Modulation of amyloid precursor protein cleavage by cellular sphingolipids.

Lipid rafts and their component, cholesterol, modulate the processing of beta-amyloid precursor protein (APP). However, the role of sphingolipids, another major component of lipid rafts, in APP processing remains undetermined. Here we report the effect of sphingolipid deficiency on APP processing in Chinese hamster ovary cells treated with a specific inhibitor of serine palmitoyltransferase, which catalyzes the first step of sphingolipid biosynthesis, and in a mutant LY-B strain defective in the LCB1 subunit of serine palmitoyltransferase. We found that in sphingolipid-deficient cells, the secretion of soluble APPalpha (sAPPalpha) and the generation of C-terminal fragment cleaved at alpha-site dramatically increased, whereas beta-cleavage activity remained unchanged, and the epsilon-cleavage activity decreased without alteration of the total APP level. The secretion of amyloid beta-protein 42 increased in sphingolipid-deficient cells, whereas that of amyloid beta-protein 40 did not. All of these alterations were restored in sphingolipid-deficient cells by adding exogenous sphingosine and in LY-B cells by transfection with cLCB1. Sphingolipid deficiency increased MAPK/ERK activity and a specific inhibitor of MAPK kinase, PD98059, restored sAPPalpha level, indicating that sphingolipid deficiency enhances sAPPalpha secretion via activation of MAPK/ERK pathway. These results suggest that not only the cellular level of cholesterol but also that of sphingolipids may be involved in the pathological process of Alzheimer's disease by modulating APP cleavage.

The amyloid ␤-peptide (A␤) 1 is the principle constituent of senile plaques found in Alzheimer's disease (AD) brains and is generated by proteolysis of an integral membrane protein, the amyloid precursor protein (APP). APP is metabolized via at least two post-translational pathways, one of which is a nonamyloidogenic pathway mediated by ␣-secretase proteolytically producing soluble APP (sAPP␣), the dominant processing product; this cleavage generates the residual 10-kDa CTF (CTF␣). Previous studies have shown that the activation of signal transduction pathways including protein kinase C (PKC) (1-3), mitogen-associated protein kinase (MAPK) (4), and growth factors (5) alter the relative amounts of sAPP␣ and A␤ production. The other cleavage is mediated by ␤-secretase, generating several proteolytic C-terminal fragments (CTFs), namely, CTF␤ and CTF⑀, and ␥-secretase, producing either a 40-residue protein (A␤40) or a 42-residue protein (A␤42) from CTF␤. The cleavages at residues 40 -42 are referred to as ␥-cleavage, and the cleavage at residues 49 -52 are referred to as ⑀-cleavage (6).
Recent studies revealed that the prevalence of AD is linked to the serum cholesterol level and that sAPP␣ secretion and A␤ generation are modulated by the cellular cholesterol level (7)(8)(9)(10)(11)(12)(13). It is also suggested that APP processing and A␤ generation are associated with membrane microdomains, known as lipid rafts, that are rich in cholesterol and sphingolipids and are also the principal compartment in which A␤ is found (13)(14)(15)(16)(17)(18).
The cholesterol content in lipid rafts has been shown to contribute to the integrity of the raft structure and the functions of the rafts in signaling and membrane trafficking (19 -21). In addition, several studies showed that sphingolipids modulate raft functions; the reduction in the cellular sphingolipid level renders glycosyl phosphatidylinositol (GPI)-anchored proteins more sensitive to phosphatidylinositol-specific phospholipase C (22) and also enhances the solubility of GPI-anchored proteins in Triton X-100 (23). The blockade of ceramide synthesis was shown to inhibit folate uptake via GPI-anchored receptors (24) and to enhance the conversion of the prion protein to its scrapie form (25)(26)(27). These studies indicate that the cellular levels of cholesterol and sphingolipids modulate the functions of lipid rafts. Therefore, the evidence that the cholesterol level in lipid rafts can modulate APP processing reasonably raises the question of whether cellular sphingolipids also modulate APP processing and A␤ generation. However, the participation of sphingolipids in APP processing remains undetermined.
To address this issue, we examined the alterations in APP processing and A␤ generation in sphingolipid-deficient cells using ISP-1 (myriocin), a potent inhibitor of serine palmitoyltransferase (SPT) (28), and a CHO-K1-cell-derived mutant cell line, the LY-B strain, which has a defect in the LCB1 subunit of SPT and is therefore incapable of de novo synthesis of any sphingolipid species (29). Our findings indicate that not only the cellular cholesterol level but also the sphingolipid level modulates APP processing.

