Ceramide 1-phosphate is a direct activator of cytosolic phospholipase A2.

Recently, we demonstrated that ceramide kinase, and its product, ceramide 1-phosphate (Cer-1-P), were mediators of arachidonic acid released in cells in response to interleukin-1beta and calcium ionophore (Pettus, B. J., Bielawska, A., Spiegel, S., Roddy, P., Hannun, Y. A., and Chalfant, C. E. (2003) J. Biol. Chem. 278, 38206-38213). In this study, we demonstrate that down-regulation of cytosolic phospholipase A(2) (cPLA(2)) using RNA interference technology abolished the ability of Cer-1-P to induce arachidonic acid release in A549 cells, demonstrating that cPLA(2) is the key phospholipase A(2) downstream of Cer-1-P. Treatment of A549 cells with Cer-1-P (2.5 microm) induced the translocation of full-length cPLA(2) from the cytosol to the Golgi apparatus/perinuclear regions, which are known sites of translocation in response to agonists. Cer-1-P also induced the translocation of the CaLB/C2 domain of cPLA(2) in the same manner, suggesting that this domain is responsive to Cer-1-P either directly or indirectly. In vitro studies were then conducted to distinguish these two possibilities. In vitro binding studies disclosed that Cer-1-P interacts directly with full-length cPLA(2) and with the CaLB domain in a calcium- and lipid-specific manner with a K(Ca) of 1.54 microm. Furthermore, Cer-1-P induced a calcium-dependent increase in cPLA(2) enzymatic activity as well as lowering the EC(50) of calcium for the enzyme from 191 to 31 nm. This study identifies Cer-1-P as an anionic lipid that translocates and directly activates cPLA(2), demonstrating a role for this bioactive lipid in the mediation of inflammatory responses.

cPLA 2 was first characterized in platelets and macrophage cells and subsequently cloned from a macrophage cDNA library (1)(2)(3)(4)(7)(8)(9). The cPLA 2 cDNA encodes for an 85-kDa protein, and the mRNA for cPLA 2 is widely expressed in the brain, lung, kidney, heart, and spleen (1)(2)(3)(4)(7)(8)(9). In vitro, cPLA 2 is activated by Ca 2ϩ ; however, the addition of salt at physiologic concentrations will also induce enzyme activation, and thus, the catalytic activity of cPLA 2 is not dependent on Ca 2ϩ (1,5,6). On the other hand, the activation/translocation of cPLA 2 in cells requires the association of cPLA 2 with membranes in a Ca 2ϩ -dependent manner (1)(2)(3)(4). Near the N terminus of cPLA 2 is a stretch of ϳ120 amino acids that constitutes the Ca 2ϩ -dependent lipid binding domain (CaLB domain) of the enzyme (1)(2)(3)(4)(5)7). However, the specific membrane lipids that regulate the association of this domain with membranes and demonstrate interaction with cPLA 2 in low calcium (300 nM), as first shown by Knopf and co-workers (3), have yet to be defined. Furthermore, it is not known whether physiologic calcium is sufficient to activate cPLA 2 , or whether activation also requires the generation of activating lipids.
Our laboratories recently reported that ceramide kinase and its product, ceramide 1-phosphate (Cer-1-P), are mediators of calcium ionophore-and interleukin-1␤-induced AA release suggesting a role for the ceramide kinase/Cer-1-P pathway in the activation of a species of PLA 2 (10). Interestingly, ceramide kinase is a calcium-activated enzyme in a similar manner as cPLA 2 , and these enzymes share similar patterns of tissue expression (7,11). In this study, mechanisms by which ceramide 1-phosphate activates cPLA 2 were investigated. Cer-1-P was found to bind cPLA 2 directly and to be an inducer of cPLA 2 activation in vitro and in cells. Thus, for the first time, a specific intracellular target for this novel bioactive lipid has been demonstrated, establishing Cer-1-P as a signaling lipid in biological systems. Furthermore, this study also demonstrates a possible lipid "missing link" in the induction of eicosanoid synthesis in response to agonists.
