A novel cytosolic calcium-independent phospholipase A2 contains eight ankyrin motifs.

We report the purification, molecular cloning, and expression of a novel cytosolic calcium-independent phospholipase A2 (iPLA2) from Chinese hamster ovary cells, which lacks extended homology to other phospholipases. iPLA2 is an 85-kDa protein that exists as a multimeric complex of 270-350 kDa with a specific activity of 1 μmol/min/mg. The full-length cDNA clone encodes a 752-amino acid cytoplasmic protein with one lipase motif (GXS465XG) and eight ankyrin repeats. Expression of the cDNA in mammalian cells generates an active 85-kDa protein. Mutagenesis studies show that Ser465 and the ankyrin repeats are required for activity. We demonstrate that iPLA2 selectively hydrolyzes the sn-2 over sn-1 fatty acid by 5-fold for 1,2-dipalmitoyl phosphatidylcholine in a mixed micelle. Moreover, we found the fatty acid preference at the sn-2 position to be highly dependent upon substrate presentation. However, iPLA2 does have a marked preference for 1,2-dipalmitoyl phosphatidic acid presented in a vesicle, generating the lipid second messenger lysophosphatidic acid. Finally the enzyme is able to hydrolyze the acetyl moiety at the sn-2 position of platelet-activating factor.


From the Genetics Institute, Cambridge, Massachusetts 02140
We report the purification, molecular cloning, and expression of a novel cytosolic calcium-independent phospholipase A 2 (iPLA 2 ) from Chinese hamster ovary cells, which lacks extended homology to other phospholipases. iPLA 2 is an 85-kDa protein that exists as a multimeric complex of 270 -350 kDa with a specific activity of 1 mol/min/mg. The full-length cDNA clone encodes a 752-amino acid cytoplasmic protein with one lipase motif (GXS 465 XG) and eight ankyrin repeats. Expression of the cDNA in mammalian cells generates an active 85-kDa protein. Mutagenesis studies show that Ser 465 and the ankyrin repeats are required for activity. We demonstrate that iPLA 2 selectively hydrolyzes the sn-2 over sn-1 fatty acid by 5-fold for 1,2-dipalmitoyl phosphatidylcholine in a mixed micelle. Moreover, we found the fatty acid preference at the sn-2 position to be highly dependent upon substrate presentation. However, iPLA 2 does have a marked preference for 1,2-dipalmitoyl phosphatidic acid presented in a vesicle, generating the lipid second messenger lysophosphatidic acid. Finally the enzyme is able to hydrolyze the acetyl moiety at the sn-2 position of platelet-activating factor.
Phospholipase A 2 enzymes are a broad group of proteins that have been intensively studied due to their potential involvement in the production of proinflammatory mediators, such as prostaglandins and leukotrienes through the release of arachidonic acid from membrane phospholipid (1). Concurrently, with fatty acid hydrolysis, lysophos-pholipid is released, the precursor for the proinflammatory molecule platelet-activating factor (PAF) 1 as well as a potent mitogen in the case of lysophosphatidic acid (2,3).
Phospholipase A 2 enzymes can be divided into several groups based on sequence similarity and subcellular localization. Groups I-III consist of low molecular weight, 13-18-kDa, extracellular (sPLA 2 ) enzymes found in snake venoms, mammalian pancreas, and synovial fluid. These enzymes require millimolar concentrations of Ca 2ϩ for maximal activity and fail to exhibit any preference for the fatty acid substituent at the sn-2 position of phospholipids (4,5). There are numerous reports that have causally linked sPLA 2 to various inflammatory con-ditions, including arthritis. However, in several mouse strains the sPLA 2 gene is naturally disrupted, yet these strains have been frequently used as antigen-induced arthritic models (6). Therefore, the role of sPLA 2 in eicosanoid production and inflammation has become less clear. Conversely, the widely expressed 85-kDa cytosolic calcium-dependent phospholipase (cPLA 2 ) of Group IV has the expected characteristics of an enzyme mediating hormonally induced eicosanoid and PAF production (7). cPLA 2 has a 20-fold preference for arachidonic acid over unsaturated fatty acids in phospholipid substrates (8) and consists of two functional domains: a calcium-dependent lipid binding domain (CaLB) and a calcium-independent catalytic domain (8,9). The former domain mediates the translocation of cPLA 2 to the nuclear envelope and endoplasmic reticulum upon treatment with extracellular factors that raise intracellular Ca 2ϩ concentrations to submicromolar levels (10,11). In the catalytic domain, Ser 228 has been shown to be essential for enzymatic activity (12,13) and is contained within a partial lipase consensus sequence, GXSG, common to lipases and esterases (14,15). Finally, cPLA 2 is activated through phosphorylation of Ser 505 (16) by mitogen-activated protein kinase (17). Other phosphorylation sites have been recently mapped, but their role in the activation of cPLA 2 is unclear (18,19).