MATERIALS AND METHODS
Antibodies-The monoclonal antibody 22C11, which recognizes amino acids 66 -81 of the N terminus of APP, was purchased from Chemicon International (Temecura, CA). The monoclonal antibodies used were BA27, which is specific for the A␤1-40 terminal site; BC05, which is specific for the A␤42 terminal site; and BNT77, which was raised against A␤11-28 but recognizes A␤11-16; all of these antibodies have been characterized previously (30). The monoclonal antibody 6E10 (raised against A␤1-17) was purchased from Senetek PLC (Maryland, MO). The rabbit polyclonal antibody, UT-18 (raised against APP695-(676 -695)) was used to detect cellular APP and its C-terminal fragments (31). The rabbit polyclonal antibody, G530, which was raised against rat A␤1-16, was used to detect rodent sAPP␣ (32). The rabbit polyclonal antibodies that recognize phospho-independent PKC␣, PKC␦, PKC⑀, and PKC␥ were purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA).
Cell Culture-The CHO-K1 cell-derived mutant cell line, the LY-B strain, has been previously established (29). LY-B/cLCB1, a corrected revertant of the LY-B strain, was previously obtained by the stable transfection of LY-B cells with the cDNA encoding the hamster LCB1 subunit of SPT (29). Ham's F-12 medium supplemented with 10% fetal bovine serum was used as the normal culture medium. CHO-K1 cells stably expressing APP751 (APP-CHO-K1) were used for determining APP processing and A␤ generation. To deplete sphingolipids, APP-CHO-K1 cells were treated with 1 M myriosin (ISP-1; purchased from BIOMOL Research Laboratories). The Nutridoma-BO medium (Ham's F-12 medium containing 1% Nutridoma-SP (Roche Applied Science), 0.1% fetal bovine serum (FBS), and 10 M sodium oleate-bovine serum albumin complex) was used as the sphingolipid-deficient medium. For cultivation in sphingolipid-deficient medium, the cells were seeded, incubated in the normal culture medium at 37°C for 1 day, and, after washing twice with serumfree Ham's F-12 medium, were cultured in the Nutridoma-BO medium for 2 days. In the experiment on the pharmacological inhibition of SPT in APP-CHO-K1 cells, the Nutridoma-BO medium was supplemented with 1 M ISP-1, and the levels of sphingolipids were recovered with concurrent treatment with 1 M D-erythro-sphingosine (Matreya, Inc., Pleasant Gap, PA) as described previously (28).
ELISA-Two-site ELISA for A␤40 and A␤42 was carried out as previously described (30,33). BNT77 was coated as the capture antibody, whereas BA27 (for A␤40) and BC05 (for A␤42) were used as the detection antibodies following conjugation with horseradish peroxidase.
Protein Preparation-Cultured cells grown in 10-cm 2 dishes were washed twice with ice-cold phosphate-buffered saline and then collected by scraper. The cells were then centrifuged at 1,000 ϫ g for 10 min, and the cell pellet was homogenized in Tris saline (50 mM Tris-HCl, pH 7.4, 150 mM NaCl), containing 1% Triton X-100 and protease inhibitors (Complete), followed by homogenization using a motor-driven Teflon homogenizer. The homogenates were then centrifuged at 200,000 ϫ g for 20 min at 4°C in a TLX ultracentrifuge (Beckman). The supernatants were collected for biochemical analyses. Protein concentrations were determined using the bicinchoninic acid protein assay kit (Pierce). Aliquots of the supernatant samples containing equal amounts of protein were subjected to 7.5% or 4 -20% SDS-PAGE for immunoblot analysis as described previously (34).
Lipid Analysis-The metabolic labeling of lipids with [ 14 C]serine in APP-CHO-K1 cells in the presence or absence of 1 M of ISP-1 was performed as described previously (35). The rate of lipid labeling was corrected for each protein concentration.
Immunoblot Analysis-The proteins separated using SDS-PAGE were electrophoretically transferred onto a polyvinylidene difluoride membrane (Millipore, Bedford, MA). Nonspecific binding was blocked with 5% fat-free milk in phosphate-buffered saline containing 0.1% Tween 20. The blots were then incubated with primary antibodies overnight at 4°C. For the detection of both primary monoclonal and polyclonal antibodies, appropriate peroxidase-conjugated secondary antibodies were used in conjunction with SuperSignal Chemiluminesence (Pierce) to obtain images that were saved on film. The primary antibodies used were as follows: monoclonal antibodies; 22C11, at a final concentration of 5 g/ml; 6E10, at a final concentration of 5 g/ml; polyclonal antibodies, G530, which recognizes rodent sAPP␣, diluted at 1:1,000; and UT-18, which recognizes the C terminus of APP, diluted at 1: 500. The membrane fractions were prepared and subjected to immunoblot analysis using anti-PKC␣, PKC␦, PKC⑀, and PKC␥ antibodies diluted at 1:1000.
