Role of Phosphorylation and Basic Residues in the Catalytic Domain of Cytosolic Phospholipase A2α in Regulating Interfacial Kinetics and Binding and Cellular Function*

Group IVA cytosolic phospholipase A2 (cPLA2α) is regulated by phosphorylation and calcium-induced translocation to membranes. Immortalized mouse lung fibroblasts lacking endogenous cPLA2α (IMLF-/-) were reconstituted with wild type and cPLA2α mutants to investigate how calcium, phosphorylation, and the putative phosphatidylinositol 4,5-bisphosphate (PIP2) binding site regulate translocation and arachidonic acid (AA) release. Agonists that elicit distinct modes of calcium mobilization were used. Serum induced cPLA2α translocation to Golgi within seconds that temporally paralleled the initial calcium transient. However, the subsequent influx of extracellular calcium was essential for stable binding of cPLA2α to Golgi and AA release. In contrast, phorbol 12-myristate 13-acetate induced low amplitude calcium oscillations, slower translocation of cPLA2α to Golgi, and much less AA release, which were blocked by chelating extracellular calcium. AA release from IMLF-/- expressing phosphorylation site (S505A) and PIP2 binding site (K488N/K543N/K544N) mutants was partially reduced compared with cells expressing wild type cPLA2α, but calcium-induced translocation was not impaired. Consistent with these results, Ser-505 phosphorylation did not change the calcium requirement for interfacial binding and catalysis in vitro but increased activity by 2-fold. Mutations in basic residues in the catalytic domain of cPLA2α reduced activation by PIP2 but did not affect the concentration of calcium required for interfacial binding or phospholipid hydrolysis. The results demonstrate that Ser-505 phosphorylation and basic residues in the catalytic domain principally act to regulate cPLA2α hydrolytic activity.

Group IVA cytosolic phospholipase A 2 (cPLA 2 ␣) 3 specifically hydrolyzes arachidonic acid (AA) from the sn-2-position of membrane phospholipids in response to diverse cellular stimuli (1)(2)(3)(4)(5). In releasing AA, it performs a fundamental role in regulating signaling for a multitude of lipid-mediated pathways that control many important cell processes. Leukotrienes and prostaglandins are products of AA metabolism via 5-lipoxygenase and cyclooxygenase pathways, respectively (6). These metabolites regulate diverse processes in normal and diseased tissues, making it important to tightly regulate the release of free AA through cPLA 2 ␣. cPLA 2 ␣ is widely expressed in mammalian tissues, reflecting its central role in regulating lipid mediator production in diverse cell types. cPLA 2 ␣ is regulated post-translationally by submicromolar levels of calcium and by phosphorylation (1,2,7). Agonist-induced increase in the [Ca 2ϩ ] i leads to a loading of the C2 domain with calcium, and this mediates cPLA 2 ␣ translocation to Golgi, endoplasmic reticulum, and nuclear envelope to access substrate (8 -12). cPLA 2 ␣ has multiple phosphorylation sites in the catalytic domain. Analysis of cPLA 2 ␣ expressed in baculovirus-infected Sf9 cells revealed constitutive phosphorylation of Ser-454, Ser-437, and Ser-505 and phosphorylation on Ser-727 in response to okadaic acid (13). In mammalian cells, cPLA 2 ␣ is phosphorylated on Ser-505, Ser-727, and Ser-515 by mitogen-activated protein kinases (MAPKs), MAPK-activated protein kinase MNK1 (or a related kinase), and calcium/calmodulin-dependent kinase II (CamKII), respectively (14 -19). Phosphorylation of Ser-505 and Ser-727 are functionally important for re-gulating cPLA 2 ␣-mediated AA release from stimulated cells (14,17). It has recently been shown that phosphorylation of cPLA 2 ␣ on Ser-515 and Ser-505 is required for AA release in vascular smooth muscle cells stimulated with norepinephrine (20).
Phosphorylation of cPLA 2 ␣ and physiological increases in [Ca 2ϩ ] i synergistically promote the full activation of cPLA 2 ␣ for releasing AA (21)(22)(23). Phosphorylation of cPLA 2 ␣ on Ser-505 increases its catalytic activity (14,15,24); however, the role of phosphorylation in regulating calcium-induced translocation in cells has not been resolved. It has been reported that phosphorylation of cPLA 2 ␣ on Ser-505 enhances the phospholipid binding affinity at low physiological calcium levels in vitro and in cells (25). This is consistent with another study showing that the inability of cPLA 2 ␣ phosphorylation site mutants to release AA is overcome by inducing supraphysiological [Ca 2ϩ ] i (17). However, it has previously been shown that cPLA 2 ␣S505A translocates to membrane in response to calcium ionophore, although it releases less AA (26). In addition, a direct comparison of wild type cPLA 2 ␣ and phosphorylation site mutants by time lapse imaging demonstrated similar translocation properties in response to physiological increases in calcium induced by ATP (27).
Calcium binding to the cPLA 2 ␣ C2 domain increases its affinity for membrane through hydrophobic interactions (28 -32). This positions the catalytic domain for interaction with the membrane by hydrophobic and electrostatic mechanisms (33). A tryptophan (Trp-464) on the membrane binding face of the catalytic domain stabilizes cPLA 2 ␣ and prolongs membrane binding after decreases in [Ca 2ϩ ] i (34). A patch of basic residues (Lys-488, Lys-541, Lys-543, and Lys-544) in the cPLA 2 ␣ catalytic domain is the site of interaction with phosphatidylinositol 4,5-bisphosphate (PIP 2 ), which increases catalytic activity in vitro (33,(35)(36)(37). However, the role of these basic residues in regulating translocation in cells stimulated with physiological agonists has not been investigated.
The goal of this study is to investigate the role of phosphorylation and basic residues in the catalytic domain in regulating cPLA 2 ␣. We compare the interfacial binding and kinetic properties of phosphorylated and mutant forms of cPLA 2 ␣ in vitro as a function of calcium concentration with the behavior of these enzymes in a cellular reconstitution model. By expressing wild type and mutant forms of cPLA 2 ␣ in lung fibroblasts lacking cPLA 2 ␣, we investigated the functional role of phosphorylation and the PIP 2 binding site in regulating calcium-dependent cPLA 2 ␣ translocation and AA release without interference of endogenous wild type enzyme.
Preparation of cPLA 2 ␣-Full-length wild type and mutant forms of human cPLA 2 ␣ containing the affinity peptide YHH-HHHH fused to the C-terminal Ala-749 were prepared by a modification of the procedure described previously (38) (supplemental material). cPLA 2 ␣ concentrations were determined from the absorbance at 280 nm (⑀ 280 ϭ 0.827 mg Ϫ1 ml Ϫ1 cm Ϫ1 ) (38). Mutagenesis of the coding region in the baculovirus transfer plasmid was carried out first by deletion of the desired region and then by insertion of the mutated region. Both steps were carried out using the QuikChange kit (Stratagene). Full coding regions were sequenced to verify the products.
Preparation of Phosphorylated Forms of cPLA 2 ␣-cPLA 2 ␣-505P and cPLA 2 ␣-515P were prepared by a modification of the previously reported methods (16, 24) (supplemental material). cPLA 2 ␣-505P/727P was prepared from activated platelets based on the method reported by Kramer (39). Full details are given as supplemental material along with data showing that the phosphorylated forms are stoichiometrically phosphorylated.
