Sphingosine mediates the immediate negative inotropic effects of tumor necrosis factor-alpha in the adult mammalian cardiac myocyte.

To determine whether activation of the neutral sphingomyelinase pathway was responsible for the immediate (<30 min) negative inotropic effects of tumor necrosis factor-α (TNF-α), we examined sphingosine levels in diluent and TNF-α-stimulated cardiac myocytes. TNF-α stimulation of adult feline cardiac myocytes provoked a rapid (<15 min) increase in the hydrolysis of [14C]sphingomyelin in cell-free extracts, as well as an increase in ceramide mass, consistent with cytokine-induced activation of the neutral sphingomyelinase pathway. High performance liquid chromatographic analysis of lipid extracts from TNF-α-stimulated cardiac myocytes showed that TNF-α stimulation produced a rapid (<30 min) increase in free sphingosine levels. Moreover, exogenous D-sphingosine mimicked the effects of TNF-α on intracellular calcium homeostasis, as well as the negative inotropic effects of TNF-α in isolated contracting myocytes; time course studies showed that exogenous D-sphingosine produced abnormalities in cell shortening that were maximal at 5 min. Finally, blocking sphingosine production using an inhibitor of ceramidase, n-oleoylethanolamine, completely abrogated the negative inotropic effects of TNF-α in isolated contracting cardiac myocytes. Additional studies employing biologically active ceramide analogs and sphingosine 1-phosphate suggested that neither the immediate precursor of sphingosine nor the immediate metabolite of sphingosine, respectively, were likely to be responsible for the immediate negative inotropic effects of TNF-α. Thus, these studies suggest that sphingosine mediates the immediate negative inotropic effects of TNF-α in isolated cardiac myocytes.

Tumor necrosis factor-alpha (TNF-␣) 1 is a proinflammatory cytokine that has been implicated as a potential pathogenetic mechanism for cardiac disease states wherein left ventricular dysfunction supervenes, including systemic sepsis (1), acute viral myocarditis (2), cardiac allograft rejection (3), myocardial reperfusion injury (4), and congestive heart failure (5). The long-standing interest in defining the mechanisms responsible for the cardiodepressant effects of TNF-␣ has been intensified recently by experimental studies that have shown that TNF-␣ produces negative inotropic effects in the intact left ventricle (6,7), in thin strips of myocardial tissue (8), and in isolated contracting cardiac myocytes (6). Although the exact cellular signaling pathways that are responsible for the negative inotropic effects of TNF-␣ are not known, a careful inspection of the literature suggests that TNF-␣ modulates myocardial function through at least two different pathways.
It is quite clear, for example, that TNF-␣ can produce immediate negative inotropic effects in myocardial tissue within 10 -30 min (6,8). Similarly, it is equally clear that TNF-␣ exerts delayed effects on myocardial function that appear to be related to uncoupling of the ␤-adrenergic receptor from cyclic AMP, rather than from a direct depression in basal myocardial contractility per se; moreover, these effects occur only after prolonged TNF-␣ exposure (24 -72 h) (9,10). Given the recognition that TNF-␣ increases nitric oxide (NO) levels in myocardial tissue through increased transcription of the inducible Ca 2ϩ -independent form of nitric oxide synthase (NOS) (11,12), given that NO directly mediates myocardial depression (13,14), and given that NO is likely responsible for the uncoupling of the ␤-adrenergic receptor following TNF-␣ stimulation (15), the logical assumption has been that NO mediates the full spectrum of cytokine-induced cardiodepressant effects. However, no previous report to date has provided direct evidence that shows that TNF-␣ stimulates NO production in cardiac myocytes with a time course that is rapid enough to explain the immediate negative inotropic effects of TNF-␣ (6). As a case in point, a recent study in which thin strips of myocardial tissue from Syrian hamsters were treated with TNF-␣ provided indirect evidence that suggested that the immediate (Ͻ5 min) negative inotropic effects "appear(ed) to result from enhanced activity of a constitutive (Ca 2ϩ -dependent) NO synthase enzyme in the myocardium." (8) Nonetheless, although combinations of cytokines may increase Ca 2ϩ -dependent NOS activity indirectly over 24 h by increasing the synthesis of NOS cofactors (16), the demonstration of a rapid increase in Ca 2ϩ -dependent NOS activity by TNF-␣, or by any other cytokine, has not been observed thus far (17). Moreover, we have found that the immediate negative inotropic effects of TNF-␣ were not abrogated by NOS inhibition (6), suggesting that TNF-␣ may produce myocardial depression through a NOS-independent pathway.
