Purification, Characterization, Molecular Cloning, and Subcellular Distribution of Neutral Ceramidase of Rat Kidney

Previously, we reported two types of neutral ceramidase in mice, one solubilized by freeze-thawing and one not. The former was purified as a 94-kDa protein from mouse liver, and cloned (Tani, M., Okino, N., Mori, K., Tanigawa, T., Izu, H., and Ito, M. (2000) J. Biol. Chem. 275, 11229--11234). In this paper, we describe the purification, molecular cloning, and subcellular distribution of a 112-kDa membrane-bound neutral ceramidase of rat kidney, which was completely insoluble by freeze-thawing. The open reading frame of the enzyme encoded a polypeptide of 761 amino acids having nine putative N-glycosylation sites and one possible transmembrane domain. In the ceramidase overexpressing HEK293 cells, 133-kDa (Golgi-form) and 113-kDa (endoplasmic reticulum-form) Myc-tagged ceramidases were detected, whereas these two proteins were converted to a 87-kDa protein concomitantly with loss of activity when expressed in the presence of tunicamycin, indicating that the N-glycosylation process is indispensable for the expression of the enzyme activity. Immunohistochemical analysis clearly showed that the ceramidase was mainly localized at the apical membrane of proximal tubules, distal tubules, and collecting ducts in rat kidney, while in liver the enzyme was distributed with endosome-like organelles in hepatocytes. Interestingly, the kidney ceramidase was found to be enriched in the raft microdomains with cholesterol and GM1 ganglioside.

Over the past decade, sphingolipids and their metabolites have emerged as a new class of lipid biomodulators of various cell functions (1,2). Ceramide (N-acylsphingosine; Cer), 1 a common lipid backbone of sphingolipids, functions as a second messenger in a variety of cellular events including apoptosis and cell differentiation (3,4). Sphingosine (Sph) has bifunctional effects on cell growth, i.e. it exerts mitogenic (5) and apoptosis inducing (6) activities, depending on the cell type and cell cycle. Sph-1-phosphate (S1P) was found to function as an intra-and intercellular second messenger to regulate cell growth (7), motility (8), and morphology (9). Interestingly, S1P inhibits the apoptosis induced by Cer and Fas ligand (10), indicating that the balance of Cer/Sph/S1P affects cell phenotype.
Ceramidase (CDase, EC 3.5.1.23) is an enzyme that catalyzes hydrolysis of the N-acyl linkage of Cer to produce Sph, which can be phosphorylated to S1P by sphingosine kinase (11). Sph is not produced by de novo synthesis (12), and thus the activity of CDase is crucial not only for switching off the Cer-induced signaling but also for generation of Sph and S1P. CDase is classified into two categories: acid and neutral/alkaline enzymes depending on pH optimum. Acid CDase is thought to be a housekeeping enzyme to catabolize Cer in lysosomes. The enzyme was purified from human urine (13), and cDNA encoding the enzyme was isolated from cDNA libraries of human (14) and mouse (15). A deficiency of the enzyme could cause Farber disease in which Cer is accumulated in lysosomes (16). Neutral/alkaline CDase seems to change the balance of Cer/Sph/S1P in response to various stimuli including cytokines and growth factors, and could modulate the sphingolipid-mediated signaling. For example, the activity of membrane-associated neutral CDase was shown to be up-regulated by platelet-derived growth factor in rat glomerular mesangial cells (17), and the enzyme activity was modulated in a bimodal manner by interleukin-1␤ in rat hepatocytes (18), resulting in a decrease of Cer concomitantly with an increase of Sph. However, the biological function of the enzyme is still not clear. Recently, cDNAs encoding sphingomyelinase, Sph kinase (19), and S1P receptors (Edg family) (20) have been successively cloned. The functions of sphingolipids are now open for elucidation at the molecular level.
In the past few years, molecular cloning of neutral/alkaline CDases, one of the missing links of sphingolipid signaling, has been performed in mice (21), human (22), bacteria (23), and yeast (24). In mice, we found two types of neutral CDase, one solubilized by freeze-thawing and the other not. The former was purified as a 94-kDa protein from mouse liver (25), and the cDNA encoding the enzyme was cloned (21). In the present paper, we report the purification, characterization, and cDNA cloning of a 112-kDa membrane-bound CDase of rat kidney, which was absolutely resistant to extraction with freeze-thawing and had an optimum pH of 6 -7. It is worth noting that neutral/alkaline CDase of human brain is specifically localized in mitochondria, suggesting the existence of a Cer pool in this organelle (22). On the other hand, we show here using a specific antibody against the neutral CDase that the enzyme was mainly localized at apical membranes of proximal tubules, distal tubules, and collecting ducts in rat kidney, while in rat liver the enzyme was distributed with endosome-like organelles in hepatocytes. Furthermore, the kidney CDase was recovered in the detergent-insoluble, cholesterol, and GM1-enriched fractions by sucrose density gradient centrifugation, suggesting that the enzyme is present in the raft microdomains.
CDase Assay-CDase activity was measured using C12-NBD-Cer as a substrate (25). Briefly, 550 pmol of C12-NBD-Cer was incubated at 37°C for 30 min with an appropriate amount of the enzyme in 20 l of 25 mM Tris-HCl buffer, pH 7.5, containing 0.25% Triton X-100. The reaction was stopped by heating in a boiling water bath for 5 min. After being dried up with a Speed Vac concentrator (Savant Instruments, Inc.), the sample was dissolved in 30 l of chloroform/methanol (2/1, v/v), and applied to a TLC plate, which was developed with chloroform, methanol, 25% ammonia (90/20/0.5, v/v). The NBD-dodecanoic acid released and C12-NBD-Cer produced were quantified with a Shimadzu CS-9300 chromatoscanner (Shimadzu, Japan). One enzyme unit was defined as the amount capable of catalyzing the release of 1 mol of NBD-dodecanoic acid/min from C12-NBD-Cer under the conditions described above. A value of 10 Ϫ3 and 10 Ϫ6 units of enzyme was expressed as 1 milliunit and 1 microunit, respectively. To characterize the CDase, various cations at 5 mM as a final concentration were added, or 150 mM GTA buffer at different pH values was used instead of Tris-HCl buffer. In some cases, 100 pmol of [ 14 C]Cer (C16:0/d18:1) was used instead of the fluorescent Cer as a substrate. The reverse hydrolysis reaction of the CDase was determined by the method described in Ref. 25.
