Limited Role of Ceramide in Lipopolysaccharide-mediated Mitogen-activated Protein Kinase Activation, Transcription Factor Induction, and Cytokine Release*

The involvement of ceramide in lipopolysaccharide-mediated activation of mouse macrophages was studied. Lipopolysaccharide, cell-permeable ceramide analogs, and bacterial sphingomyelinase led to phosphorylation of the extracellular signal-regulated kinases, c-Jun NH2-terminal kinases, and p38 kinase and induced AP-1 DNA binding in C3H/OuJ (Lps n ) but not in C3H/HeJ (Lps d ) macrophages. Lipopolysaccharide and ceramide mimetics showed distinct kinetics of mitogen-activated protein kinase phosphorylation and AP-1 induction and activated AP-1 complexes with different subunit compositions. Lipopolysaccharide-activated AP-1 consisted of c-Fos, Jun-B, Jun-D, and c-Jun, while C2-ceramide induced Jun-D and c-Jun only. Lipopolysaccharide and, less potently, C2-ceramide or sphingomyelinase, stimulated AP-1-dependent reporter gene transcription in RAW 264.7 cells. Unlike lipopolysaccharide, C2-ceramide failed to activate NF-κB and did not induce production of tumor necrosis factor or interleukin-6. The lipopolysaccharide antagonist, Rhodobacter sphae-roides diphosphoryl lipid A, inhibited lipopolysaccharide activation of NF-κB and AP-1 but did not block C2-ceramide-induced AP-1. Pretreatment of C3H/OuJ macrophages with C2-ceramide greatly diminished AP-1 induction following subsequent C2-ceramide stimulation. However, lipopolysaccharide-induced transcription factor activation and cytokine release were not influenced. In contrast, lipopolysaccharide pretreatment inhibited both lipopolysaccharide- and C2-ceramide-mediated responses. Thus, ceramide partially mimics lipopolysaccharide in activating the mitogen-activated protein kinases and AP-1 but not in mediating NF-κB induction or cytokine production, suggesting a limited role in lipopolysaccharide signaling.

Lipopolysaccharide (LPS) 1 is the major constituent of the outer membrane of Gram-negative bacteria that plays a beneficial role in the course of bacterial infection due to its ability to stimulate the host immune response (1). However, in excess, LPS is harmful because it causes increased secretion of proinflammatory cytokines (e.g. TNF, IL-1␤, and IL-6) that contribute to septic shock (2). Several LPS-binding proteins have been identified, yet the true LPS signaling receptor(s) and the mechanism by which LPS transduces a signal across the cell membrane remain poorly understood. Studies with blocking antibodies (3,4) and transfection experiments (5,6) as well as knockout and transgenic models (7)(8)(9) have underscored the importance of CD14, a 55-kDa glycosyl phosphatidylinositollinked protein, for enabling cellular responses to low and intermediate concentrations of LPS. However, CD14 lacks transmembrane and cytoplasmic regions, and, in addition, high LPS concentrations elicit cellular activation in CD14 knockout mice (7,8). Thus, although CD14 is considered as a component of the LPS receptor complex, it does not seem to represent a true LPS signal transducing molecule. Among other LPS binding molecules, complement receptors type 3 and 4 (CR3 and CR4) have been suggested to mediate LPS signaling as their expression in CHO cells conferred upon them LPS sensitivity, as judged by NF-B activation (10,11). However, mutant CR3 molecules that lack cytoplasmic regions also mediate NF-B activation in response to endotoxin (11). Moreover, monocytes obtained from patients with a deficiency in expression of CD18, the common ␤-chain of the leukocyte integrins, show normal LPS binding and responsiveness (12). In light of these findings, CD14 and complement receptors have been proposed to bind LPS and associate with other molecules that mediate signal transduction. Recently, Yang et al. (13) have reported that transfection of human embryonic kidney 293 cells with Toll-like receptor 2 (TLR2) imparts LPS responses that are dependent on LBP and enhanced by CD14. LPS-induced NF-B activation has been demonstrated to require a region in the intracellular domain of TLR2 that is homologous to the intracellular region of the IL-1 receptor implicated in the activation of the IL-1 receptor-associated kinase (13). Interestingly, a TLR2 deletion mutant with a truncation of a region in the intracellular domain appears to act as a dominant-negative receptor by inhibiting the "endogenous" LPS-mediated reporter gene activation in U373 astrocytoma cells (13), further suggesting LPS signaling functions for TLR2. In a related study, genetic and physical mapping of the mouse Lps d mutation, which is responsible for LPS unresponsiveness of C3H/HeJ mice, has revealed a single intact gene within the entire Lps critical region on mouse chromosome 4 that encodes Toll-like receptor 4 (TLR4) (14). Thus, TLR2 and TLR4, members of the IL-1 receptor family, are likely to represent LPS signal-transducing molecules in humans and in the mouse, respectively. However, further studies will be required to confirm LPS signaling functions for TLR2 and to demonstrate functional significance of TLR4 in monocytes and macrophages.
