followed by its release and that the redox state of cytoc influences its detachment by ceramide from the inner mitochondrial membrane.

Recently, the importance of ceramide in cell metabolism has been broadly investigated. It is now evident that ceramide is involved as a second messenger in what has become known as the sphingomyelin cycle (1), apoptosis, and differentiation in many cell types (2). The mechanisms by which ceramide mediates apoptosis have not yet been fully addressed, however, it is known that mitochondria are targets of ceramide. Thus, direct inhibition of complex III of the mitochondrial respiratory chain by ceramide (3), ceramide-induced generation of reactive oxygen species in intact mitochondria (4) and in cells (5,6), and ceramide-induced cell death via disruption of mitochondrial functions (7) are lines of evidence of the strong influence of ceramide on mitochondria.
Cytochrome c (cyto c) 1 plays a dual role in cell homeostasis.
As a part of the respiratory chain, it is needed for cell life, and as one of the triggers of apoptosis, it is needed for cell death. It is now well accepted that many apoptogenic factors induce cell death via mitochondrial cyto c release (8). The released cytochrome switches on the death machinery, for example, by activation of caspases (9,10). The anti-apoptoic protein, Bcl-2, was shown to prevent apoptosis both upstream (8) and downstream (11) of cyto c release.
In the present study we show that 1) C 2 -ceramide (N-acetylsphingosine), C 6 -ceramide (N-hexanoylsphingosine), and, to a much lesser extent, DHC (C 2 -dihydroceramide) release cyto c from isolated mitochondria, 2) ceramide-induced cyto c release occurs when cyto c is oxidized but not when it is reduced, 3) this release is prevented by Bcl-2, 4) cyto c release causes a decrease in mitochondrial oxygen consumption, transmembrane potential (⌬⌿), and Ca 2ϩ retention, all of which are prevented by preincubation of mitochondria with Bcl-2 and reversed by addition of cyto c, and 5) ceramide interacts with cyto c and changes its reducibility. EXPERIMENTAL PROCEDURES C 2 -and C 6 -ceramide were obtained from Alexis Biochemicals (Lä ufelfingen, Switzerland), DHC from Calbiochem (Luzern, Switzerland), horse heart cytochrome c from Sigma, mouse monoclonal cyto c antibody from RDI (Flanders, NJ), anti-mouse Ig and horseradish peroxidase from Amersham and His 6 -human Bcl-2 from Novartis (Basel, Switzerland). Ceramide stock solutions were prepared at a 500 times concentration in ethanol (containing 1% Me 2 SO) and kept at Ϫ20°C. The vehicle always served as control.
Mitochondrial Preparation-Isolation of rat liver mitochondria was performed by differential centrifugation as described (12). The protein content of mitochondria and the mitochondrial supernatants were determined by the Biuret method with bovine serum albumin as standard.
Detection of Cytochrome c Release-Freshly isolated mitochondria (10 mg protein/ml) were incubated at room temperature in 0.1 M HEPES buffer, pH 7.0, containing aprotinin, pepstatin A, and leupeptin (1.5 g/ml each). To investigate the effect of ceramide when cyto c is oxidized, mitochondrial respiratory chain complex III was blocked by 50 nM antimycin A (AA), and after 1 min, ceramide (20 M) or the vehicle was added. Mitochondria were incubated for 2 min and then energized with 1 mM ascorbate (Asc) plus 0.4 mM tetramethyl-1,4-phenylenediamine (TMPD) (Asc/TMPD). The effect of ceramide on mitochondria when cyto c is reduced was studied by addition of ceramide 1 min after Asc/TMPD. Bcl-2 (40 nM) was added 5 min before AA. After 10 min of incubation at room temperature, mitochondria were spun at 12,000 ϫ g for 10 min at 4°C, and the resulting supernatant was spun at 100,000 ϫ g for 15 min at 4°C. The supernatant of the second centrifugation was used for the detection of cyto c either spectrophotometrically or by gel electrophoresis. Spectrophotometric measurements were done in a Varian Cray spectrophotometer. As the blank sample, 10 mg of mitochondrial protein was diluted in 1 ml of the buffer, mixed gently, * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. and spun immediately in 2 steps as mentioned above, and the supernatant was considered as the blank. The absorption of the supernatants of mitochondria incubated with ceramide or the vehicle was then recorded before and after reduction with sodium dithionite. The concentration of cyto c was calculated using an extinction coefficient of 19.0 mM Ϫ1 cm Ϫ1 for reduced cyto c at 550 -540 nm (13) and is expressed as the percentage of the control. For gel analysis, 20 l of the supernatant of mitochondria incubated with ceramide or the vehicle was separated by 15% SDS-PAGE and stained with Coomassie Blue. For Western blotting, 20 g of protein of supernatants were separated by 15% SDS-PAGE, blotted onto a nitrocellulose membrane, probed by a monoclonal mouse anti-cyto c antibody, and developed by enhanced chemiluminescence.
