Reconstitution of ATP- and Cytosol-dependent Transport of de Novo Synthesized Ceramide to the Site of Sphingomyelin Synthesis in Semi-intact Cells*

Transport of ceramide synthesized at the endoplasmic reticulum to the Golgi compartment, where sphingomyelin (SM) synthase exists, was reconstituted within semi-intact Chinese hamster ovary cells. When [3H]ceramide that had been produced from [3H]sphingosine at 15 °C in perforated cells was chased at 37 °C, [3H]ceramide-to-[3H]SM conversion occurred in a cytosol-dependent manner. In various aspects (i.e. kinetics, ATP dependence, and temperature dependence), [3H]ceramide-to-[3H]SM conversion in perforated cells was consistent with that in intact cells. The cytosol from LY-A strain, a Chinese hamster ovary cell mutant defective in endoplasmic reticulum-to-Golgi transport of ceramide, did not support [3H]ceramide-to-[3H]SM conversion in perforated wild-type cells, whereas the wild-type cytosol rescued the conversion in perforated LY-A cells. Brefeldin A-treated cells, in which the endoplasmic reticulum and the Golgi apparatus were merged, no longer required cytosol for conversion of [3H]ceramide to [3H]SM. These results indicated that the assay of [3H]ceramide-to-[3H]SM conversion in semi-intact cells is a faithful in vitro assay for the activity of cytosol-dependent transport of ceramide and that LY-A cells are defective in a cytosolic factor involved in ceramide transport. In addition, conversion of [3H]ceramide to [3H]glucosylceramide in semi-intact cells was little dependent on cytosol, suggesting that ceramide reached the site of glucosylceramide synthesis by a cytosol-independent (or less dependent) pathway.