Compared with cPLA 2 , there is a paucity of information on the structure and regulation of intracellular calcium-independent phospholipases (iPLA 2 ) as well as the role of these enzymes in the inflammatory cascade, lipid metabolism, and other signaling pathways. A wide variety of poorly defined iPLA 2 activities have been found in different tissues (Ref. 20 and references therein), but only three groups have attributed cytosolic iPLA 2 activities to distinct polypeptides.
Gross and colleagues have assigned an activity found in myocardium (21)(22)(23)(24), to a 40-kDa protein that exists as a 350-kDa complex by gel permeation chromatography (25). Moreover, the level of calcium-independent activity was found to increase during induced myocardial ischemia (22). In contrast, others failed to observe the same activation under similar experimental conditions; indeed, a decrease in calcium-independent activity was observed with prolonged ischemia (26).
Since the 40-kDa species copurified with an 85-kDa phosphofructokinase isoform, it was suggested that the activity exists as a complex of the 40-kDa species and four units of the 85-kDa polypeptide. The activity was irreversibly inactivated by a bromoenol lactone (BEL; (E)-6-(bromomethylene)-tetrahydro-3-(1naphthalenyl)-2H-pyran-2-one) (27), a suicide inhibitor originally designed to inhibit the serine esterase, ␣-chymotrypsin (28). Similar calcium-independent BEL-sensitive activities were reported to exist in pancreatic islets (29), aortic smooth muscle cells (30), RAW 264.7 cells (31), and rat hippocampus (32). The existence of a 40-kDa PLA 2 or a functional high molecular weight complex was not demonstrated in these cell lines except in the case of pancreatic islets (33). The activity was stimulated by the presence of ATP (23) and had a slight preference, 2-fold, for 1-ether over 1-acylphospholipid substrates as well as a 3-fold preference for arachidonic over palmitic acid at the sn-2 position (21), but there were no significant phospholipase A 1 or lysophospholipase activities associated with the complex.
Dennis and colleagues (34) have identified an activity from a macrophage cell line, P388D 1 , correlating with a single 80-kDa polypeptide by SDS-polyacrylamide gel electrophoresis (PAGE) where the phospholipase activity was substantially enhanced in the presence of detergent (Triton X-100), giving a specific activity 5 mol/min/mg. Moreover, this preparation showed a 2-fold selectivity for hydrolysis of palmitic acid over arachidonic acid, favored 1-acyl-over 1-ether-containing phospholipids, and has an intrinsic lysophospholipase activity in a mixed micelle assay. However, analogous to the 40-kDa species studied by Gross and co-workers (21)(22)(23)(24)(25), the 80-kDa activity exists as a 337-kDa complex that was stimulated by ATP (and other nucleoside triphosphates), reversibly inhibited by palmitoyl and arachidonyl trifluoromethyl ketones, and irreversibly inhibited by BEL and methyl arachidonyl fluorophosphonate (35)(36)(37). Notably, in the absence of detergent, i.e. a vesicle assay, the substrate selectivity at the sn-2 position was reversed compared with the mixed micelle assay, favoring arachidonic over palmitic acid. Also, there was no stimulation of activity by ATP when phospholipid was presented in a vesicle. Recently, the 80-kDa activity has been implicated in phospholipid remodeling in the cell line P388D 1 (38,39).
More recently, a 28-kDa polypeptide has been purified to homogeneity from rabbit kidney cortex that preferentially hydrolyzes arachidonyl-containing plasmalogen phospholipids in a vesicle assay (40).
In our efforts to molecularly define the cytosolic enzymes that release fatty acid from membrane phospholipids, we describe the purification of an 85-kDa cytosolic calcium-independent phospholipase A 2 , iPLA 2 , and expression of a cDNA encoding this activity. The deduced amino acid sequence has one lipase motif (GXSXG) containing the putative active site serine and eight ankyrin repeat sequences and lacks homology with any other known phospholipase A 2 enzymes. M Triton X-100, 5 mM EDTA, and 10% glycerol (v/v) and was sonicated for 2 ϫ 30 s at 4°C. Substrate was incubated with aliquots of iPLA 2 at 37°C for defined time periods, and released fatty acid was measured as described (41). The release of [ 3 H]acetyl from [acetyl-3 H]PAF was measured as described (42). Alternatively, quantification of products by tlc (34) was used for determining regioselectivity of iPLA 2 and hydrolysis of 1,2-dipalmitoyl PA (43), where the lipids were separated in the basic solvent system, chloroform:methanol:ammonium hydroxide:water (45: 30:3:5, v/v/v/v) or the acidic system (34), respectively.