Ceramide is an intracellular lipid derived by hydrolysis of sphingomyelin with sphingomyelin-specific forms of phospholipase C, neutral and acidic sphingomyelinases (SMase), which is known to mediate cell activation, proliferation, differentiation, and apoptosis in response to a number of cytokines and environmental stresses (15)(16)(17). Ceramide activates multiple intracellular targets, including a ceramide-activated protein kinase (18), identified as the mammalian homologue of kinase suppressor of Ras (19), a ceramide-activated protein phosphatase (20), and protein kinase C- (21). It has been suggested that LPS may activate cells via its direct stimulation of the sphingomyelin pathway (22) due to a close structural similarity of the lipid A portion of the LPS molecule and ceramide (23). Indeed, LPS activates the ceramide-activated protein kinase in myeloid cells directly, without stimulation of SMase (23). Phosphorylated ceramide-activated protein kinase activates Raf-1 (24), which then phosphorylates and activates the two dual specificity protein kinases, mitogen-activated protein (MAP) kinase/extracellular signal-regulated kinase (ERK) kinases 1 and 2, triggering the ERK MAP kinase pathway (25). Consistent with this scenario, stimulation of the sphingomyelin pathway has been reported to induce ERK activation (26) and NF-B translocation (27,28). However, others have been unable to reproduce these results (29 -32), necessitating further studies on the involvement of the sphingomyelin pathway in ERK activation and NF-B stimulation, responses that are strongly stimulated by LPS (5,6,8,10,11,33,34). Both LPS and ceramides have been found to activate the c-Jun NH 2terminal kinases (JNK) and the transcription factor ATF2 (35)(36)(37) and to induce the expression of a novel monocyte/macrophage differentiation-dependent gene (38). Furthermore, stimulation of C3H/OuJ (Lps n ) peritoneal macrophages with exogenous SMase or cell-permeable ceramide analogs activates a subset of LPS-inducible genes, whereas C3H/HeJ (Lps d ) macrophages fail to respond to either LPS or ceramide (39,40). C3H/HeJ (Lps d ) macrophages also exhibit a defective intracellular transport of LPS and ceramides from the membrane into the perinuclear region (41), suggesting that this defect may account for their hyporesponsiveness to both LPS and ceramide (40). In contrast to LPS, however, cell-permeable ceramide analogs were poor inducers of interferon-␥-inducible protein 10 and interferon consensus sequence-binding protein gene expression and interferon secretion in C3H/OuJ (Lps n ) macrophages (39). In addition, preexposure of macrophages with LPS significantly decreases TNF release in response to their subsequent stimulation with LPS, whereas pretreatment of macrophages with SMase does not suppress LPS-induced TNF production (39). These results suggest the existence of LPS signal transduction mechanisms that are distinct from the sphingomyelin pathway.
In the present study, we sought to evaluate further the involvement of ceramide in LPS-induced macrophage activation. To do so, LPS and ceramide were compared with respect to their capacities to elicit MAP kinase phosphorylation, NF-B and AP-1 transcription factor activation, and TNF and IL-6 production. The data indicate that ceramide partially mimics LPS-mediated phosphorylation of the ERK1/2, JNK1/2, and p38 MAP kinases, as well as AP-1 induction. In contrast to LPS, ceramide mimetics fail to elicit NF-B translocation, NF-B-controlled reporter gene expression, or production of TNF and IL-6. Strong inhibition of LPS-induced activation of NF-B and AP-1 was observed by using the LPS structural antagonist, Rhodobacter sphaeroides diphosphoryl lipid A (RsDPLA), yet RsDPLA did not antagonize ceramide-induced AP-1 induction. Preexposure of macrophages with LPS renders them tolerant to subsequent stimulation with either LPS or C 2 -ceramide, whereas ceramide treatment does not induce LPS hyporesponsiveness. Taken together, these results indicate a limited role of ceramide in LPS-mediated MAP kinase activation, transcription factor induction, and cytokine production in mouse macrophages, implying that LPS utilizes both the sphingomyelin pathway and additional intracellular mechanisms for mediating optimal signal transduction.
