Positively Charged Ceramide Is a Potent Inducer of Mitochondrial Permeabilization*

Ceramide-induced cell death is thought to be mediated by change in mitochondrial function, although the precise mechanism is unclear. Proposed models suggest that ceramide induces cell death through interaction with latent binding sites on the outer or inner mitochondrial membranes, followed by an increase in membrane permeability, as an intermediate step in ceramide signal propagation. To investigate these models, we developed a new generation of positively charged ceramides that readily accumulate in isolated and in situ mitochondria. Accumulated, positively charged ceramides increased inner membrane permeability and triggered release of mitochondrial cytochrome c. Furthermore, the positively charged ceramide-induced permeability increase was suppressed by cyclosporin A (60%) and 1,3-dicyclohexylcarbodiimide (90%). These observations suggest that the inner membrane permeability increase is due to activation of specific ion transporters, not the generalized loss of lipid bilayer barrier functions. The difference in sensitivity of ceramide-induced ion fluxes to inhibitors of mitochondrial transporters suggests activation of at least two transport systems: the permeability transition pore and the electrogenic H+ channel. Our results indicate the presence of specific ceramide targets in the mitochondrial matrix, the occupation of which triggers permeability alterations of the inner and outer mitochondrial membranes. These findings also suggest a novel therapeutic role for positively charged ceramides.

teins, which, in turn, alter the outer mitochondrial membrane permeability for cytochrome c and other pro-apoptotic molecules. In this pathway, targets for ceramide are non-mitochondrial molecules: cathepsin D, which triggers translocation of Bax to the mitochondria (5,7,8), and serine/threonine protein phosphatase 2A, which dephosphorylates Bcl-2, thereby decreasing its antiapoptotic activity (9). An additional substrate for protein phosphatase 2A is the serine/threonine kinase Akt/protein kinase B (5). Protein phosphatase 2A inactivation of Akt results in dephosphorylation and activation of pro-apoptotic Bad, an Akt immediate substrate (10). The overall effect of ceramide on this pathway is an increase in pro-apoptotic proteins bound to mitochondria.
Evidence is also accruing to implicate direct actions of ceramide on mitochondria. Specifically, the selective hydrolysis of a mitochondrial pool of sphingomyelin by bacterial sphingomyelinase targeted to the mitochondrial matrix results in apoptosis, whereas production of ceramide in the plasma membrane, endoplasmic reticulum, nucleus, and Golgi apparatus by bacterial sphingomyelinase targeted to these compartments has no effect on cell viability (11). This study therefore pointed to a local action of ceramide on mitochondria in intact cells.
In this context, modulation by ceramide of mitochondrial functions at the level of isolated organelles has provided further evidence in support of this second mechanism. It has been reported that ceramides directly suppress respiratory chain activity (12)(13)(14). Moreover, current research is focused on the ability of ceramides to release cytochrome c from the intermembrane space and to induce permeability of the inner mitochondrial membrane, permitting passage of low molecular solutes. In the models of Siskind et al. (15,16) and Ghafourifar et al. (17), the outer mitochondrial membrane is considered a primary target for ceramides when inducing cytochrome c release, whereas the inner membrane is viewed as being ceramideinsensitive. In contrast, Pastorino et al. (18) and Szalai et al. (19) suggested that the opening of a PTP 1 of the inner mitochondrial membrane could be a primary event in initiation of cytochrome c release and in increased solute permeability of the inner membrane in the presence of ceramides. Yet another report by Di Paola et al. (14) provides evidence for the role of ceramide as a nonspecific modulator of ionic permeability of the lipid component of the inner membrane. Thus, the proposed mechanism by which ceramides alter mitochondrial membrane permeabilization are varied, and the localization of functionally significant ceramide targets (outer mitochondrial membrane, inner membrane, or matrix space) is also unclear.
Given these intimate and direct connections between ceramide and mitochondria, we sought to develop a strategy by which ceramide can be selectively delivered to the mitochondrial matrix to probe its mechanisms of action. In this study, we report on the development of derivatives of ceramide with a fixed positive charge (Fig. 1). These molecules are expected to accumulate in the mitochondrial matrix based on their electrochemical potential and thus would serve as direct probes of ceramide functions on mitochondria. Our results show that these novel agents do localize selectively in mitochondria in living cells. We next investigated the effects of these ceramides on permeability of the inner and outer mitochondrial membranes. Our study shows that these positively charged ceramides increase permeability of the inner membrane (decrease in ⌬⌿ and large amplitude swelling), which, in turn, results in release of cytochrome c. In addition, we provide evidence that ceramideinduced permeabilization of the inner mitochondrial membrane is mediated by specific ion transport systems, viz. the PTP and electrogenic H ϩ transporter.

EXPERIMENTAL PROCEDURES
Materials-RPMI 1640 medium, Dulbecco's modified Eagle's medium, and fetal bovine serum were from Invitrogen. TMRM was from Molecular Probes, Inc. C 6 -NBD-ceramide was from Matreya. Ceramides and their derivatives were from the Lipidomics Core of the Medical University of South Carolina. All other reagents were from Sigma.
Preparation of Mitochondria from Rat Liver-Mitochondria were prepared from livers of male Sprague-Dawley rats (220 -250 g) fasted overnight. Livers from two rats were homogenized in 100 ml of isolation medium containing 230 mM mannitol, 70 mM sucrose, 2 mM EDTA, and 10 mM HEPES (pH 7.4 adjusted with KOH). The homogenate was centrifuged at 579 ϫ g max for 10 min to pellet the nucleus and unbroken cells. The supernatant from the previous step was centrifuged at 8000 ϫ g max for 10 min to pellet the mitochondria. The mitochondrial pellet was washed with 25 ml and then with 12.5 ml of isolation medium without EDTA. The final mitochondrial pellet was resuspended in the above medium to provide a protein concentration of 60 mg/ml. Mitochondrial protein concentration was determined by the BCA assay using bovine serum albumin as the standard (20).
