Stimulation of Cellular Sphingomyelin Import by the Chemokine Connective Tissue-activating Peptide III*

The selective import of phospholipids into cells could be mediated by proteins secreted from the cells into the extracellular compartment. We observed that the supernatants obtained from suspensions of thrombin-acti-vated platelets stimulated the exchange of pyrene (py)-labeled sphingomyelin between lipid vesicles in vitro . The proteins with sphingomyelin transfer activity were purified and identified as the chemokine connective tis-sue-activating peptide III (CTAP-III) and platelet basic protein. Isolated CTAP-III stimulated the exchange of py-sphingomyelin between lipid vesicles but did not affect the translocations of py-labeled phosphatidylcholine and phosphatidylethanolamine. CTAP-III rapidly increased the transfer of py-sphingomyelin from low density lipoproteins into peripheral blood lymphocytes, other immune cells, and fibroblasts. In the Platelets were activated with thrombin (0.5 unit/ml), and the platelet free supernatants were recovered and added to a mixture of donor vesicles (enriched with either py-sphingomyelin (py-SM) or py-phos- phatidylcholine (py-PC)) and acceptor vesicles. After a 30-min incubation, the acceptor vesicles were isolated and their pyrene contents were determined. In other experiments, the transfer of the py-labeled choline phospholipids between lipid vesicles (3.8 m g of egg phosphatidylcholine and 0.08 m g of py-phospholipid) and PBLs was assessed within an 8-min incubation period in the presence of the supernatants obtained from platelet suspensions. Mean values are from four to six experiments.

The central role of phospholipids in intracellular signaling processes is well established. Degradation of membrane phospholipids yields water-soluble (e.g. eicosanoids, inositol trisphosphate) and lipophilic messenger molecules (such as diacylglycerol and ceramide), which initiate several signal transduction chains. Phospholipids are also essential for the biogenesis of caveolae, lateral domains of the cell membrane consisting of sphingolipids, cholesterol, and specific proteins (reviewed in Ref. 1). Specific phospholipids are also required for membrane fusion processes, in particular phosphatidylethanolamine (2). The intramonolayer and intrabilayer distribution of the phospholipids can be rapidly changed in response to a particular physiological situation. During apoptosis, phosphatidylserine is transferred to the outer leaflet of cell membranes, where it serves as a recognition signal for the clearance of apoptotic cells (3,4). In view of their prominent roles in cellular functions, it is evident that the concentration and localization of the individual phospholipids within the membrane bilayer need to be carefully regulated. This is accomplished by enzymes involved in the remodeling and synthesis of the phospholipids (acyltransferases, sphingomyelin synthase, etc.) and by proteins transferring phospholipids between the different intracellular membrane systems (interbilayer transfer). Furthermore, a group of proteins catalyzes the intrabilayer movements of phospholipids between the leaflets of the cellular membrane bilayer (5). Early work on human blood cells suggested that cells may also acquire phospholipids from extracellular sources (6). Among the phospholipid donors, low density lipoproteins (LDL) 1 and other lipoprotein particles are considered to be of particular relevance. The lipoproteins are major phospholipid carriers within the plasma compartment. Their phospholipid contents are only slightly lower than their cholesterol concentrations. By means of the endocytosis of the LDL particles, among the multiple components of the lipoproteins, cells also take up phospholipids.
Recent data point to the existence of a selective, endocytosisindependent phospholipid uptake pathway. Through this import, the phospholipid composition of the cells can be rapidly changed. This was shown to be relevant for the assembly of protein complexes on the platelet surface necessary for the synthesis of thrombin (7). The uptake of phosphatidylinositol may modulate intracellular signal transduction processes (8). The data thus indicated that specific proteins mediate the selective phospholipid import into the cells. Proteins catalyzing the transfer of phospholipids between the plasma lipoproteins (9) are apparently not involved in the cellular phospholipid uptake.
We demonstrate in the present study that stimulated platelets release proteins that specifically exchange sphingomyelin between phospholipid vesicles in vitro. Among these proteins, connective tissue-activating peptide III (CTAP-III), a cytokine belonging to the CXC-subfamily of chemokines, exerted the strongest stimulation of the cellular sphingomyelin uptake. CTAP-III, which is massively secreted by the activated platelets (10,11), represents an N-terminally extended precursor of the chemokine neutrophil-activating peptide 2 (NAP-2). The latter protein is generated from CTAP-III through the cleavage of a N-terminal peptide, which is mediated by proteases associated with monocytes and neutrophil granulocytes (12,13). CTAP-III, which is devoid of the stimulatory functions on neutrophils characteristic for NAP-2 (14), has been reported to act as a weak growth factor for fibroblasts (15). Thus, the physiological role of CTAP-III has not been clearly evident so far. The results of our study identify CTAP-III as a transcellular mediator of the cellular sphingomyelin import.

