Highly Selective Hydrolysis of Fatty Acyl-CoAs by Calcium-independent Phospholipase A2β

Calcium-independent phospholipase A2 β (iPLA2 β) participates in numerous diverse cellular processes, such as arachidonic acid release, insulin secretion, calcium signaling, and apoptosis. Herein, we demonstrate the highly selective iPLA2β-catalyzed hydrolysis of saturated long-chain fatty acyl-CoAs (palmitoyl-CoA ≈ myristoyl-CoA ≫ stearoyl-CoA ≫ oleoyl-CoA ≈ arachidonoyl-CoA) present either as monomers in solution or guests in host membrane bilayers. Site-directed mutagenesis of the iPLA2β catalytic serine (S465A) completely abolished acyl-CoA thioesterase activity, demonstrating that Ser-465 catalyzes both phospholipid and acyl-CoA hydrolysis. Remarkably, incubation of iPLA2β with oleoyl-CoA, but not other long-chain acyl-CoAs, resulted in robust stoichiometric covalent acylation of the enzyme. Moreover, S465A mutagenesis or pretreatment of wild-type iPLA2β with (E)-6-(bromomethylene)-3-(1-naphthalenyl)-2H-tetrahydropyran-2-one unexpectedly increased acylation of the enzyme, indicating the presence of a second reactive nucleophilic residue that participates in the formation of the fatty acyl-iPLA2β adduct. Radiolabeling of intact Sf9 cells expressing iPLA2β with [3H]oleic acid demonstrated oleoylation of the membrane-associated enzyme. Partial trypsinolysis of oleoylated iPLA2β and matrix-assisted laser desorption ionization mass spectrometry analysis localized the acylation site to a hydrophobic 25-kDa fragment (residues ∼400–600) spanning the active site to the calmodulin binding domain. Intriguingly, calmodulin-Ca2+ blocked acylation of iPLA2β by oleoyl-CoA. Remarkably, the addition of low micromolar concentrations (5 μm) of oleoyl-CoA resulted in reversal of calmodulin-mediated inhibition of iPLA2 β phospholipase A2 activity. These results collectively identify the molecular species-specific acyl-CoA thioesterase activity of iPLA2 β, demonstrate the presence of a second active site that mediates iPLA2 β autoacylation, and identify long-chain acyl-CoAs as potential candidates mediating calcium influx factor activity.

Multiple fatty acyl-CoA thioesterases have been purified from mammalian cytosol, peroxisomes, and mitochondria and have been cloned and characterized with respect to substrate selectivity, enzyme kinetics, and sensitivity to various inhibitors (27). In general, fatty acyl-CoA thioesterases have been classified as those that are induced by peroxisome proliferators (Type-I or Type-II thioesterases) and those that do not share significant sequence homology with these isoforms (27). Several other mammalian enzymes, such as lysophospholipases (28), secretory phospholipase A 2 (29), and palmitoyl-protein thioesterases (30 -32), have also been shown to exhibit acyl-CoA hydrolase activity. Interestingly, hepatocyte nuclear factor-4␣ has been recently demonstrated to hydrolyze fatty acyl-CoAs, followed by binding of the fatty acid product to hepatocyte nuclear factor-4␣, thereby allowing cross-talk between the acyl-CoA and free fatty acid binding domains (33).

EXPERIMENTAL PROCEDURES
Construction of Recombinant Baculoviruses Encoding His-tagged Wild-type and S465A Mutant iPLA 2 ␤ Proteins-cDNA encoding wildtype C. griseus (Chinese hamster) iPLA 2 ␤ (9) was utilized as template for PCR amplification of the sequence to introduce six in-frame His codons followed by a stop codon and XhoI restriction site (3Ј primer) for subcloning into pFASTBac1. The 5Ј primer contained a SalI site and Kozak sequence (GCCACC) 5Ј of the ATG start codon. The His-tagged S465A iPLA 2 ␤ construct was created by PCR amplification of the mutant cDNA (9) utilizing a 5Ј primer containing an EcoRI restriction site and a 3Ј primer encoding a His 6 tag followed by a SalI site for subcloning into pFASTBac1. The 2.4-kb products were each inserted into a bacmid shuttle vector and sequenced in both directions to confirm the integrity of the constructs. S. frugiperda (Sf9) cells in 35-mm plates containing supplemented Grace's medium were transfected and incubated at 27°C for 72 h to produce high titer baculoviral stocks for infection of Sf9 cells.