Mitochondrial Incubation Medium-Unless otherwise specified, incubations of isolated mitochondria were conducted at 25°C with 1 mg/ml protein in medium containing 250 mM sucrose, 10 mM HEPES (pH 7.4 adjusted with KOH), 10 mM succinate, 5 mM KH 2 PO 4 , and 2 M rotenone. Deviations from this medium and other reagents employed are described in the figure legends.
Mitochondrial Respiration-Oxygen consumption by mitochondria was measured in a chamber equipped with a Clark-type oxygen electrode (Instech Laboratories) under the conditions described under "Mitochondrial Incubation Medium." Synthesis of Mitochondrially Targeted Ceramide Molecules-The mitochondrially targeted compounds consisted of the lipophilic cation pyridinium covalently linked to ceramide. These pyridinium-ceramides were prepared by N-acylation of D-erythro-sphingosine with -bromo acid chlorides following quaternization of pyridine with the formed -bromoceramides. The detailed synthesis of pyridinium-ceramides has been described. 2 Measurement of Mitochondrial Permeabilization-Inner membrane permeabilization was assayed by measurements of ⌬⌿ and mitochondrial swelling and by changes in mitochondrial ultrastructure. ⌬⌿ was estimated from the accumulation of TPP ϩ in the mitochondrial matrix as described by Kamo et al. (22). TPP ϩ (2 M) was added to the incubation medium as indicated in the figure legends. Mitochondrial swelling was measured by changes in absorbance at 520 nm using a Brinkmann PC 900 probe colorimeter and fiberoptic probe.
Changes in mitochondrial ultrastructure were examined by electron microscopy. Mitochondria were fixed with 3% glutaraldehyde for 15 min, followed by sedimentation and additional fixation overnight. The fixed mitochondria were washed three times with 0.1 M sodium cacodylate (pH 7.4), treated with 2% osmium tetroxide for 1 h, dehydrated through a graded ethanol series, and embedded in Embed 812 resin. Thin sections (70 nm) were stained with uranyl acetate and lead citrate and subsequently examined using a Jeol/JEMI 1010 electron microscope.
Cytochrome c Release from Mitochondria-Aliquots of mitochondrial suspension were taken as indicated in the figure legends and centrifuged at 15,000 ϫ g for 3 min. The supernatant and mitochondrial pellet were frozen and stored at Ϫ20°C. Cytochrome c in the supernatants and pellets was quantified using the Quantikine cytochrome c enzymelinked immunosorbent assay kit (R&D Systems, Minneapolis, MN).
Cell Culture-HepG2 cells (obtained from American Type Culture Collection) were cultured in Eagle's minimal essential medium supplemented with 10% fetal bovine serum, 2 mM L-glutamine, nonessential amino acids, 1 mM sodium pyruvate, and 1.5 g/liter sodium bicarbonate in humidified air (5% CO 2 ) at 37°C. For confocal microscopy, cells were plated onto poly-D-lysine-coated 35-mm glass bottom microwell dishes at a density of 20,000 -25,000/cm 2 and were grown for 2 days. MCF7 cells (obtained from American Type Culture Collection) were cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum and 2 mM glutamine in humidified air (5% CO 2 ) at 37°C. All media were supplemented with 100 units/ml penicillin and 100 g/ml streptomycin.
Isolation of Mitochondria from HepG2 Cells-For studies with mitochondria isolated from HepG2 cells, cells were cultured in the medium described under "Cell Culture" for 3 days in 75-cm 2 flasks until 70% confluent. Cells were detached by treatment with 3 ml of 0.05% trypsin and 0.53 mM EDTA, diluted to 13 ml with incubation medium, and sedimented at 900 ϫ g for 10 min. The pellet was washed with 1 ml of ice-cold phosphate-buffered saline, and cells were resuspended in 300 l of isolation medium and then disrupted by 20 passages through a 28-gauge needle ( 1 ⁄2 inch). The homogenate was centrifuged at 900 ϫ g for 10 min to pellet the nucleus and unbroken cells. The supernatant from the previous step was centrifuged at 10,000 ϫ g for 10 min to pellet the mitochondria, which were then resuspended in incubation medium to provide a protein concentration of ϳ10 mg/ml.
Measurement of Cell Viability-HepG2 or MCF7 cells were plated at a density of 10 4 cells/well in 96-well plates in the medium described under "Cell Culture." After 24 h of incubation, the cells were treated with ceramides for 46 h in 2% fetal bovine serum. Cell viability was determined by the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (Sigma) following the manufacturer's instructions. Confocal Microscopy-Plated cells were washed once with serum-free medium and treated with 2 ml of 100 nM TMRM, 2 M C 6 -pyridinium-DMAS-ceramide, or 2 M C 6 -NBD-ceramide dissolved in the culture medium supplemented with 2% fetal bovine serum. After 30 min, unbound dyes were washed out, and images were collected using a Zeiss LSM 510 META system equipped with a krypton/argon laser and a ϫ63 oil objective (numerical aperture of 1.4). In parallel experiments, after initial loading of cells with TMRM or ceramides, cells were treated with uncoupling mixture (10 M FCCP, 5 g/ml antimycin A, and 10 g/ml oligomycin A) for an additional 30 min to the discharge mitochondrial inner membrane potential. The TMRM images were collected by excitation at 543 nm and emission at 560 nm with a long path emission filter. The C 6 -pyridinium-DMAS-ceramide and C 6 -NBD-ceramide images were collected by excitation at 488 nm and emission at 505 nm with a long path emission filter.