MATERIALS AND METHODS
Reagents and Antibodies-Platelet basic protein (PBP) and CTAP-III were isolated from released supernatants of thrombin-activated platelets by sequential immunoaffinity chromatography, cation exchange chromatography, and reversed phase chromatography (13,16). NAP-2 was generated from CTAP-III by limited digestion with chymotrypsin and purified by reversed phase chromatography (17). Recombinant CTAP-III and IL-8 bearing a His tag at the N terminus were produced in Escherichia coli according to previously described methods (18), the His tag being removed from IL-8 to yield the 72-amino acid form of the chemokine. All chemokine preparations exceeded 99% purity according to overloaded silver-stained SDS-PAGE and automated N-terminal sequence analysis. The peptide corresponding to the 15 N-terminal amino acids of CTAP-III was synthesized using solid phase methods and Fmoc (N-(9-fluorenyl)methoxycarbonyl) chemistry and purified on a Poros R2/H column (PerSeptive Biosystems, Wiesbaden, Germany). The monoclonal antibody reactive to all variably truncated isoforms of ␤-TG Ag (monoclonal antibody C-24) was induced in mice, cloned, and purified as described (13). 1-Palmitoyl-2-pyrenedecanoyl-sn-3-glycerophosphorylcholine (py-phosphatidylcholine), 1-palmitoyl-2-pyrenedecanoylsn-3-glycerophosphorylethanolamine (py-phosphatidylethanolamine), and (N-pyrenedecanoyl)sphingomyelin (py-sphingomyelin) were from Sigma (Deisenhofen, Germany) and from Molecular Probes (Eugene, OR). [  Cells-Human platelets were isolated as described previously (7). To gain PBL and monocytes, a mononuclear cell suspension was prepared from human blood according to Fogelman et al. (19). Monocytes and PBL were separated from each other by the use of anti-CD14 antibodies conjugated to microbeads and passaged over a positive selection column (Miltenyi Biotech, Bergisch-Gladbach, Germany). Granulocytes were isolated using anti-CD15 antibodies conjugated to microbeads. Skin fibroblasts from a healthy individual (LDLR ϩ/ϩ), a patient with homozygous familial hypercholesterolemia (LDLR Ϫ/Ϫ), and a patient with Niemann-Pick disease A lacking A-SMase activity were kind gifts of Dr. Joachim Thiery (University of Munich) and Dr. Erich Gulbins (University of Tü bingen), respectively. The 70Z/3 pre-B cells with transfected 55-kDa TNF receptor (TNF-R55) have been described previously (20). The Jurkat cell line was kindly provided by Dr. Hartmut Engelmann (University of Munich).
Purification of Sphingomyelin Transfer Proteins-A total of 45 buffy coat preparations (from about 350 ml of human blood each), obtained from the local blood bank (Gesundheitsamt, Stadt Mü nchen), were used for the isolation of platelets. The platelets were stimulated for 10 min at 37°C with 0.5 unit of thrombin/ml in a suspension buffer (138 mM NaCl, 3 mM KCl, 1 mM MgCl 2 , 5 mM glucose, 15 mM Hepes; pH 7.4). The suspensions were centrifuged at 1500 ϫ g for 10 min, and the supernatants were recovered. After concentration in Centriplus 10 tubes (Amicon, Witten), the supernatants were centrifuged for 10 min at 10,000 ϫ g. The supernatants thus obtained were dialyzed against the column buffer (200 mM NaCl, 10 mM imidazol, 2 mM mercaptoethanol, 1 mM EGTA, 0.5 mM phenylmethylsulfonyl fluoride, and 0.02% sodium azide; pH 7.6). The dialysate was applied onto a Sephadex G-75SF column (Amersham Pharmacia Biotech, Freiburg). 3-ml fractions were analyzed for protein content by UV absorption at 280 nm. Aliquots of the fractions were tested for their effects on py-sphingomyelin transfer between donor and acceptor lipid vesicles (see below). The fractions showing the highest sphingomyelin exchange were pooled and dialyzed against an elution buffer (10 mM Tris, 2 mM mercaptoethanol, 1 mM EGTA, 0.5 mM phenylmethylsulfonyl fluoride, 0.02% sodium azide; pH 8.0). The dialysate was loaded onto an anion exchange Resource Q column (Amersham Pharmacia Biotech), equilibrated with the elution buffer, and eluted with a linear gradient of 0 -1.0 M NaCl. Before analysis for the py-sphingomyelin exchange activity, the fractions were dialyzed against a Tris buffer (10 mM Tris, 0.02% sodium azide; pH 7.2). The proteins of the fractions from both columns were analyzed by SDS-PAGE.