Plasmid Construction and Transient Transfection-Control luciferase reporter pL d 40-Luciferase (Luc) was constructed as described previously (44) by cloning the 40-bp H-2L d basic promoter into the pGL basic luciferase plasmid (Promega). Reporter plasmids containing an NF-B enhancer from the murine TNF promoter 5Ј-CAAACAGGGGG-CTTTCCCTCCTC-3Ј, its mutated version 5Ј-CAAACAGAGAGCCTTG-GCTCCTC-3Ј (mutated nucleotides are underlined) (45), and an AP-1 enhancer 5Ј-CGCTTGATGAGTCAGCCGGAA-3Ј (Promega) were cloned by inserting three copies of the respective response elements between the XhoI-BgIII sites of the pL d 40-luciferase vector. The resulting plasmids were referred to as p(NF-B) 3 L d Luc, pMut(NF-B) 3 L d Luc, and p(AP-1) 3 L d Luc. For transient transfections, RAW 264.7 cells were seeded into 12-well plates (Costar) at 2 ϫ 10 5 cells/well in DMEM/10% FCS, incubated overnight, and co-transfected for 3 h with the reporter plasmids (0.3 g/well of the NF-B reporters or 0.1 g/well of the AP-1 reporter construct) and 0.1 g/well of pCH110 eukaryotic ␤-galactosidase assay vector (Amersham Pharmacia Biotech) by using 7.5 l/well of SuperFect transfection reagent (QIAGEN Inc., Chatsworth, CA). The total amount of plasmid DNA was equalized to 1.5 g/well by adding corresponding amounts of pBluescript II SK (ϩ/ϩ) phagemid (Stratagene). Following transfections, cells were stimulated for 20 h, washed twice with ice-cold PBS, and lysed in a lysis buffer (Analytical Luminescence Laboratory, Sparks, MD) for 30 min with constant shaking. Fifty microliters of the supernatant was assayed in 200 l of assay buffer (25 mM glycylglycine, 15 mM MgSO 4 , 1% Triton X-100, 1 mM ATP) using a Monolight 2010 luminometer (Analytical Luminescence Laboratory). Following injection of 100 l of luciferin (0.3 mg/ml, Analytical Luminescence Laboratory), light emission was measured in 10-s intervals. To assay ␤-galactosidase activity, 20 l of the supernatant was mixed with 200 l of reaction buffer (Galacton-Plus substrate diluted 1:100 with reaction buffer diluent; Analytical Luminescence Laboratory), and incubated for 1 h. After injection of 300 l of light emission accelerator (Analytical Luminescence Laboratory), the sample was counted for 5 s. Luciferase activity was normalized to ␤-galactosidase activity (normalized relative light units), as described by Haas et al. (46). Normalized relative light units obtained in reporter-transfected cells were divided by those detected in either pMut (NF-B) 3 L d Luctransfected cells (NF-B transactivation) or in pL d 40Luc-transfected cells (AP-1 transactivation) for each treatment, respectively, and all data were normalized to those calculated for medium-treated cells. The resulting parameter was referred to as "relative fold stimulation" and reflects the transactivation potential of NF-B and AP-1.
Preparation of Nuclear Extracts-Nuclear extracts were prepared according to Dignam et al. (47) with small modifications. Briefly, cells were washed twice with ice-cold PBS, harvested using a rubber policeman, transferred to Eppendorf tubes, and centrifuged (800 ϫ g, 10 min, 4°C). Cells were resuspended in 0.4 ml of ice-cold buffer A (10 mM HEPES, pH 7.9, 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, and 1.0 g/ml each of the following protease inhibitors: aprotinin, antipain, leupeptin, chymostatin, and pepstatin), incubated on ice for 15 min, and lysed by the addition of Nonidet P-40 to a final concentration of 0.5%. Nuclei were pelleted (1,000 ϫ g, 10 min, 4°C) and resuspended in 50 l of ice-cold buffer C (20 mM HEPES, pH 7.9, 0.4 M NaCl, 1 mM EDTA, 1 mM EGTA, 25% glycerol, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, and 1 g/ml concentration of the protease inhibitors indicated above). Following a 30-min incubation, the tubes were centrifuged (10,000 ϫ g, 10 min, 4°C), and supernatants were collected and stored at Ϫ80°C. Protein concentration was determined by the Bio-Rad protein assay with bovine serum albumin as a standard (Bio-Rad).
Electrophoretic Mobility Shift Assay (EMSA)-NF-B-specific oligonucleotide 5Ј-AGTTGAGGGGACTTTCCCAGGC-3Ј from the murine IgB light chain gene enhancer and AP-1-specific oligonucleotide 5Ј-CGCTTGATGAGTCAGCCGGAA-3Ј probes (synthesized by the BIC Synthesis and Sequencing Facility, Uniformed Services University of the Health Sciences, Bethesda, MD) were 32 P-end-labeled with T4 polynucleotide kinase (Promega). Nuclear extracts (5 g) were incubated with 0.2 ng DNA probe in a binding buffer containing 2 g poly (dI-dC) (Amersham Pharmacia Biotech), 20 mM HEPES, pH 7.9, 50 mM KCl, 1 mM dithiothreitol, 0.25 mg/ml bovine serum albumin, 4% glycerol for 30 min at room temperature. For supershift assays, 5 g of nuclear extracts were first preincubated with 0.4 g of antibodies against members of the Fos/Jun/Fra family for 45 min at room temperature in the above mentioned binding buffer. The DNA-protein complex was resolved from free oligonucleotide by electrophoresis in a 5% polyacrylamide gel (0.25ϫ Tris borate/EDTA, 150 V/2 h). The gels were dried (80°C, 2 h) and exposed to x-ray films (X-OMAT AR; Eastman Kodak Co.).
