A Novel Pathway for Tumor Necrosis Factor-α and Ceramide Signaling Involving Sequential Activation of Tyrosine Kinase, p21 ras , and Phosphatidylinositol 3-Kinase*

Treatment of confluent rat2 fibroblasts with C2-ceramide (N-acetylsphingosine), sphingomyelinase, or tumor necrosis factor-α (TNFα) increased phosphatidylinositol (PI) 3-kinase activity by 3–6-fold after 10 min. This effect of C2-ceramide depended on tyrosine kinase activity and an increase in Ras-GTP levels. Increased PI 3-kinase activity was also accompanied by its translocation to the membrane fraction, increases in tyrosine phosphorylation of the p85 subunit, and physical association with Ras. Activation of PI 3-kinase by TNFα, sphingomyelinase, and C2-ceramide was inhibited by tyrosine kinase inhibitors (genistein and PP1). The stimulation of PI 3-kinase by sphingomyelinase and C2-ceramide was not observed in fibroblasts expressing dominant-negative Ras (N17) and the stimulation by TNFα was decreased by 70%. PI 3-kinase activation by C2-ceramide was not modified by inhibitors of acidic and neutral ceramidases, and it was not observed with the relatively inactive analog, dihydro-C2-ceramide. It is proposed that activation of Ras and PI 3-kinase by ceramide can contribute to signaling effects of TNFα that occur downstream of sphingomyelinase activation and result in increased fibroblasts proliferation.

Ceramides are important lipid second messengers that are generated through sphingomyelin hydrolysis by sphingomyelinases (1). Agonists such as ␥-interferon, TNF␣ (2), 1 interleukin-1 (3), Fas ligand (4), and nerve growth factor (5) activate sphingomyelinases leading to ceramide accumulation. The ability of ceramides to initiate apoptosis in myeloid and lymphoid tumor cell lines as well as in normal lymphocytes is well known (6,7). Depending upon the cell type, however, ceramides can display other effects. Ceramides play an important role in the differentiation of HL-60 cells induced by vitamin D 3 (8), TNF␣, and ␥-interferon (2). Furthermore, TNF␣ and ceramides can cause cell proliferation depending on the target cells (9 -13). For example, ceramides stimulate cell division in confluent quiescent Swiss 3T3 fibroblasts (12,13). TNF␣-induced proliferation of fibroblasts has been implicated in the pathogenesis of diseases such as rheumatoid arthritis (14), neuroma formation after peripheral nerve damage (15), pulmonary fibrosis (16), and chronic intestinal inflammatory disorders such as ulcerative colitis and Crohn's disease (17).
There are several proposed downstream targets for ceramide action including ceramide-activated protein phosphatase (18), ceramide-activated protein kinase (19), and protein kinase C- (20). Additionally, we and others found that ceramides inhibit the agonist-induced activation of phospholipase D (21,22). We recently showed that treatment of 3T3-L1 adipocytes for 12 h with C 2 -ceramide increased the PI 3-kinase activity that was physically associated with IRS-1, and this increased glucose uptake in the absence of insulin (23). These effects of ceramides mimic those of TNF␣, which can increase the tyrosine phosphorylation of IRS-1, its binding of PI 3-kinase (24), the synthesis of GLUT1 (23,25), and the basal uptake of glucose by cells (26). PI 3-kinase phosphorylates the 3-position of the inositol ring to produce a family of 3-phosphoinositides that play important roles in cell signaling. In various cell types, PI 3-kinase is implicated in regulating cell growth and inhibiting apoptosis (27,28), intracellular vesicle trafficking and secretion (29 -31), and cytoskeletal organization (32)(33)(34). PI 3-kinase exists as a heterodimer consisting of a p110 catalytic subunit and a p85 regulatory subunit. The p85 subunit contains two SH2 domains, one SH3 domain, a Bcr homology domain, and prolinerich sequences (for review see Refs. [35][36][37][38], which suggests that PI 3-kinase is regulated by multiple mechanisms. Tyrosine phosphorylated proteins such as the EGF and PDGF receptors, IRS-1 and CD28 bind the SH2 domains of the p85 subunit (35)(36)(37)(38). This binding increases PI 3-kinase activity and, in some situations, causes translocation of PI 3-kinase to plasma membranes to bring it into proximity with its lipid substrates (39). Other possible mechanisms regulating PI 3-kinase activation include binding of the SH3 domains of the Src family of tyrosine kinases to the proline-rich sequences on the p85 subunit (40 -42), interaction of Cdc42 or Rac with the Bcr homology domain (43,44), tyrosine phosphorylation of p85, and autophosphorylation of p85 by the p110 kinase at serine 608 (45). In addition, binding of Ras-GTP to the catalytic domain p110 increases PI 3-kinase activity (46 -48).
