Tumor Necrosis Factor-α, Sphingomyelinase, and Ceramide Inhibit Store-operated Calcium Entry in Thyroid FRTL-5 Cells*

Tumor necrosis factor α (TNF-α) is a potent inhibitor of proliferation in several cell types, including thyroid FRTL-5 cells. As intracellular free calcium ([Ca2+] i ) is a major signal in activating proliferation, we investigated the effect of TNF-α on calcium fluxes in FRTL-5 cells. TNF-α per se did not modulate resting [Ca2+] i . However, preincubation (10 min) of the cells with 1–100 ng/ml TNF-α decreased the thapsigargin (Tg)-evoked store-operated calcium entry in a concentration-dependent manner. TNF-α did not inhibit the mobilization of sequestered calcium. To investigate whether the effect of TNF-α on calcium entry was mediated via the sphingomyelinase pathway, the cells were pretreated with sphingomyelinase (SMase) prior to stimulation with Tg. SMase inhibited the Tg-evoked calcium entry in a concentration-dependent manner. Furthermore, an inhibition of calcium entry was obtained after preincubation of the cells with the membrane-permeable C2-ceramide and C6-ceramide analogues. The inactive ceramides dihydro-C2 and dihydro-C6 showed only marginal effects. Neither SMase, C2-ceramide, nor C6-ceramide affected the release of sequestered calcium. C2- and C6-ceramide also decreased the ATP-evoked calcium entry, without affecting the release of sequestered calcium. The effect of TNF-α and SMase was inhibited by the kinase inhibitor staurosporin and by the protein kinase C (PKC) inhibitor calphostin C but not by down-regulation of PKC. However, we were unable to measure a significant activation of PKC using TNF-α or C6-ceramide. The effect of TNF-α was not mediated via activation of either c-Jun N-terminal kinase or p38 kinase. We were unable to detect an increase in the ceramide (or sphingosine) content of the cells after stimulation with TNF-α for up to 30 min. Thus, one mechanism of action of TNF-α, SMase, and ceramide on thyroid FRTL-5 cells is to inhibit calcium entry.


Tumor necrosis factor ␣ (TNF-␣) is a potent inhibitor of proliferation in several cell types, including thyroid FRTL-5 cells. As intracellular free calcium ([Ca 2؉ ] i ) is a major signal in activating proliferation, we investigated the effect of TNF-␣ on calcium fluxes in FRTL-5 cells. TNF-␣ per se did not modulate resting [Ca
] i . However, preincubation (10 min) of the cells with 1-100 ng/ml TNF-␣ decreased the thapsigargin (Tg)-evoked store-operated calcium entry in a concentration-dependent manner. TNF-␣ did not inhibit the mobilization of sequestered calcium. To investigate whether the effect of TNF-␣ on calcium entry was mediated via the sphingomyelinase pathway, the cells were pretreated with sphingomyelinase (SMase) prior to stimulation with Tg. SMase inhibited the Tg-evoked calcium entry in a concentration-dependent manner. Furthermore, an inhibition of calcium entry was obtained after preincubation of the cells with the membrane-permeable C 2 -ceramide and C 6 -ceramide analogues. The inactive ceramides dihydro-C 2 and dihydro-C 6 showed only marginal effects. Neither SMase, C 2 -ceramide, nor C 6 -ceramide affected the release of sequestered calcium. C 2 -and C 6 -ceramide also decreased the ATP-evoked calcium entry, without affecting the release of sequestered calcium. The effect of TNF-␣ and SMase was inhibited by the kinase inhibitor staurosporin and by the protein kinase C (PKC) inhibitor calphostin C but not by down-regulation of PKC. However, we were unable to measure a significant activation of PKC using TNF-␣ or C 6 -ceramide. The effect of TNF-␣ was not mediated via activation of either c-Jun N-terminal kinase or p38 kinase. We were unable to detect an increase in the ceramide (or sphingosine) content of the cells after stimulation with TNF-␣ for up to 30 min. Thus, one mechanism of action of TNF-␣, SMase, and ceramide on thyroid FRTL-5 cells is to inhibit calcium entry.