Purification Procedure-Chinese hamster ovary (CHO) cells were harvested at about 2.4 ϫ 10 6 cells/ml (44) as a by-product of large scale cultures for the production of secreted proteins. The level of calciumindependent activity in these cell lines was equivalent to the parental CHO-DUKX cell cultured as described (44). CHO cells were collected by centrifugation, washed with PBS, rapidly frozen in liquid nitrogen, and stored at Ϫ80°C. Frozen pellets (3.5 kg, about 1.4 ϫ 10 12 cells) were resuspended in 20 mM imidazole, pH 7.5, 0.25 M sucrose, 2 mM EDTA, 2 mM EGTA, 1 g/ml leupeptin, 5 g/ml aprotinin, 5 mM DTT, and 1 mM phenylmethylsulfonyl fluoride (extraction buffer). The cells were lysed by N 2 cavitation at 600 p.s.i., and the lysate was clarified by centrifugation at 10,000 ϫ g for 30 min followed by 100,000 ϫ g for 60 min. The lysate was made 10% (v/v) in glycerol and diluted to 5 mg/ml total protein with 20 mM imidazole, pH 7.5, 5 mM DTT, 1 mM EDTA, 1 mM EGTA, and 10% glycerol (v/v) (buffer A). Lysate (10 g of total protein) was applied (16 ml/min) to a DEAE toyopearl (TosoHaas) column (1liter volume) equilibrated with buffer A. The column was developed with a gradient of 0.0 -0.5 M NaCl. Active fractions (peak I) were diluted to 3 mg/ml with buffer A and made 0.5 M in (NH 4 ) 2 SO 4 and loaded (4 g of total protein) onto a phenyl toyopearl column (300-ml volume) at 10 ml/min. The column was developed with a reverse gradient of 0.5-0.2 M followed by 0.2-0.0 M (NH 4 ) 2 SO 4 over 15 min and 85 min, respectively, and at this point the activity was still bound. The activity was eluted with buffer A at 1.5 ml/min for 18 h and loaded directly onto a heparin toyopearl column (10-ml volume) connected in tandem to the phenyl column. The phenyl column was disconnected, and the activity was eluted from the heparin column with 0.5 M NaCl in buffer A at 2 ml/min. Active fractions were exhaustively dialyzed against 20 mM Bis-Tris, pH 7, 1 M urea, 5 mM DTT, and 10% glycerol (v/v) and applied (1 ml/min) to a Mono P 10/30 (Pharmacia Biotech Inc.) column equilibrated with the same buffer. Activity was eluted by establishing a pH gradient with 10% polybuffer 74, pH 5, 1 M urea, 5 mM DTT, and 10% glycerol (v/v) (1 ml/min). The combined active fractions were loaded onto a heparin toyopearl column (10 mg of protein/ml of resin) and eluted as above. Portions of active fractions were further purified using two TSK G3000SW XL (TosoHaas) columns (7.8 mm ϫ 30 cm) in tandem, developed in buffer A with 150 mM NaCl at 0.3 ml/min.
Recombinant iPLA 2 was partially purified from the CHO line c4.20.5s (6 ϫ 10 9 cells; see below) in a similar manner except that the Mono P step was omitted.
Generation of Tryptic Peptides-Active fractions from the Mono P/heparin step (63 g of total protein) were made 10% in SDS and concentrated in a Microcon-30 (Amicon). The sample was loaded onto a 4 -20% gradient SDS-PAGE minigel (Novex) run under reducing conditions. The gel was stained (45), and the 85-kDa band was excised. Tryptic peptides were generated as described previously (8).
1 ϫ 10 6 plaques from a human Burkitt lymphoma (RAJI) library (Clonetech) were screened with the 2-kb SmaI probe. Ten positive clones were isolated upon replating positive plaques. One of these, number 19, was subcloned as two EcoR I fragments, A (2 kb) and B (1.3 kb), and sequenced.
Expression of iPLA 2 -Clone 9 was excised from the vector and subcloned into the SalI site of pED (47), yielding pC9. iPLA 2 in pC9 was transiently expressed in COS cells as described (48). A stable CHO cell line was established, c4.20.5, by selection and amplification of iPLA 2 in 20 nM methotrexate as described (44). c4.20.5 was adapted to suspension culture by changing the growth media from ␣ (Life Technologies) to R1 (Mediatech) and initially maintaining the cell density at 3 ϫ 10 5 cells/ml. Activity levels were determined by detaching transfected COS or c4.20.5 cells with phosphate-buffered saline, 2 mM EDTA, collecting the cells by centrifugation, and washing with phosphate-buffered saline. The cell pellet was resuspended in extraction buffer (1 ϫ 10 7 cells/ml) and lysed by 20 strokes in a glass Dounce homogenizer, and the supernatant was collected by centrifugation. The supernatant was analyzed for activity by the release of radiolabeled fatty acid from phospholipid. The expression of iPLA 2 was analyzed by immunoblot with a polyclonal antibody raised against iPLA 2 at 1:200 dilution or, in the case of the Flag-tagged proteins, the anti-Flag M2 (IBI) monoclonal antibody (0.75 g/ml) and visualized with the ECL System (Amersham).