In this work we investigated whether ceramides are able to stimulate PI 3-kinase in rat2 fibroblasts. We demonstrated that C 2 -ceramide but not dihydro-C 2 -ceramide activates PI 3-kinase transiently after 5-20 min through a pathway that involves tyrosine kinase activity and the activation of Ras. Treatment of the fibroblasts with TNF␣ or sphingomyelinase for 20 min also stimulated PI 3-kinase activity through activation of tyrosine kinase activity and Ras. This work therefore identifies a novel TNF␣ and ceramide signaling pathway that could contribute to cellular responses such as proliferation of fibroblasts.
[␥-32 P]ATP, [1-14 C]acetic anhydride, anti-rabbit IgG linked to horseradish peroxidase, and enhanced chemiluminescence kit (ECL) were purchased from Amersham Life Science. Sphingomyelinase and [ 3 H]thymidine were from ICN Biomedicals. N-[ 14 C]Acetylsphingosine was synthesized essentially by the method of Ohta et al. (49) and then purified by thin layer chromatography on plates of silica gel G using chloroform/methanol/NH 4 OH/water (80:20:0.5:0 Cell Culture-Rat2 fibroblasts (50) were normally plated at 10,000 cells/cm 2 in 10-cm culture dishes. Fibroblasts were then cultured as described previously (51) for 4 -5 days until confluence. Cells were maintained in DMEM supplemented with 10% fetal bovine serum, 100 units/ml penicillin, and 100 g of streptomycin/ml in a humidified atmosphere of 5% CO 2 , 95% air at 37°C. Rat2 fibroblasts stably expressing wild type Ha-Ras and the dominant-negative mutant Ha-Ras (N17 Ras) were generated by infection with retroviruses containing the respective cDNA sequences and appropriate drug selection using 0.6 mg of Geneticin (G418)/ml or 2.5 g of puromycin/ml, respectively. The N17 Ras containing vector was obtained from Dr. S. E. Egan, and it was subjected to DNA sequence analysis to verify the presence of the N17 mutation (52). N17 Ha-Ras is thought to form a stable inactive complex with Ras exchange factors and thereby prevents activation of endogenous Ras (53). The N17-expressing fibroblasts contained about 28-fold more total Ras than did the Rat2 cells (Fig. 1A). It should be noted that we purposely loaded five times more lysate protein for the parental rat2 and vector control fibroblasts to visualize the relatively low levels of expression of endogenous Ras. The increase in the number of fibroblasts expressing N17 after the third day in culture was about 30% lower than for vector control or parental rat2 fibroblasts (Fig. 1B). The same growth trend was observed for DNA, measured by a fluorescent assay (54), and when analyzed for protein content (results not shown). In these experiments the initial seeding density was about 7,000 cells/cm 2 , and confluence was reached after about 6 days.
Preparation of Cell Membranes-Confluent cells were cultured overnight in DMEM containing 15 M lipid-free bovine serum albumin followed by the addition of ceramide or dihydro-C 2 -ceramide in Me 2 SO (final concentration, 0.1%) as indicated. The final concentration of Me 2 SO was 0.08%. Cells were then washed twice with ice-cold phosphate-buffered saline, harvested by centrifugation, and resuspended in buffer A, which contained 137 mM NaCl, 2.7 mM KCl, 8.1 mM Na 2 HPO 4 , 1.5 mM KH 2 PO 4 , 2.5 mM EDTA, 1 mM dithiothreitol, and 0.1 mM phenylmethylsulfonyl fluoride. Cells were then sonicated twice for 10 s each and then centrifuged for 5 min at 800 ϫ g. After discarding the nuclei and unbroken cells, membranes were prepared by ultracentrifugation at 250,000 ϫ g for 60 min. The pellets (membranes) were washed and resuspended in buffer A.