An abundance of reports has shown that the cytokine tumor necrosis factor-␣ (TNF-␣) 1 has diverse effects upon several cell systems. TNF-␣ also potently modulates thyroid function, especially growth and differentiation. In humans, the injection of TNF-␣ decreases serum triiodothyronine and TSH levels (1), whereas in rats and mice TNF-␣ decreases both serum triiodothyronine and thyroxine levels and serum TSH levels (2,3). In human thyroid cells in culture, TNF-␣ decreases the TSHevoked incorporation of 125 I and the secretion of triiodothyronine and thyroxine (4). TNF-␣ attenuates also the production of thyroglobulin and cAMP in these cells (5). In rat FRTL-5 cells, TNF-␣ inhibits the TSH-evoked uptake of iodide and inhibits mitogen-evoked cell proliferation (6 -8). Furthermore, TNF-␣ inhibits the TSH-evoked type I 5Ј-deiodinase activity, the expression of both the thyroid peroxidase gene and the thyroglobulin gene (9 -11), and TSH-evoked hydrogen peroxide production (12).
TNF-␣ binds to two membrane receptors, a 55-and a 75-kDa receptor. Of these two forms, the 55-kDa receptor apparently is the most important (13,14). Binding of TNF-␣ to FRTL-5 cells has also been reported (3), although the receptor types have not been characterized. TNF-␣ activates different sphingomyelinases in cells, resulting in the hydrolysis of sphingomyelin to ceramide and stimulation of the mitogen-activated protein kinase cascade, or the Jun kinase 1 (JNK-1) cascade (15). TNF-␣ may also activate protein kinase C (PKC) (13). The ceramideevoked activation of NFB is probably important in linking the TNF-␣-evoked stimulus to transcriptional activity in the nucleus (16,17). In human papillary thyroid carcinoma cells TNF-␣ has been shown to activate NFB (18), indicating that this signaling pathway also is present in thyroid cells. The type of SMase activated upon stimulation is apparently crucial for the ultimate fate of the cells, as the SMase-evoked production of ceramide may lead to activation of apoptosis (via JNK-1), stimulate proliferation (via mitogen-activated protein kinase), or protection against cytotoxicity (15,19). Of the reported effects of TNF-␣ on FRTL-5 cells, the inhibition of proliferation (20), the inhibition of type I 5Јdeodinase activity (11), and the inhibition of TSH-evoked production of hydrogen peroxidase (12) have also been induced by exogenous ceramide, suggesting that these events are the result of the TNF-␣-evoked activation of a sphingomyelinase and the hydrolysis of sphingomyelin to ceramide.
Other sphingomyelin breakdown products, like sphingosine (SP) and sphingosine 1-phosphate (SPP), potently stimulate proliferation and mobilize sequestered calcium in several cell types (21). These effects of SP and SPP have also been observed in thyroid FRTL-5 cells (22)(23)(24). Recent reports also show that SP attenuated store-operated calcium entry (25,26). Sphingosines and ceramides seem to have mostly opposite effects on cellular proliferation. As both SP and SPP mobilize sequestered calcium and stimulate calcium entry in FRTL-5 cells, two important events in the initiation of proliferation, we thought that it would be of interest to investigate whether ceramides could have any effect on the regulation of calcium fluxes in these cells. Our results showed that, in FRTL-5 cells, TNF-␣, SMase, and ceramides potently attenuated calcium entry. Thus, one mechanism of action of TNF-␣ on thyroid cells is an inhibition of calcium entry.
Cell Culture-Rat thyroid FRTL-5 cells were a generous gift of Dr. Egil Haug (Akers Hospital, Oslo, Norway). The cells were grown in Coon's modified Ham's F-12 medium, supplemented with 5% calf serum, and six hormones (27) (insulin, 10 g/ml; transferrin, 5 g/ml; hydrocortisone, 10 nM; the tripeptide Gly-L-His-L-Lys, 10 ng/ml; TSH, 0.3 milliunits/ml; somatostatin, 10 ng/ml) in a water-saturated atmosphere of 5% CO 2 and 95% air at 37°C. Before an experiment, cells from one donor culture dish were harvested with a 0.2% trypsin solution and plated onto plastic 100-or 35-mm culture dishes. The cells were grown for 7-8 days before an experiment, with 2-3 changes of the culture medium. Fresh medium was always added 24 h prior to an experiment. In the current-clamp experiments, the cells were grown on round coverslips in 24-well culture dishes.
