Purification and properties of a phospholipase A2/lipase preferring phosphatidic acid, bis(monoacylglycerol) phosphate, and monoacylglycerol from rat testis.

Phospholipase A(2) (PLA(2)) was purified to homogeneity from the supernatant fraction of rat testis homogenate. The purified 63-kDa enzyme did not require Ca(2+) ions for activity and exhibited both phosphatidic acid-preferring PLA(2) and monoacylglycerol lipase activities with a modest specificity toward unsaturated acyl chains. Anionic detergents enhanced these activities. Serine-modifying irreversible inhibitors, (p-amidinophenyl) methanesulfonyl fluoride and methylarachidonyl fluorophosphonate, inhibited both activities to a similar extent, indicating a single active site is involved in PLA(2) and lipase activities. The sequence of NH(2)-terminal 12 amino acids of purified enzyme was identical to that of a carboxylesterase from rat liver. The optimal pH for PLA(2) activity (around 5.5) differed from that for lipase activity (around 8.0). At pH 5.5 the enzyme also hydrolyzed bis(monoacylglycerol) phosphate, or lysobisphosphatidic acid (LBPA), that has been hitherto known as a secretory PLA(2)-resistant phospholipid and a late endosome marker. LBPA-enriched fractions were prepared from liver lysosome fractions of chloroquine-treated rats, treated with excess of pancreatic PLA(2), and then used for assaying LBPA-hydrolyzing activity. LBPA and the reaction products were identified by microbore normal phase high performance liquid chromatography/electrospray ionization ion-trap mass spectrometry. These enzymatic properties suggest that the enzyme can metabolize phosphatidic and lysobisphosphatidic acids in cellular acidic compartments.

Lysophosphatidic acid (LPA) 1 is a key intermediate for de novo synthesis of phospholipids and triacylglycerol. In addi-tion, recent studies have established that LPA serves as an intercellular signaling molecule that mediates diverse cellular functions, such as cell growth and cytoskeletal remodeling (1), via G protein-coupled receptors (2). These receptors comprise several isoforms (2), which fulfill distinct functions through different signaling pathways depending on isoforms. On the other hand, there is little information on the metabolic pathways and enzymes responsible for LPA synthesis; several pathways have been proposed depending on tissue and cell types (3). Phospholipase A 2 (PLA 2 ) or phospholipase A 1 (PLA 1 ) directly produces LPA from phosphatidic acid (PA) generated by combined action of phospholipase C and diacylglycerol kinase or by direct action of phospholipase D (PLD). Diacylglycerol lipase deacylates diacylglycerol produced in response to stimulation and then the product 2-monoacylglycerol can be phosphorylated by monoacylglycerol kinase, generating LPA. Finally, plasma lysophospholipase D can hydrolyze lysophosphatidylcholine (LPC), yielding LPA.
Of these enzymes PLA 2 s are ubiquitous and have been studied most intensively. A variety of isozymes have been known, and some isoforms exhibit specificity for PA, including group IIA PLA 2 (4), rat brain 58-kDa PLA 2 (5), and intracellular Ca 2ϩ -independent PLA 2 (6). They might be candidates for LPAsynthesizing enzymes, but relevance to LPA synthesis in vivo has not yet been established, although it was suggested that group IIA PLA 2 attacks PA-containing microvesicles, shed from damaged cells in inflammation, to produce LPA (7). In the course of a study on tissue distribution of PLA 2 activity toward mixed micelles of acidic phospholipid and cholate, we found that rat testis contained appreciable amounts of Ca 2ϩ -independent PLA 1 and PLA 2 (8). Recently, PA-preferring PLA 1 was purified and cloned from bovine testis (9,10). In this study, we purified to homogeneity a PLA 2 with a substantial specificity to acidic phospholipids, PA and phosphatidylglycerol (PG) from rat testis supernatant and characterized it enzymatically. The purified enzyme (PA-PLA 2 /MGL) exhibited both PLA 2 and monoacylglycerol lipase activities as a single 63-kDa molecule but did not exhibit PLA 1 and lysophospholipase activities.