Immunoprecipitation and Immunoblotting-PI 3-kinase was immunoprecipitated from cell lysates (300 g of protein) by adding 2 g of anti-p85␣ or 5 g of anti-phosphotyrosine antibodies and incubating for 6 h at 4°C with constant gentle rocking followed by adding 40 l of 50% protein A-Sepharose in phosphate-buffered saline. The mixtures were incubated for overnight at 4°C. The immunoprecipitates were then washed three times with 50 mM HEPES, 150 mM NaCl, 100 mM NaF, 1 mM phenylmethylsulfonyl fluoride, 2 mM Na 3 VO 4 , 1% Triton X-100, and 0.1% SDS. Identical amounts of proteins were separated by SDS-polyacrylamide gel electrophoresis as described by Laemmli (55) and were transferred to nitrocellulose membranes. The membranes were blocked overnight with Tris-buffered saline (pH 7.5) containing 1% bovine serum albumin and 1% dry skim milk at 4°C and then incubated with the primary antibody at room temperature. The immunoreactive proteins were detected using anti-rabbit IgG linked to horseradish peroxidase and enhanced chemiluminescence. Western blots were exposed for various length of times to ensure the proportionality of the response and then analyzed densitometrically.
PI 3-Kinase Assay-PI 3-kinase activity was determined by measuring the formation PI 3-[ 32 P]phosphate (56). Confluent cells were washed twice with buffer containing 50 mM HEPES, 137 mM NaCl, 1 mM MgCl 2 , 1 mM CaCl 2 , and 100 M Na 3 VO 4 and then harvested in lysis buffer (1% Nonidet P-40, 10% glycerol, 50 mM HEPES, 137 mM NaCl, 1 mM MgCl 2 , 1 mM CaCl 2 , 2 mM Na 3 VO 4 , 10 mM Na 4 P 2 O 7 , 100 mM NaF, 5 g/ml aprotinin, 1 mM phenylmethylsulfonyl fluoride, and 1 g/ml leupeptin). PI 3-kinase was immunoprecipitated by incubating the cell lysates with 2 g of anti-p85␣, 5 g of anti-phosphotyrosine, or 5 g of anti-pan Ras antibodies for 6 h at 4°C with constant gentle rocking followed by adding 40 l of 50% protein A-Sepharose in phosphatebuffered saline. The mixtures were incubated overnight at 4°C with constant gentle rocking. Immunoprecipitates were washed three times with buffer I (phosphate-buffered saline containing 1% Nonidet P-40 and 100 M Na 3 VO 4 ) and then three times with buffer II (100 mM Tris-HCl, pH 7.5, 500 mM LiCl, and 100 M Na 3 VO 4 ) followed by three times with buffer III (Tris-HCl, pH 7.5, 100 mM NaCl, 1 mM EDTA, and 100 M Na 3 VO 4 ). Immunoprecipitates were then resuspended in 50 l of buffer III. After adding 0.5 mM PI, the samples were incubated for 10 min at room temperature with constant shaking. Then 10 l of 100 mM MgCl 2 and 10 l of 440 M ATP containing 30 Ci of [␥-32 P]ATP were added, and the incubation was continued for 10 min with constant shaking. Reactions were stopped with 20 l of 8 M HCl and 160 l of chloroform/methanol (1:1). Lipids were separated on thin layer silica gel plates (pretreated with 10% (w/v) potassium oxalate) by development with chloroform/methanol/water/NH 4 OH (60:47:11 ration of 32 P into PI 3-phosphate was detected by autoradiography, and the activity was measured by scraping the corresponding region from the plates followed by scintillation counting (23).
Activation of Ras-Ras-GTP was measured by using a nonisotopic method (57). Briefly, fibroblasts overexpressing wild type Ha-Ras were treated with different agents and then lysed in buffer containing 25 mM HEPES, pH 7.5, 150 mM NaCl, 1% Nonidet P-40, 0.25% deoxycholate, 10% glycerol, 25 mM NaF, 10 mM MgCl 2 , and 1 mM EDTA. Cleared lysates were incubated for 30 min at 4°C with 20 -30 g of glutathione S-transferase-Ras binding domain (amino acids 1-149 of cRaf1 fused to glutathione S-transferase) that was bound to glutathione-Sepharose beads. After three washes with lysis buffer, bound proteins were resolved by SDS-polyacrylamide gel electrophoresis on 11% gels, and Ras was detected by immunoblotting with a pan Ras polyclonal antibody. Control experiments confirmed that Ras-GTP does not bind to glutathione-Sepharose beads alone.
