Stimulation-dependent recruitment of cytosolic phospholipase A 2 -a to EA.hy.926 endothelial cell membranes leads to calcium-independent association Seema Grewal, Jennifer Smith, Sreenivasan Ponnambalam and John Walker School of Biochemistry and Molecular Biology, University of Leeds, UK Cytosolic phospholipase A 2 -a (cPLA 2 -a) is a calcium- activated enzyme involved in agonist-induced arachidonic acid release. In endothelial cells, free arachidonic acid is predominantly converted into prostacyclin, a potent vaso- dilator and inhibitor of platelet activation. As the rate-lim- iting step in prostacyclin production is the generation of free arachidonic acid by cPLA 2 -a, this enzyme has become an attractive pharmacological target and the focus of many studies. Following stimulation with calcium-mobilizing agonists, cPLA 2 -a translocates to intracellular phospholipid membranes via its C2 domain. In this study, the calcium- induced association of cPLA 2 -a with EA.hy.926 endothelial cell membranes was investigated. Subcellular fractionation and immunofluorescence studies showed that following stimulation with histamine, thrombin or the calcium ionophore A23187, cPLA 2 -a relocated to intracellular membranes. Treatment of A23187-stimulated cells with EGTA or BAPTA-AM demonstrated that a substantial pool of cPLA 2 -a remained associated with membrane frac- tions in a calcium-independent manner. Furthermore, immunofluorescence microscopy studies revealed that cells stimulated for periods of greater than 10 min showed a high proportion of calcium-independent membrane-associated cPLA 2 -a. Calcium-independent membrane association of cPLA 2 -a was not due to hydrophobic or cytoskeletal inter- actions. Finally, the recombinant C2 domain of cPLA 2 -a exhibited calcium-independent membrane binding to mem- branes isolated from A23187-stimulated cells but not those isolated from nonstimulated cells. These findings suggest that novel mechanisms involving accessory proteins at the target membrane play a role in the regulation of cPLA 2 -a. Such regulatory associations could enable the cell to dis- criminate between the varying levels of cytosolic calcium induced by different stimuli. Keywords: endothelium; cPLA 2 -a; arachidonic acid; calcium; C2 domain. Cytosolic phospholipase A 2 -a (cPLA 2 -a) belongs to a growing family of phospholipase A 2 enzymes that catalyse the hydrolysis of the sn-2 fatty-acyl bond of phospholipids to liberate free fatty acids [1]. In the endothelium, cPLA 2 -a plays a pivotal role in releasing free arachidonic acid from membrane phospholipids. This arachidonic acid is the precursor for prostacyclin, a member of the eicosanoid family of inflammatory mediators, which acts as a potent vasodilator and inhibitor of platelet aggregation [2]. As the rate-limiting step in the production of prostacyclin is the generation of arachidonic acid by cPLA 2 -a, it can be seen that cPLA 2 -a plays a crucial role in several endothelial functions such as haemostasis, angiogenesis, control of vascular tone and prevention of thrombosis formation. Consequently, cPLA 2 -a has become an attractive target for the development of novel pharmacological therapeutics against various pathological conditions [3,4]. To date, however, the exact mechanisms involved in the control of this important enzyme remain unclear. cPLA 2 -a is an 85 kDa, calcium-sensitive protein and is subject to regulation at both the transcriptional and post- translational level [5]. Early studies on the cloning, expres- sion and purification of cPLA 2 -a show that purified recombinant cPLA 2 -a binds to natural membranes in the presence of physiological concentrations of calcium [6]. More recently, studies on cultured mammalian cells have shown that cPLA 2 -a is present in the cytosol of resting cells and relocates to intracellular membranes following stimu- lation with a variety of agonists that cause an increase in cytosolic calcium levels [7–10]. In accordance with this, studies on platelets and endothelial cells have shown that this membrane relocation is consistent with an increased PLA 2 activity in membrane fractions [11,12]. Several studies have also shown that cPLA 2 -a activity in endothelial cells is regulated by phosphorylation [13–15], however the role of these modifications in the translocation and membrane association of cPLA 2 -a is unclear. The translocation of cPLA 2 -a to membrane phospho- lipids has been shown to be mediated by its calcium- dependent lipid binding or C2 domain, which promotes binding to phospholipids upon elevation of intracellular calcium concentrations [16]. C2 domains are remark- able modules present in over 100 proteins including the Correspondence to J. Walker, School of Biochemistry and Molecular Biology, University of Leeds, Leeds LS2 9JT, UK. Fax: + 44 1133433167, Tel.: + 44 1133433119, E-mail: j.h.walker@leeds.ac.uk. Abbreviations: cPLA 2 -a, cytosolic phospholipase A2 alpha. Note: A departmental web site is available at http://www.bmb. leeds.ac.uk (Received 3 September 2003, revised 27 October 2003, accepted 30 October 2003) Eur. J. Biochem. 