Immunoblotting-Cells were fractionated into cytosol and membrane-associated proteins as described previously for cPLA 2 (12). Protein fractions (10 g) from each sample were resolved on 7.5% SDS-PAGE under denaturing conditions and then transferred to 0.20-m polyvinylidene difluoride membranes. After blocking overnight at 4°C with 5% nonfat milk in phosphate-buffered saline, 0.05% Tween 20 (M-PBS-T) and washing, the membranes were incubated with anti-cPLA 2 (4-4B-3C) (Santa Cruz Biotechnology) in M-PBS-T for 2 h at room temperature. The membranes were washed extensively in phosphate-buffered saline, 0.05% Tween 20 (washing buffer). The bands were visualized using the appropriate horseradish peroxidase-conjugated anti-mouse IgG antibody and the ECL Western blotting detection system (Amersham Biosciences).
Lipid-Protein Overlay Assay-Various amounts of the indicated lipids (dioleoyl-PC, dioleoyl-PA, dioleoyl-PE, dioleoyl-PI, dioleoylcardiolipin, dioleoyl-PS, D-e-sphingosine, D-e-sphingosine 1-phosphate, D-e-C 16 -ceramide, D-e-C 16 ceramide 1-phosphate, and dioleoylglycerol) were spotted onto Hybond C membrane (Amersham Biosciences) and dried under vacuum. The membrane was rewet in water and then blocked for 1 h in 3% fatty acid-free bovine serum albumin (FAF-BSA)/TBS-T (0.1% Tween 20). The membrane was then exposed to 0.2 g/ml cPLA 2 -GST in 3% FAF-BSA/TBS-T overnight at 4°C. The membrane was washed six times for 5 min with TBS-T and then exposed to a 1:1000 dilution of anti-human cPLA 2 mouse IgG (4 -4B-3C) in 3% FAF-BSA/TBS-T for 1 h at room temperature. The membrane was again washed six times for 5 min with TBS-T and then exposed to a 1:2500 dilution of anti-mouse IgG-horseradish peroxidase (Santa Cruz Biotechnology) in 3% FAF-BSA/TBS-T for 1 h at room temperature. The membrane was then washed 12 times for 5 min with TBS-T and then developed by ECL. Purified recombinant GST did not bind any of the lipids using this protocol (data not shown).
Immunofluoresence-Cells were fixed and permeabilized in 3.7% formaldehyde, 0.1% Triton X-100, phosphate-buffered saline for 15 min. Cells were immunostained with 10 g/ml anti-human cPLA 2 antibody (C20) in phosphate-buffered saline, 5% human serum for 2 h at room temperature. After extensive washing with phosphate-buffered saline, cells were immunostained with 10 g/ml anti-goat IgG-rhodamineconjugated antibody in phosphate-buffered saline, 5% human serum for 1 h at room temperature. After extensive washing with phosphatebuffered saline, cPLA 2 localization was visualized by confocal microscopy at 568 nm using a 60ϫ oil immersion lens with a 1.5ϫ enhanced magnification.
Immunolocalization of cPLA 2 -5 ϫ 10 5 A549 cells were plated onto number 1 coverslips. 24 h later, the cells were starved overnight and then treated with either vehicle (0.1% dodecane:ethanol, 49:1) or 5 M Cer-1-P. Cells were then fixed in 3.7% formalin in phosphate-buffered saline for 20 min at room temperature and then stained essentially as described (34). Primary antibodies were mouse anti-GM130 for the Golgi apparatus (BD Transduction Laboratories) and goat anti-cPLA 2 , and secondary antibodies were from Jackson ImmunoResearch. Images were collected on a Nikon TE-300 microscope at 60ϫ using a CoolSnap (Roper Scientific) CCD camera driven by MetaMorph software (Universal Imaging Corp.). cPLA 2 Assay-cPLA 2 -GST (0.2 g) was assayed as described (8). Briefly, a liposome substrate consisting of labeled PC (1-stearoyl-2-[ 14 C]arachidonoyl (100,000 dpm/assay) (PerkinElmer Life Sciences)), unlabeled 1-stearoyl-2-arachidonoyl PC (Avanti) (30 M/assay), and dioleoylglycerol (10 M/assay) in 50 mM Tris-HCl, pH 7.4, was produced by sonication on ice for 3 min. D-e-C 16 -ceramide 1-phosphate vesicles were generated separately by sonication on ice three times for 30 s each in 50 mM Tris-HCl, pH 7.4. cPLA 2 -GST (0.2 g) was then assayed in a base buffer consisting of 50 mM Tris-HCl, pH 7.4, and 150 mM NaCl, the substrate liposome, 100 M free calcium, and varying concentrations of D-e-C 16   a Time from the end of the RNAi transfection to analysis of cPLA 2 expression and the start of the assay for C-1-P-induced AA release. b Mean percentage of the number of cells demonstrating intracellular fluorescence of RNAi (1 ϫ 10 3 cells scored). c Mean percentage of the down-regulation of immunoreactive cPLA 2 normalized to total protein and ␣-tubulin as compared to scrambled (control) RNAi (100%).
