Cloning and Characterization of a Lysosomal Phospholipase A2, 1-O-Acylceramide Synthase*

Recently, a novel enzyme, 1-O-acylceramide synthase (ACS), was purified and characterized from bovine brain. This enzyme has both calcium-independent phospholipase A2 and transacylase activities. The discovery of this enzyme led us to propose a new pathway for ceramide metabolism in which the sn-2-acyl group of either phosphatidylethanolamine or phosphatidylcholine is transferred to the 1-hydroxyl group of ceramide. In this study, the partial amino acid sequences from the purified enzyme revealed that the enzyme contains amino acid sequences identical to those of human lecithin:cholesterol acyltransferase-like lysophospholipase (LLPL). The coding sequences of the mouse, bovine, and human genes were obtained from the respective kidney cDNAs by PCR. The open reading frames of LLPL were cloned into pcDNA3 to generate carboxyl-terminally tagged proteins. The expression of mouse LLPL in COS-7 cells demonstrated that transfected cells had higher transacylase and phospholipase A2 activities than did non-transfected cells. Immunoprecipitation confirmed that LLPL had ACS activity. There were no significant lecithin:cholesterol acyltransferase and lysophospholipase activities in the mouse LLPL-transfected cells under either acidic or neutral conditions. Amino acid sequences from cDNAs of mouse, human, and bovine LLPLs demonstrated a signal peptide cleavage site, one lipase motif (AXSXG), and several N-linked glycosylation sites in each LLPL molecule. The replacement of serine with alanine in the lipase motif of mouse LLPL resulted in elimination of enzyme activity, indicating that the serine residue is part of the catalytic site. Deglycosylation of mouse, human, and bovine LLPLs yielded core proteins with a molecular mass of 42 kDa without change in enzyme activities. LLPL was post-translationally modified by signal peptide cleavage and N-linked glycosylation, and each mature LLPL had the same size core protein. Subcellular fractionation demonstrated that ACS activity co-localized withN-acetylglucosaminidase. Therefore, LLPL encodes a novel lysosomal enzyme, ACS.

For the last decade, ceramide has been thought to play an important role in cell signal transduction involving cell growth, proliferation, differentiation, stress responses, and apoptosis (1). The ceramide levels within cells are regulated by several well defined metabolic pathways. We recently studied the metabolism of N-acetylsphingosine (NAS) 1 in Madin-Darby canine kidney (MDCK) cells (2). In that study, NAS was actively metabolized and was not an inert compound, as had been previously suggested (3). NAS was converted to other sphingolipids, including sphingosine, C 2 -sphingomyelin, C 2 -glucosylceramide, long-chain ceramide, long-chain sphingomyelin, and long-chain glucosylceramide. An unexpected product was also detected. This metabolite was a highly nonpolar compound and identified as 1-O-acyl-NAS.
This discovery led to the discovery of a new enzyme activity, one that catalyzes the esterification of the hydroxyl group at C-1 in the ceramide molecule under acidic conditions. The enzyme does not require divalent cations for its activity. Glycerophospholipids (in particular, phosphatidylethanolamine (PE) and phosphatidylcholine (PC)) were identified as acyl group donors in the reaction. The acyl group at the sn-2-position in the phospholipid is transferred to an acceptor molecule, e.g. ceramide or water. If the acceptor is ceramide, 1-O-acylceramide is formed. However, if the acceptor is water, free fatty acid is released. It was also observed that a short-chain rather than a long-chain ceramide is preferred as an acceptor. These observations raised the possibility that this new enzyme regulates a novel pathway of ceramide metabolism.
The new enzyme, named 1-O-acylceramide synthase (ACS), was purified from bovine brain and further characterized (4). ACS is a water-soluble glycoprotein with a molecular mass of 45 kDa and a single polypeptide chain, which specifically binds to concanavalin A-conjugated agarose. The enzyme has a pH optimum at 4.5 and has both phospholipase A 2 and transacylase activities. The enzyme activity is calcium-independent. Therefore, ACS may be classified as a calcium-independent phospholipase A 2 .
In this study, we report that a BLAST search revealed that the peptides obtained from ACS purified from bovine brain share amino acid sequences with a recently reported human gene termed lecithin:cholesterol acyltransferase (LCAT)-like lysophospholipase (LLPL) (5). To understand the biological function of ACS, the ACS gene was sequenced on the basis of its similarity to LLPL, and the gene products were characterized.