Expression and Affinity Purification of iPLA 2 ␤His 6 from Sf9 Cells-Sf9 cells were grown as 100 ml of suspended cultures (in 1-liter plastic bottles) in supplemented Grace's insect medium containing 10% heatinactivated fetal bovine serum and 0.1% Pluronic F-68. Following infection of 3 ϫ 100-ml cultures of Sf9 cells (1.5 ϫ 10 6 cells/ml) with baculovirus encoding iPLA 2 ␤His 6 for 48 h, cells were harvested by centrifugation (250 ϫ g for 10 min), washed once with Grace's insect medium without serum, and resuspended in 30 ml of 25 mM sodium phosphate, pH 7.8, 20% glycerol, 2 mM ␤-ME, 5 g/ml aprotinin, 5 g/ml leupeptin. After lysing the cells by sonication (30 1-s bursts), the homogenate was centrifuged at 100,000 ϫ g for 1 h to obtain the cytosol, to which NaCl was added to a final concentration of 250 mM. The cytosol was then mixed by inversion with 3 ml of HIS-Select-Co 2ϩ affinity resin (Sigma) for 1 h, and the cytosol-resin suspension was poured into an Amersham Biosciences 1 ϫ 10-cm column. Following washing of the settled resin with 30 ml of Buffer A (25 mM sodium phosphate, pH 7.8, containing 500 mM NaCl, 20% glycerol, and 2 mM ␤-ME), bound protein was eluted from the column at a flow rate of 0.25 ml/min utilizing a 250 mM imidazole gradient in Buffer A (50-ml total volume) generated using an Amersham Biosciences fast protein liquid chromatography system. Column fractions were assayed for iPLA 2 activity as described below, pooled, and dialyzed overnight against Buffer B (25 mM imidazole, pH 7.8, containing 20% glycerol, 1 mM dithiothreitol, and 1 mM EGTA). The dialyzed sample was applied to a 2.5-ml column of ATP-agarose equilibrated with Buffer B and washed with Buffer B containing 1 mM AMP and 50 mM NaCl. Bound iPLA 2 ␤His 6 was eluted with Buffer B containing 2 mM ATP and 100 mM NaCl, dialyzed against Buffer B (EGTA concentration was reduced to 0.1 mM) containing 100 mM NaCl, flash frozen in liquid nitrogen, and stored at Ϫ80°C. Approximately 1 mg (65% yield) of iPLA 2 ␤His 6 with a specific activity of 0.5 mol of oleic acid/min/mg, utilizing 5 M [ 14 C]POPC as substrate, was typically recovered from 300 ml of cultured Sf9 cells by this procedure.
Phospholipase A 2 and Acyl-CoA Hydrolase Enzymatic Assays-Purified recombinant iPLA 2 ␤His 6 (0.1-1 g) was incubated with radiolabeled phospholipid or acyl-CoA in 25 mM Tris-HCl, pH 7.2, containing 1 mM EGTA (200 l final volume) for 1-2 min at 37°C. In experiments using acyl-CoAs as guests in host phospholipid bilayers, radiolabeled acyl-CoAs were incorporated into POPC/DOPS (90:10 mol %) vesicles (100 M) before addition to the reaction mix. Long-chain acyl-CoAs have been previously demonstrated to integrate into lipid bilayers within seconds (37). Reactions were terminated by extraction of the released radiolabeled fatty acids into 100 l of butanol, separation of fatty acids from unreacted substrate by thin layer chromatography, and quantitation by scintillation spectroscopy as previous described (38). For experiments examining the effects of acyl-CoAs on calmodulinmediated inhibition of iPLA 2 ␤, phospholipase A 2 activity was continuously measured utilizing a SPECTRAmax GEMINI XS Dual-Scanning Microplate Spectrofluorometer (Molecular Devices). BODIPY-PC substrate (1.17 mM in Me 2 SO, 5 M final concentration) was co-sonicated (10 min at 40% power, 50% duty cycle) with POPC (95 M final concentration) in 25 mM HEPES, pH 7.2. Oleoyl-CoA and CaCl 2 were added at the indicated concentration to the lipid vesicles before addition to iPLA 2 ␤ with or without CaM (preincubated on ice for 10 min) present in individual wells of a black 96-well microtiter plate. Fluorescence readings were acquired at 15-s intervals for 2 min at 37°C utilizing excitation/emission wavelengths of 495/515 nm, respectively.
Electrospray Ionization (ESI)/Mass Spectral (MS) Analyses-Purified iPLA 2 ␤ (0.25 g) was incubated with POPC (95 M) vesicles containing BODIPY-PC (5 M) for 5 min at 37°C as described above. The reactions were stopped by the addition of 4 ml of chloroform/methanol (1:1, v/v) containing internal standards (i.e. 14:1-14:1 PC and 17:0 LPC), and the lipid species were extracted into the chloroform layer as described previously (39). Extracted lipid samples were filtered with Millex-FG 0.20-m filters (Millipore, Bedford, MA) and were routinely stored in 200 l of chloroform/methanol (1/1, v/v) under nitrogen at Ϫ20°C. ESI/MS analysis was performed using a Thermo Finnigan TSQ Quantum Plus spectrometer (San Jose, CA) equipped with an electrospray ion source as previously described in detail (40,41). Samples were diluted with chloroform/methanol (1:1, v/v) (ϳ5 pmol/l) prior to direct infusion into the ESI ion source at a flow rate of 4 l/min with a 2-min period of signal averaging. Tandem mass scanning of neutral loss of 59 atomic mass units (loss of trimethylamine) was performed at a collision energy of 24 eV and a collision gas pressure of 1.0 millitorr.
Covalent Modification of iPLA 2 ␤ with 14 C-labeled Long-chain Acyl-CoAs-Purified recombinant iPLA 2 ␤His 6 was incubated with POPC vesicles (90 M) containing 10 mol % [1-14 C]acyl-CoA for 1 h at 37°C. In some experiments, iPLA 2 ␤His 6 was preincubated with BEL (3 min at 23°C), N-ethylmaleimide (5 min at 30°C) or iodoacetamide (5 min at 30°C) prior to the addition of radiolabeled acyl-CoA. Chloroform/methanol precipitation of some samples was performed as described (42), utilizing 15 g of bovine serum albumin as carrier. In experiments to examine the nature of the covalent linkage between oleic acid and iPLA 2 ␤, acid (HCl), base (NaOH), and hydroxylamine were added to the indicated concentrations, and the samples were incubated at 30°C for 1 h. Bovine serum albumin (15 g) and SDS-PAGE loading buffer were then added to each sample prior to dialysis against 50 mM Tris-HCl, pH 6.8, containing 10% glycerol and 1% SDS for 4 h. Samples were electrophoresed by SDS-PAGE, fixed (40% methanol containing 10% acetic acid), stained with Coomassie Blue R-250, incubated in Amplify fluorographic reagent, dried, and exposed to Eastman Kodak Co. Biomax MR film for 2-5 days at Ϫ80°C.