Extraction of the CDase from Rat Kidney-Fresh rat kidneys (141 g wet weight from 53 rats) were homogenized with Polytron-RT3000 in 300 ml of 0.25 M sucrose containing 1 mM EDTA. The homogenate was centrifuged at 600 ϫ g for 5 min to remove debris. The supernatant was centrifuged at 27,000 ϫ g for 30 min and the pellet obtained (membrane fraction) was re-suspended in 200 ml of 10 mM potassium phosphate buffer, pH 7.0, containing 1% Triton X-100 and 1% Tween 20. After sonication for 2 min on ice, the homogenate was then centrifuged at 105,000 ϫ g for 90 min. The supernatant was used as a detergentextracted fraction.
Freeze-thawing Experiment-The membrane fractions of rat kidney and liver were suspended in 100 l of 0.25 M sucrose containing 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, and 2 g/ml leupeptin. The solutions were frozen in ethanol-dry ice immediately and thawed in warm water at 37°C. The procedure was repeated the number of times indicated. After centrifugation at 105,000 ϫ g for 60 min, the supernatant obtained was used as the soluble fraction. The precipitate was suspended in 100 l of 20 mM Tris-HCl buffer, pH 7.5, containing 0.2% Triton X-100 and used as a membrane-bound fraction.
Purification of Neutral CDase-The CDase (1.14 units), extracted from the membrane fraction of rat kidney with a mixture of 1% Triton X-100 and 1% Tween 20, was loaded onto a DEAE-Sepharose FF column (200 ml) equilibrated with 20 mM Tris-HCl buffer, pH 7.5, containing 0.1% Lubrol PX (buffer A) at a flow rate of 5 ml/min using a BPLC-600FC HPLC system (Yamazen Co., Osaka, Japan). After sample loading, the column was washed with 300 ml of buffer A followed by a linear gradient of 0 -1 M NaCl in buffer A at a flow rate of 5 ml/min. The eluted fractions containing the CDase activity were pooled, and then applied to a phenyl-Sepharose 6FF column (100 ml) at a flow rate of 5 ml/min using a BPLC-600FC HPLC system. The column was washed with 200 ml of 20 mM Tris-HCl buffer, pH 7.5, and then the CDase was eluted using a linear gradient of 0 -2% Lubrol PX in 20 mM Tris-HCl buffer, pH 7.5, at a flow rate of 5 ml/min. The eluate was pooled and loaded onto a column of chelating Sepharose FF (50 ml) at a flow rate of 5 ml/min using a BPLC-600FC HPLC system. The column was washed with 200 ml of buffer A and 100 ml of 20 mM Tris-HCl buffer, pH 7.5, containing 0.5 M NaCl and the CDase activity was eluted with 20 mM Tris-HCl buffer, pH 7.5, containing 2 M NH 4 Cl at a flow rate of 5 ml/min. The fractions containing the enzyme activity were pooled and concentrated about 10-fold using a MiniTan ultrafiltration system (Millipore). The buffer was exchanged to 20 mM Tris-HCl buffer, pH 7.5, using the same apparatus. The enzyme solution was loaded onto a HiTrap lentil lectin column (1 ml, Amersham Pharmacia Biotech) equilibrated with 20 mM Tris-HCl buffer, pH 7.5, containing 1 mM MnCl 2 , 1 mM CaCl 2 , and 0.5 M NaCl at a flow rate of 0.4 ml/min using a BioCAD system (Applied Biosystems). The CDase was eluted with 20 mM Tris-HCl buffer, pH 7.5, containing 0.5 M methyl-␣-D-glucoside. The active fractions were pooled and concentrated using a Centriprep (Millipore) and loaded onto a HiLoad 16/60 Superdex 200 pg column (Amersham Pharmacia Biotech) equilibrated with 20 mM Tris-HCl buffer, pH 7.5, containing 0.15 M NaCl and 0.3% Lubrol PX at a flow rate of 0.8 ml/min using a BioCAD system.
Amino Acid Microsequencing-The purified CDase was concentrated with a Y-shaped gel a modified form of a funnel-shaped gel (27). After the concentration, the 112-kDa protein band localized with Coomassie Brilliant Blue was cut out, equilibrated with SDS sample buffer, and loaded again on a 7.5% SDS-PAGE gel. After electrophoresis, the gel was blotted on a polyvinylidine difluoride membrane and stained with Coomassie Brilliant Blue. The 112-kDa protein (about 3 g) was cut out and treated in situ with lysylendopeptidase AP-1 (Wako Pure Chemical Industries, Osaka, Japan). Peptides released from the membrane were fractionated by reversed-phase HPLC using a C8 column (1.0 ϫ 100 mm), and sequenced using a pulse-liquid phase protein sequencer (Procise cLc, Applied Biosystems).