Transport and sorting of lipids from cellular sites of their synthesis to their appropriate destinations are essential events for membrane biogenesis in cells. Various pathways for intracellular transport of newly synthesized lipids have been suggested. Phosphatidylserine produced in the endoplasmic reticulum (ER) 1 is sorted to the mitochondria, where phosphatidylserine is converted to phosphatidylethanolamine, which is translocated to the cytoplasmic layer of the plasma membrane lipid bilayer (1). Cholesterol synthesized in the ER is likely delivered to the plasma membrane in an ATP-dependent manner by a non-Golgi pathway (2,3). Ceramide (Cer), a common precursor for both sphingomyelin (SM) and glycosphingolipids, is synthesized at the cytosolic surface of the ER, translocated from the ER to the Golgi apparatus, and then converted to SM by the enzyme phosphatidylcholine:ceramide cholinephosphotransferase (SM synthase) in the lumenal side of the Golgi apparatus or to glucosylceramide (GlcCer) on the cytosolic surface of the Golgi apparatus (4 -6). After translocation into the Golgi lumen, GlcCer is further converted to lactosylceramide and more complex glycosphingolipids. SM and glycosphingolipids produced in the Golgi lumen are predominantly delivered to the ectoplasmic layer of the plasma membrane lipid bilayer. However, the mechanisms underlying transport and sorting of lipids are unknown.
In vitro analysis of lipid transport between different intracellular organelles by using isolated organelles or semi-intact cells permeable to macromolecules would be a powerful approach to determine whether cytosolic macromolecules are involved in this function. Although several attempts to reconstitute ER-to-Golgi apparatus transport of Cer in cell-free and semi-intact cell systems have been made, the findings from these studies with in vitro systems have varied. Some investigators reached the conclusion that Cer transport does not require ATP or cytosol (7,8), whereas others argued that fusion of ER-derived transport vesicles enriched in Cer with the Golgi membrane requires cytosol (9). The contradictory results most likely reflect the fact that the nature of intracellular transport of long chain Cer for de novo SM synthesis in intact cells was little known at that time, so that it was difficult to test how accurately the in vitro systems mimicked Cer trafficking events occurring in intact cells.
Recently, we have isolated several types of Chinese hamster ovary (CHO) cell mutants resistant to lysenin, an SM-directed cytolysin (10), and found that one strain, designated LY-A, is defective in transport of Cer from the ER to the Golgi compartment for SM synthesis (11). In vivo analysis of the mutant and wild-type CHO cells has revealed that the main pathway for Cer trafficking from the ER to the site of SM synthesis is the ATP-dependent pathway, which is impaired in LY-A cells, and suggested that Cer reached GlcCer synthase by an ATP-independent (or less ATP-dependent) pathway in CHO cells (11). These findings, together with the observation that LY-A and wild-type cells were almost identical in the rate of processing of glycoproteins to endoglycosidase H-resistant forms (11), have suggested the existence of specific machinery involved in Cer transport. Recent advances in the study of intracellular Cer transport now allow us to achieve a faithful in vitro assay of Cer transport from its synthesis site to the site of SM synthesis after carefully evaluating the accuracy of the in vitro system.
In the present study, we devised an in vitro reconstitution system of Cer trafficking from the ER to the site of SM synthesis using perforated CHO cells. In various aspects (i.e. kinetics, temperature dependence, ATP dependence, and the phenotypic difference between wild-type and LY-A strains), the conversion of pulse-labeled Cer to SM in the in vitro system is consistent with Cer-to-SM conversion in intact cells. Analysis with this reconstitution system demonstrates that the ATP-dependent trafficking of Cer requires cytosol and that the phenotype of LY-A cells is due to the recessive deficiency of a heat-labile macromolecule of cytosol.
Cells and Cell Culture-The CHO-K1 cell line was obtained from the American Type Culture Collection (ATCC CCL 61). Strain LY-A, a CHO-K1-derived mutant cell line, was previously established by us (10). Cells were routinely maintained in Ham's F-12 medium (Life Technologies, Inc.) supplemented with 10% newborn bovine serum, penicillin G (100 units/ml), and streptomycin sulfate (100 g/ml) at 33°C in a 5% CO 2 atmosphere.
Preparation of Cytosol Fraction-CHO cells were cultured in ES medium (Nissui Co., Tokyo, Japan) supplemented with 5% fetal bovine serum, 2 mM L-glutamine, 0.1% sodium hydrogen carbonate, and 10 mM Hepes-NaOH, pH 7.4, at 37°C in spinner bottles to a density of ϳ4 ϫ 10 5 cells/ml. Hereafter, all manipulations were carried out at 4°C or on ice. Cells were precipitated by centrifugation (250 ϫ g for 10 min), suspended in 10 mM Tris-HCl, pH 7.4, containing 0.25 M sucrose at a ratio of 1 volume of the cell pellet to 4 volumes of the buffer, and homogenized with a stainless steel ball bearing homogenizer as described previously (12). The cell homogenate was centrifuged (900 ϫ g for 10 min) to precipitate nuclei. The postnuclear supernatant was centrifuged (10 5 ϫ g for 1 h). The supernatant fluid was centrifuged (10 5 ϫ g for 1 h) once more to remove the particulate fraction completely. The resultant supernatant fluid as a cytosol fraction was rapidly frozen in liquid nitrogen in aliquots and stored at -80°C until use. The protein concentration of the cytosol fraction was ϳ8 mg/ml.
Preparation of Semi-intact Cells-Semi-intact CHO cells were prepared as reported previously (12), with several modifications. Cells harvested by trypsinization were seeded in 10 ml of F-12 medium containing 10% newborn bovine serum at a density of 3.3 ϫ 10 6 cells per 10-cm dish (Corning) and cultured at 37°C overnight to reach subconfluence. Hereafter, all manipulations were carried out at 4°C or on ice. The subconfluent cell monolayer was washed twice with a hypotonic buffer (10 mM Hepes-KOH, pH 7.2, 15 mM KCl, 0.1 mM MgCl 2 ) and incubated for 10 min in 5 ml of the same buffer. Then, the medium was replaced with 5 ml of H/KCl buffer (25 mM Hepes-KOH, pH 7.2, 115 mM KCl), and the cell monolayer was scraped with a rubber policeman and collected by low speed centrifugation (250 ϫ g for 5 min). The pellet as semi-intact cells was washed with 5 ml of H/KCl buffer and resuspended in 100 l of H/KCl buffer. Before using the semi-intact cells for the in vitro assay of Cer transport, the protein concentration of the suspension was determined as described below. The protein concentration