Partial Trypsinolysis of Oleoylated iPLA 2 ␤-Purified iPLA 2 ␤His 6 (10 M) was incubated with 50 M [1-14 C]oleoyl-CoA or unlabeled oleoyl-CoA in 25 mM imidazole, pH 7.8, containing 50 mM NaCl, 0.1 mM EGTA, 1 mM dithiothreitol, and 20% glycerol for 1 h at 37°C. Excess [1-14 C]oleoyl-CoA was removed by using a Micro Bio-Spin (Bio-Rad) column equilibrated with the above buffer. Recovered iPLA 2 ␤ was partially digested with trypsin (1:25, w/w) for 1-30 min at 37°C. Tryptic peptides were separated by SDS-PAGE, fixed in 40% methanol, 10% glacial acetic acid, stained with Coomassie Blue, and destained in the fixation solution. Gels containing the radiolabeled peptide fragments were soaked in Amplify fluorogenic reagent (Amersham Biosciences), dried, and exposed to film. In parallel samples utilizing unlabeled oleoyl-CoA, the band corresponding to the ϳ25-kDa radiolabeled fragment was excised, cut into ϳ1 ϫ 1-mm pieces, and destained further by washing with 50% acetonitrile at 37°C. The gel pieces were then dried in a SpeedVac, resuspended in 50 mM ammonium bicarbonate (100 l) containing 0.5 g of sequencing grade modified trypsin (Promega), and incubated for 12 h at 37°C. After aliquoting the supernatant solution to a separate tube, residual peptides in the gel pieces were extracted into 50% acetonitrile, 20% isopropyl alcohol, 0.1% trifluoroacetic acid, combined with the supernatant solution, and concentrated utilizing a SpeedVac.
MALDI-TOF Analysis of iPLA 2 ␤ Tryptic Fragments-Concentrated peptide samples were diluted with 0.5% trifluoroacetic acid, absorbed to a C18 Zip-Tip (Millipore), and desorbed with a solution composed of 50% acetonitrile, 20% isopropyl alcohol, 0.1% trifluoroacetic acid, and containing in addition 5 mg/ml ␣-cyano-4-hydroxycinnamic acid. Samples were applied to 192-spot sample plates (ABI) and allowed to air-dry. MS analyses were performed utilizing an Applied Biosystems 4700 Proteomics Analyzer (Framingham, MA), which possesses a 200-Hz Nd:YAG laser operating at 355 nm. The mass accuracy of the instrument was externally calibrated to the 4700 Proteomics analyzer calibration mixture of peptides. For MALDI-MS analysis, spectra were obtained by the accumulation of 2500 consecutive laser shots at a collision energy of 1 kV with air serving as the collision gas. Calculations of predicted peptide and peptide fragment masses were performed using programs developed at the University of California, San Francisco, Mass Spectrometry Facility (available on the World Wide Web at prospector.ucsf.edu).
Other Procedures-SDS-PAGE was performed according to Laemmli (43). ECL Western analyses for iPLA 2 ␤His 6 were performed utilizing a monoclonal anti-His antibody (BD Biosciences) in conjunction with an anti-mouse IgG-horseradish peroxidase conjugate. Silver staining of SDS-polyacrylamide gels was performed as described (44). Protein concentration was determined by a version of the Bradford protein assay (Bio-Rad) with bovine serum albumin as a standard.

iPLA 2 ␤ Hydrolyzes Palmitoyl-CoA and Is Inhibited at Submicellar
Concentrations of Substrate-Due to the structural similarity between ATP and the 3Ј-phosphoadenosine moiety of CoA, we hypothesized that iPLA 2 ␤ could bind to, and potentially hydrolyze, the thioester linkage of long-chain fatty acyl-CoAs. Accordingly, we overexpressed iPLA 2 ␤His 6 in Sf9 cells and purified the enzyme to apparent homogeneity (as determined by SDS-PAGE/silver staining) by sequential cobalt and ATP affinity chromatographies as described under "Experimental Procedures." Initial assays with iPLA 2 ␤ utilizing supramicellar concentrations of palmitoyl-CoA (100 M) typically used for acyl-CoA thioesterases revealed very low rates of hydrolysis (Fig. 1A). Remarkably, robust palmitoyl-CoA thioesterase activity was demonstrated at low micromolar concentrations of substrate with a maximal rate of ϳ250 nmol of palmitic acid/min/mg at 2.5 M palmitoyl-CoA (Fig. 1A). Similar acyl-CoA-mediated substrate inhibition was observed in previous studies of a purified rabbit heart mitochondrial thioesterase (45) and peroxisomal acyl-CoA thioesterase 2 (46). It should be noted that significant inhibition of iPLA 2 ␤ by palmitoyl-CoA occurs below the critical micelle concentration (10 -20 M) (47) and implies the presence of a second acyl-CoA binding site on the enzyme (i.e. SES intermediate) (48).
iPLA 2 ␤ Catalyzes the Highly Selective Hydrolysis of Myristoyl-CoA and Palmitoyl-CoA Present as Guests in Host Phospholipid Bilayers-Since iPLA 2 ␤ would probably be expected to encounter acyl-CoAs in a membrane bilayer in vivo, we examined whether the enzyme could hydrolyze palmitoyl-CoA present as a guest (at a low mol %) in phospholipid host vesicles. Purified iPLA 2 ␤ effectively hydrolyzed palmitoyl-CoA at physiologically relevant concentrations of acyl-CoA (i.e. 1-5 mol %) present in POPC/DOPS vesicles (Fig. 1B). This was surprising, since POPC is an excellent substrate for iPLA 2 ␤ and would be expected to efficiently compete with the palmitoyl-CoA as substrate. To deter-mine the acyl-CoA substrate selectivity of iPLA 2 ␤, incubations were performed with a series of different long-chain acyl-CoAs in host POPC/DOPS bilayers. A dramatic selectivity for myristoyl-and palmitoyl-CoA was observed (up to 20-fold) in comparison with stearoyl-, oleoyl-, and arachidonoyl-CoAs (Fig. 2). Thus, iPLA 2 ␤ displays a marked preference for hydrolysis of saturated acyl-CoA substrates (14 -16 carbons in length) in comparison with longer unsaturated acyl-CoA molecular species (C18:1 and C20:4) in the presence of membrane bilayers.