Reagents-
The reagents and sources were as follows: hemagglutinin (HA) and c-Myc peptides and mouse anti-HA monoclonal antibody * This work was supported by National Institutes of Health Grant RO1 DK55823 (to J. A. S.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AY072914.
Determination of the Amino-terminal Amino Acid Sequence and Partial Amino Acid Sequences of Tryptic Fragments of Bovine Brain ACS-ACS was purified from bovine brain as previously described (4). The protein content was determined by the bicinchoninic acid protein assay (Pierce) with bovine serum albumin as a standard. SDS-PAGE was performed following the method of Laemmli and Favre (6).
For determination of the amino-terminal amino acid sequence of ACS, the purified protein was separated using a 10% acrylamide gel and transferred to a PVDF membrane using transfer buffer (10 mM CAPS (pH 11) in 10% methanol) at a constant voltage of 50 V for 60 min at room temperature. The protein on the membrane was briefly stained with 0.1% Coomassie Brilliant Blue R-250 in 40% methanol and 1% acetic acid and destained with 50% methanol. The membrane was extensively rinsed with deionized water, and the protein band was excised. The excised PVDF sample was air-dried and analyzed in the Howard Hughes Medical Institute Biopolymer/W. M. Keck Foundation Biotechnology Research Laboratory at Yale University.
For determination of partial amino acid sequences of tryptic fragments of ACS, the protein band in the gel was stained with 0.1% Coomassie Brilliant Blue R-250 in ethanol/acetic acid/water (9:2:9), extensively destained with ethanol/acetic acid/water (25:8:65), and excised. The gel slice was washed twice with 50% acetonitrile. Tryptic digestion of the band and sequence analysis were carried out at the Harvard Microchemistry Facility by microcapillary reverse-phase high performance liquid chromatography nanoelectrospray tandem mass spectrometry on a Finnigan LCQ DECA quadrupole ion trap mass spectrometer.

Construction of LLPL Expression Plasmids-
The entire open reading frames of the individual LLPLs were excised at HindIII and XhoI sites from the plasmids described above. They then were subcloned into the HindIII and XhoI sites of pcDNA3-FLAG, pcDNA3-HA, or pcDNA3-c-Myc (all three generously provided by Dr. Naohiro Inohara) (7) to generate carboxyl-terminally tagged LLPL proteins. The mutation of serine to alanine in the putative lipase motif sequence of mouse LLPL was generated by the overlap extension method using primers Mt1-F (5Ј-GCCCACGCTATGGGCAAC-3Ј) and Mt1-R (5Ј-TTGCCCAT-AGCGTGGGCGACCA-3Ј) (8).
Cell Culture and Transfection-COS-7 cells were grown in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% fetal bovine serum. For transient expression, COS-7 cells were cultured in 35-mm dishes. When the cells reached 80% confluency, they were transfected with 1 g/ml purified plasmid using LipofectAMINE Plus TM (Invitrogen) in 1 ml of Opti-MEM medium (Invitrogen). 1 ml of Dulbecco's modified Eagle's medium containing 20% fetal bovine serum was added after 3 h of incubation at 37°C and 5% CO 2 . 24 h after transfection, the cells were washed three times with 2 ml of phosphate-buffered saline, incubated with 2 ml of 1 mM EDTA in phosphate-buffered saline for 20 min at 37°C and transferred into an ultracentrifuge tube. The following procedures were carried out at 4°C. The cells were collected by centrifuge at 800 ϫ g for 10 min. The pellet of the cells was dispersed into 0.5 ml of 0.25 M sucrose and 10 mM HEPES (pH 7.4) by sonication. Sonication was carried out for 4 ϫ 10 s at 0°C in a probe-type sonicator. The suspension was centrifuged for 1 h at 100,000 ϫ g. The resultant supernatant was collected as a soluble fraction and used in protein analysis and enzyme assay.
Enzyme Assay (Transacylase Activity)-The assay conditions are described in the figure legends. In general, liposomes consisting of dioleoylphosphatidylcholine (60.5 mol %), PE (27.3 mol %) and dicetyl phosphate (12.3 mol %) were used as the acyl group donor. Constituent lipids in chloroform were mixed and dried under a stream of nitrogen. 50 mM sodium citrate (pH 4.5) was added to the dried lipids at a volume of 1 ml/128 nmol of lipid phosphorus. The lipids were dispersed into the buffer for 8 min in an ice-water bath using a probe sonicator.
