INTRODUCTION

Membrane phospholipids of activated platelets serve as precursors for second messengers, such as eicosanoids, and also as footholds for blood coagulation. Platelets contain at least three types of phospholipase A; cytosolic phospholipase A₂ (cPLA₂)¹ (1, 2), secretory 14-kDa type II phospholipase A₂ (sPLA₂) (3), and another serine-phospholipid-selective phospholipase which has not yet been fully characterized (4-6). cPLA₂ is involved in the production of eicosanoid by cleaving arachidonic acid at the sn-2 position of phospholipids in various types of cells. sPLA₂ is found in inflammatory sites (7, 8) and is considered to be involved in inflammatory response progression and may participate in eicosanoid formation in certain cells, such as vascular endothelial cells, mast cells, neutrophils, hepatocytes, and others (9-13). Thus sPLA₂ and cPLA₂ are well characterized, and their cDNA and genomic structures are also known (2, 14, 15), whereas no information about the structure of the last enzyme is available.

Like sPLA₂ this new member of the phospholipase A family is secreted from activated rat platelets (4); partially purified preparations specifically act on phosphatidylserine (PS) (6) and lysophosphatidylserine (lyso-PS) (4, 5). This activity is inhibited both by diisopropyl fluorophosphate (DFP) and dithiothreitol to release a fatty acid of PS and lyso-PS, respectively (5, 6). On several chromatography columns, PS-hydrolyzing activity co-migrates with lysophospholipase activity detected using lyso-PS as the substrate, suggesting that a single polypeptide may possess both PS-phospholipase A and lyso-PS-lysophospholipase activities. Therefore, throughout the present study we call this enzyme "PS-PLA₁."

In this study, we report the purification and cDNA cloning of PS-PLA₁. The predicted amino acid sequence did not show any homology to those of phospholipases that have been reported previously, but surprisingly it showed some homology to mammalian lipases such as hepatic lipase (HL), lipoprotein lipase (LPL), and pancreatic lipase (PL). Unlike these "conventional" lipases, the present platelet-derived enzyme did not exhibit appreciable activity to triacylglyceride (TG). Platelets release a novel phospholipase A, upon cell stimulation, that specifically acts on serine-phospholipids and not on any other lipids.

EXPERIMENTAL PROCEDURES

1-Palmitoyl-2-[¹⁴C]arachidonoyl-sn-glycero-3-phosphoethanolamine, 1-palmitoyl-2-[¹⁴C]arachidonoyl-sn-glycero-3-phosphoinositol, 1-palmitoyl-2-[¹⁴C]arachidonoyl-sn-glycero-3-phosphocholine 1,2-di[¹⁴C]oleoyl-sn-glycero-3-phosphocholine, tri[¹⁴C]oleoyl-glycerol, and 1,3-[³H]diisopropyl fluorophosphate were from Dupont NEN. 1,2-Di[¹⁴C]oleoyl-sn-glycero-3-phosphoserine and 1-[¹⁴C]oleoyl-sn-glycero-3-phosphoserine were prepared as described previously (5, 6). [-³²P]dCTP was the product of Amersham International (Amersham, UK). PS, phosphatidylethanolamine, phosphatidylcholine, phosphatidic acid, phosphatidylinositol, 1-acyl-sn-glycero-3-phosphocholine, 1-acyl-sn-glycero-3-phospho-myo-inositol, and 1,2-diacylglycerol were obtained from Serdary Research Laboratories (London, UK). 1-Acyl-sn-glycero-3-phosphoserine, 1-acyl-sn-glycero-3-phosphoethanolamine, and lysophosphatidic acid were from Avanti Polar Lipids (Alabaster, AL). Triolein was the product of Sigma. DEAE-Sepharose CL-6B, heparin-Sepharose CL-6B, blue Sepharose CL-6B, and Sepharose 4B were bought from Pharmacia Fine Chemicals (Uppsala, Sweden). Other chemicals were purchased from Wako (Osaka, Japan).