of the semi-intact suspension was 2.5-3.0 mg/ml. Trypan blue permeability was optically determined under a phase contrast microscope after the mixing of the suspension of the semi-intact cells with 0.3% of solution of trypan blue in phosphate-buffered saline. Note that the presence of MgCl 2 in the hypotonic buffer was important for efficient perforation of CHO-K1 and LY-A cells, and that 10 -20% of cells otherwise remained unperforated after scraping, although the original formula of hypotonic buffer for perforation of CHO-15B cells does not contain MgCl 2 (12). For preparation of BFA-treated semi-intact cells, the subconfluent CHO cell monolayer was washed twice with 5 ml of serum-free F-12 medium and incubated in 5 ml of Nutridoma medium (serum-free F-12 medium supplemented with 1% Nutridoma-SP (Roche Molecular Biochemicals) and 10 mM Hepes-NaOH, pH 7.4) containing 1 g/ml BFA at 37°C for 30 min. Then, the semi-intact cells were prepared as described above. Trypan blue permeability analysis showed that more than 95% of the BFA-treated cells were perforated by this method.
In To start each transport reaction, 25 l (40 g of protein) of the suspension of the fumonisin B 1 -treated semi-intact cells prelabeled was added to 65 l of a 1.4-fold concentrated transport reaction medium, and the mixture was incubated at 37°C for 30 min. The reaction was stopped by addition of 700 l of chloroform/methanol (1:2, by volume). Then, 230 l of chloroform and 420 l of 0.1 M KCl were added to the sample for phase separation to extract lipids (13). Lipids extracted were separated on high performance TLC plates with a solvent system of chloroform/ methanol/H 2 O (65:25:4, by volume). Radioactive lipids separated on the plates were detected with a BAS1800 image analyzer (Fuji Film Inc., Tokyo, Japan), and after gels were collected from the plates by scraping, the radioactivity of each lipid was determined by liquid scintillation counting using a toluene scintillation mixture. To obtain quantitative data of perforated cell-derived activity, we routinely carried out background control experiments, in which semi-intact cells were pulselabeled with [ 3 H]sphingosine in the absence of palmitoyl CoA and chased. The radioactivity incorporated to Cer, SM, and GlcCer in the control experiments, which was regarded as background activity derived from unperforated cells existing in the semi-intact cell preparation, was subtracted from the radioactivity of each lipid produced in the standard counterparts for correction of the background.
Metabolic Labeling of Sphingolipids in Intact Cells with [ 3 H]Sphingosine-For preparation of a stock solution containing 160 M [ 3 H]sphingosine (330 Ci/mmol) and 320 M BSA, 32 l of D-erythro-[3-3 H]sphingosine (20 Ci/mmol, 500 Ci/ml) and 14.4 l of 1 mg/ml nonradioactive D-erythro-sphingosine in ethanol were put in a 1.5-ml tube (Eppendorf), dried under a stream of nitrogen gas, and dispersed in 300 l of phosphate-buffered saline containing 21 mg/ml fatty acidfree BSA by mixing and brief sonication. Cells were seeded in 5 ml of F-12 medium containing 10% newborn bovine serum at a density of 1.2 ϫ 10 6 cells per 6-cm dish and cultured in the normal culture medium overnight at 37°C. After two washes with 2 ml of serum-free F-12 medium, the cell monolayer was incubated in 1.5 ml of Nutridoma medium containing 1.6 M [ 3 H]sphingosine (330 Ci/mmol) complexed with BSA for 30 min at 15°C. The pulse-labeled cells were washed twice with 2 ml of serum-free F-12 medium and incubated in 1.5 ml of Nutridoma medium in the presence of 20 M fumonisin B 1 for 15 min on ice. Then, the cells were incubated for 30 min at various temperatures for chase. After chase, the cells were washed twice with 2 ml of cold phosphate-buffered saline, lysed with 1 ml of cold 0.1% SDS, and 800 and 20 l each of the lysates were used for lipid extraction (13) and for determination of protein concentration, respectively. For metabolic labeling of BFA-treated cells, CHO cells were washed twice with 2 ml of serum-free F-12 medium and incubated in 1.5 ml of Nutridoma medium containing 1 g/ml BFA at 37°C for 30 min. Then, after addition of [ 3 H]sphingosine to the BFA-containing medium, cells were incubated for 30 min at 15°C for metabolic labeling. After separation of extracted lipids by high performance TLC, the radioactivity of each lipid was determined as described above, and normalized to cell protein.
Enzyme Assays-Enzyme assays for SM and GlcCer synthases with C 6 -NBD-Cer as the enzyme substrate were performed as described previously (14) with minor modifications. Briefly, enzyme sources were incubated in H/KCl buffer containing 10 M C 6 -NBD-Cer and 0.5 mM UDP-glucose for 10 min at 37°C. When the dependence of the activities on ATP and cytosol was examined, transport reaction mixture was used instead of H/KCl buffer (see the legend to Table I). After stopping the reactions by addition of chloroform/methanol (1:2, by volume), lipids were extracted and separated on TLC plates with a solvent system (chloroform/methanol/H 2 O, 65:25:4, by volume). SM and GlcCer metabolites of C 6 -NBD-Cer were scraped from the plates and extracted in chloroform/methanol/H 2 O (1:2:0.8, by volume). NBD fluorescence of the samples was measured with a spectrofluorometer (excitation at 470 nm; emission at 530 nm). Sphingosine-N-acyltransferase activity was determined by modification of a previous method (15). In brief, enzyme sources were incubated in 0.4 ml of H/KCl buffer containing 360 nM D-erythro-[3- 3 H]sphingosine (20 Ci/mmol), 50 M palmitoyl CoA, and 0.5 mM dithiothreitol at 37°C for 30 min. Lipids were extracted from the sample and separated by TLC with a solvent of chloroform/methanol/ acetic acid (94:5:5, by volume). Radioactive lipids on the TLC plate were visualized by image analysis with a BAS1800 image analyzer. After collecting gels from the plates by scraping, the radioactivity of the Cer produced was determined by scintillation counting. Lactate dehydrogenase activity was measured with an assay kit (MTX LDH kit, Kyokuto Seiyaku, Tokyo, Japan).
Transmission Electron Microscopy-Semi-intact cells were prepared as described above, and intact control cells were harvested by trypsinization. These samples were fixed in 2.5% glutaraldehyde and 2% paraformaldehyde in 0.1 M cacodylate buffer, pH 7.2, washed in the same buffer, and postfixed with 2% osmium tetroxide in 0.1 M cacodylate buffer, pH 7.2. The specimens were processed as described previously (16), and ultrathin sections of the specimens were observed with a Hitachi H-600 transmission electron microscope.
Determination of Protein Concentration-Protein concentrations were determined with the Pierce BCA protein assay reagent kit, using BSA as the standard.