DNA Constructs-A glutathione S-transferase fusion was generated by ligating the StuI/NotI fragment of pC9 (amino acids 437-752) into pGEX-5x-3 (Pharmacia Biotech Inc.). The amino-terminal N-Flag construct was formed by ligating a HincII/XbaI fragment of pC9 with synthetic oligonucleotides encoding the Flag epitope (IBI) and the first eight amino acids of iPLA 2 in the pED expression vector. The carboxylterminal C-Flag construct was prepared similarly by ligating a SalI/ HindIII fragment of pC9 with synthetic oligonucleotides encoding the last seven amino acids of iPLA 2 and the Flag epitope. The first 150 amino acids and ankyrin repeats were deleted by ligating the TfiI/XbaI fragment of pC9 (amino acids 422-752) with synthetic oligonucleotides encoding the Flag epitope and iPLA 2 residues (406 -421, ⌬a; 411-421, ⌬b; 416 -421, ⌬c) into the pED vector. The serine 252 and 465 to alanine mutations were performed with the Chameleon Mutagenesis kit (Stratagene) using oligonucleotides CATGGGACCCGCTGGCTTTCC and GGCAGGAACCGCCACTGGGGGC, respectively.
Antibody Generation-Cultures expressing the glutathione S-transferase fusion protein were prepared as described (Pharmacia). The cells were pelleted at 2000 g, resuspended in phosphate-buffered saline (25 ml), and lysed by French press, and the pellet was collected at 30,000 ϫ g. The pellet was resuspended in 50 mM Tris-HCl, pH 8, and passed five times through a 22-gauge needle, and the insoluble material was collected at 5000 ϫ g. The glutathione S-transferase fusion was dissolved in 7 M guanidine HCl, 50 mM Tris-HCl, pH 8, 1 mM EDTA, and 10 mM DTT (3 mg/ml), 10 mg was loaded onto a C-4 reverse-phase column (Vydac), and the fusion protein was eluted with acetonitrile (ϳ35%). Antiserum was generated by injecting rabbits with this material (Pocono Rabbit Farm).

RESULTS
Purification of a Cytosolic iPLA 2 from CHO Cells-In order to identify calcium-independent phospholipase A 2 activities, we surveyed a variety of cell lines and tissues under conditions where the activity of endogenous cPLA 2 is low, i.e. pH 7, 10% glycerol in the absence of Ca 2ϩ . Using 1-palmitoyl-2-[ 14 C]arachidonyl PC (PAPC) as the substrate, higher activity was observed in a mixed micelle assay in the presence of Triton X-100 than in a vesicle assay. Under these conditions (10 M PAPC, 500 M Triton X-100), the activity is widely distributed in extracts of brain, heart, and a variety of heart cell lines (data not shown). This wide distribution encouraged us to examine CHO cells, which can be readily grown in large quantities. Extracts of CHO cells had similar activity to heart cell lines and tissue extracts, and consequently iPLA 2 was purified from this source. Table I summarizes the purification of iPLA 2 from 500 liters of CHO cells. The majority of the activity was found in the supernatant upon lysis by nitrogen cavitation. DEAE column chromatography of the cell lysate gave two broad peaks of activity eluting at 75 mM and 300 mM NaCl (data not shown). Our efforts focused on purifying the activity from the 75 mM salt peak, since attempts to further purify the high salt peak were unsuccessful due to a lack of reproducible chromatographic behavior. The 75 mM NaCl eluate was loaded onto a  hydrophobic interaction (phenyl-TSK) column, and calciumindependent PLA 2 activity eluted as a broad peak in 0 M salt wash following a reverse gradient. To minimize losses of iPLA 2 , the activity was concentrated by an in-line heparin column. Subsequent chromatofocusing on a Mono P column, using a gradient from pH 7 to 5, resulted in a substantial purification Gaps are shown by a dashes, and X denotes the uncertainty in the amino acid sequence due to a lack of consensus splice acceptor and donor site in clone 19. The ankyrin repeats are underlined, the lipase motif is overlined, and arrows indicate Ser 252 and Ser 465 . B, sequence alignment of CHO and human iPLA 2 after the eighth ankyrin repeat with the equivalent regions in T04B2.5 and F47A4.5. C, sequence alignment of CHO and human iPLA 2 from the amino terminus to the end of the eighth ankyrin repeat with the putative ankyrin-containing region of F47A4.5. The cDNA sequences for CHO iPLA 2 , T04B2.5, and F47A4.5 can be found in the GenBank TM data base under accession numbers Il5470 (U.S. patent 5466595), Cet04b2_7, and Cef47a4_5, respectively. Sequences were aligned using the GeneWorks progressive alignment method. (Table I). SDS-PAGE of Mono P fractions indicated that the amount of iPLA 2 activity (Fig. 1A, fractions 9 -13) correlated with the staining intensity of an 85-kDa protein. To demonstrate that iPLA 2 activity corresponded to the 85-kDa species, a small portion of pooled fractions from the Mono P column was further purified using gel permeation chromatography. Interestingly, all the activity eluted in the 250 -350-kDa range (Fig.  1B, fractions 55-58), suggesting that the native enzyme is an oligomer of 85-kDa monomers. Indeed, SDS-PAGE of these fractions (Fig. 1B) confirmed the correlation of iPLA 2 activity with the intensity of the 85-kDa protein. The estimated specific activity of the iPLA 2 is approximately 1 mol/min/mg in a mixed micelle assay (10 M PAPC, 500 M Triton X-100) based on a purity of 10 -20% in the size exclusion fractions. iPLA 2 Contains One Lipase Motif and Eight Ankyrin Repeats-The Mono P/heparin fractions containing the activity were separated by SDS-PAGE, and the 85-kDa polypeptide was excised and digested in situ with trypsin. Reverse-phase high pressure liquid chromatography followed by Edman degradation yielded several peptide sequences, three of which were used to design degenerate oligonucleotide pools for screening a CHO cell cDNA library. Several full-length clones were obtained, and the amino acids of one of these were sequenced (clone 9) (Fig. 2A).