1-[1-¹⁴C]Palmitoyl-2-acyl-sn-glycero-3-phosphoserine was used to detect PS-PLA₁ activity in column chromatography fractions. Phospholipase A₁ and lysophospholipase activities were measured as described previously (5, 6). For determination of phospholipid specificity, 1-palmitoyl-2-[¹⁴C]arachidonoyl-sn-glycero-3-phosphoserine, 1-palmitoyl-2-[¹⁴C]arachidonoyl-sn-glycero-3-phosphoethanolamine, 1-palmitoyl-2-[¹⁴C]arachidonoyl-sn-glycero-3-phosphoinositol, 1-palmitoyl-2-[¹⁴C] arachidonoyl-sn-glycero-3-phosphatidic acid, and 1-palmitoyl-2-[¹⁴C] arachidonoyl-sn-glycero-3-phosphocholine (40 µM each) were incubated with recombinant PS-PLA₁ in 100 mM Tris-HCl pH 7.5, 4 mM CaCl₂ at 37 °C for 15 min, and the products were analyzed by thin layer chromatography using chloroform/methanol/acetic acid/H₂O (25/15/4/2; v/v/v/v). The radioactivities corresponding to each product were detected and analyzed using a BAS 2000 imaging analyzer (Fuji Film, Kanagawa, Japan).

Blood was collected from 200-300 Wistar rats by cardiac puncture, and platelet-rich plasma was prepared as described previously (5). The platelets were activated by incubating with 2 units/ml thrombin and 2 mM CaCl₂ for 20 min at 37 °C. Then, the plasma was centrifuged at 1,500 × g for 10 min at 4 °C, and the supernatant was used as the enzyme source.

The supernatant was loaded onto a DEAE-Sepharose CL-6B column (30 ml), which had been equilibrated with 100 mM NaCl, 10 mM Tris-HCl, pH 7.4. The flow-through fraction was then loaded onto a heparin-Sepharose CL-6B column (10 ml), which had been equilibrated with 100 mM NaCl, 10 mM Tris-HCl, pH 7.4. The column was washed thoroughly with this buffer and eluted with a linear gradient of NaCl (0.1-1.0 M). The pooled fractions containing lysophospholipase activity were applied to a blue Sepharose CL-6B column (2 ml). After the column was washed with 500 mM NaCl, 10 mM Tris-HCl, pH 7.4, the enzyme was eluted by changing the buffer to 1 M NaCl, 50 mM Tris-HCl, pH 9.0.

Aliquots of various fractions that contained PS-PLA₁ activity were mixed with [³H]DFP (5 µCi, 0.58 nmol) and incubated for 15 min at room temperature. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and fluorography were carried out as described previously (16).

The N-terminal amino acid sequence was determined by the method of Matsudaira (17). Briefly, the partially purified protein (the active fractions from the blue column) was concentrated, and the protein was separated by SDS-PAGE. After blotting onto a polyvinylidene difluoride membrane, the protein band was visualized by Coomassie staining, cut out, and subjected to automated Edman degradation using an ABI model 477A protein sequencer connected on-line to an ABI model 120A phenylthiohydantoin analyzer (Perkin-Elmer).

For determination of internal amino acid sequences, the protein was S-carboxymethylated and separated by SDS-PAGE, and in situ fragmentation either by CNBr treatment or by digestion with lysyl endopeptidase was performed. The resulting peptide fragments were separated by reverse-phase high performance liquid chromatography using a gradient of 0-64% acetonitrile in 0.1% trifluoroacetic acid. Alternatively, the peptides were separated by SDS-PAGE using a 10-24% acrylamide gel and blotted to a polyvinylidene difluoride membrane. The isolated peptides were sequenced in an ABI model 477A protein sequencer.