cDNA Cloning and Sequencing-The sequences of four peptides obtained after digestion with lysylendopeptidase AP-1 showed high identity to the mouse liver neutral CDase (21), and thus we designed two primers based on the nucleotide sequence of the mouse enzyme. PCR using sense (5Ј-AGGAAATGTTGCTAATGTGC-3Ј) and antisense primers (5Ј-GGTGACACGTCTCCGAGAT-3Ј) was performed with the cDNA library of rat kidney (Takara Shuzo Co., Otsu, Japan) as a template in a GeneAmp PCR System 9700 (Applied in Biosystems) using AmpliTaq Gold (Applied in Biosystems). The cycling parameters for PCR were 94°C for 30 s, 51°C for 30 s, and 72°C for 30 s, and the cycle number was 40. After this amplification, a 325-base pair PCR product containing the sequence of rat CDase was obtained. To obtain the full-length cDNA encoding the rat CDase, colony hybridization was performed using the ␣-32 P-labeled 325-base pair PCR product as a probe after concentration of the CDase cDNA with a CloneCapture TM Selection Kit (CLONTECH). The probe was labeled with [␣-32 P]dCTP using Ready-To-Go DNA labeling kit (Amersham Pharmacia Biotech). Colony hybridization was carried out by the standard method (28). Nucleotide sequences were determined by the dideoxynucleotide chain termination method with a Bigdye Terminator Cycle Sequencing Ready Reaction Kit (Applied Biosystems), and a DNA Sequencer (model 377A, Applied Biosystem).
Preparation of Recombinant CDase and Generation of Polyclonal Antibodies-A cDNA fragment encoding the open reading frame of rat CDase was prepared by PCR using 5Ј primer containing a HindIII site (5Ј-AGTAAGCTTATCGAAAACCACAAAGATTCAGGGA-3Ј) and 3Ј primer containing a XhoI site (5Ј-GCCGCTCGAGAGTAGTGACAATT-TCAAAAGGGGAAGA-3Ј) and the cloned rat cDNA (pAPkCD) as a template. The PCR product was inserted into the HindIII and XhoI sites of pET23b vector (Novagen) with a COOH-terminal histidine tag. Escherichia coli strain BL21(DE3) was transformed with the construct in the presence of ampicillin (100 g/ml). To obtain the recombinant CDase, 2 ml of overnight culture was inoculated into 100 ml of LB in the presence of ampicillin and incubated at 37°C. When the absorbance at 600 nm reached 0.6, isopropyl-1-thio-␤-D-galactopyranoside was added to a final concentration of 1 mM, and incubated for no more than 5 h at 37°C. Cells were harvested by centrifugation at 5,000 ϫ g for 10 min at 4°C, and the pellet was suspended in 10 mM Tris-HCl buffer, pH 7.5, containing 1% Triton X-100 and 1 mM EDTA. After centrifugation at 15,000 ϫ g for 10 min, the pellet (inclusion bodies) was lysed by sonication in 50 mM Tris-HCl buffer, pH 7.5, containing 8 M urea. Recombinant protein was purified using a HiTrap chelating column (Ni 2ϩ ) according to the manufacturer's instructions. Purified protein was dialyzed against distilled water before being used for immunization. From a rabbit immunized with the purified recombinant CDase, antiserum was obtained and purified with a HiTrap Protein A column according to the manufacturer's instructions.
Cell Culture and cDNA Transfection-CHOP cells, Chinese hamster ovary cells that express polyoma LT antigen for supporting efficient replication of eukaryotic expression vectors (29), were grown in a ␣-minimal essential medium supplemented with 10% fetal calf serum, 100 g/ml streptomycin, and 100 units/ml penicillin in a humidified incubator containing 5% CO 2 . HEK293 cells, human embryonic kidney cell, were grown in a Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, 60 g/ml kanamycin in a humidified incubator containing 5% CO 2 . cDNA transfection was carried out using LipofectAMINE TM Plus (Life Technologies, Inc.) according to the instructions of the manufacturer. To obtain Myc-tagged CDase, cDNA encoding the CDase was subcloned into pcDNA3.1/Myc-His(ϩ) vector (Invitrogen Co.) by PCR using a 5Ј primer with a KpnI restriction site (5Ј-AGGGTACCGAAATGGCAAAGCGAACCTTCTCC-3Ј) and a 3Ј primer with a XhoI restriction site and disrupted stop codon (5Ј-GC-CGCTCGAGAGTAGTGACAACTTCAAAAGGAGAAGA-3Ј). Cells were treated with tunicamycin to block the N-glycosylation of neutral CDase. Medium containing tunicamycin (10 g/ml) was added at 4 h after transfection of CDase gene and cells were harvested after 12 h.
Protein Assay, SDS-PAGE, and Western Blot-Measurement of protein was determined by the bicinchoninic acid protein assay (Pierce) with bovine serum albumin as standard. SDS-PAGE was carried out according to the method of Laemmli (30). Protein transfer onto a polyvinyldifluoride membrane was performed using TransBlot SD (Bio-Rad) according to the method described in Ref. 31. After treatment with 3% skim milk in Tris-buffered saline (TBS) containing 0.1% Tween 20 (T-TBS) for 1 h, the membrane was incubated with first antibody (anti-neutral CDase antibody, anti-Myc antibody (Invitrogen), or anti-CD71 antibody (Harian Sera-Lab)) for 1 day at 4°C. After a wash with T-TBS, the membrane was incubated with horseradish peroxidaseconjugated secondary antibody for 2 h. After another wash with T-TBS, the ECL reaction was performed for 2-3 min as recommended by the manufacturer, and chemiluminescent signals were visualized on STORM (Amersham Pharmacia Biotech).