The cDNA encodes a polypeptide of 752 amino acids with a calculated molecular mass of 85 kDa and a neutral isoelectric point. iPLA 2 does contain a consensus motif, GXS 465 XG, commonly found in serine proteases and lipases (14,15). Notably, there are seven ankyrin repeat sequences and one less conserved repeat from amino acids 251-280 ( Fig. 2A), a domain that is widely distributed through proteins with diverse functions (49,50). A partial human cDNA (clone 19) was isolated from a Burkitt lymphoma library by screening with a CHO cDNA labeled probe. Although clone 19 was the correct size for a full-length cDNA, it was found to contain potential unspliced intron sequences (data not shown). Based on sequence comparison with the CHO iPLA 2 , the putative coding sequence of clone 19 encodes 90% of human iPLA 2 and is 90% identical to the CHO protein in the region of overlap ( Fig. 2A). Importantly, iPLA 2 lacks homology to cPLA 2 , the secreted phospholipases, or the 85-kDa phosphofructokinase that was reported to be associated with the 40-kDa myocardial calciumindependent phospholipase (25). However two putative proteins of unknown function from Caenorhabditis elegans, T04B2.5 (408 amino acids) and F47A4.5 (1071 amino acids), have identical lipase motifs and significant sequence similarity in the region surrounding the lipase motif (Fig. 2B). Moreover, closer inspection of the F47A4.5 sequence amino-terminal to the putative lipase motif revealed stretches of homology throughout the equivalent region in CHO and human iPLA 2 ,
Finally, Northern analysis of several rat/mouse tissues (Fig.  3A) and cell lines (data not shown) indicates ubiquitous expression with the highest levels in testis and liver.
iPLA 2 Encodes an 85-kDa Cytosolic Calcium-independent PLA 2 -To prove the purified 85-kDa protein is an iPLA 2 , clone 9 was subcloned into the mammalian expression plasmid, pED, and transiently expressed in COS cells. Cytosolic extracts of the COS cells had over 300-fold more iPLA 2 activity than mock-transfected COS cells (Fig. 4A; Cos and iPLA 2cos ). We also established stable CHO cell lines expressing iPLA 2 through selection and amplification in methotrexate. One of these, c4.20.5, expressed 50-fold higher levels of iPLA 2 activity than the parental line, CHO-DUKX (Fig. 4A, CHO and iPLA 2CHO ).
To study the function of iPLA 2 , a polyclonal antibody was generated to a glutathione S-transferase fusion of the carboxylterminal half of the protein, amino acids 437-752, containing the lipase consensus sequence. This antibody specifically recognizes an 85-kDa polypeptide expressed in COS and CHO cells transfected with clone 9 (Fig. 4B, iPLA 2cos and iPLA 2CHO ) compared with the parental cell lines (Fig. 4B, Cos and CHO).
Although during the purification we did not observe a 40-kDa polypeptide correlating with activity (Fig. 1, A and B), we considered the possibility the the 85-kDa polypeptide might be the precursor to a 40-kDa species equivalent to the one purified by Gross and colleagues (21). To address this hypothesis, two fusion proteins of iPLA 2 tagged at either the amino or carboxyl termini with the Flag epitope, N-and C-Flag, respectively, were expressed transiently in COS cells. Both constructs yielded levels of iPLA 2 activity comparable with clone 9 (Fig.  4A, iPLA 2cos , N-Flag, and C-Flag). Moreover, polypeptides of 40 kDa were not observed by immunoblot when probed with either an anti-Flag monoclonal antibody (Fig. 4C) or with the anti-iPLA 2 polyclonal antibody (Fig. 4B). These data demonstrate that the amino acid sequence in Fig. 2A corresponds to a novel cytosolic calcium-independent PLA 2 .