The donor liposomes (64 nmol of phospholipid) were incubated with 10 nmol of NAS or 5 nmol of [ 3 H]NAS (10,000 cpm), 5 g of bovine serum albumin, and sample preparation at 37°C in a total volume of 500 l of 40 mM sodium citrate (pH 4.5). The reaction was terminated by adding 3 ml of chloroform/methanol (2:1) plus 0.3 ml of 0.9% (w/v) NaCl. After centrifugation for 5 min at 800 ϫ g, the lower layer was transferred into another glass tube and dried down under a stream of nitrogen gas. The lipid extract was applied to an HPTLC plate and developed in a solvent system consisting of chloroform/acetic acid (9:1). The location of 1-O-acyl-NAS was confirmed using 1-O-palmitoyl-NAS synthesized in our laboratory.
In the assay using nonradioactive NAS, the plate was dried, sprayed with 8% (w/v) CuSO 4 pentahydrate in water/methanol/concentrated H 3 PO 4 (60:32:8), and charred for 15 min at 150°C. An image of the plate was taken by a scanner (UMAX Astra Scanner 2200) connected to a computer and scanned by the NIH Image program (Version 1.62) to estimate the density of each band. Known amounts of ceramide were used to obtain a standard curve. In the assay using radioactive NAS, 1-O-acyl-NAS was detected under a UV light with primulin spray, scraped, and counted.
Immunoprecipitation-30 g of the soluble cellular fraction obtained from carboxyl-terminally c-Myc-tagged mouse LLPL-transfected cells was incubated with anti-c-Myc monoclonal antibody in the absence or presence of 300 g of c-Myc or HA peptide in 50 mM sodium phosphate (pH 7.4) in a total volume of 440 l for 2 h at 0°C. The reaction mixture was mixed with 60 l of protein G-agarose beads (25% suspension in 100 mM sodium phosphate (pH 7.4)) and incubated for 1 h at 4°C using a rotator. The suspension was centrifuged at 14,000 ϫ g for 30 s at 4°C. The agarose beads were washed four times with 600 l of cold 100 mM sodium phosphate (pH 7.4). The resultant beads were incubated with 450 l of assay solution (39 mM sodium citrate (pH 4.5), 11 g/ml bovine serum albumin, and liposomes containing 22 M NAS) for 1 h at room temperature using a rotator. The reaction tube was put on ice at the end of the reaction and centrifuged at 14,000 ϫ g for 30 s at 4°C. The supernatant (450 l) was mixed with 3 ml of chloroform/methanol (2:1) plus 0.35 ml of 0.9% NaCl and centrifuged for 5 min at 800 ϫ g. The resultant lower layer was analyzed as described above.
Digestion of Recombinant LLPLs with Endoglycosidase F 1 -20 g of the soluble cellular fraction obtained from the carboxyl-terminally HAtagged LLPL-transfected cells was incubated with or without 35 microunits of endoglycosidase F 1 in 32 mM sodium citrate (pH 4.5) in a total volume of 92 l for 14 h at 4°C. At the end of the reaction, the mixture was precipitated by the method of Bensadoun and Weinstein (9). The resultant pellet was dissolved in 25 l of loading buffer plus 2 l of 2 M Tris for SDS-PAGE. Proteins were separated using a 10% acrylamide gel and transferred to a PVDF membrane using transfer buffer (20 mM Tris and 150 mM glycine in 20% methanol) at a constant voltage of 100 V for 2 h at 4°C. The membrane was blocked with Tween/Tris-buffered saline consisting of 5% (w/v) skim milk in 150 mM NaCl and 20 mM Tris-HCl (pH 8.0) containing 0.05% (w/v) Tween 20. The membrane was incubated with anti-c-Myc monoclonal antibody in 5% skim milk for 1 h at room temperature in a moist chamber. The membrane was briefly rinsed twice with Tween/Tris-buffered saline, washed three times for 5 min with Tween/Tris-buffered saline, and incubated with horseradish peroxidase-conjugated goat anti-mouse IgG-antibody in 5% skim milk for 1 h at room temperature in a moist chamber. The antigen-antibody complex was visualized with diaminobenzidine and hydrogen peroxide.
