Ceramide 1-phosphate, a mediator of phagocytosis.

The agonist-stimulated metabolism of membrane lipids produces potent second messengers that regulate phagocytosis. We studied whether human ceramide kinase (hCERK) activity and ceramide 1-phosphate formation could lead to enhanced phagocytosis through a mechanism involving modulation of the membrane-structural order parameter. hCERK was stably transfected into COS-1 cells that were stably transfected with the FcgammaRIIA receptor. hCERK-transfected cells displayed a significant increase in phagocytic index in association with increased ceramide kinase activation and translocation to lipid rafts after activation with opsonized erythrocytes. When challenged with opsonized erythrocytes, hCERK-transfected cells increased phagocytosis by 1.5-fold compared with vector control and simultaneously increased ceramide 1-phosphate levels 2-fold compared with vector and unstimulated control cells. Control and hCERK-transfected cells were subjected to cellular fractionation. Utilizing an antibody against hCERK, we observed that CERK translocates during activation from the cytosol to a lipid raft fraction. The plasma membrane-structural order parameter of the transfectants was measured by labeling cells with Laurdan. Cells transfected with hCERK showed a higher liquid crystalline order than control cells with stimulation, conditions that are favorable for the promotion of membrane fusion at the sites of phagocytosis. The change in the structural order parameter of the lipid rafts probably contributes to phagocytosis by promoting phagosome formation.

The second messengers produced by membrane lipids through agonist stimulation include not only glycerolipids but also sphingolipids. Sphingolipids are comprised of lipids that contain a long chain sphingoid base. Sphingolipids, in addition to being structural components of membranes, regulate cell-cell and cell-substrate interactions, proliferation, and differentiation. Members of this diverse group of lipids have emerged as a novel class of signaling molecules that also regulate phagocytosis. The mechanisms by which sphingolipids exert these effects remain incompletely defined. More than a decade ago, it was found that ceramide can be phosphorylated to ceramide 1-phosphate (C1P) 1 (1)(2)(3).
C1P is found in brain synaptic vesicles, and it is thought to play a role in regulating the secretion of neurotransmitters by promoting the fusion of vesicle membranes (4). Ceramide kinase activity exists in HL-60 cells where C1P is derived from ceramide released from sphingomyelin (5). Human ceramide kinase (hCERK) was recently cloned based on the homology to two isoforms of mice and human sphingosine kinase (6). The expressed kinase displayed specific ceramide phosphorylating activity. BLAST search analyses using the hCERK sequence revealed that a series of putative CERKs exist in a variety of cellular organisms, including plants, nematodes, insects, and vertebrates. CERKs represent a new class of lipid kinases that are clearly distinct from sphingosine and diacylglycerol kinases (6,7).
C1P shares structural homology with phosphatidic acid, a lipid shown to be highly fusogenic (8). In an earlier study, the generation of C1P was also observed to occur through the activation of a ceramide kinase in polymorphonuclear leukocytes (7). In this model, C1P was found to be a potent fusogen. The addition of C1P to liposomes shifted the rate and extent of calcium-dependent fusion (7). Thus C1P was postulated to play a role in phagosome formation. Subsequently, it was reported that the CERK mediates Ca 2ϩ -dependent degranulation in mast cells (9). Others have also found that A549 cells transfected with the hCERK can induce arachidonate release and prostanoid synthesis (10).
Neutrophils are difficult to maintain in vitro for prolonged periods of time and are not amenable to transfection. Therefore, other model systems are often employed to probe the mechanism of phagocytosis and phagosome formation. COS-1 cells are one such model system. COS-1 cells acquire a phagocytic phenotype when transfected with the Fc␥RIIA receptor. In the present study, we evaluated the effects of transfecting COS-1 cells expressing Fc␥RIIA with the cDNA for hCERK on phagocytosis. Using the fluorescent probe Laurdan, we employed high-speed microspectrophotometry to examine the mechanism by which C1P enhances phagocytosis. We found that C1P increased during phagocytosis and that a distinct gel-like order lipid phase (Lo) was formed within the cell membrane at sites of C1P formation in the hCERK-transfected cells.
