Clostridium perfringens α-Toxin Activates the Sphingomyelin Metabolism System in Sheep Erythrocytes*

Clostridium perfringens α-toxin induces hemolysis of rabbit erythrocytes through the activation of glycerophospholipid metabolism. Sheep erythrocytes contain large amounts of sphingomyelin (SM) but not phosphatidylcholine. We investigated the relationship between the toxin-induced hemolysis and SM metabolic system in sheep erythrocytes. α-Toxin simultaneously induced hemolysis and a reduction in the levels of SM and formation of ceramide and sphingosine 1-phosphate (S1P). N-Oleoylethanolamine, a ceramidase inhibitor, inhibited the toxin-induced hemolysis and caused ceramide to accumulate in the toxin-treated cells. Furthermore, dl-threo-dihydrosphingosine and B-5354c, isolated from a novel marine bacterium, both sphingosine kinase inhibitors, blocked the toxin-induced hemolysis and production of S1P and caused sphingosine to accumulate. These observations suggest that the toxin-induced activation of the SM metabolic system is closely related to hemolysis. S1P potentiated the toxin-induced hemolysis of saponin-permeabilized erythrocytes but had no effect on that of intact cells. Preincubation of lysated sheep erythrocytes with pertussis toxin blocked the α-toxin-induced formation of ceramide from SM. In addition, incubation of C. botulinum C3 exoenzyme-treated lysates of sheep erythrocytes with α-toxin caused an accumulation of sphingosine and inhibition of the formation of S1P. These observations suggest that the α-toxin-induced hemolysis of sheep erythrocytes is dependent on the activation of the SM metabolic system through GTP-binding proteins, especially the formation of S1P.

traction of isolated rat aorta is linked to the activation of the arachidonic acid cascade via the phosphatidylinositol cycle, especially the production of thromboxane A 2 (4 -6). The toxin causes the hemolysis of various erythrocytes (7). We have reported that it induces the hemolysis of rabbit erythrocytes through the activation of glycerophospholipid metabolism (8).
In addition, we reported that the toxin-induced production of O 2 Ϫ and adhesion to the matrix of rabbit neutrophils are due to the activation of phospholipid metabolism by pertussis toxinsensitive GTP-binding protein (9).
Recently, the activation of the sphingomyelin (SM) cycle, analogous to the glycerophospholipid cycles, has been recognized as a key event in the signal transduction cascade involved in cellular proliferation, differentiation, and apoptosis (10,11). Ceramide causes the arrest of cell growth and apoptosis (10,11). Sphingosine was found to be a potent inhibitor of protein kinase C (12) and inhibited cell growth and induced apoptosis (11). S1P has been reported to promote cell growth and inhibit apoptosis (11,13). Therefore, it is speculated that the conversion of ceramide and sphingosine to S1P plays an important role in apoptotic and survival signaling. Several studies (14 -17) have reported that bacterial SMases hydrolyze cell surface SM leading to increased levels of ceramide, mimicking the effects of the activation of endogenous neutral SMase induced by physiological stimuli. Olivera et al. (18) reported that exogenous SMase induces the synthesis of DNA and potentiates the actions of known growth factors in Swiss 3T3 fibroblasts. The SM metabolites, ceramide and S1P, have been shown to play an important role in such fundamental biological processes (10).
␣-Toxin possesses SMase activity, but little is known about the relationship between this activity and the action of the toxin. Nelson (19) reported that in rabbit erythrocytes, phosphatidylcholine (PC) constitutes ϳ34% of the total amount of phospholipids, whereas in sheep erythrocytes, no PC is detected, and SM accounts for about 50% of all phospholipids. It is known that the ␣-toxin induces hot-cold hemolysis of sheep erythrocytes. Bacillus cereus SMase has also been reported to cause hot-cold hemolysis of sheep erythrocytes (20). In the present study, to clarify the mechanism involved in the ␣-toxininduced hemolysis of sheep erythrocytes, we investigated the effect of the toxin on the SM metabolic system in cells and the relationship between the toxin-induced hemolysis and SM metabolism.
Purification of C. perfringens ␣-Toxin and the Variant Toxin (H148G)-A recombinant form of the plasmid pHY300PLK harboring the structural gene of ␣-toxin was introduced into B. subtilis ISW1214 by transformation (21). The transformants were grown in Luria-Bertani broth containing 50 g of ampicillin per ml, with aeration, to an A 600 of 0.8 -0.85. The culture was centrifuged, and then ammonium sulfate (472 g/liter) was added to the supernatant fluid. The ammonium sulfate fraction obtained was used as the starting material for purifying the toxin. The purification of toxins was performed as described in detail previously (21). Purification of the variant toxin (H148G) from cultures of Bacillus subtilis transformant was performed as described in detail previously (21).