d Mean percentage of the inhibition of C-1-P-induced AA release (2.5 M C-1-P for 3 h) as compared with scrambled (control) RNAi (100%). e Percentage of the inhibition of C-1-P-Induced AA release (2.5 M C-1-P for 3 h) as compared to scrambled (control) RNAi (100%) normalized to the percent of cPLA 2 downregulation.

FIG. 1. cPLA 2 is required for Cer-1-P-induced AA release.
Down-regulation of cPLA 2 by RNAi were performed essentially as described under "Materials and Methods." 48 h after post-transfection, cells were treated with Cer-1-P (2.5 M) for 3 h and assayed for AA release. Data are presented as the percent control of AA release controlled for equivalent number of cells by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay. Data are representative of three separate determinations on two separate occasions.
Eagle's medium supplemented with 2% fetal bovine serum for 2 h. Following treatment, medium was transferred to 1.5-ml polypropylene tubes, centrifuged at 10,000 ϫ g, and [ 3 H]AA (and metabolites) cpm were determined by scintillation counting. Results were controlled for equivalent number of cells quantified by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay as described (13) and by verification of total AA labeling by scintillation counting.
RNA Interference (RNAi)-Sequence-specific silencing of cPLA 2 was performed essentially as described using sequence-specific small interfering RNA reagents (10, 14, 15) using human cPLA 2 RNAi starting 55 nucleotides from the start codon (GUUUACGGUAGUGGUGUUAdTdT and UAACACCACUACCGUAAACdTdT) for A549 cells and mouse cPLA 2 RNAi starting 300 nucleotides from the start codon (ACCCUAG-GCACAGCUACAUdTdT and AUGUAGCUGUGCCUAGGGUdTdT) for J774.1 cells. All sequences were evaluated against the data base using the NIH BLAST program to test for specificity (Xeragon). A549 or J447.1 cells (5 ϫ 10 4 ) were transfected with the 21-nucleotide duplexes using OligofectAMINE (Invitrogen) as recommended by the manufacturer. Following the 4-h transfection, cells were labeled overnight with 5 Ci/ml [ 3 H]arachidonic acid (5 nM). Cells were washed and placed in Dulbecco's modified Eagle's medium supplemented with 2% fetal bovine serum for 2 h. The optimal down-regulation time for cPLA 2 was 48 h post-transfection as judged by Western immunoblotting with an anti-cPLA 2 monoclonal antibody anti-cPLA 2 that recognizes both mouse and human cPLA 2 . Total protein was normalized using post-Amido Black staining of polyvinylidene difluoride membranes and by simultaneous ECL development of ␣-tubulin using an anti-␣-tubulin antibody (Santa Cruz Biotechnology).
Large Multilamellar Vesicle Binding Assay-Large multilamellar vesicles for Cer-1-P were produced by drying 68.3 l of a 1 mg/ml D-e-C 18:1 ceramide 1-phosphate under nitrogen for each reaction. A solution (100 l) of 50 mM Tris-HCl, pH 7.4, 150 mM NaCl solution for each reaction was added, and the lipid was vortexed vigorously for 2 min. Cer-1-P LMVs (100 l) were mixed with buffer A (200 l) (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 4 mM EDTA), and for calcium-containing reactions Cer-1-P LMVs were mixed with 4.4 mM CaCl 2 . The binding reaction was initiated by the addition of 100 l of 50 mM Tris-HCl, pH 7.4, 150 mM NaCl containing 0.2 g of cPLA 2 -GST. After 5 min at room temperature, the reaction was centrifuged at 10,000 ϫ g for 10 min, and the supernatant was transferred to a new 1.5-ml polypropylene tube. 200 l of 1ϫ Laemmli buffer was added to the lipid pellet, and 100 l of 5ϫ Laemmli buffer was added to the 400 l of supernatant. 35 l of the lipid pellet was subjected to SDS-PAGE as described in Fig. 4A. Purified recombinant GST did not bind any of the lipids using this protocol (data not shown).