During the course of previous studies we observed that TNF-␣-induced activation of the type 1 TNF receptor (TNFR1) resulted in reversible negative inotropic effects in isolated cardiac myocytes as a direct result of alterations in intracellular calcium homeostasis (6,18). Insofar as concentrations of TNF-␣ that produced negative inotropic effects did not produce dis-cernible changes in the voltage-sensitive inward calcium current, we suggested that the TNF-␣-induced alterations in intracellular calcium homeostasis were secondary to alterations in sarcoplasmic reticular handling of calcium (6). Encouraged by the observation that TNF-␣-induced oligomerization of TNFR1 leads to the rapid degradation of sphingomyelin (19) with the resultant generation of a sphingoid base termed sphingosine (20), as well as by the observation that sphingosine was not only present in cardiac and skeletal muscle (21), but was also capable of blocking calcium release from the ryanodine receptor (22,23), we investigated whether the immediate negative inotropic effects of TNF-␣ were mediated by sphingosine. In the present brief report we demonstrate that sphingosine is both necessary and sufficient to produce the negative inotropic effects of TNF-␣ in isolated cardiac myocytes, thus suggesting that TNF-␣-induced activation of the neutral sphingomyelinase pathway is responsible for the immediate negative inotropic effects of this proinflammatory cytokine.

EXPERIMENTAL PROCEDURES
Cell-free Sphingomyelinase Assay-Acidic (pH 5.0) and neutral (pH 7.5) sphingomyelinase activity were measured in adult feline cardiac myocytes according to the method described by Machleidt et al. (24). Briefly, feline cardiac myocytes were isolated, and a 2-ml suspension of cells was plated at a final concentration of 5 ϫ 10 4 cells⅐ml Ϫ1 onto laminin-coated (20 g⅐ml Ϫ1 ) polystyrene Petri dishes as described previously (25,26). On the 1st day in culture the M199 medium was changed, and the cells were treated for 0, 5, 15, 30, and 60 min with diluent (endotoxin-free 0.1% human serum albumin) or with recombinant human TNF-␣ (200 units⅐ml Ϫ1 ). The cells were then lysed, and the resultant cell supernatants were incubated at 37°C with 1 l of [methyl- 14 C]sphingomyelin (25 Ci⅐ml Ϫ1 ). The reaction was stopped after 2 h by the addition of chloroform:methanol (2:1). After thorough mixing by inversion and vortexing, ddH 2 O was added, and the two phases were separated by centrifugation. The upper aqueous phase containing [ 14 C]phosphorylcholine was removed and counted in a liquid scintillation counter. Final results were expressed as fold increase in [ 14 C]phosphorylcholine levels relative to base-line levels.
Ceramide Mass (Diacylglycerol Kinase Assay)-Ceramide mass was measured according to the method of Preiss et al. (27) Briefly, isolated cardiac myocytes were prepared as described above and stimulated for 30 min with diluent, TNF-␣ (200 units⅐ml Ϫ1 ), either in the presence or absence of a specific inhibitor of ceramidase: n-oleoylethanolamine (NOE) (28). The lipid fractions from cell extracts were extracted with chloroform/methanol (1:1 (v/v)), the samples vortexed, and then separated by centrifugation. The upper aqueous phase was aspirated and used for determination of the protein concentration, whereas the chloroform phase was dried in vacuo. The extracts were resuspended and incubated with Escherichia coli diacylglycerol and [␥-32 P]ATP using a commercial diacylglycerol kinase assay system (Amersham Corp.). Ceramide 1-phosphate was then isolated by thin layer chromatography (TLC) using chloroform/methanol/glacial acetic acid (65:15:5 (v/v/v)) in the presence of a ceramide standard, which was run simultaneously along with the unknown samples. Authentic ceramide 1-phosphate was identified by autoradiography, and the spots corresponding to ceramide 1-phosphate were scraped from the plates and then counted in a scintillation counter. Final results were expressed as cpm/mg protein.
Sphingosine Extraction and Derivatization-Sphingosine levels in adult cardiac myocytes were determined according to the method of Merrill et al. (29). Briefly, lipid extracts were obtained from freshly isolated cardiac myocytes by thoroughly mixing the cells with chloroform/methanol (1:2) for 5 min. In order to account for variability in the lipid extraction process, 0.6 nmol of tetradecylamine was added to the unknown samples as an internal standard (30). Equal volumes of chloroform and 1 M NaCl were then added, and the two phases were separated by centrifugation. The upper aqueous phase was discarded, and the chloroform phase was washed twice with 1 M NaCl and vacuumdried for 30 min. The dried lipid extracts were then subjected to a mild alkaline hydrolysis by resuspending them in 0.1 M KOH in methanol for 1 h (37°C), in order to remove ester-containing glycerolipids. After cooling to room temperature the samples were re-extracted as described above, and the free long chain sphingoid bases recovered in the chloroform phase, washed 2 ϫ with 1 M NaCl, and then vacuum-dried. Thereafter, samples were derivatized with o-phthaldialdehyde, exactly as described by Merrill and colleagues (29).