Preparation of Sheep Erythrocytes-Sheep erythrocytes were suspended in 0.02 M Tris-HCl buffer (pH 7.5) containing 0.9% NaCl (TBS) and centrifuged at 1,100 ϫ g for 3 min. The sedimented erythrocytes were then washed three times in the same buffer. The number of erythrocytes was determined with a cell counter (Celltac; Nihon Kohden Co., Tokyo, Japan). The erythrocyte concentration was adjusted to 6 ϫ 10 11 cells per ml.
Treatment of Sheep Erythrocytes with Saponin-Sheep erythrocytes (12 ϫ 10 10 cells per ml) were suspended in TBS and 15-20 g of ␤-escin per ml and incubated at 37°C for 10 min. After the incubation, the erythrocytes were washed by centrifugation (1,100 ϫ g, 3 min) and suspended in TBS. The ␤-escin concentration chosen was below the threshold for hemolysis as determined for each erythrocyte preparation. Under the conditions used, no hemolysis of sheep erythrocytes was observed.
Determination of Hemolytic Activity-␣-Toxin was incubated with washed sheep erythrocytes or saponin-treated erythrocytes (12 ϫ 10 10 cells per ml) in TBS at 37°C for 30 min, and then the cells were chilled at 4°C. Unlysed cells were removed by centrifugation at 1,100 ϫ g for 3 min. The lysis was determined by measuring the amount of hemoglobin released, spectrophotometrically at A 550 . All assay mixtures were supplemented with Ca 2ϩ to 3 mM. Hemolysis was expressed as a percentage of the amount of hemoglobin released from 0.1 ml of erythrocytes suspended in 0.4 ml of 0.4% NaCl.
Thin Layer Chromatography of Sheep Erythrocyte Membrane Phospholipids-Sheep erythrocytes (12 ϫ 10 10 cells per ml) were incubated with various concentrations of ␣-toxin in a total volume of 0.2 ml of TBS containing 3 mM CaCl 2 at 37°C for 30 min. The reaction was terminated by the addition of 0.72 ml of chloroform/methanol (1:2, v/v). The lipids were extracted by the method of Bligh and Dyer (22) except that 0.2 M KCl, 5 mM EDTA was used instead of water (23). The final organic phase was concentrated to 0.1 ml under a N 2 stream and spotted on a Silica Gel 60 plate. The plate was developed in a solvent system consisting of chloroform/methanol/acetic acid/distilled water (25:15:4:2, v/v). After drying, the phospholipids were visualized by exposure to Dittmer-Lester reagent (24) and identified by comigration with standard lipids.
Determination of 1,2-Diacylglycerol in Sheep Erythrocytes-Erythrocyte suspensions (12 ϫ 10 10 cells per ml) were incubated with ␣-toxin at 37°C for 30 min. Reactions were terminated by the addition of chloroform/methanol (1:2, v/v). The lipids were extracted by the method of Bligh and Dyer (22) except that 0.2 M KCl, 5 mM EDTA was used instead of water. The final organic phase was dried under a stream of N 2 and used for analysis of the mass amount of 1,2-diacylglycerol. The 1,2-diacylglycerol content of crude lipid fractions was measured as described by Sakurai et al. (8).
Determination of Inositol 1,4,5-Trisphosphate in Sheep Erythrocytes-Erythrocyte suspensions (12 ϫ 10 10 cells per ml) were incubated with ␣-toxin at 37°C for 30 min, and the reaction was terminated by the addition of ice-cold 10% perchloric acid. The samples were kept on ice for 20 min and then centrifuged at 2,000 ϫ g for 15 min at 4°C. The pH of the supernatant was adjusted to 7.5 with 10 N KOH. The solution was kept on ice for 30 min and centrifuged at 2,000 ϫ g for 15 min at 4°C. The concentration of inositol 1,4,5-trisphosphate (IP 3 ) in the supernatant was measured as described by Sakurai et al. (8).