Preparation of Cellular Extracts and Western Blotting-Cellular extracts were obtained as described (48), and 50 g of total protein was added in Laemmli buffer, boiled for 5 min, and loaded on SDS-10% polyacrylamide gels for electrophoresis in Tris/glycine/SDS buffer (25 mM Tris, 250 mM glycine, 0.1% SDS). Proteins were blotted onto Immobilon-P transfer membranes (150 V, 1.5 h, 4°C). The filters were blocked for 20 h at 4°C in TBS-T (20 mM Tris HCl, 150 mM NaCl, 0.1% Tween 20), containing 1% gelatin and 5% nonfat milk. Thereafter, the filters were washed three times in TBS-T and probed for 1 h with anti-phospho-MAP kinase antibodies (Anti-active TM antibody; Promega) diluted 1:2,000 in TBS-T/0.5% nonfat milk, washed three times with TBS-T, and incubated with secondary horseradish peroxidase- conjugated donkey anti-rabbit antibody (1:5,000 dilution). The blots were washed 5 times with TBS-T, and bands were detected by the enhanced chemiluminescence (ECL) detection method (Amersham Pharmacia Biotech).
TNF and IL-6 Assays-TNF activity in supernatants was measured in the WEHI 164 clone 13 bioassay as described previously (49). The lower limit of detection in this assay was 0.35 pg/ml. IL-6 content of the supernatants was determined on the basis of induction of proliferation of the IL-6-dependent B.13.29 clone 9 cell line (generously provided by Dr. T. Espevik) as described elsewhere (50).

RESULTS
Both LPS and Ceramide Cause Phosphorylation of the ERK1/2, JNK1/2, and p38 MAP Kinases-To analyze involvement of the sphingomyelin pathway in LPS signal transduction, the abilities of LPS, cell-permeable ceramide analogs, and SMase to activate the ERK1/2, JNK1/2, and p38 MAP kinases in C3H/OuJ (Lps n ) peritoneal macrophages were compared. A close correlation exists between phosphorylation of the MAP kinases and activation of enzyme activity (51). Therefore, MAP kinase phosphorylation was assessed by Western blotting, us-ing antibodies specific for phosphorylated forms of these kinases. Time course experiments revealed that LPS-mediated phosphorylation of the ERK1/2, JNK1/2, and p38 MAP kinase was evident after 2-5 min, reached a maximum response within 15-30 min, and declined by 60 min (Fig. 1). As shown in Fig. 1, incubation for 15 min was required to permit detectable phosphorylation of all three MAP kinases in response to SMase or C 2 -ceramide, which was significantly lower than that activated by LPS at this time point. After 30 min of incubation, SMase and C 2 -ceramide stimulated maximum levels of MAP kinase phosphorylation, and the responses decreased by 60 min (Fig. 1). As little as 1 ng/ml LPS induced phosphorylation of the MAP kinases, and the minimal effective concentration of SMase was in the range of 31.25-62.5 milliunits/ml, while C 2 -ceramide was active at concentrations of Ն5 M (data not shown). These results indicate that although triggering of the sphingomyelin pathway mimics LPS activation of the ERK1/2, JNK1/2, and p38 MAP kinases qualitatively, C 2 -ceramide and SMase mediate delayed and more transient responses relative to LPS-induced effects.
LPS, but Not C 2 -ceramide or SMase Induces NF-B Activation-The ability of LPS and ceramide to activate the transcription factor NF-B was next examined by measuring NF-B DNA binding and transcriptional activities. Fig. 2A demonstrates that LPS induced NF-B translocation in C3H/OuJ (Lps n ) macrophages after 15 min and exerted a maximal effect at 30 -60 min, and the response decreased by 180 min of stimulation. Dose-response experiments showed evident NF-B translocation caused by 0.1 ng/ml LPS, which reached a plateau at 1-1,000 ng/ml LPS (data not shown). Endotoxinhyporesponsive C3H/HeJ (Lps d ) macrophages did not exhibit NF-B activation following LPS treatment ( Fig. 2A). In contrast to LPS, neither C 2 -ceramide ( Fig. 2A) nor C 8 -and C 6ceramide analogs (data not shown) led to NF-B translocation in macrophages from either mouse strain throughout 180 min of stimulation. Similarly, no NF-B translocation was seen after macrophage treatment with SMase or DL-PPMP (data not shown), agents that increase intracellular concentrations of endogenous ceramide. When Gram-positive heat-killed GBS was used as a stimulus, similar patterns of NF-B induction were observed in macrophages from both mouse strains ( Fig.  2A), ruling out the possibility of generalized hyporeactivity of C3H/HeJ cells. To extend the EMSA data, we also studied the capacities of LPS and ceramide mimetics to activate NF-B-dependent transcription in RAW 264.7 cells transiently transfected with the p(NF-B) 3 L d Luc reporter construct. As depicted in Fig. 2B, 100 ng/ml LPS potently activated NF-B-dependent Luc gene expression (relative fold stimulation compared with medium-treated cells is 15.0 Ϯ 1.5). Ten and 1 ng/ml LPS caused 10 Ϯ 1.2 and 7.7 Ϯ 2.1 stimulation of the NF-B reporter, respectively, whereas 0.1 ng/ml LPS caused a 2-fold stimulation. Importantly, LPS did not activate the construct with a mutated NF-B consensus sequence, indicating that the response was NF-B-specific. Similar to the failure to cause NF-B translocation in mouse macrophages ( Fig. 2A) and in RAW 264.7 cells (data not shown), neither C 2 -ceramide, SMase, nor DL-PPMP activated NF-B-dependent transcription (Fig.  2B). Thus, in contrast to LPS, triggering of the sphingomyelin pathway does not activate NF-B in mouse macrophages.