RESULTS
Ceramide 1-Phosphate Activates cPLA 2 in Cells-We have shown previously (10) that Cer-1-P specifically induces AA release in cells, and the generation of Cer-1-P is required for cytokine-and calcium ionophore-induced eicosanoid synthesis. These results suggested the activation of cPLA 2 in response to Cer-1-P. Therefore, at first, it became important to establish that cPLA 2 is the specific enzyme responsible for AA release in response to Cer-1-P. To determine whether cPLA 2 is downstream of Cer-1-P, RNAi technology was employed to knock down cPLA 2 . OligofectAMINE transfection of fluoresceintagged RNAi gave a transfection efficiency of 85% at 24 and 48 h in A549 cells (Table I). Using these conditions, cPLA 2 expression was down-regulated 81.3% with specific RNAi after 48 h post-transfection in A549 cells (Table I). This resulted in a 79.4% inhibition of Cer-1-P-induced (2.5 M for 3 h) release of AA and metabolites in these cells as compared with scrambled RNAi and sham controls (Fig. 1). Normalization of these results to down-regulation of total cPLA 2 showed a nearly total abrogation of the Cer-1-P effects.
In another model of inflammation, J774.1 macrophages, cPLA 2 expression was down-regulated by 61% using specific RNAi after 48 h as judged by Western immunoblotting (Table  II). As with the A549 cells, AA release elicited by treatment with Cer-1-P (2.5 M for 2 h) was inhibited to a similar extent (59%) (Table II). Thus, normalization of these results to downregulation of total cPLA 2 again showed a nearly total abrogation of the Cer-1-P effects (Table II). These data indicate that Cer-1-P induces AA release via activation of cPLA 2 in cells.
Because the above results implicated cPLA 2 as a downstream target for the action of Cer-1-P, studies were undertaken to determine whether Cer-1-P activates cPLA 2 in cells.
To this end, we first examined whether Cer-1-P affects the association of cPLA 2 with cellular membranes because this is a requisite for cPLA 2 to act on its membrane phospholipid substrates. Treatment of A549 cells with Cer-1-P (2.5 M) induced a significant and time-dependent increase in the membrane content of cPLA 2 ( Fig. 2A). Cer-1-P did not significantly increase the total cellular content of immunoreactive cPLA 2 . Thus, Cer-1-P induces the translocation of cPLA 2 to cell membranes without effects on cPLA 2 expression.
To verify that cPLA 2 translocated in intact cells, the effects of Cer-1-P treatment on the cellular localization of cPLA 2 were examined by immunofluorescence to detect endogenous cPLA 2 in fixed cells as well as using cPLA 2 fused to GFP to study the localization of the overexpressed protein in live cells (Fig. 2B). Treatment of A549 cells with Cer-1-P caused the translocation of cPLA 2 to membranes in a pattern consistent with the Golgi apparatus and perinuclear membranes (12,16). This was shown for both the endogenous enzyme (Fig. 2B, IF) and the GFP fusion (Fig. 2B, GFP). To demonstrate that Cer-1-P treatment recapitulates the reported pattern of translocation of cPLA 2 to Golgi membranes in response to pro-inflammatory a Time from the end of the RNAi transfection to analysis of cPLA 2 expression and the start of the assay for C-1-P-induced AA release. b Mean percentage of the number of cells demonstrating intracellular fluorescence of RNAi (1 ϫ 10 3 cells scored). c Mean percentage of the down-regulation of immunoreactive cPLA 2 normalized to total protein and ␣-tubulin as compared with scrambled (control) RNAi (100%).
d Mean percentage of the inhibition of C-1-P-induced AA release (2.5 M C-1-P for 2 h) as compared with scrambled (control) RNAi (100%). e Percentage of the inhibition of C-1-P-induced AA release (2.5 M C-1-P for 2 h) as compared with scrambled (control) RNAi (100%) normalized to the percent of cPLA 2 down-regulation.
f ND, not determined.