Determination of Ceramide-Erythrocyte suspensions (12 ϫ 10 10 cells per ml) were incubated with or without ␣-toxin in a total volume of 0.5 ml of 0.02 M Tris-HCl buffer (pH 7.5) containing 3 mM CaCl 2 at 37°C for 30 min. The reaction was terminated by the addition of 1.8 ml of chloroform/methanol (1:2, v/v). The lipids were extracted by the method of Bligh and Dyer (22) except that 0.2 M KCl-5 mM EDTA was used instead of water (23). The final organic phase was dried under a N 2 stream and used for the analysis of ceramide. The ceramide content of the crude lipid fraction was measured based on the conversion of ceramide into [ 32 P]ceramide 1-phosphate by Escherichia coli 1,2-diacylglycerol kinase in the presence of [␥-32 P]ATP (25). The lipids were separated on Silica Gel 60 plates in a solvent system consisting of diisobutyl ketone/acetic acid/distilled water (62:32.5:5.2, v/v). Labeled lipids on the plate were visualized with a Bio-Imaging Analyzer FLA-2000 (Fujifilm Co., Tokyo, Japan). The area corresponding to [ 32 P]ceramide 1-phosphate was scraped into a scintillation vial, to which 5 ml of scintillation mixture was added. Radioactivity was measured in a liquid scintillation counter (Aloka Co., Tokyo, Japan). Known amounts of ceramide were assayed as described above, and a standard curve was drawn. The amount of ceramide in 1 ml of the original sample was calculated from the standard curve.
Determination of Phosphorylcholine-Erythrocyte suspensions (12 ϫ 10 10 cells per ml) or SM-cholesterol liposomes were incubated with various concentrations of ␣-toxin in a total volume of 0.5 ml of 0.02 M Tris-HCl buffer (pH 7.5) containing 3 mM CaCl 2 at 37°C for 30 min. The reaction was terminated by the addition of 1.8 ml of chloroform/methanol (1:2, v/v), and the phases were separated by the addition of 0.49 ml of chloroform and 0.49 ml of 1 M KCl. The upper phase was divided into two tubes, evaporated with a Centrifugal Concentrator CC-181 (Tomy Seiko Co., Tokyo, Japan), and incubated at 37°C for 60 min in a total volume of 0.5 ml of 0.05 M Tris-HCl buffer (pH 9.0) containing 1 mM MgCl 2 and 0.1 mM ZnCl 2 in the presence (phosphatase-treated sample) or absence (untreated sample) of 10 units of phosphatase alkaline per ml. After incubation, 0.24 mM 4-aminoantipyrine, 3.9 units of ascorbate oxidase per ml, 0.77 mM N-ethyl-N-(2-hydroxy-3-sulfopropyl)-3,5-dimethoxyaniline, 2 units of choline oxidase per ml, and 4.2 units of peroxidase per ml were added to the reaction mixture (final volume; 1 ml) and then incubated at 37°C for 5 min. The A 600 of the reaction mixture was measured. Known amounts of phosphorylcholine were assayed as described above, and a standard curve was drawn. The amount of phosphorylcholine produced by treatment with the toxin was obtained by subtracting the value for the untreated sample from the value for the phosphatase-treated sample.
Determination of Sphingosine-Erythrocyte suspensions (12 ϫ 10 10 cells per ml) were incubated with or without dimethyl sulfate in a total volume of 0.5 ml of TBS at 37°C for 60 min. After the incubation, the erythrocytes were washed by centrifugation (1,100 ϫ g, 3 min) and suspended in TBS. The treated erythrocytes (12 ϫ 10 10 cells per ml) were then incubated with ␣-toxin in a total volume of 0.5 ml of TBS at 37°C for various periods. The reaction was terminated by the addition of 1.8 ml of chloroform/methanol (1: 2, v/v). The lipids were extracted by the method of Bligh and Dyer (22) except that 0.2 M KCl, 5 mM EDTA was used instead of water (23). The final organic phase was dried under a N 2 stream and used for the analysis of sphingosine. The sphingosine content of the crude lipid fraction was determined by the method of Van Veldhoven et al. (26). Sphingosine in the crude lipid fraction was converted to C 6 -ceramide (N-hexanol-sphingosine) by N-acylation with hexanoic anhydride. The C 6 -ceramide formed by this reaction was quantitatively phosphorylated using E. coli 1,2-diacylglycerol kinase in the presence of [␥-32 P]ATP. [ 32 P]C 6 -ceramide 1-phosphate was then separated from other phosphorylated products by reverse-phase thin layer chromatography employing chloroform/methanol/acetic acid (10: 85:5, v/v). Labeled lipids on the plate were visualized with a Bio-Imaging Analyzer FLA-2000. The area corresponding to [ 32 P]C 6 -ceramide 1-phosphate was scraped into a scintillation vial, to which 5 ml of scintillation mixture was added. Radioactivity was measured in a liquid scintillation counter. Known amounts of sphingosine were assayed as described above, and a standard curve was drawn. The amount of sphingosine present in 1 ml of the original sample was calculated from the radioactivity of the original sample and the standard curve.