Subcellular Fractionation-A Percoll density gradient centrifugation was performed according to the method described in Ref. 32. The tissues were homogenized for 15-20 strokes in 20 mM Tris-HCl buffer, pH 7.0, containing 0.25 M sucrose and 1 mM EDTA with a glass-Teflon homogenizer. The homogenate was centrifuged at 750 ϫ g for 10 min. The supernatant obtained was centrifuged at 20,000 ϫ g for 10 min, and the pellet was resuspended in 20 mM Hepes-KOH buffer, pH 7.0, containing 0.25 M sucrose. This suspension was mixed with isotonic Percoll solution (1 ml of 200 mM Hepes-KOH, pH 7.0, containing 0.25 M sucrose was mixed with 9 ml of Percoll) at a ratio of 55/45 (v/v) and then centrifuged at 35,000 ϫ g for 90 min using a 10-ml centrifuge tube. The resulting gradient was divided into 10 fractions from top to bottom. Activities of ␤-galactosidase and alkaline phosphatase were measured using chlorophenol red ␤-D-galactopyranoside (Roche Molecular Biochemicals) and NBT/BCIP (Roche Molecular Biochemicals) as substrates, respectively. An appropriate amount of each fraction was incubated with 50 l of 5 mM chlorophenol red ␤-D-galactopyranoside containing 2.5 mM MgCl 2 or NBT/BCIP (3.75/1.88 g) in 100 l of 100 mM Tris-HCl buffer, pH 9.5, containing 100 mM NaCl and 5 mM MgCl 2 at 37°C for a given period. After incubation, absorbance at 574 and 550 nm were measured for quantification of ␤-galactosidase and alkaline phosphatase activities, respectively.
Immunohistochemistry and Fluorescence Microscopy-Samples (rat kidney and liver) were fixed with 4% paraformaldehyde in PBS overnight at 4°C, rinsed with PBS and 50 mM NH 4 Cl in PBS, and then infiltrated with 20% sucrose in PBS overnight at 4°C. The materials were embedded in OCT compound, rapidly frozen using liquid nitrogen, and stored at Ϫ80°C. The frozen materials were cut into 8-m thick sections using a cryostat (Leica CM1850) and mounted on poly-L-lysinecoated glass slides. After treatment with 5% skim milk in PBS (blocking buffer) for 20 min at room temperature, the samples were incubated with the anti-neutral CDase antibody diluted 1:100 with a blocking buffer for 2 h at room temperature followed by incubation with Cy3labeled anti-rabbit IgG (Amersham Pharmacia Biotech) at room temperature for 1 h. For controls, the primary antibody was replaced by preimmune serum IgG. For double labeling, the samples were stained with anti-neutral CDase and anti-LGP85 followed by incubation with a mixture of Cy3-labeled anti-rabbit IgG and fluorescein isothiocyanate (FITC)-conjugated anti-mouse Ig (Amersham Pharmacia Biotech). Immunostained samples were examined with a confocal laser scanning microscope (Olympus LSM-GB200). A part of immunostained samples were incubated with FITC-conjugated phalloidin (Sigma) to visualize actin filaments.
Northern Blot Analysis-A Northern blot membrane loaded with ϳ2 g of poly(A) ϩ RNA per lane from 8 different rat tissues (CLONTECH) was hybridized with a 2.0-kilobase EcoRI fragment of pAPkCD, which was gel-purified and labeled with [␣-32 P]dCTP using the Multiprime DNA labeling system (Amersham Pharmacia Biotech). Hybridization was carried out at 42°C for 20 h. Detection and quantification were performed using a BAS 1500 imaging analyzer (Fuji Film, Tokyo, Japan). Preparation of a Lipid Raft from Rat Kidney-A detergent-insoluble lipid raft was prepared as described in Refs. 33 and 34 with some modification. In brief, a fresh rat kidney (200 mg wet weight) was homogenized in 5 ml of 50 mM Tris-HCl buffer, pH 7.5, containing 150 mM NaCl, 2 g/ml leupeptin, and 1 mM phenylmethylsulfonyl fluoride (homogenized buffer) with a manually operated Teflon-glass homogenizer (20 strokes). The homogenate was centrifugation at 500 ϫ g for 5 min, and the resulting supernatant was then centrifuged at 48,000 ϫ g for 30 min to obtain a pellet of total membrane. The pellet was resuspended in 1.5 ml of the homogenized buffer containing 0.5% Lubrol 17A17 (serva), extracted for 30 min on ice, and mixed with an equal volume of 80% sucrose in the same buffer. The raft fraction was prepared by layering 5-30% sucrose on top of the extract, followed by centrifugation at 200,000 ϫ g for 20 h. After centrifugation, 10 fractions were collected from top to bottom of the gradient and dialyzed against distilled water. The free-cholesterol in each fraction was measured using a cholesterol oxidase/peroxidase (35), with N-acetyl-3,7-dihydroxyphenoxazine (Amplex Red, Molecular Probe) as substrate instead of p-hydroxyphenylacetic acid. Detection of ganglioside GM1 was performed using a biotin-labeled cholera toxin B subunit as described below. An aliquot of 50 l of each fraction was dried up, dissolved in 10 l of chloroform/methanol (2/1, v/v), and applied to a Polygram Silica G TLC plate (Macherey-Nagel, Germany). After being developed with chloroform, methanol, 0.02% CaCl 2 (5/5/1, v/v), the TLC plate was blocked with 1% skim milk-TBS for 1 h, and incubated with biotinlabeled cholera toxin B subunit (1 g/ml in T-TBS) for 2 h and then streptavidin-alkaline phosphatase (Sigma) for 1 h at room temperature. Detection of alkaline phosphatase was conducted using NBT/BCIP as a substrate as described above. Fig. 1, the CDase of rat liver was solubilized from the membrane fraction by freeze-thawing (A), whereas the enzyme of rat kidney was completely resistant to extraction with freeze-thawing (B). This discrepancy was also observed when the neutral CDase was extracted from mouse liver and kidney (25). The neutral CDase of mouse liver was easily solubilized by freeze-thawing, and purified as a 94-kDa protein, and thus cDNA encoding the enzyme was cloned (21). However, the CDase resistant to solubilization by freeze-thawing had yet to be purified from mouse liver or kidney, because of the limited amount of the enzyme. In this study, we found that the microsome fraction (27,000 ϫ g pellet) of rat kidney contained a large amount of membrane-bound neutral CDase which was solubilized with a mixture of 1% Triton X-100 and 1% Tween 20 with high yield as described under "Experimental Procedures." Following