Preparation of Lysosomes-The lysosome preparation was performed using the method of Rohrer et al. (10) with slight modifications. Confluent MDCK cells in a 15-cm dish were cultured with 21 ml of Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum and then treated for 15 h with 100 M leupeptin and 100 M pepstatin A. After treatment, the cells were washed twice with 21 ml of cold phosphate-buffered saline, scraped with 3 ml of cold 0.25 M sucrose and 1 mM EDTA (pH 7.4) (buffer A), and transferred into a 15-ml plastic tube. The following procedures were subsequently carried out at 4°C. Another 3 ml of buffer A was added to recover the remaining cells, and the cells were collected by centrifugation at 200 ϫ g for 10 min and suspended with 2.5 ml of buffer A. The cell suspension was homogenized with 20 strokes of a ground-glass pestle homogenizer. The homogenate was diluted 2-fold with buffer A and centrifuged for 10 min at 400 ϫ g. The post-nuclear supernatant was centrifuged at 12,000 ϫ g for 20 min. The pellet obtained from the post-nuclear supernatant was resuspended in 1 ml of buffer A and then diluted with 1 ml of buffer A. A total of 9.4 ml of 18% (v/v) Percoll, 0.25 M sucrose, and 1 mM EDTA (pH 7.4) was overlaid on 1 ml of 2.1 M sucrose and 1 mM EDTA (pH 7.4) on the bottom of the centrifuge tube (14 ϫ 89 mm). The suspension was overlaid on the Percoll layer and centrifuged for 30 min at 47,500 ϫ g. One faint band was observed between the sucrose cushion and the Percoll layer. A second band that was white and dense was observed between the Percoll and top layers. A total of 1.1 ml of fraction was collected from the bottom using a P-1 pump (Amersham Biosciences, Inc.) and a fraction collector.
Transacylase activity was measured as described above. ␤-Hexosaminidase activity was determined using p-nitrophenyl-N-acetyl-␤glucosamine as described by Rohrer et al. (10). Cytochrome c oxidase activity was determined by the method of Walton and Tzagoloff (11).

RESULTS
cDNA Cloning of the LLPL Gene-ACS was purified from bovine brain, and the tryptic fragments were sequenced by mass spectrometry. A BLAST search revealed that the partial amino acid sequences of ACS are highly homologous to the amino acid sequence deduced from cDNA encoding human LLPL ( Fig. 1) (5). Therefore, the gene product expressed from the LLPL gene was presumed to be identical to ACS.
A T BLAST N data base search of the GenBank TM /EBI Data Bank was performed to find ESTs that could align with the human LLPL sequence (accession number AB017494). Mouse EST clones (accession numbers AI385866 and BE912746 and accession numbers BI079146 and BE285166) overlapped with the 5Ј-and 3Ј-ends, respectively, of the human LLPL coding region (Fig. 2). Bovine EST clones (accession numbers AV597297 and BE753473 and accession number AW326965) overlapped with the 5Ј-and 3Ј-ends, respectively, of human LLPL (Fig. 2). As a next step, primers were designed based on these EST data to obtain the entire coding sequence. This sequence has been submitted to the GenBank TM /EBI Data Bank under accession number AY072914.
According to a previous report, LLPL mRNA expression is higher in kidney (5). Therefore, kidney cDNA was used for further investigation. PCR products of mouse and bovine kid-ney cDNAs with these primers gave rise to single bands that were 1.3 kb in length. The PCR products were subcloned into pCR4-TOPO and sequenced in both directions. The sequence obtained from mouse is identical to the gene reported in the DDBJ Data Bank (accession number E26773).
Predicted Protein Structures of LLPL-The deduced amino acid sequences of human, mouse, and bovine LLPLs were compared (Fig. 1). The human and mouse LLPL cDNAs encode 412-amino acid polypeptides with an acidic isoelectric point. On the other hand, the bovine LLPL cDNA encodes a 407-amino acid polypeptide with a basic isoelectric point. A rule for signal peptide sequence suggests that human and mouse LLPLs have a signal sequence cleavage site between proline 33 and alanine 34. The corresponding cleavage site for bovine LLPL is between proline 28 and alanine 29 (12). Each mature LLPL contains 379 amino acids, corresponding to a molecular mass of 43 kDa. As previously reported, alanine 34 of precursor human LLPL is the amino-terminal amino acid of human recombinant LLPL (5). Amino-terminal sequencing of the purified bovine protein yielded a sequence of GSRPPVVLVPGDM, indicating that the signal peptide cleavage site of bovine LLPL may lie between alanine 29 and glycine 30. In addition, each deduced LLPL sequence contains a putative lipase motif of AXSXG, which is also found in Bacillus lipase (5,13), and N-linked glycosylation sites, three in bovine and four in human and mouse.