Determination of DNA Synthesis and Cell Proliferation-Because rat2 cells are thymidine kinase negative, we first generated a population of rat2 derivatives by infection with a retrovirus vector expressing herpes simplex virus type I thymidine kinase (21). Confluent rat2 fibroblasts containing nuclear thymidine kinase were cultured in DMEM containing 15 M lipid-free bovine serum albumin and 0.1% fetal bovine serum for 24 h. Cells were then incubated for 24 h with TNF␣ or C 2 -ceramide, where indicated, and [ 3 H]thymidine (1 Ci/ml) was present in the last 6 h of the incubation. Cells were then washed, and the incorporation of [ 3 H]thymidine into DNA was measured (21). In a separate cell proliferation assay, fibroblasts were treated with TNF␣ or C 2 -ceramide for 48 h and then released from the dishes by trypsinization, and cell number was determined using a Coulter counter.
Measurement of Ceramidase Activity and Protein Concentration-Ceramidase activity was measured by incubating fibroblasts for 30 min with 40 M N-[ 14 C]acetylsphingosine in DMEM. The cells were washed with ice-cold incubation buffer, and the combined media and resuspended cells were extracted separately with chloroform/methanol/water (final volumes, 1:1:0.9). The distribution of [ 14 C]ceramide and [ 14 C]acetate in the organic and aqueous phases, respectively, were determined by scintillation counting. Protein concentrations were measured by the Bio-Rad or BCA method using bovine serum albumin as a standard (58). DNA concentrations were determined by a fluorescent assay using calf thymus DNA as a standard (54).

C 2 -ceramide Activates PI 3-Kinase in Rat2 Fibroblasts-PI
3-kinase activity in cell lysates was assessed using an immune complex kinase assay that employed an anti-p85 antibody to precipitate the enzyme and using PI as a substrate. Treatment of rat2 fibroblasts with 40 M C 2 -ceramide stimulated PI 3-kinase activity by 4 -6-fold after 10 -20 min, and the activity then decreased to the basal level by 60 min (Fig. 2A). The stimulation at 10 min was evident between 10 and 80 M C 2 -ceramide with a peak at about 40 M (Fig. 2B). It should be noted that these incubations contained 15 M albumin, which increases the amount of ceramide required to produce a biological effect. By comparison, the optimum increase in PI 3-kinase activity obtained after treating cells with 5 ng of PDGF/ml for 10 min was 6 -8-fold (results not shown). PI 3-kinase was also assessed using anti-phosphotyrosine antibodies for precipitation. Under the latter conditions, PI 3-kinase activity in the immunoprecipitate increased by about 4.3-fold after 10 min following treatment with 40 M C 2 -ceramide ( Fig. 2A). By comparison, the use of an optimum dose of PDGF (5 ng/ml) produced about an 8-fold increase after 10 min using this procedure (results not shown). Control experiments showed that similar amounts of precipitated PI 3-kinase protein were used to measure PI 3-kinase activity (Fig. 2C).
We also tested whether PI 3-kinase activation by ceramide is correlated with translocation of the protein to cell membranes. Treatment with ceramide caused an approximately 2-3-fold increase in the amount of the p85 subunit detected in the particulate fraction, indicating translocation from the cytosol (Fig. 3A). Furthermore, when cell lysates were immunoprecipitated with anti-p85 and analyzed by Western blotting with anti-phosphotyrosine antibody, C 2 -ceramide treatment in-creased the tyrosine phosphorylation of p85␣ subunit of PI 3-kinase in a time-dependent manner (Fig. 3B). To demonstrate that these effects were specific, we treated rat2 fibroblasts with dihydro-C 2 -ceramide, which is a relatively inactive analog of C 2 -ceramide. Dihydro-C 2 -ceramide did not cause a significant increase in PI 3-kinase activity (0.7 Ϯ 0.2-fold, mean Ϯ S.E. from six independent experiments), translocation of PI 3-kinase to membranes (Fig. 3A), or tyrosine phosphorylation of PI 3-kinase.