A rat megakaryocyte gt11 phage cDNA library (18) was used for screening. The cDNA fragment that contained internal amino acid sequences was amplified by the polymerase chain reaction (PCR) using the megakaryocyte gt11 phage cDNA library as the template, and the following degenerated oligonucleotide primers; GTICCICCICCIACNCARCC and ACICGIGADATRTGGAA, based on the VPPPTQP and FHISSRV sequences of the two peptides (N-terminal and Lys-end-2) (see Table II), respectively. The resulting PCR fragments were subcloned into a pT7Blue T-vector (Novagen, Madison, WI), and the nucleotide sequences were determined. A DNA fragment which contains internal amino acid sequences was chosen and used for identifying  phages which contained same sequence as the DNA fragments by plaque hybridization. The insert DNA was cut off by digestion from the phage DNA with EcoRI and was subcloned into the EcoRI site of the pBluescript II SK phagemid vector (Stratagene, La Jolla, CA). After the restriction enzyme map (Fig. 3A) was determined, the nucleotide sequence was determined by DNA sequencing. 5-RACE was performed using a Marathon cDNA amplification kit (Clontech, Palo Alto, CA) with rat platelet total RNA as the template for first strand cDNA synthesis according to the manufacturer's protocol. The antisense oligonucleotide TTCCGGATGTCACTGTCCTC (nucleotides 205-224 of PS-PLA₁, see Fig. 3B) was used as a primer for first strand synthesis.

Table II. Amino acid sequence of N-terminal and internal peptides Capital letters signify clearly indentified sequence; small letters represent sequences analyzed with ambiguity. Peptide Sequence Position N-terminal GNVPPpTQPKCTDFQSANLLRGtnxKVQF 25 -53 Lys-end-1 (K)ITIPK 372 -376 Lys-end-2 (K)FHISSRvwRK 401 -410 Lys-end-3 (K)AFLAGDsLdsFNP 286 -299 Lys-end-4 (K)LSLEISRFLsK 139 -149 CNBr-1 (M)AFPCASYKAFLAGDSLDSFNPFLL 279 -302 CNBr-2 (M)VGHFYKGQLGRITGLDPAGP 175 -194

cDNA encoding the coding region of PS-PLA₁ (nucleotide positions 13-1729) was inserted into the SmaI and EcoRI sites of a baculovirus transfer vector pVL1393 (PharMingen, San Diego, CA). The plasmid obtained was designated PS-PLA₁-pVL1393. A recombinant virus was prepared using the BaculoGold system (PharMingen) according to the manufacturer's protocol. Cells (6 × 10⁵ cells/ml) were mixed with recombinant or wild-type Autographa californica nuclear polyhedrosis virus (AcNPV) (multiplicity of infection = 10) and incubated for 96 h at 27 °C. Phospholipase A₁ and lysophospholipase activities in the cell culture supernatant were determined as described above.

[¹⁴C]PS (2 µM) or [¹⁴C]TG (2 µM) was incubated with the enzyme in 4 mM CaCl₂, 100 mM Tris-HCl, pH 7.5, at 37 °C for 15 min in the presence of 300 µM Triton X-100. After extraction from the mixture by Bligh and Dyer's method, the release of lyso-PS and diglyceride was measured by TLC using chloroform/methanol/acetic acid/H₂O (25/15/4/2; v/v/v/v) and hexane/ether/acetic acid (70/30/1; v/v/v), respectively. Radioactivity was detected with BAS 2000 imaging analyzer. In some experiments, [¹⁴C]PS (2 µM) was mixed with nonlabeled TG (2 µM) in the presence of 300 µM Triton X-100. Similarly, [¹⁴C]TG (2 µM) was mixed with nonlabeled PS (2 µM) in the presence of 300 µM Triton X-100. The lipid mixture was then incubated with the enzyme in 4 mM CaCl₂, 100 mM Tris-HCl, pH 7.5, at 37 °C for 15 min.

Protein concentrations were determined using the BCA protein assay reagent (Pierce). SDS-PAGE was performed by the method of Laemmli. DNA sequencing was performed using an ALFred DNA sequencer with an AutoCycle Sequencing Kit (Pharmacia Biotech Inc.).