Expression of Mouse LLPL in COS-7 Cells-The mouse LLPL gene was used in the first expression study of LLPL. The entire open reading frame of LLPL was cloned into the HindIII and XhoI sites of pcDNA3 to generate carboxyl-terminally tagged proteins with FLAG, HA, or c-Myc peptides. Each vector was transfected into COS-7 cells with LipofectAMINE. ACS activity in the soluble fraction of the cell homogenate was observed using liposomes consisting of PC, PE, dicetyl phosphate, and NAS under acidic conditions.
The soluble fraction prepared from mouse LLPL-transfected cells catalyzed marked formation of 1-O-acyl-NAS and release of fatty acid as a function of incubation time (Fig. 3). Also, a similar enzyme activity was observed in each soluble fraction from cells transfected with carboxyl-terminally FLAG-, HA-, and c-Myc-tagged LLPLs. The soluble fraction from each mouse LLPL-transfected cell preparation had ϳ30-fold increased transacylase activity compared with that from non-transfected cells (Fig. 3). The carboxyl-terminal tags did not affect LLPL activity or expression. These results indicate that recombinant LLPL has the same transacylase and phospholipase A 2 activities as ACS.
Confirmation of ACS Activity of LLPL Expressed in LLPLtransfected Cells-Immunoprecipitations were employed to demonstrate that recombinant LLPL itself has ACS activity. The soluble fraction of c-Myc-tagged mouse LLPL-transfected COS-7 cells was incubated with anti-c-Myc monoclonal antibody and protein G-agarose beads. In this assay, the resultant protein G-agarose beads were incubated with liposomes containing NAS.
When the soluble fraction was incubated with anti-c-Myc antibody, formation of 1-O-acyl-NAS and release of fatty acid were observed with protein G-agarose beads (Fig. 4, lane 3). However, when the soluble fraction was incubated with anti-c-Myc antibody in the presence of an excess of c-Myc peptide, no significant products were found with protein G-agarose beads (Fig. 4, lane 4). On the other hand, when the soluble fraction was incubated with anti-c-Myc antibody in the presence of an excess of HA peptide, formation of 1-O-acyl-NAS and release of fatty acid were observed with protein G-agarose beads (Fig. 4,  lane 5). These results demonstrate that the recombinant protein expressed in cells transfected with c-Myc-tagged mouse LLPL specifically binds to anti-c-Myc monoclonal antibody and shows dual transacylase and phospholipase A 2 activities at pH 4.5. This experiment confirmed that LLPL expressed in cells transfected with cDNA encoding LLPL has ACS activity. Therefore, ACS is a gene product of the LLPL gene.
LCAT and Lysophospholipase Activities of Mouse LLPL- FIG. 4. LLPL has both transacylase and phospholipase activities. The soluble fraction obtained from carboxyl-terminally c-Myctagged mouse LLPL (mLLPL)-transfected cells was incubated with anti-c-Myc monoclonal antibody in the absence or presence of c-Myc peptide or HA peptide for 2 h at 0°C. The reaction mixture was mixed with protein G-agarose beads and incubated for 1 h at 4°C. The agarose beads were collected and washed four times with 100 mM sodium phosphate (pH 7.4). The resultant beads were incubated with liposomes containing NAS for 1 h at room temperature using a rotator. The reaction products were analyzed as described under "Materials and Methods."

FIG. 2. Mapping of mouse and bovine ESTs overlapping with the human LLPL sequence.
Amino acid numbering starting from the most 5Ј-potential translation initiation site is indicated at the top, with a schematic representation of LLPL below. ESTs were used to map the full-length coding sequences of mouse and bovine LLPLs. The accession numbers for each EST are indicated above each solid line. The arrows indicate the position and orientation of PCR primers used to amplify LLPL coding sequences from human, mouse, and bovine kidney cDNAs. According to Taniyama et al. (5), the deduced amino acid sequence of human LLPL is 49% homologous to human LCAT, although human LLPL did not display LCAT activity under their neutral assay conditions. We determined whether mouse recombinant LLPL has LCAT activity. As expected from the previous results (Fig. 4), a marked phospholipase A 2 activity was observed in the soluble fraction from LLPL-transfected cells under acidic conditions (Fig. 5). However, no significant esterification of cholesterol was observed in the same soluble fractions under either acidic or neutral conditions (Fig. 5).