Preparation of Membrane Fractions-APP C-terminal fragment generation was performed in cell-free systems as described previously (36). CHO-K1 cells were suspended in Buffer H (20 mM HEPES, 150 mM NaCl, 10% glycerol, 5 mM EDTA, pH 7.4) containing protease inhibitors (Complete), and thereafter the postnuclear supernatant was collected. The microsomal membrane was precipitated from the postnuclear supernatant by centrifugation at 100,000 ϫ g for 1 h at 4°C and resuspended in Buffer H containing protease inhibitors and incubated 37°C for 2 h to generate CTF⑀. The aliquots of the samples kept on ice for 2 h were used as negative controls. At the end of the assay, the microsomal membrane samples were separated into pellet and supernatant fractions by ultracentrifugation at 100,000 ϫ g for 1 h at 4°C. Each pellet fraction was suspended in SDS sample buffer, and each supernatant fraction was diluted with an equal volume of 2ϫ SDS sample buffer to be used for immunoblot analysis.
PKC Translocation Assay-The PKC translocation assay was carried out as described previously (37). APP-CHO-K1 cells were incubated for 2 days in the Nutridoma-BO medium in the absence or presence of 1 M ISP-1 or 1 M ISP-1 plus 1 M sphingosine for 48 and 73 h as indicated. Wild-type CHO, LY-B, and LY-B/cLCB1 cells were also incubated in Nutridoma-BO medium for 48 h. Thereafter, the cells were washed and scraped into 200 l of homogenization buffer (20 mM Tris-HCl, pH 7.4, 1 mM EDTA, 1 mM EGTA, protease inhibitors (Complete)), lysed by homogenizer, and centrifuged at 100,000 ϫ g for 1 h at 4°C. The pellets were resonicated in 200 l of homogenization buffer supplemented with 1% Triton X-100 and centrifuged at 100,000 ϫ g for 1 h at 4°C, yielding solubilized particulate fractions. The protein concentration was determined, and the fractions were analyzed by immunoblotting using antibodies, which recognize phosphorus-independent PKC␣, PKC␦, PKC⑀, and PKC␥.
Purification of the Lipid Raft Fraction-The lipid raft fraction was obtained from each cell line according to an established method previously reported (33,38). One milliliter of each fraction was sequentially collected from the top of the gradient. The extraction of lipids and subsequent determination of the amount of cholesterol and phospholipids in each sample were carried out according to previously described methods (39).
Statistical Analysis-Statistical analysis was carried out using Stat-View computer software (Macintosh, version 5.0; Abacus Concepts Inc., Berkeley, CA). A p value Ͻ 0.05 was considered to indicate statistical significance.

Sphingolipid Deficiency Induced by SPT Enhanced sAPP␣
Secretion in CHO-K1 Cell Lines-We used CHO-K1 cells stably transfected with human APP751 cDNA (APP-CHO-K1) (40) and treated these cells with myriosin (designated as ISP-1), which is a potent inhibitor of SPT. Using this inhibitor, we obtained the pharmacological cell model of sphingolipid deficiency (28). We determined the level of sphingolipid synthesis in APP-CHO-K1 cells treated with ISP-1 in a sphingolipiddeficient medium. Fig. 1a shows that the rate of de novo sphingomyelin synthesis in ISP-1-treated APP-CHO-K1 cells decreased significantly as previously reported (28,29). Using this culture system, we determined APP levels secreted from ISP-1-treated and nontreated APP-CHO-K1 cells and cellular APP levels in these cells. Immunoblot analysis using the 6E10 antibody, which recognizes the C terminus of human sAPP␣, showed that the secreted sAPP␣ level in sphingolipid-deficient cells is significantly higher (about 3.1-fold) than that in nontreated cells (Fig. 1, b and c). These results indicate that ␣-cleavage is activated in CHO-K1 cells treated with ISP-1. The immunoblot analysis of cellular APP using the 22C11 antibody, which recognizes the N terminus of APP, and the UT-18 antibody, which recognizes the C terminus of APP, showed that treatment with ISP-1 does not seem to affect cellular total APP level but significantly reduces the levels of the mature forms (N-and O-glycosylated forms) of APP (Fig. 1,  b and c). Treatment with ISP-1 does not affect the cellular ␣-tubulin level.