271, 69–77 (2004) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03903.x GTPase-activating protein, phospholipase C and the syn- aptic vesicle protein, synaptotagmin [17]. The Ca 2+ -binding properties of these domains allow them to act as electro- static switches that bind phospholipid membranes in the presence of calcium without requiring large conformational changes [18]. Based on this, the binding of cPLA 2 -a to membranes is believed to be reversible, with translocation occurring only in the presence of calcium. In agreement with this, binding of cPLA 2 -a to synthetic liposomes was shown to be fully reversible, and the addition of an excess of the calcium chelator, EGTA, abolished binding [19]. Recently, however, in Chinese hamster ovary cells transfected with GFP-cPLA 2 -a, there is evidence for a translocation of cPLA 2 -a to membranes that persists even after calcium levels have returned to resting levels [20]. Here, using biochemical subfractionation and immuno- fluorescence localization studies, we investigated the relo- cation of cPLA 2 -a to membranes in EA.hy.926 endothelial cells. Our results demonstrate an association of cPLA 2 -a with endothelial cell membranes that is not consistent with a simple reversible calcium-dependent interaction of cPLA 2 -a with phospholipids. These results imply that some mechan- ism, other than simple C2-dependent association with lipids, must be involved in the regulation of cPLA 2 -a. Experimental procedures Materials Tissue culture media, enzymes and antibiotics were pur- chased from Gibco BRL (Paisley, Scotland). Goat poly- clonal antibodies to cPLA 2 -a were obtained from Santa Cruz Biotechnology Inc. (CA, USA). Secondary fluorescein isothiocyanate-conjugated secondary antibodies were from Sigma and anti-goat horseradish peroxidise-conjugated Igs were from Pierce (Cheshire, UK). All other standard reagents and chemicals were from Sigma (Poole, Dorset, UK) or BDH (Poole, Dorset, UK). Cell culture The EA.hy.926 cell line, a hybrid of human umbilical vein endothelial cells (HUVEC) and A549 human lung carci- noma epithelial cells [21], was a generous gift from C. J. Edgell (University of North Carolina, USA). Cells were cultured on plasticware at 37 °C in a humid atmosphere containing 5% (v/v) CO 2 in air. Cells were grown in Dulbecco’s modified Eagles medium supplemented with 10% fetal bovine serum, penicillin (100 UÆmL )1 ), strepto- mycin (100 lgÆmL )1 )andHAT(100l M hypoxanthine, 0.4 l M aminopterin, 16 l M thymidine). Subcellular fractionation This method was carried out as described previously [22] and cells were grown to confluence. Medium was removed and the cells were washed twice with prewarmed NaCl/P i (Dulbecco A, Oxoid Ltd, Hampshire, UK). For stimula- tions, the cells were then incubated for the appropriate time at 37 °Cwith5l M A23187 in Hepes/Tyrode’s buffer containing 1 m M CaCl 2 . Cells were then washed twice with NaCl/P i , scraped into ice-cold Buffer A (100 m M KCl, 10 m M Pipes, 1 m M NaN 3 ,1m M phenylmethanesulfonyl fluoride, 1 m M sodium orthovanadate, 50 m M benzamidine, 0.1 mgÆmL )1 leupeptin) and lysed either by freeze-thawing or by homogenization with a Dounce homogeniser. Veri- fication of lysis was performed using Trypan Blue staining (Sigma) according to the manufacturer’s instructions. The cell lysate was centrifuged at 200 000 g for 10 min at 4 °C and the resultant soluble cytosolic fraction (C1) was removed. The insoluble pellet was washed twice in Buffer A (generating cytosol washes C2 and C3) and finally solubilized in Buffer B [Buffer A containing 1% (v/v) Triton X-100]. Insoluble material was removed by centrifugation and the Triton-soluble membrane fraction (M) was collec- ted. Preparations were carried out at various free calcium concentrations by adding the appropriate amounts of CaCl 2 and EGTA to Buffer A as determined by the METLIG pro- gram [23]. Equivalent amounts of the subcellular fractions were analysed by SDS/PAGE and Western blotting. Preparation of cell cytoskeletons This method was carried out as described previously [22]. Cells were grown to confluence, washed three times with NaCl/P i and then collected by scraping into ice-cold Buffer B. Cytoskeleton fractions were isolated by centrifugation at 200 000 g for 2 h at 4 °C. The supernatant (representing cytosol and membrane proteins) was removed and the insoluble pellet (representing the cytoskeletal fraction) was solubilized in Laemmli sample buffer. Various free calcium concentrations were maintained by adding the appropriate amounts of CaCl 2 and EGTA to Buffer B, as determined by the METLIG program. Equivalent amounts of the fractions were analysed by SDS/PAGE and Western blotting. Temperature-induced phase separation of Triton X-114 The separation procedure was carried out as described in a previous report [24]. EA.hy.926 cell membrane fractions were prepared as described above. Membrane fractions were resuspended in 1% (w/v) Triton X-114, 150 m M NaCl, 10 m M Tris, 2 l M CaCl 2 ,pH7.4.Sampleswerethen incubated at 30 °C for 10 min and centrifuged at 3000 g for 3 min. The aqueous upper phase (representing hydro- philic proteins) was separated from the detergent-rich lower phase (representing hydrophobic proteins) and equivalent amounts of the two were analysed by Western blotting. Immunofluorescence microscopy The method for immunofluorescence microscopy was adapted from previous reported methods [25,26]. Cells were grown on glass coverslips overnight. Media was removed and the cells were washed three times with prewarmed (to 37 °C) NaCl/P i and fixed in prewarmed 10% (v/v) formalin in neutral buffered saline [approximately 4% (v/v) formal- dehyde, Sigma] for 5 min. All subsequent steps were performed at room temperature. After fixation, the cells were permeabilized with 0.1% (v/v) Triton X-100 in NaCl/ P i for 5 min and fixed once again for 5 min. The cells were then washed three times with NaCl/P i and incubated in freshly prepared sodium borohydride solution (1 mgÆmL )1 in NaCl/P i ) for 5 min to reduce autofluorescence. Following 70 S. Grewal et al.(Eur. J. Biochem. 