Vesicle Binding Assays-All phospholipids were stored in CHCl 3 in Teflon septum-lined screw cap vials under argon at Ϫ20°C. Concentrations of phospholipids were determined by the standard phosphate assay with ammonium molybdate(VI) tetrahydrate (for PIP 2 , the weight specified by the manufacturer was used). Stock solutions of phospholipids were mixed in a polypropylene microcentrifuge tube, and most of the solvent was removed (leaving ϳ50 l) with a stream of argon with the tube placed in a 37°C bath. The tube was placed in a SpeedVac concentrator under vacuum for 1-2 h. All phospholipid mixtures used for binding studies contained 0.5 mol % lissamine rhodamine B 1,2-dihexadecanoyl-sn-glycero-3-phosphoethanolamine. Vesicles were prepared by freeze-thawing the lipid suspension (typically 4 -6 mM phospholipid) in 10 mM MOPS, pH 7.2, 176 mM sucrose followed by extrusion through two 0.2-m polycarbonate membranes (VWR catalog number 110606) using a LiposoFast device (Avestin), as described previously (40). A small aliquot of lipid suspension prior to extrusion was removed for fluorimetric analysis (see below). Extruded vesicles were submitted to a prespin procedure (1 h, 100,000 ϫ g average , 21°C) in a 1.5-ml polyallomar centrifuge tube (Beckman catalog number 357448). The supernatant was removed with a pipette and discarded, and the vesicles in the pellet were resuspended by gentle mixing (gentle up and down passage with a Pipetman, 8 -10 times) in buffer A (10 mM MOPS, pH 7.2, 100 mM KCl, 0.5 mM EGTA). A small aliquot of vesicles was diluted into 50 mM Tris-HCl, pH 8.0, 1% sodium cholate and submitted to fluorometry (excitation, 550 nm; emission, 590 nm). The fluorescence value was compared with that of an aliquot of the vesicle suspension prior to extrusion to calculate the percentage yield of phospholipid, and this was used with the phospholipid concentrations in the original stock solutions to obtain the total phospholipid concentration of final vesicle suspension. Vesicles were used on the same day of preparation (stored at room temperature). cPLA 2 ␣ binding reactions were prepared in 1.5-ml polyallomar microcentrifuge tubes and contained buffer A with 0.5 mg/ml BSA containing various amounts of free calcium (see below), phospholipid (added from the calibrated, prespun stock in buffer A), and 50 ng of cPLA 2 ␣ in a total volume of 1 ml. Samples were centrifuged at 100,000 ϫ g average at 21°C for 1 h, and two 100-l aliquots of supernatant were submitted to duplicate radiometric cPLA 2 ␣ assays (see below).
Buffer A containing various amounts of free calcium (0 -5 M) were prepared using EGTA/CaCl 2 mixtures and calciumbinding fluorophores, as described (41) (supplemental material). The amount of cPLA 2 ␣ in the supernatant above pelleted vesicles was determined using a modification of the previously reported radiometric assay (41) (supplemental material).
An interesting feature of adding PIP 2 to PAPC vesicles that we observed is that when the total phospholipid concentration was dropped from 200 to 25 M, most of the PIP 2 was found in the aqueous phase above vesicles that were pelleted by ultracentrifugation (see supplemental material for more details). This presumably reflects the relative high aqueous phase solubility of PIP 2 compared with phospholipids that lack highly polar, phosphorylated inositol headgroups. cPLA 2 ␣ Kinetic Studies-Vesicles contained [ 14 C]PAPC at a specific activity of 2.7 Ci/mol (made by mixing [ 14 C]PAPC (50 Ci/mol) with PAPC) and other phospholipids as noted. Phospholipids were mixed together as CHCl 3 solutions, and solvent was removed as described above. Buffer A was added to the dry lipid film, and vesicles were prepared by freeze-thawing followed by extrusion as described above. Extruded vesicles were diluted typically 20-fold into buffer A containing 0.5 mg/ml BSA and various concentrations of calcium (see above) to give a volume of 0.1 ml. Tubes were prewarmed in a 37°C bath for 5 min, and reactions were started by adding a 2-l aliquot of enzyme stock containing 200 -250 ng of cPLA 2 ␣ (enzyme stocks were made by fresh dilution into buffer A containing 1 mg/ml BSA). Reactions were quenched after 2 min, and released 14 C-labeled AA was measured after extraction and silica chromatography as previously described (41).
Cell Culture and AA Release Assay-Lung fibroblasts were isolated from wild type (MLF ϩ/ϩ ) and cPLA 2 ␣ knock-out (MLF Ϫ/Ϫ ) mice, and SV40-transformed MLF (IMLF) were generated as described previously (42). IMLF ϩ/ϩ and IMLF Ϫ/Ϫ were plated at 1.25 ϫ 10 4 cells/cm 2 in 250 l of DMEM containing 10% fetal bovine serum, 0.1% nonessential amino acids, 1 mM sodium pyruvate, 100 units/ml penicillin, 100 g/ml streptomycin, 0.29 mg/ml glutamine (growth medium) in 48-well plates. After 18 h in 5% CO 2 at 37°C, cells were washed with and incubated in 100 l of serum-free and antibiotic-free DMEM containing 0.1% BSA (stimulation medium) and adenoviruses. After 1.5 h, 150 l of stimulation medium containing 0.2 Ci/ml [ 3 H]AA was added to each well. After 26 h, cells were washed twice, and fresh stimulation medium was added. When inhibitors were tested, they were added (10 M SB203580, 10 M U0126, 1 M wortmannin, 10 M KN93, 10 M GF109203X, and 15 ng/l CHX) at various times before stimulation with 100 nM PMA or 10% mouse serum. Culture medium was collected after stimulation and centrifuged for 10 min at 15,000 rpm, and radioactivity was determined by scintillation counting. Cells were scraped into 50 l of 0.1% Triton X-100 containing protease inhibitors, and the lysates were used to determine the total cellular radioactivity and to determine expression levels via immunoblotting. The amount of AA release into the culture medium was calculated as a percentage of the total radioactivity (cells plus medium) in each well.
Microscopy-IMLF ϩ/ϩ and IMLF Ϫ/Ϫ were plated at 1.25 ϫ 10 4 cells/cm 2 in 250 l of growth medium in glass-bottomed MatTek plates and infected with adenoviruses for expression of wild type and mutant cPLA 2 ␣ as described above. After 26 h, cells used for calcium analysis were loaded with 5 M Fura-Red-AM for 30 min at room temperature in the presence of 0.02% pleuronic acid, washed, and incubated for 30 min in Hanks' balanced salt solution containing 25 mM HEPES, pH 7.4, and 2 mM probenecid. For some microscopy, cells were fixed for 15 min with 3% paraformaldehyde in PBS containing 3% sucrose, permeabilized for 30 min with 0.1% Triton X-100 in PBS, and blocked in 5% fetal bovine serum in PBS for 1 h. Golgi was labeled using anti-giantin primary antibody (1:200 in blocking solution, 1 h), followed by a Texas Red-conjugated secondary antibody (1:100 in blocking solution; 1 h). Microscopy was conducted on an inverted Zeiss 200 M microscope driven by Intelligent Imaging Innovations Inc. (3I) software (Slidebook 4.1). Fluorescence data were calculated by subtract-ing background fluorescence, and correcting for differential bleaching at each wavelength. Calcium ratios were corrected for background fluorescence values.
Immunoblotting-For Western blotting, cell lysates were prepared in ice cold buffer containing 50 mM Hepes, pH 7.4, 150 mM sodium chloride, 1.5 mM magnesium chloride, 10% glycerol, 1% Triton X-100, 1 mM EGTA, and protease inhibitors. Lysates were centrifuged at 15,000 ϫ g for 10 min at 4°C, and protein concentration was determined using the bicinchoninic acid reagent. Lysates were diluted in Laemmli buffer and boiled for 5 min at 100°C. Proteins were separated on 10% SDS-polyacrylamide gels, transferred to nitrocellulose, and blocked for 1 h in Tris-buffered saline containing 0.25% Tween 20 and 5% nonfat dry milk. Nitrocellulose membranes were incubated overnight with a 1:5,000 dilution of antiserum to total cPLA 2 ␣, 1:1,000 of phosphospecific cPLA 2 ␣ antiserum, or 1:1,000 of anti-p38 or anti-phospho-ERK antibodies. Antibodies were diluted in blocking buffer. Immunoreactive protein was detected using the Amersham Biosciences anti-rabbit secondary antibody and ECL system.