Effect of Sphingolipid Deficiency on Generation of CTF␣, CTF␤, and CTF⑀ in CHO-K1 Cell Lines-Next, we determined the sAPP␤ levels in the conditioned media of the SPT-treated and nontreated APP-CHO-K1 cells. We first immunoprecipitated sAPP␣ in the conditioned media of APP-CHO-K1 cells using the 6E10 antibody, and the remaining supernatant was used for sAPP␤ detection. The immunoblot analysis of the remaining supernatant using the 22C11 antibody showed that ISP-1 treatment does not have any significant effect on sAPP␤ levels (Fig. 2a). We next examined whether the level of Cterminal fragments of APP differ between ISP-1-treated and nontreated APP-CHO-K1 cells. We prepared microsomal fractions from each cell line and incubated them at 0 or 37°C for 2 h. In APP-CHO-K1 cells, CTF␣ was mainly detected at 10 kDa, and CTF␤ was weakly detected (Fig. 2b). When the cells were treated with ISP-1, the intensity of the band representing CTF␣ increased significantly at 4°C (Fig. 2, b and c). Compatible with the data shown in Figs. 1b and 2a, these findings indicate that ␣-cleavage is activated in CHO-K1 cells treated with the SPT inhibitor. It was reported that the incubation of the microsomal fraction generates CTF⑀ detected at ϳ6.5 kDa, which migrates below the major APP C-terminal fragments arising from CTF␣ and CTF␤ (6,36,41). We also observed that CTF⑀ was generated in the microsomal membranes of APP-CHO-K1 cells after a 2-h incubation (Fig. 2b). The level of CTF⑀ generated in the membrane fraction of the ISP-1-treated APP-CHO-K1 cells decreased significantly compared with that of the nontreated cells (Fig. 2b), and the level of CTF⑀ in the ISP-1treated cells decreased to 40% of that of the nontreated cells (Fig. 2c). These results indicate that the extent of ⑀-cleavage decreases in CHO-K1 cells treated with the SPT inhibitor.
Altered Processing of APP Was Restored by Adding Exogenous Sphingosine in ISP-1-treated Cells-To determine that Twenty-four hours after plating, the cells were washed twice with Ham's F-12, refed with sphingolipid-deficient medium (Nutridoma-BO medium) in the presence or absence of 1 M ISP-1, and maintained for another 2 days. a, using these cells, sphingomyelin incorporation rate was determined as described under "Materials and Methods." b, for the determination of the levels of released sAPP␣ and cellular APP in cultured cells with or without treatment, the cultured media were collected, and the cells were harvested. The levels of secreted sAPP␣ in the medium and intracellular APP were determined by immunoblot analysis using 6E10 (for sAPP␣) and 22C11 and UT-18 (for intracellular APP). c, the immunoreactivities of each sample to the 6E10 antibody in the medium and to the UT-18 antibody in the cell lysate were quantified using a Macintosh computer with software (National Institutes of Health Image) for densitometric analysis. The data represent the means Ϯ S.E. for triplicate experiments. *, p Ͻ 0.005 versus ISP-1 (Ϫ). Three independent experiments showed similar results.