Measurement of [Ca 2ϩ ] i -The medium was aspirated, and the cells were harvested with HEPES-buffered saline solution (HBSS, in millimolar concentrations: NaCl, 118; KCl, 4.6; glucose, 10; CaCl 2 , 1.0; HEPES, 20; pH 7.2) lacking Ca 2ϩ but containing 0.02% EDTA and 0.1% trypsin. After washing the cells three times by pelleting, the cells were incubated with 1 M Fura 2-AM for 30 min at 37°C. Following the loading period, the cells were washed twice with HBSS buffer and incubated for at least 10 min at room temperature and washed once again. Fluorescence was measured with a Hitachi F2000 fluorimeter. The excitation wavelengths were 340 and 380 nm, and emission was measured at 510 nm. The signal was calibrated by addition of 1 mM CaCl 2 and Triton X-100 to obtain maximal fluorescence. Chelating extracellular Ca 2ϩ with 5 mM EGTA and the addition of Tris-base was used to elevate pH above 8.3 to obtain minimal fluorescence. [Ca 2ϩ ] i was calculated as described by Gryenkiewicz et al. (28), using a computer program designed for the fluorimeter with a K d value of 224 nM for Fura 2.
Measurement of Ceramide and Sphingosine Production-Cells grown on 35-mm dishes were labeled with L-[3-3 H]serine (5 Ci/ml) for 48 h in 6H medium. The plates were then washed twice with PBS (in millimolar concentrations; NaCl, 137; KCl, 2.7; Na 2 HPO 4 , 8, KH 2 PO 4 , 1.5; pH 7.4) and incubated for 1 h in serum-free Ham's F-12 medium in a water bath at 37°C. Then TNF-␣ (final concentration 100 ng/ml) was added, and the plates were incubated for 3-30 min. In some experiments SMase (final concentration 100 milliunits/ml diluted in Ham's F-12) was added to the plates, and the plates were incubated for 30 min. After the incubation, the plates were rapidly washed with ice-cold PBS and then frozen. Lipids were extracted by two 20-min incubations in 2 ml of hexane:propanol (3:2, v/v) on a shaker at room temperature. For protein measurements, the proteins were hydrolyzed in 1 ml of 0.1 M NaOH and determined according to Lowry et al. (29). The lipid extracts were pipetted to glass tubes and dried in a gentle stream of air. The dried lipids were dissolved in 70 l of hexane:2-propanol. Sphingomyelin was determined by application of the lipid extract to plastic-backed TLC plates (Whatman). The lipids were separated using chloroform:methanol:concentrated acetic acid:water (50:30:8:3) (30). After detection using iodine vapor, the appropriate bands were cut, and radioactivity was measured in a scintillation counter. Ceramide and sphingosine were separated using high performance TLC plates (Merck), and the lipids were separated by two elutions using chloroform:methanol:2 N NH 4 OH (40:10:1) (31). The plates were allowed to dry between the separate runs. The lipids were detected using iodine vapor, and the appropriate bands were scraped into scintillation vials, and the radioactivity was determined.
Activation of Protein Kinase C-Immediately before exposure of the cells, the 6H medium was removed from the wells and the treatment was started by adding 0H medium containing various concentrations of the substances. Protein kinase C (PKC) activity was measured by the method of Kikkawa et al. (32) and Roskoski (33), with some modifications (34). The experiments were terminated by removing the medium and washing the cells three times with an ice-cold Ca 2ϩ -free salt solution (in millimolar concentrations: NaCl, 145; KCl, 5.2; NaH 2 PO 4 , 1; glucose, 11.2; HEPES, 15; pH 7.4). The cells were scraped from the plates and homogenized by sonication (2 ϫ 15 s) in an ice-cold lysis buffer (containing in millimolar concentrations: EDTA, 2; phenylmethylsulfonyl fluoride, 1; Tris-HCl, 20; pH 7.5, and 50 g/ml leupeptin). Homogenates were centrifuged for 60 min at 100,000 ϫ g at 4°C. The supernatant served as the soluble fraction. The pellet was dispersed into the same buffer containing 0.1% Triton X-100, and the homogenate was incubated on ice for 60 min. The mixture was centrifuged at 100,000 ϫ g for 60 min at 4°C. This supernatant constituted the particulate PKC activity. The protein content in both subcellular fractions was measured according to Bradford (35). In the PKC assay, the final reaction mixture (100 l) contained in millimolar concentrations the following: Tris-HCl, 35; pH 7.5; mM EGTA, 0.25; EDTA, 0.5; MgCl 2 , 6; phenylmethylsulfonyl fluoride, 0.25; PKC-specific substrate peptide FKKSFKL-NH 2 , 34 nM (36); CaCl 2 , 1; and 0.1 mM [ 32 P]ATP (100 -200 cpm/pmol). The mixture also contained leupeptin (12.5 g/ml), phosphatidylserine (PS, 40 g/ml), and diacylglycerol (DAG, 8 g/ml). PKC activity was calculated as the difference in the activity in the presence and absence of CaCl 2 , PS, and DAG. The activity in the absence of CaCl 2 , PS, and DAG was the same as that obtained when only PS and DAG were omitted. The reaction was started by adding protein (0.7-1.5 g). The samples were incubated for 5 min at 30°C, and the reaction was stopped by spotting 25 l of each reaction mixture onto Whatman P81 phosphocellulose paper (1.5 ϫ 1.5 cm). The papers were washed three times in 75 mM phosphoric acid. After air-drying, the radioactivity measured was determined. The results are expressed as nanomoles of inorganic phosphate incorporated to substrate peptide/mg of protein/min.