purification using chromatography on DEAE-Sepharose FF, phenyl-Sepharose 6FF, chelating Sepharose FF, HiTrap lentil lectin, and HiLoad Superdex 200 pg, 48 g of CDase was obtained from 106 rat kidneys with 3.1% recovery ( Table I). The specific activity increased 3,640-fold in the microsome fraction. On gel filtration using a HiLoad Superdex 200 pg column, the final step of the purification, the CDase activity was eluted in fractions 78 -84 as shown by shadow in Fig. 2A. Aliquots of the fractions were subjected to SDS-PAGE followed by staining with silver solution (Fig. 2B). Among several staining bands, a 112-kDa protein is likely to be the CDase (Fig. 2B), since the elution profile of the CDase activity from the gel filtration most coincided with that of the 112-kDa band (Fig. 2C). To determine whether the 112-kDa protein is the CDase or not, SDS-PAGE of the final preparation of CDase (fraction 82 of Fig. 2A) was performed under nonreducing conditions at 5°C. After electrophoresis, the gel was cut into 2-mm slices, from which CDase was extracted and measured using C12-NBD-Cer as a substrate at pH 7.5. The gel slice extract showing the highest CDase activity was then subjected to SDS-PAGE after reduction of the sample with 2-mercaptoethanol. As a result, the 112-kDa band was detected as a major staining band under the reducing conditions (Fig. 2D), indicating that the 112-kDa protein is likely to be a CDase. The rat 112-kDa CDase seems to be a glycoprotein with N-glycans, because treatment of the enzyme with glycopeptidase F resulted in the generation of a 89-kDa protein (Fig. 2E, lanes 1 and 2). On the other hand, the molecular mass of the mouse neutral CDase was found to be 94 kDa on SDS-PAGE, which was reduced to 75 kDa after digestion with glycopeptidase F (Fig. 2E, lanes 3 and 4). The molecular mass of the deglycosylated CDase of rat kidney is clearly different from that of mouse liver (Fig. 2E), suggesting that the difference in molecular mass on SDS-PAGE is not only due to the heterogeneity with N-glycans.

Purification of Membrane-bound CDase from Rat Kidney-As shown in
Characterization of Membrane-bound CDase from Rat Kidney-Characterization of the rat CDase was conducted using the enzyme preparation after HiLoad Superdex 200 pg chromatography. The enzyme exhibited a pH optimum at 6 -7, although the pH dependence of the enzyme was quite broad and about 50% activity was observed at pH 8 -9 (Fig. 3A), indicating that the enzyme should be classified as a neutral or neutral/alkaline CDase. The activity was completely inhibited by Hg 2ϩ , whereas Zn 2ϩ and Cu 2ϩ inhibited the activity by 80% (Fig. 3B). In contrast to the bacterial CDase (36), the rat CDase was not activated by Ca 2ϩ . EDTA, Mn 2ϩ , and Mg 2ϩ had little effect on the rat CDase. The enzyme activity was greatly enhanced by addition of detergents such as sodium cholate and sodium taurodeoxycholate. The optimum concentration of detergents differed markedly depending on the detergent used. For sodium taurodeoxycholate and sodium cholate, the optimum concentrations were found to be 0.1-0.2 and 0.4 -2%, respectively, which increased the enzyme activity about 4 -5fold in comparison with that in the absence of the detergent (Fig. 3C). Triton X-100 at 0.1-0.2% also enhanced the enzyme activity by about 2-fold, although the detergent showed an inhibitory effect beyond the optimum concentration. The substrate specificity of the CDase was examined at pH 7.0 using various 14 C-labeled Cers (Table II). Among various Cers tested, N-lauroylsphingosine (C12:0/d18:1) was most efficiently hydrolyzed by the enzyme followed by N-palmitoylsphingosine (C16: 0/d18:1) and N-stearoylsphingosine (C18:0/d18:1). Cers containing sphinganine (d18:0) and phytosphingosine (t18:0) as a long chain base were somewhat resistant to the enzyme. It is of note that glycosphingolipids such as GalCer, sulfatide, and GM1a or sphingomyelin were not hydrolyzed by the enzyme. NBD-N-dodecanoylsphingosine (C12-NBD-Cer) was hydrolyzed much faster than N-lauroylsphingosine (C12:0/d18:1), indicating that attachment of NBD to the fatty acid residue at the -position increased the susceptibility of the enzyme to the substrate.
Molecular Cloning, Sequencing, and Alignment of Rat Neutral CDase-The four peptide sequences (C1-4 in Fig. 4A) were determined using the purified 112-kDa protein by protein sequencer after digestion with lysylendopeptidase as described under "Experimental Procedures." We found that the four peptide sequences of the rat CDase were homologous to the se-

Neutral Ceramidase of Rat Kidney
quence of mouse neutral CDase (21); C1 for amino acid 223-245 in the mouse enzyme, C2 for amino acid 261-273, C3 for amino acid 601-619, and C4 for amino acid 701-729. Therefore, we designed two primers based on the nucleotide sequence of mouse CDase (sense primer corresponding to amino acid 223-230 of the mouse enzyme and antisense primer corresponding to amino acid 325-331) and performed PCR amplification using a rat kidney cDNA library. The 325-base pair amplified fragment was then used as a probe for colony hybridization to screen for the cDNA encoding the CDase from a rat kidney cDNA library. Finally, the clone (pAPkCD) containing the full-length cDNA encoding the CDase was obtained. Fig. 4A shows the cDNA and deduced amino acid sequences of the neutral CDase from rat kidney. The pAPkCD contained one open reading frame of 2283 base pairs coding 761 amino acids, 86 amino acid residues of which matched the amino acid sequence of the purified CDase (Fig. 4A, C1-4). The predicted molecular mass and pI of the enzyme were 83,483 and 6.55, respectively, judging from the deduced amino acid sequence. The open reading frame of pAPkCD contained nine potential N-glycosylation sites (Fig. 4A, underlines). This result is well consistent with the fact that the CDase is highly glycosylated with N-glycans  1 and 2) and mouse liver (3 and 4). The samples were treated with glycopeptidase F as described in manufacturer's instruction (lanes 2 and 4). Proteins were stained with silver staining solution.