Cell Culture-COS-1 cells, stably transfected with Fc␥RIIA receptor cDNA, were maintained in Dulbecco's modified Eagle's medium containing 4.5 mg/ml glucose, 25 mg/ml glutamine, 100 units/ml streptomycin, 100 mg/ml penicillin, and 10% heat-inactivated fetal calf serum (11). To create double expression transfectants with Fc␥RIIA and hCERK, COS-1 cells transfected stably with Fc␥RIIA were cultured in 35-mm dishes. At ϳ80% confluence, the cells were transfected with 1 g/ml pcDNA-hCERK expression vector using Lipofectamine Plus in 1 ml of Opti-MEM medium. 1 ml of Dulbecco's modified Eagle's medium containing 20% fetal bovine serum was added after 3 h of incubation at 37°C, 5% CO 2 . To establish stable transfectants, cells were cultured with medium containing 200 g/ml G-418 and 200 g/ml Zeocin to select double transfectants.
Sheep Erythrocytes-Sheep red blood cells were sensitized with rabbit IgG anti-sheep erythrocyte antibody as previously described (12). Phagocytosis assays were conducted as previously described (13).
hCERK Expression Vector-A plasmid carrying the hCERK gene was kindly provided by Dr. Takafumi Kohama (Sankyo Co, Ltd.). PCR amplifications were employed for 35 cycles with steps at 94°C for 1 min, 60°C for 1 min, and 72°C for 1.5 min with Platinum Pfx DNA polymerase (Invitrogen) and CERK plasmid as templates. Primers used for PCR included hCERK-HindIII (5Ј-TTATATTTATCCCAAGCTTGG-GATGGGGGCGACGGGGGCGGCGGA-3Ј) and hCERK-XhoI (5Ј-ATACCGCTCGAGCGGTGACTCTTCCTCCTCGATTCCCCGAGCAAA-3Ј). PCR products were ligated into the pCR4-TOPO vector (Invitrogen) followed by cloning, purification of plasmids, and sequencing. The entire open reading frame of the hCERK was excised at HindIII and XhoI sites from the plasmid described above. The open reading frame was then subcloned into the HindIII and XhoI sites of a pcDNA3.1/Zeo(ϩ) (Invitrogen) vector, which was modified to encode a C-terminal c-Myc tag to express a c-Myc-hCERK fusion protein.
Construction of Catalytically Inactive hCERK-A search of the hCERK amino acid sequence for signaling domains using the SMART search tool revealed a similarity in residues 132-278 of hCERK to the putative diacylglycerol kinase (DGK) catalytic domain. One region of the hCERK sequence showing highest similarity corresponded to the proposed ATP-binding site of the DGK catalytic domain. This region in DGK displays some resemblance to the glycine-rich loop within the ATP-binding site of many protein kinases but is more divergent in hCERK. The glycine residue in this region is known to be essential for DGK catalytic activity, because mutation of this residue to aspartate ablates activity in all of the DGK family members examined, including the DGK region of sphingosine kinase. Therefore, in an attempt to produce a catalytically inactive hCERK, the corresponding glycine (Gly 198 ) was mutated to aspartate in hCERK. A mutation was introduced in Myc-tagged hCERK in the pcDNA3 expression vector via mutagenic oligonucleotides using primers G198D-F (forward, 5Ј-CG-GAGATGATATGTTCAGCGAG-3Ј) and G198D-R (reverse, 5Ј-GCT-GAACATATCATCTCCGCCGAC-3Ј) by overlap extension methodology.
Stably transfected COS-1 cells were labeled with D-erythro-[ 3 H]sphingosine (2 mCi/ml) for 24 h in Dulbecco's modified Eagle's medium at 37°C. After incubation, the cells were removed with trypsin EDTA. The reaction was terminated by the centrifugation of the COS-1 cells, removal of the supernatant, and the addition of 2 ml of methanol. After transferring the samples to a glass tube, methanol and chloroform were added and the samples were sonicated and centrifuged for 30 min at 3000 ϫ g. The supernatant was saved, and the pellets were reextracted with chloroform/methanol (2:1, v/v). The supernatants were combined with those of the first extraction, and 1 M NaCl was added. After centrifugation at 2000 ϫ g, the lower layer was transferred to another tube and subjected to mild alkaline hydrolysis. The lipids were dried under nitrogen, dissolved in chloroform/methanol (2:1), applied to high performance TLC plates, and developed using the system described above. The corresponding spots for C1P were identified with a radiolabeled standard. Cells were labeled with D-erythro-[ 3 H]sphingosine for 24 h. Following phagocytosis, the medium was changed with phosphatebuffered saline containing Ca 2ϩ and Mg 2ϩ and cells were challenged with erythrocytes (EIgG). The extraction was performed as outlined for phosphate labeling.