271) Ó FEBS 2003 three further NaCl/P i wash steps, the cells were blocked in 5% (v/v) rabbit serum in NaCl/P i for 3 h. The cells were then incubated with primary antibody [diluted 1 : 100 into NaCl/P i , 5% (v/v) serum] overnight followed by fluorescein isothiocyanate-conjugated secondary antibody for 3 h, with eight NaCl/P i washes performed in between incubations. Sodium azide (1 m M ) was included in all incubations to prevent bacterial growth. The cells were then washed eight times with NaCl/P i and mounted onto slides in Citifluor mounting medium (Agar Scientific, Hertfordshire, UK). Confocal imaging Confocal fluorescence microscopy was performed using a Leica TCS SP spectral confocal imaging system coupled to a Leica DM IRBE inverted microscope. Each confocal section was the average of four scans to obtain optimal resolution. The system was used to generate individual sections that were 0.485 lm thick. All figures shown in this study represent 0.485 lm sections taken through the nucleus. SDS/PAGE and Western blotting Proteins (20 lg per well) were separated on SDS/polyacryl- amide gels using a discontinuous buffer system [27]. For Western blot analysis, proteins were transferred to nitrocel- lulose [28]. Subsequently, the nitrocellulose blots were blocked in 5% (w/v) nonfat milk in NaCl/P i ,0.1%(v/v) Triton X-100 for 1 h. Primary antibody incubations (1 : 1000) were carried out overnight at room temperature, followed by 1 h incubations with the appropriate horse- radish peroxidase-conjugated secondary antibody. For antigenic adsorption, the antibody was incubated with its corresponding blocking peptide (1 : 5 ratio of lg antibody to lg antigen) for 30 mins at room temperature prior to being incubated with the nitrocellulose blot. Immunoreac- tive bands were visualized using an enhanced chemilumines- cence detection kit (Pierce) according to the manufacturer’s instructions. Following this, the developed films were photographed and captured using the FujiFilm Intelligent dark Box II with the Image Reader Las-1000 package. The intensity of the bands was quantified densitometrically using the AIDA (advanced image data analyzer) 2.11 software package. In general, the average 1 and SEM from three independent experiments were calculated. C2 domain binding assays The binding assays were based on those described previ- ously [29]. Membrane fractions from nonstimulated and A23187-stimulated cells were prepared as described above. Membrane fractions (corresponding to approximately 100 lg of total protein) were incubated with 0.1 lg purified C2 domain for 30 min at 30 °C.Thesamplewasthen centrifuged at 200 000 g for 10 min at 4 °C to sediment the membranes. Any unbound C2 domain in the supernatant was removed whilst any bound material was solubilized in Buffer B. Insoluble material was removed by centri- fugation and the soluble membrane fraction was collected. Samples were analysed by SDS/PAGE and Western blotting. Results CPLA 2 -a relocates to intracellular membranes following an elevation in cytosolic calcium concentration The subcellular location of cPLA 2 -a in EA.hy.926 endo- thelial cells was investigated by immunofluorescence micro- scopy using an antibody that specifically recognizes the a-isoform of cPLA 2 (Fig. 1A, lane 1). Antigenic adsorption of this antibody with the appropriate blocking peptide abolished detection of cPLA 2 -a by both Western blotting (Fig. 1A, lane 2) and immunofluorescence microscopy (data not shown). Using this specific antibody, a comparison of Fig. 1. Relocation of cPLA 2 -a to intracellular membranes following A23187-stimulation. (A) cPLA 2 -a was detected by Western blotting of EA.hy.926 lysates (20 lgprotein)usingagoat polyclonal antibody (i). Also shown are con- trol lanes corresponding to antigen-adsorbed antibody (ii), and horseradish peroxidase conjugated anti-(goat IgG) controls (iii). (B) Cells were grown on coverslips and incubated with buffer alone (i) or stimulated with 5 l M A23187 (ii), 10 l M histamine (iii) or 1 UÆmL )1 thrombin (iv) in the presence of 1 m M extra- cellular calcium for 1 min. Cells were then fixed and permeabilized, and cPLA 2 -a was detected using immunofluorescence micros- copy. Scale bar, 10 lm. Ó FEBS 2003 Calcium-independent binding of cPLA2-a (Eur. J. Biochem. 271)71 the location of cPLA 2 -a in resting and stimulated EA.hy.926 cells was carried out. In nonstimulated cells, cPLA 2 -a was present throughout the cytosol and the nucleus (Fig. 1B, panel i). Following elevation of the cytosolic calcium concentration in response to the physio- logical stimulus histamine or thrombin, or the calcium ionophore, A23187, a specific relocation of cPLA 2 -a to intracellular membranes resembling the endoplasmic reti- culum and nuclear envelope was evident (Fig. 1B, panels ii–iv). Both secondary antibody and peptide-adsorbed antibody controls gave no staining (data not shown) confirming that the staining observed corresponded specifi- cally to cPLA 2 -a. Measurement of intracellular calcium concentrations using Fura-2-AM demonstrated that expo- sure to either 10 l M histamine, 1 UÆmL )1 thrombin or 5 l M A23187 in the presence of 1 m M extracellular calcium led to an increase in cytosolic calcium concentration from a resting value of 100 n M to approximately 1–2 l M (data not shown). These values were consistent with those obtained from other studies on endothelial cells [30,31]. Based on these findings, future experiments were performed primarily with A23187, to avoid the complication of agonist-specific signalling events. To investigate the calcium-dependency of relocation further, fractionations were performed under the resting and elevated calcium levels observed in endothelial cells. Firstly, cytosol and membrane fractions were obtained from resting cells that were lysed in the presence of various free calcium concentrations. Analysis of the resultant samples indicated that, with increasing concentrations of free calcium, an increasing amount of cPLA 2 -a was found associated with membranes (Fig. 2A). In the complete absence of calcium, no cPLA 2 -a was present in the membrane fraction. In contrast, all of the endogenous cPLA 2 -a was found to be membrane-bound at calcium concentrations of 800 n M or above (Fig. 2). Subfractionation experiments were also performed in the presence of either 100 n M free calcium for resting cells or 2 l M