RESULTS

As rat platelets released both PS-selective phospholipase A₁ and lyso-PS-selective lysophospholipase activities upon cell activation, the extracellular medium from thrombin-activated platelet cultures was used as the enzyme source. Lyso-PS was used as a substrate for assessing enzyme activity throughout the present study. First the medium was subjected to DEAE-Sepharose column chromatography. The enzyme passed through the column, whereas 40% of the total protein was absorbed. Next the flow-through fraction was applied to a heparin-Sepharose column, and the enzyme activities were eluted with linear gradient of NaCl. The activities were eluted in fractions at about 250 mM NaCl. The pooled fractions were then applied to a blue Sepharose column, and the activities were eluted by 1 M NaCl. Typical elution profiles of the activities from the heparin-Sepharose and blue Sepharose columns, and the SDS-PAGE pattern of the fractions collected from the heparin-Sepharose and blue Sepharose columns are shown in Fig. 1. A major 55-kDa protein co-migrated with the lysophospholipase activity on the blue Sepharose column (Fig. 1B). This 55-kDa protein also co-migrated with lysophospholipase activity on the heparin-Sepharose column (Fig. 1A) and on other types of chromatography columns that we tested, such as a gel filtration column and a CM-Sepharose column (data not shown). The purification procedure is summarized in Table I. The overall purification was about 1800-fold, and the recovery was 31%.

Table I. Partial purification of lysophospholipase active on lyso-PS from supernatant of activated platelets Lysophospholipase activities were measured by the modified Dole's method using 1-[14C]oleoyl-sn-glycero-3-phosphoserine as a substrate (see "Experimental Procedures"). Procedure applied Total protein Total activity Specific activity Recovery mg µmol/min µmol/min·mg -fold % Supernatant of activated platelets 571 4.46 0.008 1 100 DEAE-Sepharose 345 4.00 0.012 1.5 90 Heparin-Sepharose 5.9 3.62 0.614 77 81 Blue Sepharose 0.1 1.46 14.0 1800 31

DFP inhibited both PS-PLA₁ and lyso-PS-lysophospholipase activities in the purified preparation. To further confirm that the 55-kDa protein was the enzyme itself, we next performed a [³H]DFP-labeling experiment. DFP inhibits the activities of various serine esterases, including lipases, by reacting covalently with an active serine residue of these enzymes. Fractions positive for the enzyme activity that were obtained during the purification procedure were then incubated with [³H]DFP, and incorporation of the radioisotope was analyzed by SDS-PAGE and fluorography. As shown in Fig. 2, the purified 55-kDa polypeptide did incorporate [³H]DFP, confirming that it possesses an active serine residue. Labeling of the 55-kDa polypeptide with [³H]DFP was seen in the fraction obtained after the heparin-Sepharose and blue Sepharose chromatography (Fig. 2). Other bands which were detected in these partially purified preparations were not components of the enzyme, because they did not co-migrate with the enzyme activities on several chromatography columns (data not shown). Thus we conclude that the 55-kDa protein is the enzyme itself.

We next determined the N-terminal and internal amino acid sequences of purified PS-PLA₁. Table II shows the sequences of the peptides obtained. On the basis of these sequences, degenerated oligonucleotides were synthesized and used in a series of PCRs with a rat megakaryocyte gt11 cDNA library as a template. One set of primers yielded an amplified product containing sequences corresponding to several peptides (Fig. 3A). Therefore we concluded that this amplified product was generated from the PS-PLA₁ cDNA and utilized it to screen a rat megakaryocyte gt11 cDNA library. As a result two  clones were obtained and sequenced (Fig. 3A). However, these two clones did not contain the entire coding region, so we amplified cDNA corresponding to the 5-region by 5-RACE. The nucleotide sequence of the PS-PLA₁ cDNA is shown in Fig. 3B. The deduced amino acid sequence contains all amino acid sequences obtained from N-terminal and internal amino acid sequence analysis. It contains an open reading frame of 1368 base pairs which encodes a 456-amino acid polypeptide. A hydrophobic sequence, composed of 24 amino acid residues, is present at the N-terminal of the deduced amino acid sequence. This hydrophobic sequence is probably a signal sequence because Gly²⁵ is an N-terminal amino acid of purified protein (Table II). The deduced amino acid sequence contains three possible sites for N-linked glycosylation, a putative active serine (in a Gly-X-Ser-X-Gly motif), and 16 cysteine residues. The presence of the motif of the active serine residue explains the inhibition of the activity of this enzyme by the serine esterase inhibitor DFP (5).