agonists (12,16), the immunofluorescent signal of Cer-1-Pactivated cPLA 2 (Fig. 2C, red) was overlaid with the fluorescent organelle marker for the Golgi apparatus, GM130 (Fig. 2C, green) (12,16,34). Significant co-localization was observed between GM130 and Cer-1-P-activated cPLA 2 demonstrating that cPLA 2 is translocated to Golgi membranes in response to Cer-1-P (Fig. 2C, Overlay, yellow). Because the translocation/activation of cPLA 2 in cells may also be regulated by phosphorylation or by the interaction of lipids with the Ca 2ϩ -dependent lipid binding domain (CaLB/C2 domain) of cPLA 2 (16 -19), the cellular localization of the CaLB/C2 domain of cPLA 2 fused to GFP (lacking the catalytic domain and, thus, the regulatory phosphorylation sites of cPLA 2 ) was examined in response to Cer-1-P treatment. Cer-1-P induced the translocation of the GFP-CaLB/C2 domain to Golgi and perinuclear membranes (Fig. 2D) similar to the translocation of endogenous cPLA 2 and GFP-cPLA 2 in response to Cer-1-P. Furthermore, Cer-1-P treatment had no effect on the phosphorylation of the cPLA 2 regulatory site, Ser-505, as judged by Western immunoblotting (data not shown). Thus, Cer-1-P likely induces the translocation of cPLA 2 through the CaLB/C2 domain, independent of increased phosphorylation of cPLA 2 on Ser-505 and independent of the catalytic activity of cPLA 2 or other functionalities outside the CaLB/C2 domain.
Ceramide 1-Phosphate Directly Binds and Activates cPLA 2 in Vitro -The CaLB/C2 domain of cPLA 2 has homology to the C2 domain of protein kinase C, which binds anionic phospholipids (3). Because Cer-1-P, an anionic phospholipid, induced the translocation of cPLA 2 through the CaLB/C2 domain, we wondered whether Cer-1-P directly activated cPLA 2 through this domain. To test this hypothesis, we first examined whether Cer-1-P increased the enzymatic activity and affected the Ca 2ϩ affinity of cPLA 2 in vitro. In the presence of saturating Ca 2ϩ and near physiological salt (150 mM NaCl), Cer-1-P increased the activity of cPLA 2 over 2-fold in a dose-dependent manner (Fig. 3A). Furthermore, the effect of Cer-1-P on cPLA 2 enzymatic activity was calcium-dependent, because in the absence of calcium, Cer-1-P was unable to activate the enzyme (basal activity in Fig. 3B). The removal of salt from the assay had no effect on the ability of Cer-1-P to increase the activity of cPLA 2 (data not shown). Using the concentration of Cer-1-P that induced maximal activation (50 M) of cPLA 2 , the enzymatic activity of cPLA 2 was examined with varying concentrations of Ca 2ϩ . In the presence of Cer-1-P, the EC 50 of calcium was lowered from 191 to 31 nM (Fig. 3B). Thus, Cer-1-P can directly activate cPLA 2 in Ca 2ϩ -dependent manner, and Cer-1-P increases the affinity of cPLA 2 for calcium analogous to the effect phosphatidylserine imparts on protein kinase C (20).
The second approach was to examine whether Cer-1-P bound directly to cPLA 2 using a protein-lipid overlay assay. In the presence of Ca 2ϩ , cPLA 2 was found to bind as little as 5 nmol of Cer-1-P (Fig. 4). cPLA 2 was also found to bind as little as 5 nmol of dioleoyl-PI although to a lesser extent than Cer-1-P. On the other hand, cPLA 2 did not bind to dioleoyl-PA, dioleoyl-PC, or dioleoyl-PS when 5 nmol was bound to the membrane (Fig. 4,  left). Weak binding of cPLA 2 to dioleoyl-PA, dioleoyl-PS, and dioleoyl-PC was observed at lipid concentrations greater than 25 nmol (data not shown). cPLA 2 did not bind to diacylglycerol, sphingosine 1-phosphate, sphingosine, or ceramide at any lipid concentration (Fig. 4, right).
Next, we used a LMV assay described previously for cPLA 2 (17). cPLA 2 did not significantly bind Cer-1-P in the absence of Ca 2ϩ , but the binding of cPLA 2 to Cer-1-P was dramatically increased in the presence of saturating Ca 2ϩ (Fig. 5A). Using this assay and varying the free Ca 2ϩ concentration, the K Ca was determined to be 1.54 M, which is almost a magnitude lower than the published K Ca for PC (10 -15 M) and PS (14 M) (Fig. 5B) (18). Thus, to date, Cer-1-P demonstrates the lowest K Ca for cPLA 2 . Furthermore and importantly, the binding of cPLA 2 to Cer-1-P was significantly enhanced in the presence of physiological calcium.