The PLA 2 activity of the enzyme toward PA showed acidic pH optimum of 5.5, suggesting that the enzyme might work in cellular acidic compartments including endosome/lysosome system. The late endosomes specifically contain another acidic phospholipid, bis(monoacylglycerol) phosphate (lysobisphos-phatidic acid, LBPA), a structural isomer of PG, that takes part in protein and cholesterol sorting in this system (11,12). The acidic pH optimum of the enzyme led us to examine its ability to hydrolyze LBPA, which has been known as a unique PLA 2resistant phospholipid because of its sn-1:sn-1Ј stereo-configuration (13). In the present study, we assayed LBPA-hydrolyzing activity of the purified enzyme using LBPA-rich lipid extracts prepared from liver lysosomal fractions of the rats treated with chloroquine (14) as substrate. LBPA and the reaction products were separated and identified by microbore normal phase HPLC/electrospray ionization ion-trap mass spectrometry.
Assay for Lipolytic Activities-PLA 2 , lysophospholipase, and lipase activities were determined by a non-radiometric HPLC method based on precolumn derivitization with 9-anthryldiazomethane (ADAM) as described previously (8). Individual fatty acids released from mixedacyl glycerophospholipids and tri-, di-, and monoacylglycerols were determined simultaneously by this method. Substrate stock solutions used were as follows: mixed micelles of 5 mM diradyl-phospholipid with various concentrations of taurocholate, cholate, deoxycholate (DOC), or Triton X-100; emulsions of 5 mM triacylglycerol or diacylglycerol; and 5% gum arabic for lipase. In a typical experiment, the assay mixtures contained 10 mM EDTA, substrate micelles, or emulsion (10-l stock solution), 0.1 M NaCl, 0.1 M Tris-HCl (pH 8.5 for lipase activity), or MES (pH 5.5 for PLA 2 and LBPA activity), 6 mM DTT, and the enzyme sample in a final volume of 50 l. The addition of DTT was essential for activity. In inhibition studies, enzyme solutions were preincubated for 1 h with APMSF and MAFP (the final concentrations of 500 and 50 M, respectively, in the assay mixture) and then enzyme activity was assayed.
CoA-independent Transacylase Assay-The transacylase activity was measured with mixed micelles of 6 mM taurocholate and the following acyl donor and acceptor combinations: mixtures of 1 mM LPC and 1 mM LPE as acceptor and either 1 mM SAPA or 1 mM POPA as donor; 2 mM LPA as acceptor and 1 mM POPG as donor. The assay mixtures contained above substrates and 0.1 M NaCl, 0.1 M MES (pH 5.5), 6 mM DTT, 10 mM EDTA, and enzyme sample in a final volume of 100 l. These mixtures were incubated for 3 h and extracted by a modified Bligh and Dyer method (16). The reaction products were analyzed by HPLC/electrospray ionization ion-trap mass spectrometry as described below.
Assay for Esterase Activities-Esterase activities were spectrophotometrically determined with p-nitrophenyl valerate and p-nitrophenyl myristate as substrates. The assay mixtures contained 0.5 mM substrate, 10 mM deoxycholate, 0.1 M Tris-HCl (pH 8.0), 6 mM DTT, and enzyme sample in a final volume of 0.5 ml. After the addition of enzyme solution an increase in absorbance at 400 nm was monitored continuously with a Jasco M550 spectrophotometer at 25°C.
Purification of PA-PLA 2 /MGL-The frozen rat testes (50 g) were homogenized in 500 ml of 20 mM Tris-HCl (pH 7.5) containing 2 mM EDTA, 0.5 mM APMSF, and 1 mM DTT and then sonicated for 5 min on ice. After filtrating the resultant homogenate with a stainless mesh, the extract was centrifuged at 23,000 ϫ g for 90 min. The pH of the supernatant was adjusted to 8.5 and then the solution was applied to a TEAE-cellulose column (6 ϫ 15 cm) pre-equilibrated with 20 mM Tris-HCl (pH 8.5) containing 0.5% Triton X-100 and 1 mM DTT. The PLA 2 activity was eluted with 20 mM Tris-HCl (pH 7.5) containing 0.1% Triton X-100 and 1 mM DTT by increasing the NaCl concentration from 0 to 1 M. To the pooled PLA 2 fractions was added 1 M lithium sulfate and then the solutions were applied to a phenyl-Sepharose column (3 ϫ 10 cm) pre-equilibrated with 20 mM Tris-HCl (pH 7.5) containing 1 M lithium sulfate and 1 mM DTT. The PLA 2 activity bound to this column under these conditions and was eluted with 10 mM Tris-HCl containing 10% ethylene glycol, 