RESULTS
Regulation of cPLA 2 ␣ in IMLF ϩ/ϩ Stimulated with Serum and PMA-To study the role of calcium and phosphorylation in regulating cPLA 2 ␣, we chose to use serum and PMA as agonists. They both trigger AA release and activate extracellular regulated protein kinases (ERKs) in fibroblasts (43,44). Serum increases [Ca 2ϩ ] i , but PMA has been reported to have little effect on [Ca 2ϩ ] i (45). This allows us to compare the role of phosphorylation sites and the PIP 2 binding residues in cPLA 2 ␣ at different [Ca 2ϩ ] i . In addition, cPLA 2 ␣ phosphorylation may be functionally more important at low [Ca 2ϩ ] i (17,25). Serum is a complex pathophysiological fluid produced in response to tissue injury as a result of platelet aggregation. It is relevant to lung fibroblasts, since it leaks into the lung during acute lung injury and contributes to fibroproliferation (46,47). Our first approach was to define the characteristics of endogenous cPLA 2 ␣ stimulation by serum and PMA in IMLF ϩ/ϩ . A time course of AA release by cPLA 2 ␣ in IMLF ϩ/ϩ was conducted. When stimulated with 10% serum, AA was rapidly released in the first 10 min and then slowed through 80 min (Fig. 1A, right). Serum stimulated high levels of AA release that were 10% of total cell-associated radiolabeled AA at 10 min and reached 20% at 80 min. In contrast, PMA stimulated a slower mobilization of AA that increased linearly through 80 min (Fig. 1A, left). Compared with serum, PMA stimulated much less AA release, which was less than 1% of total cell-associated radiolabeled AA at 10 min and reached 4 -5% at 80 min.
IMLF ϩ/ϩ were stimulated with serum and PMA to determine the effect on [Ca 2ϩ ] i . Serum caused a rapid increase in [Ca 2ϩ ] i at 30 s (Fig. 1B, panel 1). The initial calcium transient decreased by 60 s, and this was followed by a sustained calcium increase characterized by low amplitude oscillations that continued for at least 30 min (data not shown). The sustained phase but not the initial spike in calcium was eliminated by incubation of the cells in medium containing EGTA to chelate extracellular calcium (Fig. 1B, panel 2). The results demonstrate that serum stimulation of IMLF ϩ/ϩ induces an initial increase in [Ca 2ϩ ] from intracellular stores followed by capacitative influx of extracellular calcium. In contrast, stimulation of cells with PMA did not induce an initial high amplitude increase in [Ca 2ϩ ] i but promoted low amplitude oscillations in many cells that lasted at least 30 min (Fig. 1B, panel  3). The oscillations did not occur in cells incubated in medium containing EGTA (Fig. 1B, panel 4) or in control cells treated with DMSO vehicle (Fig. 1B, panel 5).
Activation of ERKs and p38 plays a role in regulation of cPLA 2 ␣-mediated AA release by many agonists in part by phosphorylation of cPLA 2 ␣ on Ser-505. Western blot analysis using phospho-ERK antibodies indicated that serum and PMA rap- idly activated ERKs by 0.5-1 min in IMLF ϩ/ϩ ( Fig. 2A). The use of antibodies that detected total ERK protein indicated equal loading of samples (data not shown). Serum also activated p38 ( Fig. 2A); however, no detectable activation of p38 occurred during stimulation with PMA (not shown). Additional experiments demonstrated that activation of ERKs in response to serum and PMA and activation of p38 in response to serum were evident up to 80 min after stimulation (data not shown).
Phosphorylation of cPLA 2 ␣ on Ser-505 was determined by Western blotting using phosphospecific antibody (Fig. 2B). cPLA 2 ␣ was constitutively phosphorylated in unstimulated IMLF ϩ/ϩ , and phosphorylation was not significantly increased by serum or PMA. cPLA 2 ␣ remained phosphorylated on Ser-505 for up to 80 min in unstimulated cells and in cells treated with serum or PMA (data not shown). Phosphorylation of cPLA 2 ␣ on Ser-505 causes a decrease in electrophoretic mobility (gel shift) of cPLA 2 ␣ on SDS-polyacrylamide gels (21). In unstimulated IMLF ϩ/ϩ , most of the cPLA 2 ␣ was gel-shifted (data not shown), consistent with results of Western blots using phosphospecific antibody. This is not unexpected, since we found that cPLA 2 ␣ is constitutively phosphorylated on Ser-505 in immortalized cell lines (27). 4 EGFP-cPLA 2 ␣ expressed in IMLF Ϫ/Ϫ was also constitutively phosphorylated on Ser-505 when assayed by Western blotting with phosphospecific antibody (data not shown). The specificity of the phosphospecific antibody was determined by comparing lysates of IMLF Ϫ/Ϫ expressing either wild type ECFP-cPLA 2 ␣ or the phosphorylation site mutant ECFP-cPLA 2 ␣S505A. Western blots of cell lysates probed with antibody to total cPLA 2 ␣ demonstrated that the wild type and mutant enzymes were expressed at equivalent levels (Fig. 2C). However, using the phosphospecific antibody, a signal for cPLA 2 ␣ was only observed in lysates of IMLF Ϫ/Ϫ expressing wild type cPLA 2 ␣ but not the S505A mutant, confirming that the antibody is specific for cPLA 2 ␣ phosphorylated on Ser-505 (Fig. 2C).
Inhibitors were used to determine if the activation of p38 and ERKs plays a role in cPLA 2 ␣-mediated AA release in IMLF ϩ/ϩ . Inhibition of p38 by SB203580 and of MEK1 by U0126, to block ERK activation, decreased serum-induced AA release by 64 and 91%, respectively (Fig. 3A). AA release stimulated by PMA, which does not activate p38, was not affected by SB203580, as expected. The MEK1 inhibitor blocked AA release stimulated by PMA by 83% (Fig. 3B). The results demonstrate that the MEK1/ERK pathway regulates cPLA 2 ␣-mediated AA release in response to serum and PMA, and p38 contributes to the regulation of AA release in serum-stimulated cells.
In order to elucidate signaling cascades that might be involved in the activation of cPLA 2 ␣ by serum and PMA, inhibitors of kinases implicated in cPLA 2 ␣ activation in previous studies were employed. CaMKII regulates cPLA 2 ␣ through a MAP kinase-dependent pathway involving phosphorylation of cPLA 2 ␣ on Ser-515 in vascular smooth muscle cells (16,20,48). The CaMKII inhibitor KN93 had no affect on AA release stimulated by serum or PMA (Fig. 3, C and D). Polyphosphoinositides, including phosphatidylinositol 3,4,5-trisphosphate, activate cPLA 2 ␣ in vitro and may regulate cPLA 2 ␣-mediated AA release (36). However, AA release in IMLF ϩ/ϩ was not blocked by the phosphatidylinositol 3-kinase inhibitor, wortmannin (1 M) in response to serum or PMA stimulation (Fig. 3, C and D). The ability of PMA to stimulate AA release implicates a role for PKC in regulating cPLA 2 ␣ activation. To determine if PKC activation is required for serum-stimulated AA release, the general PKC inhibitor GF109203X was tested. PMA activation of cPLA 2 ␣ in IMLF ϩ/ϩ was inhibited by 95% by GF109203X (Fig.  3D). Serum-stimulated AA release, however, was not significantly inhibited by blocking PKC (Fig. 3C).