FIG. 2. Effect of sphingolipid deficiency on secretion of sAPP␤and generation of CTFs in APP-CHO-K1 cells treated
with ISP-1. APP-CHO-K1 cells were seeded in the 10% FBS-containing medium in 10-cm 2 culture dishes. Twenty-four hours after plating, the cells were refed with the Nutridoma-BO medium, treated with or without 1 M ISP-1, and maintained for another 48 h. a, the culture media were harvested to detect sAPP␣ and sAPP␤ as described under "Materials and Methods." sAPP␣ in each culture medium was immunoprecipitated with the 6E10 antibody. The resultant supernatant was analyzed to detect sAPP␤ by immunoblot analysis using the 22C11 antibody. b, the pellet and supernatant fractions from microsomal membrane were obtained by further ultracentrifugation as described under "Materials and Methods." The generation of CTF␣ and CTF␤ were detected by immunoblot analysis with the UT18 antibody at ϳ10 kDa in the pellet fraction sample, and the generation of CTF⑀ was detected with the UT18 antibody at ϳ6 kDa in the samples of the supernatant fraction after 2 h of incubation. c, the intensities of the bands for CTF␣ at 4°C (left panel) and CTF⑀ at 37°C (right panel) after 2 h incubation were determined using a Macintosh computer with software (National Institutes of Health Image) for densitometric analysis. The data represent the mean Ϯ S.E. for triplicate experiments. *, p Ͻ 0.05 versus ISP-1 (Ϫ). Three independent experiments showed similar results. WB, Western blot; sup, supernatant; ppt, pellet. the increased sAPP␣ level is due to sphingolipid deficiency, we examined whether the addition of exogenous sphingosine restores the altered APP level in ISP-1-treated CHO-K1 cells. As shown in Fig. 3a, when exogenous sphingosine was concurrently treated with ISP-1, the sAPP␣ level decreased to that of the nontreated cells. Similarly, the CTF⑀ level decreased in the ISP-1-treated cells, which was also restored by adding exogenous sphingosine (Fig. 3a). There was no difference in the cellular ␣-tubulin level between these cultures. We further determined the levels of A␤40 and A␤42 in the nontreated and ISP-1-treated cultures by two-site ELISA. ISP-1 treatment increased A␤42 level (about 1.6-fold that of the nontreated cells), whereas it had no effect on the A␤40 level (Fig. 3b). When exogenous sphingosine was concurrently treated with ISP-1, the A␤42 level in the culture medium was restored to that of the nontreated cells (Fig. 3b).

MAPK/ERK, but Not PKC Activity Is Involved in the Enhancement of sAPP␣ Secretion Caused by Sphingolipid Deficiency-It
has been demonstrated that sAPP␣ secretion is regulated in either a PKC-dependent or -independent manner that involves the activation of tyrosine kinases (1, 2, 42). Moreover, the MAPK signaling pathway has recently been implicated in both PKC and tyrosine kinase receptor regulations of APP processing (4,43). We therefore assessed which, if any, of these kinases mediates the effect of sphingolipid deficiency on sAPP␣ secretion. We found that the ERK activity and the sAPP␣ secretion level increased in the APP-CHO cells treated with ISP-1 without a change in the level of total ERK, and these increases were restored by concurrent treatment with PD98059, a specific inhibitor of MAPK/ERK kinase (Fig. 4a). We also examined the levels of PKC that translocated into the plasma membrane in sphingolipid-deficient cells and noted that the levels of PKC␣, PKC␦, PKC⑀, and PKC␥ translocated into the plasma membrane fraction are not altered in sphingolipid-deficient cells (Fig. 4b). Furthermore, we carried out experiments to confirm whether sphingolipid deficiency also modulates the processing of endogenous APP in wild-type CHO cells. As shown in Fig. 4c, the sAPP␣ secretion level and ERK activity increased when sphingolipid level decreased following ISP-1 treatment, without a change in the level of total ERK; these increases were restored by concurrent treatment with PD98059, a specific inhibitor of MAPK/ERK kinase, and sphingosine sAPP␣ Secretion Level Also Increased in CHO-K1 Cell Mutant Strain (LY-B) Defective in the LCB1 Subunit of SPT-To confirm whether sphingolipid deficiency induced by ISP-1 treatment is a sphingolipid-specific phenomenon, we further examined the effect of sphingolipid deficiency on APP processing using the CHO-K1 cell mutant strain, LY-B, defective in the LCB1 subunit of SPT, which is unable to synthesize any sphingolipid species de novo (29). Another mutant CHO-K1 cell line, the LY-B/cLCB1 strain, which is the complemented transformant of LY-B, was also used (29). Because these cell lines express only endogenous hamster APP, we used G530 antibody, which recognizes rodent APP. When these cells were cultured in a sphingolipid-deficient medium for 2 days, the sphingomyelin level in LY-B cells decreased to ϳ15% of that in wild-type CHO-K1 cells (44). The sAPP␣ secretion level significantly increased in LY-B cells, whereas the total APP level and the level of ␣-tubulin, an internal control, remained unchanged; however, such an increase was not noted in LY-B/ cLCB1 cells (Fig. 5a). We further examined the effect of PD98059 on the secreted sAPP␣ level to determine whether the increase in the sAPP␣ level is mediated by MAPK/ERK activity. Similarly to the case of ISP-1-treated cells shown in Fig. 4, both ERK activity and the sAPP␣ secretion level increased in the sphingolipid-deficient cells, LY-B cells, whereas total ERK levels and the level of ␣-tubulin, an internal control, remained unchanged (Fig. 5b). These increases in ERK activity and the sAPP␣ level were restored by transfection with cLCB1 or treatment with PD98059 or sphingosine (Fig. 5b). These results indicate that sphingolipid deficiency increases sAPP␣ secretion level via ERK activation. In contrast, the levels of PKC␣, PKC␦, PKC⑀, and PKC␥ that translocated into the plasma membrane fraction were not altered in these three cell lines (Fig. 5c).