Effect of pH and Determination of the Active Site Nucleophile Mediating Palmitoyl-CoA Hydrolysis-Hydrolysis of either POPC or palmitoyl-CoA was examined over a wide pH range (5.5-8.5) to determine potential mechanistic differences between the two substrates relative to the surrounding hydrogen ion concentration. A plateau of maximal activity was observed with both substrates between pH 6.5 and 7.5, indicating a similar iPLA 2 ␤-mediated hydrolytic mechanism for POPC and palmitoyl-CoA (data not shown).
Site-directed mutagenesis of the lipase catalytic serine (GTS 465 TG) of iPLA 2 ␤ to alanine has been previously demonstrated to ablate PLA 2 activity (10). To determine if Ser-465 was equally crucial for iPLA 2 ␤ acyl-CoA thioesterase activity, we generated and purified the identical mutant (S465A) and compared the PLA 2 and palmitoyl-CoA hydrolase activities with those of its wild-type counterpart. Importantly, the S465A iPLA 2 ␤His 6 bound to ATP-agarose (as determined by Western analysis), indicating that the mutant protein was properly folded around the nucleotide binding site ( 431 GGGVKG 436 ), which is ϳ30 amino acid residues away from the lipase site. As expected, calcium-independent PLA 2 activity was abolished in the S465A mutant utilizing POPC as substrate (Fig. 3A). Hydrolysis of palmitoyl-CoA incorporated into POPC/DOPS vesicles also was virtually eliminated in the S465A mutant (Fig. 3A), indicating that the iPLA 2 ␤ active site serine hydroxyl probably serves as the primary nucleophile for both phospholipase A 2 and acyl-CoA thioesterase reactions catalyzed by iPLA 2 ␤.
Chiral Mechanism-based Inhibition of Acyl-CoA Hydrolysis by (R)and (S)-BEL-In previous work, we have demonstrated that racemic BEL potently inhibits both iPLA 2 ␤ (IC 50 ϳ0.2 M) and iPLA 2 ␥ (IC 50 ϳ3 M) phospholipase A 2 activities (11,49,50). Through resolving the enantiomers of BEL by chiral high pressure liquid chromatography, we have further shown that (S)-and (R)-BEL are selective for iPLA 2 ␤ and iPLA 2 ␥, respectively (36). To determine if (R)-and (S)-BEL had similar effects on iPLA 2 ␤ palmitoyl-CoA thioesterase activity, iPLA 2 ␤His 6 was preincubated with each BEL enantiomer prior to the addition of radiolabeled palmitoyl-CoA incorporated into POPC/DOPS bilayers. As seen in Fig. 3B, (S)-BEL inhibited iPLA 2 ␤ palmitoyl-CoA hydrolase activity with an IC 50 of ϳ0.1 M, whereas (R)-BEL was ϳ8-fold less effective (IC 50 ϭ 0.8 M). Thus, the selectivity of the BEL enantiomers for inhibiting iPLA 2 ␤ palmitoyl-CoA thioesterase activity is virtually identical to that previously observed for inhibition of PLA 2 activity (36). Collectively, these results suggest that both long-chain acyl-CoA and phospholipid substrates utilize the same mechanism and hydrolytic site (binding domain and catalytic residue(s)) in iPLA 2 ␤ for catalysis.
In order to determine the stoichiometry of iPLA 2 ␤ acylation with oleoyl-CoA, we compared the radiolabeling intensity of iPLA 2 ␤ incubated with increasing amounts of [1-14 C]oleoyl-CoA with that of a standard curve of [1-14 C]methyl-human serum albumin of known specific activity (Fig. 4B). Approximately 1 mol of [1-14 C]oleic acid was incorporated per mol of iPLA 2 ␤ in the presence of POPC vesicles containing up to a 5-fold molar excess of [ 14 C]oleoyl-CoA relative to iPLA 2 ␤. One potential consequence of iPLA 2 ␤ oleoylation is alteration of catalytic activity, either toward phospholipid or acyl-CoA substrates. To address this possibility, iPLA 2 ␤ was incubated with or without oleoyl-CoA and then purified by Co 2ϩ affinity chromatography to remove residual oleoyl-CoA. Results from these experiments indicated that oleoylation did not significantly affect either iPLA 2 ␤-catalyzed POPC or palmitoyl-CoA hydrolysis (data not shown). Thus, iPLA 2 ␤ autoacylation with oleoyl-CoA occurs at a site that does not block accessibility of substrate or inhibit release of products from the active site.