Lipid Raft Isolation-Fc␥RIIA, Fc␥RIIA/vector, and Fc␥RIIA/ hCERK-transfected COS-1 cells were challenged with opsonized EIgG and the EIgG not ingested were lysed. COS-1 cells were harvested by scraping into buffer containing 20 mM Tricine (pH 7.8), 0.25 M sucrose, and 1 mM EDTA. The cells were washed and then disrupted in the same buffer with 30 strokes in a Wheaton tissue grinder. The postnuclear supernatant was obtained by centrifugation of the cell lysates at 100 ϫ g for 10 min. The plasma membrane fraction was removed from 30% Percoll 0.25 M sucrose buffer after centrifugation at 84,000 ϫ g for 30 min. Caveolae membrane fractions were isolated from the purified plasma membrane fractions using Opti-Prep gradients as previously described (14). Aliquots of membrane fractions and the caveolae/lipid raft fractions were analyzed by SDS-PAGE. Proteins were transferred to a Polyvinylidene difluoride membrane and subjected to Western blotting using anti-caveolin-1, anti-c-Myc antibodies, and an antibody against hCERK.
Ceramide Assay-Lipids were extracted by the method of Van Veldhoven et al. (15). Total cellular ceramide was assayed by the method of Preiss et al. (16).
Antibody Generation and Immunoblotting-Human embryonic kidney 293 cells were transfected with vector or V5-CERK using Lipofectamine 2000 (Invitrogen) according to the manufacturer's recommendations. Cells were lysed after 24 h in modified radioimmune precipitation assay buffer for 20 min at 4°C. The cell lysates were centrifuged for 15 min at 15,000 ϫ g at 4°C. The supernatant proteins were separated on SDS-PAGE and transblotted to nitrocellulose. The blots were blocked with 5% nonfat dry milk for 1 h at room temperature followed by incubation with a polyclonal rabbit antiserum raised against a C-terminal hCERK peptide (RLFARGIEENPKPDSHS, 1:1000). After washing, the blots were incubated with a secondary antibody conjugated to horseradish peroxidase for 1 h at room temperature and immunopositive bands were visualized by chemiluminescence.
Treatment of COS-1 Cells with Methyl-␤-cyclodextrin (m␤CD)-COS-1 cells were treated with m␤CD as described by Ottico et al. (17). The medium was discarded, and the dishes were washed three times with 1 ml of Locke's balanced saline solution (156 mM NaCl, 5.6 mM KCl, 3.6 mM NaHCO 3 , 2.3 mM CaCl 2 , 1 mM Mg Cl 2 , 5.6 mM glucose, 5 mM HEPES (pH 7.4)) prewarmed at 37°C. The cells were incubated with 2 ml of Locke's solution in the presence or absence of 5 mM m␤CD at 37°C for 30 min.
Measurement of Membrane Fluidity-Excitation microscopy shows coexisting regions of different generalized polarization in red blood cells, in a renal tubular cell line, and in purified renal brush border labeled with a fluorescent probe, 2-dimethylamino-6-lauroylnaphthalene (Laurdan) (18). COS-1 cells, transfected with Fc␥RIIA, Fc␥RIIA/ hCERK, Fc␥RIIA/vector, and Fc␥RIIA/G198DhCERK, were labeled with Laurdan. Following EIgG binding, microspectrophotometry was performed to assess local changes in membrane fluidity. As a control, regions of the Laurdan emission spectrum that are not sensitive to membrane fluidity were evaluated. The labeling of the COS-1 cells followed the methods of Parasassi et al. (18). Laurdan emission spectra were collected from specific regions of single cells using an imaging spectrophotometer system. Cells were labeled with 2.5 M Laurdan for 25 min at room temperature and then washed at least three times with buffer. To increase light collection efficiency, the bottom port of the microscope was fiber-coupled optically to the input side of an Acton-150 (Acton Instruments, Acton, MA) imaging spectrophotometer. The exit side was connected to a liquid nitrogen-cooled intensifier attached to a Peltier-cooled I-MAX-512 camera (approximately Ϫ20°C) (Princeton Instruments) (19,20). The camera was controlled by a high-speed Princeton ST-133 interface and a Stanford Research Systems (Sunnyvale, CA) DG-535 delay gate generator. Microspectrophotometry used a 355-nm bandpass discriminating filter for excitation and a 405-nm long-pass dichroic mirror. Winspec software (Princeton Instruments) was used to analyze the data.