To verify that the activation of PI 3-kinase was caused by C 2 -ceramide itself as opposed to its conversion to sphingosine, we incubated rat2 fibroblasts with [ 14 C]acetylsphingosine and measured its recovery and the production of [ 14 C]acetate. Only 0.047 Ϯ 0.019% (mean Ϯ S.E. from three independent experiments) of the total C 2 -ceramide was converted to [ 14 C]acetate after a 30-min incubation. Rat2 fibroblasts were also pretreated for 20 min with 5 M D-MAPP or 500 M N-oleoylethanolamine to inhibit alkaline and acid ceramidases, respectively (59). The two inhibitors decreased the formation of [ 14 C]acetate by 66 -70% when added separately and by about 80% when added together, demonstrating their efficacies in these experiments. Treatment of rat2 fibroblasts with ceramidases inhibitors alone caused a marginal increase (1.2-1.6-fold) of PI 3-kinase activity, which is compatible with the expected increase in endogenous ceramide concentrations. The ceramidase inhibitors did not significantly affect the activation of PI 3-kinase by C 2 -ceramide after a 10-min incubation (Fig. 4). Taken together, our results demonstrate that the activation of PI 3-kinase is caused by ceramide and not its metabolites.
Activation of PI 3-Kinase by C 2 -ceramide Involves the Stimulation of Ras-p21 ras can activate PI 3-kinase through binding to the p110 catalytic subunit (46 -48). We therefore investigated whether C 2 -ceramide stimulates PI 3-kinase through Ras by using fibroblasts expressing N17 Ras, which interferes with activation of endogenous Ras by exchange factors (53).   (Fig. 5A). We also tested the effects N17 Ras on the activation of PI 3-kinase by EGF and PDGF. The expression of N17 Ras blocked the activation of PI 3-kinase by 5 ng/ml PDGF by about 80%. By contrast, there was no significant effect of N17 Ras expression on PI 3-kinase activation that was produced by 100 ng/ml of EGF (Fig. 5A). Our results therefore implicate Ras-GTP in the ceramide-induced activation of PI 3-kinase. This conclusion was confirmed in experiments that used rat2 cells that overexpressed wild type Ha-Ras. C 2 -ceramide increased the amount of Ras-GTP in a timedependent manner with a peak (2-3-fold) after a 10-min incubation (Fig. 5B). Treatment of cells with the relatively inactive dihydro-C 2 -ceramide (40 M) for 10 min did not significantly increase Ras-GTP levels (Fig. 5B). As a positive control, treatment of cells with 100 ng of EGF/ml for 5 min caused a 10.3 Ϯ 1.9-fold increase in Ras-GTP levels.
We also tested whether C 2 -ceramide would increase the amount of p21 ras physically associated with PI 3-kinase in rat2 fibroblasts stably expressing wild type Ha-Ras. PI 3-kinase was immunoprecipitated from cell lysates with anti-p85 antibodies, and the precipitate was resolved by SDS-polyacrylamide gel electrophoresis and then immunoblotted with anti-pan Ras antibodies. Treatment with C 2 -ceramide but not dihydro-C 2ceramide increased the co-precipitation of Ras with PI 3-kinase by about 8-fold after 5 min (Fig. 5C). By comparison, treatment with 100 ng of EGF/ml for 5 min caused a 27-fold increase in this assay (results not shown). These results prompted us to assay the PI 3-kinase activity in anti-pan Ras immunoprecipitates. Treatment of rat2 fibroblasts overexpressing wild type Ha-Ras with 40 M C 2 -ceramide increased the PI 3-kinase activity that co-precipitated with Ras by about 3-fold at 20 min (Fig. 5D). The relatively inactive analog, dihydro-C 2 -ceramide (40 M), had no significant effect on the association of Ras with PI 3-kinase (results not shown). In control experiments, no PI 3-kinase activity was associated with beads in the absence of anti-Ras antibody.