To confirm that the cDNA obtained encoded PS-PLA₁, a recombinant baculovirus was prepared and used to infect Sf9 insect cells. As shown in Fig. 4A, the culture medium of Sf9 cells infected with the recombinant baculovirus exhibited appreciable lysophospholipase activity, whereas the culture medium obtained with wild-type virus did not. The medium of cells infected with the recombinant virus also showed PS-hydrolyzing activity (Fig. 4B), confirming that both lysophospholipase and PS-hydrolyzing activities were mediated by the same protein. The activity was mostly recovered from the medium, suggesting that the enzyme was spontaneously secreted from Sf9 cells. In platelets the enzyme might be stored in granules and secreted upon cell activation.

We searched the protein data base for similar sequences using the BLASTN program (19). LPL, HL, and PL of various species such as human, rat, and mouse showed significant homology with the enzyme (Fig. 5A). The enzyme showed 30.8, 31.1, and 29.1% identity to rat LPL (20), HL (21), and PL (22), respectively. Each of these lipases (LPL, HL, and PL) is composed of two domains (the N- and C-terminal domains) (23, 24). As shown in Fig. 5A, the homologous regions (amino acids 1-280 in PS-PLA₁) are in the first half of the molecule, which corresponds to the N-terminal domain of lipases. In particular, amino acid residues surrounding Ser¹⁴², Asp¹⁶⁶, and His²³⁶ in PS-PLA₁ are well conserved in these three lipases and may form a "catalytic triad" in PS-PLA₁, because the amino acids corresponding to these three amino acids in LPL, HL, and PL make up catalytic triads (25). In addition, six cysteine residues (amino acid positions 221, 234, 258, 269, 272, and 280 in PS-PLA₁) are completely conserved in LPL, HL, and PL (Fig. 5A). A potential N-glycosylation site (Asn⁵⁵ in PS-PLA₁) is also conserved in LPL and HL. It has been reported that glycosylation at the conserved asparagine of LPL and HL is essential for the synthesis of fully active enzymes (26, 27). The latter half of the PS-PLA₁ molecule (amino acids 281-426) did not show significant homology with LPL, HL, or PL (Fig. 5A). Lipase lids exist in LPL and HL, and are surrounded by two cysteine residues (216 and 239 in LPL, and 239 and 262 in HL) (24). These lids are reported to be involved in determining the substrate specificity of the two enzymes (28). In PS-PLA₁ the region corresponding to the lipase lids (amino acids 222-233, the putative lid of PS-PLA₁) is composed of 12 amino acid residues, whereas the lids of both LPL and HL are composed of 22 amino acid residues. Thus the putative lid of PS-PLA₁ is very short. The phylogenetic relationship between the lipase family and PS-PLA₁ is shown in Fig. 5B. This shows that PS-PLA₁ is more closely related to LPL and HL than PL.

From 500 ml of culture medium of Sf9 cells infected with recombinant virus we purified about 160 µg of recombinant enzyme by the same method as that used for rat platelet enzyme. This recombinant enzyme exhibited a smaller molecular mass (50 kDa) on SDS-PAGE analysis compared with enzyme derived from rat platelets (data not shown). The specific activity of recombinant enzyme toward PS was 2.6 µmol/min/mg, whereas those toward phosphatidylcholine, phosphatidylethanolamine, phosphatidic acid, and phosphatidylinositol were less than 0.1 µmol/min/mg (Table III). The substrate specificity of the platelet enzyme was the same as that of the recombinant enzyme (data not shown).