Semi-intact CHO Cells Permeable to Macromolecules-
The plasma membrane of CHO cells was perforated by the scraping of cell monolayers after a transient exposure to a hypotonic buffer containing 0.1 mM MgCl 2 , a modified version of the method of Beckers et al. (12). By this protocol, more than 95% of both LY-A and wild-type CHO cells lost the ability to exclude trypan blue dye. Enzyme assays of precipitated and supernatant fractions after low speed centrifugation of scraped cells showed that more than 85% of lactate dehydrogenase activity was released from the cells, but that more than 80% of both sphingosine-N-acyltransferase and SM synthase activities were recovered within the perforated cells. The enzyme activities of sphingosine-N-acyltransferase and SM synthase in LY-A cells are identical to the levels in wild-type cells (11).
Transmission electron microscopic analysis indicated that the electron-dense cytosol observed in intact cells was not present in perforated cells but that various intracellular organelles including the ER, Golgi apparatus, mitochondria, and nucleus were maintained in the perforated cells, although the ER structure found in the perforated cells was swollen (data not shown), consistent with a previous study (12). Considering these observations, we refer to the perforated cells, which retain ER-and Golgi apparatus-bound enzymes but not cytosolic soluble proteins, as semi-intact cells henceforth.

Reconstitution of ATP-dependent Transport of Long Chain Cer from Its Synthesis Site to the Site of SM Synthesis in
Semi-intact CHO Cells-To determine whether transport of Cer from the ER to the site of SM synthesis was able to be reconstituted in semi-intact CHO cells, we first developed pulse-labeling conditions, where radioactive Cer was formed in semi-intact cells. De novo sphingolipid biosynthesis is initiated by condensation of serine and palmitoyl-CoA to generate ketodihydrosphingosine, which is converted to dihydrosphingosine (for a review of sphingolipid biosynthesis, see Ref. 17). Acyl transfer from acyl-CoA to the amino group of dihydrosphingosine produces dihydroceramide, which is desaturated to generate ceramide. These reactions are catalyzed by ER-embedded enzymes, the catalytic sites of which face the cytosol (18 -20). Because the dihydrosphingosine-N-acyltransferase enzyme catalyzes acylation of sphingosine as well as dihydrosphingosine (21), production of Cer from sphingosine at the ER occurs under sphingosine-supplied conditions. In addition, conversion of metabolically labeled Cer to SM in intact cells can be blocked at low temperatures (22). Thus, radioactive Cer was specifically produced in semi-intact cells of both wildtype and LY-A strains by incubation with 10 M palmitoyl CoA and 1.6 M D-erythro-[ 3 H]sphingosine at 15°C for 30 min, whereas the pulse reaction produced only a very small amount of [ 3 H]SM (Fig. 1). Note that neither ATP cytosol nor UDPglucose was added to the pulse reaction mixture. Thus, no [ 3 H]GlcCer was produced by the pulse reaction ( Fig. 1), although [ 3 H]Cer-to-[ 3 H]GlcCer conversion in semi-intact cells did not require ATP or cytosol (see below). Omission of palmitoyl CoA from the pulse reaction mixture reduced [ 3 H]Cer production by ϳ85%, indicating that [ 3 H]Cer was produced largely in perforated cells but not unperforated cells under the standard pulse conditions that we used (Fig. 1).
We next developed chase conditions in which pulse-labeled [ 3 H]Cer was converted to [ 3 H]SM in semi-intact cells at a rate similar to that observed in intact cells. The prelabeled semiintact cells (40 g of protein) were incubated with the cytosol fraction (100 g of protein) and an ATP-regenerating system at 37°C for 30 min. Fumonisin B 1 , an inhibitor of sphingosine-N-  (Fig. 2B), which is consistent with the phenotype of intact LY-A cells (11 (Fig. 2, A and B).
Depletion of ATP in the chase reaction mixture by using apyrase reduced the conversion of [ 3 (Fig. 2D). Assay for SM and GlcCer synthases with the enzyme substrate C 6 -NBD-Cer, which moves spontaneously between and across membranes (5), indicated that neither enzyme activities of semi-intact cells were ATP-dependent (Table I).