Activation of PI 3-Kinase by C 2 -ceramide Involves the Stimulation of Tyrosine Kinases-Protein tyrosine phosphorylation
could be involved at a number of levels in ceramide-induced activation of PI 3-kinase For example, PI 3-kinase is activated through the binding of SH2 domains on the p85 subunit to tyrosine phosphorylated receptor or nonreceptor proteins (35)(36)(37)(38). Therefore, we tested the effect of tyrosine kinase inhibitors on the activation of PI 3-kinase by C 2 -ceramide. Rat2 fibroblasts were pretreated with 50 M genistein or 1 M PP1 for 1 h before adding 40 M C 2 -ceramide for 10 min. These tyrosine kinase inhibitors blocked PI 3-kinase activation by C 2 -ceramide (Fig. 6A). C 2 -ceramide treatment also increased protein tyrosine phosphorylation as indicated in Fig. 6B including a protein of approximately 120 kDa. We therefore determined whether this protein was FAK because this has been implicated in activation of PI 3-kinase (60). C 2 -ceramide-and TNF␣induced tyrosine phosphorylation of FAK in a time-dependent manner. Dihydro-C 2 -ceramide produced no significant increases in the tyrosine phosphorylation of FAK (Fig. 6C).
Treatment of Fibroblasts with TNF␣ or Sphingomyelinase Activates PI 3-Kinase-Many signaling responses of TNF␣ can be mimicked by treating cells with cell-permeable ceramides or sphingomyelinase, which generates long chain ceramides (61). We therefore tested whether TNF␣ and sphingomyelinase would also activate PI 3-kinase in rat2 fibroblasts. Treatment of rat2 fibroblasts with 10 ng of TNF␣/ml or with 0.1 units of sphingomyelinase/ml increased PI 3-kinase activity after about 20 min (Fig. 7A). Pretreatment of rat2 fibroblasts with the tyrosine kinase inhibitors genistein (50 M) or PP1 (1 M) decreased the stimulation of PI 3-kinase by TNF␣ and sphingomyelinase by 88 and 82%, respectively (Fig. 7B). This indicates that PI 3-kinase activation by TNF␣ and sphingomyelinase requires a tyrosine kinase activity, as does the effect of C 2 -ceramide. Similarly, the activations of PI 3-kinase by TNF␣ and sphingomyelinase, respectively, were inhibited by about 70 and 99%, in fibroblasts expressing N17 Ras compared with control cells (Fig. 7C). This indicates the involvement of Ras-GTP in PI 3-kinase activation by these agonists.
TNF␣ and C 2 -ceramide Induce Cell Proliferation in Confluent Rat2 Fibroblasts-PI 3-kinase activation is often implicated in the control of cell proliferation. We therefore investigated whether this is the case for the TNF␣-and C 2 -ceramidemediated increases in PI 3-kinase. TNF␣ and C 2 -ceramide caused about a 2-fold increase in the incorporation of [ 3 H]thymidine into DNA after 24 h, and this effect was completely blocked by PI 3-kinase inhibitor, Ly 294002 (Fig. 8A). Furthermore, incubation of fibroblasts with 10 ng/ml TNF␣ or 40 M C 2 -ceramide for 48 h caused 1.68 Ϯ 0.05-and 1.28 Ϯ 0.1-fold increases in cell number, respectively (Fig. 8B). DISCUSSION PI 3-kinase can be activated in different cell types by proteintyrosine kinase-and Ras-dependent pathways (for review see Refs. [35][36][37][38]. These two mechanisms can activate PI 3-kinase synergistically (47). The present work establishes that TNF␣, sphingomyelinase, and C 2 -ceramide all activate PI 3-kinase in rat2 fibroblasts. This process is dependent on tyrosine kinase activity, and it involves an increase in the amount of Ras-GTP. The novelty of the present study is that TNF␣, sphingomyelinase, and ceramides should initiate such a signaling pathway.
The activation of PI 3-kinase observed after incubating rat2 fibroblasts for 10 min with 5-80 M C 2 -ceramide was demonstrated by assaying the activity of the enzyme after immunoprecipitation with anti-p85 or anti-phosphotyrosine antibodies. Furthermore, ceramide caused p85 to translocate to membranes where its associated p110 subunit would have access to its lipid substrates. C 2 -ceramide also increased the tyrosine phosphorylation of p85. The activation of PI 3-kinase was ascribed to C 2 -ceramide itself rather than metabolism to sphingosine because this rate of conversion was very low (estimated as Ͻ0.02% over the 10 min incubations), and the ceramide effect was not diminished by inhibitors of ceramidases. The specificity of the ceramide effect was confirmed because dihydro-C 2 -ceramide had no significant effect on PI 3-kinase activity.