Characterization of Lipid Rafts of Wild-type CHO, LY-B, and LY-B/cLCB1 Cells and APP Localization in the Lipid Rafts in
These Cell Lines-We finally examined the effect of sphingolipid deficiency on lipid composition in lipid raft fractions and APP localization in lipid rafts. We treated cell lysate of CHO, LY-B, and LY-B/cLCB1 cells with Triton X-100, separated them in a sucrose density gradient (33,38), and determined the levels of cholesterol, phospholipids, and GM1, a marker for lipid rafts, in each fraction. As shown in Fig. 6a, the raft fraction (fraction 4) enriched in GM1 contained 20 and 18% of total phospholipids in the wild-type CHO and LY-B/cLCB1 cells, respectively, whereas 12% of total phospholipids was recovered in fraction 4 of the LY-B cells. In contrast, the dis-

FIG. 3. Cellular sphingolipid level modulated sAPP␣secretion, the cellular CTF⑀level, and the secretion of A␤40 and A␤42 in APP-CHO-K1 cells.
Twenty-four hours after plating in the 10% FBScontaining medium in 10-cm 2 culture dishes, the culture medium was changed with the sphingolipid-deficient medium with no ISP-1, 1 M ISP-1, or 1 M ISP-1 plus 1 M D-erythro-sphingosine. The cultures were then maintained for another 48 h, followed by washing with phosphate-buffered saline twice, and the culture medium was harvested, and the microsomal fractions were prepared as described above. a, sAPP␣ in the cultured medium was detected by immunoblot analysis with the 6E10 antibody, and CTF⑀ generated in the membrane pellet after incubation at 37°C for 2 h was detected with the UT18 antibody. Three independent experiments showed similar results. b, the levels of A␤40 and A␤42 were quantified by sandwich ELISA using the BNT77 and BC05 antibodies. The data represent the means Ϯ S.E. for triplicate experiments. *, p Ͻ 0.05 versus ISP-1 (Ϫ)/sphingosine (Ϫ) and ISP-1 (ϩ)/sphingosine (ϩ). tribution peak of cholesterol in the raft fraction in LY-B cells remained similar or rather higher levels compared with those for the other two genotypes (Fig. 6a). These results suggest that the structure of the raft domain may have been altered, and thus their function deteriorated. We further determined the FIG. 4. Effect of sphingolipid deficiency on MAPK/ERK activity and PKC translocation in CHO cells. APP-CHO-K1 cells were seeded in the 10% FBS-containing medium in 10-cm 2 culture dishes. Twenty-four hours after plating, the cells were refed with the Nutridoma-BO medium and treated with or without 1 M ISP-1 or other reagent and maintained for another 48 h. Thirty minutes before changing the medium, PD98059 was added to the cultures, and then the medium was changed to a fresh Nutridoma-BO medium containing the indicated reagent(s). a, four hours after the medium change, the culture medium and the cells were harvested to detect sAPP␣, phospho-ERK, and total ERK as described under "Materials and Methods." b, the membrane fractions were prepared and subjected to immunoblot analysis using anti-PKC␣, PKC␦, PKC⑀, and PKC␥ antibodies as described under "Materials and Methods." c, six hours after the medium change, the culture medium and the cells were harvested to detect sAPP␣, phospho-ERK, and total ERK as described under "Materials and Methods."

FIG. 5. Effect of MAPK/ERK activity on the levels of APP secreted from wild-type CHO, LY-B, and LY-B/cLCB1 cells.