Assay of Sphingosine Kinase Activity-Sphingosine kinase activity in sheep erythrocytes was evaluated using the method described by Yatomi et al. (27) with some modification. Saponin-treated erythrocytes (12 ϫ 10 10 cells per ml) were incubated with ␣-toxin (20 ng per ml) in the presence of 10 Ci of [␥-32 P]ATP per ml and 2.5 M sphingosine in a total volume of 0.5 ml of TBS at 37°C. The reaction was terminated by the addition of 3 ml of chloroform/methanol (1:2, v/v) and 3 g of S1P per ml, followed by thorough mixing and sonication for 10 min. The phases were separated by the addition of 2 ml of chloroform, 2 ml of 1 M KCl, and 0.1 ml of 7 N NH 4 OH, and the resultant upper phase was transferred. The lipids were extracted by the addition of 3 ml of chloroform and 0.2 ml of concentrated HCl. The organic phase was dried under a stream of N 2 and analyzed for the formation of [ 32 P]S1P from sphingosine by thin layer chromatography developed in n-butanol/acetic acid/distilled water (7:1:1, v/v). The plate was exposed to a Fuji-Imaging Plate (BAS-SR-2040) for 30 min. Labeled S1P was visualized with a Bio-Imaging Analyzer FLA-2000. The area corresponding to [ 32 P]S1P was scraped into a vial and measured with a liquid scintillation counter.
Others-Protein kinase C activity was measured using a PepTag non-radioactive protein kinase assay kit (Promega Co., Madison, WI). Multilamellar liposomes composed of SM and cholesterol were prepared according to the method of Nagahama et al. (28). The release of 5(6)carboxyfluorescein diacetate from liposomes was measured by the method of Nagahama et al. (28). Protein concentrations were determined by the method of Lowry et al. (29), using bovine serum albumin as a standard.
Statistical Analysis-All mean values are shown with their calculated standard error. Student's t test was used to determine the significance of differences between controls and experimental groups; a p value of 0.05 or less was considered significant.

The Relationship between ␣-Toxin-induced Hemolysis and Variations in Phospholipids Levels in the Membranes of Sheep
Erythrocytes-When ␣-toxin was incubated with sheep erythrocytes in TBS at 37°C for 30 min and the mixture was then chilled at 4°C for 10 min, the toxin at concentrations exceeding 5 ng/ml caused a dose-dependent increase in hemolysis with a maximum at 20 ng/ml (Fig. 1A). No hemolysis was evident at 37°C, until the mixture had been cooled to 4°C, even when the cells were incubated with 30 ng/ml of the toxin. This result shows that the toxin induced a typical hot-cold hemolysis of sheep erythrocytes, as reported by MacFarlane (30).
To determine whether incubation of sheep erythrocytes with the toxin at 37°C leads to a change in the levels of phospholipids in the cells, the cells were treated with various concentrations of the toxin at 37°C, and then the phospholipids extracted from the cells were analyzed using one-dimensional thin layer chromatography. As shown in Fig. 2, the SM content decreased with an increase in the dose of ␣-toxin, but no effect of the toxin on phosphatidylethanolamine and phosphatidylserine levels was observed under these conditions. As reported by Nelson (19), phosphatidylcholine was not detected in sheep erythrocytes. SM is composed of a phosphorylcholine head group linked to ceramide by a phosphodiester bond. SMase catalyzes the cleavage of this bond and produces ceramide and phosphorylcholine. Thus, ceramide and phosphorylcholine levels were determined in cells treated with various concentrations of the toxin. As shown in Fig. 1B, ␣-toxin at concentrations of 5-20 ng/ml caused a dose-dependent increase in the production of ceramide and phosphorylcholine. These results show that the dose-dependent increase in the toxin-induced hemolysis was identical to that in the toxin-induced formation of ceramide and phosphorylcholine.