To further establish that the modification of iPLA 2 ␤ with oleoyl-CoA was covalent, [1-14 C]oleoyl-iPLA 2 ␤ was precipitated with CHCl 3 / CH 3 OH and extensively washed with 70% acetone before SDS-PAGE (Fig. 4C). This treatment did not result in an appreciable decrease in signal intensity, indicating that the [1-14 C]oleate is covalently bound to iPLA 2 ␤. Since esterification of fatty acids to proteins has been shown to occur through either amide, oxyester, or thioester bonds that can be distinguished through chemical treatment with strong acid (HCl), strong base (NaOH), or neutral hydroxylamine, additional experiments  were incorporated into host POPC (100 M) vesicles and incubated for 1 h at 37°C with equal amounts of either wild-type (WT) iPLA 2 ␤His 6 , S465A iPLA 2 ␤His 6 , or BEL-pretreated wild-type iPLA 2 ␤His 6 enzyme as described under "Experimental Procedures." Samples were resolved by SDS-PAGE (10% gel), fixed, and dried before visualization by autoradiography. B, purified iPLA 2 ␤His 6 (2 M) was incubated with the indicated concentrations of [1-14 C]oleoyl-CoA present as guest in host POPC (100 M) vesicles for 1 h at 37°C. Samples were electrophoresed in parallel with standard amounts (0.5-10 nCi) of [methyl-14 C]bovine serum albumin of known activity (not shown). The fixed and dried gel was exposed to film, and the resultant signals from the [methyl-14 C]bovine serum albumin were quantified utilizing an Eastman Kodak Co. Imagestation (1D software) to generate a standard curve to determine the incorporation of [1-14 C]oleate into iPLA 2 ␤. C, purified iPLA 2 ␤His 6 was incubated with were performed to determine the nature of the covalent linkage. In the case of [1-14 C]oleoyl-iPLA 2 ␤, the addition of either 1 N HCl or 1-2 N neutral hydroxylamine did not result in a significant decrease in radiolabeling, whereas the addition of 1 N NaOH completely eliminated the majority of covalently bound [1-14 C]oleate (Fig. 4C). The insensitivity of [1-14 C]oleoyl-iPLA 2 ␤ to hydroxylamine and HCl would indicate the absence of thioester and oxyester linkages, respectively, whereas the disappearance of radiolabeling in the presence of NaOH is consistent with an amide linkage. Pretreatment of iPLA 2 ␤ with N-ethylmaleimide or iodoacetamide decreased [1-14 C]oleoylation (Fig. 4C), indicating that free thiol (cysteine) groups are important for either the formation of oleoyl-iPLA 2 ␤ intermediate(s) or for subsequent transfer to the terminal acceptor residue(s). Interestingly, acylation of iPLA 2 ␤ was not detectable utilizing 1-palmitoyl-2-[1-14 C]oleoyl-sn-glycero-3-phosphocholine (with or without CoASH) (Fig. 4C). This suggests that the acyl-enzyme intermediate formed with POPC, which is readily hydrolyzed by water, is fundamentally distinct from that which leads to the relatively stable iPLA 2 ␤ acyl adduct(s) formed with oleoyl-CoA.
To determine whether oleoylated iPLA 2 ␤ could undergo acyl group exchange in the presence of additional amounts of oleoyl-CoA, we first preincubated iPLA 2 ␤ with a molar excess of unlabeled oleoyl-CoA guest in host POPC vesicles and then monitored incorporation of [1-14 C]oleic acid utilizing [1-14 C]oleoyl-CoA. Remarkably, robust radiolabeling of iPLA 2 ␤ still occurred in a time-dependent manner after preacylation with unlabeled oleoyl-CoA (Fig. 4D), indicating the ability of iPLA 2 ␤ to catalyze acyl cycling in the presence of additional exogenous oleoyl-CoA.
Oleoylation of iPLA 2 ␤ in Sf 9 Cells-Although iPLA 2 ␤ was robustly acylated by oleoyl-CoA in vitro, it remained to be determined if the enzyme could be oleoylated in an intact cell. Accordingly, we incubated control and iPLA 2 ␤His 6 -containing Sf9 cells with [ 3 H]oleic acid for 3 h and analyzed the cytosolic and membrane fractions by SDS-PAGE followed by autoradiography. A unique radiolabeled protein band was observed in the membrane fraction of cells expressing iPLA 2 ␤ but not in the membranes of control cells (Fig. 5, left). A corresponding radiolabeled protein was not observed in the cytosol of cells containing iPLA 2 ␤ (data not shown), indicating that the majority of the acylated protein is localized to cellular membranes. To confirm the identity of the radiola-beled band, Western blotting for iPLA 2 ␤ was performed. A robust immunoreactive band representing iPLA 2 ␤ was detected in Sf9 cell membranes at the same molecular weight as that of the unique radiolabeled band (Fig. 5, right).
Localization of the iPLA 2 ␤ Oleoylated Peptide by Partial Trypsinolysis and Mass Spectrometric Identification of the Oleoylated Domain of iPLA 2 ␤-To initially localize the region of iPLA 2 ␤ modified by oleoyl-CoA, we partially trypsinized [1-14 C]oleoyl-iPLA 2 ␤ in solution and separated the radiolabeled peptide fragments by SDS-PAGE. Results from these experiments revealed that the majority of the radioactivity was contained within a 25-kDa proteolytic fragment (Fig. 6). In-gel tryptic digestion and subsequent MALDI-MS analysis of the 25-kDa polypeptide determined that it encompassed residues 408 -578, which contains both the nucleotide binding domain and the active site (Table 1). Next, we utilized MALDI-MS to examine tryptic digests of the protein, specifically searching for unique peptide peaks that were 264.2 mass units (i.e. C18:1-H 2 O) greater than their respective parent peak. Despite multiple attempts utilizing a wide range of conditions (e.g. in-gel digests, solution digests, multiple proteases, combinations of proteases, organic solvent and detergent extraction/solubilization techniques, etc.), we were unable to identify potential candidate peaks for MALDI-MS/MS analysis despite achieving 70% sequence coverage of iPLA 2 ␤. The addition of oleate to would be expected to increase the calculated water-octanol partition coefficient (log(P) value) of the modified peptide by 2.23, representing a significant increase in nonpolarity (for reference, log(P) for phenylalanine ϭ 1.000). The difficulty in ionization of hydrophobic peptides is well documented (55)(56)(57) and has recently been discussed in detail (58). Potential nucleophilic amino acid residues within the 25-kDa peptide acylated by oleoyl-CoA include Cys-428, Lys-435, Lys-448, and Lys-455 near the nucleotide binding and lipase consensus sequence motifs.
