Tumor Necrosis Factor-α Autoregulates Interleukin-6 Synthesis via Activation of Protein Kinase C

We investigated the mechanism of interleukin-6 (IL-6) synthesis induced by tumor necrosis factor-α (TNF) in osteoblast-like MC3T3-E1 cells. TNF stimulated the synthesis of IL-6 dose dependently in the range between 1 and 30 ng/ml. Staurosporine and calphostin C, inhibitors of protein kinase C (PKC), significantly enhanced the TNF-induced synthesis of IL-6. 1-Oleoyl-2-acetylglycerol, a specific activator of PKC, inhibited the TNF-induced IL-6 synthesis. The stimulative effect of TNF was markedly increased in the PKC down-regulated cells. TNF produced diacylglycerol. TNF had little effect on the formation of inositol phosphates and choline. On the contrary, TNF significantly stimulated the formation of phosphocholine dose dependently. D-609, an inhibitor of phosphatidylcholine-specific phospholipase C, suppressed the TNF-induced diacylglycerol production. The TNF-induced IL-6 synthesis was significantly enhanced by D-609. TNF induced sphingomyelin hydrolysis. Neither C2-ceramide nor sphingosine but sphingosine 1-phosphate significantly stimulated the synthesis of IL-6. PKC down-regulation amplified the IL-6 synthesis by sphingosine 1-phosphate. These results strongly suggest that sphingosine 1-phosphate may act as a second messenger for TNF-induced IL-6 synthesis and that TNF autoregulates IL-6 synthesis due to PKC activation via phosphatidylcholine-specific phospholipase C in osteoblast-like cells.

Tumor necrosis factor-␣ (TNF) 1 is a multifunctional cytokine responsible for inflammation, infection, and cancer, and TNF induces numerous physiological effects on a wide variety of cells (1,2). As for intracellular signaling of TNF, it has been reported that TNF stimulates breakdown of sphingomyelin through sphingomyelinase activation, which results in the formation of ceramide (2). It is subsequently metabolized to sphingosine and sphingosine 1-phosphate. Ceramide has been reported to induce apoptosis in several cells, whereas sphingosine and sphingosine 1-phosphate are mitogenic. Accumulating evidence suggests that these sphigomyelin metabolites mediate TNF-induced biological effects (3)(4)(5). In addition, TNF has been shown to catalyze phosphatidylcholine (PC) hydrolysis via activation of PC-specific phospholipase C, resulting in the production of phosphocholine and diacylglycerol (DAG), which is generally recognized to be a physiological activator of protein kinase C (PKC) (6,7). However, the precise intracellular signaling system of TNF has not yet been fully clarified.
Bone metabolism is maintained by two types of functional bone cells, osteoblasts and osteoclasts, which are responsible for bone formation and bone resorption, respectively (8). Osteoclasts activity has been reported to be coupled through cytokines (such as TNF and interleukin-1), stimulation of osteoblasts, and the subsequent production of secondary peptide which activates osteoclasts (9,10). It is well known that TNF is a potent bone resorptive agent (11). Bone resorptive agents such as TNF, parathyroid hormone, interleukin-1, and plateletderived growth factor have been reported to stimulate interleukin-6 (IL-6) production and its secretion in cultured osteoblasts (11)(12)(13)(14). IL-6 is a pleiotropic multifunctional cytokine which regulates diverse cell functions (15,16), and it has been reported that IL-6 stimulates bone resorption and induces osteoclast formation (11,17). Thus, accumulating evidence suggests that IL-6 secreted from osteoblasts plays an important role in bone resorption as a downstream effector of a variety of bone resorptive agents. However, the exact mechanism of IL-6 synthesis in osteoblasts has not yet been clarified.
In the present study, we investigated the mechanism of IL-6 synthesis induced by TNF in osteoblast-like MC3T3-E1 cells. We show here that sphingosine 1-phosphate may act as a second messenger for TNF-induced IL-6 synthesis and that TNF autoregulates IL-6 synthesis due to PKC activation via PC-hydrolyzing phospholipase C in osteoblast-like cells.