FIG. 3. General properties of rat neutral CDase.
A, optimum pH of the rat neutral CDase. The enzyme activity was measured using 100 pmol of [ 14 C]Cer (C16:0/ d18:1) as a substrate in 20 l of 150 mM GTA buffer at different pH values containing 0.25% Triton X-100. The incubation was carried out using 5 microunits of the CDase at 37°C for 1 h. Effects of cations (B) and detergents (C) on CDase activity. The CDase activity was measured using C12-NBD-Cer as a substrate as shown under "Experimental Procedures" except that each reaction mixture contains 5 mM of the cation indicated (B) or various detergents at the concentrations indicated (C). In C, OE, sodium cholate; q, Triton X-100; E, taurodeoxycholate. (Fig. 2E). Computer analysis using a PSORT revealed the presence of one endoplasmic reticulum transitional signal sequence at amino acids 1-36, a signal peptidase cleavage site at amino acids 36 -37 (arrowhead), one possible transmembrane domain at amino acids 502-518 (box), and a di-Leu signal at amino acids 740 -741 (shading) (Fig. 4A). In addition, putative phosphorylation sites for casein kinase II (amino acids 10 -13, 259 -262, 261-264, 466 -469, 565-568, 566 -569, 611-614, and 757-760), and protein kinase C (amino acids 134 -136, 148 -150, 200 -202, 252-254, 261-263, 428 -430, 431-433, 466 -46, and 529 -531) were found in the sequence (Fig. 4A). Hydropathy analysis indicated the presence of two prominent hydro-  phobic segments, one in the amino-terminal region (amino acids 4 -31) predicting a putative signal sequence and the other in the middle of the sequence (amino acids 499 -518) predicting a possible transmembrane domain (Fig. 4B). Fig. 5 shows the alignment of the deduced amino acid sequence of rat neutral CDase with those of mouse (21), human (22), and bacterial (23) homologues. Human neutral CDase has a putative mitochondria-targeting sequence (Fig. 5, Box 2), whereas the rat and mouse enzymes do not. In contrast, rat and mouse CDases have a putative endoplasmic reticulum transitional signal sequence in the NH 2 -terminal region (Fig. 5, Box 1). Di-Leu signal, by which proteins would be sorted to lysosomes/late endosomes, was found in all CDases presented in this figure except bacteria (Fig. 5, Boxes 3 and 4). However, the di-Leu signal of human CDase does not seem to be functional because the human sequence lacks an acidic amino acid before di-Leu (Fig. 5, Box  3) which is indispensable for forming the di-Leu signal sequence (37), whereas the rat and mouse sequences contain a Glu before di-Leu (Fig. 5, Box 4).
Northern Blot Analysis of Rat Neutral CDase-To determine the size and expression of the rat neutral CDase in various tissues, Northern blot analysis was conducted using the EcoRI fragment of pAPkCD as a probe which contained almost the full-length of the CDase cDNA. As shown in Fig. 6, a strong 5.1-kb mRNA signal was detected in brain, kidney, and heart, whereas only weak signals were detected in other tissues including liver. In contrast, the mouse mRNA expression of neutral CDase was strong in liver and kidney, with a weak signal only observed in brain (21), indicating that the mRNA expression level differed somewhat among the tissues as well as animals tested.
Expression Analysis of Rat Neutral CDase-CHOP cells were transfected with pAPkCD and the CDase activity of cell lysates was measured using C12-NBD-Cer as a substrate at pH 7.5. As shown in Fig. 7A, the CDase activity of the lysate of pAPkCDtransfected cells (pAPkCD) increased more than 9,000-fold in comparison with that of mock transfectants (mock) or untransfected CHOP cells (data not shown). The optimum pH of the recombinant CDase was found to be pH 6 -7 (Fig. 7B), which is consistent with the result obtained using the purified enzyme from rat kidney (Fig. 3A). It is interesting to note that the recombinant rat CDase catalyzed the reversible reactions in which the amide linkage of ceramide is cleaved or synthesized (data not shown), as shown in neutral/alkaline CDases from mouse (25), yeast (24), and bacteria (36).
To verify whether or not N-glycosylation is essential for the expression of the CDase activity, a Myc-tagged CDase con- struct was expressed in CHOP cells in the presence or absence of tunicamycin which is a specific inhibitor for N-glycosylation. When the Myc-tagged CDase was expressed in HEK293 cells in the absence of tunicamycin, 1,690 microunits/mg of CDase was detected in cell lysates, concomitantly with the expression of two protein bands of molecular mass 133 and 113 kDa on SDS-PAGE after visualization with anti-Myc antibody (Fig.  7C). Interestingly, CDase activity markedly decreased when the transformation of HEK293 cells was conducted in the presence of tunicamycin (31.4 microunits/mg) (Fig. 7C). Western blotting using anti-Myc antibody revealed the generation of a single 87-kDa band instead of two glycosylated bands. Treatment of the recombinant CDase with endoglycosidase H converted the 113-kDa band to 87-kDa band on SDS-PAGE, whereas the 133-kDa band was little affected, indicating that a 113-kDa CDase has high-mannose type N-glycans and a 133-kDa enzyme mainly complex type N-glycans. As shown in Fig.  7D, the 113-kDa band decreased concomitantly with an increase of the 133-kDa band after treatment of the cells with cycloheximide, which inhibits de novo synthesis of proteins. These results suggest that the 113-kDa protein is an endoplasmic reticulum form CDase that matures into 133-kDa CDase during processing in the Golgi apparatus. However, two CDase bands were still observed after complete digestion with glycopeptidase F (Fig. 7C, lane 2), indicating that some modifications of the CDase, other than N-glycosylation, may occur during processing in the Golgi apparatus. In summary, the N-glycosylation process, in which the attachment and modification of N-glycans as well as some postglycosylational processing of CDase would be involved, is necessary for the expres-sion of the full activity of the CDase in HEK293 cells.