Analysis of Ceramides by Mass Spectroscopy-Accumulated ceramides in mitochondria were analyzed by mass spectroscopy using reverse-phase high performance liquid chromatography couple with an electrospray triple quadruple mass spectrometer, operating at positive ionization in the multiple reaction monitoring mode. Mass separations were performed using a ThermoFinnigan TSQ 7000 mass spectrometer according to the methodology described by Bielawski et al. (45).
Statistical Analysis-Standard curves and the data for cytochrome c release were computed by generation of a four-parameter logistic curve fit. The values for ceramide accumulation and cytochrome c release are expressed as the means Ϯ S.E. Differences between data were analyzed for significance by Student's t test. The results were considered significant at p Ͻ 0.05.

RESULTS
C 6 -pyridinium-ceramide Accumulates in Intact Cell Mitochondria in an Energy-dependent Manner-To determine whether whole cells will accumulate exogenous pyridiniumceramides in the mitochondrial matrix, we used a fluorescent analog of C 6 -pyridinium-ceramide, C 6 -pyridinium-DMAS-ceramide. Fig. 2 shows the distribution pattern of C 6 -pyridinium-DMAS-ceramide (panels C and D) and the specific mitochondrial reporter TMRM (panels E and F), which is known to accumulate inside the mitochondrial matrix. Both C 6 -pyridinium-DMAS-ceramide-and TMRM-treated cells demonstrated a similar punctate pattern of staining, characteristic of mitochondria. Thus, C 6 -pyridinium-DMAS-ceramide accumulates selectively in mitochondria in living cells.
Subsequent addition of the uncoupler FCCP in combination with inhibitors of the respiratory chain and ATPase (antimycin A and oligomycin, respectively) resulted in diffuse staining of the cytoplasm for both fluorophores, indicating that mitochondrial accumulation of C 6 -pyridinium-DMAS-ceramide in intact cells is indeed energy-dependent. In the presence of uncouplers, the diffuse staining of C 6 -pyridinium-DMAS-ceramide probably reflects equilibration of this molecule in cell membranes without specific concentration in any one compartment.
In contrast to C 6 -pyridinium-DMAS-ceramide, cells treated with a fluorescent analog of neutral C 6 -ceramide, viz. C 6 -NBDceramide, developed prominent fluorescence in a perinuclear region ( Fig. 2A), whereas mitochondrial staining was minimal. These results are consistent with several previous studies that have identified this compartment as the Golgi apparatus, and indeed, C 6 -NBD-ceramide has been accepted as a specific marker of this compartment (23,24). Also, in agreement with previous observations that accumulation of C 6 -NBD-ceramide in the Golgi apparatus is energy-independent (24), Fig. 2B demonstrates that uncouplers of oxidative phosphorylation did not affect staining of the perinuclear compartment by C 6 -NBDceramide. Taken together, these experiments provide evidence that exogenously added pyridinium-ceramide localizes preferentially to mitochondria and that its mitochondrial accumulation in situ is energy-dependent.
To demonstrate definitively that C 6 -pyridinium-ceramide preferentially accumulates in mitochondria, HepG2 cells were treated with equal concentrations of C 6 -ceramide and C 6 -pyridinium-ceramide (3 M) for 30 min; mitochondria were then isolated; and ceramide values were determined by mass spectroscopy. The amount of C 6 -pyridinium-ceramide in mitochondria was ϳ7-fold higher compared with the amount of C 6ceramide (985 and 142 pmol/mg of protein, respectively). C 6 -pyridinium-ceramide Is a Potent Effector of Cell Viability-Next, we tested whether C 6 -pyridinium-ceramide is a more potent cell-killing agent compared with its uncharged analog. Indeed, C 6 -pyridinium-ceramide readily induced killing of hepatocarcinoma HepG2 cells (IC 50 ϳ 8 M) (Fig. 3A, trace 2), whereas electroneutral C 6 -ceramide was much less effective at the same range of concentrations (IC 50 ϳ 31 M) (trace 1). The effect of C 6 -pyridinium-ceramide is not unique for HepG2 cells. MCF7 breast cancer cells also responded to this compound (Fig.  3B). MCF7 cells appeared to be more sensitive to C 6 -ceramide compared with HepG2 cells (IC 50 ϳ 16 M) and demonstrated a considerable increase in sensitivity to C 6 -pyridinium-ceramide (IC 50 ϳ 2 M).