Both LPS and Ceramide Activate the Transcription Factor AP-1-To evaluate further if LPS signals cell activation by mimicking ceramide, we compared the effects of LPS, C 2 -ceramide, SMase, and DL-PPMP on AP-1 activation. Treatment of C3H/OuJ macrophages with LPS for 30 min resulted in an increased AP-1 DNA binding, which reached a plateau by 60 -120 min and declined by 180 min post-stimulation (Fig. 3A). As demonstrated in Fig. 3, A and B, SMase, C 2 -ceramide, and DL-PPMP also activated AP-1 DNA binding but with slower kinetics compared with that induced by LPS. As little as 0.01 ng/ml LPS induced AP-1 transcriptional activity by 2.5-fold in RAW 264.7 cells transiently transfected with the p(AP-1) 3 L d -Luc reporter plasmid, whereas 1-100 ng/ml LPS caused a stimulation up to 6.7-fold ( Fig. 3C and data not shown). Although less potent than LPS, SMase, DL-PPMP, and C 2 -ceramide all resulted in the induction of AP-1-dependent reporter gene expression in a dose-dependent fashion (Fig. 3C). The inactive ceramide analog, C 2 -dihydroceramide, did not activate AP-1 (Fig. 3B). Neither LPS nor C 2 -ceramide stimulated AP-1 DNA binding in LPS-hyporesponsive C3H/HeJ mouse macrophages (Fig. 3), supporting a correlation between LPS hyposensitivity and defective ceramide responses described previously at the level of gene expression (39,40). These results demonstrate that ceramide partially mimics LPS-mediated activation of AP-1 DNA-binding and transcriptional activities, exhibiting delayed and less potent responses compared with LPS-medi-ated effects.

Different Members of the Jun/Fos Family Comprise AP-1 Complexes Induced by LPS Versus C 2 -ceramide-
The transactivating potential of many transcription factors, including AP-1, is determined by their subunit composition (52). Hence, it was of interest to examine whether different potencies of LPS and ceramide in mediating AP-1 trans-activation correlate with different subunit compositions of AP-1 complexes induced by these stimuli. To this end, we performed supershift analyses of nuclear extracts obtained from LPS-and C 2 -ceramide-stimulated macrophages using antibodies against members of the Fos/Jun family. Fig. 4 shows that the addition of antibodies against c-Fos, Jun-B, c-Jun, and Jun-D resulted in the appearance of slower migrating species relative to the major LPSinducible AP-1 band, whereas anti-Fos-B antibodies had no effect. In contrast, only c-Jun and Jun-D comprised AP-1 complexes induced by C 2 -ceramide (Fig. 4). Antibodies against Fra-1 and Fra-2 did not change the electrophoretic mobility of the AP-1 band in samples from either LPS-or C 2 -ceramidestimulated macrophages (Fig. 4). AP-1-DNA complexes induced by either stimulus were reduced in the presence of an excess of unlabeled AP-1, but not NF-B, consensus oligonucleotides (data not shown), demonstrating the specificity of the response. Taken collectively, these data demonstrate that LPS stimulates both Fos and Jun members of the AP-1 superfamily, while only c-Jun and Jun-D are activated by C 2 -ceramide. whereas 1,000 ng/ml LPS overcame the inhibitory effect of RsDPLA (Fig. 5, A and B). Neither NF-B nor AP-1 were induced by RsDPLA alone. Ten, 5, and 2.5 g/ml RsDPLA completely abrogated NF-B and AP-1 activation in response to 1 ng/ml LPS, whereas no inhibition was observed with concentrations of RsDPLA of Ͻ0.32 g/ml (data not shown). In contrast to its potent inhibition of LPS-induced responses, Rs-DPLA did not affect C 2 -ceramide-mediated AP-1 activation, even when used at 10 g/ml (Fig. 5C). Similar data were obtained with C 8 -ceramide (data not shown), demonstrating that RsDPLA fails to inhibit ceramide-induced AP-1 induction.