Cell Culture-Cloned osteoblast-like MC3T3-E1 cells derived from newborn mouse calvaria (18,19) were generously provided by Dr. M. Kumegawa (Meikai University, Sakado, Japan) and maintained in ␣minimum essential medium (␣-MEM) containing 10% fetal calf serum (FCS) at 37°C in a humidified atmosphere of 5% CO 2 , 95% air. The cells * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
(5 ϫ 10 4 ) were seeded into 35-mm diameter dishes in 2 ml of ␣-MEM containing 10% FCS. After 5 days, the medium was exchanged for 2 ml of ␣-MEM containing 0.3% FCS. The cells were used for experiments after 48 h. In the experiments for the measurement of inositol phosphates, the medium was exchanged for 2 ml of inositol-free ␣-MEM containing 0.3% FCS.
Assay for IL-6 -The cultured cells were stimulated by TNF, C 2ceramide, sphingosine, or sphingosine 1-phosphate in 1 ml of ␣-MEM containing 0.3% FCS for the indicated periods. The conditioned medium was collected, and IL-6 in the medium was measured by an IL-6 enzyme immunoassay kit. When indicated, the cells were pretreated with calphostin C, staurosporine, OAG, or D-609 for 20 min.
Measurement of the Formation of Inositol Phosphates-To determine phosphoinositide-hydrolyzing phospholipase C activity, the cultured cells were labeled with myo-[ 3 H]inositol (3 Ci/dish) for 48 h. The labeled cells were preincubated with 10 mM LiCl for 10 min in 1 ml of an assay buffer (5 mM HEPES, pH 7.4, 150 mM NaCl, 5 mM KCl, 5.5 mM glucose, 0.8 mM MgSO 4 , and 1 mM CaCl 2 ) containing 0.01% bovine serum albumin. The cells were then stimulated by TNF or sodium fluoride at 37°C. The reaction was terminated by adding 1 ml of 30% trichloroacetic acid. The acid supernatant was treated with diethyl ether to remove the acid and neutralized with 0.1 M NaOH. The supernatant was applied to a 1-ml Dowex AG1-X8 column (100 -200 mesh, formate from) as described by Berridge et al. (20,21) with a minor modification (22). The radioactive inositol phosphates were eluted from the column with 8 ml of 0.1 M formic acid containing 1 M ammonium formate.
Measurement of the Formation of Water-soluble Choline Metabolites-To determine PC-hydrolyzing phospholipase C and phospholipase D activities, the cultured cells were labeled with [methyl-3 H]choline chloride (2 Ci/dish) for 24 h. The labeled cells were stimulated by TNF in the assay buffer containing 0.01% bovine serum albumin for the indicated periods. The reaction was terminated by adding 0.75 ml of ice-cold methanol. The dishes were placed on ice for 10 min. The contents were transferred to tubes to which chloroform was added and placed on ice for a further 60 min. Chloroform and water were then added for a final chloroform:methanol:water ratio of 1:1:0.9. The tubes were centrifuged at 14,000 ϫ g for 5 min and the upper aqueous methanolic phase was taken for analysis of the water-soluble cholinecontaining metabolites. The methanolic phase was separated on a 1-ml Dowex 50-WH ϩ column (200 -400 mesh) as described by Cook and Wakelam (23) with a minor modification (24). In brief, the phase was diluted to 5 ml with water and applied to the column. Glycerophosphocholine was removed by 4 ml of water. Phosphocholine was then eluted by 10 ml of water, and choline was eluted with 10 ml of 1 M HCl.
Measurement of DAG Production-The cultured cells were incubated in the assay buffer containing 0.01% bovine serum albumin at 37°C for 20 min, and then stimulated by TNF for the indicated periods. The reaction was terminated by adding 0.75 ml of ice-cold methanol, and the lipids were extracted by the previously described method (25,26). DAG was quantitated using sn-1,2-DAG assay reagents system (Amersham, Japan) essentially according to the procedure of Preiss et al. (27). The radioactive spot corresponding to phosphatidic acid was analyzed by BAS2000 (Fuji, Japan) equipped with imaging plates (28). When indicated, the cells were pretreated with D-609 or propranolol for 20 min.
Assay for Sphingomyelin Turnover-Sphingomyelin levels were measured as described by Okazaki et al. (29). In brief, the cultured cells were labeled with [methyl-3 H]choline chloride (2 Ci/dish) for 24 h. The labeled cells were stimulated by TNF in the assay buffer containing 0.01% bovine serum albumin for the indicated periods. The lipids were extracted by the method of Bligh and Dyer (25). The samples were dried  Determination-The radioactivity of 3 H-labeled samples was determined with a Beckman LS-6500IC liquid scintillation spectrometer.