The region of cPLA 2 that binds Cer-1-P was also determined by applying the LMV assay using recombinantly expressed CaLB/C2 and the catalytic domains of cPLA 2 . The CaLB/C2 domain of cPLA 2 was found to bind Cer-1-P in a calciumenhanced manner comparable with full-length cPLA 2 . On the other hand, cPLA 2 did not bind Cer-1-P when the CaLB/C2 domain was removed (catalytic domain). Thus, cPLA 2 interacts with Cer-1-P via the CaLB/C2 domain (Fig. 5C).

DISCUSSION
In this study, Cer-1-P was found to be a novel and direct activator of cPLA 2 ; thus, this now suggests that Cer-1-P is a bioactive lipid in biological systems. Previously, Cer-1-P was reported to have effects on cell proliferation, DNA synthesis, and phagocytosis, but a direct target was not identified or hypothesized (21,22). This study found that cPLA 2 specifically interacted with Cer-1-P in a calcium-dependent manner and provided a mechanism for the previous observation that Cer-1-P was required for the induction of AA release in response to agonists that induce cPLA 2 activation (10). Therefore, with the identification of a direct and specific intracellular target for Cer-1-P, this sphingolipid now fulfills the following criteria of a bioactive lipid joining the ranks of other bioactive lipids such as its close brethren, ceramide and sphingosine 1-phosphate. First, exogenous Cer-1-P induces a specific biochemical and cellular response (release of AA), and this action of Cer-1-P demonstrates lipid specificity in the induction of cPLA 2 interaction/activation, AA release, and eicosanoid synthesis. Second, endogenous Cer-1-P reproduces these effects specifically, and Cer-1-P levels are regulated in response to agonists. Third, the generation of Cer-1-P is required for cPLA 2 activation, AA release, and eicosanoid synthesis. Lastly, a direct target (cPLA 2 ) that binds/interacts with and is activated by Cer-1-P has been identified. Therefore, the presented study has identified Cer-1-P as a possible new messenger lipid in biological systems.
This study also demonstrates that Cer-1-P is a direct activator of cPLA 2 through interaction with the CaLB/C2 domain. This ability of the CaLB/C2 domain of cPLA 2 to interact with an anionic lipid in a calcium-dependent manner is a somewhat unexpected finding in the cPLA 2 field. The current "dogma" of lipid/CaLB domain interaction is one that proposes this interaction to be via hydrophobic forces alone and not by electrostatic interaction. It is currently accepted that the CaLB/C2 domain of cPLA 2 binds to zwitterionic (e.g. PC) and neutral lipids in a calcium-dependent manner (17,18). An exception to this was the report that a synthetic non-natural anionic lipid, phosphatidylmethanol, interacted with the CaLB/C2 domain of cPLA 2 in a calcium-dependent and stereospecific manner (18,  23,24). Several reports have also alluded to the fact that the membrane binding of the CaLB/C2 domain of cPLA 2 cannot be explained by hydrophobic interactions alone (25,26), and the binding of cPLA 2 to PC could not be analyzed by crystallography techniques probably because of the low affinity (26). The high K Ca (10 M) of PC for the interaction with cPLA 2 also argues against a physiologic role for this lipid in regulating the membrane binding of this enzymes in cells, because the first report on cPLA 2 /membrane interactions demonstrated that the enzyme associated with membranes in as little as 300 nM free calcium (3,8,16). In this study, a significant increase (2-fold) in the binding of Cer-1-P to cPLA 2 was observed at 300 nM correlating with these early findings. Furthermore, physiologic salt induces the interaction of PC with cPLA 2 without the necessity of free calcium, thus abolishing the calcium-dependence of the interaction (5,6). On the other hand, physiologic salt had no effect on the ability of Cer-1-P to induce activation of cPLA 2 or on the ability of cPLA 2 to associate with Cer-1-P in a calciumdependent manner. Thus, to date, Cer-1-P demonstrates the lowest K Ca and the greatest calcium dependence for interaction with cPLA 2 . The results from this study coupled with our previous findings that the ceramide kinase/Cer-1-P pathway is FIG. 5. cPLA 2 binds Cer-1-P in a calcium-dependent manner. A, characterization of cPLA 2 -GST binding to natural ceramide 1-phosphate by using a LMV assay. Data are representative of three separate determinations on two separate occasions. B, the effect of the free calcium concentration on the binding of cPLA 2 to Cer-1-P. The LMV binding assay was repeated for Cer-1-P and cPLA 2 -GST using various concentrations of free calcium. Free calcium was determined for HEDTA/ CaCl 2 mixtures using the Maxchelator program. The K Ca was determined using the ligand binding module of the sigma plot program. Data are representative of three separate determinations on two separate occasions. C, determination of the cPLA 2 domain that binds Cer-1-P. Using the LMV assay for Cer-1-P and cPLA 2 , the GST-CaLB/C2 domain of cPLA 2 or the GST-catalytic domain of cPLA 2 was used in place of full-length cPLA 2 . Data are representative of three separate determinations on two separate occasions. required for PLA 2 activation in response to calcium ionophore and cytokines suggest that Cer-1-P is a "missing link" in the regulation of eicosanoid pathways (10).