The Addition of C 2 -ceramide or SMase Does Not Affect LPSinduced NF-B Activation and Cytokine Production but Enhances the LPS-mediated AP-1 Response-
The next goal of the study was to elucidate if triggering of the sphingomyelin pathway modulates LPS-initiated responses. C3H/OuJ macrophages were stimulated with suboptimal LPS concentrations in the absence or presence of C 2 -ceramide, and NF-B and AP-1 activation as well as TNF and IL-6 production were measured as functional parameters of LPS signaling. Treatment of cells with 1 ng/ml LPS for 30 min led to a marked NF-B translocation, whereas C 2 -ceramide neither induced NF-B nor affected the LPS response (Fig. 6). Similar data were obtained when LPS was used at concentrations of 0.1 and 10 ng/ml and when NF-B translocation was assessed throughout a 15-120min period of stimulation (data not shown). Interestingly, LPS or C 2 -ceramide alone activated AP-1 DNA binding, and their combination gave rise to an additive response (Fig. 6). Stimulation of macrophages with 0.1 and 1 ng/ml LPS for 6 h resulted in the production of TNF (1,000 Ϯ 150 and 10,300 Ϯ 689 pg/ml) and IL-6 (1,100 Ϯ 75 and 5,475 Ϯ 487 pg/ml), respectively. In contrast, C 2 -ceramide used at concentrations of 1, 12.5, and 25 M did not stimulate the release of TNF or IL-6 above levels seen in medium-treated macrophages (1 Ϯ 0.3 pg/ml TNF and 41 Ϯ 6 pg/ml IL-6). Consistent with the NF-B data (Fig. 6), simultaneous addition of 1, 12.5, and 25 M C 2 -ceramide did not modulate LPS-induced TNF and IL-6 responses. The inactive ceramide analogue, C 2 -dihydroceramide, had no effect on LPS-mediated transcription factor activation and cytokine production (data not shown). These results indicate that C 2 -ceramide-induced signal transduction pathways do not contribute to LPS-induced NF-B activation and production of TNF and IL-6. On the other hand, AP-1 stimulation induced by LPS and C 2 -ceramide may be triggered by distinct mechanisms that converge downstream, leading to an additive effect.

LPS Pretreatment Inhibits Macrophage Activation by LPS and C 2 -ceramide, whereas C 2 -ceramide Enables Macrophage
Tolerance to C 2 -ceramide but Not to LPS-Next, a model of macrophage tolerance to LPS in vitro (53,54) was utilized to examine the potential role of ceramide in LPS-mediated transcription factor activation and cytokine production. First, C3H/ OuJ macrophages were pretreated for 20 h with 10 ng/ml LPS, washed, rested in fresh medium for 2 h, and restimulated with either LPS or C 2 -ceramide. LPS pretreatment significantly decreased the ability of macrophages to induce NF-B and AP-1 (Fig. 7) as well as to secrete TNF and IL-6 ( Fig. 8) in response to subsequent stimulation with LPS. Likewise, prior exposure of macrophages to LPS markedly inhibited their AP-1 response induced by C 2 -ceramide (Fig. 7). In contrast, preincubation of macrophages with C 2 -ceramide did not affect LPS-mediated NF-B translocation, AP-1 induction (Fig. 7), TNF, or IL-6 production (Fig. 8, A and B), while it completely abolished AP-1 induction in response to C 2 -ceramide (Fig. 7). Similar results were obtained when SMase was used to generate intracellular ceramide (data not shown). Thus, prior exposure of macrophages with ceramide mimetics results in down-modulation of the sphingomyelin pathway, inhibiting subsequent induction of an AP-1 response by ceramide, whereas LPS-mediated transcription factor activation and cytokine production remain unaffected. In contrast, LPS pretreatment renders macrophages unresponsive to both LPS and ceramide. DISCUSSION LPS has been proposed to signal cell activation by mimicking the intracellular second messenger ceramide (23), due to a close structural similarity between the lipid A portion of LPS and the ceramide molecule (24). One implication of this model is that similar intracellular pathways are engaged by LPS and ceramide that consequently lead to overlapping responses. This paper demonstrates the failure of LPS-hyporesponsive C3H/ HeJ macrophages to exhibit NF-B and AP-1 activation following stimulation with LPS, SMase, and C 2 -ceramide. Our results extend the correlation between LPS and ceramide hyporesponsiveness of C3H/HeJ macrophages found previously at the level of gene expression (40) and support involvement of the sphingomyelin pathway in LPS signaling. However, several observations shown in the present study strongly suggest that activation of the sphingomyelin pathway is one of multiple signals initiated by LPS in mouse macrophages. First, LPS and ceramide were found to exhibit qualitative differences in the elicitation of several cellular responses. Indeed, in contrast to LPS, cell-permeable ceramide analogs, DL-PPMP, and SMase did not stimulate NF-B translocation in C3H/OuJ macrophages and failed to trigger NF-B-dependent