Statistical Analysis-The data were analyzed by Student's t test and a p Ͻ 0.05 was considered significant. All data are presented as the mean Ϯ S.D. of triplicate determinations.

Effect of TNF on IL-6 Synthesis in MC3T3-E1 Cells-TNF
(30 ng/ml) significantly induced the synthesis of IL-6 in a time-dependent manner up to 48 h (Fig. 1A). The stimulative effect of TNF was dose-dependent in the range between 1 and 30 ng/ml (Fig. 1B). The maximum effect of TNF was observed at 30 ng/ml. Our findings are consistent with a previous report in osteoblasts (11).
Effect of Calphostin C or Staurosporine on Synthesis of IL-6 Induced by TNF in MC3T3-E1 Cells-PKC is well known to play a pivotal role in the regulation of various cellular functions (7). To investigate the role of PKC in the mechanism of TNFinduced IL-6 synthesis in MC3T3-E1 cells, we first examined the effect of calphostin C, a highly potent and specific inhibitor of PKC (30), on the TNF-induced IL-6 synthesis. Calphostin C, which alone had little effect on IL-6 synthesis, significantly enhanced the TNF-induced IL-6 synthesis (Fig. 2). The effect of calphostin C was dose-dependent in the range between 30 nM and 0.3 M. Staurosporine (10 nM), an inhibitor of protein kinases (31), also enhanced the TNF-induced IL-6 synthesis (data not shown). It seems that PKC has an inhibitory effect on TNF-induced IL-6 synthesis.
Effect of OAG on Synthesis of IL-6 Induced by TNF in MC3T3-E1 Cells-We next examined whether the activation of PKC exogenously by addition of cell-permeant DAG would further inhibit the IL-6 synthesis in response to TNF. OAG (0.1 mM), a synthetic DAG known to be a specific activator of PKC (7), which alone had no effect on IL-6 synthesis, inhibited the TNF-induced IL-6 synthesis (Table I). Thus, it is probable that PKC activation suppresses the IL-6 synthesis by TNF.
Effect of Down-regulation of PKC on Synthesis of IL-6 Induced by TNF in MC3T3-E1 Cells-It has been shown that 24 h pretreatment of TPA (0.1 M) down-regulates PKC in osteoblast-like MC3T3-E1 cells (32). We also found that the binding capacity of phorbol-12,13-dibutyrate, a PKC-activating phorbol ester (7), in PKC down-regulated MC3T3-E1 cells is reduced to approximately 30% of the capacity in intact cells (33). To further clarify the role of PKC in the TNF-induced IL-6 synthesis, we next examined the effect of TPA (0.1 M) long-term pretreatment on the IL-6 synthesis stimulated by TNF. The effect of TNF on IL-6 synthesis was significantly enhanced in the PKC down-regulated cells compared with that in the cells without TPA pretreatment (Table II). These findings suggest that TNF activates PKC in MC3T3-E1 cells, and the PKC activation negatively regulates the TNF-induced IL-6 synthesis.
Effect of TNF on Formation of Inositol Phosphates in MC3T3-E1 Cells-We next investigated the intracellular signaling pathway responsible for TNF-activated PKC in MC3T3-E1 cells. It is well known that phosphatidylinositol hydrolysis by phospholipase C results in the formation of DAG. To test the effect of TNF on phosphatidylinositol-specific phospholipase C, we examined whether TNF affects the formation of inositol phosphates. TNF had no effect on the formation of inositol phosphates (control, 2,045 Ϯ 99 cpm; 30 ng/ml TNF, 2,111 Ϯ 106 cpm, as measured after 30 min stimulation). It is well recognized that heterotrimeric GTP-binding proteins are coupled to phosphatidylinositol-specific phospholipase C (34). So, we examined the effect of sodium fluoride, an activator of heterotrimeric GTP-binding proteins (34), on phosphatidylinositol hydrolysis, as a positive control. Sodium fluoride significantly stimulated the formation of inositol phosphates (control, 2,122 Ϯ 101 cpm; 40 mM sodium fluoride, 30,300 Ϯ 887 cpm, as measured after 30 min stimulation). It seems unlikely that TNF induces phosphatidylinositol hydrolysis by phospholipase C.