In Vitro Phospholipid Exchange-For the preparation of phospholipid donor vesicles, 250 g of egg phosphatidylcholine were dissolved together with 45 g of phosphatidic acid, and 7.5 g of either pysphingomyelin, py-phosphatidylcholine, or py-phosphatidylethanolamine in ethanol. The mixture was dispersed by very slow injection into 300 l of Tris buffer. For the production of acceptor vesicles, 10 mg of egg phosphatidylcholine in ethanol was dispersed in 1 ml of Tris buffer. 50 l of acceptor vesicle solution was mixed with 10 l of donor vesicles and 40 l of Tris buffer. To this suspension, 200 l of either Tris buffer or column fractions was added. After 0 and 30 min, 100 l of the suspensions was loaded onto a small anion exchange column (DEAE-Sepharose C6-LB), and the column was eluted with 1.5 ml of Tris buffer. The eluted acceptor vesicles were solubilized with 2% Triton X-100 at 37°C, and their pyrene monomer contents were analyzed (see below).
Loading of Lipoproteins with Labeled Phospholipids-LDL and high density lipoprotein were labeled with fluorescent and radioactive phospholipids by incubation of fresh human plasma with py-phospholipids, [ 14 C]-and [ 3 H]phospholipids (present in lipid vesicles), and subsequent isolation of lipoprotein fractions by ultracentrifugation as described (21). The specific activities ranged between 1.0 and 9.4 ϫ 10 4 cpm/nmol of phospholipid. Py-labeled phospholipids were present at 10 -25 ng of pyrene/g of LDL protein.
Incubation of Cells with Labeled Lipid Donors-In general, cells were suspended with lipoproteins or vesicles (labeled with 14 C-, 3 H-and py-phospholipids) in the suspension buffer at 37°C. In the case of py-phospholipid-containing donors, the fluorescence was monitored directly in the suspensions every 60 s under on-line conditions. Incorporation of py-labeled phospholipids into the cells was followed by the increase in monomer intensity after addition of the cells to the donors. In all experiments shown under "Results," the increase in monomer intensity was accompanied by a decrease of the excimer to monomer ratio of py-fluorescence intensity. These changes in the py-fluorescence monomer and excimer intensities specifically indicate the selective (endocytosis-independent) phospholipid transfer into the cells. Monomer and excimer fluorescence of the suspensions were determined at emission wavelengths of 380 nm and 480 nm, respectively, with excitation at 340 nm (excitation and emission slits of 5 and 10 nm). Following incubation with radioactively labeled lipoproteins, the cells were separated from the donors by centrifugation and washed once, and the cell associated radioactivity was determined.
Lipid Separations-After incubation with the [ 14 C]sphingomyelinlabeled lipoproteins, the cell suspensions were separated into aqueous and organic phases by the procedure of Bligh and Dyer (22). The upper phase was analyzed for its amount of 14 C reflecting the quantity of [ 14 C]phosphocholine liberated from [ 14 C]sphingomyelin. Concomitantly, the quantity of cell-associated [ 14 C]sphingomyelin was estimated by separating the phospholipids of the lower phase by onedimensional thin layer chromatography using the solvent CHCl 3 / CH 3 OH/NH 3 /H 2 O (45/37/12/8, v/v). In the case of incubation with [ 3 H]sphingomyelin-LDL, following the Bligh and Dyer separation procedure, the lower phase was subjected to one-dimensional thin layer chromatography using the solvent CHCl 3 /CH 3 OH/H 2 O (60/35/8, v/v) to separate the sphingolipids ceramide, sphingosine, and sphingomyelin. The area corresponding to standards of these sphingolipids was scraped off, and the radioactivity was determined in a scintillation counter.