Sphingosine Analysis by Liquid Chromatography-O-Phthaldialdehyde derivatives of the unknown samples were separated by reverse phase HPLC using an isocratic elution with methanol, 5 mM potassium phosphate (pH 7.0) (90:10), employing a Beckman 112 solvent delivery module, a 250 ϫ 4.6 mm C18 (5 m) Brownlee column with a Dynamax C18 guard column; all samples were run at 1 ml/min. Fluorescent derivatives were detected using Waters 420-AC fluorescent detector with excitation and emission wavelengths of 340 and 455 nm, respectively. Under the above conditions, incorporated fluorescence is linearly related to sphingoid base levels between 2 to 400 pmol. To determine the absolute level of sphingosine in unknown samples, the amount of fluorescence was compared with a standard curve generated from known quantities of sphingoid base. To compare the level of sphingosine in diluent and TNF-␣-stimulated samples, we expressed the amount of free sphingosine present in the unknown samples as the ratio of the area under the sphingosine peak to the area under the tetradecylamine peak.
Effect of TNF-␣ on Free Sphingosine Levels-To determine whether stimulation with TNF-␣ would increase the level of free sphingosine in isolated adult cardiac myocytes, freshly isolated cells were treated for 30 min with TNF-␣ (20 and 200 units⅐ml Ϫ1 ), as well as TNF-␣ mutants that bind selectively either to the type 1 (TNFR1) TNF receptor (corresponding mutant ϭ TNFM1) or the type 2 (TNFR2) TNF receptor (corresponding mutant ϭ TNFM2). Both the TNFM1 and TNFM2 mutants were the generous gifts of W. Lesslauer (F. Hoffman-La Roche, Basel, Switzerland) (31). The specificity of the mutated TNF ligands for binding to feline TNFR1 and TNFR2 has been validated previously (18). For these studies we used a single concentration of TNFM1(2 ng⅐ml Ϫ1 ) and TNFM2 (2 ng⅐ml Ϫ1 ), since this concentration is equivalent to the concentration of wild-type TNF-␣ (200 units⅐ml Ϫ1 ) that produces welldefined negative inotropic effects in adult feline cardiac myocytes (6) Isolated Cell Mechanics-Cell motion was characterized by videoedge detection at a stimulation frequency of 0.25 Hz, using experimental conditions identical to those we have described elsewhere (6,32). Four separate series of experiments were performed. First, to determine whether blocking the conversion of ceramide to sphingosine would abrogate the negative inotropic effects of TNF-␣, freshly isolated feline cardiac myocytes were allowed to stabilize for 1 h and were then treated for 30 min at 37°C with 200 units⅐ml Ϫ1 of TNF-␣, either in the presence or absence of a specific inhibitor of ceramidase, n-oleoylethanolamine (NOE) (28). In preliminary control experiments we determined that concentrations of NOE Ͼ10.0 M completely inhibited cell motion; therefore, 1.0 M NOE was chosen as the maximal concentration of NOE to inhibit ceramidase. As an additional control for the above experiments NOE-pretreated cells (1.0 M) were stimulated with 200 units⅐ml Ϫ1 of TNF-␣ and 1 M D-sphingosine for 30 min. Second, to determine whether exogenous D-sphingosine would mimic the negative inotropic effects of TNF-␣, cardiac myocytes were incubated for 30 min with a broad range of concentrations (0.1 nM to 10 M) of D-sphingosine or a closely related sphingoid base, 10 M dihydrosphingosine. To determine the time course for the effects of D-sphingosine (1 M), cell shortening was examined at 5, 15, and 30 min. The reversibility of the effects of D-sphingosine (1 M) was determined by treating the cells for 30 min and then washing the cells free from exogenous D-sphingosine; cell motion was examined 30 min after the D-sphingosine was removed from the cells. Given that sphingosine 1-phosphate, the phosphorylated metabolite of sphingosine, is thought to mediate many of the biological effects of D-sphingosine (33), we performed parallel experiments to those described above for sphingosine, by incubating the cells for 30 min with 0.1-10 M sphingosine 1-phosphate. To determine the time course for the effects of sphingosine 1-phosphate (1 M), cell shortening was examined at 5, 15, 30, and 60 min after stimulation. Third, to