transcription of the Luc reporter gene in RAW 264.7 cells. These data support and extend earlier publications on the inability of ceramide to induce NF-B found in other cell types (30 -32, 36). However, conflicting reports exist that demonstrate the capacities of ceramide analogs to activate NF-B in HL60 promyelocytic and EL4 lymphoma cells (27,28) and to potentiate TNF-or LPSinduced NF-B responses in human vein endothelial cells and in primary rat astrocytes (55,56). Therefore, to determine unequivocally the role of ceramide in NF-B activation in mouse macrophages, we analyzed how independent triggering and down-modulation of the sphingomyelin pathway affect LPS-initiated NF-B translocation. Not only did C 2 -ceramide lack the ability to activate NF-B by itself, but also it failed to influence LPS-initiated NF-B responses when present in the cell cultures simultaneously with LPS. Furthermore, pretreatment of macrophages with C 2 -ceramide did not affect LPSmediated NF-B translocation, while it completely abolished subsequent induction of AP-1 by ceramide mimetics. In contrast, LPS pretreatment rendered macrophages hyporesponsive to subsequent challenge with either LPS or ceramide mimetics. These data demonstrate that ceramide-elicited refractoriness of macrophages to itself has no effect on the ability of LPS to activate NF-B. Taken together, these results indicate that, in mouse macrophages, ceramide is not involved in LPS-mediated NF-B activation. Similar to the NF-B data, LPS, but not C 2 -ceramide, stimulated C3H/OuJ macrophages to secrete TNF and IL-6. Moreover, no modulatory effect of C 2 -ceramide on LPS-mediated cytokine release was observed when these reagents were added simultaneously or when cells were first pretreated with C 2 -ceramide. Thus, at least three LPS responses exist, i.e. NF-B activation as well as TNF and IL-6 release, that could not be induced or modulated by ceramide mimetics.
Second, to delineate whether LPS and ceramide involve similar mechanisms for transcription factor activation, we employed the LPS antagonist, RsDPLA (54,57). Importantly, RsDPLA, as well as other LPS partial structures, e.g. deacylated LPS or lipid IV A , suppress LPS activities at concentrations that do not affect LPS binding to CD14 (58), suggesting that they compete with LPS for a putative signaling receptor distinct from CD14 (58). An alternative hypothesis postulates that RsDPLA could trigger negative intracellular events that interfere with LPS signaling further downstream at the postreceptor level. Thus far, however, no evidence exists to support the latter model, since RsDPLA fails to induce transcription factor activation (Fig. 4), MAP kinase phosphorylation, or cytokine release (data not shown). Regardless of the exact mechanism of action of RsDPLA, we reasoned that if LPS-mediated macrophage activation occurs via the sphingomyelin pathway, then RsDPLA should inhibit both LPS-and ceramide-mediated effects. However, this is not the case, as treatment of C3H/OuJ macrophages with RsDPLA did not affect AP-1 activation in response to C 2 -ceramide, whereas this compound significantly inhibited LPS-induced NF-B and AP-1. Thus, RsDPLA-inhibitable molecules engaged in LPS-mediated transcription factor activation do not represent ceramide-triggered signaling components.
Third, this paper shows that even when overlapping responses are elicited, LPS and ceramide seem to involve divergent intracellular mechanisms. Both LPS and ceramide mediated AP-1 activation; however, different kinetics of AP-1 induction, subunit compositions of AP-1 complexes, and differ- ent potencies of AP-1 trans-activation were seen in response to LPS versus ceramide. Furthermore, an additive effect on AP-1 DNA-binding activity was observed when macrophages were stimulated with a combination of LPS and C 2 -ceramide. Finally, macrophages pretreated with LPS manifested a significantly lower AP-1 response upon subsequent stimulation with LPS, C 2 -ceramide, or SMase, whereas prior exposure to C 2ceramide conferred tolerance only to C 2 -ceramide. Our results extend earlier published observations on the ability of the sphingomyelin pathway to signal AP-1 activation in the human leukemia cell line HL-60 (59) and in the insulin-producing cell line RINm5F (36), studies where a comparative analysis of LPS versus ceramide responses was not undertaken. Furthermore, they suggest that quantitative differences between LPS and ceramide with respect to AP-1 trans-activation are likely to reflect their capacities to stimulate AP-1 complexes with different subunit compositions. Indeed, both Fos and Jun proteins comprised LPS-induced AP-1, which have been reported to form more stable complexes and activate transcription more efficiently than Jun subunits (60,61), whose up-regulation was caused by C 2 -ceramide.