Table III. Substrate specificity of PS-PLA1 Phospholipase activities of recombinant enzyme against various phospholipids were decided as described under "Experimental Procedures." Substrate Specific activity µmol/min/mg Phosphatidylcholine <0.1 Phosphatidylethanolamine <0.1 Phosphatidylinositol <0.1 Phosphatidic acid <0.1 Phosphatidylserine 2.61

Next we examined whether the enzyme hydrolyzed TG, because the structure of the enzyme resembles the lipases (Fig. 5A). No appreciable hydrolysis was observed against triolein in the Triton X-100 micelle system, whereas PS was hydrolyzed under the same conditions (Fig. 6, lanes 1 and 2). Hydrolysis of triolein was not detected even under the conditions in which the same amount of lipids was introduced in the assay (Fig. 6, lanes 3 and 4). Thus PS-PLA₁ cannot digest TG in spite of its similarity in structure to the mammalian lipases.

PS-PLA₁ releases fatty acid from 1-acyl-lyso-PS, so it can cleave the fatty acid of lyso-PS at the sn-1 position. To test whether PS-PLA₁ has phospholipase B activity toward PS, the ability of PS-PLA₁ to cleave a fatty acid at the sn-2 position was investigated using 1-palmitoyl-2-[¹⁴C]arachidonoyl-PS as the substrate. As shown in Fig. 7 radioactivity was not detected at the TLC migration position of free fatty acid, but was detected at the position of lyso-PS. Thus the acyl chain at the sn-2 position in PS is not hydrolyzed by PS-PLA₁. When 1,2-di[¹⁴C]oleoyl-PS was used as the substrate for a PS-PLA₁ assay, radioactivity was detected in spots corresponding to both free fatty acid and lyso-PS (data not shown). These results suggest that PS-PLA₁ releases fatty acid from the sn-1 position in PS and lyso-PS.

DISCUSSION

In this study we purified PS-PLA₁ that was secreted from activated platelets and established its primary structure by cDNA cloning. The primary structure of the enzyme did not exhibit homology to any phospholipase A₂ or lysophospholipase that has been reported previously (type I PLA₂ (29), type II PLA₂ (14), cPLA₂ (2), hepatic lysophospholipase (30), Escherichia coli lysophospholipase (31), human eosinophil lysophospholipase (32), and rat pancreas lysophospholipase (33)). Unexpectedly PS-PLA₁ resembles some mammalian lipases (LPL, HL, and PL) although it did not show any detectable lipase activity.

The recombinant form of PS-PLA₁ hydrolyzed PS and lyso-PS at the sn-1 position of the glycerol backbone to release a fatty acid (Fig. 7), but did not hydrolyze other phospholipids (Table III). A previous study using partially purified enzyme showed that the enzyme hydrolyzed lysophosphatidylserine but did not hydrolyze other lysophospholipids such as lysophosphatidylcholine, lysophosphatidylethanolamine, and lysophosphatidic acid (5). The enzyme did not hydrolyze TG, even though it did hydrolyze PS under the same conditions (Fig. 6), and porcine pancreatic lipase hydrolyzed TG (data not shown). Thus PS-PLA₁ cannot hydrolyze TG, even though it shares some sequence homology with mammalian lipases. HL is known to possess both lipase and phospholipase activities. In addition, a phospholipase A₁ from hornets, which has a sequence similarity of about 40% with mammalian lipases, has phospholipase A₁ activity (using phosphatidylcholine as the substrate) and weak lipase activity (34). The substrate specificity of PS-PLA₁ is different from that of lipases because it cannot react with TG, but does recognize the glycerophosphoserine structure and hydrolyzes a fatty acyl residue bound at the sn-1 position of the glycerol backbone.

Which part of the molecular structure of PS-PLA₁ determines its substrate specificity? Under aqueous conditions, lipases are inactive, as the "catalytic triad" is buried beneath a short amphipathic -helix (the "lid") (23). Compared with the lid amino acid sequences of other lipases, the putative lid of PS-PLA₁ is extremely short (Fig. 5A). Rat LPL and HL have lids composed of 22 amino acids between two conserved cysteine residues, whereas the corresponding region in PS-PLA₁ is composed of only 12 amino acid residues. It has been reported that the substrate specificities of LPL and HL are determined by the lids of the molecules (28). Guinea pig PL has a short lid and exhibits both phospholipase A₁ and lipase activities (35), and hornet phospholipase A₁, which has only 7 amino acids in the corresponding region, has weak lipase activity (34). Thus the "phospholipase A₁ activity" of lipase family molecules may depend on the length and amino acid sequence of the lid.

