Ceramide l-Phosphate, a Novel Phospholipid in Human Leukemia (HL-60) Cells SYNTHESIS VIA CERAMIDE FROM SPHINGOMYELIN*

The present, studies define the novel phospholipid ceramide l-phosphate in these cells and present evidence for formation of this compound by preferential utilization of ceramide derived from sphingomyelin.

Prior studies demonstrated that conversion of sphingomyelin to ceramide via sphingomyelinase action resulted in the generation of free sphingoid bases and inactivation of protein kinase C in human leukemia (HL-60) cells  J. Biol. Chem. 264,7617-7623).
The present, studies define the novel phospholipid ceramide l-phosphate in these cells and present evidence for formation of this compound by preferential utilization of ceramide derived from sphingomyelin.
A ceramide l-phosphate standard, prepared enzymatically via diacylglycerol kinase, was utilized for localization.
In cells labeled to equilibrium with 32Pi to label the head group of the molecule, the basal ceramide l-phosphate level was 30 f 2 pmol/106 cells. Generation of ceramide via the use of exogenous sphingomyelinase resulted in time-and concentrationdependent formation of ceramide l-phosphate. As little as 3.8 x lo-' units/ml was effective and a 3-fold increase was observed with a maximal concentration of 3.8 x  In sum, these studies demonstrate the novel phospholipid ceramide l-phosphate in HL-60 cells and suggest the possibility that a path exists from sphingomyelin to ceramide l-phosphate via the phosphorylation of ceramide.
A renewed interest in the metabolism of sphingomyelin has developed over the past few years (l-lo).
This is based primarily on the observation that sphingomyelin metabolism *This study was supported by Grants ROl-Ca-42385 from the National Institutes of Health and FRA-345 from the American Cancer Society. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. may initiate a sequence of events resulting in the generation of free sphingoid bases, potential inhibitors of protein kinase C (11). Prior studies from this laboratory demonstrated that sphingomyelinase treatment of GH, pituitary cells resulted in the quantitative generation of ceramide from sphingomyelin (1). A portion of the ceramide was subsequently converted to sphingoid bases. This correlated with inhibition of phorbol ester-mediated activation of protein kinase C (2). Sphingomyelinase treatment of HL-60 cells similarly stimulated the generation of free sphingoid bases and inactivated protein kinase C (7). Under these conditions, phorbol ester-induced differentiation of these cells into macrophages was also inhibited. These studies were interpreted as evidence for the concept that sphingomyelin metabolism might serve to initiate a feedback inhibitory pathway for protein kinase C. Little is actually known of the regulation of ceramide metabolism. This compound not only serves as precursor to the generation of free sphingoid bases, but is of central importance in the formation of glycosphingolipids and sphingomyelin. Another possible path for the metabolism of ceramide would be to ceramide phosphate. This event has not been previously demonstrated in intact cells. However, the recent report of the existence of a specific ceramide kinase in rat, brain synaptic vesicles suggests this event might be of physiologic significance (12). The present studies demonstrate the existence of ceramide l-phosphate in HL-60 cells and show that it may be generated via ceramide metabolism.
Further, these studies suggest, the possibility of a pathway from sphingomyelin to ceramide l-phosphate via the generation of ceramide. 14917 was prepared enzymatically as described by Schneider and Kennedy (13) and adapted by Preiss et al. (14). Briefly, Type III ceramide (0.5 mg) was solubilized in 5 mM cardiolipin, 7.5% octyl$-glucopyranoside, 1 mM diethylenetriaminepentaacetic acid by bath sonication and resuspended in a reaction mixture (50 mM imidazole HCI, pH 6.6, 50 mM NaCl, 12.5 mM MgCl,, 1 mM EGTA) containing diacylglycerol kinase (0.7 unit/ml). The reaction was started by addition of [r-"'P]ATP to a final concentration of 10 mM. After 90 min at room temperature, the reaction was stopped by extraction with chloroform:methanol:concentrated HCl (100:100:1, v/v/v) and 10 mM EDTA. The organic phase was dried under Np, resuspended in chloroform:methanol (l:l, v/v), and resolved by TLC using chloroform:methanol:acetic acid (65:15:5, v/v/v) as solvent as described (14). Ceramide l-phosphate was identified by autoradiography as a single spot at RF 0.25, eluted with chloroform:methanol (l:l, v/v) from the silica gel, and stored at -20 "C under Nz until use.