determine whether exogenous ceramide would mimic the effects of TNF-␣, the cells were treated for 30 min with two different ceramide analogs: C 2 -ceramide (N-acetyl-D-sphingosine) and C 6 -ceramide (N-hexanoyl-Dsphingosine). In preliminary control experiments we established that concentrations Ն100 M C 2 -ceramide and Ն10 M C 6 -ceramide were overtly toxic to the cells; therefore, we employed 10 M C 2 -ceramide and 1 M C 6 -ceramide for these studies. Fourth, to determine whether TNF-␣-induced stimulation of the phosphatidylcholine-dependent phospholipase C pathway (PC-PLC), with resultant activation of protein kinase C (PKC), was responsible for the immediate negative inotropic effects of TNF-␣, we pretreated the cells with D609 (60 min), an inhibitor of the PC-PLC pathway (34). In preliminary control experiments we determined that concentrations of Յ1 M D609 did not depress cell motion significantly; therefore, we used 0.4 M D609 to inhibit PC-PLC activity. Since the maximal concentration of D609 used may not have inhibited PC-PLC activity completely, we also pretreated the cells with 20 M sangivamycin, which has been shown to significantly inhibit diacylglycerol-dependent protein kinase C (PKC) activity (35); 20 M sangivamycin did not significantly inhibit cell motion in preliminary control experiments. For the studies with D609 and sangivamycin, the cells were pretreated with inhibitor for 60 min and were then stimulated with TNF-␣ or diluent for an additional 30 min prior to assessing cell motion. In addition to these indirect studies, we performed direct measurements of PKC activity in TNF-␣ and sphingosine-stimulated cardiac myocytes. Evidence for cytokine-induced activation of PKC (translocation) was determined in freshly isolated myocytes stimulated for 1, 7.5, or 30 min either with 200 units⅐ml Ϫ1 TNF-␣ using differential membrane centrifugation and immunoblot analysis, as described previously (36); cells stimulated with 100 nM phorbol 12-myristate 13acetate (PMA) served as the appropriate positive controls. Two separate primary antibodies were employed for immunoblotting; the first antibody was specific for PKC⑀, the major PKC isoform in myocytes, whereas the second antibody recognized PKC␣, -␤, -␥, and -␦. The relative amounts of protein kinase C in the membrane and cytosolic fractions were assessed by laser densitometry and expressed as a ratio of membrane-to-cytosolic PKC. Evidence for inhibition of PKC activity was sought by determining the incorporation of [ 32 P]ATP into a PKCspecific pseudosubstrate peptide, as described previously (37). Briefly, cardiac myocytes were homogenized and then centrifuged for 1 h at 1,000 ϫ g; the resultant crude homogenate was then separated into cytosolic and membrane fractions by centrifugation at 70,000 ϫ g. Insofar as preliminary studies showed that Ϸ75% of the PKC activity resided in the cytosolic fraction, this fraction was employed for all further studies. The PKC inhibitory activity of TNF-␣ (200 units⅐ml Ϫ1 ) and sphingosine (0.1-10 M) was compared with that of l00 nM staurosporine in assays wherein myocyte cytosolic extracts were stimulated with mixed micelles containing 0.3 mg/ml phosphatidyl-L-serine and 24 g/ml PMA, in order to simulate PKC activity.
Intracellular Calcium Homeostasis-Intracellular calcium transients were determined in isolated contracting cardiac myocytes using the fluorescent indicator fluo-3 AM (20 M), exactly as we have described previously (6). A time-intensity curve for fluorescence brightness was determined for a single cardiac myocyte contraction by measuring the total fluorescence brightness over the surface area of individual cell; the final time-intensity curves were determined as the average values for 10 consecutive contraction sequences after cell shortening had stabilized. For the purpose of comparison between D-sphingosine (1 M) and diluent-treated cells, peak levels of intracellular fluorescence brightness were compared.
Statistical Analysis-Data are expressed as mean Ϯ S.E. Data were analyzed by one-way analysis of variance, with post hoc testing where appropriate (Dunnett's), or by non-paired t tests.