Miscellaneous-Protein concentrations were measured according to the Bradford procedure by means of a kit using ␣-globulin as a standard (Bio-Rad, Munich). The concentrations of ␤-TG Ag and related proteins in platelet releasates were determined according to a previously described sandwich-enzyme-linked immunosorbent assay system (23). The mean values under "Results" are given Ϯ S.D.

RESULTS
Platelet Basic Protein and Related Proteins Stimulate Sphingomyelin Transfer in Vitro-The extracellular media recovered from suspensions of thrombin-activated platelets were shown to contain proteinaceous factors that enhance the transfer of ethanolamine phospholipids between lipid vesicles in vitro (7). In the present study, we analyzed whether the media would influence the exchange of the choline phospholipids phosphatidylcholine and sphingomyelin. Supernatants obtained from suspensions of thrombin-activated platelets were added to a mixture consisting of donor vesicles, containing py-labeled phospholipids, and acceptor vesicles. Extracellular media from suspensions of untreated platelets did not affect the transfer of the py-labeled choline phospholipids between the two types of vesicles (Table I). The supernatants recovered from thrombinactivated platelets stimulated the exchange of py-sphingomyelin by 3.4-fold compared with the one obtained from untreated platelets. The exchange of py-phosphatidylcholine was unaffected. The acceleration of py-sphingomyelin transfer was enhanced after dialysis of the supernatants from activated platelets, suggesting that low molecular weight components partially suppress the sphingomyelin transfer (Table I). Briefly boiling the supernatant and precipitation of its proteins with ammonium sulfate strongly reduced the exchange of py-sphingomyelin (not shown). Transfer of py-sphingomyelin from lipid vesicles to the PBL was also stimulated by the presence of the extracellular media gained from thrombin-activated platelet suspensions (Table I). Again, the transfer of py-phosphatidylcholine was not affected. The results suggested that platelet releasates contained proteins mediating the sphingomyelin exchange.
To isolate and identify these proteins, supernatants from thrombin-stimulated platelet suspensions were fractionated by gel filtration on a Sephadex column. The fractions eluted from the column were tested for their capacity to affect the pysphingomyelin exchange in vitro. The sphingomyelin exchange activity was mostly present in a broad peak between fractions 28 and 31 (Fig. 1A). The fractions were pooled and their proteins separated by gel electrophoresis. Several bands were observed. A prominent band was noted in the 8-kDa range (Fig.  1A, right-hand side). To further purify the proteins catalyzing sphingomyelin exchange, the combined fractions 28 -31 were applied onto an anion exchange column. The proteins were eluted from this latter column with a linear NaCl gradient. The highest sphingomyelin transfer activity was found in fraction 5 ( Fig. 1B). Gel electrophoretic separation of fraction 5 revealed a single band, which was again in the 8-kDa range (Fig. 1B, right-hand side). Amino acid analysis of tryptic peptides obtained after enzymatic digestion of the material contained in fraction 5 identified a peptide fragment that corresponded to amino acid positions 69 -79 of a protein known as platelet basic protein (PBP (11)). PBP (94 amino acids) is a precursor protein of several N-terminally truncated derivatives such as the connective tissue-activating peptide III (CTAP-III; 85 residues) and neutrophil-activating peptide 2 (NAP-2; 70 residues). Together, these proteins are termed ␤-thromboglobulin antigen (␤-TG Ag). We measured the concentration of total ␤-TG Ag in the extracellular media recovered from suspensions of thrombin-activated platelets. By using an enzyme-linked immunosorbent assay method, the concentration was determined to be 4.2 M. The majority of ␤-TG Ag (up to 90%) contained in the platelets consists of CTAP-III, the rest being almost exclusively PBP (24).