H]Cer to [ 3 H]SM in wildtype cells to the level in LY-A cells in vitro, whereas the ATP depletion did not affect conversion of [ 3 H]Cer to [ 3 H]GlcCer
Temperature  (Fig. 3B). The absence of efficient Cer-to-SM conversion at 15°C was not due to an inactive SM synthase at this temperature, because BFAtreated cells, in which the ER and the Golgi apparatus were merged (24), produced 14-fold more [ 3 H]SM without any increase in the [ 3 H]Cer level, compared with the BFA-untreated control (Fig. 3). Therefore, the inability of cells to convert Cer to SM at 15°C indicated that transport of Cer from its synthesis site to the site of SM synthesis occurred little, if at all, in CHO cells at 15°C. BFA treatment enhanced conversion of [ 3 H]Cer to [ 3 H]GlcCer at 15°C in intact cells to ϳ2-fold of the BFAuntreated control level (Fig. 3B). The weak effect of BFA on enhancement of [ 3 H]GlcCer production compared with [ 3 H]SM at 15°C will be discussed below.
When intact wild-type CHO cells prelabeled at 15°C were chased in the presence of fumonisin B 1 at various temperatures for 30 min, more than 15% of the pulse-labeled [ 3 H]Cer was converted to [ 3 H]SM at Ͼ25°C, whereas less than 5% was converted at Ͻ15°C (Fig. 4A). A similar pattern of temperature dependence of [ 3 H]Cer-to-[ 3 H]SM conversion was observed when semi-intact cells prelabeled at 15°C were chased at various temperatures under the cytosol-and ATP-supplied conditions (Fig. 4B).
Cytosol Is Required for Transport of Long Chain Cer from Its Synthesis Site to the Site of SM Synthesis-Conversion of [ 3 H]sphingosine at 15°C for 30 min, were chased at 37°C for various time periods. Lipids were extracted from semi-intact and intact cells after chase, and the radioactivity of each lipid was determined. WT, wild-type. D, semi-intact cells pulsed with [ 3 H]sphingosine were chased at 37°C for 30 min in the complete transport reaction mixture (ϩ) or an ATP-depleted transport reaction mixture (-), which contained apyrase (11 units/ml) in place of ATP. Semi-intact cells and cytosol were from the indicated strains. After chase, the radioactivity of each sphingolipid indicated was determined. The data (means Ϯ S.D. from three experiments) are expressed as a percentage of the mean values of the wild-type control, in which wild-type semi-intact cells (40 g of protein) were chased in the complete transport reaction mixture containing wild-type cytosol (100 g of protein).

TABLE I Effects of cytosol and ATP on SM and GlcCer synthase activities in
semi-intact cells Semi-intact cells (40 g of protein) of indicated strains were incubated in 90 l of the complete or incomplete reaction mixture for 10 min at 37°C. The amount of SM and GlcCer metabolites of C 6 -NBD-Cer (C 6 -NBD-SM and C 6 -NBD-GlcCer, respectively) that were produced were determined. Values (means Ϯ S.D. from three experiments) are expressed as a percentage of the mean amounts of C 6 -NBD-SM and C 6 -NBD-GlcCer formed in the complete reaction mixture.

Reaction mixture
Wild-type LY-A C 6 -NBD-SM C 6 -NBD-GlcCer C 6 -NBD-SM C 6 -NBD-GlcCer a Percentage of complete reaction mixture control. b Complete reaction mixture consisted of the transport reaction mixture (see under "Experimental Procedures") supplemented with 10 M C 6 -NBD-Cer complexed with 0.1% BSA.
c The cytosol fraction was not added to the reaction mixture. d 11 units/ml apyrase in place of ATP was added to the reaction mixture.

[ 3 H]Cer to [ 3 H]SM in semi-intact wild-type cells required the cytosol fraction prepared from wild-type cells. The level of [ 3 H]SM
produced in the presence of 300 g of protein of the cytosol fraction was ϳ10-fold the cytosol-negative control level, and the cytosol dependence reached a plateau around 100 g of protein/assay (Fig. 5A). In contrast, the activity of Cer-to-SM conversion in semi-intact LY-A cells was little enhanced by the cytosol fraction from LY-A cells, and the activity in semi-intact LY-A cells with 300 g of protein of the LY-A cytosol remained at ϳ20% of the wild-type level (Fig. 5B)

. Conversion of [ 3 H]Cer to [ 3 H]GlcCer in semi-intact wild-type and LY-A cells was less dependent on cytosol, and enhancement of [ 3 H]
GlcCer production by cytosol was maximally ϳ1.5-fold, compared with the control level (Fig. 5, A and B). BFA-treated semi-intact cells did not require cytosol for conversion of Cer to SM (Fig. 5C), suggesting that a rate-determining step in the cytosol-dependent conversion of [ 3 H]Cer to [ 3 H]SM in semi-intact cells was intermembrane transport, but not intramembrane transport, of Cer. The cytosol fraction, which was the supernatant fluid obtained by high speed centrifugation of a postnuclear supernatant frac-tion of cells, had no appreciable activities of SM synthase, GlcCer synthase, or sphingosine-N-acyltransferase. In addition, enzyme assays showed that activities of SM and GlcCer synthases within semi-intact cells were not affected by cytosol ( Table I). The activity of the wild-type cytosol to support Certo-SM conversion in semi-intact cells was inactivated by heat treatment (Table II). This activity was little affected by dialysis at 4°C for 16 h, compared with the control cytosol kept for the same time without dialysis (Table II). Treatment of the wildtype cytosol (100 g) with trypsin (5 g) at 30°C for 30 min resulted in almost complete loss of the cytosol activity. 2 These results revealed that transport of Cer from its synthesis site to the site of SM synthesis requires a heat-labile cytosol protein(s) and suggested that Cer reached the site of GlcCer synthesis by a cytosol-independent (or less dependent) pathway.
When semi-intact cells were pulse-labeled with D-erythro-