Accumulation of C 6 -pyridinium-ceramide in Isolated Rat Liver Mitochondria Is Energy-dependent-Next, we set out to determine whether the accumulation of C 6 -pyridinium-ceramide by isolated mitochondria is energy-dependent. The addition of C 6 -pyridinium-ceramide (10 M) to mitochondria resulted in 95% association with mitochondria ( Fig. 4). Dissipation of mitochondrial ⌬⌿ by simultaneous addition of the complex III inhibitor antimycin A and the protonophore FCCP suppressed accumulation of C 6 -pyridinium-ceramide by 66.8% (Fig. 4). The difference in the amount of ceramide bound in the absence and presence of uncouplers of oxidative phosphorylation provides the amount of ceramide accumulated by mitochondria in an energy-dependent manner, whereas the component resistant to uncouplers indicates the ceramide that may be partitioned into the lipid phase of mitochondrial membranes or associated with nonspecific binding sites. Calculating the approximate mitochondrial matrix volume as 1.6 l/mg of protein (25) and the ⌬⌿-dependent uptake of pyridinium-ceramide as 6.28 nmol/mg of protein, the concentration of pyridinium-ceramide in the matrix space can reach 3.9 mM. On the other hand, the addition of uncharged C 6 -ceramide (10 M) also resulted in its considerable association (79.3%) with mitochondria. The striking difference between the association of positively charged C 6 -pyridinium and electroneutral C 6 -ceramides is that the association of the latter is insensitive to dissipation of ⌬⌿. Thus, the association of C 6 -ceramide with mitochondria is exclusively related to its partitioning into the lipid phase of mitochondria and/or its association with nonspecific mitochondrial binding sites. Therefore, C 6 -ceramide is evenly redistributed between the lipid phase of the inner and outer membranes with equal concentration of free ceramide in the intermembrane space and the matrix. In contrast, C 6 -pyridinium-ceramide is highly en-riched in the inner membrane of energized mitochondria, and its free concentration in the matrix space is considerably elevated compared with that in the intermembrane space. C 6 -pyridinium-ceramide Is a Potent and Specific Inducer of Inner Mitochondrial Membrane Permeabilization-The results shown above suggest that, because of its greater accumulation in the mitochondrial matrix, C 6 -pyridinium-ceramide should affect mitochondrial function more potently compared with neutral ceramides. To this end, we compared the effects of C 6 -pyridinium-ceramide and its neutral derivative on permeability of the inner mitochondrial membrane for low molecular mass solutes. Respiring liver mitochondria, which contain ϳ10 -15 nmol of endogenous Ca 2ϩ /mg of protein, maintained accumulated TPP ϩ , an index of ⌬⌿, for Ͼ30 min (Fig. 5A, trace 1). Only slight decreases in the absorbance of the mitochondrial suspension (indicative of swelling) were observed under these conditions (Fig. 5B, trace 1), consistent with previous results on isolated mitochondria.
The addition of 40 M C 6 -pyridinium-ceramide induced a biphasic release of accumulated TPP ϩ (Fig. 5A, trace 2). The initial partial release of TPP ϩ was accomplished within 4 min and was followed by a slower phase of total TPP ϩ release FIG. 4. Accumulation of C 6 -ceramide and C 6 -pyridinium-ceramide in isolated rat liver mitochondria. Mitochondria were incubated under the conditions described under "Experimental Procedures," except that 1 M CSA and 1 mM EGTA were present from the beginning of the experiment, and 10 M C 6 -ceramide or C 6 -pyridinium-ceramide was added 2 min later after the addition of mitochondria. The plot shows the amounts of C 6 -ceramide (C-6) and C 6 -pyridinium-ceramide (C-6 pyr) that accumulated in energized and de-energized mitochondria. First and third bars (controls), binding of ceramides to mitochondria that developed high ⌬⌿ as a result of succinate oxidation under standard conditions; second and fourth bars (ϩFCCP), dissipation of ⌬⌿ by the addition of FCCP (1 M) and antimycin A (0.5 g/mg of protein) at the beginning of the experiment. Results are expressed as the means Ϯ S.E. (n ϭ 3). *, p ϭ 0.01 versus the control.

FIG. 5. Effects of ceramides on ⌬⌿ (A) and mitochondrial large amplitude swelling (B and C).
Mitochondria were incubated under the conditions described under "Experimental Procedures," except that 2 M TPP ϩ was present from the beginning of the experiment. Alamethicin (ALA; 7 g/mg of protein), a pore-forming peptide, was added as indicated to induce permeabilization and to determine the full extent of the potential changes in the parameters of interest. Where indicated, ceramides (40 M) were present from the beginning of the experiment. A and B, time courses of ceramide effects on ⌬⌿ and mitochondrial swelling. Traces 1, no addition; traces 2, C 6 -pyridinium-ceramide (C-6 pyrid); traces 3, C 6 -ceramide (C-6); traces 4, C 2 -pyridinium-ceramide (C-2 pyrid). RLM, rat liver mitochondria. C, dose-response curves of ceramide effects on mitochondrial swelling. The degree of mitochondrial swelling was determined 30 min after ceramide treatment. Trace 1, C 6 -pyridinium-ceramide; trace 2, C 6 -ceramide. reflecting complete dissipation of ⌬⌿. This later phase was accompanied by a rapid decrease in absorbance, which indicates stimulation of large amplitude swelling caused by increased permeability of the inner membrane to the components of the incubation medium (Fig. 5B, trace 2). Indeed, the effects of C 6 -pyridinium-ceramide were very similar to those of the pore-forming peptide alamethicin, the addition of which to the mitochondrial suspension produced essentially the same lightscattering response as C 6 -pyridinium-ceramide (Fig. 5B, trace  3), suggesting that this ceramide enhances pore formation. Importantly, examination of mitochondrial ultrastructure by electron microscopy before (Fig. 6A) and 30 min (Fig. 6B) after the addition of C 6 -pyridinium-ceramide revealed the typical picture of large amplitude mitochondrial swelling, whereas in the absence of C 6 -pyridinium-ceramide, mitochondria remained in the aggregated configuration characterized by a shrunken matrix space and a large intracristal space (Fig. 6A). Incubation of mitochondria with C 6 -pyridinium-ceramide resulted in an extensive increase in matrix volume and unfolded cristae, characteristic of colloid/osmotic swelling (Fig. 6B). The inner membrane remained apparently intact, whereas the outer membrane was mostly ruptured and detached from the inner membrane. Thus, the results show that C 6 -pyridiniumceramide exerts significant effects on isolated mitochondria, which are characterized by a relatively specific increase in permeability of the inner membrane. In contrast to C 6 -pyridinium-ceramide, which induced dissipation of ⌬⌿ (Fig. 5A, trace 2) as well as mitochondrial swelling (Fig. 5C, trace 1) with IC 50 ϳ 27.5 M, neutral C 6 -ceramide failed to induce dissipation of ⌬⌿ (Fig. 5A, trace 3) or mitochondrial permeabilization (Fig. 5, B, trace 3; and C, trace 2) at concentrations up to 60 M.