For studying the possible influence of ⌬⌿ on mitochondrial cyto c release, ⌬⌿ was fully abrogated by incubation of mitochondria with 5 M rotenone, 50 nM AA, and 1 M carbonyl cyanide m-chlorophenylhydrazone. The absence of ⌬⌿ was verified as explained below. After 10 min, mitochondria were spun as described above, and the supernatant was tested for the presence of cyto c by Western blotting.
Analysis of Mitochondrial Functions-For studying mitochondrial functions, 1 mg of mitochondrial protein/ml was incubated in 0.1 M HEPES buffer, pH 7.0, with the following compounds present where appropriate: 10 -20 M ceramide, 50 nM AA to block complex III, Asc/ TMPD to reduce cyto c, 1 mM KCN to block complex IV, 1 mM ATP as the substrate of ATPase, 1 M carbonyl cyanide m-chlorophenylhydrazone to uncouple mitochondria, and 1.7 g/ml oligomycin to block ATPase. Ceramide was added when cyto c was oxidized, i.e. after AA, or when it was reduced, i.e., after Asc/TMPD. Bcl-2 was added 5 min prior to the above mentioned compounds.
Oxygen consumption was measured at room temperature with a Clark-type oxygen electrode (Yellow Spring Instruments, Yellow Spring, OH) under continuous stirring. Mitochondrial transmembrane potential was measured in an Aminco DW-2A spectrophotometer at 511-533 nm in the presence of 10 M safranin as described (14). Mitochondrial Ca 2ϩ uptake and release was measured in 0.1 M HEPES buffer, pH 7.0, containing 10 M CaCl 2 (10 nmol of Ca 2ϩ /mg of mitochondrial protein) at 685-675 nm in the presence of 50 M Arsenazo III as described (15).
Assessment of Cytochrome c Reduction-The optical density of a 10 M oxidized horse heart cyto c solution in the presence of 10 M ceramide or the vehicle was recorded between 380 and 600 nm at 1-nm intervals with a Varian Cray spectrophotometer. Oxidized cyto c was then reduced stepwise by the repeated addition of 2 M ascorbate (taken from a 2 mM freshly prepared stock solution) to the cuvette. After each reduction step, the optical density was again recorded. The recorded optical density of the oxidized cyto c in each wavelength was subtracted from the corresponding one of the reduced cyto c and plotted against the wavelengths.

RESULTS
Ceramide-induced mitochondrial cyto c release was determined by SDS-PAGE (Fig. 1A), by Western blotting (Fig. 1B), and spectrophotometrically (Fig. 1C). In these experiments, ceramide was added when cyto c was mainly oxidized, i.e. after blocking the complex III. Addition of ceramide to mitochondria when cyto c was mainly reduced, i.e., in the presence of Asc/ TMPD, did not increase the released cyto c, compared to the control (Fig. 1D). Preincubation of mitochondria with 40 nM Bcl-2 (4 pmol of Bcl-2/mg of mitochondrial protein) fully prevented the release of cyto c induced by ceramide (see Fig. 1B). When equal volumes (20 l) of the supernatants of mitochondria incubated with ceramides were separated by SDS-PAGE, an increase in the total protein amount released into the supernatant was detected (Fig. 1A). When equal amounts (20 g) of the released proteins were analyzed by Western blot, a specific increase in the cyto c was found (Fig. 1B).
To investigate the consequences of cyto c release on mitochondrial functions, we measured mitochondrial oxygen consumption, ⌬⌿, and Ca 2ϩ homeostasis. Fig. 2A shows that addition of ceramide to mitochondria when cyto c was oxidized decreased the oxygen consumption supported by Asc/TMPD. Conversely, ceramide added to mitochondria when cyto c was reduced did not change the oxygen consumption (Fig. 2B). The decreased oxygen consumption caused by ceramide was prevented by preincubation of mitochondria with 40 nM Bcl-2 ( Fig.  2C) and was reversed by the addition of 200 nM exogenous cyto c (200 pmol of cyto c/mg of mitochondrial protein) (Fig. 2D).
A decrease in ⌬⌿ is considered important when cells commit suicide (16). Fig. 3A shows that addition of ceramide to mitochondria when cyto c was oxidized caused a decrease in ⌬⌿. This figure also shows that addition of 200 nM exogenous cyto c resulted in a full gain of ⌬⌿. Incubation of mitochondria with 40 nM Bcl-2 prevented the loss of ⌬⌿ caused by ceramide (not shown). Fig. 3B shows that ceramide did not alter ⌬⌿ when cyto c was mainly reduced. When ⌬⌿ was built up as a consequence of ATP hydrolysis instead of respiration, ceramide did not change ⌬⌿, neither when cyto c was oxidized nor when it was reduced (not shown).