. Effects of BFA on conversion of [ 3 H]Cer to [ 3 H]SM and [ 3 H]GlcCer in semi-intact and intact cells. A,
semi-intact wild-type cells, prepared from cells pretreated with (ϩ) or without (-) 1 g/ml BFA for 30 min, were pulsed with [ 3 H]sphingosine at 15°C for 30 min in the presence (ϩ) or absence (-) of 1 g/ml BFA. SI, semi-intact. B, intact wild-type cells were pretreated with or without 1 g/ml BFA for 30 min, after which they were pulsed with [ 3 H]sphingosine at 15°C for 30 min. After pulse, the radioactivity of each sphingolipid was determined. The data are shown as means Ϯ S.D. from three experiments.

FIG. 4. Temperature dependence of [ 3 H]Cer metabolism in intact and semi-intact cells.
A, intact wild-type cells, which had been prelabeled with [ 3 H]sphingosine, were chased for 30 min at various temperatures. B, semi-intact wild-type cells pulsed with [ 3 H]sphingosine were chased for 30 min at various temperatures in the transport reaction mixtures containing wild-type cytosol. Lipids were extracted from chased semiintact and intact cells, and the radioactivity of each lipid was determined. cluded saturated dihydro forms in addition to desaturated forms as assessed by TLC analysis of sphingoid bases liberated by acid hydrolysis. 2

LY-A Cells Are Deficient in a Cytosolic Factor Involved in Cer
Trafficking for SM Synthesis-To address the question of whether the impairment of Cer-to-SM conversion in LY-A cells was due to deficiency of a cytosolic soluble factor, we carried out cytosol exchange experiments using the in vitro system. When prelabeled semi-intact cells of the wild-type strain were chased in the presence of the cytosol fraction of LY-A cells instead of the wild-type cytosol, the activity of Cer-to-SM conversion was ϳ20% of the complete wild-type control level (Fig. 6). More interestingly, when semi-intact LY-A cells were chased with the wild-type cytosol, the activity was ϳ100% of the control level (Fig. 6). Exchange of the cytosol fraction did not affect conversion of Cer to GlcCer (Fig. 6). These results demonstrated that LY-A cells are deficient in a cytosolic factor involved in transport of Cer for SM synthesis.
The Deficiency of the LY-A Cytosol Is Recessive-We next examined whether the deficiency of the LY-A cytosol was recessive or dominant to the wild-type cytosol. For this, we car-ried out cytosol mixing experiments in vitro. Semi-intact cells with both the wild-type and LY-A cytosol fractions (50 g each of protein) were almost identical in the activity of Cer-to-SM conversion to semi-intact cells with 50 g of protein of the wild-type cytosol alone (Fig. 7). Cer-to-GlcCer conversion was not affected by mixing the two cytosol types (Fig. 7). When the wild-type cytosol that had been incubated at 95°C for 3 min was mixed with the LY-A cytosol, the mixed cytosol did not show any activity to support Cer-to-SM conversion to semiintact cells (Table II). Dialysis of the wild-type cytosol did not affect the activity to complement the LY-A cytosol, compared with the nondialyzed control level (Table II). These results indicated that the phenotype of LY-A cells defective in Certo-SM conversion is due to the recessive defect in a heat-labile macromolecule of cytosol. This conclusion was supported by in vivo metabolic labeling experiments using hybrid cells; the conversion rate of Cer to SM in wild-type ϫ LY-A hybrids was an intermediate value between the levels in wild-type ϫ wildtype hybrids and LY-A ϫ LY-A hybrids. 2

DISCUSSION
To address the question of whether cytosolic macromolecules were involved in intracellular transport of Cer, we here attempted to reconstitute transport of Cer from its synthesis site to the site of SM synthesis within semi-intact CHO cells, which retain the ER and Golgi apparatus but not cytosolic soluble proteins. In the reconstitution system that we devised,  (Fig. 2, A and C). Third, Cer-to-SM conversion in semi-intact cells is dependent on ATP (Fig. 2D),