Several studies have shown that protein synthesis is required for AA release (49 -51). To determine if protein synthesis is necessary for serum-and PMA-stimulated AA release, IMLF ϩ/ϩ were treated with CHX before stimulation. CHX blocked both serum-and PMA-induced AA release by 71 and 98%, respectively, indicating that nascent protein production is needed for cPLA 2 ␣ activation in IMLF ϩ/ϩ (Fig. 3, E and F). CHX does not prevent translocation of cPLA 2 ␣ to the Golgi in IMLF ϩ/ϩ stimulated with serum (data not shown). IMLF ϩ/ϩ 4 C. C. Leslie, unpublished observation. FIGURE 2. ERK and p38 activation by serum and PMA in IMLF ؉/؉ . Cell lysates of unstimulated (US) IMLF ϩ/ϩ or cells stimulated with 10% serum or 100 nM PMA were prepared at given times after stimulation. Activation of ERKs or p38 (A) or phosphorylation of cPLA 2 ␣ on Ser-505 (B) was determined by Western blotting using phosphospecific antibodies. Sample loading was determined using total ERK antibodies (data not shown) or antibodies to total cPLA 2 ␣ (B). Results are representative of three independent experiments. C, Western blots of lysates of serum-stimulated IMLF Ϫ/Ϫ expressing either wild type ECFP-cPLA 2 ␣ (WT) or EYFP-cPLA 2 ␣S505A were probed with antibodies to total cPLA 2 ␣ or phosphospecific antibodies to cPLA 2 ␣ phosphorylated on Ser-505. The Western blot confirms the specificity of the phosphospecific antibodies for cPLA 2 ␣ phosphorylated on Ser-505.
were treated with the inhibitors (CHX, U0126, and SB203580) that blocked AA release to determine their effect on phosphorylation of Ser-505 or, particularly relevant for CHX, the effect on total cPLA 2 ␣ protein levels. As shown by Western blotting (Fig. 3G), CHX had no effect on the total amount of cPLA 2 ␣ in IMLF ϩ/ϩ . CHX also did not affect the phosphorylation of cPLA 2 ␣ on Ser-505 that is observed in unstimulated IMLF ϩ/ϩ or in serum-or PMA-stimulated IMLF ϩ/ϩ . U0126 inhibited AA release by 80 -90% in response to PMA and serum, and this correlated with the inhibition of ERK activation by PMA and serum in IMLF ϩ/ϩ (Fig. 3G). There was a low level of constitutive activation of ERKs in IMLF ϩ/ϩ not treated with serum or PMA that was inhibited by U0126. This correlated with a decrease in the constitutive phosphorylation of cPLA 2 ␣ on Ser-505 observed in unstimulated IMLF ϩ/ϩ . However, neither U0126 nor SB203580 inhibited the phosphorylation of Ser-505 in IMLF ϩ/ϩ treated with PMA or serum. The results suggest that MAPK activation regulates cPLA 2 ␣mediated AA release by a novel mechanism in addition to Ser-505 phosphorylation.
Reconstitution of AA Release in IMLF Ϫ/Ϫ by Expression of cPLA 2 ␣-To study mechanisms for cPLA 2 ␣ activation by PMA and serum in more detail, wild type and mutant cPLA 2 ␣ were expressed in IMLF Ϫ/Ϫ using adenovirus. Expression of fluorescent protein-tagged cPLA 2 ␣ in cells lacking endogenous cPLA 2 ␣ enabled us to directly compare the effect of specific mutations on AA release and on translocation of cPLA 2 ␣ to Golgi when stimulated with PMA and serum. AA release, in response to serum and PMA, increases in proportion to levels of fluorescent protein-tagged cPLA 2 ␣ expression up to a saturation point (Fig. 4, A and B). To determine the role of extracellular calcium on AA release, IMLF Ϫ/Ϫ expressing EGFP-cPLA 2 ␣ were incubated in medium containing EGTA. Serum-and PMA-induced AA release was blocked to nearly basal levels by chelating extracellular calcium (Fig.  5A). Since activation of the MEK1/ ERK pathway is required for cPLA 2 ␣-mediated AA release in response to serum and PMA in IMLF ϩ/ϩ , we determined if chelation of calcium by EGTA affected the activation of MAPK pathways. Western blots illustrated that chelation of extracellular calcium with EGTA does not block p38 or ERK activation (Fig. 5B).
To confirm a role for a functional C2 domain in cPLA 2 ␣mediated AA release stimulated by serum and PMA, parallel cultures of IMLF Ϫ/Ϫ expressing matching amounts of wild type ECFP-cPLA 2 ␣ and the C2 domain mutant EYFP-cPLA 2 ␣D43N were compared. Asp-43 interacts with both Ca 2ϩ ions that bind the C2 domain and plays a critical role in mediating cPLA 2 ␣ membrane binding (28,52,53). AA release from IMLF Ϫ/Ϫ expressing the D43N mutant was significantly attenuated in response to serum and PMA, indicating that calcium binding to the C2 domain is required for [ 3 H]AA released into the medium was measured and presented as a percentage of total cellular [ 3 H]AA. G, IMLF ϩ/ϩ were preincubated with SB203580, U0126, or CHX or no treatment followed by stimulation with serum for 10 min or PMA for 45 min, as described above. Cell lysates of unstimulated (US) or stimulated IMLF ϩ/ϩ were analyzed by Western blotting to determine activation of ERKs (p-ERK) or phosphorylation of cPLA 2 ␣ on Ser-505 (p-cPLA 2 ␣) using phosphospecific antibodies. Sample loading was determined using antibodies to total cPLA 2 ␣ or antibodies to ␤-tubulin. cPLA 2 ␣-mediated AA release in response to these agonists (Fig. 5C).
Translocation of cPLA 2 ␣ in Response to Serum and PMA Stimulation-Live cell imaging of IMLF Ϫ/Ϫ expressing EGFP-cPLA 2 ␣ demonstrated that translocation of cPLA 2 ␣ to the perinuclear region occurred within seconds of serum stimulation (Fig. 6A). In response to serum, EGFP-cPLA 2 ␣ co-localized with the Golgi marker giantin (Fig. 6B). Most IMLF are bi-or multinucleated, a characteristic of SV40 transformation (54). We observed that the Golgi apparatus is often sandwiched between the nuclei (Figs. 6B; see also Fig. 7, A and B). Translocation of EGFP-cPLA 2 ␣ to Golgi occurred in parallel with the sharp increase in [Ca 2ϩ ] i induced by serum (Fig. 6C). There was cell to cell variation in the amount of EGFP-cPLA 2 ␣ that dissociated from the Golgi after the initial spike in calcium subsided, but most cells showed a partial (20 -30%) decrease in Golgi fluorescence by 60 s after serum addition (Fig. 6C). Chelating extracellular calcium with EGTA did not block the initial calcium spike induced by serum (see Fig. 1B). However, in most cells, we did not observe translocation of cPLA 2 ␣ to Golgi when cells were incubated in medium with EGTA. In a few cells incubated with EGTA, only a transient increase in cPLA 2 ␣ translocation to Golgi occurred, and fluorescence returned to base line by 60 s, as shown for three cells in Fig. 6D. In contrast, in cells incubated without EGTA, a larger proportion of EGFP-cPLA 2 ␣ remained associated with Golgi throughout the 160-s time course (Fig. 6C). In response to serum, ϳ7% of the total cellular EGFP-cPLA 2 ␣ translocated to Golgi in cells incubated without EGTA, and this decreased to 3% in the few cells where translocation was visually evident when incubated with EGTA (Fig. 6E). The data indicate that the sustained increase in [Ca 2ϩ ] i induced by serum is required for stable binding of cPLA 2 ␣ to Golgi and AA release. Dual imaging of IMLF Ϫ/Ϫ co-expressing wild type ECFP-cPLA 2 ␣ and EYFP-cPLA 2 ␣D43N illustrated that the D43N mutant did not translocate upon serum stimulation (Fig.  6F). Thus, a sustained increase in [Ca 2ϩ ] i is required for the C2 domain-dependent translocation to Golgi and AA release in serumstimulated IMLF. We next investigated the effect of PMA on translocation of ECFP-cPLA 2 ␣. In response to PMA stimulation, ECFP-cPLA 2 ␣ translocated to the perinuclear region (Fig. 7A), where it co-localized with giantin at the Golgi (Fig. 7B). In contrast to serum, translocation of ECFP-cPLA 2 ␣ was not detected until 8 -10 min after PMA addition and reached maximal translocation ϳ15 min after stimulation (Fig. 7, A and  C). Dual imaging of ECFP-cPLA 2 ␣ and the C2 domain mutant EYFP-cPLA 2 ␣D43N co-expressed in IMLF Ϫ/Ϫ revealed little translocation of the mutant (Fig. 7C). Collectively, the results suggest that the oscillations induced by PMA from the influx of extracellular calcium, promote the C2 domain-dependent translocation of cPLA 2 ␣ to Golgi and AA release.