Tyrosine kinases could contribute to the activation of PI 3-kinase by several mechanisms. For example, binding the p85 SH2 domains to tyrosine phosphorylated receptors or nonreceptor proteins increases PI 3-kinase activity. This binding also FIG. 5

. Activation of PI 3-kinase by C 2 -ceramide requires activated Ras.
Panel A shows the effects of 40 M C 2 -ceramide (15 min), 5 ng/ml PDGF (10 min), and 100 ng/ml EGF (5 min) on PI 3-kinase activity in fibroblasts expressing dominant-negative (N17) Ras compared with vector control cells. Panel B illustrates the concentration of Ras-GTP, measured as described under "Experimental Procedures," following stimulation of rat2 fibroblasts overexpressing wild type Ha-Ras for 0-20 min with 40 M C 2ceramide or with dihydro-C 2 -ceramide for 10 min. The representative Western blot illustrates Ras-GTP that precipitated with cRaf1-gluthione S-transferase. Panel C is a representative Western blot performed with anti-pan Ras antibody on immunoprecipitates obtained with anti-p85␣ antibody to show the relative concentration of Ras that was co-precipitated by the anti-p85␣ antibody. In these experiments, rat2 fibroblasts expressing wild type Ha-Ras were treated with 40 M C 2 -ceramide for 1 or 5 min or with 40 M dihydro-C 2 -ceramide for 5 min. Panel D demonstrates the time course for the relative increase in PI 3-kinase activity in anti-pan Ras precipitates using rat2 fibroblasts overexpressing wild type Ha-Ras. Results are shown as the means Ϯ S.E. from three independent experiments. DMSO, dimethyl sulfoxide.
localizes PI 3-kinase to the membrane bringing it into proximity with its lipid substrates thus contributing to the increased PI 3-kinase activity in vivo (39). Alternatively, or additionally, tyrosine phosphorylation events can cause the Ras guanyl nucleotide exchange protein, Sos, to be recruited to the plasma membrane where it can activate Ras and facilitate binding of Ras to p110. Tyrosine phosphorylation of p85 might represent yet another mechanism of PI 3-kinase activation by protein-tyrosine kinases, but this remains controversial. The role of tyrosine kinases in the C 2 -ceramide-induced activation of PI 3-kinase was confirmed by the inhibition of the ceramide-induced activation by genistein and PP1 and also our observation of increased tyrosine phosphorylation of the p85 subunit of PI 3-kinase (Fig. 3B), FAK, and various proteins in cell lysates after treatment with C 2 -ceramide (Fig. 6, B and C).
There is growing evidence linking PI 3-kinase to Ras-medi- ated signaling. Ras interacts with the p110 catalytic subunit in a GTP-dependent manner (46 -48), and PI 3-kinase co-immunoprecipitates with Ras (62,63). The involvement of Ras in the activation of PI 3-kinase by C 2 -ceramide is established by: 1) the lack of PI 3-kinase activation in cells expressing dominantnegative (N17) Ras, 2) the ceramide-induced increase in Ras-GTP in rat2 fibroblasts stably overexpressing wild type Ha-Ras, and 3) the ceramide-induced increase in the physical association of Ras with PI 3-kinase. This last point was demonstrated both by co-precipitation of Ras with PI 3-kinase using the anti-p85 antibody and also by the increase in PI 3-kinase activity found in anti-Ras immunoprecipitates. The idea that PI 3-kinase activation by ceramide involves an association with both tyrosine phosphorylated cellular proteins and Ras-GTP is significant. Our results show that C 2 -ceramide causes maximal activation of Ras after 5 min. However, the maximum PI 3-kinase activity found in anti-Ras immunoprecipitates occurs after 20 min. This could be because Ras activation is necessary, but alone is not enough to cause a detectable increase in PI 3-kinase activity. It may require concomitant tyrosine phosphorylation of proteins, such as FAK, which is maximal after a 20-min incubation with C 2 -ceramide (Fig. 6C). Such a dual regulatory mechanism in other systems causes the synergistic stimulation of PI 3-kinase activity (47).