Immunoblotting of PKC Isoenzymes-SDS-polyacrylamide gel electrophoresis was run using a minigel apparatus (Midget Electrophoresis Unit, Pharmacia, Sweden). Proteins (1 and 3 g per well for soluble and particulate proteins, respectively) were loaded onto 8% polyacrylamide-SDS gels and separated according to molecular weight. The proteins were electrophoretically transferred to methylcellulose membranes. The membranes were incubated three times for 15 min at 45°C in Tween/TBS (TTBS, containing in millimolar concentrations: NaCl, 500; Tris-base, 20; pH 7.5, and 0.1% Tween 20) containing 5% fat-free dry milk and 15 min in TTBS. Then the membranes were incubated for 2 h with 1:4000 -1:80,000 dilution of rabbit polyclonal anti-rat PKC antibodies that recognize ␣, ␤1, ␤2, ␥, ␦, ⑀, and subtypes of PKC (37). A horseradish peroxidase-labeled goat anti-rabbit antibody (Bio-Rad) was used as the secondary antibody, and the immunoreactive bands were visualized by enhanced chemiluminescence (ECL, Amersham Corp., UK). The localization of immunoreactive proteins was compared with those of prestained molecular weight markers (Life Technologies, Inc.).

Electrophoresis and Western Blotting of c-Jun N-terminal Kinase-
The cells were grown on 60-or 100-mm plates as described above. The cells were harvested as above and were allowed to rest for 30 min at 37°C in HBSS. After incubation with TNF-␣ for 1-30 min, the cells were centrifuged and extracted in ice-cold lysis buffer (containing in mM concentrations: NaCl, 100; Na 3 VO 4 , 2; EDTA, 2; Tris-base, 20 mM, pH 8.0; and 3% Nonidet P-40). A sample of the extract was mixed with an equal volume of boiling SDS buffer (glycerol, 20%; 2-mercaptoethanol, 10%; SDS, 4%; bromphenol blue, 0.02%; Tris-base, 0.125 mM, pH 6.8). Equal amounts of protein were separated by SDS-polyacrylamide gel electrophoresis on 10% polyacrylamide gels. The proteins were transferred electrophoretically to nitrocellulose membranes (Schleicher & Schuell, Dassel, Germany). The membrane was incubated with 5% nonfat dry milk for 1 h at room temperature in Tris-buffered saline (TBS, in mM concentrations: NaCl, 500; Tris-base, 20, pH 7.5) to block the remaining binding sites. The blots were incubated with anti-active JNK antibody (1:5000) diluted in TBS containing 5% nonfat dry milk at 4°C overnight. The blots were then incubated with peroxidase-conjugated anti-rabbit antibody (1:10 000) for 2 h at room temperature, and the proteins were detected using the ECL Western blotting detection kit according to the manufacturer's instructions.

Measurement of [ 3 H]Thymidine Incorporation in FRTL-5 Cells-
The cells were plated onto 35-mm dishes and grown in 6H medium for 2-3 days. Then the cells were washed twice with PBS and grown in 0H (Coon's medium without hormones or serum) containing 0.2% bovine serum albumin for 2 days. The medium was then changed to 0H/bovine serum albumin containing the appropriate concentrations of the test compounds and [ 3 H]thymidine (0.4 Ci/ml), and the cells were incubated for 24 h (38). The cells were washed twice with cold PBS solution and once with cold 5% trichloroacetic acid. The trichloroacetic acidinsoluble precipitate was dissolved in 0.1 N NaOH, and the radioactivity was measured by scintillation counting.
Statistics-The results are expressed as the means Ϯ S.E. Statistical analysis was made using Student's t test for paired observations. When three or more means were tested, analysis of variance was used.