Following addition of the purified CTAP-III to the in vitro system of donor and acceptor vesicles, the transfer of py-sphingomyelin was dose dependently enhanced ( Fig. 2A). The con- Platelets were activated with thrombin (0.5 unit/ml), and the platelet free supernatants were recovered and added to a mixture of donor vesicles (enriched with either py-sphingomyelin (py-SM) or py-phosphatidylcholine (py-PC)) and acceptor vesicles. After a 30-min incubation, the acceptor vesicles were isolated and their pyrene contents were determined. In other experiments, the transfer of the py-labeled choline phospholipids between lipid vesicles (3.8 g of egg phosphatidylcholine and 0.08 g of py-phospholipid) and PBLs was assessed within an 8-min incubation period in the presence of the supernatants obtained from platelet suspensions. Mean values are shown from four to six experiments.  1. Isolation of platelet releasate proteins mediating sphingomyelin exchange. A, separation of platelet-secreted proteins by Sephadex chromatography. The supernatants obtained from suspensions of thrombin-stimulated platelets were separated by passage through a Sephadex G-75SF column. The fractions eluted were added to a mixture of donor vesicles (containing py-sphingomyelin) and acceptor vesicles. The py contents of the acceptor vesicles were determined as described under "Materials and Methods." The values given refer to control suspensions without platelet releasates. The separation of the proteins of the pooled fractions 28 -31 by SDS-PAGE (18% gel) is shown on the right. B, further purification of the proteins with sphingomyelin exchange activity. The pooled fractions 28 -31 eluted from the Sephadex column were passaged over an anion exchange Resource Q column. A continuous NaCl gradient was used to elute the proteins. Subsequently, the py-sphingomyelin exchange activity was determined by means of the in vitro system of donor and acceptor vesicles. Fraction 5, showing the strongest stimulation of py-sphingomyelin transfer, was subjected to gel electrophoresis (right). The arrow indicates a single band with an apparent molecular mass of 8 kDa.
centration dependence of the stimulation indicated a steep increase between 0 and 4.4 M. At the latter concentration, at which an apparent saturation was reached, the transfer of py-sphingomyelin was enhanced by 6.5-fold. The exchange of py-labeled phosphatidylcholine and of py-phosphatidylethanolamine were unaffected ( Fig. 2A). Addition of NAP-2 (4.4 M) stimulated the py-sphingomyelin transfer by 2.9-fold (Fig. 2B). Equimolar concentrations of PBP were nearly equally effective as native and recombinant CTAP-III. A synthetic 15-amino acid peptide encompassing the residues that are cleaved off during the degradation of CTAP-III to NAP-2, as well as the structurally unrelated chemokine interleukin 8 (IL-8), did not influence the movements of py-sphingomyelin to the acceptor vesicles (Fig. 2B). Accordingly, CTAP-III selectively stimulated the exchange of sphingomyelin.
Enhancement of Cellular Sphingomyelin Import by CTAP-III-We next evaluated the effect of CTAP-III on the selective cellular uptake of sphingomyelin. LDL particles, the lipoprotein fraction with the highest amount of sphingomyelin in human plasma, were employed as donors for the sphingolipid. Blood cells and fibroblasts, which could be targets for the proteins secreted by the activated platelets under in vivo conditions, were analyzed for their capacity to incorporate sphingomyelin in a CTAP-III-dependent way. The specific transfer of sphingomyelin was assessed by following the changes in the py-fluorescence characteristics under on-line conditions (see "Materials and Methods"). CTAP-III (3.3 M) rapidly stimu-lated the translocation of LDL derived py-sphingomyelin into the PBL and skin fibroblasts (Fig. 3A). After 10 min, 2-3% of the sphingomyelin initially present in the donor vesicles had been transferred to the cells in the absence of the chemokine. With CTAP-III, the amount taken up by the cells was increased by 4-to 5-fold. The concentration dependence of the stimulation of sphingomyelin uptake by CTAP-III showed a sharp increase between 0 and 3.3 M, an apparent saturation being observed at higher concentrations (Fig. 3B). At 3.3 M, CTAP-III did not affect the transfer of py-phosphatidylcholine and py-phosphatidylethanolamine from the LDL particles into the PBL within a 10-min interval (data not shown).