Taniyama et al. (5) also reported that human recombinant LLPL has lysophospholipase activity. We investigated lysophospholipase activity in the soluble fraction of mouse LLPLtransfected cells. Interestingly, the soluble fraction did not show any significant increase in lysophospholipase activity under either acidic or neutral conditions (Fig. 6). In this study, we used the same assay conditions of lyso-PC (40 M), buffer, and ion strength under neutral conditions as reported by Taniyama et al. (5). In addition, no lysophospholipase activity was found in the medium cultured with LLPL-transfected cells, in contrast with the results of Taniyama et al. (5).
The lyso-PC concentration (40 M) used was lower than the critical micellar concentration of lyso-PC (ϳ100 M). In general, lipase activity is markedly enhanced as the phospholipid substrate concentration reaches the critical micellar concentra-tion. Therefore, 400 M lyso-PC was employed to determine whether recombinant LLPL or the cultured medium in our study has lysophospholipase activity. No significant increase in lysophospholipase activity was observed in the cultured medium of the LLPL-transfected cells under either acidic or neutral conditions. We found a very slight but significant increase in lysophospholipase activity in the soluble fraction of the LLPL-transfected cells under acidic conditions (data not shown). However, the specific activity of ACS in the same soluble fraction was Ͼ100 times higher than that of a lysophospholipase.
Point Mutation of a Putative Lipase Motif in LLPL-LLPL has a putative lipase motif, AXSXG (Fig. 1). A serine-to-alanine substitution in the AXSXG sequence of LLPL was generated by the overlap extension method using PCR (8). The soluble fraction of the mutated LLPL-transfected cells did not have any significant enzyme activity (Fig. 7A). Western blot analysis showed that the protein expression of mutated LLPL in COS-7 cells was comparable to that of LLPL (Fig. 7B). These results strongly support the conclusion that the serine residue in the lipase motif is essential for enzyme activity.
Post-translational Modification of LLPL-There is a disparity between the apparent molecular mass reported for human LLPL (57 kDa) expressed in COS-7 cells and that observed for purified bovine ACS (45 kDa) upon SDS-PAGE. To resolve this discrepancy, COS-7 cells were transfected with HA-tagged mouse, human, and bovine recombinant LLPL DNAs, and the expressed LLPLs were studied.
The soluble fraction of the transfected cells had comparable levels of ACS activity (Fig. 7A). The protein expression of mouse, human, and bovine LLPLs was also comparable as confirmed by Western blotting with anti-HA monoclonal antibody (Fig. 8B). HA-tagged mouse, human, and bovine LLPLs expressed in COS-7 cells have molecular masses of 51, 50, and 47 kDa, respectively (Fig. 8B). According to their deduced amino acid sequences, both mouse and human LLPLs have four putative N-linked glycosylation sites. Bovine LLPL has three putative N-linked glycosylation sites. Therefore, the differences in their molecular masses were thought to be due to differences in glycosylation in LLPLs. Bovine brain ACS binds to concanavalin A-agarose beads and is specifically released with methyl-␣-D-mannopyranoside from the beads (4). The treatment of each soluble fraction with endoglycosidase F 1 , an enzyme that cleaves asparagine-linked oligomannose and hybrid oligosaccharides, resulted in a decrease in molecular mass for each LLPL. Following endoglycosidase F 1 treatment, bands were detected at a molecular mass of 42 kDa (Fig. 8B). This value is smaller than that predicted from the deduced entire amino acid sequence for each LLPL. These results suggest that the deduced entire amino acid sequence for each species consists of a precursor protein of LLPL with a signal peptide sequence. In addition, each mature protein of LLPL is a high mannose-type glycoprotein with the same size protein core. In addition, deglycosylation did not have any significant effect on enzyme activity (data not shown). The N-linked oligosaccharides in LLPL may be involved in the protein sorting.
Localization of ACS in Cells-Previous studies showed that ACS has an acidic pH optimum and N-linked oligomannose (2,4). These properties indicate that ACS is a probably a lysosomal enzyme. A preliminary study showed that ACS activity is relatively high in MDCK cells compared with other animal tissues (data not shown). Therefore, MDCK cells were used to determine the subcellular localization of ACS using the method of Percoll gradient fractionation (10). 11 and 89% of the ␤-hexosaminidase activity in the postnuclear supernatant was recovered in the post-mitochondrial supernatant and crude mitochondrial fractions, respectively, after centrifugation at 12,000 ϫ g. Also, 17 and 83% of the ACS activity in the post-nuclear supernatant was recovered in the post-mitochondrial supernatant and crude mitochondrial fractions, respectively. Cytochrome c oxidase activity was not detected in the post-mitochondrial supernatant. The crude mitochondrial fraction was applied to Percoll gradient fractionation (Fig. 9). 75 and 14% of the ␤-hexosaminidase activity was recovered in fractions 1-3 and 9 -11, respectively, indicating that fractions 1-3 represent the lysosome-enriched fraction. 54 and 29% of the ACS activity was recovered in fractions 1-3 and 9 -11, respectively.