RESULTS
hCERK distinct from that of diacylglycerol kinase was recently cloned (6). In a preliminary study following transient transfection of Fc␥RIIA-expressing COS-1 cells with hCERK plasmid, a 40% increase of Fc␥RIIA-mediated phagocytosis of EIgG compared with control cells and 20% compared with vector control cells was observed (data not shown). The efficiency of the transient transfection was between 40 and 50% as detected by cotransfection with a ␤-galactosidase expression construct. These results raised the possibility that C1P potentiates phagocytosis. To test this possibility, stable transfectants simultaneously carrying Fc␥RIIA and hCERK or Fc␥RIIA and a catalytically inactive mutant, G198DhCERK, were created. CERK activity in control Fc␥RIIA cells was 0.39 Ϯ 0.11 pmol/ min/10 6 cells. The CERK activity was significantly increased in hCERK-transfected cells (0.7 Ϯ 0.2 pmol/min/10 6 cells, p Ͻ 0.03). However, the activity was not elevated in the G198DhCERK mutant-transfected cells (0.12 Ϯ 0.04 pmol/min/ 10 6 cells). The stably transfected cells revealed a marked in- crease in ingested red cells compared with the Fc␥RIIA control (Fig. 1A). Stably transfected cells containing hCERK displayed a 2.5-fold increase in their phagocytic index compared with control cells and a 1.5-fold increase compared with vector control cells at 30 min following challenge with EIgG (Fig. 1B). There was no increase in phagocytosis in Fc␥RIIA/ G198DhCERK-transfected cells beyond that observed for the other controls. The percentage of COS-1 cells ingesting at least one EIgG increased from the control value of 43 Ϯ 11 in the Fc␥RIIA cells to 70 Ϯ 9 (p Ͻ 0.001; n ϭ 6) in the hCERKtransfected cells. The percent of control vector-transfected cells ingesting EIgG was 50 Ϯ 11, and the percent of Fc␥RIIA/ G198DhCERK mutant was 45 Ϯ 9 (Fig. 1C).
Ceramide kinase activity was measured to ascertain whether hCERK transfectants would exhibit enhanced activity during phagocytosis. Ceramide kinase activity increased to maximal levels by 7 min following challenge with EIgG ( Fig.  2A). Ceramide kinase activity increased in the Fc␥RIIA control cells from a resting value of 0.39 Ϯ 0.1 to 0.86 Ϯ 0.1 pmol/min/ 10 6 cells. Ceramide kinase increased from 0.29 Ϯ 0.1 to 0.82 Ϯ 0.4 in the vector control cells. However, in hCERK-transfected cells, the CERK activity increased from a resting value of 0.7 Ϯ 0.2 to 1.46 Ϯ 0.24 pmol/min/10 6 cells (p Ͻ 0.05) during the phagocytic challenge. Ceramide kinase activity (pmol/min/10 6 cells) of the catalytically inactive mutant G198D was 0.12 Ϯ 0.04 at rest, 0.16 Ϯ 0.1 at 7 min, and 0.097 Ϯ 0.07 at 30 min with a phagocytic challenge. These differences were not statistically significant. The CERK activity rapidly fell by 10 min to levels approaching the basal-inactivated state in both the Fc␥RIIA and Fc␥RIIA/hCERK-transfected cells, but the activity rose significantly by 30 min in the Fc␥RIIA/hCERK-transfected cells (Fig. 2A).