free calcium following A23187 stimulation. These calcium concentrations were representative of the intracel- lular cytosolic calcium concentration in the absence and presence of stimulation, respectively (as described above). The results of these fractionation studies (Fig. 3A) indicated that under resting calcium levels of 100 n M ,cPLA 2 -a was predominantly cytosolic. In contrast, when cells were stimulated with 5 l M A23187 for 10 min and fractionated in the presence of 2 l M calcium, most of the cPLA 2 -a was membrane-bound. The cytosolic location of lactate dehy- drogenase confirmed that cells were lysed sufficiently. Quantification of the relative levels of cPLA 2 -a in the fractions showed that in resting cells 79.9% ± 6.9 of the total amount was present in the cytosol with only 19.1% ± 6.0 present in the membrane fraction. In contrast, following stimulation only 28.5% ± 1.9 remained in the cytosolic fraction whereas 71.5% ± 1.9 was found to be membrane-associated (Fig. 3B). The association of cPLA 2 -a with membranes is EGTA- resistant and time-dependent To further characterize the binding of cPLA 2 -a to endo- thelial cell membranes, the effects of the calcium chelator EGTA on membrane association were studied. Cells were stimulated and fractionated as above, and membrane fractions were washed with 5 m M EGTA. The results (Fig. 4) show that, as demonstrated above, only approxi- mately 30% of the total amount of cPLA 2 -a remained cytosolic following stimulation with A23187 and homo- genization in the presence of 2 l M Ca 2+ . 2 Remarkably, however, only 14.8% ± 2.6 of the total protein could be eluted from the membrane pellet by washing with EGTA, with greater than half the amount of total cPLA 2 -a (52.5% ± 4.65) remaining tightly associated with the membrane in a manner that resisted extraction with EGTA. Similar results were seen following treatment of cells with 10 l M histamine or 1 UÆmL )1 thrombin with minimal loss of cPLA 2 -a from stimulated membrane fractions following washing with EGTA (Fig. 4C). The effects of calcium chelation on the subcellular location of cPLA 2 -a were examined using immunofluores- cence microscopy. Cells were stimulated with 5 l M A23187 for various time periods. To test the effects of reducing cytosolic calcium levels, cells were stimulated in the same way then the extracellular and intracellular calcium chela- tors, EGTA and BAPTA-AM respectively, were added to the cells 3 . After a 1 min stimulation period followed by calcium chelation, a cytosolic staining pattern resembling Fig. 2. Calcium dependency of membrane binding. (A) Cells were grown to confluence in flasks, and scraped into Buffer A containing the free calcium levels indicated. The cells were homogenized and fract- ionated into cytosol and membrane. Fractions were separated by SDS/ PAGE and Western blotted, and cPLA 2 -a was detected. (B) The rel- ative amount of cPLA 2 -a in each fraction was quantified and expressed as a percentage of the total amount of cPLA 2 -a. The amounts in the respective cytosol and membrane fractions were plotted against the corresponding calcium concentration. The data is representative of results obtained from three independent experiments. 72 S. Grewal et al.(Eur. J. Biochem. 271) Ó FEBS 2003 that of a nonstimulated cell was evident (Fig. 5A). In contrast, cells stimulated for 10 min showed a high proportion of membrane-relocated cPLA 2 -a that was resistant to the calcium chelation. Consistent with this, subfractionation of cells directly into EGTA following A23187-stimulation showed that increased stimulation time led to an increase in the EGTA-resistant pool of membrane- bound cPLA 2 -a (Fig. 5B). Under these conditions, more than 20% of the total cPLA 2 -a pool was found to be EGTA-resistant when directly solubilized in a Triton/ EGTA buffer following a 10 min stimulation period (Fig. 5C). EGTA-resistant membrane binding is not dependent on a change in hydrophobicity or the cytoskeletal association of cPLA 2 -a It was possible that the binding of calcium to the C2 domain results in a change in the overall hydrophobicity of the protein, allowing it to partially insert itself into the lipid bilayer. To address this question, temperature-induced phase separation of Triton X-114 was performed. A solution of Triton X-114 is homogenous at temperatures below 20 °C. Above this temperature, the solution separates into an aqueous phase and a detergent phase. Previous studies have shown that integral membrane proteins and proteins with exposed hydrophobic regions partition into the detergent phase [21]. Analysis of cPLA 2 -a in membrane fractions prepared from resting and stimulated cells showed that no change in hydrophobicity occurred, and all the protein was exclusively in the aqueous phase of a Triton X-114 solution (Fig. 6A). The possibility that the observed EGTA-resistant binding of cPLA 2 -a to membranes was due to a cytoskeletal interaction was investigated. Cytoskeleton fractions were isolated by direct solubilization and sedi- mentation from nonstimulated and A23187-stimulated cells. The results from these studies showed that there was no association of cPLA 2 -a with the cytoskeletal pellet in either resting cells isolated in EGTA or 100 n M calcium, or in A23187-stimulated cells isolated in 2 l M calcium (Fig. 6B). Fig. 3. Calcium-induced relocation of cPLA 2 -a. (A) Resting EA.hy.926 cells scraped into 100 n M free calcium buffer, or A23187-stimulated cells (10 min at 37 °C) scraped into 2 l M free calcium buffer were homogenized and subfractionated into cytosolic (C) and membrane (M) fractions, including intermediate wash steps (C2, C3 and M2). Samples, including total lysates (T), were separated by SDS/PAGE and Western blotted, and cPLA 2 -a was detected. The distribution of the cytosolic marker, lactate dehydrogenase (LDH) in resting cells was