Analysis of Phosphorylation Site Mutants Expressed in IMLF Ϫ/Ϫ -The functional role of cPLA 2 ␣ phosphorylation was investigated by expressing wild type cPLA 2 ␣ and phosphorylation site mutants in IMLF Ϫ/Ϫ and comparing AA release and translocation. In experiments analyzing AA release, equal expression of wild type and mutant cPLA 2 ␣ was achieved by expressing three dilutions of adenovirus in neighboring wells. AA release experiments were conducted on all wells, and Western blotting was used to determine the relative expression level of cPLA 2 ␣ in each well for comparison of mutant and wild type cPLA 2 ␣. IMLF Ϫ/Ϫ expressing EYFP-cPLA 2 ␣S505A showed a 38% decrease in AA FIGURE 6. Serum-stimulated cPLA 2 ␣ translocation to Golgi correlates with [Ca 2؉ ] i increase and is dependent on a functional C2 domain. A, IMLF Ϫ/Ϫ expressing EGFP-cPLA 2 ␣ were incubated in phenol red free DMEM and stimulated with 10% serum. Live cell images were collected every 3 s using an FITC filter and a ϫ40 oil immersion objective. Images are representative of 10 individual experiments. B, IMLF Ϫ/Ϫ expressing EGFP-cPLA 2 ␣ were incubated in phenol red-free DMEM, fixed 2 min after stimulation with serum, and then probed with anti-giantin primary antibody and Texas Red secondary antibody to visualize Golgi. C, IMLF Ϫ/Ϫ expressing EGFP-cPLA 2 ␣ were loaded with FuraRed-AM, and live cell images were collected using FITC, F403, and F470 filters after stimulation with serum (arrow). D, IMLF Ϫ/Ϫ expressing EGFP-cPLA 2 ␣ were incubated in medium containing EGTA, and then images were collected using a FITC filter after stimulation with serum (arrow). E, translocation data from C and D were analyzed to determine the percentage of EGFP-cPLA 2 ␣ bound to Golgi at the peak of serum-induced translocation (ϳ30 s) in cells incubated with and without extracellular EGTA. Translocation data were calculated based on average fluorescence intensity of EGFP-cPLA 2 ␣ on the Golgi in each cell. Values were corrected for background fluorescence and differential bleaching at each wavelength through the duration of the imaging and expressed relative to time 0 (F T /F 0 ). Calcium ratios (F403/F470) were calculated and corrected for background fluorescence and expressed relative to time 0 (R T /R 0 ). Graphs are representative of 10 cells from three independent experiments. F, IMLF Ϫ/Ϫ co-expressing ECFP-cPLA 2 ␣ and EYFP-cPLA 2 ␣D43N were stimulated with serum, and translocation was determined as described in A. Data are presented relative to time 0 (F T /F 0 ). The graph is representative of 15 cells from three independent experiments. release in response to serum compared with IMLF Ϫ/Ϫ expressing wild type ECFP-cPLA 2 ␣ (Fig. 8A). Analysis of cPLA 2 ␣ phosphorylation in several cell lines has revealed that phosphorylation of Ser-505 occurs with phosphorylation of Ser-727 (13,17,18). Therefore, the ability of EYFP-cPLA 2 ␣S727A and EYFP-cPLA 2 ␣S505A/S727A to release AA was tested. AA release from IMLF Ϫ/Ϫ expressing EYFP-cPLA 2 ␣S727A was not significantly different from IMLF Ϫ/Ϫ expressing wild type ECFP-cPLA 2 ␣ (Fig. 8A). Equal expression of wild type cPLA 2 ␣ and the phosphorylation site mutant was confirmed by Western blotting (Fig. 8A, inset). IMLF Ϫ/Ϫ expressing EYFP-cPLA 2 ␣S505A/ S727A released ϳ50% less AA than IMLF Ϫ/Ϫ expressing equivalent levels of wild type ECFP-cPLA 2 ␣ at both 10 and 20 min poststimulation by serum (Fig. 8B). AA release from IMLF Ϫ/Ϫ expressing the S505A or S505A/S727A mutants was not reduced to the extent observed using the MEK1 inhibitor U0126 (see Fig. 3A). This observation is consistent with our previous studies suggesting that ERKs also play a role in regulating cPLA 2 ␣-mediated AA release by a mechanism independent of phosphorylation on Ser-505 (27,49). This was confirmed by data showing that U0126 inhibited the residual AA release from IMLF Ϫ/Ϫ expressing EYFP-cPLA 2 ␣S505A/S727A by ϳ51% (average of two experiments) (data not shown).
Interfacial Binding and Kinetics of cPLA 2 and Its Phosphorylated Forms in Vitro-Our results show that suboptimal AA release by phosphorylation site mutants is not due to a defect in translocation to the Golgi or stable membrane binding in response to physiological increase in calcium induced by serum. These data are not consistent with a previous report showing that phosphorylation on Ser-505 increases the membrane affinity of cPLA 2 ␣ in cells and in vitro (25). Therefore, we measured the interfacial kinetics and binding of different phosphorylated forms of cPLA 2 ␣ in vitro. To directly measure the binding of cPLA 2 ␣ to vesicles in vitro, we measured the amount of enzyme in the supernatant above vesicles that were pelleted using ultracentrifugation (vesicles were loaded with sucrose to allow them to pellet). As shown in Fig. 9A, the amount of dephosphorylated cPLA 2 ␣ (cPLA 2 ␣-PAP) bound to PAPC vesicles increases as the concentration of free Ca 2ϩ increased from 0 to 22 M. A concentration of PAPC of 200 M was used, which approximates the estimated concentration of phospholipid that cPLA 2 ␣ encounters inside of mammalian cells (55). The buffer was chosen to give physiological pH and ionic strength. Centrifugation studies carried out in the absence of vesicles showed no loss of cPLA 2 ␣-PAP from the buffer solution at all concentrations of Ca 2ϩ (supplemental Fig. 2); thus, depletion of enzyme from the supernatant is the result of interfacial binding to vesicles. From the data in Fig. 9A, we obtain the concentration of calcium that allows 50% of the cPLA 2 ␣-PAP to bind to vesicles. We denote this value as app K Ca to reflect the fact that it is an apparent dissociation equilibrium constant that is composed of all calcium-dependent steps (56). From the same fig- ure, we determine the fraction of enzyme bound to vesicles in the absence of Ca 2ϩ and the fraction of enzyme bound at saturating Ca 2ϩ . This interfacial binding experiment was repeated using cPLA 2 ␣ that was stoichiometrically phosphorylated on Ser-505 only (cPLA 2 ␣-505P) (Fig. 9A) or on S515 (cPLA 2 ␣-515P) (supplemental Fig. 3). Results for all enzyme forms are summarized in Table 1. cPLA 2 ␣-PAP, cPLA 2 ␣-505P, and cPLA 2 ␣-515P are all about 50% bound to 200 M PAPC vesicles in the absence of Ca 2ϩ , and increasing Ca 2ϩ causes virtually all of the proteins to become fully vesicle-bound. Phosphorylation of cPLA 2 ␣ on Ser-505 or on Ser-515 causes little, if any, statistically significant change in app K Ca (Table 1).