The difference in the observed ␤-hexosaminidase and ACS activities is likely due to the inhibition of ACS by Percoll. 40% of the transacylase activity in the crude mitochondrial fraction was inhibited by 0.72% Percoll. In addition, the lipid extract obtained from the reaction mixture containing fraction 1, 2, or 3, but not fraction 10 or 11, yielded a large Percoll spot at the TLC plate origin after development in a solvent system consisting of chloroform/acetic acid (9:1). These findings indicate that ACS activity in fractions 1-3 is likely lowered by the presence of Percoll, and the actual profile of ACS activity is similar to that of ␤-hexosaminidase activity. Furthermore, the cytochrome c oxidase activity in fraction 10 was ϳ4-fold higher than that in fraction 1, indicating that most of the mitochondria were recovered in fractions 9 -11. These results support the conclusion that ACS is a lysosomal enzyme.

DISCUSSION
This cloning study revealed that the previously reported LLPL gene encodes ACS. The gene product expressed in LLPLtransfected cells has both transacylase and phospholipase A 2 activities under acidic conditions, but lacks significant LCAT and lysophospholipase activities under either acidic or neutral conditions.
Previously, Taniyama et al. (5) reported that human LLPL expressed in COS-7 cells is a secreted protein that has lysophospholipase activity under neutral conditions. In this study, we failed to observe any significant increase in lysophospholipase activity in the cultured medium of COS-7 cells transfected with recombinant LLPL DNA. This was true even when lysophospholipase activity was assayed at concentrations exceeding the critical micellar concentration of lyso-PC. A very slight but significant increase in lysophospholipase activity in the soluble fraction of the LLPL-transfected cells was observed when assayed at lyso-PC concentrations greater than the critical micellar concentration. The specific activity of lysophospholipase was significantly lower than that of ACS. Purified ACS from bovine brain has weak enzyme activity as a phospholipase A 1 (4). Therefore, the lysophospholipase activity ob-served in recombinant LLPL probably reflects the minor phospholipase A 1 activity of ACS. We conclude that the LLPL gene product mainly functions as ACS or as a phospholipase A 2 , but not as a lysophospholipase.
Human recombinant LLPL is inactivated with diisopropyl fluorophosphate, indicating that the active site contains a serine residue (5). In our study, the replacement of serine with alanine within the lipase motif (AXSXG) of recombinant LLPL resulted in a loss of the phospholipase A 2 and transacylase activities. Therefore, the serine residue in the lipase motif must be essential for enzyme activity. As reported for human LLPL (5,14), the catalytic triad of serine 181, aspartic acid 345, and histidine 377 of LCAT is also conserved in mouse and bovine LLPLs (Fig. 1). This observation is consistent with the view that the serine residue in the motif is an active site and that the enzyme forms an acyl-enzyme intermediate via the hydroxyl group of the serine (Fig. 10).
The deduced amino acid sequences of mouse, human, and bovine LLPLs indicate that each entire sequence has a signal sequence cleavage site and N-linked glycosylation sites (Fig. 1). Based on the presence of the cleavage site, the processed LLPL would be predicted to consist of 379 amino acid residues, cor-FIG. 8. Expression of mouse, human, and bovine LLPLs in COS-7 cells. COS-7 cells were transiently transfected with carboxyl-terminally HAtagged mouse, human, and bovine LLPL cDNAs as described under "Materials and Methods." Each soluble fraction obtained from non-transfected or transfected cells was used for ACS activity assay (A) and Western blotting (B). In A, 3 g of each soluble fraction obtained from non-transfected or transfected cells was incubated for 10 min at 37°C with liposomes containing NAS as described under "Materials and Methods." In B, 20 g of each soluble fraction obtained from carboxylterminally HA-tagged LLPL-transfected cells was incubated with or without endoglycosidase F 1 for 14 h at 4°C. After the reaction, the proteins were separated using a 10% acrylamide gel and transferred to a PVDF membrane. The membrane was incubated with anti-HA monoclonal antibody. The antigen-antibody complex on the membrane was visualized with horseradish peroxidase-conjugated goat anti-mouse IgG antibody using diaminobenzidine and hydrogen peroxide. responding to a molecular mass of 43 kDa. This value is in agreement with the observed results following treatment of each recombinant LLPL with endoglycosidase F 1 . Based on amino-terminal sequence analysis of bovine brain LLPL, bovine LLPL is presumed to consist of 378 amino acid residues. These data support the interpretation that the precursor protein of LLPL is post-translationally modified by both signal peptide cleavage and N-linked glycosylation.