Wildtype CHO (WT), LY-B, and LY-B/cLCB1 cells were seeded in the 10% FBS containing medium in 10-cm 2 culture dishes. Twenty-four hours after plating, the cells were refed with the Nutridoma-BO medium and maintained for another 48 h. a, the culture medium was again changed to a fresh Nutridoma-BO medium. The cultures were then further incubated for 6 h, and the culture media and the cells were harvested to detect sAPP␣ and cellular APP using 22C11 antibody. b, for determination of the effect of ERK activity on sAPP␣ secretion, PD98059 was pretreated 30 min before changing to a fresh Nutridoma-BO medium containing the indicated reagent(s). Six hours following the medium change, the culture medium and the cells were harvested to detect sAPP␣, phospho-ERK, and total ERK as described under "Materials and Methods." c, the membrane fractions were prepared and subjected to immunoblot analysis using anti-PKC␣, PKC␦, PKC⑀, and PKC␥ antibodies as described under "Materials and Methods." APP levels in these fractions by immunoblot analysis using the monoclonal antibody, 22C11, and the intensities of APP signal in the raft fraction were quantified by densitometric analysis. As shown in Fig. 6 (b and c), the amount of APP recovered in the raft fraction of each cell line was at a similar level.

DISCUSSION
In this study, we found that the sphingolipid deficiency induced by the SPT inhibitor enhances ␣-cleavage of APP without altering the total amount of cellular APP. These alterations are restored by adding exogenous sphingosine. The enhanced ␣-cleavage of APP caused by sphingolipid deficiency is confirmed in cells whose sphingolipid synthesis is genetically defective, indicating that cellular sphingolipid level is a critical modulator of APP processing to secrete sAPP␣. We also found that MAPK/ERK is activated in sphingolipid-deficient cells and that the inhibition of MAPK/ERK pathway restores sAPP␣ level, suggesting that sphingolipid deficiency enhances sAPP␣ secretion via activation of the MAPK/ERK pathway.
It was shown that APP cleavage by ␣-secretase is dependent on the cellular cholesterol level (12), and sAPP␣ secretion and A␤42 generation are determined by the dynamic interactions of APP with lipid rafts (15,16,18), probably because of the alteration of both APP and ␤-secretase partitioning into lipid rafts (13,17). Because cholesterol depletion is postulated to disrupt raft functions (45), our present results suggest that the depletion of sphingolipids, another major component of lipid rafts, affects lipid raft functions, thereby altering APP processing as noted in cholesterol-depleted cells.
A question arises of how sphingolipid deficiency alters APP processing to enhance sAPP␣ secretion. Because PKC and ERK modulate sAPP␣ secretion (1-4, 43, 46), we examined whether PKC and ERK are responsible for the increased levels of sAPP␣ secreted from sphingolipid-deficient cells. We found that the decreased sphingolipid level in sphingolipid-deficient cells enhances ERK activity (Figs. 4 and 5) and that the inhibition of ERK activity by PD98059 restores the increase in sAPP␣ level in sphingolipid-deficient cultures, suggesting that increased levels of ERK activity associated with sphingolipid deficiency enhance sAPP␣ secretion. In contrast, we did not observe any alteration in the amount of PKC that translocated to the plasma membrane in the sphingolipid-deficient cells. Although ERK is located in the downstream of the PKC signaling cascade (4,43,46,47), these results suggest that sphingolipid deficiency activates ERK in a pathway different from the PKC pathway. The mechanism by which sphingolipid deficiency causes ERK activation is not yet known; however, it is important to note that many raft-associated proteins mediate signal transduction (20,45,48) and that cholesterol depletion also stimulates ERK activity in neurons (49) and non-neuronal cells (50,51). Interestingly, cholesterol deficiency is reported to increase sAPP␣ secretion level, although it is not clear whether the MAP/ERK pathway is involved in the cholesterol depletionmediated increase in sAPP␣ secretion level (12). These lines of evidence suggest that altered levels of cholesterol or phospholipids in lipid rafts may affect the raft-mediated signal transduction pathway, ERK, leading to an increase in the sAPP␣ secretion level.