0.1%Triton X-100, and 1 mM DTT. The pooled PLA 2 fractions were applied on a SuperQ-Toyopearl column (1 ϫ 15 cm) after pre-equilibration in 20 mM Tris-HCl (pH 8.5) containing 0.1% Triton X-100. The column was washed with 20 mM Tris-HCl (pH 8.5) containing 0.1% C 12 E 8 to remove strongly UV-absorbing Triton X-100, then connected to a HPLC system, and developed with a linear gradient of NaCl concentration in 20 mM Tris-HCl (pH 8.5) containing 0.1% C 12 E 8 and 1 mM DTT (Buffer A), from 0 to 0.15 M for 120 min, and then from 0.15 to 1 M for 125 min. The flow rate was 0.7 ml/min, and 1.4-ml fractions were collected. The resultant PLA 2 fractions were loaded onto a Biogel A-0.5m column (2 ϫ 50 cm) pre-equilibrated with 20 mM Tris-HCl (pH 7.5) containing 0.3 M NaCl, 1 mM DTT, and 0.1% C 12 E 8 . The active PLA 2 fractions were purified further by HPLC on a Cosmogel QA column (7.5 ϫ 75 mm; Nacalai Tesque) pre-equilibrated with Buffer A. The PLA 2 activity was eluted with a concentration gradient of NaCl in Buffer A from 0 to 0.2 M for 60 min and then from 0.2 to 1 M for 80 min. The pooled PLA 2 fraction was rechromatographed on the same column; the pH of the eluent was decreased to 8.0, and a shallower gradient of NaCl from 0 to 0.1 M for 90 min was used. The flow rate was 0.5 ml/min, and 1-ml fractions were collected. The active PLA 2 pools were purified further by HPLC on a Super SW 3000 column (4.6 ϫ 300 mm; Tosoh Corporation) pre-equilibrated with 20 mM Tris-HCl (pH 7.0) containing 300 mM NaCl, 0.1% C 12 E 8 , and 1 mM DTT at 0.1 ml/min. The enzyme activity was eluted at the retention time of 29 min, coinciding well with a 280-nm peak.
Preparations of LBPA-rich Lipid Fractions-Male albino rats of the Sprague-Dawley strain were given an aqueous solution of chloroquine at 100 mg per kg of body weight per day through a stomach tube for 1 week (14). The rats were anesthetized with pentobarbital and killed by drawing blood from the abdominal aorta. The liver was removed and homogenized with Buffer B (0.25 M sucrose containing 5 mM Tris-HCl, 1 mM MgCl 2 , and 2 mM EDTA (pH 7.4)). The homogenate was centrifuged at 1,000 ϫ g for 10 min and then the supernatant was centrifuged at 10,500 ϫ g for 20 min. Lipids in the resultant pellets (crude lysosomal fractions) were extracted by the same method as used for extracting synthesized PAs as described above, and their phosphorus contents were determined and then used for assays as soon as possible. In some experiments the extracted lipids were analyzed by HPTLC with a one-dimensional two-solvent system using chroloform/methanol/30% ammonium hydroxide (65:35:8) (v/v/v) followed by hexane/diethyl ether/ acetic acid (16:4:2) (v/v/v). LBPA and its molecular species were identified by a normal phase HPLC/ESI ion-trap mass spectrometry. An aliquot of extracted lipids was injected to a Lichroshere Si-100 column (1 ϫ 150 mm) pre-equilibrated with Solvent A at the flow rate of 50 l/min, and the column was developed with a linear gradient of Solvent C, Solvent A/1 M ammonium formate/water (100:2.24:9.76) (v/v/v). The effluent was monitored with a ThermoFinnigan LCQ mass spectrometer equipped with an ESI ion source in alternate positive and negative ion full scan, and data-dependent negative ion MS/MS modes on a single run.
Assay for LBPA-hydrolyzing Activity-The mixed micelles of freshly reprepared LBPA-rich lipids (1 mM of total phospholipids) and 6 mM taurocholate was used for substrate. Because LBPA is not a substrate for pancreatic PLA 2 , almost all phospholipids other than LBPA included in the micelles were first hydrolyzed thoroughly by pancreatic PLA 2 (4.8 g/ml) purified from guinea pig stomach (18) for 30 min in the presence of 5 mM CaCl 2 and then the resulting solution was immediately used for an LBPA assay. The assay mixtures contained the pancreatic PLA 2 -treated mixed micelles (20 l) and 0.1 M NaCl, 0.1 M MES (pH 5.5), 6 mM DTT, 10 mM EDTA, and purified PA-PLA 2 /MGL in a final volume of 100 l. The mixtures were incubated at 37°C and then extracted as reported (16). Fatty acid release from LBPA was determined by the ADAM method described above. Another reaction product, lysophosphatidylglycerol, and its molecular species were identified by HPLC/negative ion MS/MS spectrometry as described above.