We considered the possibility that the ankyrin repeats might be required for activity, since this motif is postulated to be involved in protein-protein interactions (49,50). To test this hypothesis, the carboxyl-terminal half of iPLA 2 , lacking the first 150 amino acids and the eight ankyrin repeats but retaining the lipase consensus sequence, was expressed as a fusion protein with the Flag epitope at the amino terminus. Although expression of the truncated form was similar to full-length iPLA 2 (Fig. 4, B and C), it lacked activity (Fig. 4A). Similar results were obtained when amino acids 411-752 or 416 -752 of iPLA 2 where expressed as fusion proteins (data not shown). . All assays were in duplicate. B, aliquots of cytosolic extracts (same order as in A) were separated by SDS-PAGE on a 4 -20% gel. Proteins were transferred to nitrocellulose, probed with the anti-iPLA 2 polyclonal sera, and visualized as described under "Experimental Procedures." C, cytosolic extracts were analyzed as in B except that the nitrocellulose blot was probed with an anti-Flag monoclonal antibody. iPLA 2 contains two serine residues that could be at the active site. The first, GXS 465 XG, is a perfect consensus sequence found in many lipases (14,15). The other potential active site is located in the sequence GPSGF, although this is a poor site due to the presence of the large Phe residue in the sequence ( Fig.  2A). To identify the catalytic serine, Ser 252 3 Ala and Ser 465 3 Ala were expressed in COS cells as amino-terminal fusions with the Flag epitope. Both constructs were expressed at levels comparable with that of the full-length N-Flag iPLA 2 (Fig. 4, B and C), indicating that there were no gross conformational changes in the proteins due to the respective mutations. However, detailed studies will be required to confirm the structural integrity of the mutants. Ser 465 3 Ala had no activity above background levels, i.e. 1-2% of residual activity compared with wild-type iPLA 2 , whereas Ser 252 3 Ala exhibited activity comparable with that of the full-length N-Flag iPLA 2 (Fig. 4A). These data are consistent with Ser 465 being the active site nucleophile and with the 85-kDa polypeptide as the catalytic unit of the 250 -350-kDa complex.
iPLA 2 Selectively Hydrolyzes Phospholipids at the sn-2 Position-We investigated the positional selectivity of iPLA 2 using material purified from the overexpressing CHO cell line, c4.20.5s, and the substrate 1,2-dipalmitoyl PC presented in a Triton X-100 mixed micelle. Initially, the hydrolysis of 1,2dipalmitoyl PC was followed by measuring the release of palmitic acid, 1-and 2-palmitoyl lysophosphatidylcholine, where both the sn-1 and sn-2 fatty acids were labeled. The results in Fig. 5A show that the ratio of palmitic acid to lyso products was 1:1 over 30 min. This observation suggests that iPLA 2 has little lysophospholipase activity under the assay conditions. Based on these results, we examined the regioselectivity of hydrolysis using 1,2-dipalmitoyl PC labeled at the sn-2 fatty acid. Over a 30-min time period, we observed that 5-fold more palmitate was formed compared with 2-palmitoyl lysophosphatidylcholine (Fig. 5B), demonstrating that iPLA 2 is selective for the hydrolysis of sn-2 fatty acids in phospholipid substrates.
Fatty Acid Selectivity of iPLA 2 Depends upon Substrate Presentation-We investigated the preference for the fatty acid at the sn-2 position when the phospholipid substrate (10 M) was dispersed in either 50 or 500 M Triton X-100. These concentrations of detergent were chosen, since we observed an abrupt drop in activity at the critical micelle concentration of Triton X-100 (0.24 mM in water) when substrate concentration was kept constant (10 M) and the detergent concentration was varied (data not shown). This phenomenon has also been observed for cPLA 2 (58). As the results in Table II indicate, iPLA 2 failed to demonstrate any significant preference for a particular fatty acid, although the rate of hydrolysis of unsaturated fatty acids was greater at 50 M Triton. Additionally, the overall rates of hydrolysis were, at least for some substrates, very sensitive to lipid presentation, e.g. 1,2-dipalmitoyl PC was 4 times better when substrate was dispersed in 500 M Triton, whereas 1-palmitoyl-2-arachidonyl (linoleyl or oleyl) PC were equivalent in both assays (Table II).