Subcellular Fractionation of Rat Neutral CDase by Percoll Density Gradient Centrifugation-To elucidate the subcellular distribution of neutral CDase in rat kidney and liver, both microsome fractions were fractionated by Percoll density gradient centrifugation according to the method described in Ref. 32. Almost all activity of the kidney neutral CDase was found in the fractions with alkaline phosphatase which is a marker enzyme for plasma membrane (Fig. 8A), whereas the liver neutral CDase was recovered in higher density fractions with ␤-galactosidase (Fig. 8B). Because ␤-galactosidase is a marker enzyme for lysosomes, the liver neutral CDase is most likely to distribute with late endosomes/lysosomes. Western blot analysis using anti-neutral CDase antibody confirmed the different subcellular distribution of the neutral CDase between kidney and liver (Fig. 8, A and B).
Distribution of the Neutral CDase in Rat Kidney and Liver-We raised anti-neutral CDase antibody (IgG) in rabbit using recombinant neutral CDase as antigen expressed in E. coli as described under "Experimental Procedures." Using this specific antibody and Cy3-labeled anti-rabbit IgG antibody as a second antibody, we examined the distribution of the CDase in the cortex (Fig. 9, A-C) and medulla (Fig. 9D) of rat kidney. Phalloidin-FITC (green) was used to visualize the actin filaments of brush borders in the proximal tubule cells (Fig. 9, B and C, arrowheads). Phalloidin-positive signals were also discernible at the bottom region of urinary tubules (Fig. 9, C and  D, double arrows). A strong signal for neutral CDase (red) was observed in the luminal surface in most urinary tubules, such as proximal and distal tubules (Fig. 9, A-C), and collecting The rat neutral CDase cDNA was subcloned into pcDNA3.1/Myc-His(ϩ) vector (Invitrogen Co.) and named pcDNAkCD. At 4 h after transfection with pcDNAkCD, the medium was changed to that containing tunicamycin (10 g/ml), and then cultured at 37°C for 12 h. Cells were harvested, lysed in standard assay solution, and the activity was determined using C12-NBD-Cer. The lysate was also subjected to the treatments with endoglycosidase H (Calbiochem) and glycopeptidase F as described in manufacturer's instructions. Aliquots of samples were subjected to SDS-PAGE, followed by staining with anti-Myc antibody. D, the effect of cycloheximide on 113-kDa (endoplasmic reticulum form) and 133-kDa (Golgi form) CDases. The medium was changed to that containing cycloheximide (50 g/ml) at 16 h after transfection of HEK293 with pcDNAkCD, and then cultured in the indicated times. Cells were harvested and the Myc-tagged CDases were detected by Western blotting with anti-Myc antibody. tubules (Fig. 9D). Positive signal for the CDase was, however, hardly detectable in cells of glomerulus (not shown). Counterstaining with FITC-phalloidin revealed that the CDase was localized on top of the microvilli in the proximal tubule cells (Fig. 9, B and C, arrows). In hepatocytes in liver, the CDase positive signal (red) appeared as many dot-like structures that distributed throughout the cytoplasm (Fig. 9E, center). To specify the organelles containing the CDase in hepatocytes, we performed double immunostaining using monoclonal antibody YA 30 which reacts with LGP85, a marker protein for lysosomes/late endosomes (Fig. 9E, left). The CDase signals (red) were found to be partially co-localized with the signal of LGP85 (green) (Fig. 9E, right, arrows). In summary, in rat kidney the neutral CDase seems to localize to apical membranes of urinary tubule such as proximal tubules, distal tubules, and collecting ducts, while the enzyme is distributed in endosome-like Tissue sections of A-D were stained with rabbit anti-CDase IgG followed by staining with goat anti-rabbit IgG-Cy3 (red). Specimens shown in B-D were counterstained with phalloidin-FITC (green) after immunostaining. Phalloidin-labeled actin filaments in microvilli of proximal tubule cells were indicated by arrowheads in B and C. In D, the CDase signal (left, red) was merged with phalloidin signal (green) in the right panel. Phalloidin signals were detectable at the bottom region of urinary tubules (C and D, double arrows). E, tissue sections of hepatocytes of rat liver. They were stained with monoclonal mouse anti-LGP85 IgG (left) and rabbit anti-CDase IgG (center). Confocal images produced by superposition of left (green) and center (red) are shown in the right panel. Arrows in the right panel indicate co-localization of the two antigens. N, nuclei. Details are described under "Experimental Procedures." organelles of hepatocytes in rat liver.