HPLC-The HPLC system consisted of two Gilson model 302 liquid delivery modules, a Gilson model 811 dynamic mixer, and a model 611 programmable UV detector. For mass spectrometry, the UV detector was not used, but the column was connected directly to a mass spectrometer. For a column of 1-mm diameter, the use of Accurate (LC Packings) as a mixer greatly improved mixing efficiency. Lichrosphere Si-100 (Merck) was slurry-packed into a column (1 ϫ 150 mm) in our laboratory.
Protein Sequencing-The purified enzyme was reduced, S-carboxymethylated, and purified as described in the legend for Fig. 4. The amino acid sequences were analyzed with an Applied Biosystems 477A sequencer and a 120A PTH analyzer. To identify a phenylthiohydantoin S-carboxymethylated Cys precisely, both S-carboxymethylated and unmodified proteins were analyzed, and the resultant data were compared.

RESULTS
Purification of PA-PLA 2 /MGL-We purified to homogeneity PA-PLA 2 /MGL as described under "Experimental Procedures," and the results of purification are summarized in Table I. We focused on and purified the PLA 2 activities with preference to PA. Rat testis contains a PLA 1 activity toward PA (9), which was co-purified with PLA 2 at earlier steps of purification but was not detectable after a SuperQ Toyopearl chromatography presumably because of its nonspecific adsorption onto the column. Among several chromatographic steps involved in this purification strategy, Cosmogel QA HPLC at pH 8.0 was the most effective. At the final step of purification gel chromatography on a Super SW 3000 column allowed us to remove minor contaminants reproducibly. To ensure its high resolution, the HPLC was operated at a low flow rate of 0.1 ml/min, leading to the overall purification of 538-fold (see Table I and Fig. 1B). Inclusion of DTT and a non-ionic detergent in the eluent of chromatographies was essential for improving the recovery.
During purification we found that monoacylglycerol lipase activity was co-purified with PLA 2 activity. Almost all of PLA 2 activity toward PA bound to a TEAE-cellulose column, whereas a significant fraction of monoacylglycerol lipase activity flew through. The ratios of lipase to PLA 2 activities after Biogel A-0.5m gel chromatography were rather constant (Table I). Silver-stained SDS-PAGE analysis showed a single band of 63-kDa protein (Fig. 1C). This molecular mass was similar to that estimated by gel chromatography on a Super SW 3000 column (72 kDa). To check whether this 63-kDa protein represented PA-PLA 2 /MGL, aliquots of purified enzyme were separated by polyacrylamide gel electrophoresis on two adjacent lanes of a native gel. A gel strip of one lane was stained with Coomassie Brilliant Blue, and the other lane was sliced into pieces of 3-mm length. Monoacylglycerol lipase and PLA 2 activities were assayed with materials extracted from these gel slices, and both activities were recovered in the same strip containing a protein band. The omission of DTT from the eluent on gel chromatography at the final step of purification caused almost complete losses of both PLA 2 and monoacylglycerol lipase activities. Similarly, the respective specific activities of purified enzyme for POPA and monooleoylglycerol decreased significantly in a similar extent, from 0.22 and 4.0 mol/ min/mg (the specific activity ratio of lipase/PLA 2 of 18.2) in the presence of DTT to 0.13 and 2.2 mol/min/mg (the ratio of 16.9) in its absence. These results indicate that a single enzyme catalyzes PLA 2 and monoacylglycerol lipase activity.