We reported that endogenous phosphatidylinositol 4,5bisphosphate (PIP 2 )-specific PLC activated by the toxin is essential for the hemolysis of rabbit erythrocytes, and that the hemolysis is closely related to the generation of phosphatidic acid via the formation of 1,2-diacylglycerol in the cell membranes (8). To clarify whether the toxin stimulates PIP 2 -specific PLC activity in sheep erythrocytes, the production of 1,2diacylglycerol in sheep erythrocytes treated with various concentrations of the toxin was measured. When the erythrocytes were incubated with 20 ng/ml of the toxin at 37°C, no 1,2-diacylglycerol was detected in the cells (data not shown). However, when cell lysates of the erythrocytes were incubated with 20 ng/ml toxin at 37°C, the formation of 1,2-diacylglycerol and IP 3 began simultaneously 20 min after the incubation and peaked after 30 min (data not shown). Treatment of the lysates with 20 ng/ml of the toxin for 30 min stimulated the formation of 1,2-diacylglycerol and IP 3 (about 3.8 and 2.7 pmol/12 ϫ 10 10 cells, respectively), and the toxin at concentrations exceeding 5 ng/ml induced a rise in production in a dose-dependent manner (Fig. 3), suggesting that it activates PIP 2 -specific PLC in the cells. It appears that the concentration of PIP 2 hydrolyzed by this PIP 2 -specific PLC activity is at least 2.7 pmol/12 ϫ 10 10 cells. Thus, the concentration of PC calculated in terms of the amounts of 1,2-diacylglycerol and IP 3 is predicted to be about 1 pmol/12 ϫ 10 10 cells, 10,000 times less than the level of SM calculated in terms of the amount of ceramide (Fig. 1B), if PC exists in the cells. It is likely that the toxin-induced formation of phosphorylcholine is mostly due to the hydrolysis of SM. Therefore, the level of phosphorylcholine was about three times that of ceramide (Fig. 1B), suggesting that ceramide is rapidly converted to other products. Next, we tested if the toxin-induced production of 1,2-diacylglycerol and IP 3 is related to the hemolysis of sheep erythrocytes. Fig. 4 shows that U73122, an endogenous PLC inhibitor, at concentrations of 5-20 M, inhibited the toxin-induced formation of 1,2-diacylglycerol and IP 3 in a dose-dependent manner but not the toxin-induced hemolysis under these conditions. We reported that a variant ␣-toxin (H148G), which loses PLC and SMase activities, binds rabbit erythrocytes but does not lyse the cells, suggesting that the activity of ␣-toxin is essential for the toxin-induced hemolysis of rabbit erythrocytes (21). Guillouard et al. (31) also reported that variant ␣-toxins that lose SMase activity had no hemolytic activity. The variant toxins (H148G) bound to sheep erythrocytes but did not lyse the cells (data not shown). It therefore appears that SMase activity of the toxin is required for hemolysis of sheep erythrocytes.
Effect of Ceramide Metabolism on the Toxin-induced Hemolysis of Sheep Erythrocytes-Ceramide is known to be deamidated to sphingosine by ceramidase. However, little sphingosine (2 nmol/12 ϫ 10 10 cells) was detected in the toxin-treated cells. It therefore is possible that sphingosine is also rapidly metabolized. To clarify the relationship between ␣-toxin-induced hemolysis and the deamidation of ceramide to sphingosine, the effect of N-oleoylethanolamine (32), a ceramidase inhibitor, on the toxin-induced hemolysis and formation of phosphorylcholine and ceramide was investigated (Fig. 5). The ceramide levels in the cells increased from 8 to 13 nmol/ml with an increase in the amount of agent.
Spinedi et al. (33) reported that N-oleoylethanolamine is effective in inhibiting glucosylceramide synthase at micromolar concentrations. Therefore, we examined the effect of Nbutyldeoxynojirimycin and N-(butyl)deoxygalactonojirimycin, glucosylceramide synthase inhibitor, on the toxin-induced hemolysis. N-Butyldeoxynojirimycin (0.025-1 mM) and N-(butyl)deoxygalactonojirimycin (0.025-1 mM) had no effect on the toxin-induced hemolysis (data not shown). These results suggested that N-oleoylethanolamine specifically blocked the toxin-stimulated deamidation of ceramide to sphingosine, resulting in the accumulation of ceramide.