TNF-␣, SMase, and Cell-permeable Ceramides Inhibit DNA
Synthesis-Previous studies have shown that TNF-␣ and C 2and C 6 -ceramides potently inhibit both the TSH-and the insulin-evoked incorporation of [ 3 H]thymidine in DNA, i.e. DNA synthesis (6 -8, 20). We confirmed these results and further showed that SMase also inhibited the incorporation of [ 3 H]thymidine in response to TSH and insulin (data not shown). The inactive ceramides dihydro-C 2 and dihydro-C 6 had only marginal effects (data not shown).
TNF-␣ Inhibits Calcium Entry-In FRTL-5 cells, as well as in other cell types, changes in [Ca 2ϩ ] i are probably important events in the initialization of cell proliferation (39,40). We thus investigated the effect of TNF-␣ on [Ca 2ϩ ] i and on the entry of calcium in FRTL-5 cells. TNF-␣ (100 ng/ml, the highest dose tested) did not per se affect [Ca 2ϩ ] i in these cells (data not shown). To avoid any possible effects of TNF-␣ on receptormediated events, we activated calcium entry by stimulating the cells with the Ca 2ϩ ATPase inhibitor thapsigargin (Tg) (41). Tg activates a rapid store-operated calcium entry in FRTL-5 cells (42). Pretreatment of the cells with TNF-␣ for 10 -30 min potently attenuated the Tg-evoked calcium entry in both a calcium-containing buffer and in a calcium-free buffer in a concentration-dependent manner (Fig. 1). We also observed that TNF-␣ did not inhibit the Tg-evoked mobilization of sequestered calcium. In cells pretreated with 100 ng/ml TNF-␣ for 10 min, the Tg-evoked release of intracellular calcium was 178 Ϯ 15 nM, compared with 155 Ϯ 10 nM in control cells. These experiments were performed in a calcium-free buffer to avoid any interference of Tg-evoked calcium entry.
To investigate whether the observed effect of TNF-␣ was due to activation of a sphingomyelinase, we preincubated the cells with the phosphatidylcholine-phospholipase C inhibitor D609. Previous investigations have shown that D609 effectively inhibits TNF-␣-evoked events (16). However, we observed that D609 was a very potent modulator of calcium entry in FRTL-5 cells (data not shown). Furthermore, D609 also mobilized sequestered calcium in our cells (data not shown). Thus, D609 is apparently not a suitable compound for studies using intact cells, as its effects on calcium fluxes probably will affect a multitude of cellular events.
Activation of protein kinases, including PKC, is an important part of the signaling cascade evoked by TNF-␣ (15). We prein-cubated the cells with 200 nM staurosporin for 10 min prior to addition of 100 ng/ml TNF-␣. In these cells, we were unable to detect any TNF-␣-evoked inhibition of calcium entry in cells stimulated with Tg (Fig. 2). We next investigated the effect of the PKC inhibitor calphostin C, and we treated the cells with 100 nM calphostin C for 10 min prior to addition of 100 ng/ml TNF-␣. In these cells, the effect of TNF-␣ on the Tg-evoked increase in [Ca 2ϩ ] i was abolished (Fig. 2). We have also shown earlier that stimulating FRTL-5 cells with the phorbol ester PMA attenuates store-operated calcium entry (42). In the present study, pretreatment with 200 nM PMA significantly decreased the plateau level of [Ca 2ϩ ] i after readdition of calcium to cells stimulated with Tg in a calcium-free buffer (Fig. 2). Pretreatment of the cells with both PMA and 100 ng/ml TNF-␣ decreased both the transient increase in [Ca 2ϩ ] i as well as the new plateau level of [Ca 2ϩ ] i (Fig. 2). In addition, in these cells the increase in the plateau level of [Ca 2ϩ ] i was of lower magnitude than in cells treated with PMA only, suggesting an additive effect of PMA and TNF-␣ on calcium entry (Fig. 2).
SMase and Cell-permeable Ceramides Inhibit Calcium Entry-To investigate further whether the effect of TNF-␣ on calcium entry was mediated via activation of SMase, we pre- incubated the cells with different concentrations of exogenous SMase for 30 min. As shown in Fig. 3, SMase inhibited calcium entry in a concentration-dependent manner very similar to that of TNF-␣. Furthermore, SMase did not affect the amount of sequestered calcium. In cells treated with SMase (1 units/ml) for 30 min, the increase in [Ca 2ϩ ] i evoked by Tg in a calciumfree buffer was 130 Ϯ 23 nM, compared with 155 Ϯ 10 nM in control cells.