The enhancement of the py-sphingomyelin incorporation by CTAP-III was comparable in LDL receptor-positive (LDLR ϩ/ϩ) and LDL receptor-negative (LDLR Ϫ/Ϫ) fibroblasts, as well as in Jurkat cells (Fig. 3C). In the presence of the anti-␤-TG Ag antibody, the increase in the sphingomyelin import into the LDLR ϩ/ϩ fibroblasts and Jurkat cells as induced by CTAP-III was diminished by 93% and 86%, respectively. The presence of an irrelevant mouse IgG antibody did not affect the CTAP-III-mediated sphingomyelin uptake into the PBL (not shown). CTAP-III barely augmented the py-sphingomyelin incorporation into the platelets (Fig. 3C). Equimolar concentrations of PBP and CTAP-III elicited a comparable increase of the py-sphingomyelin transfer from the LDL particles to the PBL (Table II). The stimulation by CTAP-III was completely prevented by the anti-␤-TG Ag antibody. NAP-2 enhanced the py-sphingomyelin uptake by 4-fold, whereas IL-8 was ineffective (Table II). IL-8 differed from the ␤-TG Ag proteins in the electrostatic properties of the C-terminal, positively charged, and neutral amino acids prevailing in the ␤-TG Ag proteins, whereas more residues with negative charges were present in IL-8. The basic, water-soluble lysozyme did not promote the uptake of py-sphingomyelin (Table II). This excludes the notion that the presence of the positive charges per se was sufficient for the phospholipid transfer function of CTAP-III (and related peptides). When lipid vesicles were employed as phospholipid donors instead of the lipoproteins, the transfer of py-sphingomyelin into the cells was accelerated by 7-fold (Table II). Thus, CTAP-III promoted the cellular uptake of py-sphingomyelin.
Basic amino acids, in particular lysine, may play a role for the interaction of CTAP-III with plasma membrane glycosaminoglycans, thereby physically approaching CTAP-III to the cells. The CTAP-III-mediated translocation of py-sphingomyelin from LDL to the PBL was reduced by 94% in the presence of heparin (Fig. 4A). Interactions of CTAP-III with cell surface glycosaminoglycans might thus be implicated in the phospholipid transfer activated by the chemokine. The temperature dependence for the selective uptake process mediated by CTAP-III was determined. PBL were equilibrated at four different temperatures between 4°C and 37°C. Arrhenius plots were generated from the temperature dependence of LDL derived py-sphingomyelin transfer in the presence of CTAP-III (Fig. 4B). The activation energies calculated after linearization of the curve amounted to 7.4 kcal/mol. This indicates that water was excluded from the CTAP-III promoted transfer of sphingomyelin (see "Discussion").
CTAP-III-delivered Sphingomyelin as Substrate for Cellular Sphingolipid Production-Cells were incubated in the presence of CTAP-III with LDL containing either [ 14 C]sphingomyelin or [ 14 C]phosphatidylcholine. After the end of the incubation in the presence of [ 14 C]sphingomyelin-LDL, the cellassociated 14 C of Jurkat cells and of LDLR ϩ/ϩ fibroblasts was 2-to 3-fold higher in the CTAP-III-treated samples (Table III). The increase of 14 C was completely abolished when the anti-TG Ag antibody was present during the preincubation period (not shown). In cells that had been incubated with LDL supplemented with [ 14 C]phosphatidylcholine, the cell-associated radioactivity was unchanged by CTAP-III (Table III). In further experiments, the LDL particles were double labeled with [ 3 H]phosphatidylcholine plus [ 14 C]sphingomyelin, and the lipoproteins were incubated with the Jurkat cells. After the end of the incubation, the cellular amounts of 3 H and 14 C were determined. Although the quantity of cell-associated 3 H was unaffected by the presence of CTAP-III, the cellular amount of 14 C was augmented. Accordingly, the chemokine specifically increased the transfer of labeled sphingomyelin into the cells, the CTAP-III-stimulated import being independent of the endocytosis of the particles.
To analyze whether the sphingomyelin delivered to the cells   under the control of CTAP-III was available for the production of sphingolipid messenger molecules, the formation of the sphingomyelin hydrolysis product phosphocholine was measured (Fig. 5A) (Fig. 5B). Preincubation with N-oleoylethanolamine, which inhibits the deacylation of ceramide by ceramidases, tended to increase the generation of [ 3 H]ceramide in cells treated with CTAP-III plus TNF␣. The amount of [ 3 H]sphingosine was lowered by 72% (Fig. 5B). This suggested that indeed ceramide degradation had been inhibited. Thus, the sphingomyelin delivered to the cells by CTAP-III was rap-idly degraded to sphingolipid messengers after cytokine activation.