Previous work has demonstrated the existence of phagocytosis-dependent ceramide formation in Fc␥RIIA-transfected COS-1 cells (11). Ceramide mass was measured to determine whether the changes in ceramide kinase activity might second- arily lower ceramide levels in hCERK transfectants. There were no significant differences in ceramide levels between vector and CERK-transfected cells (Fig. 2B). Because ceramide levels were no lower in the hCERK transfectants, these data support the view that changes in C1P and not ceramide accounted for differences in phagocytic response. To ascertain whether C1P was also formed during phagocytosis, hCERK COS-1 cells were labeled with the ceramide precursor D-erythro-[ 3 H]sphingosine. The lipids were then extracted, and the product profile was evaluated in several thin layer chromatography solvent systems by comparison to known standards. High performance thin layer plates were activated with 2% sodium borate for 1 h at 100°C. Using a solvent system consisting of chloroform/acetone/methanol/acetic acid/ water (40/16/12/8/4, v/v), a labeled product comigrating with C1P (R f 0.50) was detected. The product was resistant to alkaline hydrolysis and was clearly separable from sphingosine (R f 0.56), phosphatidic acid (R f 0.80), and lysophosphatidic acid (R f 0.27).
C1P radioactivity increased by 3-fold in the hCERK-transfected cells compared with the vector control and non-transfected Fc␥RIIA cells (Fig. 3, A and B). Labeling stably transfected COS-1 cells with D-erythro-[ 3 H]sphingosine and challenging them with EIgG for 30 min resulted in a further increase in the C1P formed. The phagocytic challenge increased C1P levels by 2-fold in the hCERK-transfected cells compared with Fc␥RIIA or Fc␥RIIA/vector control cells (Fig.  3B). D-Erythro-[ 3 H]sphingosine-labeled hCERK-transfected COS-1 cells were also challenged with EIgG for different peri-ods of time, and C1P formation was evaluated. C1P increased significantly at 7 min simultaneously with the increase of ceramide kinase activity and remained elevated to a 30-min time point (Fig. 4). Assuming that the cell sphingolipids were labeled to equilibrium by 24 h, the change in C1P activity would represent an increase in C1P from 11.52 Ϯ 0.9 to 20.67 Ϯ 6 pmol/10 6 cells at 7 min and to 21.1 Ϯ 8 pmol/10 6 cells at 30 min. There was a significant difference for C1P levels at the 7-and 30-min time points.
Lipid rafts are believed to be small membrane microdomains containing cholesterol and sphingolipids (21). Due to the high degree of saturated fatty acyl content of these sphingolipids, lipids rafts are thought to form a distinct gel-like ordered lipid phase within cell membranes. These domains have been reported to contain a variety of signaling molecules.
Therefore, it was next confirmed that lipid rafts could be identified and isolated from COS-1 cells. Lipid rafts were subsequently identified by a caveolin-1 marker (Fig. 5A). Caveolin-1 was expressed in the lipid rafts but only weakly in nonmembrane raft fractions obtained from Fc␥RIIA, vector control, and hCERK-transfected cells. Utilizing an antibody to c-Myc, c-Myc-tagged hCERK was identified in the appropriate raft fraction generated from Myc-tagged hCERK vector but not from cells generated with the control vector. An anti-hCERK antibody was employed to evaluate the subcellular distribution of hCERK. Immunoblots yielded bands consistent with the known molecular weight of hCERK (Fig. 5C). Using this antibody, the intracellular localization of CERK during activation was evaluated. COS-1 cells transfected with hCERK were stimulated with EIgG for different periods of times and then subjected to fractionation. Immunoblots revealed that, under resting conditions, CERK was distributed within the cytosol and only weakly in the raft fraction. With stimulation, CERK was translocated to the raft fraction and diminished in the cytosolic fraction (Fig. 5C). The ceramide kinase activity was higher in the cytosolic fraction at the zero time point (11.23 Ϯ 0.05 pmol/min/mg protein). As time progressed, the activity in the raft fraction increased from 4.95 Ϯ 0.1 to 8.76 Ϯ 0.07 pmol/ min/mg protein at 7 min and to 14.86 Ϯ 0.2 pmol/min/mg protein at 30 min following stimulation (Fig. 5D). Correspondingly, ceramide kinase activity was enriched in the caveolar fractions generated from non-stimulated transfected cells. The activity in hCERK-transfected cells was 4.9 Ϯ 0.1 pmol/min/mg protein compared with 1.0 Ϯ 0.08 pmol/min/mg protein for Fc␥RIIA controls and 1 Ϯ 0.04 pmol/min/mg protein from caveolae obtained from vector control-transfected cells.
To establish the critical role of the lipid rafts in phagocytosis, the lipid rafts were disrupted with 5 mM m␤CD (17). After the metabolic labeling, the cells with D-erythro-[ 3 H]sphingosine, COS-1 cells were treated with m␤CD for 30 min at 37°C followed by a challenge with EIgG. The incubation buffer was removed, the cells were collected, and the lipids were extracted. C1P diminished in the cells and was present in medium. Under these conditions, phagocytosis was decreased by 50% (Fig. 6).