FIG. 5. Cytosol dependence of conversion of [ 3 H]Cer to [ 3 H]SM and [ 3 H]GlcCer in the semi-intact cell system. A and B, semi-intact cells pulsed with [ 3 H]sphingosine
were chased at 37°C for 30 min in the transport reaction mixture containing various amounts of the cytosol fraction. After chase, the radioactivity of each sphingolipid was determined. Both semi-intact (SI) cells and cytosol fractions were derived from wild-type (WT) CHO cells (A) and LY-A cells (B). C, semi-intact wild-type cells, prepared after pretreatment with 1 g/ml BFA, were pulsed with [ 3 H]sphingosine and chased at 37°C for 30 min in the complete transport reaction mixture in the presence of 1 g/ml BFA. After chase, the radioactivity of each sphingolipid was determined. In the experiments without cytosol (-), no cytosolic fraction was added to the reaction mixture. For no chase, lipids were extracted from the transport reaction mixture containing the pulse-labeled semi-intact cells without chase. The bars each show the mean of duplicate experiments, with actual activities for individual experiments indicated by points. Open bars, no chase; gray bars, chase in the presence of cytosol; hatched bars, chase in the absence of cytosol. being consistent with our previous observation that ATP-depletion in intact CHO-K1 cells by energy inhibitors inhibits ERto-Golgi apparatus trafficking of Cer (11). Fourth, a rate-determining step for production of [ 3 H]SM from [ 3 H]Cer in vitro as well as in vivo is Cer transport but not enzyme reaction of SM synthase, because efficient conversion of Cer to SM occurs in BFA-treated intact and semi-intact cells even under low temperature and ATP-depleted conditions, unlike in untreated controls (Fig. 3A). Furthermore, the semi-intact cell system reproduces the phenotype of mutant LY-A cells defective in Cer trafficking from the ER to the site of SM synthesis; the rate of Cer-to-SM conversion in semi-intact LY-A cells is only ϳ20% of the wild-type level, whereas the rate of Cer-to-GlcCer conversion in semi-intact LY-A cells is near the wild-type level (Fig.  2B). From these observations, we conclude that the in vitro system reconstitutes intracellular Cer transport events occurring in intact CHO cells for de novo SM synthesis.
Analysis with the in vitro reconstitution system demonstrated that the ATP-dependent conversion of Cer to SM requires cytosol (Fig. 5A). In addition, cytosol exchange and cytosol mixing experiments demonstrated that the phenotype of LY-A cells defective in Cer-to-SM conversion results from the recessive deficiency of the LY-A cytosol (Figs. 6 and 7). The activity of the wild-type cytosol to rescue the LY-A cytosol in terms of Cer-to-SM conversion in semi-intact cells is heat-labile and is not permeable by a dialysis membrane (Table II). SM synthase activity itself, which was assessed with C 6 -NBD-Cer as the enzyme substrate, was not dependent on ATP or cytosol (Table I). Collectively, these results indicate that ATP-dependent transport of Cer from the ER to the site of SM synthesis in CHO cells requires cytosol and that LY-A cells are defective in a heat-labile cytosol macromolecule involved in the Cer transport.
BFA treatment rendered Cer-to-SM conversion in semi-intact cells independent of cytosol (Fig. 5C). Our preferred interpretation of this observation is that cytosol is required for intermembrane transport of Cer from the ER to the Golgi compartment for SM synthesis. However, because catalytic sites of SM synthase and GlcCer synthase have been suggested to exist in the lumenal and cytoplasmic sides, respectively, of the Golgi complex (25)(26)(27)(28)(29), there is the alternative interpretation that cytosol is required for transport of Cer across the Golgi membrane but that cytosol-independent transbilayer movement of Cer occurs in the merged organelle formed by BFA treatment. Unfortunately, the effects of BFA treatment on transbilayer movement of natural Cer are unknown. Thus, at this time, the latter interpretation is conceivable, and it is also possible that both transbilayer and intermembrane transport of Cer for de novo SM synthesis require cytosol.
The precise sites of SM synthase and GlcCer synthase in Golgi subcompartments are unknown. If GlcCer is synthesized in more proximal Golgi subcompartments compared with SM, the possibility exists that transport of Cer from the ER to the proximal Golgi subcompartment, where GlcCer but not SM is synthesized, does not require ATP or cytosol but that intra-Golgi apparatus transport of Cer from the proximal Golgi subcompartment to the subcompartment for SM synthesis requires ATP and cytosol. We cannot currently exclude this possibility. Wattenberg (30) previously developed a cell-free assay system of intra-Golgi apparatus transport of glycosphingolipid by using donor and acceptor Golgi membranes prepared types. For the experiments without cytosol (-), no cytosolic fraction was added to the reaction mixture. After chase, the radioactivity of each sphingolipid was determined. The data (means Ϯ S.D. from three experiments) are expressed as a percentage of the mean values of the wild-type control, in which wild-type semi-intact cells (40 g of protein) were chased in the transport reaction mixture containing wild-type cytosol (100 g of protein).