In the next series of experiments, the cells were incubated with 200 nM staurosporin for 10 min prior to addition of SMase (1 units/ml for 30 min). In these experiments, pretreatment with staurosporin totally abolished the effect of SMAse (Fig. 4), in a manner similar to what was observed in cells treated with both staurosporin and TNF-␣. To investigate whether the effect of SMase was mediated via activation of PKC, we pretreated the cells with 100 nM calphostin C. In these experiments calphostin C also abolished the effect of SMase (Fig. 4). However, in cells in which PKC was down-regulated by preincubating the cells with 2 M PMA for 24 h, SMase (1 milliunit/ml for 30 min) still attenuated the Tg-evoked calcium entry (Fig. 4).
To investigate whether the observed effect of SMase was due to inhibition of calcium entry, and not due to an enhanced calcium extrusion (43), we tested the effect of SMase on barium entry. As barium is not a substrate for Ca 2ϩ ATPase, barium cannot be extruded from the cells (44). In cells pretreated with SMase (1 units/ml for 30 min), and then stimulated with Tg, the entry of barium was clearly attenuated (Fig. 5). Thus, our results suggest that pretreatment of the cells with SMase indeed resulted in a decreased store-operated entry of calcium.
We then investigated the effect of two membrane-permeable ceramide analogs, C 2 -and C 6 -ceramide. These compounds per se were without any significant effects on basal [Ca 2ϩ ] i levels. As seen in Fig. 6, C 6 -ceramide attenuated Tg-evoked calcium entry in FRTL-5 cells in a concentration-dependent manner. The inactive analogue dihydro-C 6 was without an effect. In addition, neither C 6 -ceramide nor dihydro-C 6  30 M dihydro-C 6 . Furthermore, in PKC down-regulated cells, C 6 -ceramide attenuated the Tg-evoked calcium entry (data not shown). We also tested C 2 -ceramide and obtained a decreased calcium entry in Tg-stimulated cells. However, the effect of C 2 was smaller than that observed with C 6 -ceramide (data not shown). Dihydro-C 2 had no effect on calcium entry (data not shown).
TNF-␣, C 6 -ceramide, and DHC 6 -ceramide and the Activation of PKC-In addition to activating the sphingomyelinase pathway, TNF-␣ may activate PKC (15). Recent studies have shown that FRTL-5 cells express the ␣, ␦, ⑀, and isoforms of PKC (45), and our initial experiments confirmed these findings (data not shown). However, we were unable to show an activation of PKC in cells stimulated with neither TNF-␣, C 6 -ceramide, nor DHC 6 -ceramide (Table I). In control experiments PMA significantly activated PKC (Table I).
Importance of c-Jun N-terminal Kinase and p38 Mitogenactivated Protein Kinase-TNF-␣ may activate JNK (15) and p38 mitogen-activated protein kinase (46) in several cell types. Furthermore, at least in human thyroid cells, JNK may be activated by a PKC-mediated mechanism (47). When our cells were stimulated with TNF-␣ (100 ng/ml), a transient activation of JNK was observed after 1 and 3 min of stimulation (data not shown). This effect was absent in cells pretreated with PMA (1 M for 24 h, data not shown). As the effect of TNF-␣ on Tg- Production of Ceramide and Sphingosine in FRTL-5 Cells in Response to TNF-␣-Stimulating the cells with TNF-␣ (final concentration 100 ng/ml) for up to 30 min did not result in a significant increase in ceramide production, although SMase (100 milliunits/ml) potently increased ceramide content of the cells (Table II). Furthermore, TNF-␣ did not decrease the amount of sphingomyelin in the cells, although this was clearly obtained with SMase (Table II). In addition, we could not see an increase in cellular sphingosine content in response to a 30-min incubation with TNF-␣. Thus, the lack of an effect of TNF-␣ on ceramide production was not the result of a rapid conversion of ceramide to sphingosine (data not shown). A similar lack of an increase in sphingosine content was obtained when the cells were stimulated with SMase (data not shown).