Structural Specificity of C 6 -pyridinium-ceramide Action-To verify that the effect of C 6 -pyridinium-ceramide is due to its acting as a ceramide analog, we compared its effect with the action of a number of structurally related and unrelated positively charged compounds. First, we determined the effect of the pyridinium moiety on mitochondrial permeabilization. To this end, the effects of short chain C 2 -pyridinium-ceramide and cetylpyridinium were evaluated. Fig. 5 (A, trace 4; and B, trace 4) shows that, employed at the same concentration as C 6pyridinium-ceramide (40 M), C 2 -pyridinium-ceramide caused only minor changes in the magnitude of mitochondrial swelling and the value of ⌬⌿ compared with the control. Even employed at 60 M (binding of 29.8 Ϯ 1.4 nmol/mg of protein at 4 min) (Fig. 7A, trace 2), C 2 -pyridinium-ceramide failed to induce the same degree of swelling that was observed with 30 M C 6pyridinium-ceramide (binding of 5 Ϯ 0.8 nmol/mg of protein at 4 min) (Fig. 8A, trace 2).
When used at 60 M, cetylpyridinium provided only moderate mitochondrial swelling (Fig. 7B, trace 4). In line with this notion, two other hydrophobic cations (viz. TPP ϩ and TMRM) that readily accumulated in the mitochondrial matrix driven by ⌬⌿ (negative inside) failed to induce large amplitude swelling even at concentrations twice as high as C 6 -pyridinium-ceramide (TMRM (Fig. 7, A, trace 3) and TPP ϩ (Fig. 7B, trace 5), concentration of 60 M, binding of 50 Ϯ 2.6 nmol/mg of protein at 4 min; and C 6 -pyridinium-ceramide (Fig. 8A, trace 2), concentration of 30 M, binding of 5 Ϯ 0.8 nmol/mg of protein at 4 min). To the contrary, inhibition of the basal swelling rate was observed.
To further confirm that the effect of C 6 -pyridinium-ceramide is specific with respect to the structure of this molecule, we investigated the permeabilizing properties of its structural analog, viz. C 6 -pyridinium-dihydroceramide, which differs only by the lack of a 4,5-trans-double bond in the sphingoid backbone. Fig. 8A shows an ϳ3-fold increase in the lag period of the induction of mitochondrial swelling in the presence of C 6 -pyridinium-dihydroceramide (trace 3; concentration of 30 M, binding of 9.8 Ϯ 0.7 nmol/mg of protein at 4 min) compared with C 6 -pyridinium-ceramide (trace 2; concentration of 30 M, binding of 5 Ϯ 0.8 nmol/mg of protein at 4 min). Moreover, the dose-response curves (Fig. 8B) demonstrate that increases in the different ceramide concentrations shortened the lag periods of C 6 -pyridinium-and C 6 -pyridinium-dihydroceramide- induced swelling. These data indicate that the unsaturated pyridinium-ceramide analog is somewhat more effective than the pyridinium-dihydroceramide analog. Overall, these results indicate that C 6 -pyridinium-ceramide can be considered as an analog of the uncharged ceramide and that its action does not reflect nonspecific mitochondrial perturbation that could be expected with any cationic hydrophobic compound.

Inhibitors of Mitochondrial Ion Transporters (CSA and DCCD) Suppress the C 6 -pyridinium-ceramide-induced Mitochondrial
Permeability Increase-The permeability increase observed in the presence C 6 -pyridinium-ceramide could arise from the formation of lipid channels as a result of perturbation of the hydrophobic portion of the inner membrane, or alternatively, C 6 -pyridinium-ceramide could regulate specific transport pathways, resulting in equilibration of small molecules and ions across the inner membrane, large amplitude swelling, and dissipation of ⌬⌿. To discriminate between these two possibilities and to address the mechanism by which C 6 -pyridinium-ceramide induces mitochondrial permeability, we investigated the effects of the transpotent PTP inhibitor CSA and the mitochondrial ion transporter nonselective inhibitor DCCD on C 6 -pyridinium-ceramideinduced permeabilization of the inner membrane.
As shown in Fig. 9, CSA substantially suppressed (60%) and delayed the pyridinium-ceramide-induced decreases in ⌬⌿ and large amplitude swelling (A, trace 3; and B, trace 3, respectively). Chelation of the PTP activator Ca 2ϩ by EDTA as well as the use of another PTP inhibitor, bongkrekic acid, resulted in a similar degree of suppression of ceramide-induced mitochondrial alterations (data not shown). Although not as specific as CSA, the carboxylic group modifier DCCD is also known to be an inhibitor of PTP opening (26 -28). DCCD also suppressed the permeability increase induced by C 6 -pyridinium-ceramide by 90% (Fig. 10A, trace 2). This inhibition reached a maximum at a DCCD concentration of ϳ40 nmol/mg of protein (Fig. 10,  inset).
In contrast to the slow phase of ⌬⌿ discharge, the initial fast phase was insensitive to CSA (Fig. 9A, traces 1 and 3) and was accompanied by the shrinkage of mitochondria rather than by large amplitude swelling (Fig. 9B, traces 1 and 3). This rapid discharge of ⌬⌿ could be explained by ceramide-induced suppression of respiratory chain activity that could occur directly as described (12)(13)(14) or indirectly as a result of cytochrome c release from the intermembrane space (17,29,30). However, the addition of 10 M cytochrome c (an amount exceeding that for maximum activation of respiratory chain activity (30)) to the incubation medium did not modify the mitochondrial response to C 6 -pyridinium-ceramide (data not shown). Moreover, measurement of oxygen consumption of the mitochondrial suspension showed nearly maximum acceleration of respiration within the first minutes after C 6 -pyridinium-ceramide addition (Fig. 10B). These data strongly suggest activation of an electrogenic H ϩ leak across the inner membrane as a cause of Trace 3 was corrected for the absorbance of TMRM. For determination of cation and C 2 -pyridinium-ceramide binding to mitochondria, the mitochondria were incubated under essentially the same conditions, but 100 M DCCD was present from the beginning of the experiment. Four minutes after the addition of ceramides, mitochondria were sedimented, and the amount of ceramides in the pellet was determined by mass spectroscopy. TPP ϩ binding was determined using a TPP ϩ -selective electrode as described under "Experimental Procedures." Because TMRM (similar to TPP ϩ ) rapidly equilibrates across the inner membrane according to its electrochemical potential, its accumulated amount was assumed to be equal to that of TPP ϩ . RLM, rat liver mitochondria.