The demonstration that Cer-1-P is a direct activator of cPLA 2 provides insight into the mechanism of cPLA 2 translocation in response to agonists. Previously, it has been reported (27) that sphingomyelin inhibits if not repulses cPLA 2 from membranes. Numerous reports have established that sphingomyelin is hydrolyzed to ceramide by sphingomyelinase in response to inflammatory cytokines such as interleukin-1␤ and tumor necrosis factor ␣ (28,29). Because Cer-1-P is produced via the phosphorylation of ceramide, the production of ceramide in response to cPLA 2 -activating agonists is logical. Thus, one can hypothesize that in response to an inflammatory agonist (e.g. cytokines), sphingomyelin is hydrolyzed to ceramide followed by conversion of the ceramide to Cer-1-P by ceramide kinase. This possible mechanism would remove an inhibitor/repulsor from the membrane, reduce the rigidity of the membrane by producing ceramide, thereby increasing the penetration ability of the enzyme, and produce the activating lipid, Cer-1-P. This hypothesis correlates with our recent demonstration that Cer-1-P levels and ceramide kinase activity are increased in response to interleukin-1␤. Furthermore, this hypothesis explains the reports that exogenous ceramide and sphingomyelinase C, which have little to no effect on their own, synergize with cytokines and calcium ionophore to induce cPLA 2 translocation/activation, AA release, and prostanoid synthesis (27,30,31).
The question remains whether the generation of Cer-1-P is sufficient to elicit the translocation/activation of cPLA 2 or whether calcium mobilization in conjunction with the generation of Cer-1-P is required. Clearly, our previous findings (10) demonstrated that Cer-1-P is required, because the down-regulation of ceramide kinase, and thus Cer-1-P abolishes calcium ionophore-induced AA release as well as inhibiting ATP-induced AA release. Cer-1-P has been reported to induce calcium mobilization in thyroid FRTL-5 cells, but another recent report demonstrated that Cer-1-P did not mobilize calcium in neutrophils (32,33). Findings from our laboratory 2 are in agreement with the latter report in that Cer-1-P did not induce calcium mobilization in A549 cells using a FURA-2-based assay. This does not, however, conclusively demonstrate that Cer-1-P is not affecting calcium homeostasis in a very compartmentalized manner in A549 cells as in calcium microdomains where calcium concentrations can be increased to several hundred micromoles. Because the hydrolysis of sphingomyelin with a subsequent increase in ceramide (the metabolic precursor of Cer-1-P) would affect membrane fluidity, a compartmentalized increase in calcium is certainly possible. Furthermore, our data clearly support a calcium-dependent mechanism of cPLA 2 activation by Cer-1-P. Thus, a co-activation mechanism where calcium concentrations are increased in a highly compartmentalized manner by Cer-1-P and then Cer-1-P in conjunction with calcium activates cPLA 2 is a logical hypothesis.
The possibility that ceramide kinase and Cer-1-P generation are upstream of cPLA 2 activation raises the possibility of the development of a new generation of therapeutics for inflammatory disorders. Therapeutics based on ceramide kinase would have the benefit of blocking AA liberation as well as the un-wanted formation of leukotrienes and COX-1-derived prostanoids (e.g. thromboxanes) possibly lowering the problems of side effects associated with both selective and non-selective COX inhibitors. Because non-steroidal anti-inflammatory drugs are also being used to treat cancer, this newly described pathway may have widespread applications.