Ceramides Inhibit ATP-evoked Calcium Entry-Previous studies have shown that ATP evokes calcium entry in FRTL-5 cells (48). It was of interest to investigate whether the tested ceramides also could attenuate ATP-evoked calcium entry. Neither C 2 -ceramide nor dihydro-C 2 affected the transient increase in [Ca 2ϩ ] i in response to 100 M ATP (Fig. 7). The ATP-evoked increase in [Ca 2ϩ ] i in control cells was 880 Ϯ 71 nM, in cells treated with 30 M C 2 -ceramide 796 Ϯ 85 nM, and in cells treated with 30 m dihydro-C 2 733 Ϯ 69 nM. However, C 2 -ceramide clearly attenuated the plateau phase of the ATPevoked change in [Ca 2ϩ ] i , i.e. calcium entry. Dihydro-C 2 was without an effect (Fig. 7). A similar lack of an effect of TNF-␣ and C 2 -ceramide on the GTP-evoked transient increase in [Ca 2ϩ ] i has also been reported (12). Furthermore, in cells stimulated with ATP in a calcium-free buffer, the ATP-evoked increase in [Ca 2ϩ ] i was 53 Ϯ 11 nM in control cells, in cells treated with 30 M C 2 -ceramide 55 Ϯ 10 nM, and in cells treated with 30 M dihydro-C 2 the response to ATP was 65 Ϯ 10 nM. Similar results were obtained using C 6 and dihydro-C 6 (data not shown). It is interesting to note that the ATP-evoked receptor-mediated, transient calcium entry (48) was not affected by the tested compounds. DISCUSSION In the present investigation we show that TNF-␣ inhibits store-operated calcium entry in FRTL-5 thyroid cells. The same effect was observed when the cells were treated with SMase and membrane-permeable ceramide derivatives. In a recent observation, Barger et al. (19) showed that TNF-␣ inhibits calcium entry in hippocampal neurons in response to glutamate. However, in that study the importance of SMase or ceramides was not evaluated. Although we were unable to observe an increase in ceramide production after incubating the cells with TNF-␣ for periods relevant for the inhibition of calcium entry, our observation is the first to suggest that TNF-␣, SMase, and ceramides acutely inhibit calcium entry. Another mechanism of action has been shown in osteoblasts. In these cells, several cytokines, including TNF-␣, inhibited a parathyroid hormone-evoked increase in [Ca 2ϩ ] i by abrogating the parathyroid hormone-induced formation of inositol 1,4,5trisphosphate (49). However, this effect required at least 8 h of incubation with TNF-␣.
In thyroid cells, TNF-␣ inhibits an array of different functions. Some of these, like the inhibition of proliferation (20), the inhibition of type I 5Ј-deiodinase (11), and the inhibition of TSH-evoked production of hydrogen peroxide (12), have clearly been shown to be mediated via the production of ceramides. Of these events, at least the activation of proliferation is crucially dependent on intracellular calcium, and especially on calcium entry (39,40). Based on our findings in the present study it is  The results are given as the mean ϩ S.D. of triplicate determinations. The amount of lipids at 30 min in plates treated with vehicle were considered as Control, and the results are given as the percent change in lipid content compared with Control. We could not observe any significant change in lipid content in control plates after 30 min of incubation, compared with plates whose lipids were extracted immediately after the pretreatment period. thus tempting to speculate that the TNF-␣-evoked inhibition of calcium entry is one important mechanism in inhibition of proliferation. Clearly it cannot solely explain the effects of TNF-␣ on FRTL-5 cells, especially as TNF-␣ has cytotoxic effects that probably are mediated via calcium-independent signaling pathways. Furthermore, TNF-␣ inhibits the activation of thyroid peroxidase and the production of thyroid hormones (4, 6 -8, 10, 11). Both the activation of thyroid peroxidase and the production of thyroid hormones involve calciumdependent events in FRTL-5 cells (50 -54). Thus, it is possible that the inhibitory effect of TNF-␣ on these processes also is, at least in part, mediated via inhibition of calcium entry. It is also interesting to note that TNF-␣ may be produced by thyroid epithelial cells (55), suggesting an autocrine function for TNF-␣.
Our results suggest that the mechanism by which TNF-␣ attenuated calcium entry could involve activation of SMase and the production of ceramides. This signaling pathway is usually connected to the binding of TNF-␣ to the p55 TNF-␣ receptor (13,14). Although binding of TNF-␣ to FRTL-5 cells has been shown (3), there presently exists no information on the type of TNF-␣ receptors present in these cells. Our results show that TNF-␣ did not induce a measurable increase in ceramide for up to 30 min of incubation. In a recent study in FRTL-5 cells an effect of TNF-␣ on ceramide production was observed, but the first measurements were made 2.5 h after stimulation (12). Thus, we cannot exclude the possibility that TNF-␣ induced a small or localized increase in ceramide production which we were unable to detect.