FIG. 8. Time course (A) and dose-response curves (B) of the effect of C 6 -pyridium-ceramide versus C 6 -pyridinium-dihydroceramide on mitochondrial large amplitude swelling.
Mitochondria were incubated under the conditions described under "Experimental Procedures." Alamethicin (ALA; 7 g/mg of protein), a pore-forming peptide, was added as indicated to induce permeabilization and to determine the full extent of mitochondrial swelling. Where indicated, ceramides (30 M) were added to the incubation medium. A, time courses of ceramide effects on mitochondrial swelling. Trace 1, no addition; trace 2, C 6 -pyridinium-ceramide (C-6 pyr); trace 3, C 6 -pyridinium-dihydroceramide (C-6 Dh pyr). B, dose-response curves of ceramide effects on mitochondrial swelling. The degree of mitochondrial swelling was determined 15 min after ceramide treatment. Trace 1, C 6 -pyridinium-dihydroceramide; trace 2, C 6 -pyridinium-ceramide. Determination of C 6 -pyridinium-ceramide and C 6 -pyridinium-dihydroceramide binding to mitochondria was performed as described in the legend to Fig. 7 for C 2 -pyridinium-ceramide. RLM, rat liver mitochondria. decreased ⌬⌿. Similar to the PTP opening described above, electrogenic ion fluxes arising in the first minutes after C 6pyridinium-ceramide addition were sensitive to DCCD, as can be seen from suppression of acceleration of oxygen consumption (Fig. 10B). Therefore, electrogenic H ϩ flux activated by ceramide is mediated by some specific transporter, not by disturbance of the lipid phase of the inner membrane.
Next, we set out to investigate the possible mitochondrial sites of ceramide action. The results in Figs. 2 and 3 indicate that uncoupling of mitochondria correlates with loss of C 6pyridinium-ceramide from the mitochondrial matrix. Under these same conditions, the addition of FCCP suppressed the swelling phase of the mitochondrial response to ceramide (Fig.  9B, trace 4). These observations therefore indicate that the sites of action of C 6 -pyridinium-ceramide for activation of PTP are localized in the inner membrane or matrix space of the mitochondria.
C 6 -pyridinium-ceramide Induces Cytochrome c Release in an Energy-dependent Manner-Studies aimed at elucidating the mechanisms of ceramide-induced cell death showed that ceramide acts at least in part by inducing the release of cytochrome c from mitochondria. Formation by ceramides of specific pores for cytochrome c and molecules of up to 60 kDa in the outer mitochondrial membrane was suggested as a preferential mechanism for cytochrome c release (16). Yet induction by ceramides of the classical permeability transition of mitochon-dria, which is accompanied by the their osmotic swelling, rupture of the outer membrane, and, as a result, release of cytochrome c from the intermembrane space, was proposed as an alternative model (18,19). To determine whether C 6 -pyridinium-ceramide is able to release cytochrome c from mitochondria and to address the mechanism by which this occurs, experiments were conducted to evaluate its effect on cytochrome c release under conditions that result in mitochondrial swelling versus conditions when C 6 -pyridinium-ceramide large amplitude swelling was suppressed by FCCP (Fig. 9B, trace 4).
As shown in Fig. 11B, incubation of mitochondria with C 6pyridinium-ceramide resulted in progressive large amplitude swelling. After 20 min of incubation with C 6 -pyridinium-ceramide, ϳ40% of the cytochrome c was released from mitochondria (Fig. 11A). When C 6 -pyridinium-ceramide-induced mitochondrial swelling was suppressed by the addition of FCCP plus antimycin A (Fig. 11B), an ϳ3-fold decrease in cytochrome c release was observed (Fig. 11A), which was comparable with the control value. C 6 -ceramide exerted no effect on cytochrome c release in both the absence and presence of uncouplers of oxidative phosphorylation compared with the control (Fig. 11A). Under the same conditions, C 6 -ceramide failed to increase large amplitude swelling (Fig. 11B). The addition of the pore-forming peptide alamethicin provided a 100% response in the parameters of interest that can be observed under the conditions employed. These results indicate that the preferential mechanism of cytochrome c release by C 6 -pyridinium-ceramide is permeabilization of the inner membrane as an initial step, with subsequent swelling and rupture of the outer membrane.

DISCUSSION
In this study, we have shown that positively charged C 6pyridinium-ceramide readily permeates the lipid bilayer and specifically targets the inner mitochondrial membrane and matrix space. Because of the large mitochondrial inner membrane potential (negative inside), these molecules accumulate inside isolated mitochondria and within mitochondria in cultured cells. Moreover, accumulation of these molecules is reversible and can be prevented by discharge of ⌬⌿. In addition, the accumulation of these ceramides in the mitochondrial matrix space increases permeability of mitochondrial membranes by activating putative ion porters of the inner mitochondrial membrane: PTP and the electrogenic H ϩ channel.