Neutral CDase in Lipid Microdomain Raft of Rat Kidney-From the brush border of Madin-Darby canine kidney cells, a cholesterol-enriched lipid microdomain raft was isolated as a non-ionic detergent-insoluble fraction using Lubrol 17A17 (34). In this study, we thus prepared a Lubrol-insoluble lipid microdomain from rat kidney and examined whether or not the neutral CDase is associated with the lipid microdomain. As shown in Fig. 10A, neutral CDase activity was found in fractions 6 -8, in which free cholesterol and GM1 ganglioside were abundant, indicating that these fractions contain the lipid microdomain raft. Western blotting also confirmed that the CDase was concentrated in fractions 6 -8 (B), whereas CD71 (transferrin receptor, a marker for the non-raft membrane fraction) was solely detected in the high density fraction, number 10 (C). In conclusion, the neutral CDase of rat kidney is likely to associate with a lipid microdomain raft in which cholesterol and glycosphingolipid GM1 are enriched. DISCUSSION Purification, Characterization, and cDNA Cloning of the Neutral CDase from Rat Kidney-Previously, we purified the neutral CDase from mouse liver, which was solubilized from the membrane fraction by freeze-thawing (25). However, the enzyme in mouse kidney was not solubilized by freeze-thawing and therefore had not been purified. In the present study, we succeeded in purifying the membrane-bound neutral CDase from the microsome fractions of rat kidney and cloned the cDNA encoding the enzyme. The rat kidney enzyme was classified as a neutral or neutral/alkaline CDase, based on its optimum pH. The enzymatic properties of the rat kidney CDase are somewhat different from those of mouse liver (25) and rat brain (38) CDases in the cation requirement, substrate specificity, and molecular weight. While the rat brain enzyme was activated by Mn 2ϩ (38), the kidney enzyme was not (Fig. 3B). The mouse liver enzyme hydrolyzed N-palmitoylsphingosine (C16:0/d18:1) most efficiently, the rat kidney enzyme, N-lauroylsphingosine (C12:0/d18:1) (Table II). The molecular mass of the rat kidney CDase was estimated to be 112 kDa on SDS-PAGE, which is clearly different from the neutral/alkaline CDases isolated from other origins; 94 kDa for the enzyme from mouse liver (25), 95 kDa from rat brain (38), 60 kDa from guinea pig skin (39), and 70 kDa from Pseudomonas aeruginosa (36). The molecular mass of acid CDase from human urine was reported to be 50 kDa (13).
However, the deduced amino acid sequence from the cDNA encoding the CDase of rat kidney is homologous to that of mouse liver (21) and human brain (22): 92 and 76% identity were found, respectively. The molecular mass estimated from the deduced amino acids was also similar: 83,483 for the rat enzyme (this study), 83,504 for the mouse enzyme (21), and 83,193 for the human enzyme (22). We speculated that this contradiction may stem from cell/tissue/organ-specific posttranslational modification of the enzyme including N-glycosylation. It has been reported that the lysosomal proteins were occasionally truncated in the COOH-terminal region (40). Since the anti-Myc antibody reacts with the Myc tag of the CDase at the COOH terminus, we considered that the visualized CDase was not truncated in the COOH-terminal region (Fig. 7C). The possible modification of the enzyme, such as phosphorylation, acylation, and sulfation, should be further investigated and it is of interest that several phosphorylation sites are found in the deduced amino acid sequence of the rat neutral CDase.
To investigate the presence of the human-type CDase homologues in tissues other than human, we performed 5Ј rapid amplification of cDNA ends-PCR with a primer designed using amino acid sequence 83-89 from the initiation Met of mouse CDase (21) against cDNA libraries prepared from mRNAs of mouse brain, liver, kidney, and spleen. Four splicing variants were found in mouse brains, 3 variants in liver, 1 variant in kidney, and 2 variants in spleen in which the 5Ј noncoding regions were different from each other, although 88 amino acids from the initiation Met in open reading frames were exactly the same. These results strongly suggest that the human-type CDase homologues are not present or not as a major homologue in mice.
Subcellular Localization of Neutral CDase in Rat Tissue-The liver neutral CDase was efficiently extracted with freezethawing whereas the kidney enzyme was not. This paper clearly indicated the reason why, i.e. in hepatocytes the neutral CDase was localized in late endosomes/lysosomes, whereas the enzyme was associated with a lipid microdomain raft on the apical membrane of urinary tubule cells in the kidney. It has been already reported that some lysosomal enzymes could be released from membrane fractions by freeze-thawing (41).
Why does the enzyme distribute in tissue-specific manner? One possible explanation is the tissue-specific expression of a receptor for di-Leu signal, which is a sorting signal for vesicular transport from plasma membrane to endosomes/lyosomes (37) or for targeting to basolateral membranes (42), because a functional di-Leu motif was found in the putative amino acid sequences in CDases of rat and mouse, but not human. Di-Leu receptors are a component of adaptor proteins (AP) associated with clathrin-coated vesicles and the organ-specific expression of AP has been reported (43). It is likely that di-Leu receptors would be expressed in rat heptocytes but not in rat urinary tubule cells. Thus, in rat hepatocytes the CDase would be sorted from plasma membranes to late endosomes/lysosomes by vesicle transport using the di-Leu motif, whereas in rat urinary tubule cells the CDase would be retained at the apical sites of the plasma membrane possibly due to the lack of a receptor for the di-Leu signal. Tyr signal is also thought to be a conventional sorting signal (44), but is not present in the putative amino acid sequence of the rat and mouse neutral CDases.
Recently, Bawab et al. (22) reported the presence of a mitochondria-targeting signal in the deduced amino acid sequence of human neutral CDase (Fig. 5) and showed that the overexpressing green fluorescent protein-tagged CDase was exclusively localized to mitochondria in HEK293 and MCF7 cells. However, the rat CDase seems not to be present in mitochondria, since the deduced sequence lacks the mitochondria-targeting signal and the enzyme is highly glycosylated with Nglycans that are not usually present in mitochondrial enzymes (45). It was also revealed in this study that tunicamycin treatment inhibited the generation of matured CDase with full activity, indicating that the N-glycosylation process is indispensable for the expression of the enzyme activity.
The localization of neutral CDase of rat liver is similar, but not identical, to that of LGP85(Limp-2) which is a marker protein for late endosomes/lysosomes. The liver CDase could be transiently localized in intermediate vesicles such as secreted vesicles, endocytotic vesicles, and early endosomes during vascular transport. It should be noted that the neutral/alkaline as well as acid CDases are actively released by murine endothelial cells (46).
It is believed that a genetic deficiency of acid CDase could cause Farber disease, since an acid CDase is localized in lysosomes where Cer is accumulated in those with this disease. However, the present study showed that the subcellular localization of neutral/alkaline CDase depends on the cell/tissue/ organ and in hepatocytes, the enzyme actually localizes to late endosomes/lysosomes. Thus, the roles of the neutral/alkaline CDases in the catabolism of Cer in late endosomes/lysosomes and possible participation in Farber disease should be clarified.