also determined. (B) Quantification of the amount of cPLA 2 -a present in each of the indicated fractions, expressed as a percentage of the total amount of cPLA 2 -a (± SEM, n ¼ 3). Fig. 4. EGTA-resistant binding of cPLA 2 -a to EA.hy.926 cell mem- branes. (A) Cells were stimulated with 5 l M A23187 for 10 min in the presence of 1 m M extracellular calcium. Cells were then scraped into lysis buffer containing 2 l M free calcium, homogenized and subfract- ionated. Following removal of the cytosolic fraction, the remaining pellet containing membrane proteins was washed twice in lysis buffer containing 5 m M EGTA. The remaining pellet was solubilized in EGTA/Triton X-100 to give the membrane fraction. The samples were then immunoblotted to detect cPLA 2 -a. (B) Quantification of the amount of cPLA 2 -a present in each of the indicated fractions, expressed as a percentage of the total amount of cPLA 2 -a (± SEM, n ¼ 3). (C) Subcellular fractionation was also carried out following 10 min stimulation with 1 UÆmL )1 thrombin and 10 l M histamine, as described above. The amount of cPLA 2 -a in the samples was detected by Western blotting. Ó FEBS 2003 Calcium-independent binding of cPLA2-a (Eur. J. Biochem. 271)73 The C2 domain of cPLA 2 -a demonstrates calcium-independent binding to membranes To determine whether the C2 domain alone was able to confer EGTA-resistant membrane binding, in vitro binding studies of purified recombinant C2 domain to EA.hy.926 cell membranes were performed. The results of these studies (Fig. 7A) show that the calcium dependency of binding of purified C2 domain to membranes prepared from nonstim- ulated cells is identical to that of the binding of the endogenous protein. To determine whether this binding was reversible, membrane fractions containing the bound C2 domain were washed with EGTA. As expected, the C2 domain could be removed from nonstimulated membrane fractions by washing with 5 m M EGTA, and was found exclusively in the EGTAwashfraction (Fig. 7B). To examine whether any changes occurred in the membrane following stimulation, studies were also performed using membrane fractions isolated from A23187-stimulated cells (in the pres- ence of 2 l M free calcium). Using this approach, remarkable differences in the binding properties of the C2 domain were observed (Fig. 7B). In contrast to the data shown above, only a small proportion of the C2 was able to bind to the membranes, resulting in a large pool that remained in the soluble fraction. Furthermore, these studies showed that the protein that did bind could not be removed from the membrane by EGTA washing, hence was found tightly associated with the EGTA-resistant membrane fraction. Discussion To date, the association of cPLA 2 -a with cellular mem- branes has been attributed to the calcium-dependent binding of its C2 domain to membrane phospholipids. This domain promotes the reversible binding of proteins to phospholipids in the presence of calcium. The results shown here, however, demonstrate a novel mode of binding of cPLA 2 -a to EA.hy.926 cell membranes in a manner that resists extraction with the calcium chelator EGTA. Using subcellular fractionation experiments, it was observed that endogenous cPLA 2 -a binds to EA.hy.926 cell membranes in a calcium-dependent manner. At con- centrations below 200 n M the protein was largely cytosolic, whereas it was completely membrane-associated at physio- logically elevated calcium concentrations of 800 n M and above. In resting endothelial cells, the basal levels of arachidonic acid release and prostacyclin production [32] imply that a pool of cPLA 2 -a is constitutively membrane- associated and catalytically active. Not surprisingly there- fore a small proportion of cPLA 2 -a was found to be associated with a nonstimulated membrane fraction. The exact role and nature of this constitutively membrane- associated cPLA 2 -a requires further investigation. Most interestingly, more than 50% of the total cellular pool of cPLA 2 -a relocated to membranes following stimu- lation with A23187 and remained associated with a membrane fraction even after extraction with EGTA. Similar results were also seen using 10 l M histamine or 1UÆmL )1 thrombin (Fig. 4). Immunofluorescence micros- copy also confirmed that membrane-relocated cPLA 2 -a remained associated with membranes even in the presence of intracellular and extracellular calcium chelators. Fur- thermore, the amount of cPLA 2 -a present in the EGTA- resistant membrane fraction was seen to increase with stimulation time, suggesting that prolonged activation leads to membrane association that resists extraction by the removal of calcium. These findings are consistent with those published by Hirabayashi and coworkers [20] which suggested that stimulation periods of less than 2 min caused only partial activation and reversible relocation of cPLA 2 -a Fig. 5. Effects of EGTA and BAPTA-AM on the relocation of cPLA 2 -a. Cells were stimulated with 5 l M A23187 in the presence of 1 m M extracellular calcium for the times indicated. (A) Cells were fixed and permeabilized, and cPLA 2 -a was detected by fluorescence microscopy. For EGTA/BAPTA-AM treatment, cells were stimulated and processed as above. However, following stimulation, cells were washed with 5 m M EGTA/5 m M BAPTA-AM for 5 mins directly prior to fixation. Scale bar, 5 lm. (B) Cells were treated as indicated, scraped into a 5 m M EGTA buffer, homogenized and subfractionated into cytosolic and membrane fractions and cPLA 2 -a was detected by immunoblotting. (C) Quantification oftheamountofcPLA 2 -a present in each of the indicated fractions, expressed as a percentage of the total amount of cPLA 2 -a (± SEM, n ¼ 3). 74 S. Grewal et al.(Eur. J. Biochem. 