Several observations are in favor of this conclusion. First, C 6 -pyridinium-ceramide induced a light-scattering response (indicative of change in mitochondrial ultrastructure) that was similar in magnitude to that observed upon conventional Ca 2ϩ treatment (data not shown) or in the presence of the poreforming peptide alamethicin (Fig. 5B, trace 3). This suggests that the light-scattering response observed in the presence of C 6 -pyridinium-ceramide reflects mitochondrial large amplitude swelling, which is colloid/osmotic in nature as opposed to nonspecific amphiphilic compound-mediated solubilization of mitochondrial membranes. Additional support for the relative specificity of the permeability defect created by C 6 -pyridiniumceramide in the inner membrane comes from examination of mitochondrial ultrastructure by electron microscopy (Fig. 6, A  and B). Comparison of mitochondrial ultrastructure before and after ceramide treatment revealed all the features of classical permeability transition: increased mitochondrial volume, unfolded cristae, ruptured outer membranes, and apparent intactness of the inner membrane.
A second observation in favor of PTP opening came from the use of the PTP inhibitors CSA and DCCD. These inhibitors suppressed or delayed mitochondrial large amplitude swelling and discharge of ⌬⌿ by 60 and 90%, respectively. This indicates that the permeability transition observed in the presence of C 6 -pyridinium-ceramide is likely attributed to the activation of protein transporters of the inner mitochondrial membrane rather than the formation of lipid channels created by segregation of ceramides in a special domain, as was proposed previously for the outer membrane (16).
Our data also provide evidence that C 6 -pyridinium-ceramide activates additional ion transport pathways distinct from PTP. Indeed, the shrinkage phase observed during the first minutes after ceramide addition and the accompanying discharge of ⌬⌿ indicate selective loss of cations from the mitochondrial matrix and activation of electrogenic ion fluxes without a simultaneous increase in permeability to sucrose, which is usually observed in classical models of permeability transition. Perhaps these relatively specific cation fluxes reflect operation of PTP in a low conductance (impermeable to sucrose) state, as has been demonstrated previously (31,32). However, the lack of sensitivity of these fluxes to CSA does not support this notion. It also should be kept in mind that, although DCCD suppresses both these selective fluxes and the nonspecific permeability increase, this does not unequivocally indicate operation of PTP because, in contrast to CSA, this compound can modify other mitochondrial proteins such as the K ϩ /H ϩ exchanger (33) of the inner mitochondrial membrane and the F 0 channel of mitochondrial ATPase (34,35).
The best explanation for the initial mitochondrial response to ceramide treatment seems to be simultaneous activation of selective electrogenic K ϩ and H ϩ fluxes. K ϩ is known to be the most abundant ion in the mitochondrial matrix, playing a major role in regulation of mitochondrial volume (36). In this model, increased H ϩ permeability across the inner membrane FIG. 10. DCCD suppresses C 6 -pyridinium-ceramide-induced large amplitude swelling (A) and electrogenic ion fluxes (B) in isolated rat liver mitochondria. A, mitochondria were incubated under the conditions described under "Experimental Procedures." Traces 1 and 2, C 6 -pyridinium-ceramide (C-6 pyr; 40 M) was added were indicated. For trace 2, DCCD (100 M) was added at the beginning of the experiment. Alamethicin (ALA) was added as indicated to determine the full degree of permeabilization. The inset shows the dose-response curve of the DCCD effect on C 6 -pyridinium-ceramide-induced permeabilization. The curve was generated from experiments similar to those depicted in traces 1 and 2 with the indicated concentration of DCCD present from the beginning of the experiment. The degree of mitochondrial swelling was assessed 30 min after the addition of C 6 -pyridinium-ceramide. B, respiration of rat liver mitochondria at state 4 was measured as described under "Experimental Procedures." DCCD (100 M) was present from the beginning of the experiments. C 6 -pyridinium-ceramide (40 M) was added 2 min after the addition of mitochondria. For bars 1-4, the data show the respiratory rate 1 min after the addition of C 6 -pyridinium-ceramide. It should be noted that the respiratory rate in the presence of DCCD was linear for at least 8 min. For bar 5, FCCP (1 M) was added 8 min after the addition of C 6 -pyridinium-ceramide. RLM, rat liver mitochondria.
dissipates ⌬⌿, which allows K ϩ to be lost from the matrix according to its electrochemical potential, which, in turn, results in mitochondrial shrinkage.
One of the interesting questions is the structural specificity of C 6 -pyridinium-ceramide action in the induction of mitochondrial permeabilization. Some of the cellular responses and enzymes (e.g. apoptosis (37) and ceramide-activated protein phosphatase (38)) demonstrate a high degree of specificity for ceramide versus dihydroceramide. At the same time, generation of reactive oxygen species by mitochondria appears to lack this specificity (12). Also, a report from Richter and co-workers (17) demonstrates a Ͼ3-fold increase in cytochrome c release from isolated mitochondria under the effect of C 2 -dihydroceramide compared with the control. These considerations are important with regard to ceramide interactions with PTP, which has been implicated by this and previous studies to mediate the ceramide effect on mitochondria. Previous work by Gudz et al. (39) and Walter et al. (40) postulated the presence of low and high affinity binding sites that can nonspecifically interact with a variety of hydrophobic compounds, resulting in PTP opening or closure. The natural effectors of these sites are unknown, but ubiquinones of the respiratory chain (40) or ceramides (18,19) may be good candidates for this role. This could explain our observation that the selectivity for C 6 -pyridinium-ceramide versus C 6 -pyridinium-dihydroceramide is not absolute.