For short term labeling studies, cells were resuspended (35 X lo6 cells/ml) in BSS containing 5 mCi/ml "'Pi. After 60 min, cells were resuspended in BSS without radiolabel and studied as above. Experiments involving endoglycoceramidase contained a 10 mM sodium acetate buffer with 0.003% bovine serum albumin, pH 6 (diluent) or diluent alone.

Identification
of Ceramide I-Phosphate-For these studies, cells were labeled with "'Pi for 1 h and processed as above. After mild alkaline hydrolysis and resolution by TLC, ceramide phosphate was eluted from the silica gel, dried under Nz, and deacylated in 6 N HCI, l-butanol (l:l, v/v) for 1 h at 100°C according to the method of Kaller (16). Sphingosine phosphate was resolved by TLC using chloroform:methanol:acetic acidwater (25:15:4:3.5, v/v/v/v) as solvent. Recovery of ceramide phosphate as sphingosine phosphate was 65% via this procedure. Sphingosine phosphate was eluted from the silica gel and subjected to periodate oxidation (12). Briefly, the eluted material was dried under N, and resuspended in 0.30 ml of 0.03% methanolic Triton containing 0.03 M aaueous NaIOI. After 24 h at 22 "C, products were extracted by the addition of 0.6 ml of methylene chloride and 0.15 ml of water. The aqueous phase was dried under reduced pressure and glycoaldehyde phosphate was resolved by TLC using l-butanol:acetic acidwater (6:2:2, v/v/v) as solvent (12). Recovery of sphingosine phosphate as glycoaldehyde phosphate was 70% via this procedure. A similar value was reported by Schneider and Kennedv (13). The organic uhase was dried under NV and linid productsofthisreaction were analyzed by TLC as above usingchloroform:methanol:acetic acidwater (25:15:4: Other Procedures-Assuming "isotopic equilibrium," the specific activity of serine as determined from the radioactivity and phosphorus content of sphingomyelin (7,17) was used to quantitate serinecontaining compounds'. A similar approach was used for [3H]palmitate and "'P-labeled compounds.
Statistics-Statistical analysis was performed by t test and linear regression analysis by the method of least squares.
' An alternative approach validated the results obtained with radiolabeled serine. These studies utilized the diacylglycerol kinase reaction to directly measure ceramide levels. Both methods yielded similar values. These were 180 +. 20 pmol/lOG cells for ["Hlserine and 209 f 29 pmol/lO" cells for the radioenzymatic method.

AND DISCUSSION
Initial studies determined the existence of ceramide lphosphate in HL-60 cells. These studies utilized cells labeled for 1 h with "Pi. Fig. 1 compares the ceramide l-phosphate generated enzymatically to a total "'P-labeled phospholipid extract of HL-60 cells. As seen in this autoradiogram (left lane), authentic ceramide l-phosphate migrates as a single spot with an RF 0.25 in this system. A small portion of the total lipid extract (lane 1) co-migrates with the ceramide lphosphate standard. This compound is partially obscured by the mass of radiolabeled phospholipid. If this material is eluted from the silica gel, it co-migrates with authentic ceramide l-phosphate in two additional solvent systems: chloroform:pyridine:formic acid (50:30:7, v/v; RF 0.25) and chloroform:methanol:acetic acidwater (25:15:4:3.5, v/v/v; RF 0.7). To separate this material from the mass of cellular phospholipid, the total lipid extract was subjected to mild alkaline hydrolysis (15). This removes the majority of the glycerophospholipids accounting for 93% of labeled phospholipids under these conditions. The material that co-migrates with ceramide l-phosphate standard, however, was resistant to degradation (lane 2); recovery of material was 90%. HL-60 cells (50 x 106/ml) were resuspended in BSS containing 5 mCi/ml "Pi. After 60 min, cells were washed with BSS without radiolabel and centrifuged at 800 x g for 5 min. The cell pellet was extracted with 1 ml of chloroform:methanol:HCl (lOO:lOO:l, v/v) and 0.3 ml of BSS containing 10 mM EDTA. The organic phase was dried under Nz, resuspended in chloroform:methanol (l:l, v/v), and resolved by TLC using chloroform:methanol:acetic acid (65:15:5, v/v) as solvent as described (14). A portion of the organic phase was subjected to mild alkaline hydrolysis (0.1 M methanolic KOH at 37 "C for 1 h)(l). Lane S, authentic ceramide l-phosphate standard (3000 dpm) prepared enzymatically; lane I, total cellular "'P-labeled phospholipid extract (1.5 x 10 6 cells); lane 2, cellular phospholipid extract after mild alkaline hydrolysis (3 X 10" cells). material co-migrated with glycoaldehyde phosphate derived from authentic ceramide l-phosphate (13). The small amount of material that remained in the organic phase was sphingosine phosphate that had not been oxidized. material from twice as many cells was applied to lane 2 as lane I). Based on this evidence, the cellular material was tentatively identified as ceramide l-phosphate. To further define the cellular phospholipid as ceramide lphosphate, the material was first eluted from the silica gel and then subjected to acid butanol hydrolysis by the method of Kaller (16). This procedure deacylates ceramide l-phosphate to sphingosine l-phosphate. Fig. 2  These studies clearly identify the phospholipid derived from HL-60 cells as ceramide phosphate.