RESULTS AND DISCUSSION
TNF-␣-induced oligomerization of TNFR1 has been shown to activate membrane-bound neutral sphingomyelinase, with the resultant generation of ceramide, which can then be deacylated to sphingosine by the enzyme ceramidase (20). Three series of experiments were performed to determine whether TNF-␣ activated this pathway in the adult mammalian cardiac myocyte. Table I shows that TNF-␣ stimulation resulted in time-dependent increase in neutral sphingomyelinase activation in adult cardiac myocytes. As shown, [ 14 C]phosphorylcholine levels were significantly different from control values by 15 min (p Ͻ 0.05) and were Ϸ2-fold greater than control values 60 min following stimulation. TNF-␣ stimulation also activated the acidic sphingomyelinase pathway, consistent with reports from other laboratories (34). Next we measured ceramide mass in diluent and TNF-␣-stimulated cardiac myocytes (n ϭ 4 dishes per group). When compared with diluent-treated controls, TNF-␣ stimulation provoked a significant (p Ͻ 0.05) 1.8-fold increase in ceramide mass (10,458 Ϯ 571 versus 18,955 Ϯ 831 cpm/g protein). In contrast, stimulating the cells with diluent in the presence of NOE resulted in Ϸ8-fold increase in ceramide mass (82,998 Ϯ 5,140 cpm/g protein), whereas stimulating the cells with TNF-␣ in the presence of NOE resulted in Ϸ12fold increase in ceramide mass (121,155 Ϯ 24,086 cpm/g protein) Thus, these studies are consistent with the notion that NOE acts, at least in part, by inhibiting ceramidase (28). Fi-nally, to determine whether TNF-␣ activation would stimulate increased levels of free sphingosine in cardiac myocytes, freshly isolated cells were stimulated with TNF-␣ for 30 min before lipid extraction. Fig. 1 shows the elution profiles for the ophthaldialdehyde derivatives of the lipid extracts from cardiac myocytes stimulated with diluent (Fig. 1A) and 200 units⅐ml Ϫ1 of TNF-␣ (Fig. 1B). As shown, a major sphingosine peak was resolved at 14 min in the diluent and TNF-␣-treated myocytes, consistent with previous studies that have demonstrated constitutive levels of sphingosine in cardiac and skeletal muscle (21). The identity of the sphingosine peak was confirmed by its comigration on HPLC with authentic sphingosine standards, as well as by the observation that closely related exogenous lipids (dihydrosphingosine and psychosine) had different elution times when added to the unknown samples. In preliminary control experiments we established that the amount of free sphingosine present in isolated cardiac myocytes was Ϸ23 pmol/10 6 cells, consistent with reports in cardiac and skeletal muscle, as well as other cell types (29,38,39). As shown in Fig.  1 the elution profile for tetradecylamine, which was added as an internal standard, had a retention time of 30 min. In the representative example depicted in Fig. 1B, TNF-␣ stimulation led to a 1.4-fold increase in the sphingosine/TDA ratio when compared with values obtained in diluent-treated cardiac myocytes. Fig. 2 shows two important findings with respect to group data for sphingosine levels in isolated cardiac myocytes. First, stimulating the cells with a concentration of TNF-␣ (200 units⅐ml Ϫ1 ) that consistently depresses cell shortening (6,32) resulted in a significant (p Ͻ 0.05) increase in free sphingosine levels within 30 min, whereas stimulating the cells with a concentration of TNF-␣ (20 units⅐ml Ϫ1 ) that does not depress cell shortening (6) did not stimulate increased (p Ͼ 0.05) levels of free sphingosine levels relative to control values. Moreover, the time course for the TNF-␣-stimulated increase in sphingosine levels was sufficiently rapid to explain the temporal development (i.e. Ͻ 30 min) of TNF-␣-induced contractile dysfunction in isolated cardiac myocytes (6), thin strips of isolated muscle (8), as well as in the intact left ventricle (6). Both the time course and absolute values for the TNF-␣-induced increase in free sphingosine levels in intact myocytes are consistent with previously reported values for sphingosine levels obtained from whole cell lipid extracts following TNF-␣ stimulation (40). A second important point shown by Fig. 2 is that stimulating the myocytes with the TNFM1 ligand produced a significant (p Ͻ 0.05) Ϸ1.4-fold increase in free sphin- TNF-␣ for the times indicated above (n ϭ 5 dishes/time point), and the extent of sphingomyelin hydrolysis was examined under neutral (pH 7.5) and acidic (pH 5.0) conditions (see "Experimental Procedures"). One-way analysis of variance indicated that there were significant overall