The observations that ACS has an acidic pH optimum and N-linked oligomannose and co-localizes with ␤-hexosaminidase in MDCK cells strongly indicate that ACS is a lysosomal enzyme. Deglycosylation of recombinant LLPLs had no effect on ACS activity. Thus, the oligomannose in ACS is probably involved in sorting ACS to lysosome.
The family of phospholipase A 2 enzymes has expanded greatly in recent years (15). With this expansion, the criteria for recognition of an enzyme as a phospholipase A 2 and its assignment to one of the 11 currently recognized groups have become more stringent. Currently, there are no lysosomal phospholipases A 2 that meet such criteria. In 1997, a calciumindependent phospholipase A 2 activity that was inhibited by serine hydrolase inhibitors was described (16). This enzyme was subsequently identified as a 1-cysteine peroxiredoxin (17); and thus, its characterization as a phospholipase A 2 has been challenged.
We conclude that the product of the gene encoding LLPL is not a lysophospholipase, but ACS. Moreover, ACS is identified as a lysosomal enzyme and has the dual enzyme activities of a calcium-independent transacylase and phospholipase A 2 . A reaction mechanism for ACS is proposed (Fig. 10). In this model, the enzyme reacts with phospholipid and forms an acyl-enzyme intermediate at serine 165 of the mature protein based on the putative cleavage site. The acyl group of the intermediate is then transferred to a hydroxyl group of water or lipophilic alcohol such as ceramide.
We believe that ACS is a lysosomal phospholipase A 2 , but its FIG. 9. Localization of 1-O-acylceramide synthase in MDCK cell. Percoll gradient fractionation was carried out as described under "Materials and Methods." For the ␤-hexosaminidase assay, the reaction mixture consisted of 36 mM sodium citrate (pH 4.5), 2 mM p-nitrophenyl-N-acetyl-␤-glucosamine, 0.1% (w/v) Triton X-100, and 17.5 l of Percoll fraction in a total volume of 250 l. The reaction was initiated by adding the fraction, kept for 20 min at 37°C, and terminated by adding 250 l of 0.2 M Na 2 CO 3 . The formation of p-nitrophenol was measured by absorbance at 400 nm. ACS activity was determined by formation of 1-O-acyl-NAS. For the transacylase assay, the reaction mixture consisted of 42 mM sodium citrate (pH 4.5), 10 g/ml bovine serum albumin, 10 M [ 3 H]NAS, liposomes (64 nmol of phospholipid, PC/PE/dicetyl phosphate (7:3:1)), and 20 l of Percoll fraction in a total volume of 500 l. The reaction was initiated by adding the fraction, kept for 20 min at 37°C, and terminated by adding 3 ml of chloroform/methanol (2:1). biological function is unknown. Several possible functions may be entertained, but remain to be tested. ACS may play a primary role in regulating the levels of ceramide within cells. Many investigators have postulated that ceramide may mediate cell differentiation and death responses. A means for regulating ceramide content, particularly through lysosome-mediated pathways, may be important in such a system. In this regard, ACS may function to terminate the biological activities of ceramide generated through signaling pathways perhaps by sequestering ceramide within lysosomes or by facilitating its movement across membrane leaflets. Alternatively, ACS produces a novel metabolite, 1-O-acylceramide. We have previously observed that phospholipids containing arachidonate can serve as the acyl donor for the transacylase reaction (2). It will be interesting to determine whether 1-O-arachidonylceramide is a source for biologically active arachidonic acid in cells. Finally, it is also conceivable that water is the usual acceptor for ACS and that its primary function is to serve as a lysosomal phospholipase A 2 with PE and PC as preferred substrates and their respective lysolipids and free fatty acids as products.
Based on the characterization reported in this study, the nomenclature used to refer to this lysosomal phospholipase is inaccurate. LCAT-like lysophospholipase does not describe the primary activity of this protein and should probably be discarded. Similarly, 1-O-acylceramide synthase may not reflect the primary lipid pathway catalyzed by the phospholipase because its in vivo activity has yet to be established. We propose therefore that the name lysosomal phospholipase A 2 be used in reference to this enzyme.