Catalytic Properties of Purified PA-PLA 2 /MGL-The addition of either Ca 2ϩ ions up to 50 or 10 mM EDTA in the assay mixtures did not affect the enzyme activities toward POPA and monoolein at pH 5.5 and 8.5, respectively. The addition of DTT in the assay mixtures enhanced both PLA 2 and lipase activities to a similar extent (1.7-fold for PLA 2 and 1.8-fold for lipase). Anionic detergents enhanced both PLA 2 and monoacylglycerol lipase activities greater (taurocholate Ͼ cholate Ͼ DOC) than non-ionic Triton X-100; activity was measured at pH 8.5 to ensure solubility of cholate and DOC. The pH dependence of the enzyme action toward POPA and monoolein was intriguingly different; the optimal pH values were 5.5 for POPA and 8.5 for monoolein (Fig. 2B), but both activities extended broadly over a physiologically relevant pH range from pH 5.5 to 7.5. Fig. 3 shows the substrate specificity of the purified enzyme. The enzyme preferred anionic phospholipids POPA and POPG in this order but hardly hydrolyzed zweiterionic POPE and POPC. It released oleic acid from the sn-2 position of POPG, but not palmitic acid esterified at its sn-1 position, and can hydrolyze 1-O-hexadecyl-OPA as efficiently as POPA, confirming the A 2 regiospecificity for these phospholipids (see Fig. 2A and Fig. 3A). As to the specificity for the sn-2 acyl groups, PA-PLA 2 /MGL much preferred unsaturated than saturated acyl chains (Fig. 3A). The enzyme also hydrolyzed bisphosphatidic acid, an anionic and more bulky substrate, to the extent similar to POPA, but we did not address the regiospecificity for this substrate. PA-PLA 2 /MGL hardly exhibited lysophospholipase activities toward 1-acyl lysophospholipids including LPA, LPC, LPI, and LPS at pH 5.5. PA-PLA 2 /MGL exhibited monoacylglycerol lipase activity with modest preference for polyunsaturated acyl groups at pH 8.5, with the specificity order linolenoyl Ͼ arachidonoyl Ͼ linoleoyl Ͼ oleoyl Ͼ stearoyl, whereas its di-and tri-acylglyc-erol lipase activities were very low (Fig. 3B). This order was the case with an assay using mixtures of stearoyl-, oleoyl-, linoleoyl-, linolenoyl-, and arachidonoylglycerols (1 mM each) in the presence of 6 mM taurocholate (pH 8.5), as substrates to ensure the similar surface quality for each substrate. Cholesterol oleate, which was emulsified by three different methods as described under "Experimental Procedures" was not a substrate for PA-PLA 2 /MGL. CoA-independent transacylase activities were not detectable. Serine-modifying irreversible inhibi- tors, APMSF and MAFP, inhibited PLA 2 and lipase activities to a similar extent (Table II). PA-PLA 2 /MGL also exhibited esterase activity toward p-nitrophenyl esters of short and long chain fatty acids. The specific activity for the valerate ester (580 mol/min/mg) was significantly greater than the myristate ester (12 mol/min/mg). After this treatment the amounts of diacylphospholipids were less than 0.5% of those in untreated lipids except for 1-stearoyl-2-arachidonoyl-phosphatidylinositol and LBPA as revealed by mass spectrometry. Signals of PA were not detectable after the treatment. LBPA is a structural isomer of PG, but tandem mass spectrometry can discriminate both isomers; negative ion MS/MS spectrum of PG contains a peak (m/z ϭ 791) for di-docosahexaenoyl (C22:6)-PG arisen from neutral loss of a glycerol moiety but not that of LBPA (Fig. 4B). The major LBPA molecular species were found to be di-C22:6-and C22:6-linoleoyl (C18:2)-LBPAs as revealed in the presence of the respective fatty acid anions in negative ion MS/MS spectra (Fig. 4, A and B). Other phospholipids with such fatty acyl groups were not detectable after treatment with pancreatic PLA 2 ; major lysophospholipid molecular species found were either palmitoyl-and stearoyl-LPC or LPE and stearoyl-LPS, further supporting the assignment of unsaturated fatty acids into sn-2 position as described above. These unique fatty acyl compositions of LBPA and the substrate specificity of PA-PLA 2 /MGL allowed us to assay LBPA hydrolyzing activity by following the C22:6 and C18:2 release on incubation of purified PA-PLA 2 /MGL with the pancreatic PLA 2 -treated LBPA-rich lipid fractions. Fig. 5A shows the time course of C22:6 and C18:2 release. The specific activity for the total of C18:2 and C22:6 release (0.33 mol/min/mg) was comparable with that toward PA, but C18:1 and C20:4 release was very low. Release of lysophosphatidylglycerol (LPG), another product of LBPA hydrolysis by PA-PLA 2 /MGL, was examined by negative ion HPLC/ESI-MS spectrometry (Fig. 5, B-D). The major species was found to be C22:6-LPG (m/z ϭ 555). For comparison, we used endogenous N-palmitoyl-sphingomyelin (m/z ϭ 747 of its formate adduct) as an internal standard (Fig.  5B). On MS/MS analysis the presence of a peak (m/z ϭ 463) derived from the neutral loss of glycerolϪH 2 O and that of C22:6 anion (m/z ϭ 327) confirmed its structure.