differences in the extent of activation of the neutral (p Ͻ 0.004) and acidic sphingomyelinase (p Ͻ 0.003) pathways. The base-line values (mean Ϯ S.E.) for [ 14  gosine levels, similar to that which was observed with the wild type TNF-␣. As shown, the TNFM2 ligand, which does not affect cell motion (18), did not increase levels of free sphingosine in isolated contracting cardiac myocytes. The findings obtained with the TNFM1 and TNFM2 ligands are consistent with previous reports that have shown that TNFR1, as opposed to TNFR2, is coupled to the neutral sphingomyelinase pathway (19) and that activation of TNFR1, as opposed to TNFR2, produces negative inotropic effects in isolated contracting cardiac myocytes (18). Next, to determine whether blocking the conversion of ceramide to sphingosine would abrogate the negative inotropic effects of TNF-␣, the cells were pretreated (60 min) with a specific inhibitor of ceramidase: n-oleoylethanolamine (NOE) (28). The important finding shown by Fig. 3 is that the negative inotropic effects of TNF-␣ were abrogated completely by pretreatment with 1.0 M NOE. To confirm that NOE blocked the generation of free sphingosine, we measured free sphingosine levels in TNF-␣-stimulated cells in the presence and absence of NOE. HPLC analysis showed that NOE pretreatment blunted the TNF-␣-induced increase in free sphingosine levels by Ϸ75%. Moreover, when the NOE-pretreated cells were stimulated concurrently with sphingosine and TNF-␣, we observed a significant depression in cell short-ening, suggesting that NOE did not act by interfering with the negative inotropic effects of sphingosine in isolating contracting cardiac myocytes. Fig. 3 also shows that pretreating the cells with C 2 and C 6 -ceramide analogs had no significant effect on cell motion.
To determine whether sphingosine itself was sufficient to mimic the effects of TNF-␣ in isolated cardiac myocyte shortening, the cells were treated with 0.1 nM to 10 M D-sphingosine. The salient finding shown by Fig. 4A is that treating the cells with exogenous D-sphingosine for 30 min resulted in a concentration-dependent decrease in myocyte shortening that was significantly different from control for Ն0.001 M sphingosine. Importantly, the concentrations of exogenous D-sphingosine that were necessary to depress cell shortening fell within the theoretically calculated range for sphingosine levels in TNF-␣-stimulated cardiac myocytes (Ϸ0.1-1.0 M, assuming a cell of density of 1.23 g⅐ml Ϫ1 ) (21,38). To confirm the specificity of the observed effects with D-sphingosine, the cells were treated with dihydrosphingosine, which differs from sphingosine structurally by the absence of a double bond in the carbon 4 -5 position (41). Fig. 4A shows that dihydrosphingosine had no effect on isolated cell shortening, thus arguing against a nonspecific lipid membrane effect of sphingosine. The inset of Fig. 4A shows that the sphingosine-induced (1 M) depression in cell shortening was maximal at Յ5 min, congruent with the overall rapid time course for the development of the immediate negative inotropic effects of TNF-␣ (6,8). Finally, as shown in the inset, the negative inotropic effects of D-sphingosine were shown to be completely reversible within 30 min after D-sphingosine was washed out of the cells (Fig. 4), consistent with previous observations that the immediate negative inotropic effects of TNF-␣ are completely reversible (6,8). Fig. 4B shows that when the cells were pretreated with Ն1 M sphingosine 1-phosphate, the extent of cell shortening was reduced significantly (p Ͻ 0.05) compared with control values. The inset of Fig.  4B shows that the time course for the onset of negative inotropic effects with sphingosine 1-phosphate was delayed relative to that observed with D-sphingosine and was significantly different (p Ͻ 0.05) from control only after 30 min of continuous stimulation. Thus, the time course for the onset of negative inotropic effects in sphingosine 1-phosphate-treated myocytes FIG. 2.). Free sphingosine levels in isolated cardiac myocytes. Freshly isolated cardiac myocytes were treated with diluent (n ϭ 8), 20 units⅐ml Ϫ1 TNF-␣ (n ϭ 3), 200 units⅐ml Ϫ1 TNF-␣ (n ϭ 6), TNFM1 (n ϭ 4), TNFM2 (n ϭ 4). Data are expressed as fold increase in the ratio of (sphingosine/g myocyte protein)/TDA in agonist-stimulated cells compared with the (sphingosine/g myocyte protein)/TDA ratio in diluenttreated cells. The amount of sphingosine present in diluent-treated cells was Ϸ23 pmol/10 6 cells (n refers to the number primary myocyte isolations).