We also observed an intrinsic lysophospholipase activity with the substrate 1-palmitoyl lysophosphatidylcholine where the relative rate of hydrolysis was lower with 500 M than 50 M Triton X-100, 1.4 and 10.7 pmol/min, respectively (Table II). We posed the question of whether the lysophospholipase/phospholipase A 1 activity of iPLA 2 , at the low concentration of Triton, was required for cleavage of an sn-2 fatty acid, as opposed to 500 M Triton, where it is not required (Fig. 5). This is clearly not the case, since 1-hexadecyl-2-arachidonyl PC, containing a nonhydrolyzable ether-linked sn-1 fatty acid, was readily cleaved at twice the rate compared with 1-palmitoyl-2arachidonyl PC (Table II). Significantly, we found the sn-1 ether-linked substrate PAF to be a substrate for iPLA 2 in both assay formats (Table II), indicating that the phospholipase A 2 activity of iPLA 2 is not restricted to long chain fatty acids. This is in contrast to cPLA 2 and sPLA 2 , which have no significant PAF acetylhydrolase activity under either assay condition in the presence of calcium (data not shown).
Finally, we investigated the effect of the phospholipid head  (33) and then separated by tlc using the basic solvent system (see "Experimental Procedures"). Radioactive spots were excised and counted. All points were in duplicate. B, 1-palmitoyl-2-[ 14 C]palmitoyl PC was incubated with iPLA 2 , and products were quantified as in A.
group on relative rates of hydrolysis. The enzyme showed no preference between choline and ethanolamine head groups for the substrates tested (Table II). Yet there were clear differences in rates of hydrolysis between choline and negatively charged head groups, such as PI and PA (Table II). For instance, the rate of hydrolysis of 1-stearoyl-2-arachidonyl PI was 5-fold lower than the corresponding PC substrate in both assay systems. Conversely, the immediate precursor to the mitogen lysophosphatidic acid, 1,2-dipalmitoyl PA, was hydrolyzed at a 20-fold higher rate than 1,2-dipalmitoyl PC when the lipid was dispersed in 50 M Triton, but no significant difference in rates was observed at 500 M Triton. Indeed, for the substrates 1,2-dipalmitoyl PC and PA, at least part of this rate difference is attributable to the charge of the head group, since the hydrolysis of 1,2-[ 14 C]dipalmitoyl PA (10% molar fraction) in a bilayer of 1,2-dipalmitoyl PC (90% molar fraction) is reduced 2-fold compared with 1,2-dipalmitoyl PA alone. In contrast, 1,2-[ 14 C]dipalmitoyl PC (10% molar fraction) is hydrolyzed 5-fold faster in a background of 1,2-dipalmitoyl PA (90% molar fraction) (data not shown). DISCUSSION We have demonstrated the existence of a novel cytosolic calcium-independent iPLA 2 in the cytosol of CHO cells whose sequence is not related to any other phospholipase. We found that the iPLA 2 activity correlated with an 85-kDa species in active fractions from two consecutive chromatographic steps. However, we observed that iPLA 2 in the native state exists as a oligomeric complex of molecular mass 250 -350 kDa, a result that has been found for other calcium-independent phospholipase activities (23,25,34). We obtained several full-length cDNA CHO clones, and one of these was sequenced and found to encode a 752-amino acid polypeptide. Expression in COS cells generates a single 85-kDa species and 300-fold increase of iPLA 2 activity in cytosolic extracts.
Although the inferred amino acid sequences of CHO and human proteins are 90% identical, they lack homology to cPLA 2 , to any sPLA 2 , or to the 85-kDa phosphofructokinase isoform associated with the 40-kDa myocardial calcium-independent phospholipase (25,33) but are homologous to the murine iPLA 2 isolated from P388D 1 cells (accompanying manuscript (Ref. 59)). Significantly, iPLA 2 is homologous to two putative proteins from C. elegans, T04B2.5 and F47A4.5, through two distinct sets of sequence motifs. Amino acids surrounding the consensus lipase motif and the relative position of the motif (GTS 465 TG) to the carboxyl termini are conserved among CHO, human, and C. elegans proteins. Indeed, mutagenesis of Ser 465 3 Ala in CHO iPLA 2 abrogates activity, while Ser 252 3 Ala in a partial lipase motif, which is not conserved from CHO to human, had no effect. Second, iPLA 2 contains eight consecutive ankyrin repeats of approximately 30 amino acids in length. Likewise, one of the C. elegans proteins (F47A4.5) has the two-domain structure of iPLA 2 with eight putative ankyrin repeats followed by the region containing the lipase motif. In contrast, T04B2.5 (408 amino acids) has no ankyrin repeats and corresponds to the carboxyl-terminal half of iPLA 2 (340 amino acids). This raises the possibility that F47A4.5 is the C. elegans homolog of iPLA 2 . More speculatively, T04B2.5, which has a theoretical molecular size of 45 kDa, may be a related lipase that does not require the ankyrin repeat motif for lipolytic activity and, potentially, could be related to the 40-kDa calcium-independent activities observed by others (21,25).
The presence of ankyrin repeats in iPLA 2 is a novel and intriguing result, since this structural motif has been found in many proteins of diverse function (49,50), including transcriptional regulation, cell cycle control, and cell differentiation.