Sequence Analysis-The sequence of the NH 2 -terminal 12 amino acids of the purified enzyme was YPSSPPVVNTVK and is identical to those of rat liver serine esterases CES1A including microsomal serine carboxylesterase ES-10 (19) and cholesterol esterase (20). DISCUSSION We purified to homogeneity a phospholipase A 2 from rat testis that preferred both PA and monoacylglycerol. Purified enzyme showed a single band on a silver-stained gel (Fig. 1), ratios of the specific activity of PLA 2 to lipase were constant at the final purification steps (Table I), and the two activities co-migrated on native gels required the presence of DTT in the assay mixtures and were similarly inhibited by MAFP and APMSF (Table II), suggesting that a single enzyme catalyzes PLA 2 and monoacylglycerol lipase activity. Although mammalian PLA 2 s are rather specific for diradyl phospholipids, phos-  pholipase B/lipase expressed in rat testis and intestine exhibits high PLA 2 , lysophospholipase, and triacylglycerol lipase activities (15). PA-PLA 2 /MGL is the second mammalian enzyme with PLA 2 and lipase activities. Interestingly, both enzymes hardly show PLA 1 activity, which is often associated with lipases (21).
PA-PLA 2 /MGL displayed broad pH profiles with acidic pH optimum (5.5) toward PA, an acidic substrate, and with alkaline one (8.5) toward monoacylglycerol, a neutral substrate (Fig. 2B). MAFP, a potent irreversible inhibitor for lipases and PLA 2 s with an active site serine, strongly inhibited the two lipolytic activities of PA-PLA 2 /MGL (Table II) at a concentration of 50 M. This suggests that a serine nucleophile participates in its catalysis, which is further supported by the identity of its NH 2 -terminal amino acid sequence to a serine carboxyesterase. In this mechanism an active site base, usually an unprotonated His N ⑀2 , accepts a proton from the Ser O ␥ to facilitate its concerted nucleophilic attack on a carbonyl carbon at a sessile bond. Because a pK a of the catalytic His is typically around 7 in serine enzymes (22), lowering pH to 5.5 causes its protonation, thereby abolishing activity. This is not compatible with a reaction mechanism for broad pH optimum involving a single active site. However, serine carboxypeptidases including prohormone-processing carboxypeptidases working in the secretory pathway with acidic milieu (see below) and preduodenal acid lipases are active at acidic pH, even though they belong to ␣/␤ hydrolase fold enzymes with a typical "catalytic triad" in their active sites; it still remains elusive how their catalytic His keeps unprotonated at low pH (23,24). Interestingly, the carboxypeptidases exhibit pH activity profiles similar to that of PA-PLA 2 /MGL, acidic pH optimum toward peptide substrates with a negative charge at COOH terminus (versus a phosphate group of PA) and alkaline one toward neutral ester substrates (23). Alternatively, different active sites might be responsible for the two activities. Lung PLA 2 with an active site Ser is active at acidic pH with an optimal pH of ϳ4, whereas it exhibits glutathione peroxidase activity through a different active site at alkaline pH (25). Whichever mechanism operates in PA-PLA 2 /MGL, it will provide an additional good model for studying the mechanism of enzymes with broad pH optimum.