FIG. 1. Elution profiles of sphingosine in diluent and TNF-␣stimulated cardiac myocytes. Cells were treated with diluent or TNF-␣ for 30 min, the lipids extracted, and o-phthaldialdehyde derivatives separated isocratically by reverse phase HPLC (see "Experimental Procedures"). In order to account for variations in the lipid extraction process, tetradecylamine (TDA) was added to the samples as an internal control; free sphingosine (SPH) levels were then normalized by the amount of TDA in the samples. As shown, sphingosine and TDA were eluted with retention times of 14 and 30 min, respectively. The overall efficiency of the lipid extraction process was 35-40%, consistent with previous reports in other cell types (29). is inconsistent with the rapid onset of negative inotropic effects observed with TNF-␣ (Ͻ10 -15 min) (6). Moreover, two additional lines of evidence suggest that the negative inotropic effects of sphingosine are not mediated by sphingosine 1-phosphate. First, if the conversion of sphingosine to sphingosine 1-phosphate is necessary for negative inotropism in cardiac myocytes, then one would predict that exogenous sphingosine 1-phosphate would depress cell shortening at an earlier or at least equivalent time point to that observed with exogenous D-sphingosine, particularly given that exogenous sphingosine and sphingosine 1-phosphate are taken up rapidly by mammalian cells (Ͻ1-5 min) (33,42), and given that both amphipathic molecules share a similar time to onset of action when applied exogenously (42). However, as shown by Fig. 4, A and B, the time course for the onset of action for sphingosine 1-phosphate was Ϸ30 min and was 6-fold slower than was observed when the cells were stimulated with D-sphingosine (5 min). Second, if sphingosine 1-phosphate mediates the negative inotropic effects of sphingosine, then one would predict sphingosine 1-phosphate would be more potent than sphingosine on a molar basis. However, as shown by Fig. 4, A and B, sphingosine 1-phosphate is 1000-fold less potent than sphingosine in terms of producing negative inotropic effects in isolated cardiac myocytes. Thus, while we cannot exclude a potential contributory role for sphingosine 1-phosphate in terms of mediating the negative inotropic effects of TNF-␣, the data do not support a primary role for this molecule.
We next examined the effects of exogenous D-sphingosine on intracellular calcium transients in isolated contracting cardiac myocytes to determine whether sphingosine would mimic the effects of TNF-␣ on intracellular calcium homeostasis (6). Fig.  5 shows representative time-intensity curves for fluorescence brightness in cardiac myocytes treated either with 1 M Dsphingosine or with diluent. As shown, treatment with D-sphingosine produced a striking decrease in the peak levels of intracellular fluorescence brightness, consistent with previous observations that 1 M sphingosine is sufficient to inhibit calcium release by the sarcoplasmic reticular ryanodine receptor (22,23). Similar findings with respect to the effects of sphingosine on intracellular calcium homeostasis have also been observed in neonatal cardiac myocytes (43). The inset of Fig. 5 summarizes the results for the studies, wherein peak fluorescence brightness was examined for groups of diluent and Dsphingosine-treated cells. As shown, there was Ϸ50% decrease (p Ͻ 0.05) in the peak intensity of fluorescence brightness for the D-sphingosine (1 M) -treated cells compared with diluenttreated controls, again consistent with the previous findings from this laboratory that have shown that TNF-␣ suppressed peak intracellular fluorescence brightness by Ϸ40% (6).