Effect of ␣-Toxin on Hemolysis and Sphingosine Metabolism in Sheep Erythrocytes-Sphingosine is phosphorylated by sphingosine kinase. To test whether or not the phosphorylation of sphingosine is required for the hemolysis induced by the toxin, we investigated the effect of DHS (34) and B-5354c (35), inhibitors of sphingosine kinase, on the toxin-induced hemolysis and production of S1P in sheep erythrocytes. DHS (Fig. 6) and B-5354c (Fig. 7A) blocked the toxin-induced hemolysis in a dose-dependent manner, and at concentrations above 60 and 20 M, respectively, inhibited it by more than 90% relative to the control. Incubation of sheep erythrocytes with the toxin resulted in an increase in sphingosine levels with an increase in the doses of DHS (Fig. 6) and B-5354c (Fig. 7B). The sphingosine levels in the cells sharply increased after exposure to the toxin in the presence of 60 M DHS or 20 M B-5354c and reached a maximum within 5 min of incubation (data not shown). It has been reported that DHS inhibits protein kinase C activity (12). We therefore examined the effect of B-5354c on protein kinase C activity using PepTag protein kinase assays as follows. Cells were preincubated with B-5354c (250 M) and protein kinase C at 37°C for 10 min, and then substrate was added. After 30 min, phosphorylated and non-phosphorylated substrates were examined by agarose gel electrophoresis. The data show that the phosphorylated substrate was present, compared with the substrate in the absence of the blocker, indicating that B-5354c has no effect on protein kinase C (data not shown). However, DHS inhibited protein kinase C under these conditions (data not shown). Furthermore, we investigated the effect of protein kinase C inhibitors (staurosporine and H-7) on the toxin-induced hemolysis of sheep erythrocytes. The hemolysis induced by the toxin (10 ng/ml) was not inhibited by staurosporine (1-50 nM) or H-7 (1-100 M) at all (data not shown). In addition, B-5354c (250 mM) had no effect on the toxin-induced degradation of SM and formation of sphingosine in sheep erythrocytes (data not shown), suggesting that B-5354c selectively inhibits sphingosine kinase in the cells. From these results, it appears that sphingosine is quickly phosphorylated to S1P by sphingosine kinase. Next, the effect of sphingosine on the toxin-induced hemolysis was investigated. Sheep erythrocytes permeabilized by saponin were preincubated with various concentrations of sphingosine (1-5 M) at 37°C for 30 min and then incubated with a sub-hemolytic dose (5 ng/ml) of ␣-toxin at 37°C for 30 min. Sphingosine dose-dependently potentiated the hemolysis induced by the toxin under these conditions (Fig. 8).

Effect of ␣-Toxin on Hemolysis and Sphingosine 1-Phosphate in Sheep
Erythrocytes-To obtain direct evidence that the toxin induces the formation of S1P in sheep erythrocytes, lysated sheep erythrocytes were incubated with the toxin in the presence of [␥-32 P]ATP at 37°C for 30 min. Fig. 9A shows that the toxin induced S1P production in a dose-dependent manner under these conditions. Next, we tested whether or not S1P affects the hemolysis induced by the toxin. The incubation of sheep erythrocytes permeabilized by saponin with 5 ng/ml of the toxin in the presence of various concentrations of S1P promoted increased hemolysis with an increase in the dose of S1P (Fig. 9B). However, S1P had no effect on the toxin-induced hemolysis of intact erythrocytes. Moreover, incubation of the saponin-permeabilized cells or intact cells with S1P in the absence of the toxin resulted in no hemolysis, indicating that S1P itself does not induce hemolysis of sheep erythrocytes permeabilized by saponin and intact erythrocytes. In addition, DHS (60 M) or B-5354c (20 M) inhibited the hemolysis and formation of S1P induced by the toxin (20 ng/ml) (Fig. 7C). However, staurosporine (1-50 nM) and H-7 (1-100 M) had no effect on the toxin (20 ng/ml)-induced S1P formation (data not shown). It has been reported that the biological activities elicited by stimuli depend on S1P released from the cells. However, even when sheep erythrocytes were incubated with the toxin (20 ng/ml) at 37°C for 30 min, no S1P was detected outside the cells (data not shown). It therefore is likely that S1P is not released from the cells treated with the toxin.