To establish the position of the phosphate group as either in the first or third position, ["Plsphingosine phosphate was subjected to periodate oxidation (12,13). This procedure converts sphingosine l-phosphate to glycoaldehyde phosphate, a hydrophilic molecule extracted into the aqueous phase. In contrast, this procedure converts ceramide 3-phosphate to radiolabeled long chain aldehydes which remain in the organic phase. In fact, 70% of the sphingosine phosphate derived either from the cellular material or authentic sphingosine l-phosphate was converted to an aqueous extractable form. A similar result was obtained by Schneider and Kennedy (13). When this aqueous material was analyzed by TLC using l-butanol:acetic acidwater (6:2:2, v/v) as solvent, the cellular 'I The lightly labeled upper portion of the sphingosine l-phosphate spot most likely represents an allylic rearrangement of the molecule known to occur during acid hydrolysis of this class of compounds (18).
clearly establish the existence of ceramide l-phosphate in HL-60 cells.
Previous studies in HL-60 cells demonstrated that exogenous sphingomyelinase induced a time-and concentrationdependent reduction in the level of sphingomyelin and a quantitative increase in the level of ceramide (7). Sphingomyelin decreased from 560 to 350 pmol/lO" cells and ceramide increased from 180 to 370 pmol/lO" cells. As little as 3.8 x lo-" units/ml sphingomyelinase was effective and a maximal effect occurred with 3.8 x lo-' units/ml. These events were detectable by 5 min and maximal at 15 min of stimulation.
Since ceramide may serve as precursor to ceramide lphosphate in uitro, the ability of cells to utilize endogenous ceramide, generated via sphingomyelinase (3.8 x lo-' units/ ml) action, was assessed. These studies used cells labeled to isotopic equilibrium with "P> and resuspended in medium without radiolabel.
Under these conditions, the content of radiolabel correlates directly with the phospholipid mass. Fig.  3 shows that in control incubations the basal level of ceramide l-phosphate decreased from 30 to 26 pmol/lO' cells whereas sphingomyelinase induced a time-dependent increase of ceramide l-phosphate from 30 to 57 pmol/lO" cells 0, < 0.001). The increase in the level of ceramide l-phosphate was detectable by 5 min and maximal at 30 min. Fig. 4 demonstrates the concentration dependence of sphingomyelinase-induced generation of ceramide l-phosphate.
The upper panel demonstrates that as little as 3.8 x lo-" units/ml sphingomyelinase was effective and a maximal effect to 3-fold of control was achieved with 3.8 x lo-' units/ml sphingomyelinase; EDso E 2 x lo-' units/ml. An autoradiogram of a typical experiment is shown in the lower panel.
Only three phosphorylated spots were detected; ceramide lphosphate, a spot that often but not always separated from 05   the origin and contained sphingomyelin and other lipids, and a faint spot at RF 0.61. This latter spot co-migrated with phosphatidic acid and may represent a dialkyl form resistant to mild alkaline hydrolysis. Only ceramide l-phosphate was consistently observed and responsive to sphingomyelinase in this system. An additional set of studies was performed with cells radiolabeled with [3H]serine to preferentially label the sphingoid base backbone of the molecule. For these studies, cells were labeled to "isotopic equilibrium" with [3H]serine and resuspended in medium without radiolabel. Fig. 5 again shows that sphingomyelinase induced a concentration-dependent increase in the level of ceramide l-phosphate at 45 min of stimulation. The basal level of ceramide l-phosphate was 39 f 5 pmol/lO" cells. This value is very similar to that derived with cells labeled with "P,. An increase in the level of ceramide l-phosphate was detectable with as little as 3.8 x 10e6 units/ml sphingomyelinase and a maximal effect to 86 pmol/ lo6 cells was achieved with 3.8 X 10e3 units/ml; EDso E 5 X lo-" units/ml. Similarly, sphingomyelinase induced a maximal 2.5-fold increase in ceramide l-phosphate in studies (n = 2) performed with cells labeled to equilibrium with ["HI palmitate.