In addition to engaging the neutral sphingomyelinase pathway, TNF-␣-induced oligomerization of TNFR1 activates phosphatidylcholine-specific phospholipase C (PC-PLC), with increased activity of diacylglycerol-dependent PKC, as well as the phospholipase A2 pathway, with increased formation of arachidonic acid (44). Previously, we have shown that inhibiting arachidonic acid cyclooxygenase did not abrogate the TNF-␣induced negative inotropic effects (6), suggesting that prostag- , and cell shortening was examined using video edge detection (6). The inset of A shows the time course for the negative inotropic effects of D-sphingosine. For these studies the cells were treated for 5, 15, or 30 min with 1 M D-sphingosine, and cell motion was examined. In addition, the cells were treated with 1 M D-sphingosine for 30 min and then washed to remove the exogenous D-sphingosine; cell motion was determined 30 min after washing the cells. The inset of B shows the time course for the negative inotropic effects for cells treated for 5, 15, 30, and 60 min with sphingosine 1-phosphate. For these studies, cell motion is expressed as the ratio of experimental to control values, in order to facilitate comparisons between myocytes from different isolations. The mean Ϯ S.E. value for cell shortening for the cells treated with diluent was 9.3 Ϯ 0.3%, which is similar to the values we have reported previously (6). (* ϭ p Ͻ 0.05 compared with control.) landins were not responsible for producing the negative inotropic effects of TNF-␣. To determine whether the PC-PLC pathway was important in terms of mediating the negative inotropic effects of TNF-␣, we stimulated isolated contracting cardiac myocytes with TNF-␣ in the presence of specific inhibitors of the PC-PLC and the diacylglycerol-dependent PKC pathways. These studies showed that the negative inotropic effects of TNF-␣ were not abrogated by PC-PLC inhibition with D609 nor by PKC inhibition with sangivamycin: that is TNF-␣-induced a 21.2 Ϯ 2% and 18 Ϯ 6% decrease in myocyte shortening, respectively, in cells pretreated with 0.4 M D609 (p Ͻ 0.002 compared with control; n ϭ 10 cells) and 20 M sangivamycin (p Ͻ 0.009 compared with control; n ϭ 9 cells). We also directly measured PKC activity in TNF-␣ and sphingosine-stimulated cells. In unstimulated myocytes the PKC⑀ membrane-to-cytosol ratio was Ϸ0.4. Stimulating the cells with 100 nM PMA resulted in a rapid (Ͻ1 min) increase in the PKC⑀ membrane-to-cytosol ratio to 0.8, which was maintained at 7.5 min. In contrast, there was no change in the membrane-associated PKC⑀ at any time point up to 30 min for the cells stimulated with 200 units⅐ml Ϫ1 TNF-␣. Identical results were obtained when an antibody that recognized PKC␣, -␤, -␥, and -␦ was used. Insofar as sphingosine has been shown to decrease PKC activity in certain cell types (41), we also measured incorporation of 32 P into a specific PKC pseudosubstrate (n ϭ 6 experiments/group) in PMA-stimulated cytosolic extracts from cells that had been pretreated with TNF-␣ (200 units⅐ml Ϫ1 ) or sphingosine (0.1-10 M). This study showed that there was no significant difference in radiolabeling of the pseudosubstrate in cytosolic extracts pretreated with TNF-␣ (64 Ϯ 4 pmol/min/mg) or with 0.1-10 M sphingosine (76.0 Ϯ 5.6, 72 Ϯ 8.0, 76 Ϯ 6.0 pmol/min/mg, respectively), when compared with the values obtained in control cytosolic extracts (68 Ϯ 3.2 pmol/min/mg). In contrast, there was a significant decrease in PKC activity (p Ͻ 0.05) in the cytosolic extracts that had been pretreated with 100 nM staurosporine (32.0 Ϯ 2 pmol/min/mg). Taken together, these latter studies suggest that neither activation of PKC through the PC-PLC pathway nor inhibition of PKC activity by sphingosine play a major role in mediating the negative inotropic effects of TNF-␣.
In summary, we have provided evidence that shows that TNF-␣ and sphingosine both control the same events in isolated contracting cardiac myocytes: that is alterations in intracellular calcium homeostasis and negative inotropism. We have also provided data that show that the time course for and degree of free sphingosine production following cytokine stimulation is sufficient to completely mimic the negative inotropic effects of TNF-␣ in isolated cardiac myocytes. Finally, we have shown that TNF-␣-induced sphingosine production is necessary for the negative inotropic effects of this cytokine, insofar as blocking sphingosine production through inhibition of ceramidase with NOE abrogates the negative inotropic effects of TNF-␣. In contrast to the findings for sphingosine, we have shown that ceramide, the immediate precursor for sphingosine, is neither necessary nor sufficient to mimic the negative inotropic effects of TNF-␣ and that sphingosine 1-phosphate, the phosphorylated metabolite of sphingosine, is 1000-fold less potent than sphingosine in terms of producing negative inotropic effects; moreover, the delayed onset for the biological effects of sphingosine 1-phosphate in cardiac myocytes is inconsistent with the time course for the negative inotropic effects in TNF-␣. Taken together, the above results provide a rational basis for concluding that sphingosine mediates the immediate negative inotropic effects of TNF-␣. Although we cannot exclude a potential contributory role for NO in terms of modulating the immediate negative inotropic effects of TNF-␣, it bears reem-phasis that TNF-␣ has not yet been shown to produce NO with a sufficiently rapid time course to explain the immediate negative inotropic effects of this cytokine (6). The importance of the above issues notwithstanding, perhaps the more intriguing biological issue that arises from these studies, particularly in view of the pleiotropic nature of sphingosine signaling in mammalian cells (20), is that of understanding what other role(s) sphingosine might play in cytokine-stimulated adult cardiac myocytes.