Effect of TNF on Formation of Water-soluble Choline Metabolites in MC3T3-E1 Cells-It is recognized that PC hydrolysis by PC-specific phospholipase C or D results in the formation of DAG and subsequent activation of PKC (35,36). To clarify the effect of TNF on PC hydrolysis, we examined the effects of TNF on the formations of phosphocholine and choline, the products of PC hydrolysis by phospholipases C and D, respectively. TNF did not affect the formation of choline, but significantly stimulated the formation of phosphocholine (Fig. 3). The stimulative effect of TNF was dose-dependent in the range between 10 and 30 ng/ml, and the maximum effect of TNF was observed at 30 ng/ml (Fig. 3). Thus, these findings suggest that TNF activates not phospholipase D but PC-specific phospholipase C.
Effect of TNF on Production of DAG in MC3T3-E1 Cells-Based on our findings, it is probable that TNF activates PKC through PC hydrolysis by PC-specific phospholipase C in MC3T3-E1 cells. Thus, we examined the effect of TNF on DAG production. TNF (30 ng/ml) time dependently stimulated the production of DAG (Fig. 4A). The effect reached submaximum  within 30 min, and decreased after 60 min. The stimulative effect of TNF was dose-dependent in the range between 1 and 30 ng/ml, and the maximum effect of TNF was observed at 30 ng/ml (Fig. 4B). The pattern of dose-response curve in TNFinduced DAG production was similar to that in TNF-induced IL-6 synthesis. D-609, a specific inhibitor of PC-specific phospholipase C (37), significantly inhibited the production of DAG induced by TNF (Table III). On the contrary, propranolol, an inhibitor of phosphatidic acid phosphohydrolase (38), had little effect on the TNF-induced DAG production (Table III). It is most likely that TNF induces DAG production through PC hydrolysis by phospholipase C, and then activates PKC.
Effect of D-609 on TNF-induced Synthesis of IL-6 in MC3T3-E1 Cells-To clarify the role of PC-specific phospholipase C in TNF-induced IL-6 synthesis, we examined the effect of D-609 on the TNF-induced IL-6 synthesis. D-609, which alone had little effect on IL-6 synthesis, significantly enhanced the synthesis of IL-6 induced by TNF (Fig. 5). The stimulative effect of D-609 was dose-dependent in the range between 0.1 and 0.3 ng/ml. This finding suggests that TNF-induced IL-6 synthesis is negatively regulated by PC-specific phospholipase C activated by TNF itself.
Effect of TNF on Sphingomyelin Turnover in MC3T3-E1 Cells-It has been reported that sphingomyelin hydrolysis takes part in the signaling mechanism of TNF in several types of cells (2). Thus, we examined the effect of TNF on sphingomyelin levels in MC3T3-E1 cells. TNF (30 ng/ml) decreased sphingomyelin levels to 77% of control 30 min after the stim-ulation (Fig. 6). The levels then returned to control levels by 60 min. Our finding indicates that TNF truly induces sphingomyelin hydrolysis in MC3T3-E1 cells.
Effects of C 2 -Ceramide, Sphingosine, and Sphingosine 1-Phosphate on IL-6 Synthesis in MC3T3-E1 Cells-To clarify whether sphingomyelin hydrolysis is involved in TNF-induced IL-6 synthesis, we next examined the effects of sphingomyelin metabolites on IL-6 synthesis. C 2 -ceramide, a cell-permeable ceramide analogue, had no effect on the synthesis of IL-6 in the range between 0.1 and 30 M (Fig. 7). In addition, sphingosine did not stimulate IL-6 synthesis between 0.1 and 30 M (Fig. 7). On the contrary, sphingosine 1-phosphate significantly induced the synthesis of IL-6 (Fig. 7). The stimulative effect was dosedependent in the range between 0.1 and 30 M. The long-term pretreatment with 0.1 M TPA markedly enhanced the IL-6 synthesis by sphingosine 1-phosphate compared with that in intact cells (Table IV). Therefore, it is probable that sphingosine 1-phosphate mediates the stimulative effect of TNF on IL-6 synthesis.

FIG. 4. Effect of TNF on DAG production in MC3T3-E1 cells.