[3H]Palmitate measures both the fatty acid moiety and the sphingoid base backbone (8). A basal level of 39 pmol/ 10" cells was again derived with this probe. Hence, similar basal and sphingomyelinase-stimulated levels of ceramide l- These studies were performed as described in Fig. 4  To determine whether the phosphate moiety might be derived from sphingomyelin directly, perhaps via contamination of the sphingomyelinase preparation with a phospholipase D, or via phosphorylation of ceramide, studies were performed with cells short term labeled with 32Pi. Under these conditions, sphingomyelin is virtually unlabeled and hence cannot contribute radiolabeled phosphate to ceramide l-phosphate. Fig.  6 demonstrates the concentration-dependent effects of sphingomyelinase on 32Pi incorporation into ceramide l-phosphate and sphingomyelin. Sphingomyelinase-induced "*Pi incorporation into ceramide l-phosphate was detectable with as little as 3.8 x lO-'j units/ml sphingomyelinase to 250 & 60% of control and a maximal 12-fold increase in labeling was achieved with 3.8 X 10m3 units/ml; EDso E 7 X lo-" units/ml. In contrast, 32P; incorporation into sphingomyelin was minimal in basal incubations and did not change with sphingomyelinase stimulation. Hence, sphingomyelin does not serve as the direct precursor for the phosphate moiety of ceramide l-phosphate.
The specificity of the ceramide phosphorylation reaction for ceramide derived from sphingomyelin was evaluated. These studies compared the effect of sphingomyelinase and endoglycoceramidase on ceramide and ceramide l-phosphate formation. The upperpanel of Fig. 7 demonstrates that sphingomyelinase (2 X 10e4-1 X 10m3 units/ml) and endoglycoceramidase (3 x 10m5-1 X 10e3 units/ml) induce similar elevations in the level of ceramide as measured by the diacylglycerol kinase assay (14). In contrast, only sphingomyelinase action induced "*Pi incorporation into ceramide l-phosphate in cells short term labeled with "*Pi and resuspended in medium without radiolabel (lower panel). These studies demonstrate selective utilization of ceramide derived from sphingomyelin and suggest the possibility of a pathway from sphingomyelin to ceramide l-phosphate via the generation of ceramide.
In sum, the present studies describe the novel phospholipid to mild alkaline hydrolysis as described in Fig. 3. Ceramide was quantitated by the diacylglycerol kinase assay as described (14). These data (mean f S.E.) represent duplicate determinations from three experiments. Lower panel, "P, incorporation into ceramide l-phosphate. HL-60 cells, short term labeled with '*P,, were handled as described in Fig. 6. These data (mean + S.E.) represent duplicate determinations from three experiments.
systems and was converted to sphingosine phosphate by acidic butanol hydrolysis. Sphingosine phosphate was further oxidized to glycoaldehyde phosphate by periodate treatment indicating the site of the phosphorylation as the first position. Additionally, elevation of ceramide levels from endogenous sphingomyelin stores resulted in ceramide l-phosphate formation. Identical results were obtained whether the probe utilized measured the backbone or the head group of the molecule. Acute labeling studies with 32Pi clearly demonstrated that the phosphate group was not derived directly from sphingomyelin but rather by phosphorylation of ceramide. This was anticipated since sphingomyelin degradation was achieved by sphingomyelinase action. Additionally, this effect appeared specific for ceramide derived from sphingomyelin since ceramide from glycosphingolipids was not converted to ceramide l-phosphate. Thus, these studies suggest the possibility that a pathway exists from sphingomyelin to ceramide l-phosphate via the generation of ceramide. Similarly, Slife et al (19) found that ceramide derived from sphingomyelin but not glycosphingolipid served as precursor for sphingoid base formation in rat liver plasma membranes.
Although the present studies do not address the enzymatic mechanism involved in this process, preliminary studies4 suggest that HL-60 cells contain a specific ceramide kinase activity distinct from diacylglycerol kinase activity (12). This series of events appears analogous to the initial arm of the phosphoinositide pathway. In both instances, phospholipid hydrolysis via a phospholipase C-like mechanism yields a compound of similar structure which is subsequently phosphorylated. Whether ceramide l-phosphate may be synthesized in response to physiologic agonists such as l-a,25 dihydroxyvitamin Da (10) or may serve as a precursor to the formation of other phosphosphingolipids (20) is presently under investigation.