Our results are not limited to the effects of exogenously added ceramide but probably reflect the normal biological signaling events downstream of cell activation by TNF␣. This cytokine exerts its effects, in part, by stimulating sphingomyelinase with subsequent accumulation of ceramides (2). TNF␣ and sphingomyelinase stimulated PI 3-kinase, and these ef-fects were blocked by the tyrosine kinase inhibitors, genistein and PP1. In addition, activation of PI 3-kinase by sphingomyelinase was completely abolished in fibroblasts that expressed N17 Ras, as was the case for C 2 -ceramide. However, the activation of PI 3-kinase by TNF␣ in the same cell line was inhibited by only 70%. This indicates that a maximum of about 30% of the TNF␣-induced increase in PI 3-kinase is probably mediated by a sphingomyelinase-independent pathway and the formation of ceramides and Ras-GTP. Taken together, our results demonstrate that tyrosine kinase stimulation and formation of Ras-GTP is upstream of PI 3-kinase activation by TNF␣ as well as by C 2 -ceramide and long chain ceramide.
The complete inhibition of the ceramide-induced activation of PI 3-kinase in rat2 cells expressing N17 Ras is striking. Warner et al. (64) showed that stable expression of N17 Ras in rat1 fibroblasts using the same vector as described here causes a significant but partial block to MAP kinase activation in response to EGF treatment. However, in these cells, N17 Ras expression completely blocked activation of phospholipase A 2 by EGF. By contrast, Burgering et al. (65) found that N17 Ras expressed by means of a vaccinia vector in rat1 fibroblasts did not block EGF stimulation of MAP kinase. Burgering et al. argued that rat fibroblasts possess both Ras-dependent and Ras-independent pathways that function in EGF signaling to MAP kinase. Likewise, we found that N17 Ras expression does not block PI 3-kinase activation by EGF, although there was a 80% inhibition of the PDGF activation of PI 3-kinase (Fig. 5A). Also N17 Ras expression in rat2 cells decreased the stimulation of MAP kinase by 0.1-100 ng/ml EGF by only about 8%. 2 In contrast, stimulation of MAP kinase activity by 5 ng/ml PDGF was decreased by 33-43% in rat2 fibroblasts expressing N17 Ras (results not shown). The existence of Ras-independent pathways of MAP kinase activation might explain why expression of N17 Ras had only a modest effect on the growth rate of rat2 cells in our studies. In any case, our results with N17 Ras indicate that sphingomyelinase and C 2 -ceramide activate PI 3-kinase by a pathway that depends absolutely on Ras-GTP in rat2 fibroblasts. This conclusion is supported by our observation that ceramide treatment leads to activation of wild type Ras.
The activation of PI 3-kinase by ceramides and TNF␣ could play an important role in regulating several cellular functions. For example, we demonstrated that cell-permeable ceramides increase PI 3-kinase activity associated with IRS-1 in 3T3-L1 adipocytes (23). We have also shown that treatment of rat2 cells for 20 min with either C 2 -ceramide or TNF␣ leads to activation of MAP kinase. 2 These effects are substantially blocked by Ly 294002 and by expression of N17 Ras. Other workers demonstrated that ceramides can activate MAP kinase in some cell types through the stimulation of a ceramide-dependent kinase that activates Raf (66). Our studies provide an alternative pathway for the ceramide-induced activation of MAP kinase that involves the stimulation of tyrosine kinases, Ras and PI 3-kinase. The conclusion that ceramides can activate mitogenic enzymes such as PI 3-kinase and MAP kinase in confluent rat2 fibroblasts may appear counter-intuitive because ceramides are often associated with producing apoptosis.
In an attempt to explain this contradiction, Kolesnick and Fuks (67) suggested that the cellular responses to ceramides depend on the genetic component of cells as well as the microenvironment in which the signal is generated.
The stimulation of fibroblast proliferation by TNF␣ plays an important role in the pathogenesis of many autoimmune and chronic inflammatory diseases (14 -17). TNF␣-induced fibro- blast proliferation has been reported to be dependent on PDGF secretion (11), stimulation of c-raf-1 kinase (68), and MAP kinase activation (15). Our results provide another mechanism for the TNF␣-induced proliferation of fibroblasts, which is dependent on PI 3-kinase activation. PI 3-kinase plays a key role in many cell processes such as growth (27,28), intracellular vesicle trafficking, secretion (29 -31), and regulation of the cytoskeleton (32)(33)(34). The demonstration that PI 3-kinase is activated by TNF␣ and ceramides in a tyrosine kinase-and Ras-GTP-dependent manner identifies a pathway that may contribute to signal transduction by cytokines and other agonists that stimulate sphingomyelinase activities.