Having demonstrated that COS-1 cells possess lipid raft markers, a fluorescent probe was employed to determine whether these domains represent a distinct lipid environment. COS-1 cells were labeled with 2-dimethylamino-6-lauroylnaphthalene (Laurdan), a compound that is known to be embedded in the hydrophobic region of a phospholipid bilayer and to partition between crystalline and gel phases. Laurdan changes its steady-state emission spectrum when embedded in a fluid compared with gel phase membrane. Using high-sensitivity microspectrophotometry, emission spectra were collected from COS-1 cells undergoing phagocytosis both at the site of EIgG ingestion and at lateral cell body surfaces of the COS-1 cells away from the site of phagocytosis. Although no dramatic changes were noted at the sites of phagocytosis generated from COS-1 cells bearing Fc␥RIIA, Fc␥RIIA/vector or mutant G198DhCERK, the emission spectrum generated from Fc␥RIIA/hCERK transfectants exhibited a dramatic shift in emission spectra of ϳ37 nm (Ͻ0.001) at the site of phagocytosis (Fig. 7). However, no change in emission was noted at a lateral site distant from the point of contact with EIgG. These studies provide direct evidence in living COS-1 cells for the presence of L o phase lipid domains in sites engaged in phagocytosis that spatially correspond to sites where lipid rafts are present.
When Laurdan spectra were collected for 1 s with a 20-s interval between each collection, we observed that the spectra did not change at a region of the membrane distinct from the site of phagocytosis. In contrast, when spectra were obtained from the vicinity of phagocytosis, the Laurdan spectra shifted at the site of phagocytosis in the cells transfected with the Fc␥RIIA/hCERK (Fig. 8). As noted in both Figs. 7 and 8, the spectra changed only slightly at the site of phagocytosis in Fc␥RIIA control cells and in the vector control or the Fc␥RIIA/ G198DhCERK mutant. DISCUSSION An emerging theme in the control of cellular signaling and metabolic regulation is the importance of the spatial localization of the molecules participating in cell activation. Recent advances in the study of cell membrane structure have led to the concept that microdomains exist within the fluid bilayer of the plasma membrane. One type of microdomain is the lipid raft, which is enriched with tightly packed sphingolipids and cholesterol (22). These rafts present in both excitable and nonexcitable cells localize a number of proteins, including multiple signal transduction molecules, while excluding others (23). Different types of rafts are likely to exist based on the specific marker proteins and ultrastructure (24). Caveolae represent a well studied subpopulation of lipid rafts having an invaginated morphology and containing caveolin that interacts directly with several other proteins (23,25,26). Caveolin forms a scaffold onto which many classes of signaling molecules can assemble to generate signaling complexes. Signaling components that are highly enriched in caveolae include low molecular weight and heterotrimeric G proteins, src family kinases, mitogenactivated protein kinase, protein kinase C, and the p85 subunit of phosphatidylinositol 3-kinase (27,28). The Fc␥RIIA receptor also localizes to caveolae upon cross-linking (29). Ceramide generation from sphingomyelin may occur primarily in lipid rafts (30) and may facilitate the clustering of Fc␥RIIA and its association with rafts (31). These observations suggest that caveolae may provide the structure for assembling critical components of the signaling pathways that mediate the early phagocytic response.
The recent cloning of hCERK has made it possible to evaluate the physiological functions of ceramide kinase and of C1P. Because polymorphonuclear leukocytes are not readily amenable to transfection, an in vitro cell culture system was chosen to study the mechanisms of C1P-induced phagocytosis. We took advantage of the ability of COS-1 cells to express a phagocytic phenotype when transfected with the Fc␥RIIA. In the present study, COS-1 cells stably transfected with Fc␥RIIA were also stably transfected with hCERK. Using this model, two important observations were made. First, the increased expression of hCERK resulted in heightened Fc␥RIIA-mediated phagocytosis. In concert with the increase in phagocytosis, we observed that, in both Fc␥RIIA-and Fc␥RIIA/hCERK-transfected cells, ceramide kinase activity and C1P generation increased.