Our observation that suppression of C 6 -pyridinium-ceramide-induced mitochondrial swelling by FCCP also resulted in suppression of cytochrome c release indicates that mitochondrial swelling is a prerequisite for the outer membrane permeability alterations. Even at 40 nmol/mg of protein, a concentration twice that used by Siskind et al. (16), neutral C 6 -ceramide failed to induce considerable cytochrome c release compared with the control. It has been reported that loss of cytochrome c by mitochondria under the effect of C 2 -ceramide is highly dependent on the redox state of this protein, with the oxidized state favoring the release (17). However, we found no substantial release (compared with the control) of cytochrome c by C 6 -ceramide and C 6 -pyridinium-ceramide under conditions in which the respiratory chain downstream of complex III is completely oxidized by the presence of oxidative phosphorylation uncouplers. On the contrary, suppression of cytochrome c release was observed. This provides evidence that, in our experiments, the limiting step in cytochrome c release is not a redox state value, but the formation of a permeability pathway for cytochrome c across the outer membrane. This conclusion fits well with the observation of Kristal and Brown (41), who suggested that, under conditions in which C 2 -ceramide (100 nmol/mg of protein) is unable to induce the permeability transition of the inner membrane, no cytochrome c release is observed. In agreement with these data, in the experiments of Szalai et al. (19), conditions that resulted in C 2 -dependent increase in permeability of the inner membrane were also found to trigger cytochrome c release from mitochondria in a CSA-sensitive manner.
Notably, previous studies suggested that either Ca 2ϩ at 100 -150 M or Bax is required in addition to ceramide to cause permeability change in the outer and inner membranes (18,19). In contrast, in our experiments, C 6 -pyridinium-ceramide by itself induced permeabilization of the mitochondria, or the requirement for Ca 2ϩ was extremely low. (The estimated endogenous Ca 2ϩ concentration is ϳ10 nmol/mg of protein.) This effectiveness of C 6 -pyridinium-ceramide is best explained by its greater accumulation in the mitochondrial matrix. In addition, the low potency of C 2 -pyridinium compared with C 6 -pyridinium-ceramide likely excludes the possibility of a nonspecific effect of the pyridinium group on mitochondrial membranes and underscores the importance of the length of the N-fatty acylsphingosine moiety in mitochondrial permeabilization.
Our results obtained by in vitro experiments indicate that mitochondria are the primary targets for C 6 -pyridinium-ceramide in cell death and that the mechanism of cell death involves disruption of mitochondrial function. Indeed, by confocal microscopy, we observed preferential accumulation of C 6 -pyridinium-ceramide in the mitochondrial compartment, and the relative potency of C 6 -pyridinium-ceramide to induce permeabilization of isolated mitochondria corresponds well with its ability to kill cells. One of the factors that should be kept in mind while considering the effect of ceramide treatment on cell viability is the concentration of ceramide in the vicinity of its target. Electroneutral ceramides redistribute preferentially in the Golgi apparatus (Fig. 2, A and B), which decreases their FIG. 11. C 6 -pyridinium-ceramide-induced large amplitude swelling (B) is accompanied by cytochrome c release (A). Mitochondria were incubated under the conditions described under "Experimental Procedures." C 6 -pyridinium-ceramide (C-6 pyr) or C 6 -ceramide (C-6) at 40 M was added at 2 min, and mitochondria were incubated for an additional 20 min, followed by the addition of CSA (1 M) and EGTA (1 mM) to prevent further permeabilization. Two minutes after the addition of CSA and EGTA, samples were collected and treated for cytochrome c analysis as described under "Experimental Procedures." Alamethicin (7 g/mg of protein) was added as indicated to determine the full degree of permeabilization and maximum cytochrome c release. Where indicated, FCCP (1 M) and antimycin A (0.5 g/mg of protein) were present from the beginning of the experiment. The total amount of cytochrome c is 1.95 g/mg of protein. Data are expressed as the means Ϯ S.E. (n ϭ 3). *, p Ͻ 0.05 versus the control. effective concentration in mitochondria. In contrast, positively charged ceramides are specifically concentrated within their immediate target, the inner mitochondrial membrane, whereas redistribution to other compartments is relatively small (Fig.  2C). This specific redistribution of positively charged ceramide correlates well with its higher potency in cell killing compared with its neutral counterpart. In this way, our results support the hypothesis that the mechanism by which ceramides induce cell killing is permeabilization of the inner mitochondrial membrane with subsequent release of cytochrome c. With respect to the mechanism of pyridinium-ceramide-induced cell death, it should be noted that the permeability alterations of the inner membrane and the subsequent release of cytochrome c observed in isolated mitochondria under the effect of pyridiniumceramide are compatible with both apoptotic and necrotic pathways. Also, the results from 3-(4,5-dimethylthiazol-2-yl)-2,5diphenyltetrazolium bromide assay of cell viability based on measurement of mitochondrial dehydrogenase activities reflect both necrotic and late apoptotic cell death. Therefore, experiments in our laboratory are ongoing to determine the mechanism of pyridinium-ceramide-induced cell death and the particular steps involved in ceramide signal propagation.
Irrespective of the mechanism of cell death, our data suggest that positively charged ceramides could be effective in selective killing of cancer cells. The basis for this selectivity is a substantial difference in ⌬⌿ between normal and tumor cells (42)(43)(44). The difference in ⌬⌿ between carcinoma and control epithelial cells can be Ͼ60 mV higher in carcinoma cells (43,44), a difference that may allow for 10-fold greater accumulation of positively charged ceramides in tumor mitochondria. Thus, future studies are aimed at a better understanding of the nature of molecular targets for ceramide in mitochondria and toward optimization of the molecular structure of positively charged ceramides to increase their accumulation in the mitochondrial matrix. Overall, our results indicate the presence of specific ceramide targets in the mitochondrial matrix, the occupation of which alters permeability of the inner and outer membranes; these findings support a possible novel therapeutic role for positively charged ceramides.