Involvement of GTP-binding Proteins in the SM Metabolic Pathway Activated by the Toxin-We reported that the toxininduced hemolysis of rabbit erythrocytes is due to glycerophospholipid metabolism through the activation of GTP-binding protein (36). The effect of GTP␥S on the hemolysis and formation of ceramide and S1P induced by the toxin was investigated. When sheep erythrocytes permeabilized by saponin were incubated with the toxin (5 ng/ml) in the presence of various concentrations (1-5 M) of GTP␥S, GTP␥S dose-dependently stimulated the toxin-induced hemolysis (Fig. 10). Furthermore, when lysates of sheep erythrocytes were incubated with the toxin in the presence of GTP␥S, the formation of ceramide and S1P increased with an increase in the dose of GTP␥S (Fig. 11). These results suggest that the toxin-induced hemolysis and SM metabolism are linked to the activation of GTP-binding proteins. Therefore, we tested the effect of pertussis toxin (37) and C3 exoenzyme (38), in which ADP ribosylates a heterotrimeric GTP-binding protein and a small molecular GTP-binding protein, respectively, on the toxin-activated SM metabolism. To clarify the effect of pertussis toxin on SMase in the cells, the lysates were incubated with pertussis toxin at 37°C for 120 min and then additionally with ␣-toxin at 37°C for 30 min, and the contents of SM and ceramide were analyzed. Fig. 12 shows that the toxin-induced hydrolysis of SM and formation of ceramide were dose-dependently inhibited with increasing concentrations (1-20 g/ml) of pertussis toxin. In particular, incubation of the lysates treated with 20 g/ml of pertussis toxin with ␣-toxin caused little change in the SM content (Fig. 12). Furthermore, pertussis toxin (1-10 g/ml) dose-dependently inhibited S1P production induced by the toxin (data not shown). However, pertussis toxin had no effect on the SMase activity of ␣-toxin (data not shown). These observations show that pertussis toxin specifically blocks the ␣-toxin-stimulated conversion of SM to ceramide, resulting in the inhibition of S1P production. Next, the lysates pretreated with C3 exoenzyme at 0.5-10 g/ml were incubated with 1.0 g/ml ␣-toxin. C3 exoenzyme at concentrations exceeding 1.0 g/ml caused a dramatic accumulation of sphingosine and inhibited the production of S1P, as shown in Fig. 13. However, C3 exoenzyme had no effect on the SMase activity of ␣-toxin (data not shown). These results show that C3 exoenzyme specifically inhibited the ␣-toxin-activated formation of S1P from sphingosine.

DISCUSSION
The present study provides evidence that ␣-toxin-induced hemolysis of sheep erythrocytes is linked to the activation of SM metabolism, especially the formation of S1P. It is known that the enzymatic activities of ␣-toxin are essential for hemolysis (39). We have reported that the PLC activity of the toxin plays a role in the activation of endogenous PIP 2 -specific PLC in rabbit erythrocytes, which stimulates the glycerophospholipid system in the membrane, and that activation of the system leads to hemolysis of the erythrocytes (8). It is known that the N-terminal domain (residues 1-249) of the toxin contains PLC activity but not hemolytic activity (40,41). The PLC from B. cereus is nonhemolytic (42). However, B. cereus SMase induces the hemolysis of sheep erythrocytes (20). It has therefore been speculated that the toxin-induced hemolysis of sheep erythrocytes is closely related to the hydrolysis of SM in the membrane by the SMase activity of ␣-toxin. The present study shows that endogenous SMase activated by the enzymatic activity of ␣-toxin in sheep erythrocytes plays an important role in hemolysis in sheep erythrocytes.
No 1,2-diacylglycerol was detected in intact sheep erythrocytes treated with the toxin. It therefore appears that the toxin does not stimulate hydrolysis of glycerophospholipids in intact cells. The toxin induced the formation of 1,2-diacylglycerol and IP 3 in the lysates of sheep erythrocytes, suggesting that it stimulates and activates PIP 2 -specific PLC activity in the cells. However, U73122, an endogenous PLC inhibitor, inhibited the toxin-induced formation of these products, but not the hemolysis. These observations indicate that the toxin-activated PIP 2specific PLC activity in sheep erythrocytes is independent of the hemolysis.
It has been reported that the catabolic generation of ceramide is mediated by the hydrolysis of SM by SMase, that cera- mide is deamidated to sphingosine by ceramidase, and that sphingosine is phosphorylated to S1P by sphingosine kinase (10). Treatment of sheep erythrocytes with the toxin induced a decrease in SM in the membranes and an increase in ceramide and S1P in the cells, indicating that the toxin activates the SM metabolic system. Moreover, the level of phosphorylcholine markedly increased in the cells treated with the toxin, compared with that of ceramide. Phosphatidylcholine was not detected in sheep erythrocytes as reported by Nelson (19). From our study, it is possible that sheep erythrocytes contain traces of phosphatidylcholine (10,000 times less than the amount of SM). These results show that the phosphorylcholine released is mostly derived from SM under our experimental conditions. The levels of phosphorylcholine released by treatment with the toxin were significantly higher than those of ceramide, suggesting that ceramide is rapidly metabolized to sphingosine. However, a markedly lower level of sphingosine, compared with S1P, was detected, although S1P increased with an increase in the dose of the toxin. It therefore is likely that sphingosine is rapidly metabolized to S1P in the cells treated with the toxin.