We also determined the Ca 2ϩ dependence of hydrolysis of 200 M [ 14 C]PAPC vesicles by phosphorylated forms of cPLA 2 ␣. As shown in supplemental Fig. 4, the amount of 14 C-labeled arachidonate released from [ 14 C]PAPC vesicles rises steadily from 0 to 30 min; thus, studies carried out by quenching the reaction after a 2-min incubation provide the initial reaction velocity. As shown in Fig. 9B, the enzymatic activity of cPLA 2 ␣-PAP rises as a function of increasing concentration of Ca 2ϩ . As for interfacial binding, app K Ca was obtained as the concentration of Ca 2ϩ that causes 50% of maximal activation of enzyme. Results for cPLA 2 ␣-PAP, cPLA 2 ␣-505P, and cPLA 2 ␣-505P/727P are also shown in Fig. 9B, and those for cPLA 2 ␣-515P are shown in supplemental Fig. 5. Interfacial kinetic values are summarized in Table 2. All three enzymes show only a small amount of activity in the absence of calcium (12% or less of that seen at saturating Ca 2ϩ ) and display very similar app K Ca values. The specific activities of cPLA 2 ␣-505P and cPLA 2 ␣-505P/727P at saturating Ca 2ϩ are about 2-fold higher than that of cPLA 2 ␣-PAP, whereas cPLA 2 ␣-515P shows a specific activity that is not statistically different from that for nonphosphorylated enzyme. Treatment of plateletderived cPLA 2 ␣-505P/727P with PAP to remove phosphates yielded an enzyme that showed values of app K Ca and maximum activity at saturating Ca 2ϩ statistically indistinguishable from those of recombinant cPLA 2 ␣-PAP (data not shown), strongly suggesting that the altered properties of the enzyme isolated from platelets is due solely to phosphorylation. The fact that values of app K Ca observed for interfacial binding and kinetic studies are statistically indistinguishable suggests that the sole function of Ca 2ϩ is to support interfacial binding of enzyme. The data also indicate that phosphorylation of cPLA 2 ␣ on Ser-505 functions to increase the catalytic activity but does not affect calcium-dependent membrane affinity, consistent with the results in IMLF.
Analysis of Basic Residue Mutants Expressed in IMLF Ϫ/Ϫ -In addition to phosphorylation, basic residues in the catalytic domain (Lys-488, Lys-541, Lys-543, and Lys-544) are implicated in cPLA 2 ␣ regulation through interaction with PIP 2 (33,37). Basic residues Lys-488, Lys-543, and Lys-544 in cPLA 2 ␣ were mutated to asparagines to determine their role in regulating cPLA 2 ␣. AA release from IMLF Ϫ/Ϫ expressing the EYFP- [ 3 H]AA released into the medium is expressed as a percentage of the total cellular radioactivity in each well. Immunoblotting was conducted to determine expression levels of wild type and mutant cPLA 2 ␣ in each well (insets). [ 3 H]AA release is shown from wells with matching expression levels. The release of [ 3 H]AA by the mutants was significantly less (p Ͻ 0.05) than by wild type cPLA 2 ␣, as indicated (*). Live cell images of IMLF Ϫ/Ϫ co-expressing wild type ECFP-cPLA 2 ␣ and either EYFP-cPLA 2 ␣S505A/S727A (C) or EYFP-cPLA 2 ␣S505A (D) were collected every 3 s after serum stimulation, using CFP and YFP filters and a ϫ40 oil immersion objective. Translocation to Golgi in cells expressing wild type (gray lines) or mutant cPLA 2 ␣ (black lines) is shown for two representative cells. Values are corrected for background fluorescence and differential bleaching and are presented relative to time zero (F T /F 0 ). Data are representative of 20 individual cells from three independent experiments. cPLA 2 ␣K488N/K543N/K544N triple mutant was decreased by 78%, compared with cells expressing an equivalent amount of wild type ECFP-cPLA 2 ␣ in response to serum (Fig. 10A). In contrast, there was no significant decrease in AA release from IMLF Ϫ/Ϫ expressing the triple mutant in response to PMA (Fig.  10B). Live cell imaging showed a similar initial rate of translocation of EYFP-cPLA 2 ␣K488N/K543N/K544N and wild type ECFP-cPLA 2 ␣ in response to serum, but the mutant exhibited an increase in Golgi binding after the first 30 s (Fig. 10C). The results demonstrate that the decreased AA release by the basic residue mutant does not correlate with a defect in translocation.
Effect of PIP 2 on the Interfacial Properties of cPLA 2 ␣ in Vitro-As a correlate to the cellular experiments, we studied the effect of PIP 2 on interfacial binding and kinetics of wild type cPLA 2 ␣ and cPLA 2 ␣ containing mutations in basic residues in the catalytic domain. The role of phosphorylation on cPLA 2 ␣ activation by PIP 2 was also investigated. The results are summarized in Tables 1 and 2 (binding and kinetic curves are presented as supplemental Figs. 6 and 7). The extent of interfacial binding of wild type cPLA 2 ␣ to PAPC vesicles containing 10 mol % PIP 2 in the absence of Ca 2ϩ is similar to that seen with vesicles that lack PIP 2 (compare supplemental Figs. 6 and 7 with Fig. 9). The presence of PIP 2 in vesicles led to a modest, ϳ2-fold decrease in values of app K Ca (Table 1). Similar trends were seen with nonphosphorylated and various phosphorylated cPLA 2 ␣ proteins ( Table 1).
Interfacial kinetic studies summarized in Table 2 show that inclusion of 10 mol % PIP 2 in PAPC vesicles led to a ϳ5-fold increase in the specific activity of cPLA 2 ␣-PAP and all of its phosphorylated forms in the presence of saturating Ca 2ϩ (see also supplemental Figs. 5 and 8). Values of app K Ca were similar in the presence and absence of PIP 2 in PAPC vesicles.
We also studied cPLA 2 ␣ mutants in which basic residues possibly involved in interaction with PIP 2 were mutated to asparagine. Enzyme activity data versus the concentration of Ca 2ϩ for cPLA 2 ␣-PAP-K488N/K543N/K544N and cPLA 2 ␣-PAP-K271N/K273N/R274N are summarized in Table 2 (kinetic curves are shown in supplemental Fig. 8). cPLA 2 ␣-PAP-K488N/K543N/K544N showed an ϳ2-fold increase in specific activity compared with wild type enzyme on PAPC vesicles, and the mutant did not show rate enhancement when 10 mol % PIP 2 was added to PAPC vesicles (virtually identical results were obtained with cPLA 2 ␣-PAP-K541N/K543N/ K544N, a second triple site mutant involving basic residues occupying a similar region of the enzyme's surface; data not shown). The mutant cPLA 2 ␣-PAP-K271N/K273N/R274N displayed the same activity as wild type enzyme on PAPC vesicles and showed a rate enhancement by 10 mol % PIP 2 about half of that seen for wild type enzyme. Values of app K Ca for all mutants were similar to those for wild type enzymes with both PAPC and PAPC/PIP 2 vesicles ( Table 2). The results of the in vitro experiments support the cellular studies and implicate a role for After ultracentrifugation, two 100-ml aliquots of the supernatant were each submitted to the standard radiometric cPLA 2 ␣ assay to determine the amount of enzyme not bound to vesicles. The latter is plotted as the percentage of enzyme added to each binding solution, where 100% corresponds to the radiometric assay signal measured for 5 ng of cPLA 2 ␣ added directly from the stock solution to the radiometric assay mixture (see "Experimental Procedures"). Each experimental condition was carried out in duplicate, and both data points are plotted.  the basic residues in the catalytic domain of cPLA 2 ␣ in regulating hydrolytic activity through interaction with anionic phospholipids and not by enhancing calcium-dependent membrane binding.

DISCUSSION
cPLA 2 ␣ activity is controlled by complex post-translational mechanisms. The importance of lipid mediators produced as a result of cPLA 2 ␣ activation necessitates tight multidimensional regulation (6). Since cPLA 2 ␣ is expressed in most normal tissues, its function must be controlled in response to diverse stimuli by means appropriate to that cell type and circumstance. Reflecting this, AA release is stimulated by a variety of agonists that increase [Ca 2ϩ ] i and activate MAPKs, thereby initiating the signaling pathways for activation of cPLA 2 ␣ (1, 3, 7). In this study, we shed light on the mechanisms of cPLA 2 ␣ regulation by calcium, phosphorylation, and basic residues in the catalytic domain in fibroblasts stimulated with serum and PMA. The results are supported by our findings studying the role of phosphorylation and PIP 2 in regulating calciumdependent interfacial binding and kinetics of cPLA 2 ␣ in vitro.