The amino acid sequence of the first 12 NH 2 -terminal residues of the purified enzyme was identical to that of rat liver carboxylesterase ES-10 (19) belonging to a CES1A subfamily of serine carboxyesterase with broad substrate specificity (26). Indeed, PA-PLA 2 /MGL exhibited greater esterase activities toward p-nitrophenyl esters of both short and long chain fatty acids than monoacylglycerol lipase and PLA 2 activities. The ES-10 esterase reportedly exhibited monoacylglycerol lipase activity, and broad pH 6 -10 optimum (27), consistent with the catalytic properties of PA-PLA 2 /MGL but had not been tested for PLA 2 activity toward acidic phospholipids including PA and LBPA at pH 5.5. Significant differences between the properties of the two enzymes should, however, be noted; unlike PA-PLA 2 / MGL, the ES-10 esterase is not active toward monoacylglycerol with a long chain acyl group and behaves as trimers on gel chromatography, a characteristic of the ES-10 esterase. There exists the ES-10 species with only a few amino acid substitutions, and this microheterogeneity in sequence caused it to switch substrate specificity. Rat lung CES1A esterase and liver cholesterol esterase differ in sequence by only four residues, three of which were important to confer cholesterol esterase activity on the lung enzyme as revealed by site-directed mutagenesis (28). Recent x-ray crystallography of rabbit CES2 esterase suggested not only the importance of the equivalent residues in maintaining substrate specificity but also the participation of a sugar chain in the binding site of the acyl intermediate (29). Moreover, rat liver ES-10 esterase functions as a triacylglycerol lipase that mobilizes cytoplasmic triacylglycerol stores on lipoprotein assembly (30), whereas PA-PLA 2 / MGL hardly reacted with triacylglycerol (Fig. 3B). Hence, intensive protein structural analysis by mass spectrometry will be required to detect a few amino acid substitutions and posttranslational modifications including glycosylation in PLA 2 / MGL, compared with the ES-10 enzyme, helping us clarify its mechanisms of controlling substrate specificity.
Recently, several PLA 2 s that are optimally active at acidic pH have been characterized (5,25,31,32). The lung 26-kDa PLA 2 with the peroxidase activity at alkaline pH as mentioned above can hydrolyze dipalmitoyl phosphatidylcholine, a lung surfactant, and showed a more acidic pH optimum (4ϳ4.5), although its selectivity for acidic phospholipids has not been tested (25). A 45-kDa PLA 2 partially purified from macrophage-like RAW 264.7 cells preferred PG, but not PA, and has a weak but significant PLA 1 activity (31). These two PLA 2 s differed considerably from PA-PLA 2 /MGL in substrate specificity and molecular mass (see Fig. 1 and Fig. 3). An ϳ58-kDa PApreferring PLA 2 purified from rat brain (5) exhibited pH 6.0 optimum for hydrolyzing PA and is rather similar to PA-PLA 2 / MGL, but significant differences should be noted; unlike PA-PLA 2 /MGL, the brain enzyme did not prefer unsaturated acyl groups at sn-2 position at all and strongly binds to cationexchange and heparin-liganded gels, suggesting its more basic nature (5). It was not tested whether the brain enzyme hydrolyzed monoacylglycerol. An ϳ40-kDa PLA 2 with an optimum pH of 4.5 purified from calf brain catalyzed transacylation of the acyl group at sn-2 position of phosphatidylcholine or phosphatidylethanolamine to 1-hydroxyl group of N-acetyl ceramide, as well as PLA 2 reaction. This enzyme is different from PA-PLA 2 /MGL in substrate specificity, reactivity with serinemodifying irreversible inhibitors, net charge on the protein, and molecular mass (32).
The testis expresses the enzymes that have been purified and cloned and display the activities related to those of PA-PLA 2 /MGL, i.e. 97.6-kDa PA-preferring PLA 1 (9, 10) and 33.2-kDa monoacylglycerol lipase (33). Apparent molecular mass (63 kDa) of PA-PLA 2 /MGL is quite different from those of the latter two enzymes. These three enzymes do not require Ca 2ϩ ions for activity but show substrate specificities different from one another. PA-PLA 2 /MGL exhibited high monoacylglycerol lipase and PLA 2 activities with a specific activity toward the mixed micelles of taurocholate and either monooleoylglycerol (8.21 mol/min/mg) or POPA (0.43 mol/min/mg). In contrast, PApreferring PLA 1 with a specific activity of 2.95 mol/min/mg toward PA did not exhibit PLA 2 activity (9), and monoacylglycerol lipase with a specific activity of 350 mol/min/mg toward monooleoylglycerol did not exhibit appreciable PLA 2 , tri-and diacylglycerol lipase and lysophospholipase activities (34).
There are several biosynthetic pathways of LPA as mentioned in the Introduction, and the major pathway varies from tissue to tissue. One of the major sources of LPA in the serum was reportedly platelet in which the combined action of phospholipase C and kinase, a PLA 2 -independent pathway, mainly contributes to LPA synthesis (35), but a study with a PLA 2 inhibitor suggested for involvement of a PLA 2 -dependent pathway in LPA synthesis, depending on molecular species of LPA (36). We found that human platelets contained PA-PLA 2 /MGLlike activity, PLA 2 activities toward POPA of 0.21 and 0.41 nmol/min/mg at pH 8.5 and 5.5, respectively, and the activity hydrolyzing monoleoylglycerol of 3.34 nmol/min/mg at pH 8.5. Note that the PLA 2 specific activity at pH 5.5 was greater than at pH 8.5, suggesting that this PA-PLA 2 /MGL-like activity could be involved more significantly in producing LPA than PLA 2 being active at alkaline pH. Under the assay conditions used group IIA PLA 2 , which prefers anionic phospholipids including PA (4,8), accounted for most of the measurable PLA 2 activity at pH 8.5 in platelet homogenate (8). PA-PLA 2 /MGL is a candidate for enzymes producing LPA in the testis that was one of the rich sources of LPA (37).