Mitochondria are important calcium buffers in eukaryotic cells, and mitochondrial calcium release is involved in apoptosis (17). Fig. 4 shows that C 6 -ceramide, added to mitochondria when cyto c was mainly oxidized, caused a decrease in Ca 2ϩ retention by mitochondria, in a Bcl-2 sensitive manner. The same results were obtained with C 2 -ceramide and, to a minor extent, with DHC (not shown). Addition of ceramide to mitochondria when cyto c was reduced did not change the mitochondrial Ca 2ϩ homeostasis (not shown).
Reduction of cyto c can be followed photometrically. Fig. 5 shows that in the presence of C 2 -ceramide, the reduction of cyto c by Asc was hampered, as evidenced by the smaller increase in the optical densities of the ␥and ␣-regions. Again, C 2 -ceramide was most effective, followed by C 6 -ceramide and DHC (not shown).

DISCUSSION
The present study shows that ceramide induces cyto c release from isolated mitochondria, an event strongly influenced by the redox state of cyto c, and that incubation of mitochondria with Bcl-2 prevents the cyto c release. Release of cyto c decreases mitochondrial oxygen consumption, ⌬⌿, and the Ca 2ϩ buffering capacity of mitochondria, all of which are reversed by addition of exogenous cyto c. For all parameters measured, the observed rank order of potency is C 2 Ͼ C 6 Ͼ DHC. This study also provides evidence for a possible direct interaction of ceramide and cyto c.
There is growing evidence that mitochondria are involved in apoptosis (16). The release of cyto c from mitochondria triggers apoptosis (18), and ⌬⌿ decreases during apoptosis (16, 19 -21). Accordingly, prevention of cyto c release (22) as well as stabilization of ⌬⌿ (23) prevent apoptosis. Furthermore, mitochondria carry the pro-and antiapoptotic proteins, cyto c and Bcl-2. The presence of procaspase-3 in mitochondria was also shown recently (24).
Several mediators, pathways, and factors are involved in apoptosis (25,26). Among them, ceramide has been shown to directly target mitochondria. Zhang et al. (27) showed that in Molt-4 leukemic cells, 6 h of incubation with C 6 -ceramide increased the cytosolic concentration of cyto c, which was preventable by overexpression of Bcl-2. In a study by Amarante-Mendes et al. (28) 6 h of incubation with C 2 -ceramide caused the cytosolic accumulation of cyto c in control but not in Bcr-Abl-overexpressing HL-60 cells. In the present study, we show that ceramide directly causes the release of cyto c from isolated mitochondria and accordingly hypothesize that cyto c is the  prime mitochondrial target of ceramide. The fact that incubation of mitochondria with a low concentration of Bcl-2, used in this study, prevents ceramide-induced cyto c release and its consequences suggests a specific function of the oncoprotein in ceramide-mediated apoptotic signals.
To investigate a possible direct interaction of ceramide with cyto c, we primed Sepharose columns with the biologically active D-C 2 -ceramide or biologically inactive L-C 2 -ceramide. 2 We observed that D-C 2 -ceramide but not L-C 2 -ceramide columns selectively retained cyto c and that the retained protein could be eluted with D-C 2 -ceramide. We also observed that reduced cyto c had a much lower affinity to D-C 2 -ceramide column, compared to the oxidized cyto c. Because of that observation together with the fact that ceramide affects the reducibility of cyto c (Fig. 5), we hypothesize that ceramide may directly interact with cyto c, with a higher affinity for the oxidized protein, and that this interaction changes the physicochemical properties of cyto c, leading to its rejection from mitochondria. The fact that cyto c has multiple lipid binding sites and that the lipid-bound cyto c shows a lower affinity for attachment to the artificial membranes (29) strengthens this hypothesis.
Under our experimental conditions, prevention of mitochon-drial cyto c reduction is paralleled by the disappearance of ⌬⌿.
To distinguish which of these two is decisive for ceramideinduced cyto c release, we investigated by Western blot analysis whether cyto c is released because of ⌬⌿ collapse. We found that the absence of ⌬⌿, achieved by blocking respiration and uncoupling of mitochondria, does not result in cyto c release, indicating that binding of cyto c to the outer side of the inner mitochondrial membrane does not require ⌬⌿. These results also indicate that the release of cyto c by ceramide is not a consequence of a nonspecific solubilization of mitochondrial membranes due to the lipophilicity of ceramide, but rather is a specific event.