A, the cultured cells were stimulated by 30 ng/ml TNF (q) or vehicle (E) for the indicated periods. B, the cultured cells were stimulated by various doses of TNF for 20 min. Each value represents the mean Ϯ S.D. of triplicate determinations. Similar results were obtained with two additional and different cell preparations. *, p Ͻ 0.05, compared with the value of control.

DISCUSSION
In the present study, we showed that TNF induced the production of DAG in osteoblast-like MC3T3-E1 cells. This finding suggests that TNF activates PKC in MC3T3-E1 cells. In addition, we demonstrated that calphostin C enhanced the TNFinduced synthesis of IL-6. Staurosporine also amplified the IL-6 synthesis by TNF. Thus, it seems that the TNF-induced synthesis of IL-6 is inhibited by PKC, which is activated by TNF itself. We next examined the effect of TNF on IL-6 synthesis in the PKC down-regulated cells. We demonstrated here that the stimulative effect of TNF on IL-6 synthesis was markedly amplified in the PKC down-regulated cells compared with that in intact cells. Additionally, we showed that the PKC activation by OAG inhibited the TNF-induced IL-6 synthesis. Therefore, these results suggest that PKC activated by TNF regulates the overstimulation of IL-6 synthesis by TNF itself in osteoblast-like MC3T3-E1 cells.
We next investigated the exact mechanism behind the TNFinduced activation of PKC in osteoblast-like MC3T3-E1 cells. We first showed that TNF had no effect on the formation of inositol phosphates. It is well known that phosphatidylinositol is hydrolyzed by phospholipase C, resulting in the formation of inositol phosphates and DAG (39). Thus, it seems unlikely that TNF activates phosphatidylinositol-specific phospholipase C. Namely, our finding suggests that TNF does not activate PKC through phosphoinositide hydrolysis. In addition, we demonstrated that TNF did not affect the formation of choline, and that propranolol, did not affect the DAG production induced by TNF in MC3T3-E1 cells. It is recognized that phospholipase D hydrolyzes PC to yield phosphatidic acid and choline (35,36). Phosphatidic acid is subsequently degraded into DAG by phosphatidic acid phosphohydrolase. Thus, it seems unlikely that TNF simulates PC-specific phospholipase D in MC3T3-E1 cells. On the contrary, TNF stimulated the formation of phosphocholine. PC is also hydrolyzed by phospholipase C, resulting in the formation of DAG and phosphocholine (35,36). In addition, we found that D-609 reduced the DAG production by TNF. Therefore, these results suggest that TNF activates PKC via stimulating PC-specific phospholipase C in osteoblast-like MC3T3-E1 cells. Furthermore, we demonstrated here that D-609 enhanced the TNF-induced IL-6 synthesis. Taking our findings into account, it is most likely that TNF regulates IL-6 synthesis due to PKC activation through PC-hydrolyzing phospholipase C in osteoblast-like MC3T3-E1 cells.
We showed here that TNF induced sphingomyelin turnover in osteoblast-like MC3T3-E1 cells and that neither C 2 -ceramide nor sphingosine but sphingosine 1-phosphate significantly stimulated the synthesis of IL-6. It is well known that sphingomyelin hydrolysis results in the production of ceramide. Ceramide can be subsequently metabolized to sphingosine and sphingosine 1-phosphate (1,2). Recent evidence suggests that these sphingomyelin metabolites mediate TNF-induced biological effects (3)(4)(5). Thus, it seems that sphingosine 1-phosphate is involved in the TNF-induced IL-6 synthesis in MC3T3-E1 cells. In addition, we showed that PKC down-regulation significantly enhanced the sphingosine 1-phosphate-induced IL-6 synthesis as well as TNF-induced IL-6 synthesis. Therefore, our findings suggest that TNF exerts the stimulative effect on IL-6 synthesis through sphingosine 1-phosphate in MC3T3-E1 cells, and that PKC acts as a regulator of IL-6 synthesis at a point downstream from sphingosine 1-phosphate.
In conclusion, these results strongly suggest that TNF induces both PC hydrolysis by phospholipase C and sphingomyelin hydrolysis in osteoblast-like cells. Sphingosine 1-phosphate, a metabolite of the latter hydrolysis, may act as a second messenger for TNF-induced IL-6 synthesis, and that PKC ac-  tivation due to the former hydrolysis autoregulates the TNFinduced IL-6 synthesis.