However, the possibility cannot be excluded that cholesterol and sphingolipid depletion enhances APP ␣-cleavage in a different manner, because previous reports suggested that disruption in the formation of lipid rafts and their clustering caused by the depletion of cholesterol in lipid rafts lead to the inhibition of APP partitioning into lipid rafts, a decrease in the ␤-cleavage activity, and an increase in ␣-cleavage activity of APP (7,13). In contrast, our findings show that under the conditions in which cellular cholesterol level is unchanged, lipid raft dysfunctions caused by sphingolipid depletion may enhance the ␣-cleavage activity of APP without affecting the ␤-cleavage activity of APP or APP level in lipid rafts (Fig. 6b). These results imply that sphingolipid depletion may enhance APP ␣-cleavage activity without shifting the intracellular trafficking of APP from the A␤-generating site (lipid rafts) to the A␤-nongenerating site (outside lipid rafts). These results suggest that cholesterol and sphingolipids play entirely different roles in determining the properties of lipid rafts. In support of this notion, the effect of sphingolipid depletion opposite to that Wild-type CHO (WT), LY-B, and LY-B/cLCB1 cells were seeded in the 10% FBS containing medium in 10-cm 2 culture dishes. Twenty-four hours after plating, the culture medium was changed with the Nutridoma-BO medium, and the cells were maintained for another 48 h. The cells were then harvested, homogenized in the presence of 1% Triton X-100, and fractionated by sucrose density gradient centrifugation as described previously (33,38). The fractions were collected from the top gradient, and 12 fractions were obtained. a, the levels of cholesterol and phospholipids in each fraction were determined as described previously (39). b, for immunoblot analysis using the 22C11 antibody, the cells were harvested, homogenized, and subjected to immunoblot analysis as described under "Materials and Methods." c, the intensity of APP signal in the lipid rafts fraction (fraction 4) of each cell line was analyzed. The data are the means Ϯ S.E. for triplicate experiments. There is no significant difference between these three lines. a-c, three independent experiments showed similar results. of cholesterol depletion on the formation of the scrapie prion protein, which is assumed to occur in lipid rafts, was demonstrated (27,52). However, the mechanism(s) underlying the different effects of cholesterol and sphingolipid depletion on APP processing remains unclarified, and further studies are required elucidate the regulation of ␣-secretase, ADAM 10 (a disintegrin and metalloprotease) (53), based on lipids present in rafts and outside lipid rafts.
Recently, it has been shown that ceramide enhances the biogenesis of A␤ by modulating APP ␤-cleavage (54). It was also shown that an increased level of cellular ceramide increases the level of ␣and ␤-APP-CTFs, indicating that the sAPP␣ secretion level also increases. In our experiments, sAPP␣ secretion level increased in the cells in which the levels of sphingolipid including ceramide are reduced. One cannot explain this discrepancy at present; however, it may be possible that the profiles of the cellular levels of phospholipids in our experiments and others are different. For example, both ISP-1 (our study) and fumonisine B1 (54) reduce the cellular levels of ceramide, glycosphingolipids, and sphingomyelin; however, fumonisine B1 increases the level of dehydrosphingosine, which is a precursor of ceramide, whereas ISP-1 reduces its level. Because dehydrosphingosine has various biological effects on cells such as PKC activity (55), the different effect of fumonisine B1 from that of ISP-1 on the dehydrosphingosine level may in part explain the contradictory result of sAPP␣ secretion caused by fumonisine B1 to that induced by ISP-1 and SPT deficiency.
Our data show that an increase in the level of secreted A␤42 is accompanied by a decreased activity of APP ⑀-cleavage in sphingolipid-deficient cells, supporting in part our previous finding that A␤42-specific elevation accompanied by the significant reduction of sphingolipids in lipid rafts are noted in the mutant presenilin 2 transgenic mouse brains (33). Interestingly, the different effects of sphingolipid deficiency on A␤42 generation and APP ⑀-cleavage agree with the findings reported in previous studies using cells with PS mutations (56,57). In those studies, it was shown that an increase in the level of APP ␥42-cleavage is accompanied by a decrease in the ⑀-cleavage level in various PS1 mutant cells. These data suggest that ␥-secretase at residue 42 and ⑀-cleavage are likely to be reciprocally regulated in PS mutant cells and that cellular sphingolipids may be involved in these regulations. It has been also shown that the activation of PKC stimulates ␣-cleavage of APP at the expense of ␤-secretase cleavage (58 -60). In contrast, other studies demonstrated that PKC activation enhances sAPP␣ release without decreasing A␤ production (43,61,62). Similarly, cholesterol depletion increases sAPP␣ secretion level and reduces A␤ production (7,12,13), whereas the data presented here indicate that the phospholipid deficiencyinduced activation of ERK enhances sAPP␣ secretion, although it does not inhibit A␤ generation but rather increases A␤42 secretion level. These different results provide evidence of the sAPP␣ production and of A␤ being derived from distinct metabolic pathways that can be differently regulated by cholesterol or phospholipids.
Finally, our present study raises the caution that not only cholesterol but also sphingolipids should be focused on when one discusses the relationship between lipid rafts and AD development. Further studies are required to clarify whether PS mutations alter the sphingolipid metabolism and whether alterations in sphingolipid metabolism are associated with sporadic AD development. However, our observations in the present study provide a new insight into one of the central issues concerning AD pathogenesis, that is, the relationship between altered lipid metabolism and the development of AD.