Analysis of MAP kinase phosphorylation also underscored the kinetic differences between LPS and ceramide mimetics. To the best of our knowledge, this paper shows for the first time that ceramide is capable of mediating phosphorylation of p38 MAP kinase in mouse macrophages, extending a similar observation made recently in Jurkat cells (62). In addition, we have found that LPS, C 2 -ceramide, or SMase shared the ability to mediate phosphorylation of ERK1/2 MAP kinases, an observation for which conflicting data have been reported (26,29). In agreement with earlier results (32)(33)(34)(35), C 2 -ceramide or SMase also mimicked LPS-induced phosphorylation of the JNK1 and two MAP kinases. However, despite the fact that both LPS, C 2 -ceramide, and SMase caused phosphorylation of ERK1/2, JNK1/2, and p38 MAP kinases, slower and less sustained responses were induced by ceramide mimetics compared with LPS. It is tempting to speculate that differential activation of protein phosphatases by LPS versus ceramide accounts for this phenomenon, limiting MAP kinase-mediated downstream signaling triggered by ceramide. Since different patterns of cellular responses could be induced depending on whether transient or sustained MAP kinase activation occurs (63), this could have functional consequences for differential effect of LPS and ceramide on both transcription factor activation and cytokine production. Experiments are in progress to examine this hypothesis.
NF-B is a prerequisite for the expression of many cytokine genes, including TNF (45), IL-6 (64), granulocyte-macrophage colony-stimulating factor (65), and interferon-␥-inducible protein 10 (66). Therefore, the failure of ceramide to activate NF-B is likely to reflect its inability to elicit TNF and IL-6 production (data herein, and see Refs. [67][68][69] and to induce interferon-␥-inducible protein 10 gene expression (39). In addition, recent reverse transcription-polymerase chain reaction analyses of cytokine gene expression in C3H/OuJ macrophages have demonstrated that SMase and cell-permeable ceramide analogs are very poor inducers of the expression of mRNA for another NF-B-dependent gene, granulocyte-macrophage colony-stimulating factor. 2 Furthermore, ceramide is incapable of activating C3H/OuJ macrophages to secrete nitric oxide (70), a process that requires NF-B activation for the induction of inducible nitric-oxide synthase mRNA (71). These results are consistent with recent findings in which unimpaired cytokine production in acid SMase knock-out mice (68,69) as well as normal NF-B activation in acid SMase-deficient cells (32) have been demonstrated. On the other hand, stimulation of the sphingomyelin pathway with cell-permeable ceramide analogs (39) or SMase (data not shown) mimics LPS-induced expression of IL-1␤ mRNA in mouse C3H/OuJ macrophages. It is important to note that in mice, IL-1␤ gene expression is NF-Bindependent but is controlled by several transcription factors, including AP-1, NF-IL6, and PU-1 (72)(73)(74). Therefore, it is plausible that ceramide-induced AP-1 activation contributes to stimulation of IL-1␤ gene expression in our system. More generally, it suggests a functional role of the sphingomyelin pathway in mediating expression of NF-B-independent genes, whose stimulation requires AP-1. Another important consequence of AP-1 activation by ceramide underlies apoptotic death of human leukemia HL-60 cells (59), implying that ceramide-induced cell death of C3H/OuJ mouse macrophages (70) could also be AP-1-mediated. However, C3H/OuJ macrophages exhibit only necrotic cell death in response to ceramide analogs and mimetics, whereas LPS induces both necrosis and apoptosis (70), again suggesting the existence of additional intracellular pathways activated by LPS.
In summary, several novel findings presented herein indicate that LPS involves not only the sphingomyelin pathway but also alternate intracellular pathways to evoke MAP kinase 2 A. E. Medvedev and S. N. Vogel, unpublished observation.
FIG. 8. Effect of macrophage pretreatment with LPS or C 2ceramide on TNF and IL-6 production after subsequent LPS challenge. C3H/OuJ macrophages were pretreated for 20 h with medium, 10 ng/ml of LPS, 10 M C 2 -ceramide, or 10 M C 2 -dihydroceramide. After washing, cells were rested in fresh medium for 2 h, washed, and stimulated with a serial dilution of LPS. Following incubation for 6 h, cell-free supernatants were collected and analyzed for TNF (A) and IL-6 (B). The data (mean Ϯ S.D.) of a representative experiment (n ϭ 4) are shown. activation, transcription factor induction, and production of TNF and IL-6 in mouse macrophages. These include the lack of capacity of ceramide to mimic LPS-mediated NF-B activation, TNF, and IL-6 production, its failure to tolerize macrophages against LPS, and the ability of RsDPLA to inhibit LPS-induced transcription factor activation without affecting ceramide-induced AP-1. Furthermore, even in the case where a certain response, e.g. AP-1 activation, is elicited by both LPS and ceramide, they are likely to utilize distinct intracellular pathways. Identification of upstream regulators of MAP kinase cascades triggered by LPS and ceramide as well as characterization of their downstream targets (e.g. kinases and transcription factors) will help bring about a better understanding of the contribution of the sphingomyelin pathway to LPS signal transduction.