We do not yet know how the attenuation of calcium entry occurs in response to stimulation with TNF-␣, SMase, or ceramide. In the recent report by Barger et al. (19), it was suggested that NFB transcription factor may be involved in the TNF-␣-evoked attenuation of calcium entry. However, in their experiments the cells were treated with TNF-␣ for 24 h prior to testing for an inhibition of calcium entry. Our experiments show that TNF-␣ is effective within 10 min of application to the cells and the C 2 -and C 6 -ceramides within a few minutes. Such a rapid activation of NFB has been reported in HL-60 cells (17,56). As TNF-␣ has been reported to activate NFB in human thyroid cells (18), we cannot exclude the possibility that NFB mediates the TNF-␣-evoked inhibition of calcium entry in FRTL-5 cells. In another recent report, a short (1-2 min) preincubation with C 2 -ceramide was also shown to attenuate calcium influx evoked with N-formyl-methionyl-leucyl-phenylalanine (57). The mechanism by which this inhibition was obtained was not established.
Our experiments performed in the presence of staurosporin suggest that a kinase apparently is of importance in mediating the effect of TNF-␣. Some effects of TNF-␣ and sphingomyelinase have been shown to be mediated via activation of PKC (13,58). In FRTL-5 cells, the ␣, ␦, ⑀, and isoforms of PKC have been detected (45). Of these isoforms, ␣, ␦, and ⑀ can be downregulated by PMA, whereas the isoform is insensitive to PMA (45). Previous studies have suggested that the PKC isoform activated by TNF-␣ is the isoform (13,58). This finding could explain why TNF-␣ and SMase were effective in PKC downregulated cells but ineffective in cells treated with calphostin C. This could also explain why the abrogating effect of PMA and TNF-␣ on calcium entry was additive. In other cell types, specific isozymes of PKC regulate calcium entry (59,60). There is, however, a discrepancy between these observations and the fact that we could not measure an activation of PKC with either TNF-␣ or C 6 -ceramide (although PMA did so in control experiments). The PKC experiments might be hampered by the fact that about 60% of the PKC in our cells was already associated with the particulate fraction prior to stimulation, making a small effect of either TNF-␣ or C 6 -ceramide difficult to detect.
We also observed that TNF-␣ evoked a transient activation of JNK, and this effect was absent in cells pretreated with PMA. However, as the effect of TNF-␣ on thapsigargin-evoked calcium entry still occurred in cells pretreated with PMA, it is not likely that JNK is involved in the TNF-␣-evoked attenuation of store-operated calcium entry. Furthermore, as the p38 kinase inhibitor SB203580 did not inhibit the effect of TNF-␣, we think it is unlikely that p38 kinase is mediating the effect of TNF-␣ on store-operated calcium entry.
The effect of ceramide was not due to conversion of ceramide to sphingosine, as we were unable to detect an increase in sphingosine content after stimulating the cells with either TNF-␣ or SMase. However, we have recently shown that the PMA-evoked activation of PKC depolarizes the membrane potential, resulting in decreased calcium entry due to a decreased electrochemical driving force for calcium (42). As the effect of TNF-␣ on store-operated calcium entry was abolished by inhibitors of PKC activity, an effect of TNF-␣ (and ceramide) on membrane potential cannot be excluded. Indeed, preliminary results suggest that TNF-␣ and ceramide evoke a depolarization of the membrane potential and that this effect is attenuated by calphostin C. 2 These observations are consistent with a recent report showing that ceramide depolarizes the membrane potential in oligodendrocytes by inhibiting inwardly rectifying K ϩ channels (61). Furthermore, we cannot exclude the possibility that TNF-␣ also could modulate store-operated calcium channels, especially as a protein kinase has been shown to inhibit the calcium release-activated calcium current (I CRAC ) (62). This possibility appears unlikely as a recent report shows that ceramide does not modulate I CRAC (26). The identification of the steps involved in the TNF-␣/ceramide-evoked signaling pathway will be of crucial importance in understanding the mechanism(s) by which TNF-␣ inhibits calcium entry. Furthermore, this information may help in understanding the mechanisms regulating calcium entry in cells.
In conclusion, we have defined a novel mechanism of action for TNF-␣, i.e. an inhibition of calcium entry. This observation will probably help in understanding the effects of this cytokine (and probably also of SMase and of ceramide) in thyroid cells and in other cell systems.