In a reconstituted system we tested whether coactivation of the platelets and PBL enhanced the sphingolipid messenger generation by stimulation of the sphingomyelin transfer. Control experiments verified that the supernatants recovered from thrombin-activated platelets increased the transfer of pysphingomyelin from the LDL particles (1 g/ml) to the PBL (1 ϫ 10 6 ; by 4.7-Ϯ 1.3-fold compared with the addition of extracellular media from untreated platelets (n ϭ 3)). To exclude that the LDL-associated [ 3 H]sphingomyelin was hydrolyzed by extracellular secretory SMases (27), supernatants recovered from suspensions of the activated platelets, as well as from the TNF␣-stimulated PBL, were added to the suspensions of [ 3 H]sphingomyelin-LDL. After the end of a 10-min incubation period, no 3 H was detected in the ceramide and sphingosine fractions of the lipid phases indicating that there was no extracellular hydrolysis of the [ 3 H]sphingomyelin. The generation of the [ 3 H]sphingolipids was enhanced by 3.0-(ceramide) and 2.4-fold (sphingosine) due to the presence of the activated platelets (Fig. 6C). The formation of the degradation products was prevented when the anti-␤-TG Ag antibody was added. Thus, platelet-secreted ␤-TG Ag proteins, predominantly CTAP-III, stimulated sphingomyelin import into the PBL, which was subsequently degraded in response to cytokine activation.

DISCUSSION
Sphingomyelin represents a key precursor for intracellular messenger molecules implicated in the signaling chains mediating the cellular response to different forms of stress, as well as inducing proliferation, cellular differentiation, and apoptosis (28,29). Sphingomyelin is also an important constituent of the lateral domains of the plasma membrane known as caveolae (1) and lipid rafts (30). To fulfill these functions, the intrabilayer localization and concentration of sphingomyelin is subject to extensive regulation. The selective import of lipoproteinderived phospholipids represents a means to rapidly modify the cellular phospholipid composition. We observed that the releasates of thrombin-activated platelets contained proteins mediating the selective sphingomyelin exchange between lipid vesicles. Following purification of the active proteins, they were identified as the chemokines CTAP-III and its precursor PBP, members of the ␤-TG Ag family. Under in vivo conditions, CTAP-III is quantitatively the major ␤-TG Ag protein secreted by the activated platelets. The uptake of py-sphingomyelin into PBL, fibroblasts, and Jurkat cells was strongly increased by physiologically relevant concentrations of CTAP-III. The exchange of py-labeled glycerophospholipids (phosphatidylcholine, phosphatidylethanolamine) remained unchanged by the presence of CTAP-III. Thus, CTAP-III specifically stimulated the sphingomyelin transfer. It represents the first water-solu-ble protein known to mediate the selective cellular import of sphingomyelin.
Well characterized phospholipid transfer proteins with a high affinity toward specific phospholipids are those promoting the intracellular exchange of phosphatidylcholine (31) and of phosphatidylinositol (32). Most other phospholipid-translocating proteins such as, for example, the plasma phospholipid transfer protein (33) catalyze the movements of a considerable variety of structurally different (phospho)lipids. The amino acid sequences of the known phospholipid transfer proteins do not exhibit substantial homologies (34). A data base analysis, including the phospholipid binding proteins mentioned above, did not yield any considerable homologies with the CTAP-III sequence (35). Previous data on the structure of the ␤-TG Ag proteins might allow some predictions regarding the putative phospholipid binding site within the CTAP-III molecules.
The N-terminally extended forms of the ␤-TG Ag proteins (CTAP-III, PBP) accelerated the sphingomyelin transfer more efficiently than did the proteolytically truncated derivative NAP-2. A synthetic peptide encompassing the N-terminal amino acids cleaved from CTAP-III to yield NAP-2, when present alone, did not affect cellular sphingomyelin uptake. Accordingly, this stretch of sequence in CTAP-III is necessary for the optimal stimulation of sphingomyelin exchange, but does not itself mediate the transfer. The N-terminal amino acids of CTAP-III were previously shown to be involved in stabilizing the association of CTAP-III into homo-oligomers (predominantly dimers) (36). In line with this view, dimer formation in NAP-2 is less favored than in CTAP-III (37,38). Dimerization of CXC chemokines is accompanied by the formation of a shallow hydrophobic groove in these molecules (39,40), which could accommodate the sphingomyelin molecule during the transfer reaction. The role of this hydrophobic compartment as potential sphingomyelin binding site will be tested in further studies.