This study provides evidence for the ability of PA-PLA 2 /MGL to hydrolyze LBPA, a unique anionic phospholipid localized specifically in intravesicular vesicles of late endosomes (11), at the optimal pH of 5.5, which is interestingly the same as a typical pH value of the lumen of late endosomes (38). LBPA is unstable because of 2,3-acyl migration catalyzed by acid or base and chromatographic silica supports. This hinders isolation and structural determination of its natural form(s) (39). In this study we used LBPA-rich lipids, extracted from lysosomal frac-tions of chloroquine-treated rat liver and then hydrolyzed by group IB PLA 2 , as substrate for assaying LBPA-hydrolyzing activity without purifying LBPA. The molecular species of phospholipids included in the substrates and products were identified by microbore HPLC/ESI ion-trap mass spectrometry, and fatty acids released from LBPA by the action of PA-PLA 2 / MGL were determined by the ADAM method (8). Although this method has its inherent limitation that other lipids included in the substrate might affect PA-PLA 2 /MGL activity, it can avoid acyl migration and be used to measure activity toward the natural form of LBPA. Both chiral centers of the bisglycerolphosphate scaffold of LBPA have S absolute configurations opposite those of usual, natural phospholipids (13). LBPA has hitherto been thought to be resistant to PLA 2 actions, because secretory PLA 2 s used widely for structural studies of phospholipids specifically recognize R configuration of phospholipids but not the other configuration. PA-PLA 2 /MGL is the first PLA 2 that hydrolyzes LBPA at the physiological pH of the lumen of late endosomes.
As to the acyl chain location of LBPA, namely at sn-2 (2Ј) or sn-3 (3Ј) positions, controversial results have been reported (40,41). A recent study suggested that rat testis LBPA fractions contained three stereoisomeric species with different acyl chain locations, which could be separated reproducibly on a normal phase NH 2 -silica column (42). We observed broad and unresolved multiple peaks of LBPA on a silica column, but in contrast, peak shapes varied from experiment to experiment, suggesting for acyl migration catalyzed by silica gels during chromatography (39). The sn-2 specificity of PA-PLA 2 /MGL for phospholipids and a considerable decrease in the activity during storage of LBPA-rich lipids presumably because of acyl migration suggested that the sessile acyl chain of LBPA is located at sn-2 position.
Chloroquine, a lysosomotropic drug, is a weak base that causes alkalization of endosome/lysosome lumen where its protonated forms are trapped. In addition it directly inhibits H ϩ v-ATPase (43), facilitating more alkalization. The drug also inhibited lysosomal PLA 1 (44), but LBPA was not a substrate for the PLA 1 (45). In contrast, the drug did not inhibit PA-PLA 2 /MGL up to the concentration of 0.5 mM (data not shown). This alkalization to decrease PA-PLA 2 /MGL activity is therefore one of the mechanisms for accumulation of LBPA by treatment with lysosomotropic drugs (14).
There is a gradient in pH of the lumen of organelles along the secretory pathway from endoplasmic reticulum (pH 7.4) through trans-Golgi network (pH 6.2) to secretory granules (pH 5.5) (46) or along the endocytotic pathway from early endosomes (pH 6.6) through late endosomes (pH 5.5) to lysosomes (pH 4.5-5.0) (38). The pH 5.5 optimum of PA-PLA 2 /MGL for PA and LBPA hydrolysis suggests that it can function in these acidic compartments. Its pH activity profile, however, extends from acidic to alkaline regions (Fig. 2B), and monoacylglycerol lipase activity is much greater than PLA 2 activity at neutral pH. Hence, the enzyme can function in the neutral environments of cells, and it should be examined whether monoacylglycerol lipase activity of PA-PLA 2 /MGL functions in vivo. It is important to determine the precise subcellular localization and physiologically relevant substrate of an enzyme with broad substrate specificity.