Second, in labeling cells with Laurdan, we observed that Fc␥RIIA/hCERK transfectants exhibited a dramatic shift in emission during phagocytosis, corresponding to a distinct change in lipid-ordered structure. The change in Laurdan emission provides strong evidence for the existence of lipid raft like L o domain in COS-1 cells at the site of phagosome formation. L o phase lipids have been observed in reconstituted lipid rafts using Laurdan (32). Their demonstration in living cells has been limited to neutrophils. In this study, optical microspectrophotometry of Laurdan-labeled polarized neutrophils revealed the presence of large lipid raft-like domains at the lamellipodium (33).
Phospholipid composition is known to play a significant role in membrane fusogenicity. The fusion of apposed membranes requires destabilizing of the membranes to render them susceptible to fusion. This can be the result of the Ca 2ϩ -induced phase separation of rigid crystalline domains of acidic phospholipids (e.g. phosphatidic acid) within mixed lipid membrane (34). Fusion can be initiated between closely opposed membranes at the boundaries between such crystalline and the surrounding non-crystalline domains. Such boundaries represent structurally unstable points and thus offer focal points for the mixing of molecules from opposed membranes. A calciumdependent ceramide kinase was reported previously in synaptic vesicles, which led to the hypothesis that C1P may attenuate membrane charge, regulate vesicle transport, or mediate membrane fusion (1). This possibility was evaluated in neutrophils in a previous study in which we reported that C1P is formed during phagocytosis and that C1P promotes the fusion of complex liposomes (7). Recently, Igarashi and co-workers (9) found that Ca 2ϩ -dependent C1P formation was involved in the degranulation pathway of mast cells. The degranulation process includes the fusion of the plasma membrane with secretory granules (9). Their findings can potentially be explained by the alteration in membrane fusogenicity elicited by an increase of C1P. In combination with the present report, these data suggest that C1P may promote phagosome formation by enhancing the rapidity by which pseudopods engage and merge with each other in lipid rafts. FIG. 8. Kinetic study of lipid order membrane domains. Each spectrum was collected for 1 s with a 20-s period between each collection. A-C, spectra were collected from a region of the membrane distant from the site of phagocytosis denoted by the dark rectangle. D-F, spectra obtained from the vicinity of phagocytosis as denoted by the dark rectangles. Panels A and B show cells before and after the collection of the spectra in panel C, respectively. Similarly, panels D and E show cells before and after collection of the spectra shown in panel F. As time passes, the Laurdan spectra far from the site of phagocytosis do not change (panel C). However, the emission spectra at the site of phagocytosis change over time (panel F).
Cellular fractionation of COS-1 cell transfectants with hCERK also demonstrated that ceramide kinase activity is localized in cytosol and transferred to lipid rafts during stimulation with EIgG. Using an antibody raised against CERK, the cytosolic distribution of CERK was reported in mast cells (9). However, in another study, CERK was found in the membrane fraction in human embryonic kidney 293 cells (6). The observation of the translocation of CERK during activation in the present study may provide and explanation for these discrepant observations (30,31).
Lipid rafts are also sites of signal transduction. Cross-linking of receptors may increase their affinity for rafts. Partitioning of receptors into rafts results in a new microenvironment where their phosphorylation state can be modified by local kinases and phosphatases modulating downstream signaling. The coalescence of several rafts may result in an amplification of a transduced signal.
One approach for studying the function of lipid rafts involves depleting cells of cholesterol. Lipid rafts are held together to a great extent via interactions between cholesterol and sphingolipids. COS-1 transfectants were treated with methyl-␤-cyclodextrin, and phagocytosis decreased significantly. COS-1 transfectants were labeled with D-erythro-[ 3 H]sphingosine and then treated with methyl-␤-cyclodextrin. Treatment with methyl-␤-cyclodextrin presumably disturbed the lipid-ordered structure of COS-1 transfectants and reduced phagocytosis.
Lipid and protein interactions play an important role in keeping the plasma membrane intact and functionally active. Many membrane-bound enzymes require a special conformation in order to display their optimal activity. There are a large number of lipids present with the potential to form rafts in the plasma membrane including C1P (35). Because of the presence of C1P, a liquid-order domain with higher rigidity than the surrounding plasma membrane was identified in Fc␥RIIA/ hCERK-transfected cells. This study supports the view that a regulated increase in C1P during phagocytosis may destabilize the membrane and lead to fusion.