FIG. 7. Effects of mixing of wildtype and LY-A cytosol on conversion of [ 3 H]Cer to [ 3 H]SM and [ 3 H]GlcCer
in the semi-intact cell system. Semiintact wild-type and LY-A cells pulsed with [ 3 H]sphingosine were chased at 37°C for 30 min in the transport reaction mixture containing the wild-type (WT) cytosol (50 g of protein) and/or LY-A cytosol (50 g of protein) fractions as indicated. After chase, the radioactivity of each sphingolipid was determined. The data (means Ϯ S.D. from three experiments) are expressed as a percentage of the mean values of the wild-type control, in which wild-type semi-intact cells (40 g of protein) were chased in the transport reaction mixture containing wildtype cytosol (50 g of protein). from two types of CHO cell mutants defective in sugar metabolism. From analysis with this system, he has shown that ATP and cytosol are required for transport of glycosphingolipid from the Golgi subcompartment for galactosylation to the subcompartment for sialylation in G M3 ganglioside synthesis (30). LY-A cells are normal in de novo synthesis of G M3 and also in processing of glycoproteins to endoglycosidase H-resistant forms (11). Therefore, the cytosolic factor that is impaired in LY-A cells appears to be specifically responsible for Cer trafficking, even if LY-A cells are defective in intra-Golgi apparatus, but not ER-to-Golgi apparatus, transport of Cer.
Phosphatidylinositol-transfer protein ␤ isoform (PI-TP␤) is able to transfer not only phosphatidylinositol and phosphatidylcholine but also SM between membranes in cell-free systems (31,32). Interestingly, when PI-TP␤ is overexpressed in mouse NIH3T3 fibroblast cells, replenishment of SM in the plasma membrane upon degradation by bacterial sphingomyelinase is accelerated, suggesting that PI-TP␤ is involved in membrane or lipid flow between the plasma membrane and the Golgi apparatus (33). Nevertheless, varying the expression levels of PI-TP␤ in cells has no effects on de novo synthesis of SM (33), whereas both de novo synthetic rate and steady-state content of SM are lower in LY-A cells than in wild-type CHO cells (11). Thus, the transport pathway of Cer from the ER to the Golgi apparatus for de novo SM synthesis appears to be distinct from the PI-TP␤-dependent pathway.
One might expect that if both SM and GlcCer synthases use the same pool of Cer, dramatic elevation of GlcCer synthesis should occur when SM synthesis is blocked. However, conversion of [ 3 H]Cer to [ 3 H]GlcCer in LY-A cells is similar to the wild-type level in both intact and semi-intact cell systems (this study and see also Ref. 11). One possible explanation is that during chase, [ 3 H]Cer that has a final destination of the site of SM synthesis becomes incapable of being redistributed to the site of GlcCer synthesis. For example, Cer used for SM synthesis might be first translocated into the lumen of the ER or to a post-ER compartment devoid of GlcCer synthase.
In contrast to Cer-to-SM, Cer-to-GlcCer conversion in vitro was little dependent on ATP or cytosol (Figs. 2D and 5). The ATP independence (or less ATP dependence) is consistent with our previous observation that depletion of intracellular ATP by energy inhibitors affected conversion of Cer to GlcCer in vivo by only a little (11). These in vitro and in vivo observations suggested that the access of de novo synthesized Cer to the site of GlcCer synthesis was largely independent of ATP and cytosol in CHO cells. Some populations of GlcCer synthase might exist transiently or permanently in the ER, and so intermembrane transport of Cer may not be required for production of GlcCer, because a previous study has suggested that GlcCer synthase was not strictly localized in the Golgi apparatus but was more widely distributed among the microsome (28). Alternatively, transport of Cer from the ER to the Golgi compartment for GlcCer synthesis might not require ATP or cytosol. Protein transport from the ER to the ER-Golgi intermediate compartment is not blocked at 15°C, but this process requires cytosol (for reviews, see Refs. 34 and 35). Under assay conditions for Cer transport in semi-intact cells, pulse reaction was carried out at 15°C without any supply of cytosol. Therefore, the lack of any obvious requirement of cytosol for Cer-to-GlcCer conversion in pulse and chase experiments within semi-intact cells suggests that if GlcCer synthesis occurs at the ER-Golgi intermediate compartment, the mechanism for Cer transport from the ER to this compartment for GlcCer synthesis differs from the protein transport mechanism.
Kok et al. (8) previously concluded that conversion of metabolically labeled Cer (with [ 14 C]serine) to SM as well as to GlcCer did not require ATP or cytosol by analysis using streptolysin O-permeabilized HT29 Gϩ cells (8). A possible explanation for the discrepancy between their conclusion and ours is that the streptolysin O-permeabilized cell system reconstitutes the ATP-or cytosol-independent pathway but not the ATP-and cytosol-dependent pathway of Cer transport. Another possibility is that the HT29 Gϩ strain, a permanent cell line derived from human colonic carcinoma, naturally lacks the ATP-and cytosol-dependent pathway of Cer transport, although we previously showed that conversion of metabolically labeled Cer (with [ 3 H]sphingosine and [ 3 H]dihydrosphingosine) to its SM metabolites was inhibited by depletion of intracellular ATP in HeLa cells and normal human skin fibroblasts as well as in CHO cells (11).
The in vitro reconstitution system described in this study has revealed that cytosol is required for intracellular transport of long chain Cer and that the phenotype of LY-A cells defective in Cer transport results from the recessive defect in a cytosolic factor. Thus, the in vitro reconstitution system will also be a useful assay system for purification of the cytosolic factor that rescues the deficiency of LY-A cells.