Effect of Calcium-activated Calmodulin on iPLA 2 ␤-mediated Acyl-CoA Hydrolysis-Calcium-bound calmodulin has been previously demonstrated to bind to iPLA 2 ␤ and potently inhibit the phospholipase A 2 activity of the enzyme (22,59). We were therefore interested to determine if Ca 2ϩ -CaM would have a similar effect on the acyl-CoA thioesterase activity of iPLA 2 ␤. Although Ca 2ϩ -CaM inhibited the PLA 2 activity of recombinant iPLA 2 ␤ by ϳ70 -80%, the palmitoyl-CoA thioesterase activity was unaffected under similar conditions (Fig. 7A). Thus, whereas the phospholipase A 2 activity of iPLA 2 ␤ is responsive to changes in intracellular calcium (via calmodulin), iPLA 2 ␤  would be expected to hydrolyze acyl-CoA thioesters independent of calcium concentration or the presence of calmodulin.
Calmodulin-mediated Protection of iPLA 2 ␤ against Oleoylation by Oleoyl-CoA-The proximity of the oleoylated iPLA 2 ␤ 25-kDa tryptic fragment to the calmodulin binding domain next led us to investigate whether Ca 2ϩ -CaM could protect the enzyme against covalent acylation by oleoyl-CoA. Although the addition of either Ca 2ϩ or CaM in the presence of EGTA alone did not alter the extent of oleoylation of iPLA 2 ␤ (Fig. 7B), the combination of Ca 2ϩ and CaM significantly decreased autoacylation of the enzyme. From these results, acyl-CoAmediated acylation would be predicted to primarily occur after dissociation of the iPLA 2 ␤⅐CaM complex.
Oleoyl-CoA-mediated Reversal of the Inhibition of iPLA 2 ␤ by Calmodulin-Depletion of intracellular calcium stores has been previously shown to initiate the influx of extracellular calcium through a poorly understood process, potentially involving an unknown cellular metabolite referred to as calcium influx factor (CIF) (25,26,60). Recent studies have demonstrated that CIF activates iPLA 2 ␤ by reversal of Ca 2ϩ -calmodulin inhibition of the enzyme (19,20,24). To determine if acyl-CoA could mitigate the inhibition of iPLA 2 ␤ by CaM, we utilized a real time fluorescence assay employing the PLA 2 substrate, BODIPY-PC, to measure the kinetic effects of oleoyl-CoA (guest in POPC (95 mol %)/BODIPY-PC (5 mol %) host vesicles) on CaM inhibition of iPLA 2 ␤ phospholipase A 2 activity. In the absence of calmodulin, iPLA 2 ␤ efficiently hydrolyzes BODIPY-PC present at 5 mol % in a POPC background as demonstrated by a robust time-dependent increase in fluorescence intensity (Fig. 8A). The presence of calcium ion did not appreciably affect the phospholipase A 2 activity of iPLA 2 ␤ under these conditions (data not shown). In contrast, the presence of Ca 2ϩ -bound CaM inhibited iPLA 2 ␤-catalyzed hydrolysis of BODIPY-PC by ϳ70 -80% (Fig. 8A). Remarkably, the addition of 1 mol % oleoyl-CoA could activate CaM-inhibited iPLA 2 ␤ (ϳ40% of initial activity) (Fig. 8B), and the presence of 2.5-5 mol % oleoyl-CoA completely eliminated CaMmediated inhibition of iPLA 2 ␤ (Fig. 8, C and D) under these conditions. To confirm that iPLA 2 ␤ was in fact hydrolyzing BODIPY-PC and that the increase in fluorescence observed was not due to either proteinfluorophore or acyl-CoA-fluorophore interactions, the reaction substrates and products were extracted into chloroform/methanol in the presence of internal standards and subsequently quantified by shotgun lipidomics (61). As anticipated, the production of lyso-BODIPY-PC, 16:0-LPC, and 18:1-LPC was dependent upon the presence of iPLA 2 ␤, and the amount of each product was diminished (ϳ80%) by the presence of Ca 2ϩ -bound CaM (Fig. 8, E-G). Importantly, the addition of 5 mol % of oleoyl-CoA to the POPC/BODIPY-PC vesicles in the presence of Ca 2ϩ /CaM/iPLA 2 ␤ completely reversed the inhibition of iPLA 2 ␤ by Ca 2ϩ /CaM as evidenced by the recovery of similar amounts of 16:0-LPC, 18:1-LPC, and lyso-BODIPY-PC to that observed with iPLA 2 ␤ alone (Fig. 8H). Finally, to establish that the effects of oleoyl-CoA on Ca 2ϩ /CaM-mediated inhibition of iPLA 2 ␤ were not dependent on the presence of BODIPY-PC, similar experiments were performed with 1-palmitoyl-2-[1-14 C]oleoyl-sn-glycero-3-phosphocholine as substrate. Similar to the real time fluorescence assays, CaM inhibited iPLA 2 ␤ activity by ϳ70%, and the presence of oleoyl-CoA alone caused moderate inhibition (20%) of iPLA 2 ␤ activity (Fig. 8I ), presumably due to interactions with iPLA 2 ␤ at or near the substrate binding site. As anticipated, the presence of oleoyl-CoA significantly attenuated the CaM-mediated inhibition of iPLA 2 ␤, resulting in 75% of the activity observed with oleoyl-CoA alone (Fig. 8I ). Collectively, these results demonstrate the rescue of the calmodulin-inhibited iPLA 2 ␤ activity by oleoyl-CoA by three independent methods and identify fatty acyl-CoAs as potential candidates for calcium influx factor.