271) Ó FEBS 2003 in CHO cells, whereas longer stimulations caused binding that persisted even after reduction of cytosolic levels of calcium to resting values. This may be a mechanism for allowing the cell to discriminate appropriate signals from small transient fluctuations in intracellular calcium concen- trations. Hence, once the calcium transient exceeds a critical level, a tight-binding state of cPLA 2 -a couldleadtoa continuous membrane localization and arachidonic acid production for prolonged periods, even after the calcium levels return to their resting value. A recently published study also demonstrates a prolonged ionophore-stimulated, perinuclear membrane association of wild type cPLA 2 -a for several minutes after the return of intracellular calcium to unstimulated levels [33]. This phenomenon was seen to be dependent on the phosphory- lation of S505, which enhanced the hydrophobic interaction of catalytic domain residues to membrane phospholipids. In support of this, Evans and colleagues [34] demonstrated that full-length cPLA 2 -a dissociated more slowly from membranes than the C2-domain alone, also indicating that the catalytic domain may be involved in prolonged mem- brane binding. In the present study, however, no change in the overall hydrophobicity of cPLA 2 -a in ionophore-treated cells was observed by the Triton X-114 phase separation method (Fig. 6). However it may be possible that this method is insufficiently sensitive to detect subtle changes in hydrophobicity. Analysis of the calcium dependency of binding of the purified C2 domain of cPLA 2 -a to EA.hy.926 membranes demonstrated that it exhibited calcium-dependent mem- brane binding properties identical to that of the endogenous full-length protein. Interestingly, it was observed that recombinant C2 domain could bind to nonstimulated membranes in a reversible manner, whereas binding to stimulated membranes was EGTA-resistant. Most import- antly, it was noticed that only partial binding of the C2 domain to stimulated membranes occurred. This raises the possibility that under stimulated conditions, the binding site Fig. 7. In vitro binding of pure re-folded C2 domain to to EA.hy.926 cell membranes. (A) Cells were grown to confluence in flasks, scraped into EGTA buffer and fractionated into cytosol and membrane fractions. Membrane fractions were incubated with 0.1 lgpureC2domainat 30 °C for 30 min in the presence of the indicated levels of free calcium. Following centrifugation at 200 000 g any unbound protein (soluble) was collected and the pellet (membrane) was solubilized. (B) Cell membranes were prepared from resting (in EGTA) and stimulated (5 l M A23187 for 10 min, in 2 l M calcium) cells. Membranes were incubated with 0.1 lg pure C2 domain at 30 °Cfor30minsinthe presence of 2 l M free calcium. Following centrifugation at 200 000 g any unbound protein (soluble) was collected and the membrane pellet was washed in 5 m M EGTA. Following a further centrifugation step, the EGTA-elutable fraction was collected (EGTA wash) and the EGTA-resistant cell pellet (membrane) was solubilized in Triton X-100. Fractions were analysed by Western blotting using a mouse anti-(cPLA 2 -a)mAb. Fig. 6. Temperature-induced phase separation of Triton X-114 and extraction of EA.hy.926 cytoskeletons. (A) Cytosol and membrane fractions from nonstimulated cells and A23187-stimulated cells were prepared in the presence of 100 n M and 2 l M free calcium levels, respectively. The membrane fractions were resuspended in 1% (v/v) Triton X-114 and temperature-induced phase separation was per- formed. The aqueous and detergent phases were separated and made up to equal volumes. Samples were immunoblotted for cPLA 2 -a.(B) Triton X-100 soluble (S, representing cytosol and membranes) and insoluble (P, representing cytoskeleton) fractions were prepared from nonstimulated cells (in EGTA or 100 n M free calcium levels) and cells stimulated with A23187 for 10 mins (in 2 l M free calcium). Equivalent amounts of the fractions were immunoblotted for cPLA 2 -a. Ó FEBS 2003 Calcium-independent binding of cPLA2-a (Eur. J. Biochem. 271)75 for cPLA 2 -a may be partially blocked or saturated by endogenous cPLA 2 -a. These findings imply that, following stimulation, the membrane fraction undergoes a change that allows the anomalous binding of cPLA 2 -a. It is possible that this may be due to a change in the protein or lipid composition, implying that some other protein or lipid interaction is involved in the EGTA-resistant binding of cPLA 2 -a to membranes. Previous studies have shown that cPLA 2 -a is able to bind to ceramide, cholesterol and phosphatidylinositol 4,5-bisphosphate [35–37] thus it is possible that such interactions mediate the calcium-inde- pendent membrane associations observed here. This, and the possible involvement of binding proteins, is further supported by the observation that cPLA 2 -a relocates to specific cellular membranes indicating that a specific mech- anism for targeting is present. In particular there is no relocation of cPLA 2 -a to the plasma membrane whereas C2 domains in other proteins (e.g. protein kinase C-c [38]), result in these proteins moving exclusively from the cytosol to the plasma membrane. Overexpression of the C2 domain of cPLA 2 -a alone [8] or GFP–C2–cPLA 2 -a fusion proteins [34,39] demonstrate that this truncated protein exhibits the same relocation patterns as the full-length protein, indica- ting that the targeting information or mechanism lies within this region of the protein. C2 domains are also known to mediate protein–protein interactions hence the presence of a C2domainincPLA 2 -a further supports the possibility that an accessory protein is involved in the regulation of cPLA 2 -a. Previous far Western studies identified the inter- mediate filament protein vimentin as an adaptor protein that interacts with the C2 domain of cPLA 2 -a in a calcium- dependent manner [40]. Whether vimentin plays a role in the calcium-independent association