The mechanism by which PS-PLA₁ specifically recognizes serine-phospholipids is not known at present. A cluster of positively charged amino acids (Lys and Arg) is present in the sequence of PS-PLA₁ (348-353). It is interesting that the amino acid sequences around this cluster (VENEKRKDT) resemble the PS-binding motif found in protein kinase C and PS decarboxylase (FXFXLKXXXXKXR) (36). Amino acid sequences of lipases around these regions are full of variety. This region may be responsible for determining the substrate specificity of PS-PLA₁. Mutation studies will test this hypothesis.

Of the known mammalian phospholipase As, rat platelet PS-PLA₁ is the only one that is a glycoprotein. Three potential N-glycosylation sites are present in the deduced amino acid sequence of PS-PLA₁. The idea that PS-PLA₁ is a glycoprotein was confirmed by two additional observations: 1) in our preliminary experiments, PS-PLA₁ and lyso-PS lysophospholipase activities were retained on concanavalin A-, WGA-, and RCA-Sepharose columns and were eluted by the hapten sugars of each lectin; and 2) the calculated molecular mass of PS-PLA₁ is 47 kDa, whereas the purified protein had a molecular mass of 55 kDa by SDS-PAGE analysis. This difference may be due to the addition of an asparagine-linked sugar moiety because treatment of the purified protein with N-glycosidase F, which removes N-linked sugar moieties, reduced the molecular mass of platelet PS-PLA₁ to 48 kDa (data not shown). The molecular mass of the recombinant protein, produced in a baculovirus system, was 50 kDa. The difference in molecular mass observed between rat platelet and recombinant PS-PLA₁ may be due to differences in the sugar structure because the molecular mass of recombinant PS-PLA₁ after N-glycosidase F treatment was also 48 kDa. A strict carbohydrate structure seems not to be required for PS-PLA₁ activity. This fact is in contrast with LPL and HL, since it has been reported that an N-linked sugar on LPL and HL at the N-terminal (amino acid 75) is necessary for enzyme activity (26, 27). In addition, PS-PLA₁ has no homology with phospholipase B of Penicillium notatum (37) or Saccharomyces cerevisiae (38), both of which are glycoproteins.

PS-PLA₁ was released from rat platelets when the cells were stimulated with thrombin. The enzyme is also released in response to other stimuli such as ADP and A23187 (5). In unstimulated platelets PS-PLA₁ may be stored in -granules, since type II PLA₂, which is known to be present in -granules, is released from rat platelets in a similar manner to PS-PLA₁ upon stimulation with thrombin (5). The mechanism of PS-PLA₁ sorting to secretory granules is not known at present. The N-terminal hydrophobic sequence of PS-PLA₁, which consisted of 24 amino acids, may be a signal sequence because Gly²⁵ is an N-terminal amino acid of the purified enzyme (Table II). According to the PSORT program, which predicts protein sorting from the primary amino acid sequence (39), the 24 hydrophobic amino acid sequence of PS-PLA₁ does not show features typical of the signal sequences of "secretory" proteins. The enzyme was secreted from infected Sf9 cells, supporting the fact that the signal observed in the present enzyme functions as a signal sequence.

The biological function of PS-PLA₁ is unknown at present. It is well known that membrane phospholipids of activated platelets serve as precursors for second messengers such as eicosanoids and also as footholds for blood coagulation. PS-PLA₁ is released from activated platelets. Under the conditions, 2-acyl lyso-PS may accumulate on the membrane, since PS-PLA₁ attacks only the fatty acyl residue at the sn-1 position of serine-phospholipids. PS and lyso-PS may play important roles in either intracellular signal transduction or intercellular blood coagulation processes. PS-PLA₁ may be involved in the regulation of these processes through control of the concentration of PS or lyso-PS.