fragment tryptic peptides identified by MALDI-MS analysis
Unlabeled oleoylated iPLA 2 ␤His 6 was prepared and partially trypsinized for 10 min as described under "Experimental Procedures." The band corresponding to the 25-kDa radiolabeled fragment in Fig. 8 was excised and in-gel trypsinized, and the resultant peptides were identified by comparison of mass peaks obtained by MALDI-TOF with predicted iPLA 2 ␤His 6 tryptic peptide masses. Letters in parentheses indicate the neighboring amino acid residues in the primary sequence before and after the tryptic cleavage sites.  (23), ankyrin repeats (10,69), splice variants (70,71), proteolytic products (38,72,73), phosphorylation (9,74), and interaction with calmodulin kinase II ␤ (75), each of which serve as potential regulators of the pleiotropic functions of iPLA 2 ␤. In this paper, we demonstrate that iPLA 2 ␤ efficiently hydrolyzes saturated fatty acyl-CoAs at physiologically relevant concentrations, is selectively autoacylated by oleoyl-CoA, is protected from autoacylation by Ca 2ϩ -CaM, and is rescued from calmodulin-mediated inhibition of phospholipase A 2 activity by oleoyl-CoA.
Site-directed mutagenesis of Ser-465 or pretreatment of iPLA 2 ␤ with BEL inhibit both thioesterase and phospholipase A 2 activities to identical degrees, indicating that the same active site and nucleophile (Ser-465) are utilized for both hydrolytic reactions. Remarkably, iPLA 2 ␤ autoacylation by saturated acyl-CoAs is dramatically increased by either mutagenesis of Ser-465 or pretreatment of the enzyme with BEL. One possible explanation for these effects is that a greater effective saturated acyl-CoA concentration for acylation is achieved near the iPLA 2 ␤ active site, since hydrolysis is reduced or eliminated under these conditions. These results demonstrate the existence of a second nucleophilic site(s) (distinct from Ser-465 and the proximal nucleophilic residue(s) modified by BEL) in iPLA 2 ␤. Furthermore, the presence of oleoylated iPLA 2 ␤ in the membranes of an intact cell indicates that acylation may facilitate localization of iPLA 2 ␤ to specific membrane compartments or microdomains.
A combination of partial trypsinolysis and MALDI-MS was utilized to localize the region of acylation to amino acid residues ϳ400 -600 (adjacent to the calmodulin binding domain), which includes the nucleotide and lipase consensus sequence motifs. Thus, acylation by oleoyl-CoA occurs near the catalytic domain of iPLA 2 ␤, although it does not appear to inhibit or block substrate (i.e. palmitoyl-CoA or POPC) access to the active site serine (Ser-465) for catalysis. Interestingly, Ca 2ϩbound CaM blocked covalent acylation of iPLA 2 ␤ by oleoyl-CoA. In contrast, Ca 2ϩ -CaM significantly inhibited only the phospholipase A 2 activity of iPLA 2 ␤, whereas the acyl-CoA thioesterase activity of the 1 PC and 17:0 LPC for phosphatidylcholine and lysophosphatidylcholine species, respectively) and were subjected to ESI tandem mass analysis as described under "Experimental Procedures." Spectra were acquired by scanning for the neutral loss of 59 atomic mass units in the positive ion mode at a collision energy of 24 eV and a collision gas pressure of 1.0 millitorr. The relative abundance of lipid species in each spectrum was normalized to that of the LPC internal standard (I.S., 17:0 LPC). I, purified iPLA 2 ␤His 6 in the presence or absence of Ca 2ϩ /CaM was incubated with 1-palmitoyl-2-[1-14 C]oleoyl-sn-glycero-3-phosphocholine (100 M) host vesicles with or without guest oleoyl-CoA (10 M) for 3 min at 37°C. Radiolabeled fatty acid was extracted into butanol, resolved by TLC, and quantified by scintillation spectrometry as described under "Experimental Procedures." Results represent the average Ϯ S.E. of four separate determinations. enzyme was unaffected under similar conditions. This result suggests that the less sterically bulky hydrophobic portion of the acyl-CoA substrate may have greater access to the iPLA 2 ␤ active site (Ser-465) than the bulkier diacyl phospholipid substrate in the presence of Ca 2ϩ -bound CaM.
To our knowledge, iPLA 2 ␤ is the only intracellular phospholipase A 2 to exhibit substantial amounts of long-chain acyl-CoA thioesterase activity and represents the first acyl-CoA thioesterase identified at the molecular level shown to efficiently hydrolyze membrane-associated acyl-CoAs. In contrast, in vitro assays with purified recombinant cytosolic PLA 2 ␥ (76) and iPLA 2 ␥ 3 in our hands did not detect appreciable long-chain acyl-CoA hydrolase activities. 3 A 54-kDa acyl-CoA hydrolase from rat intestinal microsomes was found to cleave long-chain acyl-CoAs in the presence of phosphatidylcholine vesicles, although the sequence identity of this enzyme has not been described since its original purification (77).