of cPLA 2 -a with phos- pholipids remains to be investigated. Furthermore, a grow- ing body of evidence supports the functional coupling of cPLA 2 -a to its downstream cyclo-oxygenase enzymes, COX-1 and COX-2 [41,42]. It is possible that these enzymes, which show similar subcellular localization [43,44], may specifically interact with cPLA 2 -a and act as accessory proteins. The identification and characterization of these and/or other binding partners or adapter proteins would give further insight into the novel mechanism of cPLA 2 -a regulation identified here. In conclusion, it has been demonstrated here that cPLA 2 - a relocates to cellular membranes following elevations in cytosolic free calcium concentration; however, it is able to remain tightly associated with the membrane in a calcium- independent manner. Prolonged association with mem- branes, despite a return of cytosolic calcium to resting levels, could be of physiological significance in prolonging arachi- donate production in response to cell stimulation. These results indicate that this novel binding is not due simply to the calcium-dependent lipid binding capacity of the C2 domain, and that some other binding partner or accessory protein may be involved in the regulation of cPLA 2 -a. Acknowledgements This work was funded by the British Heart Foundation and the BBSRC. We thank Dr C. J. Edgell for the gift of the EA.hy.926 cells, Dr R. Williams for purified C2 domain and Dr E. E. Morrison for assistance with confocal imaging. References 1. Dennis, E.A. (1997) The growing phospholipase A2 superfamily of signal transduction enzymes. Trends Biochem. Sci. 22, 1–2. 2. Vane, J.R., Anggard, E.E. & Botting, R.M. (1990) Regulatory functions of the vascular endothelium. N.Engl.J.Med.323, 27–36. 3. Balsinde, J., Balboa, M.A., Insel, P.A. & Dennis, E.A. (1999) Regulation and inhibition of phospholipase A2. Annu. Rev. Pharmacol Toxicol. 39, 175–189. 4. Yedgar, S., Lichtenberg, D. & Schnitzer, E. (2000) Inhibition of phospholipase A2 as a therapeutic target. Biochim. Biophys. Acta 1488, 182–187. 5. Clark, J.D., Schievella, A.R., Nalefski, E.A. & Lin, L.L. (1995) Cytosolic phospholipase A2. J. Lipid Mediat. Cell Signal. 12, 83–117. 6. Clark, J.D., Lin, L.L., Kriz, R.W., Ramesha, C.S., Sultzman, L.A., Lin, A.Y., Milona, N. & Knopf, J.L. (1991) A novel ara- chidonic acid-selective cytosolic PLA2 contains a Ca 2+ -dependent translocation domain with homology to PKC and GAP. Cell 65, 1043–1051. 7. Glover, S., de Carvalho, M.S., Bayburt, T., Jonas, M., Chi, E., Leslie, C.C. & Gelb, M.H. (1995) Translocation of the 85-kDa phospholipase A2 from cytosol to the nuclear envelope in rat basophilic leukemia cells stimulated with calcium ionophore or IgE/antigen [published erratum appears in J. Biol. Chem. 1995 September 1; 270 (35): 20870], J. Biol. Chem. 270, 15359– 15367. 8. Schievella, A.R., Regier, M.K., Smith, W.L. & Lin, L.L. (1995) Calcium-mediated translocation of cytosolic phospholipase A2 to the nuclear envelope and endoplasmic reticulum. J. Biol. Chem. 270, 30749–30754. 9. Peters-Golden, M., Song, K., Marshall, T. & Brock, T. (1996) Translocation of cytosolic phospholipase A2 to the nuclear envelope elicits topographically localized phospholipid hydrolysis. Biochem. J. 318, 797–803. 10. Sierra-Honigmann, M.R., Bradley, J.R. & Pober, J.S. (1996) ÔCytosolicÕ phospholipase A2 is in the nucleus of subconfluent endothelial cells but confined to the cytoplasm of confluent endothelial cells and redistributes to the nuclear envelope and cell junctions upon histamine stimulation. Laboratory Invest. 74, 684–695. 11. Kramer, R.M., Roberts, E.F., Manetta, J.V., Hyslop, P.A. & Jakubowski, J.A. (1993) Thrombin-induced phosphorylation and activation of Ca 2+ -sensitive cytosolic phospholipase A2 in human platelets. J. Biol. Chem. 268, 26796–26804. 12. SaG., Murugesan, G., Jaye, M., Ivashchenko, Y. & Fox, P.L. (1995) Activation of cytosolic phospholipase A2 by basic fibro- blast growth factor via a p42 mitogen-activated protein kinase- dependent phosphorylation pathway in endothelial cells. J. Biol. Chem. 270, 2360–2366. 13. Wheeler-Jones, C., Abu-Ghazaleh, R., Cospedal, R., Houliston, R.A., Martin, J. & Zachary, I. (1997) Vascular endothelial growth factor stimulates prostacyclin production and activation of cyto- solic phospholipase A2 in endothelial cells via p42/p44 mitogen- activated protein kinase. FEBS Lett. 420, 28–32. 14. Gliki, G., Abu-Ghazaleh, R., Jezequel, S., Wheeler-Jones, C. & Zachary, I. (2001) Vascular endothelial growth factor-induced prostacyclin production is mediated by a protein kinase C (PKC)- dependent activation of extracellular signal-regulated protein kinases 1 and 2 involving PKC-d and by mobilization of intra- cellular Ca 2+ . Biochem. J. 353, 503–512. 15. Houliston, R.A., Pearson, J.D. & Wheeler-Jones, C.P. (2001) Agonist-specific cross talk between ERKs and p38 (MAPK) reg- ulates PGI2 synthesis in endothelium. Am. J. Physiol. Cell Physiol. 281, C1266–C1276. 76 S. Grewal et al.(Eur. J. Biochem. 271) Ó FEBS 2003 16. Gijon,M.A.,Spencer,D.M.,Kaiser,A.L.&Leslie,C.C.(1999) Role of phosphorylation sites and the C2 domain in regulation of cytosolic phospholipase A2. J. Cell Biol. 145, 1219–1232. 17. Rizo, J. & Sudhof, T.C. (1998) C2-domains, structure and func- tion of a universal Ca 2+ -binding domain. J. Biol. Chem. 273, 15879–15882. 18. Davletov, B., Perisic, O. & Williams, R.L. (1998) Calcium- dependent membrane penetration is a hallmark of the C2 domain of cytosolic phospholipase A2 whereas the C2A domain of synaptotagmin binds membranes electrostatically. J. Biol. Chem. 273, 19093–19096. 19. Nalefski, E.A., Sultzman, L.A., Martin, D.M., Kriz, R.W., Towler, P.S., Knopf, J.L. & Clark, J.D. (1994) Delineation of two functionally distinct domains of cytosolic phospholipase A2, a regulatory Ca 2+ -dependent lipid-binding domain and a Ca 2+ -independent catalytic domain. J. Biol. Chem. 269, 18239– 18249. 