The release of other mitochondrial proteins apart from cyto c is also increased upon treatment with ceramide (Fig. 1A). We argue that this apparently nonspecific protein release by ceramide is due to a general weakening of mitochondria upon cyto c release. Fig. 1D supports this argument by showing that when ceramide does not cause cyto c loss, other mitochondrial proteins are also not released. This notion, together with the fact that addition of exogenous cyto c reverses all the altered mitochondrial functions, leads us to conclude that mitochondrial cyto c is a prime target for ceramide. This conclusion does not rule out other possibilities, for example, the modification of Bcl-2 by ceramide.
In contrast to the finding that binding of cyto c to the inner mitochondrial membrane is not a function of ⌬⌿, the formation and the maintenance of ⌬⌿ is dependent on the presence of a functional cyto c (cf. Fig. 3A). Disruption of ⌬⌿ was shown to be involved in ceramide-induced apoptosis (30). According to our study, cyto c has a critical role in ⌬⌿ formation, in that stabilization of cyto c stabilizes ⌬⌿. Bcl-2 is located at the outer mitochondrial membrane (31) and has been shown to stabilize ⌬⌿ (23). It is known that Bcl-xL, another oncoprotein with mitochondrial location, binds cyto c (32). Based on the present results and the close proximity of Bcl-2 to mitochondrial cyto c, it may be speculated that Bcl-2 stabilizes ⌬⌿ by preventing cyto c loss.
Nonspecific solute transport across the inner mitochondrial membrane, operated via a Bcl-2-sensitive megachannel (the "permeability transition pore") is considered to be the reason for the collapse of ⌬⌿ and many other features of apoptosis (reviewed in Ref. 33). It was also reported that operation of such a pore causes cyto c release, and thus it was concluded that cyto c release occurs downstream to pore formation and ⌬⌿ collapse (34). Because addition of cyto c reverses the ceramide-induced decrease in mitochondrial oxygen consumption, ⌬⌿, and Ca 2ϩ -maintaining capacity (see Figs. 2-4), we argue that ceramide-induced apoptosis is not mediated via opening a nonspecific megachannel or pore. The fact that ceramide does not affect ⌬⌿ when it is supported by ATP gives further support to this argument.
It was reported by Gudz et al. activity of complex III of the mitochondrial respiratory chain, as deduced from the measurement of mitochondrial oxygen consumption supported by Asc/TMPD. Cyto c shuttles electrons between complex III and IV, and oxygen is consumed at the level of complex IV. Figs. 2D and 3A show that the addition of cyto c to mitochondria recovered the decreased respiration and the reduced ⌬⌿. Therefore, we conclude that parts of the ceramide-induced reduction of complex III activity is due to cyto c loss.
Garcia-Ruiz et al. (4) showed that ceramide induces superoxide radical formation in isolated rat liver mitochondria. From the elegant recent study by Cai and Jones (35), it is evident that superoxide formation by mitochondria is a consequence, and not the cause, of cyto c loss. Release of cyto c from mitochondria produces a gap in the hierarchically arranged mitochondrial respiratory complexes III and IV, and therefore a site for electron leakage. Present finding explains the ceramideinduced superoxide formation, reported by Garcia-Ruiz et al. (4), and supports the hypothesis that cyto c is the prime mitochondrial site of action of ceramide.
We reported (23) that TNF-␣ added to cells induces a drop in ⌬⌿ that is accompanied by increased reactive oxygen species formation in mitochondria and that both events are prevented in Bcl-2-overexpressing cells. It is known that de novo ceramide synthesis is part of the transmembrane signaling transduction mechanism of TNF-␣ receptor (36 -40). It was also reported that stimulation of TNF-␣ receptors causes apoptotic cell death by releasing mitochondrial cyto c (41). The present study makes the link: after de novo ceramide synthesis upon TNF-␣ receptor stimulation, mitochondrial cyto c is released, ⌬⌿ is decreased, and reactive oxygen species formation is favored. Accordingly, overexpression of Bcl-2 prevents the apoptogenic properties of TNF-␣ (37) by prevention of ceramide-induced cyto c release.
It was reported by Hampton et al. (42) that cyto c does not necessarily need to be reduced to activate caspases. The investigators acknowledged, however, that they were unable to keep cyto c oxidized in the presence of cytosolic extracts. In the present study, we show that oxidized cyto c is released by ceramide (Fig. 1, A-C) and that ceramide decreases the reducibility of oxidized cyto c (Fig. 5). Regarding the fact that both ceramide and cyto c are strong mediators of caspase-induced apoptosis (43), the possible interaction of ceramide and cyto c may thus amplify the progression of the apoptotic cascade.