N-Oleoylethanolamine (32), a ceramidase inhibitor, significantly inhibited the toxin-induced hemolysis and accumulation of ceramide in sheep erythrocytes treated with the toxin. This agent has also been reported to inhibit glucosylceramide synthase (33). However, the inhibition of glucosylceramide synthase by N-butyldeoxynojirimycin or N-(N-butyl) deoxygalactonojirimycin had no effect on the toxin-induced hemolysis of sheep erythrocytes. These results show that the agent specifically suppresses ceramidase to block hemolysis. Furthermore, DHS (34) and B-5354c (35), sphingosine kinase inhibitors, blocked the toxin-induced hemolysis and production of S1P and caused sphingosine to accumulate. DHS blocked protein kinase C activity, as reported by Hannun et al. (12), but B-5354c did not. Therefore, it appears that as a consequence of the specific blocking of sphingosine kinase activity by these inhibitors, the toxin-induced hemolysis is inhibited. These observations show that the activation of the SM metabolic pathway is essential for the toxin-induced hemolysis, suggesting that metabolites derived from the SM metabolic system are required for the toxininduced hemolysis. In addition, there is evidence that treatment of sheep erythrocytes with protein kinase C inhibitor had no effect on the formation of S1P and hemolysis induced by the toxin. This observation shows that the enzyme is not involved in the hemolysis induced by the toxin.
Several studies have reported that S1P is produced in the cells, secreted from the cells, and binds to EDG family receptors such as EDG-1, EDG-3, and AGR16/H218 on the surface of the cells (43). It has been reported that binding of S1P to the receptors results in the activation of mitogen-activated protein kinase (44,45), inhibition of cAMP production (46,47), and release of Ca 2ϩ (48). Furthermore, it has been reported that S1P stimulates the release of Ca 2ϩ from intracellular stores (49), that it acts intracellularly to regulate the ERK1/2 pathway (45,50), and that it controls mitogenesis (51) and apoptosis (52,53). In addition, inhibition of the ERK1/2 activation in Swiss 3T3 fibroblasts (54) and Ca 2ϩ signals in rat mast cells by DHS has been reported to support a second messenger role for S1P in biological events (55). Accordingly, S1P has been proposed to play a role in intracellular action and extracellular actions (56). However, Pyne and Pyne (43) reported that the key question is whether S1P can function as an intracellular second messenger, because it is possible that S1P is released from cells to act at EDG receptors. In the present study, no S1P was released from erythrocytes treated with a hemolytic dose of the toxin. Moreover, S1P stimulated the toxin-induced hemolysis of saponin-permeabilized erythrocytes but not that of intact erythrocytes, suggesting that intracellular S1P is important for the toxin-induced hemolysis and that the hemolytic effect is not dependent on the action of S1P outside the cells. Our observations indicate that S1P plays an intracellular role on the toxin-induced hemolysis, not an extracellular one, suggesting that it functions as a second messenger in the process. S1P itself caused no hemolysis of saponin-permeabilized cells, suggesting that hemolysis may be induced by a combination of S1P and other events elicited by the toxin in the cells.
GTP␥S stimulated ␣-toxin-induced hemolysis, hydrolysis of SM, and formation of S1P in sheep erythrocytes. It therefore appears that the activation of GTP-binding proteins is required in the toxin-activated SM metabolic system. Alemany et al. (57) reported that the direct activation of heterotrimeric GTP-binding protein increased production of S1P in HL-60 cells, and sphingosine kinase was activated by GTP␥S. Pertussis toxin inhibited the ␣-toxin-induced hydrolysis of SM in the cell ly-sates but had no effect on the SMase activity of ␣-toxin, suggesting that ␣-toxin activates endogenous SMase through pertussis toxin-sensitive G i type GTP-binding protein. Also, C3 exoenzyme inhibited the formation of S1P in the cells treated with the toxin and caused sphingosine to accumulate, indicating that sphingosine kinase is controlled by Rho (small molecular GTP-binding protein) in the cells. Therefore, it appears that Rho activated directly or indirectly by ␣-toxin stimulates sphingosine kinase.
In conclusion, we have shown that activation of the SM metabolic system in sheep erythrocytes, especially the formation of S1P, is involved in ␣-toxin-induced hemolysis. Our observations provide new insights into the molecular pathogenesis of gas gangrene caused by C. perfringens.