Changes in [Ca 2ϩ ] i play a fundamental role in regulating cPLA 2 ␣ by promoting C2 domain-mediated translocation to membrane. Comparing translocation and AA release in response to serum and PMA allowed us to dissect out how cPLA 2 ␣ is regulated by differences in the mode of calcium mobilization. Serum induces a typical capacitative increase in [Ca 2ϩ ] i . A rapid, transient increase in [Ca 2ϩ ] i from intracellular stores is followed by the influx of extracellular calcium, resulting in a sustained, low amplitude [Ca 2ϩ ] i increase (57,58). We have previously suggested that PMA stimulates AA release without increasing [Ca 2ϩ ] i in macrophages (22,49). However, we found that PMA induces low amplitude [Ca 2ϩ ] i oscillations in IMLF ϩ/ϩ that were only detected by analyzing a large number of individual cells using live cell imaging, revealing considerable cell-to-cell heterogeneity in the extent of calcium mobilization. Our findings demonstrate that the low amplitude sustained phase of [Ca 2ϩ ] i due to influx of extracellular calcium is essential for AA release triggered by both serum and PMA in IMLF ϩ/ϩ . This is supported by results showing that mutation of a residue (Asp-43) in the C2 domain essential for binding calcium compromises the ability of cPLA 2 ␣ to translocate and release AA in response to serum and PMA in IMLF. In addition, chelating extracellular calcium with EGTA abolished AA release.
The initial high amplitude calcium transient induced by serum promotes the rapid translocation of cPLA 2 ␣ to Golgi and AA release. The maximal amount of cPLA 2 ␣ bound to Golgi occurred ϳ20 s after adding serum and represents ϳ7% of the total cellular ECFP-cPLA 2 ␣. In contrast to physiological agonists, such as serum, calcium ionophores used at concentrations that induce a high, sustained influx of calcium promote  translocation of a much larger proportion of cPLA 2 ␣ to membrane (10). As we previously reported, the release of calcium from intracellular stores promotes more rapid translocation of cPLA 2 ␣ to Golgi than does influx of extracellular calcium, suggesting the importance of local high calcium increases released from intracellular membrane stores for rapid translocation (10,34). Chelating extracellular calcium did not block the initial increase in [Ca 2ϩ ] i induced by serum but prevented AA release. Translocation of cPLA 2 ␣ to Golgi in response to serum was only briefly evident in a few cells when incubated in medium containing EGTA. This indicates that the influx of extracellular calcium triggered by store depletion is required to promote stable binding of cPLA 2 ␣ to membrane, which is necessary for mediating AA release. This is consistent with previous reports demonstrating the important role for capacitative calcium influx and the duration of calcium elevation in regulating cPLA 2 ␣-mediated AA release (59 -61). Thus, the two phases of calcium mobilization induced by serum serve to induce rapid and stable binding of cPLA 2 ␣ to membrane. We also demonstrate that the low amplitude sustained calcium oscillations induced by PMA promote translocation of cPLA 2 ␣ in the absence of a rapid calcium transient from intracellular stores. This occurs more slowly, taking several minutes for cPLA 2 ␣ to accumulate on the Golgi. There was considerable variation in the extent of cPLA 2 ␣ translocation in cells stimulated with PMA, reflecting the cell-to-cell variability in the extent of PMA-induced [Ca 2ϩ ] i oscillations. In contrast to PMA-stimulated cells, translocation of cPLA 2 ␣ was evident in over 90% of the serum-stimulated cells. Also, serum stimulates a greater proportion of cellular cPLA 2 ␣ to translocate to Golgi due to the rapid high amplitude release of calcium from intracellular stores. This may account for the higher levels of AA release induced by serum compared with PMA.
Previous studies identified a role for CaMKII activation and phosphorylation of cPLA 2 ␣ on Ser-515 in regulating AA release in smooth muscle cells stimulated with norepinephrine (16,20,48). However, we found no evidence for this pathway in regulating AA release in IMLF ϩ/ϩ . The CaMKII inhibitor had no effect on AA release from IMLF ϩ/ϩ , and AA release from IMLF Ϫ/Ϫ expressing the S515A mutant was the same as in cells expressing wild type cPLA 2 ␣. Therefore, the regulation of cPLA 2 ␣ by CaMKII is cell type-specific. We could not demonstrate a change in specific activity or app K Ca after stoichiometric phosphorylation of cPLA 2 ␣-PAP on Ser-515 by CaMKII. In a previous study, it was shown that treatment of cPLA 2 ␣ with CaMKII in vitro led to a 2-3-fold increase in specific activity (16). The reason for this discrepancy is not known. Therefore, it remains unclear how phosphorylation of cPLA 2 ␣ on Ser-515 in cells functions to regulate cPLA 2 ␣, especially since it has been shown that phosphorylation of Ser-515 or Ser-505 does not regulate cPLA 2 ␣ translocation in smooth muscle cells stimulated with norepinephrine (20).
Serum-and PMA-stimulated AA release is dependent on activation of MAPKs in IMLF ϩ/ϩ , as observed in many cell types (1). Activation of MAPKs by serum and PMA was not blocked by extracellular EGTA, indicating that it is not dependent on the influx of calcium. cPLA 2 ␣ is constitutively phosphorylated on Ser-505 in IMLF Ϫ/Ϫ , as we have observed in other cells lines (27). 4 This may be a consequence of culturing cells in the presence of serum, which activates MAPKs, resulting in cPLA 2 ␣ phosphorylation. We found that after overnight serum starvation, a protocol commonly used to quiesce cells, MAPKs exhibit only weak activation and can be strongly activated upon subsequent serum addition. However, cPLA 2 ␣ remains phosphorylated on Ser-505 during serum starvation, indicating that phosphorylation at Ser-505 is very stable. This is consistent with our previous data in macrophages showing that activation of ERKs in response to CSF-1 is very transient but leads to stable phosphorylation of cPLA 2 ␣ on Ser-505 (22). Our results in IMLF also confirm that phosphorylation on Ser-505 is not sufficient for cPLA 2 ␣ to mediate AA release, which requires an increase in [Ca 2ϩ ] i for translocation to membrane. Although MAPKs play a role in regulating cPLA 2 ␣ by phosphorylation of Ser-505, they clearly play an additional novel role in regulating cPLA 2 ␣, since agonist-induced AA release is blocked by MAPK inhibitors without affecting Ser-505 phosphorylation. We have previously made this observation in other cell types, suggesting that it is a commonly used regulatory pathway, although the mechanism involved remains to be determined.
Phosphorylation of cPLA 2 ␣ on Ser-505 enhances its activity in vitro and its ability to release AA in cells (14,15,17,24). We found that IMLF Ϫ/Ϫ expressing EYFP-cPLA 2 ␣S505A released ϳ38% less AA than cells expressing wild type ECFP-cPLA 2 ␣, indicating that phosphorylation on Ser-505 augments AA release but is not essential in IMLF. AA release in IMLF Ϫ/Ϫ expressing the double phosphorylation site mutant S505A/ S727A was attenuated to a similar extent (50%). A comparison of the translocation properties of wild type ECFP-cPLA 2 ␣ and the double EYFP-cPLA 2 ␣S505A/S727A and single EYFP-cPLA 2 ␣S505A phosphorylation site mutants showed similar initial rates of translocation of the mutants to Golgi in response to serum as wild type cPLA 2 ␣ that correlated with the initial rise in [Ca 2ϩ ] i . The phosphorylation site mutants tended to accumulate on Golgi to a greater extent than wild type cPLA 2 ␣. The reason for this is not known, but despite greater levels of the mutants on Golgi, their ability to release AA is not as efficient as that of wild type cPLA 2 ␣. Our data demonstrate that phosphorylation on Ser-505 does not augment translocation of cPLA 2 ␣ in response to a capacitative calcium increase (serum stimulation) or in response to a low oscillatory rise (PMA stimulation) that is not preceded by a high amplitude transient increase in calcium from intracellular stores. Thus, we find no evidence that phosphorylation on Ser-505 affects the calciumdependent membrane binding affinity of cPLA 2 ␣.