Indeed, the predominant function of ankyrin repeats, although poorly defined, appears to be protein-protein interactions in three broad categories: first, the coupling of the spectrin network on the cytoplasmic face of the plasma membrane in neural cells and erythrocytes with the cytoplasmic domain of various integral membrane proteins (51); second, the formation of functionally active protein complexes, e.g. the Ets family of nuclear proteins (52); and third, the masking of regulatory signals, e.g. the retention of NF-B by IB␣ or the p105 precursor of p50 in the cytoplasm (53,54).
The demonstration that iPLA 2 , with or without the Flag epitope, is not posttranslationally modified to a smaller catalytically active protein, at least when expressed in COS or CHO cells, implies the ankyrin repeats are not masking activity. This observation is supported by the loss of lipolytic function when the ankyrin repeats and the first 150 amino acids before the repeats are deleted, although the lipase consensus sequence is retained. Although further deletion experiments will be required to fully define the functional boundaries of the domains. However, the results imply that some or all of the ankyrin repeats are required for function, possibly to enable the ordered association of 85-kDa subunits to give the 250 -350-kDa oligomeric complex found by gel permeation chromatography. Also, this region may contain amino acid residues that support catalytic function at the active site of iPLA 2 . Overall these results suggest that iPLA 2 is not the precursor to a 40-kDa species analogous to the one found by others (21,25), but rather that the 85-kDa protein is similar to the 80-kDa activity from the murine macrophage-like cell line, P388D 1 (34). Indeed, the 80-kDa protein is recognized by the anti-iPLA 2 polyclonal sera and has the same electrophoretic mobility as the 85-kDa species (accompanying manuscript (Ref. 59)).
In preliminary experiments we found the phospholipid selectivity of iPLA 2 and the relative reactivity for a particular substrate to be highly dependent upon substrate presentation and detergent concentration. Indeed, our results parallel those found for the 80-kDa activity from P388D 1 cells, including the 5:1 regioselectivity for sn-2 versus sn-1 fatty acids (34). At 500 M Triton X-100, the apparent substrate preference is 1,2dipalmitoyl PC Ͼ 1-palmitoyl-2-arachidonyl PC Ͼ 1-hexadecyl-2-arachidonyl PC. Conversely, at 50 M Triton X-100, the order of preference for these three substrates was reversed, analogous to the substrate selectivity exhibited by the 40-kDa myocardial activity in a vesicle assay (21). iPLA 2 does exhibit an intrinsic lysophospholipase and PLA 1 activity, depending on the assay conditions. However we found no stimulation of lipolysis by the presence of ATP in either assay system compared with others (23,34).
Notably, iPLA 2 is a PAF acetylhydrolase indicating a unique flexibility at the active site in being able to accommodate not only long but also short chain fatty acids. This is in contrast to the known intracellular PAF acetylhydrolases that are specific for short chain sn-2 substituents (42,55,56). Further, the enzyme has a marked preference, at 50 M Triton X-100, for the phosphatidic acid substrate 1,2-dipalmitoyl PA producing the lysophospholipid, lysophosphatidic acid, a second messenger implicated in a number biological events (2,3). Only two other distinct PA-specific phospholipase A 2 activities have been identified, neither of which appears to be biochemically or biophysically similar to iPLA 2 ; the first isolated from activated platelets is calcium-dependent (43), and the second, a 58-kDa protein from rat brain, is calcium-independent and specific for phosphatidic acid over other phospholipid head groups (57).
Additionally, iPLA 2 has similar sensitivity to inhibition by BEL, palmitoyl trifluoromethyl ketone, and methyl arachidonyl fluorophosphonate (accompanying manuscript (Ref. 59)) as the P388D 1 iPLA 2 (35)(36)(37). However, although BEL at 30 M completely inhibits iPLA 2 and has no effect on cPLA 2 activity in cell lysates, caution should be exercised in using this reagent on intact cells. For instance, the above concentration of BEL used to inhibit iPLA 2 will suppress cPLA 2 -mediated arachidonic acid release by 50% using a CHO cell line overexpressing cPLA 2 , E5 (16). Similarly, calcium ionophore-induced production of prostaglandin E 2 from COS cells transiently transfected with cPLA 2 and COX-2 is reduced by 70%. 2 In summary, we have described the isolation, molecular cloning, and characterization of a novel 85-kDa cytosolic calciumindependent phospholipase A 2 , iPLA 2 . The unanticipated discovery is the presence of the ankyrin repeats, an important structural and functional motif for many proteins. Clearly, there is ample precedent to suggest that this motif has two roles that are not mutually exclusive: first, to cause the selfassociation of iPLA 2 to form a catalytically competent species that can bind a lipid bilayer, and second, to mediate the interaction with other proteins that may regulate or direct lipolytic activity in particular intracellular locations. Defining these interactions and determining whether the PAF acetylhydrolase activity, as well as the selectivity for phosphatidic acid are biologically relevant will answer the question of involvement of iPLA 2 in the inflammatory cascade, the production of lipid second messengers, and general lipid metabolism.