Myocardial ischemia is known to be accompanied by activation of iPLA 2 ␤ (21), although the mechanism(s) contributing to such activation have been the subject of much debate. It is known that iPLA 2 ␤ largely exists as an inhibited complex in transgenic mice, since vast overexpression does not result in substantive increases in phospholipolysis (21). However, induction of myocardial ischemia results in the dramatic activation of the enzyme demonstrated by fatty acid release and lysolipid accumulation. The present results identify a likely mechanism contributing to the activation of iPLA 2 ␤, since acyl-CoA is known to dramatically increase during myocardial ischemia due to the blockade of mitochondrial fatty acid oxidation (78). Accordingly, these results potentially link the activation of iPLA 2 ␤ in hearts during ischemia with that which occurs during the influx of extracellular calcium (see below) in other tissues.
Depletion of intracellular calcium stores in smooth muscle cells has been previously demonstrated to activate iPLA 2 ␤ through a mechanism hypothesized to involve the dissociation of CaM from the enzyme (13,36). Store-operated calcium (cation) channels in the plasma membrane are then activated in response to agonist-stimulated intracellular calcium pool depletion for the purpose of replenishing the emptied calcium stores. Recent work by Bolotina and co-workers has provided additional details of this process by showing that iPLA 2 ␤ is required for activation of store-operated calcium (cation) channels through generation of lysophospholipids (20). Furthermore, the inhibitory complex between CaM and iPLA 2 ␤ could be disrupted by a partially purified preparation of CIF (20). Although attempts to elucidate the molecular identity of CIF over the past 10 years have not been successful, these studies have determined various chemical properties of calcium influx factor. In general, CIF is believed to be a nonprotein, diffusible, phosphorylated "sugar nucleotide," which is resistant to heat, alkaline pH, and protease treatment and is retained on a C18 reverse phase matrix (79). In this work, we demonstrate that oleoyl-CoA is able to mimic the properties of CIF by restoring the phospholipase A 2 activity of Ca 2ϩ / CaM-inhibited iPLA 2 ␤. The acute production of acyl-CoAs due to fatty acid influx or release (e.g. from phospholipids or triacylglycerol) and subsequent thioesterification in specific membrane microenvironments containing complexes of Ca 2ϩ /CaM-inhibited iPLA 2 ␤ would probably be sufficient to mediate iPLA 2 ␤ activation through displacement of calmodulin in a temporally and spatially specific manner. Thus, this process would be predicted to initiate an amplification cascade by a subset of activated iPLA 2 ␤, which initially releases fatty acids (from surrounding phospholipids) for thioesterification to CoA, thereby acti-vating adjacent iPLA 2 ␤⅐Ca 2ϩ /CaM complexes. We specifically point out that many other CIF-like cellular constituents capable of reversing Ca 2ϩ /CaM-iPLA 2 ␤ inhibition may exist and that other membrane components and conditions that occur in vivo (accessory proteins, membrane surface charge and curvature, membrane electrochemical potential, etc.) may facilitate this process.
Multiple acyl-CoA thioesterases have been cloned from mammalian sources and are classified on the basis of their subcellular localization (cytosolic, mitochondrial, or peroxisomal), sequence similarity, and ability to be induced by peroxisome proliferators. The majority of these thioesterases, as well as all known intracellular phospholipases A 2 , contain the canonical GXSXG lipase (esterase) consensus sequence motif. Amino acid sequence alignments of iPLA 2 ␤ with the known mammalian acyl-CoA thioesterases did not reveal any significant sequence homology outside of the GXSXG consensus motif. This is not completely unexpected, given the diversity among the different classes of acyl-CoA thioesterases. Some of the established acyl-CoA thioesterase family members (e.g. mitochondrial acyl-CoA thioesterases and cytosolic acyl-CoA thioesterases) possess conserved putative nucleotide binding sequences (GXGXXG). Interestingly, calcium-independent phospholipase A 2 ␤ displays an acyl-CoA substrate selectivity (C14-C20) similar to the cytosolic Type-I thioesterase (27). In addition, iPLA 2 ␤, like this cytosolic acyl-CoA thioesterase, is not inhibited by high concentrations of CoASH, 3 indicating that these enzymes are probably not involved in "sensing" and regeneration of free CoASH through acyl-CoA hydrolysis, as has been ascribed to peroxisomal acyl-CoA thioesterase-2 (46). Calcium-independent phospholipase A 2 ␤ and other acyl-CoA thioesterases probably serve multiple pleotropic roles in metabolism and cellular signaling.
In this study, we demonstrate that purified recombinant iPLA 2 ␤ possesses robust palmitoyl-CoA hydrolase activity in addition to its previously well characterized lysophospholipase and phospholipase A 2 activities. Moreover, iPLA 2 ␤ is covalently autoacylated in a highly substrate-specific fashion (by oleoyl-but not palmitoyl-CoA), which occurs at a second active site distinct from the hydrolytic lipase site (GXSXG). Thus, iPLA 2 ␤ could potentially have multiple effects on the production of lipid metabolites (arachidonic acid and lysolipids) or alternatively through removal of saturated acyl-CoAs from specific cellular membrane compartments. Importantly, calcium-independent phospholipase A 2 ␤ is present in multiple subcellular compartments, most notably plasma and nuclear membranes, cytosol, and mitochondria (68,74,80). Collectively, these results identify novel enzymatic (acyl-CoA thioesterase), covalent (oleoylation), and regulatory (acyl-CoA reversal of CaM inhibition) properties of iPLA 2 ␤ that probably contribute to the multiple diverse signaling roles of this enzyme in cellular functions.