We found that mutating the phosphorylation site Ser-727 to alanine did not affect cPLA 2 ␣ translocation or AA release in cells. Also, Ser-727 phosphorylation did not influence cPLA 2 ␣ binding or catalytic activity in vitro. Unlike the role of Ser-505 phosphorylation in enhancing cPLA 2 ␣ catalytic activity, a recent study has identified a novel function for Ser-727 phosphorylation in regulating protein-protein interactions (62). Previous studies had identified a cPLA 2 ␣-binding protein, p11 (also called S100-A10 or calpactin I light chain) that forms a complex with annexin A2 in cells (63,64). p11 was reported to bind to the catalytic domain of cPLA 2 ␣ and inhibit its activity (63). A recent study found that the p11-annexin A2 complex binds to the hydroxyl group of Ser-727 and prevents binding of cPLA 2 ␣ to membrane (62). The interaction of cPLA 2 ␣ with the p11-annexin A2 complex is disrupted by phosphorylation of cPLA 2 ␣ on Ser-727, thus relieving inhibition. Mutating Ser-727 to alanine also disrupts interaction of cPLA 2 ␣ with the p11annexin A2 complex and mimics phosphorylation of Ser-727 (62). Our results showing that cPLA 2 ␣S727A behaves similarly to wild type cPLA 2 ␣ when expressed in IMLF Ϫ/Ϫ suggest that wild type cPLA 2 ␣ is constitutively phosphorylated on Ser-727 (as observed for Ser-505) or phosphorylated on Ser-727 in response to serum and PMA and relieved from potential inhibition by endogenous p11.
Our findings do not corroborate a report suggesting that phosphorylation of cPLA 2 ␣ on Ser-505 increases the membrane binding affinity of cPLA 2 ␣ particularly at low calcium concentrations induced by adding ionomycin (25). The reasons for the different findings are not known. However, our results in IMLF Ϫ/Ϫ correlate well with results of our in vitro experiments. Phosphorylation of cPLA 2 ␣ at Ser-505 and at Ser-505/Ser-727 (from platelets) leads to a 2-fold increase in the specific activity of the enzyme acting on PAPC vesicles, which is in line with the level of activation reported previously (e.g. see Refs. 24 and 39). We see no evidence that phosphorylation of cPLA 2 ␣ at Ser-505 dramatically alters the concentration of Ca 2ϩ needed to support interfacial binding. It has been reported that the K d for the nonphosphorylated S505A mutant of cPLA 2 ␣ is ϳ60-fold lower than that for wild type cPLA 2 ␣ purified from Sf9 cells (which is shown to be mainly phosphorylated on Ser-505) or the S505E mutant, which is suggested to be a mimic of Ser-505phosphorylated cPLA 2 ␣ (25). Even if we use the same vesicles and buffer reported in this previous study, we fail to see a significant effect of cPLA 2 ␣ phosphorylation on the amount of calcium required to support interfacial binding. The basis for this discrepancy between the two studies is not known. Our values of app K Ca have been measured by two independent methods (centrifugation and initial velocity versus [Ca 2ϩ ]). A calculation shown in the supplemental material shows that our values of app K Ca are close in magnitude to those reported in the earlier study (25).
Phosphorylation of cPLA 2 ␣ has been suggested to influence the conformation of the cPLA 2 ␣ catalytic domain on the membrane for optimal interaction with phospholipid substrate, thus augmenting the hydrolytic activity of cPLA 2 ␣ (65). Conformational effects may also be the basis for enhanced activity of cPLA 2 ␣ by interaction of the basic residues (Lys-488/Lys-541/ Lys-543/Lys-544) in the catalytic domain with PIP 2 (33). We found that the addition of PIP 2 (10 mol %) to PAPC (200 M) resulted in a 4-fold decrease in app K Ca and ϳ8-fold increase in the maximal activity of wild type cPLA 2 ␣ in the presence of saturating calcium but did not activate cPLA 2 ␣-K488N/ K543N/K544N. Our results are more in line with data showing 3.5-8-fold stimulation of cPLA 2 ␣ activity with PIP 2 using PC vesicles (33,35) and not the 120-fold increase observed using PC/Triton X-100 mixed micelles (36). In the absence of calcium, we found that PIP 2 did not change the amount of cPLA 2 ␣ catalytic activity, which remained low at about 7-10% of the amount of cPLA 2 ␣ bound to vesicles. It has been reported that the inclusion of PIP 2 in PAPC/Triton X-100 mixed micelles substrate allows the enzyme to act in a calcium-independent manner, yet the enzyme showed no activity in the absence of calcium when PIP 2 was omitted from mixed micelles (36,37). Since it is thought that the main role of calcium is to allow interfacial binding of cPLA 2 ␣, this result would suggest that PIP 2 allows the enzyme to bind to PAPC/Triton X-100 mixed micelles with similar affinity in the presence and absence of calcium. It is clear from the results of our work and a previous study that calcium-independent activation of cPLA 2 ␣ by PIP 2 does not occur with phosphatidylcholine liposomes (33). This suggests that with an aggregate that is predominantly composed of a neutral detergent, Triton X-100, and only a small amount of PAPC, small amounts of anionic phospholipid, such as PIP 2 , can induce calcium-independent binding, presumably via interactions of the PIP 2 with cationic residues in the catalytic domain of cPLA 2 ␣. Perhaps binding of the C2 domain to detergent micelles is not favorable compared with binding of this domain to phospholipid vesicles. In this case, electrostatic interaction of cationic residues of cPLA 2 ␣ and anionic PIP 2 in detergent micelles becomes a dominant mechanism for interfacial binding. It would seem that the use of phospholipid vesicles is more physiologically relevant than the use of detergent micelles.
Our results confirm that Lys-488/Lys-543/Lys-544 are important for regulating cPLA 2 ␣ activation in cells in response to serum stimulation. Expression of EYFP-cPLA 2 ␣K488N/ K543N/K544N in IMLF Ϫ/Ϫ resulted in a 78% decrease in AA release in response to serum compared with cells expressing wild type cPLA 2 ␣. The triple mutant and wild type cPLA 2 ␣ exhibit a similar initial rate of translocation to Golgi in response to serum-induced [Ca 2ϩ ] i increase. Surprisingly, the triple mutant accumulated on Golgi to a greater extent than wild type cPLA 2 ␣. This may explain our results and a previous report that the basic residue mutant has ϳ2-fold greater activity in vitro using PC vesicles (33). The cellular data suggest that interaction of the basic residues with anionic components on the membrane regulates catalytic activity and not calcium-dependent membrane binding, consistent with our in vitro experiments. The cell results do not establish that PIP 2 is the endogenous component in membranes that activates cPLA 2 ␣; nor does it establish whether the component is constitutively present or rapidly increased in response to serum. It has not been confirmed that PIP 2 is the endogenous activator of cPLA 2 ␣ in cells. It has recently been demonstrated that cPLA 2 ␣-mediated AA release is stimulated when intracellular levels of polyphosphoinositides are increased by feeding cells either PIP 2 or phosphatidylinositol 3,5-bisphosphate, suggesting that the response is not specific to PIP 2 , but again this does not establish that phosphorylated phosphatidylinositol is the endogenous activator of cPLA 2 ␣ (66). cPLA 2 ␣ preferentially targets Golgi, which contains very low levels of PIP 2 compared with the plasma membrane (67). A role for other anionic lipids or binding proteins in regulating cPLA 2 ␣ through interaction with basic residues in the catalytic domain remains a possibility. We found that blocking nascent protein synthesis using CHX diminished cPLA 2 ␣-mediated AA release in IMLF ϩ/ϩ , supporting previous reports that an unidentified, rapidly turning over protein aids in cPLA 2 ␣-mediated AA release (49 -51). CHX did not block cal-cium mobilization or translocation of cPLA 2 ␣ to Golgi, suggesting that the rapidly turning over protein plays a role in enhancing cPLA 2 ␣ activity once it has translocated to membrane.
In summary, our ability to reconstitute IMLF lacking endogenous cPLA 2 ␣ with wild type and mutant forms of functionally active, fluorescent protein-tagged cPLA 2 ␣ allowed us to directly compare for the first time translocation and AA release in response to the physiological agonist serum in the absence of endogenous wild type cPLA 2 ␣. The results demonstrate that phosphorylation of cPLA 2 ␣ and the interaction of basic residues in the catalytic domain with membrane components largely act to regulate catalytic activity and not calcium-dependent membrane binding.