LDL are the particles with the highest sphingomyelin content among the plasma lipoproteins and might, therefore, provide substantial proportions of the sphingomyelin delivered to the cells by means of CTAP-III in vivo. Our findings exclude that the enhancement of sphingomyelin import by CTAP-III is mediated by the endocytosis of the lipoproteins. In the case of stimulation of phospholipid uptake via endocytosis, one would expect that the transfer of all phospholipid fractions present in the lipoproteins should be enhanced by the chemokine. However, this was not the case. The chemokine increased the cellular uptake of py-sphingomyelin and acted as transfer protein for the sphingolipid under in vitro conditions but did not enhance the transfer of other py-phospholipids. Furthermore, after incubation of Jurkat cells with LDL particles double labeled with [ 3 H]phosphatidylcholine plus [ 14 C]sphingomyelin, CTAP-III selectively increased the amount of cell-associated 14 C. The cellular uptake of LDL-derived sphingomyelin was also independent of the interaction of the lipoprotein with the classic LDL receptor, indicating that the effect of CTAP-III is unrelated to the LDL receptor pathway. The results reveal a previously unrecognized function of LDL: its role as a selective donor for sphingomyelin.
The sphingomyelin translocation across the aqueous medium as catalyzed by the water-soluble CTAP-III might occur either exclusively within a hydrophobic environment or at least partially proceed through the water phase. The temperature dependence of the sphingomyelin transfer promoted by CTAP-III yielded a value for the activation energy of 7 kcal/mol. This value is considerably lower than the activation energy for sphingomyelin diffusion across an aqueous medium, which is 21-25 kcal/mol (41). Within the transfer process, the desorp-tion of the sphingomyelin molecule from the donor particle represents the step requiring most of the energy. The low activation energy measured suggests that the desorption of the sphingomyelin molecule is greatly facilitated by CTAP-III, with the translocation pathway proceeding mostly in a hydrophobic environment. In principle, the stimulation of the sphingomyelin uptake may be facilitated by the physical interactions of the ␤-TG Ag proteins with the plasma membrane of the cellular acceptors. No protein receptors for PBP and CTAP-III have so far been characterized. Platelet factor-4, a chemokine presenting sequence homology with the ␤-TG Ag proteins, is known to interact with cell membrane heparan sulfate and chondroitin sulfate proteoglycans (42,43). CTAP-III (and PBP) are also known to bind to plasma membrane proteoglycans. Addition of heparin inhibited the insertion of sphingomyelin into the cell membrane. Thus, the binding of CTAP-III to the plasma membrane proteoglycans is apparently required for the efficient insertion of the sphingolipid into the cell membrane. Structural variations in the glycosaminoglycan chains might be responsible for the differential efficiency of the cellular acceptors to incorporate sphingomyelin in a CTAP-III-dependent way.
Subsequent to the CTAP-III-mediated import, the newly delivered sphingomyelin was found to be hydrolyzed to ceramide and sphingosine as a consequence of TNF␣-induced cellular activation. The sphingomyelin degradation elicited by the cytokine could in principle be catalyzed by acid and neutral sphingomyelinases (A-SMase and N-SMase), because both enzymes are principally stimulated by the interaction of TNF␣ with the TNF-R55 receptor (44,45). Our data show that the sphingomyelin transferred to the cells by CTAP-III is exclusively hydrolyzed by TNF␣-stimulated A-SMase. The direct processing of the CTAP-III-delivered sphingomyelin by the activated cells indicates that the generation of the sphingolipid messengers can be modulated by the cellular sphingomyelin import. Stimulation of the SMases induces substantial degradation of the cellular sphingomyelin (46 -48). The reduction in the plasma membrane sphingomyelin content itself, as well as the depletion of cholesterol resulting therefrom (46,49), are expected to cause pronounced alterations in the structural organization and functional properties of the cell membrane. Thereby, the biogenesis of sphingolipid-rich membrane microenvironments (caveolae, lipid rafts) might be severely impaired. The fast CTAP-III-mediated supply of sphingomyelin to the cells might contribute to rapidly restore the integrity of the sphingolipid-containing membrane domains under these conditions. This could be of particular importance under conditions where platelets, immune cells, and fibroblasts are coactivated such as during inflammation and wound healing.
In conclusion, our data reveal a novel function for the chemokines CTAP-III (and PBP) as mediators of the selective cellular uptake of sphingomyelin. The sphingomyelin delivered to the cells by platelet-secreted CTAP-III rapidly mixes with an intracellular pool. Following activation of the acceptor cells with TNF␣, the newly supplied sphingomyelin is rapidly hydrolyzed by acid sphingomyelinase. By secreting CTAP-III, activated platelets are thus able to modulate the sphingolipid signaling of their neighboring cells.