20. Hirabayashi, T., Kume, K., Hirose, K., Yokomizo, T., Iino, M., Itoh, H. & Shimizu, T. (1999) Critical duration of intracellular Ca 2+ response required for continuous translocation and activa- tion of cytosolic phospholipase A2. J. Biol. Chem. 274, 5163–5169. 21. Edgell, C.J., McDonald, C.C. & Graham, J.B. (1983) Permanent cell line expressing human factor VIII-related antigen established by hybridization. Proc.NatlAcad.Sci.USA80, 3734–3737. 22. Koster, J.J., Boustead, C.M., Middleton, C.A. & Walker, J.H. (1993) The sub-cellular localization of annexin V in cultured chick- embryo fibroblasts. Biochem. J. 291, 595–600. 23. Denton, R.M., Richards, D.A. & Chin, J.G. (1978) Calcium ions and the regulation of NAD+-linked isocitrate dehydrogenase from the mitochondria of rat heart and other tissues. Biochem. J. 176, 899–906. 24. Bordier, C. (1981) Phase separation of integral membrane proteins in Triton X-114 solution. J. Biol. Chem. 256, 1604–1607. 25. Barwise, J.L. & Walker, J.H. (1996) Annexins II, IV, V and VI relocate in response to rises in intracellular calcium in human foreskin fibroblasts. J. Cell Sci. 109, 247–255. 26. Heggeness, M.H., Wang, K. & Singer, S.J. (1977) Intracellular distributions of mechanochemical proteins in cultured fibroblasts. Proc.NatlAcad.Sci.USA74, 3883–3887. 27. Laemmli, U.K. (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227, 680–685. 28. Towbin, H., Staehelin, T. & Gordon, J. (1979) Electrophoretic transfer of proteins from polyacrylamide gels to nitrocellulose sheets: procedure and some applications. Proc. Natl Acad. Sci. USA 76, 4350–4354. 29. Chow, A., Davis, A.J. & Gawler, D.J. (1999) Investigating the role played by protein-lipid and protein–protein interactions in the membrane association of the p120GAP CaLB domain. Cell Signal. 11, 443–451. 30. McIntyre,T.M.,Zimmerman,G.A.,Satoh,K.&Prescott,S.M. (1985) Cultured endothelial cells synthesize both platelet- activating factor and prostacyclin in response to histamine, bradykinin, and adenosine triphosphate. J. Clin. Invest. 76, 271–280. 31. Weksler, B.B., Ley, C.W. & Jaffe, E.A. (1978) Stimulation of endothelial cell prostacyclin production by thrombin, trypsin, and the ionophore A23187. J. Clin. Invest. 62, 923–930. 32. Suggs, J.E., Madden, M.C., Friedman, M. & Edgell, C.J. (1986) Prostacyclin expression by a continuous human cell line derived from vascular endothelium. Blood 68, 825–829. 33. Das, S., Rafter, J.D., Kim, K.P., Gygi, S.P. & Cho, W. (2003) Mechanism of group IVA cytosolic phospholipase A2 activation by phosphorylation. J. Biol. Chem. 5 278, 41431–41442. 34. Evans, J.H., Spencer, D.M., Zweifach, A. & Leslie, C.C. (2001) Intracellular calcium signals regulating cytosolic phospholipase A2 translocation to internal membranes. J. Biol. Chem. 276, 30150–30160. 35. Huwiler, A., Johansen, B., Skarstad, A. & Pfeilschifter, J. (2001) Ceramide binds to the CaLB domain of cytosolic phospholipase A2 and facilitates its membrane docking and arachidonic acid release. FASEB J. 15, 7–9. 36. Klapisz, E., Masliah, J., Bereziat, G., Wolf, C. & Koumanov, K.S. (2000) Sphingolipids and cholesterol modulate membrane sus- ceptibility to cytosolic phospholipase A2. J. Lipid Res. 41, 1680–1688. 37. Balsinde, J., Balboa, M.A., Li, W.H., Llopis, J. & Dennis, E.A. (2000) Cellular regulation of cytosolic group IV phospholipase A2 by phosphatidylinositol bisphosphate levels. J. Immunol. 164, 5398–5402. 38. Sakai, N., Sasaki, K., Ikegaki, N., Shirai, Y., Ono, Y. & Saito, N. (1997) Direct visualization of the translocation of the gamma- subspecies of protein kinase C in living cells using fusion proteins with green fluorescent protein. J. Cell. Biol. 139, 1465–1476. 39. Perisic, O., Paterson, H.F., Mosedale, G., Lara-Gonzalez, S. & Williams, R.L. (1999) Mapping the phospholipid-binding surface and translocation determinants of the C2 domain from cytosolic phospholipase A2. J. Biol. Chem. 274, 14979–14987. 40. Nakatani, Y., Tanioka, T., Sunaga, S., Murakami, M. & Kudo, I. (2000) Identification of a cellular protein that functionally inter- acts with the C2 domain of cytosolic phospholipase A2a. J. Biol. Chem. 275, 1161–1168. 41. Croxtal, J.D., Newman, S.P., Choudhury, Q. & Flower, R.J. (1996) The concerted regulation of cPLA2, COX2, and lipocortin 1 expression by IL-1b in A549 cells. Biochem. Biophys. Res. Commun. 220, 491–495. 42. Murakami, M., Shimbara, S., Kambe, T., Kuwata, H., Winstead, M.V., Tischfield, J.A. & Kudo, I. (1998) The functions of five distinct mammalian phospholipase A2S in regulating arachidonic acid release. Type IIa and type V secretory phospholipase A2S are functionally redundant and act in concert with cytosolic phos- pholipase A2. J. Biol. Chem. 273, 14411–14423. 43. Morita, I., Schindler, M., Regier, M.K., Otto, J.C., Hori, T., DeWitt, D.L. & Smith, W.L. (1995) Different intracellular loca- tions for prostaglandin endoperoxide H synthase-1 and -2. J. Biol. Chem. 270, 10902–10908. 44. Grewal, S., Ponnambalam, S. & Walker, J.H. (2003) Association of cPLA2-a and COX-1 with the Golgi apparatus of A549 human lung epithelial cells. J. Cell Sci. 116, 2303–2310. Ó FEBS 2003 Calcium-independent binding of cPLA2-a (Eur. J. Biochem. 271)77 . Stimulation-dependent recruitment of cytosolic phospholipase A 2 -a to EA. hy. 926 endothelial cell membranes leads to calcium-independent association Seema Grewal, Jennifer. (1996) Cytosolic phospholipase A2 is in the nucleus of subconfluent endothelial cells but confined to the cytoplasm of confluent endothelial cells and redistributes to the nuclear envelope and cell junctions. 7. In vitro binding of pure re-folded C2 domain to to EA. hy. 926 cell membranes. (A) Cells were grown to confluence in flasks, scraped into EGTA buffer and fractionated into cytosol and membrane fractions. Membrane